Evaluating spatially explicit density estimates of unmarked wildlife detected by remote cameras

Abstract

Remote cameras have become a promising, cost‐effective tool for monitoring wildlife populations. Yet, for species where individuals are indistinguishable, remote cameras’ ability to provide robust and precise density estimates has been limited without the use of invasive marking. Using the American black bear as a model species, we evaluated methods for estimating wildlife densities using remote camera detections of unmarked individuals against estimates from spatial capture–recapture (SCR) models using individual detections. We also tested the effect of incorporating varying proportions of marked individuals on model accuracy and precision. Spatial count (SC) models using unmarked individuals produced estimates of bear density within 0.6% of those from SCR. We extended SC models to incorporate variation in density as a function of land use/land cover, and identified identical relationships between variation in bear densities and housing density as obtained using SCR. Incorporating individual detection data from simultaneous non‐invasive genetic sampling lead to more precise, but biased estimates. Synthesis and applications. Our results identify contexts in which camera count data can be used as an alternative to spatial capture–recapture (SCR) when individual identification is prohibitive. Spatial count models provided an accurate, but less precise replication of spatial capture–recapture density estimates and may provide consistent insights into spatial variation in density. Mixed samples of camera counts and auxiliary individual detections are likely to be of limited use, but fitting spatial count models to populations with partial visual markings could improve their precision.

Drawn out of the shadows: Surveying secretive forest species with camera trap distance sampling

“…SECR methods have been extended allowing density estimates of unmarked populations (Chandler & Royle, 2013). Here, sampling effort must be spatially intensive and estimates lack precision unless supplemented with auxiliary data such as genetic sampling or telemetry (Evans & Rittenhouse, 2018;Linden, Sirén, & Pekins, 2018;Sollmann et al, 2014). Random encounter models (REMs; Rowcliffe, Field, Turvey, & Carbone, 2008) were considered a promising development.…”

Evaluating and integrating spatial capture-recapture models with data of variable individual identifiability

“…Including habitat covariates that influence spatial density may in some cases offset bias (e.g. Evans and Rittenhouse 2018), but our results suggest that fully unmarked SCR models (i.e. SC) are often unsuitable for density estimation of the carnivores we studied, particularly when detectors are widely spaced, count data are sparse, and auxiliary data are not available.…”

Density Estimation of Unmarked Populations Using Camera Traps in Heterogeneous Space

“…Obtaining population density of wildlife is critical for effective conservation and management (Nakashima et al 2018). Development and validation of methods for estimating density using remote cameras in the absence of uniquely identifiable individuals can extend the technology's application to a wide variety of species, and reduces costs associated with population monitoring (Evans and Rittenhouse 2018). Based on the framework by Pollock et al (2004), our model provided a potential approach of how we can add species’ space‐use information into population density estimations using camera‐trap data.…”

“…They demonstrated that neither spatial independence nor individual recognition is needed to estimate population density. Evans and Rittenhouse (2018) evaluated estimates from SCR models. While N‐mixture models address the inability to uniquely identify individuals, SCR models address nonclosure by accommodating information on the juxtaposition of animal activity centers and traps (Chandler and Royle 2013).…”