Effect of imperfect detection on the estimation of niche overlap between two forest dormice

Abstract: Quantification of nicheoverlap represents an important topic in several aspects of ecology and conservation biology, although it could be potentially affected by imperfectdetection, i.e., failure to detect a species at occupied sites. We investigate the effect of imperfectdetection on nicheoverlap quantification in two arboreal rodents, the edible dormouse(Glis glis) and the hazel dormouse( Muscardinusavellanarius). For both species, we used Generalized Linear Mixed Models(GLMM) to estimate the occurrence probabilityand OccupancyModels (OM) to calculate occurrence and detection probabilities. By comparing these predictions through nicheequivalency and similarity tests, we first hypothesised that methods correcting for imperfectdetection (OM) provide a more reliable estimate of nicheoverlap than traditional presence/ absence methods (GLMM). Furthermore, we hypothesised that GLMM mainly estimate species detectability rather than actual occurrence, and that a low number of sampling replicates provokes an underestimation of species nicheby GLMM. Our results highlighted that GLMM-based nicheoverlap yielded significant outcomes only for the equivalency test, while OM-based nicheoverlap reported significant outcomes for both nicheequivalency and similarity tests. Moreover, GLMM occurrence probabilities and OM detectabilities were not statistically different. Lastly, GLMM predictions based on single sampling replicates were statistically different from the average occurrence probability predicted by GLMM over all replicates. We emphasized how accounting for imperfectdetection can improve the statistical significance and interpretability of nicheoverlap estimates based on occurrence data. Under a habitat management perspective, an accurate quantification of nicheoverlap may provide useful information to assess the effects of different management practices on species occurrence.