Restoring biodiversity and ecosystem function: will an ... - LERF

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177CHAPTER 12Restoring biodiversity and ecosystemfunction: will an integrated approachimprove results?Justin Wright, Amy Symstad, James M. Bullock, KatharinaEngelhardt, Louise Jackson, and Emily Bernhardt12.1 IntroductionTwenty years ago, Bradshaw (1987) stated thatecological restoration should be an acid test ofecological understanding. In fact, the practice ofrestoration has developed more through trial anderror than by the application of any scientificframework. Since Bradshaw’s statement, restorationecology has undergone a rapid increase inconceptual development and basic research, asindexed by the rising number of peer-reviewedpublications (Young et al. 2005), the creation of thejournal Restoration Ecology in 1993, and the recentpublication of a number of edited volumes dedicatedto exploring the conceptual underpinning ofrestoration ecology (Falk et al. 2006, Van Andel andAronson 2006). In addition, meta-analyses of restorationstudies (e.g. Pywell et al. 2003) are beginningto draw out some general patterns and relatethem to broader ecological theory. In this chapterwe contribute to this development by exploring theapplicability of the biodiversity-ecosystem functioning(BEF) framework to restoration science.Restoration ecology is the subdiscipline of ecologythat informs the ‘intentional activity that initiatesor accelerates the recovery of an ecosystemwith respect to its health, integrity and sustainability’(Ser2004). Like the broader field of ecology,restoration ecology is an integrative discipline,having drawn important influences from fields asdiverse as applied sciences such as agronomy andengineering (Mitsch 1993); social sciences such associology (Geist and Galatowitsch 1999) and landscapearchitecture (Fabos 2004); and earth sciencessuch as soil science (Bradshaw 1997) and hydrology(Morris 1995); as well as the subfields of population(Rosenzweig 1987) and landscape ecology (VanDiggelen 2006). This diversity of influences has ledto many different approaches and goals for restorationprojects. However, given the primary focuson restoring the structure and function of ecosystems,the strongest conceptual basis for most ofrestoration ecology stems from community andecosystem ecology (Ehrenfeld and Toth 1997, Palmeret al. 1997, Young 2000, Falk et al. 2006). Simplyput, restoration ecology is typically interested inrestoring either biodiversity, ecosystem functioning,or both.Despite these overlapping areas of interest, littlecrossover is evident between restoration ecology and‘classical’ biodiversity-ecosystem function research(Naeem 2006a). With few exceptions (Bullock et al.2001, Callaway et al. 2003, Bullock et al. 2007), BEFexperiments have not taken place in restoration settings.Although BEF research might inform restoration(Aronson and Van Andel 2005, Young et al. 2005,Naeem 2006a), thus far few concrete suggestions havebeen offered on how restoration ecology might benefitfrom a consideration of BEF research and theoryor vice versa. In this chapter, we start by comparingcommunity and ecosystem approaches to restorationand suggest how a BEF approach might differ fromthese. We then draw on BEF theory and empiricalresearch to suggest three broad areas whereunderstanding the links between biodiversity and167

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177168 BIODIVERSITY, ECOSYSTEM FUNCTIONING, AND HUMAN WELLBEINGecosystem functioning could have significantimpacts on the success of restoration: ‘classical’ BEFimpacts (i.e. higher diversity leads to improvedfunctioning); the effects of biodiversity on thestability of ecosystem functioning; and the effectsof biodiversity on the provisioning of multipleecosystem services.For this chapter, we constrain the scope of whatwe consider to be restoration to management activitieswhose primary goal is to improve ecosystemservices other than provisioning services, althoughimproved provisioning services may result from theactivities. Management activities whose primarygoal is to improve provisioning services are consideredin Chapter 13, which formally examines the roleof BEF research in managed ecosystems.12.2 Community, ecosystem, and BEFapproaches to restorationBEF research combines elements from community andecosystem ecology. Consequently, Naeem (2006a)contrasted community and ecosystem approaches torestoration with an approach based on the BEF perspective.A community approach largely focuses onthe restoration of the biotic components of an ecosystem– which species are present, their relative abundance,and their interactions (trophic, competitive,facilitative). It is often used when starting fromessentially bare ground, as in tallgrass prairie or haymeadow plantings in former agricultural fields, andwhen the goal of restoration is to enhance the conservationvalue of a protected landscape – restorationfor biodiversity’s sake. This approach has evolvedover time from assuming that restoration is essentiallythe speeding up of a linear approach to a specific,predictable, equilibrium state, to accepting thedynamic nature of communities, possible alternativestable states (Bullock et al. 2002, Hobbs 2006), and theinfluences of disturbance and dispersal limitation ondiversity (Pywell et al. 2003, Walker et al. 2004).This evolution may help overcome some of theproblems encountered when using the communityapproach. For example, restoring the target communityis often more difficult than expected, even insystems where community restoration has beenpractised for decades (Kindscher and Tieszen 1998).Furthermore, restoration of one part of a communityhas not always yielded restoration of the wholecommunity. In both California grasslands andEnglish meadows, plant composition and relativeabundance of species have been restored to resemblethe reference condition, but the soil microbial communityhas remained significantly different, implyingthat functions such as decomposition and nutrientturnover may not have been restored (Smith et al.2003, Potthoff et al. 