Original investigationComparative evaluation of three sampling methods to estimate detection probability of American red squirrels (Tamiasciurus hudsonicus)

Measuring changes in species distribution and understanding factors influencing site occupancy are recurring goals in wildlife studies. Imperfect detection of species hinders such studies, resulting in the underestimation of the number of sites occupied by the species of interest. American red squirrels (Tamiasciurus hudsonicus) are sampled traditionally with live-traps that require substantial resources to deploy and monitor. Here, we assessed whether auditory methods yield similar detection probabilities. We compared the detection probability of American red squirrels in boreal forest using point counts, playback counts, and live-trapping. Over the summer of 2014, we conducted three trapping sessions in 60 sites within black spruce forests of northwestern Quebec, Canada. We also conducted 10 min point counts in the same sites, together with playback counts using recordings of American red squirrel alarm and territorial calls. Using dynamic occupancy models to analyse three primary periods, all composed of three secondary periods, we found that the detection probability of squirrels from point counts was as high as with live-trapping. Our results thus highlight the value of the point count method in measuring American red squirrel occupancy.