Landscape structure indices for assessing urban ecological networks

Landscape and Urban Planning 58 (2002) 269–280Landscape structure indices for assessing urbanecological networksEdward A. CookSchool of Planning and Landscape Architecture, Arizona State University,P.O. Box 872005, Tempe, AZ 85287-2005, USAAbstractAnalysis and planning of ecological networks is a relatively new phenomenon and is a response to fragmentation anddeterioration of quality of natural systems. In urban areas, the problems of land use intransigence, political and jurisdictionalissues create a difficult environment for implementing ecological networks. The specific questions addressed in this researchprogram revolve around the viability of planning an ecological network in an urban landscape. The development andarticulation of an ecological network plan was undertaken previously and in this paper, a series of assays of landscapestructure are used to examine the viability of an ecological network in the Phoenix, Arizona urban area. Three principalanalyses were utilized: (1) patch content analysis, (2) corridor content analysis, and (3) network structure analysis. Patch andcorridor content analyses examined the internal characteristic and immediate context for each of the 89 ecological networkelements. The network structure analysis incorporates a process for aggregating results of patch and corridor analyses andincorporates indicators that describe interrelationships between landscape elements. For each of these analyses the existingcondition was compared to the optimal plan to demonstrate the level of change that can be expected. The results andconclusions of this research are that an ecological network plan provides modest but important improvement in ecologicalsystems in the Phoenix urban area. # 2002 Elsevier Science B.V. All rights reserved.Keywords: Landscape structure; Ecological networks; Urban ecosystems; Urban greenways1. IntroductionThis paper demonstrates how landscape structurecan be used to assess the viability of urban ecologicalnetworks. Ecological networks are increasingly beingconsidered as a suitable approach to improving theecological value of urban open space systems (Cook,1991a; Cook and van Lier, 1994). The objective of thisanalysis is to measure the difference between alternativeecological network scenarios using landscapestructure indicators. Three analyses; patch contentanalysis, corridor content analysis, and network structureanalysis, are undertaken to describe differences.Patch content analysis uses indicators including patchtype, size, perimeter/area ratios, degree of naturalness,matrix utility factors, and isolation. Corridor contentanalysis uses indicators of corridor type, size, interior/upland ratios, matrix utility factors, connectivity, anddegree of naturalness. Network structure analysis usesindicators of mesh density connectivity, circuitry,matrix utility factors and degree of naturalness.An ecological network plan (Fig. 1) prepared inearlier phases of this study (Cook, 2000) includesrecommendations for establishing an ecological networkin the Phoenix, Arizona urban area. The planfocuses on three scales, landscape or regional, localor municipal, and site. At each scale optimal conditionsare described for establishment of an ecological0169-2046/02/$20.00 # 2002 Elsevier Science B.V. All rights reserved.PII: S 0169-2046(01)00226-2

270 E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280Fig. 1. Patches and corridors and landscape scale.network. The plan is designed to enhance the overallecological viability of the study area by reducingfragmentation, maintaining naturalness, and improvingcompatibility of the landscape matrix with theecological network. The analysis demonstrates thatthe ecological network plan provides notable improvementover the current situation and managementschemes.2. Criteria for assessing viabilityAssessment of ecological network viability can beundertaken by analyzing the inherent characteristicsof the landscape elements, the interrelationshipsbetween landscape elements and external factorsaffecting the functioning of the ecological network.A range of landscape element attributes are assessedthrough patch and corridor content analysis. Inherentcharacteristics include size and type, vegetativestructure and diversity, and naturalness. Interrelationshipsbetween individual landscape elements and thelandscape matrix are assessed through a number ofindicators that are described as context. Networkstructure analysis considers the overall effect of theinterrelationship of patches and corridors within thecontext of the urban matrix. The network structureanalysis includes mesh density, circuitry and connectivity.This process is undertaken using plans describedin previous studies for the Phoenix urbanarea, principally at landscape scale. Both the existingsituation and an optimal situation are assessed.2.1. Patch content analysisThe patch content analysis incorporates four principlevariables, (1) patch size and type, (2) vegetativestructure and diversity, (3) patch context, and (4) patchnaturalness. Each variable has several sub-variables orattributes that are used to describe and assess patch

