new spatial connectivity moderates the effect of predatory fish on … · 2020. 8. 29. · spatial...

Spatial connectivity moderates the effect of predatory fishon salamander metapopulation dynamics

BRADLEY J. COSENTINO,1,� ROBERT L. SCHOOLEY,1,2 AND CHRISTOPHER A. PHILLIPS1,3

1Program in Ecology, Evolution, and Conservation Biology, University of Illinois, Urbana, Illinois 61801 USA2Department of Natural Resources and Environmental Sciences, University of Illinois, Urbana, Illinois 61801 USA

3Illinois Natural History Survey, University of Illinois, Champaign, Illinois 61820 USA

Citation: Cosentino, B. J., R. L. Schooley, and C. A. Phillips. 2011. Spatial connectivity moderates the effect of predatory

fish on salamander metapopulation dynamics. Ecosphere 2(8):art95. doi: 10.1890/ES11-00111.1

Abstract. In predator-prey metapopulations, persistence of prey in patches with predators may depend

on the rescue effect in which immigration from nearby sources prevents local extinction. Thus, constraints

on spatial connectivity may have important implications for predator-prey coexistence. We tested the

hypothesis that metapopulation dynamics of Ambystoma tigrinum (tiger salamander) depend on combined

effects of predatory fish and spatial connectivity. Because matrix heterogeneity can influence dispersal, we

also considered how a proximate constraint on amphibian dispersal—desiccation risk—scales up to

influence metapopulation dynamics for A. tigrinum. Occupancy and subsequent turnover patterns were

quantified in a network of 90 wetlands for three years in an agricultural landscape in Illinois. Our previous

field experiments demonstrated that desiccation risk varies among matrix habitats, and that individuals

orient movements towards habitat with low desiccation risk. We used cost-distance modeling to generate a

connectivity metric that accounted for desiccation risk. Occupancy and colonization probabilities were

related negatively to fish occupancy and positively to connectivity. Matrix structure had a strong influence

on colonization, and the connectivity metric based on desiccation risk was a better predictor of colonization

than alternative metrics. The positive effect of desiccation-informed connectivity on colonization was

strongest in wetlands with fish, indicating matrix composition can moderate the effects of predation on

amphibians. We detected a rescue effect in which extinction probability was related negatively to

connectivity, and this effect was strongest in sites with fish. The matrix did not have a strong effect on

occupancy or extinction probabilities, and we discuss why matrix effects may vary for different aspects of

population turnover. Our results suggest effects of fish predators on metapopulation dynamics of

amphibians depend on spatial connectivity, and that immigration may be essential for maintaining

persistence of amphibians in systems with fish. This study also demonstrates that the mechanisms

underlying dispersal limitation for A. tigrinum may include desiccation risk.

Key words: Ambystoma tigrinum; amphibian; connectivity; dispersal; fish; Illinois; landscape structure; least-cost;

metapopulation; predation; rescue effect; salamander.

Received 4 May 2011; revised 5 July 2011; accepted 21 July 2011; published 25 August 2011. Corresponding Editor: S.

Collinge.

Copyright: � 2011 Cosentino et al. This is an open-access article distributed under the terms of the Creative Commons

Attribution License, which permits restricted use, distribution, and reproduction in any medium, provided the original

author and sources are credited.

� E-mail: [email protected]

INTRODUCTION

Metapopulation theory predicts a balance

between stochastic extinction and subsequent

recolonization, underscoring the importance of

dispersal for regional persistence (Hanski and

Gaggiotti 2004). In systems in which patch area

and isolation are poor predictors of extinction

v www.esajournals.org 1 August 2011 v Volume 2(8) v Article 95

and colonization (Baguette 2004, Pellet et al.2007, Prugh et al. 2008), habitat heterogeneitymay be a primary factor driving observedturnover dynamics (With 2004, Schooley andBranch 2009, Cosentino et al. 2010). For example,extinction can be deterministic in patches withpoor habitat quality (Thomas 1994), and recolo-nization and immigration can depend on effectsof matrix heterogeneity on dispersal (Ricketts2001). For prey species with patchily distributedpredators, local habitat suitability, includingpredator presence, and landscape connectivityare likely important for predicting occupancyand turnover dynamics.

In predator-prey metapopulations, the patchyoccurrence of predators can create strong spatialvariation in survival and reproductive success forprey. Persistence of prey in patches with preda-tors can depend on the rescue effect (Brown andKodric-Brown 1977, Stacey et al. 1997) or source-sink dynamics (Pulliam 1988) in which immigra-tion from predator-free sources prevents localextinction (Amezcua and Holyoak 2000, Caudill2003, 2005, Woodford and McIntosh 2010; wefollow Stacey et al. 1997 and use the term ‘‘rescueeffect’’ as a general form of source-sink dynam-ics). Thus, the degree to which patches areconnected through dispersal is likely to affectpredator-prey coexistence. Prey populations inisolated sites should be more susceptible todeterministic extinction by predators than popu-lations in connected sites. Furthermore, therescue effect may be reinforced or inhibited incomplex landscapes depending on how matrixstructure facilitates or impedes dispersal (Croninand Haynes 2004).

Fish predation is an important source ofmortality for pond-breeding amphibians in per-manent or semi-permanent wetlands (Wellbornet al. 1996), and introduction of nonnative fish isa significant cause of amphibian declines (Katsand Ferrer 2003). Predatory fish consume theeggs and larvae of many pond-breeding amphib-ians, and fish negatively influence amphibianreproductive behavior, survival, abundance, andspecies richness (Sexton et al. 1994, Hecnar andM’Closkey 1997, Werner et al. 2007, Pope 2008).Although predatory fish can strongly limit thespatial distribution of amphibians (Pilliod et al.2010), the rescue hypothesis may explain thepersistence of amphibians in wetlands with fish

(Sjogren Gulve 1994, Pilliod and Peterson 2001).If dispersal moderates the risk of local extinctionat sites with fish, conservation efforts may benefitfrom a spatial perspective on fish-amphibianinteractions.

We collected three years of data on wetlandoccupancy and spatial turnover for Ambystomatigrinum tigrinum (eastern tiger salamander,Ambystomatidae) to address the hypothesis thatmetapopulation dynamics depend on the com-bined effects of predatory fish occupancy andconnectivity. Like many pond-breeding amphib-ians, A. tigrinum is susceptible to predatory fishand can be excluded from sites occupied by fish(Sexton and Phillips 1986). We investigated howconnectivity among breeding populations of A.tigrinum interacts with the spatial distribution ofpredators to influence occupancy, colonization,and extinction probabilities of A. tigrinum. Wepredicted fish occupancy would have negativeeffects on occupancy and colonization, and apositive effect on extinction. If immigrationcounteracts the deterministic effects of predators,then wetland connectivity should have positiveeffects on occupancy and colonization, and anegative effect on extinction for sites with fish.

