Robustness of fecal DNA spatial capture‐recapture to clustered space‐use by tule elk

Abstract

AbstractSpatially explicit capture‐recapture (SCR) approaches using noninvasive fecal DNA (fDNA) are applied increasingly to obtain statistically robust abundance estimates for various wildlife species. But SCR has not been widely used for more gregarious species, such as elk (Cervus canadensis). Because of their heterogeneous use of the landscape and grouping behavior, elk present novel challenges to sampling efficiency and statistical assumptions. We employed fDNA SCR and a stratified random sampling approach to estimate abundance in 3 northern California tule elk (C. c. nannodes) populations concurrent with global positioning system (GPS)‐telemetry monitoring of 66 elk (32 male, 34 female) in Colusa and Lake counties, California, USA, during June–August 2017–2019. We collected 1,616 fecal pellet groups from the 3 populations, resulting in 1,002 fDNA genotypes (≥19 microsatellite loci, 1 sex marker) of 425 unique individuals. Based on SCR estimates from a model incorporating both sexes, elk density ranged from 0.31 (95% CI = 0.17–0.55) elk/km2to 1.7 (95% CI = 1.3–2.2) elk/km2, translating approximately to 650 individuals (evenly split between sexes) among the 3 populations. Independent telemetry data from concurrently tracked individuals indicated that activity centers of females, but not males, were clustered on the landscape. This finding was corroborated using fDNA to infer activity centers. Comparison of SCR estimates to non‐spatial estimates using physically captured individuals suggested that combined‐sex SCR models were robust to spatial clustering of females in all 3 populations. Additionally, single‐sex models were robust to female clustering in the 2 populations with multiple female groups, but not in Lake Pillsbury, where all females effectively shared a single activity center. In that case, the estimate of female abundance was deemed an overestimate. Thus, results indicate that SCR was robust to gregariousness in elk in all but the most extreme cases of clustering (single‐sex female estimation when all individuals share a single activity center), and that the inclusion of both sexes was sufficient to offset biases stemming from aggregation of females in all cases. Altogether, our findings suggest that SCR methods can be gainfully applied to socially gregarious ungulates such as elk.