E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280 271Table 1Summary of patch content analysisPatch size and typeOrigin 17 Natural 18 CulturalType 17 Remnant 18 IntroducedArea (mean)4650 haPerimeter length (mean) 22051 mPerimeter/area ratio (mean) 1.48Patch vegetative structure and diversityVegetative cover (mean) 0.33Native cover (mean) 0.17Structural diversity (mean) 0.62Total vegetation value (mean) 1.17Patch contextMatrix utility (mean) 0.59Isolation index (mean) 1.51T-links (mean)1.23 (43 Total)Patch naturalnessNative vegetation (mean) 0.17Soils (mean) 0.42Flows (mean) 0.35Total naturalness0.94value (mean)characteristics. Each of the patches were analyzedusing these variables. A summary of the results ofthe patch content analysis is included in Table 1.2.1.1. Patch size and typePatch size and type have been shown to affectinteractions of wildlife and affect ecological value.Forman (1995) uses five terms to classify patches,primarily based on origin and function. A remnantpatch is one that is left in its original characterfollowing widespread disturbance that changes thesurrounding matrix. A regenerated patch has beenpreviously disturbed or changed and has since naturallyreestablished vegetative cover. An introducedpatch is similar to a regenerated patch in that it waspreviously disturbed, but the establishment of newvegetation is of human origin. An environmental patchis one that exists within a predominantly naturalcontext and has significant natural values, but differsfrom the surrounding matrix. A disturbance patchis established by local disturbance of an area thatchanges the character, resulting in a patch that differsfrom its surroundings.The size of patches also affects their viability andecological value. Large patches are more valuablebecause they support large, persistent populations.Thus, there is a relationship between patch area andwildlifeabundance,persistenceanddiversity(Duerksonet al., 1997). Numerous researchers have demonstratedthis species area relationship. Large patchespreserve greater areas of interior habitat which isbeneficial to many indigenous species that are nottolerant of edge habitats. Smaller patches have moreedge area. Forman (1995, p. 47) lists ecological valuesof large and small patches.Large patches:1. Water quality protection for aquifer and lake.2. Connectivity of a low-order stream network forfish and overland movement.3. Habitat to sustain populations of patch interiorspecies.4. Core habitat and escape cover for large-homerangevertebrates.5. Source of species dispersing through the matrix.6. Microhabitat proximities for multihabitat species.7. Near-natural disturbance regimes at where manyspecies evolved with and require disturbance.8. Buffer against extinction during environmentalchange.Small patches:1. Habitat and stepping stones for species dispersal,and for recolonization after local extinction ofinterior species.2. High species densities and high population sizesof edge species.3. Matrix heterogeneity that decreases fetch (run)and erosion, and provides escape cover frompredators.4. Habitat for small-patch-restricted species. Occasionalexamples are known of species that do notpersist in larger patches.5. Protect scattered small habitats and rare species.The bottom line: large patches, large benefits, andsmall patches, small supplemental benefits.Three measures of patch size and type are used toevaluate the various attributes of patches within thissystem; patch area, perimeter length and perimeter/area ratio. Patch area is a simple measurement ofthe patch area in hectares and, with other factors,may be a good measure of long term viability of apatch. Perimeter length is used in combination with

272 E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280patch area to calculate the perimeter/area ratio. Perimeter/arearatio describes the relationship betweenpatch area and boundary length and provides usefulinformation about the potential edge effect that may bepresent in a specific patch. Perimeter/area is a simpleratio and is calculated using the following formula:D i ¼ pP2 ffiffiffiffiffiffiApwhere D i is the perimeter area ratio for patch i (alsoreferred to as patch shape index); P the perimeter ofthe patch; A the area of the patch.2.1.2. Vegetative structure and diversityVegetative structure and diversity is described usingthree principle attributes that provide some considerationof the inherent qualities of the vegetation of eachpatch. Included in this analysis are net percent areacovered by vegetation or vegetative cover, percentnative vegetation, and vegetative structural diversity.Vegetative cover is one factor that contributes to theecological health and can be used as wildlife habitat(Matthews et al., 1988). Forman (1995) notes thatplant growth (or biomass, the overall weight of livingtissue) is an attribute that can be measured as anindicator of productivity, one of four componentsForman believes can be used as a measure of ecologicalintegrity. Vegetative cover was calculated foreach patch using aerial photographs and field inventory.Multiple layers manifested as vertical stratificationis another indicator of habitat value in particular.Overlap of vegetation is not measured in the computationof coverage. Alone, net area covered by vegetationis not sufficient to fully evaluate the vegetativestructure. In urban areas, vegetation often consistslargely of exotic species and while these may providemany ecological benefits, exotic species are usuallynot as important to native wildlife species as habitat.Hence, percent of vegetation in each patch that isindigenous is an indicator of ecological value. Nativevegetation (Kearney and Peebles, 1960) coverage wascalculated for each patch as a percentage of total patcharea (Shaw et al., 1998). Representative patches weresampled for each type and values applied over similarpatch types.Structural diversity is another