River Restoration Managing the Uncertainty in Restoring ... - Inecol

River Restoration
Managing the Uncertainty in
Restoring Physical Habitat
Editors
Stephen Darby
School of Geography, University of Southampton, UK
and
David Sear
School of Geography, University of Southampton, UK

River Restoration

River Restoration
Managing the Uncertainty in
Restoring Physical Habitat
Editors
Stephen Darby
School of Geography, University of Southampton, UK
and
David Sear
School of Geography, University of Southampton, UK

Copyright © 2008 John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester,
West Sussex PO19 8SQ, England
Telephone (+44) 1243 779777
Email (for orders and customer service enquiries): cs-books@wiley.co.uk
Visit our Home Page on www.wileyeurope.com or www.wiley.com
All Rights Reserved. No part of this publication may be reproduced, stored in a retrieval system or transmitted in any form or by any
means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms of the Copyright, Designs and
Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road, London W1T
4LP, UK, without the permission in writing of the Publisher. Requests to the Publisher should be addressed to the Permissions Department,
John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England, or emailed to permreq@wiley.co.uk, or
faxed to (+44) 1243 770620.
Designations used by companies to distinguish their products are often claimed as trademarks. All brand names and product names used in
this book are trade names, service marks, trademarks or registered trademarks of their respective owners. The Publisher is not associated
with any product or vendor mentioned in this book.
This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the
understanding that the Publisher is not engaged in rendering professional services. If professional advice or other expert assistance is
required, the services of a competent professional should be sought.
The Publisher and the Author make no representations or warranties with respect to the accuracy or completeness of the contents of this
work and specifi cally disclaim all warranties, including without limitation any implied warranties of fi tness for a particular purpose. The
advice and strategies contained herein may not be suitable for every situation. In view of ongoing research, equipment modifi cations,
changes in governmental regulations, and the constant fl ow of information relating to the use of experimental reagents, equipment, and
devices, the reader is urged to review and evaluate the information provided in the package insert or instructions for each chemical, piece
of equipment, reagent, or device for, among other things, any changes in the instructions or indication of usage and for added warnings and
precautions. The fact that an organization or Website is referred to in this work as a citation and/or a potential source of further information
does not mean that the author or the publisher endorses the information the organization or Website may provide or recommendations it
may make. Further, readers should be aware that Internet Websites listed in this work may have changed or disappeared between when this
work was written and when it is read. No warranty may be created or extended by any promotional statements for this work. Neither the
Publisher nor the Author shall be liable for any damages arising herefrom.
Other Wiley Editorial Offi ces
John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA
Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA
Wiley-VCH Verlag GmbH, Boschstr. 12, D-69469 Weinheim, Germany
John Wiley & Sons Australia Ltd, 42 McDougall Street, Milton, Queensland 4064, Australia
John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop #02-01, Jin Xing Distripark, Singapore 129809
John Wiley & Sons Ltd, 6045 Freemont Blvd, Mississauga, Ontario L5R 4J3, Canada
Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books.
Library of Congress Cataloging-in-Publication Data
Darby, Stephen.
River restoration : managing the uncertainty in restoring physical habitat / Stephen Darby and David Sear.
p. cm.
Includes bibliographical references and index.
ISBN 978-0-470-86706-8 (cloth)
1. Stream restoration. 2. Riparian restoration. I. Sear, David (David A.) II. Title.
TC409.D28 2008
627′.12–dc22
2007030210
British Library Cataloguing in Publication Data
A catalogue record for this book is available from the British Library
ISBN-13 978-0-470-86706-8 (HB)
Typeset in 9/11 pt Times by SNP Best-set Typesetter Ltd., Hong Kong
Printed and bound in Great Britain by Antony Rowe Ltd, Chippenham, Wiltshire

Contents
Preface vii
List of Contributors xi
Section I Introduction: The Nature and Signifi cance of Uncertainty in
River Restoration
1 Uncertainty in River Restoration 3
J. Lemons and R. Victor
2 Sources of Uncertainty in River Restoration Research 15
W. L. Graf
3 The Scope of Uncertainties in River Restoration 21
J.M. Wheaton, S.E. Darby and D.A. Sear
Section II Planning and Designing Restoration Projects
4 Planning River Restoration Projects: Social and Cultural Dimensions 43
G.M. Kondolf and C-N. Yang
5 Conceptual and Mathematical Modelling in River Restoration: Do We Have
Unreasonable Confi dence? 61
M. Stewardson and I. Rutherfurd
6 Uncertainty in Riparian and Floodplain Restoration 79
F.M.R. Hughes, T. Moss and K.S. Richards
7 Hydrological and Hydraulic Aspects of Restoration Uncertainty for Ecological Purposes 105
N.J. Clifford, M.C. Acreman and D.J. Booker
8 Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 139
M.R. Perrow, E.R. Skeate, D. Leeming, J. England and M.L. Tomlinson
Section III The Construction and Post-Construction Phases
9 Constructing Restoration Schemes: Uncertainty, Challenges and Opportunities 167
J. Mant, R. Richardson and M. Janes

vi Contents
10 Measures of Success: Uncertainty and Defi ning the Outcomes of River Restoration Schemes 187
K. Skinner, F.D. Shields, Jr and S. Harrison
11 Methods for Evaluating the Geomorphological Performance of Naturalized Rivers:
Examples from the Chicago Metropolitan Area 209
B.L. Rhoads, M.H. Garcia, J. Rodriguez, F. Bombardelli, J. Abad and M. Daniels
12 Uncertainty and the Management of Restoration Projects: The Construction and Early
Post-Construction Phases 229
A. Brookes and H. Dangerfi eld
Section IV Uncertainty and Sustainability: Restoration in the Long Term
13 The Sustainability of Restored Rivers: Catchment-Scale Perspectives on Long Term Response 253
K.J. Gregory and P.W. Downs
14 Uncertainty and The Sustainable Management of Restored Rivers 287
M.D. Newson and M.J. Clark
Index 303

For many years scientists and river practitioners have
recognised the severity and extent to which aquatic ecosystems
have been degraded by a variety of human disturbances
and activities (Gregory and Park, 1974; Sear and
Arnell, 2006). In turn, realisation of the widespread nature
of the problem has more recently elicited a surge of interest
in the possibility of undertaking corrective interventions,
such as fl ow restoration and channel modifi cations,
to restore or rehabilitate lost and/or damaged ecosystem
functions (Brookes and Shields, 1996; Wissmar and
Bisson, 2003). Indeed, there is now a substantial volume
of literature on the broad topic of river restoration, much
of which suggests that, to be sustainable, river restoration
projects should be designed to recreate functional characteristics
within a context of physical (i.e. geomorphic)
stability. It is true that the emphasis on stable channel
design may refl ect the traditional disciplines of many of
the river engineers who have now turned their attentions
to restoration. Whatever the provenance and merits of this
approach, a focus on stable channel design requires the
application of geomorphic and engineering design tools
(models) that are for the most part either entirely empirical
or empirically calibrated. As a result, different results are
obtained when different models are applied to the same
problem. Furthermore, the data required to apply morphological
models to restoration design are often absent,
incomplete, or subject to measurement error. Finally, even
when a restoration design is completed, it is usually not
possible to predict the precise sequence of fl ood events.
In the long term further variability is introduced by climatic
or catchment changes (e.g. in land use), or unanticipated
social or cultural changes, all of which might shift
the basic premise(s) of the design. It is evident that the
designers and managers of stream restoration projects are
inevitably confronted with uncertainty.
Despite, or perhaps because of, this challenging situation
the restoration literature, albeit with some notable
Preface
exceptions (Wissmar and Bisson, 2003), has not yet
devoted consideration to identifying associated uncertainties,
let alone seeking to quantify, manage, or – where
appropriate (see below) – constrain them. Rather, the discipline
has instead tended to focus on management
responses (e.g. post-project appraisal, adaptive management
strategies) that only implicitly confront assumed
sources of variability and uncertainty. Our concern is that
a collective disciplinary failure to recognise, communicate
and deal appropriately with uncertainties might, at some
time in the future, undermine institutional and public confi
dence in river restoration. In a fi rst attempt to address
these issues, we (together with Dr Andrew Collison and
Dr Sean Bennett) convened a special session on Uncertainty
in River Restoration at the 2002 Fall Meeting of the
American Geophysical Union (AGU) in San Francisco,
California. While recognising that no single volume can
ever cover all aspects of such a multi-faceted discipline as
river restoration, the positive response to the topic at that
AGU symposium prompted us to seek to explore it further
in this volume. All the chapters for this book were, therefore,
specially commissioned in an attempt to provide a
coherent narrative structure that offers a rational theoretical
analysis of the uncertain basis of restoration, while
simultaneously providing practical guidance on managing
the implications of that uncertainty.
The resulting book is structured into four main sections.
Each offers a range of case studies in an attempt to ensure
a wide geographic coverage. Likewise, the authorship is
drawn from a range of countries and disciplines, in an
attempt to bring a range of perspectives to the table.
Section I comprises three chapters that review the nature
and signifi cance of uncertainty in river restoration, providing
a context for the remainder of the book. In Chapter 1
Lemons and Victor focus on the specifi c nature of scientifi
c uncertainty in restoration, while Graf (Chapter 2)
expands on this theme, identifying a series of sources of

viii Preface
uncertainty in theory, research and communication. In
Chapter 3 Wheaton et al. synthesise and extend these
analyses, presenting a classifi cation that suggests uncertainty
fundamentally arises either through limited knowledge
or through natural system variability. This is an
important distinction, not least because it helps to discriminate
between those sources of uncertainty (limited
knowledge) which should, where possible, be constrained
(e.g. by scientifi c progress) from those sources (e.g. natural
variability) that should be embraced to promote healthy
system functioning. The management implications associated
with each form of uncertainty are therefore distinct,
but recognition that embracing certain types of uncertainty
may be both necessary and desirable to assure sustainability
is a theme that runs throughout many of the contributions
herein.
The book is subsequently structured to address the discrete
stages in the life span of a typical restoration project,
covering the planning and design activities associated with
the pre-construction phase (Section II), the construction
phase itself (Section III) and the long term post-construction
phase (Section IV). Section II (Chapters 4 to 8) presents
contributions covering various aspects of planning
and design associated with restoration projects. In Chapter
4 Kondolf and Yang’s review reminds us that restoration
is fundamentally a social and cultural process, with variability
in cultural values acting as a signifi cant contributor
of uncertainty. Presenting an Australian case study, where
the aim was to restore fl ows capable of fl ushing fi ne sediment
from river gravels, Stewardson et al. (Chapter 5)
identify the limits in our understanding of hydrological,
hydraulic and geomorphic processes and how these constrain
our ability to model river system dynamics. In terms
of the uncertainty classifi cation discussed in Chapter 3,
their focus is essentially on quantifying the magnitude of
uncertainty due to limited knowledge. Their results (that
the magnitude of designed fl ushing fl ows is subject to
uncertainty estimates approximately twice that of the fl ow
itself) reinforce the earlier suggestion that restoration is
indeed an uncertain discipline. Whether this really means
that we should have ‘unreasonable confi dence’ in restoration,
as suggested by their provocative sub-title, is a theme
that is continued throughout the book.
In contrast to the focus on uncertainty due to limited
knowledge expounded in Chapter 5, Chapter 6 (Hughes
et al.) reviews some of the diffi culties associated with
extending restoration into complex riparian and fl oodplain
habitats, emphasising that in these systems uncertainty (in
this case in the form of physical diversity and variability)
is necessary to underpin the successful restoration of
forest fl oodplain ecosystems. In Chapter 7, Clifford et al.’s
comprehensive review of how the restoration of fl ow
hydrology and hydraulics can be used to enhance aquatic
habitats also recognises the importance of restoring natural
variability, and provides recommendations on how such
variability can be interpreted in modelling investigations.
The theme of uncertainty associated with ecological
targets is explored further in Chapter 8 (Perrow et al.),
where the paradox that uncertainty is often viewed as a
pejorative term is again highlighted, even if it is uncertainty
(in the form of natural variability) that is the key
mechanism for sustaining healthy ecosystems. How uncertainty
due to the lack of understanding of a discipline
(ecology) by river managers has led to a lack of using
available science within restoration process is also
highlighted.
Section III (Chapters 9 to 12) addresses the construction
phase of a restoration project, which is defi ned in this book
as extending up to one or two years after completion of
the project. The contributions in this section are written
primarily by river practitioners, who employ their collective
experience to offer a range of perspectives on uncertainties
encountered during this key stage of restoration.
Mant et al. (Chapter 9) review the diffi culties encountered
during construction and note that strong teamwork skills
are required to ensure that the design concepts provided
by geomorphologists and ecologists are correctly translated
into practice by those responsible for construction.
A diffi culty here is that restoration is seen as a relatively
new facet of civil engineering, such that contractors may
not always have the experience necessary to recognise that
variability, rather than uniformity (their experience to
date), is often necessary. To this end it is essential that
designers inform the workforce of the specifi c requirements
of the river restoration project, while project managers
must also take responsibility for monitoring
construction as it progresses. This points to the importance
of ensuring that the constructed project does indeed
conform to the design specifi cations, raising the issue of
evaluating project outcomes.
This subject is the theme of the next three chapters.
Skinner et al. (Chapter 10) review post-project appraisals
with reference to both physical and ecological measures
of success, whilst Rhoads et al. (Chapter 11) focus on
methods for evaluating the geomorphological performance
of restored rivers, providing examples from heavily urbanised
catchments in Illinois in the USA. Both contributions
emphasise the key need for both pre and post-project monitoring,
even if only to a minimum standard. This is viewed
as necessary to verify that projects are constructed according
to their design, as well as an integral tool of adaptive
management that can make project adjustments in the face
of uncertainties introduced by variable post-project conditions.
This theme is further explored by Brookes and

Dangerfi eld (Chapter 12), who propose that managers
should adopt continuous improvement as an overall operational
philosophy for restoring rivers. The term continuous
improvement is widely documented in human resource,
organisational management and environmental management
literature, and is taken to be a philosophy of learning
during the construction and post-construction phases (and
making adaptations to a particular project as necessary)
for the benefi t of the continuing work and future practice.
This appears to be a robust approach that has the potential,
over time, to reduce uncertainties associated with limited
knowledge while simultaneously providing a framework
for adaptive response to uncertainties associated with
natural variability.
Section IV (Chapters 13 and 14) addresses the challenge
of the need to assure the long term sustainability of restoration
projects in the face of uncertain futures. There is a
clear recognition that as the time scales over which project
outcomes should be considered increase, there is a concomitant
need to address increased spatial scales. Specifi -
cally, there is a need to consider how catchment-scale
processes (which infl uence the fl uxes of water and sediment
supplied to restoration reaches) are to be sustained
in the long term. Clearly, as spatial and temporal scales
increase, then so do the uncertainties particularly, but not
exclusively so, those associated with increases in spatial
and temporal variability. In Chapter 13 Downs and Gregory
bring a hydromorphological perspective to these issues,
suggesting that the bounds of these uncertainties can be
evaluated with reference to long term (palaeo)hydrological
and geomorphological evidence of past catchment response
– in effect advocating a more precise defi nition of the
uncertainty due to natural variability.
The fi nal chapter (Chapter 14; Newson & Clark) provides
an apt conclusion. Recognising that uncertainty (in
the form of natural variability) is both endemic and necessary,
it is noted that there is a confl ict between the precautionary
principle – a cornerstone of sustainable thinking
– and uncertainty. Newson and Clark recognise that all
restorations have outcomes that are to some extent unpredictable,
and the precautionary principle thus becomes a
recipe for inaction. Uncertainty is therefore simultaneously
necessary for, but also a barrier to, sustainability.
They attempt to resolve this particular problem by identifying
management and restoration opportunities that are
sustainable despite being uncertain, noting that in practical
terms it is to adaptive management that we most often turn
for a way forward, reinforcing a series of conclusions from
earlier chapters.
Do we have unreasonable confi dence in restoration,
based on the state of the art, or are we happy to boldly go
Preface ix
with the uncertain ebb and fl ow of natural variability?
Perhaps the way forward is through a clearer and more
transparent approach to communicating uncertainty, such
that all participants – and especially stakeholders – understand
that while every effort can and should be made to
identify, and where possible reduce, scientifi c and methodological
uncertainties, uncertainty due to natural variability
is both welcome and necessary to sustain healthy
aquatic ecosystems. This implies a concerted approach on
two fronts: scientists can continue to refi ne the knowledge
base (and managers need to recognise the value of this
research for their (adaptive) management practices), while
managers must work harder to build and accommodate
variability into projects. Uncertainty in river restoration is
endemic but clearly offers opportunities, not just as a
rationale for further research, but fundamentally for more
sustainably managed restoration projects. Just as life goes
on with little confi dence or ability to predict the future, so
restoration must continue to evolve and adopt an approach
that is consistent with the uncertain functioning of riverine
ecosystems.
In closing this preface, we would like to acknowledge
those who have made signifi cant contributions during the
production of this book. Firstly, we would like to thank
those numerous professionals who provided detailed peer
reviews of each chapter, often working to tight deadlines.
The anonymous nature of peer review means that we are
unable to identify them here, but you know who you are!
Finally, Tim Aspden and the staff of the Cartographic Unit
at the School of Geography, University of Southampton,
provided guidance on, and help with, the production of
much of the artwork. A fi nal acknowledgment must go to
our families who have put up with longer hours than
normal (or natural!) in the drive for completion.
Stephen Darby and David Sear
December 2007
REFERENCES
Brookes A, Shields FD. 1996. River Channel Restoration:
Guiding Principles for Sustainable Projects. John Wiley &
Sons Ltd: Chichester, UK.
Gregory KJ, Park CC. 1974. Adjustment of river channel capacity
downstream from a reservoir. Water Resources Research 10:
840–873.
Sear DA, Arnell NW. 2006. The application of palaeohydrology
to river management. Catena 66: 169–183.
Wissmar RC, Bisson PA (Eds). 2003c. Strategies for Restoring
River Ecosystems: Sources of Variability and Uncertainty in
Natural and Managed Systems. American Fisheries Society:
Bethesda, Maryland, USA.

Jorge Abad, Department of Civil and Environmental Engineering,
University of Illinois at Urbana-Champaign,
Urbana, Illinois 61801, USA.
Dr Mike Acreman, Water Resources & Environment Division,
Centre for Ecology & Hydrology, Maclean Building,
Crowmarsh Gifford, Wallingford, Oxfordshire OX10 8BB,
UK (man@ceh.ac.uk).
Professor Fabian Bombadelli, Department of Civil and
Environmental Engineering, University of California at
Davis, Davis, California 95616, USA (fabombardelli@
ucdavis.edu).
Dr Douglas Booker, National Institute of Water and
Atmospheric Research, 10 Kyle St., Riccarton, Christchurch,
8011, New Zealand (d.booker@niwa.co.nz).
Dr Andrew Brookes, Jacobs (UK), School Green, Shinfi
eld, Reading RG2 9HL, UK (andrew.brookes@jacobs.
com).
Professor Mike Clark, School of Geography, University of
Southampton, Highfi eld, Southampton SO17 1BJ, UK
(M.J.Clark@soton.ac.uk).
Professor Nick Clifford, School of Geography, The University
of Nottingham, University Park, Nottingham NG7
2RD, UK (Nick.Clifford@nottingham.ac.uk).
Dr Helen Dangerfi eld, Royal Haskoning, 4 Dean’s
Yard, London, SW19 3NL, UK (Helen.dangerfi eld@
royalhaskoning.com).
List of Contributors
Dr Melinda Daniels, Department of Geography, Kansas
State University, USA (melinda.daniels@uconn.edu).
Dr Stephen Darby, School of Geography, University of
Southampton, Highfi eld, Southampton SO17 1BJ, UK
(S.E.Darby@soton.ac.uk).
Dr Peter Downs, Stillwater Sciences, 2855 Telegraph
Avenue, #400, Berkeley, California 94705 USA (downs@
stillwatersci.com).
Dr Judy England, Environment Agency, Apollo Court, 2
Bishop’s Square, St. Albans Road West, Hatfi eld AL10
9EX, UK.
Professor Marcelo Garcia, Department of Civil and Environmental
Engineering, University of Illinois at Urbana-
Champaign, Urbana, Illinois 61801, USA (mhgarcia@
uiuc.edu).
Professor William Graf, Department of Geography, University
of South Carolina, Columbia, South Carolina
29208, USA (graf@sc.edu).
Professor Ken Gregory, School of Geography, University
of Southampton, Highfi eld, Southampton SO17 1BJ, UK
(Ken.Gregory@btinternet.com).
Dr Simon Harrison, Department of Zoology, Ecology and
Plant Sciences, University College Cork, Lee Maltings,
Prospect Row, Cork, Ireland (s.harrison@ucc.ie).

xii List of Contributors
Dr Francine Hughes, Department of Life Sciences, Anglia
Ruskin University, East Road, Cambridge CB1 1PT, UK
(f.hughes@anglia.ac.uk).
Martin Janes, River Restoration Centre Manager, The
River Restoration Centre, Building 53, Cranfi eld University,
Cranfi eld, Bedfordshire MK43 0AL, UK (rrc@therrc.
co.uk).
Professor Mat Kondolf, Department of Landscape Architecture
and Environmental Planning, University of California,
Berkeley, California 94720-2000, USA (kondolf@
uclink.berkeley.edu).
David Leeming, Consultant Ecologist, Spindlewood, 45
West End, Ashwell, Hertfordshire SG7 5QY, UK.
Professor John Lemons, Department of Environmental
Studies, University of New England,11 Hills Beach Road,
Biddeford, Maine 04005, USA (jlemons@une.edu).
Dr Jenny Mant, The River Restoration Centre, Building
53, Cranfi eld University, Cranfi eld, Bedfordshire MK43
0AL, UK (rrc@therrc.co.uk).
Dr Tim Moss, Institute for Regional Development and
Structural Planning (IRS), Flakenstrasse 28–31, 15537
Erkner, Germany. (MossT@irs-net.de).
Professor Malcolm Newson, Department of Geography,
University of Newcastle-Upon-Tyne, Newcastle-Upon-
Tyne NE1 7RU, UK (M.D.Newson@ncl.ac.uk).
Dr Martin Perrow, ECON, Ecological Consultancy,
Norwich Research Park, Colney Lane, Norwich, Norfolk
NR4 7UH, UK (m.perrow@econ-ecology.com).
Professor Bruce Rhoads, Department of Geography, University
of Illinois at Urbana-Champaign, Room 220 Davenport
Hall, 607 South Mathews Avenue, Urbana, Illinois
61801-3671, USA (brhoads@uiuc.edu).
Professor Keith Richards, Department of Geography, University
of Cambridge, Downing Place, Cambridge CB2
3EN, UK (keith.richards@geog.cam.ac.uk).
Dr Roy Richardson, Scottish Environment Protection
Agency, Burnbrae, Mossilee Road, Galashiels TD1 1NF,
UK.
Dr Jose Rodriguez, Faculty of Engineering and Built Environment,
University of Newcastle, Newcastle, New South
Wales 2308, Australia (jose.rodriguez@newcastle.edu.
au).
Dr Ian Rutherfurd, Geography Program, School of
Resource Management, University of Melbourne, Melbourne,
Victoria 3010 Australia (idruth@unimelb.edu.au).
Professor David Sear, School of Geography, University of
Southampton, Highfi eld, Southampton SO17 1BJ, UK
(D.Sear@soton.ac.uk).
Dr F. Doug Shields Jr, Water Quality and Ecology
Research Unit, National Sedimentation Laboratory, USDA
Agricultural Research Service, National Sedimentation
Laboratory, PO Box 1157, Oxford, Mississippi, USA
(dshields@ars.usda.gov).
Eleanor R. Skeate, ECON, Ecological Consultancy,
Norwich Research Park, Colney Lane, Norwich, Norfolk
NR4 7UH, UK (e.skeate@econ-ecology.com).
Dr Kevin Skinner, Principal Geomorphologist, Jacobs
Babtie, School Green, Shinfi eld, Reading RG2 9HL UK
(Kevin.skinner@jacobs.com).
Dr Michael Stewardson, Department of Civil and Environmental
Engineering and eWater CRC, University of
Melbourne, Melbourne, Victoria, 3010 Australia (mjstew@
unimelb.edu.au).
Mark L. Tomlinson ECON, Ecological Consultancy,
Norwich Research Park, Colney Lane, Norwich, Norfolk
NR4 7UH, UK (m.tomlinson@econ-ecology.com).
Professor Reginald Victor, Centre for Environmental
Studies and Research, c/o Department of Biology, Sultan
Qaboos University, PO Box 36, Al-Khod, PC 123, Muscat
Sultanate of Oman (rvictor@squ.edu.om).
Joseph M. Wheaton, Institute of Geography and Earth
Sciences, University of Wales, Aberystwyth SY23 3DB,
UK (Joe.Wheaton@aber.ac.uk).
Dr Chia-Ning Yang, Department of Landscape Ar -
chitecture, California State Polytechnic University,
Pomona, California 91768, USA (cnyang@csupomona.
edu).

SECTION I
Introduction: The Nature and Signifi cance
of Uncertainty in River Restoration

River Restoration: Managing the Uncertainty in Restoring Physical Habitat Edited by Stephen Darby and David Sear
© 2008 John Wiley & Sons, Ltd
1
Uncertainty in River Restoration
John Lemons 1 and Reginald Victor 2
1 Department of Environmental Studies, University of New England, USA
2 Center for Environmental Studies and Research, Department of Biology, Sultan Qaboos University, Sultanate of Oman
1.1 INTRODUCTION
As we are well aware, rivers fundamentally shape the
planet and human life. Both ancient and modern societies
have developed and fl ourished in the proximity of rivers
and this trend has continued till modern times. Nienhuis
and Leuven (2001) summarize how humans have spatially
and temporally altered rivers over a 6000-year period by
various anthropogenic activities. For example, intensive
use of European rivers started over 500 years ago leading
to the loss of their ecological integrity (Smits et al., 2000).
Some rivers were altered for navigation, fl ood control,
agriculture and reclamation of land for urban development,
while most were used as chutes for waste disposal
including sewage, thermal effl uents and both nontoxic and
toxic chemicals; some rivers were also routinely dredged
to facilitate the transport and storage of timber, while
others were heavily fi shed (Ward and Stanford, 1979; De
Wall et al., 1995; Eiseltova and Biggs, 1995).
Large river systems (stream order >8) all over the world
have been extensively dammed for hydroelectric power,
recreation, fl ood control and to divert water to support
agriculture. Impacts of large dams include the loss of
fi sheries and the ecological collapse of the entire river
regime (Balon and Coche, 1974; Rzoska, 1976; Obeng,
1981). Extensive series of levees built along large rivers
have caused major losses of ecosystem structure and function.
Along the Mississippi River, the largest river in North
America, levees threaten federal plans to protect endangered
species (EPA, 2004). The effects of impounding
small rivers (stream order 4–8) are even more drastic. In
some West African small rivers entire fi sh communities
had changed due to impoundment and the ecological perturbations
extended for considerable distances downstream
(Victor and Tetteh, 1988; Victor and Meye, 1994; Victor
and Onomivbori, 1996). Gopal (2003) describes how
rivers in arid and semi-arid regions in Asia are being
degraded due to overexploitation of natural resources,
salinization, pollution and introduction of exotic species.
Just as rivers have undergone alteration, so too have
there been efforts to restore them in order to provide benefi
ts to the environment and/or human health, as this book
attests (see also MacMahon and Holl, 2001). Obviously,
scientifi c research contributes to river restoration by: providing
reliable and needed explanatory or heuristic knowledge
and understanding of restoration problems; helping
to identify and defi ne new research needs and directions
through the acquisition of factual information; and informing
policy and decision making (Caldwell, 1996).
A major premise of this book is that to be sustainable,
river restoration projects need to effectively recreate a
rivers’ functional characteristics taking into account the
dynamic geomorphic characteristics. While many restoration
projects have benefi ted environmental and/or human
health, understudied sources of uncertainty limit confi -
dence in predicting the outcomes of restoration activities
and programs. Specifi c examples of uncertainty in river
restoration discussed in this book include those inherent
in: river management processes; the planning and design
phases of restoration projects; hydraulic and hydrological
aspects of restoration; water quantity issues; identifying
appropriate ecological characteristics and predicting their

4 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
responses in restoration designs; and the construction and
post-construction phases of restoration projects. The
sources of uncertainties include: lack of scientifi c and
other information; limitations of analytical methods
and tools; complexities of river systems; and needs
to make value-laden judgments at all stages of river restoration
problem identifi cation, analysis and solution
implementation.
Beginning in and since the early 1990s some philosophers,
scientists and public policy experts concluded that
the sources and implications of scientifi c and other uncertainty
in environmental problem solving, including restoration,
have been understudied and, as a consequence,
not suffi ciently taken into account by researchers, public
policy makers and decision makers (Mayo and Hollander,
1991; Cranor, 1993; Shrader-Frechette and McCoy, 1993;
Funtowicz and Ravetz, 1995; Lemons and Brown, 1995;
Lemons, 1996; EEA, 2001; Kriebel et al., 2001; Tickner,
2002, 2003).
In agreeing with this conclusion, the objective in this
chapter is therefore to fi rst discuss various broad views
about scientifi c uncertainty and indicate how and why
these need to be taken into greater account by scientists,
policy makers and decision makers. (Other chapters
address uncertainty and analyze in more concrete detail
how it interacts with the specifi c theories and practices of
river restoration.). Discussion then focuses on what might
constitute ‘good’ science when science is used to inform
policy and decision making under conditions of scientifi c
uncertainty. Value-laden sources and implications of
uncertainty in river restoration are then discussed because
they are both important but understudied. Discussion of
the value-laden sources and implications of uncertainty is
followed with: a brief discussion of some of the practical
and policy implications of uncertainty in river restoration,
and, fi nally, a brief case study of river restoration in order
to communicate our views with a practical example. For
reasons of brevity the case study communicates views
about some, but not all, aspects of uncertainty in river
restoration.
Parenthetically, here it is necessary to comment on defi -
nitions of ‘restoration’ when used in the context of river
restoration. The fi eld of restoration ecology suffers from
a lack of conceptual clarity concerning its meaning, goals
and objectives. Since about the mid-1980s, the fi eld of
river restoration has increasingly evolved in an attempt to
better meet societies’ needs to more effectively repair
damage to rivers (e.g., Cairns and Heckman, 1996; Karr
and Chu, 1999; Cairns, 2001). The Society of Wetlands
Scientists (SWS, 2000) defi ned restoration as ‘actions
taken in a converted or degraded natural wetland that
result in the re-establishment of ecological processes,
function, and biotic/abiotic linkages and lead to a persistent,
resilient system integrated within its landscape.’ In
2002, the Society for Ecological Restoration (SER, 2002)
defi ned restoration as the ‘. . . process of assisting the
recovery of an ecosystem that has been degraded, damaged,
or destroyed.’ Regardless of these defi nitions, the goals
and objectives of river restoration are not clear.
Rolston (1988) believes that where possible ecosystems
should be returned to their ‘natural’ or ‘original’ condition.
Westra (1995) argues that restoration should focus on
restoring ecosystems’ abilities to continue their ongoing
change and development unconstrained by human interruptions
past or present. The United States National
Research Council (NRC, 1999) defi ned restoration as
‘the return of an ecosystem to a close approximation of
its condition prior to disturbance.’ This defi nition was
expanded by Cairns (2001), who asserted that the goal of
restoration should be devoted to ‘returning damaged ecosystems
to a condition that is structurally and functionally
similar to the predisturbance state.’
Alternatively, others involved in the fi eld of restoration
ecology provide defi nitions for restoration that more
explicitly focus on historical, social, cultural, political,
aesthetic and moral aspects. For example, Sweeney (2000)
argues that restoration should focus on the value-laden
social and ethical perspectives regarding what constitutes
a ‘restored’ ecosystem. Some others maintain that conservation
and, by implication, restoration goals should take
into account the views and practices of rural and indigenous
people who depend on the ecosystems for their
physical and cultural subsistence, and should also include
scientifi c and nonscientifi c considerations (Gomez-Pompa
and Kaus, 1992; Westra, 1995; Light and Higgs, 1996;
Higgs, 1997; Chauhan, 2003). Regier (1995) proposes an
abstract defi nition for restoration that is dependent on
what people believe as fostering a state of ‘well-being.’
Obviously, lack of conceptual clarity about restoration
introduces an element of uncertainty into restoration
problem solving. In this chapter, while being mindful of
the unresolved problems of conceptual clarity regarding
‘restoration’ other sources and implications of uncertainty
and their relevance to river restoration are focused upon.
1.2 BROAD PHILOSOPHICAL VIEWS ABOUT
SCIENTIFIC UNCERTAINTY
During the 19th century there was a high degree of confi -
dence in the methods and tools of science and technology
to increase understanding of the natural world and enable
robust predictions of its future states. This confi dence in
science contributed to beliefs that ‘nature’ could be controlled
and rendered useful to humankind (Latour, 1988).

Contributing to these beliefs were philosophers and scientists
(so-called ‘logical positivists’) who proposed that
an important goal of science should focus on formulating
hypotheses and conducting observations to test them,
developing an understanding of processes and linkages
among variables, and developing conclusions and predictions
about which there is a high degree of confi dence.
More specifi cally, the logical positivistic view of science
assumes that: knowledge is founded on experience; concepts
and generalizations only represent the particulars
from which they have been abstracted; meaning is
grounded in observation; the sciences are unifi ed according
to the methodology of the natural sciences and the
ideal pursued in knowledge is the form of mathematically
formulated universal science deducible from the smallest
number of possible axioms; and values are not facts
grounded in observation and therefore cannot be included
as a part of scientifi c knowledge. One the one hand, while
logical positivism has infl uenced the thinking of modern
scientists public policy makers, and decision makers, on
the other it does not enjoy wide support from contemporary
scientifi c philosophers (Hull, 1974).
Scientists typically are conservative insofar as they provisionally
reject a null hypothesis only if the probability
of making a type I error is fi ve percent or less (Cranor,
1993; Lemons et al., 1997). This scientifi c conservatism
is consistent with the logical positivist goal of developing
conclusions about which there is a high degree of confi -
dence. With respect to the use of science as a basis for
public policy and decision making, there are those who
hold that scientifi c methods and tools are capable of yielding
information about which there is a high degree of scientifi
c confi dence and, therefore, it is this information and
not more speculative information that should be used as
the basis for policy and decision making (Peters, 1991;
Sunstein, 2002). This latter view is a component of the
fi eld of environmental and human health risk assessment,
which has developed to help inform public policy and
decision makers about the risks from threats from both
natural phenomena and human activities, including assessing
whether to undertake some river restoration projects.
Components of risk analysis include: identifying the
sequence of events through which exposure to risk could
occur; determining the number and kinds of people or
environmental resources exposed to the risk; determining
the adverse effects of exposure to the risks; and communicating
risk assessment fi ndings to decision makers and
the public. Although risk assessors acknowledge scientifi c
uncertainty, they often hold that scientifi c methods and
tools can identify the risks and enable the calculation of
the probabilities of their occurrence, including the bounding
of the probabilities with confi dence limits. For in-
Uncertainty in River Restoration 5
depth discussions on the role of scientifi c information in
policy and decision making, see Peters (1991), Shrader-
Frechette, (1994), Caldwell (1996), Lemons (1996), and
Kaiser and Storvik (2003).
Historically, logical positivism and its outgrowths also
have infl uenced the thinking of some scientists and policy
makers in other ways by inculcating the view that ‘good’
science is objective insofar as it is not biased by the values
of the scientists. Accordingly, this view holds that the
proper role of science in policy and decision making is to
provide factual information to decision makers, and that
any controversies about the factual information should be
left to members of the scientifi c community competent
in evaluating the scientifi c bases of the controversies
(Shrader-Frechette, 1982). Consequently, the conclusions
of scientifi c analyses do not become a part of broader
public policy debates such as those that might pertain to
such issues as what level of risk is acceptable. Practically
speaking, proponents of this view believe that the scientifi
c and technical problems of managing large scale and
complex problems are enormous and that the public cannot
be expected to grasp the many scientifi c and technical
issues inherent in understanding and resolving the problems.
Further, the fundamental differences people have
about how problems should be handled generate endless
debate and controversy. This implies that while people and
local governmental representatives with different interests
may review and comment on scientifi c and technical
documents, they would not be brought into the actual
decision- making process regarding the complex scientifi c
dimensions of problems (Lemons et al., 1997).
Despite the high degree of confi dence held by some
people in scientifi c methods, confi dence in the power of
science to understand and predict natural phenomena has
been undermined by general relativity theories, quantum
theories and chaos theories (Brown, 1987). Rorty (1979)
notes that there is no evidence that science develops better
and more accurate ‘mirrors’ with which to view nature. In
his classic work, Kuhn (1962) describes how on the one
hand the level of confi dence in models used by members
of the scientifi c community increases with evidence that
supports the underlying hypotheses of the models, and on
the other the scientists’ use of the models cannot be
expected to produce consistently better and cumulatively
more truthful descriptions of the way the world works.
According to Kuhn, the reason is because predictive successes
of scientifi c theories do not guarantee their metaphysical
accuracy because ‘paradigm shifts’ subsequently
change scientists’ views of nature. Other critics have
pointed out that so-called scientifi c truths of historical
periods are social constructs infl uenced by the dominant
cultural and political powers of those periods (Briggs and

6 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Peat, 1982; Funtowizc and Ravetz, 1995). Some postmodern
critics argue that Western science has been permeated
by a variety of biases (e.g., ‘free market’ economics and
industrialism, racism, religion, patriarchy) that while
serving powerful interests have not led to the generation
and use of more ‘objective’ or value-free scientifi c
knowledge (Sirageldin, 2002).
More practically speaking, scientifi c institutions as well
as individual scientists increasingly hold the view that
scientifi c uncertainty regarding environment and human
health problems is so pervasive and value laden that many
conclusions about the problems cannot be made with
a high degree of scientifi c confi dence (Cranor, 1993;
Shrader-Frechette and McCoy, 1993; Lemons and Brown,
1995; Lemons, 1996; EEA, 2001; Kriebel et al., 2001;
Tickner, 2002, 2003). This view is based on empirical
studies focusing on: exposure to radiation from nuclear
facilities and nuclear waste; managing large-scale ecosystems
such as the Florida Everglades, agricultural lands,
marine and freshwater oil spills; biodiversity protection
and management of biological reserves; ocean dumping
of sewage sludge; sulfur dioxide and protection of human
lungs to remote lake restoration; antifouling paints on
ships (e.g. tributyltin); estuarine eutrophication; protection
and management of marine fi sheries; extrapolating
from toxicological responses in laboratory systems to both
human health and to the responses of natural systems;
management of fresh water resources; benzene in occupational
settings; the use and health impacts of asbestos;
risks from polychlorinated biphenyls (PCBs); halocarbons
and the ozone layer; diethylstilbestrol (DES) and longterm
consequences of prenatal exposure; human health
effects of lead in the environment; methyl tertiary-butyl
ether (MBTE) in petrol as a substitute for lead; chemical
contamination in the Great Lakes; hormones as growth
promoters in animals used for food; and global climate
change.
1.3 WHAT IS ‘GOOD’ SCIENCE UNDER
CONDITIONS OF UNCERTAINTY?
Here, the question discussed is: What is ‘good’ science
when science is used in trying to solve river restoration
problems under conditions of scientifi c uncertainty?
A traditional and commonly accepted goal of science is
that the probabilities of adding speculative information to
the body of scientifi c knowledge should be minimal (Hull,
1974; Peters, 1991). For this reason, scientists typically
are conservative insofar as they provisionally reject a null
hypothesis if there is a fi ve percent or less chance of
rejecting it when it is true; this criterion is known as a
normal standard of scientifi c proof or so-called ‘ninety-
fi ve percent confi dence rule.’ With respect to the science
used to inform certain types of river restoration policies
and decisions, an example of a null hypothesis is that there
is no effect on rivers or their resources from existing or
proposed human activities. A type I error is to accept a
false positive result, that is, to conclude that there is harm
to rivers or their resources when in fact there is none. A
type II error is to accept a false negative result, that is, to
conclude there is no harm when in fact there is.
Many environmental laws and regulations place the
burden of proof for demonstrating harm to the environment
or human health on government regulatory agencies
or others attempting to demonstrate harm from development
activities and, often, the standard that is used to meet
the burden of proof test is the normal standard of scientifi c
proof (Brown, 1995). When this standard is adopted as a
basis for environmental decisions the scientifi c uncertainty
that pervades many environmental problems means
that the burden of proof usually will not be met, despite
the fact that some information or even the weight of evidence
might indicate the existence of harm to the environment
or human health. Consequently, in public policy and
decision making if the data show that some factor or perturbation
has had an effect on the environment or human
health but, say, only at the 70–90% confi dence level the
null hypothesis that there is no effect from the factor or
perturbation is accepted. In such cases there is a tendency
by decision makers and others to assume not only that
there was not enough evidence to reject the null hypothesis
but that there was no effect when, in fact, the experimental
design or test could have been too weak or the data too
variable or too close for an effect to be demonstrated even
if there had been one (a type II error).
Minimizing a type II error requires the statistical power
of a research design or hypothesis test to be calculated. In
contrast to confi dence, which is designed to minimize type
I error, power depends on the magnitude of the hypothesized
change to be detected, the sample variance, the
number of replicates and the signifi cance value. The power
of a test is the probability of rejecting a null hypothesis
when it is in fact false and should be rejected. The larger
the detected change, the larger is the power. In situations
where the detected changes are relatively small, statistical
power is increased by increased sampling size but this
involves additional costs, research facilities and time.
Analysis of variance in assessing threats to environmental
and human health problems shows that the number of
samples required to yield a power of 0.95 increases rapidly
if changes smaller than 50% of the standard deviation are
to be detected (Cranor, 1993). If the sample size stays
the same the probability of a type I error is increased if
the probability of a type II error is decreased. A practical

problem in river restoration is that a desired emphasis on
avoiding type II error must be balanced against other
opportunities to use limited scientifi c resources to address
other environmental and human health problems.
Decisions about water management in the Klamath
Basin along the California and Oregon border in the
United States show some of the types of consequences that
can happen when the law or decision makers require the
use of scientifi c information that meets the normal standard
of scientifi c proof. In decades long disputes about
water management in the basin, federal biologists have
been trying to save three species of endangered fi sh by
calling for diversions of water from irrigation into the
basin to reduce the frequency of fi sh kills during low water
periods (over 30 000 Chinook salmon died during a fi sh
kill in 2002) (Service, 2003). As would be expected, a
recommendation to reduce the amount of water available
for irrigation met with strong opposition by ranchers and
farmers in the basin. However, failure of the biologists to
meet normal scientifi c standards of proof demonstrating
that releasing more water into the basin would help the
fi sh has been cited by the United States Department of
Interior (DOI) in its recent refusal to restrict the amount
of water farmers can remove from waterways in the basin
(NRC, 2004). It is important to understand that the DOI
was not criticizing the scientists for doing poor science;
rather, it concluded that the normal standard of proof was
not met. The DOI noted that factors such as nutrient runoff
from natural sources as well as farms and ranches, algae
blooms and dams that restrict access to fi shes’ spawning
grounds complicate and in fact might preclude demonstrating
the relation of water fl ow into the basin and the
health of the fi sh populations with a higher degree of
scientifi c confi dence.
The question of how to protect endangered species in
the Klamath Basin and manage water resources raises a
fundamental dilemma that those involved in river restoration
have to confront. On the one hand, traditional scientifi
c norms call for making conclusions on information
about which there is a high degree of confi dence. In the
Klamath Basin example, adhering to traditional scientifi c
norms constrains decisions to protect endangered fi sh
under conditions of uncertainty but, at the same time, in
the absence of decisions to protect endangered fi sh the
threats continue. In this type of situation, when science is
used for public policy and decision making, scientists
might wish to consider whether and to what extent they
should be more comfortable with making conclusions
based on the weight of evidence rather than based solely
or primarily on high levels of confi dence, especially since
public policy decisions are not based simply upon probabilistic
considerations but rather involve making discrete
Uncertainty in River Restoration 7
and explicit choices among specifi c alternatives, including
those with political, economic and ethical ramifi cations
(Bella et al., 1994; Lemons et al., 1997). Admittedly, this
could create a tension between doing ‘good’ science as
traditionally defi ned because scientists would be making
more speculative conclusions; however, in their attempt to
make science rigorous in the sense of not wanting to add
speculation to the body of scientifi c knowledge as required
by the scientifi c profession the regulatory questions for
which the studies are done may be frustrated.
1.4 VALUE-LADEN DIMENSIONS OF SCIENCE
AND UNCERTAINTY
In addition to the policy and management problems that
arise from the use of traditional scientifi c norms for
making conclusions in river restoration, other value-laden
dimensions of science and policy both contribute to uncertainty
and raise complicated questions about how it should
be handled in public policy.
Westra and Lemons (1995) and Lemons (1996) contain
papers analyzing both philosophical and scientifi c concepts
used to inform ecological restoration science and
practice. The concepts are diverse and include basing restoration
on: ecosystems’ abilities to function successfully
in a way deemed satisfactory by society; ecosystems’
abilities to maintain a balanced, integrated, adaptive
community of organisms having species composition,
diversity and functional organization comparable to that
of ‘natural’ habits of the region; ecosystems’ abilities to
regenerate themselves and withstand anthropogenic stress;
and ecosystems’ abilities to approach optimum capacity
for ecological succession development options. One
problem with all these defi nitions is that they are incomplete,
general and qualitative insofar as they fail to provide
precise principles that would make them operational.
In his analysis of value-laden issues in restoration
for ecological as opposed to primarily or exclusively
economic development goals, Cairns (2003) focuses on
several types of problems. Firstly, some restoration projects
are carried out on habitats different in kind from those
altered or destroyed. For example, an upland forest may
be destroyed in order to partially restore river systems and
wetlands that once occupied a particular lowland area.
Despite the fact that restoration of rivers and/or wetlands
has ecological value, sacrifi cing a relatively undamaged
habitat to restore another kind may cause unanticipated
ecological change or harm. Secondly, with few exceptions
most river and other ecological restoration projects are
done to support the anthropocentric commodity or utilitarian
values they offer humans and this poses confl icts with
restoration goals for nonanthropcentric reasons. Thirdly,

8 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
river restoration has uncertain outcomes because of unpredictable
events like fl oods or droughts, and because of the
limitations of the methods and tools of science to predict
long-term outcomes. Fourthly, restoration efforts focusing
on single species or ecosystem attributes might eliminate
those species that had initially colonized disturbed areas
and were at the same time able to tolerate anthropocentric
stress. However, restoration projects might result in the
displacement of species tolerant to human activities with
those less tolerant, at least in the short term. Fifthly, ecological
restoration often takes place with species that
tolerate anthropocentric stress and the ultimate succession
processes and states will be human dominated or dependent.
Most likely, a return to indigenous species would
require continual intervention by researchers and
environmental decision makers on behalf of their reestablishment.
While science is not determinative to how
the issues are resolved, robust scientifi c information is
needed to help inform satisfactory policy judgments.
Mayo and Hollander (1991), Cranor (1993), Shrader-
Frechette and McCoy (1993) and Lemons and Brown
(1995) analyzed how and why numerous value-laden
judgments, evaluations, assumptions and inferences are
embedded in scientifi c methods pertaining to the study
and management of ecosystems, including geohydrological
and other water resources. For example, people have
to decide the ecosystem parameters that are more important
to base judgments on, often with little or no empirical
information available. Assumptions have to be made, often
without direct empirical evidence, whether ecosystem
parameters should be considered independently or synergistically,
and whether threshold values for environmental
or health impacts exist and, if so, what such values should
be. In addition, a lack of empirical data cannot be separated
entirely from practical limitations imposed on environmental
scientists. Decision makers require information
in a relatively short period and at reasonable cost. These
factors constrain the focus of most restoration studies to
the short term, relatively small spatial areas and measurement
of a relatively small number of samples and parameters.
Further, the above commentators conclude that
many of the value-laden dimensions of scientifi c methodology
and information not only are not fully recognized
by scientists, policy and decision makers, but that the
failure to suffi ciently recognize the value-laden dimensions
of science casts serious doubts about even the best
and most thorough scientifi c and technical studies used to
inform decisions about problems such as river restoration.
In other words, unless the value-laden dimensions of scientifi
c studies are disclosed the positions of decision
makers will appear to be justifi ed on value-neutral scientifi
c reasoning and will appear to be more certain than
warranted when, in fact, the positions will be based, in
part, on often controversial and confl icting values of scientists
and decision makers (see also Fleck, 1979).
One of the most common ways in which value issues
are hidden in public policy concerning issues such as river
restoration develops out of the expectation that technical
analysts can isolate and apply the facts under dispute in
a manner consistent with policy directives or legislative
mandates. This separation of facts and values is highly
problematic. For example, consider the use of safety
factors in river water quality regulations as a means of
extra protection for human or environmental health.
Implicit in the choice of safety factors is an asymmetric
cost function with health costs rising more steeply than
costs for over-treatment. Implicit in the magnitude of a
safety factor are signifi cant uncertainties in health impacts
and a steeper cost function for health effects from undertreatment
than for over-treatment. When these issues
remain implicit in the use of safety factors (as they typically
are) the real issues of knowledge and uncertainty are
obscured for decision makers and the public. Often, these
issues remain implicit or hidden because safety factors
and cost factors are described in quantitative terms pertaining
to risks or cost–benefi t calculations. This increases
the likelihood of the misuse of conclusions by decision
makers who do not understand the basis for deriving safety
factors (Brown, 1987).
1.5 PRACTICAL AND POLICY ASPECTS
OF UNCERTAINTY
Cairns (2001) analyzed how most complex environmental
problems transcend the capabilities of any single discipline
but at the same time and all too often research teams
are not suffi ciently interdisciplinary to deal adequately
with the problems. In addition, problem solving often does
not provide a balanced mix of academicians, public policy
and decision makers, representatives from private industry
or business and nongovernmental organizations. As a
result, the framing of problems and their solution is too
often fragmented and ineffectual and biased towards one
or a few disciplinary approaches or stakeholder groups
(Nienhuis and Leuven, 2001; Benyamine, 2002).
Some scientists and policy makers involved in environmental
problem solving have argued for synthesizing
analyses and alternatives to solutions of environmental
resource problems (Lubchenco et al., 1991; Bella et al.,
1994; Lemons and Brown, 1995; Caldwell, 1996). In practice,
at least three levels of synthesis may be identifi ed.
The fi rst is conceptual synthesis and occurs when the
diverse and often disparate elements of a problem situation
are pulled together intuitively, then tested and integrated

to form a coherent research design. Following analysis of
the problem and identifi cation of its causes and consequences,
a second level of synthesis involves delineation
of the fi ndings of the scientifi c research. A third level of
synthesis can occur when research fi ndings are evaluated
and consolidated in deciding a course of action by
decision makers.
Despite the need for greater synthesis of research
methods and information, synthesis itself introduces additional
value-laden dimensions and uncertainties into environmental
problem solving. Caldwell (1996) and Brown
(1995) discuss how decision makers must synthesize a
policy (in part) from the scientifi c information available
even when the information often is incomplete. When
science is used to inform policy decisions such decisions
also include economic, legal, administrative and cultural
parameters and, therefore, are based on human values
and judgments. Benyamine (2002) discusses how disagreements
about scientifi c theories that are used as a
basis for informing public policy and decision making
become entangled with economic, legal and ideological
issues. Sometimes, the disagreements remain largely confi
ned to the scientifi c community, while at other times the
public knows about them. When scientists and/or decision
makers know the underlying theoretical bases for disagreements,
this knowledge can infl uence the scientifi c
arguments about the disagreements. However, some confl
icting arguments and their underlying theoretical support
can be under recognized or little understood by the nonscientifi
c communities as well as by scientists whose specialized
fi elds are outside the discipline where debates
about theories are taking place. When this happens, confl
icting scientifi c arguments will not have much infl uence
on the disagreements.
There is debate within the scientifi c and public policy
communities regarding approaches to deal with uncertainties
(Bradshaw and Borchers, 2000). For example, one
approach might be to attempt to increase scientifi c confi -
dence by increasing scientifi c confi rmation of hypotheses.
In this way, scientists can decrease uncertainty suffi ciently
to allow more precise estimates of risk for policy and
decision makers. A second approach might be to increase
the knowledge of sources of uncertainty by enhancing
education and communication between scientists, policy
and decision makers and the general public. A benefi t of
this approach is that when scientists and decision makers
are involved with the public there is greater opportunity
for consensus building and less risk of legal challenges
from disaffected stakeholders. A third approach might be
to foster the view that scientifi c uncertainty should be
regarded in public policy and decision making as it is
within the scientifi c community, namely, as information
Uncertainty in River Restoration 9
for hypothesis building and testing. Consequently, calls
for faster and more ‘certain’ scientifi c conclusions to
inform public policy and decision making would be tempered
with a better understanding of the limitations and
capabilities of science to provide information about which
there is a high degree of confi dence.
Still another approach might be for society to require
procedural rules for making decisions under conditions
of scientifi c uncertainty to take into account confl icting
points of view, possible consequences to welfare, as well
as various ethical and legal obligations such as those
involving free informed consent and due process (Shrader-
Frechette, 1996). This approach could include greater use
of the precautionary principle by helping to ensure that
when there is substantial scientifi c uncertainty about the
risks and benefi ts of a proposed activity, policy decisions
should be made in a way that errs on the side of caution
with respect to the environment and the health of the
public (Kriebel et al., 2001; Tickner, 2003).
1.6 CASE STUDY OF SCIENTIFIC
UNCERTAINTY IN RIVER RESTORATION
The example discussed here is based on ecological studies
conducted from 1980–1989 in a small (4th order), black
water West African river, the River Ikpoba fl owing through
Benin City, Southern Nigeria (Victor and Dickson, 1985;
Victor and Ogbeibu, 1985, 1986, 1991; Victor and Tetteh,
1988; Ogbeibu and Victor, 1989; Victor and Brown, 1990;
Victor and Meye, 1994; Victor and Onomivbori, 1996;
Victor, 1998). The stretch of river studied was affected by
a variety of urban perturbations such as damming, water
extraction, point and nonpoint source pollution, sand
dredging and agriculture. As a result of government
policies and directives mandating river clean-up activities,
there was a rare opportunity to study river restoration by
recovery processes. Scientifi c results of this study were
published in the series of publications listed above and
provide one of the bases of our focus on uncertainties
associated with the restoration process.
The fi rst logical step was to investigate recovery processes.
Geomorphologic changes of the river channel and
the entire riparian corridor infl uenced by urban development
could not be reversed (e.g. the presence of a dam,
water extraction for human consumption) and therefore
complete restoration would not be possible. Removal of
human infl uences where possible would permit recovery,
but the rates limiting recovery in different sections would
not only depend on the type of infl uence (e.g. sand extraction,
car washing), but would also be complicated by
natural events such as fl oods. Thus the optimum threshold
for the recovery process in this study at various sections

10 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
of the river continuum was unpredictable and uncertain.
Other signifi cant uncertainties were: the role of early
recolonizing species affecting the trajectory of recovery;
the successional sequence of species re-establishing; and
the establishment of appropriate abiotic conditions and
the establishment of previously non-existing non-native
species like the water hyacinth.
The next group of uncertainties was related to the analysis
and synthesis of data. Removal of a particular human
infl uence (e.g. discharge of untreated sewage) in one
section signifi cantly increased the presence of a parameter,
say i (P < 0.05), showing that this parameter was a good
indicator of recovery. But the same parameter did not
increase signifi cantly in an adjacent section with a similar
problem (P > 0.05) showing its uncertain predictive status.
Graphical examination of associations between specifi c
human infl uences (e.g. removal of detergent contamination)
and biological parameters like taxa richness and
abundance showed positive relationships, but statistically
these relationships as evaluated by Pearson’s r or Spearman’s
r s were not signifi cant (P > 0.05). Thus, correlation
matrices generated for evaluating relationships between
the removal of perturbation infl uences and the recovery of
both biotic and abiotic parameters were diffi cult to interpret.
Interpretation using traditional statistical norms and
acceptable levels of signifi cance were ecologically and
rationally highly problematic.
Further uncertainties arose while considering the temporal
and spatial scale of the recovery process. The recovery
process was happening in an urban setting with a new
land use matrix, far different from pristine or semipristine
natural conditions that previously existed. Therefore, comparison
of the restored river sections to that of ‘undisturbed’
sections upstream was not valid and new baseline
standards had to be established for future monitoring.
Even these were extremely site specifi c with very limited
potential for use in other sections of the study stretch.
Because of the uncertainties involved, the scale needed for
managing temporal and spatial variability in restoration
was not apparent. ‘Rules of thumb’ based on value judgments
had to be made to evaluate recovery in specifi c
sections of the river stretch with specifi c types of perturbations.
The magnitude of uncertainties involved render the
combination of tools used here (e.g. sampling duration,
sampling frequencies, choice of methods, size of samples,
analytical models) inadequate to evaluate recovery processes
in other rivers of similar stream order, larger rivers
with higher stream order and even the same river 100 km
downstream where its stream order is >8.
Implementation and analysis of monitoring were also
wrought with uncertainties. For example, fi ve different
sections of the river stretch were monitored for restoration
by recovery. Each section was characterized by its own set
of physical and biological parameters that were good
indicators of recovery at the time of the study. Due to
limitations of funding, personnel and the required cost
effectiveness of the monitoring program, proposals had
to identify common parameters that would monitor the
overall health of the study stretch in the long term. As
discussed earlier, uncertainties associated with the analysis
and synthesis of data did not permit the ready identifi
cation of common parameters. Even if there was an
agreement on using different sets of parameters for different
sections of the stretch, there was no certainty that these
parameters (e.g. BOD, nitrate–N, fi sh diversity) will continue
to serve as good indicators of recovery in the long
term. It was also possible that a parameter considered
trivial and not included in the monitoring program (e.g.
dissolved organic matter, haptobenthos) may become
important in the long term, which in itself cannot be
defi ned clearly. ‘Long term’ in this case at least did not
refer to an indefi nite period and envisaged monitoring
programs were not relatively open-ended, as often is the
case in countries with limited resources. Policy and decision
makers considered what seemed to be a comprehensive
proposal for monitoring in the view of scientists as
not being practical.
Policy questions concerning river restoration in the
geopolitical context were plagued with more uncertainties
than scientifi c questions. The political climate of the study
area at that time was unstable and government changed
hands frequently. For example, one government downgraded
the priority given to environmental issues, such as
river restoration, by the previous government if personal
interests and political expediency demanded it. Assuming
no change in policies with change in governments, there
were uncertainties concerning funding tools that would
ensure the long term success of restoration, design of
legislation to accommodate river restoration without compromising
sustainable development and coordination of
policies and legislation to devise strategies for river restoration
in a broader context of the administrative region
(e.g. district, state, country). The management of restored
or recovered river as a water resource for domestic use,
agriculture, fi sheries and recreation was not considered
intentionally. For scientifi c uncertainty concerning water
resources management, see Canter (1996).
1.7 CONCLUSION
Scientifi c and other uncertainty is pervasive in environmental
problem solving, and river restoration is no exception.
When the traditional scientifi c standard of proof is
used as a basis for river restoration decisions, the scientifi c

uncertainty that pervades many restoration problems
means that the standard usually will not be met, despite
the fact that some information or even the weight of evidence
might indicate the existence of harm and therefore
the need for restoration. A high degree of confi dence in
river restoration science, as in other sciences, unfortunately
seems to hinge on conventional statistical decision
rules such as when, for example, river monitoring during
restoration strives to detect human-infl uenced factors
that caused deviations from baseline conditions. The major
concern here will be ecological change and not how large
or small the P-values are (Yoccoz, 1991; Stewart-Oaten,
1996). Most statistical decision rules are too simplistic and
misleading insofar as their assumptions that lack of statistical
signifi cance means lack of environmental signifi -
cance (Karr and Chu, 1999). According to Yoccoz (1991),
Kriebel et al. (2001), and Lemons et al. (1997) ecologists
tend to over-use tests of signifi cance and restoration ecologists
are no exception to this rule. Karr and Chu (1999)
suggest that it would be wiser to decide what is ecologically
relevant fi rst and then use hypothesis testing to detect
ecologically relevant effects; the use of other statistical
tools such as power analysis and decision theory also is
recommended (Hilborn, 1997).
Cairns and Heckman (1996) state that restoration
ecology in general ‘is a bridge between the social and
natural sciences.’ In this chapter it has been shown that it
is impossible to separate scientifi c and policy questions in
restoration ecology and this, in and of itself, introduces
uncertainty into what otherwise might be viewed as value–
neutral or ‘objective’ scientifi c conclusions.
As discussed more generally in this chapter and shown
more specifi cally in the case study section, scientifi c
research is both value-laden and is used to support politically-driven
river restoration policies and decision making
(see also Shrader-Frechette, 1994). For example, historical
or descriptive research is intended to reveal or explain the
dynamics of a given policy and to explore its origin and
evolution. Prescriptive or advocacy research defends a
conclusion or possibly even a preconceived policy, and
also is characterized by publicized disputes among, e.g.,
scientists. Decision-informing or predictive research typically
is fi nanced by grants or contracts leading to conclusions
supportive of a predetermined policy preference,
sponsor bias, or predilections within a research peer group.
Consequently, the focus of this research does not attempt
to analyze all feasible alternative policy choices and the
probable consequences. Because the focus of this research
is on applicability for a particular policy its fi ndings are
presented in the form of propositions upon which decisions
can be made. The effi cacy of the policy towards
which the research is focused depends on the validity,
Uncertainty in River Restoration 11
reliability and persuasiveness of the research and the
extent of political public receptivity.
It is important to clearly distinguish between the use of
methods and tools of science to understand the phenomena
of nature and the acquisition of scientifi c information
about a restoration issue and the setting of policy; but in
practice, there is not always an unambiguous demarcation.
Policy makers set agendas that determine the questions
that are asked of scientists; scientists formulate hypotheses
in ways limited by their tools and their imaginations
and disciplinary conventions. Consequently, the information
they provide to the policy makers is limited and
socially determined to a degree and therefore there is a
complicated feedback relation between the discoveries of
science and the setting of policy. While attempting to be
objective and focus on understanding river restoration
phenomena, scientists and other researchers should be
aware of the policy uses of their work and of their social
responsibility to carryout science that protects the environment
and human health (Kriebel et al., 2001). In trying
to fulfi ll this responsibility, scientifi c and other uncertainty
needs to be taken into greater account.
The discussion of some of the value-laden decisions and
judgments scientists and other researchers make is not a
criticism. Rather, the issue is discussed because a failure to
recognize the existence of the value-laden dimensions of
science casts serious doubt about even the best and most
thorough of scientifi c and technical studies used to inform
decisions about river restoration. In other words, unless the
value-laden dimensions of scientifi c and technical studies
used to derive information are disclosed, the positions
of policy makers and decision makers will appear to be
justifi ed on objective or value–neutral scientifi c reasoning
when, in fact, they will be based in part on often controversial
or confl icting values of scientists themselves.
REFERENCES
Balon EK, Coche AG. 1974. Lake Kariba, A Man-Made Tropical
Ecosystem in Central Africa. DW Junk Publishers: The Hague,
The Netherlands.
Bella DA, Jacobs R, Hiram L. 1994. Ecological indicators of
global climate change: A research framework. Environmental
Management 18: 489–500.
Benyamine M. 2002. Theoretical Disputes and Practical Environmental
Dilemmas. Orebro Studies in Environmental Science
3. Orebro University: Orebro.
Bradshaw GA, Borchers JG. 2000. Uncertainty as information:
Narrowing the Science – Policy gap. Conservation Ecology
4: 7. [online] URL: http://www.consecol.org/vol4/iss1/art7/)
(accessed 23 March 2004)
Briggs J, Peat F. 1982. Looking Glass Universe. Simon and
Schuster: New York.

12 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Brown D. 1987. Ethics, science, and environmental regulation.
Environmental Ethics 9: 331–350.
Brown D. 1995. The role of ethics in sustainable development and
environmental protection decisionmaking. In Sustainable
Development: Science, Ethics, and Public Policy, Lemons J,
Brown D (Eds). Kluwer Academic Publisher: Dordrecht, The
Netherlands; 39–51.
Cairns (Jr) J. 2001. Rationale for restoration. In Handbook of
Ecological Restoration, Vol. 1: Principles and Restoration,
Davy AJ, Perry, J (Eds). Cambridge University Press:
Cambridge; 10–23.
Cairns (Jr) J. 2003. Ethical issues in ecological restoration. Ethics
in Science and Environmental Politics 3: 50–61.
Cairns J (Jr), Heckman JR. 1996. Restoration ecology: The
state of an emerging fi eld. Annual Review of Energy and
Environment 21: 167–187.
Caldwell LK. 1996. In Scientifi c Uncertainty and Environmental
Problem Solving, Lemons J (Ed). Blackwell Science,
Cambridge; 394–422.
Canter LW. 1996. Scientifi c uncertainty and water resources management.
In Scientifi c Uncertainty and Environmental Problem
Solving, Lemons J (Ed). Blackwell Science Inc.: Cambridge;
264–297.
Chauhan M. 2003. Conserving biodiversity in arid regions: Experiences
with protected areas in India. In Conserving Biodiversity
in Arid Regions. Lemons J, Victor R, Schaffer D (Eds).
Kluwer Academic Publishers: Boston.
Cranor C. 1993. Regulating Toxic Substances: A Philosophy of
Science and Law. Oxford University Press, New York.
De Wall LC, Large ARG, Gippel CJ, Wade PM. 1995. River and
fl oodplain rehabilitation in Western Europe: Opportunities and
constraints. Archives Hydrobiologia Supplement 101 – Large
Rivers 9: 679–693.
Eiseltova M, Biggs J. 1995. Restoration of stream ecosystems: An
integrated approach. International Waterfowl and Wetlands
Research Bureau Publication 37: 1–170.
(EEA) European Environment Agency. 2001. Late Lessons From
Early Warnings: The Precautionary Principle 1896–2000,
Environmental Issue Report No. 22. European Environment
Agency: Luxembourg.
(EPA) U.S. Environmental Protection Agency. 2004. Lower Mississippi
Valley Ecosystem Restoration Initiative. [online] URL:
http://www.epa.gov/region4/programs/cbep/lowmiss.html
(accessed 25 March 2004).
Fleck L. 1979. Genesis and Development of a Scientifi c Fact.
University of Chicago Press: Chicago.
Funtowicz SO, Ravetz JR. 1995. Science for the post normal age.
In Perspectives on Ecological Integrity, Westra L, Lemons
J (eds). Kluwer Academic Publishers: Dordrecht, The
Netherlands, 146–161.
Gomez-Pompa A, Kaus A. 1992. Taming the wilderness myth.
BioScience 42: 271–279.
Gopal B. 2003. Aquatic biodiversity in arid and semi-arid zones
of Asia and water management. In Conserving Biodiversity in
Arid Regions, Lemons J, Victor R, Schaffer D (Eds). Kluwer
Academic Publishers: Norwell; 199–216.
Higgs ES. 1997. What is good ecological restoration? Conservation
Biology 11: 338–348.
Hilborn R. 1997. Statistical hypothesis testing and decision
theory in fi sheries science. Fisheries 22: 19–20.
Hull D. 1974. Philosophy of Biological Science. Prentice–Hall:
Englewood Cliffs.
Kaiser M, Storvik H (Eds). 2003. The Precautionary Principle:
Between Research and Politics. The National Research
Ethical Committee for Natural Science and Technology:
Oslo, Norway.
Karr JR, Chu EW. 1999. Restoring Life in Running Waters. Better
Biological Monitoring. Island Press: Washington, DC.
Kriebel D et al. 2001. The precautionary principle in environmental
science. Environmental Health Perspectives 109: 871–876.
Kuhn T. 1962. The Theory of Scientifi c Revolutions. University of
Chicago Press: Chicago.
Latour B. 1988. The politics of explanation: An alternative. In
Knowledge and Refl exivity: New Frontiers in the Sociology
of Knowledge, Woolgar S (Ed). Sage Publications: London;
155–176.
Lemons J (Ed). 1996. Scientifi c Uncertainty and Environmental
Problem Solving. Blackwell Science: Cambridge.
Lemons J, Brown D (Eds). 1995. Sustainable Development:
Science, Ethics, and Public Policy. Kluwer Academic
Publishers: Dordrecht, The Netherlands.
Lemons J, Shrader-Frechette KS, Cranor C. 1997. The precautionary
principle: Scientifi c uncertainty and type I and type II
errors. Foundations of Science 2: 207–236.
Light A, Higgs ES. 1996. The politics of ecological restoration.
Environmental Ethics 18: 227–247.
Lubchenco J et al. 1991. The sustainable biosphere initiative: An
ecological research agenda. Ecology 72: 371–412.
MacMahon JA, Holl KD. 2001. Ecological restoration. A key to
conservation biology’s future. In Conservation Biology.
Research Priorities for the Next Decade, Soule ME, Orians GH
(Eds). Island Press: Washington, DC; 245–269.
Mayo DG, Hollander RD (Eds). 1991. Acceptable Evidence.
Oxford University Press: Oxford.
(NRC) National Research Council. 1999. New Strategies for
America’s Watersheds. National Academy Press, Washington,
DC.
(NRC) National Research Council. 2004. Endangered and
Threatened Fishes in the Klamath River Basin: Causes of
Decline and Strategies for Recovery. Committee on Endangered
and Threatened Species in the Klamath River Basin,
Washington, DC.
Nienhuis PH, Leuven RSEW. 2001. River restoration and fl ood
protection: Controversy or synergism? Hydrobiologia 444:
85–99.
Obeng LE. 1981. Man’s Impact on Tropical Rivers. In Perspectives
in Running Water Ecology, Lock MA, Williams DD (Eds).
Plenum Press: New York; 265–288.
Ogbeibu AE, Victor R. 1989. The effects of road and bridge
construction on the bank–root macrobenthic invertebrates
of a southern Nigerian stream. Environmental Pollution 56:
85–100.

Peters RH. 1991. A Critique for Ecology. Cambridge University
Press: Cambridge.
Regier HA. 1995. Ecosystem integrity in a context of ecostudies
as related to the Great Lakes Region. In Perspectives on Ecological
Integrity, Westra L, Lemons J (Eds). Kluwer Academic
Publishers: Dordrecht, The Netherlands, 88–101.
Rolston H (III). 1988. Environmental Ethics, Duties to and Values
in the Natural World. Temple University Press: Philadelphia.
Rorty R. 1979. Philosophy and the Mirror of Nature. Princeton
University Press: New Jersey.
Rzoska J. 1976. The Nile: Biology of an Ancient River. DW Junk
Publishers: The Hague, The Netherlands.
(SER) Society for Ecological Restoration Science and Policy
Working Group. 2002. The SER Primer on Ecological
Restoration. [online] URL: www.ser.org (accessed 25 August
2004).
Service RF. 2003. NRC backs ecosystem-wide changes to save
Klamath fi sh. Science 302: 765.
Shrader-Frechette KS. 1982. Environmental impact assessment
and the fallacy of unfi nished business. Environmental Ethics
4: 37–48.
Shrader-Frechette KS. 1994. Ethics of Scientifi c Research. Rowan
& Littlefi eld Publisher, Inc.: Lanham.
Shrader-Frechette KS. 1996. Methodological rules for four
classes of scientifi c uncertainty. In Scientifi c Uncertainty and
Environmental Problem Solving, Lemons J (Ed). Blackwell
Science: Cambridge; 12–39.
Shrader-Frechette KS, McCoy E. 1993. Method in Ecology.
Cambridge University Press: Cambridge.
Sirageldin I. 2002. Human prospects in an age of uncertainty and
decline of rationality. Newsletter of the Economic Research
Forum for the Arab Countries, Iran and Turkey 2: 9–11.
Smits AJ, Niehuis PH, Leuven RSEW. 2000. New Approaches
to River Management. Backhuys Publishers: Leiden, The
Netherlands.
Stewart-Oaten A. 1996. Goals in environmental monitoring. In
Detecting Ecological Impacts: Concepts and Applications in
Coastal Habitats, Schmitt RJ, Osenbert CW (Eds). Academic
Press: San Diego; 17–28.
Sunstein C. 2002. Risk and Reason. Cambridge University Press.
Cambridge.
Sweeney S. 2000. Different means, shared ends: Environmental
restoration and restoration ecology. Biologica 38: 129–139.
(SWS) Society of Wetland Scientists position paper on the defi nitions
of wetlands restoration. 2000. Ecological Applications 6:
84–93.
Uncertainty in River Restoration 13
Tickner J. (Ed). 2002. Environmental Science and Preventive
Public Policy. Island Press: Washington DC.
Tickner JA. 2003. Precaution: Environmental Science and
Preventive Public Policy. Island Press: Washington, DC.
Victor R. 1998. Fish community changes associated with a small
African reservoir. International Review of Hydrobiologia 83:
603–610.
Victor R, Brown CA. 1990. The food and feeding habits of two
species of characid fi sh in a perturbed West Africa river.
Journal of African Zoology 104: 97–108.
Victor R, Dickson DT. 1985. Macrobenthic invertebrates of a
perturbed stream in Southern Nigeria. Environmental Pollution
Series A 38: 99–107.
Victor R, Meye J. 1994. Further studies on the fi sh communities
of a perturbed stream in Southern Nigeria. Journal of Tropical
Ecology 10: 627–632.
Victor R, Ogbeibu AE. 1985. Macrobenthic invertebrates of
a stream fl owing through farmlands in Southern Nigeria.
Environmental Pollution Series A 39: 337–349.
Victor R., Ogbeibu AE. 1986. Reconlonization of macrobenthic
invertebrates in a Nigerian stream after pesticide treatment and
associated disruption. Environmental Pollution Series A 41:
125–137.
Victor R, Ogbeibu AE. 1991. Macrobenthic invertebrates in the
erosional biotope of a Nigerian river. Tropical Zoology 4:
1–12.
Victor R. Onomivbori O. 1996. The effects of urban perturbations
on the benthic macroinvertebrates of a southern Nigerian
stream. In Perspectives in Tropical Limnology, Schiemer F,
Boland T (Eds). SPB Academic Publishing: Amsterdam, The
Netherlands; 233–238.
Victor R, Tetteh JO. 1988. Fish communities of a perturbed
stream in Southern Nigeria. Nigeria Journal Tropical Ecology
4: 49–59.
Ward JV, Stanford JA. 1979. The Ecology of Regulated Streams.
Plenum Press: New York.
Westra L. 1995. Ecosystem integrity and sustainability: The foundational
value of the wild. In Perspectives on Ecological Integrity,
Westra L, Lemons J (Eds). Kluwer Academic Publishers:
Dordrecht, The Netherlands; 12–33.
Westra L, Lemons J (Eds). 1995. Perspectives on Ecological
Integrity. Kluwer Academic Publishers: Dordrecht, The
Netherlands.
Yoccoz NG. 1991. Use, overuse and misuse of signifi cance tests
in evolutionary biology and ecology. Bulletin Ecological
Society America 71: 106–111.

