the role of landscape connectivity in planning and ... · pdf fileesa published by the...

esa

Published by the Ecological Society of America

esa

The Role of Landscape Connectivityin Planning and ImplementingConservation and Restoration

PrioritiesDeborah A. Rudnick, Sadie J. Ryan, Paul Beier, Samuel A. Cushman, Fred Dieffenbach,

Clinton W. Epps, Leah R. Gerber, Joel Hartter, Jeff S. Jenness, Julia Kintsch,Adina M. Merenlender, Ryan M. Perkl, Damian V. Preziosi, and Stephen C. Trombulak

Fall 2012 Report Number 16

The Role of Landscape Connectivityin Planning and ImplementingConservation and Restoration

Priorities

Issues in EcologyIssues in Ecology

© The Ecological Society of America • [email protected] esa 1

ISSUES IN ECOLOGY NUMBER SIXTEEN FALL 2012

The Role of Landscape Connectivity in Planning andImplementing Conservation and Restoration Priorities

SUMMARY

Landscape connectivity, the extent to which a landscape facilitates the movements of organisms and their genes, facescritical threats from both fragmentation and habitat loss. Many conservation efforts focus on protecting and

enhancing connectivity to offset the impacts of habitat loss and fragmentation on biodiversity conservation, and toincrease the resilience of reserve networks to potential threats associated with climate change. Loss of connectivity canreduce the size and quality of available habitat, impede and disrupt movement (including dispersal) to new habitats, andaffect seasonal migration patterns. These changes can lead, in turn, to detrimental effects for populations and species,including decreased carrying capacity, population declines, loss of genetic variation, and ultimately species extinction.

Measuring and mapping connectivity is facilitated by a growing number of quantitative approaches that can integratelarge amounts of information about organisms’ life histories, habitat quality, and other features essential to evaluatingconnectivity for a given population or species. However, identifying effective approaches for maintaining and restoringconnectivity poses several challenges, and our understanding of how connectivity should be designed to mitigate theimpacts of climate change is, as yet, in its infancy.

Scientists and managers must confront and overcome several challenges inherent in evaluating and planning for con-nectivity, including:

•characterizing the biology of focal species;

•understanding the strengths and the limitations of the models used to evaluate connectivity;

•considering spatial and temporal extent in connectivity planning;

•using caution in extrapolating results outside of observed conditions;

•considering non-linear relationships that can complicate assumed or expected ecological responses;

•accounting and planning for anthropogenic change in the landscape;

•using well-defined goals and objectives to drive the selection of methods used for evaluating andplanning for connectivity;

•and communicating to the general public in clear and meaningful language the importance ofconnectivity to improve awareness and strengthen policies for ensuring conservation.

Several aspects of connectivity science deserve additional attention in order to improve the effectiveness of design and imple-mentation. Research on species persistence, behavioral ecology, and community structure is needed to reduce the uncertaintyassociated with connectivity models. Evaluating and testing connectivity responses to climate change will be critical toachieving conservation goals in the face of the rapid changes that will confront many communities and ecosystems. All ofthese potential areas of advancement will fall short of conservation goals if we do not effectively incorporate human activitiesinto connectivity planning. While this Issue identifies substantial uncertainties in mapping connectivity and evaluatingresilience to climate change, it is also clear that integrating human and natural landscape conservation planning to enhancehabitat connectivity is essential for biodiversity conservation.

Cover photos: Examples of ways different species move through landscapes and depend on connectivity. Clockwise starting on the upper left: a) The interconnection ofocean surface current patterns provides pathways for dispersal of larvae between coral reefs. b) A network of riparian corridors used by wildlife to move through an agri-cultural landscape. c) Continuous grasslands are used by migrating wildebeest in eastern Africa. d) Intact lowland forest is used by endemic forest birds for dispersalbetween mountain ranges.

Photos credits: a) NASA Goddard's Scientific Visualization Studio. b) Adina Merenlender. c) Flickr user Abeeeer. d) Flickr user Daniel Lane.

© The Ecological Society of America • [email protected] esa

Introduction

What Is Landscape Connectivity andHow Does It Affect ConservationObjectives?

Connectivity is the extent to which move-ments of genes, propagules (pollen and seeds),individuals, and populations are facilitated bythe structure and composition of the land-scape. A landscape’s connectivity is definedrelative to the requirements of the organismsthat live within it and move through it.Therefore, connectivity is species and contextdependent. Consider the interconnection ofocean surface current patterns providing path-ways for dispersal of larvae between coral reefs;a network of riparian corridors used by wildlifeto move through an agricultural landscape;the continuity of grasslands used by migratingwildebeest; intact lowland forest throughwhich endemic forest birds move (see coverphotos). Each of these examples demonstratesthat connectivity is measured relative to theease or difficulty with which a particularspecies is able to move across a particular landor seascape.

Connectivity has both structural and func-tional components. Structural connectivitydescribes the physical characteristics of a land-scape that allow for movement, includingtopography, hydrology, vegetative cover, andhuman land use patterns. Functional connectivitydescribes how well genes, propagules, individu-als, or populations move through the landscape.Functional connectivity results from the waysthat the ecological characteristics of the organ-ism, such as habitat preference and dispersalability, interact with the structural characteris-tics of the landscape. The examples providedon the cover depict the ways that differentspecies move through and depend on the land-scape and demonstrate both functional andstructural connectivity, whereby ecological

requirements of individual species interact withthe composition and configuration of the land-scape. These interactions influence the abilityof individuals and populations to move amonglocations to find key resources, such as food,water, appropriate substrates for sessile organ-isms, or breeding partners.

The destruction and degradation of naturalhabitats on which all organisms rely – includ-ing humans – is occurring at an unprece-dented rate across most regions of our planet.As humans convert land for resource extrac-tion and for urban and agricultural uses, and asour impacts on global climate continue togrow, we profoundly change the physical,chemical, and biological character of theselandscapes. Land use changes may reduce theamount of a habitat or fragment it, breaking itup into smaller or differently arranged units.This process changes not only the size of habi-tat patches but also other landscape features,such as patch geometry or the amount of edgehabitat, that may be of fundamental impor-tance to species, communities, and ecologicalfunctions. Because human-caused disturbancesoften occur in shorter timeframes and overlarger areas than do natural disturbances, eco-logical communities face challenges of how toadapt and respond to novel rates and scales ofdisturbances that are quite different fromthose with which they may have evolved.

Fragmentation, habitat degradation, andhabitat loss are the dominant mechanisms bywhich connectivity is reduced or lost, and arewidely recognized as major drivers of the pre-sent global biodiversity crisis. Fragmentation,the subdivision of habitat into smaller or moreisolated remnants, can directly impact speciespersistence and accelerate local extinctionrates. Habitat fragmentation is frequently asso-ciated with habitat loss. However, fragmenta-tion can also eliminate dispersal or gene flowwithout causing impacts on a population’s core

The Role of Landscape Connectivity in Planning andImplementing Conservation and Restoration PrioritiesDeborah A. Rudnick, Sadie J. Ryan, Paul Beier, Samuel A. Cushman, Fred Dieffenbach, Clinton W. Epps,

Leah R. Gerber, Joel Hartter, Jeff S. Jenness, Julia Kintsch, Adina M. Merenlender, Ryan M. Perkl,Damian V. Preziosi, and Stephen C. Trombulak




habitat – for example, a highway bisecting amovement corridor.

For any given species, some parts of thelandscape provide better opportunities thanothers to fulfill its ecological requirements,such as food, breeding habitat, or refuge frompredation. Fragmentation and degradation canfurther increase the patchiness of the land-scape in terms of meeting a species’ needs.Conserving connectivity in this contextrequires identifying, maintaining, and possiblyenhancing the linkages between suitablepatches of habitat in the landscape. Corridors,which are generally linear spaces that facili-tate movement between patches, are fre-quently used as a tool for conserving orenhancing linkages. The creation or protec-tion of corridors can maintain connectivity formobile species, such as ungulates or largefelines that typically have large territories.

Corridors provide structural connectivityand are consistent with the functional con-nectivity needs of animals that can takeadvantage of linear spaces to move among dis-parate habitat patches. However, landscapeconnectivity is highly diverse and species-dependent, and other forms of connectivity

may be more relevant to other types of organ-isms; for example, a linked mosaic of smallwetlands for breeding populations of amphib-ians, continuity of vegetated intertidal rockysubstrate along a coastline for a marine snail,or a heterogeneous assemblage of meadowplant communities with different floweringtimes for a population of pollinators. Thechallenge of matching connectivity patternsto ecological requirements becomes evengreater when we expand our thinking to con-sider maintaining or restoring connectivity formultiple species or entire communities.

