Evaluation of four different methods for assessing bee diversity as ecological indicators of agro-ecosystems

Monitoring of wild bees is becoming more and more popular in nature conservation because of the high indicator value of this insect group. However, uncertainties about the sampling performance of different methods still exist, especially in areas of limited accessibility. We therefore compared four commonly applied sampling methods across vineyard fallows in a species diverse study area over two successive years: hand netting along variable transects (HN), pan traps (PT), trap nests (TN) and Malaise traps (MT). The chosen method significantly affects the number of sampled bee species and individuals, with PT sampling by far the largest number of species and individuals, and HN sampling the most diverse bee fauna. HN samples contained a significantly higher proportion of males, red-listed, large and social Bombus species than PT, but HN and PT samples contained a significantly lower proportion of male individuals compared to MT and TN. PT colour had a significant effect on the number of sampled individuals and species, with yellow PT sampling the largest numbers, while blue PT sampled the largest number of individuals of social Bombus species. The HN sampling results of an experienced and a less-experienced observer differed remarkably, with the turnover component of the Jaccard distance being significantly higher compared to the nestedness component. Our findings indicate that PT was the most efficient method for sampling bees in our study system. Due to species-specific differences in attractivity, sets of different PT colours should be used. However, if the study focus is on red-listed species or male individuals, HN represents a more efficient method. When HN is applied, observer bias should be considered as much as possible, especially with regard to differences in sampling experience. Due to different shortcomings, MT and TN cannot be seen as appropriate methods for standard monitoring of bees.