Abstract Accurate and reliable estimation of wildlife population density is fundamental to effective conservation and management. While camera traps show potential for monitoring ground‐dwelling mammal densities, labour‐intensive data processing remains a significant constraint. Ideally, conservation efforts would benefit from the continuous monitoring of multiple species across broad spatial scales; however, the time and effort required for data processing make achieving this challenging. We developed the RAD‐REST (Random Encounter and Staying Time model Relying on All Detections) model, an extension of the Random Encounter and Staying Time model, to enable multi‐species density monitoring with substantially reduced effort. By explicitly modelling the probabilistic process of animals entering a predefined focal area, our model allows density estimation by analysing entry counts and staying times in only a subset of videos, while incorporating trapping rates from all video data. We also developed a user‐friendly R package ‘ctrest’ to implement the model's fully Bayesian framework, including estimation of activity levels. Using data from the Boso Peninsula, Japan, we conducted Monte Carlo simulations to determine the number of videos required to achieve (1) standard precision for density estimation (coefficient of variation, ) and (2) more stringent wildlife management standards (). The RAD‐REST model produced unbiased estimates with appropriate coverage rate. Although the original REST model showed superior precision, RAD‐REST model achieved acceptable estimation precision with dramatically reduced analytical effort. To reach acceptable values (≤0.35) using a 200‐camera array, most species (10 of 12) required analysing only 100 videos for entry counts and measuring staying times in videos with ≥1 entry (approximately 60%–70% of analysed videos). For higher precision (), a larger 400‐camera network was necessary, requiring analysis of 7.00% (single‐species) or 6.45% (multi‐species) of total expected videos (>ca. 40,000 videos). Although precise estimates might still demand substantial effort, the RAD‐REST model enables cost‐effective monitoring by either supplementing photo‐trap arrays with a subset of video data or deploying dedicated video cameras across a portion of the network. This approach marks a critical advance towards large‐scale monitoring efforts, enabling more robust, multi‐species density estimation across diverse ecological contexts.
Nakashima et al. (Sat,) studied this question.