Can camera traps predict habitat-species associations? A comparison with GPS-based habitat selection in red deer

Abstract Context Camera traps are widely used in ecological studies, including analyses of habitat use. However, the relationships between animal spatial behavior and landscape features obtained with camera traps have seldom been compared to those obtained with more established methods. Objectives We evaluated the use of camera trap data to predict habitat use in red deer Cervus elaphus by comparing them to GPS telemetry data, used as a benchmark. Methods Within a 100 km 2 study area in the Stelvio National Park (Italy), we collected red deer presence data from 50 camera traps from 2019 to 2023. In addition, 23 deer (15 females and 8 males) were equipped with global positioning system collars. For both types of data, we modeled presence/absence as a function of environmental variables using a generalized linear mixed modeling approach. Model estimates were compared in terms of directionality of the effects and of predicted habitat suitability. Results We found a higher concordance of effect directionality and predicted suitability for females than for males, possibly due to limited sample size in the latter. Discrepancies were found in specific habitat relationships and may be due to inherent differences between telemetry-based and camera trap-based data. Conclusions Our results suggest that camera trap data can capture coarse-scale patterns of habitat use, whereas telemetry data are more suitable for finer-scale analyses. The choice of which method to use depends on the research objectives.