Because the distribution and movement pat-terns of amphibians can depend on matrixheterogeneity (Joly et al. 2001, Stevens et al.2004, Mazerolle and Desrochers 2005, Ritten-house and Semlitsch 2006), we also evaluatedsupport for different models of connectivity.Desiccation risk has been identified as animportant cost of movement for juvenile dispers-ers due to their small bodies, permeable skin, andhigh surface-area-to-volume ratio compared toadults (Spight 1968, Rothermel and Semlitsch2002). In a field experiment on A. tigrinumconducted in our study area, desiccation ratesvaried significantly among dominant matrixhabitats. Desiccation was greatest in corn andgrassland and lowest in forest and soybean(Cosentino et al. 2011). A second field experimentdemonstrated salamanders consistently movedinto habitats that minimize desiccation risk whengiven a choice at habitat boundaries (Cosentinoet al. 2011). These results indicate desiccation forA. tigrinum varies among matrix habitats, andmovement decisions of individuals are influ-enced by desiccation risk. To evaluate conse-quences of this physiological constraint at the


COSENTINO ET AL.

population-level, we used cost-distance model-ing to examine how well a connectivity metricbased on habitat-specific desiccation risk ex-plained site occupancy and turnover comparedto connectivity metrics representing Euclideandistance and expert opinion.

METHODS

Study species and siteAmbystoma tigrinum is mainly subterranean,

using mammal burrows and self-excavated bur-rows in upland forests and grasslands for refugeduring most of the year. Breeding migrationsoccur between late fall and early spring, andadults breed in ponds from February to April.Juveniles emigrate from ponds between July andSeptember and can become sexually maturewithin two years (Petranka 1998). Consistentwith other pond-breeding amphibians (Semlitsch2008), interpond dispersal for this species is rarefor adults but more common for juveniles(Church et al. 2007).

Our study was conducted at a 9300-ha area innorthern Illinois centered at the RichardsonWildlife Foundation property (West Brooklyn,IL; 41842026.600 N, 89811025.000 W). The landscapeis dominated by row-crop agriculture (corn,soybean) and patches of grassland and forest(Fig. 1). Yearly crop rotation between corn andsoybean is common, but corn cover increasedwhile soybean cover declined between 1999 and2008 (B. J. Cosentino, unpublished data). In 2008,cover by corn was 50% and cover by soybeanwas 21%. Suitable breeding habitat for A.tigrinum in this landscape consisted of freshwaterwetlands with variable hydroperiods, whichrepresented only 0.8% of the landscape.

We documented the occupancy status of A.tigrinum in 90 wetlands (median area ¼ 0.56 ha;range¼0.01–5.29 ha) in 2007, 2008, and 2009. Themedian nearest-neighbor distance between wet-lands sampled in this study was 245 m (range ¼36–2830 m). Wetlands were identified using1:24000 National Wetland Inventory quadranglesand aerial photographs, and all temporary pools(i.e., those that held water only after heavy rains)were excluded from sampling. Wetland emergentvegetation was dominated by Alisma subcorda-tum, Eleocharis spp., Phalaris arundinacea, Polygo-num spp., Pontederia cordata, Sagittaria spp.,

Scirpus spp., and Typha spp. Other amphibiansencountered during surveys included Acris crep-itans, Anaxyrus americanus, Hyla chrysoscelis,Pseudacris crucifer, Lithobates catesbiana, L. clami-tans, and L. pipiens.

Wetland surveys: salamanders and predatory fishWe surveyed each wetland a single time in

sampling sessions of 1–4 consecutive daysbetween late May and early August in each year.Sites were surveyed for �3 consecutive days in97% of sampling sessions across seasons. Siteswere not sampled randomly due to logisticalconstraints. Instead, wetlands were grouped byspatial proximity, and we randomized the se-quence in which we surveyed groups in eachyear. Collapsible minnow traps (45.7 3 30.5 cmwith two 6.4-cm entrances; Promar, Gardena,California, USA) were used to detect the presenceor absence of A. tigrinum larvae at each wetland.We set 1–13 traps within 10 m of the shoreline,and sampling effort was standardized by wet-land area. Traps were checked within 24 hr ofdeployment.

We documented adult and juvenile fish at eachwetland using minnow traps and collapsiblehoop traps. The most common fish encounteredwere yellow bullhead (Ameiurus natalis), greensunfish (Lepomis cyanellus), and bluegill (L. macro-chirus). Ameiurus spp. and Lepomis spp. are bothdocumented predators of pond-breeding am-phibians (e.g., Sexton and Phillips 1986, Hecnarand M’Closkey 1997, Babbitt et al. 2003). We usedoccupancy models that account for imperfectdetection (MacKenzie et al. 2006) to estimate theprobability of fish predators occupying a wet-land. The program PRESENCE 3.1 was used tobuild single-season occupancy models (MacKen-zie et al. 2006) to estimate wetland occupancyprobability given a site’s detection history for fishin each year (Appendix). The conditional occu-pancy probability (wfish) for wetlands in whichfish were detected was 1. When fish were notdetected, conditional occupancy probabilitieswere less than one (0 � wfish , 1). Predatory fishoccurred in �31% of wetlands each year, and fishoccupancy was temporally dynamic (27 observedturnover events).

Wetland area and connectivityPatch area and connectivity metrics were


COSENTINO ET AL.

measured with ArcMap 9.3 (ESRI, Redlands,California, USA) after digitizing wetlands usingaerial photographs from 2007. To measureconnectivity of each wetland, we used a metricthat includes a negative exponential dispersalkernel and accounts for distances to potentialwetlands that function as sources of dispersers(Hanski 1994, Moilanen and Nieminen 2002). Theconnectivity (Ci ) of wetland i was defined as

Ci ¼X

j 6¼i

pjexpð�adijÞ ð1Þ

where pj is the probability of occupancy by A.tigrinum larvae at source wetland j, a is aparameter scaling the effect of distance ondispersal (1/a is the mean dispersal distance),and dij is the distance between target wetland iand source wetland j.

Observed colonizations were used to estimatethe mean dispersal distance (1/a) for A. tigrinum(Prugh 2009). The ability of three connectivitymetrics to explain occupancy dynamics was

assessed when a was set to represent the meandistance 6 1 SE (550 6 99 m) between colonizedwetlands and the nearest occupied source (451,550, and 649 m). A metric with a¼ 0.0022 (meandispersal distance¼451 m) was a better predictorof occupancy patterns than alternative metrics(Cosentino 2011), so we specified a as 0.0022 insubsequent analyses.

