Detecting spatiotemporal changes in harvester ant disks in response to fire and rodent competition using small unoccupied aerial systems imagery

North American deserts are changing due to expansion and increasing intensity of human activities. Disturbances and landscape change affect harvester ants, a keystone species in dryland ecosystems. Imagery acquired from small Unoccupied Aerial Vehicles (sUAVs) was analyzed to assess above-ground ant responses to change over time in the context of disturbance ecology and biological community interactions. We assessed the effectiveness of using sUAV imagery to detect spatiotemporal trends in ant disks over time and evaluated how ant disk density and area changed due to fire and rodent exclusion. In 2011, we implemented a controlled experiment in Tooele County, Utah, USA, to test the effects of experimental fire and rodent reduction on desert communities. We applied supervised object-based image classification to sUAV images from 2016, 2021, and 2023 and assessed classification accuracy and how fire and rodent removal affected ant disk density and area over time. Mapping accuracy of ant disk detection using sUAV imagery was consistent across all years. Our classification method detected ant disks at a lower density than manual methods. We failed to detect ant disks < 2.5 m in diameter; however, disks above 2.5 m were accurately identified. Burning increased disk density by 2.33 disks per plot, and rodent exclusion increased disk density by 1.73 disks per plot. Ant disk area increased by an average of 5.7 m2 per plot per year but was not affected by fire or rodents. sUAV imagery can be used to monitor harvester ants’ response to disturbances, with limitations. Our remote sensing methods have the potential to assess dryland ecosystem resilience to abiotic and biotic change by evaluating the responsiveness of harvester ant communities using mound and disk characteristics.