2005, Steenwerth et al. 2006, Pywellet al. 2007). Although greater understanding of factorsinfluencing community assembly may overcomesome of these problems, it may not overcome thecommon assumption of the community approach thatfunctioning will be restored if the community isrestored. This was explicitly studied in two constructedSpartina marshes. In both, vegetation composition,cover, and biomass of planted sites weresimilar to those of natural marshes within 18 monthsof planting, but in one case, the height structure failedto meet the needs of the endangered bird for whichthe marshes were constructed (Zedler 1993), and inanother, soil organic matter and nutrient accumulation,denitrification rates, and tidal export of nutrientsrequired much longer to resemble reference levels(Craft et al. 1999).The ecosystem approach makes the oppositeassumption, in that a habitat template that restoresecosystem processes (e.g. hydrology of a wetland)is created, but most or all species are left to colonizeon their own in a process of self-assembly (e.g.Bradshaw 2000). The approach, which is characteristicof many early restoration efforts (Bradshawand Chadwick 1980), may more accurately bedescribed as ‘rehabilitation’ rather than ‘restoration’(Ser 2004). Some examples of this approach includeconstructing mechanical barriers in eroded gulliesin an overgrazed rangeland in order to slow surfacewater flow and enhance water percolation into thesoil (King and Hobbs 2006) and increasingmeandering of a stream in order to reduce flashflooding and increase habitat complexity and sedimentand nutrient retention (Rosgen 1994). Otheractivities using the ecosystem approach includeecological engineering and reclamation. Ecologicalengineering strives to achieve and maintain a specificecosystem service within a relatively strictrange, such as reducing nitrogen in wastewater toregulator-accepted standards via a constructed

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177RESTORING BIODIVERSITY AND ECOSYSTEM FUNCTION 169wetland. In contrast, reclamation strives to make ahighly altered system serve some useful purpose byachieving an acceptable level of safety and aesthetics,such as stabilizing mine tailings withvegetation that can tolerate the harsh conditions butare not necessarily native to the site (Ser 2004). Bothuse ecological processes and biotic components toachieve their goals, but these goals do not includecreating a system whose functioning or compositionresemble a reference ecosystem.Although economically attractive, the ‘Field ofDreams’ approach of solely manipulating the physicalenvironment (‘If you build it, they will come’;Hilderbrand et al. 2005) is risky because dispersalbarriers limit the colonization of desired bioticcomponents and, even if these can be overcome,interactions with species not yet present (e.g. pollinatorsor mycorrhizae) may be necessary for establishmentor reproduction of important species.Furthermore, initial seeding with native or exoticspecies that may grow quickly and provide goodcover can prevent establishment of desired latersuccessional species. For example, in old field sites inEngland, sowing of a few native grasses limitedsuccessful colonization by desired species (Pywellet al. 2002). In addition, projects using the ecosystemapproach with no attention to the composition of thebiotic component may achieve their physio-chemicalgoals, but the narrow focus of these goals may leadto long-term problems. For example, some of today’smost troublesome invasive species were introducedas a rehabilitation measure in over-grazed, droughtstressedrangelands (Christian and Wilson 1999,Clarke et al. 2005, Schussman et al. 2006, Williamsand Crone 2006). Finally, the approach does not takeadvantage of the multiple functionality and resiliencepotentially provided by systems with anactively restored biotic community.In contrast, a restoration approach based on BEFtheory and empirical results stresses the relationshipbetween the biotic community and ecosystemfunctioning. Although the BEF perspective does notencompass all of the topics relevant to ecologicalrestoration, it does cover the majority of a moregeneral framework recently proposed for restorationecology (King and Hobbs 2006). A BEFapproach to restoration is based on the asymptoticrelationship between biodiversity and ecosystemfunctioning. This relationship is what sets the BEFapproach to restoration apart from the other twoapproaches. Restoration strives to restore an ecosystemto that relationship by removing anthropogenicinputs that maintain high functioning but lowdiversity in managed systems (e.g. fertilizer or pesticides)or by enhancing functioning in degradedsystems by adding key sets of species (Naeem2006a). Debate about the applicability of BEF theoryoutside of experimental settings (Huston 1997,(a)Ecosystem functionManagedDegraded(b)Ecosystem functionBiodiversityBiodiversityFigure 12.1 The BEF perspective for ecological restoration. (a) The fundamental assumption of the BEF perspective is that biodiversity affects ecosystemfunction, often in the manner depicted here, where adding diversity when diversity is low has a stronger effect on ecosystem functioning than adding diversitywhen diversity is high, until the point at which increasing diversity has no effect of functioning. Restoration (vertical arrows), in this perspective, is therestoration of this relationship in degraded systems, which have lower functioning than expected, or in managed systems, in which anthropogenic inputsproduce functioning at a higher level than expected given the system’s diversity. (b) The theoretical realm of possible ecosystem functioning in relation tobiodiversity. The hatched area incorporates the assumption, supported by empirical evidence, that it is possible to have greater functioning with a low numberof species than with the most diverse system, as well as the assumption that the range of variability in functioning decreases as biodiversity increases. Thelatter assumption has less support from empirical evidence (see text). Figure is adapted from Naeem (2006).