We calculated Ci by setting pj equal to 0 forsource wetlands in which A. tigrinum wasundetected in all three years, 0.33 for sourcewetlands occupied in one year, 0.67 for sourcewetlands occupied in two years, and 1 for sourcewetlands occupied in all three years.

Cost-distance modelingWe used cost-distance modeling to incorporate

matrix heterogeneity into our connectivity metric(Adriaensen et al. 2003). To account for variationin dispersal cost among matrix habitats, costsurfaces were created in ArcMap 9.3 by assigningmovement resistances to each habitat in the

Fig. 1. Naıve occupancy pattern of Ambystoma tigrinum among 90 wetlands surveyed for three years (2007–

2009) in northern Illinois. Wetlands were unoccupied by salamanders in all years (open circles), occupied in one

year (closed triangles), occupied in two years (closed squares), or occupied in all three years (closed circles).


COSENTINO ET AL.

landscape. Then, distances along paths betweensource and target wetlands that minimize costwere estimated (i.e., effective distances) usingPATHMATRIX in ArcView 3.2 (Ray 2005).Finally, effective distances were substituted forEuclidean distances (dij) in the formula for Ci togenerate a connectivity measure that accountedfor matrix heterogeneity.

We used USDA Cropland Data Layer Maps togenerate cost surfaces (hhttp://www.nass.usda.gov/research/Cropland/SARS1a.htmi). We col-lapsed land-cover types into seven categories:corn, soybean, other crop (mainly alfalfa andwinter wheat; ,1% of landscape), forest, grass-land, developed or road, and water. We devel-oped three resistance sets that representeddifferent hypotheses about how the landscapeinfluences movement. In all resistance sets, weassigned the value of 1.0 to water, assuming thatwetlands function as stepping stones for dispers-al. For set REUC (Euclidean), resistances were setas 1.0 for each habitat to represent a homoge-neous matrix. For set REXP (Expert Opinion), weassigned resistances to reflect expert opinion foragricultural landscapes in which the commonassumption is that all crops impose a uniformlyhigh cost on dispersal (e.g., Compton et al. 2007,Greenwald et al. 2009). Thus, we assigned anequally high resistance value of 500 to allagricultural crops, and forested habitats wereassigned a lower resistance value than grasslandhabitats (Fig. 2; forest¼ 100, grassland¼ 200; e.g.,Rothermel and Semlitsch 2002). For set RDES

(Desiccation), the relative rankings of movementresistances were directly informed by experimen-tal data on A. tigrinum desiccation risk andmovement behavior (Cosentino et al. 2011).Soybean and forested areas received equallylow resistances of 100, and corn and grasslandareas received equally high resistances of 500(Fig. 2). We assigned a resistance value of 500 tocrops other than corn and soybean and a value of1000 to roads and developed areas in all sets. Thethree resistance sets were used to generate threemeasures of connectivity: CEUC, CEXP, and CDES.Resistance values for each set are relative values(Adriaensen et al. 2003), and our results were notsensitive to changes in absolute cost values whenthe relative rankings remained the same (B. J.Cosentino, unpublished data).

Due to temporal variation in crop cover

associated with crop rotations, we created year-specific cost surfaces for sets REXP and RDES (Fig.2). When evaluating the effect of connectivity ondemographic parameters, we used connectivitymetrics generated from cost surfaces in the yearprevious to the parameter-year combinationbeing modeled. For example, the 2007 costsurfaces were used to generate connectivitymetrics to explain colonization patterns in 2008.That is, we assumed most dispersal contributingto spring colonization occurs during the previousgrowing season when agricultural crops are onthe landscape. However, we note that costsurfaces based on REXP exhibited minimal yearlyvariation because corn and soybean were as-signed equally high resistance values (Fig. 2). Ifmost dispersal contributing to colonization oc-curs in spring when agricultural landscapes arebarren, REXP may be a good representation ofdispersal cost because amphibians avoid movingin barren areas associated with high desiccation(e.g., Mazerolle and Desrochers 2005).

Data analysisData from repeated occupancy surveys within

each year were used in a multiple-seasonoccupancy model that accounts for imperfectdetection probability (q) to assess how metapop-ulation factors (i.e., area and connectivity) andfish occupancy influenced initial occupancyprobability (w2007), colonization probability (c),and extinction probability (e) for A. tigrinumlarvae. A repeated survey was defined as a groupof minnow traps within a wetland open for onenight. We considered using a two-species param-eterization to model the co-occurrence of fish andA. tigrinum, but that parameterization was notavailable in a multiple-season framework (Mac-Kenzie et al. 2006). We conducted all analyses inthe program PRESENCE 3.1 using a logit linkfunction to model effects of covariates on among-wetland variation for each rate parameter.Covariates did not exhibit strong multicollinear-ity (all r , 0.31). Initial analyses were conductedto select a model accounting for variation in qwhile holding w2007, c, and e constant. Potentialdetection covariates included survey day withinsampling session (1–4), Julian date (calendar dateof the first survey day), and year. We thenmodeled w2007, c, and e sequentially, starting withw2007 and ending with e. The most supported


COSENTINO ET AL.

model for each rate parameter was maintainedwhen modeling the remaining parameters. Whilewe could have considered all combinations ofmodel structures on each parameter in a singlecandidate set, modeling each parameter sepa-rately allowed us to greatly reduce the overallnumber of models, which minimizes the chanceof spurious results (Burnham and Anderson2002).

The Akaike Information Criterion corrected forsmall sample size (AICC) was used to rank thesupport of 31 candidate models representing theeffect of different combinations of metapopula-tion factors and fish occupancy on w2007, c, and e.The candidate set was the same for each rateparameter. The first model constrained theparameter of interest to be equal among sites

(constant). The next three models included theeffects of fish occupancy (F) and metapopulationfactors (A ¼ area, C ¼ connectivity) individually(F, A, C). The next four models included additiveeffects of metapopulation factors and fish occu-pancy (Aþ C, FþA, Fþ C, FþA þ C). Becausethe importance of connectivity may vary betweensites with and without fish, we then added a fish3 connectivity interaction effect (F3C) to modelsthat contained fish occupancy and connectivity(F þ C þ F 3 C, F þA þ C þ F 3 C). Finally, weadded an area 3 connectivity interaction effect(A3C) to models that contained wetland areaand connectivity (AþCþA3C, FþAþCþA3

C, F þ A þ C þ F 3 C þ A 3 C). Importantly,models including connectivity were constructedseparately using each metric: CEUC, CEXP, or

Fig. 2. Dispersal cost surfaces for Ambystoma tigrinum in a northern Illinois landscape for 2006, 2007, and 2008.