vegetation attributethat is an indication of ecological value. Verticalstratification of vegetation is manifested in the formof tree canopy, shrub layer and ground cover. Structuraldiversity gives more possibilities for greaterdiversity of habitat types, yielding increased diversityand greater possibilities for species survival. Smallmammals, lizards and birds will use shrub layers asescape cover, while insects may be more reliant uponground cover. Evaluation of structural diversity wasaccomplished by assessing the area covered by eachlayer. In this assessment, overlap of vegetative layerswas included, potentially yielding a value of 3.0.Sampling occurred in a representative patch for eachtype.2.1.3. Patch contextSeveral external attributes can also be used todescribe a patch and affect its ecological value. Manyexternal factors can make a patch more or less ecologicallyviable. These are analyzed with relation tothe effect on each patch. Cumulative effects of externalfactors are analyzed later in the network structureanalysis. Important considerations include the compatibilityof the matrix (matrix utility) adjacent to thepatch, the isolation of the patch and accessibility of thepatch (T-links). Matrix utility refers to the extentspecies may utilize the landscape adjacent to andimmediately surrounding patch boundaries. An idealcondition is for a patch to be buffered with compatibleland use types and vegetation. The absence of asuitable buffer of compatible land uses effectivelyincreases the edge effect and reduces the interior orcore area of a patch. Conditions that change the matrixutility value are intense land uses, significant humanimpact, absence of vegetation (in particular indigenousvegetation), presence of pollutants of toxic materials,and domestic predators (cats, dogs, etc.). Matrixutility is indexed on a scale in which a wholly compatibleadjacent matrix landscape is valued at 1 andwholly incompatible matrix is 0. Matrix utility iscalculated by measuring percent adjacent land useof the patch perimeter multiplied by the land useintensity factor.M ¼ X a i u iwhere M is the matrix utility index of patch; a thepercentage of perimeter of land use i; u the intensityvalue of land use i.Patch isolation is a function of the proximity ofpatches or corridors within a landscape mosaic. While

E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280 273connectivity is measured among a series of landscapeelements, isolation addresses the connectivity or lackof connectivity of an individual patch. It is simply anindex of the distance between patches and the numberof neighboring patches. Isolation is calculated usingthe following formula (Bowen and Burgess, 1981).R i ¼ 1 nXdijwhere R i is the isolation index of the patch; n thenumber of neighboring patches considered and d ij isthe distance between patch i and any neighboringpatch j.Accessibility is a simple measure of the number ofphysical connections (described here as T-links) ofcorridors or patches included within the ecologicalnetwork.2.1.4. Naturalness indexPatch naturalness is essentially a function of theabsence of human impact and is a measure of variabilityfrom the original ecological system. Indicatorsof human impact include the presence of exotic species,soil compaction or extent of paving, disruption offlow and function, disconnection from original locationand form. Presence of native vegetation species asan indicator was discussed in the section on vegetativestructure and diversity. In the absence of extensivebiological wildlife surveys, known habitat preferences,threatened, rare or endangered species are used.Soil compaction and extent of pavement providesinformation concerning a patch’s fundamental nutrientcycles. Compacted soils, pavement or areas of soilerosion are unproductive and cause secondary problemssuch as increased runoff, loss of nutrients andminerals and increased sediment in streams, reducingwater quality. Disruption of flows and functions can beassessed using landscape structure by comparing originaland current condition of movement of air, waterand species. Disturbed streams, migration corridorsand presence of barriers are all examples of ways flowsand function are disrupted.2.2. Corridor content analysisThe viability of corridors is assessed using similarlandscape structure indicators as with patches. Thedifferent types of corridors can be described using thesame five terms Forman (1995) uses to classifypatches, based primarily on origin and function; remnant,regenerated, introduced, environmental and disturbance.A remnant corridor is typically one that is aleft over strip after a widespread disturbance. Aregenerated corridor has evolved naturally followingsome other use or form, such as a hedgerow that hasevolved naturally. Corridor’s functions include habitat,conduit, filter, source and sink (Forman, 1995).These basic functions can be further articulated ascontributors to a range of more specific goals. Biologicaldiversity is enhanced through the preservationand restoration of riparian corridors that are criticalhabitat and important dispersal routes for species.Water resource management including flood controlerosion control and water quality are facilitated indrainage corridors. Agroforestry production isenhanced with windbreak corridors that also help tocontrol erosion. Recreational activities typically occurin corridors such as hiking, game management, bicyclingand greenbelts. Important attributes include sizeand type, vegetative structure and diversity, corridorcontext and naturalness. A summary of the results ofthe corridor content analysis in included in Table 2.2.2.1. Corridor size and typeCorridor size and type affects the five basic functions(Forman, 1995) in various ways. Two principalcharacteristics of width indicate how well a corridormay accommodate these functions. The internal areais important to accommodate stream flows, as conduitfor species migration, as source pool, and as sink,controlling erosion nutrient runoff, sedimentation andflooding. In addition, the upland zone of a corridormay perform many of the same ecological functionsbut also filtering and buffering. In urban matrices,the filter and buffer function is essential to maintainthe usability of the internal entities. Dramstad et al.