Many populations and ecosystem functionsare dependent on extensive, well-connectedhabitats; however, understanding the factorsthat contribute to landscape connectivity forspecific populations, species, or communitiesis challenging. This Issue reviews the impor-tance of habitat connectivity, summarizes cur-rent science-based strategies for mitigating thenegative ecological effects of fragmentation,explores data gaps and limitations of connec-tivity models, and describes obstacles andopportunities for developing policies and man-agement approaches that improve connectiv-ity and reach conservation goals.

Case Study 1. Managing for Marine Connectivity: Marine Protected Areas in the Gulf of California,Mexico

The growing movement toward ecosystem-based management, including networks of no-take zones in marine ecosystems (marineprotected areas, or MPAs) requires that these conservation areas be deliberately and adequately spaced to allow for connectivity. Theperformance of a network of sites designed with the two-fold purpose of protecting commercial species and allowing for spillovereffects (movement of organisms from protected areas into harvestable areas) will largely depend on whether sites in a network arefunctionally and structurally linked to each other by both biological (e.g., dispersal of organisms) and physical (e.g., currents)processes. Although the number and extent of MPAs has increased recently, studies have shown that, on a global scale, average dis-tance between neighboring MPAs exceeds the distance of reef organism propagule dispersal. This distance suggests that some taxacould become genetically isolated if populations cannot reach each other, undermining the viability of populations in the MPAs.

The conservation of species, habitats, and ecoregions depends on developing practical, efficient, and effective planning strategies.This is especially true in the marine realm, where threats are diffuse and difficult to both identify and quantify. Well-designed networksshould include MPAs and other conservation and management areas that support each other by taking advantage of oceanic currentsand movement/migration capabilities of species. They also provide much-needed resilience against a range of threats. Because estab-lishment of isolated marine reserves may not alone suffice for the conservation of biodiversity, identifying the level of connectivitybetween the areas is a critical aspect in network design.

In the Gulf of California, Mexico (GOC), two organizations, Comunidad y Biodiversidad and The Nature Conservancy, recently com-pleted a marine ecoregional assessment to identify priority conservation sites and establish a network of conservation areas. Thisanalysis identified 54 conservation areas that are deemed critical to marine conservation objectives, which cover 26% of the ecoregion.An important step towards implementing the assessment will be to account for connectivity between putative sites. To move from con-nectivity assessments based exclusively on structural attributes of connectivity to a detailed assessment of actual connectivity, mod-els are used to track species’ dispersal from site to site as well as movement through the matrix (for example, satellite tracking and thedevelopment of oceanographic models for the entire ecoregion). Pop-up satellite archival tags are being used globally for many marinespecies (e.g., the Tagging of Pacific Predators Program, http://www.topp.org/) and can greatly enhance knowledge of the dispersal offocal species. For example, sea turtle, cetacean, and whale shark tagging programs are already underway in the ecoregion, andexpanded versions of these programs are expected to provide a more complete understanding of connectivity throughout the ecore-gion. Integrating data from these tagging programs into the GOC ecoregional assessment is an important priority in understanding therole of connectivity in marine spatial planning.


How Fragmentation AffectsMovement: From Genes toSpecies

Landscape fragmentation affects ecologicalcommunities at multiple levels of organiza-tion. Here, we briefly explore these effects,ranging from the movement of individuals andgene flow within and between populations toshifts in species range and species persistence.

Landscape connectivity is important for dis-persing or migrating individuals. Dispersalincreases resilience to disturbances by allowingorganisms to track their shifting habitats, andit promotes the spread and expansion of popu-lations. In some species – for example, wilde-beest in Africa, bison in North America, awide variety of bird species – seasonal migra-tion has evolved as a means of maximizingaccess to critical resources as ecological condi-tions change throughout the year. Habitat frag-mentation can disrupt dispersal and migrationin several ways. First, edges of the remnanthabitat patches may act as filters or barriersthat discourage or impede movement. Second,increased distances between suitable habitatpatches may influence the likelihood of suc-cessful movement. Last, the composition andstructure of the intervening landscape mosaicmay influence the permeability of the land-scape to movements by different organisms.If fragmentation impedes seasonal migration,wildlife may be cut off from seasonal resources.If dispersal routes are blocked or altered, organ-isms may experience higher rates of mortalitywhen trying to disperse, or they may bestopped completely, leading to unsustainablyhigh densities of organisms in remnantpatches, resulting in increases in mortality.

Habitat fragmentation may impede geneflow and lead to genetic isolation. Gene flowis critical to population viability, as it helpsmaintain local genetic variation and spreadspotentially adaptive genes. Genetic isolationcan be a mechanism for the creation of newpopulations and even species; however, smallremnant populations inhabiting fragmentedlandscapes are more likely to suffer frominbreeding and low genetic variation, whichcan increase vulnerability to other stressorsand lead to local extinctions. Retaining orrestoring connectivity counteracts these nega-tive effects of genetic isolation.

Landscape connectivity is essential acrosslarge areas (connectivity across ecoregions orcontinents is critical for some species) andover long timeframes (connectivity over many

years or generations) to allow species’ rangeshifts in response to long-term ecologicalchange. Projected climate change over thenext few decades will change ecosystem struc-ture, species composition, and diversity.Changes in biophysical conditions will likelylead to species replacement in communities(community turnover) and latitudinal and ele-vational shifts in geographic ranges. Duringepisodes of climate change since thePleistocene, vegetation zones or communitiesdid not move as a whole in response to cli-mate shifts; rather, species responded individu-ally to climate change, according to their ownindividual and largely independent environ-mental tolerances, dispersal abilities, andresponses to biotic interactions. Current cli-mate change appears to be occurring substan-tially faster than in the pre-historical record,meaning that the ecological conditionsrequired by many species (their niches) maybe shifting faster than species can adapt.These pressures, caused by changes in climaticconditions encountered by species in theircurrent distributions, are compounded byhabitat loss and fragmentation. The resultingobstacles to migration may impede species’abilities to adapt to climate change, to such anextent that many species could be driven toextinction.

Effects of Connectivity onDisease and Biotic Invasions

The extent to which landscapes are connectedor fragmented may also affect the rate and pat-tern of disease spread and invasion by non-native species. Species introductions, which canradically alter ecosystems, include plant and ani-mal diseases as well as competitors and predatorsagainst which native communities may not haveevolved defenses. It is important for managers toconsider how changes in connectivity and frag-mentation may influence the spread of diseasesand invasive species, and to recognize that theseinfluences are not necessarily unidirectional, butrather depend on the characteristics of the par-ticular species and landscape in question.

Intact, well-connected landscapes can serveas conduits for many invasive species if theydisperse in similar ways to native species. Inother cases, processes that fragment habitatsfor native species may simultaneously provideconnections that can facilitate biotic inva-sions. For example, the recent Asian carpinvasion (including grass carp (Ctenopharyn-godon idella), silver carp (Hypophthalmichthys




molitrix), and bighead carp (H. nobilis)) in theMississippi River watershed, and their potentialspread into the Great Lakes, illustrate howhuman-constructed connections (canals) canboth fragment the terrestrial environment andprovide new corridors between aquatic systems.Similarly, while roads can fragment vegetatedhabitats, they can simultaneously serve as con-duits for some invasive species, such as cheat-grass (Bromus tectorum), yellow star thistle(Centaurea solstitialis), and other invasive speciesthat benefit from the openings created by roads.

Connectivity for pathogens and parasites islargely a function of host distribution and abun-dance. While disease persistence benefits fromincreased host connectivity, it does not neces-sarily follow that these conditions are optimizedonly in well-connected landscapes. Landscapedisturbance and fragmentation can increasehost abundance and alter host distribution, andthese changes can increase connectivity for

pathogens and parasites for which the hostsconstitute the true “landscape” across whichmovement occurs (Box 1). However, fragmen-tation may also lead to the isolation of smallerhost subpopulations, which then may becomemore susceptible to disease or invasions. Inother situations, isolation resulting from land-scape fragmentation may protect a populationfrom disease. For example, plague (Yersinia pestis)in Colorado prairie dog (Cynomys ludovicianus)populations was shown to be less prevalent inmore remote, isolated populations than in thosemore closely grouped together.

Measuring, Analyzing andDesigning LandscapeConnectivity

Measuring structural connectivity has increas-ingly become a routine objective of researchersand policy makers, as Geographic Information

Box 1. Habitat Fragmentation and Increased Disease Transmissivity

An important consequence of fragmentation inforested habitats is the loss of species diversity.Those species that thrive in fragmented habitatstend to be more generalist or opportunistic, orhave traits such as smaller home range require-ments and tolerance for higher densities.Fragmentation can actually increase connectiv-ity from the perspective of a disease-causingpathogen. Higher densities of hosts increaseopportunities for transmissivity, and the hostpopulation is the true “landscape” across whichpathogen movement occurs. This is the case forthe tick-transmitted bacterium (Borrelia burgdor-feri) that causes Lyme disease. Its host, thewhite-footed mouse (Peromyscus leucopus),has become increasingly common in small forestfragments (<2 ha) in New England, likely result-ing from its small home range requirementscombined with release from competitors andpredators in smaller forest patches. P. leucopusis the principal natural reservoir for Lyme dis-ease. Higher densities of ticks infested with B.burgdorferi are found in smaller forest fragments(Figure 1), which may result from higher densitiesof white-footed mouse in these smaller frag-ments, presenting more opportunities for ticks tofeed on the mice. Consequently, humans livingnear these small forest fragments may have ahigher risk of exposure to Lyme disease relativeto those near larger forest fragments.