Movement resistances were either representative of expert opinion in agricultural landscapes (A–C) or based on

empirical data on desiccation risk and movement behavior (D–F).


COSENTINO ET AL.

CDES.Akaike weights (wi ) were estimated for each

model, and model-averaged point estimates ofw2007, c, and e were calculated for each wetlandusing all models from candidate sets specific toeach rate parameter (Burnham and Anderson2002). To evaluate the hypothesis that desiccationrisk scales up to influence A. tigrinum occupancydynamics, we compared the relative support ofCEUC, CEXP, and CDES in explaining w2007, c, ande. Each connectivity metric was in nine modelsfor each candidate set, and Akaike weights weresummed across models within each set toevaluate the relative importance of each connec-tivity metric (Burnham and Anderson 2002).

RESULTS

Ambystoma tigrinum was detected at 29 wet-lands (naıve occupancy ¼ 0.32) in 2007, 43wetlands (0.48) in 2008, and 45 wetlands (0.50)in 2009. Using a constant model of w in single-season occupancy models, yearly occupancyprobabilities after accounting for imperfect de-tection were 0.42 (SE ¼ 0.06) in 2007, 0.55 (SE ¼0.06) in 2008, and 0.51 (SE¼ 0.05) in 2009. Of the90 sites, 35 sites (38.9%) were never occupied, 12sites (13.3%) were occupied once, 23 sites (25.6%)were occupied twice, and 20 sites (22.2%) wereoccupied during all three years (Fig. 1). Turnoverwas common in both transition periods. Ob-served local colonizations outnumbered localextinctions between 2007 and 2008 (NC ¼ 20, NE

¼ 6) and between 2008 and 2009 (NC¼ 9, NE¼ 7).

Multiseason occupancy and turnover modelsA constant model of q indicated the daily

detection probability for A. tigrinum was gener-ally high (q ¼ 0.82, SE ¼ 0.02). The mostsupported model of detection probability in ourmultiseason models included effects of Juliandate in 2007 and 2008, and survey day in 2009(accounting for three parameters). Specifically,Julian date had a negative effect on detectionprobability in 2007 and 2008. In 2009, detectionwas greatest on the second day of each samplingsession.

The top model of initial occupancy probabilityfor salamanders included additive effects of fishoccupancy, wetland area, and Euclidean connec-tivity (Table 1). Initial occupancy was related

negatively to fish occupancy and positively towetland area and connectivity (beta estimate 6 1SE for the top model, F¼�1.63 6 0.43, A¼0.78 6

0.36, CEUC ¼ 0.67 6 0.40). A competing model(DAICC � 2) included only effects of fish andwetland area (Table 1).

The top model of colonization probabilityincluded effects of fish occupancy and connec-tivity accounting for desiccation risk (CDES; Table1). In general, colonization probability wasgreatest in fishless sites and lowest in sites withfish, and colonization increased with connectiv-ity (Fig. 3A–B; beta estimate 6 1 SE for the topmodel without interaction effects, F ¼ �1.30 6

0.37, CDES ¼ 0.76 6 0.43). However, connectivityinteracted with both fish occupancy and wetlandarea (Table 1). Connectivity interacted positivelywith fish occupancy (beta estimate 6 1 SE, F 3

CDES¼ 3.90 6 1.95) indicating the positive effectof connectivity on colonization was strongest inwetlands with fish. Connectivity interacted neg-atively with wetland area (beta estimate 6 1 SE,A 3 CDES¼�4.14 6 1.80) indicating the positiveeffect of connectivity on colonization was stron-gest in small wetlands. Colonization probabilityalso increased with wetland area in fishless siteswith low to moderate connectivity (Fig. 3A).

The most supported model of extinctionprobability included additive effects of fish andconnectivity based on Euclidean distance (Table1). Local extinctions were most common inwetlands with fish and low connectivity (Fig. 4;beta estimate 6 1 SE for the top model, F¼0.80 6

0.35, CEUC ¼�0.84 6 0.44). There was marginalsupport for a positive interaction between fishoccupancy and connectivity (DAICC ¼ 1.04 formodel F þ CEUC þ F 3 CEUC), suggesting theeffect of connectivity on extinction probabilitywas strongest in wetlands with fish.

Cost-distance connectivity metrics: summed Akaikeweights

Given model selection uncertainty (Table 1),the summed Akaike weights were the mosteffective way to assess relative support for thethree connectivity metrics. Cost-distance modelsaccounting for matrix heterogeneity (CEXP, CDES)did not have more support than Euclideanconnectivity (CEUC) for predicting initial occu-pancy and extinction probabilities (Fig. 5). Incontrast, a connectivity metric accounting for


COSENTINO ET AL.

upland desiccation risk was more supported than

Euclidean and expert opinion models for pre-

dicting colonization probability (Fig. 5).

DISCUSSION

Predatory fish strongly limited the spatial

distribution of A. tigrinum, which is consistent

with previous studies on fish-amphibian interac-

tions (Hecnar and M’Closkey 1997, Pilliod and

Peterson 2001, Pilliod et al. 2010). By explicitly

focusing on metapopulation processes, our work

illustrates how distributional constraints result

from high extinction risk and low colonization

success at sites with fish predators. However,

these deterministic effects of fish on amphibians

were moderated by spatial connectivity in two

ways: (1) extinction probability was related

negatively to connectivity, suggesting a rescue

effect in which immigration prevents local

Table 1. Model selection statistics for initial occupancy, colonization, and extinction probabilities of Ambystoma

tigrinum from 90 wetlands in northern Illinois.

Rate Parameter Model DAICC xi �2l K

Occupancy2007 F þ A þ CEUC 0.00 0.19 634.51 9F þ A þ CDES 0.95 0.12 635.46 9F þ A þ CEXP 0.96 0.12 635.47 9F þ A 1.31 0.10 637.96 8

Colonization2008,2009 F þ CDES þ F 3 CDES þ A þ A 3 CDES 0.00 0.45 602.61 14F þ CEXP þ F 3 CEXP þ A þ A 3 CEXP 0.79 0.30 603.40 14

Extinction2008,2009 F þ CEUC 0.00 0.10 590.95 16F þ CEUC þ F 3 CEUC þ A 0.18 0.09 586.55 18F þ CEXP 0.21 0.09 591.16 16F þ CEUC þ A 0.64 0.07 589.31 17F þ CDES 0.77 0.07 591.72 16F þ CEXP þ A 0.92 0.06 589.59 17

Notes: Years are indicated as subscripts for each rate parameter. Main effects include fish occupancy probability (F), wetlandarea (A), Euclidean connectivity (CEUC), expert opinion connectivity (CEXP), and desiccation risk connectivity (CDES). Summaryincludes relative difference between model AICC and AICC for the best model (DAICC), Akaike weights (xi ), twice the negativelog-likelihood (�2l ), and number of parameters (K). Effects of Julian date and survey day on detection probability wereincluded in all models (3 parameters). Only models with DAICC � 2 are presented for occupancy and colonization and DAICC

� 1 for extinction.

