Field estimates of detectability and site occupancy show that northern Italy forest molluscs are spatially rare and poorly detectable

Abstract Estimations of site occupancy of species and their vital rates, and of species richness of a community, are biased if detection probability is not taken into account. Models that allow the joint estimation of occupancy and detectability were used in a land mollusc survey in six forest habitats in northern Italy, allowing estimations for 55 species of land snails. Most species were poorly detectable (p ≤ 0.3). Detectability varied among habitats and collecting techniques. Taking into account detectability, occupancy estimations may be from 15 to 85% larger than the naïve occupancy values computed by dividing the number of stations at which the species was detected by the total number of stations (0.13–0.42 in absolute terms), significantly altering our understanding of species rarity. In several habitats the share of spatially rare species was remarkably high, above 30%, with peaks above 60%; at the regional scale 62.7% for visual sampling and 46.7% for leaf litter and soil sampling. The sampling effort to monitor occupancy for the numerous uncommon to rare species is well above what would be satisfactory in terms of saturation of species inventory. Visual search gave higher estimates of detectability for species above 1 cm in size, while leaf litter and soil sorting gave higher estimates for species below 1 cm. Consideration of fresh empty shells along with living individuals allowed us to improve the estimates, but when interpreting results the bias induced by taphonomic processes has to be considered. When detection probabilities are unknown, an estimate of their order of magnitude may be obtained by relative abundances, while shell size does not correlate significantly with detectability.