Using Spatial Capture–Recapture to Elucidate Population Processes and Space-Use in Herpetological Studies
The cryptic behavior and ecology of herpetofauna make estimating the impacts of environmental change on demography difficult; yet, the ability to measure demographic relationships is essential for elucidating mechanisms leading to the population declines reported for herpetofauna worldwide. Recently developed spatial capture–recapture (SCR) methods are well suited to standard herpetofauna monitoring approaches. Individually identifying animals and their locations allows accurate estimates of population densities and survival. Spatial capture–recapture methods also allow estimation of parameters describing space-use and movement, which generally are expensive or difficult to obtain using other methods. In this paper, we discuss the basic components of SCR models, the available software for conducting analyses, and the experimental designs based on common herpetological survey methods. We then apply SCR models to Red-backed Salamander (Plethodon cinereus), to determine differences in density, survival, dispersal, and space-use between adult male and female salamanders. By highlighting the capabilities of SCR, and its advantages compared to traditional methods, we hope to give herpetologists the resource they need to apply SCR in their own systems.Abstract
(A) Traditional encounter histories with individuals i and survey occasions k. Ones indicate captures and zeroes indicate otherwise (e.g., in red, individual caught on first and last occasion). (B) For the spatial encounter history with a multicatch encounter model, a response of 1 indicates not captured, and each number represents trap in which an animal was caught. Because “trap 1” is no capture, possible responses are 1 plus the number of traps. In red, the individual was caught in the second trap during the first survey and the eighth trap in the last survey. (C) The spatial encounter history type used in our robust design analysis. There are six secondary surveys within three primary seasons. In red, the individual was caught in the first trap on survey 4 and the fifth trap on survey 6 during the first primary season. It was not caught again until the last survey of the third primary season.
(A) The Gaussian encounter probability is one of the simplest ways to model space-use. The baseline detection probability, p0, acts as the intercept in the detection function and represents the detection when an individual's activity center (black dot) aligns at a trap location. The spatial scaling term, σ, determines how detection decreases with distance from an activity center. A larger σ indicates larger space-use. A 95% movement radius (dashed lined) is then r = σ
. (B) This encounter probability translates into a circular area of use where 95% of movement outcomes occur within the dashed line (via πr2). Across this area, detection decreases (gradient) as distance increases from activity center.
The process in which SCR estimates the number of individuals within a defined study area. (A) As an example, the state-space (study area boundary denoted in red) can be thought of as having M potential animals and their true activity centers (black dots) that occur inside or outside of the trapping area (dashed line). (B) This individual has a unique spatial encounter history based on the traps (gray squares) in which it was caught (connected by lines). The model then estimates its activity center (black dot). (C) Collectively, the location of each activity center and the movement exhibited in the spatial encounter history inform estimates from the encounter model. (D) The encounter model is then used to estimate density, and the M potential activity centers are evaluated. Triangles represent captured individual's activity centers, circles represent unobserved individual's activity centers, and gray represents individual activity centers that do not exist in the population.
Representations of different SCR study designs. (Left) Unstructured surveys, say by a dog team, follow some arbitrary and free-roaming path (gray line) through a state-space (red boundary). An array of gridded cells is overlain on the path, and the length of the path per cell is calculated and used as a covariate to account for effort. (Center) Assuming researchers and managers have a maximum number of traps, a clustered grid of 48 traps (gray squares) can be used to cover a larger amount of area while still getting high spatial recaptures. The area covered by the trapping area (dashed line) is 88 square units. (Right) Comparatively, a systematic grid uses 50 traps to cover 50 square units.
Our study site in Union Co., Pennsylvania. Pictured is one of our coverboard arrays. Each coverboard is given a unique identification based on its row (A–E) and column (e.g., A6, D10) location. Coverboard arrays measured 5 × 10 m and contained 50 pine (Pinus strobus) or hemlock (Tsuga canadensis) boards spaced 1 m apart.
An example of how an individual Plethodon cinereus was marked using visual implant elastomer. We used five colors (no mark, red, yellow, orange, and blue) to mark individuals encountered in the coverboard array. This individual was given no mark in position 1, an orange mark in position 2, a red mark in position 3, and a blue mark in position 4, earning the identification xORB. For clarity, marking locations are emphasized by circles.
Plethodon cinereus density estimates (left) and space-use estimates (right) across the three closed-population seasons for both females (black) and males (gray). Means (squares) and 95% BCIs (lines) for density do not vary greatly among seasons. For space-use, females used significantly less space than males. Males estimates are not provided for autumn 2013 because of no male recaptures.
Contributor Notes