Fig. 3. Relationships of predicted colonization probability of Ambystoma tigrinum to desiccation-informed

connectivity and wetland area in sites (A) without and (B) with fish predators.


COSENTINO ET AL.

extinction, and (2) colonization probability wasrelated positively to connectivity at sites withfish, indicating a substantial number of immi-grants can overcome fish predation duringcolonization. Furthermore, we found that desic-cation risk for A. tigrinum dispersers scales up toinfluence colonization probability, thereby link-ing population-level isolation effects to a physi-ological constraint on dispersal. Overall, theseresults support the view that predator-preyinteractions and resulting patterns of communitystructure can be mediated by dispersal andlandscape context in spatially structured systems(Urban 2004, Richter-Boix et al. 2007).

Fish predators and spatial connectivityIf fish typically have strong deterministic

effects on pond-breeding amphibians, whatexplains the pattern of co-occurrence betweenA. tigrinum and fish predators in highly connect-ed wetlands? One hypothesis is that source-sinkdynamics could generate this pattern (Pulliam1988). In connected networks, persistence of A.tigrinum at sites with fish may depend oncontinual dispersal from nearby sources withoutfish (e.g., Caudill 2003, 2005). If resources arelimited in fishless sites because of large popula-tion sizes, intense competition for resources may

force some individuals to colonize low-qualitysites with fish.

Although source-sink dynamics may be gener-ated by active dispersal and habitat selection(Pulliam 1988), optimal habitat selection isunlikely for A. tigrinum. Adults are generallyphilopatric to breeding ponds (Church et al.2007), and ambystomatid juveniles have limitedperceptual range (e.g., ,50 m; Rothermel 2004)and are not known to exhibit sophisticatedsearching strategies. In addition, some ambysto-matids are unable to detect fish predators usingolfaction (Sexton et al. 1994). If A. tigrinumindividuals disperse and select habitat randomly(Semlitsch 2008), an expected strategy when thedirection of suitable habitat is unpredictable(Hawkes 2009), then colonization probability atsites with fish may be a simple function of thetotal number of dispersers moving through anarea. Connected networks may also generate therescue effect in which a large number ofimmigrants reduces the overall chance of localextinction probability (Fig. 4; Brown and Kodric-Brown 1977). If isolated sites receive fewerimmigrants than connected sites, isolated sitesmay be more susceptible to extinction due to thecombined effects of fish predation and demo-

Fig. 4. Relationship of predicted extinction probabil-

ity to Euclidean connectivity in sites without (open

circles) and with (closed circles) fish predators. For

purposes of presentation, sites with fish occupancy

probabilities .0.85 were defined as present, and sites

with ,0.15 were defined as absent.

Fig. 5. Relative support of cost-distance connectivity

metrics based on summed Akaike weights for models

of initial occupancy probability, colonization probabil-

ity, and extinction probability of Ambystoma tigrinum.

Cost-distance models represented Euclidean distance

(white bars), expert opinion (gray bars), and desicca-

tion risk in matrix habitats (black bars).


COSENTINO ET AL.

graphic stochasticity. In contrast, colonization ofconnected wetlands without fish may be lessdependent on high propagule size, which wouldexplain why the effect of connectivity on coloni-zation probability was strongest for sites withfish. Although oviposition by a single female mayresult in successful colonization in fishless sites(Trenham et al. 2001), a greater breeding effort islikely required in sites with fish predators.

Interestingly, colonization probability was re-lated positively to wetland area when connectiv-ity was low. For ground-dispersing animals, theprobability of intercepting a patch increases withthe linear size of the patch, and linear size scalespositively with patch area (Bowman et al. 2002).Thus, our results suggest a target effect forisolated sites in which larger wetlands functionas larger targets for dispersers (Lomolino 1990).However, the effect of area on colonization wassubstantial only among fishless wetlands (Fig.3A–B). In isolated sites, the deterministic effectsof fish on A. tigrinum may overwhelm thestructural effect of large area on colonization,particularly if population size or diversity of fishincreases with wetland area. In our study area,some permanent, large wetlands were stockedwith large numbers of fish, whereas smallwetlands were often colonized naturally byrelatively few individuals (B. J. Cosentino,personal observation). In connected networks witha high density of dispersers, a target effect onsuccessful colonization may not occur becausethe probability of at least one propagule inter-cepting a patch is not strongly related to patcharea. Patch area may be more important forcolonization in isolated locations where thedensity of dispersers is low.

Although persistence of A. tigrinum in siteswith fish may depend on immigration, sources ofemigrants underlying the rescue effect are notlikely to be stable over time. During our study,the spatial distribution of fish varied temporallydue to local extinctions and colonizations, sug-gesting that A. tigrinum should exhibit strongspatiotemporal variation in recruitment resultingfrom fish metapopulation dynamics. Spatiotem-poral variation in recruitment may also be drivenby climate and local hydrology (Church 2008).Precipitation was above average during the lasttwo years of our study, and hydroperiod was nota strong driver of A. tigrinum turnover (Cosenti-

no 2011). However, survival and reproductivesuccess of A. tigrinum can be low in dry years andin wetlands with short hydroperiods (Church etal. 2007). Thus, breeding populations may shiftbetween functioning as sources and sinks overtime (Stacey et al. 1997, Johnson 2004) due tostochastic climatic conditions and fish turnover.Effective conservation in such systems likelyrequires maintaining an assemblage of wetlandsconnected via dispersal that vary in hydrologyand suitability for predatory fish.

In our system, fish occupancy was relatedpositively to wetland hydroperiod and negative-ly to both emergent vegetation cover and canopycover (Appendix). Thus, one possible interpreta-tion is that A. tigrinum turnover dynamics can beexplained by habitat variables as opposed to fishpresence. However, occupancy modeling of fishand habitat variables indicated that hydroperiod,emergent vegetation, and canopy cover were notstrong predictors of A. tigrinum turnover com-pared to fish presence (Cosentino 2011). Thus, weare confident that fish predation is a major factordriving A. tigrinum colonization and extinctiondynamics.

