Optimising occupancy models and detection probability for conservation monitoring in a forest-dwelling small mammal

The ability to determine the survey effort required to detect species presence is critical for the conservation of populations in order to monitor changes in distribution or abundance, a challenge for rare and elusive species. While designing an effective population survey may be of concern to scientists, it may also be a legal requirement for protected species. We analysed how spatial and temporal variation in sampling effort impacts detection probability in the hazel dormouse (Muscardinus avellanarius), a small mammal that naturally occurs at low density in forest habitat. We used presence and absence data from 144 sites in England, UK. We found that detection probability was strongly affected by survey intensity, with a 37.5% increase in detection probability when sampling effort was increased from 16 to 50 nest tubes per site. We also show that detection probability is highly temporally dependent; monitoring early in the year results in low detection probability (21%-53%; April-June), whilst detection dramatically increased later in the year (89%; September). Our results suggest that variation in detection probability can be complex and influenced by effects at temporal and spatial scales, and by sampling effort. We show that sampling effort can be minimised depending on seasonal variation in detection probability, however this is likely to be species and method dependent.