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177170 BIODIVERSITY, ECOSYSTEM FUNCTIONING, AND HUMAN WELLBEINGBox 12.1 Should the BEF restoration approach apply to all ecosystems?Ecological restoration, like most things ecological, isinherently and frustratingly context-dependent. Restorationis being carried out in habitats as diverse as streams,rivers, forests, wetlands, grasslands, estuaries, deserts,and alpine habitats to achieve an equally diverse suite ofgoals. Given the diversity of contexts in which restorationoccurs, it is appropriate to ask whether the BEF approachis relevant to all forms of restoration. We see greatpotential for the BEF perspective to improve the practiceof ecological restoration in many cases. However, it isimportant to recognize that in systems where abioticstructure serves as the dominant control of ecosystemproperties, the BEF approach may not be as relevant tosuccessful restoration.Streams and rivers are an instructive case study ofrestoration in ecosystems that are strongly controlled byabiotic forces. River restoration is one of the most extensiveforms of restoration in the USA, with at least $1 billiondollars invested annually (Bernhardt et al. 2005). A growingbody of scientific literature has documented biodiversityeffects on ecosystem function in stream ecosystems (e.g.Jonsson and Malmqvist 2000, Cardinale et al. 2002). Yetbecause the dominant taxa in river ecosystems are easilydispersed, short-lived, and small it is difficult to apply theseBEF perspectives to river restoration projects. Seedingstreams with the appropriate algae, macroinvertebrates,and fish is no guarantee that those organisms will establish.In terrestrial ecosystems, where vegetation itself providesmuch of the physical structure of the environment on whichother organisms depend, it is easy to see how plantingdiverse native species assemblages can effectively ‘restore’not only an ecosystem function (¼ productivity) but also keycomponents of ecosystem structure (e.g. canopyarchitecture). This is not the case in rivers, where thephysical template is primarily controlled by hydrology andgeomorphology.However, we would argue that a BEF perspective can andshould inform river restoration in two important respects. First,terrestrial vegetation can play very important roles in riverecosystems, and thus BEF approaches can directly informriparian management aspects of river restoration. The majorityof small stream ecosystems are primarily fueled by leaf litterinputs (Fisher and Likens 1973), and higher diversity litterinputs can lead to greater secondary production (Swan andPalmer 2006). Riparian trees themselves contribute largewood to stream channels which can play important roles asboth habitat and biogeochemical hotspots in rivers (Wallaceet al. 1995, Valett et al. 2002, Wright and Flecker 2004,Warren and Kraft 2006).Second, a BEF perspective should inform the evaluationof river restoration projects. Even when aquatic organismscannot be directly introduced, monitoring changes incommunity composition following restoration activities canprovide important insights into what is and what is notworking. For example, the absence of shredding functionalfeeding groups from restored streams relative to referenceconditions may indicate that organic matter dynamics havenot been effectively restored, the presence of nitrogen-fixingblue–green algae may indicate that phosphorus loads areexcessively high, or the absence of a diverse hyporheicmeiofauna may suggest that groundwater and surfacewaters have not been effectively reconnected. Whilerestoration practitioners may not be able to actively managethe diversity of all ecosystems to affect services, recognizingthe links between biodiversity and ecosystem functioningthat exist even in ecosystems strongly controlled by abioticforces can lead to improved assessment of the success orfailure of restoration projects.Wardle 1999, Naeem 2000, Grace et al. 2007) hashighlighted that restoring a highly diverse communitydoes not necessarily guarantee a high level offunctioning – environmental factors such as soilfertility and climate are also crucial determinants ofecosystem functioning (Huston and Mcbride 2002,Naeem 2002b). However, when environmental factorsare held constant, the BEF approach suggeststhat greater biodiversity provides a high level ofecosystem functioning, although achieving the highestlevel of functioning does not require the restorationof the entire community (Lehman and Tilman2000, Naeem 2006a). The rest of this chapter exploresthese implications for restoration in greater detail.12.3 ‘Classical’ BEF implicationsfor restorationSince its earliest inception BEF research has largelybeen focused on testing the hypothesis that the lossof diversity of species (or functional groups) leads tochanges in ecosystem functions such as productivity

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177RESTORING BIODIVERSITY AND ECOSYSTEM FUNCTION 171Box 12.2 Diversity of grassland plantingsGrassland restoration in the central portion of theUSA illustrates the relatively low level of diversity used insome restorations compared to their reference systemsand the implications. Native northern mixed-grass prairie in westernNebraska and South Dakota has approximately 37–80native plant species per 0.1 ha (Symstad et al. 2006),whereas the recommended seed mixtures for nativerangeland plantings (> 0.1 ha) in this region have amaximum of 13–25 species and a minimum of fourspecies (http://www.nrcs.usda.gov/technical/efotg/). The federal Conservation Reserve Program (CRP) paysfarmers to plant or maintain perennial vegetative cover,grassland being one allowable type, in areas that wouldotherwise be used for agricultural production. The program,which affects more than 10 million ha of grasslandnationwide, rewards plantings that provide ecosystemservices such as reduced soil erosion, increased wildlifehabitat, water quality protection, soil salinity reduction, andcarbon sequestration (Barbarika 2005). The diversity