Figure 1. Relationship between measures of Lyme disease risk and forest patch area in a frag-mented landscape in New York state. a) Density of nymphal ticks is higher in smaller forestfragments. b) Percentage of nymphal ticks infected with the bacterium Borrelia burgdorferi is higherin smaller forest fragments. c) Density of nymphal ticks infected with the B. burgdorferi is higher insmaller forest fragments. (Source: Allan, B.F., F. Keesing, and R.S. Ostfeld. 2003. Effect of forestfragmentation on Lyme disease risk. Conservation Biology. 17: 267–272). Image used withpermission of John Wiley and Sons.

(a)

(b)

(c)



System (GIS) and remote sensing toolsbecome more widely available, affordable, andscalable. However, measuring functional con-nectivity using the movements of individualorganisms can be logistically complicated.Even the largest studies using the most appro-priate technologies can track only relativelyfew individuals over modest time periods, andcontrolled experiments addressing movementsand dispersal at relevant scales are extremelydifficult to implement. One way to addressthis difficulty is to measure gene flow, whichmay more accurately and efficiently reflectfunctional connectivity across large landscapes.Genetic studies avoid the logistic and financialcosts of tracking individual animals and inte-grate only those movements that producemeaningful population impacts – dispersalsthat result in breeding or emigration. A short-coming of this approach is that current geneticpatterns may not reflect the impact of currentlandscape features, especially for species withlarge population sizes or long generation times,or species affected by unobserved events, suchas genetic bottlenecks caused by past epidemicsor human persecution. In addition, geneticconnectivity may be masked in some instancesby local adaptation, which can drive geneticdistinctiveness even in a well-connected land-scape, by selecting for particular characteristicsof the local environment.

A common product of connectivity analysisis a map of predicted core areas, linkage zones,or barriers. Such maps often become the basisfor management actions. Several tools can beused to map these features, and each hasunique strengths and weaknesses. All of theapproaches described in the next sectiondepend on accurately defining landscaperesistance (an indication of how well a land-scape can be traversed by a given species), achallenging task when only a limited amountof information about species habitat prefer-ences is available. Furthermore, connectivitymodels can be difficult to validate.

Several research teams are working todevelop methods to rigorously estimatespecies-specific resistance from data on geneflow, genetic distances, habitat use, andmovement paths. Simple estimates of resis-tance, based on the extent to which land-scapes are impacted by roads, loss of naturalland cover, increased edge effects, spread ofinvasive species, and other direct humanimpacts measures may be useful for some gen-eralist species, but are insufficient for address-ing species-specific movements and habitat

needs. Thus, recent developments in connec-tivity modeling combine a structural land-scape approach, identifying both the potentialfor and obstacles to long-term habitat shifts,with a functional approach that highlightsthe specific connectivity needs of species withrestricted habitat requirements.

Modeling Approaches forIdentifying and QuantifyingLandscape Connectivity

We describe five widely-used analyticalapproaches, all implemented in a GIS envi-ronment, to assist planners in mapping andprioritizing landscape connections. Eachapproach has specific data requirements thatoften require input from biologists to helpdefine model parameters. In addition, eachapproach is designed to meet different objec-tives and will, therefore, produce different out-comes.

Least-cost analysis identifies the leastcostly route that an animal can take from onearea to another. The method assumes that theanimal incurs a cost as it moves over an area,where “cost” may reflect the actual energyexpended to move over the area, mortality risk,or impact on future reproductive potential. Inpractice, cost is usually estimated simply as theinverse of habitat suitability. Habitats that theanimal favors are assigned low cost whileunsuitable habitats are assigned high cost.

The least-cost path is the contiguous collec-tion of cells that has the lowest cumulativecost as the path crosses from one endpoint(such as a park, natural area, or known popu-lation; sometimes referred to as a node orpatch) to the other endpoint. Computersusing GIS software can easily identify thispath. Because the least-cost path is only onecell wide (for example, the center panel inFigure 2), it is often not a realistic area to pro-pose for conservation. Therefore, analysts usu-ally identify the least-cost corridor (shown inred in the panel on the right in Figure 2),which is a swath of cells expected to provide alow-cost route for movement.

Increased distance between two nodes orpatches also results in higher costs. This latterassumption is important, in that some speciesmay be able to identify and take advantage ofshorter linkages, while others operate at a finerscale of perception and therefore may not beable to consider total corridor length. Correctlyassigning these cost values (also referred to as



resistance or permeability values) is the mostproblematic aspect of least-cost analysis and theother approaches described here.

Factorial least-cost paths address onemajor limitation of traditional least-cost pathand least-cost corridor analyses in that theyare limited to predictions of connectivitybetween single sources and single destinations.While this may be ideal in the case where oneis interested in the lowest cost routes betweentwo focal conservation areas, many situationsrequire a more comprehensive analysis of con-nectivity. For example, corridor connectivitymay need to be calculated between thousandsof sources and a single destination, or betweenhundreds of sources and hundreds of destina-tions distributed across a complex landscape.A factorial implementation of least-cost pathsintegrates a vast number of paths to show anetwork of connectivity across large and com-plex landscapes, such as a factorial least-costpath analysis among hundreds of points acrossa resistance surface (Figure 3). Densities ofpaths are shown in a gradient from yellow tored, with red paths representing routes that arepredicted to contain the least-cost pathsbetween many pairs of source and destinationpoints. Additionally, while factorialapproaches are most common among least-cost approaches, they can also be integratedinto graph and circuit analysis as well.

Circuit theory treats the landscape as if itwere a large electrical circuit, in which allcells in the landscape can support movement.An important distinction between this andother methods is that while circuit approachescan be used to delineate corridors (with addi-tional processing), they are most useful in ana-lyzing and describing how well connectedsource and destination habitat patches maybe, given multiple movement pathways. Well-connected habitat patches have wide, contin-uous habitat between them, while paths

between poorly-connected habitat patchesmight have constrictions and bottlenecks,each of which can be identified using a circuitbased approach.

Current maps (Figure 4) can be a usefulway to visualize a circuit-theoretic analysis.Current strength reflects the predicted proba-bility of movement between the two pointsor habitat patches. Current maps can be diffi-cult to interpret: higher current (usuallydepicted in yellow) may occur in a cellbecause resistance is low, because most pathsare forced through that area because of highresistance elsewhere, or because the analysisis spatially constrained (Figure 4a). Lowercurrent (usually depicted in blue), in turn,may imply either high resistance in theunderlying layer, or simply that there aremany equally good alternate paths for move-ment. In this way, circuit models may moreaccurately approximate how organisms movethrough real landscapes. Despite their rela-tive complexity, these maps are useful forevaluating connectivity and identifying con-strained areas (bottlenecks) for possible con-servation action. In Figure 4a, where the

Figure 2. A least cost pathanalysis.

Figure 3. A factorial least-costpath analysis, evaluating least-cost paths (lines in blue to red,with red showing paths with thelowest costs) among manysource areas (green points).



model was confined to a particular habitatcorridor, high current values clearly indicatea bottleneck where movement is funneledinto a narrow space. In Figure 4b, where theanalysis was conducted in a different land-scape and not constrained to a corridor, fewmajor bottlenecks are apparent but someareas have higher current due to lower resis-tance or proximity to the nearest edges of thehabitat patches, which are shown in beige.

Graph theory combined with least-costmodeling or circuit theory provides several

useful enhancements to landscape connectiv-ity assessment and modeling. The landscapeitself can be likened to an interlaced web ornetwork that is composed of habitat patches(graph theory modeling uses centers, or“nodes,” of these patches as points of connec-tion) and the connections between thesepatches (the linear representations of whichare described in graph theory language as“edges”) (Figure 5). Once identified, nodes andedges can be prioritized based on their overallcontribution to the landscape network, forinstance by evaluating how many potentialconnections rely on each node or edge. Thisapproach allows for multiple least-cost path-ways to be evaluated for their contribution tothe configuration of the overall network. Thisapproach is particularly useful when modelingconnectivity between large reserve sets (assem-blages of patches, parks, or protected areas),identifying isolated reserve sets within the con-text of the modeled landscape, evaluating therobustness of multiple connections within thelandscape network, node/connection prioritiza-tion, and evaluating the consequences of los-ing nodes due to competing factors such asdevelopment pressure or fiscal constraints.

