Camera trapping for density estimation: comparing the TIFC model to aerial surveys for multiple ungulate populations

Population density estimates are invaluable to wildlife managers, but difficult to attain. Several methods have been developed to estimate density using camera traps, many of which require further testing. The time-in-front-of-the-camera (TIFC) approach allows for density estimation when "unmarked" individuals are monitored using camera traps. We applied the TIFC model to populations of bison (Bison bison, B. bison athabascae), elk (Cervus elaphus canadensis), and moose (Alces alces) in Elk Island National Park (EINP) and to populations of elk and moose in Cooking Lake - Blackfoot Provincial Recreation Area (BPRA). EINP and BPRA are fully fenced natural areas in the Beaverhills Region of central Alberta, Canada. Our objectives were to (i) use the TIFC model to estimate ungulate densities in EINP and BPRA, and (ii) compare the performance of TIFC density estimates against aerial ungulate survey estimates. Camera trap data were collected from 43 cameras in EINP between December 2016 and October 2020, and 23 cameras in BPRA from April 2019 to August 2020. Annual densities were estimated in EINP north and south (2017-2019) and in BPRA (2019). Moose density estimates had the lowest discrepancy between approaches. Bison TIFC density estimates were lower than AUS densities, and elk TIFC density estimates were higher than AUS densities. In addition to the density estimates evaluated for the three focal species, the TIFC approach also was applied to white-tailed deer (Odocoileus virginianus) and mule deer (O. hemionus) in EINP and BPRA, in the absence of aerial survey data. We conclude that the TIFC model and AUS were complementary, with pros and cons of the two approaches varying based on focal species ecology. Careful consideration is required for several factors related to camera study design for TIFC density estimation that can affect the accuracy and precision of estimates.