Boone, Michelle2024-07-242024-07-242023-05https://hdl.handle.net/11299/264289University of Minnesota Ph.D. dissertation. May 2023. Major: Entomology. Advisors: Sujaya Rao, Daniel Cariveau. 1 computer file (PDF); xx, 155 pages + 5 supplementary files.Insect declines are of mounting concern, yet evidence for widespread declines is limited due to a lack of standardized, long-term datasets. Furthermore, practitioners often fail to account for heterogeneity in the detection process when making inferences from survey data. As more insects are petitioned for listing under the Endangered Species Act, it is imperative that monitoring schemes implement standardized sampling protocols and adopt analytical methods that account for imperfect detection of target species during surveys. To optimize sampling for endangered insects, we must better understand the effects of biotic and abiotic factors on occupancy and detection probability of target species. In addition to understanding key factors that influence detectability of rare species to improve monitoring, it is crucial to identify habitat preferences of endangered insects to implement effective recovery and conservation planning. Bumble bees (Hymenoptera: Apidae: Bombus) are among the insect taxa with the best evidence of widespread declines. Bumble bee declines were reported in the United Kingdom as early as the 1970’s. In North America, the rusty patched bumble bee (B. affinis Cresson) was listed as federally endangered in Canada in 2012 and the United States in 2017. Franklin’s bumble bee (B. franklini Frison) was also listed in the U.S. in 2021. Additional species have been petitioned and are under consideration for listing. This has led to calls from scientists for a national bee monitoring framework to support conservation planning. For my doctoral research, I investigated the relationships between site, weather, and survey covariates with detection and occupancy probabilities for a suite of bumble bee species in Minnesota, USA, including B. affinis. I also investigated associations between habitat type and occupancy and detection probabilities. I collected bumble bee data during the summers of 2018, 2019, and 2021 for three distinct studies and used single-season, multi-species occupancy models to address my research questions.Assessing factors related to Bombus occupancy and detection probabilities is an emerging area of research, thus basic information about these relationships is sparce. For my first study, I investigated the effects of impervious surface and floral area on Bombus occupancy probability, and whether date, time of day, or observer were related to detection probabilities. I conducted roadside surveys in the seven-county metropolitan area of Minneapolis-St.Paul. This study was among the first to quantify detection uncertainty for B. affinis and to use multi-species occupancy models to draw inferences about imperiled Bombus detection. For the second study, I tested the efficacy of single-species versus multi-species occupancy models for estimating species-specific detection and occupancy probabilities. I also investigated whether bumble bees in our study system exhibited associations between occupancy probability and landscape habitat type (developed, natural, and agricultural) from surveys conducted across the state of Minnesota. I found that developed habitat had the most variable effect on Bombus occupancy probability out of the three landscape types investigated, which led to my third study, in which I delved deeper into developed landscapes. In this study, I investigated whether detection probabilities were related to adjacent habitat type (woodland, wetland, or crops) in the heterogenous, mixed-use landscape of Washington County, one of the seven metropolitan counties that serve as the human population center of Minnesota and which represents a transitional zone from a dense urban environment to a rural agricultural landscape. The results of my dissertation research provide some of the first investigations into the effects of biotic and abiotic factors on the bumble bee detection process. This information can be used to develop survey and monitoring protocols for community and species-specific monitoring. The results can also inform recovery monitoring for the endangered B. affinis or be adapted to investigate factors that influence the detection process for other imperiled insects. The finding that B. affinis occupancy probability is associated with developed areas lends support to the idea that urban areas may provide important refuge for this species in our study region. The associations between Bombus detection probabilities and survey, weather, and site covariates can be used to optimize sampling design for rare species. Taken together, this dissertation provides a framework for imperiled species monitoring that accounts for uncertainty in the detection process and can be adapted to other insect species.enBombus affinisbumble beesdetection probabilityendangered speciesmonitoringoccupancy modellingThesis or Dissertation