(1996) note that 2nd–4th order stream corridors shouldmaintain upland habitat on both sides, enough tocontrol dissolved-substance inputs from the matrix.This also provides conduit for interior species andsuitable habitat for flood plain species displaced duringflood events. For larger corridors such as rivers of5th–10th order, Dramstad et al. (1996) recommendupland zones on both sides, but also a ladder pattern oflarger patches crossing the flood plain. This structureprovides a hydrologic sponge that traps sediment,

274 E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280Table 2Summary of corridor content analysisCorridor size and typeOrigin 15 Natural 17 CulturalType15 Environmental 17 IntroducedLength (mean)28792 mAverage width (mean) 245 mAverage interior0.58width (mean)Average filter0.11width (mean)Area (mean)1079 haCorridor vegetative structure and diversityVegetative cover (mean) 0.20Native cover (mean) 0.16Structural diversity (mean) 0.35Total vegetation0.71value (mean)Corridor contextMatrix utility (mean) 0.56Diversity gradient (mean) 86.40 mT-gaps (mean)1.84 (59 Total)Longest continuous 18870 mdistance (mean)Link segments (mean) 2.78 (89 Total)Corridor naturalnessNative vegetation (mean) 0.16Soils (mean) 0.37Flows (mean) 0.40Total naturalness0.93value (mean)provides nutrients and habitat for flood plain species.Forman (1995) notes that, overall, wider corridors arebetter and will enhance all five functions of habitat,conduit, filter, source and sink. He also notes thatresearch is still needed to determine how wide acorridor needs to be to function effectively. Oneimportant indicator is known and that is that a corridorshould be wide enough to accommodate an uplandzone of vegetation to control and filter runoff.2.2.2. Vegetative structure and diversityInternal structure of corridors has been discussed tosome extent in the section describing patch structureand also in this section on corridors. One importantdistinction that can be made concerning the internalstructure of corridors is that this attribute is one ofseveral that is critical to improve connectivity ofcorridors. The presence of interior entities (i.e. water,flood plains, native vegetation, etc.) improves theconnectivity value of a corridor. The physical structureof the upland zone is also important in that it contributesto the viability of the interior as well asproviding buffer and filter functions. The type, structuraldiversity and preponderance of vegetative coverare used as an indicator and measured for each corridor.2.2.3. Corridor contextExternal characteristics of corridors are examinedin a similar manner to those of patches. Matrix utilityvalues are assigned according to intensity of theadjacent land uses along the corridor. Corridors thattraverse different habitat zones have potential to functionas important conduits for many species. Ascorridors (principally riparian corridors) traverse thelandscape through the upper reaches of a watershedto the lower elevations where they converge and formrivers and streams, they pass through a variety ofecosystem types. Thus, there is great potential forincreased diversity, transmission of nutrients and continualexchange. Most often this condition yieldsgreater ecological health. This condition is a diversitygradient and is another indication of the relativeecological value of a corridor. Values are assignedfor the purpose of this study according to the elevationchange within the reach of the corridor that falls withinthis study area.Continuity is a factor that when combined withothers provide an indication of overall connectivityof a corridor. The presence or absence of gaps in acorridor can be evaluated to assess the continuitypresent in a corridor. In urban areas, continuity is oftenbroken by roads, channel reconfiguration or filling, orsimply by disturbance or human impact resultingfrom use or abuse. Particularly for those species thatare exclusively terrestrial, gaps in a corridor representsignificant barriers to migration and other importantfunctions. Each corridor is examined and a continuityvalue is assigned corresponding with the presence ofgaps or barriers in the corridor. In the network analysis,the concept of continuity is extended and a measurementof connectivity is computed.2.2.4. Naturalness indexThe naturalness index for corridors uses the sameindicators as patches; native species, soils and flows.

