Linking bee species traits to detection probabilities in pan trap surveys

Bee community monitoring data provides a foundation for conservation and management needed to address growing concerns over bee population declines. Various bee survey methods offer different tradeoffs. Pan traps prove effective when field time is limited and standardized collections are needed. However, pan traps may introduce biases, which can affect the detection probabilities of different taxa. Understanding how detection varies across species will aid in the interpretation of the occurrence data collected. Recently, occupancy models have been applied to the analysis of bee data from netting surveys. Application to pan trap surveys could address the challenges of variable detection probabilities. Unfortunately, most occupancy models cannot simultaneously account for environmental conditions and species traits influencing detection. One solution is to conduct a post-hoc analysis of the relationship between species traits and detection probability. Our study used a multi-species occupancy framework to estimate detection probabilities for 107 bee species collected from pan traps at 143 sites around Ohio, USA, in 2020. After breaking up the data into species with similar phenological active periods, we modeled occupancy based on site characteristics (forest cover, development, and land protection status) and detection probability based on time of year. We then analyzed how detection probabilities relate to species' traits (body size, nesting behavior, sociality). Preliminary results suggest that detectability varies with time of year, body size, and sociality. By incorporating detectability into our understanding of occupancy from pan traps, we can better assess how pan-trap data inform our knowledge of bee distributions.