Quantifying Floral Resource Availability Using Unmanned Aerial Systems and Machine Learning Classifications to Predict Bee Community Structure
Date of Award:
Master of Science (MS)
James P. Pitts
James P. Pitts
Joseph S. Wilsom
R. Douglas Ramsey
Jonathan B. Uhuad Koch
Bees are important for agricultural and non-agricultural ecosystems because they pollinate both wild plants and commercial crops. Flowers provide pollen and nectar resources that bees use to survive and reproduce. Measuring the relationship between the floral community and bee community may help apiarists and land managers to make informed decisions in managing wild and domesticated bee species. Manual methods to describe and count flowering vegetation is costly in time and personnel. Unmanned aerial vehicle (UAV) technology may be an efficient way to describe and count flowering vegetation on a large scale. UAVs with classification analysis and ground transect surveys were used to describe the variation in the flower communities at three field sites in non-agricultural environments. The variation in bee communities were also recorded at the field sites. Seven unique flower species were quantified using UAVs. Using the UAV imagery, it was determined that the period of flowering and changes of flower coverage for different species varied. Twenty-two unique flower species were described and counted using the ground transect surveys and 136 bees from 11 genera were recorded using net surveys. I tested the hypothesis that increased bee diversity and abundance would positively correlate with increased floral diversity and abundance using seven simple linear regression models. I found that the floral resource data collected from ground transect surveys predicts bee diversity, bee richness, and bee abundance. I also found floral abundance data captured by UAVs predicts bee abundance at the field sites. Finally, I found UAV floral abundance predicts ground transect floral abundance suggesting a positive relationship between different sampling methods. My results support previous research that suggests a high diversity of resources will support a high diversity of insects; and habitats with abundant flowers have greater possibilities for partitioning of available resources. My results also support UAVs as an efficient method for describing and counting floral resources in non-agricultural settings. Further research should include using UAV imagery to count flowers to predict bee communities on a landscape scale.
Tabor, Jesse Anjin, "Quantifying Floral Resource Availability Using Unmanned Aerial Systems and Machine Learning Classifications to Predict Bee Community Structure" (2022). All Graduate Theses and Dissertations. 8637.
Copyright for this work is retained by the student. If you have any questions regarding the inclusion of this work in the Digital Commons, please email us at .