N-mixture models reliably estimate the abundance of small vertebrates

Accurate measures of species abundance are essential to identify conservation strategies. N -mixture models are increasingly used to estimate abundance on the basis of species counts. In this study we tested whether abundance estimates obtained using N- mixture models provide consistent results with more traditional approaches requiring capture (capture-mark recapture and removal sampling). We focused on endemic, threatened species of amphibians and reptiles in Italy, for which accurate abundance data are needed for conservation assessments: the Lanza’s Alpine salamander Salamandra lanzai , the Ambrosi’s cave salamander Hydromantes ambrosii and the Aeolian wall lizard Podarcis raffonei . In visual counts, detection probability was variable among species, ranging between 0.14 (Alpine salamanders) and 0.60 (cave salamanders). For all the species, abundance estimates obtained using N -mixture models showed limited differences with the ones obtained through capture-mark-recapture or removal sampling. The match was particularly accurate for cave salamanders in sites with limited abundance and for lizards, nevertheless non-incorporating heterogeneity of detection probability increased bias. N- mixture models provide reliable abundance estimates that are comparable with the ones of more traditional approaches, and offer additional advantages such as a smaller sampling effort and no need of manipulating individuals, which in turn reduces the risk of harming animals and spreading diseases.