ofthese plantings is difficult to track, but typical values are4–10 species. The species planted in restorations like these are oftendominants (e.g. warm-season bunchgrasses) that drivemajor aspects of ecosystem functioning (Camill et al. 2004),but low functional diversity, particularly the lack of nitrogenfixinglegumes in some plantings, may limit a restoration’sfunctioning potential (Kindscher and Tieszen 1998).or nutrient cycling (Naeem et al. 1994, Hooper andVitousek 1997, Tilman et al. 1997b). Since these earlystudies, the field of BEF research has flourished, withover 100 published experiments testing this generalhypothesis (Cardinale et al. 2006a). Contentiousdebates have ensued about the proper experimentaldesign or how best to interpret the results of theseexperiments (Garnier et al. 1997, Huston 1997,Wardle et al. 1997a, Thompson et al. 2005, Wrightet al. 2006), and considerable work still needs to bedone to determine the mechanisms that might linkdiversity to ecosystem functioning. Despite theseuncertainties, several recent meta-analyses havedemonstrated that, on average, ecosystem functioningdoes increase with increasing numbers of speciesin BEF experiments (Balvanera et al. 2006, Cardinaleet al. 2006a). For example, Cardinale and colleagues(2006a) found that diversity enhanced both plantproductivity and nutrient uptake. These biodiversityeffects can be tied to two ecosystem services that areoften the focus of restoration efforts. It should benoted that while the most represented ecosystem inthis meta-analysis were grasslands, these comprisedonly 34% of the studies, with the rest coming from abroad diversity of other terrestrial and aquatic ecosystems.Thus there is a growing body of workaddressing BEF questions in other ecosystems suchas streams (Jonsson and Malmqvist 2000, Cardinaleet al. 2002), wetlands (Engelhardt and Ritchie 2001,Callaway et al. 2003, Sutton-Grier et al. In Review),forests (Bunker et al. 2005, Scherer-Lorenzen et al.2005a), and marine systems (Duffy et al. 2001, Solanet al. 2004, France and Duffy 2006b, Worm et al.2006). For most systems in which restoration is beingconducted, potentially relevant BEF experimentshave been conducted. Applying the findings fromBEF research to restoration practices should be anatural extension of existing research. Indeed, astudy that examined the consequences of restoringplant communities on ecosystem functioning in aCalifornia estuarine salt marsh demonstrated thatincreasing plant richness led to higher rates ofnitrogen uptake and greater above- and belowgroundbiomass (Callaway et al. 2003). Grasslandand wetland restorations typically start from bareground and try to recreate natural communitiesthrough the addition of seed. The seed mixes usedare not particularly diverse relative to the naturalecosystems that serve as restoration targets (Box12.2). Thus, many terrestrial restoration activities areoperating in the region of diversity where varyingspecies richness is most likely to have a significanteffect on ecosystem functioning, since most ecosystemfunctions saturate at relatively low levels ofspecies richness (Cardinale et al. 2006a).However, the application of classical BEFresearch to restoration activities may still be limitedgiven our current knowledge. First, while averageecosystem function has been shown to increasewith increasing diversity (Balvanera et al. 2006,Cardinale et al. 2006a), meta-analysis of BEFexperiments has also demonstrated that the highestperforming species in monoculture produces a levelof ecosystem function that cannot be distinguished

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177172 BIODIVERSITY, ECOSYSTEM FUNCTIONING, AND HUMAN WELLBEINGfrom the level observed in the highest diversitytreatment (Cardinale et al. 2006a). Second, due tothe constraints of experimental design, BEFexperiments do not always include low-diversitypolycultures capable of outperforming the highestdiversity treatment (Chapter 2). Thus, if the goal ofrestoration is to provide a maximum level of ecosystemservices, arguments could be made forestablishing a high-performing species to rapidlyachieve a high level of functioning early in therestoration process. However, because these highperformersare usually dominant species, subordinatespecies may be more difficult to establish afterthe fact, as observed in tallgrass prairie plantings(Weber 1999), hay meadows (Pywell et al. 2002) andcoastal marshes (Keer and Zedler 2002). This isparticularly important because there is some evidencethat the effects of greater diversity on functioningtake some time to develop (Tilman et al.2001, Cardinale et al. 2007), and because thesesubordinate species may contribute to stability offunctioning.12.4 BEF and stability of servicesin restorationThe question of whether biodiversity contributes tostability (e.g. resistance to disturbance, resilienceafter disturbance, and moderate range of variabilitythrough time) has been a topic in ecology for morethan half a century (e.g. Macarthur 1955, May1972). The recent development of the BEF perspectivehas provided resolution to some aspects ofthis question by suggesting that species that sharefunctional effects traits (characteristics that affectecosystem functioning in a specific way) often differin their functional response traits (characteristicsthat determine how they respond to a specificperturbation). As a result, a relatively low numberof species may provide a certain level of an ecosystemfunction in a constant environment, but ifthese species are adversely affected by a perturbation(e.g. drought, flood, fire, or herbivory), thatlevel of functioning will only be maintained ifspecies with a similar effect on functioning respondpositively to this perturbation. For example, Evineret al. (2006) identified a strong seasonality in theeffects of individual species on N and P cyclingin northern California grasslands. Their resultssuggest that when in mixture, the species’ contributionsto ecosystem functions would varythroughout the season, providing