The influence of sample size on two approaches to estimate Black Grouse Lyrurus tetrix population size using non‐invasive sampling methods

Abstract

Population size is an important parameter to monitor for species conservation and management. This is especially important for rare and endangered species, as declines can give information about anthropogenic impacts and the need for new conservation measures. To estimate population size, various methods of analysis can be used, for which sample size is an important factor. Sample size is particularly important to consider when applying non‐invasive sampling strategies such as sampling faeces or feathers/hairs as a source of DNA, as a means to limit disturbance and stress for the species of concern. We investigated a Black Grouse Lyrurus tetrix population in the eastern part of the Alps, in East Tyrol (Austria), and estimated population size using two approaches: capture–recapture and rarefaction. With a set of 12 polymorphic microsatellite markers, we identified genotypes from faeces and feathers (backed up with 23 tissue samples) and checked for population substructure and gene flow among sampling sites. We estimated population size using four different models from the two approaches (molecular capture–recapture: TIRM, TIRMpart; rarefaction: hyperbolic function – Kohn, exponential function – Eggert). To evaluate the impact of sample size on the estimations, we used the full dataset of 500 samples (‘complete’ dataset) and half the dataset of 250 samples (‘half’ dataset). We also estimated the population size for each sex separately using complete and half datasets to check for sex‐specific differences in population size. We found similar results in three of four models (capture–recapture: capwire TIRM, capwire TIRMpart; rarefaction: rarefaction Kohn). Using just half of the data increased the uncertainties in the estimation of population size in all models used and deviations were particularly large in females, which indicated a sex bias. Only the complete dataset of males had an observation rate of more than two observations/individual, and this observation rate meets the recommendation for using the capwire models. This indicates that, for species with different sex‐specific detectability, larger sample sizes do not generally imply higher observation rates. We conclude that calculating the observation rates and population‐size estimations for each sex separately can improve overall population‐size estimation, especially in species with biased sex ratios and those that exhibit sex‐specific behaviour.