Vantage point counts and monitoring roe deer

Vantage point counts (VPCs) are adopted to estimate the densities of roe deer (Capreolus capreolus) populations for harvest management. Thus, counts should be performed only within blocks without woodlands where it is possible to relate counts to block sizes. Alternatively, if VPCs are simply carried out on all blocks in a study region, the expectations of total counts could be used as relative abundance indices. In most cases, surveying all blocks is too demanding because of the number of observers required, time, and organization. Therefore, VPCs are performed only on a portion of the blocks, and relative abundance indices are estimated from these counts. If the blocks are selected by means of probabilistic sampling schemes, then statistically sound estimators of total count expectations can be adopted. Therefore, the estimation of the sampling errors, construction of confidence intervals, and assessment of change are possible, together with a post hoc power analysis for evaluating the probability of failing to detect a change in the expectations. Our objective in this study is to consider sampling strategies that allow the performance of all these statistical steps and to check the performance of these strategies on a hunting district located in Tuscany, Italy, in which all the blocks were surveyed in 2013 and 2014. The results provide evidence of the imprecision of the estimators. Even for large sampling fractions of 40-50%, the relative standard errors never decreased below 20%, and the corresponding powers in detecting a change of 30% at a level alpha = 0.05 were <0.65. Our results highlight the need for efficient and robust alternative strategies.