Use of vocalisations to estimate population size of roe deer

Estimating population abundance of large mammals generally requires substantial time and effort, and low-investment alternatives are needed. We present a novel application of a capture-mark-recapture (CMR) method, using vocalization frequency to estimate the size of roe deer (Capreolus capreolus) populations. The technique involves walking along fixed transects to disturb all animals present on a study plot, potentially provoking a vocal response. Those animals heard to vocalize (whether observed or not) are then considered the total number of 'marked' individuals in the population. The proportion of 'marked' individuals in the population is estimated from the proportion of animals that vocalize in the subsample of individuals observed (the vocalization frequency). Population size is estimated by dividing the number of marked individuals by the vocalization frequency, correcting for bias, which is directly analogous to the Lincoln-Petersen (L-P) estimate for CMR. We used this method to estimate population size for roe deer inhabiting a 150-ha forest plot for 8 separate surveys, and we used the L-P estimator to compare our estimates to mark-resight estimates. We estimated deer density as 23.45±7.80 deer/100 ha by CMR and 19.87 ± 7.92 deer/100 ha by the vocalization frequency (VF) method (x¯ ± SE). If the assumptions are met, this method provides estimates of absolute population size at low cost and with little material investment, because physical capture and marking of animals is not necessary. Further research is required to validate the assumption that visually observed animals have the same probability of vocalizing as those disturbed but not seen.