The resistant kernel approach to con-nectivity modeling is based on least-cost dis-persal from a defined set of sources. The modelcalculates the expected relative density of dis-persing individuals in each cell around thesource, given the dispersal ability of thespecies, the nature of the dispersal function,and the resistance of the landscape. Once theexpected density around each source cell (thesmallest unit of space that is modeled contain-ing individuals dispersing to other parts, orcells, in the landscape) is calculated, the ker-nels surrounding all sources are summed togive the total expected density at each cell.The results of the model are surfaces ofexpected density of dispersing organisms at anylocation in the landscape, in contrast to thephysical delineation of linkages or corridors.

The resistant kernel approach has a numberof advantages for assessing population connec-tivity. First, unlike most corridor predictionefforts, but similar to circuit-based approaches,it is spatially comprehensive, provides predic-tion and mapping of expected migration ratesfor every cell in the study area extent, and cando so for large geographic extents rather thanonly for a few selected linkage zones. Second,this approach allows assessment of how specieswith different movement patterns and disper-

Figure 4. Current maps,illustrating a) a landscape where

analysis was confined to acorridor and b) a different

landscape where the analysiswas not confined to a corridor.

Figure 5. A graph theory modeldepicting connectivity between

habitat patches.

(a)

(b)



sal abilities are affected by a range of land-scape change and fragmentation scenarios.This approach is useful for characterizing con-nectivity across continuous surfaces but doesnot identify individual linkages or corridorswithout additional analyses.

Figure 6 shows an individual resistant kernelaround a) a single source cell and b) thecumulative resistant kernel surface createdfrom summing all individual kernels for allhabitat cells. Red areas are predicted to havehigh frequency of occupancy by dispersers,while blue areas are predicted to experiencelow rates of dispersal. Case Study 2 providesan example of the use of the resistant kernelapproach in evaluating fragmenting effects ofroads on amphibian populations.

Considering Resolution andFocus in Connectivity Design

The resolution at which connectivity is ecolog-ically meaningful varies enormously, dependingon the species in question. For example, con-sider the scale of connectivity relevant to a bee-tle versus a bison. In practice, we tend to designand plan for connectivity at a human scale,meaning that we visualize connectivity in termsof landscape management units in a policyframework. Spatially extensive maps (thou-sands of kilometers), with coarse grained resolu-tion (for example, in the hundreds of metersper pixel or measurement unit) can depict anetwork of numerous habitat blocks and theconnections among them. Such maps mayserve as decision-support tools for managers, orprovide a high-level vision of landscape con-nectivity. They may be used to alert decision-makers to potential threats to large-scale con-nectivity as well as conservation opportunities.However, these types of maps are often too

coarse to inform specific conservation actionplans. Examples include the Yellowstone toYukon initiative, Arizona Wildlife LinkageAssessment, California Essential HabitatConnectivity, Two Countries-One Forest (CaseStudy 3), Washington Connected Landscapes,and the Bhutan Biological Corridor Complex.The two largest challenges for coarse-grainedanalysis and mapping are the identification anddelineation of core habitat blocks (areas whoseconservation value derives from the species andecological processes within them) and deter-mining which pairs or sets of blocks can feasiblybe connected in a way that promotes functionalconnectivity and meets conservation goals.Once habitat blocks have been identified, vari-ous techniques may be used to map the connec-tions (or linkages) among them, includingleast-cost path analysis, graph theory, or indi-vidual-based movement models.

Finer-grained linkage designs can guide site-specific actions to conserve connectivitybetween specific habitat areas that are rele-vant to the distances and ways in whichspecies of interest move across the landscape.To develop maps for these plans, landscapeconnectivity planners typically select a suite offocal species and use the union of their corri-dors or movement pathways (usually producedby least-cost modeling) to serve as a prelimi-nary linkage design for the entire biota. Forinstance, each of the 27 linkage plans inCalifornia (South Coast Missing Linkages pro-ject, available at www.scwildlands.org) andArizona (the Arizona Missing Linkages pro-ject, www.corridordesign.org) was designed tomeet the needs of several focal species includ-ing mammals, reptiles, amphibians, plants,and invertebrates. Focal species included area-sensitive species, species with short or habitat-restricted dispersal movements, and species

Figure 6. Resistance kernelmodeling: a) single-kernelanalysis and b) the cumulativeresistance surface of all kernelsacross the landscape.

(a) (b)



reluctant to traverse barriers in the planningarea. Large carnivores are commonly used asfocal species because they can require large areasof habitat, but many are habitat generalistswhose ecological requirements may not suffi-ciently encompass other focal species. To createa final multi-species linkage design, this unionof pathways is expanded to 1) include patcheslarge enough to support successful breeding inspecies for which corridors could not be mod-eled, 2) minimize edge effects, 3) provide suffi-cient space for animals and plants that requiremultiple generations to achieve gene flow, and

4) include physical and biological elements thatmay help support ecological processes that aremore complex or operating at scales that differfrom those of animal movements.

A complementary approach to restoring orretaining landscape connectivity based on theperceived requirements of focal species isbased instead on the abiotic drivers of landcover and species distributions. This approachis grounded in the ecological concept that bio-diversity at any point in time is determined bythe interaction of the recent species pool withclimate, soils, and topography. For example,

Case Study 2. Effects of Habitat Fragmentation on Vernal Pool Amphibian Populations

The resistant kernel approach for modeling connectivity is well suited to assessing how species with different dispersal abilities will beaffected by landscape change and fragmentation. Cushman and colleagues used the resistant kernel modeling approach to evaluate theeffect of habitat fragmentation by roads and residential development on a broad range of hypothetical population sizes and dispersal abil-ities of vernal pool breeding amphibians in western Massachusetts. The analysis compared habitat connectivity among 100 combinationsof population size and dispersal ability, across three scenarios. The scenarios included a null scenario, in which the landscape is uniformlysuitable for movement, and two scenarios of landscape fragmentation. The fragmentation scenarios included the effects of roads andeffects of roads and land use combined.

The amount of habitat that was predicted to be occupied in the null scenario was strongly related to population size and dispersal abil-ity. Figure 7 shows cumulative resistant kernel surfaces for a small portion of the study area for one combination of dispersal ability andpopulation size for (a) the null scenario, (b) the roads scenario and (c) the roads and land use scenario. Areas in red are predicted to havehigh densities of dispersing individuals, while dark blue areas are predicted to have very low occupancy rates. The amount of habitat pre-dicted to be connected by dispersal in the roads scenario decreased dramatically, with most simulated populations experiencing at leasta 75% reduction in connectivity compared to the null scenario. Somewhat counter-intuitively, the largest reductions in the extent of con-nected habitat were for species with relatively large dispersal abilities; this finding is consistent with a number of empirical studies thathave examined the effects of fragmentation on species with varying dispersal abilities. With the combined effects of roads and residen-tial/urban development, the proportion of the simulated populations predicted to experience over 85% reduction in habitat connectivityincreased from less than 10% to nearly 50%.

These results suggest that past road building and land use change may have had profound effects on the population connectivity ofsome vernal pool breeding species. This modeling exercise highlights the importance of landscape-level studies that explicitly includespecies-specific movement, abundance parameters, and the spatial patterns of the environment in a representation relevant to theorganisms in question.

Figure 7. Cumulative resistant kernel surfaces for amphibian dispersal under a) a scenario in which the landscape is assumed to beuniformly suitable for movement, b) a scenario in which roads are resistant to movement, and c) a scenario where roads and differentland-cover types (forest, agriculture, urban, residential, for example) are differentially resistant to movement. In all three scenarios,areas with the highest densities of organisms are in red, and those with lowest densities are in blue. The x and y axes representlongitude and latitude, and the z axis (height) represents expected density of dispersing individuals in each cell. The figure showsdifferences in densities of dispersing individuals across a 16 km2 area under the three different scenarios described above.



Case Study 3. Northern Appalachian/Acadian Ecoregion-Scale Connectivity Assessment

The Northern Appalachian/Acadian ecoregion spans 330,000 square kilometers across four states within the U.S. and all or part of fourprovinces in Canada, and contains large expanses of wilderness within close proximity to large human populations. As development withinthe region continues, and the demand on forest resources continues to increase, the ecoregion faces the very real threat of large-scalelandscape fragmentation. Two Countries, One Forest (2C1Forest) is a highly collaborative international consortium of 50 conservationorganizations, researchers and foundations dedicated to using landscape conservation to protect and maintain the forests and natural her-itage of the ecoregion. To date, initiatives have been undertaken to inventory natural resources, evaluate human impact, project futuregrowth, and identify priority locations for conservation action. Recent efforts have focused on identifying priority linkages among key port-folio conservation areas within the ecoregion.

