Local abundance and observer's identity affect visual detectability of Sardinian mountain newts

Visual counts gathered within citizen science programs are increasingly used to determine distribution and abundance of species of conservation concern. However, to obtain reliable patterns from counts, imperfect detection should always be considered, with particular reference to rare and elusive species. By analysing data from a citizen science monitoring program based on multiple simultaneous observers, we studied detection probability of the Sardinian mountain newt, Euproctus platycephalus. Detectability of individual newts widely varied among observers, and was positively affected by the number of newts exposed to during sampling. Training, although appearing to improve detectability, did not accommodate for differences among trained observers. No effect of sampling hour, tree shade, cloud cover, water flow, turbidity, and temperature was found, possibly due to standardisation of sampling conditions. Depending on observer’s skills and the population exposed to during sampling, detection probability of newt populations can widely vary. Most of the sampling units (pools) had few newts exposed to during sampling, with a high probability of recording false absences. Herpetological surveys could be more extensively based on multiple simultaneous observers to reduce observer heterogeneity bias in the detection process, and obtain more reliable patterns of species abundance and distribution for conservation purposes