E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280 275With stream or river corridors in particular, soil erosionand disruption of flows are very important factors.Because of the intermittent nature of many arid regionriver and stream corridors, extreme hydrologic eventsoccur frequently and significantly change the structureof a channel. While these events may be of naturalorigin, the increases in runoff within watersheds due tourbanization, soil compaction and encroachment onflood plains exacerbates problems and rapidly reducesthe natural character of the corridor and the functionsthat occur within.2.3. Network structure analysisThe analysis of patch and corridor content is onlysignificant at the landscape scale if data are integratedinto a larger more comprehensive analysis. Networkstructure analysis introduces a process for aggregatingresults of patch and corridor analysis and incorporatesindicators that describe interrelationships betweenlandscape elements. The collection of these elementsforms the whole landscape mosaic and as patches andcorridors connect and intersect, they form a network ofpotentially high value elements in which ecologicalfunctions are preserved and/or restored. Several factorscontribute to making an ecological network moreviable. In urban landscapes, the competition of variousland uses and exotic species is intense. Fragmentationand isolation contribute to deterioration of ecologicalfunctions and ultimately decline in native species.Thus, establishing or maintaining linkage betweenpatches and corridors is essential to facilitate ecologicalfunctions. Two factors largely contribute to thesuccessful ecological functioning of a network: (1) theoverall density of suitable patches and corridors and(2) the extent to which they are linked.2.3.1. Mesh density/naturalness indexThe density of suitable patches and corridors ismeasured through mesh density analysis. Mesh densityhas three factors that are analyzed: (1) the actualnet area covered by patches and corridors in thenetwork, (2) the matrix utility as related to thosepatches and corridors, and (3) the relative quality ofpatches and corridors. The net area covered is computedby calculating the area occupied by networkelements within the study area or within a particularsector. This is described as percent net density of theTable 3Mesh density/naturalness indexArea (ha)Totalarea (%)Natural patch 161137.00 0.22 1.79Cultural patch 1632.00 0.00 0.15Natural corridor 30490.00 0.04 1.98Cultural corridor 4047.00 0.01 0.02Total 197306.00 0.27Study area 739218.00 1.00Naturalnessvaluetotal area. The greater the value, the more area that isavailable where ecological functions can occur. Themesh density/naturalness index is described in Table 3.2.3.2. Matrix utility indexMatrix utility has been discussed previously in thesections describing patch and corridor content analysis.The values computed for each patch and corridor areaggregated and averaged for an overall matrix utilityvalue. The relative quality of patches and corridorsrefers to the patch and corridor content analysis indicatorsof naturalness, vegetation, type and size. Thesevalues are also aggregated and averaged to assess overallnetwork quality and are summarized in Table 4.2.3.3. Circuitry and connectivityThe linkage of various elements within the networkcan be explained with the concepts of circuitry andconnectivity. Network circuitry is described as theextent to which loops or circuits are present in thenetwork. The ‘‘alpha index for circuitry’’ measures thenumber of loops present divided by the maximumnumber of loops possible. Loops are importantbecause they provide alternative migration routesfor organisms that may need to avoid disturbancesor predators. Circuitry is measured using the followingTable 4Matrix utility indexArea (ha)Totalarea (%)MatrixutilityNatural patch 161137.00 0.22 0.72 0.16Cultural patch 1632.00 0.00 0.47 0.00Natural corridor 30490.00 0.04 0.73 0.03Cultural corridor 4047.00 0.01 0.54 0.00Total 197306.00 0.27 0.19Study area 739218.00 1.00Area matrixutility

276 E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280formula (Hagget et al., 1977; Taafe and Garthier,1973; Forman and Godron, 1986)a ¼ L V þ 12V 5where a is the degree of network circuitry; L thenumber of linkages; V the number of nodes.The results of the circuitry analysis of the existingsituation yielded the presence of 43 nodes (V) and 75links (L) for a circuitry value of 0.39.Connectivity is a measure of the extent to which allnodes are connected. As in previous discussions ofnaturalness and the preference for native species, twotypes of connectivity may exist. Natural connectivityutilizes existing or restored naturally originated corridorsyielding potentially greater ecological benefits.Artificial connectivity is achieved through establishmentof culturally derived corridors and in most casesecological benefits will be less when compared to anatural corridor of similar characteristics. In urbanareas, utilization of artificial connectivity may be anessential dimension of a network. Since many naturalcorridors are likely to have been removed or destroyed,supplemental connections can be achieved throughproper design of artificial corridors. Together, the alphaindex of circuitry and the gamma index of connectivitydescribe the level of network complexity (Forman andGodron, 1986) and provides an overall index of theeffectiveness of linkages (Dramstad et al., 1996).The ‘‘gamma index of connectivity’’ is the ratio ofthe number of links in a network to the maximumnumber of links possible. The number of links presentand the number of nodes are totaled from the networkand applied in the following formula (Lowe andMoryadas, 1975; Hagget et al., 1977; Sugihara andMay, 1990; Forman and Godron, 1986)Lg ¼3ðV 2Þ ðSÞwhere g is the gamma index of connectivity; L thenumber of linkages; V the number of nodes; S the linksuitability factor.For the purposes of this analysis, another