Noninvasive genetic sampling with a spatial capture‐recapture analysis to estimate abundance of Roosevelt elk

Abstract

AbstractPrecise abundance estimates of large mammals are important for effective conservation, harvest, and conflict management. Determining abundance of wide‐ranging, herding ungulate species has presented a unique challenge to wildlife managers because of factors such as dense forest, elusive behavior, and heterogeneity in density across the landscape. Roosevelt elk (Cervus canadensis roosevelti) populations in Northern California, USA, are no exception to these challenges, and as the elk population has grown, so has human–wildlife conflict, necessitating the need for efficient and repeatable methods to determine population abundance for management decisions. We explored non‐invasive genetic sampling combined with spatial capture‐recapture (SCR) as an alternative for monitoring populations that are difficult to observe directly. We combined an SCR model with a binomial point process and an unstructured single survey search method to estimate elk abundance in Northern California via Bayesian inference. We searched open grassy hillsides for female‐calf groups and used a detection dog team to search forested areas to increase the number of detections of males and other solitary individuals. For the SCR analysis, we used sex and survey effort as detection covariates, and used a trap‐level random effect to account for overdispersion in the count data from the herding behavior of elk. Our population estimate (N ± SD) for the study area was 618 ± 36.34 individuals (95% Bayesian credible interval = 551–693) with a mean density of 1.09 ± 0.06 elk/km2. Our work demonstrates a potential method to estimate population size of ungulates in an area that is not conducive to traditional monitoring methods.