Can photosurveys track bee diversity? The successes and challenges of an urban community science program: Shutterbee

Nina S. Fogel, Department of Biology, Saint Louis University, St. Louis, MO, Gerardo R. Camilo, Saint Louis Zoo, St. Louis, MO and Nicole Miller-Struttmann, Biological Sciences, Webster University, St. Louis, MO

Background/Question/Methods
With the rise of affordable, high-quality cameras, researchers across the globe are collaborating with community scientists to monitor biodiversity via photography. Partnering with community scientists can broaden the ecological scope of a research project while simultaneously engaging the public in authentic science. However, unique challenges emerge to ensure high data quality when collaborating with individuals not formally trained in the field. Here, we examine whether photographic surveys conducted by community scientists accurately document bee diversity along an urban to exurban gradient in St. Louis, MO, U.S.A. In 2020, 168 community scientists were trained to conduct biweekly standardized, walking photo surveys in residential gardens as part of the Shutterbee Community Science Program. The bee communities of 37 gardens that occurred along the urbanization gradient were also surveyed via aerial netting. We tested for a correlation between estimates of bee diversity, richness, and evenness as quantified by photographic and netting surveys. To test if there was a size-based bias in survey results, each genus was categorized as “small” or “large” bodied (i.e., below or above 15 mm, respectively). Similarity between photographic surveys and netting surveys for each body size group was quantified via Jaccard Distance and compared via ANOVA.
Results/Conclusions
Eighty-nine percent of the 8209 observations submitted to Shutterbee iNaturalist project were identified to genus or a lower taxonomic level. Photographic surveys provided similar, though not equivalent, patterns in bee diversity relative to aerial netting surveys. Estimates of bee diversity and richness by photographic and netting surveys were correlated, though the correlation coefficients were relatively low (r = 0.32 and r = 0.44, respectively). Evenness was not correlated between photographic and netting surveys. Similarity between photographic and netting surveys varied considerably among the community scientists, and on average, was higher for large bees relative to small bees, indicating that community scientists are not reliably documenting small-bodied bees. We discuss the implication of these findings for biodiversity monitoring of mobile insects by community scientists. We explore techniques to identify “high performing” community scientists and those who require additional training and/or assistance to increase the accuracy of their survey data. By identifying and addressing barriers to successful participation, community science projects may train, retain, and theoretically recruit a greater diversity of higher performing participants, while simultaneously improving data quality.