Population monitoring of wildlife species requires techniques that produce estimates with low bias and adequate precision. Use of infraredtriggered camera (hereafter; camera) surveys for white-tailed deer (Odocoileus virginianus; deer) population density estimation is popular among land managers. However, current camera surveys do not provide an estimate of precision critical for accurate density estimation. We believed that incorporating spatial aspects of sampling into the analytical process would allow for both estimates of precision associated with density and an ability to calculate effective sample area. We conducted camera surveys for deer in Units 1 (1,385 ha) and 2 (1,488 ha) at Arnold Air Force Base, Tennessee, in August 2010. We used 1 camera per 53 and 62 ha in Units 1 and 2, respectively, and identified individual male deer based on antler criteria. We used spatially explicit capture-recapture (SECR) data with Program DENSITY to fit a spatial detection function (g0; probability of detecting an individual on a single occasion when the distance between their home range center and a trap is zero) and sigma (the scale parameter that determines the rate at which detection probability decreases with distance between a home range center and a trap) to estimate antlered male density. Density estimates were similar between camera surveys (based on recaptures of recognizable antlered males from camera images) using traditional sampling techniques (without spatial information on capture) and spatially explicit density estimation (with a record of location for each individual camera capture). Antlered male density estimates obtained via traditional sampling for Units 1 and 2 were 2.0 and 2.6 males/km2, respectively. Density estimates based on SECR models were 1.6 males/km2 (SE = 0.33, g0 = 0.24) for Unit 1 and 2.5 males/km2 (SE = 0.56, g0 = 0.14) for Unit 2. Both estimation methods indicated lower deer density in Unit 1 versus Unit 2. Analysis of camera surveys using SECR modeling uses the data from the spatial distribution of cameras and does not require the assumption of equal detectability. Use of SECR modeling can improve current camera survey methods by providing both a measure of precision that is currently lacking from traditional camera analysis methods and including spatial distribution of captured deer. Spatial modeling should be explored further to enhance our understanding of potential biases associated with behavioral responses to the use of bait as an attractant.