factor wasadded, that of link suitability. This factor is an adjustmentfor corridor naturalness. In an urban context, thepotential for migration is significantly affected by thepresence of suitable vegetative cover and the absenceof human impact.The results of the connectivity analysis of theexisting situation yielded the presence of 43 nodes(V), 89 links (L) and a link suitability value of 0.31 fora connectivity value of 0.22.2.4. Summary of existing situation analysisThe existing landscape mosaic of the Phoenixmetropolitan area has a range of open space elementsof varying ecological value. Many of the remnantpatches and corridors have only recently becomedisconnected from the supporting landscape structure.Soule (1991) notes, however, that as isolated fragmentsage, the number of species decline and overallecological health deteriorates. The existing conditionanalysis assumes that current land use and landscapemanagement practices will continue. No plan forreestablishing connections or improvement of vegetativecover is put into effect. In essence, the status quoreigns. The indicator values outlined in Tables 1–4 forpatch content analysis, corridor content analysis andnetwork structure analysis summarize this currentcondition.2.5. Summary of analysis of optimal conditionThe implementation of an ecological network plancan yield changes in the overall ecological value. Thisanalysis builds upon the same existing patch andcorridor elements that are present in the Phoenixmetropolitan area (Fig. 1), but employs a series ofmanagement, preservation or restoration strategies toimprove the ecological value. This situation is viewedas a system or a comprehensive ecological network.Specific recommendations are assumed to be implementedthroughout the network on an individual patchor corridor basis.The examples of site improvement strategiesdescribed in the ecological network plan (Cook,2000) are typical of the types of strategies that couldbe employed. These include actions such as revegetationof disturbed sites, creating artificial or syntheticcorridors or patches, expanding buffers by extensifyingadjacent land uses, and others. Based on thesestrategies adjusted patch content, corridor contentand network structure analyses are undertaken and asummary is displayed in Tables 5–8. The changedvalues are noted in bold in each case representing the

E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280 277Table 5Summary of patch content analysis (optimal) aPatch size and typeOrigin 17 Natural 18 CulturalType 17 Remnant 18 IntroducedArea (mean)4650 haPerimeter length (mean) 22051 mPerimeter/area ratio (mean) 1.48Patch vegetative structure and diversityVegetative cover (mean) 0.38Native cover (mean) 0.27Structural diversity (mean) 0.62Total vegetation value (mean) 1.27Patch contextMatrix utility (mean) 0.73Isolation index (mean) 1.51T-links (mean)1.23 (43 Total)Patch naturalnessNative vegetation (mean) 0.27Soils (mean) 0.59Flows (mean) 0.53Total naturalness1.39value (mean)a Values in bold represent the increased ecological value in theoptimal condition plan.increased ecological value in the optimal conditionplan.2.5.1. Optimal plan patch adjustmentsFor the purpose of this study, patch size and typewere not adjusted for the optimal plan. Significantimprovement in ecological values can, however, occurwith enlargement of patch size or reduction of perimeterarea ratios to increase interior or core areas.Although, the scope of this study did not address thisstrategy, it should be a part of any ecological networkplan where feasible.Optimal patch vegetative structure and diversityvalues are adjusted by introducing native vegetationin patches, principally where it is absent in cultivatedparks. A site study focusing on the Indian Bend WashPark (Merkle et al., 1996) (an existing linear park inScottsdale, AZ), demonstrates that in cultivated urbanparks approximately 50% of the area currently coveredby exotic vegetation could be converted to nativespecies with no significant loss of traditional parkfunctions. The resulting adjustment increases themean native vegetation coverage for patches from0.17 to 0.27.Table 6Summary of corridor content analysis (optimal) aCorridor size and typeOrigin 15 Natural 17 CulturalType15 Environmental 17 IntroducedLength (mean) 28792 mAverage width (mean) 245 mAverage interior0.58width (mean)Average filter0.30width (mean)Area (mean)1079 haCorridor vegetative structure and diversityVegetative cover (mean) 0.27Native cover (mean) 0.25Structural diversity (mean) 0.48Total vegetation1.00value (mean)Corridor contextMatrix utility (mean) 0.77Diversity gradient (mean) 86.40 mT-gaps (mean)0 (0 Total)Longest continuous 18870 mdistance (mean)Link segments (mean) 2.78 (89 Total)Corridor naturalnessNative vegetation (mean) 0.25Soils (mean) 0.60Flows (mean) 0.62Total naturalness1.47value (mean)a See footnote to Table 5 for explanation of values in bold.Optimal patch context values are adjusted byincreasing native vegetation plantings in land usesadjacent to existing patches improving matrix utilityvalues. Increase in matrix utility value, in effect, helpsestablish a more significant buffer external to theboundaries of the patch. In turn, the negative aspectsTable 7Mesh density/naturalness index (optimal) aArea (ha) Total area (%) Natural valueNatural patch 161137.00 0.22 1.79Cultural patch 1632.00 0.00 1.01Natural corridor 30490.00 0.04 2.03Cultural corridor 4047.00 0.01 0.97Total 197306.00 0.27Study area 739218.00 1.00a See footnote to Table 5 for explanation of values in bold.