River Restoration: Managing the Uncertainty in Restoring Physical Habitat Edited by Stephen Darby and David Sear
© 2008 John Wiley & Sons, Ltd
2
Sources of Uncertainty in River
Restoration Research
2.1 INTRODUCTION: GENERAL SOURCES
OF UNCERTAINTY
The practice of science in support of river restoration is
subject to four primary sources of uncertainty (see Chapters
1 and 3 for additional/alternative views) so signifi cant
that they may prevent the restoration from achieving its
goals. Firstly, the underlying theory applied by investigators
to particular problem cases is imperfect and contains
substantial gaps in explanatory and predictive capability.
Secondly, the research process itself is subject to a variety
of operational problems that introduce uncertainty to the
use of science. Thirdly, the communication of scientifi c
results to decision makers is often fraught with ambiguity
derived from the scientifi c sender as well as the policy
receiver. Fourthly, the scientists themselves are subject to
bias that generates doubt in the outcome of generating
scientifi c products and applying them. In the following
sections the issues for each of these sources of uncertainty
is outlined.
2.2 UNCERTAINTY IN THEORY
All science in support of river restoration begins with
theory, because it is theory that allows the investigator to
identify what to measure and how to construct a conceptual
model that connects the measurements together.
Investigators perceive only those aspects of the river and
its operations that theory allows them to see. Practitioners
of fl uvial geomorphology tend to revere existing theory as
a sacrosanct starting point but, like all sciences, geomorphology
is in a state of constant change and revision. The
William L. Graf
Department of Geography, University of South Carolina, USA
change is sometimes gradual, as with the development and
application of fundamental hydraulics to explain river
behavior that evolved over a period of several decades
(Chang, 1998; Simons and Sentürk, 1992). Sometimes the
change is abrupt, as was the case with the introduction of
the concepts surrounding hydraulic geometry that burst
upon the fl uvial geomorphology scene, became widely
accepted in less than a decade and continued in common
use for several decades (Leopold, 1994). The result of this
constantly changing theory is that the geomorphologist
working in 2007 may perceive a very different system than
one working just a few years before or later, even though
the physical system in all cases would be the same. Uncertainty,
therefore, is included in the application of science
in its broadest sense.
Another source of ambiguity in theory for river restoration
is the regional specifi city that is built into much
of fl uvial theory. Much of what we theorize about
single-thread meandering rivers comes from research
experience in northwest Europe and eastern North America
(Knighton, 1998), yet the global applicability of this work
is largely untested. Most of the streams of northwest
Europe and eastern North America that have been intensively
investigated are relatively small on a world-wide
basis, and though some generalities certainly must apply
in many locales, the details may differ. Until the late
1990s, much of the theory for dryland rivers came from
experiences in the American Southwest (Graf, 1988), but
more recent investigations in Australia by Gerald Nanson,
Steven Tooth and others, for example, have shown that the
American experience is not applicable in all drylands
(Nanson and Knighton, 1996).

16 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
If it is true that we must theorize based on what we
know best, it must also be true that we are still limited in
the range of our collective experience. As a result, when
we apply existing theory in new geographic settings, there
is reasonable doubt about the applicability of that theory,
at least in its totality. Geomorphologists commonly recognize
that it is unwise to extend statistical models beyond
the numerical ranges of the data. It is equally risky to
extend geomorphological models beyond the geographic
ranges of their origin. The extension of theory also forces
us to consider how much of the geomorphology and
hydrology of a particular river is unique, regardless of its
geographic location. Each reach (a few kilometers long)
of a stream is likely to be unique but the overall operation
and form of a river (hundreds of kilometers long) is likely
to have many similarities with other systems of similar
magnitude. At the more extensive end of this range of
magnitudes, generalizations are possible, while at the local
end of the range of magnitudes, uniqueness becomes more
apparent.
The incompleteness of most fl uvial geomorphologic
theory is also a source of uncertainty. This incompleteness
is in part purely a function of the natural river system, for
which investigators have nine fundamental operating variables,
but for which there are only a very few connective
mathematical functions (Leopold, Wolman, and Miller,
1964). But an equally important limitation of existing
theory is its lack of recognition of human effects. Throughout
much of the twentieth century (with a few exceptions),
geomorphology as a science pursued explanation for
‘natural’ rivers and many investigators made a conscious
effort to avoid the confounding infl uence of technological
infl uences. It has only been in the last twenty years that
those human infl uences, pervasive and signifi cant in many
rivers of the world, have themselves become the objects
of study (Costa et al., 1995; Graf, 2001). By defi nition,
rivers subject to restoration have undergone changes
resulting from human management and technology, but
existing theory is remarkably weak with respect to these
issues.
The Colorado River in the Grand Canyon in the USA
provides an example of the issues related to uncertainty in
theory. Glen Canyon Dam, several kilometers upstream
from the Grand Canyon controls the fl ow of the river, and
especially reduces annual fl ood peaks to less than half of
their former magnitude. The dam also reduces the sediment
supply to the downstream canyon by more than 80%.
As a result, the river has eroded sandy beaches and bars
that once were common in the canyon (National Research
Council, 1996). River restoration for the canyon included
reintroduction of moderate fl oods to move the available
sediment from the channel fl oor to elevated positions,
restoring these ecological niches. Despite considerable
research, there were no established theories to predict the
response of the river to the artifi cial fl oods and although
there have been several fl ood-simulating releases from the
dam, the restoration results are not yet apparent.
2.3 UNCERTAINTY IN RESEARCH
Research using admittedly limited theory in support of
restoration is subject to uncertainty in the specifi cation
of variables, assumptions, sampling, measurements and
testing of hypotheses. The specifi cation or defi nition of
variables, for example, is much entangled in the vagaries
of science, law and personal perception of the researcher.
Channel width provides an example. Most geomorphologists
would agree that channel width is the distance across
the active channel from one bank to the other, but the
application of this seemingly simple proposition is devilishly
diffi cult in many rivers. How should semi-permanent
islands be taken into account? What about ephemeral
bars? How should width be determined in the common
circumstance where multiple sets of banks have resulted
from episodic incision or simply variable fl ows, which is
often the case in arid, semi-arid, arctic or alpine regions.
Many legal systems also defi ne the channel as being
‘between the banks,’ but do not specify which banks to use
for the description (Graf, 1988).
All geomorphological research includes assumptions
which form another source of uncertainty. The geographic
and ecological complexity of rivers and their environments
imply that when conducting investigations it is
essential to focus on a few components and assume away
the importance of variability in other factors that go
unmeasured. In geomorphology, hydrology and engineering
studies that support restoration, investigators often
assume stationarity of the hydro–climatic processes ruling
the river. Stationarity means investigators assume that
the underlying statistical distributions describing climatic
variables important to river processes are unchanging.
Standard magnitude/frequency analysis includes this
assumption so that the researcher can address other variables
of interest to planners, including the return intervals
for various magnitudes of discharge. However, climate is
anything but stationary, and its variation is highly likely
to infl uence the statistical distributions upon which return
interval concepts depend. This variation is also likely to
be signifi cant to the fl uvial system over time scales as
short as decades, scales that encompass the likely project
life of most restoration efforts. Predictions for the nearterm
future of a few decades are therefore uncertain
because the effects of expected climatic changes are not
part of the analysis (see Chapter 13).

Sampling processes in fl uvial geomorphology also cast
doubt on the confi dence users may have in the reliability
of the resulting data. Rivers present the researcher with
long lines or corridors, and if the investigator uses cross
sections or point samples, the selection of the locations of
these sample points may strongly infl uence the resulting
data. The spacing of meanders, riffl es and pools introduces
some regularity to the spacing of important geomorphic
and hydraulic features of rivers, so that if investigators use
regular spacing for sample cross sections or sites, that
spacing may coincide with the spacing of particular
characteristics of the channel. For example, sample sites
might occur only on riffl es, or only in pools, giving a false
picture of the system. Truly random spacing for samples
may be statistically desirable, but a realistic view of geomorphic
and hydrologic conditions demands that all the
various environments of the channel be included in the
sample, something that is not possible without some
understanding of the basic spatial framework of the
system.
Uncertainty may be lessened if scale is part of the planning
process for selecting sample sites. Sample schemes
couched in the concept of river reaches permit bracketing
of relevant parts of the stream. A river reach is from one
to a few kilometers in length and contains similar geomorphic
conditions throughout its extent, sometimes with
repetitive and alternating channel confi gurations such as
pools and riffl es, pools and rapids, or meanders. Several
reaches may make up a river segment. Geological boundaries,
confl uences with major tributaries or human structures
such as dams and diversion works create the upstream
and downstream boundaries of each segment. Sampling
schemes constructed with the reaches and segments as
geographic frameworks are more likely to be informative
for restoration work than truly regular or truly random
samples. Project design for restoration requires a clear
assessment of the appropriate dimensions and spacing of
repetitions to insure long term stability. If the dimensions
and spacing are not refl ective of the restored hydraulic
conditions, the restored system will be unstable and may
disintegrate.
The example of restoration of the Platte River in
Nebraska in the USA illustrates the role of uncertainty in
research. Upstream fl ood control dams have brought about
great changes in the hydrologic regime and geomorphology
of the Platte, a serious issue because the river
hosts several endangered bird species. The river originally
included valuable habitats for birds, particularly the
whooping crane, but the hydrologic and geomorphic
changes have reduced their habitat. River restoration
includes returning the river to conditions closer to those
that prevailed before the dams were in place. Features such
Sources of Uncertainty in River Restoration Research 17
as multiple channels, high and low islands, and complex
bars are essential to the restoration. Research to support
the restoration has been under way for more than a decade
but has produced results that are diffi cult to interpret
(National Research Council, 2004). Cross-sectional
surveys, for example, are numerous but not often conducted
at the same places through time so that sampling
is an issue. The fl ow parameters that are most important
to the species may not be the same parameters that are
important to the geomorphology. Stationarity is a particular
problem in dealing with the hydrologic records of the
Platte because the river is located on the boundary between
sub-humid and sub-arid regions and is subject to climatic
fl uctuations on decadal and century-long periods which
casts doubts on shorter term records.
2.4 UNCERTAINTY IN COMMUNICATION
The successful use of science in formulating public policy
for the restoration of rivers relies on accurate communication
between researchers and decision makers, but this
connection sometimes suffers from failings by both participants.
Scientists are usually accustomed to communicating
with each other using a specifi c shared language of
technical terms. Even terms shared with the general language
may take on specifi c shadings of meaning when
used by specialists communicating with each other. A
steep river gradient, for example, calls to mind a general
picture for most geomorphologists, perhaps of a step-pool
sequence for mountain streams or a braided channel in
other settings. For the decision maker, a steep river may
be one with water falls. More importantly, specialized
terms and terms with nuances do not work well when the
listener or reader is not trained or experienced in the scientifi
c specialty. As a result, scientists have an obligation
to use language without jargon and without assumptions
when communicating with decision makers and the
informed public, a task that seemingly challenges many
specialists.
Decision makers, on the other hand, have poorly
informed expectations regarding their scientifi c advisors.
They expect clear, unambiguous answers to their questions
and specifi c, robust predictions of future processes that
might result from a variety of potential decisions. Science,
however, rarely offers truly unambiguous conclusions, and
caveats are many in applied geomorphology. Often, it is
possible to predict the direction of change, but there is
greater trouble predicting the magnitude of that change.
To a public servant attempting to protect property values,
such predictions may lack the required level of comfort.
A common error in communication that results in
uncertainty occurs when the decision maker requests

18 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
specifi c numbers, such as ‘what will be the stable, long
term width of the restored channel’, a reasonable question
in the public policy arena. However, most scientifi c and
engineering models predict some most likely value, enveloped
in error bars. The most effective (and honest)
report includes the most probable value but also includes
some discussion of potential deviation from that expected
value. Reporting of potential errors and ranges of possible
values rather than single numbers protects the interest of
everyone involved and produces more reasonable expectations
on the part of the general public.
Restoration of the Elwha River in the state of Washington
exemplifi es the issues surrounding uncertainty in communication.
National legislators directed that the river be
restored to its ‘natural’ condition for the benefi t of endangered
salmon, which use the river for spawning during
annual migrations (US House of Representatives, 1992).
The method of restoration centers on the removal of two
large dams. However, from a scientifi c standpoint, it is
impossible to meet the requirements of the law, because
even with the removal of the dams, the river will not even
approach truly natural conditions. Upstream land use,
including logging, have altered the basic hydrologic and
sediment regimes of the river, and in downstream areas
levees and other structures prevent natural river processes.
In fact, ‘natural’ conditions are a model to which restoration
might aspire, but the principle as a true objective is
unworkable.
2.5 BIAS
A fi nal source of uncertainty in science for river restoration
lies within the intellect of the researchers themselves.
All geomorphologists are products of their cultural backgrounds,
academic training, experiences and personal
characteristics (see Chapters 1 and 4). Researchers raised
in intellectually liberal surroundings may be more questioning
of established theories than those raised in more
rule-bound settings, and each of these types is likely to
approach problems, evidence and methods with different
biases and preferences. Gender may play a subtle, but as
yet relatively unexplored role in our approaches to science.
Academic training for fl uvial geomorphology is highly
variable from one group of teachers to another, with some
groups emphasizing a stochastic rather than a deterministic
approach. Other contrasting styles include greater
emphasis on geographic approaches as opposed to those
more strongly oriented toward engineering, research
designs that emphasize small-scale (particle-sized) as
opposed to ecosystem (or landscape) scale perspectives,
or empirical rather than model-based approaches. Once
trained, the researcher is further formed by personal fi eld
experiences, with extensive backgrounds in dryland settings
producing a different perspective than experiences
dominated by humid subtropical environments, tropical
settings or polar landscapes. Finally, scientists are just like
everyone else: some are stubborn, some are open minded;
some are methodical while their colleagues make successful
leaps of logic; and some are quick to reach judgements
while others seem never to reach closure on their
conclusions.
There is nothing inherently wrong with biases, for to be
biased is to be human. In river restoration, however, and
especially when dealing with decision makers, the wise
scientists fi nd ways to communicate their biases to the
consumers of the scientifi c products. If the consumer
(policy maker or informed citizen) recognizes the biases
and takes them into account, the interpretation and application
of the scientifi c products is confi dent by all parties
involved. Admission of biases by the researcher, usually
in the form of a brief disclosure that clearly defi nes
the schools of thought and experience of the researcher,
enhances the professionalism of the investigator and fairly
discloses to the consumer the background of the knowledge
that forms the basis of decisions pertaining to public
resources. In unusual cases, critics may contest decisions
in administrative hearings or in court proceedings, two
venues where biases are likely to be revealed and explored
in a combative environment. If biases have previously
been revealed, they lack sinister overtones in a strategy
that benefi ts both researcher and consumer.
2.6 CONCLUSION
Uncertainty is a fact of life as well as an inescapable
feature of scientifi c research and decision making for river
restoration. Therefore, the researcher has two essential
options: either ignore the uncertainty and hope that it is
not debilitating for the project at hand, or accept the uncertainty
and use it as a feature of the research. The researcher
can investigate the uncertainty, quantify it in some cases,
and reveal it in explicit terms when reporting results. In
this latter approach, uncertainty becomes an integral part
of research for river restoration, a feature of the work that
is a welcome challenge to be embraced and used to achieve
a more effective end product. Project design may include
a variety of channel dimensions and characteristics, for
example, and avoid relying on a single rigidly defi ned
morphology, so that if original understandings of the
system are not exactly correct the fi nal project will have
some fl exibility. In other cases, it may be wise to simply
allot more space for channel changes in the designed
project to accommodate unforeseen adjustments. By
dealing directly with uncertainty, researcher and decision

maker increase the probability in successfully restoring a
river with enhanced environmental and social benefi ts.
REFERENCES
Chang HH. 1998. Fluvial Processes in River Engineering.
Krieger: Malabar, Florida.
Collier MP, Webb RH, Andrews ED. 1997. Experimental fl ooding
in the Grand Canyon. Scientifi c American 276: 82–89.
Costa JE, Miller AJ, Potter KW, Wilcock PR (Eds). 1995. Natural
and Anthropogenic Infl uences in Fluvial Geomorphology: The
Wolman Volume, Geophysical Monograph 89, American Geophysical
Union: Washington, DC.
Graf WL. 1988. Defi nition of fl ood plains along arid-region
rivers. In: Baker VR, Kochel RC, Patton PC (Eds), Flood
Geomorphology, John Wiley & Sons, Inc.: New York, NY;
231–242.
Graf WL. 2001. Damage Control: Dams and the physical integrity
of America’s rivers. Annals of the Association of American
Geographers 91: 1–27.
Sources of Uncertainty in River Restoration Research 19
Knighton D. 1998. Fluvial Forms and Processes: A New Perspective.
Arnold: London.
Leopold LB, Wolman MG, Miller JP. 1964. Fluvial Processes in
Geomorphology. WH Freeman: San Francisco, California.
Leopold LB. 1994. A View of the River. Harvard University Press:
Cambridge, Massachusetts.
Nanson GC, Knighton AD. 1996. Anabranching rivers: Their
cause, character and classifi cation. Earth Surface Processes
and Landforms 21: 217–239.
National Research Council. 1996. River Resource Management
in the Grand Canyon. National Academy Press: Washington,
DC.
National Research Council. 2004. Endangered and Threatened
Species of the Platte River. National Academy Press: Washington,
D.C.
Simons DB, Sentürk F. 1992. Sediment Transport Technology.
Water Resources Publications: Littleton, Colorado.
US House of Representatives. 1992. Joint Hearings on HR 4844,
Elwha River Ecosystem and Fisheries Restoration Act, 102nd
Congress, 2nd Session. Government Printing Offi ce: Washington,
DC.

3.1 INTRODUCTION
River Restoration: Managing the Uncertainty in Restoring Physical Habitat Edited by Stephen Darby and David Sear
© 2008 John Wiley & Sons, Ltd
3
The Scope of Uncertainties in
River Restoration
Joseph M. Wheaton 1 , Stephen E. Darby 2 and David A. Sear 2
1 Institute of Geography and Earth Sciences, The University of Wales, UK
2 School of Geography, University of Southampton, UK
The science and practice of river restoration are both
still very much in their adolescence (Palmer et al., 1997).
Yet, both have been graced with funding and support
from a diverse range of interest groups (Malakoff, 2004).
One of the premises of this book is that if funding is to
continue to be allocated to river restoration, it will have to
be shown that river restoration is ‘working’ (see Preface;
Wissmar and Bisson, 2003c). Defi nitions of ‘working’
(often equated with success) are understandably subjective
and vulnerable to uncertainties in the river restoration
process, societal values, the fl uvial system and ecosystem
response to restoration management activities. Davis and
Slobodkin (2004) argued that defi ning restoration goals
and objectives is rightfully a value-based activity, as
opposed to scientifi c activity. Each activity is inherently
uncertain. Paradoxically, the uncertainties infl uencing
river restoration projects are rarely recognised or quantifi
ed, much less reported to stakeholders or the public
(Walters, 1997).
The topic of uncertainty in river restoration is riddled
with complexity and confusion. Indeed, uncertainty manifests
itself in many ways, as established in Chapters 1 and
2. Lemons and Victor (see Chapter 1) have already illustrated
how deep the value-laden dimensions of uncertainty
lies, not just in decision making, but in scientifi c research
as well. Graf (see Chapter 2) expanded on this theme,
citing uncertainties from theories, the research itself, communication
and biases among investigators. He concluded
that the research can either ‘ignore the uncertainty and
hope that it is not debilitating for the project at hand,
or accept the uncertainty and use it as a feature of the
research.’ In this chapter the rich topic of uncertainty is
presented in a broader, more generic context. This foundation
is intended to help separate the sources and types of
uncertainties that the various authors in this book present,
and meanwhile unravel some of the ambiguities surrounding
uncertainty in river restoration.
As cautioned earlier, potentially signifi cant uncertainties
are rarely recognised, much less explicitly dealt
with in river restoration. Hence, a lexicon and typology
for uncertainty is outlined fi rstly in this chapter. This is
done to dispel the notion of a certain world with certain
outcomes within the broader scope of types and sources
of uncertainty. A return specifi cally to river restoration
then follows to identify types of uncertainties using the
above-mentioned typology. The tremendous diversity of
river restoration in the context of uncertainties arising
from restoration motives, notions and approaches are considered.
A case will be made that a basic strategy for
dealing with uncertainty is needed by the river restoration
community to allow both the community and individual
investigators or practitioners to:
explore the potential signifi cance (both in terms of
unforeseen consequences and welcome surprises) or
insignifi cance of uncertainties;
effectively communicate uncertainties;
eventually make adaptive, but transparent, decisions in
the face of uncertainty.

22 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Finally, it will be argued that amongst the various strategies
for dealing with uncertainty, the only strategy that might
provide these aims is one of embracing uncertainty.
3.1.1 The Status Quo in River Restoration
The rapid rise and international popularity of river restoration
is both encouraging and worrisome (Kondolf, 1996).
Although sparse examples dating back to the 1930s exist1 ,
river restoration has primarily developed on the coat tails
of the environmental awareness movement of the late
1970s (Graf, 1996; Sear, 1994). It is encouraging that so
much enthusiasm exists to restore rivers. Yet, it is interesting
to note the societal choices between some mix of
reactive restoration efforts in response to damage already
done, as opposed to pro-active conservation actions to
prevent further damage (Boon, 1998). The international
popularity of river restoration is evident in the restoration
literature (e.g. restoration in 21 different countries reported
in Nijland and Cals, 2000), restoration databases (e.g. the
United Kingdom River Restoration Centre (RRC), United
States Environmental Protection Agency (EPA)) 2 and an
International River Restoration Survey3 launched by
Wheaton et al. (2004c) with respondents from 34 different
countries. In Denmark alone, 1068 restoration projects had
been completed by Danish regional authorities by 1998
(Hansen and Iversen, 1998); whereas in the United States,
Malakoff (2004) reported that by 2004 more than $US10
billion had been spent on a total of more than 30 000
projects. The popularity of river restoration is apparent in
international, national, regional and local public policy that
actively promotes, requires and, in some cases, funds river
restoration efforts (Jungwirth et al., 2002). However, their
effectiveness is constrained by limited funds and scope to
deal with closely related land use issues and other sociopolitical
goals (Tockner and Stanford, 2002).
Despite the popularity of river restoration in the developed
nations of the world, the global decline of the physical
and ecological integrity of rivers is diffi cult to overstate
(Jungwirth et al., 2002; Vitousek et al., 1997). Indeed,
1 The United States Department of Agriculture Forest Service
started undertaking ‘stream improvement’ in the 1930s with the
intent of increasing salmonid production (Everest and Sedell,
1984).
2 RRC Database includes over 750 projects within the United
Kingdom: http://www.therrc.co.uk; the USEPA River Corridor
and Wetland Restoration Database includes over 600 projects
throughout the United States: http://yosemite.epa.gov/water/
restorat.nsf/rpd-2a.htm.
3 Complete real time results, background information and forthcoming
interpretations are available on the web: http://www.geog.
soton.ac.uk/users/WheatonJ/RestorationSurvey_Cover.asp.
most restoration efforts still pale into signifi cance relative
to expanding anthropogenic impacts on riverine landscapes
(Tockner and Stanford, 2002). Even in parts of the
world where numerous river restoration efforts are already
underway (i.e. Europe, North America and Australia), wetlands
are actively being drained and fi lled, rivers are still
diverted and regulated, urban growth is encroaching into
fl oodplains and headwaters, while we continue to permanently
alter basin hydrology and fragment habitats (Collins
et al., 2000; Moss, 2004; Mount, 1995). These problems
pose even larger threats in the developing nations of
the world (Marmulla, 2001). Over 250 new major dams
become operational worldwide annually and 75 are
planned for the Amazon Basin alone (Robinson et al.,
2002). It seems logical that preservation should be easier
to achieve than restoration (Frissell et al., 1993), but there
seems to be excessive confi dence in the ability to restore
(Stewardson and Rutherfurd, see Chapter 5), sometimes
reducing restoration to a mitigation measure justifying
planned impacts or maintaining the status quo. Both conservation
and restoration are based on the transformation
of uncertain science and uncertain notions of what is
natural, ecosystem integrity and physical integrity into
societal goals (Graf, 2001; Lemons and Victor, see Chapter
1). Additionally, the good intentions of restoration projects
may lead to unintended but often foreseeable consequences.
Even if society is willing to make diffi cult sociopolitical
decisions to support preservation and restoration
of rivers, there is no guarantee of desired outcomes
following.
Given the dynamism of rivers, it seems obvious that the
outcomes of restoration projects are uncertain. However,
the restoration community seems hesitant to admit that the
goals and science that restoration are founded upon are
uncertain too (Stewardson and Rutherfurd, see Chapter 5).
Aside from indirect references to uncertainty in adaptive
management programs, the river management community
has largely brushed uncertainties aside (Clark, 2002;
Wissmar and Bisson, 2003c). It is unclear whether this is
a conscious or passive decision, though individual decisions
to ignore uncertainty can be plausibly attributed to
one or more of the following:
ignorance of uncertainty and/or its signifi cance;
the hope that uncertainty is insignifi cant;
an acknowledgement of uncertainty, but not knowing
how to deal with it;
being misinformed about uncertainty, leading to the
assumption that it is insignifi cant;
being knowledgeable about uncertainty, but having
established its insignifi cance.

Newson and Clark (see Chapter 14) attribute the river
manager’s current treatment of uncertainty to a ‘riskaverse’
management culture that prefers to entrench itself
in ‘rituals of verifi cation’ aimed at minimising liability
(Power, 1999). Uncertainty is also frequently misunderstood
by the general public (Pollack, 2003; Riebeek, 2002)
as something negative and undesirable (Newson and
Clark, see Chapter 14). A widespread misconception that
science embodies certain knowledge persists in the reports
of the mainstream media and views of the general public
(Clark, 2002; Riebeek, 2002). Such misconceptions fuel
expectations that science-based approaches to river restoration
will yield positive outcomes. Ironically, people confront
uncertainties everyday without hostility and choose
to routinely make decisions about the future (Pollack,
2003).
Restoration science and the restoration literature are not
much further along than practitioners and decision makers.
Wissmar and Bisson (2003b) asserted that ‘a better understanding
of variability and uncertainty is critical to the
successful implementation of restoration programs for
aquatic and riparian systems.’ Yet, buried within a rich
literature on restoration are only occasional passing
mentions of uncertainty (Brookes and Shields, 1996) and
a handful of explicit treatments (Johnson and Brown,
2001; Johnson et al., 2002; Johnson and Rinaldi, 1997;
Johnson and Rinaldi, 1998; Wissmar and Bisson, 2003c).
These studies understandably tend to focus on a specifi c
type of uncertainty that might be reasonably articulated
within a specifi ed page limit, so a more holistic treatment
of uncertainty is necessary (Newson and Clark, see
Chapter 14; Van Asselt, 2000). Restoration is established
as one important component of environmental management.
It would be a shame to lose what public support
already exists for restoration if political scrutiny recasts
unrealistic expectations of river restoration as a ‘failure’,
as opposed to the inadequate consideration of uncertainty
they truly stem from.
3.2 WHAT DO WE MEAN BY UNCERTAINTY?
3.2.1 A Lexicon of Uncertainty
In the simplest sense, uncertainty is a lack of sureness
about something or someone (Merriam-Webster, 1994).
However, uncertainty can be more than simply a lack
of knowledge. It persists even in areas where knowledge
is extensive; and knowledge does not necessarily equate
to truth or certainty (Van Asselt and Rotmans, 2002).
There are at least 24 potential synonyms for the noun
uncertainty and 27 synonyms for the adjective uncertain
The Scope of Uncertainties in River Restoration 23
(Table 3.1). There are a number of concepts related to and
infl uenced by uncertainty, but which differ from uncertainty
itself. A selection of these concepts is considered
briefl y below.
Accuracy: Accuracy refers to correctness or freedom from
error. In measurement, accuracy refers to how close an
individual measurement is to the ‘true’ or ‘correct’ value
(Brown et al., 1994). The classic accuracy analogy is the
location of darts on a dart board – the closer the darts are
to the intended position (bull’s-eye) the more accurate. If
one can be certain about both the ‘true’ value (e.g. the
position of the bull’s-eye) and the value of the individual
measurement (e.g. the position of the dart), then the accuracy
is actually a certainty. In practice, accuracy statements
are uncertain because ‘true’ values are often
assumed and measurements have limited precision.
Table 3.1 Potential synonyms of the noun ‘Uncertainty’ and
the adjective ‘Uncertain’
Synonyms of Uncertainty Synonyms of Uncertain
Ambiguity Ambiguous
Indeterminacy Causeless
Capriciousness Capricious
Chance Probabilistic
– Deferred
Danger Dangerous
Disbelief Disbelieving
Equivocation Equivocal
Doubt Doubtful
– Erratic
Expectation –
Future condition –
Hesitation Hesitant
Ignorance Ignorant
Improbability Improbable
Indecision Indecisive
Indeterminacy Indeterminant
Insecurity Insecure
Irresolution –
Obscurity Obscure
Surprise Surprising
– Unauthentic
Unintelligibility Unintelligible
– Unexplained
– Questionable
Vacillation Vacillating
Vagueness Vague
– Undecided
Unsureness Unsure
Unpredictability Unpredictable

24 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Confi dence: Confi dence (e.g. in a statement, hypothesis,
measurement, feeling or notion) relates to the degree of
belief or level of certainty. Confi dence levels, for example,
describe the probability that a given population parameter
estimate falls within a designated continuous statistical
confi dence interval.
Divergence: Divergence describes a situation when similar
causes produce dissimilar effects (Schumm, 1991). Divergence
relates to uncertainty in situations where problems
of cause and process are under consideration.
Error: Error is the difference between a measured or calculated
value and a ‘true’ value. In every day conversation,
an error is a mistake. In science, error is the metric by
which accuracy is reported and is not a synonym for uncertainty
(Ellison et al., 2000). A ‘true’ value is certain by
defi nition. If the error between the ‘true’ value and a
measured or calculated value is known there is no uncertainty
in principle. However, in practice ‘true’ values are
often not known and instead are assumed to be so, while
the measured or calculated value may be uncertain. Hence
error becomes representative of uncertainty. Once errors
are calculated, it can be helpful to consider whether the
error is systematic or random. Systematic errors stem from
consistent mistakes and are often constant or predictable,
affecting the mean of a sample (i.e. bias, Trochim, 2000).
Systematic errors can potentially be constrained as their
source is identifi able. By contrast, random errors infl uence
the variability of a sample (not the mean) and are generally
unpredictable or unconstrainable (Trochim, 2000).
Exactness: Exactness is really a synonym for accuracy.
However, it is worth pointing out that exactness has quite
a different meaning to exact. Exact statements or exact
numbers, in principle, have no uncertainty about them.
They are statements of truth. By contrast, exactness is a
relative measurement assigned to inexact statements or
values (i.e. those with some uncertainty).
Expectation: Expectation has to do with anticipation of
probable or certain events. Uncertainty fundamentally
relates to expectations. When uncertainties are unknown,
not fully considered or ignored, the degree that expectations
may be unrealistic will generally increase.
Equifi nality: Equifi nality (also referred to as convergence),
arises when different processes and causes produce
similar effects (Schumm, 1991). In a modelling context,
Beven (1996a; 1996b) suggests that ‘the consequences of
equifi nality are uncertainty in inference and prediction.’ In
a social context, a potentially limitless range of possibilities
may lead to a single event, such as the election or
defeat of a politician.
Precision: Precision is a measure of how closely individual
measurements or calculations match one another
(Brown et al., 1994). Recalling the dart board analogy
from accuracy, a precisely thrown set of darts will cluster
around one another, but may be nowhere near the bull’seye.
In measurement, the precision of an instrument refers
to the fi nest scalar unit the instrument can resolve. Precision
is related to uncertainty in that it defi nes a detection
threshold, below which differences can not be discerned.
Repeatability: Repeatability can be viewed as either the
ability to reproduce the same measurement, result or
calculation or the variability in repeated measurements,
results or calculations. Uncertainty can simply limit
repeatability or increase variability.
Risk: Risk is a measure of likelihood that an undesirable
event or hazard will occur (Merriam-Webster, 1994). Ward
(1998) credited Knight (1921) for making the important
clarifi cation between risk and the type of uncertainty for
which there exists ‘no valid basis of any kind for classifying
instances’:
‘He used the term “risk” for situations in which an
individual may not know the outcome of an event,
but can form realistic expectations of the probabilities
of the various possible outcomes based either on
mathematical calculations or the history of previous
occurrences.’
Newson and Clark (see Chapter 14) contrast risk (with
‘known’ impacts and probabilities) with uncertainty (with
‘known’ impacts but ‘unknown’ probabilities) and ignorance
(with ‘unknown’ impacts and probabilities).
It is worth noting that uncertainty itself and all the
related concepts outlined above are described in terms of
their ‘degree’. That is, none of these concepts are simple
Aristotelian two-valued logic concepts (e.g. true–false).
Each concept is measured along a continuum of values
with end-members of total uncertainty (complete irreducible
ignorance) and absolute certainty. Probabilistic uncertainty
is an example of a quantifi cation of uncertainty, yet
not all uncertainty is quantifi able. To quantify uncertainty
it is necessary to estimate the degree of our limited knowledge.
Yet, if a condition of irreducible ignorance is considered
as one extreme of uncertainty, it is diffi cult at best
to estimate the degree of something we do not even know
exists. Within this broad view of uncertainty, uncertainty
might also be considered along a continuum that refl ects
our ability to quantify it (Figure 3.1).
In summary, when someone mentions uncertainty casually,
it is diffi cult to discern whether they are referring to

limited knowledge, a lack of knowledge altogether, or one
of the above-mentioned concepts that are infl uenced by
uncertainty. Moreover, the lexicon provided here contains
concepts that are highly inter-related and easily confused.
Similar to vague, pseudo-scientifi c buzzwords and catchall
phrases like holistic and integrated, the term ‘uncertainty’
alone evidently has little meaning until its details
are unravelled.
3.2.2 A Typology for Uncertainty
Since uncertainty is so hard to defi ne, a classifi cation of
uncertainty is often used (Van Asselt and Rotmans, 2002).
The utility of any typology or classifi cation is ultimately
dependent on its application (Kondolf, 1995b; Lewin,
2001). Rotmans and van Asselt (2001) astutely pointed
out ‘there is not one overall typology that satisfactorily
covers all sorts of uncertainties, but that there are many
possible typologies’. In the context of this review, a
typology was sought which considered sources of uncertainty
and did not unnecessarily ignore any type of
uncertainty. The existing van Asselt (2000) typology was
chosen over others because of its generic and inclusive
consideration of uncertainty. The typology was fi rst introduced
in detail in van Asselt (2000) and concisely reviewed
in Rotmans and van Asselt (2001) and van Asselt and
Rotmans (2002).
At the highest level, two sources of uncertainty exist:
uncertainty due to variability and uncertainty due to
limited knowledge (Figure 3.2). Van Asselt and Rotmans
(2002) presented uncertainty due to variability fi rst as
these uncertainties ultimately combine to contribute to
uncertainty due to limited knowledge. Environmental
management is concerned with the management of inherently
variable natural and managed systems. Knowledge
The Scope of Uncertainties in River Restoration 25
Figure 3.1 The quantifi able continuum of uncertainty (Once uncertainties are acknowledge as unquantifi ed uncertainties, increased
knowledge about the uncertainties will determine their position on the continuum.)
about natural change and variability in ecosystems, fl uvial
systems and hydrologic systems is incomplete and hence
contributes to uncertainty due to limited knowledge
(Wissmar and Bisson, 2003a). Five distinct subclasses
of uncertainty due to variability are proposed: inherent
natural randomness, value diversity (socio-political),
behavioural diversity, societal randomness and technological
surprise. Inherent natural randomness is attributed to
‘the nonlinear, chaotic and unpredictable nature of natural
processes’. The natural variability of river systems should
be a fundamental consideration in integrated river basin
management and restoration; it is reviewed thoroughly in
Wissmar and Bisson (2003c). Value diversity, behavioural
diversity and societal randomness each contribute to
uncertainties in environmental management, particularly
through stakeholder negotiations, public support, project
funding, policy making and individual perspectives. Technological
surprises result from new breakthroughs, which
may provide unforeseen benefi ts and/or bring unforeseen
consequences.
Van Asselt and Rotmans (2002) separated seven types
of uncertainty due to limited knowledge. Unlike uncertainties
due to variability, these are thought to map out
along a continuum that refl ects the relative degree of
uncertainty. At the highest degree of uncertainty are four
‘structural uncertainties’ (van Asselt and Rotmans,
2002):
Irreducible ignorance: ‘We cannot know.’
Indeterminacy: ‘We will never know.’
Reducible ignorance: ‘We do not know what we do not
know.’
Confl icting evidence: Knowledge is not fact but interpretation,
and interpretations frequently contradict and
challenge each other. ‘We don’t know what we know.’

26 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Figure 3.2 Typology for sources and degree of uncertainty (Adapted from Van Asselt’s (2000) proposed typology for uncertainties
in integrated assessment.)
Van Asselt and Rotmans (2002) then proposed a transition
into ‘unreliability’ uncertainties of a relatively lesser
degree:
Practically immeasurable: A lack of data or information
is always a reality in studying natural systems.
Not only are many natural phenomena incredibly diffi
cult or impossible to measure, all are fundamentally
limited by problems of temporal and spatial resolution,
up-scaling and averaging (Kavvas, 1999). ‘We
know what we don’t know’ (Van Asselt and Rotmans,
2002).
Lack of Observations and Measurements: Although
in principle this is easy to identify and augment, in
practice this is always a factor. Borrowing from van
Asselt and Rotmans (2002): ‘could have, should
have, would have, but didn’t.’
Inexactness: Related to lack of precision, lack of
accuracy, measurement and calculation errors. Under
Klir and Yuan’s (1995) typology, these are considered
‘fuzziness’ or vagueness.
The van Asselt (2000) typology is both more general
and detailed than other typologies such as Klir and Yuan
(1995). However, all provide a reasonable means to deal
with the fi rst step to understanding uncertainty. Namely,
they allow a systematic identifi cation of sources and types
of uncertainties that could work in either individual river
restoration projects or international policy making on
water and environmental management (see also Chapters
1 and 2). In practice, it is recognised that the semantics of
uncertainty will always be interpreted differently in different
professional contexts (Newson and Clark, see Chapter
14). However, within the context of this chapter, the van
Asselt (2000) typology and associated meanings will be
used consistently.
3.2.3 How do Knowledge and Uncertainty Relate?
The positivist view (Van Asselt and Rotmans, 2002) contends
that as knowledge increases, uncertainty decreases.
Brookes et al. (1998) made the more restrictive but
contradictory generalisation that ‘as knowledge relating
to rivers and their fl oodplains increases, uncertainty is
increased rather than decreased.’ So, which is it? In reality,
there is no unique relationship between uncertainty and
knowledge (Van Asselt and Rotmans, 2002), nor is uncertainty
a fi xed quantity that will always be reduced by scientifi
c research (Jamieson, 1996). It is a highly contextual
relationship dependent on the type of uncertainty (i.e.
uncertainty due to lack of knowledge versus variability)
and the specifi c circumstances under consideration. A few

examples of potential relationships between knowledge
and uncertainty using the nomenclature of the van
Asslet typology are illustrated in Figure 3.3. Having
established the basic terminology of uncertainty, it is
possible to discuss the sources of uncertainty within
river restoration.
The Scope of Uncertainties in River Restoration 27
3.3 REVISITING RIVER RESTORATION
AND UNCERTAINTY
It is diffi cult to generalise about the importance of uncertainty
simply because restoration activities and the restoration
community itself are so diverse. The stakeholders who
Figure 3.3 Some potential relationships between knowledge and uncertainty through time (Contrary to the argument of the positivist,
no unique inverse relationship between uncertainty and knowledge exists.)

28 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
initiate river restoration projects include private individuals,
non-governmental organisations (NGOs), governmental
organisations and various collaborative combinations
of the above. The restoration community is also comprised
of practitioners, decision makers and scientists. No attempt
is made here to list ‘all’ the uncertainties encountered
throughout the restoration process as the daunting list
would never be comprehensive, and is entirely perspective
and project specifi c. For example, there is little consensus
over the meaning of the term ‘river restoration’ with at
least 30 different authors proposing different defi nitions
(Lemons and Victor, see Chapter 1; NAP, 2002; Newson,
2002; Sear, 1994; Stockwell, 2000). Similar to Shields
et al. (2003), ‘river restoration’ in this book is used as a
catch-all term for a variety of management responses and
activities used to address perceived problems with rivers
(Kondolf, 1996). As a starting point, a generic decision
process, which most restoration projects loosely follow,
highlighting some of the common sources of uncertainty
is mapped out in Table 3.2.
3.3.1 Motives for Restoration
Once river restoration projects gain momentum, it is easy
to lose sight of why they were originally envisioned
(Stewardson and Rutherfurd, see Chapter 5). Here, the
motives for restoration are considered to represent more
generalised aims than formalised and specifi c restoration
objectives and activities (i.e. the ‘why’ instead of the
‘what’). Eight common types of motives for river restoration
(still others exist) are listed below:
1. Ecosystem Restoration
2. Habitat Restoration
3. Flood Control/Defence
4. Floodplain Reconnection
5. Property and Infrastructure Protection (bank
stability)
6. Sediment Management
7. Water Quality
8. Aesthetic and Recreational.
Considerable overlap exists between many of the above.
For example, fl oodplain reconnection can be a type of
fl ood control. Habitat restoration and water quality restoration
are sometimes considered forms of ecosystem
restoration. In another example, water quality restoration
could be viewed by some as sediment management or
by others as aesthetic or recreational restoration. Thus, a
hierarchical organisation of restoration motives would be
highly subjective and dependent on individual values and
perspectives. This in itself is not necessarily problematic.
However, it represents a form of communication uncertainty
arising out of value diversity which is often taken
for granted. Once the motives (why to do it) for restoration
are established, restoration aims fall into place, but more
specifi c objectives (what and how to do it) require careful
consideration.
Table 3.2 Sources of uncertainty in an environmental management decision process structure (Adapted from Chapman & Ward
(2002))
Stage in Decision Process Uncertainty About
Monitor the environment and current operations Completeness, veracity and accuracy of information received, meaning
within the organisation
of information, interpretation of implications
Recognise an issue Signifi cance of issue, urgency, need for action
Scope the Decision Appropriate frame of reference, scope of relevant organisation activities,
who is involved, who should be involved, extent of separation from
other decision issues
Determine the performance criteria Relevant performance criteria, whose criteria, appropriate metrics,
appropriate priorities and trade offs between different criteria
Identify alternative courses of action† Nature of alternatives available (scope, timing, logistics involved), what
is possible, level of detail required, time available to identify
alternatives
Predict the outcomes of courses of action† Consequences, nature of infl uencing factors, size of infl uencing factors,
effects and interactions between infl uencing factors (variability and
timing), nature and signifi cance of assumptions made
Choose a course of action How to weigh and compare predicted outcomes
Implement the chosenalternative* How alternatives will work in practice
Monitor and reviewperformance‡ What to monitor, how often to monitor, when to take further action
† = Most decision support systems only provide input at these levels; * = The precautionary principle is implemented here; ‡ = Adaptive management
starts here and feeds back through the process as necessary.

Many have argued that uncertainty in assessing restoration
success arises from inadequate, vague and unclear
restoration objectives (Jungwirth et al., 2002; Kondolf,
1995a; Skinner et al., see Chapter 10). Motives may serve
well as aims (not necessarily to be achieved by an individual
project) but they are insuffi cient to act as detailed
project objectives, which in principle should be achievable.
Using restoration motives carelessly as objectives
produces unrealistic expectations. For example, in a recent
request for proposals to fund community-based river restoration
projects by American Rivers and the National
Oceanic and Atmospheric Association, applicants were
asked to demonstrate that their project: will successfully
restore anadromous fi sh habitat, access to existing
anadromous fi sh habitat, or natural riverine functions; is
the correct approach, based on ecological, social, economic,
and engineering considerations; will minimise any
identifi able short or long term negative impacts to the river
system as a result of the project . . .’
The problem with requiring an applicant to make such
bold statements about individual projects is that it asserts
a level of confi dence in restoration simply not warranted
by current science or practice and creates unrealistic
expectations 4 (Stewardson and Rutherfurd, see Chapter 5).
Subtly rewording such requirements to account for uncertainty
could help recast river restoration in a tone commensurate
with our abilities and uncertainties. Interestingly,
these objectives are consistent with Clark’s (2002) critical
synopsis of Predictive Management as opposed to
adaptive management as the current model in river
management.
The restoration community has burdened itself with the
idea that restoration objectives should be scientifi cally
based (Davis and Slobodkin, 2004). While science surely
has an important role in restoration, Davis and Slobodkin
(2004) argued that determining restoration objectives
is fundamentally a value-based and subjective process.
Nothing is seen as inherently wrong with this reality, so
long as it is transparently recognised. From an uncertainty
perspective, this means that restoration objectives are
therefore sources of uncertainty due to variability; namely
value diversity, behavioural diversity and societal randomness.
For example, the fate of 81 000 hectares of forest
land allocated for ecosystem restoration around the city of
Chicago, Illinois has pitted two ‘environmental’ groups
against each other based on their contrasting notions of
‘what is natural’. The divergent environmental views are
essentially split between preservationists, who wish to preserve
the forest land planted in the 1800s, and restoration-
4 This is fundamentally a communication uncertainty resulting
from socio-political value diversity (see Figure 3.1).
The Scope of Uncertainties in River Restoration 29
ists, who want to restore the pre-settlement (1830s) prairie
and savannah (Alario and Brün, 2001). Both evoke emotional
arguments, which can be supported on scientifi c
grounds. ‘Which is right?’ is the wrong question to ask of
science. Alario and Brün (2001) concluded that the appropriate
arena to decide such an issue is a political decision
making process.
3.3.2 Notions that Drive Restoration
Underlying motives for river restoration and the eventual
specifi c techniques tried to achieve them are some very
basic, yet highly uncertain notions. Since these basic
notions are rarely questioned, it is important to highlight
how they introduce uncertainty. Notions are also known
as ‘Lietbilds’ – or target visions – and have gained widespread
acceptance in the restoration literature (Hughes,
1995; Jungwirth et al., 2002; Kern, 1992). Notions, such
as those in Table 3.3, that drive restoration strategies
are frequently based on societal values and beliefs, or on
popular, but by no means certain, scientifi c paradigms
(Davis and Slobodkin, 2004; McDonald et al., 2004;
Rhoads et al., 1999).
Falkenmark and Folke (2002) argued that sustainable
catchment management must be based on ethical principles.
They suggest that management based on scientifi c
principles alone is primarily concerned with ‘doing the
thing right’, whereas notions that drive restoration strategies
are actually driven by ‘doing the right thing.’ It is a
presumption that good ethical practice generally translates
into good biological practice (Pister, 2001). Hence notions
are vague ideas, perhaps based on scientifi c knowledge,
but primarily supported by ethical beliefs and societal
values. The restoration literature is rarely explicit in distinguishing
the notions it advocates from the science used
to support it. Phillip Williams (personal communication)
asserts that ‘rigour’ in restoration planning should start
with the development of an explicit conceptual model
transparently describing our notions of how the river
system functions5 . Such a conceptual model should identify
both the historical context and the present day limitations
(i.e. uncertainties). Wheaton et al. (2004a) argued
that numerous conceptual models in the scientifi c literature
already exist and can be borrowed or modifi ed to
formulate a site or basin specifi c conceptual model as the
basis for restoration. Yet, Stewardson and Rutherfurd (see
Chapter 5) describe three levels in restoration from which
5 In principle, the process of ‘rigour’ in restoration planning still
follows the generic environmental management decision process
of Table 3.2. In essence what Phillip Williams, a seasoned practitioner,
describes is an informal Decision Support System
(DSS).

30 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Table 3.3 Common motives that guide notions and drive river restoration efforts
Notion Example(s)
What is Natural?
Nature is in Equilibrium ‘the equilibrium between sediment supply and available transport capacity.’ (Soar &
Thorne 2001); ‘landforms can be considered as either a stage in a cycle of erosion or as
a system in dynamic equilibrium.’ (Schumm & Lichty 1965).
Nature is in fl ux ‘Restored ecosystems are those in which the rates and types of disturbance do not exceed
the capacity of the system to respond to them.’ (Hruby 2003).
Nature Constant ‘confi dence on global stability; there are no limitations to development’ (Levy et al. 2000).
Nature Balanced ‘the environment is forgiving of most shocks, but large perturbations can knock ecological
variables into new regions of the landscape.’ (Levy et al. 2000).
Nature Ephemeral ‘the environment can not safely tolerate human modifi cations’ (Levy et al. 2000).
Nature Resilient ‘ecosystems are adaptive, evolutionary, and self organising . . . ecological systems often
thrive under conditions of high variability’ (Levy et al. 2000).
Physical Integrity
Physical Integrity ‘Physical Integrity for rivers refers to a set of active fl uvial processes and landforms
wherein channel, fl oodplains, sediments, and overall spatial confi guration maintain a
dynamic equilibrium, with adjustments not exceeding limits of change defi ned by
societal values. Rivers possess physical integrity when their processes and forms
maintain active connections with each other in the present hydrologic regime.’
(Graf 2001).
Alluvial River Attributes Several commonly known concepts that govern how alluvial channels work have been
compiled into a set of ‘attributes’ for alluvial river integrity (Trush et al. 2000).
Ecological Integrity Ecological Integrity ‘maintenance of all internal and external processes and attributes
interacting with the environment in such a way that the biotic community corresponds
to the natural state of the type-specifi c aquatic habitat, according to the principles of
self-regulation, resilience and resistance.’ (Angermeier & Karr 1994).
High Biodiversity = Ecological Natural systems foster biodiversity and artifi cial systems are homogenized and dominated
Integrity
by invasive species (Ward et al. 2002, Lister 1998).
Morphological Diversity =
Biological Diversity
epistemological uncertainties emerge: the validity of
the conceptual model; whether the proposed intervention
results in the planned geomorphic change; and whether
the change is sustainable. They then caution that the
validity of the conceptual model is the source of the ‘most
uncertainty.’ Returning to Phillip William’s concept of
rigour in planning, he argues restoration objectives should
be based on an understanding of how the conceptual
model interacts and responds to various societal motives
Newson (2002) did not dispute the abundance of evidence supporting the linkages between
channel dynamics and biodiversity, but criticises the lack of direct collaboration
between geomorphologists and ecologists to substantiate the links in river management:
‘the mantra “morphological diversity = biodiversity” currently remains an act of faith.’
What is Sustainable?
Sustainability According to Cairns (2003), the notion of sustainability is based on ‘the assumption that
humankind has the right to alter the planet so that human life can inhabit Earth
indefi nitely.’
Geomorphic Sustainability ‘sustainability encompasses the notion of self-regulation of spontaneous functions (e.g.
sediment deposition, colonisation and succession of vegetation) with minimal
intervention and no adverse impact on the future aquatic environment whilst
maintaining the functions of the channel demanded by society (fl ood control,
navigation etc.).’ (Sear 1996).
(NRC, 1992). Based on specifi c objectives, a measurable
set of indicators and target levels can be selected (Doyle
et al., 2000; Levy et al., 2000; Merkle and Kaupenjohann,
2000; Smeets and Weterings, 1999). Finally, a comparison
of predicted indicator responses to restoration intervention
versus inaction should be used to decide whether restoration
is appropriate. Although available science may be
used to inform the steps leading up to this decision
(Lemons and Victor, see Chapter 1), the decision whether

or not to proceed is ultimately a political one (Alario and
Brun, 2001).
3.3.3 Approaches to Restoration
Generally, river restoration projects consist of three
components: planning, implementation and evaluation.
The diversity of approaches available to implement these
components rather appropriately refl ects the varied types
(motives) of restoration projects and physiographic settings
they are applied in. Thus, historical and spatial contingencies
are contributing to uncertainties due to natural
variability (Phillips, 2001). Indeed, a plethora of restoration
approaches and strategies has been formalised in both
the peer-reviewed and grey literature (Wheaton et al.,
2004a). Examples range from generalised approaches for
stream restoration (e.g. FISRWG, 1998; Jungwirth et al.,
2002; Koehn et al., 2001; NRC, 1992; RRC, 2002) to
more specifi c strategies incorporating: fl uvial geomorphology
(e.g. Brookes and Sear, 1996; Gilvear, 1999;
Kondolf, 2000; Sear, 1994), ecosystem theory (e.g.
Richards et al., 2002; Stanford et al., 1996), hydraulic
engineering (e.g. Shields, 1996) and detailed design procedures
(Miller et al., 2001; Shields et al., 2003; Wheaton
et al., 2004b). Most of the approaches have parallels in
structure and ideology (Wheaton et al., 2004a).
Popular labels used to describe restoration approaches
include holistic, science-based, integrated and multidisciplinary
(Hildén, 2000; Jungwirth et al., 2002;
Wissmar and Bisson, 2003a). Since most approaches
purport or aim to be all of these (Wheaton et al., 2004a),
and the converse of each is perceived as negative, there is
little value in discriminating approaches on these grounds.
However, their components (i.e. planning, implementation
and monitoring) can be differentiated using three descriptive
metrics: the scale of restoration; form based versus
process based; and active versus passive. These metrics
can provide insight into the types of uncertainties encountered
and expectations placed on restoration projects
during planning, implementation and monitoring.
Since the late 1990s, approaches almost unanimously
call for catchment scale planning in restoration6 . However,
confusion arises over whether this means: restore the
entire catchment; use watershed assessments to nest reach
scale restoration in a catchment context (e.g. Bohn and
Kershner, 2002; Brookes and Shields, 1996; Walker et al.,
2002) or undertake a range of management and restoration
activities across various spatial scales but nested within a
catchment context (e.g. Frissell et al., 1993; Roni et al.,
2002). Ecosystem degradation has often taken place over
6 See also Table 3.1.
The Scope of Uncertainties in River Restoration 31
many decades or centuries and extends across landscape,
catchment and regional scales (Palmer et al., 1997).
However, restoring an entire catchment is rarely viable
(Brookes and Shields, 1996). Even those who call for
ecological restoration of the entire catchment (e.g. Frissell
et al., 1993) actually advocate achieving this through a
range of targeted activities at various spatial and temporal
scales.
Most of the restoration literature also points towards a
consensus that a ‘process-based’ approach is superior to a
‘form-based’ one (Wheaton et al., 2004a). Much of the
form versus process debate simplifi es down to the diffi -
culty and/or appropriateness in selecting an analogue or
reference condition. The frequently referenced ‘Lietbilds’
or target visions (Kern, 1992) and the popular Rosgen
approach to restoration (Malakoff, 2004; Rosgen, 1996)
both rely heavily on analogues. Jungwirth et al. (2002)
suggest that at least three methods for selecting analogue
or reference conditions exist:
Select an existing reference site with ‘desirable’ conditions
(location substitution).
Select a historical reference condition for the site of
interest on the basis of historical analysis (time for space
substitution).
Create a reference condition on the basis of theoretical
models (either conceptual or mathematical).
In referring to these analogue conditions, is the desired
form or the desired process then mimicked? This seems
to be the point of departure for opinions within the restoration
literature. Some argue that any mimicking of reference
conditions is a form-based approach (McDonald et al.,
2004). Others suggest that as long as ample consideration
of sustaining processes and desired functions is made, the
use of analogue conditions can be process based (Palmer
et al., 1997; Wheaton et al., 2004a). Although exact interpretations
are themselves uncertain and will continue to
spur debate over semantics (confl icting evidence uncertainties),
most concur that consideration of sustaining
processes is fundamental (Wheaton et al., 2004c).
Fundamental methodological disagreements arise in the
restoration literature with respect to passive versus active
approaches to river restoration (Edmonds et al., 2003;
Wissmar and Beschta, 1998). Here, active approaches are
referred to as those which involve direct structural modifi
cation to the river, its fl oodplain or infrastructure therein
(e.g. channel realignment, levee removal, instream habitat
structures). By contrast, passive approaches are those that
‘rely on the river to do the work’ (e.g. fl ow augmentation,

32 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Figure 3.4 Five philosophical attitudes towards uncertainty (The Venn diagram is meant to illustrate the overlap between contemporary
attitudes towards uncertainty. Note that ignoring uncertainty shares no overlap with contemporary attitudes towards uncertainty.)
change in landuse, managing nonpoint sources of pollution,
buffer strips) (Wissmar and Beschta, 1998). Using a
‘process-based’ approach can make intuitive sense for
passive approaches to restoration. For example, providing
fl ow releases from a reservoir to mimic a natural hydrograph
and encourage mobilisation and reorganisation of
sediments, may restore the processes that ‘allow the river
to do the work’ (Stanford et al., 1996; Trush et al., 2000).
However, active approaches are considered favourable
when natural or passive recovery may take an unacceptably
long time (Montgomery and Bolton, 2003). The
choice of a passive versus active approach will depend
very much on the specifi c social, political, economic
and environmental contingencies of individual river basins
(Wissmar et al., 2003), as well as the extent to which
initial conditions matter (Phillips, 2002). Wheaton et al.
(2004b) suggested that in some spawning habitat rehabilitation
contexts, it may be appropriate to employ passive
approaches like gravel augmentation in conjunction with
active approaches like spawning bed enhancement to kickstart
recovery. Ultimately, all these choices are fuelled by
an uncertain conceptual understanding of the system and
logical ideas about how best to proceed with restoration.
Given these inherent uncertainties, adaptive management
is well suited to allow practitioners and decision makers
to make a decision in the face of uncertainty, and to adjust
that decision as time and new challenges unfold (Clark,
2002; Lister, 1998).
7
See Section 3.3.1 for the relationship between expectation and
uncertainty.
3.4 PHILOSOPHIES OF UNCERTAINTY
So, is all this uncertainty bad? By this point, it should be
clear that uncertainty in river restoration is ubiquitous.
However, different segments of society view uncertainty in
very different ways, depending on the context (Lemons
and Victor, see Chapter 1). As already mentioned, ordinary
people are quite comfortable with the uncertainties of
life in an intuitive and nonexplicit sense (Anderson et al.,
2003; Pollack, 2003). However, uncertainty in policy and
science, especially as reported in the media (Riebeek,
2002), are very different contexts. The choice of what to do
about uncertainty is a philosophical question. Five potential
philosophical treatments of uncertainty are proposed in
Figure 3.4. Each of these philosophies is reviewed in the
remaining sections and linked to current attitudes within
different segments of the river restoration community.
3.4.1 Ignore Uncertainty
It has already been argued here that the restoration community
has tended to passively ignore uncertainty and
possible explanations as to why this may be the case proposed.
For example, managers, policy and decision makers
are fearful of admitting uncertainties, as this may be seen
as a sign of weakness (Clark, 2002; Levy et al., 2000).
Now that public support exists for river restoration, so
too does the expectation7 that the problems restoration
addresses are well understood. Indeed, these problems are
reasonably well understood, but numerous uncertainties
remain. Aside from basic, and potentially reducible,

communication uncertainties the signifi cance of the vast
majority of uncertainties associated with restoration are
simply not known. Admittedly, specifi c examples of
uncertainties in restoration may indeed be insignifi cant.
However, to assume insignifi cance on both ethical and
technical grounds without fi rst establishing it might ultimately
backfi re on the restoration community.
3.4.2 Eliminate Uncertainty
The positivist view of the world has fuelled much scientifi
c progress on the notion that uncertainty is bad, absolute
knowledge is good, and it is necessary to strive to
eliminate uncertainty (Klir and Yuan, 1995; Priddy, 1999;
Van Asselt and Rotmans, 2002). This fosters an unnecessarily
narrow view of uncertainty as subsumed entirely
within the realm of science. van Asselt and Rotmans
(2002) argued this view grew out of the ‘Enlightenment
Period’ or ‘Age of Reason’ of the 17th and 18th centuries
where science was to be ‘the provider of certainty.’ Further
to this endeavour, many scientists assumed that unique
causal laws exist for all natural phenomena and ignored
the possibilities of indeterminacy and equifi nality (Wilson,
2001). Many physical scientists still subscribe to a ‘positivist’
view (Harman, 1998), implicitly associating uncertainty
with an inability to quantify the environment, rather
than acknowledging a limited understanding about the
environment itself (Klir and Yuan, 1995).
Whether specifi c types of uncertainty can be eliminated
depends on an individual’s interpretation of semantics.
Under the holistic view of uncertainty advocated in this
chapter uncertainty cannot be completely eliminated.
Pollack (2003) suggests that ‘uncertainty is always with
us and can never be fully eliminated’. Other authors (e.g.
Knight, 1921) suggest that some types of uncertainty can
be transformed into related concepts (e.g. error, expectation,
reliability, risk) with the help of mathematical constructs
and knowledge gained from historical inference.
Through this transformation, uncertainty of a specifi c type
(i.e. uncertainty for which a valid basis for classifi cation
exists) in a sense might be ‘eliminated.’ Such a transformation
represents an improved understanding of uncertainty
but does not truly ‘eliminate’ it.
With technological progress has come the expectation
of greater predictive power. Priddy (1999) suggested, ‘the
strictest standard of truth in science is that of predictability.’
Although intuitively no one expects prediction to be
completely free of uncertainty, the notion that uncertainty
can be eliminated is latent in the mainstream media
(Riebeek, 2002). Pollack (2003) argues that scientists are
accustomed to dealing with uncertainty explicitly, but the
general public’s familiarity with uncertainty is implicit
The Scope of Uncertainties in River Restoration 33
and often confused. Jamieson (1996) suggests that, particularly
with respect to decisions about increased environmental
protections, the ‘rhetorical role of uncertainty
claims’ are used to suggest no action should be taken until
uncertainty is eliminated. Hence, it is concluded that
attempts to eliminate uncertainty are misleading and
founded on ignorance of the principles of uncertainty.
3.4.3 Reduce Uncertainty
A more pragmatic view of uncertainty seeks to reduce,
rather than eliminate, those specifi c elements that are perceived
as problematic (Klir and Yuan, 1995). This approach
to uncertainty is represented diagrammatically in Figure
3.4. Notice that with regards to reducing uncertainty, the
key questions are, in order: can it be quantifi ed, is it signifi
cant and can it be constrained? So long as the answer
is ‘yes’ to all these questions, uncertainty might be reduced.
However, if the opposite is true, uncertainty is simply
ignored. To move beyond uncertainty as an ambiguous
buzzword that will forever plague scientists and decision
makers, a broader view of uncertainty as information is
appropriate (Newson and Clark, see Chapter 14).
3.4.4 Cope with Uncertainty
Coping or living with uncertainty represents a more proactive
view of dealing with uncertainty than elimination or
reduction. This approach recognises that, regardless of
the signifi cance of uncertainty and our ability/inability to
quantify or constrain it, we are always forced to cope with
it. Especially within the hydrologic and atmospheric
modelling literature, uncertainty is actively recognised
and specifi c methods to cope with it are continually being
proposed (e.g. Beven, 1996a; Beven, 1996b; Osidele et al.,
2003; Werritty, 2002).
3.4.5 Embrace Uncertainty
Despite the advantages of efforts to cope with or reduce
uncertainty over eliminating it, all the preceding still fundamentally
view uncertainty as negative. Several authors
have departed from this view towards a more progressive
view of embracing uncertainty (Johnson and Brown, 2001;
Newson and Clark, see Chapter 14). One of the earlier
proponents of this view appears to be Holling (1978), who
argued:
‘while efforts to reduce uncertainty are admirable . . . if
not accompanied by an equal effort to design for
uncertainty and obtain benefi ts from the unexpected,
the best of predictive models will only lead to larger

34 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
problems arising more quickly and more often’ (in:
Levy et al., 2000).
Klir and Yuan (1995) considered uncertainty in modelling
as ‘an important commodity . . . , which can be traded
for gains in the other essential characteristics of models.’
Others have suggested that recognising that not all uncertainty
is bad will be increasingly important to decision
makers who are forced to make decisions in the face of
uncertainty (Clark and Richards, 2002; Pollack, 2003).
Especially in long term policy analysis (the next 20–100
years) decision makers are faced with what Lempert et al.
(2003) referred to as ‘deep uncertainty’. Johnson and
Brown (2001) argued that incorporating uncertainty into
restoration design allows practitioners to consider multiple
causes and hypothesised fi xes; thereby reducing the potential
for project failure. It has been argued here that uncertainty
is not necessarily bad, but ignorance of it can foster
unrealistic expectations. Chapman and Ward (2002) argued
that uncertainty is not just as a risk, but also an opportunity.
Uncertainty due to natural variability, in say fl ow regime,
can be a particularly good thing, for example by promoting
habitat heterogeneity and biodiversity (Clifford et al., see
Chapter 7; Montgomery and Bolton, 2003).
In Figure 3.5, the notions of embracing uncertainty are
synthesised in the context of the van Asselt (2000) typology.
This approach embraces uncertainty as information
and its potential for helping avoid risks, or embracing
unforeseen opportunities. Notice that the uncertainties are
not treated uniformly, but instead are segregated by source
(i.e. due to limited knowledge or due to variability) and
type. Anderson et al. (2003) note that environmental management
problems are so diverse that a single approach is
unlikely to be appropriate for all. Thus, Chamberlin’s
(1890) idea of multiple working hypotheses is emerging
in environmental management through advocating pluralistic
approaches (e.g. Lempert et al., 2003; Van Asselt and
Rotmans, 2002).
The embracing uncertainty framework proposed here
emphasises this point by structuring a range of questions
and possible management decisions based on the specifi c
uncertainties at hand. In the spirit of ‘sustainable uncertainty’
as proposed by Newson and Clark (see Chapter 14),
this is not at all a rigid framework but instead a loose and
adaptive guide built around an uncertainty typology.
Unlike the four other philosophical treatments of uncertainty,
this allows the restoration scientist, practitioner or
decision maker to:
explore the potential signifi cance (both in terms of
unforeseen consequences and welcome surprises) or
insignifi cance of uncertainties;
effectively communicate uncertainties;
eventually make adaptive, but transparent, decisions in
the face of uncertainty.
3.5 CONCLUSION
In this chapter a very broad picture of uncertainty in river
restoration and environmental management has been
painted. This was done to unravel the ambiguities around
the notion of ‘certainty’ in restoration and recast uncertainty
as useful information. In fact, the arguments and
evidence presented challenge the view of scientifi c deterministic
‘certainty’ and societal beliefs that certainty is
necessary in restoration. A typology for discriminating
uncertainty was reviewed that can be used to separate
uncertainties that can lead to unforeseen and undesirable
consequences from uncertainties that lead to potentially
welcome surprises. Many of the uncertainties surrounding
restoration motives, notions and approaches are most seriously
manifested as communication uncertainties. That is,
instead of being expressed simply as uncertainties due
to limited knowledge, they are ignored and miscommunicated
through the restoration process in a manner that
prevents transparent decision making. The signifi cance of
the plethora of other uncertainties alluded to is largely
situation-specifi c and, to date, unexplored.
Five philosophical strategies for dealing with uncertainty
ranging from the status quo of ignoring uncertainty
to the advocated embracing uncertainty were reviewed.
Traditional scientifi c research has focused on a narrow
class of uncertainties and adopted ‘eliminate’ and ‘reduce’
uncertainty philosophies. It is argued that it is unethical to
assume that uncertainty is insignifi cant. There is an
increasing recognition in environmental management that
ethical and social dimensions are the primary drivers, with
scientifi c and technical dimensions playing a secondary
role (Falkenmark and Folke, 2002; Lister, 1998) 8 . Thus, an
emerging challenge which the restoration community is
faced with is combining these dimensions to ‘do the right
thing right.’ Out of the decision making arena has emerged
the pragmatic view of coping with uncertainty. However,
from the suggestions and examples in the more general
environmental management literature, it is concluded that
embracing uncertainty could also help transcend the scientifi
c research and decision making boundaries in river
restoration.
8 Recall the development of notions in Section 3.3.2, and the distinction
of Falkenmark and Folke (2002) between technical concerns
(e.g. ‘doing the thing right’) and ethical concerns (e.g.
‘doing the right thing’).

The Scope of Uncertainties in River Restoration 35
Figure 3.5 Framework for embracing uncertainty in the decision making process (This framework relies on the Van Asselt (2000)
typology of uncertainty.)

36 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
REFERENCES
Alario M, Brun M. 2001. Uncertainty and Controversy in
the Science and Ethics of Environmental Policy Making.
Theory and Science. 2 (1): Available on the web: \url{http://
theoryandscience.icaap.org/content/vol002.001/02alariobrun.
html}.
Anderson JL, Hilborn RW, Lackey RT, Ludwig D. 2003. Watershed
restoration: Adaptive decision making in the face of
uncertainty. In: Wissmar RC, Bisson PA, Duke M (Eds), Strategies
for Restoring River Ecosystems: Sources of Variability and
Uncertainty in Natural and Managed Systems. American
Fisheries Society: Bethesda, Maryland; 185–201.
Angermeier PL, Karr JR. 1994. Biological integrity versus biological
diversity as policy directives: protecting biotic resources.
Bioscience 44: 690–697.
Beven K. 1996a. Equifi nality and uncertainty in geomorphological
modelling. In: Rhoads BL, Thorn CE (Eds), The Scientifi c
Nature of Geomorphology. John Wiley & Sons Ltd: Chichester;
289–314.
Beven K. 1996b. The limits of splitting: Hydrology. The Science
of the Total Environment. 183: 89–97.
Bohn BA, Kershner JL. 2002. Establishing aquatic restoration
priorities using a watershed approach. Journal of Environmental
Management 64 (4): 355–363.
Boon PJ. 1998. River restoration in fi ve dimensions. Aquatic Conservation:
Marine and Freshwater Ecosystems 8 (1): 257–264.
Brookes A, Downs P, Skinner K. 1998. Uncertainty in the engineering
of wildlife habitats. Journal of the Chartered Institution
of Water and Environmental Management 12 (1): 25–29.
Brookes A, Sear DA. 1996. Geomorphological principles for
restoring channels. In: Brookes A, Shields FD (Eds), River
Channel Restoration: Guiding Principles for Sustainable
Projects. John Wiley & Sons Ltd: Chichester; 75–101.
Brookes A, Shields FD. 1996. River Channel Restoration:
Guiding Principles for Sustainable Projects. John Wiley &
Sons Ltd: Chichester.
Brown TL, LeMay HE, Bursten BE. 1994. Chemistry: The Central
Science (6th Edition). Prentice Hall: Englewood Cliffs, New
Jersey.
Cairns J. 2003. Numeracy and sustainability. Ethics in Science
and Environmental Politics 2003: 83–91.
Chamberlin TC. 1890. The method of multiple working hypotheses.
Science. Reprinted 1965: 148: 754–759.
Chapman C, Ward S. 2002. Managing Project Risk and Uncertainty:
A Constructively Simple Approach to Decision Making.
John Wiley & Sons Ltd: Chichester.
Clark MJ. 2002. Dealing with uncertainty: adaptive approaches
to sustainable river management. Aquatic Conservation:
Marine and Freshwater Ecosystems 12: 347–363.
Clark MJ, Richards KJ. 2002. Supporting complex decisions for
sustainable river management in England and Wales. Aquatic
Conservation: Marine and Freshwater Ecosystems 12: 471–483.
Collins JP, Kinzig A, Grimm NB et al. 2000. A New Urban
Ecology: Modeling human communities as integral parts of
ecosystems poses special problems for the development and
testing of ecological theory. American Scientist 88: 416–425.
Davis MA, Slobodkin LB. 2004. The science and values of restoration
ecology. Restoration Ecology 12 (1): 1–3.
Doyle MW, Harbor JM, Rich CF, Spacie A. 2000. Examining the
effects of urbanization on streams using indicators of geomorphic
stability. Physical Geography 21 (2): 155–181.
Edmonds RL, Francis RC, Mantua NJ, Petersen DL. 2003.
Sources of climate variability in river ecosystems. In: Wissmar
RC, Bisson PA, Duke M (Eds), Strategies for Restoring River
Ecosystems: Sources of Variability and Uncertainty in Natural
and Managed Systems. American Fisheries Society: Bethesda,
Maryland; 11–37.
Ellison SLR, Rosslein M, Williams A (Eds). 2000. Quantifying
Uncertainty in Analytical Measurement, Guide number 4.
Eurachem/CITAC: Teddington, UK.
Everest FH, Sedell JR. 1984. Evaluating effectiveness of stream
enhancement projects. In: Hassler TJ (Ed), Pacifi c Northwest
Stream Habitat Management Workshop. American Fisheries
Society: Humboldt State Universitiy, California; 156–164.
Falkenmark M, Folke C. 2002. The ethics of socioecohydrological
catchment management. Hydrology and Earth
System Sciences 6 (1): 1–9.
FISRWG. 1998. Stream Corridor Restoration: Principles, Processes,
and Practices. GPO Item No. 0120-A: Federal Interagency
Stream Restoration Working Group, Washington, DC.
Frissell CA, Liss WJ, Bayles D. 1993. An integrated, biophysical
strategy for ecological restoration of large watersheds. In:
Spangenborg NE, Potts DE (Eds), Changing Roles in Water
Resources Management and Policy. American Water Resources
Association: Bethesda, Maryland; 449–456.
Gilvear DJ. 1999. Fluvial geomorphology and river engineering:
future roles utilizing a fl uvial hydrosystems framework.
Geomorphology 31: 229–245.
Graf WL. 1996. Geomorphology and policy for restoration of
impounded American rivers: What is ‘Natural?’ In: Rhoads BL,
Thorn CE (Eds), The Scientifi c Nature of Geomorphology.
John Wiley & Sons Ltd: Chichester; 221–254.
Graf WL. 2001. Damage control: dams and the physical integrity
of America’s rivers. Annals of the Association of American
Geographers 91: 1–27.
Hansen HO, Iversen TM. 1998. The European Centre for River
Restoration (ECRR). In: Hansen HO, Madsen BL (Eds), River
Restoration ’96 ECRR: Denmark; 73–79.
Harman W. 1998. Global Mind Change: The Promise of the 21st
Century (2nd Ed). Berrett–Koehler Publishers, Inc.: San
Francisco, California.
Hildén M. 2000. The role of integrating concepts in watershed
rehabilitation. Ecosystem Health 6 (1): 39–50.
Holling CS. 1978. Adaptive Environmental Assessment and Management.
John Wiley & Sons Ltd: Chichester.
Hruby T. 2003. Where is the ecology in wetland restoration?
Society of Ecological Restoration Northwest and Society
Wetland Scientists Joint Regional Conference. SERNW: Portland,
Oregon.
Hughes RM. 1995. Defi ning acceptable biological status by comparing
with reference conditions. In: Davis WS, Simon RF
(Eds), Biological Assessment and Decision Making. Lewis
Publishers: Boca Raton, Florida; 31–47.