Dispersal costs and landscape connectivityOur results indicate that the mechanistic basis

of dispersal limitation for A. tigrinum likelyinvolves the physiological risk of desiccation.Wetlands separated by matrix habitats with alow risk of desiccation (forest and soybean) weremore likely to be colonized than wetlandsseparated by high-risk habitat (corn and grass-land). Although a connectivity model based onexpert opinion had competitive support (Table1), summed Akaike weights across all modelsindicated that desiccation-informed connectivityhad more overall weight for predicting coloniza-tion (Fig. 5). Given the limited timeframe of ourstudy and the similarity in most cost valuesbetween models CDES and CEXP (Fig. 2), thepower to detect strong differences in supportbetween CDES and CEXP was likely low. However,there was little support for a Euclidean model ofconnectivity (Fig. 5), which indicated that matrixheterogeneity in general had a strong effect onwetland colonization.

Previous cost-distance studies have used mod-el-fitting techniques (e.g., Sutcliffe et al. 2003) andexpert opinion to estimate habitat-specific move-


COSENTINO ET AL.

ment resistances. Less frequently, experimentaldata have been used to inform resistances byassessing movement rates in matrix habitats andmovement decisions at habitat boundaries(Schooley and Wiens 2004, Stevens et al. 2004,2006, Castellon and Sieving 2006, Desrochers etal. 2011). Our study extends this approach bymechanistically linking a physiological constrainton small-scale movement decisions to landscape-scale population dynamics (Lima and Zollner1996).

In spatially structured systems, populationdynamics can be strongly dependent on theability of animals to disperse among habitatpatches (Bowler and Benton 2005). For pond-breeding amphibians, water economy plays anessential role in determining mortality risk andoverall resistance to dispersal in matrix habitats.Desiccation can decrease locomotor performance(Preest and Pough 1989), habitat permeability(Rothermel and Semlitsch 2002, Mazerolle andDesrochers 2005), and survival (Rittenhouse et al.2009) in amphibians. Additionally, habitat choic-es at forest-clearcut edges are commonly biasedtowards forest, where desiccation risk is pre-sumed to be low (Chan-McLeod 2003, Ritten-house and Semlitsch 2006, Cosentino et al. 2011).Taken together, these studies indicate physiologycan be a major constraint on dispersal, and ourresults reveal the consequences of that constraintfor wetland colonization and predatory-preyinteractions.

Matrix effects on occupancy dynamicsFew empirical studies have assessed how

matrix effects vary among occupancy, coloniza-tion, and extinction (e.g., Schooley and Branch2009). The absence of matrix effects on initialoccupancy and extinction for A. tigrinum may berelated to the dynamic nature of landscapestructure in agricultural systems dominated byannual row crops. Matrix habitats in agroecosys-tems are temporally dynamic due to yearly croprotations and long-term trends in crop plantings.However, habitat maps in ecological studiesgenerally represent a snapshot of landscapestructure. When dispersal costs vary amongagricultural crops, snapshot maps may haveimportant limitations. Occupancy and extinctionpatterns in a given year may reflect long-termstochastic processes, delayed responses to deter-

ministic factors, or immigration patterns overmultiple years. Therefore, historical effects ofmatrix structure on immigration may go unde-tected by using a snapshot of landscape struc-ture. In contrast, matrix effects on colonizationshould be easier to detect with snapshot maps ifthe timing of colonization aligns with the periodfor which land cover is represented. For pond-breeding amphibians like A. tigrinum, wetlandcolonizations are likely punctuated events result-ing from recent dispersal across matrix habitats(Semlitsch 2008). To account for historical effectsin cost-distance modeling, cost surfaces fordifferent years may be merged using GISmethods. However, such an approach is difficultdue to uncertainties about the timeframe forecological processes and the appropriate schemefor weighting resistance values over time.

ConclusionThe introduction of non-native fish predators

is a well-documented cause of amphibian de-clines and local extinction (e.g., Kats and Ferrer2003, Pilliod et al. 2010). We show that extinctionrisk may be greatest for amphibians when fishare introduced to isolated sites where thepotential for the rescue effect is low. This studyalso demonstrates how matrix structure mayconstrain the potential for recolonization afterextinction when wetlands are surrounded byhabitats with high desiccation risk. Overall, ourresults underscore the importance of populationconnectivity and landscape structure for main-taining pond-breeding amphibians in wetlandnetworks with predatory fish. We suggest that amore complete understanding of effects of fishpredators on the distribution of amphibiansrequires a spatial perspective on demographyand turnover dynamics.

ACKNOWLEDGMENTS

This research was funded with support from theAmerican Museum of Natural History, ChicagoHerpetological Society, Illinois State Academy ofScience, and University of Illinois. We thank J.Edmonson, B. Fischman, S. Hager, Z. Johnson, A.Kritzman, E. Manofsky, R. Pardee, J. Ross, N. Smeenk,S. Strom, and J. Wolff for their assistance withfieldwork. T. Moyer and B. Towey provided invaluablelogistical support, and we are grateful to the Richard-son Wildlife Foundation and landowners of Lee


COSENTINO ET AL.

County, IL for facilitating this research. G. Batzli, J.Brawn, E. Heske, W. Lowe, M. Mazerolle, K. Paige, andthe Schooley lab group provided helpful comments onthe manuscript.

LITERATURE CITED

Adriaensen, F., J. P. Chardon, G. De Blust, E. Swinnen,S. Villalba, H. Gulinck, and E. Matthysen. 2003. Theapplication of ‘least-cost’ modelling as a functionallandscape model. Landscape and Urban Planning64:233–247.

Amezcua, A. B., and M. Holyoak. 2000. Empiricalevidence for predator-prey source-sink dynamics.Ecology 81:3087–3098.

Babbitt, K. J., M. J. Baber, and T. L. Tarr. 2003. Patternsof larval amphibian distribution along a wetlandhydroperiod gradient. Canadian Journal of Zoolo-gy 81:1539–1552.

Baguette, M. 2004. The classical metapopulation theoryand the real natural world: a critical appraisal.Basic and Applied Ecology 5:213–224.

Bowler, D. E., and T. G. Benton. 2005. Causes andconsequences of animal dispersal strategies: relat-ing individual behaviour to spatial dynamics.Biological Reviews 80:205–225.

Bowman, J., N. Cappuccino, and L. Fahrig. 2002. Patchsize and population density: the effect of immigra-tion behavior. Conservation Ecology 6(1):9.

Brown, J. H., and A. Kodric-Brown. 1977. Turnoverrates in insular biogeography: effect of immigrationon extinction. Ecology 58:445–449.

