Inferring red squirrel (Sciurus vulgaris) absence with hair tubes surveys: a sampling protocol

Hair tubes are often used to monitor red squirrel in fragmented landscapes, where presence/absence data are gathered to determine its distribution and factors affecting it. Despite many applications and evaluation of this technique for density estimation, the reliability of absence data has been overlooked, as no rigorous statistical estimate has been attempted both on the survey duration and on the reliability of absences. Accurate determination of the duration of a survey (e.g. how many visits should be carried out to consider the species absent rather than non-detected) will affect total costs and number of monitored sites; moreover, false absences will bias the distribution estimates. By applying some recently developed occupancy models, we estimated detection probability and sampling size required to infer red squirrel absence. Application of this sampling and data analysis protocol allows to infer the species absence at a reasonable cost and thus to evaluate the reliability of a presence/absence dataset.