Jamieson D. 1996. Scientifi c uncertainty: how do we know when
to communicate research fi ndings to the public? The Science
of the Total Environment 184: 103–107.
Johnson PA, Brown ER. 2001. Incorporating uncertainty in the
design of stream channel modifi cations. Journal of the American
Water Resources Association 37 (5): 1225–1236.
Johnson PA, Hey RD, Brown ER, Rosgen DL. 2002. Stream restoration
in the vicinity of bridges. Journal of the American
Water Resources Association 38 (1): 55–67.
Johnson PA, Rinaldi M. 1997. Uncertainty in the design of stream
channel restorations. In: Jr. FMH and Alsaffar A (Eds), Water
for a Changing Global Community: The 27th Congress of the
International Association for Hydraulic Research. American
Society of Civil Engineers Press: San Francisco, California;
451–457.
Johnson PA, Rinaldi M. 1998. Uncertainty in stream channel
restoration. In: Ayyub BM (Ed), Uncertainty Modeling and
Analysis in Civil Engineering. CRC Press: London; 425–
437.
Jungwirth M, Muhar S, Schmutz S. 2002. Re-establishing and
assessing ecological integrity in riverine landscapes. Freshwater
Biology 47 (4): 867–887.
Kavvas ML. 1999. On the coarse-graining of hydrologic processes
with increasing scales. Journal of Hydrology 217 (3–4):
191–202.
Kern K. 1992. Rehabilitation of streams in South-West Germany.
In: Boon PJ, Calow P, Petts G (Eds), River Conservation and
Management. John Wiley & Sons Ltd: Chichester; 321–336.
Klir GJ, Yuan B. 1995. Fuzzy Sets and Fuzzy Logic: Theory and
Applications. Prentice Hall: Upper Saddle River, New Jersey.
Knight F. 1921. Risk, Uncertainty and Profi t. Houghton Miffl in:
New York.
Koehn JD, Brierley GJ, Cant BL, Lucas AM. 2001. River Restoration
Framework, The National Rivers Consortium: Land and
Water Australia, Canberra, ACT.
Kondolf GM. 1995a. Five elements for effective evaluation of
stream restoration. Restoration Ecology 3 (2): 133–136.
Kondolf GM. 1995b. Geomorphological stream channel classifi -
cation in aquatic habitat restoration – Uses and limitations.
Aquatic Conservation: Marine and Freshwater Ecosystems
5 (2): 127–141.
Kondolf GM. 1996. A cross section of stream channel restoration.
Journal of Soil and Water Conservation 51 (2): 119–125.
Kondolf GM. 2000. Some suggested guidelines for geomorphic
aspects of anadromous salmonid habitat restoration proposals.
Restoration Ecology 8 (1): 48–56.
Lempert RJ, Popper SW, Bankes SC. 2003. Shaping the Next
One Hundred Years: New Methods for Quantitative, Long term
Policy Analysis. The Rand Pardee Center: Santa Monica,
California.
Levy JK, Hipel KW, Kilgour DM. 2000. Using environmental
indicators to quantify the robustness of policy alternatives to
uncertainty. Ecological Modelling 130 (1–3): 79–86.
Lewin J. 2001. Alluvial systematics. In: Maddy D, Macklin MG,
Woodard JC (Eds), River Basin Sediment Systems: Archives of
Environmental Change. A.A. Balkema Publishers: Steenwijik,
The Netherlands; 19–41.
The Scope of Uncertainties in River Restoration 37
Lister N-ME. 1998. A systems approach to biodiversity conservation
planning. Environmental Monitoring and Assessment 49:
123–155.
Malakoff D. 2004. The river Doctor: Profi le Dave Rosgen.
Science. 305 (13 August): 937–939.
Marmulla G. (Ed). 2001. Dams, Fish and Fisheries. Opportunities,
Challenges and Confl ict Resolution, Technical Paper no.
419. FAO Fisheries Department: Rome, Italy.
McDonald A, Lane SN, Chalk EA, Haycock NE. 2004. Rivers of
dreams: on the gulf between theoretical and practical aspects
of an upland river restoration. Transactions of the Institute of
British Geographers 29 (3): 257–281.
Merkle A, Kaupenjohann M. 2000. Derivation of ecosystemic
effect indicators – method. Ecological Modelling 130 (1–3):
39–46.
Merriam–Webster. 1994. Merriam Webster’s Collegiate Dictionary.
Merriam–Webster, Inc.: Springfi eld, Massachusetts.
Miller DE, Skidmore PB, White DJ. 2001. Channel Design, Inter-
Fluve, Inc., Prepared for Washington Department of Fish and
Wildlife: Boseman, Montana.
Montgomery DR, Bolton SM. 2003. Hydrogemorphic variability
and river restoration. In: Wissmar RC, Bisson PA, Duke M
(Eds), Strategies for Restoring River Ecosystems: Sources of
Variability and Uncertainty in Natural and Managed Systems.
American Fisheries Society: Bethesda, Maryland; 39–80.
Moss T. 2004. The governance of land use in river basins: prospects
for overcoming problems of institutional interplay with
the EU Water Framework Directive. Land Use Policy 21:
85–94.
Mount JF. 1995. California Rivers and Streams: The Confl ict
Between Fluvial Process and Land Use. University of California
Press: Berkeley, California.
NAP. 2002. Riparian Areas: Functions and Strategies for Management.
National Academy Press: Washington, DC.
Newson MD. 2002. Geomorphological concepts and tools for
sustainable river ecosystem management. Aquatic Conservation:
Marine and Freshwater Ecosystems 12 (4): 365–379.
Nijland HJ, Cals MJR (Eds). 2000. River Restoration in
Europe: Practical Applications. ECRR: Wageningen, The
Netherlands.
NRC. 1992. Restoration of Aquatic Ecosystems: Science, Technology,
and Public Policy. National Academy Press: Washington,
DC.
Osidele OO, Zeng W, Beck MB. 2003. Coping with uncertainty:
A case study in sediment transport and nutrient load analysis.
Journal of Water Resources Planning and Management-Asce
129 (4): 345–355.
Palmer MA, Ambrose RF, Poff NL. 1997. Ecological theory and
community restoration ecology. Restoration Ecology 5 (4):
291–300.
Phillips JD. 2001. Contingency and generalization in pedology,
as exemplifi ed by texture-contrast soils. Geoderma 102 (3–4):
347–370.
Phillips JD. 2002. Global and local factors in earth surface
systems. Ecological Modelling 149 (3): 257–272.
Pister EP. 2001. Wilderness fi sh stocking: History and perspective.
Ecosystems 4: 279–286.

38 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Pollack HN. 2003. Uncertain Science . . . Uncertain World. Cambridge
University Press: Cambridge, UK.
Power M. 1999. The Audit Society: Rituals of Verifi cation. Oxford
University Press: Oxford, UK.
Priddy R. 1999. Science Limited. Available on the web: http://
home.no.net/rrpriddy/indexlim.html: Oslo, Norway.
Rhoads BL, Wilson D, Urban M, Herricks EE. 1999. Interaction
between scientists and nonscientists in community-based
watershed management: Emergence of the concept of
stream naturalization. Environmental Management 24 (3):
297–308.
Richards K, Brasington J, Hughes F. 2002. Geomorphic dynamics
of fl oodplains: ecological implications and a potential modelling
strategy. Freshwater Biology 47 (4): 559–579.
Riebeek H. 2002. The perception of scientifi c uncertainty in
science news writing. URL: http://tc.eserver.org/13650.html
Accessed on: 19 September, 2003.
Robinson CT, Tockner K, Ward JV. 2002. The fauna of dynamic
riverine landscapes. Freshwater Biology 47 (4): 661–677.
Roni P, Beechie TJ, Bilby RE et al. 2002. A review of stream
restoration and a hierarchal strategy for prioritizing restoration
in Pacifi c Northwest watersheds. North American Journal of
Fisheries Management 22: 1–20.
Rosgen D. 1996. Applied River Morphology. Wildland Hydrology:
Pagosa Springs, Colorado.
Rotmans J, Van Asselt MBA. 2001. Uncertainty management
in Integrated Assessment Modelling: Towards a pluralistic
approach. Environmental Monitoring and Assessment 69 (2):
101–130.
RRC. 2002. Manual of River Restoration Techniques. The River
Restoration Centre, Silsoe, UK.
Schumm SA. 1991. To Interpret the Earth: Ten Ways to be Wrong.
Cambridge University Press: Cambridge, UK.
Schumm SA, Lichty RW. 1965. Time, space and causality in
geomorphology. American Journal of Science 263 (February):
110–119.
Sear DA. 1994. River restoration and geomorphology. Aquatic
Conservation: Marine and Freshwater Ecosystems 4 (2):
169–177.
Sear DA. 1996. The Sediment System and Channel Stability. In:
Brookes A, Shields FD (Eds), River Channel Restoration:
Guiding Principles for Sustainable Projects. John Wiley &
Sons Ltd: Chichester, 149–177.
Shields FD. 1996. Hydraulic and hydrologic stability. In: Brookes
A, Shields FD (Eds), River Channel Restoration: Guiding
Principles for Sustainable Projects. John Wiley & Sons Ltd:
Chichester; 103–126.
Shields FD, Copeland RR, Klingeman PC et al. 2003. Design for
stream restoration. Journal of Hydraulic Engineering 129 (8):
575–584.
Smeets E, Weterings R. 1999. Environmental Indicators: Typology
and Overview. Technical Report No 25, European Environment
Agency: Copenhagen, Denmark.
Soar PJ, Thorne CR. 2001. Channel Restoration Design for
Meandering Rivers. ERDC/CHL CR-01-1, U.S. Army Corps
of Engineers, Engineer Research and Development Center,
Vicksburg, Massachusetts.
Stanford JA et al. 1996. A general protocol for restoration of
regulated rivers. Regulated Rivers: Research and Management
12: 391–413.
Stockwell BR. 2000. The Mary River and Tributaries Rehabilitation
Plan: A Review of River Processes, Fluvial Geomorphology
and Ecological Concepts and Their Application to a River
Rehabilitation Plan for a Major South-East Queensland Catchment.
Masters Dissertation, University of New England, New
South Wales, Australia.
Tockner K, Stanford JA. 2002. Riverine fl ood plains: present
state and future trends. Environmental Conservation 29 (3):
308–330.
Trochim WM. 2000. The Research Methods Knowledge Base,
2nd Edition. Atomic Dog Publishing: Cincinnati, Ohio.
Trush WJ, McBain SM, Leopold LB. 2000. Attributes of an alluvial
river and their relation to water policy and management.
Proceedings of the National Academy of Sciences of the United
States of America 97 (22): 11858–11863.
Van Asselt MBA. 2000. Perspectives on Uncertainty and Risk:
The PRIMA Approach to Decision Support. Kluwer Academic
Publishers: Dordrecht, The Netherlands.
Van Asselt MBA, Rotmans J. 2002. Uncertainty in integrated
assessment modelling – From positivism to pluralism. Climatic
Change 54 (1–2): 75–105.
Vitousek PM, Mooney HA, Lubchenco J, Melillo JM. 1997.
Human domination of Earth’s ecosystems. Science 277:
494–499.
Walker J, Diamond M, Naura M. 2002. The development of
Physical Quality Objectives for rivers in England and Wales.
Aquatic Conservation: Marine and Freshwater Ecosystems
12 (4): 381–390.
Walters CJ. 1997. Challenges in adaptive management of riparian
and coastal ecosystems. Conservation Ecology [online]: http://
www.consecol.org/vol1/iss2/art1.
Ward JV, Tockner K, Arscott DB, Claret C. 2002. Riverine landscape
diversity. Freshwater Biology 47 (4): 517–539.
Ward T. 1998. Risk and uncertainty in environmental policy
evaluation. In: Dore MHI, Mount TD (Eds), Global Enviornmental
Economics: Equity and the Limits of Markets. Blackwell
Publishers: Oxford, UK; 116–135.
Werritty A. 2002. Living with uncertainty: climate change, river
fl ows and water resource management in Scotland. The Science
of The Total Environment 294 (1–3): 29–40.
Wheaton JM, Pasternack GB, Merz JE. 2004a. Spawning habitat
rehabilitation – II. Using hypothesis testing and development
in design, Mokelumne River, California, USA. International
Journal of River Basin Management 2 (1): 21–37.
Wheaton JM, Pasternack GB, Merz JE. 2004b. Spawning habitat
rehabilitation – I. Conceptual approach and methods. International
Journal of River Basin Management 2 (1): 3–20.
Wheaton JM, Sear DA, Darby SE, Milne JA. 2004c. The International
River Restoration Survey. http://www.geog.soton.ac.
uk/users/WheatonJ/RestorationSurvey_Cover.asp. Accessed
on: 15/06/04.
Wilson DW. 2001. On the Problem of Indeterminacy in Fluvial
Geomorphology. PhD Thesis, University of Southampton,
Southampton, UK.

Wissmar RC, Beschta RL. 1998. Restoration and management of
riparian ecosystems: a catchment perspective. Freshwater
Biology 40: 571–585.
Wissmar RC, Bisson PA. 2003a. Strategies for restoring river
ecosystems: Sources of variability and uncertainty. In: Wissmar
RC, Bisson PA, Duke M (Eds), Strategies for Restoring River
Ecosystems: Sources of Variability and Uncertainty in Natural
and Managed Systems. American Fisheries Society: Bethesda,
Maryland; 3–7.
Wissmar RC, Bisson PA. 2003b. Strategies for restoring rivers:
Problems and opportunites. In: Wissmar RC, Bisson PA, Duke
M (Eds), Strategies for Restoring River Ecosystems: Sources of
The Scope of Uncertainties in River Restoration 39
Variability and Uncertainty in Natural and Managed Systems.
American Fisheries Society: Bethesda, Maryland; 245–262.
Wissmar RC, Bisson PA (Eds). 2003c. Strategies for Restoring
River Ecosystems: Sources of Variability and Uncertainty in
Natural and Managed Systems. American Fisheries Society:
Bethesda, Maryland.
Wissmar RC, Braatne JH, Beschta RL, Rood SB. 2003. Variability
of riparian ecosystems: Implications for restoration. In:
Wissmar RC, Bisson PA, Duke M (Eds), Strategies for Restoring
River Ecosystems: Sources of Variability and Uncertainty
in Natural and Managed Systems. American Fisheries Society:
Bethesda, Maryland; 107–127.

SECTION II
Planning and Designing
Restoration Projects

River Restoration: Managing the Uncertainty in Restoring Physical Habitat Edited by Stephen Darby and David Sear
© 2008 John Wiley & Sons, Ltd
4
Planning River Restoration Projects:
Social and Cultural Dimensions
4.1 INTRODUCTION
G. Mathias Kondolf 1 and Chia-Ning Yang 2
1 Department of Landscape Architecture and Environmental Planning, University of California, USA
2 Department of Landscape Architecture, California State Polytechnic University, USA
Nationwide, since 1990, at least US$17 billion has been
spent on restoration projects, a fi gure that is an underestimate
because most reported costs do not include staff
time and many projects did not report their costs at all
(Bernhardt et al., 2005). In many areas, river restoration
has become an industry, with nonprofi t groups, government
agencies and consulting fi rms now depending upon
river restoration funds to support large components of
their budgets. For example, over four fi scal years from July
2000 to June 2004, over US$100 million was disbursed
by the California Department of Fish and Game to recipient
groups and agencies in the Fishery Restoration Grants
Program for restoration projects in coastal river basins
(F. Sime, personal communication, December 2003),
mostly to construct habitat enhancement structures in
salmon-bearing rivers and streams. Elsewhere in California,
stream restoration projects employ many in rural communities,
including former timber cutters (Hamilton,
1993). In the city of Bozeman, Montana, there is suffi cient
business to support six fi rms specializing in restoring
trout streams.
River and stream restoration can be viewed as a contemporary
phase of the environmental movement. Unlike
early phases of the movement, which tended to document
and draw attention to the nature and extent of environmental
degradation (Carson, 1962; Ehrlich, 1968) and which
therefore tended to be negative or pessimistic in tone,
restoring rivers and streams has a positive, pro-active connotation.
This is especially true of streams in urban neigh-
borhoods, where restoration projects can provide positive
reinforcement and a sense of empowerment to local
groups. In many respects, the greatest benefi ts to restoring
local urban creeks are probably the social benefi ts that
accrue through the process of community building and the
public environmental education achieved.
Technical specialists often assume implicitly that restoration
is a technical problem, and a glance at articles
published in restoration-related journals shows a preponderance
of papers addressing the scientifi c aspects of
project design and planning. However, restoration can be
viewed as fundamentally a social phenomenon, as it results
from a societal decision to restore some functions to a
river (Eden et al., 2000). Its goals and implementation
approaches can be informed by science, but they are
essentially social in nature. The very fact that restoration
has become such a widespread activity refl ects a change
in public attitudes towards watercourses. The current attitudes
are possible only thanks to past social investments
in waste water treatment following passage of the Clean
Water Act in the United States (Wolman, 1971) and
comparable legislation in other developed countries. The
resulting improvements in water quality now make human
contact with urban waters desirable and ecological restoration
feasible, which was not the case in the past when these
channels were open sewers. As society and culture evolve,
goals change and the realm of what is ‘feasible’ in river
restoration can change dramatically, introducing uncertainties
for restoration planning and design – uncertainties
that do not lend themselves to technical engineering
analysis.

44 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
This chapter considers social and cultural dimensions
of river restoration, and related uncertainties for restoration
planning and design. While recognized as important,
they are probably less well understood by the agencies
involved in designing and implementing restoration projects
(FISCRWG, 1998). Systematic studies of public
attitudes and expectations regarding river restoration
(Tunstall et al., 2000) have been rare in light of the enormous
societal investment in restoration projects. There is
no way this short chapter can do justice to this broad topic,
but herein we attempt to raise some issues relevant to the
enterprise of river restoration, which we hope will be
useful to scientists and practitioners in the fi eld. We fi rst
briefl y review some important social aspects of river restoration:
the land-use context of restoration projects,
underlying cultural preferences in river restoration design
and the increasing importance of public participation
in river management and restoration programs. We
draw upon recent research to consider various human
activities in urban streams and the confl icts among restoration
goals of various professionals and stakeholder
groups. Finally, we present two brief case studies from
northern California, which illustrate social issues and
attendant uncertainties in river restoration.
4.2 OVERVIEW OF SOCIAL ASPECTS OF
RIVER RESTORATION
4.2.1 An Urban–Rural–Wilderness Continuum
Appropriate goals and the solutions possible vary widely
with context, from near wilderness to dense urban settings
(Figure 4.1). Where catchment processes are relatively
unaltered and runoff and sediment load are virtually
Attributes
General approach
Examples
WILDERNESS
Unaltered watershed.
Channel may be altered.
Can restore pre-disturbance,
historical channel, by either:
1) letting river restore itself
2) ‘carbon-copy’ approach
Flow regime unchanged.
Sediment load unchanged.
No urban encroachment.
Figure 4.1 An urban–wilderness continuum in river restoration
unchanged, a restoration project can logically seek to
restore pre-disturbance channel conditions, either by
giving fl oods and sediment transport the opportunity to
recreate natural channel conditions (an approach often
termed ‘passive restoration’) or by proactively reconstructing
pre-disturbance channel form (the ‘carboncopy’
approach of Brookes and Shields, 1996). An example
would be a channel whose catchment land use has remained
constant, but whose form was altered by channel straightening
or by removal of bank vegetation and consequent
instability. At this wilderness end of the continuum, it
makes sense to either let natural processes accomplish
the restoration or to use the pre-disturbance channel as a
template, because the processes that supported the predisturbance
channel will tend to support the same channel
form again. In either case, to maintain ecological values,
the channel should be permitted to migrate freely and to
fl ood overbank areas (Ward and Stanford, 1995).
Moving towards the urbanized end of the continuum,
land use change in the catchment has altered runoff
and sediment load, so there is no reason to expect predisturbance
channel dimensions to be maintained by
current processes. At the extreme, urban development in
the catchment increases peak fl ows such that the channel
tends to incise, which, if uncontrolled, may lead to bank
collapse and channel widening. However, encroachment
of urban development to the channel margins means that
incision and channel widening are socially unacceptable.
At this urban extreme, restoration projects must be built
to convey urban runoff without fl ooding adjacent lands
and to withstand increased shear stresses of urban runoff
without erosion. Here, restoration can be viewed as a form
of gardening, in which the elements are deliberately
chosen and maintained by human input, albeit one that
HIGHLY URBAN
Transitional cases. Highly altered watershed and channel.
Encroached banks.
Can partly restore processes.
Must decide what changes to accept as
constraints, what to try to change/restore.
Increase releases from dam?
Reduce peak urban flow by detention?
Add gravel below dams?
Reduce erosion in watershed?
Remove houses along bank/floodplain?
Cannot restore historical conditions.
Restoration
as
‘ gardening:
’
choose elements to include but must account for
erosive forces, altered hydrology.
Social issues are important: potential to improve
ecology/water quality are limited, so emphasis on
community-building and environmental education.
Flow regime altered.
Sediment load altered.
Urban encroachment to banks.
Attributes
General approach
Examples

equires hard structures to resist erosive forces of urbanrunoff-augmented
fl oods. In such cases, ‘naturalness’ in
restoration may be viewed as more an aesthetic choice
than a real design approach. Such channels can be highly
successful in providing recreational and aesthetic amenities,
linking communities through walking and biking
trails, even providing for kayaking and canoeing. However,
they may be viewed as ‘water features’ (to borrow a term
from the fi eld of landscape architecture) capable of conveying
fl oodwaters within the channel and without eroding
banks, rather than large scale, dynamic ecosystems. They
can still provide ecological benefi ts at a local scale, but
without rebuilding of the urban infrastructure these waterways
are unlikely to support sensitive target species at a
large scale. In these highly urban settings, the ecological
potential from a restoration project can rarely be comparable
to that achieved in a less urban setting, and thus
urban restoration projects may be best justifi ed by their
potential social benefi ts as they respond to human needs
and uses.
Many restoration projects can be seen to fall on a continuum
between these two extremes, with constraints, but
with the potential to restore some natural processes and
functions. It is in these intermediate cases that the greatest
uncertainties arise, as one person’s ‘constraint’ may be
another’s opportunity to restore process. For example, if
an upstream reservoir has eliminated the natural magnitude
and frequency of fl oods, should we accept this as
a constraint that effectively limits the degree to which
natural ecosystem processes can be restored, or do we
seek to alter the reservoir operation rules to more closely
mimic natural fl ow patterns? Reservoir release patterns
have been altered and aquatic ecological conditions
improved on rivers such as the Green River, Kentucky
(Postel and Ritcher, 2004), the St Mary River, Alberta
(Rood and Mahoney, 2000) and Putah Creek, California
(Marchetti and Moyle, 2001). Similarly, does the existence
of human infrastructure or housing on a fl oodplain
mean we cannot inundate this fl oodplain? Or should the
restoration project include compensation for moving the
inappropriately-sited land use to higher ground, so overbank
fl ooding processes can be restored? These are social/
political decisions, which can be informed by science, but
which cannot be predicted technically – adding substantial
uncertainty to restoration planning.
4.2.2 Cultural Preferences in River
Restoration Design
Unstated and often unacknowledged cultural preferences
probably underlie many restoration design decisions. For
example, grassy banks are preferred over shrubby or
Planning River Restoration Projects: Social and Cultural Dimensions 45
wooded banks along many restored streams in northern
Europe, refl ecting the long history of pastoral land use.
Similarly, open park-like landscape seems to be broadly
preferred in western culture (Appleton, 1975), and residents
near urban stream restoration projects in northern
California have complained when restored streams become
too ‘bushy’ and woody riparian vegetation blocks visual
access to the stream bed (Purcell et al., 2002). Similarly,
large woody debris in channels imparts a messy look, to
which most people have a negative reaction (Piégay et al.,
2005).
In North America, restoration projects seeking to create
stable, symmetrically-meandering channels have proliferated.
In some cases, previously single-thread channels
have been reconstructed in attempts to create a more ideal,
symmetrical meandering form in the belief that these
would be more stable (Smith and Prestegaard, 2005). In
other cases, the channels have been reconstructed with the
goal of converting braided rivers to single-thread, meandering
rivers. In many cases, the streams so ‘restored’ were
never single-thread meandering channels under natural
conditions, and the projects can be viewed as essentially
attempts to impose an idealized meandering form onto the
river, as illustrated on Uvas Creek, California (Kondolf
et al., 2001). Many of these channel reconstructions have
washed out within months or years (Figure 4.2). Despite
its mixed record of performance, the design approach
underlying most of these projects – application of the
classifi cation scheme of Rosgen (1994) (NRC, 1992) –
continues to be popular among government agencies
responsible for funding restoration projects. This is
probably due to the ease with which the classifi cation
scheme can be used and applied by those without academic
training in fl uvial geomorphology, the availability
of commercial short courses teaching users how to apply
the scheme and – though largely unrealized and unacknowledged
– the likelihood that the channel designs that
result from applying the scheme satisfy a deep-seated
cultural preference for stable, single-thread meandering
channels.
Research on human responses to landscape form suggest
that subjects (at least in western culture) tend to prefer
the ‘defl ected vistas’ in curved paths, rivers and valleys
over straight lines (Appleton, 1975), in part because they
elicited curiosity in subjects (Ulrich, 1983). Kaplan and
Kaplan (1984) designated this landscape property as
‘mystery’, conveying the opportunity to explore and a
promise to learn more with a changing vantage point as
one moves more deeply into the scene. What is probably
a (near-) universal attraction to the form of meandering
channels was recognized in the 18th century by Hogarth
(1753), who proposed that the ‘serpentine’ line provided

46 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
a
b
Figure 4.2 (See also colour plate section) Uvas Creek viewed
downstream from Santa Teresa Road bridge: (a) January 1996,
two months after completion of the channel reconstruction
project; (b) July 1997, after the constructed channel washed out
in February 1996 during an approximately six-year fl ow (Photo
(a) courtesy of the City of Gilroy, (b) by Kondolf.)
the greatest aesthetic pleasure, and more so when actively
moving. A river moving through a meandering channel
thus has the elements needed for the experience of beauty
under Hogarth’s theory. This preference also found expression
in the work of late 18th century English landscape
designers such as Capability Brown, who built serpentine
channels on the estates of their wealthy clients.
Despite the evidence for a landscape preference for the
meandering channel form, the justifi cations for meandering
channels specifi ed in river restoration projects in North
America are almost always stated in terms of bankfull
discharge, width–depth ratios, meander wavelengths etc.
The fundamental question as to whether a meandering
channel is appropriate at all is rarely addressed. Similarly,
the notion that channels should be stable can be viewed
as largely anthropocentric. Dynamic channels with variable
fl ow regimes tend to support the greatest variety of
habitats and best ecosystem function (Ward and Stanford,
1995; Poff et al., 1997). Yet the meandering channels
constructed using the Rosgen approach have the outside
of meander bends armored by root wads and boulders,
with rock weirs at the crossovers to keep the main current
away from the banks (e.g. Uvas Creek in Figure 4.2(a)).
Indeed, the Rosgen scheme is used to select the ‘proper’
geometry for a site, ‘proper’ meaning it will be stable. We
do not argue with the need to armor channels in dense
urban areas or elsewhere when infrastructure is threatened
by channel migration, but these restoration projects typically
include armored banks even at sites where channel
migration would not threaten human works. The armoring
seems to be accepted in part because it consists of ‘natural’
materials (i.e. it is not concrete) and because those involved
in funding and designing these projects hold a belief that
a stable channel is preferable to an eroding channel, even
if in a rural or park setting.
Finally, stable, meandering channels, fl anked by grassy
banks, probably appeal to our aesthetic senses in large part
because they are ‘tidy’ landscapes. Natural riparian corridors
are frequently inaccessible thickets, which, while
great habitat for wildlife, are unappealing to our western
aesthetic sensibility. Nassauer (1995) demonstrated that
for such ‘messy’ ecosystems to be widely accepted, we
must set them off within a frame that conveys to the
viewer that the messiness is deliberate and not a sign of
neglect. She demonstrated how ‘cues to care’ such as a
neatly maintained fence around a yard of native prairie
could make the otherwise messy bit of landscape acceptable
within the context of a suburban street.
To the extent that public support for restoration is based
on culturally-driven landscape preferences that are not
recognized or articulated, this creates enormous uncertainty
in river restoration projects, as public support cannot
be predicted based on ‘logical’ analysis of how best to
improve aquatic ecology or to manage fl oods. There is
another factor, which cannot be predicted by technical
experts. The topic of human preference in landscape is an
area of active research. Many of the fi ndings probably
have relevance for river restoration, besides the few
touched upon here.
4.2.3 Public Participation and Active Stakeholders
Today, public participation has become an institutionalized
element in stream restoration. Public acceptance and
support in many cases determines the ultimate success and
sustainability of a project. For example, providing public
access to a restoration plan can substantially increase
public support for the plan (Bauer et al., 2002). Support
for restoration is important not only in advocating for the
proposed project, but also in its stewardship after construction.
Stewardship can be developed by encouraging
people to experience the restored natural areas (Ryan
et al., 2002).

Increasingly the role of stakeholders is not limited to
providing review comments on draft documents, but
to active participation in setting objectives and selecting
implementation strategies. The success of such a collaborative
planning process is often evaluated by whether or
not agreement is reached among interest groups. This
approach implicitly assumes that there is an optimal solution
that satisfi es all interests and is technically feasible.
However, there is no a priori reason to assume that this
is the case, and in fact there are good reasons to expect
it frequently will not be. Accordingly, confl icts can arise
among different actors, such as between stakeholder
groups, between professionals in different fi elds and
between design professionals and residents, all creating
uncertainties for restoration planning. These are discussed
in more detail in the following pages. Although water
policy making and planning remains a much contended
arena in California, collaborative planning or policy
making has been documented as benefi cial not simply
based on whether or not consent is reached among various
stakeholder groups, but through the long term, invisible
outcomes in terms of collective learning and accumulation
of social, political and economic capitals (Connick and
Innes, 2003).
An important feature of river restoration today is the
proliferation of local creek groups, known in the United
States as ‘Friends of’ the local creek, in the United
Kingdom as river ‘Trusts’ (e.g. the Eden Rivers Trust). In
the San Francisco Bay Area, these groups have formed a
signifi cant force in shaping the fate of restoration projects.
Friends groups not only voice their desires during restoration
planning, but in many cases they have become
the task force of implementing plans and management
regimes. To the restoration project designer, the potential
role of local creek groups is a source of uncertainty. If a
local group is active, it is important to work closely with
it, both to improve the project design with respect to its
social functioning and to improve the chances of successful
implementation and sustainability by virtue of the
public support a local group can often provide.
There are also fundamental issues with representativeness
in the stakeholder and public participation process.
These processes can be drawn-out, and the long term
active participants tend to be agency staff or industry
representatives for whom participation is part of their job,
or staff of NGOs who are often stretched thinly amongst
many such processes. To actively participate, members of
the public at large must have the time and energy to devote
to meetings over a long period (often exceeding a year) at
their own expense. Unless they are strongly motivated –
often by an imminent threat such as stopping a development
in their neighborhood – few can fi nd the time to be
Planning River Restoration Projects: Social and Cultural Dimensions 47
active in the public participation process. This is refl ected
by the survey results of the ‘befriended’ watersheds in
the San Francisco Bay Area. Neighborhoods with active
‘Friends’ groups have a much higher average income than
areas that do not form creek groups (Mozingo, 2005),
showing urban stream stewardship in the United States
still serves a clientele biased toward the upper and middle
classes.
4.3 HUMAN USES OF URBAN WATERWAYS
While the habitat requirements of fi sh have been extensively
studied (Reiser and Bjornn, 1979) and are used as
a basis for design of restoration projects oriented towards
salmon and trout (Flosi et al., 1998), the habitat requirements
of humans in the stream environment, broadly
construed, are less well understood. Recent research into
why certain activities occur spontaneously at certain parts
of the stream suggests that there are fundamental characteristics
of streams that encourage, and can be designed
for, recreational use. Here we review a range of human
uses of stream corridors, emphasizing urban and suburban
settings.
4.3.1 Camping by Homeless
Riparian corridors have long been preferred sites for
camping by homeless people. River corridors were sites of
large camps of migratory workers and tramps in North
America during the depression of the 1930s, and homeless
encampments are a common element along urban streams
in California today, offering a relatively secluded refuge for
the ‘down and out’. Homeless camps are often found under
bridges, exploiting the shelter from rain, although these
sites are more accessible and thus more likely to be visited
by others and less private (Figure 4.3). Along Ledgewood
Creek near Fairfi eld, California, a camp of fi fty residents
had tents with carpeted fl oors, furniture, and batterypowered
television; the residents reportedly left during
periodic police sweeps, only to return (Fagan, 2005).
Migrants from the provinces of Cuba have settled along
the banks of the Almendares River in Havana, forming a
squatter community known as ‘El Fangito’ (Figure 4.4).
While the streets are mud, many of these dwellings feature
cement or tiled fl oors, furniture and television sets.
Although the settlement was illegal, utilities have hooked
up electrical power and water; sewage fl ows mostly
through buried pipes directly to the river. The fl oodplain
occupied by El Fangito is fl ooded every few years. Residents
take their television sets and leave for higher ground
when the river begins to rise. The management plan of the
Metropolitan Park of Havana (Fornes, 1994) calls for

48 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Figure 4.3 (See also colour plate section) Homeless campsite,
San Pablo Creek, California (Photo by Kondolf, February
2005.)
Figure 4.4 (See also colour plate section) The squatter neighborhood
El Fangito, Havana (Photo by Kondolf, March 2005.)
moving these residents from El Fangito to better, permanent
housing and reforestation of the fl oodplain, but an
international aid agency recently granted funds to build
a levee around the settlement to protect it from
fl oodwaters.
The authors are aware of no studies focusing on homeless
use of riverine spaces, but our fi eld studies in
California suggest that this user group has little tolerance
for other user groups and vice versa. We have observed
that no other users were present near homeless camps
(usually under bridges and behind thickets on fl oodplains).
In Sonoma, California, children were often warned off by
parents or scared away when they accidentally invaded
homeless territory (Yang, 2004). In Japan, while homeless
people also frequently reside under bridges of urban
streams, the tension was not as high, as homelessness in
Japan is regarded mainly as a product of unjust industrial
structure (Dohi, 1999), with little association with drug
abuse and crime. In America, the fl ood of homelessness
during the past four decades is largely attributed to the
failure of ‘deinstitutionalization,’ a major initiative under
the Community Mental Health program that started in
1963 (CCHR, 2004).
Occupation of river corridors by homeless can be an
important source of uncertainty to the outcome of river
restoration efforts. Use of stream corridors by homeless
people has not (to the authors’ knowledge) been encouraged
by designers. However, it is clearly one of the biggest
uses along many urban rivers and streams. Because the
presence of homeless people could discourage use by
other groups, the actual use of a restored stream corridor
may be very different from that anticipated by project
designers, introducing uncertainties.
4.3.2 Fishing
Fishing is a traditional use of rivers and streams, ranging
from subsistence fi shing with traps and nets to purely sport
fi shing in which the fi sh is released back to the stream.
Fisheries in urban channels range widely from wild, anadromous
salmonids in urban channels in the Pacifi c Northwest
of North America to warm water species pulled from
the polluted waters of Asian cities. Fishing is usually well
regulated by licensing and many streams are artifi cially
stocked. Fishing is a well documented and well studied
activity in rivers, a large subject well treated elsewhere
and beyond the scope of this chapter. However, we point
out that fi shing has long been an important activity drawing
people to rivers, similar to other activities we discuss
below. Improving fi sh habitat is cited as a goal for many
restoration projects and many funding sources are available
to improve fi sheries.
4.3.3 Water Sports
Urban rivers (if not so polluted as to be unpleasant) have
long been used for canoeing and fl oating. On summer

weekends, the Chattahoochee River near Atlanta, Georgia,
is packed with young people fl oating downstream on tire
inner tubes or rafts. Increasingly, more active forms of
kayaking and canoeing are being designed for in urban
river restoration projects. For example, the steeper, upper
reaches of the restored Boulder Creek in Boulder, Colorado,
have been designed as a kayak course, and kayakers
and canoers commonly continue downstream through the
town.
4.4 SPONTANEOUS USES OF
URBAN WATERWAYS
Although recreation is commonly cited as a goal of stream
restoration projects, it is often treated perfunctorily compared
to other goals such as fl ood control and habitat,
except where its value can be expressed in monetary
terms. In the context of cost and benefi t analysis, emphasis
on recreation necessarily narrows down to the licensed,
quantifi able activities such as fi shing and boating (NRC,
1992). In contrast to such vacation-orientated uses, there
is a suite of more intuitive and unplanned activities, hereby
named ‘spontaneous uses,’ that involve direct and active
interaction with the landscape, such as skipping rocks,
catching frogs, collecting nuts and swimming. When
human uses are considered at all in urban stream restoration
projects, the focus is typically on passive uses, such
as trail walking and social gathering. The orientation is
also often towards adults uses only, whereas children may
have very different (and strongly felt) attitudes towards
stream environments (Tunstall et al., 2004; Yang, 2004).
However, a growing literature suggests that the more
interactive activities are crucial to the forming of environmental
awareness (Chawla, 1988; Harvey, 1989; Orr,
1992) and place attachment (Owens, 1988; Hester et al.,
1988; Cooper-Marcus, 1992), and are benefi cial for
healthy human development (Nicholson, 1971; Kaplan,
1977; Cobb, 1977; Hart, 1979; Moore, 1986). Whether
a restored channel encourages spontaneous use or not
is a source of uncertainty to the ‘social’ success of a
restoration project.
To understand the specifi c habitat characteristics that
permit and encourage spontaneous uses, Yang (2004)
reviewed the literature to identify probable habitat characteristics
encouraging such uses and then undertook systematic
fi eld observations, especially of children, and
interviews of children and adults in fi eld areas in California
and Japan. Although many of these interactions were
engaged mainly by children, they were not enjoyed by
children exclusively. Adults accompanying children, or
even among a group of adults, appeared to fully enjoy
such uses. From this research, we summarized the most
Planning River Restoration Projects: Social and Cultural Dimensions 49
common types of spontaneous interaction and their habitat
requirements.
4.4.1 Quiet and Secluded Use
Users who appreciate the stream environment in a transcendent
way, go to the stream for a temporary escape,
enjoy intimate relationships with signifi cant others and
those who pursue quiet reading, thinking etc., are commonly
attached to a specifi c base-point. Their territory
may seem small, but the quality demands are high and
specifi c. Since such users can stay for hours, a certain
comfort level (dry seating, foothold and shade) is normally
required. A rock, tree root, log, or a soft grassy spot
by water are particularly appealing. Yet more than anything
else they need privacy, or visual/auditory seclusion
from supervision or other users. Lewis (1995) highlighted
the value of San Leandro Creek, California, as a secret
hiding place and unsupervised play area, with many ‘fi rsttime’
events of local youth. All his interviewees who
played there appreciated this quality of nonsupervision.
For this reason, they preferred detoured or inconspicuous
access and a back screen.
The view toward dense foliage, open fi eld or expression
of water surface, the sound of trickling water, the appearance
of wildlife and easy access to water all tremendously
enhance the value of quiet and secluded base-points, as
users cite these elements as bestowing the healing power
of nature. Both adults and children have been found to
seek out space for quiet and secluded use (Yang, 2004).
Quiet and secluded base-points are easily lost, not only
because privacy is often lost with increased urbanization,
but also because planners and designers usually don’t
design for such spots, operating instead on a design model
of a cheerful park for adult socializing and playgrounds
where all children play together.
4.4.2 Adventures
Adventure connects known to unknown parts in the landscape,
expanding cognitive and physical territory. Adventurers
walk, bike, swim, leap, climb, creep and cross to
‘conquer’ a new piece of landscape. A system of basepoints
connected by diverse, usually three-dimensional
paths, plays an important role in the process of expanding
territory. Dirt paths apparently possess special values to
adventurers. On Marsh Creek, California, some adults
favor dirt paths for aesthetic reasons, but children preferred
dirt paths for the practical reasons that they are
usually avoided by adult bikers and runners, who are often
impatient with children in their way, and the dirt path
provides more interactive features (Yang, 2004). On dirt

50 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
paths children can jump on mounds, leap into muddy
puddles, bend under low branches, or crouch to stare into
a gopher hole. The mounds, puddles, low branches or
gopher holes would all be considered undesirable and
eliminated on a paved trail, but on a dirt path they provide
tempting invitations for sensorial experiences.
Adventurers are particularly keen to fi nd good stream
crossing points. In smaller streams users search shallow
and narrow spots with stepping rocks to set foot on or from
which to build a bridge. In large rivers swimming across
is a common game. If a rope and tree are available and the
stream is narrow enough, they swing across the channel.
Similarly, children in Turkey Brook, London, asked for
ropes to swing on and logs to slide down (Tunstall et al.,
2004). When the slope is right and the channel not too
wide, bikers or skateboarders fl y across on wheels.
Metal culverts are especially attractive to adventurers:
it’s easy to make loud, eerie echoes in them, they are secret
hide-outs, they offer the allure of a connection to somewhere
else and they are perceived somehow off-limits.
4.4.3 Wildlife Contact
Wildlife contact can be by simple observation or active
catching, two distinct modes of interaction. Observers are
usually interested in all life forms they see, from little bugs
to big animals such as otters and raccoons. They interact
with the stream with a highly intensive but unintrusive
way. Wildlife sightings often occur in unexpected, uncalculated
moments, producing a ‘wow’ experience.
Catchers are more physically active and focus on certain
target species, which are small enough to catch. Catching
wildlife along creeks has traditionally provided subsistence,
but in urban areas in developed nations today, catching
is usually based upon affi nity toward the target and a
sense of achievement. Fish, frogs, tadpoles, shrimps, crawdads,
crabs and insects are fascinating creatures for users
to match wits with. The habitats of catchers are as diverse
as those of their target species and their spots correspond
directly to those of their quarry. Children who actively
catch wildlife tend to be agile and willing to access diffi -
cult sites, get wet or scratched and in general are highly
adaptive to their environments (Figure 4.5). In Marsh
Creek, crawdad hunters were often observed thriving at
the least ‘user-friendly’ spots, such as among rugged
riprap under road bridges or by grassy, muddy shores.
Methods of catching are numerous, even for the same
species. They range from the bare hand to highly elaborated
means and tools. Catchers in various regions in
Japan and California often stored captured fi sh or crawdads
temporarily in a container or a little pond enclosed
with sand or rocks. Most of the trapped creatures were set
Figure 4.5 Crawdad from Marsh Creek caught by a child and
drawing of Marsh Creek wildlife by 4th grade child (both from
Yang, 2004.)
free after a short time, but some captures would become
pets to be enjoyed at home until they expired. Many catchers
have learned from experiences which animals ‘work
better’ as pets (Yang, 2004).
Profi cient catchers and observers are often knowledgeable;
they can usually identify many species and know
when and where to fi nd them. Observers and catchers have
similar habitat requirements: the environment needs to
support a suffi ciently high density of wildlife and a meaningful
human/wildlife interface. Though the former is a
widely claimed goal in restoration and greenway projects,
the latter is usually discouraged. For spontaneous users, a
meaningful wildlife/human interface provides plenty of
chances for close-up observation and hands-on catching,
without the need of specialized equipment beyond that
which can be made at home or obtained from a grocery
store. Examples of such interfaces are water edges framed
by vegetation or porous structures where different species
hide, or shallow water reaches adjacent to gravel bars
where fry of amphibians and fi sh hatch. It is important that
water edges designed to sustain a dense wildlife popula-

tion also remain accessible to users, except in cases (rare
in urban areas) where protected species need to be isolated
from human harassment. When physical access is not feasible,
visual access can be provided from the bank, bridges
etc.
4.4.4 Manipulating the Environment
The value of creeks and rivers for spontaneous use depends
largely on their provision of loose parts – elements that
can be easily manipulated in the environment (Nicholson,
1971). At least three categories of common uses rely on
contact with rocks, plants, junk and other kinds of loose
parts in stream environments: collecting, building and
clever craft:
• Collecting allows one to discern treasures from the basically
chaotic stream environment. Once purposely rummaged
or fortuitously encountered, stones and other
elements from the bed, plant parts and junk recovered
from banks may be used in drama play, building, or in
displays in the collector’s yard or room. Gravel bars are
prized sources of stones for collecting and clay banks
provide material for handcraft, mud ball fi ghts and gray
make-up.
• Building projects, whether big (e.g. tree houses, huts,
bases, dams, bridges, ponds) or small (arranging rocks
and sticks) are rooted in an innate attempt to create an
impact on the landscape (Figure 4.6). Through building,
users claim their ownership and adapt the stream to
themselves. The result of spontaneous building usually
is not durable enough to survive fl oods and other natural
processes. Building may have practical purposes, but the
process is all-important: many children build, destroy
and rebuild.
Figure 4.6 12-year-old child’s stove in drama house by Marsh
Creek (from Yang, 2004.)
Planning River Restoration Projects: Social and Cultural Dimensions 51
• Clever crafts, the skillful manipulation of materials
found in stream environments (Yang, 2004), is usually
quite precise in terms of materials and surroundings.
For example, to skip a rock (a trans-culturally popular
trick), one needs a gravel bar containing platy stones
with intermediate axes usually between about 30 and
70 mm, and a fl at pool allowing satisfactory skips. Along
Sonoma Creek in California, adults applied red algae to
skin rash and children made fl utes with deer grass.
Along Kure River in Japan, smashed mugwort was used
to heal scratches and defog goggles, while foxtail stalks
were made into knots to trap frogs and shrimp.
4.4.5 Wading and Paddling
Small children and other users who don’t want to get very
wet will wade in waters shallower than 0.5 m, with currents
20 cm/s or less, such as shallow margins or backwaters.
The range of paddling by children is usually only a
few meters from the water edge and the dry spot. In large
streams, some hints of boundary around a smaller space
(e.g. a cover or re-entrant in the bank or offshore bar) are
needed to overcome the uneasiness induced by an unlimited
expanse of water. Paddlers prefer gently sloping
access to water (rather than grassy or upright banks) and
sandy or clay bottoms (which provide comfortable footholds)
(Yang, 2004).
4.4.6 Swimming, Flushing and Diving
Swimming occurs mostly in pool reaches more than 0.5 m
deep, with velocities under 0.5 m/s and with gentle and
gradual water edges at bars or ‘ledge’ banks protected
by tree roots as entry points. In large or swift rivers,
swimmers also require ‘stopover bases’ (island bars,
bridge piers etc.) at which to rest. Warm surfaces such as
big rocks, pebble beach, concrete blocks, asphalt roads
etc. are valuable dry spots. Flushing makes clever use of
locally concentrated fl ow (>0.5 m/s) and variations in bed
form. Most commonly, fl ushing is done in riffl es: the
fl usher would start at the end of the pool where the speed
starts to pick up, allowing his body to be carried by the
accelerating current downstream to be caught at the crest
of riffl e (if shallow) or carried through the riffl e (if deeper)
to the next pool. Hard structures in the streams such as
bridge piers can also form concentrated currents for
fl ushing.
Diving in larger streams and rivers with deep pools
is popular on hot days, offering the thrill of a free fall and
the sudden impingement of cool water on the body. The

52 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
diving height is limited by pool depth and the diver’s skill
and nerve. We observed children diving from a 15-m high
treetop in the Kagami River, Japan (Figure 4.7). In stream
environments, a pool deeper than two meters is rare and
considered enough for moderate-height diving. A good
diving spot also has a landing spot with a gentler water
edge, a path to connect landing and launch spots into a loop
and, ideally, choices for different skill levels. Rock outcrops
(adjacent to deep pools because they induce scour at
high fl ow) provide steady footholds for the launch point
and outcrops with complex form and multiple take-off
points for different dive heights allow divers to practice
and build their courage and skills gradually. Diving from
the trees, one experiences the thrill of shaking footholds;
diving from a rope swing, one challenges the arm strength,
body balance and the timing to let go; diving at concrete
levees, one needs to leap forward to avoid the concrete
foundation jetting out beneath the mean water level (Yang,
2004). For ‘thrilling’ water contacts such as fl ushing and
diving, routes that connect back to the set-in points are
indispensable to support their repetitive characteristics.
Figure 4.7 Children diving into the Kagami River, Japan (from
Yang, 2004.)
4.5 CONFLICTS AMONG MULTIPLE GOALS
AND OBJECTIVES
Everybody wants more nature but there has been persistent
confusion about the meaning of ‘restoration’. The controversy
over the term refl ects disagreements over goals, even
when considered only within the physical science realm.
As causative agents, humans constantly change and control
nature to ‘help’ it, leading to fundamental questions about
goals. Given that nature is in constant fl ux and there is no
single correct condition (Hull and Robertson, 2000), the
choice of desired end state for restoration will perforce
involve societal priorities. Because river restoration projects
are now commonly undertaken with the involvement
of multiple professionals and stakeholders, and because
all the goals and objectives are essentially value-driven,
three types of confl icts often arise in project implementation,
creating uncertainties for the course of river restoration
planning and implementation.
4.5.1 Confl icts among Professionals
Engineers, fl uvial geomorphologists, ecologists and landscape
architects are trained to see the stream differently
(Figure 4.8). In the past, they have all shaped or reshaped
streams with their particular value systems and disciplinary
tools. Engineering has been the single most powerful
profession in past stream transformation, altering rivers
for fl ood control, water supply and navigation. Hydraulic
engineers model fl ows under conditions where the variables
are controlled, usually approximating channel shapes
as simpler geometric entities. Deviations from clear water
and Euclidean channel shapes are treated with adjustments
in formulas. Fluvial geomorphologists tend to approach
problems at larger scales and over longer periods. The
engineer or manager may pose a question such as, ‘What
kind of bank protection should we use along this reach
of stream?’ The fl uvial geomorphologist will tend to ask
why the bank is eroding in the fi rst place, whether it
is simply part of the natural channel migration process or
a result of changes in the catchment upstream. Especially
in the latter case, it is likely that placing bank protection
will not ‘solve’ the problem but will induce problems
elsewhere.
Ecologists tend to view streams as organic compounds
of habitats. They perceive fi ne details of leaf litter and its
decompostion, moss on boulders, food chains, and cycles
of nitrogen, carbon etc. Traditional biologists may see
unspoiled natural process as the reference condition
against which to measure degradation and a return to that
condition as a restoration goal. Human activities are
viewed as ‘impact.’ Adding a spatial structural perspective,

they view streams as ‘corridors,’ a crucial element in landscape
to allow movement of species and therefore to maintain
biodiversity and long term genetic diversity. These
professionals have long realized that to maintain a healthy
ecosystem, a river needs fl oods (Poff et al., 1997; Junk
et al., 1989). Developers and fl ood control agencies often
resist losing developable lands to fl ood inundation and,
through their infl uence on the political process, typically
succeed in implementing fl ood control measures such as
dams or levees that permit them to build on fl oodplains.
Similarly, natural channel migration is an important
process to create diverse habitats, but riverside developments
are threatened by bank erosion, resulting in pressure
to stabilize the river bank with hard structures. As a result,
although ecologists and environmental scientists have
Planning River Restoration Projects: Social and Cultural Dimensions 53
Figure 4.8 Different perceptions and attitudes towards rivers by engineers, biologists etc. (Source: Hough (1990), adapted from
drawings originally prepared by Newbury (unpublished data)).
been institutionalized into the planning process since the
1960s, they often remain ‘second-class citizens’ in affecting
the design of urban stream channels (Riley, 1998).
A traditional tenet of landscape architects is to view
landscape in abstract, formal, aesthetic terms: forms,
lines, colors, textures and their inter-relationships (Daniel
and Vining, 1983). Although visual aesthetics are usually
the paramount ‘public’ goal, designers also emphasize the
cultural and historical signifi cance of urban streams, as
well as the user’s experiences. Some landscape architects
are well trained ecologically and effectively integrate ecological
considerations in their designs; some are involved
in successful efforts to redevelop urban waterfronts to
revitalize downtown economies, attract tourists and
provide recreation opportunities for urban residents (Otto

54 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
et al., 2004). These projects provide chances to enhance
physical and visual connection with streams by placing
walkways along them, or by promoting vistas and facing
commercial fronts to the streams (Jones and Battaglia,
1989). Landscape architects also transform fl oodplains to
open spaces to accommodate civic activities such as
exhibits, concerts, fairs or sports. The design of these open
spaces, however, often takes its model from pastoral parks
or architectural plazas and while they may provide effective
urban spaces, the purported uses and the design
schemes are often in confl ict with the chaotic character of
fl oods and organic quality of riparian habitats.
4.5.2 Confl icts Among Stakeholder Groups
Humans use rivers for many different purposes, so it
should come as no surprise that human expectations and
the demands of rivers often confl ict. The same confl icts
that manifest themselves in management of existing river
channels tend to emerge when river restoration projects
are conceived and scoped.
Some of the best documented such confl icts are in the
Colorado River below Glen Canyon Dam, where cold
water releases have allowed a rainbow trout (Oncorhynchus
mykiss) fi shery, highly valued by anglers, to become
established. The rainbow trout are exotic to the river and
would not have survived the high temperatures and high
suspended sediment loads characteristic of the pre-dam
river. The post-dam river is now unfavorable to native fi sh,
such as humpback chub (Cila cypha) and razorback sucker
(Xyrauchen texanus). The native fi sh are ugly and undesirable
as sport fi sh, but they are native to the river and their
numbers have dwindled such that several species are now
listed as threatened or endangered (Schmidt et al., 1998).
Where the exotic trout and native fi sh coexist, the trout
may prey on the natives. Proposed actions to improve
conditions for the native species have encountered resistance
from trout fi shing groups. It is well established that
the reduction in high fl ows effected by Glen Canyon Dam
has had numerous ecological effects on the reach downstream
and thus deliberate high fl ow releases are planned
in efforts to restore the reach. The fi rst such release, a
much-publicized fl ow of 1300 m 3 s −1 in 1996, was only
about one-third of the average annual pre-dam high fl ow.
The fl ow was limited to avoid inundating a rare snail that
had extended its range down the canyon walls during the
post-dam period (Marzolf et al., 1998). Thus, restoration
of a dynamic fl ow regime (with attendant benefi ts for the
river ecosystem) was perceived to confl ict with protection
of the rare snail. A similar confl ict among user groups is
on the North Fork Feather River, California, where high
fl ows released periodically to provide fl ows for rafters
have scoured benthic macroinvertebrates, washing these
and other organisms downstream (Garcia and Associates,
2005).
4.5.3 Confl icts Between Professionals and
Local Groups
Perhaps the best documented example of a professional–
local group confl ict involves terrestrial habitat restoration,
the Chicago prairie restoration controversy. Efforts to
restore 7000 acres of the DuPage County forest reserves
in the Chicago metropolitan area back to the historical oak
savanna and tallgrass prairie condition were attacked by
local groups and residents who opposed removing trees
and brush. Ryan (2000) concluded that the discrepancy
between restoration planners and neighborhood users
stemmed from differences in attachment. While both
groups were attached to nature, their attachment can be
diverse and contradictory. Scientists and volunteers are
attached to a particular type of original landscape, which
is established through environmental criteria such as biodiversity
and system integrity. Such attachment is not
bound to a place – the same habitat image can be reproduced
elsewhere and still be satisfactory. On the other
hand, the attachment of local residents is intertwined in
locale and context. Individual trees, albeit non-native, bear
an identity in terms of furnishing the spot for children to
play tag or a seat for quietness or framing a magnifi cent
view toward sundown. In other words, the attachment of
local residents is composed of life memories.
A different confl ict between professionals and a local
group occurred in the northern Sierra Nevada of California
in the early 1990s. In planning a restoration project on
Jamison Creek in Plumas-Eureka State Park (the Park), a
local nonprofi t group active in implementing stream restoration
projects (but without expertise in fl uvial geomorphology)
challenged the effective discharge analysis
conducted by a university team, contending that the bankfull
discharge was only about one-third that computed by
the university team. Despite a thorough and well documented
scientifi c report supporting the university team’s
analysis, the Park rejected the analysis and sided with the
local nonprofi t group, stating that it preferred the smaller
design discharge because ‘a smaller channel is better for
fi sh habitat’. The Park also cited that fact that the local
Coordinated Resource Management Program group (composed
of local agency staff, landowners and staff of the
nonprofi t group (none of whom possessed expertise in
fl uvial geomorphology) had voted in favor of the lower
design discharge.
The notion that one can arbitrarily choose a design discharge
and build a stream channel to smaller dimensions

is a fascinating one, but not one supported by geomorphic
science. Likewise, the notion that scientifi c questions
should be put to a vote by a group without expertise in the
fi eld raises questions about the role of science in such a
restoration design process. Ultimately, the Park had the
authority and responsibility to design and construct the
channel as it saw fi t. The university team withdrew from
the project and the local nonprofi t group proceeded to
design and build a channel reconstruction in 1995. The
project was damaged by the high fl ows of 1996, repaired,
and then completely washed out by high fl ows in 1997. In
2000, the Park sent out a call for proposals to reconstruct
the channel once again.
4.6 CASE STUDIES
4.6.1 Baxter Creek, El Cerrito
Baxter Creek drains an 11-km2 urban area of El Cerrito,
California, debouching into San Francisco Bay at Richmond
(Figure 4.9). In 1997, the City of El Cerrito replaced
a 70-m reach of failing culvert (in a small neighborhood
park) with an open channel (Figure 4.10). The open
channel was stabilized with a series of boulder weirs
Pacific
Ocean
San
Francisco
San
Pablo
Bay
San
Francisco
Bay
Planning River Restoration Projects: Social and Cultural Dimensions 55
Suisun Bay
Richmond
a b
Brentwood
Berkeley
San Jose
Figure 4.9 Location map, Baxter (a) and Marsh Creeks (b),
Contra Costa County.
(which dissipated energy from the 10% gradient) and the
banks were planted with willow (salix spp). Post-project
appraisals in 1999 and 2004 (Purcell et al., 2002; Purcell,
2004) showed that the biotic condition of the restored
reach was measurably better than an unrestored control
section upstream and that the biotic condition did not
improve further between 1999 (two years post-project)
and 2004 (seven years post-project), indicating the stream
may have reached its biotic potential within two years.
Purcell et al. (2002) conducted an attitudinal survey of
the residents within one block of the daylighted section of
Baxter Creek. Of the 45 responses received, most were
positive overall about the restoration, but many expressed
concerns that the willow trees, some of which had grown
to over 6 m in height, blocked the view across the park
and potentially provided hiding places for burglars. In a
repeat survey of the neighborhood in 2004 (n = 45),
Purcell found that about half of those who had moved to
the neighborhood after the completion of the restoration
did not realize the creek had formerly been in an underground
culvert. Nearly all respondents reported they
enjoyed living near the creek, many citing the sounds of
the water, aesthetics, or accessibility for children or dogs.
69% perceived an improvement since the restoration was
completed; 31% said conditions had worsened. Overall,
the project was successful in creating a vibrant stream
corridor where formerly there had been only a relatively
sterile strip of lawn. The success of the project led to the
formation of the ‘Friends of Baxter Creek’, a group which
subsequently supported two other restoration projects in
downstream reaches of Baxter Creek (Lisa Owens-Viani,
personal communication, 2006).
Figure 4.10 Baxter Creek in Poinsett Park, El Cerrito, California.
Photo by Alison Purcell, April 2007, about 10 years after
construction. Note height of willows, some exceeding 6 m.

56 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
There was some negative reaction to ‘overgrown’
vegetation (Purcell, 2004), which is not unusual in urban
stream restorations in northern California. Blackberry
Creek in the Thousand Oaks School in Berkeley was
removed from an underground culvert and replaced with
an open channel in 1995. Some residents had negative
reactions to the density of willow growth, which precluded
access or seeing into the stream channel. In planning
the 1990 restoration of Cortland Creek in Oakland,
the plans by creek activists to extensively plant willows
(for habitat) met with resistance from local residents, who
did not want to create a place where criminals could hide
(Walter Hood, personal communication, 1995). Though
not as well documented as the Baxter Creek case, many
other urban stream restoration projects have been marked
by similar confl icts between planting willows to enhance
habitat and the desire by residents and police to see into
the channel to discourage criminals. This has been true
especially in low-income neighborhoods, where concern
about crime may be greater.
4.6.2 Marsh Creek, Brentwood, California
Marsh Creek drains 332 km2 , with its upper basin mostly
woodland, rangeland and farmland, and its lower 15 km
traversing a broad alluvial fan, which now supports the
urban areas of Brentwood and Oakley, about 60 km northeast
of San Francisco (Figure 4.9). As typical of Mediterranean-climate
streams, runoff from the catchment was
naturally intermittent in all but wet years. There is little
record of the historical channel conditions in Marsh Creek
in Brentwood, but historical maps from the late 1800s and
early 1900s show multiple, sinuous channels and active
channel migration (Robins and Cain, 2002) (Figure 4.11).
As agriculture expanded onto the fertile soils in the early–
mid 20th century, and in response to fl ooding of downtown
Brentwood in the 1950s, the channel of Marsh Creek was
straightened, the riparian corridor largely cleared and a
fl ood-control reservoir constructed about 3 km upstream
of Brentwood (Figure 4.12).
Brentwood has grown rapidly, increasing in population
from 7500 in 1990, to 23 000 in 2000, to 33 000 in 2003
(Cain et al., 2003). Many of these residents commute (oneway
travel times of over an hour) to jobs in the San Francisco
Bay Region. As Brentwood has grown, interest has
grown in enhancing the creek corridor for human uses,
removing barriers to salmonid migration and improving
stormwater detention. A watershed study (Cain et al., 2003)
documented historical changes in physical and biological
conditions, identifying signifi cant effects of straightening
on channel form and instream habitat, effects of the altered
fl ow regime on habitat, effects of former mercury mining
upstream and urban/agricultural runoff on water quality
and the loss of native plant and animal species.
To better understand the perceptions and preferences of
local residents, Yang (2004) surveyed 1800 residents
living within 400 m of Marsh Creek in Brentwood to
assess their perceptions (and ideal images) of the creek.
The residents consistently presented an ideal image of the
creek, identifying luxuriant woods, year-round running
water, bountiful wildlife and easy access as features of the
‘natural’ or ‘original’ Marsh Creek. However, this idyllic
image of the creek is largely inconsistent with the character
of the creek as documented by historical evidence, with
its Mediterranean-climate runoff regime. Similarly, many
residents delighted in contact with wildlife, but did not
realize that the most contacted species, i.e. crayfi sh, bullfrogs,
bluegill and largemouth bass, are not native, but
exotic generalists. Likewise, with vegetation, most subjects
did not distinguish native from introduced plant
species and even those who could tended to prefer vegetation
that ‘looks natural without being overgrown’ regardless
of origin (Yang, 2004).
The problems identifi ed by the residents contrasted
sharply with those presented by the professionals in the
watershed report. By far the leading concern of surveyed
residents was garbage and dumping in the creek. Many
residents considered the summer water levels too low,
evidently without understanding the highly seasonal nature
of fl ow in Mediterranean-climate streams. Residents also
regarded ‘mosquitoes/pests’ as more serious than ‘monotonous
channel form’ and ‘poor habitat value,’ both major
issues identifi ed in the watershed report (Cain et al.,
2003). Only ‘not enough shade’ was identifi ed as a concern
both by the surveyed residents and in the watershed report
(Yang, 2004).
The substantial differences in perception and landscape
preference between restoration scientists and local residents
will be a source of uncertainty in setting restoration
priorities and garnering public support for restoration
projects. As funding becomes available to plan restoration
projects in Brentwood, these gaps will need to be addressed
in a participatory context so that confl icts in restoration
goals can be reduced.
4.7 CONCLUSIONS
Uncertainties on the social and cultural fronts of stream
restoration can be viewed as signifying forward progress
in the fi eld, rather than simply further impediments
in implementing projects. Twenty years ago, when the
notions of creek restoration fi rst became widespread in the
United States, many engineers regarded ecological concerns
as obstacles to be overcome in the single-minded

a 1 0.5 0
1 km b
Planning River Restoration Projects: Social and Cultural Dimensions 57
Figure 4.11 Topographic map details of marsh Creek in Brentwood: (a) 1914 and (b) 1978. (Source: US Geological Survey topographic
maps.)
Figure 4.12 (See also colour plate section) View of Marsh Creek
channel in Brentwood (Photo by Kondolf, September 1991.)
pursuit of diking, channelizing, straightening and culverting
streams. It was only when engineers started to confront
other viewpoints that ‘uncertainties’ were introduced
in their modus operandi. While confl icts between engineers
and ecologists persist in restoration projects, by and
large the engineering profession has embraced the need to
work effectively with geomorphologists and biologists
to achieve effective ecosystem restoration. Now we see
increasingly that the ecological engineering approach is
perturbed by the uncertainties introduced by social and
cultural concerns. In other words, the current phenomenon
of restoration professionals experiencing uncertainties on
all fronts may be simply an indicator of a rapidly broadening
viewpoint and recognition of problems without commensurate
solutions.

58 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Although the possible range of goals, values, perception,
aesthetic taste, use and meanings of a population for
a stream can be overwhelming, and confl icts sometimes
unavoidable, the goal of sustainable stream restoration
increasingly requires integration of diverse points of view.
Institutionally, citizen groups have now become an integral
part of many restoration efforts. Professionals have
learnt that collaborative planning processes may not bring
about a fast solution, but it may provide substantial long
term benefi ts in terms of political and social capital, and
may yield more sustainable restoration projects.
To be successful on a sustainable basis, stream restorations
must be both technically sound and enjoy strong
public support. Although decisions in stream restoration
are essentially value driven, sound science is fundamental
to constrain the range of possible solutions and evaluate
possible alternatives. Without it, a Jamison Creek situation
can result, in which the responsible agency selects a
scientifi cally unsound option and the project fails. On the
other hand, a technically sound restoration plan is unlikely
to be funded and implemented without strong public
support, and unlikely to be sustainable if built without
local buy-in.
Where there are signifi cant uncertainties on social and
cultural aspects, these should probably be settled before
proceeding to settle technical uncertainties. For example,
until the values of large woody debris for fi sh or boaters
are established, there may be little point in quantifying
its catchment production and morphological qualities. In
cities, we fi nd that the recreational potential of spontaneous
uses is often conspicuous in its absence from the
agenda of stream restoration. Once their importance is
recognized and spontaneous uses and their implied societal
values are added to the restoration agenda, more
precise research may be needed to assess them.
Cultural preferences (commonly unacknowledged)
largely shape restoration goals. Building a culturally preferred
form (such as a stable, meandering channel) is
perfectly reasonable as a restoration goal, but we suspect
that the fi eld would benefi t from an explicit recognition of
this as motivation, rather than cloaking such projects in
seemingly scientifi c details of channel morphology and
(commonly vague) references to improved fi sh habitat. To
the degree that cultural preferences remain unacknowledged,
they introduce greater uncertainty in the trajectory
of restoration projects. Cultural preference for tidy landscapes
over messy landscapes (Piégay et al., 2005) should
be acknowledged, so that ‘overgrown’ riparian zones can
either be ‘framed’ (Nassauer, 1995) or simply avoided in
urban areas. For ecological design to be truly successful
and widely accepted, designers will need to fi nd ways to
make stream restoration compelling as designs (Mozingo,
1997). The concept of ‘eco-revelatory design’ suggests
that by accepting humans into the restored ecosystem and
designing the project to reveal ecological processes, we
may achieve ecosystem restoration (to the extent possible
in urban areas) while still gaining public acceptance
(Galatowitsch, 1998).
4.8 ACKNOWLEDGEMENTS
The research on which this chapter is based was partially
supported by a grant from the University of California,
Berkeley, Department of Landscape Architecture Beatrix
Farrand Fund. Shannah Anderson contributed substantially
with supporting research, fi gure and manuscript
preparation, and review comments. Louise Mozingo contributed
valuable ideas and references. The chapter was
improved through comments by anonymous reviewers
and the volume’s editors, Dave Sear and Steve Darby.
REFERENCES
Appleton J. 1975. The Experience of Landscape. John Wiley &
Sons Ltd: Chichester.
Bauer DM, Cyr NE, Swallow SK. 2002. Public preferences for
compensatory mitigation of salt marsh losses: a contingent
choice of alternatives. Conservation Biology 18: 401–411.
Bernhardt ES et al. 2005. Synthesizing US river restoration
efforts. Science 308: 636–637.
Brookes A, Shields FD. 1996. River Channel Restoration:
Guiding Principles For Sustainable Projects. John Wiley &
Sons Ltd: Chichester.
Cain JR, Robins JD, Beamish SS. 2003. The Past and Present
Condition of the Marsh Creek Watershed. Natural Heritage
Institute: San Francisco, California.
Carson R. 1962. Silent Spring. Houghton Miffl in Company: New
York, New York.
CCHR (Citizens Commission on Human Rights). 2004. The
real crisis on mental health today. Available at Community
Ruin: Psychiatry’s Coercive Care, CCHR: Los Angeles,
California.
Chawla L. 1988. Children’s concern for the natural environment.
Children’s Environment Quarterly 5 (3):13–20.
Cobb E. 1977. The Ecology of Imagination in Childhood. Columbia
University Press: New York, New York.
Connick S, Innes J. 2003. Outcomes of collaborative water policy
making: applying complexity thinking to evaluation. Journal
of Environmental Planning and Management 46: 177–197.
Cooper-Marcus C. 1992. Environmental memories. In: Low SM,
Altman I (Eds), Place Attachment, Human Behavior and Environment:
Advances in Theory and Research, Vol. 12. Plenum
Press: New York, New York; 87–112.
Daniel TC, Vining J. 1983. Methodological issues in the assessment
of landscape quality. In: Altman I, Wohlwill JF (Eds),
Behavior and Natural Environment, Plenum Press: New York,
New York; 39–84.

Dohi M. 1999. The community design process at Kamagasaki,
Osaka, Japan. In: Hester RT, Kweskin C (Eds), Democratic
Design in the Pacifi c Rim, Ridge Times Press: Mendocino,
California; 228–241.
Eden SS, Tunstall S, Tapsell S. 2000. Translating nature: river
restoration as nature-culture. Environment and Planning D:
Society and Space 18: 257–273.
Ehrlich PR. 1968. The Population Bomb. Buccaneer Books: New
York, New York.
Fagan K. 2005. Solano’s hidden homeless. San Francisco Chronicle.
13 November 2005, pp. 1, A10–11.
FISCRWG (Federal Interagency Stream Corridor Restoration
Working Group). 1998. Stream Corridor Restoration Handbook.
Natural Resources Conservation Service: Washington,
DC.
Flosi G et al. 1998. California Salmonid Stream Habitat Restoration
Manual (3rd edition). California Department of Fish and
Game: Sacramento, California.
Fornes J. 1994. Plan for the Metropolitan Park of Havana.
Parque Metropolitano de la Habana: Havana, Cuba.
Galatowitsch SM. 1998. Ecological design for environmental
problem solving. Landscape Journal 17: 99–107.
Garcia and Associates. 2005. Rock Creek-Cresta recreational
streamfl ow monitoring year one report: 2002 macroinvertebrate
drift sampling North Fork Feather River, Plumas County
CA. Prepared for Pacifi c Gas and Electric Company (FERC
No. 1962).
Hamilton J. 1993. Streams of hope. Sierra September/October
1993: 98–122.
Hart R. 1979. Children’s Experience of Place. Irvington Publishers:
New York, New York.
Harvey MR. 1989. Children’s experiences with vegetation.
Children’s Environment Quarterly 6 (1): 36–43.
Hester RT et al. 1988. ‘We’d like to tell you . . .’: Children’s
views of life in Westport, California. Small Town 18 (4):
19–24.
Hogarth W. 1753. The Analysis of Beauty. Reprinted in 1997 by
Yale University Press: New Haven.
Hough M. 1990. Out of Place: Restoring Identity to the Regional
Landscape. Yale University Press, New Haven.
Hull RB, Robertson DP. 2000. The language of nature matters:
we need a more public ecology. In: Gobster PH, Hull RB
(Eds), Restoring Nature, Island Press: Washington, DC;
7–118.
Jones DR, Battaglia AM. 1989. Main Street Rivers: Making Connections
Between Rivers and Towns. Pennsylvania State University:
State College, Pennsylvania.
Junk WJ, Bayley PB, Sparks RE. 1989. The fl ood pulse concept
in river-fl oodplain system. Canadian Journal of Fisheries and
Aquatic Sciences. 106: 110–127.
Kaplan R. 1977. Preference and everyday nature: method and
application. In: Stokols D (Ed), Perspectives on Environment
and Behavior: Theory, Research, and Applications, Plenum
Press: New York, New York; 235–250.
Kaplan R, Kaplan S. 1989. The Experience of Nature: A Psychological
Perspective. Cambridge University Press: New York,
New York.
Planning River Restoration Projects: Social and Cultural Dimensions 59
Kondolf GM, Smeltzer MW, Railsback S. 2001. Design and
performance of a channel reconstruction project in a coastal
California gravel-bed stream. Environmental Management 28
(6): 761–776.
Lewis SA. 1995. Design and planning implications of uses, perceptions
and attitudes of San Leandro Creek. Master’s thesis.
Univeristy of California, Berkeley.
Marchetti MP, Moyle PB. 2001. Effects of fl ow regime on fi sh
assemblages in a regulated California stream. Ecological
Applications 11 (2): 530–539.
Marzolf GR, Valdez RA, Schmidt JC, Webb RH. 1998. Perspectives
on river restoration in the Grand Canyon. Bulletin of the
Ecological Society of America 79 (4): 250–254.
Moore RC. 1986. Childhood’s Domain. Croom Helm: London.
Mozingo LA. 1997. The aesthetics of ecological design: seeing
science as culture. Landcape Journal 16: 46–59.
Mozingo LA. 2005. Community participation and creek restoration
in the East Bay of San Francisco, California. In: Hou J,
Francis M, Brightbill N (Eds), (Re)constructing Communities,
the 5th Pacifi c Rim Conference on Participatory Community
Design, Center for Design Research: University of California,
Davis; 249–251.
Nassauer JI. 1995. Messy ecosystems, orderly frames. Landscape
Journal 14: 161–170.
Nicholson S. 1971. The theory of loose parts. Landscape Architecture
62 (1): 30–34.
NRC (National Research Council). 1992. Restoration of Aquatic
Ecosystems. National Academy Press: Washington, DC.
Orr D. 1992. Ecological Literacy: Education and the Transition
to a Postmodern World. State University of New York Press:
Albany, New York.
Otto B, McCormick K, Leccese M. 2004. Ecological riverfront
design: restoring rivers, connecting communities. American
Planning Association Planning Advisory Service Report
Number 518–519, American Planning Association: Chicago,
Illinois.
Owens PE. 1988. Natural landscapes, gathering places, and prospect
refuges: characteristics of outdoor places valued by teens.
Children’s Environments Quarterly 5 (2): 17–24.
Piégay H, Mutz M, Gregory KJ. 2005. Public perception as a
barrier to introducing wood in rivers for restoration purposes.
Environmental Management 36 (5): 665–674.
Poff NL et al. 1997. The natural fl ow regime. BioScience 47:
769–784.
Postel S, Richter B. 2004. Rivers for Life. Island Press: Covelo,
California.
Purcell AH, Friedrich C, Resh VH. 2002. An assessment of a
small urban stream restoration project in northern California.
Restoration Ecology 10: 685–694.
Purcell AH. 2004. A long term post-project evaluation of an
urban stream restoration project (Baxter Creek, el Cerrito, California).
Term project in River Restoration, LA227, University of
California, Berkeley. Available through the University of California
library website (http://repositories.cdlib.org/wrca/
restoration/)
Reiser DW, Bjornn TC. 1979. Habitat requirements of anadromous
salmonids. USDA General Technical Report PNW-96.

60 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Riley AL. 1998. Restoring Streams in Cities: A Guide for Planners,
Policymakers, and Citizens. Island Press: Washington,
DC.
Robins JD, Cain JR. 2002. The Past and Present Condition of the
Marsh Creek Watershed. Report by the Natural Heritage Institute
and the Delta Science Center to the California Coastal
Conservancy: Oakland, California.
Rood SB, Mahoney JM. 2000. Revised instream fl ow regulation
enables cottonwood recruitment along the St Mary River,
Alberta, Canada. Rivers 7: 109–125.
Rosgen DL. 1994. A classifi cation of natural rivers. Catena 22:
169–199.
Ryan RL. 2000. A people-centered approach to restoration projects:
insights from understanding attachment to urban natural
areas. In: Gobster PH, Hull RB (Eds), Restoring Nature, Island
Press: Washington, DC; 209–228.
Ryan RL, Erickson DL, DeYoung R. 2002. Farmers’ motivations
for adopting conservation practices along riparian zones in a
Midwestern agricultural watershed. Journal of Environmental
Planning and Management 46 (1): 19–37.
Schmidt JC et al. 1998. Science and values in river restoration in
the Grand Canyon. BioScience 48 (9): 735–747.
Smith SM, Prestegaard KL. 2005. Hydraulic performance of a
morphology-based stream channel design. Water Resources
Research 41 (W11413): 1–17.
Tunstall SM, Penning-Rowsell EC, Tapsell SM, Eden SE. 2000.
River restoration: public attitudes and expectations. Journal of
the Institution of Water and Environmental Management 14
(5): 363–370.
Tunstall SM, Tapsell SM, House M. 2004. Children’s perceptions
of river landscapes and play: what children’s photographs
reveal. Landscape Research 29 (2): 181–204.
Ulrich RS. 1983. Aesthetic and effective response to
natural environment. In: Altman I, Wohlwill J (Eds), Behavior
and the Natural Environment, Plenum: New York, New
York.
Ward JV, Stanford JA. 1995. Ecological connectivity in alluvial
river ecosystems and its disruption by fl ow regulation. Regulated
rivers: Research and Management 11: 105–119.
Wolman MG. 1971. The nation’s rivers. Science 174: 905–918.
Yang C-N. 2004. Inviting Spontaneous Use into Urban Streams.
Doctoral dissertation. University of California, Berkeley.
Available online at http://www.lib.berkeley.edu/WRCA/
restoration/theses.html.