As part of these efforts, Perkl and colleagues developed and evaluated landscape networks connecting target habitat areas arisingfrom four plausible conservation scenarios for the Northern Appalachian/Acadian ecoregion (Figure 8). A graph-theoretic approach wasused, applying the best available data on human settlement, access, land use change, and electrical power infrastructure, as a cost sur-face. Models indicated that while local connectivity was potentially retained at several sub-ecoregion scales, widespread ecoregionalconnectivity was not evident even in this extensive, forest-dominated region. Furthermore, the spatial dimensions of these modeled land-scape networks were staggering in scale and pose substantial challenges to implementation. Among the four scenarios, total networklengths ranged from 2,589 to 4,190 km, with total corridor areas ranging from 13 to 18 million hectares.

This analysis was a first pass at evaluating ecological connectivity for the region and among the first to assess and model connec-tivity at this scale. While this initial work was not intended to serve as a prescription for landscape or connectivity design, it has proveda valuable first step in developing plausible scenarios that can be further refined to test assumptions for large swaths of the ecoregionthat might serve a connectivity function.

Figure 8. Evaluation of landscape networks using four different scenarios to select habitat targeted for conservation: a) biodiversityconservation (n=95), b) Last of the Wild areas (an evaluation of human influence across ecosystems to identify remaining large wildplaces, n=120), c) modeled habitat patches for mature-forest focal species (n=105), and d) a composite viability scenario which wascomposed of selected sites from each of the previous scenarios (n=143).

(a) (b)

(c) (d)

N



least-cost path analysis can be used to identifycorridors that optimize continuity amongrecurring landscape units – areas with specifictopographic, bedrock and soil attributes.Similar to identifying corridors for individualspecies, landscape units can be included in alinkage design with multiple, broad corridorsthat are likely to facilitate the movements ofmultiple species. These areas will, theoreti-cally, support ecological and evolutionaryprocesses, including species’ range shifts inresponse to climate change, because they con-tain both the resources that species need, andthe means for individuals to move among mul-

tiple potentially appropriate habitat blocks tobuffer against the loss or degradation of habi-tat due to landscape change.

The challenges of moving from mapsand designs to implementation

Connectivity maps and linkage designs are use-ful for conservation only to the extent thatthey can support decision-making, guide man-agement, or otherwise be implemented. Ifmunicipalities, transportation agencies, andland management agencies are to integratethese designs into their own land use and plan-

Case Study 4. Evaluating Landscape Connectivity for Prioritizing Restoration Opportunities in theDelaware Estuary

Landscape connectivity analysis can be avaluable decision tool for prioritizingrestoration opportunities, helping to iden-tify areas that hold the greatest potentialfor increasing connectivity for focalspecies. In the Mid Atlantic region of theU.S., interest in restoration opportunitiesin the Delaware Estuary is increasing. Forscientists, planners, and the public, achief challenge has been to plan for theestuary in a manner that ensures that thevalue of regional restoration efforts ismaximized to improve habitat quality forand persistence of wildlife populations,against a backdrop of an increasinglyurbanized ecosystem.

To address this need, researchersmodeled landscape connectivity for sixcandidate restoration sites under con-sideration by the Delaware EstuaryRegional Restoration Work Group (DER-RWG). The approach integrated data onhome ranges, dispersal requirements,and habitat quality to evaluate suitablehabitat for black duck (Anas rubripes),least bittern (Ixobrychus exilis), andmarsh wren (Cistothorus palustris). Therelative connectivity value for each habi-tat patch was determined through thecalculation and comparison of three value parameters for each potential restoration site: (1) production, defined as the relative ability of apatch to contribute to overall recruitment as determined by local natality or mortality rates, which are influenced by patch area or habitatquality; (2) dispersal, the relative importance of a patch to the dispersal flux of individuals away from their natal patches or as part of ahome range, and (3) traversability, the relative importance of a patch as a stepping stone between isolated patches. Species-specific infor-mation was used to assign habitat suitability scores based on several characteristics, including dominant vegetation community, similarityto target species preferences for vegetative types, proximity to wetlands, and species habitat and area requirements.

The ability of the six restoration sites to provide high-quality habitat was species-specific: all restoration sites showed potential for provisionof high-quality habitat for black duck, three restoration sites showed particularly good potential habitat quality for marsh wren, and all six sitesshowed high traversability scores for this species. However, only two of the six sites met the dispersal requirements of the least bittern (St.Vincent’s and Pennypack Park, Figure 9), and these two sites were highlighted as potentially serving an important function for connecting iso-lated populations of least bitterns across the landscape. These landscape-scale measures of the species-specific ecological value of eachrestoration site were provided for consideration by the DERRWG in the evaluation of potential restoration sites across the Estuary.

Figure 9. Maps identifying habitat quality based on dispersal requirements of the LeastBittern, used in evaluation of candidate restoration sites in the Delaware Estuary. Habitat qualityscores are created by evaluating multiple, species-specific habitat characteristics, includinghow the patch contributes to overall population growth, the ability to provide opportunities fordispersal, and the extent to which the habitat patch may serve as a stepping stone to otherpatches (traversability).

Pennsylvania

New York

New JerseyMaryland

Habitat Quality Score

33

58

83

100

0 1 2 4 6 8 Kilometers

Pennypack ParkSaint Vincent’s

K&T Trail, Frankford LaunchLardner’s Point

Bridesburg



ning efforts, conservation professionals mustprovide specific guidance on how to use con-nectivity maps and designs in land use, zoning,transportation, and other types of plans.

Teams that engage stakeholders from theproject outset and remain focused on the pur-pose and need for connectivity analyses aremore likely to produce useful maps and data(for example, Case Studies 4 and 5). It isimportant to understand how organizationsinteract to achieve connectivity goals, particu-larly where different agendas may operate inthe same landscape. If groups operate indepen-dently, and the fundamental needs of eachorganization are widely divergent, simplyembracing the same long term objectives maynot be sufficient. Differences in organizationalneeds or agendas can be constrained by politi-cal or land ownership boundaries. Recognizingthese differences up front and explicitlyaddressing them in a connectivity strategy isvital to stakeholder agreement and projectsuccess. Finding agreement and compromiseamong stakeholder agendas helps in reducingconflict and the time spent on decision-mak-ing to pursue a conservation strategy, which inturn can reduce the overall costs and time-frame of implementation.

Recognizing and AddressingUncertainties

Connectivity modeling involves uncertaintiesand challenges, all of which should be directlyaddressed. All models are abstract and partialrepresentations of reality. However, if models ordata are wildly incorrect, modeled corridors andother inferences about landscape connectivity

may do more harm than good. Here, we discusspoints to consider when applying connectivitymodels to achieve conservation goals.

Uncertainty and error can be introducedinto the modeling process through a numberof avenues, including inaccurate or incom-plete characterization of species’ biology,incomplete theoretical basis for model devel-opment, inappropriate data extrapolation,uncertainty in future landscape change, andchanges in design goals during the planningprocess. Taken together, these sources of errorincrease uncertainty and signal the need fortargeted collection of empirical data or, at aminimum, indicate that greater caution mustbe exercised when crafting conservationactions from the models.

The following points provide a framework forexamining and addressing potential uncertain-ties in the context of connectivity planning.

Pay special attention to correct charac-terization of the biology of focal speciesin the ecosystem under analysis. Forexample, core habitat requirementsshould be as correctly and completelyidentified as possible. If habitat require-ments are not well understood and are notaccurately parameterized, even good mod-els of movement or gene flow will resultin erroneous linkage designs. If the corehabitat requirements of a given focalspecies are not well understood, it may bea good idea to substitute a different focalspecies whose biology is better under-stood, so that the model can be parame-terized with more confidence.

Case Study 5. Using Science to Evaluate Compromises During Implementation of a Linkage Design

In June 2009, the Arizona state wildlife agency released a detailed plan to conserve a wildlife corridor between the Tortolita Mountains and theSanta Catalina Mountains, just north of Tucson, Arizona. The plan (linkage design) included corridors for 9 focal species, as well as habitatmaps for several other focal species for which corridor models could not be created. In a proactive response, the local land-use planningagencies (Pima County, Town of Oro Valley, and the State Land Department) modified a proposed 14,000-acre urban development project sothat the project would conserve one of the three main corridors of the linkage design. Their modified development plan also proposed todestroy another corridor and defer the fate of the third.

The planners asked ecologists to assess how well their compromise plan would serve each of the nine focal species, compared to the full3 corridors in the optimal linkage design. Using newly developed corridor evaluation tools (available at www.corridordesign.org), the ecologistsprovided biologically meaningful comparisons of both designs. They concluded that the compromise was virtually as good as the optimum for7 of the 9 focal species. They quantified the degradation in utility for the other 2 species in terms of the gap lengths between patches of breed-ing habitat that each species would have to cross. The analysis allowed stakeholders and decision-makers to act with detailed appreciationof the ecological costs of compromise. The agencies proceeded with the compromise corridor plan. From a conservation perspective, theoutcome was a fully protected corridor nearly a mile wide for its entire length, an $8 million commitment to build one wildlife overpass wherea highway crosses the corridor, high confidence that the corridor is as good as possible for 7 species, and moderate degradation in corridorutility for the other 2 species. The cost of land acquisition was reduced by > $50 million, and the decision to proceed was made less than 6months after the start of deliberations. Although not everyone agreed, most conservation advocates, developers, and land-use planners feltthat a good compromise had been reached without prolonged litigation and clashes of expert opinion.