a mechanism formaintaining these nutrient cycles throughout theseason. In addition, they investigated the relationshipbetween a variety of plant traits (live tissueand litter chemistry and biomass, modification ofbioavailable C and soil microclimate) and the ecosystemfunctions. The influence of individual traitson N mineralization and nitrification also variedthroughout the growing season, illustrating howresponse traits (to seasonal changes in moisture andtemperature) are not necessarily correlated withfunctional effect traits (Landsberg 1999, Lavorel andGarnier 2001, Hooper et al. 2002, Naeem andWright 2003).Several reviews describe this and other mechanismsin greater detail and show the strong theoreticalsupport for the diversity–stability hypothesis(McCann 2000, Cottingham et al. 2001, Loreau et al.2002, Hooper et al. 2005, Chapter 6), and empiricalsupport for the hypothesis exists from experimentsin systems relevant to ecological restoration. Inplots in which plant species richness varied becauseof nitrogen manipulations, aboveground biomasswas more resistant to, and recovered more fullyfrom, a major drought in more diverse grasslandplots in central Minnesota, USA (Tilman andDowning 1994). In the same system, but in plots inwhich plant species richness was directly manipulated,temporal stability (measured as temporalmean/standard deviation) of aboveground plantbiomass over ten years increased linearly, fromapproximately 3.5 to 5.8, as planted species richnessincreased from one to 16 species, with the diverseplots having lower temporal standard deviationsfor a given mean biomass than the monocultures(Tilman et al. 2006b). Despite these encouragingexamples, there are also counter-examples. Morediverse plots had lower resistance of primary productivityto drought in a Swiss grassland experiment(Pfisterer and Schmid 2002); individualspecies, rather than species richness, affected biomassstability in constructed or natural wetlands(Rejmankova et al. 1999, Engelhardt and Kadlec2001); and rocky shore intertidal communities withthe greatest diversity were most severely affected

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177RESTORING BIODIVERSITY AND ECOSYSTEM FUNCTION 173by heat stress (i.e. had low resistance) but thequickest to recover from the stress (i.e. high resilience)(Allison 2004). A recent meta-analysis of alarge number of studies confirmed this inconsistency– averaged across experiments, more diversesystems were more resistant to nutrient perturbationsor invasions, but diversity had either a neutralor negative effect on resistance to drought, responseto warming, and variation in response to long-termenvironmental variability (Balvanera et al. 2006).Maintaining ecosystem services within a reasonablerange of variability is an important componentof restoration projects focused on restoring functioning.The balance of evidence is currently notvery strong that biodiversity plays a large role indetermining this variability. However, relativelyfew field-scale studies have investigated this question,and there is no evidence that diversitystrongly negatively affects stability, particularlyover the long term, so the cost of increasing thediversity of a restoration is likely only the directcost associated with adding that diversity.12.5 Biodiversity and the restorationof multiple ecosystem functionsRestoration activities are not typically conductedwith the goal of restoring a single ecosystem service.Rather, there is an implicit understanding that‘healthy’ ecosystems provide a large number ofservices (Duraiappah and Naeem 2005), and thatrestoration can serve to increase multiple ecosystemservices (Nrc 2001, Bernhardt et al. 2005). Forexample, the restoration of Iraq’s Mesopotamianmarshes has been assessed based on five separateecosystem functions: productivity of the dominantplant Phragmites australis, redox status, hydrologicfunction, salinity, and bird diversity (Richardsonand Hussain 2006). Similarly, the US NationalResource Council suggested that five major functionsof wetlands need to be considered in theconstruction or restoration of wetlands: hydrologicfunctions, water quality functions, support ofvegetation, support of fauna, and soil functions(Nrc 2001).This demand for multiple ecosystem servicesfrom a single restoration may be the strongestargument for incorporating greater biodiversityinto ecological restorations. As was discussedabove, high diversity plots in BEF experiments donot, on average, significantly outperform the highestperforming monoculture (Cardinale et al. 2006a).However, what is not clear from this analysis iswhether a single species can maximize the provisioningof multiple ecosystem functions. Evidencefrom a biodiversity, carbon dioxide, and nitrogenmanipulation experiment in a Minnesota grasslandcontext (Reich et al. 2001, Reich et al. 2004) suggeststhat this is not the case. In this experiment, the totalabove- and belowground biomass produced by aspecies and the amount of inorganic nitrate left inthe soils after plant uptake were strongly correlatedin the first year after planting (Fig. 12.2(a)). Such aresult would suggest that a single species might becapable of both fixing high amounts of carbon andimproving water quality by reducing nitrogenexport. However, in subsequent years of theexperiment, this correlation broke down (Fig. 12.2(b,c)), making it more difficult to suggest a singlespecies maximizes both ecosystem services. In fact,Hector and Bagchi (2007) found that maximizingseven ecosystem functions required between 8 and16 species at eight grassland sites across Europe. Ina recent review, Gamfeldt et al. (2008) found thatmultifunctional redundancy (i.e. the degree towhich multiple species could sustain multiplefunctions) was generally lower than single-functionredundancy (i.e. the degree to which multiple speciescould sustain a single function).Species effects on ecosystem functions are functionsof key morphological and ecophysiologicaltraits (Engelhardt and Kadlec 2001, Eviner andChapin 2003, Diaz et al. 2004), and BEF research isincreasingly focused on how particular traits interactto determine the effects of diversity on ecosystemfunctioning (Eviner and Chapin 2003, Naeemand Wright 2003, Solan et al. 2004, Bunker et al.2005). While attempts to determine which traits aremost important in regulating particular ecosystemfunctions are still in the early stages, it seems reasonableto assume that different processes might beaffected by different combinations of traits. Giventhis assumption, the extent to which single speciescan maximize multiple functions will depend onthe extent to which the traits responsible for regulatingthe ecosystem functions of interest are