River Restoration: Managing the Uncertainty in Restoring Physical Habitat Edited by Stephen Darby and David Sear
© 2008 John Wiley & Sons, Ltd
5
Conceptual and Mathematical Modelling in
River Restoration: Do We Have
Unreasonable Confi dence?
Michael Stewardson 1 and Ian Rutherfurd 2
1 Department of Civil and Environmental Engineering and eWater CRC, The University of Melbourne, Australia
2 Geography Program, School of Resource Management, The University of Melbourne, Australia
5.1 INTRODUCTION
5.1.1 Geomorphic Modelling in
Restoration Planning
From the 1960s to the 1980s, applied fl uvial geomorphology
described the degradation of streams in response to
human disturbance. This endeavor was part of a larger
quest to explain the controls on stream form. From the
1990s, the discipline has found new vigor as skills have
been turned to the restoration of those degraded streams.
However, this change does not simply represent a new job
opportunity; it represents a fundamental test of our knowledge
of processes controlling stream form.
It is relatively easy to describe the degradation of fl uvial
systems, and much of this work has identifi ed associations
with human disturbances rather than strict causations.
Recall the protracted debates about whether arroyo incision
was caused by clearing, channelisation or climate
change (Cooke and Reeves, 1976). It is much more challenging,
fi rstly to recommend priorities for rehabilitating
streams and, secondly, to implement these changes. Geomorphologists
have moved from being observers of human
impact, to advisors and managers, actively intervening in
stream channels for environmental outcomes. Most of this
intervention has been in the areas of: channelised streams
(Larson and Goldsmith, 1997), sediment slugs (Rutherfurd,
2001) and mitigating the effects of dams and fl ow
regulation. Numerous articles describe the contribution
that geomorphologists can make to stream restoration
endeavors and there is no shortage of admonitions to
‘include a geomorphologist on every restoration project’
(Sear, 1994; Brookes, 1995; Brierley et al., 1996; Brookes
and Sear, 1996) and on every stream engineering project
(Gilvear, 1999).
Fluvial geomorphologists produce conceptual and
mathematical models that are central to many stream restoration
projects. In this chapter it is argued that managers,
ecologists, and even geomorphologists themselves, can
have a false sense of confi dence in these models. As communities
around the world invest in stream restoration
projects, false confi dence can have a huge direct and
opportunity cost. We believe it is better to be frank about
model uncertainties from the outset to promote realistic
expectations of project success and a balanced and welltargeted
investment in investigations to reduce these
uncertainties. Fluvial geomorphologists contribute to
stream restoration projects in three main ways:
1. The most basic work of a fl uvial geomorphologist in
a restoration project is to describe how a stream has
changed its form over time, and to identify the factors
that are responsible for these changes. Such reconstructions
have now passed from being a curiosity-driven
activity for scientists, to basic consulting practice. The
result of these investigations is usually a conceptual
model [or a perceptual model in the parlance of Beven
(2001)] that may be based on space-for-time substitution
or historical coincidence between geomorphic and

62 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
other catchment changes. Perhaps the best established
geomorphic conceptual models are the incised stream
models (Schumm et al., 1984; Simon, 1989) where
there may be some empirical guidance for the magnitude
of change (e.g. a threshold width–depth ratio for
incised streams Schumm et al., 1984). The conceptual
model serves to describe the ‘original’ condition of a
system, which often becomes the ‘target’ for the restoration
project. Further, the conceptual model is used to
predict the trajectory of channel change if there is no
intervention.
2. Geomorphologists recommend specifi c actions (or
interventions) that will alter process and form, to move
the stream toward the ‘target’ state, identifi ed from the
conceptual model. These actions usually involve changing
the fl ow or sediment regime, or modifying the
boundary of the channel to cause a change in some
geomorphic variable (e.g. width, scour frequency,
erosion rate). Design of project specifi cs is often based
on a mathematical model of geomorphic response
related to the specifi c action proposed.
3. They also predict the ‘secondary’ (perhaps unintended)
consequences of stream restoration projects. This
could also be described as the ‘sustainability’ of the
intervention.
Thus, uncertainty emerges at three levels: the validity
of the conceptual model; whether the proposed intervention
results in the planned geomorphic change; and, fi nally,
whether the change is sustainable. Typical examples of
in-stream geomorphic actions in restoration projects are:
• Incised, channelised streams are the classical example of
stream restoration. A geomorphologist reconstructs
the original dimensions of a channelised stream in
Denmark by superimposing old maps and photos, and by
making painstaking fi eld observations (Neilsen, 1996).
Working with an engineer, a ‘stable’ re-meandered
stream path that links the palaeochannels for a rehabilitated
stream is then designed. The dimensions of the
channel and the variations in depth (the pool-riffl e
sequence) are designed to be scaled to catchment area
(Newbury and Gaboury, 1993). Finally, the restoration
that will lead to increased erosion downstream of the
reach is predicted.
• Gullying has dumped a large pulse of sand into a stream.
A geomorphologist applies the classical ‘wave’ model
of sand slug migration (Gilbert, 1917; Neilsen, 1996)
and, on the basis of historical movement of the sand
front, concludes that it will take over 50 years for the
sand to move through the reach. Extracting the sand
at a defi ned rate will protect the downstream reaches
and accelerate recovery (Rutherfurd, 2001). Building
artifi cial spur dikes will also create habitat pools
(Kuhnle et al., 2002).
5.1.2 Why do we Care about Uncertainty?
Regan et al. (2002) defi ne epistemic uncertainty as uncertainty
associated with knowledge of the state of a system;
it can be classifi ed into six main types including random
measurement error, natural variation and model uncertainty
(compare to the classifi cation presented in Chapter
3). This chapter is specifi cally concerned with these three
sources of uncertainty in the context of conceptual and
mathematical geomorphic models used in restoration projects.
Measurements of various sorts are used to establish
the state of a system. Measurements may be used directly
(e.g. the diameter of an individual grain of sediment), but
can also be transformed using some kind of calibration
(e.g. stream discharge, which is often estimated from an
observation of stage and transformed to discharge using a
rating curve). Measurement error is the result of errors in
the direct measurement and subsequent transformations.
Natural variation presents a challenge for observing the
true state of a system. To address uncertainty, assumptions
are often made about the statistical properties of environmental
variations. Model uncertainty can arise both from
the choice of variables and processes to be represented in
the model and from the method used to represent the relations
between variables.
Uncertainty is defi ned here as the range of possible
values for a model variable (input or response). One part
of uncertainty is accuracy, which is the difference between
a measurement and the ‘true value’. Uncertainties in
response variables are the consequence of the need to
choose a conceptual and mathematical representation of
river processes, and uncertainties in input parameters for
the model estimated from fi eld measurements, previous
studies or by calibration. Uncertainties can also exist as a
consequence of unknown future environmental conditions
in particular climatic conditions. It is possible that restoration
works are destroyed by an extreme fl ood event not
considered in the planning process. Design models often
consider a range of expected conditions based on conditions
experienced in the past, but uncertainties in the
actual conditions over the life the project create further
uncertainty in geomorphic design.
Discussion of uncertainty, in the realm of management,
can quickly degenerate into an unfocused quest for greater
accuracy, greater precision, more samples and more effort.
It is usually scientists who write about certainty and they
may have a completely different perspective on the issue

Conceptual and Mathematical Modelling in River Restoration: Do We Have Unreasonable Confi dence? 63
to managers. The notion of ‘certainty’ is not simply a
statistical/scientifi c issue; it is at the heart of the social
process of decision making. The implications of estimating
geomorphic uncertainty can be unexpected to those
focused primarily on improved confi dence in model prediction.
Whilst more certain predictions should improve
the chance of project success, consideration of uncertainties
does not in itself improve model performance and can
reduce public support for a project.
It is necessary to be clear about why we care about
geomorphic uncertainty in stream restoration. In reality,
most scientifi c geomorphologists revel in the uncertainties
of river process and form. Most of our endeavour is
directed at situations where accepted models do not work,
where we are uncertain. If it was all certain, we would do
something else. But in the realm of management, uncertainty
is seldom welcome. Whilst the enthusiasm to rehabilitate
streams will not disappear, the initial fl ush of
community and government support for these endeavors
could be lost if geomorphologists and other scientists
foster unrealistic expectations of these projects. As geomorphologists
engage with engineers and managers, they
must decide how to deal with uncertainty (Volkman,
1999), when admitting to too much uncertainty can weaken
support for a project and possibly stall it. However, underplaying
uncertainties will undermine confi dence in geomorphic
advice when some projects inevitably ‘fail’.
Another reason to care about uncertainty is that it provides
the justifi cation for improved geomorphic investigations
prior to completion of a restoration plan. As will be
demonstrated, in some cases it may be relatively cheap and
easy to reduce uncertainty, but it can also be very expensive
to do so. A strong conclusion of this chapter is that,
given the large cost of stream restoration projects, sound
geomorphic advice has often tended to be undervalued.
Large projects are sometimes launched on the basis of
fl imsy conceptual models, possibly because their uncertainty
has not been properly considered.
Finally, adaptive management is often proposed as the
only reasonable way forward in the face of uncertain restoration
outcomes (see Chapter 14). The adaptive approach
is to treat the restoration project as an experiment designed
to inform our knowledge of how rivers respond to restoration.
This knowledge is used to improve river restoration
decisions for the particular river and presumably elsewhere.
Despite the frequent calls for adaptive management
of river restoration, actual examples of success are rare for
a number of reasons (Walters, 1997; Ladson and Argent,
2002). We argue that systematic consideration of uncertainties
in geomorphic modelling is essential to the
application of adaptive approaches in river restoration.
Management experiments and associated monitoring need
be targeted to reducing these uncertainties if they are to
feed back effectively into future restoration practice.
5.1.3 Introduction to Case Studies
Geomorphologists normally acknowledge the uncertainties
in their models but it is rare for these uncertainties to
be quantifi ed or systematically examined. There could be
a perception that these uncertainties are relatively small
and have limited signifi cance in restoration decisions.
There may be limited experience amongst geomorphologists
in the quantitative aspects of uncertainty analysis
which could discourage attempts to handle these explicitly.
There may also be a concern that quantifying uncertainties
will undermine support for a project. In any case,
there is currently very little published information on the
scale of uncertainties in geomorphic studies for river
restoration. Uncertainty analyses in case study projects are
needed to inform discussion of how best to handle these
uncertainties in the future.
This chapter examines uncertainties in geomorphic
modelling for two river restoration projects. The two case
studies involve, respectively, conceptual and design models
for restoration planning. The scale of uncertainties and
also the potential benefi ts of a systematic analysis of
uncertainties in these projects are examined. Based on
these case studies, it is suggested how uncertainties might
best be handled in the future by geomorphologists developing
conceptual and design models for river restoration
planning.
The fi rst case study concerns the development of a
conceptual model for planning Australia’s largest stream
restoration project, on the Snowy River. The infl uence of
the conceptual model on proposed plans is examined and,
in particular, the implication of subsequent changes in the
geomorphic model. The second case study concerns the
design of a fl ushing fl ow in the Goulburn River, Victoria,
using a one-dimensional hydraulic modelling approach. In
this design problem, multiple sources of uncertainties are
quantifi ed and the key uncertainties in designing the fl ushing
fl ow identifi ed. A two-dimensional numerical model
may have had some advantages over a one-dimensional
model approach in this problem. However, a onedimensional
model (similar to HEC RAS) is used because
it is currently the standard approach used throughout the
industry. Thus, we are concerned with the uncertainty associated
with current practice in stream restoration projects
rather than with the level of certainty that is theoretically
possible with detailed scientifi c study. It is often
assumed that uncertainties associated with modelling the
physical response of channels are small (c.f. biological
responses, see Chapter 8). The case studies demonstrate

64 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
that uncertainties associated with the ‘accepted’ approaches
to geomorphic modelling can be substantial.
5.2 CASE STUDY OF A GEOMORPHIC
CONCEPTUAL MODEL
5.2.1 Introduction
The following section describes the chronological development
of the geomorphic conceptual model for the
restoration of the lower 32 km of the Snowy River in
south-eastern Australia. The Snowy River drains a catchment
of 15 800 km2 situated in New South Wales and Victoria
in south-east Australia, before discharging into Bass
Strait (Brizga and Finlayson, 1994). For most of its course,
the river fl ows in a narrow valley, that only widens around
Bete Belong (Figure 5.1). The river below Bete Belong is
a perched, sand-bed, channel between 70 m and 170 m
wide. The downstream end of the Snowy River is estuarine,
with tidal infl uences persisting upstream to Orbost.
Figure 5.1 The lower Snowy River in Victoria (from Finlayson and Bird, 1989)
When the fi rst European settlers arrived in the late
1840s, the Snowy River fl oodplain was swampy wetlands
away from the channel, with the higher levees along the
channel covered in warm temperate rainforest (described
as ‘jungle’) (Owen, 1997). By the 1880s, the stream banks
had been cleared and the clearing and draining of the
wetlands had begun. By the 1930s, artifi cial fl ood levees
had been built along much of the river and much of the
swampy wetlands had been drained for grazing. Over
the same period, large woody debris was removed from
the stream. This began in the 1880s to aid navigation
(Seddon, 1994) and reached a peak in the 1950s to combat
perceived aggradation of the bed and to reduce fl ood peaks
(Finlayson and Bird, 1989).
A second phase of disturbance was initiated by construction
in the river’s headwaters under the Snowy Mountain
Scheme. Completed in 1967, the Snowy Mountain Scheme
diverts water out of the Snowy River catchment for hydroelectric
power generation and for the supply of water for
irrigation in the neighboring Murray and Murrumbidgee

Conceptual and Mathematical Modelling in River Restoration: Do We Have Unreasonable Confi dence? 65
catchments. Erskine et al. (1999) concluded that the Snowy
Mountain Scheme had reduced median and low fl ows by
60–70%, as far downstream as Jarrahmond.
5.2.2 The Geomorphic Model
The following is a chronological summary of the history of
stream channel change recorded by various workers on the
lower Snowy River and explanations for these changes.
1. There are anecdotal suggestions that the lower Snowy
River increased its width at the end of the nineteenth
century. For example, the Snowy River Mail (July 1,
1884) wrote:
‘The river which, when settlement fi rst appeared
here, was not half its present width, is increasing at
every fl ood, widening and shallowing the channel
and precipitating huge trees into its bed . . .’ (cited in
Seddon, 1994).
In reviewing these claims, Finlayson and Bird (1989)
concluded that there was not suffi cient evidence to substantiate
this catastrophic increase in width.
2. Since the 1930s, people living along the Snowy River
have been convinced that the river was becoming
shallower and fi lling with sand (Strom, 1936). Flushing
out this sand was probably the earliest ‘restoration’
target for the river, leading to desnagging and other
works in the stream in the 1950s. Despite the local
conviction that the bed has aggraded, comparisons of
16 repeat cross-sections, dating from the 1920s, along
the Jarrahmond reach, do not support this view (Gippel,
2002). Instead the bed level fl uctuates over a range of
±2 m. Brizga and Finlayson (1994) suggested that the
reduced fl ows of the river since regulation mean that
more of the bed is now visible, leading to the illusion
of aggradation. This is a controversial suggestion
amongst the local people, who still see ameliorating the
effects of sedimentation as the major restoration target
for the river. This perception may come, in part, from
the fact that some deep pools in the river have certainly
fi lled-in since the 1970s, particularly around Bete
Bolong (Gippel, 2002).
3. In the 1990s, attention turned to the effect of the Snowy
Mountain Scheme on the geomorphology of the river.
Brizga and Finlayson (1992) mentioned the loss of
lateral bars, and their associated pools in the lower
Snowy River, and speculated that the cause could be
fl ow regulation (Figure 5.2). Erskine and Tilleard
(1997) described the loss of the bars and related it more
Figure 5.2 1940 aerial photograph of the Lower Snowy River
upstream of Lynn’s Gulch, showing well developed lateral bars
and associated pools
strongly to the loss of some formative discharges after
regulation in 1967. ‘. . . rhythmically spaced, bankattached,
alternate side bars with well defi ned poolriffl
e sequence were present above the estuary at Bete
Bolong before 1967 and they have never reformed
since then . . .’ (Erskine et al., 1999). Stewardson
(1998) concluded that, not only has the frequency of
fl ows that are thought to form bars and pools fallen
dramatically, but the incidence of low fl ows that can
potentially in-fi ll the pools with sediment has increased.
Replacing deep pools with a ‘plane bed’ of sand is
thought to provide poor habitat, particularly for migrating
fi sh (Raadick and O’Connor, 1997). Twelve of the
seventeen fi sh species found in the river are migratory.
Returning the pools to the lower Snowy River presented
an elegant and achievable goal for stream restoration
and was the main recommendation of the fi rst
restoration plan (ID&A, 1998). The recommendation

66 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
was also supported by the Snowy Water Inquiry
(1998a,b). Following these reports, the effort in the
project swung to designing timber pile fi elds (retards)
at the historical locations of the bank attached side
bars. It was hoped that these retards would encourage
sand deposition and eventually scour pools (ID&A,
1998; Gippel et al., 2002).
4. Gippel (2002) has comprehensively reviewed the evidence
for channel change in the lower Snowy River,
adding the evidence from two student theses. Gippel
concludes that it is likely that the Snowy River did
dramatically increase in width after the 1870 fl ood, in
common with other rivers in Gippsland (Brooks and
Brierley, 1997; Brooks et al., 2003). This suggests that
the pre-1967 un-regulated river (that has up till now
formed the ‘reference’ for the restoration strategy) was
in far from ‘natural condition’. Gippel concludes that
the evidence that regulation removed the naturally
occurring lateral bars, is weak. An alternative possibility
is that bars did not form until the river widened after
disturbance, during the early period of European settlement
and, even then, the bars occurred only when
hydrological conditions were ideal. Regulation caused
these ideal hydrological conditions to be less likely. In
addition, Gippel concluded that:
After 1940, the alternate bar and pool morphology
only ever existed in the straight Jarrahmond/Bete
Bolong reach (10.2 km of the 32 km lowland section
of the river).
It has been incorrectly assumed that the alternate bars
always occurred in the same location, with the same
wavelength, when in fact they moved and changed
form.
Even more fundamentally, the link between pools
and fi sh diversity has never been well established.
Even in the original report by Raadik and O’Connor
(1997), the abundance and diversity of fi sh in the
reaches without pools were found to be higher than
in the reference reaches with pools.
There is some question about whether the pools can
be sustained. The persistent low fl ows that characterise
the regulated regime could quickly in-fi ll any
pools that are scoured by favourable fl ow events.
At present, lateral bars and pools remain the major
focus of the restoration plan on the Snowy River. The
Victorian Government is developing a major trial (US$ 1.5
million) of retard structures (in the fi eld and in fl umes), as
a pilot project before building the major pile-fi elds. The
trial has been held up because the local community would
not accept the assurances of the engineers and geomor-
phologists that the proposed log structures would not
cause any change in fl ood stage or duration. Ironically, the
project has been stopped, not because of scientifi c uncertainty
about the goals of the project, but because the
community remains uncertain about something that the
engineers and scientists are certain of: the minor hydraulic
effect of the works on fl oods.
5.2.3 Analysis of Uncertainties
The foundation for stream restoration projects are conceptual
models of physical and biological change. This case
study demonstrates that the problem with much of the
proposed restoration comes from a false sense of certainty
in these models. Several geomorphologists have contributed
to a conceptual model of channel change. The restoration
plan has proceeded on the basis of a conceptual
model that dismissed aggradation by sand as a geomorphic
change, but did identify a clear coincidence between
fl ow regulation and the loss of lateral bars. This neat conceptual
model then formed the basis for a major restoration
project involving engineering works to artifi cially
recreate lateral bars and pools. Over the last fi ve years, the
engineering aspects of the project have taken hold. The
key question for managers now is not whether pools are
an appropriate goal but which engineering design will
develop pools most effi ciently.
It is reasonable to imagine that managers begin by being
uncertain about how a geomorphic system functions. They
then commission investigations that lead to progressively
greater certainty, until restoration decisions can be made
with confi dence. In the turbulent boundary between
science and management (Cullen, 1989), however, certainty
is a fi ckle commodity. Consider the chronology of
certainty in the Snowy River project:
1. The local community and the river managers were
completely certain that the river was aggrading and that
the appropriate restoration strategy was to somehow
remove the sand. Theories of aggradation were subsequently
dismissed following the geomorphic investigation
of Brizga and Finlayson (1994).
2. Early geomorphic assessments concluded that regulation
had led to the loss of alternate bars and restoration
plans were developed to restore them. Gippel’s (2002)
subsequent review revealed strong evidence of channel
widening in the late 1800s. Since alternate bars have
only been found in wider reaches of the lower Snowy
River, and only during ideal hydrological periods,
alternate bars were probably not a natural feature of the
channel prior to European settlement.

Conceptual and Mathematical Modelling in River Restoration: Do We Have Unreasonable Confi dence? 67
3. Finally, the restoration trial is being delayed because
scientists are not able to convince local stakeholders
about the minimal hydraulic effect of the proposed
works.
There are many examples where geomorphic conceptual
models have been found wanting when used as a basis
for management. Much of the criticism in the literature
involves situations where ‘real geomorphology’ has been
replaced by ‘cookbook’ approaches (Sear, 1994). Much of
this criticism has been directed at restoration projects
based on classifi cation systems, such as that of Rosgen
(1996). In one example, Kondolf et al. (2001) described
a meandering channel inappropriately constructed in a
naturally braided stream. Miller and Ritter (1996) provide
a more general review of the Rosgen method. The Snowy
River Project, by contrast, is not superfi cial geomorphology,
but uses the earnest principles recommended by
geomorphologists (Kondolf, 2000). Even ‘real geomorphology’
can be uncertain. Kondolf and Micheli (1995)
were correct when they argued that every stream restoration
project must be considered an experiment. This sits
well with the sometimes stark uncertainties that surround
diagnosis of geomorphic ‘problems’. However, it does not
sit well with multi-million dollar intervention projects.
The lower Snowy River restoration project, for example,
is planned to cost US$20 million, and this is before the
environmental fl ow component is included.
5.2.4 Discussion
Process based conceptual models of stream channel
change (with or without biological conceptual models) are
one of the key contributions of geomorphology to stream
restoration. However, they are also a major source of
uncertainty. A restoration project will proceed on the basis
of a geomorphic conceptual model about which everybody
is initially confi dent. However, the Snowy River case
study shows that this confi dence can be misleading.
Further investigation can reduce confi dence in a model,
but by the time geomorphologists have settled on a conceptual
model that deserves confi dence, the restoration
process has moved on. Changing the model becomes
increasingly diffi cult in the political and management
process. Multi-million dollar restoration projects can be
launched on the basis of uncertain conceptual models. Not
surprisingly, managers and engineers are impatient to
proceed to the ‘real’ business of restoration, which is
building things and changing things. At least here the
uncertainty can be quantifi ed.
Geomorphologists have argued strongly the central
importance of understanding the geomorphic context of
restoration projects to ensure their sustainability. If this
is the case, a restoration plan will be undermined if the
conceptual model from which it was developed is wrong.
In the case of the Snowy River, the initial model of aggradation
led to calls for artifi cial sand extraction to restore
the river. However, such efforts would have been unsustainable
since the sediment in the Snowy River bed was
not a discrete slug of sand. Any pool formed by sand
extraction would have been infi lled during subsequent
storms.
The accuracy of a conceptual model can be critical to
the success of a project. However, there is often only a
poor basis for judging the levels of uncertainty in those
models. Some conceptual models are reasonably simple
and have been tested in numerous situations. The channelised
stream and sand-slug models are examples of welltested
models in general terms, although applying them in
specifi c cases is challenging because they provide a direction
of change rather than either rates or magnitudes.
So, given that each stream requires a variant of a conceptual
model, how is the certainty of that model judged?
At present the model will be presented, with more or less
confi dence, after a geomorphic study. That confi dence is
based on either the weight of reconstructed evidence (‘all
12 of the cross-sections changed in the same way’), on
precedents provided by analogous cases (‘a very similar
model has been described on three nearby rivers’), or on
the reputation and forcefulness of the investigator. Given
that most of these restoration projects cannot wait for
models to be published in the peer-reviewed literature,
how can managers judge the uncertainty of these conceptual
models? Here are fi ve proposals:
1. The most obvious way to improve confi dence is to
subject a conceptual model to anonymous, external
review, by disinterested ‘experts’. Whilst this sounds
easy, the pool of appropriate reviewers can be small
and, from a manager’s perspective, plans can get bogged
down in what appear to be petty, academic debates.
2. Simple guidelines could be established for evaluating
the uncertainty of a conceptual model based on the type
and strength of evidence provided. Evidence in the
form of a mathematical expression of the conceptual
model tested on long term geomorphic data should
provide more confi dence in the model than a model
based on anecdotal accounts. The guidelines could be
applied by the geomorphologists developing the model
to advise managers on model uncertainty. It would
also be possible to use the guidelines to recommend
additional investigations that could reduce model
uncertainty. Alternate conceptual models could also be
compared using these guidelines.

68 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
3. A rule-of-thumb in engineering projects is that about
10% of the total cost of a project should be spent on
design. Given that stream restoration projects tend to
cost AUS$10 5 –$10 6 , perhaps a good guide is to argue
that all of the design would be costing around $US
10 000 to $US 100 000, with the conceptual model
absorbing from 20–40% of that amount. This might
avoid the situation described in the Snowy River
Project, where a simple/elegant conceptual model
based on a modest study sets in train a management
response that it is hard to modify. A problem with this
approach is that the geomorphic investigation that
establishes the conceptual model may be commissioned
before the full costs of the restoration project
are known. In this case, additional geomorphic investigations
may be commissioned later in the planning of
the project after preliminary estimates of project costs.
The challenge is to avoid advancing too far with the
design before the conceptual model is fi nalised.
4. Modellers often develop the model with one set of data
and test it on another set. Thus, some data are held back
for verifi cation. A similar approach could be used with
conceptual models. For example, in the Snowy River
project, a proportion of aerial photographs could have
been held back and used to verify the model later
(given that there are eight sets of photos, this may be
possible).
5. All of the data and information could be collected and
collated and then passed onto an appropriate third party
(or more than one), without interpretation, so that an
independent interpretation of the data could be developed.
The diffi culty with this is that the conceptual
model developed by the third party may require a different
type of data for testing, which would require
further data collation, adding substantially to project
costs.
5.3 CASE STUDY OF A GEOMORPHIC
DESIGN MODEL
5.3.1 Introduction
So far, the role of geomorphologists in using conceptual
models to identify ‘reference’ states and actions that can
move a stream toward the ‘reference’ state have been discussed.
The conceptual model will help to identify a restoration
action, but a mathematical model can be required
to design specifi c aspects of the intervention. The most
common restoration interventions recommended in the
northern hemisphere relate to structural and fl ow changes
that will improve the success of fi sh populations. Having
been involved in many such projects in Australia, we have
noted the tendency for managers (and ecologists) to
assume that the ‘physical stuff’ – the hydrology, hydraulics
and geomorphology – is well understood, with low errors
and low uncertainty. They assume that most uncertainty
lies in the biological responses to intervention. The following
section suggests that they may be giving the geomorphological
studies we examine too much credit!
In this section we examine the uncertainties in a mathematical
model used to design a restoration action. The
case study concerns the design of fl ushing-fl ows for a
gravel-bed stream. Numerous authors have identifi ed the
problem of fi ne sediment infi ltration into a stream bed
known as colmation (Sear, 1993; O’Neill and Kuhns,
1994; Milhous, 1995; Kondolf and Wilcock, 1996). The
loss of competent fl ows below dams means that coarse
beds are infi ltrated by fi ne sediment, damaging habitat for
macroinvertebrates and fi sh. The only way to clean out the
bed is to initiate movement of the coarse fraction. A
common geomorphic goal of fl ow management is to ‘turnover
the bed’. The uncertainty in this question is whether
the predicted fl ow will turn-over the bed. The cost of the
uncertainty is the chance that the bed does not move when
the target amount of water is released. If the bed is not
turned over, then that volume of water has been wasted.
Similarly, if the bed moves at a discharge below the target
discharge, then the extra volume of water released has
been wasted. Thus, in this example, the uncertainty can
be expressed in terms of the extra water that has to be
released to be confi dent that the bed will fl ush. This also
allows the relative saving in water to be estimated if managers
do various things to reduce the uncertainty.
The case study considered is that of fl ushing fl ows for
the mid-Goulburn River, downstream of the Eildon Dam,
in northern Victoria, Australia. The Goulburn River is the
largest stream in Victoria (catchment of 20 000 km 2 ) and
the largest Victorian tributary to the Murray River. It is a
meandering, anabranching river, that is regulated for irrigation
by the Eildon Dam. The bankfull discharge is not
clearly defi ned but fl ow along anabranch channels begins
at about 120 m 3 /s in the reach below the Eildon Dam.
Abundant data is available for the stream, compiled for a
recent environmental fl ow study (DNRE, 2002).
5.3.2 The Geomorphic Model
The volume of water required to deliver the fl ushing fl ow
is estimated in the following three stages. It must be
emphasised that this is an approach used for designing a
fl ushing fl ow. This approach is not advocated for use in
other studies. Alternate methods for calculating fl ushing
fl ows might be considered, particulary given the magni-

Conceptual and Mathematical Modelling in River Restoration: Do We Have Unreasonable Confi dence? 69
tude of uncertainties in this analysis described later. The
three stages are:
1. Estimating the critical bed shear stress for incipient
motion of the bed sediments using Shields’ entrainment
function; the critical shear stress is given by:
τ crit = (γs − γw)d50τ* (5.1)
where γ s and γw are the specifi c weights of the bed sediment
and water respectively, d 50 is the median grain size
of the bed sediments and τ* is Shields’ dimensionless
shear stress, often estimated as 0.045 for gravel bed
rivers (Buffi ngton and Montgomery, 1997).
2. Estimating the discharge at which the mean shear stress
for the reach is equal to the critical shear stress for
incipient motion, using a one-dimensional hydraulic
model (applied over a 2 km reach of the Goulburn
River); and
3. Estimating the volume of water required to mimic the
natural duration and frequency of fl ow spells during
which bed sediments are mobilised based on an analysis
of a modelled natural fl ow series.
Importantly, a one-dimensional hydraulic model is
used rather than the more sophisticated two- or threedimensional
models now available because it is the
approach widely used in practice for restoration projects.
The uncertainties associated with a two-dimensional
model will be different, but it is not possible to suggest
whether they would be smaller or larger than those using
a one-dimensional analysis without a proper investigation.
The modelling procedure also includes the preparation of
input data which includes fi eld surveys and selection of
an appropriate value for Shields’ dimensionless shear
stress (Table 5.1). 0.045 is used as the value for Shields’
dimensionless shear stress in the initial model, but the
implications of uncertainty in this parameter are also considered.
In this case study the concern is not so much with
the correct value of this parameter as the effect of parameter
uncertainty.
Using a standard Wolman count (sample size = 100),
the median bed material sediment size is estimated as
30 mm. Using Shields’ entrainment function, incipient
motion for the grain size occurs at 22 N/m 2 . Seventeen
cross-sections are surveyed along a 2 km reach of the mid-
Goulburn river. A one-dimensional hydraulic model is
calibrated for these cross-sections using water levels surveyed
at a fl ow close to the mean. Calibration was achieved
by adjusting the Manning n roughness parameter. Assuming
Manning n is invariant with discharge, the hydraulic
model estimates that the mean shear stress for the reach
is equal to 22 N/m 2 at a fl ow of 360 m 3 /s.
To estimate the natural frequency of bed mobilisation
events the natural fl ow regime is needed, but there is no
record of the natural fl ow regime in the mid-Goulburn
River because it has been regulated since fl ow gauging
commenced. Instead a natural fl ow series is modelled
using available streamfl ow data upstream of Eildon Dam.
This requires estimation of fl ows from ungauged portions
of the catchment by scaling fl ows in the gauged catchments,
combining fl ows from the various sub-catchments
and routing fl ows to the study reach. Using this modelled
natural fl ow series, it is estimated that 157 GL/year is
required to mimic the natural frequency and duration of
fl ows exceeding 360 m 3 /s.
5.3.3 Analysis of Uncertainties
There are a number of sources of uncertainty in the estimated
channel fl ushing discharge and volume of water
required to mimic the natural frequency and duration of
channel fl ushing events (Table 5.1). In this study, uncertainty
in the estimated threshold discharge for sediment
fl ushing is assessed based on a consideration of:
Table 5.1 Summary of errors and uncertainties considered in this analysis (the text in italics shows the values used in the
analysis)
Source of Error Channel Hydraulics Critical Shear Flow Regime
Sample uncertainty Cross-section sampling Number of particles in sample Number of years in the record
(used n = 17)
(used n = 100)
(used n = 25)
Measurement error (Neglected here) Measurement of grain
diameter (± 5 mm)
Error in rating curve (r2 = 0.95)
Model error Manning n (Based on data Shields entrainment function* Estimating fl ow from ungauged
presented in Hicks and
catchments (20% of catchment
Mason, 1998)
area) and fl ow routing*
* see text for explanation.

70 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
• errors in measurement of grain sizes;
• sampling bed particles to estimate median grain size;
• estimating Shields’ dimensionless shear stress (τ*).
• assuming the channel roughness parameter (i.e. Manning
n) is constant with discharge;
• sampling of cross-sections from the 2 km reach.
Uncertainty in the volume of water to mimic the duration
and frequency of sediment fl ushing fl ows is assessed based
on a consideration of these same factors in addition to:
• errors in measuring discharge using a rating curve at
each gauge;
• using a sample of the fl ow record to represent the long
term fl ow regime;
• errors in modelling natural fl ows at the survey site.
In this section the effect of these uncertainties is quantifi
ed using a Monte Carlo Analysis, which involves running
the fl ushing fl ow model many times (1000 in this case)
with different, but equally plausible sets of input parameters
to generate a range of plausible model outputs (Manly,
1997). For each of the 1000 replicates values were randomly
chosen for each of the input parameters from the
range of possible values. These were selected from a probability
distribution centred on the best estimate used in the
original analysis (described in the previous section). The
combined uncertainty is estimated by combining random
selections of values for each input variable.
It is necessary to evaluate the uncertainty for each
input parameter so that it is known how it should be
varied in the replicate model runs. A highly uncertain
parameter should be varied over a bigger range than a more
accurately known parameter. The methods used to evaluate
parameter uncertainties are described below. In some
cases, these methods are obvious. For the case of the
median particle size, it is relatively straight forward to
express the estimate of the median grain size as a distribution
of possible values, based on the size of the sample of
bed grains. It is not so straightforward to evaluate uncertainties
in recorded discharge series or calibrated values
of the Manning channel roughness parameter (n). A best
effort is made to quantify these uncertainties, but these are
best described as models of uncertainty, albeit models
which cannot be truly tested. The result is a stochastic
model where some components are deterministic and some
are random.
Median Particle Size
The median grain size is estimated from a sample of 100
bed particles (a standard Wolman count). To estimate the
uncertainty associated with sampling grain sizes, 1000
replicate samples (each with 100 grain sizes) are synthetically
generated by assuming that the natural log of
grain sizes (in millimetres) are distributed normally with
a mean of 3.4 and standard deviation of 0.51. This gives
a true median particle size of 30 mm, although sample
medians for each replicate will vary about this value.
Errors in measurement of grain size diameter are modelled
as normally distributed with a mean of zero and standard
deviation of 3 mm. This gives 90% confi dence intervals
on grain size measurements of ±5 mm. To represent
measurement errors, each grain size in each of the 1000
replicate samples is perturbed by adding this random
component.
Entrainment Threshold
Buffi ngton and Montgomery (1997) collated values provided
in the literature for the Shields’ dimensionless shear
stress value for incipient motion. A value of 0.045 is often
used as the best estimate of τ* for gravel-bed rivers. The
uncertainty in τ* is estimated from the range of values
compiled by Buffi ngton and Montgomery that are likely
to occur in gravel-bed streams with Reynolds roughness
number in the range 25 to 1000 (Figure 5.3). The distribution
of residuals for the regression in Figure 5.3 was replicated
by assuming that the distribution was log-normally
distributed with a standard deviation in the log-error of
0.03, and randomly generating 1000 replicate values using
this error model.
Dimensionless shear stress
0.1
y = 0.017x 0.145
R 2 = 0.150
n = 98
0.01
10 100 1000
Reynolds roughness number
Figure 5.3 Values of Reynolds roughness number (for the range
25 to 1000) and dimensionless shear stress obtained from published
incipient motion studies and collated by Buffi ngton and
Montgomery (1997)

Conceptual and Mathematical Modelling in River Restoration: Do We Have Unreasonable Confi dence? 71
Survey Errors
Uncertainty associated with cross-section survey errors
were neglected in this study and likely to be small given
the high accuracy of survey equipment relative to other
sources of uncertainty. The survey of the Goulburn River
reach included a sample of 17 evenly-spaced crosssections
over a reach length of 2 km. There is uncertainty
in the estimate of reach-averaged bed shear stress depending
on how well the sampled cross-sections represent the
variability of the reach. To estimate this uncertainty, replicate
samples of 17 cross-sections were generated using
a bootstrap procedure (Manly, 1997) in which 17 crosssections
were randomly selected from the 17 available,
‘with replacement’. This bootstrap procedure was used to
generate 1000 replicate samples.
Errors in the Hydraulic Model
The critical shear stress was converted into a discharge
using a one-dimensional hydraulic model. The major
uncertainty in this method is the roughness coeffi cient. In
the model, Manning n was calculated from a single stage
discharge measurement in the reach. The hydraulic modeling
required the assumption that Manning n was then
constant over the range of discharges considered (up to
bankfull). It is well known that Manning n varies with
discharge, sometimes increasing and sometimes decreasing
with increasing fl ow. The error associated with this
assumption was estimated using the large range of roughness
estimates for gravel-bed New Zealand rivers (Hicks
and Mason, 1998). These data reveal that the inverse of
Manning roughness parameter ( 1 – n) is approximately proportional
to the log of discharge in most cases (Stewardson
and Anderson, 2002). A log regression was fi tted to data
for each of 72 sites [provided by Hicks and Mason, 1998]
to provide a constant (c) and coeffi cient (k) in this regression
equation for each site:
1
c 1 k
n
Q ⎛ ⎞
= ⎜ + ln − ⎟
(5.2)
⎝ Q ⎠
where Q – is the mean daily fl ow at the site. About one
quarter of the coeffi cients (k) were negative, indicating
increasing Manning n with discharge at these sites. The
parameter k defi nes variation in Manning n with discharge.
For our site, c was determined by calibration to the
observed water surface profi le (using k = 0). The change
in shear stress with discharge was then modelled using the
one-dimensional model, and run 1000 times, each time
using a different value of k selected at random from the
72 values for the New Zealand rivers.
Uncertainty in the Hydrology
Flows at the Goulburn River site are regulated by operation
of Lake Eildon, a large water supply reservoir,
upstream of the site. Flow data from gauging stations
located on nine unregulated tributaries upstream of the
Goulburn River site were used to model natural daily fl ows
at this site (i.e. fl ows that would occur in the absence of
Lake Eildon, neglecting routing effects). A 25 year series
of daily fl ows was generated for the survey site from the
25 year fl ow records at nine streamfl ow gauges. These
25 year series are a sample of the long term fl ow regime.
A bootstrap procedure was used to generate 1000 replicate
25-year series by sampling random years from the 25 year
sequences. Variability in the replicates represents uncertainty
in the long term fl ow regime. This approach assumes
fl ow independence between years.
Flow modelling for the study was based on records of
discharge at nine streamfl ow gauges on unregulated tributaries
of the Goulburn River. Discharge is estimated at
the gauges from rating curves. Rating curves are fi tted to
periodic measurements of discharge and stage. There are
errors in discharges provided by the rating curve as a
consequence of errors in these periodic measurements of
stage and discharge. Rating curves are often fi tted by a
log–log regression and an r2 of 0.95 is generally regarded
as a good fi t (Clarke, 1999). Error in discharges derived
from the rating curves are represented by a random component
added to the log-discharge where the error is distributed
normally with standard deviation of 0.05 times
the mean of log-discharges recorded at the gauge. This can
be represented by:
Q′= e
ln Q+ N(
0005) ∑ ln Q
n
1
, .
(5.3)
where Q is the recorded daily discharge, N(0,0.05) denotes
a normally distributed random variable with mean of zero
and standard deviation of 0.05, and n is the number of days
of recorded fl ow. This error would result in an r 2 of 0.95
for a rating curve fi tted by log–log regression.
The 25 year natural daily fl ow series at the study site
was modeled in two parts: (i) estimating fl ows from the
ungauged portion of the catchment by scaling gauged
fl ows in unregulated tributaries; and (ii) summing fl ows
recorded at the streamfl ow gauges and estimated for the
ungauged portions of the catchment. Routing effects,
which generally result in attenuation and delay of fl ood
pulses further downstream, were ignored because no data
were available to calibrate a routing model. Uncertainty
in the scaling parameter for the ungauged catchment
was evaluated using a comparison of fl ows recorded at
seven different gauges (see Appendix 5.1). There is some

72 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
uncertainty in the modelled natural fl ows as a consequence
of ignoring routing effects. This uncertainty was represented
using a routing model (described by Stewardson
and Cottingham, 2002), where the routing parameters
were randomly perturbed according to an assumed distribution
of possible values (see Appendix 5.1).
Results of Monte Carlo Analysis
The critical shear stress for incipient motion for this reach
has wide confi dence limits varying from 18 N/m2 , up to
26 N/m2 (Figure 5.4). A larger uncertainty applies to the
Shear stress (N/m 2 )
Flow volume (GL/year)
70
60
50
40
30
20
10
0
800
600
400
200
reach shear stress
critical shear stress
0 200 400 600 800
Flow (m 3 /s)
(a)
0
0 200 400 600 800
Flow threshold (m3 /s)
(b)
Figure 5.4 (a) reach-averaged shear stress estimated using a
one-dimensional hydraulic model and the critical shear stress
estimated using Shields, entrainment function; and (b) the volume
of artifi cial fl ow spells required to mimic the frequency and duration
of natural fl ow spells for varying fl ow threshold. (Dashed
lines indicate 90% confi dence intervals based on consideration of
all errors described in Table 5.1.)
possible range of shear stresses that occur in the reach for
a given discharge. The fl ushing fl ow is estimated to be
360 m 3 /s, but this has 90% confi dence limits of between
250 m 3 /s and well over 500 m 3 /s.
Uncertainty in the critical discharge for bed fl ushing is
reported using the magnitude of the 90% confi dence interval
(Table 5.2). Combining all sources of uncertainty, the
magnitude of this confi dence interval is 560 m 3 /s. The
major source of uncertainty in estimating the critical discharge
comes from error in the estimate of Manning n in
the hydraulic model. This is because n is calibrated to a
single discharge (effectively the mean discharge of 60 m 3 /
s) and becomes increasingly uncertain as discharge
increases. This provides a massive 430 m 3 /s range of discharge
(Table 5.2). By comparison, if the only source of
uncertainty in the estimate was error in measuring the
particle size, then the range in the estimate would only be
50 m 3 /s. The implication of Table 5.2 is that ever more
detailed estimates of the bed-particle size distribution, or
even increasingly sophisticated bed-load transport threshold
models, will not remove the major uncertainty associated
with the channel hydraulics.
In a good environmental fl ow project, a manager would
not simply specify a fl ow magnitude, the frequency and
duration of that critical fl ow would also be specifi ed. A
popular approach used to defi ne the frequency and duration
is the ‘natural fl ow paradigm’ (Poff et al., 1997;
Richter et al., 1997), which suggests that an environmental
fl ow should mimic the natural regime. To estimate this the
average duration and frequency of critical transport events
in the natural regime is calculated. The volume of water
required to mimic these events is the average duration
multiplied by the frequency, multiplied by the threshold
discharge. The curve in Figure 5.4(b) shows that, as the
fl ow threshold increases, the total fl ow required decreases
substantially. If, for example, the threshold fl ushing fl ow
Table 5.2 Size of 90% confi dence interval (expressed in m 3 /s)
for threshold discharge for incipient motion for different
sources of error (best estimate of threshold discharge is
360 m 3 /s)
Source of error Channel Hydraulics Critical Shear
Measurement error Neglected 50
Model error 430 180
Sample error 200 120
Combined 540 220
All sources combined 560
Note: The numbers are not additive. The combined uncertainty comes
from MonteCarlo simulations using the full distributions. This table
should be compared with Table 5.1.

Conceptual and Mathematical Modelling in River Restoration: Do We Have Unreasonable Confi dence? 73
Table 5.3 Size of 90% confi dence interval (in GL/year) for volume of discharge required to mimic natural frequency and duration
of bed scouring events for different sources of error (best estimate of volume is 157 GL/year)
Source of Error Channel Hydraulics Critical Shear Flow Regime
Measurement error neglected 37 20
Model error 234 167 13
Sample error 164 104 107
Combined 267 209 125
All sources combined 350
is 260 m 3 /s, then to generate the natural duration and frequency
of these fl ow pulses would require about 160 GL/
yr, but it could range between 10 GL/year and 360 GL/yr
(i.e. the 90th percentile confi dence interval).
The numbers in Table 5.3 are the confi dence intervals on
this volume of the fl ushing fl ow. For example, considering
all sources of error, the 90% confi dence limits are 10 GL/
year and 360 GL/year so the magnitude of the interval is
350 GL/year which is close to 10% of the mean annual
fl ow! Expressed another way, 360 GL have to be allocated
for the fl ow component in order to be 95% confi dent that
there is suffi cient water that the bed will be turned over at
a natural frequency and duration, given all of the sources of
uncertainty in the method. How much is this water worth?
One way to estimate this is to consider the cost of infrastructure
projects that would be required to achieve comparable
water savings (pipelines etc.). The Snowy Water
Inquiry (1998a,b) estimated that environmental fl ows,
costed in this way, were worth $US 0.45 million per GL/
year. Clearly this uncertainty has economic signifi cance.
5.3.4 Discussion
In this simple analysis three other sources of uncertainty
have not been considered. The fi rst is the simplifi cation
of the fl ushing fl ow problem to a single threshold shear
stress. The range of τ* used accounts for many of the
well-known problems of hiding, packing and other particle
interactions that affect bed material transport, but there is
still the issue of hysteresis. The same discharge on the
rising and falling limb of the hydrograph has very different
transporting capacity (Gomez, 1991). It is also possible
that a fl ow exceeding the threshold for incipient motion
will be required to mobilize bulk quantities of the bed
sediments.
The second uncertainty not considered is the longitudinal
variation in the effect of the fl ushing fl ow. In this case
a threshold fl ow in a single reach has been estimated. The
target fl ow may well turn the bed over at the sample site,
but what will the fl ow do up and downstream of that site?
Given storage effects, a larger fl ow might have to be
released to turn the bed over further downstream. The
result might be that the upstream site is not just ‘turnedover’
but it is progressively scoured away. This introduces
the third uncertainty which is the sustainability of the
process target.
The fl ushing fl ow may work the fi rst time, but the bed
will then presumably progressively armour, as there may
be no source of coarse sediment below the dam. Either the
fl ushing fl ow will have to be progressively increased over
time, or the idea of a fl ushing fl ow is simply not sustainable
in this situation. Such analyses are completed as
steady-state models when in reality the release of environmental
fl ows will lead to changes in the input variables.
An example of this problem is the major fl ushing-fl ow
experiment carried out on the Colorado River (Collier
et al., 1997). The fl ood was successful in scouring and
rejuvenating the point-bars of the river below the Glen
Canyon Dam but this does not mean that it can be successful
indefi nitely.
Major sources of uncertainty have been considered in a
reasonably simple geomorphic problem: a fl ushing fl ow.
Most discussion of uncertainty in this type of analysis is
associated with the entrainment function, or in the measurement
of the particle size (Kondolf and Wilcock, 1996).
It is interesting to note the many sources of uncertainty
(and error) in such analyses. In fact, for this example, the
main uncertainty comes from the hydraulics, particularly
the estimation of roughness.
It is pertinent to ask what options are available to reduce
the uncertainty? For a geomorphologist the response may
be to argue for the use of a more sophisticated bed-load
threshold function, but this will not reduce the uncertainty
associated with the hydraulics and hydrology. The geomorphologist
could also argue that more research would
reduce uncertainty. This may be true, but it is also fair to
say that 150 years of research into bed-load transport
reinforces the view that there is unlikely to be a simple
principle that will dramatically change our capacity to
predict this process. Instead the best option is to make
extra measurements at the site. The geomorphologists
could either measure variables that reduce the range of

74 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
uncertainty in each of the modelled variables discussed
above, or they could directly measure bed-load movement
to observe the discharge at which the bed turns over.
Reducing Model Uncertainty
Extra fi eld measurements can reduce the uncertainty and
so the amount of water required for the fl ushing fl ow. The
options include:
• extra cross-section surveys;
• recording stages along the reach over the range of fl ows
being considered to avoid the need to extrapolate
Manning n from calibration at a single discharge; or
• larger bed material samples and more careful measurement
of particle diameter.
Most of the error in this example lies in estimates
of the channel hydraulics (Table 5.4). The most effective
way to reduce uncertainty in these modelled estimates
is to survey water levels over a range of discharges to
allow calibration of Manning n for a range of fl ows. Some
uncertainty would remain for extrapolations outside this
range of discharges. Similarly, surveying more cross-sections
will produce a greater decrease in error than will a
more accurate estimate of the entrainment function.
Overall, the estimates are insensitive to uncertainties in
the fl ow regime, such as an increase in the length of fl ow
record, or improvements in the accuracy of the rating
curves.
Another way to reduce uncertainty (and the waste of
water) is not to mimic a ‘natural’ fl ow regime. The uncertainty
(and cost) that this can introduce must be appreciated.
A reductionist approach is to understand the target
processes, and release fl ows when required, to achieve this
goal. In the case of a fl ushing fl ow, it would be better to
monitor the bed processes so that managers knew when
the bed needed to be fl ushed (say after a particular series
of lower fl ows). Thus, the cost of lost fl ow would be
replaced by the cost of monitoring the bed.
A clear conclusion from this analysis is that a few extra
measurements can dramatically reduce uncertainty. In this
particular case, it may be cheaper and more accurate to
simply observe the processes directly, rather than rely on
modelling. The most effi cient way to achieve this is by
trial releases of water from dams (in the fl ushing-fl ow
case). Dam managers may argue that the water is too
valuable to ‘waste’ on such exercises. However, at least
in our example, far more water (and hence money) could
be wasted as a consequence of the uncertainty.
However, river restoration projects are often carried out
under tight time constraints for political or economic
reasons. There is rarely time to monitor the processes that
underpin the modelling or to check the predictions. In a
typical stream rehabilitation project, a consultant is
engaged and given perhaps a month or two to report. Any
fi eld investigations are expected to be brief inspections to
develop a conceptual model.
5.4 CONCLUDING DISCUSSION
A central contribution of geomorphology to the new practice
of stream restoration is in developing conceptual
models that describe the change of stream form and
process over time. This exercise also provides a ‘reference’
condition that can be the target for restoration
actions. If the conceptual model is wrong, or is too simplistic,
then it does not matter how well executed the
management actions are, the project outcomes remain
highly uncertain.
In this chapter the development of the conceptual geomorphic
model for the restoration of the lower Snowy
River has been described. This is an interesting example
because it illustrates the tension between the typically
iterative process of geomorphic discovery and the need
for action by managers. It also shows the dangers of too
much confi dence in early interpretations, when these can
trigger expensive management actions that are diffi cult to
reverse even with adaptive management (Stankey et al.,
2003).
Restoration can be a very expensive exercise but there
is often little basis for managers to assess how much confi
dence they should have in conceptual models that are
presented to them. Five methods have been proposed that
Table 5.4 Size of 90% confi dence interval (expressed in GL/year) for volume of discharge required to mimic natural frequency
and duration of bed scouring events with one source of uncertainty removed (e.g. if completely certain about the hydraulic
roughness of the channel, then the confi dence interval would be reduced from 350 GL/yr down to 280 GL/year, as shown in the
bottom left cell of the table)
Source of Error Omitted from Analysis Channel Hydraulics Critical Shear Flow Regime
Sample error Cross-sections 280 No. of particles 330 Years of record 340
Measurement error – Grain size 350 Rating curve 350
Model error Manning n 270 Shields entrainment 320 Extrapolating from gauged
catchments 350

Conceptual and Mathematical Modelling in River Restoration: Do We Have Unreasonable Confi dence? 75
can be used to test the validity of conceptual models that
often fall outside the area of peer review: independent
review of the model; establishing simple guidelines for
evaluating the uncertainty of a conceptual model based on
the type of strength of evidence provided; ensuring that
an appropriate percentage of the total cost of the project
is committed to the development of the conceptual model;
holding back some of the information and data that underpin
the model, and using these data to verify the hypothetical
model; and passing all of the data and information to
a third party, without interpretation, so that a competing
model can be developed.
The feasibility has been demonstrated of quantifying
uncertainty in a geomorphic design model, using the case
study of a fl ow to fl ush the fi ne sediments from the bed of
the Goulburn River. In this case, the uncertainty was very
large. Surprisingly, the main source of uncertainty came
from the hydraulic modelling and relatively little from
either the bed-load entrainment function or the hydrological
modelling of the fl ow series. It is also striking how
adding extra fi eld measurements can reduce the uncertainty
in the model estimates, particularly in estimating
Manning n. In fact, the key to uncertainty often comes
down to modelling versus monitoring.
It is necessary to include some form of data gathering
with any geomorphic modelling exercises, including
modelling geomorphic response to river restoration. It has
been shown how an uncertainty analysis might be used to
direct this data gathering effort to most effectively reduce
uncertainty in model predictions. It seems logical that
modelling and data gathering should be integrated to minimise
uncertainties in restoration. In many cases, some
form of monitoring is carried out as part of the restoration
implementation to check that the project achieves what
was intended. However, these monitoring programs are
rarely designed to verify or improve the models used in
planning the restoration. Uncertainty analysis appears to
be useful for optimising such monitoring programs to
provide improved models for subsequent restoration
decisions.
In some cases the most effective way to reduce uncertainties
is to run trial programs in the actual systems being
restored. Even in these cases, the model and uncertainty
analysis can provide the basis for designing the trial to
ensure that it tackles the key sources of uncertainty in
model predictions. In the case of the lower Snowy River,
the state government has recognised the value of additional
geomorphic measurements and analysis, and
invested in a restoration trial which includes fi eld measurements
and physical modelling to improve the geomorphic
design. It is possible to design monitoring activities
without regard to uncertainties during the planning phase
of a restoration project. However, it is proposed that
stronger integration of monitoring and modelling activities
will lead to greater improvements in the knowledge
underpinning river restoration.
Geomorphic models, like models developed in any
science, are subject to some uncertainty. Although this
uncertainty is widely acknowledged it is rarely evaluated.
We conclude that there is unreasonable confi dence in
geomorphic models used for river restoration. Certainly
we were surprised by the magnitude of uncertainty in the
fl ushing fl ow example and we have been involved in many
of these modelling studies in Australia. There is rarely any
thought given to the best approach to modelling, including
the calculation of input parameters to minimise uncertainties
in restoration decisions. There is a need to rethink our
approach to geomorphic modelling in the context of river
restoration, in particular the benefi ts of evaluating uncertainties
in both our conceptual and mathematical models.
This requires more expertise, information and funding
but the result will be more realistic expectations by those
involved in the restoration project and a more careful
approach to optimising data gathering for calculating
model parameters and verifying model structure.
REFERENCES
Beven K. 2001. Rainfall–Runoff Modelling: The Primer. John
Wiley & Sons Ltd: Chichester.
Brizga SO, Finlayson BL. 1992. The Snowy River Sediment
Study: Investigation into the Distribution, Transport and
Sources of Sand in the Snowy River between Lake Jindabyne
and Jarrahmond. Department of Water Resources: Melbourne,
Victoria, Australia.
Brizga SO, Finlayson BL. 1994. Interactions between upland
catchment and lowland rivers: an applied Australian case
study. Geomorphology 9: 189–201.
Brookes A. 1995. Challenges and objectives for geomorphology
in UK river management. Earth Surface Processes and Landforms
20: 593–610.
Brookes A, Sear DA. 1996. Geomorphological principles for
restoring channels. In: Brookes A, Shields FD (Eds), River
Channel Restoration: Guiding Principles for Sustainable
Projects. John Wiley & Sons Ltd: Chichester; 75–101.
Brooks AP, Brierley GJ, Millar RG. 2003. The long-term control
of vegetation and woody debris on channel and fl oodplain
evolution: insights from a paired catchment study in southeastern
Australia. Geomorphology 51: 7–29.
Brooks AP, Brierley GJ. 1997. Geomorphic responses of Lower
Bega River to catchment disturbance, 1851–1926. Geomorphology
18: 291–304.
Buffi ngton JM, Montgomery DR. 1997. A systematic analysis of
eight decades of incipient motion studies, with special reference
to gravel-bedded rivers. Water Resources Research 33
(8): 1993–2029.
Clarke RT. 1999. Uncertainty in the estimation of mean annual
fl ood due to rating curve indefi nition. Journal of Hydrology
222: 185–190.

76 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Collier MP, Webb RH, Andrews ED. 1997. Experimental fl ooding
in Grand Canyon. Scientifi c American 276: 82–89.
Cooke RV, Reeves RW. 1976. Arroyos and Environmental
Change in the American Southwest. Oxford Research Studies
in Geography, Oxford.
Cullen P. 1989. The Turbulent Boundary Between Science and
Management. Canberra College of Advanced Education:
Canberra, Australia.
DNRE 2002. The FLOWS method: A method for determining
environmental water requirements in Victoria Victorian
Department of Natural Resources and Environment,
Australia.
Erskine WD, Terrazzolo N, Warner RF. 1999. River restoration
from the hydrogeomorphic impacts of a large hydro-electric
power project: Snowy River, Australia. Regulated Rivers 15:
3–24.
Erskine WD, Tilleard JW. 1997. Formative processes of alternating,
bank-attached side bars and associated pool-riffl e
sequences on sand-bed streams similar to the Snowy River at
Jarrahmond, Victoria. Department of Natural Resources and
Environment: Melbourne, Victoria, Australia.
Finlayson BL, Bird JF. 1989. Initial investigation into the extent
and nature of the current sedimentation problem on the lower
Snowy River. Centre for Environmental Applied Hydrology,
The University of Melbourne: Melbourne, Australia.
Gilbert GK. 1917. Hydraulic mining debris in the Sierra Nevada.
United States Geological Survey Professional Paper 105:
1–154.
Gilvear DJ. 1999. Fluvial geomorphology and river engineering:
future roles utilizing a fl uvial hydrosystems framework. Geomorphology
31: 229–245.
Gippel CJ. 2002. The Victorian Snowy River: Review of Historical
Environmental Change and Proposed Restoration Options.
East Gippsland Catchment Management Authority: Victoria,
Australia.
Gippel CJ, Anderson BA, Marsh N. 2002. Trial of Snowy River
Restoration Concept Plan: Scoping Study to Review Struc -
tural Restoration Options. Report by Fluvial Systems
Pty Ltd, Stockton, to Snowy River Restoration Project, East
Gippsland Catchment Management Authority: Victoria,
Australia.
Gomez B. 1991. Bedload transport. Earth-Science Reviews 31:
89–132.
Hicks DM, Mason PD. 1998. Roughness characteristics of New
Zealand Rivers, National Institute of Water and Atmospheric
Research, Water Resource Publications: Englewood, New
Zealand.
ID&A 1998. River restoration concept plan for the Snowy River
in Victoria. Wangaratta, Victoria, Consultants report to East
Gippsland Catchment Management Authority: Victoria,
Australia.
Kondolf GM. 2000. Some suggested guidelines for geomorphic
aspects of anadromous salmonid habitat restoration proposals.
Restoration Ecology 8 (1): 48–56.
Kondolf GM, Micheli ER. 1995. Evaluating stream restoration
projects. Environmental Management 19 (1): 1–15.
Kondolf GM, Smeltzer MW, Railsback SF. 2001. Design and
performance of a channel reconstruction project in a coastal
California gravel-bed stream. Environmental Management 28
(6): 761–776.
Kondolf GM, Wilcock PR. 1996. The fl ushing fl ow problem:
defi ning and evaluating objectives. Water Resources Research
32 (8): 2589–2599.
Kuhnle RA, Alonso CV, Shields FD. 2002. Local scour associated
with angled spur dikes. Journal of Hydraulic Engineering
128 (12): 1087–1093.
Ladson AR, Argent RM. 2002. Adaptive management of environmental
fl ows: lessons for the Murray-Darling Basin from
three large North American Rivers. Australian Journal of
Water Resources 5 (1): 89–102
Larson M, Goldsmith W. 1997. Incised channel stabilization and
enhancement integrating geomorphology and bioengineering.
In: Wang SSY, Langendoen EJ, Shields FD (Eds), Management
of Landscapes Disturbed by Channel Incision, University
of Mississippi: Oxford, Mississippi; 458–465.
Manly BFJ. 1997. Randomization, Bootstrap and Monte Carlo
Methods in Biology. Chapman and Hall: London.
Milhous RT. 1995. Flushing fl ows for habitat restoration. In:
Espey W (Ed) Water Resources Engineering, American
Society of Civil Engineers: San Antonio, Texas; 663–667.
Miller JR, Ritter JB. 1996. An examination of the Rosgen classifi
cation of natural rivers. Catena 27: 295–299.
Neilsen MB. 1996. Lowland stream restoration in Denmark. In:
Brookes A, Shields FD (Eds), River Channel Restoration:
Guiding Principles for Sustainable Projects. John Wiley &
Sons Ltd: Chichester; 269–290.
Newbury R, Gaboury M. 1993. Stream analysis and fi sh habitat
design – a fi eld manual. Newbury Hydraulics Ltd: British
Columbia, Canada.
O’Neill MP, Kuhns MR. 1994. Stream bank erosion and fl ushing
fl ows. Stream Notes (July), USDA Forest Service.
Owen R. 1997. The Lower Snowy River Revegetation Strategy.
Snowy River Improvement Trust: Nungurner Hills Nursery,
Victoria, Australia.
Poff NL, Allan JD, Bain MB et al. 1997. The natural fl ow regime,
a paradigm for river conservation and restoration. BioScience
47: 769–84.
Raadick TA, O’Connor JP. 1997. Fish and Decapod Crustacean
Survey, and Habitat Assessment, of the Lower Snowy River,
Victoria. Report to the Snowy River Improvement Trust by the
Freshwater Ecology Division of the Department of Natural
Resources.
Regan HM, Colyvan M, Burgman M. 2002. A taxonomy of
uncertainty for ecology and conservation biology. Ecological
Applications 12 (2): 618–628.
Richter BD, Baumgartner JV, Wigington R, Braun DP. 1997.
How much water does a river need? Freshwater Biology 37:
231–249.
Rosgen DL. 1996. Applied River Morphology. Wildland Hydrology:
Pagosa Springs, Colorado.
Rutherfurd ID. 2001. Storage and movement of slugs of sand in
a large catchment: developing a plan to rehabilitate the Glenelg

Conceptual and Mathematical Modelling in River Restoration: Do We Have Unreasonable Confi dence? 77
River, SE Australia. In: Anthony DJ, Harvey MD, Laronne JB,
Mosley MP (Eds), Applying Geomorphology to Environmental
Management. Water Resources Publications: Denver, Colorado;
309–332.
Schumm SA, Harvey MD, Watson CC. 1984. Incised Channels:
Morphology, Dynamics and Control. Water Resources Publications:
Littleton, Colorado.
Sear DA. 1993. Fine sediment infi ltration into gravel spawning
beds within a regulated river experiencing fl oods: Ecological
implications for salmonids. Regulated Rivers 8: 373–390.
Sear DA. 1994. River restoration and geomorphology. Aquatic
Conservation: Marine and Freshwater Ecosystems 4:
169–177.
Seddon GS. 1994. Searching for the Snowy, an Environmental
History. Allen and Unwin: St Leonards, New South Wales.
Simon A. 1989. A model of channel response in disturbed
alluvial channels. Earth Surface Processes and Landforms 14:
11–26.
Snowy Water Inquiry 1998a. Snowy Water Inquiry: Draft Options
for Discussion. Snowy Water Inquiry: Sydney, NSW,
Australia.
Snowy Water Inquiry1998b. Appendix of Resource Materials
(Part 2). Snowy Water Inquiry: Sydney, NSW, Australia.
Stankey GH, Bormann, BT, Ryan C. 2003. Adaptive management
and the northwest forest plan: Rhetoric and reality.
Journal of Forestry 40: 40–46.
Stewardson MJ. 1998. Pool formation – fl uvial processes, Section
6 in ID&A, River restoration concept plan for the Snowy River
in Victoria. Report to East Gippsland Catchment Management
Authority by ID&A: Wangaratta, Victoria, Australia.
Stewardson MJ, Anderson B. 2002. Variations in the fl ow resistance
of natural channels with discharge. Proceedings of the
Hydrology and Water Resources Symposium. Institution of
Engineers: Melbourne, Australia.
Stewardson MJ, Cottingham P. 2002. A demonstration of the
fl ow events method: Environmental fl ow requirements of the
Broken River. Australian Journal of Water Resources 5 (1):
35–48.
Strom HG. 1936. The Flood Problem in Gippsland, 2. The Snowy
River. State Rivers and Water Supply Commission: Melbourne,
Victoria, Australia.
Volkman JM. 1999. How do you learn from a river? Managing
uncertainty in species conservation policy. Washington Law
Review 74: 719–762.
Walters C. 1997. Challeges in adaptive management of riparian
and coastal ecosystems. Conservation Ecology [online] URL
http://www.consecol.org/vol1/iss2/art1.
APPENDIX 5.1
Flows at the Goulburn River site are regulated by operation
of Lake Eildon, a large water supply reservoir, upstream of
the study site. Flow data for gauging stations located on
nine unregulated tributaries upstream of the Goulburn
River site were used to model natural daily fl ows at this site
(i.e. fl ows that would occur in the absence of Lake Eildon).
Seven of these tributaries are upstream of Lake Eildon and
two other tributaries have confl uences with the Goulburn
River between Lake Eildon and the survey site (Table 5.5).
In total 80% of the 4225 km2 catchment at the survey site
is upstream of these fl ow gauges (510 km2 upstream Eildon
and 314 km2 between Eildon and the study site). Flows in
the ungauged portion of the catchments upstream and
downstream of Lake Eildon are estimated by scaling fl ow
in the gauged portion of the catchments upstream and
downstream of Lake Eildon respectively. The scaling
factor was estimated using a linear function of the ratio of
gauged and ungauged catchment areas, derived from the
available streamfl ow data (Figure 5.5). Daily fl ows at
the survey site are estimated as the sum of daily fl ows (for
the same day) at the upstream gauges and estimated in the
ungauged catchments. Routing effects (i.e. travel times
and attenuation of fl ood peaks) are neglected because there
Table 5.5 Gauged tributaries of the Goulburn River upstream of the Goulburn River
survey site
Tributary Catchment area (km 2 )
Upstream of Lake Eildon
Delatite River 368
Howqua River 368
Jamison River 368
Goulburn River at Dohertys Infl ow 694
Big River 619
Ford Creek 115
Brankeet Creek 121
Between Lake Eildon and the surveys site
Rubicon River 129
Acheron River 619

78 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
is no information with which to establish a routing
model.
The regression equation in Figure 5.5 is used to estimate
a scaling factor for estimating fl ows in the ungauged catchments.
Data points in this plot show the mean ration of fl ow
Statistic of daily flow ratio
0.5
0.45
0.4
0.35
0.3
0.25
0.2
0.15
0.1
0.05
0
y = 1.33x - 0.05
R 2 = 0.87
Table 5.6 Characteristics of the ratio of daily fl ows at each of seven streamfl ow gauges to the sum of daily fl ows at the other six
gauges (Gauges are all located upstream of Lake Eildon)
Gauge Catchment
Area (km 2 )
y = 0.35x - 0.0037
R 2 = 0.83
0 0.1 0.2 0.3 0.4
Ratio catchment areas
mean of daily flow ratio
standard dev. of daily flow ratio
Figure 5.5 Mean and standard deviation of daily fl ows ratio and
catchment area ratios for seven streamfl ow gauges upstream of
Lake Eildon. Ratios are calculated at each gauge in turn by dividing
daily fl ows and catchment area at the gauge by the sum of
daily fl ows and catchment areas at the other six gauges
Mean Flow
(ML/day)
Catchment
Area Ratio
Mean of
Daily Flow
Ratio
Standard
Deviation of Daily
Flow Ratio
Characteristics for the streamfl ow gauges upstream of Lake Eildon
Delatite River 368 297 0.16 0.10 0.037 0.66
Howqua River 368 477 0.16 0.18 0.051 0.94
Jamison River 368 560 0.16 0.21 0.056 0.91
Goulburn River 694 882 0.35 0.35 0.097 0.93
Big River 619 843 0.30 0.46 0.14 0.93
Ford Creek 115 34 0.045 0.0049 0.011 0.44
Brankeet Creek 121 48 0.048 0.022 0.018 0.56
Characteristics estimated for the ungauged catchments using regression equation in Figure 5.5
Upstream of Eildon 510 – 0.19 0.21 0.063 0.9*
Eildon to study site 314 – 0.42 0.51 0.14 0.9*
* selected based on serial correlations derived from streamfl ow gauge data
at one of the gauges to sum of fl ow at all other gauges. In
reality this ratio varies from day to day. The mean, standard
deviation and lag-1 serial correlation for this ratio
were calculated for each of the seven sites upstream of
Lake Eildon (Table 5.6). These characteristics were estimated
for the ungauged catchments using the regression
equation in Figure 5.5. A stochastic model is used to represent
uncertainty in the fl ow estimated in the ungauged
catchment. This model gives fl ow from the ungauged
catchment on the ith day of the 25 year record as:
Q = [ mr . + ( 1−m) r 1] Q and r = N(
ms , ) (5.4)
ui , i i− ∑
j
ji ,
i
where Qj,i is the fl ow at the jth gauge site on the ith day.
Values of m were calibrated as 0.1 to provide a lag-1 serial
correlation of 0.9 for the fl ow ratios in ungauged catchments
upstream and downstream of Lake Eildon. The
daily parameter r i is a normally distributed random variable.
The mean (µ) of r i was set equal to the mean of daily
fl ow ratios estimated from the regression in Figure 5.5 The
standard deviation (σ) was adjusted to replicate the standard
deviations calculated from the regressions in Figure
5.5. Standard deviations chosen for upstream and downstream
of Lake Eildon were 0.27 and 0.61 respectively.
Note that these are higher than the values estimated
directly from the regression equations to account for the
effect of lag correlation in the stochastic model. This stochastic
model is used to generate replicate 25-year timeseries
of fl ow from the ungauged catchments.
Lag-1 Serial Correlation
of Daily Flow Ratio

6.1 INTRODUCTION: THE CASE
FOR RESTORATION
6
Uncertainty in Riparian and
Floodplain Restoration
Francine M.R. Hughes 1 , Timothy Moss 2 and Keith S. Richards 3
1 Department of Life Sciences, Anglia Ruskin University, UK
2 Institute for Regional Development and Structural Planning (IRS), Germany
3 Department of Geography, University of Cambridge, UK
The riparian and fl oodplain zones of rivers are physically
dynamic places, subject to the delivery and removal of
water and sediments during fl ood events. Ecosystems that
occupy these places have evolved a tolerance to these
natural disturbance processes and many of their component
species have become dependent on them for completion
of their life cycles. In many river valleys, riparian and
fl oodplain zones would once have been occupied by forested
ecosystems, composed of a dynamic mosaic of
forest types in different successional stages, interspersed
with more open wetland communities with emergent vegetation.
Such fl oodplain forests have high levels of biodiversity
because they are at the interface between terrestrial
and lotic ecosystems (Petts, 1990). In addition, they experience
frequent disturbance from fl oods which create the
conditions for a heterogeneous mosaic of habitats across
the fl oodplain, each supporting a varied mix of species
(Nilsson et al., 1991a; Hughes et al., 2005). A high plant
species diversity has been recorded on fl oodplains from
rivers in many different bioclimatic zones. For example,
fl oodplains in the Amazon basin account for 20% of tree
species diversity (Junk et al., 1989). In the Tana River
fl oodplain forests in Kenya, which stretch for only 200 km
of river length and average only 1 km in width, 175 woody
plant species, over 250 species of birds and at least 57
species of mammals have been found, including two
endemic primates (Medley and Hughes, 1996).
Riparian and fl oodplain ecosystems not only have
intrinsic and intangible values associated with these high
levels of biodiversity, and with their diverse landscape
character, but they can also be valued because they contribute
to timber production and carbon sequestration,
fl oodwater storage, groundwater recharge, pollution
control and even recreation. The importance of riparian
and fl oodplain zones can thus lie in this wide range of
natural functions and services that they provide, although
there is uncertainty about how these can be valued over
time, given their dynamic and varying nature. Such valuation
is today increasingly necessary, since hydrological
pathways in river basins have often been altered indirectly
through land-use change and directly through management
of the fl ow regime. In downstream fl oodplain zones
in particular there have been many engineered changes to
river channels, leading to isolation of fl oodplains from
their channels and to severe damage or eradication of
fl oodplain ecosystems. All these changes have severely
limited or even destroyed the capacity of fl oodplains to
deliver their natural functions and services.
The disappearance of these ecosystems from the landscape
is poignantly illustrated by the case of fl oodplain
forests in Europe. 90% of these forested ecosystems have
disappeared and remaining patches are often in critical
condition. They are listed in Annexe I of the European
Habitats Directive (92/43/EEC, 1992) as a priority habitat
type and are included in the Natura 2000 network of nature
reserves. In western Europe they are more reduced in
extent than in eastern and central Europe where some
River Restoration: Managing the Uncertainty in Restoring Physical Habitat Edited by Stephen Darby and David Sear
© 2008 John Wiley & Sons, Ltd

80 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Figure 6.1 Map of remaining European fl oodplain forests (based on data from UNEP – World Conservation Monitoring Centre in
UNEP–WCMC, 2000 and Girel et al., 2003)
impressive patches remain (Figure 6.1). Even here, their
extent is not easy to gauge as many of the areas marked
as fl oodplain forest have in fact been converted to areas
of forestry on fl oodplains, without any of the characteristic
dynamic features of a naturally functioning fl oodplain
forest and frequently dominated by non-native species.
For example, in Hungary, where fl ood control works have
reduced the fl oodplain area across all river systems from
2.3 billion km 2 to only 1500 km 2 , 40% of the remaining
areas of fl oodplain forests have been converted to forestry
plantations (Haraszthy, 2001).
6.2 POLICY-RELATED WINDOWS
OF OPPORTUNITY
However, since the mid-1990s, shifts in policy content and
style in the fi elds of fl ood protection, nature conservation
and agriculture at European Union (EU) and national
levels are creating ‘windows of opportunity’ for fl oodplain
restoration (Table 6.1). In the fi eld of fl ood protection,
recent major fl ood events in France, Germany and the
United Kingdom, for instance, have accelerated the willingness
of authorities to entertain catchment-oriented
approaches and soft-engineering techniques of fl ood pro-
tection, creating new opportunities for fl oodplain restoration.
The sheer cost of improving and maintaining physical
fl ood defences, in particular in rural areas, is raising interest
in alternative strategies. These alternative, integrated
fl ood management strategies developed at the catchment
scale are now considered viable to provide appropriate
‘standards of service’ in areas of particular risk, while also
ensuring no net loss of ecosystem status, and even allowing
enhancement of aquatic, riparian and fl oodplain environments.
The evaluation of such strategies refl ects a
policy shift from fl ood defence to fl ood risk management,
and includes the possibility of increasing the frequency of
fl ooding and reducing the standard of service in some
fl oodplain locations where land is of relatively low value.
This can have the effect of storing fl oodwater and attenuating
hydrograph peaks, reducing fl ood potential in downstream
high-value urban fl oodplains that are otherwise at
risk.
Water protection agencies, concerned at water shortages
and motivated by the EU Water Framework Directive,
are also showing increased interest in water fl ow regimes
across whole catchments and in the potential of fl oodplains
to improve water quality as part of a policy shift
from downstream protection to upstream river basin man-

Table 6.1 Recent policy shifts conducive to fl oodplain restoration
Policy fi eld Forces for change Policy issues
agement. For nature conservationists restored fl oodplains
represent important habitats that can contribute to meeting
biodiversity targets in accordance with the EU Habitats
and Birds Directives; here the policy shift is in part from
species protection to habitat enhancement. Political pressure
is growing for more environmentally-sensitive forms
of agriculture and forestry, creating new funding opportunities
for extensive practices more suited to fl oodplain
restoration, in a policy shift from agricultural support to
integrated rural development. Agricultural policy focused
on agri-environmental management means that fl oodplains
originally expensively drained and protected can
now be considered sites for reinstating other functions
whose relative values are perceived to have increased.
Finally, land-use planning regulations are being modifi ed
to offer more effective protection of existing fl oodplains
and, in some instances, earmarking land for the future
restoration of fl oodplains. Spanning these sectoral policy
shifts is a trend towards greater policy integration and
stakeholder participation over schemes of this kind (Pahl-
Wostl, 2002, 2004), informed in part by debates on sustainable
development and new forms of governance
(Bressers and Kuks, 2003).
Managing these changes is a challenge for the responsible
agencies, which must grapple with questions of
multi-functionality, multiple and nested scales, crosssectoral
activity, policy interplay, actor collaboration and
issues of considerable complexity. One institutional framework
for managing the changes in values, and the associated
restoration of natural functions in fl oodplains, is
provided by the Water Framework Directive in Europe,
and by the tools developed for its implementation – River
Basin Management Plans. In the United Kingdom, existing
tools for strategic fl ood management (such as Catchment
Flood Management Plans) now need adaptation to
allow for ecosystem enhancement, combining the Envi-
Uncertainty in Riparian and Floodplain Restoration 81
Flood protection Flooding events; climate change;
Risk management, soft engineering techniques,
infrastructure costs; environmental quality natural fl ood storage
Water protection EU Water Framework Directive; water
Catchment-oriented approaches, fl ow regimes,
quality/quantity problems
wetlands, geomorphology
Nature conservation EU Habitats Directive; concerns for
biodiversity
Functional fl oodplain ecosystems
Land-use planning Linkage of fl ooding events to land use Planning mechanisms for protecting and creating
areas for fl ood retention
Rural development EU Rural Development Regulation; spatial Integrated approaches to rural economic
disparities
development
Agriculture Agenda 2000; public health concerns;
environmental degradation
Improved agri-environmental schemes
ronment Agency’s roles in relation to fl ood management
and conservation (and implementation of the EU Habitats
Directive). These tools will thus necessarily focus on
multi-functional management across a catchment-reachhabitat
scale hierarchy, which will imply a variation in the
level of detailed planning at different scales. At the catchment
scale, the planning is essentially a strategic assessment
for managing priorities in relation to budgetary
provision, while at the reach scale planning is focused on
the implementation of specifi c policies. This has implications
for the meaning of uncertainty, since qualitative
general goals set at the catchment strategic scale may
allow greater fl exibility, while more restrictive quantitative
goals may be set at the reach scale. However, underlying
this is the governance requirement for stakeholder involvement,
exposing the question of the accountability, in democratic
societies, of those institutions responsible for
goal-setting in relation to restoration initiatives. This introduces
another level of uncertainty.
Decision makers must thus manage complexity and
uncertainty, and cope in an adaptive manner with unintended
negative effects which emerge alongside the advantages
derived from the policies they enact and implement.
In the context of fl oodplain restoration, they may seek to
identify ‘reference’ conditions towards which the Water
Framework Directive encourages a shift; however, there is
considerable uncertainty in the defi nition of such reference,
or target, conditions. This ‘goal’ of a restoration
initiative may suggest identifi cation of a set of performance
indicators, but here again there is uncertainty. If the
strategic goal of restoration is to recover the dynamics of
natural process and function (by letting the river do the
work again), the performance criteria are very different,
and necessarily less rigid given the time-variability of
fl ows, than if the goal is set as a restoration of a specifi c
(static) form.

