The objective of this study was to improve our understanding of source-sink dynamics for subpopulations or populations of species of relevance to Department of Defense resource managers. Specifically, this project sought to better understand the impact of landscape heterogeneity on the source-sink dynamics of a species of conservation concern (the ringed salamander, Ambystoma annulatum) and a co-occurring species (the spotted salamander, A. maculatum) at Fort Leonard Wood, Missouri. Investigators sought to (1) define source habitat patches by identifying key environmental conditions associated with varying levels of reproductive success of A. annulatum, (2) determine whether species interactions influence source-sink dynamics, (3) identify habitat features associated with effective dispersal of A. annulatum among populations to define habitat connectivity, and (4) develop models to predict and manage population connectivity for salamanders in heterogeneous landscapes.

Technical Approach

This study involved a combination of intensive ecological field studies, genetic analyses, and statistical modeling. Between 2011 and 2014, larvae were surveyed at each pond for three consecutive days within each sampling period. Field teams recorded the abundance of each species and that of invertebrate predators, other amphibians, and fish. During each year, visual encounter surveys and quadrat sampling for egg masses were also conducted. These surveys involved a single visit to each pond, during which data were recorded for the number of eggs observed and the number of metamorphs captured, respectively.

To study dispersal, investigators reared salamanders from hatchlings and manipulated larval densities to generate juveniles that differed in size. The activity level of juvenile salamanders was repeatedly measured to determine if movement behavior was repeatable, indicative of dispersers being intrinsically different from residents. Nine morphological traits were measured and regressed against distance moved. Whether habitat type or the interaction between body size and habitat affected initial movement distance was tested in a field study using replicated field enclosures to isolate the effects of natal and juvenile habitat on initial movement behavior. Investigators measured the surface area and maximum depth of each pond each time it was sampled. Pond attributes assumed to be constant over time (pond basin area, hydroperiod, canopy closure, emergent and submergent vegetation) were measured once during 2012.

Species-specific microsatellite loci were developed for A. annulatum and A. maculatum. After analysis of the 2012 data revealed limited genetic differentiation among ponds for both species, A. opacum (marbled salamander), a fall-breeding species, was added for comparison to the fallbreeding A. annulatum. After analysis of the 2013 data, the spring breeding Notophthalmus viridescens (eastern newt) was also added for comparison to the spring-breeding A. maculatum. Genotypes for all species were tested for genetic differentiation using traditional measures of genetic distance (FST and G'ST) as well as Bayesian analyses to infer genetic partitions.

Landscape resistance was assessed using resistance surfaces generated in ArcGIS. The effects of distance and landscape resistance surfaces were assessed using CIRCUITSCAPE v4 and optimized using the "ResistanceGA" package for R. Further, pairwise G’ST was used as the response variable and scaled effective resistance surfaces as the independent variable in a linear mixed effects model framework implemented in the "lme4" package for R. Model fits were assessed using AICC calculated from the linear mixed effects models.


Habitat features, especially pond hydroperiod, forested habitat surrounding ponds including canopy cover, and the presence/absence of fish were found to be the strongest predictors of A. annulatum and A. maculatum distribution and population dynamics. Evidence was weak for species interactions being important factors in population dynamics in natural ponds, but strong evidence was observed for density-dependence and food web organization as critical components in experimental studies. Although the field data do not clearly support an important role for species interactions in source-sink dynamics in this system, the experimental data suggest that caution is warranted, and species interactions should not be disregarded as important contributors to source-sink dynamics.

The results indicate that metamorph size was strongly affected by intraspecific density and field observations indicated extensive interpond variability in metamorph size. Size affects dispersal in that intermediate sized metamorphs dispersed the farthest. A better understanding of the habitat and ecological factors affecting metamorph size in the field is an important next step to ensuring population viability and connectivity.

The absence of fish is a key characteristic of source ponds, and it is important that ponds be located within forested habitat. Field observations support a pattern of most ponds having few larvae. Spatial variation in larval density was four to ten times greater than temporal variation, suggesting that few source populations likely occur on the landscape, making their conservation priority even greater. Because growth in salamanders was affected by pond hydroperiod, it is recommended that ponds with a mosaic of hydroperiod regimes should be maintained.

Although genetic differentiation was limited, no support was found for land cover or topographic features to affect gene flow in any species, making such connectivity models irrelevant. Isolation by distance was evident and significant in A. annulatum, A. maculatum, and A. opacum. For both A. annulatum and A. maculatum, effective dispersal distances were estimated from the genetic data. The estimated dispersal distances were used to parameterize a demographic network model, which was used to assess the local importance of a population to maintaining connectivity, regional connectivity, and global contribution to gene flow of the entire network.


This project’s demographic modelling allowed for the simulation of the likely number of immigrants, emigrants, and philopatric individuals at each pond, determining average source-sink status, and assessing more integrative measures such as metapopulation capacity. Modeling results suggest that populations occur on a continuum of source to sink and that most populations fluctuate in their contributions.

Results from this model can ultimately be used to improve management concerning removal of anthropogenic stressors, mitigation of habitat loss, or facilitation of terrestrial and pond restoration for A. annulatum across FLW at the landscape level. The in-depth understanding of the source-sink dynamics of this representative species can be used to direct management at FLW to ensure species persistence, link management at this site to regional dynamics within the adjacent U.S. Forest Service Mark Twain National Forest, and can be applied to other ambystomatid salamanders of concern at other DOD facilities (e.g., Ambystoma cingulatum and Ambystoma bishopi at Fort Stewart, GA or Eglin Air Force Base, FL).