Understand the strengths and limitationsof the connectivity models. For example,does the model predict movement solelyfrom habitat-use data? If so, the model mayunderestimate long-distance movements ofindividual dispersers. Models based oninferences from movement behavior orpatterns of landscape genetics may providebetter predictions of connectivity.

Consider the effects of spatial and temporalextent in all of the analytical approachesdescribed above. Temporal aspects of eco-logical function may range from dailychanges in foraging habitat to seasonalmigrations, to inter-annual fluctuations inclimate, disturbance, and site productivity.Spatial considerations can range fromremoving specific barriers to local animalmovement, to facilitation of long-distancedispersal, to regional plans that considerregional connectivity. For example, individ-ual property owners may remove fences atlot lines. Neighborhood considerations mayinclude street design, such as reducing curbheight, which can be a significant barrier forsmaller, less mobile species such as turtles.Municipal connectivity initiatives mayincorporate greenway delineation as part ofthe comprehensive planning process. Stateand regional initiatives may incorporate sys-tem-wide processes, such as watershed plan-ning and ecoregional connectivity assess-ments. The spatial extent of connectivityplanning may even transcend nationalboundaries, as in the Yellowstone to Yukoninitiative linking wildlands in westernNorth America, or an ecoregional connec-tivity assessment in the Appalachian/Acadian region (Table 1 and Case Study 3).The challenge for connectivity managementis ensuring that multiple levels of effort com-plement each other in a coordinated way,and that the selected analytical method ormethods are appropriate relative to theintended management or conservationobjectives. Mismatches between the scale ofthe ecological processes of interest and thescale of analysis may result in failure to con-serve connectivity.

Be aware of uncertainties that emergewhen trying to extrapolate results outsideof the originally observed conditions.While some degree of extrapolation isinevitable, given that comprehensive infor-mation about the needs and responses of all

individuals of all species in every locationcould never be obtained, extrapolation toother locations or species should be basedon empirical results relevant to the speciesand landscape under consideration. Forexample, if the goal of connectivity model-ing is to understand the movements of arare temperate forest songbird, but a surro-gate species is identified for modeling dueto the constraints of available data, oneshould look first to another temperate for-est songbird whose taxonomy and ecologi-cal requirements are likely to be similar tothe target species. Extrapolation acrossscales, both temporal and spatial, is anadditional challenge. Knowing the physicalcharacteristics that influence patterns ofmovement across short distances or shortperiods of time may not be representative ofhow an organism makes decisions regardinglong-distance movements or movementsthat extend over periods of time measuredin seasons or generations.

Be aware that relationships among ecologi-cal and landscape variables may not belinear. Extrapolation requires an assump-tion about the nature of the relationshipamong factors outside the range of what isactually known. Because nonlinearities cantake almost any form, assumed relationshipsstand a great chance of being incorrect.

Try to account for anthropogenic land-scape change and the processes that driveit. Anthropogenic influences will continueto greatly affect many species’ habitats andtheir movements; modeling should incor-porate anthropogenic drivers of changeand how these drivers affect habitat andmovements.

Address the random variation that is inher-ent in many biological processes. It often ishelpful to present a range of potential out-comes or the likelihood of certain outcomes(such as population persistence) under dif-ferent assumptions about such variation,rather than a single “best” solution.

Regularly refer back to the stated goals ofthe analysis throughout the project dura-tion so that data inputs, assumptions, andmethods remain consistent with thesegoals. While researchers and practitionersalike may seek to produce a linkage mapthat captures the movement needs of all



wildlife species and can be applied to land-scape-scale and local planning efforts, onelinkage design cannot encompass all possi-ble connectivity goals and objectives at allplanning scales. For example, a linkagedesign that captures landscape-scale con-nectivity of natural habitats across largeecoregions (for example, Case Study 3) mayaddress broader or different conservationgoals than a design based on local-scale con-nectivity between pairs of core areas (forexample, Case Study 4). Similarly, an ecore-gional analysis may prioritize connectionsthat do not emerge in a continental-scaleanalysis. Through regular reference to thestated goals, the tendency for the focus toshift or the scope of a project to widen(“mission creep”) can be avoided.

Plan for increased connectivity and con-serve existing corridors to account forchanging landscape conditions and threats.Because many changes are not predictable,adaptive management is critical.Monitoring is essential for adaptive man-

agement, because monitoring helps man-agers to keep track of what is changing inreal time – such as the arrival of a newinvasive species in the landscape, orchanges in stream discharge that may affectfish habitat connectivity – and informsquick and appropriate responses. Managersand planners should be willing to expendresources on the adaptive managementprocess, including not only making deci-sions about priority linkages, but also fol-lowing through to monitor the status, func-tion, and trends of these linkages, evaluateand assess the observed changes, and adaptfuture connectivity planning accordingly.

In the long term, data gaps and areas of highuncertainty can be reduced with better dataand models or a stronger conceptual approach.At a minimum, practitioners should exercisecaution when crafting conservation actionsfrom connectivity models. Articulation of theseuncertainties does not, however, diminish theusefulness of these tools. Explicit acknowledg-ment of uncertainties allows results to be inter-

Table 1. Examples of connectivity maps and their utility.

Extent Project* Map Type Application

Individual South Coast Linkage design and specific conservation Over 25 partners have used the plan to helplinkage Missing Linkages plans for each of 11 key linkages in secure over 100,000 ha of land, modify

California’s south coast ecoregion. current land-use plans, and plan wildlife-friendly infrastructure.

State-wide Washington Coarse-grained analysis identifying broad Used to identify where highway mitigation Connected connectivity patterns for Washington and dollars can provide greatest wildlife benefitsLandscapes adjacent areas. Maps depict suitable habitat and to inform actions by state and federal

and linkages for 16 focal species comple- land management agencies, NGOs, and mented by an ecological integrity analysis. other parties.

Ecoregional Staying Connected Coarse-grained identification of critical Used by 21 public and private partners asin the Northern movement areas for wildlife spanning a framework to engage local stakeholders inAppalachians several states and provinces, especially in each area to further refine and identify

across the U.S./Canadian border. key areas of local connectivity and reduce Two Countries, risks of habitat fragmentation to wildlife One Forest movement.

National Wild LifeLines Coarse-grained analysis of potential wildlife Prioritizes a network of naturalness-baseddispersal pathways across the coterminous connections in terms of the contribution of 48 states. Uses multiple layers including land each pathway to the flow of connectivity cover, distance to roads and housing density across the network. Model favors pathwaysas proxies for habitat permeability and that avoid fragmented and human-modifiedpotential for wildlife movement. landscapes.

* For more information about each project:South Coast Missing Linkages http://scwildlands.orgWashington Connected Landscapes http://waconnected.orgStaying Connected in the Northern Appalachians http://www.conservationregistry.org/projects/3837Two Countries, One Forest http://www.2c1forest.org/en/mainpageenglish.htmlWild LifeLines http://www.wildlandsnetwork.org/what-we-do/scientific-approach/wild-lifelines

16 esa © The Ecological Society of America • [email protected]


landscape according to their influence on dis-persal between patches and the importance ofthose dispersal pathways to species persistence.However, the construction of complete modelscan require many years for a single species,making these methods infeasible for applica-tions to large numbers of species. An impor-tant way to address this constraint is todevelop spatially explicit, stochastic, demo-graphic meta-population models that can beparameterized for many species. Simulationmodels can be used to compare how the sub-traction of each patch and linkage in a com-plete network influences mean time to extinc-tion. Only through this type of persistencemodeling will it be possible to examine thetrade-offs between corridor conservation andaugmenting the size of existing protected areasfor long-term biodiversity conservation.Importantly, this approach can incorporateeconomic trade-offs, allowing prioritization ofhow conservation dollars are applied within anetwork of core areas and corridors.