82 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
There has thus been a growing interest in restoration of
riparian and fl oodplain ecosystems, driven by increasing
knowledge of the biophysical linkages between parts of
river basins and, in Europe, by shifting policy directions
and an increased frequency and severity of fl ood events.
Awareness of the potential impacts of global climate
change and of the impacts of river management activities
such as dam building on hydrological patterns in river
basins (Montgomery and Boulton, 2003; World Commission
on Dams, 2000) have led to a broadening of approaches
to fl ood management. Coupled with a wider acceptance of
multi-functional fl oodplains and more integrated policy
contexts for these, this has led to the proliferation of river
restoration projects in many countries (Palmer et al., 2005).
However, river restoration that involves the planned return
of fl ooding and its associated geomorphological processes
also reduces the predictability of river behaviour for river
managers. The uncertainty associated with this development
is the subject of this chapter, which continues with a
consideration of the spatial and temporal dynamics that
underpin the natural function of rivers and their riparian
and fl oodplain zones.
6.3 THE NATURAL FUNCTIONAL DYNAMICS
OF RIPARIAN ENVIRONMENTS
To achieve the goal of integrated management of fl oodplains,
including restoration of their physical and ecological
functions, it is necessary to understand some of the
key relationships that determine the health of the aquatic,
riparian and fl oodplain ecosystems. This section accordingly
briefl y reviews this understanding, focusing particularly
on scale and spatial relationships within the drainage
basin, and on the dynamics of inter-related processes that
defi ne the functional status of rivers and their fl oodplains
(Malanson, 1993).
6.3.1 Scale and Spatial Relationships: Longitudinal
and Lateral
It is fi rst necessary to recognise that a river responds to
conditions within the upstream catchment draining to it,
and that the river corridor is therefore the terrestrial low
point which receives water fl ows, sediments, nutrients and
plant propagules from its contributing area (Brierly and
Fryirs, 2000). Accordingly, the status of a river reach is
dependent on its catchment environment. In a natural river
reach, there will be a diversity of habitats – pools, riffl es,
gravel bars, point bars, steep banks, levees, side channels
and chutes, abandoned channels, ox-bow lakes, back-
swamps and fl oodplains. These habitats will be preferentially
occupied by particular species of fauna and fl ora
(see, for example, Marston et al., 1995; Richards et al.,
2002). It is often the case that conservation focuses on
particular species, but a key to general ecological health
is the maintenance of habitats (Ward and Tockner, 2001).
However, the habitats refl ect the behaviour of the river at
the reach scale. For example, if a river meanders without
constraint, and inundates its fl oodplain roughly once every
1–5 years, it is likely that it will create and maintain a
natural diversity of habitat. Whether this behaviour occurs
will depend on the way in which the catchment is occupied
and managed, and delivers water and sediment to the
reach in question. Thus the crucial connections to understand
are those between the hydrology, sediment supply
and ecology of the contributing catchment area, and the
character of the river reach to which it drains. This results
in a structure of longitudinal relationships and upstream–
downstream connectivity, while lateral connections are
also critical in linking the aquatic (river) and terrestrial
(fl oodplain) environments across the riparian zone.
The ecological status of a river reach is strongly dependent
on longitudinal (downstream) connectivity, and this
is refl ected in the river continuum concept (Vannote et al.,
1980; Petts et al., 2000). This is based on the idea that
rivers transport water and sediment downstream through
a systematic continuum of conditions from steep, headwater
reaches with coarse bed material, and shallow,
tumbling fl ow, in narrow valleys, to more gently-sloping
lowland reaches with fi ne silty, sandy beds, deeper, slow
fl ow and wide fl oodplains. Related to this, the ecology
changes systematically from upstream to downstream
reaches, and a continuum here refl ects the changing
nature of biological processes. Upstream, woody debris is
introduced from the riparian vegetation and is ‘processed’
by ‘shredders’ (invertebrates that consume leaves and
woody material) in the stream, while also supplying nutrients
and dissolved organic carbon. Further downstream,
these products of upstream processes are used by other
species, as when fi ne particulate organic matter is either
collected or fi ltered by species which are ‘gatherers’. This
implies that management which prevents upstream biological
functions will have downstream effects because of
the continuous relationships existing along the river
course. Similarly, managing the river by introducing dams
and weirs will interrupt the continuity of migration of
species by inhibiting upstream fi sh migration to spawning
sites, and the downstream fl ow of both seeds and plant
material from which vegetative reproduction may take
place.
There may be some debate about the details of the
‘river continuum’; an alternative is the process domains

concept (Montgomery, 1999), which argues that localscale
geomorphological and biological processes determine
the stream habitat, the disturbance regimes and the
species interactions that infl uence stream communities
and biodiversity. However, for practical purposes, whether
there are longitudinal zones, or a longitudinal continuum,
is less important than that there are strong upstream–
downstream relationships. Aquatic organisms have
evolved to be adjusted to the most probable set of physical
conditions arising from the fl uvial geomorphology and
hydrology. Downstream reaches rely on carbon inputs
from upstream, while middle reaches are sites of primary
production (with clear water and less shading by
vegetation).
Longitudinal relationsips are supplemented by lateral
connections (Ward and Stanford, 1995). The high biodiversity
of river corridors is, in part, a refl ection of the
diversity of habitat at the margin between two distinct
ecosystems, those of the river and the fl oodplain. The
riparian zone is an aquatic–terrestrial ecotone, and is a
zone of transition and exchange which benefi ts from, and
regulates, the processes and functions of the ecosystems
it connects. Interference with one inevitably affects the
other, and the ecotone is itself a high priority for conservation.
When the channel is deepened and dredged, or when
embankments are created, the lateral connectivity between
river and fl oodplain is disrupted. In some cases this
restricts access to lateral channels, side-arms and ox-bow
lakes, which convert to terrestrial status. The bi-directional
connection between the river and groundwater is
inhibited and fl oodplain recharge is restricted. The potential
for fi sh species to use fl oodplain woodland as a fl oodperiod
refuge is prevented and the fi sh population suffers.
Bank-side vegetation provides shade, which regulates
water temperature, and tree roots a diversity of microhabitats,
which benefi t the aquatic fauna; it also supplies
organic material to the stream. Removal of the vegetation
removes these functions and causes a loss of productivity,
habitat and biodiversity.
6.3.2 Dynamics: Variable Flows, Sediment Delivery
and Channel Migration
Aquatic and riparian species are adapted to and require
variable fl ows (including overbank fl ows), and are sensitive
to the timing of that variation. They also require different
fl ows at various stages of their life cycles. For
example, salmonidae require a stable gravel substrate
when spawning, but at other times depend on fl ows
which dilate the gravel and fl ush out fi ne sediment that
has accumulated in the pore spaces. The preferred water
Uncertainty in Riparian and Floodplain Restoration 83
depth, velocity and substrate for a given species will also
change between the fry, juvenile and adult life stages, and
this requires that there is access to spatial variation of
habitat, which provides suitable refuge locations during
high fl ows for individuals which cannot survive extreme
conditions.
Some riparian tree species, such as black poplar
(Populus nigra) and willows, require occasional large
fl oods to create new sedimentary surfaces for colonisation
and regeneration, but also need a gradual recession
during the period of seedling establishment for their survival.
The timing of high fl ow needs to match the release
of seeds in late spring and early summer; this is an issue
in river basins with reservoirs or water transfers that
reduce fl ows and change the natural timing of high fl ows.
The role of high fl ows is represented in the fl ood pulse
concept (Junk et al., 1989; Middleton, 1999), which
emphasises that fl oods cause channel migration and create
new habitat, and supply nutrients and genetic material
(seeds and plant material) to the riparian environment.
However, these processes can occur across a range of
fl ows, so a general idea of a range of fl ow pulses (Tockner
et al., 2000) is also found in the literature on riparian
ecology.
Of course, fl ows that are too high may both
cause damage and create anaerobic conditions which
prevent germination and cause mortality in early seedlings.
This illustrates that the process of managing
fl ows for ecological benefi ts requires a subtle understanding
of the impacts of fl ows on the life histories of a
range of species. This has given rise to the concept of
‘environmental fl ows’, which include a wide range of fl ow
levels at different times, and with different recurrences,
each of which contributes to regeneration and to different
life-cycle stages of different aspects of the aquatic and
riparian ecology (Hill and Platts, 1991; Mahoney and
Rood, 1998). The management of such variable fl ows,
seasonally and inter-annually, is a complex and uncertain
process demanding an adaptive response to the unfolding
of the unpredictable climatically-driven fl ow regime.
However, it is a process which is now occurring in many
climatic regions and is considered further in the following
section.
An important characteristic of natural, unmanaged
rivers is that they migrate across their fl oodplains at rates
that depend on the energy of their fl ows and the resistance
of their perimeter sediments to entrainment, erosion and
transport. This means that they gradually turn over their
fl oodplain sediments, and that the fl oodplain consists of a
mosaic of surfaces of varying age and sedimentology
giving rise to a mosaic of diverse habitats (Salo et al.,
1986; Nilsson et al., 1991b).

84 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
The most well-known form of such migration is that of
river meandering, which arises because of bank erosion
on the outside of bends and the deposition of point bars
on the insides. This process is what creates the diversity
of habitat and is therefore very important for biodiversity.
It is commonly understood that disturbance of ecosystems
is critical for the structuring and maintenance of biodiversity,
as it stops a plant succession from progressing to a
uniform mature state and continually resets the succession
locally and re-introduces pioneer species. In some ecosystems,
disturbance may be caused by wind, fi re or human
infl uences (in shifting cultivation, for example). In river
corridors, it is caused by erosion, deposition and river
migration. The intermediate disturbance hypothesis
(Connell, 1978; Ward et al., 1999) argues that maximum
biodiversity occurs under conditions of intermediate disturbance.
Too much disturbance (as in very dynamic
braided rivers) results in a preponderance of pioneer
species, while too little disturbance allows a succession to
progress towards a mature and relatively uniform plant
community. Both have lower biodiversity than occurs
with rates of disturbance that are intermediate. If in river
corridors it is high fl ows, erosion, deposition and channel
migration (the river dynamics) that promote biodiversity,
it follows that practices of river management designed to
inhibit movement of the river (such as embankment and
bank protection) are likely to reduce biodiversity. Ecological
status will improve therefore when rivers are given the
freedom to move. This is a concept which is now enshrined
in policy for the management of the tributaries of the
Rhine and the Meuse in The Netherlands, and involves
relocation of embankments, increased off-channel fl ood
storage and restoration of the functions of abandoned
channels and side-arms.
From this outline it is evident that the characteristics of
the natural biophysical system of the river-fl oodplain
environment are connectivity, spatial variability and temporal
instability (dynamics). It is generally acknowledged
that restoration of these characteristics is a necessary component
of any restoration initiative although the scale at
which they can be restored is very variable. The uncertainty
posed by restoring the very characteristics that were
removed in order to make rivers more manageable and by
extension less uncertain in their behaviour also needs to
be accommodated (see also Gregory and Downs, Chapter
13). This is as much a question of changing attitudes
(personal, public and institutional) through education as it
is a question of understanding more about the biophysical
processes, although it remains the case that such understanding
is highly uncertain when it requires prediction of
future behaviour at a specifi c location within a river
system (e.g. Chapter 5).
6.4 THE SCALE AND PRACTICE
OF RESTORATION
The practice of river restoration is informed by the continuing
scientifi c investigation of the vital links, at different
spatial and temporal scales, between biotic and abiotic
components across a catchment. However, it is also limited
by competing needs for resources within a catchment.
The net result in many river basins is that river restoration
takes place at relatively small, confi ned sites where reinstatement
of the links between the river and its fl oodplain
is limited in degree and extent and primarily focussed
on re-establishing lateral connectivity. It is important to
distinguish between river restoration that takes place at
this spatially limited scale of the reach or section of a
reach and river restoration that takes into consideration the
longitudinal linkages within a river basin and manages
the primary inputs of water and sediment across the whole
catchment. There are many thousands of river restoration
projects worldwide that have been implemented at the fi rst
scale but far fewer examples of resource management
taking place at the scale of the catchment (see Chapter 13).
At both scales there is a considerable challenge for scientists
to defi ne ecosystem needs in a way that can guide
policy formulation and management action (Poff et al.,
2003). However, the challenge is considerably greater at
the scale of the catchment because levels of uncertainty
about the ecological and other outcomes and the number
and range of stakeholders that need to engage with the
process increase rapidly as spatial scale increases. Furthermore,
until quite recently, the water needs of humans
and those of ecosystems have been seen in competition
(Richter et al., 2003) and the quantum leap in human
perception from this to viewing ecosystems as legitimate
users of water (King and Louwe, 1998; Naiman et al.,
2002) has largely still to be made.
To illustrate the issues and practice of working at
different scales, the example of the needs of fl oodplain
forest ecosystems and the different scales at which those
needs can be provided is considered here. There are many
inter-related variables operating in a fully functional fl oodplain
forest ecosystem but, nevertheless, their diverse,
mobile vegetation mosaics can be said to have four essential
requirements to be self-regenerating (Table 6.2). To
get all of these it is necessary to manage the disturbance
processes that arrive at a fl oodplain and this can be done
in a number of ways at both the reach and catchment
scales.
6.4.1 Catchment-Scale Management
At the scale of the catchment, restoration initiatives are
likely to involve management of physical processes in one

or more places in the catchment upstream of the fl oodplain,
so that they eventually have an effect on the disturbances
arriving in the fl oodplain zone. This type of
disturbance management is ‘indirect’ but it is also the
most desirable for long term successful and self-sustaining
restoration or management of fl oodplain forests. It allows
the river to fl ood and the channel to move and create its
own sites for the regeneration of trees. It is, however, quite
diffi cult to achieve for a variety of reasons:
• It is not easy to predict where disturbances will have an
infl uence in the fl oodplain zone and it is therefore an
uncertain form of management.
• Unpredictability and reduced control have not been
desirable features for river managers and as such they
tend to be avoided.
• In highly managed and fragmented river systems there
may be too many human interventions for it to work.
Uncertainty in Riparian and Floodplain Restoration 85
Table 6.2 The four essential requirements for a self-regenerating fl oodplain forest (from Hughes and Muller, 2003; Hughes et al.,
2005)
Requirement Rationale
Flows needed by fl oodplain forests • Regular fl ows which replenish and maintain fl oodplain water tables.
These fl ows allow established trees to grow.
• Periodic high fl ows which cause channel movement and sediment
deposition. These provide potential regeneration sites and should be
variable between years.
• Well-timed fl ows through the fi rst growing season which allow
delivery of seeds to the fl oodplain and establishment of seedlings.
Unseasonal high fl ows can cause high mortality to seedlings in their
fi rst growing season.
Regeneration sites needed by fl oodplain forests • Open sites as many pioneer tree species typical of fl oodplain forests
cannot tolerate competition.
• Sites that are moist through the fi rst growing season to facilitate
regeneration.
• Sites near the water’s edge because these tend to be moister and catch
organic debris. However, sites right on the water’s edge tend to suffer
from fl ow disturbance and waterlogging.
• A variety of sediment types to provide regeneration niches for a
variety of species.
Water table conditions needed by fl oodplain forests • Water tables accessible to the roots of seedlings through their fi rst
growing season.
• Gradual recession of water tables following a fl ood.
• Limited waterlogging.
Propagation materials needed by fl oodplain forests • Seeds which are carried by the river and deposited during fl oods. The
phenology of seed release and the timing of fl ood peaks are critical in
any year for successful establishment of seedlings.
• Vegetative material which arrives by fl ood or is deposited locally.
• Seeds that are carried in the wind. Whereas seeds carried in the river
always move from upstream areas to downstream areas, seeds carried
in the wind tend to move in the direction of prevailing winds.
• It requires consensus among a huge number of
stakeholders.
The ways in which this catchment-scale management
of water and sediment resources takes place is varied
but there are now a number of both established and
emerging methodologies which have been put into practice,
many of which are reviewed more fully elsewhere
(Arthington, 1998; Hughes and Rood, 2003; Postel and
Richter, 2003):
• Managed releases downstream from impoundments.
This methodology is relatively simple in concept and
involves planning fl ow releases from structures such as
dams so that they provide maximum benefi t to downstream
aquatic and riparian ecosystems as well as to
other users. In practice this can only be contemplated
where the engineering design of the dam structure gives

86 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
suffi cient control to release fl ows. The approach has
been well-tried in North America and particularly good
examples can be found in the St Mary River in Alberta,
Canada (Rood and Mahoney, 2000) (Case Study 6.1)
and in the Truckee River of Nevada, USA (Rood et al.,
2003). Both examples use the Recruitment Box Model
of Mahoney and Rood (1998) which allows quantitative
description of the combined requirement for appropriately
timed high fl ows to create and saturate suitable
fl oodplain sites downstream and subsequent gradual
fl ow recession (ramping rates) to permit seedling survival.
Detailed knowledge of the requirements of the
germination and seedling establishment phases of the
life-cycles of target tree species have to be known and
ecologically relevant fl ows have to be characterised.
There is considerable debate on how best to characterise
the most ecologically critical aspects of fl ow regimes
(Olden and Poff, 2003) such that an optimal balance is
struck between a minimum of hydrological indices and
maximum explanation of ecosystem function by these
indices. The process becomes more diffi cult as projects
go beyond prescribing fl ows for a single ecosystem, like
Case Study 6.1
The ‘Recruitment Box’ model developed by Mahoney
and Rood (1998) delineates a zone on a fl oodplain,
defi ned by elevation and time, in which riparian cottonwood
seedlings are likely to become successfully
established if streamfl ow patterns are favourable
(Figure 6.2). Along the St Mary River in Alberta,
Canada, fl ow regulation from a headwater dam built in
1953 led to high mortality of established cottonwood
(Populus deltoides) trees and no recruitment of new
trees in the fl oodplain downstream due to insuffi cient
fl ows at critical times in the growing season (Rood
et al., 1995) (Figure 6.3). During the 1990s, after identifi
cation of the cause of the problem, regional water
resource managers implemented changes in the operation
of the St Mary Dam (Rood and Mahoney, 2000)
(Figure 6.4). In particular, fl ows were designed to
provide a gradual reduction in the falling limb of the
hydrograph after the spring snow-melt fl ood rather
than the abrupt fall that occurred during the post-dam
period. This occurred because the dam operators shut
the spillway gates at the dam to divert water into irrigation
channels. The gradual recession of fl oods was
considered vital for replenishment of water tables in
the fl oodplain zone and for the establishment of cottonwood
seedlings whose roots are unable to maintain
contact with the water table if it falls too rapidly.
Figure 6.2 The recruitment Box model applied to the lower St
Mary River (modifi ed from Mahoney and Rood, 1998)
(a)
(b)
Figure 6.3 (a) A photograph of the St Mary River fl oodplain
taken in 1991 shows dead cottonwoodtrees before successful
fl ood releases were implemented (Photograph by Francine
Hughes). (b) (See also colour plate section) By 2002 there is
signifi cant growth of young cottonwood trees on the fl oodplain
following planned releases (Photograph by Stewart Rood)

Figure 6.4 The two pairs of graphs depict a stage hydrograph
with superimposed recruitment box at the ‘ideal’ time and elevation
and with ideal drawdown rates. (a) In 1964, a post-dam fl ood
was managed for maximal cut-back rather than naturalised recession.
Regeneration of trees did not occur in that year. (b) In 1995,
a managed fl ow, using the recruitment box as a guide, successfully
promoted regeneration with a well-timed fl ood peak and
suitable fl ood recession rates through the fi rst growing season
(from Rood and Mahoney, 2000)
a fl oodplain forest, into satisfying multiple ecosystem
needs.
• Flow allocation methodologies. River fl ows are altered
as soon as water is used for human purposes. Richter
et al. (2003) state that ‘the ultimate challenge of ecologically
sustainable water management is to design and
implement a water management program that stores and
diverts water for human purposes in a manner that does
Case Study 6.2
Uncertainty in Riparian and Floodplain Restoration 87
not cause affected ecosystems to degrade or simplify’.
There is necessarily a limit to how much water can be
taken out of a river, and at what times of the year, if an
ecosystem like a fl oodplain forest is to remain selfsustainable.
The question of how to allocate water to
achieve sustainable water use for a range of ecosystems
and human purposes is addressed by a series of holistic
fl ow allocation methodologies, many of which have
been developed in semi-arid areas in Australia, South
Africa and North America. Exampes of these models
include the Bench Marking Methodology (Arthington,
1998; Brizga, 2000) and the Flow Restoration Methodology
(Arthington et al., 2000) in Australia; The Building
Block Methodology (King and Louwe, 1998) and
the DRIFT methodology (King et al., 2003; Brown and
Joubert, 2003) in South Africa. Closely allied to these
models is the Adaptive Water Management Framework
for initiating ecologically sustainable water management
programmes in the United States and elsewhere
in the world by Richter et al. (2003). The general
approach of these methodologies is to determine the
‘environmental fl ows’ necessary to sustain aquatic and
riparian ecosystems and then to integrate the identifi ed
fl ow needs with other fl ow requirements within the river
basin on seasonal, inter-annual and even decaded timescales.
They are multi-stage, consensual or round-table
approaches requiring the input of many experts and data
on a range of aquatic and riparian ecosystems, on hydrological
and geomorphological aspects, on modelling of
relationships between hydrological and biological attributes
and eventually on all the other water uses in the
river basin. It is at the stage when environmental fl ows
are determined that fl oodplain forest requirements are
included in the process. The applicability of these methodologies
to the restoration of fl oodplain forests in
Europe is discussed by Hughes and Rood (2003).
Another holistic approach to the management of water
resources is the ‘alternative futures’ approach. This
approach includes consideration of land uses across a
catchment as well as fl ow allocation and among other
places has been applied in the Willamette River Basin in
Oregon, USA (Case Study 6.2).
In the Willamette Basin, in western Oregon, USA, an alternative futures analysis has been carried out to inform community
decision making regarding land and water use in the river basin. This is a participatory approach to river basin
planning that presents stakeholders with a range of alternative scenarios involving higher or lower levels of land and
water use in the basin and their resultant environmental impacts. In the Willamette Basin, the current and historical
landscapes were analysed and then three future scenarios were generated, refl ecting varying assumptions about land

88 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
and water use. Historical data on land and water use and population levels dating from 1850 were used and scenarios
were projected to 2050 (Baker et al., 2004). Scenarios were evaluated on four areas of resource endpoints that were
considered to be of value to stakeholders: water availability; river attributes such as channel structure, riparian and
instream ecosystem richness; ecological condition of streams (using fi sh and benthic invertebrates as indicators); and
terrestrial wildlife in the basin (Dole and Niemi, 2004) (Figure 6.5).
The fl oodplain of the Willamette River has been studied specifi cally with regard to prioritising parts of the historical
fl oodplain that might be particularly suitable for restoration using a landscape modelling approach (Hulse
and Gregory, 2004). This categorises the fl oodplain into areas of high and low potential for a range of both biophysical
properties and socio-economic constraints considered at three spatial scales: the river network, the reach and
the focal area. Scale was considered important because interactions between bio-physical and socio-economic
factors and with them priorities, change at different spatial scales. Biophysical factors included characteristics like
channel complexity and hydrology and fl oodplain vegetation type. Socio-economic constraints included factors such
as private or state ownership of fl oodplain land and population density. The units used for applying these categories
are ‘slices’ of fl oodplain at right angles to fl ow, each one kilometre long. Sections of the river with low constraints
and high opportunity were identifi ed using this process and enabled prioritisation of candidate river and fl oodplain
restoration sites. The results are presented as maps with marked areas that integrate representation of processes with
patterns (Hulse and Gregory, 2004) (Figure 6.6). This methodology was used in the generation of scenarios for the
alternative futures analysis in the Willamette Basin and particularly in assessment of the sensitivity of endpoints in
the river valley (Baker et al., 2004).
Figure 6.5 Percentage change in the ‘forested riparian’ indicator of natural resource condition in the Willamette River Basin, in the
historical and three future scenarios. The indicator is the percentage of a 120-metre wide riparian buffer strip with forest vegetation
along all streams in the valley ecoregion (Hulse et al., 2002). The future scenario labels are described as: Conservative – a low level
of development involving a high degree of natural resource protection; Plan Trend – a level of development consistent with current
policies and trends; Development – refl ects a loosening of current policies to allow a freer rein to market forces across all landscape
components but still within the range of what stakeholders considered plausible (From Ecological Application (2004) EA14-2,
pp. 313–324, fi gure 4. Reprinted with permission from The Ecological Society of America.)

Uncertainty in Riparian and Floodplain Restoration 89
Figure 6.6 Graphical example of river reaches (1, 2 and 3) with coincident low constraint and high opportunity to restore channel
complexity and native fl oodplain forest. The units of (a) Population density, are people per square kilometre circa 1990 within each
one kilometre slice of the fl oodplain. The units of (b), number of structures, are rural buildings per square kilometre circa 1990 within
each one kilometre slice of the fl oodplain. The units of (c), loss of channel complexity, are net increase or decrease in channel length
within each one kilometre slice of the fl oodplain between 1850 and 1995. The units of (d), loss of fl oodplain forest, are net decrease
in area of fl oodplain forest within each one kilometre slice of the fl oodplain between 1850 and 1990. (From Hulse & Gregory (2004)
Urban Ecosystems 7 (3): 295–314. Reprinted with kind permission from Springer Science and Business Media.)

90 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Although the number of river basins to which these
emerging holistic methodologies have been applied is
increasing, they remain the exception and not the rule.
Richter and Postel (2004) give 350 as the number of river
basins worldwide that are listed on a data base held by the
United States’ Nature Conservancy and are optimistic that
obstacles to implementing these approaches are increasingly
being overcome. Very positive steps have been taken
in Europe to apply more holistic approaches to river basin
management through the European Union Water Framework
Directive and its required River Basin Management
Plans. However, management of total water volume and
its seasonal distribution (currently carried out by most
national river management agencies in Europe) tend to be
limited to the management of water quantities in rivers to
satisfy requirements for pollution dilution and the maintenance
of specifi ed minimum fl ows. This management is
largely effected through control of water abstraction
licences (Figure 6.7). It is usual for management of low
fl ows to be separated from management of fl oods and this
is an area that needs to be addressed to achieve a more
holistic approach.
• Sediment management. While a huge amount of literature
has appeared on the management of fl ows, far less
research has been carried out on the management of
sediments and there are also few examples of projects
that involve restoration of sediment loads in rivers. In
the European context, sediment loads have drastically
changed over the last 200 years in response to changes
in mass movements in upper catchments, to sand and
gravel extraction in fl oodplain zones, to the installation
of upstream impoundments and to the armouring of
river banks with artifi cial dykes. In the Drôme River in
France measures are proposed to restore sediment loads
to the river. These include a moratorium on gravel
extraction and clearance of river bank vegetation to remobilise
sediment through bank erosion (Michelot,
1995). The aim is to improve the delivery of sediment
to the Ramieres du Val de Drôme Nature Reserve, which
is designated for its high quality fl oodplain forest in two
active, braided river reaches.
6.4.2 Reach-Scale Management
At the scale of the reach, there are two main approaches
to carrying out river and riparian restoration. The fi rst
involves managing physical processes locally in the fl oodplain
zone, for example by introducing sluices into side
channels to control water levels in selected parts of a
fl oodplain. The second involves managing the landforms
in the fl oodplain so that the physical disturbances act differently
on each part of the fl oodplain. Both types of disturbance
management are more directed to specifi ed
reaches of a river than managing fl ows of water and sediment
at a catchment scale. However, they have the disadvantage
that they can only have a relatively local effect. In
most parts of Europe it is the type of management that can
most readily be promoted and in many river basins it has
already taken place through a number of river restoration
projects. Restoration at this scale is easier to achieve than
catchment-scale management for a variety of reasons:
Figure 6.7 A typical environmental allocation for a UK catchment is shown in this graph (from Environment Agency, 2002). It
varies considerably through the seasons and is determined through a technical approach called the Resource Assessment and Management
(RAM) framework. First each river reach in the catchment is assessed on the basis of its physical characteristics, fi sh populations,
macrophytes and macro-invertebrates to produce an environmental weighting. Five environmental weighting bands are used to classsify
the sensitivity of each reach to the effects of water abstraction. The environmental weighting is used with long term fl ow duration
data to derive an Ecological Flow Objective and the percentage left for abstraction. The Ecological Flow Objective seeks to protect
low fl ows and fl ow variability by allowing percentages of fl ow bands to be abstracted. The impacts of groundwater abstraction (both
seasonally and spatially) on river fl ows is also incorporated into the process. To preserve water levels in fl oodplain sites adjacent to
rivers, a site-scale approach has been used in parts of the UK using Water Level Management Plans. These are usually applied to sites
designated as wetland nature reserves whose water tables are related to those of adjacent river courses in either a direct or indirect
way

• It is more predictable than catchment-scale management
of disturbances and as such is relatively easy to control
and poses less uncertainties.
• River managers are already used to managing rivers at
a local scale through the engineering of fl ood defences.
• It can neatly be fi tted into chosen river sections between
reaches where other human interventions dominate.
• It requires consensus among a much smaller group of
stakeholders.
Exampes of restoration at this scale are now common
and can be categorised into a series of activities which may
be carried out singly or in combination at any individual
site (Hughes and Muller, 2003):
• Restoration of the channel wetted perimeter. These projects
have as a main aim the improvement of instream
habitats, often for fi sh populations. They emphasise
increasing the heterogeneity of physical habitat. For
example, the Ecological recovery of the Vindel and Pitte
Rivers (EVP) project, in northern Sweden involves
removal of instream structures installed during the nineteenth
century for driving logs for the timber industry.
In addition, large boulders that were removed for smooth
fl oating of log rafts are being put back to diversify
instream habitats (Nilsson et al., 2005).
• Re-connection of side arms in rivers. Most projects in
this category aim to remove one or more sections of
artifi cial embankments to allow fl ows to penetrate fl oodplain
zones that have been cut off. There are a number
of examples of this approach that are well documented
in the literature, such as the Regelsbrunner Au project
on the River Danube in Austria (Scheimer et al., 1999)
or the L’Ile de la Platière and Rosillon Channel sites on
the River Rhône in France (Michelot, 1995; Downs
et al., 2002). These projects had as a major objective the
restoration of both the quality of fl oodplain forests and
improved opportunites for forest regeneration by
increasing lateral connectivities between the channel
and the fl oodplain.
• Increase in fl oodplain storage capacity through setting
back defences, lowering fl ood defences or lowering the
fl oodplain. There are major and well-documented plans
for a range of these activities in the distributaries of the
River Rhine in The Netherlands and also in the River
Meuse. They form part of a master plan to increase fl ood
storage capacity in these rivers in anticipation of sealevel
rise and increasing frequency of fl oods from
upstream (Middelkoop and van Haselen, 1999). These
activities are currently restricted to a series of discrete
sites, such as at the Millingerwaard Nature Reserve in
Uncertainty in Riparian and Floodplain Restoration 91
The Netherlands where fl ood defences have been
lowered. The general approach in The Netherlands is
shown in Figure 6.8. In the United Kingdom fl ood
storage washlands are proposed along some rivers but
in all these projects the design of drainage dykes and
control structures will be very important in determing
the amount of control on the length of time and depth
that water stays in the washland. Creative confi guration
of the fl oodplain surface can mimic natural fl oodplain
habitat heterogeneity to give conservation gains as well
as fl ood storage gains if water control is fl exible enough
to operate for the good of wetland ecosystems as well
as for fl ood control.
• Management of the river’s sediment load. Sedimentation
is an essential process along the margins of river channels
as newly created alluvial bars are prime regeneration
sites for many species of fl oodplain vegetation.
Groynes can be used to create artifi cial ‘beaches’
although their primary purpose is usually to maintain a
channel for navigation. Re-activation of erosion in sites
where embankments have been removed will alter the
sediment loads downstream.
• Management of riparian vegetation. This can take the
form of planting fl oodplain forests or waiting for natural
regeneration to take place. In either case, management
of the vegetation that grows may be considered necessary,
though this is not in the spirit of self-sustainability.
In many river basins, grazing by domestic animals in
riparian zones prevents natural regeneration and fencing
Figure 6.8 In The Netherlands, large parts of the Rhine fl oodplain
will be lowered as part of the Flood Action Plans drawn up
by the International Rhine Committee. The aim is to increase
fl ood storage capacity of the Rhine fl oodplains. The proposed
works are shown schematically in this diagram (from Middelkoop
and Van Haselen, 1999)

92 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
off the riparian zone becomes an important form of
management. Management of the vegetation also has
implications for the volume of woody debris that arrives
in a river. MacNally et al. (2002) suggest that in the
Murray-Darling River system in Australia only 15% of
the former woody debris load is now present. There are
projects where woody debris has been put back into
rivers and some where artifi cial log jams have been built
to test their restoration effect (Dewberry et al., 1998).
6.4.3 The Use of Reference Systems in River
Restoration Projects
It is common to set objectives for river restoration projects
by using reference systems. In the EU Water Framework
Directive (2000/60/EC) it is a major and perhaps unrealistic
aim to use a network of near-natural reference systems
so that the ecological status of rivers can be measured
against them. In essence this involves fi nding a river
system that has the attributes considered desirable in the
restored system and using it as a template on which to base
the restoration activities. Criteria specifi cally for establishing
riparian reference conditions are proposed by Harris
(1999) using multivariate analyses of aspects of vegetation
community composition and structure. Typically a reference
system will be part of or the whole of a less-damaged
river system, preferably located in a similar bioclimatic
zone and in a river basin exhibiting similar physiographic
characteristics. Alternatively it can be an historic system,
whose attributes are known from maps, old photographs
or written accounts. A critique of the use of reference
systems to defi ne objectives in river restoration is given in
Hughes et al. (2005), where the following categories of
problems encountered with the use of reference systems
are discussed:
• There are often no appropriate reference systems to
use.
• Many catchment parameters have changed since the
times of chosen historic reference systems.
• Climate change has been continuous through the
Holocene.
• Projected climate change is of uncertain magnitude.
• Alien species have become common in the landscape
and cannot be avoided.
• Landscape context changes through time.
Using reference systems can give river managers a misplaced
confi dence in the predictability of ecological outcomes
in river restoration projects (Hughes et al., 2005),
although the degree to which the project outcomes and the
reference system coincide can largely be decided at the
outset when objectives are set (Simons and Boeters, 1998).
Evaluation of restoration projects is also usually carried
out against the reference system, although with highly
variable levels of rigour (Anderson and Dugger, 1998;
Stream Corridor Working Group, 1998).
Nevertheless, reference systems can provide useful,
broad guiding images for restoration. In the United
Kingdom, a number of sources of information are used.
At the catchment scale, the National Vegetation Classifi cation
(Rodwell, 1991a, 1991b, 1992, 1995, 2000), The
National Biodiversity Network (NBN), the Multi-Agency
Geographic Information for the Countryside (MAGIC)
databases on Habitats and Sites of Special Scientifi c interest
(SSSI’s), provide information on the distribution of
habitats, communities and species across the United
Kingdom. For river reaches, there may be River Morphology,
River Habitat and River Corridor Surveys, and
Hydro-Morphology Quality Assessments that have already
been undertaken in some relatively undisturbed reaches
(RSPB, NRA, RSNC, 1994). Using data on the catchment
area, slope and river corridor width of the reaches, it may
be possible to extrapolate to other river reaches with
similar characteristics. Overall, the aim should be to
develop a sense of how the spatial structure of the catchment
hydrology and river morphology and ecology are
related, and how the catchment and river function under
‘natural’ conditions, bearing in mind that any system is in
a transient state over time.
6.5 MONITORING AND EVALUATING
RESTORATION
6.5.1 What is Ecological Success and How Do We
Evaluate It?
A widely applicable scheme for evaluating the ecological
success of river restoration projects does not currently
exist, though many authors have emphasised the need for
such a scheme (NRC, 1992; Downs et al., 2002). Part of
the problem is in defi ning exactly what is meant by ecological
success and recently an attempt has been made to
identify fi ve criteria that can be used for its measurement
(Palmer et al., 2005):
1. The design of an ecological river restoration project
should be based on a specifi ed guiding image of a more
dynamic, healthy river that could exist at the site.
2. The river’s ecological condition must be measurably
improved.
3. The river system must be more self-sustaining and
resilient to external perturbations so that only minimal
follow-up maintenance is needed.

4. During the construction phase, no lasting harm is
infl icted on the ecosystem.
5. Both pre- and post-construction ecological assessment
is carried out and the information made available.
Refi nements to this scheme have been proposed by
Jansson et al. (2005) and Gillilan et al. (2005), including
the need to include stated ecological mechanisms by
which any intended restoration will reach its goal. A major
diffi culty with this scheme is giving practical meaning to
the concepts of resiliency and self-sustainability, principally
because these are entirely relative terms that are also
closely linked to the concept of dynamic equilibrium.
The term ‘resilience’ was fi rst coined in the 1970s in
the fi eld of population ecology and used to characterise
the magnitude of population perturbations a system could
tolerate before changing into some qualitatively different
dynamic state (Holling, 1973; May, 1976a). More recently
it has been used to describe the ability of an ecosystem to
regain a functional state following disturbance and the
rapidity of this process is a measure of its resilience
(Waring, 1989). The disturbance is usually temporary and
if it is suffi ciently regular, component species of the ecosystem
often evolve a dependence on it to complete their
life cycles. Both ecological and evolutionary time scales
must be considered in assessing the signifi cance of disturbances
of different magnitudes and frequencies in river
and riparian ecosystems (Poff, 1992; Hughes, 1994; Dodds
et al., 2004). As described earlier in this chapter, in the
case of riparian and fl oodplain ecosystems, the disturbance
is provided by fl oods and many riparian species
have indeed evolved a dependence on these fl oods for the
regeneration phase of their life cycles. ‘Dynamic equilibrium’
encompasses the notion of changing parameters
within a stable framework and has often been used in
descriptions of whole ecosystems. If the stable framework
changes then a qualitatively different dynamic state again
prevails. The defi nition of a stable framework is usually
determined by the temporal and spatial scales of an investigation
and often limited by an investigator’s consideration
of change through time and over space. However, as
stated by May (1976b), in the real world there are no fi xed
parameter values; environmental parameters and interactions
between organisms and between organisms and their
environment are constantly fl uctuating. It follows that
stable frameworks do not exist in the real world unless
they are artifi cially delimited spatially, temporally or
both.
Measuring resilience to evaluate how successful we
have been in returning it to a riparian or fl oodplain ecosystem
during a river restoration project is very diffi cult
to do. If a framework has been identifi ed to describe a
Uncertainty in Riparian and Floodplain Restoration 93
dynamic equilibrium, then within the defi ned framework
it is possible to predict and measure recovery of a disturbed
ecosystem. However, it is a huge challenge to
predict the functions of ecosystems when they begin to
fl uctuate along new trajectories caused by environmental
change or include new species arrivals. Since all ecosystems,
including restored ecosystems, are moving along
fl uctuating trajectories, it follows that, conceptually, it is
impossible to measure resilience and therefore to evaluate
ecological success using this criterion (see Chapters 8 and
11 for further discussion). In practical terms it is possible
to delineate a framework for measuring resilience though
different types of ecologists might defi ne very different
frameworks for the same restoration project. A fl oodplain
forest ecologist might defi ne a framework where selfsustainability
is measured in terms of turnover rates
for fl oodplain habitats (10 2 –10 3 years) related to fl ood
return periods (Mahoney and Rood, 1998; Hughes and
Rood, 2001). A fi sheries ecologist might assess selfsustainability
in terms of provision of instream habitats
that permit completion of fi sh life cycles over annual or
10 1 year time frames (e.g. Frissel and Nawa, 1992).
The evaluation of success using such measurements
also changes with scale and is discussed with respect to
fl oodplain forests by Hughes et al. (2005). When viewed
at a small spatial scale (10 2 metres), habitat patches in the
forest will change from year to year as a response to
channel movement and species arrival, death or migration.
At a broader spatial scale of a reach or whole fl oodplain,
variability becomes less pronounced because the balance
of different habitat patches remains more constant (Figures
6.9(a) and 6.9(b)), particularly over short time frames of
10 1 to 10 2 years (see Ward et al., 2002). However, over
longer periods (10 3 years), this balance at the reach scale
might shift in response to climate change, sea level change,
isostatic uplift, change in availability of propagules or
changes in the biophysical attributes of the catchment
(Figure 6.9(c)). In this scenario, shifting geomorphological
processes (for example from aggradation to downcutting
of the river valley) and changed channel patterns
cause major changes to ecological vectors and patterns in
the fl oodplain. Evaluation of success can only be relative
to these changing frameworks or else within the context
of a framework that has been held stable.
6.5.2 What is Ecological Quality and How Does it
Relate to Ecological Success?
Closely related to the consideration of how to measure
ecological success is the need to give meaning to the
concept of ecological quality. Both concepts are dependent
on the monitoring or surveillance of a series of physical

94 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Figure 6.9 At the scale of a whole fl oodplain, progressive or rapid changes can take place in the distribution of fl oodplain vegetation
communities. Thus in 6.9(a), t1 is biodiversity at an initial time period and consists of vegetation communities a and b, present
in the proportion of 4a to 3b. In 6.9(b), t2 is biodiversity at a later time. Vegetation communities a and b are still present in the same
balance but they are all in different locations following shifts in channel location. Over much longer time frames (or over rapid time
frames following extreme events or human intervention), there may be changes in catchment parameters that alter the geomorphological
patterns and hydrological activity of the river. In 6.9(c), the meandering river has become braided and a new vegetation community
c has arrived. The biodiversity (t3) of fl oodplain vegetation communities has now changed (from Hughes et al., 2005)
and biological parameters that are representative of ecosystem
function. Whereas the fi rst is tied to the evaluation
of a restoration project, the second is descriptive of the
state of an ecosystem. However, they are interrelated in
functional terms and decisions on what to monitor to
measure ecological quality will have a signifi cant impact
on our ability to evaluate ecological success in river restoration,
since this will depend on pre- and post-project
measurements of ecosystem-relevant parameters. In
Europe, the Water Framework Directive (WFD) (2000/60/
EC) provides a legislative framework for water policy and
among other things aims to maintain and improve the
aquatic environment through attention to both water
quality and quantity. It requires that member states ensure
good ‘chemical status’ and ‘ecological status’ of surface
waters. Groundwaters must meet ‘good groundwater
status’. There is no objective for riparian and fl oodplain
ecosystems; however, the health of groundwaterdependent
wetlands is an indicator of ‘good groundwater
status’. ‘Ecological status’ is described as ‘an expression
of the quality of the structure and functioning of aquatic
ecosystems associated with surface waters’ and member
states are required to monitor this status. To do this there
has to be clear understanding of what ‘ecological status’
means, and much debate about this has followed publication
of the Water Framework Directive (WFD) both at
European Union and national levels.
The main descriptors of ecological quality for rivers are
grouped into biological, hydrogeomorphological and
physico-chemical elements (Table 6.3). Soon after adoption
of the WFD (22nd December, 2000), the fi rst volume
of Common Implementation Strategy (CIS) guidance was
produced. Its main aim is to help member states share
expertise on implementation of the technical aspects of the
WFD, including achievement of good ecological status for
surface waters. Annex V of the WFD includes normative
defi nitions of the three ecological status categories (High,
Good and Moderate). The levels at which these three categories
will be set is the subject of a major EU intercalibration
process that should have been completed in

2006, but which remains contentious and partial. This
process is being carried out by the ‘European Centre for
Ecological Water Quality and Intercalibration’ (EEWAI),
which aims to compare the different national classifi cation
systems for ecological status assessment. Two key parts of
the process are identifi cation of reference conditions and
establishment of monitoring protocols. It is already known
that river hydromorphology will have to be restored
(unless the river stretch is designated ‘heavily modifi ed’,
i.e. restoration would adversely affect its function for
navigation, fl ood protection or power generation) if it is
preventing biological elements from achieving ‘good ecological
status’ (Withrington, Natural England, personal
communication).
Individual member states have to integrate implementation
of the WFD with their already established protocols
for monitoring and evaluating the status of their environment,
species and habitats. In the United Kingdom, the
statutory conservation agencies have recently introduced
a system of ‘Common Standards Monitoring’ (CSM) for
sites designated under national legislation and European
directives. Site-based conservation is a signifi cant part of
biodiversity and earth science conservation in the United
Kingdom and evaluation of the effectiveness of measures
put into place to achieve biodiversity conservation is the
Uncertainty in Riparian and Floodplain Restoration 95
Table 6.3 Quality elements for the classifi cation of Ecological Status-Rivers (EU Water Framework Directive, 2000/60/EC,
Annex V)
Quality Element Characteristics
Biological elements • Composition and abundance of aquatic fl ora
• Composition and abundance of benthic invertebrate fauna
• Composition, abundance and age structure of fi sh fauna
Hydrogeomorphological elements supporting
the biological elements
Chemical and physico-chemical elements
supporting the biological elements
• Hydrological regime
– quantity and dynamics of water fl ow
– connection to groundwater bodies
• River continuity
• Morphological conditions
– river depth and width variation
– structure and substrate of the river bed
– structure of the riparian zone
• General
– thermal conditions
– oxygenation conditions
– salinity
– acidifi cation status
– nutrient conditions
• Specifi c pollutants
– pollution by all priority substances identifi ed as being
discharged into the body of water
– pollution by other substances identifi ed as being discharged in
signifi cant quantities into the body of water
main aim of the CSM process. The CSM guidance for
monitoring rivers designated as important by national legislation
(Sites of Special Scientifi c Interest (SSSIs) under
the Wildlife and Countryside Act, 1981) recommends the
use of fl uvial geomorphological audit to defi ne modifi cations
to rivers and identify options for river restoration
(JNCC, 2004). Riparian zones are included in the CSM
guidance for rivers, both in their own right as woodland,
grassland or swamp communities and for the contribution
they make to in-channel river communities. Flow levels
and patterns are mostly assessed in terms of their importance
to instream species and habitats rather than to adjacent
terrestrial ecosystems. The functional attributes of
riparian zones is considered through measurement of
factors such as production of woody debris and leaf litter
as is their intrinsic habitat value, for example bird, mammal
or invertebrate habitat. Such monitoring schemes for designated
sites will be integrated within broader-scale, river
basin wide monitoring of ecological quality, under the
WFD, of both aquatic and groundwater-dependent terrestrial
ecosystems.
The design and density of monitoring networks eventually
established on European rivers will determine
how useful they are for also evaluating river restoration
success. Numerous research projects funded by the

96 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
European Commission are in the process of exploring
monitoring protocols for particular groups of species, such
as macroinvertebrates and fi sh, with a particular emphasis
on their transferability, comparability and ability to indicate
ecological quality. For example, the LIFE in UK
Rivers Project (1998–2003) produced monitoring protocols
for rivers with Callitricho-Batrachion vegetation and
for 10 river-dependent species such as the otter (Lutra
lutra). Others are concerned with defi nition of reference
conditions, taking into account climate change (e.g.
EUROLIMPACTS).
6.5.3 How Do We Value Biodiversity?
Most commonly, use of the term biodiversity indicates a
measure of species numbers, although habitat diversity
and genetic diversity are also vitally important (Heywood,
1995; Gopal and Junk, 2001). It is relatively straightforward
to measure species diversity compared with ecological
and genetic diversity, although the signifi cance of
changing biodiversity remains elusive in all cases. In
riparian and fl oodplain habitats, levels of species diversity
are related to both the ecotonal situation of these habitats
and to their physical heterogeneity. Lateral connectivity
between riparian habitats and channels and longitudinal
connectivities between upper and lower parts of river
basins all contribute to high levels of biodiversity (Tabacchi
et al., 1996). Less well understood are the vertical
connectivities between groundwater zones and riparian
zones but they contribute in complex ways to habitat
quality in interstices of fl oodplain sediments (Lambs,
2004) and to diversity of organisms in the hyporheic zone
(Gibert et al., 1997; Hancock et al., 2005).
Species composition on fl oodplains has changed over
long and short time scales as shown by changing pollen
profi les from peat deposits in fl oodplain zones (Godwin,
1941), responding to climate change and the progression
of species through the landscape. In the last few centuries,
many new or alien species have arrived on fl oodplains as
a result of introductions and garden escapes. Because
riparian and fl oodplain zones are physically highly
dynamic, and are populated by species able to cope with
habitat mobility, they provide excellent habitats for ecological
pioneers, which many invasive species can be classifi
ed as (Planty-Tabacchi et al., 1996). These species have
often out-competed native species because hydrological
and geomorphological factors have changed and the
species that have evolved their life cycles to fi t in with
natural hydrological cycles are no longer favoured. In river
restoration projects today, the presence of alien species
can be viewed in several ways. They can either be eradicated
by active processes, such as clearance, or they can
have conditions for their regeneration made diffi cult by
altering physical inputs, such as fl ood timing to reduce
their competitive edge. On the other hand, the number of
alien species present in many river systems makes their
complete eradication impossible, and acceptance of some
of them as components of the ecosystems found in riparian
zones is another approach. It is easier to justify this
approach in view of predicted climate change, which will
signifi cantly alter the range that many species can occupy
(Jensen, 2004) and the hydrological patterns in river basins
(Montgomery and Boulton, 2003) but in ways that are
currently uncertain. It is certain that species assemblages
in the landscape will change and perhaps have no presentday
analogues and that an open-minded approach to
acceptable community composition will have to be taken
(Hughes et al., 2005).
6.6 INTERCONNECTIVITY AND BOUNDARY
CROSSING: INSTITUTIONAL COMPLEXITIES
OF RESTORING FLOODPLAINS
The task of restoring fl oodplains poses multi-dimensional
challenges to policy makers and project managers alike.
Involving essentially a reconfi guration of the interaction
between a river and adjacent low-lying land, fl oodplain
restoration has far-reaching implications for existing
forms of water and land use. Floodplains provide multiple
functions and services for humans as well as the natural
environment. These can range from valuable artefacts for
socio-economic reproduction, such as crops, timber, water
or prime land for development, to less tangible but equally
valuable functions, such as protection from fl ooding,
attractive landscapes or opportunities for recreational pursuits.
The restoration of functional fl oodplains requires
changes to existing activities on the site of the fl oodplain
itself, but also – particularly in the case of larger schemes
– along whole reaches of a river and even a whole catchment.
On this wider scale it can signifi cantly infl uence, for
instance, levels of fl ood protection, the navigability of a
river reach or the viability of current farming practices. In
this way, fl oodplain restoration affects a wide range of
stakeholders and interests (Adams and Perrow, 1999;
Adams et al., 2004; Turner et al., 2000; Adger and Luttrell,
2000), making it potentially highly controversial.
Behind these stakeholders and interests lie institutions
– understood here as rule systems – designed to
protect and provide a variety of private and public goods,
ranging from commercial products to rights of access. For
each of the policy fi elds affected by fl oodplain restoration
– primarily water protection, fl ood defence, nature conservation,
recreation, navigation, urban and rural development
– complex institutional arrangements have been

designed and adapted over the years. Each institutional
arrangement comprises a set of codifi ed norms (such as
laws, regulations and contractual obligations), planning
instruments and funding mechanisms, as well as standardised
procedures of operation, values and accepted
practices of the relevant organised and individual actors.
A scheme to restore a fl oodplain requires the successful
enrolment of these institutions and organisations in such
a way as to create a result acceptable to the principal
stakeholders. This is a highly complex process, involving
multiple uncertainties (on institutional constraints, see
WWF, 2000).
Managing the interdependence of multiple functions,
actors and institutions is, however, not the only major
socio-political challenge of fl oodplain restoration. Considerable
uncertainty is generated by the diverse spatial scales
and time frames involved. As described above, restoring a
fl oodplain requires consideration of the longitudinal connectivity
of a fl oodplain to river uses both up- and downstream
as well as of the lateral connectivity to ways in
which adjacent land is used (Adams and Perrow, 1999).
Even on site, interventions generally cut across several
functional and administrative boundaries. These can relate
to the spatial remit of local landowners and farmers, planning
authorities, government agencies, protected areas or
infrastructure networks (e.g. rail, roads, canals). Temporally,
functional fl oodplains are characterised by their
dependence on fl ooding events which are, by their nature,
periodic and unpredictable, and by signifi cant time lags
between changes in biotic and abiotic systems (see above;
Adams and Perrow, 1999). Socio-economically, too, the
process of restoring a fl oodplain is framed by diverse time
scales, ranging from the payback periods for investments
in altered practices of agriculture and forestry to the electoral
periods of key public authorities. Floodplain restoration
is, therefore, not only highly complex but also highly
unpredictable. The institutional challenge is further complicated
by the fact that many institutional arrangements
– in particular for nature conservation – exhibit a strong
tendency to protect existing conditions rather than encourage
change, and those which do pursue change are generally
geared towards achieving specifi c targets rather than
creating suitable frameworks for open-ended processes, as
is required for functional fl oodplains.
6.6.1 Effective Institutions: The Search for Optimal
Fit, Interplay and Scale
Our knowledge of institutions which can support – or
obstruct – the protection of public goods such as water,
fl ood defence and biodiversity has been developing rapidly
over the past decade (cf. Breit et al., 2003). However, rela-
Uncertainty in Riparian and Floodplain Restoration 97
tively little is known about the institutional dimensions of
fl oodplain restoration itself. Exceptions include studies
of institutional constraints and complexities (Adams and
Perrow, 1999), competing discourses of fl oodplain restoration
(Adams et al., 2004), relevant European Union policies
(WWF, 2000, 2004) and European case studies (e.g. Zöckler,
2000) and economic valuations of the functions and services
provided by fl oodplains or wetlands (Gren et al.,
1995; Turner et al., 2000; Adger and Luttrell, 2000).
The Science Plan of the Institutional Dimensions of
Global Environmental Change Project (IDGEC) of the
International Human Dimensions Programme (IHDP)
offers useful analytical frameworks for conceptualising
some of the essential institutional challenges of resource
management in general and fl oodplain restoration in particular
(Young, 1999, 2002). It identifi es three generic
factors infl uencing the effectiveness of environmental
institutions: problems of fi t, problems of interplay and
problems of scale.
The issue of fi t addresses the need to develop institutional
arrangements that match the properties of the biogeophysical
systems they are designed to regulate. Fit can
relate to a variety of ecosystem properties. The following
are identifi ed in the IDGEC Science Plan: closed vs open
systems; heterogeneity/homogeneity; interdependencies
among subsystems; simplicity/complexity; productivity/
metabolism; cyclicity/periodicity; resilience; equilibria;
dynamics (Young, 1999, p. 47). Problems of spatial misfi t
are a particularly common cause of institutional ineffectiveness.
The territories covered by institutions rarely
match those of biogeophysical systems, resulting in an
inability of the institutions to internalise external effects
(both positive and negative) effectively. The management
of fl oodplains is fraught with boundary problems of this
kind. Floodplain restoration not only works across a
variety of physical spaces along the river and across the
catchment, but also involves institutions and organisations
from multiple policy fi elds – from nature conservation and
fl ood defence to agriculture – each with their own spatial
remits.
Interplay relates, by contrast, to interdependencies
between different institutions. The assumption here is that
the effectiveness of an institution depends not only on its
inherent qualities but also on how well it builds on, and is
connected to, the broader institutional context. Institutional
interplay can be horizontal, between different policy
fi elds, and vertical, between different levels of social
organisation. A further distinction is made between functional
linkages emanating from the properties of the institutions
involved and political linkages as expressions of
deliberation (Young, 2002). Problems of interplay are very
familiar to efforts to restore fl oodplains, which are often

98 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
confounded by the inability to bridge differences in the
objectives, power structures and modes of action of the
various key organisations. Horizontal interplay is complicated
by the number of policy fi elds affected and vertical
interplay by the increasing role of the European Union and
catchment-scale approaches to fl oodplain management.
Problems of scale can be of a spatial and a temporal
nature. Spatially the effectiveness of an institution depends
on fi nding the appropriate level of social organisation for
specifi c instruments and measures, taking consideration of
the connectivity between scales and the needs this creates
for multi-level and multi-directional forms of governance.
Temporally the issue of scale is about managing the
diverse time frames within which actors operate, institutions
work, projects are implemented, ideas are generated
etc. Here, too, the relevance to fl oodplain restoration is
self-evident. Identifying the appropriate spatial scale for
policy development, strategic guidance, operational management,
public participation and so on is of paramount
importance. Similarly, actors operate according to very
different time scales, some of which are rigid and predictable,
others much less so.
Research on problems of fi t, interplay and scale suggests
that solutions are rarely straightforward. For instance,
efforts to overcome problems of spatial fi t by institutionalising
river basin management can create new misfi ts and
disturb existing modes of interplay (Moss, 2003). Success
would appear to be dependent less on attempting to reduce
the given complexities and more on fi nding ways of
accommodating complexity. It is argued here that a similar
approach is required when dealing with uncertainty. Given
that substantial uncertainties will continue to surround
fl oodplain restoration despite advances in our knowledge
of the physical, biological and socio-political systems, it
makes sense to consider possible coping strategies. The
remainder of this section addresses ways of coping with
institutional uncertainty in fl oodplain restoration with an
empirically based study of three different approaches.
6.6.2 Coping with Uncertainty I:
Keeping Restoration Simple
The fi rst option for coping with uncertainty is to limit the
scope and scale of restoration. This was particularly
common of the earlier schemes to restore fl oodplains in
Europe. Up until the late 1990s most fl oodplain restoration
schemes were small-scale and site-based. They were
typically single-issue projects, targeting environmental
improvements as a rule. They involved only a small
number of actors and policy instruments, often relying on
a single source of funding for the physical interventions.
It is generally true to say that these early generation res-
toration schemes were conducted largely in isolation from
national or regional policy initiatives, whether for fl ood
protection, biodiversity enhancement or rural development.
Examples include the Rheinvorland-Süd project in
Germany, a scheme to improve hydrological and ecological
conditions by widening ducts and removing structures
in a section of the Rhine fl oodplain near Rastatt, the Long
Eau project in England, in which fl ood banks were set back
for primarily environmental benefi ts, and the Bourret
project in France, a scheme to reconnect an old arm to the
River Garonne and restore an alluvial forest – again primarily
for environmental benefi ts.
Being relatively unambitious and straightforward,
schemes of this kind tend to avoid the most pressing problems
associated with high levels of uncertainty and complexity.
They have succeeded in restoring fl oodplain
habitats with limited resources and, in some cases, within
a short period. With the benefi t of relatively straightforward
administrative procedures, organisational structures
and funding mechanisms it has been demonstrated how
fl oodplain ecosystems can be restored on a small scale.
These schemes do, however, have several critical limitations.
They rarely incorporate a catchment perspective on
restoration, but concentrate on the site itself. Being predominantly
single-issue schemes, they regularly overlook
potential benefi ts for other policy areas, such as fl ood
protection, recreation or rural development. Little attention
is generally paid to cultivating support for the project
in the wider policy making domain, scientifi c communities
or even in the local community. The performance of
many such projects is rarely monitored or evaluated
systematically.
In terms of resolving problems of fi t, interplay and scale
it is possible to observe how schemes of this kind are illequipped
to meet the principal institutional challenges to
fl oodplain restoration set out above. Spatially, the small
scale and site focus of the projects offer little opportunity
to consider the catchment dimensions of fl ow regimes and
biophysical connectivity beyond the immediate reach.
Institutional interplay may be less diffi cult but only
because the number of policy fi elds and organisations
involved is kept small. Integrating the schemes into the
development of the locality or region may well prove diffi
cult at a later date for this reason. Problems of scale are
reduced to site-based perspectives, missing valuable
opportunities to explore multi-scalar solutions.
6.6.3 Coping with Uncertainty II: Embracing the
Challenge of Open Outcomes
Following recent shifts in policies towards fl ood risk management,
integrated water resources management and

ural development, more favourable framework conditions
are creating windows of opportunity for ambitious forms
of fl oodplain restoration (see previously). In response to
these institutional drivers, and also to a growing recognition
of the inadequacies of current ways of managing
rivers, a new generation of fl oodplain restoration schemes
is emerging which are of a quite different scope to those
of the early to mid-1990s. These schemes deliberately set
out to address some of the complex challenges to largescale,
integrated fl oodplain restoration. Distinctive features
of the new generation schemes are their multiple
objectives (covering, for instance, fl ood defence, biodiversity,
rural development and water quality management),
their wide actor engagement (including the relevant policy
fi elds, local authorities, non-government organisations
and the general public), their use of various instruments
from different policy fi elds (e.g. joint funding from fl ood
defence and agri-environment budgets) and their interaction
with policy making processes, serving for instance as
pilot projects for national policy development. In addition,
they often have a long term vision for the measures envisaged
and take a catchment – or at least large-scale – perspective
on the fl oodplain. Restoration sites are selected
according to their suitability for the catchment and not
primarily because they are available.
Examples include: the Lenzen project in Germany, a
major scheme to set back fl ood banks along a length of
the Elbe river allowing fl ooding primarily for nature conservation,
but also fl ood defence and regional development,
benefi ts; the Parrett Catchment Project in England,
an ongoing project to promote more sustainable techniques
of fl ood management in the whole catchment
serving multiple purposes (fl ood protection, water level
regulation, biodiversity targets, rural development); and
the La Bassée project in France, a planned, large-scale
fl ood retention scheme on the Seine upstream of Paris with
multiple benefi ts (fl ood protection, biodiversity, regional
development). Since these new generation schemes were
only launched from the late 1990s onwards and are all at
very early stages of implementation it is at present impossible
to judge their effectiveness. They would at least
appear to have the potential to overcome some of the
principal institutional constraints to fl oodplain restoration
which have thwarted or curtailed efforts in the past.
Our research fi ndings, however, caution against overoptimistic
expectations from the new generation of projects.
Early signs suggest that the sheer complexity of the
tasks they are tackling and the uncertainties they are
exposing are posing a major problem for project management.
Building and maintaining the large partnerships
takes time and care. Striking an acceptable balance and
negotiating trade-offs between diverse policy objectives
Uncertainty in Riparian and Floodplain Restoration 99
is very demanding. Accessing multiple funding sources
requires a high degree of fl exibility to satisfy different
funding agencies. Attempts to enroll instruments from
different policy fi elds can reveal serious incompatibilities
and inconsistencies. As a result, project design and
implementation has become more complex, more timeconsuming
and more expensive, endangering effective
implementation.
Floodplain restoration schemes of this kind are making
substantial steps towards addressing problems of fi t, interplay
and scale. Their catchment orientation and long term
perspective create a better fi t between ecosystem properties
of the fl oodplain and the institutional arrangements
for its restoration, both in spatial and temporal terms.
Building on better interplay between institutions is central
to the new generation projects, as is exploiting different
scales of action – from national pilots to local management
teams – for different purposes. What the schemes
are revealing, though, are serious secondary problems
associated with this more ambitious and integrated
approach to fl oodplain restoration. Efforts to take on problems
of fi t, interplay and scale are proving, in many cases,
too demanding for project management. This does not
query the desirability of addressing these core institutional
problems but, rather, raises questions about how project
managers can be assisted in doing so.
6.6.4 Coping with Uncertainty III: Tightening
Controls to Secure Better Policy Delivery
A third way of coping with uncertainty can be identifi ed
not at the level of individual schemes of fl oodplain restoration
but in the development and pursuit of policy. Here the
uncertainty addressed relates to the outcomes of new policies
and the strategy is to attempt to minimise uncertainties
of policy delivery by tightening controls over those
entrusted with implementation. As described earlier, in
several policy fi elds of direct relevance to fl oodplain
restoration a more integrated and holistic approach to
problem solving can be detected. This applies particularly
to fl ood protection, following recent fl ooding events, water
resources management, in response to the Water Framework
Directive, and nature conservation. Changes in
policy content potentially conducive to fl oodplain restoration
can also be observed – if to a lesser degree – across
Europe in spatial planning, agriculture, forestry and
rural development. Many of these supranational and
national policy initiatives are characterised by a more
comprehensive problem analysis, longer term visions for
improvements, better cross-sectoral policy integration,
more strategic guidance and stronger and broader local
partnerships. The guiding principles underpinning

100 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Integrated Water Resources Management (IWRM) on a
river basin scale are a case in point (WWF, 2004, p. 24).
Ironically, whilst policy content is, generally speaking,
taking a broader perspective, modes of policy implementation
are to some extent becoming more restrictive. In
recent years a growing array of instruments have been
introduced which set out in detail not only what policy is
to be pursued but also how this is to be done at the operational
level. Targets are set to measure progress, strict
consultation procedures must be followed to gain planning
approval, match-funding is required to demonstrate multifunctionality
and audits are conducted to assess the performance
of projects and programmes alike. These controls
are widely justifi ed by governments in terms of their value
in improving accountability, policy integration and, above
all, cost effectiveness.
Our research suggests, however, that measures of this
kind are having important (unintended) negative effects on
the ability of project managers to implement fl oodplain
restoration schemes. The fi rst problem relates to the cumulative
effect of the new policy initiatives. Each of the new
requirements – whether on policy content or style – may
individually make a lot of sense. Experiences of policy
implementation show that the combined effect of multiple
new requirements can be to create a degree of management
complexity that can severely delay the progress of
some restoration projects and cause others to be shelved.
Ironically, therefore, effective policy delivery is being
jeopardised by the sheer extent of policy reform.
The second problem is more fundamental, having to
do with an emergent culture of control in policy making
circles. The measures to increase accountability, policy
integration and cost effectiveness refl ect not only very
justifi able concerns about effective policy implementation
and effi cient use of public funds but also the concerns of
senior management in many government agencies that the
policy rethinking described above is not fi ltering down
effectively to the operational level. This argument is used
by senior offi cers to justify tighter control – or ‘guidance’
– to assure more effective implementation. Whilst the need
for greater strategic guidance over such complex issues as
a catchment-scale approach to fl ood protection is undisputed
by all those involved, one of the effects has been to
restrict the freedom of action of project and programme
managers. In the past their judgement – for example on
whether to fund a restoration project – was based on their
individual expertise, local knowledge and professional
experience; today it is framed much more by targets
devised at regional, national or even supranational levels.
Consequently, the nature of their work is adapting in order
to meet what Michael Power has termed the ‘rituals of
verifi cation’ required by auditing processes (Power, 1997).
For the task of restoring fl oodplains this poses a particular
dilemma: whilst recent policy shifts and new generation
schemes are encouraging fl oodplain restoration to entertain
greater risks and uncertainties, administrative procedures
to assure policy implementation are becoming
increasingly risk-averse.
In terms of fi t, interplay and scale the picture here is
more differentiated. Recent policy initiatives relating to
fl oodplain restoration are certainly addressing very clearly
problems of spatial and temporal fi t, taking a more catchment-oriented
and long term perspective on the river and
land management. Inter-sectoral interplay is also strong.
Vertical interplay and issues of scale appear more problematic,
however. The rhetoric of policy documents tend
to be very supportive of multi-level and multi-direction
governance. The reality – whether intentional or not – is
often very different, with control mechanisms of central
government agencies reaching down into the operational
level of project management to an unprecedented extent.
This, it appears, is undermining both project implementation
and – ultimately – policy delivery.
6.6.5 Making Policy More Sensitive to the Challenges
of Project Management
It has been observed here how policy makers, in their
efforts to encourage integrated, cross-sectoral and multiagency
action, often overlook the implications for implementation
at the operational level. In future, more
consideration needs to be given to how individual policy
incentives work in conjunction with others; that is, how
they alter the existing institutional setting. In addition,
policy makers need to be more sensitive to the contexts of
action in which their instruments operate. What makes a
policy instrument effective is not the assumed preferences
of individuals acting according to a rational choice logic
but the real scope and willingness of stakeholders to alter
their practices. More feedback into policy making processes
is needed about the real-life experiences of project
management and stakeholder involvement at the operational
level. This will require more monitoring of how and
why policy instruments do or do not work in practice. In
addition, if uncertainty and complexity are unavoidable
features of ambitious schemes of fl oodplain restoration, as
it appears, then project managers and other stakeholders
need assistance in assessing the situation and their own
ability to meet the challenge. Policy needs to provide not
only targets for orientation but also frameworks for developing
the necessary economic, social and institutional
capital at local and regional level, and instruments capable
of adapting to the dynamics of a fl oodplain restoration
process. On this basis policy makers and project managers

alike should be better equipped to identify and exploit
windows of opportunity for the restoration of fl oodplains
in the future.
6.7 CONCLUSION
A case has been made for the re-introduction of diversity
and variability in the riparian zone. It is advocated that
this is done through the management of physical processes
and within a policy context that is sensitive to the complexities
of successfully implementing river restoration
initiatives. Inevitably the re-introduction of diversity and
variability will also introduce a signifi cant level of uncertainty
for river managers. The challenge is to decide what
are acceptable levels of uncertainty for different stakeholders,
for different scales of involvement (such as catchment
strategies or planning for a particular reach) and for
different purposes (such as biodiversity targets or fl ood
management). In the light of projected climate changes it
is suggested that the key is to fi nd ways of adapting to
uncertainty (see Chapter 14), rather than aiming to reduce
uncertainty in the scientifi c sense of the word.
6.8 ACKNOWLEDGEMENTS
Much of the thinking that has gone into this chapter has
arisen from the work of the EC-funded project FLOBAR2
(EVK1-CT-1999-00031). We thank all our colleagues on
that project for stimulating and enjoyable discussions over
many years of working together. We would also like to
thank Stewart Rood, John Mahoney, David Hulse and Joan
Baker for permission to use their work in the two case
study boxes; David Withrington of Natural England for
reading and commenting on this manuscript; and Ian
Agnew of the Department of Geography at the University
of Cambridge for drawing the fi gures.
REFERENCES
Adams WM, Perrow M. 1999. Scientifi c and institutional constraints
on the restoration of European fl oodplains. In: Marriott
S, Alexander J, Hey, R (Eds), Floodplains: Interdisciplinary
Approaches. Geological Society of London Special Publ. 163,
Geological Society: London; 89–97.
Adams WM, Perrow MR, Carpenter A. 2004. Conservatives and
champions: river managers and river restoration discourses in
the United Kingdom. Environment and Planning A 36: 1929–
1942.
Adger WN, Luttrell C. 2000. Property rights and the utilisation
of wetlands. Ecological Economics 35: 75–89.
Anderson DH, Dugger BD. 1998. A conceptual basis for evaluating
restoration success. In: Wadsworth KG (Ed.), Changing
Resource Values in Challenging Times. Transactions, 63rd
Uncertainty in Riparian and Floodplain Restoration 101
North American Wildlife and Natural Resources Conference,
Wildlife Management Institute: Washington, DC; 111–120.
Arthington AH. 1998. Comparative Evaluation of Environmental
Flow Assessment Techniques:Review of Holistic methodologies.
LWRRDC Occasional Paper 26/98.
Arthington AH, Brizga SO, Choy SC et al. 2000. Environmental
Flow Requirements of the Brisbane River Downstream from
Wivenhoe Dam. South East Queensland Water Corporation
Ltd. and Centre for Catchment and In-stream Research, Griffi th
University: Brisbane, Queensland, Australia.
Baker JP, Hulse DW, Gregory SV et al. 2004. Alternative futures
for the Willamette River Basin, Oregon. Ecological Applications
14 (2): 313–324.
Breit H, Engels A, Moss T, Troja M. 2003. How Institutions
Change. Perspectives on Social Learning in Global and Local
Environmental Contexts. Leske and Budrich: Opladen,
Germany.
Bressers H, Kuks S. 2003. What does governance mean? From
conception to elaboration. In: Bressers H, Rosenbaum W (Eds),
Achieving Sustainable Development: the Challenge of Governance
Across Spatial Scales. Praeger: New York.
Brierley GJ, Fryirs K. 2000. River styles, a geomorphic approach
to catchment characterization: implications for river rehabilitation
in Bega Catchment, New South Wales, Australia. Environmental
Management 25: 661–679.
Brizga SO. 2000. Burnett River water allocation management
plan:Proposed environmental fl ow performance measures.
Department of Natural Resources: Brisbane, Australia.
Brown CA, Joubert A. 2003. Using multicriteria analysis to
develop environmental fl ow scenarios for rivers targeted for
water resource management. Water South Africa 29 (4):
365–378.
Connell JH. 1978. Diversity in tropical rain forests and coral
reefs. Science 199: 1302–1310.
Dewberry C, Burns P, Hood L. 1998. After the fl ood: the effects
of the storms of 1996 on a creek restoration. Restoration and
Management Notes 16: 174–182.
Dodds WK, Gido K, Whiles MR et al. 2004. Life on the edge:
The ecology of Great Plains prairie streams. BioScience 54:
205–216.
Dole D, Niemi E. 2004. Future water allocation and in-stream
values in the Willamette River Basin: a basin-wide analysis.
Ecological Applications 14: 355–367.
Downs PW, Skinner KS, Kondolf GM. 2002. Rivers and streams.
In: Perrow MR, Davy AJ (Eds), Handbook of Ecological Restoration,
Volume 2 Restoration in Practice. Cambridge University
Press: Cambridge; 267–296.
EEC. 1992. Directive 92/43/EEC of the European Parliament and
of the Council of 21 May 1992 on the conservation of natural
habitats and of wild fauna and fl ora. OJ L 206, 22 July 1992,
The European Parliament: Brussels, Belgium.
Environment Agency. 2002. A Framework for Catchment Abstraction
Management Strategies. R&D Technical Report W6-066/
TR, RAM version 3 Summary Report. Environment Agency:
Bristol.
European Union. 2000. Directive 2000/60/EC of the European
Parliament and of the Council of 23 October 2000 Establishing

102 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
a Framework for Community Action in the Field of Water
Policy. OJ L 327, 23 October 2000, The European Parliament:
Brussels, Belgium.
Frissell CA, Nawa R. 1992. Incidences and causes of physical
failure of artifi cial habit structures in streams of Western
Oregon and Washington. North American Journal of Fisheries
Management 12: 187–197.
Gibert J, Fournier F, Mathieu J. 1997. The groundwater/surface
waters ecotone perspective: State of the art. In: Gibert J,
Mathieu J, Fournier F (Eds), Groundwater/Surface Water Ecotones:
Biological and Hydrological Interactions and Management
Options. International Hydrology Series, Cambridge
University Press: Cambridge; 3–8.
Gillilan S, Boyd K, Hoitsma T, Kauffman M. 2005. Challenges
in developing and implementing ecological standards for geomorphic
river restoration projects: A comment on Palmer et al.
(2004). Journal of Applied Ecology 42 (2): 223–227.
Girel J, Hughes FMR, Moss T et al. 2003. A case for fl oodplain
forests? In: Hughes FMR. (Ed) The Flooded Forest: Guidance
for Policy Makers and River Managers in Europe on the Restoration
of Floodplain Forests. FLOBAR2, Department of
Geography, University of Cambridge: Cambridge; 6–23.
Godwin H. 1941. Studies of the post-glacial history of British
Vegetation. II Fenland Pollen Diagrams. Philosophical Transactions
of the Royal Society Series B 230: 239–284.
Gopal B, Junk WJ. 2001. Assessment, determinants, function and
conservation of biodiversity in wetlands: present status and
future needs. In: Gopal B, Junk WJ, Davis JA (Eds), Biodiversity
in Wetlands: Assessment, Conservation and Function,
Volume 2. Backhuys Publishers: Leiden, The Netherlands;
277–302.
Gren I-M, Groth K-H, Sylvén M. 1995. Economic Values of
Danube Floodplains. Journal of Environmental Management
45: 333–345.
Hancock PJ, Boulton AJ, Humphreys WF. 2005. Aquifers and
hyporheic zones: Towards an ecological understanding of
groundwater. Hydrogeology Journal 13: 98–111.
Haraszthy L. 2001. The fl oodplain forests in Hungary. In: Klimo
E, Hager H (Eds), The Floodplain Forests in Europe: Current
Situation and Perspectives. Koninklijke Brill NV, Leiden and
European Forest Institute Research Report No. 10: Leiden, The
Netherlands; 17–24.
Harris RR. 1999. Defi ning reference conditions for restoration of
riparian plant communities: examples from California, USA.
Environmental Management 24: 55–63.
Heywood VH (Ed). 1995. Global Biodiversity Assessment. Cambridge
University Press: Cambridge.
Hill M, Platts WS. 1991. Ecological and geomorphological concepts
for instream and out-of-channel fl ow requirements.
Rivers 2: 319–343.
Holling CS. 1973. Resilience and stability of ecological systems.
Annual Review of Ecology and Systematics 5: 25–37.
Hughes FMR. 1994. Environmental change, disturbance and
regeneration in semi-arid fl oodplain forests. In: Millington AC,
Pye K (Eds), Environmental Change in Drylands: Biogeographical
and Geomorphological Perspectives. John Wiley &
Sons Ltd: Chichester; 321–346.
Hughes FMR, Muller E. 2003. How can fl oodplain forests be
restored. In: Hughes FMR (Ed), The Flooded Forest: Guidance
for Policy Makers and River Managers in Europe on the Restoration
of Floodplain Forests. FLOBAR2, Department of
Geography, University of Cambridge: Cambridge; 49–70.
Hughes FMR, Rood SB. 2001. Floodplains. In: Warren A, French
JR (Eds), Habitat Conservation: Managing the Physical Environment
John Wiley & Sons Ltd: Chichester; 105–121.
Hughes FMR, Rood SB. 2003. The allocation of river fl ows for
the restoration of woody riparian and fl oodplain forest ecosystems:
a review of approaches and their application in Europe.
Environmental Management 32: 12–33.
Hughes FMR, Colston A, Mountford JO. 2005. Restoring riparian
ecosystems: the challenge of accommodating variability and
designing restoration trajectories. Ecology and Society 10(1),
article 12 Online URL: http://www.ecologyandsociety.org/
vol10/iss1/art12/
Hulse DW, Gregory SV. 2004. Integrating resilience into fl oodplain
restoration. Urban Ecosystems 7: 295–314.
Hulse DW, Gregory SV, Baker JP (Eds), 2002. Willamette River
Basin: Trajectories of Environmental and Ecological Change.
Oregon State University Press: Corvallis, Oregon.
Jansson R, Backx H, Boulton AJ et al. 2005. Stating mechanisms
and refi ning criteria for ecologically successful river restoration:
A comment on Palmer et al. (2005). Journal of Applied
Ecology 42 (2): 218–222.
Jensen MN. 2004. Climate warming shakes up species. BioScience
54: 722–729.
JNCC. 2004. Common Standard Monitoring Guidance for Rivers.
Joint Nature conservation Committee: Peterborough, UK.
Johnson WC. 2000. Tree recruitment and survival in rivers: infl uence
of hydrological processes. Hydrological Processes 14:
3051–3074.
Junk WJ, Bayley PB, Sparks RE. 1989. The fl ood-pulse concept
in river-fl oodplain systems. Canadian Special Publication of
Fisheries and Aquatic Sciences 106: 110–127.
King J, Louwe D. 1998. Instream fl ow requirements for regulated
rivers in South Africa using the building block methodology.
Aquatic Ecosystem Health and Management 1: 109–124.
King J, Brown C, Sabat H. 2003. A scenario-based holistic
approach to environmental fl ow assessments for rivers. River
Research and Applications 19: 619–639.
Lambs L. 2004. Interactions between groundwater and surface
water at river banks and the confl uence of rivers. Journal of
Hydrology 288: 312–326.
MacNally R, Parkinson A, Horrocks G, Young M. 2002. Current
loads of coarse woody debris on southeastern Australian fl oodplains:
Evaluation and change and implications for restoration.
Ecological Restoration 10 (4): 627–635.
Mahoney JM, Rood SB. 1998. Streamfl ow requirements for cottonwood
seedling recruitment-An integrative Model. Wetlands
18: 634–645.
Malanson CP. 1993. Riparian Landscapes Cambridge University
Press: Cambridge.
Marston RA, Girel J, Pautou G et al. 1995. Channel metamorphosis,
fl oodplain disturbance, and vegetation development –
Ain River. Geomorphology 13: 121–131.