278 E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280Table 8Matrix utility index (optimal) aAreaTotalarea (%)MatrixutilityNatural patch 161137.00 0.22 0.82 0.18Cultural patch 1632.00 0.00 0.64 0.00Natural corridor 30490.00 0.04 0.84 0.03Cultural corridor 4047.00 0.01 0.70 0.00Total 197306.00 0.27 0.22Study area 739218.00 1Area matrixutilitya See footnote to Table 5 for explanation of values in bold.of edge effect (i.e. loss of interior or core habitat areaof patches) is reduced. The site study for DobsonRanch residential community (Brenda et al., 1997),illustrates how existing land uses can be ‘‘extensified’’to make them more usable as habitat and for otherecological functions. The adjustment is an increase inmatrix utility values from the existing condition (0.59)to the optimal (0.73) of 14%. As with delimitationsnoted concerning patch size and type, context indicatorssuch as isolation and terrestrial links can beimproved by adding more usable patches or corridorsto the network. This was beyond the scope of thisstudy.Optimal patch naturalness adjustments includeincrease in area covered by native vegetation, reductionof area with paving, compacted or eroded soils,and site adjustments to improve flows (i.e. removal ofobstructions or barriers) to return site functions to amore natural condition. Adjustments, as with thoseprevious, are generally confined to cultivated patchessince natural patches have retained much of theiroriginal condition. Adjustments for native vegetationare reflective of those noted previously. Soils adjustmentsare introduced by specific improvements to siteconditions, principally by eliminating or reducingpaving where feasible or managing parks and othersites to limit the areas where soil becomes compacteddue to excessive use. The increase in the mean of soilsvalues is 0.17 (0.42–0.59) and is linked to the introductionof native vegetation in areas currently in amore cultivated condition. Flows are improved incultivated patches by reconnecting severed links ordrainage corridors or providing alternatives for movementof water, air, nutrients and other elements. Thisis also accomplished through cultivated park siteredesign (Merkle et al., 1996). Overall, patch naturalnessvalues increase from 0.94 to 1.39.2.5.2. Optimal plan corridor adjustmentsAs with patches, the overall size and type of corridorshave not been adjusted for this study. Optimalcorridor size and type, also includes data concerningcorridor interior and filter widths. Adjustments to thefilter width of corridors represent an increase in meancorridor width of 19% (0.11–0.30). The most commonadjustment is for canals. The site study for canals(Section 5.4.5) demonstrates that currently underutilizedcanal banks can become important filters andbuffers of incompatible use and flows. In addition,restoration of river and stream corridors (Cook,1991a,b) revitalizes buffer and filter functions andprovides significant ecological benefits.Optimal corridor vegetative structure and diversityvalues are adjusted in similar ways as the patch values.Specifically, native vegetation is introduced wherefeasible, and in the case of canals, adjustments tovegetative cover and structural diversity are notedbecause in the current situation, there is no vegetationpresent. Further illustration of adjustments areincluded in the site studies of the Salt River corridor(Cook, 1991b) and the canal system (Fifield et al.,1990)). The most notable adjustment is the increase inthe mean of percent of area covered by native vegetationfrom 0.16 to 0.25.Optimal corridor context values are adjusted in twoways. First, matrix utility values are adjusted in thesame fashion as with patches. The mean value ofmatrix utility is increased from 0.62 to 0.77. Second,site scale corridor studies (Bushbacher et al., 1996;Merkle et al., 1996; Cook, 1991b; Fifield et al., 1990)all recommend the removal of barriers or gaps incorridors that would negate flow of species, nutrientsand other important elements. With all gaps eliminatedor alternative connections established, the totalof 59 gaps in the existing condition is reduced to 0 inthe optimal plan.Optimal corridor naturalness is adjusted essentiallythe same way as patch naturalness. Existing culturalcorridors are those with most significant adjustments.Specifically, the values for canals have the greatestpromise for improvement since existing naturalnessvalues are practically non-existent. Mean values fornative vegetation (0.17–0.27), soils (0.37–0.60) and

E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280 279flows (0.40–0.62) are improved and the total naturalnessvalue increases from 0.94–1.47.2.5.3. Optimal plan network adjustmentsAdjustments for mesh density/naturalness (Table 7)and the matrix utility index (Table 8) are extrapolatedfrom the previous patch and corridor adjustments.Since, actual area devoted to the ecological networkdoes not change from the existing to the optimal plan,the change in the overall mesh density/naturalnessindex is not significant. The principle reason is thatin this context, over 82% of the area within theecological network is in natural patches that requiremanagement strategies that support the current functioningecosystems as opposed to restoring or introducingviable ecosystems. More importantly, thematrix utility index (Table 8) works toward preservingthese important ecological nodes. The adjustment inthe matrix utility index is modest but notable (0.19–0.22) and could provide important buffering and filteringfunctions from adjacent land uses.The computation of the degree of network circuitryin the existing situation includes a network of 45 nodesand 75 links. This tabulation eliminates any links(such as canals) from the network that have negligibleecological values. The adjusted values based on theoptimal plan allow for 14 (total 89) additional links,boosting the degree of network circuitry from 0.39 to0.59.Connectivity in the optimal plan is increased in twoways. First, adjustments are made to the suitability oflinks or