Burnham, K. P., and D. R. Anderson. 2002. Modelselection and multimodel inference: a practicalinformation-theoretic approach. Second edition.Springer, New York, New York, USA.

Castellon, T. D., and K. E. Sieving. 2006. An experi-mental test of matrix permeability and corridor useby an endemic understory bird. ConservationBiology 20:135–145.

Caudill, C. C. 2003. Empirical evidence for nonselec-tive recruitment and a source-sink dynamic in amayfly metapopulation. 2003. Ecology 84:2119–2132.

Caudill, C. C. 2005. Trout predators and demographicsources and sinks in a mayfly metapopulation.Ecology 86:935–946.

Chan-McLeod, A. C. A. 2003. Factors affecting thepermeability of clearcuts to red-legged frogs.Journal of Wildlife Management 67:663–671.

Church, D. R. 2008. Role of current versus historicalhydrology in amphibian species turnover withinlocal pond communities. Copeia 2008:115–125.

Church, D. R., L. L. Bailey, H. M. Wilbur, W. L.Kendall, and J. E. Hines. 2007. Iteroparity in thevariable environment of the salamander Ambystomatigrinum. Ecology 88:891–903.

Compton, B. W., K. McGarigal, S. A. Cushman, andL. R. Gamble. 2007. A resistant-kernel model ofconnectivity for amphibians that breed in vernalpools. Conservation Biology 21:788–799.

Cosentino, B. J. 2011. Connectivity, metapopulationdynamics, and spatial genetics of tiger salamandersin a heterogeneous landscape. Dissertation. Uni-versity of Illinois, Urbana, Illinois, USA.

Cosentino, B. J., R. L. Schooley, and C. A. Phillips. 2010.Wetland hydrology, area, and isolation influenceoccupancy and spatial turnover of the paintedturtle, Chrysemys picta. Landscape Ecology 25:1589–1600.

Cosentino, B. J., R. L. Schooley, and C. A. Phillips. 2011.Connectivity of agroecosystems: dispersal costs canvary among crops. Landscape Ecology 26:371–379.

Cronin, J. T., and K. J. Haynes. 2004. An invasive plantpromotes unstable host-parasitoid patch dynamics.Ecology 85:2772–2782.

Desrochers, A., M. Belisle, J. Morand-Ferron, and J.Bourque. 2011. Integrating GIS and homing exper-iments to study avian movement costs. LandscapeEcology 26:47–58.

Greenwald, K. R., J. L. Purrenhage, and W. K. Savage.2009. Landcover predicts isolation in Ambystomasalamanders across region and species. BiologicalConservation 142:2493–2500.

Hanski, I. 1994. A practical model of metapopulationdynamics. Journal of Animal Ecology 63:151–162.

Hanski, I., and O. E. Gaggiotti, editors. 2004. Ecology,evolution, and genetics of metapopulations. Aca-demic Press, San Diego, California, USA.

Hawkes, C. 2009. Linking movement behaviour,dispersal and population processes: is individualvariation a key? Journal of Animal Ecology 78:894–906.

Hecnar, S. J., and R. T. M’Closkey. 1997. The effects ofpredatory fish on amphibian species richness anddistribution. Biological Conservation 79:123–131.

Johnson, D. M. 2004. Source-sink dynamics in atemporally heterogeneous environment. Ecology85:2037–2045.

Joly, P., C. Miaud, A. Lehmann, and O. Grolet. 2001.Habitat matrix effects on pond occupancy in newts.Conservation Biology 15:239–248.

Kats, L. B., and R. P. Ferrer. 2003. Alien predators andamphibian declines: review of two decades ofscience and the transition to conservation. Diversi-ty and Distributions 9:99–110.

Lima, S. L., and P. A. Zollner. 1996. Towards abehavioral ecology of ecological landscapes. Trendsin Ecology and Evolution 11:131–135.

Lomolino, M. V. 1990. The target area hypothesis: theinfluence of island area on immigration rates ofnon-volant mammals. Oikos 57:297–300.

MacKenzie, D., J. Nichols, J. A. Royle, K. H. Pollock,L. L. Bailey, and J. E. Hines. 2006. Occupancy


COSENTINO ET AL.

estimation and modeling: inferring patterns anddynamics of species occurrence. Academic Press,Burlington, Massachusetts, USA.

Mazerolle, M. J., and A. Desrochers. 2005. Landscaperesistance to frog movements. Canadian Journal ofZoology 83:455–464.

Moilanen, A., and M. Nieminen. 2002. Simple connec-tivity measures in spatial ecology. Ecology 83:1131–1145.

Pellet, J., E. Fleishman, D. S. Dobkin, A. Gander, andD. D. Murphy. 2007. An empirical evaluation of thearea and isolation paradigm of metapopulationdynamics. Biological Conservation 136:483–495.

Petranka, J. W. 1998. Salamanders of the United Statesand Canada. Smithsonian Institute Press, Wash-ington, D.C., USA.

Pilliod, D. S., B. R. Hossack, P. F. Bahls, E. L. Bull, P. S.Corn, G. Hokit, B. A. Maxwell, J. C. Munger, andA. Wyrick. 2010. Non-native salmonids affectamphibian occupancy at multiple spatial scales.Diversity and Distributions 16:959–974.

Pilliod, D. S., and C. R. Peterson. 2001. Local andlandscape effects of introduced trout on amphibi-ans in historically fishless watersheds. Ecosystems4:322–333.

Pope, K. L. 2008. Assessing changes in amphibianpopulation dynamics following experimental ma-nipulations of introduced fish. Conservation Biol-ogy 22:1572–1581.

Preest, M. R., and F. H. Pough. 1989. Interaction oftemperature and hydration on locomotion of toads.Functional Ecology 3:693–699.

Prugh, L. R. 2009. An evaluation of patch connectivitymeasures. Ecological Applications 19:1300–1310.

Prugh, L. R., K. E. Hodges, A. R. E. Sinclair, and J. S.Brashares. 2008. Effect of habitat area and isolationon fragmented animal populations. Proceedings ofthe National Academy of Sciences 105:20770–20775.

Pulliam, H. R. 1988. Sources, sinks, and populationregulation. American Naturalist 132:652–661.

Ray, N. 2005. PATHMATRIX: a geographical informa-tion system tool to compute effective distancesamong samples. Molecular Ecology Notes 5:177–180.

Ricketts, T. H. 2001. The matrix matters: effectiveisolation in fragmented landscapes. AmericanNaturalist 158:87–99.

Richter-Boix, A., G. A. Llorente, and A. Montori. 2007.Structure and dynamics of an amphibian meta-community in two regions. Journal of AnimalEcology 76:607–618.

