PS 4-57 - Mass mortality: Using UAV's to understand a die off of sclerocactus on the Colorado Plateau

Abstract: Sclerocactus are among the most visually striking and charismatic species on the Colorado Plateau and are an important pollen resource for insects. Low summer precipitation and nearly absent winter snow cover in 2020-2021 contributed to a significant mortality event in the sclerocactus communities of Millcreek Canyon, east of Moab, Utah. Although regional drought is presented as the main driver for mortality among dryland plants, other biophysical factors may also be contributing to the underlying patterns of cactus loss and topography and community structure may modify the influence of drought on plant survivorship. Individual cactus size and clustering can introduce competition, and the physical ground cover (rocky, biocrust, dirt, vegetation) might be indicators of differences in localized soil climate conditions. Our goal in this work is to overlay these biophysical metrics over sites that differ in cactus survival and mortality. We hypothesized that cactus sites with a southern exposure on rockier soils will have higher rates of mortality due to warmer microclimates and lower water availability. We also hypothesized that mortality increased as size of individuals decreased, with size being a proxy for resource storage. To address these hypotheses, we will utilize UAV imagery to identify and georeference all living and dead sclerocactus, in addition to installing RFID identification tags for ongoing monitoring. We have conducted initial surveys in the Millcreek Canyon Wilderness study area in 2022-2023 and surveyed over 1500 sclerocactus individuals, while final UAV flights will be carried out Fall 2023. We recorded size class, clustering patterns, and mortality status for each cactus, then used RFID identification to coordinate these mortality and size values with precise geographic locations using a single-band RTK GNSS receiver. We used ArcGIS Pro to analyze the impacts of slope and aspect, and neighborhood density. The results from this groundwork have shown that there is no significant difference in mortality patterns across slope, aspect, and clustering, which refutes our first hypothesis about landscape driven mortality. To further evaluate this die off, we plan on building a high-resolution model of elevation and terrain (cm-scale accuracy) using overlapping images acquired with a UAV, allowing us to implement a supervised classification algorithm to classify the ground cover of the study area. This follow up work will be centered on testing our second hypothesis, the concept of ground cover being indicative of different soil-growth conditions.