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177174 BIODIVERSITY, ECOSYSTEM FUNCTIONING, AND HUMAN WELLBEINGTotal inorganic N (0–60 cm)(a) (b) (c)59543 ASTU2100r = 0.78 r = 0.18 LUPEr = 0.158LUPE47AMCALECAPEVI653ANCYLECAPOPRASTUPEVIANCY42AMCAANGEKOCRBOGRPEVILECAAMCABRINSONUSCSC POPRKOCR SORH SOREADRE ANOE POPR3ANCY KOCR BOIP AGREASTUACMIAGRE1ACMIBOGRSORISCSO2ACMIPOGRANOESONUSONUSONUSORI0200 400 600 800 1000 1200 1400 0 200 400 600 800 1000 1200 1400 1600 0 200 400 600 800 1000 12001Total biomass (g/m 2 )LUPEAQ: Please notethat the mergedlabels are notclear. Kindlyadvise.Figure 12.2 Relationship between two ecosystem functions, total above- and belowground biomass (a surrogate for carbon storage) and total extractablesoil inorganic nitrogen (a surrogate for nitrogen removal from groundwater) from monocultures of different species in control plots (ambient levels ofatmospheric CO 2 and nitrogen) of the BioCON BEF experiment from (a) 1999, (b) 2000, and (c) 2001. Species codes of the monocultures are: Achilleamillefolium ACMI; Agropyron repens AGRE; Amorpha canescens AMCA; Andropogon gerardi ANGE; Anemone cylindrical ANCY; Asclepias tuberose ASTU;Bouteloua gracilis BOGR; Bromus inermis BRIN; Koeleria cristata KOCR; Lespedeza capitata LECA; Lupinus perennis LUPE; Petalostemum villosum PEVI; Poapratensis POPR; Schizachyrium scoparium SCSC; Solidago rigida SORI; Sorghastrum nutans SONU. Note that the scale changes on the axes between years.correlated. Several recent large-scale analyses havefound significant correlations between severalimportant traits in plants (Diaz et al. 2004, Wrightet al. 2004, Reich et al. 2006) and animals (Brown et al.2004). These correlations yield suites of traits (e.g.those contributing to rapid acquisition of resourcesvs. those that contribute to conservation of resourcesin well-protected tissues in plants) that typicallyoccur together in organisms. The uniformity of thesesuites across broad taxonomic groups and geographicalgradients argue for the existence of fundamentaltradeoffs in organismal development andlife history. Whether or not these tradeoffs hold upwithin local species pools, where selection mightpush for diversification along trait axes (Grime 2006,Ackerly and Cornwell 2007), and for the traitsactually responsible for ecosystem functions ofinterest in restoration (Eviner 2004), remains an openand important question.12.6 The economics of BEF in restorationFor BEF research to be useful for ecological restoration,ecosystem functions must be related to theecosystem services desired as the outcome of restoration.A key issue in performing this translationis that BEF research often does not assign a specificworth to a level of ecosystem functioning otherthan a vague ‘more is better’ for primary productivityor nutrient capture (Srivastava and Vellend2005). In contrast, in restoration, the relative worthof various levels of ecosystem services must beassessed and agreed to by many stakeholders whenany but the simplest restoration project is commenced(Fig. 12.3). Provisioning services, such asforage production, can be easily converted intocurrency value (Bullock et al. 2007). However, whileWorm et al. (2006) showed dramatic increases intourism-related revenue following the closure offisheries in marine protected areas, which theyattribute to increases in diversity, this conversion isnot as easy for other types of ecosystem servicesresulting from ecological restorations. For example,results from BEF research might predict grams ofcarbon fixed per square metre, grams of nitrogenremoved through denitrification, and grams of thegreenhouse gasses N 2 O and methane produced bydifferent combinations of species used in a wetlandrestoration project. Until markets exist that allowtranslation of these ecosystem processes into acommon currency, determining which mixture ofspecies maximizes benefits while minimizing costsis, at best, just a guess. The growth of carbontrading markets (Bonnie et al. 2002) and early

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177RESTORING BIODIVERSITY AND ECOSYSTEM FUNCTION 175Carbon storageTotal benefit ?Monocultureattempts at nitrogen trading schemes such as theEPA’s Water Quality Trading Program (active inareas of 10 states in the USA) suggests that economicvaluations of at least some of the nonprovisioningecosystem services provided by restoredecosystems are being developed. An even more difficulttask is deriving a common currency for servicessuch as protecting higher trophic levels or maintaininghydrologic services, although tools for evaluatingdifferent options in the face of such uncertainties doexist (Lynam et al. 2007).12.7 RecommendationsPolycultureNitrogen removalFigure 12.3 Hypothetical restoration scenario involving two ecosystemservices: carbon storage (circles) and nitrogen removal (squares). Managerscould choose between planting a monoculture of species 1 (filled symbols)that provides high levels of carbon storage but low levels of nitrogenremoval, a monoculture of species 2 (open symbols) that provides lowlevels of carbon storage but high levels of nitrogen removal, or a mixture ofboth species that provides intermediate levels of both carbon storage andnitrogen removal. The total benefit of the ecosystem services provided bythese different management choices (straight lines) is unknown and willdepend