(1st Place) Taxonomic bias of bee sampling methods varies seasonally and between landscape types

Representative sampling of entire wild bee communities is a common goal of both pollinator monitoring efforts and ecological studies, but common active and passive sampling methods are not equivalent in their abilities to detect different bee taxa. While reported biases with respect to particular bee taxa vary among previous studies, the drivers of this variation are unclear. We analyzed 25,324 bee specimens captured by paired sampling using pan traps and either blue vane traps or hand netting across varied landscape types and times of year in New Jersey, New York, and Pennsylvania. To establish whether sampled bee communities differ between pan trap and each alternative method, we performed rarefaction randomization tests to compare species richness, composition, and relative abundance of sampled communities. We then used linear modeling to evaluate whether biased sampling of individual bee genera between each pair of methods is consistent across seasons and between sites in predominately forested and predominately open landscapes. For both pairs of comparisons, sampled assemblages of bees differed significantly between methods. The degree of bias between pan traps and vane traps differed significantly between open and forested areas for two of five common bee genera. Lastly, four of these five genera showed significantly different degrees of bias between pan traps and vane traps at different times of year. These results support cautious use of bee community data, suggesting that detectability of individual genera, even by a specific sampling method, may not be constant between locations or between surveys conducted in different seasons.