A case of underestimation of density by direct line transect sampling in a hunted roe deer (Capreolus capreolus) population

Ungulate population estimates from direct line transect counts are sometimes considered unreliable, but the causes and the degree of potential biases are often unknown. In this study, we compare population estimates of roe deer Capreolus capreolus from line transect data with those from camera traps and harvest statistics. Using a systematic line transect sampling approach, we walked 22 transects covering 1026 km 2 in Central Germany (198 km survey effort). We also used random encounter modelling (REM) to estimate roe deer populations from data collected with 20 camera traps which were primarily installed to assess a recently established population of lynx Lynx lynx. From line transect data, the density of roe deer was estimated at 2.7 individuals per km 2 (95% confidence interval CI, 1.4–5.4). Using camera trap data, REM estimated much higher roe deer densities, which ranged from 14.6 (CI 6.8–24.5) to 19.4 (CI 8.7–36.2) individuals per km 2 . Estimates obtained from camera trap data and REM are consistent with harvest statistics and density estimates from other European forest landscapes. Moreover, a careful examination of the detection probability plot from line transect data suggests that evasive movement of roe deer could have biased density estimates derived from the line transect survey. In our case study, this bias could have led to an underestimation in the hunted roe deer population in the range of 81 to 86%.