The quality of citizen scientists’ bee observations: An evaluation of PollinatorWatch at Royal Botanical Gardens and the rare Charitable Research Reserve
MetadataShow full item record
Citizen science engages members of the nonscientific community in academic research, contributing to our collective knowledge of the natural environment through biological monitoring and environmental observations. Observation plots are often used to assess pollinator diversity and abundance in citizen science monitoring programs. To ensure that data collected are reliable, citizen observations should be evaluated against controlled scientific studies. I designed this project to assess the accuracy of citizen observations of bees in order to enhance the efficacy of PollinatorWatch, a Canadian pollinator monitoring program. PollinatorWatch engages volunteers in collecting observational data on bees visiting flowers but the program’s effectiveness at reporting on bee faunal information has not been evaluated. Specifically, I was interested in determining how PollinatorWatch could be standardized to validate the efforts of participants. Research took place in mixed meadow habitats at two urban conservation areas, the Royal Botanical Gardens in Burlington, ON and the rare Charitable Research Reserve in Cambridge, ON. I trained 19 citizen scientists to observe and record bees visiting flowers using broad species-groups based on recognizable features (e.g. Green bee) or familiar bees (e.g. Bumble bee). Over the course of one summer, I conducted a survey of bees using pan-trapping and sweep netting at eleven sites. I collected 1864 bees of 74 species, verified by experts. Additionally, volunteers made observations at six of the eleven sites. To evaluate the reliability of citizen science data, I compared observations (observation data set, 590 bees) to specimens (specimen data set, 1041 bees) collected from the same sites. I found positive correlations in bee abundance among the two data sets (Spearman’s ρ ranged from 0.8 to 1, p-values 0.017 to 0.333), though information collected by volunteers was more robust over the long-term (season-wide observations) than the short-term (single observations). Observations more closely matched netted + pan-trapped bees than netted bees alone but observers recorded approximately half as many bees as were collected. Discrepancies between observational and specimen-based data were greatest for species-groups that lumped a large variety of bees (e.g. Small bee), so I propose changes to the PollinatorWatch protocol to reduce identification errors. Although the scope of this project was limited by the number of participants and the habitats surveyed, I suggest that PollinatorWatch can be improved by further studies that examine a revised, standardized observation protocol that would serve to improve data quality. In this way, citizen science contributions may more reliably complement more localized, hypothesis-driven bee research while also enhancing participants’ own understanding of environmental monitoring.