May RM. 1976a. Models for two interacting populations. In: May
RM (Ed.), Theoretical Ecology: Principles and Applications.
Blackwell Scientifi c Publications: Oxford; 49–70.
May RM. 1976b. Patterns in multi-species communities. In: May
RM (Ed.), Theoretical Ecology: Principles and Applications.
Blackwell Scientifi c Publications: Oxford; 142–162.
Medley K, Hughes FMR. 1996. Riverine forests. In: McClanahan
TR (Ed.), Ecosystems and their Conservation in East Africa.
Oxford University Press: Oxford; 361–384.
Michelot J-L. 1995. Gestion Patrimoniale des Milieux Naturels
Fluviaux: Guide Technique. Reserves Naturelles de France.
Middelkoop H, van Haselen COG (Eds), 1999. Twice a River –
Rhine and Meuse in the Netherlands. RIZA Report no. 99.003,
Institute for Inland Water Management and Waste Water Treatment/RIZA:
Arnhem, The Netherlands.
Middleton B. 1999. Flood Pulsing and Disturbance Dynamics.
John Wiley & Sons Inc.: New York.
Montgomery DR. 1999. Process domains and the river continuum.
Journal of the American Water Resources Association 35:
397–410.
Montgomery DR, Boulton SM. 2003. In: Wissmar RC, Bisson
PA (Eds), Strategies for Restoring River Ecosystems: Sources
of Variability and Uncertainty in Natural and Managed Systems.
American Fisheries Society: Bethesda, Maryland; 39–80.
Moss T. 2003. Solving problems of ‘fi t’ at the expense of problems
of ‘interplay’? The spatial reorganisation of water management
following the EU Water Framework Directive. In:
Breit H, Engels A, Moss T, Troja M (Eds), How Institutions
Change. Perspectives on Social Learning in Global and Local
Environmental Contexts. Leske and Budrich: Opladen,
Germany; 85–121.
Naiman RJ, Bunn SE, Nilsson C et al. 2002. Legitimizing fl uvial
ecosystems as users of water: An overview. Environmental
Management 30: 455–467.
Nilsson C, Grelsson G, Johansson M, Sperens U. 1991a. Small
rivers behave like large rivers: effects of postglacial history on
plant species richness along river banks. Journal of Biogeography
18: 533–541.
Nilsson C, Ekblad A, Gardfjell M, Carlberg B. 1991b. Long term
effects of river regulation on river margin vegetation. Journal
of Applied Ecology 28: 963–987.
Nilsson C, Lepori F, Malmqvist B et al. 2005. Forecasting
environmental responses to restoration of rivers used as log
fl oatways: An interdisciplinary challenge. Ecosystems 8:
779–800.
NRC (National Research Council). 1992. New Strategies for
America’s Watersheds. National Academic Press: Washington,
DC.
Olden JD, Poff NL. 2003. Redundancy and the choice of hydrologic
indices for characterizing streamfl ow regimes. River
Research and Applications 19: 101–121.
Pahl-Wostl C. 2002. Participative and stakeholder-based policy
design, analysis and evaluation processes. Integrated Assessment
3: 3–14.
Pahl-Wostl C. 2004. Information, public empowerment, and the
management of urban watersheds. Environmental Modelling
and Software 20: 457–467.
Uncertainty in Riparian and Floodplain Restoration 103
Palmer MA, Bernhardt ES, Allen JD et al. 2005. Standards for
ecologically successful river restoration. Journal of Applied
Ecology 42 (2): 208–217.
Petts GE. 1990. The role of ecotones in aquatic landscape management.
In: Naiman RJ, Decamps H (Eds), The Ecology and
Management of Aquatic-Terrestrial Ecotones. UNESCO/MAB
Series 4: Parthenon, New Jersey; 227–261.
Petts GE, Gurnell AM, Gerrard AJ et al. 2000. Longitudinal
variations in exposed riverine sediments; a context for the
ecology of the Fiume Tagliamento, Italy. Aquatic Conservation:
Marine and Freshwater Ecosystems 10: 249–266.
Planty-Tabacchi A-M, Tabacchi E, Naiman RJ et al. 1996. Invasibility
of species-rich communities in riparian zones. Conservation
Biology 10: 598–607.
Poff NL. 1992. Why disturbance can be predictable: A perspective
on the defi nition of disturbance in streams. Journal of the North
American Benthological Society 11: 86–92
Poff NL, Allan JD, Palmer MA et al. 2003. River fl ows and water
wars: emerging science for environmental decision making.
Frontiers in Ecology and Environment 1 (6): 298–306.
Postel S, Richter B. 2003. Rivers for Life: Managing Water for
People and Nature. Island Press: Washington, DC.
Power M. 1997. The Audit Society. Rituals of Verifi cation. Oxford
University Press: Oxford.
Richards KS, Brasington J, Hughes FMR. 2002. Geomorphic
dynamics of fl oodplains: Ecological implications and a potential
modelling strategy. Freshwater Biology 47: 1–22.
Richter B, Postel S. 2004. Saving Earth’s rivers. Issues in Science
and Technology Spring issue: 31–36.
Richter BD, Mathews R, Harrison DL, Wigington R. 2003. Ecologically
sustainable water management: managing river fl ows
for ecological integrity. Ecological Applications 13 (1):
206–224.
Rodwell JS (Ed.). 1991a. British Plant Communities, Volume 1:
Woodlands and Scrub. Cambridge University Press:
Cambridge.
Rodwell JS (Ed.). 2000. British Plant Communities, Volume 2:
Mires and Heath. Cambridge University Press: Cambridge.
Rodwell JS (Ed.). 1991b. British Plant Communities, Volume 3:
Grasslands and Montane Communities. Cambridge University
Press: Cambridge.
Rodwell JS (Ed.). 1992. British Plant Communities, Volume 4:
Aquatic Communities, Swamps and Tall-Herb Fens. Cambridge
University Press: Cambridge.
Rodwell JS (Ed.). 1995. British Plant Communities, Volume 5:
Maritime Communities and Vegetation of Open Habitats. Cambridge
University Press: Cambridge.
Rood SB, Mahoney JM. 2000. Revised instream fl ow regulation
enables cottonwood recruitment along the St Mary River,
Alberta, Canada. Rivers 7: 109–125.
Rood SB, Mahoney JM, Reid D, Zilm L. 1995. Instream
fl ows and the decline of riparian cottonwoods along the
St. Mary River, Alberta. Canadian Journal of Botany 73:
1250–1260.
Rood SB, Gourley C, Ammon EM et al. 2003. Flows for fl oodplain
forests: successful riparian restoration along the lower
Truckee River, Nevada, U.S.A. BioScience 53: 647–656.

104 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
RSPB, NRA and RSNC. 1994. The New Rivers and Wildlife
Handbook. RSPB: Sandy, Bedfordshire.
Salo J, Kalliola R, Häkkinen I et al. 1986. River dynamics and
the diversity of Amazon lowland forests. Nature 322:
254–258.
Scheimer F, Baumgartner C, Tockner K. 1999. Restoration of
fl oodplain rivers: the Danube restoration project. Regulated
Rivers, Research and Management 15: 231–244.
Simons J, Boeters R. 1998. A systematic approach to ecologically
sound river bank management. In: de Waal LC, Large ARG,
Wade PM (Eds), Rehabilitation of Rivers: Principles and
Implementation. John Wiley & Sons Ltd: Chichester; 57–85.
Stream Corridor Working Group. 1998. Stream Corridor Restoration:
Principles, Processes and Practices. Federal Stream
Interagency Stream Restoration Working Group, USA.
Tabacchi E, Planty-Tabacchi AM, Salinas MJ, Decamps H. 1996.
Landscape structure and diversity in riparian plant communities:
A longitudinal comparative study. Regulated Rivers:
Research and Management 12: 367–390.
Tockner K, Malard F, Ward JV. 2000. An extension of the Flood
Pulse concept. Hydrological Processes 14: 2861–2883.
Turner RK, van den Bergh JCJM, Söderqvist T et al. 2000. Ecological-economic
analysis of wetlands: scientifi c integration
for management and policy. Ecological Economics 35: 7–23.
UNEP – World Conservation Monitoring Centre. 2000. European
Forests and Protected Areas: Gap Analysis. UNEP – WCMC:
Cambridge, UK.
Vannote RL, Minshall GW, Cummins KW et al. 1980. The river
continuum concept. Canadian Journal of Fisheries and Aquatic
Sciences 37: 130–137.
Ward JV, Stanford JA. 1995. Ecological connectivity in
alluvial river ecosystems and its disruption by fl ow regulation.
Regulated Rivers: Research and Management 11: 105–
119.
Ward JV, Tockner K. 2001. Biodiversity: towards a unifying
theme for river ecology. Freshwater Biology 46: 807–819.
Ward JV, Tockner K, Schmeier F. 1999. Biodiversity of fl oodplain
ecosystems: ecotones and connectivity. Regulated Rivers:
Research & Management 15: 125–139.
Ward JV, Tockner K, Arscott DB, Claret C. 2002. Riverine landscape
diversity. Freshwater Biology 47: 517–539.
Waring RH. 1989. Ecosystems: fl uxes of matter and energy. In:
Cherrett JM (Ed.), Ecological Concepts: The Contribution of
Ecology to an Understanding of the Natural World. British
Ecological Society and Blackwell Scientifi c Publications:
Oxford; 17–42.
World Commission on Dams 2000. Dams and Development: A
New Framework for Decision Making. Earthscan Publications
Ltd: London.
WWF 2000. Wise Use of Floodplains. Policy and Economic
Analysis of Floodplain Restoration in Europe. Opportunities
and Obstacles. Report, November 2000.
WWF 2004. Living with Floods: Achieving Ecologically Sustainable
Flood Management in Europe. Report, June 2004.
Young O. 1999. Institutional Dimensions of Global Environmental
Change. Science Plan. IHDP Report No. 9, IHDP: Bonn,
Germany.
Young O. 2002. The Institutional Dimensions of Environmental
Change. Fit, Interplay, and Scale. MIT Press: Cambridge,
Massachusetts.
Zöckler C. 2000. Wise Use of Floodplains – LIFE Environment
Project: A Review of 12 WWF River Restoration Projects
Across Europe. WWF European Freshwater Programme:
Copenhagen, Denmark.

River Restoration: Managing the Uncertainty in Restoring Physical Habitat Edited by Stephen Darby and David Sear
© 2008 John Wiley & Sons, Ltd
7
Hydrological and Hydraulic Aspects of
River Restoration Uncertainty for
Ecological Purposes
7.1 INTRODUCTION
N.J. Clifford 1 , M.C. Acreman 2 and D.J. Booker 2,3
1 School of Geography, University of Nottingham, UK
2 Centre for Ecology and Hydrology, UK
3 (now at) National Institute of Water and Atmospheric Research, New Zealand
This chapter presents river restoration as a hybrid activity,
involving hydrological, geomorphological and ecological
expertise. It introduces a range of techniques and methods
that are appropriate to each of these areas, gives examples
of their application and reviews some of the fundamental
opportunities and limitations of current restoration practice.
River restoration is not only a hybrid activity but also
an emerging one (both scientifi cally and practically).
‘Uncertainties’ are present at every stage of restoration
intervention. These span such basic issues as: the ability
to support ideas of catchment hydrology or fl ow regime
within which to frame restoration designs; providing fi eld
evidence for conceptual and numerical simulation models;
capturing the natural range of variability inherent in
complex and dynamic physical–ecological systems; and
incorporating all of these uncertainties themselves into
restoration design and appraisal practice. The chapter concludes
that fi ve basic sources of uncertainty (also see
Chapters 1 to 3) underpin contemporary river restoration
theory and practice: data (type, quality and quantity);
characterisation (of those physical and ecological phenomena
involved in the restoration attempt); coupling (of
physical environmental and ecological dynamics); awareness
(of opportunities and limitations) and fl exibility (in
the approach to design and evaluation). Evaluating each
of these sources at every stage of the restoration is,
perhaps, the best way of managing such uncertainties and,
too, of improving prospects for the incorporation of more
sophisticated simulation approaches to river restoration
design and appraisal in the future.
7.2 THE RIVER AND CATCHMENT AS AN
UNCERTAIN SYSTEM
Restoration of rivers may be viewed as the joint product
of the physical structure of the channel and its fl oodplain,
their hydrological integration and their ecological value
and function. Channels and fl oodplains are characterised
by the timing and quantity of fl ows and sediments which
the channel conveys, and which the fl oodplain stores. As
a result, channel restoration in its widest sense encompasses
structural modifi cations of channel form, the reestablishment
of natural fl ow regime and the reconnection
(or even recreation) of channel and fl oodplain areas (see
Chapter 6). Both the problems of, and solution to, hydrological
and hydraulic uncertainties in river restoration thus
arise from considering the channel as an embedded part
of the wider fl uvial hydrosystem (Petts and Amoros,
1996). Incorporating hydrological connections or disturbances
requires an holistic approach to streamfl ow management
(Hill et al., 1991) in recognition of the many
scales of connection or ‘dimensions’ of the fl uvial system
(Boon, 1998; and for general review see Clifford 2001).
Approaches to cost effective and multifunction river
rehabilitation works have increasingly emphasised the
need to include an ecological perspective. Attention
has focused on restoring sustainable hydrological and

106 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
hydraulic habitats using principles of fl uvial behaviour
(Newbury and Gaboury, 1993) as shown in Figure 7.1.
This fi gure illustrates the hierarchy of feedbacks (or interconnections)
between physical structure, hydrological
function and ecological responses in an alluvial river–
fl oodplain system which are relevant when designing and
assessing river restoration or rehabilitation schemes. Flow
and sediment transport depend upon catchment inputs of
material and energy. These in turn maintain and confi gure
the channel shape, determine the sediment and bedform
environments and create diversity of fl ow patterns and
structures with varying degrees of coherence and spatial
coverage. Ecological function of the channel is primarily
a response to local (imposed) conditions of velocity, depth
and substrate (the hydraulic variables), whose spatial
(cross-section, reach-scale) and temporal (event-specifi c
Human Land
Use and Flow
Regulation
• water
• sediment
• nutrients
Watershed Inputs
Fluvial Geomorphic Processes
• sediment transport/deposition/scour
• channel migration and bank erosion
• floodplain construction and inundation
• surface and groundwater interactions
Geomorphic Attributes
• channel morphology (size, slope, shape,
bed and bank composition)
• floodplain morphology
• water turbidity and temperature
Habitat Structure, Complexity, and Connectivity
• instream aquatic habitat
• shaded riparian aquatic habitat
• riparian woodlands
• seasonally inundated floodplain wetlands
Biotic Responses
(Aquatic, Riparian, and Terrestrial Plants and Animals)
• abundance and distribution of native and exotic species
• community composition and structure
• food web structure
and seasonal) characteristics refl ect wider reach-scale,
inter-reach and catchment controls (which are primarily
hydrologically determined).
While the nature of the linkages in Figure 7.1 is well
understood, giving precise values to quantities and timings
of material and energy transfers, and accounting for the
feedbacks between them, gives rise to uncertainties at all
scales. These uncertainties are compounded by the recognition
that climate and land use, which determine catchment
rainfall–run-off response, are nonstationary (that is,
they are changing in their mean level and variance) through
time. In the United Kingdom, for example, Prudhomme
et al. (2003) examine the implications of no less than
25 000 climate scenarios for four typical fl ood events as
applied to fi ve catchments. Most scenarios show an
increase in both the magnitude and frequency of fl ood
• energy
• large woody debris
• chemical pollutants
Natural
Disturbance
Figure 7.1 The nested hierarchy of the channel system and its associated habitat potential and biotic response (Tuolumne River
restoration program summary report, summary of studies, conceptual models, restoration projects, and ongoing monitoring. Prepared
for the CALFED/AFRP Adaptive Management Forum, with assistance from the Tuolunine River Technical Advisory Committee,
(2001). Reproduced with permission from Stillwater Sciences.)

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 107
events, but the largest uncertainty arises from the type
of Global Climate Model used: the magnitude of modelled
change varies by a factor of nine in Northern England
and Scotland! At scales below the level of climatic
input, major uncertainties remain in the modelling and
prediction of rainfall–run-off relationships, upon which
channel fl ow ultimately depends. Catchment run-off modelling
is increasingly able to incorporate greater physical
parameterisation, and to distribute parameter combinations
around the catchment and between surface and subsurface
fl ows. Similar predictions may, however, be
obtained using models that incorporate differing levels of
sophistication and differing parameter values, giving rise
to model ‘equifi nality’. This raises concerns about the
degree to which fundamental understanding of process
can, in fact, lead to better information on future (or
designed) outcomes (Beven, 2001). Another major source
of concern is the degree to which model predictions may
be appropriately up- or down-scaled (for a comprehensive
review of these issues, see Beven, 2000 and Bierkens
et al., 2000).
Gilvear et al. (2002) and Hendry et al. (2003) point out
that the hydrological basis for determining future fi sheries
stocks is complicated by issues of water and sediment
quality, as well as quantity. Water quality necessitates
consideration of longer- and shorter-term land use histories,
run-off behaviour and the monitoring or modelling
of diffuse as well as point-source pollutants. ‘Hydrology’
itself, therefore, has many uncertain components,
and Clarke SL et al. (2003) call for the development
of ‘hydrogeomorphological knowledge’ of catchments
supporting ‘tools’ for water resource management. The
need to service ever-more complex models and to comply
with increasingly complex (but frequently competing) legislative
requirements will also place growing pressures for
the provision of data (both quantity and quality). This is
likely to demand changes to long-standing methods of
data acquisition and processing (Marsh, 2002). All of
these essentially hydrological issues are encountered
before the channel-scale is addressed from a hydraulic
standpoint!
Within the channel at reach and subreach scales, it is
hydraulics which underpin connections between the physical
and ecological environments. Determining an appropriate
channel morphology and the characteristics of fl ow
behaviour in response to changing discharge and sediment
transport are key factors in designing restoration schemes.
Such schemes must be both sustainable in a physical
sense, and functional in an ecological sense. Both considerations
require some allowance for natural dynamics and
post-design evolution. Over the last two decades, a new
subject of ‘eco-hydraulics’ has developed (Leclerc, 2002),
in which a range of monitoring and modelling strategies
link the expertise of engineers, geomorphologies and
ecologists to design and assess restoration or rehabilitation
schemes. Yet, from very basic stages of fl ow ‘characterisation’
through to fl ow modelling, habitat simulation and
post-project appraisal, numerous uncertainties exist. This
chapter reviews the sources and implications of these
uncertainties, and provides case study examples of current
and prospective restoration practice motivated by ecohydraulic
considerations.
7.3 HYDROLOGICAL ASPECTS AND
UNCERTAINTY IN CHANNEL RESTORATION
7.3.1 Connectivity, Disturbance and the Ecological
Flow Regime in Channels
A river ecosystem and its associated benefi ts to humankind
are strongly conditioned by the pattern of fl ows
between days, seasons and years (including fl oods and
droughts) that occur within the drainage basin. For
example, fl oods maintain river structure and sediment
distribution (Hill et al., 1991), medium fl ows trigger
fi sh migration (Junk et al., 1989) and low fl ows maintain
species diversity (Everard, 1996). The pattern of fl ows
required to support a river ecosystem is called the environmental
fl ow requirements (Dyson et al., 2003). If the
river fl ow pattern is altered from its natural regime, then
the river ecosystem will change from its natural state. Too
much fl ow at the wrong time can be as damaging as too
little fl ow. The fl ow regime of a river describes the temporal
variability of run-off within a single hydrological
year (inter-annual, such as between winter fl oods, spring
snowmelt run-off, summer basefl ow), and from year-toyear
(intra-annual, such as dry years, wet years or alternating
periods or cycles of drought and higher fl ows).
Recognition of the importance of dynamism and adjustment
in fl ow regime for the physical and biotic system is
something of a recent paradigm shift in river management
and restoration (Bergen et al., 2001; Newson, 2002), but
poses areas of additional design and management uncertainty.
Traditional engineering intervention was guided by
principles of control and stability, which led to ‘certain’
or fi xed design and performance criteria. Designing-in
variability, or allowing for a degree of post-modifi cation
‘naturalisation’ to channel works is not only a novel
concept but is also largely untried or tested. It requires the
setting of generous tolerances on design functional requirements,
to refl ect the degree of ignorance and paucity of
research (Bergen et al., 2001). In this respect, ‘uncertainty’
is itself something of an essential design criterion
(also see Chapter 14)!
The most extreme changes to a fl ow regime are often
associated with construction of a dam, where the entire

108 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
pattern of fl ows may be replaced by a constant low fl ow.
This can have serious negative impacts on the river ecosystem
downstream (Petts, 1984). A key part of river restoration
is often the reduction of abstractions (Barker and
Kirmond, 1998) or release of water from dams (Acreman,
2002) to restore river fl ows to a more natural pattern.
While geomorphologists and hydrologists have tended to
emphasise connectivity and transfer of water and materials
through the fl uvial system, the ‘real’ state of catchments
in the developed world is of highly fragmented, and
largely modifi ed transfers. Graf (2001) estimates that only
2% of the 3.5 million miles of streams in the United States
are unaffected by dams, and that 18% of this is actually
under the waters of reservoirs. The picture of fragmentation,
threat and change is similar in the United Kingdom,
where less than 10% of rivers are free from structural
modifi cation of channel and banks and 53% of rivers have
fl ow regimes altered by more than 20% (Acreman, 2000).
Indeed, such is the scale of interference with natural conditions
of catchments in developed countries, that the most
convincing strategy for environmental enhancement in the
short to medium term may be to use the capacity for
regulation to remediate fl ows. The alternative, to remove
expensive, fi xed infrastructure at the larger scale (i.e.
restoration of fl ows), may require unrealistic cultural,
political and economic shifts (Graf, 2001).
Remediation of fl ows might be accomplished by changes
to the operating rules of dams, by redesign and by physical
renovation of structures. A high profi le case of relaxing
impoundment occurred on the Colorado River in the United
States in 1998 (Schmidt et al., 1998) and illustrates
Principle 1
variation in flow regime
Habitat complexity biotic diversity
discharge
Principle 4
time
natural regime discourages invasions
the importance of uniting otherwise confl icting policies
to achieve environmental benefi ts. Acreman (2002)
addresses the implications of modifi ed natural systems for
the practice of scientifi c hydrology, arguing that the realities
of modifi ed systems must be incorporated into traditional
hydrological training and practice (for review see Clifford,
2002).
The natural fl ow regime of a river is a function of the
magnitude, duration, frequency and timing of precipitation;
the form of the precipitation (rain or snow) and the
characteristics of the drainage basin (which determines
how precipitation translates into streamfl ow via surface
and subsurface run-off). Each stream has a unique fl ow
regime, characterised by the stream fl ow hydrograph. Four
principles which might be used to guide restoration or
enhancement of fl ow for ecologically motivated restoration
schemes based upon common hydrograph characteristics
are described in Figure 7.2 (Bunn and Arthrington,
2002).
From this it can be seen that almost all aspects of the
fl ow hydrograph have some importance either to individual
species at various stages of their life cycle, or to the
determination of species assemblages and hence biotic
diversity and abundance more generally. These hydrograph
components may become more or less important in
the context of both inter- and intra-annual streamfl ow
variation. Hypothesised relationships between water
years, hydrograph components and ecological processes
developed for the restoration of gravel-bed stream
environments downstream of the Sierra Nevada foothills,
California, USA, are illustrated in Table 7.1.
Principle 3
access to floodplains (lateral
connectivity; longitudinal
connectivity) essential
Principle 2
regular variation and stable
baseflows determine life
history patterns
(spawning/recruitment)
Figure 7.2 Basic principles governing ecological response to the stream hydrograph (Source: modifi ed after Bunn and Arthrington,
2002)

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 109
Table 7.1 General hypothesised relationships between hydrograph components and ecosystem processes for gravel-bed streams
downstream of the Sierra Nevada foothills (Modifi ed from San Joaquin River Restoration Study Background Report, prepared for
Friant Water Users Authority, Lindsay, CA, and Natural Resources Defense Council, San Francisco, CA, (2002). Reproduced with
permission from McBain & Trush, Inc.)
Hydrograph
component
Geomorphic-hydrologic
processes
Snowmelt peak Wetter years: bed mobility,
long duration fl oodplain
inundation, moderate
channel migration,
groundwater recharge
Normal years: bed
mobility, short duration
fl oodplain inundation
Seasonal recession
limb
Summer–fall
basefl ow
Winter–spring
rainfall run-off
Gradual decrease in water
stage, maintain
fl oodplain soil moisture
Wetter years: channel
avulsion, signifi cant
channel migration, bed
scour and deposition,
bed mobility, fl oodplain
scour, fl oodplain
inundation, fi ne
sediment deposition on
fl oodplains, large
woody debris
recruitment
Normal years: Some
channel migration,
minor bed scour, bed
mobility, fl oodplain
inundation, some fi ne
sediment deposition on
fl oodplains
Riparian processes Salmonid life-history processes
Wetter years: riparian
seedling scour within
bankfull channel, riparian
seedling initiation on
fl oodplains, discourages
riparian seedling initiation
within bankfull channel
Normal years: periodic
riparian seedling initiation
on fl oodplains
Wetter years: Allow riparian
seedling establishment on
fl oodplains
Normal and drier years:
Discourages riparian
seedling establishment on
fl oodplains by desiccating
them, encourage seedling
establishment within
bankfull channel
Encourages late seeding
riparian vegetation
initiation and
establishment within
bankfull channel
Wetter years: mature
riparian removal within
bankfull channel and
portions of fl oodplain,
scour of seedlings within
bankfull channel, seedbed
creation on fl oodplains
for new cohort initiation,
microtopography from
fl oodplain scour and fi ne
sediment deposition
Normal years: scour of
seedlings within bankfull
channel, some fi ne
sediment deposition on
fl oodplains
Wetter years: Increase juvenile growth rates by
long-term fl oodplain inundation, increase
stranding by inundating fl oodplains,
stimulate outmigration, reduce predation
mortality by reducing smolt density and
increasing turbidity
Normal years: Increase juvenile growth rates
by short-term fl oodplain inundation,
increase stranding by short-term fl oodplain
inundation, stimulate outmigration, reduce
predation mortality by reducing smolt
density and increasing turbidity
Drier years: Increase outmigration predation
mortality by increasing density and reducing
turbidity
Wetter years: Increase outmigration success by
reducing water temperatures and extending
outmigration period
Normal years: Increase outmigration success
by reducing water temperatures and
extending outmigration period
Drier years: Increase outmigration mortality by
increasing water temperatures and
shortening outmigration period
Water temperature for over-summering
juveniles and spring-run adults, immigration
for fall-run adults
Wetter years: partial loss of cohort due to redd
scour or entombment from deposition,
improve spawning gravel quality by
scouring/redepositing bed and transporting
fi ne sediment, mortality by fl ushing fry and
juveniles, mortality by stranding fry and
juveniles on fl oodplains, reduce growth
during periods of high turbidity, reduce
predation during periods of high turbidity,
creation and maintenance of high quality
aquatic habitat
Normal years: improve spawning gravel
quality by mobilising bed and transporting
fi ne sediment, low mortality by fl ushing fry
and juveniles, low mortality by stranding fry
and juveniles on fl oodplains, reduce growth
during periods of high turbidity, reduce
predation during periods of high turbidity,
maintenance of high quality aquatic habitat
Winter basefl ows Fine sediment transport Increase habitat area in natural channel
morphology

110 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
The ‘Hydrograph Component Analysis’ illustrated in
Table 7.1 enables the user to connect the geomorphic and
ecological consequences (or responses) to the results of
a hydrological analysis. For example, snowmelt peaks
and winter run-off occurring in wetter years may have
signifi cantly different consequences in terms of enhanced
channel change than those occurring in drier years.
Channel changes have associated ecological detriments in
the shorter term, but provide ecological benefi ts in the
longer term. Reducing hydrological uncertainties before
restoration (for design purposes) and after restoration (for
monitoring and appraisal purposes) thus requires the
deployment of techniques that span the determination of
low fl ows, which characterise the occurrence and magnitudes
of high fl ows, and which identify any temporal
changes or trends with respect to these. Subsequently,
linkage of these hydrological parameters to physical (sediment
transport, channel morphological) and ecological
responses may be undertaken. Whereas ‘classical’ hydrology
has focused on determining catchment ‘signatures’
with respect to the unit hydrograph (the average shape of
storm hydrographs with equal distributions of rainfall),
minimising uncertainty in restoration applications entails
a more sophisticated paradigm for application. This
requires consideration of hydrograph variability, the
potential links between this and other properties of the
catchment network, as well as knowledge of land use
management history. Collectively, it is these which defi ne
a ‘natural fl ow’ as the alternative paradigm for hydrological
assessment and river restoration schemes.
While that natural fl ow is now accepted as fundamental
to improved management and study of rivers (Richter et
al., 1996; Poff et al., 1997), it is also becoming clear that
the response of organisms to fl ood and drought may be
evidenced over different time scales, and may be fundamentally
different in character. Human modifi cations to
the fl ow regime may also alter the course and viability of
otherwise natural adaptive strategies (Lytle and Poff,
2004). Thus, while general principles or paradigms can be
identifi ed, application to specifi c regions, case studies or
species, may require considerable fi eld calibration, and/or
engender considerable uncertainty in assessing the output
of models and scenarios. The following sections describe
some of the techniques used, and identify areas of uncertainty
associated with their application and interpretation
in channel restoration designs.
7.3.2 Determining the Environmental Flow Regime
for River Restoration: Some Alternative Frameworks
There is no simple formula or fi gure that can be given for
the environmental fl ow requirements or instream fl ow
requirements of rivers. Much depends on the desired
future character of the river ecosystem under study, which
may be set by legislation or negotiated as a trade-off
between water users. The fl ow allocated to a river may thus
be primarily a matter of social choice, with science providing
technical support to help determine the river ecosystem
response under various fl ow regimes. Most scientifi c
efforts have been directed to the determination of minimum
fl ows in rivers, particularly during dry periods of the year
(or in drier years). These have been thought to be the most
important limiting factor in the ecological status or health
of the river environment. During the past 20 years, a range
of methods has been developed to help set environmental
fl ows, each with advantages and disadvantages in particular
circumstances of application.
Where fl ow data exist, methods for the determination
of minimum fl ows have been primarily statistical. Criteria
for method selection include: the type of issue (abstraction,
dam, run-of-river scheme); expertise, time and money
available; and the legislative framework within which the
fl ows must be set. Ungauged catchments give rise to particular
uncertainties, which may be reduced by constructing
regional regression curves from rivers where data do
exist, supplemented by statistical or geomorphologcal
models of response (see Smakhtin, 2001 for a comprehensive
review of most methods and applications). Most
recently, means of connecting fl ow to modelled or observed
ecological response have been developed to inform hydrological
analyses. There are thus four broad categories of
methods (Acreman and King, 2003) which may be used
to reduce uncertainty in assessing the hydrological basis
for channel restoration designs: look-up tables; desk top
analysis; functional analyses; and habitat modelling. Each
of these methods may involve different degrees of ‘expert’
involvement and may address all or just parts of the river
system. Consequently, the use of experts and the degree
to which methods are holistic are considered as crosscutting
issues. The basic principles, strengths and weaknesses
of each of these methods are outlined below.
Look-up Tables
Worldwide, the most commonly applied methods to defi ne
target river fl ows are rules-of-thumb based on simple
indices given in look-up tables. The most widely employed
indices are purely hydrological but those employing ecological
data were also developed in the 1970s. Many early
applications of environmental fl ow setting focused on
single species or single issues. For example, much of the
demand for environmental fl ows in North America and
northern Europe was from sport fi shermen concerned
about the decline in numbers of trout and salmon follow-

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 111
ing abstractions and dam operations. Environmental fl ows
were then set to maintain critical levels of habitat (including
sediment, fl ow velocity, depth) for these species. Part
of the justifi cation was that these species are very sensitive
to fl ow and if the fl ow is appropriate for them, it will be
for other parts of the ecosystem.
Engineers have traditionally used hydrologicallydefi
ned indices for water management rules to set compensation
fl ows below reservoirs and weirs. Examples are
percentages of the mean fl ow or exceedence percentiles
from a fl ow duration curve. (The fl ow duration curve is a
water resources tool that defi nes the proportion of time
that a given fl ow is equalled or exceeded). This approach
has been adopted for environmental fl ow setting to determine
simple operating rules for dams or off-take structures
where few or no local ecological data are available.
A hydrological index is used in France, where the Freshwater
Fishing Law of 1984 required that residual fl ows in
bypassed sections of river must be a minimum of 1 – 40 of the
mean fl ow for existing schemes and 1 – 10 of the mean fl ow
for new schemes (Souchon and Keith, 2001). In Brazil,
fl ows below dams must be at least 80% of minimum
monthly average fl ow (Benetti et al., 2002). In regulating
abstractions in the United Kingdom, an index of natural
low fl ow has been employed to defi ne the environmental
fl ow. Q 95 (i.e. that fl ow which is equalled or exceeded for
95% of the time) is often used. The fi gure of Q 95 was
chosen purely on hydrological grounds. However, the
implementation of this approach often includes ecological
information (Barker and Kirmond, 1998).
The Tennant Method (Tennant, 1976) was developed to
specify minimum fl ows to protect a healthy river environment.
This employed calibration data from hundreds of
rivers in the mid-Western states of the United States. Percentages
of the mean annual fl ow are specifi ed that provide
different quality habitat for fi sh, e.g. 10% for poor quality
(survival), 30% for moderate habitat (satisfactory) and
60% for excellent habitat. This approach can be used elsewhere
but the exact indices need to be re-calculated for
each new region. The indices are modifi ed where run-off
in the spring is important and are widely used in planning
at the river basin level.
Matthews and Bao (1991) concluded that methods
based on proportions of mean fl ow were not suitable for
the fl ow regimes of Texan rivers as they often resulted in
an unrealistically high fl ow. Instead, they devised a method
using variable percentages of the monthly median fl ow,
based on fi sh inventories and known life history requirements,
fl ow frequency distributions and conditions for
special periods and processes (e.g. migration).
The advantage of all look-up approaches is that once
developed, application requires relatively few resources.
However, simple hydrological indices are not readily
transferable between regions without re-calibration. Even
then, they do not take account of site-specifi c conditions.
In particular, adjusting hydrological indices does not
ensure ecological validity without a corresponding
ecological re-assessment, but ecological data may be
much more costly and time consuming to collect. In
general, look-up tables are thus particularly appropriate
for low controversy situations. They also tend to be
precautionary.
Desk Top Analysis
Desk top analyses tend to focus on analysis of existing
routine data (such as river fl ows from gauging stations
and/or fi sh data from regular surveys) although data may
be collected as part of a specifi c project at a particular site
or sites on a river. Some desk top methods are purely
hydrological. For example, Richter et al., (1996, 1997)
developed a hydrological method intended for setting
benchmark fl ows on rivers, where a natural ecosystem is
the primary objective. Development of the method relies
upon identifi cation of the components of a natural fl ow
regime, indexed by magnitude (of both high and low
fl ows), timing (indexed by monthly statistics), frequency
(number of events) and duration (indexed by moving
average minima and maxima). The method uses gauged or
modelled daily fl ows and a set of 32 indices. A range of
variation of the indices may then be set, based upon ±1
standard deviation from the mean or between the 25th and
75th percentiles. Variability in stream fl ow is essential in
sustaining ecosystem integrity (long term maintenance of
biodiversity and productivity) and resiliency (the capacity
to endure natural and human disturbances – Stanford
et al., 1996). This method is intended to defi ne interim
standards, which can be monitored and revised. However,
so far, there has not been enough research to relate the
fl ow statistics to specifi c elements of the ecosystem.
A particular aspect of desk top analysis relates to the
characterisation of high fl ows in rivers, especially the
identifi cation of groupings of higher fl ow events, either
within or between years. The emphasis on higher fl ows is
something of a counter to the previous concentration of
hydrological analyses on low fl ow requirements. It refl ects
the importance of higher magnitude fl ows in connecting
the various parts of the catchment–fl oodplain–channel
system, in sustaining sediment transport through the
fl uvial system to maintain channel and fl oodplain structure
(including larger-scale riparian woodlands – Gregory et
al., 2003) and in servicing particular stages of species’ life
cycles. In this context, fl ow variability or hydrological
disturbance is thus both a potential indicator of land use

112 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
change and an important control on river ecology. Knowledge
of the changing attributes of such disturbance is
important in setting the appropriate historical and contemporary
context for stream fl ow restoration (through consideration
of land use and climatic histories), and for
designing-in crucial aspects of fl ow variability.
By way of illustration, Archer and Newson (2002) focus
on the number and frequency of rises and falls above
selected threshold levels (pulses) expressed as multiples
of the median fl ow, together with the duration of the events
obtained when analysing the discharge history of the Coalburn
catchment, UK (Figure 7.3). The pulse duration is
the time from rising above the threshold to falling below
the same threshold, and the number of pulses refl ects the
magnitude of the chosen threshold fl ow. Provided that the
data series are extensive enough, of suffi cient resolution
(for example, daily mean fl ows are unlikely to be an adequate
basis for analysis when catchment lag is much less
than one day) and are accompanied by other appropriate
documentary/archival information, the analysis may be
extended to consider periods of land use or climatic
change, as shown in Figure 7.4.
The Coalburn is a small upland catchment with an area
of 1.5 km 2 and an altitudinal range from 270–330 m. The
natural surface material comprises a cover of blanket peat
(0.5–3 m thick) overlying glacial till up to 5 m in thickness.
This was ploughed in 1972, resulting in a drainage density
60 times greater than the original stream network. In the
spring of 1973, 90% of the catchment was planted with
Sitka spruce and, since then, growth rates have been variable,
reaching 1 m height in 1978 and 7–12 m in 1996, by
which time some 60% of the catchment had reached
canopy closure. With this knowledge of land use management,
the hydrological data can be analysed as shown in
Discharge
1 2
1
1
2
Time
Figure 7.4 to reveal phases of hydrological response to the
land use change. Thus, immediately following ploughing
and planting, pulse numbers and total pulse duration
increase, and subsequently decline with vegetation maturity,
whereas the average duration of individual events
increases as vegetation matures. These patterns in hydrological
response are most clearly shown when the analysis
is applied to events defi ned by thresholds of 2–6 times the
median fl ow. Importantly, the method further demonstrates
that, even where peak and time-to-peak of the unit hydrograph
are similar in pre- and post-disturbance situations,
there are differences in pulse numbers and pulse magnitudes,
which are thought to help determine habitat potential
and ecological response. Not surprisingly given
its emphasis on higher magnitude threshold events, the
method is least satisfactory with respect to disclosing
behaviour of low fl ows, and this limitation is important in
relation to use of the technique to ‘set’ regulated low fl ows
as described above. In such cases, it may, however, be
supplemented by other procedures and the method should
be thought of as complementing traditional approaches so
as to include hydrological variability as a key ecological
determinant.
Another area of research that may be considered under
the heading of desk top analysis (but which underpins
other methodologies described below) is the prediction of
hydrograph characteristics from catchment channel
network properties. Here, the focus of attention has been
on the effects of network scale and the relative timing
of the hillslope hydrograph and channel water routing.
Because the density of channels in a river network refl ects
climate and land use, there should be a network response
(transformation) to any changes in these driving variables
(Kirkby, 1993). The goal has been to incorporate network
2 3
3 × Median
2 × Median
Median
3
4 5
Figure 7.3 Defi nition diagram of pulses above selected thresholds and pulse duration (between arrows) in hydrological time series
(Reprinted from D. Archer et al. (2002), Journal of Hydrology 268, 244–258, with permission from Elsevier.)

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 113
Annual number of pulses
Total duration (hrs) above threshold
Average duration of flow above
80
70
60
50
40
20
10
0
5000
4500
4000
3500
3000
2500
2000
1500
1000
500
0
threshold (hrs) 30
35
30
25
20
15
10
5
0
>0.5M
>M
>2M
>3M
>4M
>0.5M
>M
>2M
>3M
>4M
>0.5M
>M
>2M
>3M
>4M
>5M
>5M
>5M
>6M
>7M
>8M
>10M
>15M
(a)
(b)
(c)
>20M
Multiples of median flow
>6M
>7M
>8M
>10M
>15M
>20M
>30M
>40M
>30M
Multiples of median flow
>6M
>7M
>8M
>10M
>15M
>20M
>30M
Multiples of median flow
>50M
>60M
1967-99
1967-71
1974-82
1983-90
1992-99
>80M
>40M
>50M
>60M
>80M
>40M
>50M
>60M
>80M
>100M
1967-99
1967-71
1974-82
1983-90
1992-99
>100M
1967-99
1967-71
1974-82
1983-90
1992-99
Figure 7.4 Pulse number (a), total pulse duration (b) and average pulse duration (c) for the Coalburn catchment over the full range
of fl ows in pre- and post-drainage and planting periods (Reprinted from D. Archer et al. (2002), Journal of Hydrology 268, 244–258,
with permission from Elsevier.)
>100M

114 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
geometry into a catchment hydrograph forecasting model
and this has given rise to the Geomorphological Unit
Hydrograph (GUH). This defi nes the average shape of
storm hydrographs with equal distributions of rainfall as
a function of bifurcation ratio, basin area ratio, stream
length ratio, watershed order and a mean waiting time for
each order (Rodriguez-Iturbe and Valdes, 1979):
and
peak discharge = 1.31/ L ΩRL 0.43
time to peak = 0.44 L Ω RB 0.55 RA −0.55 RL −0.38
where RB = bifurcation ratio; RA = basin area ratio; RL =
stream length ratio; and basin order = Ω.
The GUH may be criticised because of its dependence
on network properties that are apparently insensitive to
changes and its neglect of some aspects describing network
shape, topology and topography. More recent developments
of this approach are directed to incorporating
measures of channel network width and network link concentration
functions (for detailed review, see Mousouridis,
2001).
Other research has also related river fl ow directly to
ecological data, such as population numbers or indices
of community structure calculated from species lists.
However, it is diffi cult to derive biotic indices that are only
sensitive to fl ow and not to other factors (e.g. habitat
structure or water quality). At the minimum, biotic indices
designed for water quality monitoring purposes should be
used with extreme caution (Armitage and Petts, 1992).
Generally, such data are scarce, and interpretation is complicated
where methods of collection encompass pointspecifi
c measurements obtained at different times, and
spatially-distributed methods picking-out more persistent
patterns refl ecting longer term aspects of land use and
water quality (Dakova et al., 2000). In addition, it is not
always clear which fl ow variables to choose to represent
different aspects of the fl ow regime which are of
most ecological relevance. Studies which directly relate
‘responses’ such as macroinvertebrate abundance to
hydro-climatological and sediment variables confi rm the
importance of intermediate levels of disturbance as underpinning
greatest habitat diversity, and thus imply that
modelling must incorporate more than one aspect of
hydrological diversity (Reice et al., 1990).
The lack of complementary hydrological and biological
data is often a limiting factor, and sometimes routinely
collected data gathered for other purposes turn out to be
unsuitable. In addition, time series may not be independent
(which can violate assumptions of classical statistical
techniques) while the representation and characterisation
of disturbances (‘extremes’) necessitates longer, higherquality
time series. For example, Clausen and Biggs (2000)
examined 35 fl ow variables using daily mean fl ows for a
seven-year record common to 62 perennial rivers in New
Zealand. Based upon a covariance analysis among the sites
through a principal components analysis, the 35 variables
could be collapsed into four variable groupings relating
to: size of river; overall variability of the fl ow; volume of
high fl ow; and frequency of high fl ow. Signifi cantly, the
statistical properties (particularly intra-annual variability)
varied between groups, necessitating the use of a suite of
different variables from each group to adequately represent
the facets of fl ow of most ecological relevance. This illustrates
the requirements for high quality data and, too, the
uncertainties of interpretation and application.
A method recently developed in the United Kingdom
involving ecological data is the Lotic Invertebrate Index
for Flow Evaluation (LIFE; Extence et al., 1999). The
LIFE score is based on the abundance and sensitivity to
water velocity of different taxa, collected in routine macroinvertebrate
monitoring data. Moving averages of preceding
fl ow have shown good relationships with LIFE scores
over a range of sites (Figure 7.5), but it is as yet uncertain
as to how the approach can be used to manage river fl ows.
Nevertheless, the principle is believed to be sound and
LIFE has the major advantage of utilising the data collected
by existing bio-monitoring programmes.
Functional Analysis Methods – the Building
Block Methodology
The third group of methods to determine fl ow requirements
for restoration and ecological purposes includes
those that build an understanding of the functional links
between hydrology and ecology in the entire river system.
These methods take a broad view and combine hydrological
analysis, hydraulic rating information (to estimate
sediment transport and channel capacity) and many aspects
of the river ecosystem. Perhaps the best known is the
Building Block Methodology (BBM) developed in South
Africa (Rowntree and Wadeson, 1998; King et al., 2000).
The basic premise of the BBM is that riverine species are
reliant on three basic elements (building blocks) of the
fl ow regime, each of which has a particular ecological and
geomorphological signifi cance: low fl ows, freshets and
fl oods. An acceptable fl ow regime for ecosystem maintenance
can thus be constructed by combining these building
blocks. In a typical exercise to determine instream fl ow
requirements through the building block method, fi ve
stages may be involved (Rowntree and Wadeson, 1998): a
general assessment of catchment condition to determine

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 115
Discharge (m³/s)
10.00
1.00
0.10
0.01
0.00
6.0
86 87 88 88 89 90 91 92 93 94 95 96 97 98 99 00
Year
Flow
Figure 7.5 Example River Flow (logarithmic scale) and LIFE Score time series
potential for geomorphological change in a river; an evaluation
of river network characteristics to help identify representative
river reaches to determine the instream fl ow
requirements; an assessment of the hydraulic diversity at
various stages for each of the morphological units present
in these reaches (see below for discussion of biotopes); an
assessment of the magnitude and importance of freshet
and fl oods; and an assessment of any inter-basin water
transfers.
The BBM thus revolves around a team of physical scientists
(e.g. hydrologists and geomorphologists) and biological
scientists (e.g. botanists and fi sh biologists) who
follow a series of structured analyses to come to a consensus
on the building blocks of the fl ow regime. The BBM
has a detailed manual for implementation (King et al.,
2000); is routinely used in South Africa to comply with
the 1998 Water Act; has been applied in Australia (Arthington
and Long 1997, Arthington and Lloyd, 1998); and is
being trialled in the United States.
To overcome the diffi culties in relating changes in the
fl ow regime directly to the response of multiple species
and communities, approaches have been developed that
use habitat for target species as an intermediate step. Of
those environmental conditions required by an individual
animal living in a river, it is the physical aspects that are
most heavily impacted by changes to the fl ow regime.
Habitat models seek to link data on the physical conditions
(such as water depths and velocities) in rivers at different
fl ows (either measured data or derived from computer
models) with data on those physical conditions which key
animal or plant species (or their individual developmental
stages) require. Once functional relationships between
physical habitat and ecology have been defi ned, they are
linked to fl ow scenarios in river restoration and design (see
Section 7.3 for further details on habitat modelling).
Life Score
8.0
7.5
7.0
6.5
LIFE Score
The Holistic Approach
Where intervention seeks fl ow remediation or restoration,
there is an implicitly holistic approach insofar as all elements
of the river ecosystem are likely to be supported.
However, more and more methods now adopt an overtly
holistic assessment of the whole ecosystem, such as
associated wetlands, groundwater and estuaries; all
species that are sensitive to fl ow (invertebrates, plants and
animals); the human context; and all aspects of the hydrological
regime, including fl oods, droughts, and water
quality. A fundamental principle is to maintain natural
variability of fl ows and to allocate fl ow based upon likely
habitat impact. This ecosystem approach is especially
important in the management and restoration of dryland
environments, where inherent variability of hydrological
‘extremes’ is a limiting factor ecologically (Thoms and
Sheldon, 2002). Ecosystem approaches inherently allow
for complexity and diversity, and may also capitalise on
features such as persistence and evolution – ecological
systems may have self-organising or self-designing
attributes (Bergen et al., 2001).
Generally, holistic approaches make use of teams of
experts and may also involve the participation of stakeholders,
thus extending holism beyond scientifi c issues.
Where methods have the characteristic of being holistic
they clearly have the advantage of covering the whole
hydrological–ecological–stakeholder system. The disadvantage
is that it is diffi cult to identify and to collect the
relevant data which capture such diverse perspectives.
7.3.3 Choice of Method
Selection of the most appropriate method for linking
hydrological criteria to river restoration designs, where a

116 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Table 7.2 Some advantages and disadvantages of different methods and characteristics of setting environmental fl ows
Method type Sub-type Advantages Disadvantages and Uncertainties
Look-up table Hydrological Cheap, rapid to use once calculated Not site-specifi c. Hydrological indices not
valid ecologically
Ecological Ecological indices need region-specifi c data
to be calculated
Desk top Hydrological Site specifi c
Limited new data collection
Long time series required
Hydraulic No explicit use of ecological data
Ecological Ecological data time consuming to collect
Functional analysis Flexible, robust, more focused on
whole ecosystem
range (or set) of methods is available, is considered briefl y
here. Some of the advantages and disadvantages which
underlie the uncertainties in using different approaches are
summarised in Table 7.2.
Broadly, Table 7.2 illustrates two trade-offs: that between
local sensitivity or regional coverage, and that between
data availability and complexity or simplicity of data interpretation.
Moving through the table, the more integrated
and hence more complex methods are characterised by a
greater degree of connection between physical and biological
parameters. The key element of this integration is to
describe or model the effects of fl ows on habitat structure
and function. In turn, this necessitates the addition of
hydraulic and hydrodynamic considerations to supplement
the hydrological foundations described above.
7.4 HYDRAULIC ASPECTS AND UNCERTAINTY
IN CHANNEL RESTORATION
7.4.1 Connecting Flows, Sediments and Ecological
Response: the Physical Habitat and Eco-Hydraulics
Over the past decade, there has been a trend towards
approaching river habitat assessment and rehabilitation
design using combinations of fi eld survey and predictive
hydraulic models (Chapter 5). ‘Eco-hydraulics’ (Leclerc,
2002) is now a commonplace term in both the academic
and practitioner literature, and the linkage between physical,
chemical and biotic components of the river environment
is central in efforts to restore and maintain ecological
habitat and function (Kemp et al., 2000). To realise these
benefi ts, however, cost-effective (but scientifi cally sound)
means of combining traditional fi eld monitoring and
Expensive to collect all relevant data and to
employ wide range of experts. Consensus
of experts may not be achieved.
Habitat modelling Replicable, predictive Expensive to collect hydraulic and ecological
data
survey with emerging modelling and design-support
approaches are required. There remains much work to be
done both to encourage multi-disciplinary co-operation
and to ‘invent’ new trans-disciplinary areas of research
and practical expertise (Janauer, 2000).
Physical habitat can be defi ned as a set of physical
conditions that can be measured and compared to the
conditions that may be suitable for specifi c species or
individuals at a particular stage of their life cycle. The
fundamental aspects of the eco-hydraulic approach to river
characterisation and restoration are summarised in Figure
7.6. This is based upon the idea that species assemblages
and/or abundances are organised to refl ect physical conditions
at a range of scales: catchment conditions (such as
slope, geology and rainfall regime) determine at the
most fundamental level the kind of species which may be
present; at the reach scale, assemblages or abundances
from within this broader range are most likely to be
observed depending on fl ow regime; at the sub-reach
scale, particular species or individuals at particular stages
of their life cycle, are found in niches with particular fl ow
and sedimentological conditions.
While the catchment scale conditions the range of
habitat potential, at the smaller scales physical habitat is
commonly defi ned by combinations of depth, velocity,
substrate and distance to cover. In this respect, physical
habitat quality is normally considered independently of
water quality issues. It may be approached (at its simplest)
from visual survey, through statistical models based upon
measured associations, and via complex deterministic
modelling, in which the physical habitat is fi rst modelled
and then linked to ecological response. Although the
potential of eco-hydraulic modelling has been generally

DRAINAGE
BASIN
10 6 -10 5 years
Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 117
10 4 -10 3 m
10 3 -10 2 m
FLOODPLAIN
10 4 -10 3 years
10 2 -10 1 m
REACH
10 2 -10 1 years
HABITAT
10 1 -10 0 years
10 –1 m
Sand silt
over cobbles
Gravel
MICROHABITAT
10 0 -10 –1 years
Aquatic and
semi-aquatic
vegetation
Leaf and stick
detritus in
margin
Figure 7.6 Spatial scale and distribution of stream habitats based upon fl ow–sediment interactions (Source: Naiman et al., 1992
after Frissell et al., 1986)
Fall
Rapid
Marginal
Deadwater
Chute
acknowledged positively, some signifi cant areas of uncertainty
remain. A fundamental requirement is the ability
to demonstrate (rather than assume) a linkage between
measured parameters delimiting physical habitat and
the species occurrence and abundances (‘responses’ or
‘tolerances’) which these are supposed to condition.
7.4.2 Survey and Visual Methods: Biotopes and
Functional Habitats
At present, hydraulic performance and habitat are essentially
approached separately. Two alternatives are available:
‘bottom up’, in which in-stream habitat units are
PHYSICAL BIOTOPES FUNCTIONAL (MESO-) HABITATS
Riffle
Boll
Run
glide
Pool
Floating
leaves
Submerged
plants
(slow)
Cobbles/
pobbles
Rock Woody
debris
Tree roots/
overhanging
vegetation
(fast)
Gravel Emergent
plants
Sand
Silt
Marginal
plants
Figure 7.7 Representation of physical biotopes and functional habitats in a stream sub-reach (Reproduced from M. D. Newson and
C. L. Newson (2000) ‘Geomorphology, ecology and river channel habitat: mesoscale approaches to basin-scale challenges’ Progress
in Physical Geography 24, 195–217, with the permission of Edward Arnold (Publishers) Ltd.)
inferred from a knowledge of hydraulic conditions that
defi ne physical biotopes (channel features characterised
through hydraulic measurement or visual survey of surface
fl ow type; Padmore, 1997); and ‘top down’, in which
functional or meso- habitats are inferred from analysis of
biological communities associated with substrate and vegetation
characteristics (Kemp et al., 1999). A common
strategy is to visually classify velocity–depth combinations
(i.e. the biotopes) which might then be associated
with functional habitats, or discrete species assemblages
as indicated in Figure 7.7 (Newson and Newson, 2000).
Biotopes are easily recognised from fi eld survey
without costly fi eld measurement and, when linked with

118 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
knowledge of habitat suitability, allow some insight into
likely biological function. Despite early optimism with
respect to this approach, relating biotopes to meaningful
biotic function appears more complicated than fi rst
assumed (Kemp et al., 2000; Leclerc, 2002). Demonstrable
relationships between biotopes and functional habitats
remain scarce. While the survey or biotope method is
meant to circumvent issues relating to single-species preference
curves or single measures of channel environment
(since the biotope is an ‘integrated’ habitat), major uncertainties
in improving links between biotopes (from simple
survey or modelling) and functional habitats remain
(Newson and Newson, 2000). In part, the scarcity of data
refl ects the cost and practical limitations associated with
fi ne-scale fi eld survey (Newson et al., 1998) and diffi culties
in monitoring highly dynamic, stage-dependent fl ow
characteristics. There is also the question of the appropriate
range of supposed habitat determinants to associate
with particular species or species clusters, at differing
stages of their life cycles, and which may vary greatly
between species found even in close proximity. Key questions
which belie the uncertainty in an otherwise appealing
approach are: how consistent are our observations?
How stable are biotopes with stage? How do biotopes
relate to channel morphology and biology?
Biotope mapping has been supported by ecological
observations that species patchiness (superimposed on a
general downstream continuum) does occur, and that this
may further be related to discharge exceedence values
(Newson and Newson, 2000). This suggests that there
should be a link with similar patchiness of functional
habitat, if this, too, can be defi ned. Functional habits may
be discriminated on the basis of combined fl ow indices
such as the Froude number. (The Froude number is the
ratio of depth to the square root of depth multiplied by
gravitational acceleration). From a survey of 32 sites in
eastern England, Kemp et al. (2000) suggest that functional
habits fall into two groups: lowest Froude number
classes, and those with Froude number above 0.5. Insofar
as the Froude number itself is assumed to refl ect biotopes
derived from measurement or observation, then a connection
between biotopes and functional habitats should exist,
but the ecological signifi cance of this link is uncertain in
the absence of simultaneous, point-by-point measures of
physical and ecological indicators, and because of feedbacks
or interdependence between the apparently independent
classifi cation variables. For example, vegetation
growth affects velocities and sedimentation, and hence
helps determine the local Froude number, which is supposedly
independent. A recent investigation by Clifford
et al. (2006) identifi es the need for more consistent use of
bitope and habitat defi nitions, and improved experimental
design when assessing the linkage between fl ows, habitats
and ecology. Use of the Froude number may also be questioned
because it may obscure differences between very
different fl ow–depth combinations.
Figure 7.8 illustrates another basic area of uncertainty
in the biotope and physical habitat approach, but also
perhaps one way in which this uncertainty might be
managed or, in future, reduced. The fi gures show reaches
of the River Cole, near Birmingham, UK, and the River
Tern, Shropshire, UK, which have been ‘classifi ed’ into
zones of statistically similar velocity and/or depth behaviour
at various fl ow stages (Emery et al., 2003). Both
rivers have a well-marked riffl e-pool bedform sequence (a
common biotope), but the Cole is a less sinuous channel,
with more regular bank morphology and less bankside tree
growth. The approach uses cluster analysis (a hierarchical
statistical method of association) as a reproducible means
of assessing both the degree, location and persistence of
clustering or patchiness of physical habitat, which might
support more objective identifi cation of biotopes.
Cluster analysis is an agglomerative process to identify
homogeneous groupings (patches) based upon selected
characteristics. In this case, standardised fi eld velocity
(measured velocity scaled according to the mean and variance
of the total measured distribution) was used. Initially,
all velocity observations were considered as separate, but
were then successively combined according to ‘distance’
(i.e. differences) from neighbours and from emerging
clusters. The next stage was to assess cluster number in
relation to morphology, spatial coverage and stability.
ANOVA (analysis of variance testing) was used to test
differing cluster numbers for statistically signifi cant differences
at each fl ow stage and to assess their coverage
and coherence as fl ow stage varied. The cluster analysis is
therefore a means of defi ning biotopes on the basis of fi eld
measurements, rather than on the basis of visual survey.
Figures 7.8(a) and (b) show the spatial coverage and location
of the six distinct velocity clusters at higher fl ow stage
resulting from the analysis, while Figures 7.8(c) and (d)
detail the velocity characteristics associated with each
cluster over a range of fl ow changes from low fl ow to a
high in-bank fl ow.
Several points emerge from this analysis which indicate
that the method might help both understand and, in time,
reduce uncertainties in habitat recognition and appraisal.
The method is clearly sensitive to bedform amplitude and
planform complexity. Even where the basic biotope structure
of rivers is similar, the differing patterns of coverage
are suffi cient to indicate a degree of ‘biotope subtlety’ –
contrast Figures 7.8(a), where fl ow organisation at high
stage is marked by linearity, and 7.8(b), where patchiness
is dominant. This is supported by the stage-dependent

Northing (m)
Velocity (m/s)
1060
1040
1020
1000
980
960
940
920
Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 119
Cluster1
Cluster2
Cluster3
Cluster4
Cluster5
Cluster6
Northing (m)
1070
1060
1050
1040
1030
1020
Cluster1
Cluster2
Cluster3
Cluster4
Cluster5
Cluster6
900
4990 5000 5010 5020
Easting (m)
5030 5040
1010
4985 4990 4995
Easting (m)
5000 5005
(a) (b)
1
0.8
0.9
0.8
0.7
0.7
0.6
0.6
0.5
0.5
0.4
0.4
0.3
0.3
0.2
0.2
0.1
0.1
0
Low ‘x’ V mean Med ‘x’ V mean High ‘x’ V mean
0
Low ‘x’ V mean Med ‘x’ V mean High ‘x’ V mean
Stage class Stage class
(c) (d)
Velocity (m/s)
Figure 7.8 Results of cluster analysis to determine coverage and behaviour of velocity ‘patches’ in the Rivers Cole, Birmingham (a
and c) and Tern, Shropshire (b and d) (Source: Modifi ed from Emery, 2003)
velocity behaviour seen in Figures 7.8(c) and (d). In the
River Cole, the straighter, simpler channel, patch coherence
increases and cluster number decreases (coverage
increases) as fl ow stage rises. Thus, the river ‘simplifi es’
with increasing fl ow stage into three basic groupings:
channel margins, channel centreline, and areas above
bedform crests. However, in the River Tern, the lower
stage complexity is largely maintained as fl ow stage rises,
possibly because the fl ow now interacts with large bankside
tree roots creating new or maintaining old, biotopes
(for further discussion, see Emery et al., 2003).
The other key points emerging from this analysis are
equally important and draw attention to some pitfalls in
identifi cation and representation, which might improve
survey methods in the future. First, in assessing biotopes
and functional habitats, the entire fl ow regime should be
considered – what exists at low fl ow may or may not
change in character as stage rises. Second, there is an
obvious role for hydraulic and hydrodynamic fl ow modelling.
This might support fi eld observations (which may be
limited to single fl ows) and assist in restoration design and
appraisal, where the ability to predict and to visualise

120 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
changing aspects of fl ow and habitat performance as fl ow
stage varies is crucial.
7.4.3 Hydraulic Habitat Simulation and Modelling
Assessment of river fl ow management options, including
river restoration and changes in fl ow regime, often involves
assessing scenarios that fall outside the range of observed
conditions. This negates investigation using direct analysis
of fi eld observations and necessitates the use of predictive
models. Models employ a variety of hydrological, morphological
and hydraulic parameters to predict variations,
which might then be related to the abundance and distribution
of aquatic organisms. The parameters vary according
to the spatial scale of the simulation. Habitat modelling
frequently involves two steps: the physical analysis and
evaluation of the hydraulic and morphological aspects of
the problem, and the subsequent linkage to ecological
response.
Uncertainties in the Hydraulic and Hydrodynamic
Modelling Process: the Fundamentals
All hydraulic and hydrodynamic fl ow simulation models
begin with equations for the conservation of mass and
momentum, which are modifi ed for the bed and surface
boundary conditions found in shallow, open channel fl ow.
As a result of limitations imposed by the scale of the
smaller turbulent motions, the governing equations are not
tractable in their original form and require additional
modifi cations or ‘closure models’ to provide computational
outputs. These assumptions in turn give rise to new
terms in the governing equations, which are themselves
subsequently manipulated to satisfy the requirements of
the solution, which may be represented as a one-, two- or
three-dimensional numerical scheme (for full review, see
Lane, 1998). Generally, in eco-hydraulic and river restoration
applications, the uncertainties in model derivation are
secondary to those of model application to the particular
case study. In effect, modelling is really a recursive
process, involving uncertainties at all stages from premodel
data collection, through model verifi cation during
the numerical calculation, to post-modelling validation,
where results are compared with fi eld measurements and
other sources of evidence/expertise to assess, or appraise
the model performance.
Traditionally, hydraulic modelling has been focused on
one-dimensional representation of open channel fl ow
(Chow, 1973), in which the cross-sectionally averaged
depth and velocity are obtained. These models may be
used to simulate the passage of fl oods through reaches and
channel systems, or for the assessment of channel stability
given particular boundary characteristics (see Chapter 5
for more details). In ecological applications, they may be
of use to determine depth and timing of inundation associated
with particular events. Hydrographs are routed
through cross-sections either by running a series of steadystate
solutions for different discharge ‘bins’ in the hydrograph
or using unsteady methods to model the hydrograph.
Common one-dimensional implementations of this
sort are iSIS, HEC–RAS and MIKE11, while the most
popular hydraulically-coupled habitat suitability model –
PHABSIM (Spence and Hickley, 2000, and below) – uses
one-dimenstional approaches to predict velocity and depth
in channel cells or slices, defi ned between adjacent measured
cross-sections. In their simplest forms, these models
rely on the ability to specify the relationship between
depth, water surface slope and velocity via an empirical
‘constant’ known as Manning’s n, which is normally
assumed to represent the fl ow resistance arising from
boundary friction of the channel. The value of n is usually
determined from look-up tables (e.g. Chow, 1973). Coeffi
cients of channel expansion and contraction may be used
in addition to channel geometry to account for additional
components of energy loss arising from fl ow acceleration
and deceleration. Away from channels of very simple
geometry and boundary characteristics, the appropriate
determination of n is more subtle: to maintain predictive
success, for example, n is varied inversely with depth and,
in many situations, n is really a compound calibration
factor expressing (with more-or-less success) the various
contributions to fl ow resistance, and hence energy loss in
channels: grain, bedform, and spill (channel curvature).
Some of the issues in more complex modelling are outlined
below.
Two- and three-dimensional Numerical Schemes in
Ecological Restoration Applications
Ideally, two- and three-dimensional approaches are required
for realistic simulation of fl ow behaviour and to better represent
the diversity and variability of physical habitats
(Crowder and Diplas, 2000), thus overcoming the limitations
of popular habitat simulation models. In particular,
accounting for small scale stage-dependent variations in
fl ow is important in the maintenance and survival of the
biotic community. In this way, more complex numerical
fl ow simulation offers potential as a ‘supporting’ function
for river rehabilitation schemes. Two- and threedimensional
codes are now widely accessible and can be
run over a wide range of discharge conditions, including
over-bank fl ows, but there are numerous sources of uncertainty
at all stages of the modelling process associated with
these simulations. Models involve many underlying

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 121
assumptions at their input stage, they are sensitive to the
numerical structures involved in obtaining the fl ow solution
and outputs may be diffi cult to observe or measure in
the fi eld. Some of the more common models, model applications
and model uncertainties are listed in Table 7.3.
Perhaps the most basic but much neglected area of
uncertainty in the application of more complex models
rests with the fi eld data requirements for model calibration
and validation. This is most onerous where modelling is
used at a spatial resolution commensurate with sub-reach
detail in habitat in conditions of complex topography
(Crowder and Diplas, 2000). These have led some to argue
that the ecological and hydraulic/hydrodynamic approaches
should proceed independently, because river environments
are ‘too complex’ for a coupled approach (Kondolf et al.,
2000). Field observations indicate that physical habitat
may be structured at an extremely small scale, such as
around boulders, or extremely close to channel banks
(Railsback et al., 1999), which necessitate huge increases
in model effort and computational time, and whose characteristics
may not, in any case, be picked up by conventional
fi eld survey. Figure 7.9 for example, illustrates how
our knowledge of the character of the river changes as the
number and density of fi eld measurements at various fl ow
stages is increased or decreased.
Here, the results of an original fi eld survey of more than
350 data points obtained by survey through a 150 m river
reach of the River Cole (representing both longitudinal
and cross section variation with an average point-to-point
spacing of 1–2 m) have been plotted as a frequency histogram
of velocities, and then subsequently re-plotted in the
same way but with the original dataset systematically
degraded by a factor of one half or one third. The resultant
changes thus give an idea of the effects of varying the fi eld
data collection efforts or schemes in characterising ‘true’
fi eld velocities, both prior to modelling simulation and in
post-modelling assessment. Several points are worthy of
note from this analysis. Firstly, channel velocity characteristics
differ depending upon fl ow stage: at lower fl ow
stage, results are more skewed than at higher stage, but at
the higher stage there is the suggestion of multiple modes
in an otherwise more even velocity distribution. This
change in velocity distribution should be interpreted in
conjunction with the results of the biotope analysis in
Section 7.4.2, since the fi eld data are common to both. In
this case, it seems that, as biotopes ‘clarify’, the distribution
of velocities also becomes organised about dominant
peaks, refl ecting the infl uence of channel margins, channel
centrelines and shallows over bedform crests. The second
point to note is that sub-sampling from the original data
Table 7.3 Common numerical fl ow models, their principal applications in river restoration and eco-hydraulics, and some issues
relating to parameterisation and model interpretation
Class of model
and common
implementation
One-dimensionnal
HEC-RAS
ISIS
backwater
Two-dimensionnal
Telemac
RMA
Three-dimensionnal
SSIIM
Fluent
CFX
Principal application in
channel restoration
fl ood routing and channel
conveyance
initial water surface
elevation (in higher order
models); velocity and
depth between cross
sections in PHABSIM
dynamic simulation of
channel fl ow and
fl oodplain inundation;
cross-sectional patterns
of habitat suitability
detailed sub-reach
modelling of habitat
suitability
Required parameters Principal aspects of uncertainty
downstream discharge;
upstream water level;
Manning’s n; channel
expansion and contraction
coeffi cients; basic channel
cross section morphology
boundary roughness (ks);
fl ow information as above;
bed topography; turbulence
closure model; choice of
numerical solver and
relaxation coeffi cients;
correct estimation of n (particularly with
depth and vegetation changes) and
channel expansion and contraction
coeffi cients; loss of information on
cross-sectional fl ow distribution; no
account of channel curvature
loss of information on depth-related fl ow
properties; time-stepping in unsteady
solutions; poor representation of
secondary fl ow from channel
discontinuities; meshing issues;
representation of wetting and drying
as above most of the above, plus: meshdependence
in the vertical as well as
cross-section; determination of the free
water surface