2. Behavioral ecology

Current methods for identifying connectionsamong habitat areas and estimating the coststo organisms of moving through a landscapeare largely based on habitat suitability model-ing, which uses information on habitatrequirements or the extent to which speciesavoid human-altered landscapes, such as thebuilt environment. However, current informa-tion regarding species’ behavioral responses tohabitat modifications at the local and land-scape scale is inadequate. For example, under-standing how wildlife species are impacted byroads and how they possibly avoid them iscritical because roads are widespread andincreasing in density in many areas. Land usechanges may lead to behavioral responses thatdiffer widely even within taxa. For example,some monkey species adapt quickly to humanalteration of the landscape for agriculture,migrating through and even benefittingdirectly from agricultural crops by foraging inthem, while other species in the same areathat specialize on mature forests will not useand may actively avoid farmed areas. Otherpotential barriers, such as noise or light pollu-tion, may have profound effects on animalbehavior and functional landscape connectiv-ity, but are even less well studied and under-stood. Preliminary research suggests that evensupposedly unobtrusive forms of human activi-ties, such as low-impact recreation, can affect

preted honestly and can suggest future direc-tions for research and critical evaluation.Further, models can be useful even when under-lying data are sparse or weak, such as using pre-dictions to compare scenarios (which among aset of strategies is likely to lead to greater geneflow?) rather than stating absolute responses (ifa strategy is implemented, it will increase geneflow by a specific amount).

Future Directions forConnectivity Science

Although connectivity science has evolvedconsiderably in the last twenty years, muchstill needs to be learned in order to improvethe effectiveness of design and implementa-tion. Research on species persistence, behav-ioral ecology, and community structure isneeded to increase the accuracy and reduce theuncertainty associated with connectivity mod-els. Also, evaluating and testing connectivityresponses to climate change will be fundamen-tal to achieving conservation goals in the faceof the rapid changes that will confront manycommunities and ecosystems. And all of thesepotential areas of advancement will fall shortof conservation goals if the interdependence ofhumans with natural landscapes is not recog-nized and human activities are not effectivelyincorporated into connectivity planning.

1. Species persistence

One of the most important directions for con-nectivity science is to incorporate the likeli-hood of species or population persistence intoconservation plans. Persistence is a vital con-cept for evaluating what happens to species andcommunities when connectivity is lost andwhich connections are most important tomaintain. Most reserve design is based on staticmaps of species distributions, in an attempt tomaximize the potential number of species con-served across a reserve network. However, it isimportant to incorporate species persistenceinto conservation plans, particularly whenplanning for a network of reserves or habitatpatches in which populations of species may bedistributed as metapopulations – groups of sub-populations that are linked by some gene flow.

Persistence metrics, such as mean time toextinction or probability of extinction withina given timeframe, are necessary for determin-ing the relative merits of different connectiv-ity configurations. In theory, such methodscould also be used to prioritize linkages in the

animal behavior; for example, bird-watchingactivities have been shown to contribute tomarked declines in the detection rate of meso-carnivores in California. Equally important isthe continued study of local habitat characteris-tics that are required to maintain functionalhabitat connectivity. Some bat species, forexample, appear to be more sensitive to localhabitat features such as suitable roosting struc-tures and wet areas for foraging, than to changesin large-scale vegetation cover patterns.

The influence of many human activities onwildlife behavior and connectivity is still notwell understood. Habitat mapping and simula-tions are often less expensive than fieldworkand hence form the basis for much of the workdone in connectivity science to date; however,the need for more fieldwork to fully informsuch models cannot be overstated. Local fieldstudies are essential for improving connectiv-ity models and, ultimately, on-the-groundconservation outcomes.

3. Community structure

Habitat loss and fragmentation strongly influ-ence predator-prey and competitive interac-tions that help shape ecological communities,with variable impacts on individual species orpopulations. If habitats are reduced in size andconnectivity to the extent that apex predatorscannot persist, this process can lead to trophicdisruption or collapse, and many such trophiccascades have been documented in both ter-restrial and marine systems. For example, lossof connectivity has facilitated trophic col-lapse in some African landscapes: habitatfragmentation and loss has caused declines oflions and other apex predators, leading torapid increases in mesopredators, such asbaboons. These baboon populations are inturn able to sustain themselves by takingadvantage of human-altered landscapesthrough crop-raiding. Theoretical studies haveshown that interactions between competitorsare very sensitive to the structure of the land-scape. These studies also demonstrated thathabitat fragmentation can either stabilizeinterspecific competition, allowing for thecoexistence of similar species or, as fragmenta-tion increases, it may increase competition toa level where one of the species is eliminated.In these cases, coexistence or exclusion ofspecies results from a tradeoff between disper-sal rate and competitive ability, in relation tothe degree of habitat modification and frag-mentation. The key factors underlying each

case are the interactions of behavioral andlife-history characteristics of particular specieswith the scales and patterns of variability ofthe landscapes in which they live.

4. The challenges of climate change

Connectivity is one of the most commonlyadvocated strategies to help species adapt andsurvive rapid climate change. The idea is thatconnectivity may allow species to shift theirranges in response to changing climate, andthereby allow evolutionary and ecologicalprocesses to be sustained. However, connec-tivity designs based on current land cover pat-terns may not allow species to adapt to achanging climate and shifting ecologicalzones. During the next century of climatechange, habitats will not simply shift orshrink, but many will disappear as species shifttheir ranges, adapt, evolve, or go extinct inidiosyncratic ways.

Connectivity designs can incorporate theability to respond and adapt to climate changein several ways, although none have been rig-orously tested. One approach tracks how aspecies’ climatic envelope (suitable tempera-ture and moisture regime) moves across alandscape under several decades of simulatedclimate change. The predicted corridor is thechain of locations that were (during the simu-lation) contiguous for enough time to supportrange shifts, with new populations becomingestablished in locations that transition intothe envelope while other populations goextinct. Such models are conceptually soundbut depend on at least four other highly uncer-tain and often only partially predictive mod-els: 1) predictions of future carbon emissions,2) models of how the atmosphere and oceansrespond to these emissions, 3) climate enve-lope models for the focal species, and 4) dis-persal abilities of these species. In addition,the most problematic aspect of these models isthat, under climate change, novel types of cli-mates are expected to occur, and forecastinghow suitable these novel climates will be forexisting species cannot be reliably conducted.

A simpler alternative, which avoids theinherent uncertainties in a species-basedapproach, is to design linkages based on theexpected rates of climate change and the dis-tribution of climates across space and time.This approach examines different characteris-tics of climate (such as rate of change, diver-sity, and low temperatures) that are poten-tially influential for reserve network



resilience, based on the following assump-tions: 1) the advantages of connectivity aregreatest for areas that will experience fasterrates of change, 2) a reserve network that har-bors greater climatic diversity will allow forgreater adaptation, and 3) maintaining accessto cooler climates is a high priority. For exam-ple, if the distribution and representation ofclimates contained in a protected area isexpected to change quickly and dramatically,then species in those locations may have tomove, and corridors may be critical for theirsurvival. This approach prioritizes the mainte-nance of reserves with greater climate stabil-ity. Another equally reasonable assumption isthat a reserve network that harbors greaterclimatic diversity will provide refugia thatallow for adaptation, so targeting links thatadd climatic diversity to the network isanother possible approach.

An even simpler approach builds on the ideathat rivers and their associated valleys providea gentle and monotonic temperature gradientthat may allow species to shift their ranges bysequentially colonizing areas along that gradi-ent. Because such valleys will support riparianvegetation in nearly all climate regimes, theyshould be part of any connectivity map.

In some instances, short distance movementmay be all that is required for species to shifttheir range and persist, in which case addi-tions to existing protected areas may provemore effective and efficient than establishingcorridors. Hence, additional research isneeded to determine under what conditionsadding area to existing reserves is more effec-tive than adding linkages for increasing cli-matic diversity within a reserve network.Recent efforts by physical scientists to down-scale climate change models may allow ecolo-gists to explore these questions. Approachesusing landscape units, temperature gradients,and river valleys are coarse-filter approachesand, therefore, are unlikely to meet the needsof all terrestrial species, especially the mostextreme habitat specialists. However, even thebest fine-filter (species-based) approach willnot be able to serve the needs of species thatcannot shift their range fast enough inresponse to rapid climate change.

5. Putting people back on thelandscape

The interaction between land use and naturalsystems affects not only biodiversity but alsohuman livelihoods. Quantification of the

ecosystem service benefits that result fromwell-connected habitat, as compared to frag-mented landscapes, may help leverage publicand political support. In particular, hydrologicconnectivity can provide increased waterquality and aquatic species diversity uponwhich humans and other terrestrial speciesrely. Multidisciplinary teams can use surveys ofresource use and attitudes, and analysis ofpolitical structure to complement connectivitymaps and plans, supporting a holistic under-standing of the landscape and the organismsthat inhabit it.

Further knowledge about how connectivitybenefits ecological services can, in turn,improve the public’s understanding of theimportance of protecting connected wildlands.Ecosystem services, such as water and air qual-ity or pollination, provide additional valuebeyond species movement and persistence. Asconnectivity conservation continues tobecome more prevalent, planning efforts needto be directed beyond simply quantifying thefacilitation of species movement towardsquantifying these additional positive conserva-tion outcomes as well.