Rittenhouse, T. A. G., and R. D. Semlitsch. 2006.Grasslands as movement barriers for a forest-associated salamander: migration behavior of adultand juvenile salamanders at a distinct habitat edge.Biological Conservation 131:14–22.

Rittenhouse, T. A. G., R. D. Semlitsch, and F. R.Thompson. 2009. Survival costs associated withwood frog breeding migrations: effects of timberharvest and drought. Ecology 90:1620–1630.

Rothermel, B. B. 2004. Migratory success of juveniles: apotential constraint on connectivity for pond-breeding amphibians. Ecological Applications14:1535–1546.

Rothermel, B. B., and R. D. Semlitsch. 2002. Anexperimental investigation of landscape resistanceof forest versus old-field habitats to emigratingjuvenile amphibians. Conservation Biology16:1324–1332.

Schooley, R. L., and L. C. Branch. 2009. Enhancing thearea-isolation paradigm: habitat heterogeneity andmetapopulation dynamics of a rare wetland mam-mal. Ecological Applications 19:1708–1722.

Schooley, R. L., and J. A. Wiens. 2004. Movement ofcactus bugs: patch transfers, matrix resistance, andedge permeability. Landscape Ecology 19:801–810.

Semlitsch, R. D. 2008. Differentiating migration anddispersal processes for pond-breeding amphibians.Journal of Wildlife Management 72:260–267.

Sexton, O. J., and C. A. Phillips. 1986. A qualitativestudy of fish-amphibian interactions in 3 Missouriponds. Transactions of the Missouri Academy ofScience 20:25–35.

Sexton, O. J., C. A. Phillips, and E. Routman. 1994. Theresponse of naıve breeding adults of the spottedsalamander to fish. Behaviour 130:113–121.

Sjogren Gulve, P. 1994. Distribution and extinctionpatterns within a northern metapopulation of thepool frog, Rana lessonae. Ecology 75:1357–1367.

Spight, T. M. 1968. The water economy of salamanders:evaporative water loss. Physiological Zoology41:195–203.

Stacey, P. B., V. A. Johnson, and M. L. Taper. 1997.Migration within metapopulations: the impactupon local population dynamics. Pages 267–291in I. A. Hanski and M. E. Gilpin, editors.Metapopulation biology: ecology, genetics, andevolution. Academic Press, San Diego, California,USA.

Stevens, V. M., E. Leboulenge, R. A. Wesselingh, andM. Baguette. 2006. Quantifying functional connec-tivity: experimental assessment of boundary per-meability for the natterjack toad (Bufo calamita).Oecologia 150:161–171.

Stevens, V. M., E. Polus, R. A. Wesselingh, N.Schtickzelle, and M. Baguette. 2004. Quantifyingfunctional connectivity: experimental evidence forpatch-specific resistance in the Natterjack toad(Bufo calamita). Landscape Ecology 19:829–842.

Sutcliffe, O. L., V. Bakkestuen, G. Fry, and O. E.Stabbetorp. 2003. Modelling the benefits of farm-land restoration: methodology and application tobutterfly movement. Landscape and Urban Plan-


COSENTINO ET AL.

ning 63:15–31.Thomas, C. D. 1994. Extinction, colonization, and

metapopulations: environmental tracking by rarespecies. Conservation Biology 8:373–378.

Trenham, P. C., W. D. Koenig, and H. B. Shaffer. 2001.Spatially autocorrelated demography and inter-pond dispersal in the salamander Ambystomacaliforniense. 2001. Ecology 82:3519–3530.

Urban, M. C. 2004. Disturbance heterogeneity deter-mines freshwater metacommunity structure. Ecol-ogy 85:2971–2978.

Wellborn, G. A., D. K. Skelly, and E. E. Werner. 1996.Mechanisms creating community structure across afreshwater habitat gradient. Annual Review of

Ecology and Systematics 27:337–363.

Werner, E. E., D. K. Skelly, R. A. Relyea, and K. L.

Yurewicz. 2007. Amphibian species richness across

environmental gradients. Oikos 116:1697–1712.

With, K. A. 2004. Metapopulation dynamics: perspec-

tives from landscape ecology. Pages 23–44 in I.

Hanski and O. E. Gaggiotti, editors. Ecology,

genetics, and evolution of metapopulations. Aca-

demic Press, San Diego, California, USA.

Woodford, D. J., and A. R. McIntosh. 2010. Evidence of

source-sink metapopulations in a vulnerable native

galaxiid fish driven by introduced trout. Ecological

Applications 20:967–977.

APPENDIX

Table A1. Occupancy models for fish predators from 90

wetlands in northern Illinois in 2007, 2008, and 2009.

Year Model DAICC xi �2l K

2007 L1 þ L2 0.00 0.97 80.63 5L1 þ A 7.96 0.02 88.59 5

2008 L1 þ L2 0.00 0.90 231.80 4L1 þ A 6.44 0.04 238.24 4

2009 L1 þ L2 0.00 0.78 266.22 5L1 þ S 4.48 0.08 270.70 5

Notes: Main effects include wetland area (A), the number ofstream inlets/outlets at a wetland (S), and PCA scores for twoaxes that explained 91.8% of the variation in local habitatcharacteristics (canopy cover, emergent vegetation cover, andhydroperiod). The first PCA axis (L1) was positivelycorrelated with emergent vegetation and negatively correlat-ed with hydroperiod (factor loadings: canopy ¼ 0.03,emergent vegetation¼ 0.95, hydroperiod¼�0.79). The secondaxis (L2) was positively correlated with canopy cover andnegatively correlated with hydroperiod (factor loadings:canopy ¼ 0.98, emergent vegetation ¼�0.11, hydroperiod ¼�0.50). Models are presented with the relative differencebetween model AICC and AICC for the best model (DAICC),Akaike weights (xi ), twice the negative log-likelihood (�2l ),and the number of parameters (K). Effects of Julian date ondetection probability were included in all models for 2007 and2009 (accounting for 2 parameters). The average detectionprobability 6 1 SE was 0.94 6 .01 in 2007, 0.82 6 0.03 in 2008,and 0.73 6 0.01 in 2009. The top two models are included foreach year. Fish occupancy was related negatively to emergentvegetation and canopy cover and positively to hydroperiod ineach year (beta estimates for top models 6 1 SE: L12007 ¼�2.51 6 0.59, L22007 ¼�3.98 6 1.59; L12008 ¼�1.11 6 0.28,L22008¼�0.94 6 0.38; L12009¼�1.12 6 0.29, L22009¼�0.84 60.35).


COSENTINO ET AL.

new spatial connectivity moderates the effect of predatory fish on … · 2020. 8. 29. · spatial...

Documents