corridors. In the existing situation, the overallsuitability value for links is 0.31 (corridor naturalnessvalue), while in the optimal plan it increases to 0.48.Adjustment to the gamma index of connectivityincreases the value from 0.22 to 0.34.3. Filter for ecological integrityForman (1995) notes that ecological integrity couldbe measured as the single most important or sensitiveattribute of an ecological system. He also states that ‘‘asystem with ecological integrity has near naturalconditions for four broad characteristics: productivity,biodiversity, soil and water’’. If we accept this premise,then ecological integrity should be the goalof any sustainable environment. Productivity, as anindicator of ecological integrity, should be at nearnatural levels as an average for a landscape. Thismeans it should be maintained at a level not far fromthat which would prevail if the landscape had nativeecosystems. Specific elements that could be measuredinclude biomass, animal or secondary production,length of food chains, herbivory and decomposition.Biodiversity is represented by a number of nativespecies for each major group (i.e. mammals, trees).The condition should be relatively few extirpatedspecies and measurements could include assessmentof community types, keystone species, rare speciesand genetic diversity. Soil is best measured by theamount of soil erosion, compaction or the amount ofarea sealed by paving. Measures could include soilmoisture, runoff of minerals and nutrients and toxicmaterials in soils. Water can be assessed by quality andquantity. Quality indicators include measures of turbidity,nutrient status and fish populations. Forman(1995) notes that fish communities may be the bestoverall measure of water quality. Water quantity isbest measured through indicators such as the ability toaccommodate floods, low flow events, evapotranspiration,water tables, stream flow and aquifer recharge.The greatest challenge of using these four indicatorsis the extensive amounts of data required to makeproper assays. In the case of the Phoenix metropolitanarea ecological network, some of these indicators canbe applied. It is still uncertain what near natural levelsmay be interpreted to mean. Perhaps, the most usefulstandard may be to set levels that are appropriate goalsfor the context in which the assessment is to beundertaken.4. ConclusionThis analysis shows that notable improvement inecological value can be achieved by implementing aplanning strategy for open space systems thatembraces the concept of ecological networks. Theresults shown here use a series of landscape structureindicators at a single point in time. Longer termstudies and a broadening of data collected and analyzedmay yield additional information about theviability of ‘‘natural systems’’ in an urban context.This paper does not address the range of relatedbenefits provided by an urban ecological network

280 E.A. Cook / Landscape and Urban Planning 58 (2002) 269–280(i.e. recreational opportunities, aesthetic quality, culturalidentity), but it is important to include thesefactors in any comprehensive study to understand thefull benefits of this approach.ReferencesBowen, G.W., Burgess, R.L., 1981. A Quantitative Analysis ofForest Island Pattern in Selected Ohio Landscapes. Tennessee:Oak Ridge National Laboratory, Environmental SciencesDivision, Publication ORNL-TM-7759, Oak Ridge, p. 111.Brenda, K., Brown, B., Dioszegi, M., Gray, K., Lansing, T.,Maldondo, D., Montei, K., Walsh, J., 1997. Dobson RanchAnalysis. Mesa, AZ, Arizona State University, Tempe, AZ, p. 60(unpublished report).Bushbacher, M., Chavez, G., D’Angelo, C., Mason, G., Ramirez,M., 1996. Lower Verde River Ecovision. Arizona StateUniversity, Tempe, AZ, p. 42 (unpublished report).Cook, E.A., van Lier, H.N. (Eds.), 1994. Landscape Planning andEcological Networks. Elsevier, Amsterdam, p. 354.Cook, E.A., 2000. Ecological Networks in Urban Landscapes.Wageningen University, Wageningen, p. 201.Cook, E.A., 1991a. Urban landscape networks an ecologicalplanning framework. Landsc. Res. 16, 7–15.Cook, E.A., 1991b. Ecosystem modeling as a method for designingsynthetic fluvial landscapes: a case study of the Salt River inArizona. Landsc. Urban Planning 20, 291–308.Dramstad, W., Olson, J., Forman, R., 1996. Landscape EcologyPrinciples for Landscape Architecture and Land Use Planning.Island Press, Washington, DC, p. 89.Duerkson, C.J., Elliot, D.L., Hobbs, N.T., Johnson, E., Miller, J.R.,1997. Habitat Protection Planning Where the Wild Things Are.American Planning Association, Washington, DC. P. 82.Fifield, M., Pihlak, M., Cook, E., 1990. Metropolitan Canals: ARegional Design Framework. Colleger of Architecture andEnvironmental Design, Arizona State University, Tempe, AZ,p. 76.Forman, R.T.T., 1995. Land Mosaics. Cambridge University Press,Cambridge, p. 632.Forman, R.T.T., Godron, M., 1986. Landscape Ecology. Wiley,New York, p. 619.Hagget, P., Cliff, A.D., Fry, A., 1977. Locational Analysis inHuman Geography, 2nd Edition. Wiley, New York, p. 454.Kearney, T.H., Peebles, R.H., 1960. Arizona Flora, 2nd Edition.University of California Press, Berkeley, CA, p. 1085.Lowe, J.C., Moryadas, S., 1975. The Geography of Movement.Houghton Mifflin, Boston, p. 442.Matthews, M.J., O’Conner, S., Cole, R., 1988. Database for theNew York State urban wildlife habitat inventory. Landsc. UrbanPlanning 15, 23–37.Merkle, S., Mon, R., Socha, R., Wittig, T., Zimmerman, S., 1996.Indian Bend Wash, Systems Network and Ecological Analysis.Arizona State University, Tempe, AZ, 94 p (unpublished report).Shaw, W., Harris, L., Livingston, M., 1998. Vegetative characteristicsof urban land covers in metropolitan Tucson. UrbanEcosyst. 2, 65–73.Soule, M., 1991. Land use planning and wildlife maintenance:guidelines for conserving wildlife in an urban landscape. J. Am.Planning Assoc. 3, 313–323.Sugihara, G., May, R.M., 1990. Applications of fractals in ecology.Trends Ecol. Evol. 5, 79–86.Taafe, E.J., Garthier, H.L. Jr., 1973. Geography of Transportation.Prentice Hall, Englewood Cliffs, NJ, p. 442.