Conclusions

Landscape connectivity is of fundamentalimportance to the maintenance of populationsand species, as it enables organisms to moveamong habitat patches to access the ecologicalresources they need. In this Issue, we haveexplored the issues that conservation scientistsface in trying to evaluate, plan, and imple-ment habitat connectivity for biodiversityconservation, today and into the future. Westress the importance of protecting structuraland functional connectivity, to the extent thatfunction can be measured, at multiple scales.However, processes such as biological invasionalso highlight the complexity of understand-ing how connectivity influences the persis-tence of a given population or species.Increasing connectivity for one species canfacilitate the spread of invasive species anddisease, under some conditions, while frag-mented landscapes and high levels of distur-bance can also lead to similar results in othersituations. Various methods exist to measureconnectivity and identify priority linkages,and all point to the importance of scale foridentifying linkages that can be restored orconserved through practical on-the-groundmanagement. While increasing habitat con-nectivity remains a primary adaptation strat-


18 esa © The Ecological Society of America • [email protected]

egy for biological conservation, more researchis needed on how best to assess the effective-ness of corridors for facilitating migration andproviding overall reserve network resilience atthe necessary scales, given expected rates oflandscape change. Field studies that addresshow well linkages function for different species,and identify potential barriers to movement,are important to inform linkage design.

In the end, much remains to be learned abouthow configurations of reserves and linkages arelikely to influence species conservation, how thismay change under future climate scenarios, howspecies respond to site-level disturbances, andwhat management guidelines are most likely toprotect the connectivity functions that linkagesare designed to provide. Despite these data gapsand uncertainties, we believe that researchersand managers can successfully navigate many ofthe challenges in maintaining and restoringlandscape connectivity by using the guidance wehave outlined. First, approach connectivity atscales relevant to conservation targets. Second,incorporate flexibility and anticipatoryapproaches into connectivity planning throughthe use of tools including sensitivity analyses,uncertainty analyses, and adaptive management.Last, move forward on connectivity plans withstakeholder agreement and coordination,towards a common set of well-articulated goalsthat are scale-appropriate and account for majordrivers in landscape connectivity, includinganthropogenic influences and climate change.

For Further Reading

Baldwin, R.F., R.M. Perkl, S.C. Trombulak,and T. Burwell. 2010. Modeling Eco-regional Connectivity. In Landscape-scaleConservation Planning [Trombulak, S.C. andR.F. Baldwin (eds.)]. Springer Verlag,Chicago, IL, USA.

Beier, P., D.R. Majka, and W.D. Spencer. 2008.Forks in the road: choices in procedures fordesigning wildland linkages. ConservationBiology 22: 836-851.

Bennett, A.F. 2003. Linkages in the Landscape:The Role of Corridors and Connectivity inWildlife Conservation. Conserving ForestEcosystem Series No.1. IUCN Forest Con-servation Programme, Thanet Press Ltd,Marget, UK.

Collinge, S.K. 2009. Ecology of FragmentedLandscapes. Johns Hopkins University Press,Baltimore, MD, USA.

Compton B.W., K. McGarigal, S.A. Cushman,and L.R. Gamble. 2007. A resistant-kernelmodel of connectivity for amphibians that

breed in vernal pools. Conservation Biology21: 788-799.

CorridorDesign. 2011. GIS tools and informa-tion for designing wildlife corridors. On theweb at www.corridordesign.org.

Crooks, K.R. and M. Sanjayan (eds.). 2006.Connectivity Conservation. ConservationBiology Series No. 14. Cambridge Univer-sity Press, Cambridge, UK.

Cushman, S.A. 2006. Effects of habitat lossand fragmentation on amphibians: Areview and prospectus. Biological Conserva-tion 128: 231-240.

Dobson, A., K. Ralls, M. Foster, M. Soule, D.Simberloff, D. Doak, J. Estes, S. Mills, D.Mattson, R. Dirzo, H. Arita, S. Ryan, E.Norse, R. Noss, D. Johns. 1999. Corridors:Maintaining Flows in Fragmented Land-scapes. In Continental Conservation [Soule,M.E. and J. Terborgh (eds.).] The Wild-lands Project, Island Press, Washington,DC, USA.

Estes, J.A., J. Terborgh, J.S. Brashares, M.E.Power, J.Berger, W.J. Bond, S.R. Carpenter,T.E. Essington, R.D. Holt, J.B.C. Jackson,R. J. Marquis, L. Oksanen, T. Oksanen,R.T. Paine, E.K. Pikitch, W.J. Ripple, S.A.Sandin, M. Scheffer, T.W. Schoener, J.B.Shurin, A.R.E. Sinclair, M.E. Soulé, R.Virtanen, D.A. Wardle. 2011. Trophicdowngrading of planet earth. Science 333:301-306.

Gilbert-Norton, L., R. Wilson, J. R. Stevens,and K. H. Beard. 2010. A meta-analyticreview of corridor effectiveness. Conserva-tion Biology 24: 660-668.

Hilty, J.A., W.Z. Lidicker, and A.M.Merenlender. 2006. Corridor Ecology: Thescience and practice of linking landscapesfor biodiversity conservation. Island Press,Washington, DC, USA.

Opdam, P. and D. Wascher. 2004. Climatechange meets habitat fragmentation: link-ing landscape and biogeographical scalelevels in research and conservation.Biological Conservation 117: 285-297.

Acknowledgment

Funding for this project was provided by JointVenture Agreement 08-JV-11221633-248between the USDA Forest Service RockyMountain Research Station and theEcological Society of America.

About the Scientists

Deborah A. Rudnick, Integral ConsultingInc., 411 1st Avenue South, Suite 550,Seattle, WA 98104



Sadie J. Ryan, Department of Environmentaland Forest Biology, College of EnvironmentalScience and Forestry, State University of NewYork (SUNY-ESF), 149 Illick Hall, 1 ForestryDrive, Syracuse, NY 13210Paul Beier, School of Forestry, NorthernArizona University, Flagstaff, AZ 86011Samuel A. Cushman, USDA Forest Service,Rocky Mountain Research Station, 2500 S.Pine Knoll Drive, Flagstaff, AZ 86001Fred Dieffenbach, Appalachian NationalScenic Trail/Northeast Temperate Network,Marsh-Billings-Rockefeller NHP. U.S.National Park Service, 54 Elm Street,Woodstock, VT 05091Clinton W. Epps, Department of Fisheriesand Wildlife, Oregon State University, NashHall Room 104, Corvallis, OR 97331Leah R. Gerber, School of Life Sciences,Arizona State University, Box 874501, Tempe,AZ 85287Joel Hartter, Department of Geography,University of New Hampshire, 102Huddleston Hall, 73 Main Street, Durham,NH 03824Jeff S. Jenness, Jenness Enterprises GISAnalysis and Application Design, 3020 N.Schevene Blvd., Flagstaff, AZ 86004Julia Kintsch, ECO-resolutions LLC, 1149Downing Street, Denver, CO 80218Adina M. Merenlender, University ofCalifornia at Berkeley & Hopland Researchand Extension Center, 4070 University Road,Hopland, CA 95449Ryan M. Perkl, College of Architecture +Landscape Architecture, University ofArizona, 1040 N. Olive Road, Tucson, AZ85719Damian V. Preziosi, Integral Consulting Inc.,4D Bay Street, Berlin, MD 21811Stephen C. Trombulak, Biology andEnvironmental Studies, Middlebury College,Middlebury, VT 05753.

Layout

Bernie Taylor, Design and layout

About Issues in Ecology

Issues in Ecology uses commonly understood lan-guage to report the consensus of a panel of sci-entific experts on issues related to the environ-ment. The text for Issues in Ecology is reviewedfor technical content by external expert review-

ers, and all reports must be approved by theEditor-in-Chief before publication. This reportis a publication of the Ecological Society ofAmerica. No responsibility for the viewsexpressed by the authors in ESA publications isassumed by the editors or the publisher.

Editor-in-Chief

Jill S. Baron, U.S. Geological Survey andColorado State University,[email protected].

Advisory Board of Issues inEcology

Charlene D’Avanzo, Hampshire College Jessica Fox, Electric Power Research InstituteSerita Frey, University of New HampshireNoel Gurwick, Union of Concerned ScientistsKeri Holland, U.S. Department of StateRobert Jackson, Duke University Thomas Sisk, Northern Arizona University

Ex-Officio Advisors

Jayne Belnap, US Geological SurveyDeborah Goldberg, University of MichiganRich Pouyat, USDA Forest Service

ESA Staff

Clifford S. Duke, Director of Science ProgramsJennifer Riem, Science Programs Coordinator

Additional Copies

This report and all previous Issues in Ecologyare available electronically for free atwww.esa.org/issues.

Print copies may be ordered online or bycontacting ESA:

Ecological Society of America1990 M Street NW, Suite 700Washington, DC 20036(202) 833-8773, [email protected]



the role of landscape connectivity in planning and ... · pdf fileesa published by the...

Documents