122 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Frequency (%)
Frequency (%)
30
25
20
15
10
5
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
30
25
20
15
10
5
Resultant velocity (m 3 s –1 )
(a)
full data set
half data set
third data set
0
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1 1.2 1.3 1.4
Resultant velocity (m 3 s –1 )
(b)
full data set
half data set
third data set
Figure 7.9 Effect of degrading the velocity sample on the distribution of depth average velocities: a) low fl ow; b) high in-bank
fl ow

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 123
does not have an even effect in all velocity classes. This
may or may not be of signifi cance but is likely to impart
uncertainty into the assessment of model performance if
the affected velocity classes are those judged to be of most
relevance for habitat suitability. This arises because changing
the frequency of particular velocity class observations
is equivalent to under- or over-representing the spatial
coverage associated with that velocity. For further discussion,
see Clifford et al. (2002a, 2002b).
Once acceptable fi eld velocity data have been obtained,
these must be coupled with representations of boundary
topography and channel geometry. This should be suffi -
ciently detailed to incorporate both small-scale irregularities
in bed and banks, the major breaks of slope associated
with channel margins and more subtle fl oodplain topography.
Breaklines are notoriously diffi cult to handle, and if
fi ne-scale survey is unavailable then topography has to be
interpolated, introducing a further source of uncertainty,
which also occurs in areas of low relief, too. Simple linear
interpolation, for example, may be preferable to avoid
spurious high- and low-points (French and Clifford, 2000).
Topography forms the template of the modelling grid or
mesh which is draped over the boundary surface and which
forms the basis for the numerical solution to the governing
fl ow equations. The form of the mesh depends upon the
kind of model scheme adopted. Meshes might exploit triangular
(in fi nite element schemes) or rectangular elements
(in fi nite difference and fi nite volume schemes), of
either constant or variable size. Triangular meshes are
more versatile, but most ecological applications use rectangular
grid elements and structured grid schemes. While
allowing elements to vary in size, these always require the
same number of elements in each cross plane of solution.
In practice, the constraints of gridding normally require
some degradation in the quality of the topographic representation,
since multiplying the number of very small elements
to represent more detailed topographic variation
imparts a disproportionately large increment in computer
time required to solve the problem and may result in
instabilities in the solution. Many of these issues receive
a thorough exploration in Anderson et al. (1996). As computer
power has increased, more detailed modelling is
possible, but approaching closer representations of reality
does not simply depend upon enhanced computer speed
or memory. A key issue is that, with smaller elements, bed
grains and micro-topography may approach or exceed the
size of model elements, rendering solution impossible.
This is especially pertinent in two-dimensional modelling,
where the number and size of elements in the vertical may
also be varied, and the issue is related to the wider question
of the appropriate parameterisation of multiple scales of
fl ow resistance (Nicholas, 2001).
With respect to fl ow resistance, most two-dimensional
models often remain reliant on a boundary resistance
value specifi ed by n or by Chezy’s conveyance coeffi cient,
C. Three-dimensional models employ a boundary roughness
specifi ed in terms of an equivalent roughness length,
k s, which propagates through the boundary cells. While ks
can be determined in relation to the boundary grain size,
the numerical value required to obtain plausible solutions
is generally much larger than grain size alone would imply,
giving rise to uncertainty in its interpretation. Flow resistance
effects beyond skin friction may be incorporated
wittingly or unwittingly into the meshing strategy – for
example, by varying the size of cells around the boundary
as compared to the main body of the fl ow (Clifford et al.,
2002a). Other areas of uncertainty in both model application
and result interpretation include: whether the model
uses a fi xed water surface (lid) or allows the water
surface to freely adjust; the scheme used to allow the
simulation of turbulence effects (the closure assumptions);
and the complexity of the solver (which determines how
many and which neighbouring points are considered in the
propagation of the numerical solution through the mesh).
Many of these issues are common to both two- and threedimensional
schemes, although similar strategies and
choices will not necessarily have the same effects in the
differing model implementations. Nicholas (2001) provides
a comprehensive starting point for assessing and
managing these uncertainties. The particular limitation of
two-dimentional models is the parameterisation of secondary
circulation effects on momentum transport, which
has concentrated on the effects of channel curvature, but
not on topographic discontinuities and river confl uences
(Lane, 1998).
Rarely do the uncertainties arising with respect to each
aspect of the modelling occur independently: Figure 7.10
for example, illustrates the changing fl ow representations
arising from coarser- and fi ner-scale element sizes using
the SSIMM model of the River Cole, near Birmingham,
UK. Apart from the element size, all other model parameters
are held constant. As the results demonstrate, modelled
velocities vary in both planform and in cross-section
and vertical distributions as grid element size is varied
from approximately 2–0.5 m. The coarse grid distorts
the channel shape from generally rectangular to a trapezoidal
cross section and this contributes to biasing the faster
fl ow towards the channel centre, raising velocities above
those measured in the fi eld. This effect is also enhanced
by the interaction of grid element size and boundary
roughness, which in the model extends to 20% of the near
boundary grid element. When elements are large, a larger
proportion of the channel is thus affected by the boundary
roughness, even though the numerical value of k s is the

124 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
HIGH flow
coarse grid
Pool
Pool
1.4
1.3
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
1.4
1.3
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
Modelled
velocities
Modelled
velocities
Pool
HIGH flow
fine grid
Pool
Riffle crest
Riffle crest
(m)
(m)
1060
1040
1020
1000
980
960
940
1060
1040
1020
1000
980
960
940
N
920
4980 5000
(m)
5020 5040
N
PT 4
Flow
PT 4
Flow
PT 2
PT 2
RC 4
PT 3
RC 4
RC 2
PT 3
RC 2
RC 3
RC 3
PT 1
RC 1
PT 1
RC 1
0.3
0.25
0.2
0.15
0.1
0.05
0
-0.05
-0.1
-0.15
-0.2
-0.25
-0.3
0.3
0.25
0.2
0.15
0.1
0.05
0
-0.05
-0.1
-0.15
-0.2
-0.25
-0.3
920
4980 5000
(m)
5020 5040
Modelledfield
velocities
Modelledfield
velocities
Figure 7.10 (See also colour plate section) Changing SSIIM model output and model ‘fi t’ with as a function of mesh element size
in a reach of the River Cole, near Birmingham, UK

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 125
same. This effective increase in friction combined with the
reduced area of the channel in the coarser grid scheme
‘push’ the fl ow to the centreline. In the streamwise direction,
the coarser grid resolution is also insuffi cient to represent
the ‘patchiness’ of velocity seen in the fi ner grid
case and, hence, both misses and distorts the spatial distribution
of potential physical habitat. By contrast, use of
the fi ner mesh over-emphasises velocity patchiness and
gives rise to problems of near-bank velocity representation.
Here, apparently high velocity values cannot be corroborated
with fi eld data which were collected at coarser
resolution.
Some models allow adaptive grids which vary in complexity
as the problem evolves. The most fundamental
illustration of this is the ability (or otherwise) to cope with
wetting and drying, which will occur as fl ow stage rises
and inundates fl oodplain or channel margins, before receding.
Full simulation of this requires an unsteady solution
coupled to one of several schemes to either include or
exclude fully or partially dry elements. Dynamic solutions
are generally confi ned to two-dimensional schemes. A
simpler solution is to produce a number of grids, each of
which is fully wet for the particular fl ow stage under consideration.
In analogy with one-dimensional models, the
hydrograph may then be simulated from a series of steady
state solutions.
Much of the uncertainty in the modelling process may
be summarised by the terms verifi cation and validation.
These are fundamental considerations in numerical fl ow
modelling but have recently been re-examined as the
number of models and range of model application has
burgeoned. Verifi cation relates to the ability to correctly
solve the appropriate equations of the model problem at
hand (i.e. whether the solution is an accurate solution of
the particular choice of model); whereas validation relates
to the plausibility of the model as a whole and the testing
of parameters predicted by the model (thus involving
the degree of fi t between prediction and measurement).
As model applications have increased, the distinctions
between the two have become more blurred, and the term
model assessment or appraisal has been used (Lane and
Richards, 2001). Hardy et al. (2003) follow conventional
engineering practice, arguing that verifi cation is the essential
element in the assessment process, which must precede
attempts at validation. However, relatively little attention
has been given to developing means of model assessment
that are tailored to the application, particularly with respect
to eco-hydraulics. In environmental systems, for example,
the degree of experimental closure is much less than in
laboratory conditions or engineering problems, and the
application of models is as much (if not more) to produce
data to support or inform new fi eld strategies or explana-
tory interpretations/insights (Oreskes et al., 1994) as is it
is to judge the correspondence between measured and
modelled values. Thus, in ecological applications, where
the uncertainties of measurements (both physical and biological)
are coupled with an intrinsic dynamism and variability,
Clifford et al. (2005) have argued that assessment
of models necessarily requires a more ‘relaxed’ approach.
This is illustrated in Figure 7.11, where planform representations
of model:fi eld correspondence of velocities for
the River Cole are shown.
In Figure 7.11(a), the fi t of the model is shown for a
higher fl ow stage, purely as the absolute value of the pointby-point
modelled-fi eld velocity. On this basis, approximately
80% of the channel area is modelled to within
±0.2 ms −1 , and approximately 40% to within ±0.1 ms −1
(Figure 7.11(c)). Figure 7.11(b), however, shows the
improvement in fi t when a degree of ‘relaxation’ is applied.
In this case, velocity ‘errors’ were recoded to zero
provided that two criteria were met: fi rst, the modelled
velocity must lie within 0.1 ms −1 of the fi eld velocity and,
second, this 0.1 ms −1 difference must also lie within a 1 m
radius of the actual fi eld measurement. By allowing this
relaxation, the fi t of the model is dramatically improved:
reference to Figure 7.11(c) shows that only approximately
5% of the channel fails to fulfi l these joint criteria. The
justifi cation for the relaxation lies in the fact that: (a) fi eld
measurements are rarely precise enough to warrant absolute
comparison to modelled results, particularly where
both fi eld and modelled velocities have been interpolated
to the same grid for comparison; (b) in ecological applications,
the uncertainties in relating physical parameters to
ecological response are large; and (c) in any case plant and
animal communities are mobile, dynamic communities,
rather than static and adapt to exploit the most favourable
environments. Ecological simulation contains, then, an
essential uncertainty, which might usefully be incorporated
into numerical model assessment so as to improve
interpretation!
Ecological Uncertainties in Eco-hydraulic Modelling
Ecological uncertainties relate mainly to the simplifi cation,
observation and testing of habitat suitability criteria.
Most frequently, these have been characterised in terms of
species’ preferences or abundances, resulting in Habitat
Suitability Indices (HSIs), although more recent approaches
seek linkages to particular aspects of life cycle and growth
of organisms or communities (see the later section on
bioenergetics in this chapter). Preference criteria may be
simple (univariate) or complex (multivariate) and may be
related to the physically-determined environment through
correspondence (association) or through the application of

126 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
(m)
Cumulative Area (%)
100
90
80
70
60
50
40
30
20
10
0
-1
-0.8
(a)
PT1
RC 1
1.2
1.1
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
20 5040
relaxed model
original model
-0.6
-0.4
-0.2
(m)
1060
1040
1020
1000
980
960
940
N
PT4
Flow
PT 2
0.3
0.2
0.1
0
-0.1
-0.2
-0.3
920
4980 5000 5020 5040
(m)
Modelled - Field Resultant Velocities (m s –1 )
(c)
original model
relaxed model
0
0.2
0.4
0.6
(b)
RC 4
PT 3
Model
RC 2
RC 3
PT1
Relaxed model
Figure 7.11 Spatial (a and b) and areal (c) representations of model fi t to fi eld measurements for the River Cole, Birmingham
(Source: Based upon Figures 5 and 6 of Clifford et al., 2005)
0.8
RC 1
1

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 127
‘rules’ which allow a degree of uncertainty as an essential
element of the procedure. Once physical habitat and ecological
response are obtained, results of the simulation are
expressed in graphical or tabular form as values of the
output between limits or above thresholds, often with a
spatial component, expressed as Weighted Useable Area
(WUA). WUA is an aggregate measure of physical habitat
quality and quantity and will be specifi c to a particular
discharge and species/life stage. However, it is also clear
that patterns or responses in ecology frequently result
from persistent variability and renewal of opportunities,
rather than the ‘static’ presence/absence of clear habitat
delimiters. There have, therefore, been calls to explore the
capacity of eco-hydraulic modelling to identify capacity,
opportunity and variability rather than the prediction of
exclusion via individual tolerance boundaries (Bergen
et al., 2001, and Reynolds, 2002).
In Europe, the Water Framework Directive will require
river management at the catchment scale (Logan & Furse,
2002) and this is being mirrored in national legislation
and subsequent water management activity by organisations,
such as the Environment Agency of England and
Wales. Therefore, a major research question for habitat
modelling is whether existing reached-based methods
can be scaled up, or whether an entirely new approach
needs to be developed. Physical habitat modelling investigations
are typically confi ned to short lengths of river,
approximately 50–200 m. A representative reach approach
may be taken in which a reach is subjectively chosen
to represent a longer length of river, including the
direct proportions of habitats within that reach. A habitat
mapping approach entails classifying and recording habitat
types (pools, riffl es, glides etc) over long lengths of
river and then choosing cross-sections to represent the
identifi ed habitat types. The results from each crosssection
are then weighted according to the proportions
of the identifi ed habitat types (Morhardt et al., 1983;
Maddock, 1999).
Progress towards catchment-scale modelling has been
made in associated fi elds such as fl ood modelling, which
provides broad-scale hydraulic predictions. This provides
a potential basis for catchment-scale physical habitat
assessment using inputs of discharge, water levels and
channel geometry supplied from one-dimensional hydraulic
models. Velocity variations across each cross-section
can then be predicted using disaggregation algorithms
such as that used in the HEC–RAS model (US Army
Corps of Engineers, 2002). Velocity and depth are used to
assess habitat quality for target species. A major issue
when applying this method is the loss of local detail when
scaling up. Booker et al. (2004a) describe the application
of a method to the River Itchen, UK. This was able to
predict physical habitat for an entire catchment under different
water use strategy scenarios compared with present
and naturalised situations.
Some of the many (and complex) uncertainties in ecohydraulic
modelling are explored below in the context
of the most commonly applied numerical scheme, the
PHABSIM model.
7.4.4 The PHABSIM Model
The fi rst step in formulating a habitat modelling approach
for rivers was published by Waters (1976). This led to the
more formal description of a computer model called
PHABSIM (Physical Habitat Simulation) by the US Fish
and Wildlife Service (Bovee, 1982; Bovee et al., 1998;
Milhous, 1999). The PHABSIM system (Bovee, 1982;
Bovee et al., 1998) is a suite of numerical models that
allows quantifi cation of physical habitat for a given site,
defi ned in terms of the combination of depth, velocity
and substrate/cover at a particular discharge (e.g. Johnson
et al., 1993; Elliott et al., 1996). This system is most
commonly used to assess the availability of suitable habitat
for fi sh, although macroinvertebrates (Gore et al., 1998)
and macrophytes (Hearne et al., 1994), which have measurable
physical habitat requirements, have also been
the focus of PHABSIM studies. In the United Kingdom
the method has been used to assess changes in physical
habitat associated with alterations in fl ow regime; for
example, application to the Rivers Allen (Johnson et al.,
1995), Piddle (Strevens, 1999) and Kennet (McPherson,
1997) to aid management decisions based on the effects
of abstraction by groundwater pumping on habitat availability.
The model has also been used to assess the impact
of channel restoration (Acreman and Elliott, 1996) and
difference in habitat caused by different levels of channel
modifi cation (Booker & Dunbar, 2004; Booker et al.,
2003). Over the years, the methodology used has been
adapted by various researchers and institutions (Table 7.4).
This has lead to the development of other models that
follow basically the same approach (Parasiewicz and
Dunbar, 2001).
The approach adopted in many PHABSIM studies has
been outlined by Elliott et al. (1999) and Johnson et al.
(1995). This approach includes identifi cation of river
sectors and species of interest, identifi cation of habitats that
exist within the river length of interest, selection of crosssections
which represent replicates of each habitat type and
collection of model calibration data (water surface elevation,
depth and velocity). The distribution of depths and
velocities are then predicted for the range of discharge
required. Predicted depths and velocities and measured
substrate classes are then compared with HSIs. This allows

128 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Table 7.4 Examples of adaptations of PHABSIM in different countries
Country River Software name Adaptation Reference
France Saint Sauveur brook EVHA French version of PHABSIM Rousel et al. (1999)
Taiwan Chou-Shui Creek PHABSIM Incorporates affects of simulated
substrate changes
Wu & Wang (2002)
Italy Adda PHABSIM Utilises bivariate HSIs Vismara et al. (2001)
USA North Fork Middle PHABSIM Incorporates an individual based Van Winkle et al. (1998)
Fork Tule River
model
Canada Waterton PHABSIM Uses hydraulic output from a 2D
model
Ghanem et al. (1996)
Finland River Oulujoki FISU Uses hydraulic output from a 2D
model
Yrjänä et al. (1999)
France Rhone STATHAB Uses a statistical hydraulics model. Lamouroux et al. (1999)
Norway Mandal SSIIM Uses hydraulic output from a 3D
model
Fjeldstad (2001)
Switzerland Brenno CASIMIR Incorporates fuzzy rules for fi sh
and invertebrate HSIs
http://www.greenhydro.ch
USA Quinnebaug Meso-HABSIM Broader scale, uses more habitat
variables
Parasiewicz (2001)
Figure 7.12 Habitat suitability indices (HSIs) calculated for juvenile salmon preference using pooled results from several rivers
(Source: Data from Dunbar et al., 2001)
prediction of usable physical habitat for the species/life
stage of interest, as WUA in m 2 per 1000 m of river
channel.
The form of HSIs used in PHABSIM applications will
affect the results (Booker and Dunbar, 2004) and is therefore
one source of uncertainty affecting results. HSIs have
been categorised into several groups. These are:
• Type I: HSIs that are derived from expert opinion or
through information published in literature.
• Type II: HSIs that have been derived through frequency
analysis of physical habitat conditions used by
different species or life stages as identifi ed in fi eld
observations.
• Type III: Type II HSIs that have been corrected for
habitat availability. These HSIs are referred to as preference
curves.
• Type IV: Multi-variant HSIs that weight habitat suitability
based on depth and velocity together.
Site-specifi c HSIs can be developed for particular rivers.
However, this is costly and time consuming. HSIs that
used pooled results from several rivers in an attempt to
create generalised curves that can be transferred between
rivers are shown in Figure 7.12. Belaud et al. (1989)
reported similarities between four site-specifi c HSIs and
generalised HSIs, suggesting that generalised HSIs may
be more useful. Roussel et al. (1999) added to the debate

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 129
on HSIs by suggested that different HSIs should be used
for resting and feeding fi sh. Vismara et al. (2001) compared
univariate and bivariate HSIs, concluding that depth
is much more important than velocity in defi ning habitat
suitability requirements when using bivariate models.
Evaluation of PHABSIM
The PHABSIM modelling methodology may be applied
before and after a restoration scheme to infer changes in
habitat using a methodology such as Elliott et al. (1996).
Alternatively, predicted results may be compared with
changes in abundance of fi sh derived from surveys before
and after the scheme. In both instances, it is important to
remember the simplicity of the PHABSIM approaches,
both hydraulic and habitat modelling. Successful application
still requires considerable calibration and refi nement.
Calibration may be based on empirical (regression) fi ts in
an effort to capture stage-dependent as well as spatial
variations between vertical cross-section slices (Milhous
et al., 1989). Alternatively, local and stage-dependent
variations may be estimated from sparse velocity measurements
at a single stage by back-calculating of Manning’s
n on a slice-by-slice basis, having further adjusted to
match the modelled fl ow. In this case, however, the physical
integrity of the model is then lost and cells are not truly
associated by sound hydraulic principles (Ghanem et al.,
1996). In neither case, therefore, is the model a real substitute
for further fi eld measurement. Frequently, there
is an interaction between the uncertainties inherent in
a one-dimensional model and those arising in higherorder
model applications, since the output of the onedimensional
model is sometimes required to determine the
initial water surface profi le for the more complex model.
While PHABSIM remains something of an ‘industry
standard’ tool, the biological representation is also often
highly simplifi ed and empirical (Pusey, 1998). Wood et al.
(2001) found that the strength of ecological relationships
increased and model errors are reduced with increasing
spatial resolution, indicating that most potential is still at
the site-scale but that all modelling efforts are limited by
the lack of longer-term data sets, too. However, one advantage
of this habitat modelling approach is that there are
clear manuals that defi ne step-by-step procedures, which
allow replication of results by different researchers. The
disadvantage of this is that it has led to poor applications
by practitioners with little experience. Best results are
obtained where teams including hydraulic engineers,
hydrologists and ecologists work together, using habitat
modelling as a basis for their river-specifi c studies. Major
strengths and weaknesses of the PHABSIM methodology
are summarised in Table 7.5.
7.4.5 More Complex Hydrodynamic Modelling and
Habitat Simulation
Numerous specifi c modelling applications have been
described that demonstrate some kind of improvement on
simpler schemes. However, these have not given rise to
any single package that is the logical replacement to
PHABSIM. Greater hydraulic process representation may
be achieved using two- and three-dimensional computational
fl uid dynamics models (Alfredsen et al., 1997;
Booker, 2003). New approaches to quantifying hydraulic
habitat have been published (Peters et al., 1995; Nestler
and Sutton, 2000). New habitat models have included
additional variables and have been expanded to the community
level (Bain et al., 1988; Bain, 1995; Lamoroux
et al., 1998). All of these improvements currently come at
a cost of increased complexity, although in the future it is
hoped that they can be used to derive general rules with
which to develop improved look-up methods and to defi ne
the impacts of river fl ow regulation on populations rather
than habitats (Hardy, 1998).
Two-dimensional PHABSIM
The PHABSIM method may also be applied using spatially
continuous patterns of depth and velocity as predicted
by hydrodynamic models (e.g. Ghanem et al.,
1996). Spatially continuous information negates any problems
caused by sampling frequencies applied in fi eld
based studies. This approach also allows visualisation of
areas of suitable habitat. A comparison of habitat suitability
predicted using two different HSIs for a 50 m reach
of the Bere stream, Dorset, UK, is shown in Figure 7.13.
These maps show how truncation of the depth HSI can
lead to changes in patterns of habitat suitability. An
accompanying sensitivity analysis during the construction
of Figure 7.13 also demonstrates how uncertainty in
habitat modelling will depend on the interactions between
the sampling density and the HSI used. In analogy with
the analysis of velocity data in Figure 7.9, spatially continuous
simulations can be sub-sampled to simulate how
the number of cross-sections and point measurements
might affect results had the information been collected in
the fi eld. In general, predicted habitat decreases as more
points are added across each section. This occurs because
poor habitat at the margins of the channel is more likely
to be sampled when more points are measured along each
section. When the truncated HSIs were used, the positions
of the cross-sections in relation to a deep pool in the
middle of the reach had a strong infl uence on predicted
habitat. When small numbers of cross-sections were used,
the likelihood of the pool being over- or under-sampled

130 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Table 7.5 Major strengths and weaknesses of the application of the PHABSIM methodology
Strengths Weaknesses
The method used is replicable and therefore
does not rely on expert opinion
The method can be applied using hydraulic
predictions calculated using a range of
techniques from crude stage-discharge
relationships to spatially continuous
computational fl uid dynamics modelling.
Channel change can be incorporated through
repeated survey
Multi-variate habitat suitability indices have
been developed (Vismara et al., 2001).
Application of habitat mapping has allowed
results to be up-scaled to represent the
length of river under investigation
(Maddock and Bird, 1996; Maddock, 1999).
Recent studies have confi rmed that physical
habitat is an important factor controlling
fi sh populations in the long term (Sabaton
et al., 2003; Souchon and Capra, 2003).
No readily applicable alternative methods
have been developed which enable
prediction of population changes as the
result of water resource impacts.
was higher. Results also suggest that, as the number of
cross-section increases, there is less variation in predicted
habitat above six cross-sections when the non-truncated
HSIs were used and eight cross-sections when the truncated
HSIs were used.
Swimming Speeds
A more sophisticated example of habitat modelling using
hydrodynamic calculations is to compare patterns of
velocities predicted by three-dimensional models with the
swimming speeds of fi sh to assess the effects of channel
design on fi sh habitat. The swimming performance of fi sh
can also be analysed to assess the health of the fi sh when
exposed to sub-lethal toxic chemicals (e.g. Alsop et al.,
1999). One frequently used measurement is the Maximum
Results may be sensitive to the location of cross-sections and
the position of measurements along these cross-sections
Inaccuracies in hydraulic modelling, especially in steep boulder
rivers (Azzellino & Vismara, 2001) and where vegetation
growth causes changes in stage-discharge relationships
throughout the year (Hearne et al. 1994) will affect habitat
predictions.
Requires re-calibration to assess changes in channel morphology
over time. Incorporation of channel change would require
expensive repeated surveys or sediment transport predictions,
which have their own sources of uncertainty.
Rules for weighting depth, velocity and substrate are not readily
available.
Typically restricted to depth, velocity and substrate, although
variables such as water quality, vegetation cover can be
incorporated.
Reached-based results (covering approximately 100 m of river
length) must be scaled up.
Calculates habitat suitability and not population numbers or
presence/absence.
In most studies physical habitat availability at low fl ow has been
the main area of interest, with PHABSIM used primarily to
compare the implications of alternative fl ow regulation
scenarios on habitat. This has led to an emphasis on the
relationship between discharge and usable habitat given a
distribution of relatively shallow depths and slow velocities.
Sustained Swimming Speed (MSSS). This is defi ned
as the maximum velocity at which a fi sh can swim for
a period of more than 200 minutes (Turnpenny et al.,
2001).
Booker (2003) compared simulated velocity patterns to
MSSS of roach, dace and chub to assess habitat suitability
at high fl ows in two reaches of the River Tame, an urban
river in Birmingham, UK. The percentage of in-stream
area that was less than the MSSS was used as an indicator
of habitat quality. This was the area of river, at a specifi ed
distance from the bed, in which fi sh of a certain size and
species could theoretically sustain a position for at least
200 minutes. Due to uncertainty as to the exact distance
between the fi sh and the river bed during high fl ows, and
to simplify the three-dimensional nature of the analysis,
different heights above the bed were considered separately.

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 131
Figure 7.13 Maps of habitat suitability derived using the same hydraulic patterns but different suitability curves: (a) HSIs given in
Dunbar et al. (2001), (b) the same HSIs truncated for depth

132 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
The percentage of survivable area for chub, dace and
roach at bankfull discharge was calculated at two sites
(Figure 7.14). The ‘Highly Modifi ed’ reach was a 91 m
straightened reach with an average width of 11.9 m and
artifi cially strengthened banks contained within a larger
two-stage channel. This reach had no distinct geomorphological
features and relatively uniform bed topography. In
contrast, the ‘Less Modifi ed’ reach was 139 m in length
with an average width of 10.0 m. This reach was also
contained within a two-stage channel but had one bank
consisting of natural material, a slightly sinuous path and
an undulating bed profi le. Figure 7.14(b) shows that as the
body length increases there is an increase in the area of
habitat in which a fi sh is likely to be able to sustain a stationary
position. This is because bigger fi sh can sustain
faster swimming speeds. Similarly, as distance from the
bed increases the percentage of survivable habitat
decreases due to faster velocities. Results showed that at
the Highly Modifi ed site the uniform channel structure
supported no discrete areas of slower velocity which could
be used as refugia. At the Less Modifi ed site a deep pool
approximately 40 m from the upstream boundary provided
a fl ow refuge that could act as a niche of suitable habitat.
There was also a separate discrete area of slower velocity
created by the sheltering effect of a change in direction of
the river planform.
Bioenergetics
Bioenergetic models attempt to quantify the trade-off
between energy gained through feeding and that lost
through swimming ‘costs’. Costs arise from holding position
in a moving fl ow, in the capture of food, as well as
through digestion, faeces and urine (Hayes et al., 2000).
Fish activity may be directly observed and related to fl ow
conditions or estimated on the basis of models that use
results of physiological experiments undertaken on fi sh
which are forced to swim against fl ows of constant
velocity. For example, Booker et al. (2004b) used a threedimensional
Computational Fluid Dynamics (3D-CFD)
model to simulate hydraulic patterns in a 50 m reach of
the Bere stream, Dorset, UK. This information was then
combined with a bioenergetic model that used behavioural
and physiological relationships to quantify the spatial
pattern of energy gain when feeding on invertebrates drifting
in the river. The model was tested by comparing patterns
of predicted energy intake with observed habitat use
by juvenile salmonids at different times of day. A map of
energy intake predicted by the model compared with
observed locations of feeding and resting fi sh at the site
is shown in Figure 7.15.
There are many uncertainties that must be considered
when using this bioenergetic modelling approach as a tool
to assist river restoration. The approach relies on the accuracy
of both the hydrodynamic predictions and the accuracy
of the physiological algorithms and parameters used.
Enders et al. (2003) demonstrate that models of fi sh activity
cost based upon constant velocity experiments may
underestimate actual costs by up to a factor of 4.2, because
turbulent fl ow fl uctuations are neglected. Direct observations
of feeding behaviour illustrate close relationships
to fi sh activity and characteristics of turbulent fl ow (Enders
Figure 7.14 (a) Mean ‘maximum sustainable swimming speed’ and ‘burst swimming speed’ for roach, dace and chub at 8 °C (based
on data from Clough & Turnpenny, 2001), (b) Percentage volume of habitat less than the mean ‘maximum sustainable swimming
speed’ at different distances from the bed in each reach at 18 m 3 s −1 (Source: Figure 11 of Booker, 2003)

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 133
Figure 7.15 Predicted net energy intake for a 0.1 m fi sh and observed fi sh locations
et al., 2005). Models may also be hard to test against fi eld
observations of habitat uses because factors other than
energy intake can infl uence microhabitat selection. For
example, the proximity to other fi sh (Valdimarsson and
Metcalfe, 2001) and predation risk or distance to cover
(Mesick, 1988). Fish do not spend 100% of their time driftfeeding.
They may feed very effi ciently for short periods
and then retreat to more sheltered locations (Gries and
Juanes, 1998). Furthermore, the decision to select a certain
position may not result from conditions (e.g. drift density)
at that time but conditions at a previous time or conditions
over a longer period. Also, a fi sh may not have perfect
knowledge of its habitat. This means that fi sh feeding in an
energetically poor area may not be aware that there are
more favourable alternative positions elsewhere.
7.4.6 Alternatives to Physical Habitat Models
Several alternatives to complex eco-hydraulic simulation
have emerged. A fi eld approach known as ‘Expert Habitat
Mapping’ (EHM) is used in streams that are hydraulically
complex, particularly where the irregularity and variability
of boundary conditions makes hydraulic modelling
inappropriate (for example, in steep bedrock rivers with
large woody debris). EHM uses habitat suitability criteria
but then relies on fi sh biologists mapping habitat based on
those criteria at a range of fl ows to develop fl ow-vs-habitat
curves. The method works well provided that the HSI cri-
teria are used and verifi ed with spot measurements, and
the biologists are experienced.
Alternatively, physically-parameterised models may be
replaced by stochastic or hybrid approaches such as cellular
autometa (Chen et al., 2002) or neural network
models (Werner and Oback, 2001; Reyjol et al., 2001;
Gevrey et al., 2003) as witnessed in hydrology and fl oodplain
inundation studies. As yet, however, these approaches
have focused on simulating ecological characteristics:
appropriate scales for applications, for the development of
rules and training data sets remain to be explored.
7.5 CONCLUSIONS
Restoring and rehabilitating rivers for ecological purposes
is an essentially multi-disciplinary and multi-stage activity.
There is growing awareness that to be successful (that
is, to produce schemes which are sustainable in the
medium- to longer-term) truly functional environments
are required, where catchment hydrology provides the
background, or context, for reach and inter-reach fl ows
and sediment transport. In turn, these hydraulic variables
structure physical habitats, which themselves support a
diverse ecological response. Monitoring of fl ows and
aquatic ecology have been long-standing areas of research
and continue to be vitally important but, increasingly,
emphasis has been upon the modelling or simulation of
fl ows and the habitats which these determine. Modelling

134 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
is useful in extending the range of fi eld observation, in
scenario-type exercises to evaluate restoration schemes
before their implementation and also in providing outputs
against which schemes might be appraised after implementation.
Ideally, modelling efforts should proceed with
the closest possible coupling between their physical
(hydraulic and hydrodynamic) and biological elements,
and in close association with improved strategies for fi eld
monitoring, too. Running through this chapter are some
very basic themes. These refl ect the principal sources of
uncertainty in restoration schemes for ecological purposes
and must be addressed to improve both fi eld and modelling
aspects of restoration design:
data requirements – the appropriate amount and kind of
fi eld data required to adequately specify pre- and post
design aspects of the river;
characterisation – of fl ow coherence and species assemblages
or behaviour, which partly follows from data
considerations as above, but which also includes design
criteria relating to channel form, as well as parameterisation
of models;
coupling – of key physical and ecological aspects of river
system function in models;
awareness – of model application, limitations and sensitivity,
and of other limitations;
development – of newer, more fl exible means of model
evaluation and of restoration appraisal more generally.
While each of the above remain sources of uncertainty,
they represent, too, areas of opportunity for the rapid
development of research and intervention protocols. For
the practitioner, asking questions from each of these areas
– and questions which cut across or link between them – is
perhaps the best form of guidance for recognising, reducing
and even incorporating uncertainty into river restoration
activity from an ecological standpoint.
ACKNOWLEDGEMENTS
This chapter was produced as part of research projects
NER/A/S/1998/00009, ‘Identifi cation of physically-based
design criteria for riffl e-pool sequences in river rehabilitation’,
NER/D/S/2000/01422, ‘Formation of a new river
channel: fl ow, sediment and vegetation dynamics’, and
NER/T/S/2001/01250, ‘Vegetation infl uences on fi ne sediment
and propagule dynamics in groundwater-fed rivers:
implications for river management, restoration and riparian
biodiversity’. These were funded in the United
Kingdom by the Natural Environment Research Council.
SSIIM is produced and made available by Nils R. B.
Olsen, Department of Hydraulic and Environmental
Engineering, The Norwegian University of Science and
Technology. Scott McBain kindly provided references to
the fi gures and tabular information used in the text, and
advised on initial chapter layout and content. Nigel Wright
made additional suggestions in respect of hydraulic
modelling.
REFERENCES
Acreman MC. 2000. The changing hydrology of the UK. In:
Acreman MC (Ed), The Hydrology of the UK: A Study of
Change. Routledge: London, UK.
Acreman MC. 2002. Case Studies of Managed Flood Releases.
Environmental Flow Assessment Part III. World Bank Water
Resources and Environmental Management Best Practice Brief
No 8, World Bank: Washington, DC.
Acreman MC, Elliott CRN. 1996. Evaluation of the river Wey
restoration project using the Physical HABitat SIMuation
(PHABSIM) model. Proceedings of the 31st MAFF Conference
of River and Coastal Engineers, Keele University, 3–5
July 1996; 1.2.1–1.2.14.
Acreman MC, King J. 2003. Defi ning water requirements. In:
Dyson M, Bergkamp G, Scanlon H (Eds), Flow: The Essentials
of Environmental Flows. IUCN: Gland, Switzerland and Cambridge,
UK.
Alfredsen K, Marchand W, Bakken TH, Harby A. 1997. Application
and comparison of computer models quantifying impacts
of river regulation on fi sh habitat. In: Broch E, Lysne DK,
Flatabo N, Helland-Hansen E (Eds), Proceedings of the 3rd
International conference on hydropower Hydropower ’97, A.A.
Balkema Publishers: Rotterdam/Brookfi eld.
Alsop DH, McGeer JC, McDonald DG, Wood CM. 1999. Costs
of chronic waterborne zinc exposure and the consequences of
zinc acclimation on the gill/zinc interactions of rainbow trout
in hard and soft water. Environmental Toxicology and Chemistry
18: 1014–1025.
Anderson MG, Walling DE, Bates PD (Eds). 1996. Floodplain
Processes. John Wiley & Sons Ltd: Chichester, UK.
Archer D, Newson, M. 2002. The use of indices of fl ow variability
in assessing the hydrological and instream habitat impacts of
upland afforestation and drainage. Journal of Hydrology 268:
244–258.
Armitage PD, Petts GE. 1992. Biotic score and prediction to
assess the effects of water abstraction on river macroinvertebrates
for conservation purposes. Aquatic Conservation:
Marine and Freshwater Ecosystems 2: 1–17.
Arthington AH, Long GC (Eds). 1997. Logan River Trial of the
Building Block Methodology for Assessing Environmental
Flow Requirements: Background Papers. Centre for Catchment
and In-Stream Research and Department of Natural Resoruces:
Brisbane, Queensland.
Arthington AH, Lloyd R (Eds). 1998. Logan River Trial of the
Building Block Methodology for Assessing Environmental
Flow Requirements: Workshop Report. Centre for Catchment
and In-Stream Research and Dept Natural Resources: Brisbane,
Queensland, Australia.

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 135
Azzellino A, Vismara R. 2001. Pool quality index: New method
to defi ne minimum fl ow requirements of high-gradient, loworder
streams. Journal of Environmental Engineering 127 (11):
1003–1013.
Barker I, Kirmond A. 1998. Managing surface water abstraction.
In: Wheater H, Kirby C (Eds), Hydrology in a Changing Environment
British Hydrological Society, 249–258.
Bain MB. 1995. Habitat at the local scale: multivariate patterns for
stream fi shes. Bull. Fr. Peche Pisic. 337/338/339: 165–177.
Bain MB, Finn JT, Booke HE. 1988. Streamfl ow regulation and
fi sh community structure. Ecology 69: 382–392.
Belaud A, Chaveroche P, Lim P, Sabaton C. 1989. Probability-ofuse
curves applied to brown trout (Salmo trutta fario L.) in
rivers of southern France. Regulated Rivers: Research and
Management 3: 321–336.
Benetti AD, Lanna AE, Cobalchini MS. 2002. Current practice
for establishing environmental fl ows in Brazil. Proceedings of
the Fourth International Ecohydraulics Symposium, Southern
Waters, Cape Town.
Bergen SD, Bolton SM, Fridley JL. 2001. Design Principles for
ecological engineering. Ecological Engineering 18: 201–210.
Beven K. 2000. Rainfall–runoff Modelling. John Wiley & Sons
Ltd: Chichester, UK.
Beven K. 2001. On modelling as collective intelligence. Hydrological
Processes 15: 2205–2207.
Bierkens MFP, Finke PA, de Willigen P. 2000. Upscaling and
Downscaling Methods for Environmental Research. Kluwer
Academic Publishers: Dordrecht, The Netherlands.
Booker DJ. 2003. Hydraulic modelling of fi sh habitat in urban
rivers during high fl ows. Hydrological Processes 17: 577–599.
Booker DJ, Dunbar MJ. 2004. Application of Physical HAbitat
SIMulation (PHABSIM) modelling to modifi ed urban river
channels. River Research and Applications 20: 167–183.
Booker DJ, Dunbar MJ, Acreman MC et al. 2004a. Habitat assessment
at the catchment scale; application to the River Itchen,
UK. British Hydrological Society International Conference:
Hydrology: Science & Practice for the 21st Century, July 2004,
volume II, 10–18, British Hydrological Society, London, UK.
Booker DJ, Dunbar MJ, Ibbotson AT. 2004b Predicting juvenile
salmonid drift-feeding habitat quality using a three-dimensional
hydraulic-bioenergetic model. Ecological Modelling
177: 157–177.
Booker DJ, Dunbar MJ, Shamseldin A et al. 2003. Physical
habitat assessment in urban rivers under future fl ow scenarios.
CIWEM Journal 17: 251–257.
Boon PJ. 1998. River restoration in fi ve dimensions. Aquatic Conservation:
Marine and Freshwater Ecosystems 8: 257–264.
Bovee KD. 1982. A Guide to Stream Habitat Analysis using the
IFIM. US Fish and Wildlife Service Report FWS/OBS-82/26:
Fort Collins, Colorado.
Bovee KD, Lamb BL, Bartholow JM et al. 1998. Stream habitat
analysis using the Instream Flow Incremental Methodology. US
Geological Survey, Biological Resources Division, Information
and Technology Report USGS/BRD/ITR–1998–004.
Bunn SE, Arthington AH. 2002. Basic principles and ecological
consequences of altered fl ow regimes for aquatic biodiversity.
Environmental Management 50: 492–507.
Chen Q, Mynett AE, Minns AW. 2002. Application of cellular
automata to modelling competitive growths of two underwater
specieis Chara aspera and Potamogeton pectinatus in Lake
Veluwe. Ecological Modelling 147: 253–265.
Chow VT. 1973. Open Channel Hydraulics. McGraw-Hill: Tokyo,
Japan.
Clarke SJ, Bruce-Burgess L, Wharton G. 2003. Linking form and
function: towards an eco-hydromorphic approach to sustainable
rive restoration. Aquatic Conservation: Marine and Freshwater
Ecosystems 13: 439–450.
Clausen B, Biggs BJF. 2000. Flow variables for ecological studies
in temperate streams: groupings based on covariance. Journal
of Hydrology 237: 184–197.
Clifford NJ. 2001. Conservation and the river channel environment.
In: Warren A, French JR (Eds), Habitat Conservation.
John Wiley & Sons Ltd: Chichester, UK; 67–104.
Clifford NJ. 2002. Hydrology: the changing paradigm. Progress
in Physical Geography 26: 290–301.
Clifford NJ, Soar PJ, Emery JC et al. 2002a. Numerical fl ow modelling
for eco-hydraulic and river rehabilitation applications:
a case study of the River Cole, Birmingham, UK. In: Bousmar
D, Zech Y (Eds), River Flow 2002. Lisse: Swets & Zeitlinger/
Balkema: Rotterdam, The Netherlands; 1195–1204.
Clifford NJ, Soar PJ, Emery JC et al. 2002b. Sustaining water
related ecosystems – the role of in-stream bedform design in
river channel rehabilitation. In: Friend 2002 – Regional hydrology:
bridging the gap between research and practice. IAHS
Publication 274: 407–413, IAHS, Wallingford, UK.
Clifford NJ, Soar PJ, Harmar OP et al. 2005. Assessment of
hydrodynamic simulation results for eco-hydrological and ecohydraulic
applications: a spatial semivariance approach.
Hydrological Processes 19: 3631–3648.
Clifford NJ, Harmar OP, Harvey G, Petts GE. 2006. Physical
habitat, eco-hydraulics and river design: a review and reevaluation
of some popular concepts and methods. Aquatic Conservation:
Marine and Freshwater Ecosystems 16: 389–408.
Clough SC, Turnpenny AWH. 2001. Swimming Speeds in Fish:
Phase 1. Environment Agency R&D Technical Report W2–
026/TR1, Environment Agency: Bristol, UK.
Crowder DW, Diplas P. 2000. Using two-dimensional hydrodynamic
models at scales of ecological importance. Journal of
Hydrology 230: 172–191.
Dakova S, Uzunov Y, Mandadjiev D. 2000. Low fl ow – the river’s
ecosystem limiting factor. Ecological Engineering 16: 167–
174.
Dunbar MJ, Ibbotson AT, Gowing IM et al. 2001. Ecologically
Acceptable Flows Phase III: Further Validation of PHABSIM
for the Habitat Requirements of Salmonid Fish. Final R&D
Technical report to the Environment Agency, Environment
Agency, Bristol, UK.
Dyson M, Bergkamp G, Scanlon H (Eds). 2003 Flow: The Essentials
of Environmental Flows. IUCN: Gland, Switzerland and
Cambridge, UK.
Elliott CRN, Johnson IW, Sekulin AE et al. 1996. Guide to the
use of the Physical Habitat Simulation System, EA Release
version. Environment Agency R&D Technical Report W20.
Environment Agency: Bristol, UK.

136 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Elliott CRN, Dunbar MJ, Gowing I, Acreman MC. 1999. A
habitat assessment approach to the management of groundwater
dominated rivers, Hydrological Processes 13: 459–475.
Emery JC. 2003. Characteristics and controls of gravel-bed riffl epool
sequences for habitat assessment and river rehabilitation
design. Unpublished PhD thesis, University of Birmingham,
UK.
Emery JC, Gurnell AM, Clifford NJ et al. 2003. Classifying the
performance of riffl e-pool bedforms for habitat assessment and
river rehabilitation design. River Research and Applications
19: 533–549.
Enders EC, Boisclair D, Roy AG. 2003. The effect of turbulence
on the cost of swimming for juvenile Atlantic salmon (Salmos
salar). Canadian Journal of Fisheries and Aquatic Sciences 60:
1149–1160.
Enders EC, Buffi ngton-Belanger T, Boisclair D, Roy AG. 2005.
The feeding behaviour of juvenile Atlantic salmon in relation
to turbulent fl ow. Journal of Fish Biology 66: 242–253.
Everard M. 1996. The importance of periodic droughts for maintaining
diversity in the freshwater environment. Freshwater
Forum 7: 33–50.
Extence C, Balbi DM, Chadd RP. 1999. River fl ow indexing using
British benthic macro-invertebrates: a framework for setting
hydro-ecological objectives. Regulated Rivers Research and
Management 15: 543–574.
Frissell CA, Liss WJ, Warren CE, Hurley MD. 1986. A hierarchical
framework for stream classifi cation: viewing streams in a
watershed context. Environmental Management 10: 199–214.
French JR, Clifford NJ. 2000. Hydrodynamic modelling as a basis
for explaining estuarine environmental dynamics: some conceptual
and methodological issues. Hydrological Processes 14:
2089–2108.
Fjeldstad H-P. 2001. Numerical modelling tools for predicting
physical habitat adjustments. In: Taugbol T, L’Abee-Lund J-H
(Eds), Proceedings from the CONNECT workshop ‘Physical
habitat restoration in canalised watercourses – possibilities
and constraints, Norweigan Institute of Nature Research, Oslo,
Norway, 91–99.
Gevrey M, Dimopoulosb I, Leka S. 2003. Modelling the structure
of aquatic communities: concepts, methods and problems. Ecological
Modelling 160: 249–164.
Ghanem A, Steffl er P, Hicks F, Katopodis C. 1996. Two-dimensional
hydraulic simulation of physical habitat conditions in
fl owing streams. Regulated Rivers-Research & Management
12: 185–200.
Gilvear DJ, Heal KV, Stephen A. 2002. Hydrology and the ecological
quality of Scottish river ecosystems. The Science of the
Total Environment 294: 131–159.
Gregory S, Boyer K, Gurnell AM (Eds). 2003. The Ecology and
Management of Wood in World Rivers. American Fisheries
Society Symposium 37, American Fisheries Society: Bethesda,
Maryland.
Gries G, Juanes F. 1998. Microhabitat use by juvenile Atlantic
salmon (Salmo salar) sheltering during the day in summer.
Canadian Journal of Zoology 76: 1441–1449.
Gore JA, Crawford DJ, Addison DS. 1998. An analysis of artifi cial
riffl es and enhancement of benthic community diversity by
physical habitat simulation (PHABSIM) and direct observation.
Regulated Rivers: Research and Management 14: 69–77.
Graf WL. 2001. Damage control: restoring the physical integrity
of America’s rivers. Annals of the Association of American
Geographers 91: 1–27.
Hardy RJ, Lane SN, Ferguson RI, Parsons DR. 2003. Assessing
the credibility of a series of computational fl uid dynamic simulationsof
open channel fl ow. Hydrological Processes 17:
1539–60.
Hardy TB. 1998. The future of habitat modeling and instream
fl ow assessment techniques Regulated Rivers: Research and
Management 14: 405–420.
Hayes JW, Stark JD, Shearer KA. 2000. Development and test of
a whole-lifetime foraging and bioenergetics growth model for
drift feeding brown trout. Transactions of the American Fisheries
Society 125: 315–332.
Hearne J, Johnson IW, Armitage PD. 1994. Determination of
ecologically acceptable fl ows in rivers with seasonal changes
in the density of macrophyte Regulated Rivers: Research and
Management 9: 117–184.
Hendry K, Cragg-Hine D, O’Grady MO et al. 2003. Management
of habitat for rehabilitation and enhancement of salmonid
stocks. Fisheries Research 12: 171–192.
Hill MT, Platt S, Beschta RL. 1991. Ecological and geomorphological
concepts for instream and out of channel fl ow requirements.
Rivers 2: 198–210.
Janauer GA. 2000. Ecohydrology: fusing concepts and scales.
Ecological Engineering 16: 9–16.
Johnson IW, Elliott CRN, Gustard A et al. 1993. Ecologically
Acceptable Flows National Rivers Authority R&D Project
Record 282/1/Wx: Bristol, UK.
Johnson IW, Elliott CRN, Gustard A. 1995. Modelling the effect
of groundwater abstraction on salmonid habitat availability in
the River Allen, Dorset, England. Regulated Rivers: Research
& Management 10: 229–238.
Junk WJ, Bayley PB, Sparks RE. 1989. The fl ood pulse concept
in river-fl oodplain systems. In: Dodge DP (Ed), Proceedings
of the International Large River Symposium (LARS), Canadian
Special Publication of Fisheries and Aquatic Sciences 106,
National Research Council of Canada, Ottawa; 110–127.
Kemp JL, Harper DM, Crosa GA. 1999. Use of ‘functional habitats’
to link ecology with morphology and hydrology in river
rehabilitation. Aquatic Conservation: Marine and Freshwater
Ecosystems 9: 159–178.
Kemp JL, Harper DM, Crosa GA. 2000. The habitat-scale ecohydraulics
of rivers. Ecological Engineering 16: 17–29.
Kondolf GM, Larsen EW, Williams JG. 2000. Measuring and
modelling the hydraulic environment for assessing instream
fl ows. North American Journal of Fisheries Management 20:
1016–1028.
Kirkby MJ. 1993. Network hydrology and geomorphology. In:
Beven K, Kirkby MJ (Eds), Channel Network Hydrology. John
Wiley & Sons Ltd: Chichester, UK; 1–11.
King JM, Tharme RE, de Villiers MS (Eds). 2000. Environmental
fl ow assessments for rivers: manual for the Building Block
Methodology. Water Research Commission Report TT 131/00:
Pretoria, South Africa.

Hydrological and Hydraulic Aspects of River Restoration Uncertainty for Ecological Purposes 137
Lamouroux N, Capra H, Pouilly M. 1998. Predicting Habitat
Suitability for lotic fi sh: linking statistical hydraulic models
with multivariate habitat use models. Regulated Rivers 14:
1–11.
Lamouroux N, Doutriaux E, Terrier C, Zylberblat M. 1999.
Modélisation des impacts de la gestion des débits réservés du
Rhône sur les peuplements piscicoles. Bulletin Français de la
Pêche et de la Pisciculture 352: 45–61.
Lane SN. 1998. Hydraulic modelling in hydrology and geomorphology:
a review of high resolution approaches. Hydrological
Processes 12: 1131–1150.
Lane SN, Richards KS. 2001. The ‘validation of hydrodynamic
models: some critical perspectives. In: Anderson MG, Poates
PD (Eds), Model validation: perspectives in hydrological
science. John Wiley & Sons Ltd: Chichester, UK; 414–438.
LeClerc 2002. Ecohydraulics, last frontier for fl uvial hydraulics:
research challenges and multidisciplinary perspectives. In:
Bousmar D, Zech Y (Eds), River Flow 2002. Swets & Zeitlinger/
Balkema: Liss; 13–26.
Logan P, Furse M. 2002. Preparing for the European Water
Framework Directive – making the links between habitat and
aquatic biota. Aquatic Conservation: Marine and Freshwater
Ecosystems 12: 425–437.
Lytle DA, Poff NL. 2004. Adaptation to natural fl ow regimes.
Trends in Ecology and Evolution 19: 94–100.
Maddock I. 1999. The importance of physical habitat assessment
for evaluating river health. Freshwater Biology 41: 373–391.
Maddock IP, Bird D. 1996. The application of habitat mapping to
identify representative PHABSIM sites on the River Tavy,
Devon, UK. In: Leclerc et al. (Eds), Proceedings of the 2nd
International Symposium on habitats and hydraulics, Quebec,
Canada, June 1996.
Marsh TJ. 2002. Capitalising on river fl ow data to meet changing
national needs – a UK perspective. Flow Measurement and
Instrumentation 13: 291–298.
Matthews RC, Bao Y. 1991. The Texas method of preliminary
instream fl ow determination. Rivers 2: 295–310.
McBain & Trush, Inc. (Eds). 2002. San Joaquin River Restoration
Study Background Report. Prepared for Friant Water Users
Authority, Lindsay, CA, and Natural Resources Defense
Council: San Francisco, California.
McPherson JI. 1997. Appeal by Thames Water Utilities under
Section 43 of the Water Resources Act 1991, Inspector’s
Report, WAT/95/22. Department of the Environment: Bristol,
UK.
Mesick CF. 1988. Effects of food and cover on numbers of apache
and brown trout establishing residency in artifi cial stream channels.
Transactions of the American Fisheries Society 117:
421–431.
Milhous RT. 1999. History, theory, use, and limitations of the
Physical Habitat Simulation System. Proceedings of the 3rd
International Symposium on Ecohydraulics, Salt Lake City,
Utah, USA. Available on CD-ROM only.
Milhous RT, Updike MA, Scheider DM. 1989. Physical Habitat
Simulation (PHABSIM) Reference Manual, Version II.
Instream Flow Information Paper No. 26. US Fish and Wildlife
Service Biology Report 89 (16).
Morhardt JE, Hanson DF, Coulston CJ. 1983. Instream fl ow
analysis: Increased accuracy using habitat mapping. Waterpower
83- An International Conference on Hydropower.
Conference Proceedings Vol 3, Tennessee Valley Authority,
Knoxville; 1294–1304.
Mousouridis A. 2001. Identifying Upland Channel Network
Dynamics Over Decadal and Century Timescales. Unpublished
PhD thesis, University of London, UK.
Naiman RJ, Lonzarich DG, Beechie TJ, Ralph SC. 1992. General
principles of classifi cation and the assessment of conservation
potential in rivers. In: Boon PJ, Calow P, Petts GE (Eds), River
Conservation and Management. John Wiley & Sons Ltd:
Chichester, UK; 93–123.
Nestler J, Sutton VK. 2000. Describing scales of features in river
channels using fractal geometry concepts. Regulated Rivers:
Research & Management 16: 1–22.
Newbury R, Gaboury M. 1993. Exploration and rehabilitation of
hydraulic habitats in streams using principles of fl uvial behaviour.
Freshwater Biology 29: 195–210.
Newson MD. 2002. Geomorphological concepts and tools for
sustainable river ecosystem management. Aquatic Conservation:
Marine and Freshwater Ecosystems 12: 365–379.
Newson MD, Newson CL. 2000. Geomorphology, ecology
and river channel habitat: mesoscale approaches to basinscale
challenges. Progress in Physical Geography 24: 195–
217.
Newson MD, Harper DM, Padmore CL et al. 1998. A costeffective
approach for linking habitats, fl ow types and species
requirements. Aquatic Conservation: Marine and Freshwater
Ecosystems 8: 431–446.
Nicholas AP. 2001. Computational fl uid dynamics modelling of
boundary roughness in gravel-bed rivers: an investigation of
the effects of random variability in bed elevation. Earth Surface
Processes and Landforms 26: 345–362.
Oreskes N, Shrader-Frechette K, Belitz K. 1994. Verifi cation,
validation, and confi rmation of numerical models in the Earth
Sciences. Science 263: 641–646.
Padmore CL. 1997. Biotopes and their hydraulics: a method for
defi ning the physical component of freshwater quality. In:
Boon PJ, Howell DL (Eds), Freshwater Quality: Defi ning
the Indefi nable? The Stationery Offi ce: Edinburgh, UK;
251–257.
Parasiewicz P. 2001. MesoHABSIM: A concept for application
of instream fl ow models in river restoration planning. Fisheries
26: 6–13.
Parasiewicz P, Dunbar MJ. 2001. Physical habitat modelling
for fi sh: A developing approach. Large Rivers, 12 2–4, Arch.
Hydrobiol. Suppl. 135/2–4: 239–268.
Peters MR, Abt SR, Watson CC et al. 1995. Assessment of
restored riverine habitat using RCHARC. Water Resources
Bulletin 31: 745–752
Petts GE. 1984. Impounded Rivers: Perspectives for Ecological
Management. John Wiley & Sons Ltd: Chichester, UK.
Petts GE, Amoros C. 1996. Fluvial Hydrosystems. Chapman and
Hall: London, UK.
Poff NL, Allan JD, Bain MB et al. 1997. The natural fl ow regime.
BioScience 47: 769–784.

138 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Prudhomme C, Jakob D, Svensson C. 2003. Uncertainty and
climate change impact on the fl ood regime of small UK catchments.
Journal of Hydrology 277: 1–23.
Pusey BJ. 1998. Methods addressing the fl ow requirements of
fi sh. In: Arthington AH, Zalucki JM (Eds), Comparative evaluation
of environmental fl ow assessment techniques review of
methods. Occassional Paper 27/98, Land and Water Resources
and Development Corporation: Canberra, Australia; 64–103.
Railsback SF, Lamberson RH, Harvey BC, Duffy WE, 1999.
Movement rules for individual-based models of stream fi sh.
Ecological Modelling 123: 73–89.
Reice SR, Wissmar RC, Naiman RJ. 1990. Disturbance regimes,
resilience and recovery of animal communities and habitats in
lotic ecosystems. Environmental Management 14: 647–660.
Reyjol Y, Lim P, Belaud A, Lek S. 2001. Modelling microhabitat
used by fi sh in natural and regulated fl ows in the river Garonne
(France). Ecological Modelling 146: 131–142.
Reynolds CS. 2002. Ecological pattern and ecosystem theory.
Ecological Modelling 158: 181–200.
Richter BD, Baumgartner JV, Powell J, Braun DP. 1996. A method
for assessing hydrological alteration within ecosystems. Conservation
Biology 10: 1163–1174.
Richter BD, Baumgartner JV, Wigington R, Braun DP. 1997. How
much water does a river need? Freshwater Biology 37:
231–249.
Rodriguez-Iturbe, Valdes JB. 1979. The geomorphic structure
of hydrologic response. Water Resources Research 15:
1409–1420.
Roussel JM, Bardonnet A, Claude A. 1999. Microhabitats of
brown trout when feeding on drift and when resting in a
lowland salmonid brook: effects on Weighted Usable Area.
Archiv Fur Hydrobiologie 146: 413–429.
Rowntree K, Wadeson R. 1998. A geomorphological framework
for the assessment of instream fl ow requirements. Aquatic Ecosystem
Health and Management 1: 125–141.
Sabaton C, Souchon Y, Lascaux JM et al. 2003. The ‘Guaranteed
Flow Working Group’: A French feedback of IFIM based on
habitat and brown trout population time series observations. In:
Lamb BL, D Garcia de Jalon C, Sabaton C et al. (Eds), Proceedings
of International IFIM users conference: Colorado
State University, Fort Collins, Colorado.
Schmidt JC, Webb RH, Valdez RA et al. 1998. Science and values
in river restoration in the Grand Canyon. BioScience 48:
735–747.
Smakhtin VU. 2001. Low fl ow hydrology: a review. Journal of
Hydrology 240: 147–186.
Souchon Y, Capra H. 2003. Aquatic habitat modelling: biological
validations of IFIM/PHABSIM methodology and new perspectives
In: Lamb BL, D Garcia de Jalon C, Sabaton C et al. (Eds),
Proceedings of International IFIM users conference: Colorado
State University, Fort Collins, Colorado.
Souchon Y, Keith P. 2001. Freshwater fi sh habitat: science, management
and conservation in France. Aquatic Ecosystem Health
and Management 4: 401–412.
Spence R, Hickley P. 2000. The use of PHABSIM in the management
of water resources and fi sheries in England and Wales.
Ecological Engineering 16: 153–158.
Stanford JA, Ward JV, Liss WJ et al. 1996. A general protocol for
restoration of regulated rivers. Regulated Rivers: Research and
Management 12: 391–413.
Stillwater Sciences 2001. Tuolumne River restoration program
summary report, summary of studies, conceptual models, restoration
projects, and ongoing monitoring. Prepared for the
CALFED/AFRP Adaptive Management Forum, with assistance
from the Tuolumne River Technical Advisory Committee.
Strevens AP. 1999. Impacts of groundwater abstraction on the
trout fi shery of the River Piddle; and an approach to their alleviation.
Hydrological Processes 13: 487–496.
Tennant DL. 1976. Instream fl ow regimens for fi sh, wildlife,
recreation and related environmental resources. Fisheries 1:
6–10.
Thoms MC, Sheldon F. 2002. An ecosystem approach for determining
environmental water allocations in Australian dryland
river systems: the role of geomorphology. Geomorphology 47:
153–168.
Turnpenny AWH, Blay SR, Carron JJ, Clough SC. 2001. Literature
Review of Swimming Speeds of Freshwater Fish. Environment
Agency R&D Technical Report W2–026/TR2,
Environment Agency: Bristol, UK.
US Army Corps of Engineers. 2002. HEC-RAS River Analysis
System. CDP-68. US Army Corps of Engineers: Davis,
California.
Valdimarsson SK, Metcalfe NB. 2001. Is the level of aggression
and dispersion in territorial fi sh dependent on light intensity?
Animal Behaviour 61: 1143–1149.
Van Winkle W, Jager HI, Railsback SF et al. 1998. Individualbased
model of sympatric populations of brown and rainbow
trout for instream fl ow assessment: model description and calibration.
Ecological Modelling 110: 175–207.
Vismara R, Azzellino A, Bosi R et al. 2001. Habitat suitability
curves for brown trout (Salmo trutta fario L.) in the River Adda,
Northern Italy: Comparing univariate and multivariate approaches.
Regulated Rivers, Research & Management 17: 37–50.
Waters BF. 1976. A methodology for evaluating the effects of
different stream fl ows on salmonid habitat. In: Orsborn JF,
Allman CH (Eds), Instream Flow Needs. American Fisheries
Society: Bethesda, Maryland. 254–266.
Werner H, Oback M. 2001. New neural network types estimating
the accuracy response for ecological modelling. Ecological
Modelling 146: 289–298.
Wood PJ, Hannah DM, Agnew MD, Petts GE. 2001. Scales of
hydroecological variability within a groundwater-dominated
stream. Regulated Rivers: Research and Management 17:
347–367.
Wu FC, Wang CF. 2002. Effect of fl ow-related substrate alteration
on physical habitat: A case study of the endemic river loach
Sinogastromyzon puliensis (Cypriniformes, Homalopteridae)
downstream of Chi-Chi diversion weir, Chou-Shui Creek,
Taiwan. River Research and Applications 18: 155–169.
Yrjänä T, Lahti M, Kamula R. 1999. Effects of a habitat enhancement
experiment in the Laukka area of the river Oulujoki,
Finland. University of Oulu. Publications of Water Resources
and Environmental Engineering Laboratory. A5. (In Finnish,
with English Summary).

River Restoration: Managing the Uncertainty in Restoring Physical Habitat Edited by Stephen Darby and David Sear
© 2008 John Wiley & Sons, Ltd
8
Uncertainty Surrounding the Ecological
Targets and Response of River and
Stream Restoration
8.1 INTRODUCTION
Martin R. Perrow 1 , Eleanor R. Skeate 1 , David Leeming 2 , Judy England 3 and
Mark L. Tomlinson 1
In his seminal paper, Bradshaw (1987) described restoration
ecology as the ‘acid test’ of ecology in that if there
is suffi cient understanding of form, structure, process and
function then it should be possible to restore any particular
habitat to a close approximation of that which is desired.
The recent Handbook of Ecological Restoration in two
volumes detailing principles (Perrow and Davy, 2002a)
and practice (Perrow and Davy, 2002b) suggests that restoration
ecology has come of age as a science, although
there is great disparity in the level of understanding and
thus success in different biomes. The book also reveals
the fundamentally different approach to the restoration of
lotic and lentic environments.
In lakes, especially shallow ones, there has been a theoretical
shift from ‘bottom-up’, where the form and function
was controlled by physical and chemical properties,
to ‘top-down’, where components at the top of the food
web, especially fi sh, have great bearing on lower trophic
levels and even physical properties (Perrow et al., 2002).
The focus of lake restoration is therefore often on fi sh
(Jeppesen and Sammalkarpi, 2002) embracing and using
fundamentally ecological interactions.
In rivers and streams there has also been a shift in focus
away from water quality – which is now not seen as the
primary limiting factor for a healthy, natural system as it
was immediately after the industrial revolution and prior
1 ECON, Ecological Consultancy, UK.
2 Consultant Ecologist, UK
3 Environment Agency, UK
to the development of improved water quality standards –
to the limitations of physical habitat structure. For example
in the United Kingdom, initial restoration efforts on the
River Thames in 1858 were driven by water problems
caused by pollution from human and animal wastes
(Gameson and Wheeler, 1977), whereas now the focus is
on restoring habitat quality, especially for fi sh, in the upper
catchment where tributaries have suffered from river engineering
schemes (Robinson and Whitton, 2004).
What has become the billion dollar industry of river
restoration (Malakoff, 2004; Palmer et al., 2005) generally
uses a geomorphological (in combination with engineering)
approach (Downs et al., 2002; Malakoff 2004), in the
hierarchical gradient of process–form–habitat–biota promoted
by the highly infl uential National Research Council
(of the United States) publication Restoration of Aquatic
Ecosystems (1992). This is restore natural water and sediment
regime; restore natural channel geometry; restore
natural riparian plant communities; and restore native
aquatic plants and animals.
The widely perceived ecological focus of river restoration
(Newson et al., 2002; Downs and Skinner 2002;
Ormerod, 2004), is therefore the fi nal step in the chain of
works. As with ripples on a pond or the cascade response
through trophic levels (Carpenter and Kitchell, 1993), the
strength of the interaction and the response is likely to
weaken the further from the source action. To take a more
specifi c example, the huge amount of river restoration

140 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
focused on anadromous fi shes including sturgeon
(Waldman and Wirgin, 1998) and particularly salmonids,
upon which millions of dollars are spent annually in the
US Pacifi c Northwest alone (Roni et al., 2002), effectively
hinges on the relationship between the fi sh and limiting
habitat variables. However, such relationships are not
always clear-cut and may explain a low proportion of the
variation in abundance and biomass (Milner et al., 1985).
In very simple terms, a ‘restored’ river may be perfectly
capable of supporting a high abundance and biomass of
fi sh but may actually contain very few. In recognition of
the fact that biological and ecological limitations (i.e.
stock limited recruitment, disease, predation, competition
etc) may be more important than the generic problem of
not having enough habitat, release of large numbers of
artifi cially raised juveniles may have to be undertaken, e.g.
on the River Mattole in California (Mattole Restoration
council http://www.mattole.org/) and in rivers in North-
East England (Russell, 1994). In a nutshell, the paradigm
of getting the physical structure right and then the ‘ecology’
will then surely follow is likely to be a false premise
(Ormerod, 2004).
Geomorphological and ecological processes are inextricably
intertwined and best considered in an eco-hydromorphic
approach to restoration (Clarke et al., 2003). To
illustrate, consider a low order boulder-strewn upland
stream. Here, geological structure, gradient and geomorphological
processes expressed in fl ow and sediment
regime may initially be viewed as dominant over ecological
processes. But consider the impact of the presence or
absence of trees and their coarse woody debris (CWD)
feeding back into geomorphological processes by constraining
channel width and increasing fl ows and promoting
erosion in one place and slowing fl ows and causing
deposition in another. Allochthonous input in the form of
leaf fall provides coarse particulate matter (CPOM) to the
channel providing the basis of nutrient cycling and spiraling
(see Newbold, 1992 for an overview) and energy fl ow
(Calow, 1992 for an overview) and thus biological production,
ultimately determines the biomass of the biota.
Coarse woody debris from the trees promotes retentiveness
and decomposition of litter (Lepori et al., 2005), key ecological
functions infl uencing the abundance and diversity
of shredding invertebrates responsible for producing fi ne
particulate matter (FPOM) used by other feeding guilds,
consumed in turn by fi sh, birds and mammals (Richardson
and Jackson, 2002). In this hypothetical example, understanding
and subsequently manipulating ecological processes,
such as succession, herbivory (predation) and
competition, to infl uence the structure and composition of
the riparian tree fl ora may be as (if not more) effective as
manipulating physical habitat structure.
Moreover, whilst ecological processes may be embodied
within a particular component of the fauna and fl ora, this
is intuitively unlikely to operate over trophic scales within
biotic components as well as interact with abiotic components.
Individuals, species populations and their communities
cannot be seen as mere products of the physical (and
chemical) framework but inextricably linked with it via
particular processes. Thus, an engineering analogue of
fi tting habitat components back together (i.e. a building
block functional habitat approach; Harper et al., 1992) to
make a naturally functioning ecosystem is likely to be too
simplistic. The nub of an engineering approach, that a
habitat component or unit may be expressed as a certain
number of individuals or composition of a community,
thus takes no account of process, but rather assumes it.
Our experiences of river restoration in the United
Kingdom lead us to believe there is a widespread lack of
understanding of what ‘ecology’ is and does in river restoration.
Pertinent to this volume, we aim to explore
whether this intuitively leads to uncertainty (in this case
reducible ignorance, see Chapter 3) and a lack of confi -
dence in ecologists amongst other types of restoration
practitioner. For example, Downs and Skinner (2002) suggested
‘river restoration ecology suffers from the inability
of ecologists to defi ne their system requirements as closely
as, for instance, engineers can specify and model (although
not necessarily achieve) their fl ood defence requirements’.
This neatly encapsulates the perception that not only is
ecological restoration uncertain but that the scientifi c basis
to predict and subsequently evaluate ecological response
is severely lacking, perhaps even absent.
One of the aims of this chapter is to determine whether
such bold statements are justifi ed in the context of the
basic premise of this volume that much restoration of
rivers has and is being conducted with considerable uncertainty
in one form or another. To do this, the sources of
uncertainty surrounding ecology (and ecologists) affecting
river restoration schemes are explored in stepwise manner,
broadly akin to the planning, design and target setting,
implementation and outcome and appraisal phases in any
project (Figure 8.1). An attempt is made to determine if
there is widespread uncertainty in the ecological response
and outcome of restoration schemes over and above that
expected as a result of natural variation, and, if so, whether
uncertainty and unreliability can be reduced.
8.2 SOURCES OF ECOLOGICAL UNCERTAINTY
8.2.1 Inherent Variability
Natural systems are inherently variable in time and space,
and probably none more so than rivers. Variation in space

Figure 8.1 Is Ecological River Restoration ‘Uncertain’? Infl uences of Institutional and Ecological Drivers in Project Planning and Implementation
Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 141

142 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
is particularly well developed both longitudinally, laterally
and vertically in rivers, as recognised by attempts to
understand them within the river continuum concept
(RCC) (Vannote et al., 1980). In larger rivers with extensive
fl oodplain systems the fl ood pulse concept (Junk et
al., 1989) has been used to frame interactions between
fl oodplain and channel, which typically vary over time.
Temporal variation may also contain other important interannual
(random, periodic or semi-periodic), annual, seasonal
and diel components, not simply amongst fauna and
fl ora but also in the rates of ecosystem processes such as
productivity and nutrient and energy fl ow.
Buijse et al. (2002) outline that whilst considerable
effort and emphasis have been directed at attempting to
quantify and understand biological, hydrological and geomorphological
interactions that together determine the
functioning of a river fl oodplain system, these are inevitably
variable and complex. Complexity occurs at the
physical level (variation in types of riverine system), the
restoration level (the combination of features requiring
restoration are site-specifi c) and the ecological level (community
interactions). Hughes et al. (Chapter 6) argue that
‘complexity’ accounts for a degree of the perceived ecological
‘uncertainty’ in river restoration and it must be
accepted that much may remain unknowable (Chapter 3).
In a wider discussion of management of nature in the
United Kingdom, Adams (1997) suggested a recent shift
in ecological thinking from looking to restore a ‘stable
equilibrium’, often taken to represent historical conditions,
to a new view that a riverine system is by its nature
dynamic and changing, i.e. one that embraces natural variability.
This is the basis of ‘recovery enhancement’ where
the scope for natural recovery is exploited, after the shackling
constraints such as structures (e.g. weirs) and modifi
ed engineered banks, management actions such as
intensive vegetation management regimes or even simply
grazing by livestock are removed. The river may also be
given the space to move (i.e. during the re-connection
of the fl oodplain as on the Rivers Cole and Brede, see
Kronvang et al., 1998; Holmes and Nielsen, 1998). Such
action to favour a potentially less controllable and predictable
system (Hughes et al., Chapter 6) may appear to the
engineer, whose target is to achieve control over the system
to meet specifi ed fl ood defence and shipping access
targets, as simple acceptance of uncertainty.
8.2.2 Limited Understanding
In spite of the wealth of work carried out on riverine
systems, several authors (de Waal et al., 1995; Buijse
et al., 2002) have argued that there is actually a lack
of scientifi c literature about river restoration, and that a
lack of robust datasets has rendered much of the work
that has been carried out unpublishable in peer reviewed
journals. Although linked, there are actually two issues
here, information exchange and scientifi c substance and
credibility.
Information exchange has undoubtedly increased enormously
as the number of river restoration projects across
the globe has increased. Malakoff (2004) reports 30 000
projects in the United States alone with tens of thousands
more to come in the next few years. The literature search
for this chapter uncovered a huge variety of information
on various river restoration projects, especially via websites
such as the European Centre for River Restoration
(http://www.rws.nl/rws/riza/home/ecrr/) and its UK
(http://www.therrc.co.uk/) and Danish counterparts (http://
www2.dmu.dk/) and the National River Restoration
Science Synthesis, which operates in seven states of the
United States, with a satellite branch in Victoria, Australia
(http://www.nrrss.umd.edu/). However, much information
is produced in an abbreviated form and it diffi cult to
determine whether many projects outlined in such a
manner are of real scientifi c substance.
At the opposite end of the spectrum lies the academic
peer-reviewed process as a vehicle for information
exchange. The fact that Ormerod (2004) found only 300+
papers including the terms river or stream restoration or
rehabilitation in the title, abstract or key words in the ISI®
database, indicates that only a fraction of projects reach
the academic literature. This could illustrate that much
work is not submitted for publication and/or many submissions
are not accepted as a result of inadequate scientifi c
rigour. It may also suggest that few academic river ecologists
are involved with restoration. In the United Kingdom
at least, ecological input in many schemes is undertaken
internally by statutory bodies such as the Environment
Agency or by consulting fi rms (like the authors of this
chapter!).
Overall, whilst there is a proliferation of information,
facilitating the rapid development of project planning,
target setting, implementation and post-project appraisal
on a global scale, the question-mark over data quality and
the lack of involvement of academic ecologists may mean
the uncertainty due to limited knowledge is not reducing
as rapidly as it might.
8.2.3 Project Selection
In the United Kingdom, until very recently, virtually all
restoration was opportunistic despite the need for a strategic
approach being recognised over a decade ago (ECON,
1993). One of the reasons for this was simply that restoration
was most effectively directed at rivers (reaches) of

Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 143
highest conservation value, i.e. those closest to the predisturbance
state. Investment in the most degraded
reaches/rivers was misplaced and wasted resources, as
they could not achieve high status as a result of overwhelming
and complex limiting factors (Kern, 1992).
Moreover, as shown for lakes (Benndorf, 1992), the
pathway from highly degraded to high quality is long and
littered with the opportunity for unpredictable and indirect
effects, which may constrain or even reduce ecological
value (i.e. colonisation by alien species). In simple terms,
long restoration trajectories (see Section 8.2.6) are more
diffi cult to plan and pull off.
8.2.4 Project Structure
It is generally acknowledged that river restoration projects
require an integrated and interdisciplinary approach,
which requires a number of experts from different fi elds
working alongside each other (NRC, 1992). The scale of
the project appears to have great infl uence as whilst this
may have been achieved in the case of large, high-profi le
schemes (e.g. the Kissimme River Restoration Project in
the USA – http://www.sfwmd.gov/org/erd/krr/ – and the
schemes tackled by the River Restoration Centre in the
UK – http://www.therrc.co.uk/) it does not appear to be
routine. Taking the schemes evaluated in the UK Rivers
and Wildlife Handbook as a sample, out of the 25 schemes
where the project team was described, 80% did not employ
an interdisciplinary approach (RSPB et al., 1995).
Resources are also clearly an issue, with a common
desire to ensure that the bulk of resources available are
channeled into ‘doing’ restoration works rather than be
absorbed into project infrastructure and institutions
(Bruce-Burgess and Skinner, 2002). The involvement of
too many organisations and stakeholders with confl icting
interests may also ultimately constrain the options for
restoration and the ability to undertake it (McDonald
et al., 2004).
From an ecological perspective it seems obvious to
suggest that the more central the position of ecologists in
the project, the greater the chance of uncertainties surrounding
ecological issues being identifi ed early in the life
of the project and subsequently tackled. The perception
that much river restoration has been undertaken with the
idea of improving at least some elements of the biota,
especially fi sh (Bruce-Burgess and Skinner, 2002;
Ormerod, 2004) gives the impression that such schemes
are ‘front-loaded’ by ecologists. Our experience is more
that ecologists enter the restoration fray to ‘count bugs’ in
the process–form–habitat–biota sequence (see previously),
particularly where an opportunity for restoration such as
on the back of a maintenance scheme has been taken.
Clearly, it is too late to set meaningful ecological targets
if the ecologists are simply used in project monitoring.
8.2.5 Goals and Target Setting
The nature of the goal underpins the ecological restoration
process. Restoration in its strictest sense involves the
return of the structure and function of an ecosystem to a
condition that existed prior to disturbance (NRC, 1992)
and this forms a ready-made goal for the project (White
and Walker, 1997). However, reliable historic information
is often absent and there is also the issue that climatic
conditions, amongst other planetary processes, may have
changed suffi ciently to mean that in ecological terms the
river may no longer function in the same manner, even if
it had remained undisturbed. The use of ecologically
similar but undisturbed contemporary reference sites or
even professional expert opinion based on empirical and/
or computational models, represent an alternative means
of establishing a suitable goal (White and Walker, 1997;
Anderson and Dugger, 1998), although any method still
requires understanding of the current nature of the system,
range of natural variation, mechanisms whereby impacts
have occurred and the specifi c effects of such impacts.
Otherwise, restoration may be doomed to failure.
The identifi cation of causes, in particular, can be diffi -
cult if they are subtle and far removed in space and time
from ecological damage (NRC, 1992). Some of the most
commonly overlooked factors are: the number of stresses
on biotic components; the multi-causal nature of degradation;
and the diversity of resulting problems. These issues
may only be resolved by adequate baseline monitoring.
Poor baseline data potentially leads to poor target setting.
Setting and achieving targets ultimately provides the
bench-measure for the success of the project. Without a
clear and appropriate goal and precise and accurate targets,
there may be a lack of benefi cial impacts, and even detrimental
ones. Kondolf (1998) describes the project at Rush
Creek, California, USA, as an example of this type of
mistake. The aim to protect the banks from erosion was
inappropriate and inconsistent with geomorphological and
ecological pressures at the site, and the aim to create a wet
meadow to stop bank erosion was probably unrealistic due
to hydrologic changes. By stopping bank erosion, the end
result of the project was that the stream’s natural processes
of recovery and re-establishment of woody riparian vegetation
was arrested.
These sort of experiences may have lead Downs and
Skinner (2002) to suggest that one of the two key aspects
limiting restoration was the inability of ecologists to set
targets (the other was monitoring and evaluation, see
below). Certainly, there does seem to be a general lack of

144 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
ecological target setting in projects. In the United
Kingdom, an evaluation of the 40 different case studies by
Holmes (1998) and a further 40 in the New Rivers &
Wildlife Handbook (RSPB et al., 1995) showed that
although projects were driven by ecological goals, only
around 65% actually set any sort or target, and only on
one scheme were quantitative targets (based around the
restoration of spawning habitat for salmonid fi shes) set
and publicised (Table 8.1). Even in the recent Cole and
Skerne projects, perhaps representing the pinnacle of restoration
in the United Kingdom to date, there is little sign
of ecological target setting, neither qualitative nor qualtitative,
the benefi ts brought about by the scheme being
assessed purely on the results of monitoring (Holmes and
Nielsen, 1998; Kronvang et al., 1998; Hoffmann et al.,
1998; Biggs et al., 1998; Vivash et al., 1998).
This is in sharp contrast with large scale schemes such
as the Kissimmee River Restoration Project RRP (KRRP)
in the USA (Trexler, 1995; Toth, 1996; Toth and Anderson,
1998; Toth et al., 1998) and the Rhine in Continental
Europe (Buijse et al., 2002). In the latter, ecological conditions
in a similarly functioning section of the Danube were
used to establish target conditions. In the former, around
60 target aims were selected (performance measures), all
of which had explicit goals associated with restoring the
full range of structural and functional processes, and
which were developed with peer review by an independent
scientifi c review panel. Targets covered the range of ecosystem
components and trophic levels and included habitat
characteristics (12 – hydrology, geomorphology, water
quality), wetland vegetation (10), food base (13 – phytoplankton,
periphyton, invertebrates, herpetofauna) and
fi sh and wildlife (25) (Whalen et al., 2002). The feasibility
Table 8.1 Proportion (%) of the 80 projects reviewed by
Holmes (1998) and presented in the New Rivers and Wildlife
Handbook (1995) in England, Wales and Northern Ireland
satisfying selected criteria
Criteria Holmes (1998) NRWH (1995)
Wholly/partially
ecologically driven
65 55
Sites strategically selected
using ecological
restoration criteria
43 30
Use of an interdisciplinary
team
50 13
With specifi c ecological
targets
65 68
With some sort of appraisal 70 100
Achieving some measure of
ecological improvement
100 92
of the targets was considered in great detail. To evaluate
the restoration of the fi sh community a conceptual model
outlining aspects of ecosystem function was developed
(Trexler, 1995). This integrated information from several
levels of biotic organisation (individuals, populations,
communities and systems) and focused on the dynamics
of fl oodplain–channel nutrients and the movement of
larvae, juvenile and adult fi sh and their macroinvertebrate
prey.
The experiences from large schemes, especially the
KRRP, suggest there are other reasons rather than a lack of
ability as to why targets are not routinely set. In simple
terms, it may be diffi cult, time consuming and expensive.
Particularly if the ‘ultimate’ approach to target setting,
namely setting quantitative targets and formulating expectations
as hypotheses that can be evaluated statistically
to provide objective evaluations, is pursued (Toth and
Anderson, 1998). Such an approach requires an extensive
information base on the river in question. Where this is
absent, gathering such data may not be cost effective, especially
in the case of reach-scale projects, since understanding
a complex system could easily take several years and
exceed the lifetime of the project (Kondolf, 1998). As a
result, a more pragmatic approach may be to set more
qualitative targets, although it should be recognised that
even qualitative targets may be evaluated in a rigorous
manner. Toth and Anderson (1998) suggest that if qualitative
expectations are to be used as success criteria they must
be expressed in an objective scientifi c manner relative to
the restoration goal. Assessments of the presence or absence
of species, for example, are highly sensitive to sampling
effort (see Section 8.2.6) and an increased frequency of
occurrence may be a response that cannot be used to differentiate
restoration from habitat enhancement.
Further pragmatism to target setting may be to simply
narrow the targets (Toth and Anderson, 1998), such as
focusing on a single species, as adopted in the recovery of
endangered species populations. Many such species, from
mammals (e.g. Eurasian otter Lutra lutra; Ottino and
Giller, 2004) to fi sh (e.g. Colorado pikeminnow Ptychocheilus
lucius in the River Colorado – van Steeter
and Pitlick, 1998; Roanoke logperch Percina rex in the
Nottoway & Roanoke rivers in Virginia – Rosenberger and
Angermeier, 2003) to invertebrates (e.g. freshwater pearl
mussel Margaritifera margaritifera – Cosgrove and Hastie,
2001; White-clawed crayfi sh Austropotamobius pallipes
– Smith et al., 1996) have been subject to in-depth assessment
of habitat requirements in riverine environments.
After detailed research, specifi c habitat attributes may be
readily defi ned for particular life history stages.
Defi ning habitat relationships between any organism
and any particular physical parameter, especially fl ow, has

Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 145
been exploited by the Instream Flow Incremental Methodology
(IFIM) (Gore and Judy, 1981; Bovee, 1982) and its
underlying model, the Physical Habitat Simulation Model
(PHABSIM) (see Appendix 8.1). PHABSIM provides a
potentially powerful tool that may enable specifi c targets
to be set for particular species where enough information
of the habitat relationships of the species concerned exists.
Specifi c sets of habitat relationships, which may feed into
restoration efforts, are particularly well advanced for charismatic
and/or commercially valuable species such as salmonid
fi shes, particularly in the United States, where
detailed and structured attempts at relatively large scale
habitat restoration for salmonids has been going on for
40 years or more (White & Brynildson, 1967; Wesche,
1985) with a wealth of detailed guides and manuals on
the web (e.g. http://www.wildfi sh.montana.edu/resources/
mammals.asp).
In some cases the use of indicators such as fi sh (or birds
or mammals) at the top of the food web may at least indicate
the direction and strength of the response of lower
trophic levels as a result of habitat restoration. This has
the basic assumption that targeted improvements in higher
trophic levels as a result of improvements in habitat diversity
are also likely to have led to a response in the diversity
of lower trophic levels (i.e. species diversity is related to
habitat diversity; Gorman and Karr, 1978). Monitoring
keystone species may not always be appropriate, however,
as it may not provide enough information about other
important processes such as organic matter spiraling and
ground–surface water interactions (Tockner and Schiemer,
1997).
Nevertheless, where multiple measures are taken, the
multiple expectations generated may perversely lead to an
ambiguous evaluation of restoration success if all expectations
are not met (Toth and Anderson, 1998). In such a
case, either all expectations need to be judged collectively,
for example by tracking a cumulative count of expectations
that are achieved over time, or perhaps expectations
could also be weighted according to set priorities. Targets
that can be expressed as constants or thresholds are
perhaps easiest to evaluate but probably will be limited to
a few attributes, for example in many systems enough
information is available to formulate constant or threshold
expectations for species richness. This approach is readily
employed for invertebrates, with the recent development
of numerous assessment criteria particularly in relation to
fl ow at the assemblage level (Appendix 8.2).
Due to the natural spatial and temporal variability that
is associated with the scale of ecosystem restoration, Toth
and Anderson (1998) suggest many targets need to be
expressed as ranges, although the breadth of the range that
is used for restoration expectations cannot be so great as
to mask or preclude evaluation of restoration success. For
example, the expectation of small fi sh densities of 0.7
to 1.3 fi sh per square foot of restored marshes on the
Kissimmee fl oodplain refl ected sampling variability whilst
still providing a useful range for evaluating restoration
success. With hindsight, our own experiences on small
rivers in the United Kingdom show that such targets could
have been readily set. For example, at several sites on the
Misbourne, the fi sh community responded in a rather predictable
way to the increase in fl ow, with one assemblage
replaced by another as habitat conditions changed (Appendix
8.3).
Target setting in large and complex schemes, especially
the KRRP, but not in smaller schemes is the inverse of that
expected. Perhaps it is simply that setting meaningful
targets and goals may be diffi cult, time consuming and
expensive. In large schemes in which huge investment of
time and resources is made, it is prudent that every effort
is made to reduce the uncertainty of the outcome, particularly
where this relates to reducible ignorance in the form
of increasing the knowledge base. Conversely, smaller
schemes, where the risks of failure may not be as great and
where resources are fewer and tend to be directed at ‘doing
restoration’, may embrace all aspects of uncertainty.
In general, contrary to the conclusions of Downs and
Skinner (2002), we suggest that ecologists may set targets
as readily and as well as engineers given the opportunity,
especially since Stewardson and Rutherfurd (Chapter 5)
argue that there may be unreasonable confi dence in the
target setting ability of geomorphologists, who tend to rely
on uncalibrated (with real fi eld data) theoretical models
dogged by poor estimation of variance.
8.2.6 Time Scales and Trajectories
In any restoration, consideration should be given to the
time scale for recovery or restoration, which is infl uenced
by how far removed the system is from the goal and targets
set, and the trajectory and pathway(s) that may be undertaken
(Gregory and Downs, Chapter 13). Anand and
Desrochers (2004) observe that the lack of long term
observational studies in virtually any ecological system
often compels researchers to rely on theoretical models to
speculate upon the trajectory that will lead the damaged
system to the desired state. According to complex systems
theory, which incorporates concepts such as chaos, the
nature and direction of a trajectory is governed by the
system’s attractor, which may be thought of as its destination.
It is hoped that the destination and desired state are
one and the same thing or the restoration attempt may be
doomed at the outset. Anand and Desrochers (2004)
eloquently illustrate the different types of attractor, with

146 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
the simplest being the progression of a system to the same
stable state regardless of initial conditions, analogous to a
climax state of vegetational succession. Where a system
has more than one attractor, it may cycle between two
alternative states. However, systems may have any number
of attracting states and the system may progress to a particular
‘basin of attraction’ depending on the starting point
or the initial conditions at which restoration started. This
offers the possibility of multiple alternative stable states,
which proved to be highly infl uential in the restoration of
shallow lakes (Figure 8.2, Scheffer et al., 1993). Ecological
stability itself is also a diffi cult concept to grasp, as
large spatial and temporal variation in populations of even
key species tuned in to their life cycles or disturbance
events (Figure 8.3) may mask underlying trends. Stability
is thus best judged over decades rather than from one year
to the next.
To the best of our knowledge such concepts have yet to
be explored in rivers, although river ‘types’ are the basis
of the Rosgen classifi cation used extensively and not
without controversy in river restoration (Malakoff, 2004).
Notable gaps (i.e. unquantifi ed uncertainties) in our conceptual
understanding of ecological dynamics within
rivers appear to constrain any ability to predict how long
nature will take to return the system to a specifi ed historic
condition or even what is to be expected.
Y
X
8.3 DEVISING A MONITORING PROGRAMME
8.3.1 Basic Design
To minimise ecological uncertainty, monitoring needs to
be undertaken prior to the work in the form of collection
of a robust set of baseline data (see previously). If the
pre-restoration system is not fully understood, the results
of the scheme cannot be correctly assessed or evaluated.
Use of reference reaches (see previously) may help set
targets and, to monitor progress against those targets, a
suitable monitoring programme needs to be devised. Monitoring
of controls (e.g. unaltered reaches) undoubtedly
helps reduce uncertainty, as this should help distinguish
what degree of change is due to natural inter-annual variation
and what is due to the restoration works (Stewardson
and Rutherfurd, Chapter 5).
Choosing which measurements of biotic and abiotic
patterns and processes to monitor is then critical, as these
need to be relevant to the goals of the project and the
targets set, and must be able to be linked to progress
towards those targets. Where the target of restoration is a
particular species, this may lead to very specifi c targets
and the desire to monitor a few aspects. However, several
workers have highlighted the need for wider surveillance
rather than aiming monitoring at one aspect of the ecosystem.
For example, Boon (1998) suggested that fi sheries-
(Left) A point attractor. The same final state is reached in both instances X and Y, although the
time taken to achieve this (shown by differing arrow lengths) varies depending on the location
of the starting position on the restoration trajectory.
(Right) A dual attractor. This functions like a pendulum swinging between two magnets, or
alternative stable states (A and B). In this instance the restoration time scales are the same,
although the final state reached may depend on other ecological factors, for example composition
of the community at the point when restoration is commenced.
Figure 8.2 Types of attractors (Reproduced from M. Anand and R. Desrochers (2004) ‘Quantifi cation of Restoration Success Using
Complex Systems Concepts and Models’ Restoration Ecology 12 (1), 117–123, with kind permisison from Blackwell Publishing.)
Y
X

Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 147
(A) The potential effects of disturbance events (d) on ecosystem structure over time.
Disturbance events could constitute anything from pollution or alien species introduction to the
impacts of restoration e.g. dredging, and recovery times differ accordingly.
(B) Variation in flow at two temporal scales.
Figure 8.3 The effects of extensive spatial and temporal scale variation on ecological stability (Reproduced from P. S. White and
J. L. Walker (1997) ‘Approximating nature’s variation: selecting and using reference information in restoration ecology’ Restoration
Ecology 5, 338–349, with kind permission from Blackwell Publishing.)
led restoration projects should take a comprehensive
ecosystem approach. Tockner et al. (1998) reached a
similar conclusion ‘A key challenge in the evaluation of
the effects of restoration is the development and testing of
an appropriate monitoring scheme, which has to include
a wide range of physical, chemical, geomorphic, and ecological
parameters.’ Monitoring a range of variables provides
more information about the system response and the
potential mechanisms behind the response of the target
variable or species.
Consequently, a number of groups across the variety of
trophic levels are typically used as indicators of ecological
status and response. For example, benthic invertebrates,
fi sh and aquatic macrophytes were selected as indicators
of the ecological status of rivers in the European
Union Water Framework Directive (EU, 2000), and benthic

148 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
invertebrates, fi sh, plankton, birds, amphibians and terrestrial
and aquatic plants have been used on the project
on the Austrian Danube (Buijse et al., 2002). For evaluating
biodiversity of river–fl oodplain complexes, a number
of species-specifi c groups that differ in their responses to
hydrological connectivity, water quality and habitat heterogeneity
should be considered (Buijse et al., 2002;
Tockner et al., 1999). The absence of certain species may
also be a good indication of ecological deterioration (EU,
2000). For example, the loss of long-distance migratory
fi sh (salmonids, coregonids, shads and sturgeons) indicates
disruption of longitudinal connectivity or a deterioration
of spawning/nursery areas.
8.3.2 Sampling Requirements
Brooks et al. (2002) pointed out that many commonly
used river restoration techniques may never have been
scientifi cally tested and, in a demonstration of the need
for rigorous scientifi c testing, carried out a fi eld experiment
designed to mimic restoration to investigate the
importance of habitat heterogeneity for macroinvertebrates.
The results showed that although a diversity of
habitats was required, macroinvertebrate populations were
actually more sensitive to individual site conditions at
each riffl e than to the heterogeneity treatments. This led
to the conclusion that the extremely high variability
between replicate riffl es meant that a monitoring programme
for localised restoration projects would be
unlikely to detect gradual shifts in community structure
until the differences between the reference and treatment
sites were extreme. Brooks et al. then suggested that innovative
measurement of other parameters, such as ecosystem
function variables (e.g. production, respiration,
decomposition), may be more appropriate indicators of
change at local scales.
This illustrates the age-old problem faced by ecologists
of huge spatial and temporal variation in their subject
matter (Figure 8.3). However, rather than looking to other
variables that may be technically diffi cult to measure,
parameterise and ultimately interpret, the only option to
overcome variability is to devise appropriate sampling
regimes of suffi cient intensity and frequency and to use
robust methods. In the case of invertebrates, methods that
allow changes to be evaluated on a community rather than
species-specifi c level are likely to be useful (Appendix
8.2). For fi sh, Bohlin et al. (1990) suggested three precision
classes for fi sheries studies depending on the nature
of the change that needs to be detected, i.e. a factor as
small as 1.2, 1.5 or as large as 2.0, with about 80% probability
when using a 5% signifi cance level. Unfortunately,
when the number of fi sh is relatively low with even moder-
ate variation around a mean value, even the lowest precision
class tends to require a high sampling effort. On the
River Lambourn, UK, during an attempt to monitor the
status of bullhead Cottus gobio, one of the species for
which the river was designated as a site of international
conservation interest (candidate Special Area of Conservation
[cSAC]), it was calculated that twenty-seven 100 m
sites would need to be surveyed over the 21 km river
to detect change on a population level (Perrow and
Tomlinson, 2002). Using the same calculations to estimate
the number of points required in PASE (see Appendix 8.3)
at each site, as few as 39 points were required where the
population of bullheads was dense, but up to 150 points
were required where bullheads were at low density. The
latter probably exceeds the number of points that can be
independently sampled, in that the points are far enough
apart to not infl uence each other, at any particular site.
Where such factors cannot be calculated in advance, rules
of thumb are often generated and in the case of PASE this
is often regarded as 50 points (Garner, 1997). Below this,
the confi dence in the estimates may be reduced, making
the response of the fi sh community to restoration diffi cult,
if not impossible, to evaluate and interpret. Too few
samples may have been another factor contributing to the
lack of a signifi cant impact of the installation of channel
enhancements (i.e. artifi cial riffl es and fl ow defl ectors)
upon fi sh abundance and species richness in low energy
lowland alluvial channels (Pretty et al., 2003).
The use of standardised methods and techniques in a
‘one size fi ts all approach’ may also be a particular
problem. Standardised approaches are particularly liked
by statutory organisations and whilst the desire for comparability
is understandable, many of the standard
approaches have been designed for objectives other than
monitoring the response of a restoration scheme. Ultimately,
such approaches may simply provide qualitative
targets with no basis for statistical comparison. In relation
to the recovery enhancement of the River Misbourne, UK
(Appendix 8.3), standard River Corridor Survey (RCS),
River Habitat Survey (RHS) and River Macrophyte Survey
(RMS) were used at each site at not inconsiderable effort.
However, all these methods simply described and mapped
the nature of the vegetation present and did not routinely
provide quantitative measures that could be evaluated statistically.
With the likely difference in hydrological regime
between the different monitoring sites along the course of
the river, the response of each site had to be evaluated
independently. Thus, the use of the standard fi sheries
survey through depletion fi shing resulting in n = 1 on each
sampling occasion could provide no more than a subjective
comparison over time as fl ow was recovered. For this
reason, a more specifi c sampling regime using PASE and

Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 149
number of replicate samples within each site was also used
(Appendix 8.3).
Even where the sampling regime has been carefully and
perhaps specifi cally developed, particularly amongst fi sheries
studies, it is essential to be clear that the effects of
the scheme are at a population level with an increase (or
decrease) in the population, and not simply a change in
distribution of the fi sh present. Whilst the latter may still
be viewed as a benefi cial response, as it illustrates that the
restoration works have produced the required sort of
habitat, this may not have actually achieved the target
set.
8.3.3 Evaluation and Deviations from
Expected Outcomes
Put simply, the measure of success of any scheme is
whether the targets have been met after a specifi ed time.
Even where the planning and science were sound, poor
implementation may have jeopardized the success of a
restoration project. For example, in 1991 an attempt was
made to restore vegetation on mud fl ats bordering the
Milwaukee River, USA, which had been previously submerged
due to a dam (Drezner, 2004). Although the area
was initially seeded with native plants, and non-native
plants were regularly pulled up, an oversight resulting in
the failure to remove Reed canary grass Phalaris arundinacea
resulted in its domination in certain areas. When
the area was surveyed 11 years later, on average 89% of
plant stems per quadrat were non-native. The reasons
behind such problems are not necessarily simple carelessness.
Frequently, budgetary decisions relating to a whole
project have to be made at the planning stage, which
means that the fi nancial fl exibility required to deal with
additional and unpredictable implementation issues may
simply not be available.
Streams and rivers have an inherent ability for natural
recovery, which has been widely exploited in restoration
(Brooks and Shields, 1996; Downs and Skinner, 2002).
Consequently, there seems to be little concern of whether
a river will be restored once the limiting factors are
removed but rather when. Thus, once habitat restoration
has taken place, ecological communities are left to recolonise.
The rate of recovery following a disturbance is
dependent upon the resilience of the community to the
perturbation in the fi rst place as well as the rate of colonisation.
Many invertebrate species are highly resilient to
disturbance through physical adaptations and behavioural
change. Magoulik and Kobza (2003) concluded that many
seek refuge from disturbance and/or have adaptations that
provide refuge, whilst Townsend (1989) highlighted the
critical role played by refugia as sources of recolonisation
after spates, and therefore as buffers against disturbance.
Important refugia include low fl ow areas (Winterbottom
et al., 1997).
The extent and intensity of channel modifi cation (i.e.
disturbance) is thought to have an impact on the time scale
of recovery. Tikkanen et al. (1994) found that recolonisation
of invertebrates occurred rapidly, within ten days,
after a small-scale rehabilitation scheme. In contrast, on
the River Rib where major habitat reconstruction was
undertaken, although there was rapid colonisation by
some macroinvertebrates, the community took almost two
years before showing indications of stabilisation. Furthermore,
Niemi et al. (1990) found that many systems took
more than fi ve years to recover to the desired ‘endpoints’,
and they concluded that the longest recovery times are
associated with disturbance that leads to long term alterations
in physical habitat. For example, the effects of an
episodic pollution event are relatively short lived whilst
recovery following dredging takes much longer since geomorphic
channel readjustment is often slow. Indeed work
carried out by Laasonon et al. (1998) on the recovery of
macroinvertebrates following stream habitat restoration
works – which took the form of boulder dams, fl ow defl ectors,
excavations and channel enlargements – indicates
that even after 16 years abundances of invertebrates were
still lower than in natural streams. Abundances of shredders
were particularly low in recently restored streams in
comparison to streams restored a number of years previously,
although even some streams that had been restored
8 or 16 years ago still contained relatively sparse shredder
populations. Further work (Muotka et al., 2002) highlighted
the importance of aquatic mosses, which are frequently
uprooted during restoration works, and which
perform important ecosystem functions, e.g. the retention
of fi ne particles and the provision of fl ow refugia for
invertebrates, which ultimately infl uence the rate of recovery
of the macroinvertebrate community. At present the
recovery rates of aquatic mosses are unknown, making it
diffi cult to predict the responses of macroinvertebrate
communities in these types of systems.
The time scales implicated in these studies indicate that
all too often schemes are probably assessed too quickly,
at a point during the colonisation process rather that at the
end point target, even if this was defi ned accurately in the
fi rst place. Short term fl uctuations in species populations
are to be expected before a longer-term stable equilibrium
is reached.
There is a wide range of factors infl uencing the rate of
recolonsiation that may simply over-ride the level of disturbance
undertaken, including the source of colonists
within the river itself, the proximity of watercourses
nearby for aerial colonsiation by winged adults and the

150 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
life cycle of the invertebrates concerned and their habitat
preferences. On the Rib, there were marked differences in
the rates of recovery of different species and four groups
were readily recognised: early colonists, medium term
colonists, seasonal colonists and long term colonists
(Figure 8.4). The Olive dun mayfl y Baetis rhodani was a
typical early colonist being an active swimmer as well as
a typical component of the drift. Whilst the Riffl e beetle
Elmis aenea also used the drift, as a resident of fast fl owing
areas it is adapted to being washed away and thus tends to
colonise more slowly. Blackfl y larvae Simulium spp. are
typically abundant in the spring and only tend to colonise
newly available areas at this time. A typical long term
colonist in this system was the predatory caseless caddis
Rhyacophila dorsalis, which relies on the adult winged
stage to colonise, as the larvae are adapted to fast fl ows
Figure 8.4 The response of the mayfl y Baetis rhodani and the riffl e beetle Elmis aenea at two re-wetted sites on the River Rib, UK.

Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 151
and are unable to tolerate slower fl ows between suitable
habitat patches.
Like invertebrates, fi sh are often assumed to be capable
of rapid colonisation and, again like invertebrates, the
speed and recovery of colonisation may be primarily
dependent on the geographical isolation of the communities
and the proximity of potential colonisers. Both downstream
and upstream movement may be readily undertaken
provided there are no barriers to the latter. What constitutes
a barrier varies enormously. Bullhead, a small
benthic fi sh is effectively limited by barriers of just 10 cm
(Utzinger et al., 1998), whereas Barbel Barbus barbus a
large (to 1 m) powerful cyprinid in European rivers has
been recorded making movements of up to nearly 20 km,
crossing large weirs (to 2 m) in the process (Lucas and
Bately, 1996). The ability of many anadromous salmonids
(salmon and trout) to leap large obstacles is well known.
Even though the restored area may have been recolonised,
recruitment success can still fl uctuate owing to a wide
range of factors not related to the scheme (e.g. temperature,
predation of eggs and larvae etc). Such factors can
affect both the abundance and distribution of individuals,
and may signifi cantly slow the colonisation process.
Moreover, the major structuring forces of predation and
competition may rage for some time before a stable confi
guration is reached. In the Misbourne for example, sticklebacks
may have out-competed other species for some
time, even as physical conditions changed, simply as they
arrived fi rst and achieved high density as a result of the
abundance of potential nest sites and fl ow refugia in the
form of emergent and then submerged vegetation. Only
when these were removed, did the community shift rapidly,
with bullheads dominating through rapid recruitment
(Appendix 8.3).
In general, vagaries in the abilities of different organisms
to colonise intuitively mean that this stage of the
project is highly vulnerable to the infl uence of ecological
uncertainty. Predicting the likely pattern of colonisation
may be problematic, due to both a limited understanding
of other populations in the local area, and a general lack
of scientifi c knowledge. For example, the factors infl uencing
the colonisation of submerged macrophytes, which are
of enormous ecological importance, are poorly understood.
Clearly, better knowledge of the colonisation mechanisms
and movement patterns is vital to understand the
processes behind restoration of riverine systems (Fenoglio
et al., 2002) and allow a better understanding of the time
scale required for recovery to a stable end point. A lack of
scientifi c knowledge is not necessarily easy to overcome,
since the situation at one site may differ substantially from
another and at present there is a lack of longer term studies
that may feed into the knowledge base (Figure 8.1).
8.3.4 Appraisal and Feedback
The importance of appraisal has been highlighted by
Anderson and Dugger (1998) who concluded that ‘Failure
to evaluate projects not only precludes learning anything
about a particular restoration, but it also limits the opportunity
for improving plans for future projects’. In our experience
in the United Kingdom the number of projects that
incorporate post-project appraisal (PPA) is increasing following
early efforts to raise the importance of this aspect
(ECON, 1993). Reviewing projects from 1991–1996,
Holmes (1998) recorded that 53% of projects had undergone
PPA, 23% of projects did not incorporate any form
of PPA and for 25% of projects this information was not
known. Today, partly as a result of the efforts of the RRP
demonstration projects and subsequently the River Restoration
Centre (RRC) within the European Centre for River
Restoration (see previously), we would suggest that perhaps
only the smaller scale schemes do not routinely undertake
PPA. As well as the importance of shared information,
drivers have included the need for organisations such as
the Environment Agency to demonstrate cost effi ciency
for their ‘business’ needs, and that the publicity machines
of the organisations involved like/need to have something
to show to the public and other interested parties.
However, there is potentially little control over the
quality of the information produced, with accumulating
ecological uncertainties in the earlier stages of the project
potentially ultimately resulting in misleading information
that could be used to inform future projects. This is a
variant of the ‘Almost as bad as no evaluation are poorly
planned efforts that waste limited resources while providing
meaningless or even misleading information’ warned
by Anderson and Dugger (1998). It is diffi cult if not
impossible to assess how much this type of situation
occurs. Although ecological uncertainty may frequently
occur, the concern is that it goes unrecognised, which
could lead to inaccurate results and misinformation. If
ecological uncertainty is recognised, and if the scheme
results in a clear improvement, even if it is not exactly as
predicted, this is perhaps less of a concern. However, there
is the possibility that the project may be held in less
esteem or considered not as successful if targets are not
met, and project funding and the prospect of continued
restoration may be jeopardised, even though the work may
have ultimately had a positive impact on the ecology of
the site. Schemes are beginning to allow for this eventuality
and are constructed to allow adaptive management if
the scheme needs to be adjusted to obtain more desirable
results.
Even if appraisal is conducted, it is often not easy to
evaluate success and for this reason there is a major move

152 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
from river ecologists to establish a set of criteria for measuring
the ecological success of schemes (Palmer et al.,
2005). It is hoped that this will fi nally be endorsed by the
United Nations Environmental Programme. The criteria
specifi ed are that:
the design should be based on a specifi ed guiding image
that a more dynamic, healthy river could exist at the
site;
the river’s ecological condition must be measurably
improved;
the river system must be more self-sustaining and resilient
to external perturbations so that only minimal
follow-up maintenance is needed.
during the construction phase, no lasting harm should
be infl icted on the ecosystem.
both pre- and post-assessment must be completed and
data made publicly available.
The emphasis behind these criteria is the desire to have
projects that are funded and implemented in the name of
ecological restoration evaluated according to ecological
indicators of success rather than according to other social,
economic and institutional factors, such as cost effectiveness,
stakeholder satisfaction and aesthetic/recreational
value (Palmer et al., 2005). These criteria are fl exible
enough to apply to both large scale major projects and
smaller projects where complex and expensive design is
not appropriate. They also encourage the view of restoration
success as an adaptive process rather than a single
defi ned end point.
8.4 CONCLUSIONS AND RECOMMENDATIONS
Throughout this chapter we have sought to outline sources
of unwanted uncertainty that ultimately reduce confi dence
in restoration and hinder the development of a cost effective,
repeatable approach. The latter can only be achieved
by contributing to a shared knowledge base, the potential
for which is now enormous with a number of dedicated
restoration journals and even more specifi c websites. The
approach adopted accepts that much uncertainty resulting
from natural variation and sheer system complexity, especially
in large systems, will remain unknowable. Fortunately,
this has not dampened the desire for restoration of
large systems, which make greatest contribution to biodiversity,
as recent initiatives within the European Union to
restore rivers such as the Danube (Tockner and Schiemer,
1997), the Rhine (Buijse et al., 2002) and the Drava and
Lech rivers in Austria (Mohl, 2004) testify.
Indeed, investment in large systems such as the KRRP
has illustrated that many uncertainties may be quantifi ed
and overcome, with the accumulation of an extensive
knowledge and baseline data, gathered over a number of
years, which in turn required considerable resources. With
such effort, predictions and targets can be detailed, quantitative
and set to within a realistic time scale. This quells
the notion of the inability of ecologists to set targets which
has been raised as a particular source of uncertainty. Our
experiences, especially in the United Kingdom, suggest
that the lack of target setting cannot be attributed to a
particular inability to do so but rather appears to be more
due to a lack of resources in small projects especially.
With an increase in post-project appraisal of whether the
project succeeded or failed, it is more diffi cult to justify
not using targets at all. Project ‘culture’ thus appears to
be changing for the better.
There is probably a better knowledge base of the fauna–
habitat information than is immediately apparent and this
may be developed specifi cally where this is lacking,
although this may be expensive of time and resources.
Habitat relationships have, after all, been a fundamental
component of many species-centred restoration projects,
including in rivers. Powerful tools such as PHABSIM and
a wide range of classifi cation systems are available to
predict and quantify change even in species-rich groups
such as invertebrates, although of course the limitations
of such tools do need to be recognised.
There seems to be historical sense of ecologists being
advocates of a ‘black art’, perhaps particularly to hardened
river engineers dealing in mathematical formulae. This
may have originated from ecologists being employed to
count ‘bugs’ or fi sh or record plant abundance, originating
from engineering works. Ecology is essentially a numerate
science that can be quantifi ed by relationships and patterns
and is no more ‘uncertain’ than any other science, although
it does deal with extremely complex systems and the relationships
and interactions within them. Ecological understanding
also often takes time and resources. Bringing
ecologists into a more central role, with involvement from
the early stages of the project, may help reduce uncertainty.
Engaging a greater range of academic river ecologists
with wide understanding of catchment processes,
theoreticians and ecological modellers, as well as the
‘samplers and sorters’, who are frequently consultants, is
also thought likely to be particularly benefi cial.
Even a lack of historical reference data may be partly
alleviated by monitoring reference (control) reaches before
and after the impact of the restoration scheme in a before–
after–control–impact (BACI) design (e.g. as on the Rivers
Cole and Brede; biggs et al., 1998). Particularly where this
incorporates any historical information, it lends itself well
to target setting with the subsequent means of assessing
whether the target in the controls is reached. As more

Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 153
information is gathered, a fl exible adaptive management
approach allows targets to be adjusted (or perhaps even
hardened from qualitative to quantitative, if necessary),
enables further studies to be planned when needed and can
recognise and deal with emerging problems.
The rigorous selection or even design of appropriate
monitoring techniques is essential if the major source
of uncertainty in the monitoring phase is to be reduced.
The use of standard techniques simply because they are
routinely used by the organisation involved, is to be
guarded against, as such methods have often been developed
for a different purpose. It is far better to treat each
scheme as a unique individual experiment for which the
appropriate method, intensity and frequency of sampling
is calculated. In other words, monitoring should be seen
as detailed research rather than a mere estimation of
general changes.
Although rivers have enormous resilience and an
inherent ability for natural recovery the time scale and
trajectory of recovery/restoration represents a major
source of uncertainty, particularly where a stable endpoint
has been poorly defi ned or there is little understanding of
what that may be. The ‘how long is a piece of string’
analogy does not sit well with project budgets, which are
typically fi nite and allocated to a particular time frame.
We suggest that the trajectory of restoration may be more
direct and shortened by considered project selection in the
fi rst place, where restoration is part of a generic theme
adopted by a statutory body or interested organisation.
Tackling less damaged sites with good prospects for
recovery, particularly where this can be undertaken through
recovery enhancement and not major habitat reconstruction,
is also likely to lead to a reduced time scale for
recovery.
There is clearly a need for more work on the factors
surrounding colonisation to enable better predictions to be
ultimately made. This may need to be conducted specifi -
cally for the river concerned within the project in an adaptive
manner and may also require action to speed up the
colonisation process. For example, with problem species
such as invertebrates with particularly poor dispersal abilities,
it may be prudent to introduce these, perhaps through
‘seeding’ of substrate, to reach a stable confi guration as
soon as possible. For fi sh, it is important to defi ne any
barriers to natural movement that again, may be readily
overcome by translocation.
Obviously, the strength of post-project appraisal ultimately
depends on the uncertainties accumulated earlier
in the project. Where these are great this could deliver a
fl awed evaluation of little or no value to future projects.
Following the approach outlined it should be possible to
avoid this scenario in the future.
ACKNOWLEDGEMENTS
We are grateful to the Environment Agency for funding
several of the studies documented and to its staff for assistance
with sampling. ECON staff undertook much of the
work on the River Misbourne.
REFERENCES
Adams W. 1997. Rationalization and conservation: Ecology and
the management of nature in the UK. Transactions of the
Institute of British Geographers 22: 227–291.
Anand M, Desrochers R. 2004. Quantifi cation of restoration
success using complex systems concepts and models. Restoration
Ecology 12: 117–123.
Anderson DH, Dugger BD. 1998. A conceptual basis for evaluating
restoration success. Transactions of the North American
Wildlife and Natural Resources Conference 63.
Benndorf J. 1992. The control of indirect effects of biomanipulation.
In: Sutcliffe DW, Jones JG (Eds), Eutrophication:
Research and Application to Water Supply. Freshwater Biological
Association: Cumbria, UK; 82–93.
Biggs J, Cornfi eld A, Grøn R et al. 1998. Restoration of the rivers
Brede, Cole and Skerne: a joint Danish and British EU–LIFE
demonstration project, V – Short–term impacts on the conservation
value of aquatic macroinvertebrate and macrophyte
assemblages. Aquatic Conservation: Marine and Freshwater
Ecosystems 8: 241–255.
Bohlin T. 1990. Estimation of population parameters using electric
fi shing: aspects of the sampling design with emphasis on
salmonids in streams. In: Cowx IG, Lamarque P (Eds), Fishing
with Electricity Fishing News Books: Oxford, UK; 156–173.
Boon PJ. 1998. River restoration in fi ve dimensions. Aquatic
Conservation: Marine and Freshwater Ecosystems 8:
257–264.
Bovee KD. 1982. A Guide to Stream Habitat Analysis Using the
Instream Flow Incremental Methodology. Instream Flow Information
Paper No.12. US Fish and Wildlife Service, FES/OBS–
82/86: Washington, DC.
Bradshaw AD. 1987. Restoration: an acid test for ecology. In:
Jordan WR, Gilpin ME, Aber JD (Eds), Restoration Ecology
Cambridge University Press: Cambridge, UK; 23–29.
Brooks A, Shields FD Jr. 1996. Towards an approach to sustainable
river restoration. In: Brookes A, Shields FD Jr., River
Channel Restoration: Guiding Principles for Sustainable Projects.
John Wiley & Sons Ltd: Chichester, UK; 385–402.
Brooks SS, Palmer MA, Cardinale BJ et al. 2002. Assessing
stream ecosystem rehabilitation: Limitations of community
structure. Restoration Ecology 10: 156–168.
Bruce-Burgess L, Skinner L. 2002. Appraisal: River restoration’s
missing link. Proceedings of Conference held 27th November
2002 at the University of Nottingham, Downloadable from
http://www.therrc.co.uk.
Buijse AD, Coups H, Staras M et al. 2002. Restoration strategies
for river fl oodplains along large lowland rivers in Europe.
Freshwater Biology 47: 889–907.

154 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Calow P. 1992. Energy budgets. In: Calow P, Petts G (Eds), The
Rivers Handbook, Volume 1. Blackwell Scientifi c Publications:
Oxford, UK; 370–378.
Carpenter SR, Kitchell JF. 1993. The Trophic Cascade in Lakes.
Cambridge University Press: Cambridge, UK.
Carle FL, Strub MR. 1978. A new method for estimating population
size from removal data. Biometrics 34: 621–630.
Chandler JR. 1970. A biological approach to water quality management.
Water Pollution Control, 69: 415–422.
Clarke SJ, Bruce-Burgess L, Wharton G. 2003. Linking form and
function: towards an ecohycromorphic approach to sustainable
river restoration. Aquatic Conservation: Marine and Freshwater
Ecosystems. 13: 439–450.
Copp GH, Peñáz 1988. Ecology of fi sh spawning and nursery
zones in the fl ood plain, using a new sampling approach.
Hydrobiologia 169: 209–224.
Cosgrove PJ, Hastie LC. 2001. Conservation of threatened freshwater
pearl mussel population: river management, mussel
translocation and confl ict resolution. Biological Conservation
99: 183–190.
de Waal LC, Large ARG, Gippel CJ, Wade PM. 1995. River and
fl oodplain restoration in western Europe: opportunities and
constraints. Archiv für Hydrobiologie 101: 679–693.
Downs P, Skinner KS. 2002. Rivers and streams. In: Perrow MR,
Davy AJ (Eds), Handbook of Ecological Restoration. Volume
2: Restoration in Practice. Cambridge University Press: Cambridge,
UK; 267–296.
Drezner TD. 2004. Few native species colonize on mud fl ats ten
years after dam removal. Restoration Ecology 22: 50–51.
ECON 1993. The River Restoration Project, Phase 1: the Feasibility
Study. Report to River Restoration Project (RRP), UK.
Extence CA, Balbi DM, Chadd RP. 1999. River fl ow indexing
using British benthic macro-invertebrates: a framework for
setting hydroecological objectives. Regulated Rivers Research
and Management 15: 543–574.
EU. 2000. Directive 2000/60/EC of the European Parliament
and of the council establishing framework for the community
action in the fi eld of water policy. (Website: http://europa.eu.
int/comm/environment/water/water-framework/index_en.
html).
Fenolglio S, Agosta P, Bo T, Cucco M. 2002. Field experiments
on colonization and movements of stream invertebrates in an
Apennine river (Visone, NW Italy). Hydrobiologia 474:
125–130.
Gameson ALH, Wheeler A. 1977. Restoration and the recovery of
the Thames Estuary. In: Cairns J Jr., Dickson KL, Herricks EE
(Eds), Recovery and Restoration in Damaged Ecosystems
University Press of Virginia: Charlottesville, Virginia; 72–101.
Garner P. 1997. Sample sizes for length and density estimation
of 0+ fi sh when using point sampling by electrofi shing. Journal
of Fisheries Biology 50: 95–106.
Gore SA, Judy RD Jr. 1981. Predictive models of benthic macroinvertebrate
density for use in instream fl ow studies and
regulated fl ow management. Canadian Journal of Fisheries
and Aquatic Sciences 38: 1363–1370.
Gorman OT, Karr JR. 1978. Habitat structure and stream fi sh
communities. Ecology 59 (3): 507–515.
Harper DM, Smith CD, Barham PJ. 1992. Habitats as the building
blocks for river conservation assessment. In: Boon PJ, Calow
P, Petts GE (Eds), River Conservation and Management. John
Wiley & Sons Ltd: Chichester, UK; 311–319.
Hearne J, Johnson IW, Armitage PD. 1994. Determination of
ecologically acceptable fl ows in rivers with seasonal changes
in the density of macrophytes. Regulated Rivers: Research and
Management 9: 117–184.
Hoffmann CC, Pedersen ML, Kronvang B, Øvig L. 1998. Restoration
of the Rivers Brede, Cole and Skerne: a joint Danish and
British EU–LIFE demonstration project, IV – implications for
nitrate and iron transformation. Aquatic Conservation: Marine
and Freshwater Ecosystems 8: 223–240.
Holmes NTH. 1998. A Review of River Rehabilitation in the UK,
1990–1996. Technical Report to the Environment Agency
(W175), Environment Agency: Bristol, UK.
Holmes NTH, Nielsen MB. 1998. Restoration of the rivers Brede,
Cole and Skerne: a joint Danish and British EU–LIFE demonstration
project, I – Setting up and delivery of the project.
Aquatic Conservation: Marine and Freshwater Ecosystems 8:
185–196.
Jeppesen E, Sammalkarpi I. 2002. Lakes In: Perrow MR, Davy
AJ (Eds), Handbook of Ecological Restoration. Volume 2: Restoration
in Practice. Cambridge University Press: Cambridge,
UK; 297–324.
Junk WJ, Bayley PB, Sparks RE. 1989. The fl ood pulse concept
in river-fl oodplain systems. Special Publication of the Canadian
Journal of Fisheries and Aquatic Sciences 106:
110–127.
Kern K. 1992. Rehabilitation of streams in South-west Germany.
In: Boon PJ, Calow P, Petts GE (Eds), River Conservation and
Management. John Wiley & Sons Ltd: Chichester, UK;
321–335.
Kondolf GM. 1998. Lessons learned from river restoration projects
in California. Aquatic Conservation: Marine and Freshwater
Ecosystems 8: 39–52.
Kronvang B, Svendsen LM, Brookes A et al. 1998. Restoration
of the rivers Brede, Cole and Skerne: a joint Danish and British
EU–LIFE demonstration project, III – channel morphology,
hydrodynamics and transport of sediment and nutrients.
Aquatic Conservation: Marine and Freshwater Ecosystems 8:
209–222.
Laasonen P, Muotka T, Kilvijärvi I. 1998. Recovery of macroinvertebrate
communities from stream habitat restoration.
Aquatic Conservation: Marine and Freshwater Ecosystems 8:
101–113.
Lepori F, Palm D, Malmqvist B. 2005. Effects of stream restoration
on ecosystem functioning: detritus retentiveness and
decomposition. Journal of Applied Ecology 42: 175–189.
Lucas MC, Batley E. 1996. Seasonal movements and behaviour
of adult barbel Barbus barbus, a riverine cyprinid fi sh: implications
for river management. Journal of Applied Ecology 33:
1345–1358
Magoulick DD, Kobza R. 2003. The role of refugia for fi shes
during drought: a review and synthesis. Freshwater Biology 48:
1186–1198.
Malakoff D. 2004. The River Doctor. Science 305: 937–939.

Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 155
McDonald A, Lane SN, Haycock NE, Chalk EA. 2004. Rivers of
dreams: on the gulf between theoretical and practical aspects
of an upland river restoration. Transactions of the Institute of
British Geographers 29: 257–281.
Milner NJ, Hemsworth RJ, Jones BE. 1985. Habitat evaluation
as a fi sheries management tool. Journal of Fish Biology 27
(Supplement A): 85–108.
Mohl A. 2004. LIFE River Restoration projects in Austria. In:
Geres D (Ed) River Restoration 2004 – Principles, Processes,
Practices. Proceedings of the 3rd ECRR International Conference
on River Restoration in Europe, Zagreb: Croatia (Papers
downloadable from http://www.ecrr.org/).
Muotka T, Paavol R, Haapala A et al. 2002. Long-term recovery
of stream-habitat structure and benthic invertebrate communities
from in-stream restoration. Biological Conservation 105:
243–253.
Nestler JM, Schneider LT, Latka D, Johnson P. 1996. Impact analysis
and restoration planning using the riverine community
habitat assessment and restoration concept (RCHARC). In:
Leclerc M, Capra H, Valentin S et al. (Eds), Ecohydraulics 2000.
Proceedings of the 2nd International Symposium on Habitat
Hydraulics, INRS – Eau: Québec, Canada; A871-A876.
Newbold JD. 1992. Cycles and spirals of nutrients. In: Calow P,
Petts GE (Eds), The Rivers Handbook Volume 1. Blackwell
Scientifi c Publications: Oxford, UK; 79–408.
Newson MD, Pitlick J, Sear DA. 2002. Running water: fl uvial
geomorphology and river restoration. In: Perrow MR, Davy AJ
(Eds), Handbook of Ecological Restoration. Volume 1: Principles
of Restoration. Cambridge University Press: Cambridge,
UK: 324–354.
Niemi GJ, DeVore P, Detenheck N et al. 1990. Overview of case
studies on recovery of aquatic systems from disturbance.
Environmental Management 14: 571–587.
NRA (National Rivers Authority)/Kirkpatrick Scott Wilson.
1992. Method for Assessment of Low Flow Conditions Caused
by Abstraction. National Rivers Authority, R & D Note 45.
Environment Agency: Bristol, UK.
NRA (National Rivers Authority). 1995. Surface Water Abstraction
Licensing Policy Development. Project Report 505,
National Rivers Authority: Bristol, UK.
NRC (National Research Council US). 1992. Chapter 5: rivers
and streams. In: Restoration of Aquatic Ecosytems National
Academy Press: Washington, DC.
Ormerod SJ. 2004. A golden age of river restoration science?
Aquatic Conservation: Marine & Freshwater Ecosystems. 14
(6): 543–549.
Ottino P, Giller P. 2004. Distribution, density, diet and habitat use
of the otter in relation to land use in the Araglin Valley, Southern
Ireland. Biology and Environment: Proceedings of the
Royal Irish Academy 104B (1): 1–17.
Palmer MA, Bernhardt ES, Allan JD et al. 2005. Standards for
ecologically successful river restoration. Journal of Applied
Ecology, 42 (2): 208–217.
Perrow MR, Jowitt AJD, Zambrano González L. 1996. Sampling
fi sh communities in shallow lowland lakes: point-sample electric
fi shing vs electric fi shing within stop-nets. Fisheries
Management & Ecology 3: 303–313.
Perrow MR, Tomlinson ML, Zambrano L. 2002. Fish. In: Perrow
MR, Davy AJ (Eds), Handbook of Ecological Restoration.
Volume 1: Principles of Restoration. Cambridge University
Press: Cambridge, UK; 324–354.
Perrow MR, Davy AJ (Eds). 2002a. Handbook of Ecological
Restoration. Volume 1: Principles of Restoration. Cambridge
University Press: Cambridge, UK.
Perrow MR, Davy AJ (Eds). 2002b. Handbook of Ecological
Restoration. Volume 2: Restoration in Practice. Cambridge
University Press: Cambridge, UK.
Perrow MR, Tomlinson ML. 2002. River Lambourn cSAC baseline
survey: Bullhead and Lamprey populations. Report to
English Nature.
Pretty JL, Harrison SSC, Shepherd DJ et al. 2003. River rehabilitation
and fi sh populations: assessing the benefi t of instream
structures. Journal of Applied Ecology 40: 251–265.
Punchard NT, Perrow MR, Jowitt AJD. 2000. Fish habitat associations,
community structure, density and biomass in natural and
channelized lowland streams in the catchment of the River
Wensum, UK. In: Cowx IG (Ed.), Management and Ecology
of River Fisheries. Blackwell Science: Oxford, UK; 143–157.
Richardson JS, Jackson MJ. 2002. Aquatic invertebrates. In:
Perrow MR, Davy AJ (Eds), Handbook of Ecological Restoration.
Volume 1: Principles of Restoration. Cambridge University
Press: Cambridge, UK; 300–323.
Robinson C, Whitton S. 2004. Fisheries Action Plans: a new
approach to public consultation and the impetus for habitat
enhancement works in the upper Thames catchment (England).
In: Geres D (Ed.), River Restoration 2004 – Principles, Processes,
Practices. Proceedings of the 3rd ECRR International
Conference on River Restoration in Europe: Zagreb, Croatia.
Papers downloadable from http://www.ecrr.org.
Roni P, Beechie TJ, Bilby RE et al. 2002. A review of stream
restoration techniques and a hierarchical strategy for prioritizing
restoration in the Pacifi c Northwest watersheds. North
American Journal of Fisheries Management 22: 1–20.
Rosenberger A, Angermeier PL. 2003. Ontogenetic shifts in
habitat use by the endangered Roanoke logperch (Percina rex).
Freshwater Biology 48: 1563–1577.
RSPB (Royal Society for the Protection of Birds), NRA (National
Rivers Authority) and the RSNC (Royal Society for Nature
Conservation). 1995. The New Rivers and Wildlife Handbook
The Royal Society for the Protection of Birds: Sandy, Bedfordshire,
UK.
Russell IC. 1994. Salmon stocking in north-east England – some
factors affecting return rates. In: Cowx IG (Ed.), Rehabilitation
of Freshwater Fisheries. Fishing News Books, Blackwell Scientifi
c Publications Ltd: Oxford, UK; 255–267.
Scheffer M, Hosper SH, Meijer M-L et al. 1993. Alternative
equilibria in shallow lakes. Trends in Ecology and Evolution 8:
275–279.
Smith GRT, Learner MA, Slater FM, Foster J. 1996. Habitat features
important for the conservation of the native crayfi sh in
Britain. Biological Conservation 75: 239–246.
Tikkanen P, Laasonen O, Muotka T et al. 1994. Short term recovery
of benthos following disturbance from stream habitat rehabilitation.
Hydrobiologia 273: 121–130.

156 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Tockner K, Schiemer F. 1997. Ecological aspects of the restoration
strategy for a river-fl oodplain system on the Danube River
in Austria. Global Ecology and Biogeography Letters 6:
321–329.
Tockner K, Schiemer F, Ward JV. 1998. Conservation by restoration:
the management concept for a river-fl oodplain system on
the Danube River in Austria. Aquatic Conservation 8: 71–86.
Tockner K, Schiemer F, Baumgartner C et al. 1999. The Danube
Restoration Project: Species diversity patterns across connectivity
gradient6s in the fl oodplain system. Regulated Rivers:
Research and Management. 15: 245–258.
Toth LA. 1996. Restoring the hydrogeomorphology of the channelised
Kissimmee River. In: Brookes A, Shields FD Jr (Eds),
River Channel Restoration: Guiding Principles for Sustainable
Projects John Wiley & Sons Ltd: Chichester, UK; 369–383.
Toth LA, Anderson DH. 1998. Developing expectations for ecosystem
restoration. In: Transactions of the 63rd North American
Wildlife Natural Resources Conference: Washington, DC;
122–134.
Toth LA, Melvin SL, Arrington DA, Chamberlain J. 1998. Hydrological
manipulations of the channelised Kissimmee River:
Implications for restoration. Bioscience 48: 758–764.
Townsend CR. 1989. The patch dynamics concept of stream community
ecology. Journal of the North American Benthological
Society 8: 36–50.
Trexler JC. 1995. Restoration of the Kissimmee River: a conceptual
model of past and present fi sh communities and its consequences
for evaluating restoration success. Restoration Ecology
3: 195–210.
Utzinger J, Roth C, Peter A. 1998. Effects of environmental
parameters on the distribution of bullhead Cottus gobio with
particular consideration of the effects of obstructions. Journal
of Applied Ecology 35: 882–892.
Van Steeter MM, Pitlick J. 1998. Geomorphology and endangered
fi sh habitats of the Upper Colorado River. 1: Historic changes
in streamfl ow, sediment load and channel morphology. Water
Resources Research 34: 287–302.
Vannote RL, Minshall GW, Cummins KW et al. 1980. The river
continuum concept. Canadian Journal of Fisheries and Aquatic
Sciences 37: 130–137.
Vivash R, Ottosen O, Janes M, Sørensen HV. 1998. Restoration
of the rivers Brede, Cole and Skerne: a joint Danish and British
EU–LIFE demonstration project, II – the river restoration
works and other related practical aspects. Aquatic Conservation:
Marine and Freshwater Ecosystems 8: 197–208.
Waldman JR, Wirgin II. 1998. Status and restoration options for
Atlantic sturgeon in North America. Conservation Biology 12:
631–638.
Wesche TA. 1985. Stream channel modifi cations and reclamation
structures to enhance fi sh habitat. In: Gore JA (Ed.), The
Restoration of Rivers and Streams: Theories and Experience.
Butterworth Publishers: Massachusetts, USA; 103–163.
Whalen PJ, Toth LA, Koebel JW, Strayer PK. 2002. Kissimmee
River Restoration: A Case Study. Water Science and Technology
45: 55–62.
White RJ, Brynildson OM. 1967. Guidelines for Management of
Trout Stream Habitat in Wisconsin. Technical Bulletin No. 39,
Division of Conservation, Department of Natural Resources:
Madison, Wisconsin.
White PS, Walker JL. 1997. Approximating nature’s variation:
selecting and using reference information in restoration
ecology. Restoration Ecology 5: 338–349.
Winterbottom JH, Orton SE, Hildrew AG. 1997. Field Experiments
on fl ow refugia in streams. Freshwater Biology 37:
569–588.
APPENDIX 8.1 THE USE AND POTENTIAL
LIMITATIONS OF IFIM AND PHABSIM
The ‘Instream fl ow incremental methodology’ (IFIM) and
its underlying model, the ‘Physical habitat simulation
model’ (PHABSIM) represent the most popular approach
for simulating habitat preferences (Downs and Skinner,
2002). The relationship between streamfl ow and available
physical habitat (defi ned by depth, velocity, substrate and
cover) is used to compute the ‘weighted usable area’
(WUA) in a reach as a function of the river discharge, for
different life stages and species of fi sh. For each life
stage of the target species, the model requires expressions
of the relative suitability for that species of the full
range of values taken by these variables. These univariate
curves or habitat suitability indices may be derived from
existing literature, expert opinion or by sampling techniques
such as electro-fi shing or snorkelling. PHABSIM
also contains a number of hydraulic models that predict
values of depth and velocity at different simulation discharges.
These models require calibration using fi eld data
collected at two or more calibration discharges. Observations
of substrate and cover are recorded using a coding
system and are assumed to be independent of discharge.
Once calibrated the model can simulate values of microhabitat
variables over the full range of discharge within a
river reach. Combining the results with habitat suitability
data produces the WUA versus Discharge relationship
(CEH website: http://www.nwl.ac.uk/ih). PHABSIM,
by relating habitat to discharge, provides a quantitative
entity, allowing river ecologists to negotiate prescribed
fl ows in equivalent terms to other water resource demands,
and offers a practical means of integrating ecological
requirements of aquatic species with other water resource
demands.
However, numerous procedural, biological and physical
limiting assumptions prevented IFIM and PHABSIM
from being reliable and, in response to these criticisms, a
wide range of more complex approaches based on
advanced understanding of the bioenergetics of fi sh
biology and on more realistic representation and modelling
of fl ow patterns is now at the research stage (3rd
International Symposium on Ecohydraulics, 1999).

Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 157
At the Instream Flow Incremental Methodology Workshop
held in New Zealand (February 2004), jointly hosted
by Fish and Game New Zealand and the Department of
Conservation, Cawthron Institute NIWA (Conference
website: http://www.doc.govt.nz/Explore/Hunting-and-
Fishing/Taupo-Fishery/), several speakers (e.g. Maclean
and Death) expressed concern about some of the underlying
assumptions behind PHABSIM, including that: fi sh are
limited by habitat; that habitat preferences are adequately
described by depth, velocity and substrate type; that habitat
preferences do not change in accordance with diurnal/seasonal
patterns; and that a large area of suboptimal habitat
is preferable to a small area of optimal habitat.
It was also argued that PHABSIM was not appropriate
for use in connection with New Zealand (NZ) invertebrates
on the basis that depth, velocity and substrate type
were of minor importance in comparison with other factors
such as riparian catchment vegetation, disturbance and
food supply. Habitat requirements also appeared to be
fl exible for many NZ species. The use of habitat suitability
curves was also brought into question and it was noted
that it needs to be made explicitly clear how they were
developed, for example which sites were used, the number
of samples taken and whether diurnal patterns were
accounted for in the sampling process. The lack of variance
estimation (i.e. error bars) was also raised as a limitation.
Reduced (or increased) fl ows might have effects on
other variables that are not taken into account, such as
nutrient content, sediment build up, and temperature.
PHABSIM and variants such as the Riverine Community
Habitat Assessment and Restoration Concept
(RCHARC) (Nestler et al., 1996) in the United Kingdom
on lowland chalk streams suffering from low fl ow have
also proved to be of limited use for invertebrates. There is
the risk that habitat area/habitat response predictions will
be meaningless if the sampling transects are not widely
representative. These problems may be more marked for
smaller rivers and streams in the UK context. It is known
that for chalk streams and other nutrient-rich lowland
rivers the seasonal changes in the growth and biomass of
macrophytes require, at the very least, a careful calibration
of PHABSIM to prevent distortion of results (Hearne et
al., 1994).
Overall, whilst PHABSIM is a potentially powerful tool
it may be the most effective when evaluating and comparing
different management options, in instances where
physical habitat limits populations. The United Kingdom
method only assesses physical habitat, so factors such as
water quality and temperature and sediment transport
require complementary studies. Indeed, where such factors
are the prime constraint on populations the use of
PHABSIM is inappropriate.
Much criticism centres on this latter point and
PHABSIM like any model is only as good as the data that
is entered and makes the assumption that habitat variables
are the best descriptor of the abundance of a particular
group, which of course need not be the case. Also, the
quantity and quality of water cannot limit the distribution
or abundance of the specifi ed organism in the fi rst place.
Whilst there is an inherent desire for scientists to adopt a
quantitative approach in order to reduce uncertainty, this
is only successful if a high level of confi dence can be
placed in the predictions. In the case of PHABSIM, many
users have found that predictions for even ‘straightforward’
schemes do not come close to the quantitative
targets set. However, PHABSIM can also be used as a
qualitative tool, and may be useful when it comes to
selecting the best likely restoration option from a number
of possible schemes.
APPENDIX 8.2 CLASSIFICATION TECHNIQUES
FOR MONITORING AND POTENTIALLY
PREDICTING THE RESPONSE OF
INVERTEBRATES TO FLOW RESTORATION
In the United Kongdom low fl ows are seen to be a problem,
especially in chalk streams, which are characterised by
high invertebrate biomass and abundance and a relatively
high diversity of species. The main physical effects of low
fl ows in chalk streams are: long term drying of ephemeral
or winterbourne reaches; long term drying of spring seepages
or fl ushes; downstream migration of the perennial
head of a chalk stream or river; and reduced water levels
and/or velocity in downstream reaches. Much restoration
in the United Kingdom has been concerned with ‘Alleviation
of Low Flow’ (ALF) schemes, which were driven
by the concern of the environmental impacts of overabstraction.
The advantages and disadvantages of methodologies
developed to monitor invertebrates in low fl ow
systems are outlined below.
Scott Wilson Kirkpatrick (SWK) Methodology
The fi rst national methodology for assessing low fl ow
conditions caused by abstraction was not written until
1992 (NRA R&D Note 45, 1992 – a procedural manual
by the consultants Scott Wilson Kirkpatrick). This set out
a point-scoring prioritising procedure based on the assessment
of four indicators: Hydrological; Ecological; Landscape
and Amenity; and Public Perception.
For each indicator, individual assessments of a variety
of parameters were conducted. The scores for each indicator
were combined, with separate weightings. The ecological
indicator parameters and their individual weightings

158 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
were: invertebrate community parameter (0.4), fi shery
parameter (0.2), fi sh stocks parameter (0.3), plant parameter
(0.1) and an optional conservation parameter (0.3).
The major disadvantage was that the methodology was
based solely upon Average Scores Per Taxon (ASPT)
scores, a measure of the balance between pollutionsensitive
and pollution-tolerant invertebrates. This provides
what is essentially an index of organic pollution
without any accommodation of the overall faunal richness
of samples, the fl ow velocity associations of species
present, or their preference for either perennial or ephemeral
fl ow. The subjective manner of deducing severity
index scores, based on targets that were produced on the
basis of adjustments to a starting fi gure suited to sites
considered to offer highest potential for ASPT, was also
of major concern, since it effectively skewed most lowland
rivers towards low impact scores, despite the known severity
of impacts within chalk catchments at the time.
Instream Flow Incremental Methodology
(IFIM) and PHABSIM
Instream Flow Incremental Methodology (IFIM) and
PHABSIM have also been used to establish fl ow objectives
from direct measurement (or modelling) of resident
biotic communities and their specifi c habitat requirements,
although these models have a tendency to be inaccurate
and inadequate, and at best require careful calibration to
prevent distortion of results.
The Surface Water Abstraction Licensing Policy
(SWALP) Methodology
The SWALP methodology developed for the NRA by the
consultant engineers Sir William Halcrow & Partners Ltd
in 1995 incorporates an environmental weighting (EW)
system, for which scores for sensitivity to abstraction are
assigned to physical character, ecology and fi sheries characteristics
(NRA, 1995). The catchment is fi rst assigned
an EW score and thresholds are produced for a set of
defi ned critical assessment points (APs).
The principle behind the SWALP ecological scoring
system was that the highest scores were allocated to
species perceived to require coarse bed materials and
rapid/fast current velocities, and lower scores were given
to species which are common in still water habitats or are
capable of withstanding dessication. In contrast to the
SWK method, SWALP targeted environmental weighting
towards physical and biotic (ecology and fi sheries) attributes,
excluding use-related amenity and recreational criteria,
which were somewhat subjective. It also represented
the fi rst attempt to bring together the perceived fl ow pref-
erences of individual macrophyte and invertebrate species
in a biotic index concerned solely with fl ow dependency.
However, the environmental weightings produced by the
SWALP methodology are rather crude and offer only a
subjective basis for decision making and a limited capacity
to monitor improvement or determine the relative fl ow
requirements within or between river basins.
The Lotic–Invertebrate Index for Flow Evaluation
Index (LIFE)
Following the invertebrate sensitivities to fl ow suggested
by the SWALP methodology, Extence et al. (1999) developed
LIFE. The method sought to link changes in riverine
benthic macroinvertebrate assemblages to prevailing fl ow
regimes. The index is based upon the ‘primary fl ow associations’
of different macoinvertebrate taxa at either
BMWP1 family or mixed-species level – deduced, for the
most part, from taxonomic keys. Calculation of the LIFE
index score for a sample involves individual fl ow scores
(fs) for each scoring taxon present in a sample obtained
from a matrix. This matrix is based on the infrastructure
of the biotic score system proposed by Chandler (1970)
for assessing water quality (Extence et al., 1999). Increasing
abundance of standing water or drought resistant
species produces a lower fs score, whilst increased numbers
of individuals from running water fl ow groups produces
higher scores.
Whilst changes in LIFE score may faithfully refl ect the
observed shifts in composition at sites undergoing gross
changes in habitat provision, the depletion of fauna caused
by an earlier channel-drying event can produce spurious
LIFE scores. This may occur, for example, by the temporal
exclusion of molluscs normally associated with slack or
slow-fl owing water or by rapid colonisation of blackfl ies
(Simuliidae) and mayfl ies (e.g. Baetis spp) at a re-wetted
site (Figure 8.4). These aspects of recolonisation usually
produce a faunal composition with a higher proportion of
velocity-dependent species and an artifi cially elevated
LIFE score, even though taxon richness and diversity may
remain severely depleted. In the case of the River Misbourne
(Appendix 8.3), LIFE scores were also found to
vary signifi cantly between sites in close proximity (100 m
apart) linked to spatial differences in habitat representa-
1 The BMWP (Biological Monitoring Working Party) scores
refere to a system for assessing water quality based on the macroinvertebrate
assemblage. Each taxa is allocated a value from 1
to 10 depending on its known tolerance to organic pollution. High
scoring taxa have lower pollution tolerance, and are thus an indicator
of good water quality.

Uncertainty Surrounding the Ecological Targets and Response of River and Stream Restoration 159
tion in a relatively unmodifi ed chalk stream. Accordingly,
the LIFE score response largely depended upon the magnitude
of temporal habitat change observed at an individual
site. This highlights the need to select a representative
variety of sites to assess the linear extent of environmental
damage attributable to low fl ow.
The MISINDEX
An alternative fl ow dependency index, MISINDEX, was
developed for use on the River Misbourne (Appendix 8.3)
to accommodate cases where assemblages are transitional
and depleted by previous drying. The MISINDEX works
in a similar manner to the LIFE index, with the added
advantage that it featured extended taxonomic coverage of
aquatic species and the inclusion of water dependent (but
not fully aquatic) species of Coleoptera. As with the LIFE
index, the MISINDEX used fi ve fl ow-dependency groups
and used abundance data for taxa in a similar manner to
Extence et al. and Chandler’s matrices, but introduced a
category of water-related or marsh species that were not
necessarily aquatic but depend upon wet places within a
river corridor, lake basin or marsh. Moreover, rather than
attempting to allocate individual species to a primary fl ow
type association that was evidently subjective, the intention
was to defi ne them by the limitations of their tolerance
to fl ow cessation and to tell it like it is if a taxon was of
low or unknown fi delity.
Neither the LIFE or MISINDEX indices explicitly take
into account the dispersal ability of species, or their ability
to recolonise fl ow derogated sites after fl ow restoration, an
important factor that requires further investigation and
could improve prediction of the likely time scale of
impacts resulting from fl ow derogation or periodic channel
drying.
Detrended Correspondence Analysis (DECORANA)
As a multivariate statistical technique DECANORA can
be used to quantify similarities between samples or sites
either spatially or temporally. From such an analysis, it
may be possible to produce meaningful fl ow objectives
determined by observed empirical evidence of the links
between faunal composition or population sizes of individual
species and river fl ows or other recorded environmental
characteristics linked to hydrological variables in
a particular river reach. For this, longer term time-series
of information are required than usually exist at present,
as experience suggests that there is considerable fl ux in
the composition of assemblages, the abundance of constituent
species and instream habitat characteristics linked
to the fl ow history of sites and many other factors.
APPENDIX 8.3 MONITORING THE
ECOLOGICAL RESPONSE OF THE RECOVERY
ENHANCEMENT OF THE RIVER MISBOURNE
Introduction
The River Misbourne, a 28 km long, ecologically valuable
chalk stream situated in the River Thames catchment in
Berkshire, was believed to be amongst the rivers most
affected by abstraction in the United Kingdom. Compounded
by the effects of drought by 1996/1997, migration
of the perennial head saw the river dry from its source
to the middle reaches. A scheme to restore fl ow was implemented
by the Environment Agency and Three Valleys
Water, which comprised abstraction reduction and relocation
of abstraction points. A series of boreholes commissioned
by the Environment Agency (EA) monitored the
hydrological response of these measures, with water
quality routinely monitored at four sites.
The ecological response to the attempted fl ow recovery
was monitored at six sites along the course of the river,
including sites that had dried down and those that had
retained fl ow. Sites with differing characteristics were
chosen to assess response more thoroughly, although the
variation between sites meant that differences between
sites needed to be compared. For example, Sites 1 and 2
were dry in 1998; Site 3 remained dry until 2001, retained
fl ow until spring 2003, but had dried down again by
autumn 2003; Site 4 was dry only in autumn 1997; and
Sites 5 and 6 retained fl ow throughout.
An attempt was made to monitor ecological response
using a series of standard survey methodologies typically
applied twice annually (often spring and autumn) over a
three-year period (1996 to 1999). Changes in vegetation
within the corridor was monitored with Phase 1 and 2
Habitat Survey coupled with River Corridor Survey (RCS)
and in the channel by Macrophyte Survey (MS) according
to changes in habitat conditions shown by River Habitat
Survey (RHS). Birds were monitored using Common Bird
Census (CBC) and Winter Atlas (WA) surveys. For
mammals, bat transect surveys were conducted and the
presence of otter Lutra lutra and Water voles Arvicola
terrestris was to be recorded during RCS.
Macroinvertebrates and fi sh were monitored with more
specifi cally designed sampling regimes. In the case of fi sh
this provided an opportunity to compare two monitoring
techniques: standard depletion electric fi shing within stopnets,
and point-abundance sampling by electric fi shing
(PASE) (Copp and Peñáz, 1988; Perrow et al., 1996). Following
this, the longer term response over a further fi ve
years was undertaken using PASE.
The project had a general aim to restore a native fi sh
community of Brown trout Salmo trutta and bullhead

160 River Restoration: Managing the Uncertainty in Restoring Physical Habitat
Cottus gobio, but no specifi c targets were set; not only
was information on the fi sh population scarce, there was
also no move to gather more. However, the connection of
the Misbourne to the River Colne was thought likely to
provide a source of colonists at least to the lower reaches.
Shardloes Lake, with a direct connection to the river, was
also thought likely to operate as a refuge for fi sh in the
upper reaches, although this was also thought to have dried
out in 1997/98.
Comparison Between Fish Survey Methods
Each site was sampled twice annually (spring and autumn)
between July 1996 and October 1999 using both depletion
fi shing and PASE. The same electrofi shing equipment was
used for both methods. Depletion electric fi shing was
undertaken within a 100 m section created by the use of
stop-nets at either end to prevent the escape of fi sh. Two
operators, each with a single anode attached to a separate
100 m cable and a hand net fi shed the section, exploring
the entire area with sweeping movements. Two runs were
conducted at all sites, apart from at Site 6 in spring and
autumn in 1997, where emergent vegetation seriously
hampered fi shing and only one run was attempted. All fi sh
captured were retained in one or two bins carried by additional
personnel and processed (identifi ed and measured,
with some fi sh also being weighed to calculate length–
weight regressions) at the end of each run, before release
downstream of the stop-netted area. Using the numbers of
each species captured in each run, an estimate of the total
population size of each species was generated using the
maximum weighted likelihood model of Carle and Strub
ind. m -2
3
2.5
2
1.5
1
0.5
0
(1978). Using the area of the section sampled (100 m ×
width of the channel), quantitative estimates of numerical
(ind. m −2 ) and biomass (g m −2 ) density were calculated.
Prior to PASE, a volt meter was used to calibrate the
gear to provide a voltage gradient of 0.12 V at 45 cm away
from the 40 cm anode. Such a voltage corresponds to the
minimum effective voltage at which inhibited swimming
occurs (Copp and Peñáz, 1988). The effective sampling
area was thus calculated to be 1.3 m 2 from which quantitative
estimates of abundance and biomass were calculated.
A total of 50 points were sampled within a 200 m section
of river directly upstream of the stop-netted section at each
site. As in the depletion fi shing, the anode was attached to
a 100 m extension cable from the control box, which was
sited mid-way along the section. Points were taken in a
stratifi ed random manner at 4 m intervals, with the electric
fi shing operator moving diagonally upstream from bank
to bank, and sampling a pre-determined number of points
in the littoral margin, which was determined from the relative
width of the margin to that of the channel.
Overall, both techniques tended to sample a similar
number of species (Wilcoxon signed ranks test, n = 17,
Z = −2.84, p = ns) and produce similar estimates of total
biomass (n = 17, Z = −1.87, p = ns), but signifi cantly different
estimates of total abundance (n = 17, Z = −3.10,
p < 0.01) (Figure 8.5). PASE produced the higher estimates
of total abundance on account of its tendency to
produce signifi cantly higher estimates for bullhead (n =
13, Z = −2.69, p < 0.01) and Ten-spined stickleback Pungitius
pungitius (n = 9, Z = −2.67, p < 0.01). There was
little consequence of the production of signifi cantly higher
biomass estimates for both these species (p = 0.01 and
gm -2
Spring Autumn Spring Autumn Spring Autumn
Spring Autumn Spring Autumn Spring Autumn
1997 1998 1999
1997 1998 1999
PASE
8
6
4
2
0
DEPLETION
Figure 8.5 Comparison of numerical (ind. m −2 ) and biomass (g m −2 ) density estimates obtained by PASE (mean ± 1SE shown) and
depletion fi shing (density derived from total population estimate shown) over time at Site 5, dominated by small species such as
sticklebacks, stone loach and bullheads