both on the shape of the relationship between diversity and thesetwo functions and on the relative economic weight placed on the twoservices.We argue that understanding the relationshipbetween biodiversity and ecosystem functioning isimportant in enhancing restoration success in manyecosystems and that restoration activities can serveas powerful tools for exploring some of the centralBEF questions. However, information exchangebetween the scientific community and restorationpractitioners is too little and too slow for BEFresearch to be relevant in adaptive management ofdegraded ecosystems. Ecologists are recognizingthat performing policy-relevant science involvesmore than simply publishing in high-profile??academic journals (Palmer et al. 2004). This point isreinforced by a recent survey of stream restorationpractitioners that showed that less than 1 per centof over 300 restoration projects had specificallybeen informed by results published in scientificjournals (Bernhardt et al. 2007). So what can bedone to improve the situation?For scientists, we make a few suggestions. First,continue basic BEF research, but keep in mind theinformation needed by restoration practitioners.Deeper understanding of how functional traitsacting alone and in combination affect ecosystemfunctions that are related to ecosystem services isparticularly important. Although general relationshipsbetween diversity and functioning explainwhy the restoration of biodiversity is important forrestoring ecosystem functioning, restoration practitionersultimately need to know which specificspecies combinations to restore and have confidencein the outcome of restoration projects.However, BEF theory, while explanatory, is not yetpredictive (Hooper et al. 2005) and is therefore notyet ready to be embraced by the managementcommunity. Furthermore, experimental perturbationsin BEF experiments, and monitoring ofexperiments over long time periods in which theenvironment fluctuates naturally, will help resolvewhat role biodiversity plays in stabilizing ecosystemservices. However, because species and communitiesrespond to environmental fluctuations inseemingly idiosyncratic ways, investigations thattie these two themes together, by seeking patternsin traits that determine species’ responses to environmentalvariations, will yield the most relevantinformation for restoration.BEF research focusing on microbial diversity isparticularly crucial because many of the mostcritical ecosystem services are underpinned bymicrobial processes (e.g. the nutrient transformationsthat improve water quality). Our understandingof both how plant and animal diversityaffect microbial community structure and howmicrobial diversity directly affects ecosystemfunctioning in real systems is relatively weak(Fierer et al. 2007, Jackson et al. 2007). To date therehas been little focus on the importance of restoringmicrobial communities or how to go about doingso (Hasselwandter 1997). Restoration of the

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177176 BIODIVERSITY, ECOSYSTEM FUNCTIONING, AND HUMAN WELLBEINGmicrobial community in meadows were notachieved by a simple soil microbe addition technique(Pywell et al. 2007), but experiments haveshown that microbial community composition isrelated to both the diversity of the restored plantcommunities and to particular ‘facilitating’ plantspecies (Smith et al. 2003, Bardgett et al. 2006). Adeeper understanding of the controls and consequencesof microbial diversity is an area whereBEF research is well-poised to make importantcontributions to restoration ecology.Another aspect of diversity that may be critical torestoration success, but has been not been as extensivelystudied in BEF research, is the importance ofgenetic diversity. Many ecosystems, including manythat are important targets for restoration, are dominatedby a single species that controls ecosystemfunction, e.g. giant kelp in kelp forests, Ponderosapine in many forests of the Western US, seagrassessuch as Zostera marina in shallow estuarine systems,and Spartina in intertidal zones. A growing body ofwork is showing that genetic diversity within a speciescan be an important regulator of ecosystemfunction (Hughes et al. 2008). Work on Zostera, animportant species in estuarine restoration, showedthat clonal diversity measurably affects ecosystemprocesses or the stability of those processes (Hughesand Stachowicz 2004, Reusch et al. 2005). Williams(2001) demonstrated that reduced genetic variationin Zostera used in restoration efforts resulted indecreased population growth and individual fitness.For both cottonwood (Populus fremontii x augustifolia)(Schweitzer et al. 2005a) and apsen (Populus tremuloides)(Madritchet al. 2006), genotypic richness canaffect decomposition rates and nutrient cycling.Clonal diversity of Solidago altissima were shown toaffect, not only primary productivity, but the diversityof higher trophic levels (Crutsinger et al. 2006).Given these compelling examples, research on theconsequences of genetic diversity of species commonlyused in restoration is likely to yield benefitsboth to basic science and restoration.Third, to make the results of any BEF researchapplicable to practitioners, the connection from biodiversityto function to services needs to be stronger.For example, although the positive effects of plantdiversity on above- and belowground biomass productionin grassland experiments have been vaguelyrelated to forage production, carbon storage, and soilerosion, little attention has been paid to whether thenutrient content of the biomass is sufficient for thepurported foragers (see Bullock et al. 2007 for anexception) and direct measurements of soil movementor long-term C sequestration are rare (thoughmore common in wetland studies). Given the need ofmany restorations to restore multiple services, thisconnection needs to be made simultaneously formultiple functions and their related services.Addressing these issues is already one of the centralthrusts of the next generation of BEF research (Naeemand Wright 2003) and should not require significantchanges to how we proceed with our science beyondthe functions and properties measured in typical BEFexperiments.Fourth, evaluate the relevance of experiments toreal-world conditions. There is some conflictbetween the typical BEF experiment, which iscarefully designed to disentangle the effects ofindividual species, functional traits, and diversityper se on the functions measured, and restorationprojects, which are concerned with achieving adesired result with available materials and limitedfinancial resources. Restoration practitioners maylook at the high density of expensive species (manyforbs, for example) planted in some BEF experimentsand question the applicability of the resultsto their work: does the diversity effect require thisrelatively high input of normally subordinate orrare species? They might also question the relativelycontrolled situations of the field experiments:would the results be the same if vertebrate herbivoreswere not excluded from the plots, or if colonizingspecies were not removed? Finally, apractitioner would never think of restoring a Spartinaalterniflora marsh without Spartina alterniflora,but many BEF studies include treatments analogousto this situation. Of greater interest to a restorationpractitioner would be the question of how strongthe diversity effect is when the only portion ofdiversity manipulated is the subordinate and rarespecies. These subordinate species may be essentialto certain key functions which define the success orfailure of the restoration. For example, grasslandrestoration in the UK to meet national BiodiversityAction Plan targets requires the presence of foodplants of certain declining butterflies and other

From: Biodiversity, ecosystem functioning and human wellbeing. (2009). Oxford University Presseds S. Naeem, D. E. Bunker, A. Hector, M. Loreau & C. Perrings, pp. 167-177RESTORING BIODIVERSITY AND ECOSYSTEM FUNCTION 177insects, but these plants are often uncommon andparticularly difficult to establish (Pywell et al. 2003).Thus, BEF researchers seeking to provide answersfor restoration practitioners and other naturalresource managers need to ensure that they designtheir experiments with these issues in mind.Concurrently, restoration practitioners and ecologistscan potentially contribute significantly tostrengthen BEF research. For them, we stress theutmost importance of monitoring and reportingshort- and long-term effects of various restorationprojects and practices. In the US, the federal governmentpays private land owners millions of dollarseach year to plant and maintain perennial grasslandsthrough the Conservation Reserve Program. A landowner’sproposal is more likely to be funded if s/heplants a higher diversity of species (Usda 2006), butfollow-up on the establishment success and environmentalbenefits of individual plantings is rare, andonly recently have comparisons among the ecosystemservices provided by plantings of different diversitylevels been explored. In the USA, a major source offunding for stream and wetland restoration is associatedwith mitigation of wetland losses under theClean Water Act (Nrc 2001). Guidelines for successfulrestoration vary from state to state, but typicallyinvolve recommended species lists and some assessmentof total vegetation cover. Some states are currentlydeveloping improved vegetation assessmentsthat include repeated measurements of the cover of allplanted and unplanted species. In the case of NorthCarolina, this improved monitoring scheme has beendeveloped in coordination with the Carolina VegetationSurvey to ensure that monitoring data can bedirectly incorporated into an existing vegetationdatabase that is actively being used in ecologicalresearch (Lee et al. 2007). As a member of the EuropeanUnion, the UK Government pays out many millionsof pounds a year in funding restoration on farmlandunder the Environmental Stewardship scheme(http://www.defra.gov.uk/erdp/schemes/es/).This scheme exemplifies the biodiversity or ecosystemservice dichotomy. Certain activities aim to restore aservice, such as sowing field margins with a mix ofplants designed to provide pollen and nectar for beesand butterflies or to provide winter seed for birds. Inthese cases the plant mixture does not resemble anyseen in (semi-) natural systems and often containsnon-natives. Other activities are focused on biodiversity,such as the sowing of specific plant mixturesinto bare arable land or species-poor grasslands torestore particular target species-rich grasslands. Certainauthors have criticized the effectiveness ofEuropean agri-environment schemes (e.g. Kleijn et al.2006). This is partly because these authors have confusedthe aims of different activities (e.g. expecting thebird-seed mixtures to restore rare arable weed communities),but also because monitoring of outcomeshas been poor. The limited monitoring (e.g. Critchleyet al. 2004, Feehan et al. 2005) of vegetation has shownsome success, but suggests that better targeting andmore precise methods are required. This is leading toplanning for more extensive and repeated monitoringprograms and consideration of how biodiversity andecosystem service aims might be integrated (e.g. forsoil conservation).Clearly, BEF research still has many avenues toexplore before the majority of questions surroundingthe relationship between biodiversity and ecosystemfunctioning are answered. Just as clearly,restoration ecology still has a long way to go beforethe results of reconstructing an ecosystem can be aspredictable as the construction of a bridge or even aspace station. Although not all restoration projectsare suited to answering basic science questionsregarding the relationship between biodiversity andecosystem functioning, they are crucial to makingBEF science applicable to real-world situations.Direct partnerships between researchers testing BEFconcepts and restoration practitioners are crucial forecological restoration live up to its potential as anacid test for BEF ecology.