Title: Predicting Flower Stalk Production Of A Native Shrub Using Drone-Based Structure-From-Motion Photogrammetry
Program: Masters of Science in Biology
Advisor: Dr. Trevor Caughlin, Biological Sciences
Committee Members: Dr. Megan Cattau, College of Innovation and Design (Co-Chair); and Dr. Jennifer Forbey, Biological Sciences
Climate change is threatening rangeland ecosystems, including increasing frequency of extreme weather, wildfire, and drought. Identifying which native plants are likely to be resilient to these ongoing changes is crucial for developing climate-smart restoration plans. Sagebrush is a foundational species across western rangeland, supporting populations of native plant and animal species with structure and forage. Analyzing sagebrush resilience through flowering success will help us understand how vulnerable sagebrush populations and post-restoration sites will respond to extreme weather events. We applied high-resolution remotely sensed data to map flower stalk production in big sagebrush plants. Using cost-effective unoccupied aerial vehicles (UAVs) we collected RGB imagery that enabled canopy segmentation to inform machine learning algorithms. We applied these data to quantify flower stalk production for individual plants across our 240-acre study site in Castle Rocks State Park, Idaho in 2021 and 2022. Individual plants represent three-sagebrush subspecies: Wyoming Big Sagebrush (Artemisia tridentata ssp. wyomingensis), Mountain Sagebrush (Artemisia tridentata ssp. vaseyana), and Basin Big Sagebrush (Artemisia tridentata ssp. tridentata). We found that high-resolution imagery has potential to predict flower stalk production, including an R2 of ~50%. Structural metrics, including height differences between June and September, canopy height, and edge-to-area ratio of plant crowns, were more important than spectral data for accurate predictions. Our work demonstrates the potential for UAV data collection to quantify how individual plants respond to weather events across landscape-scale environmental gradients, including an algorithm that can predict flower stalk production. Our goal is to apply these results to enable land managers to identify locally adapted sagebrush genotypes that will thrive in future climate regimes.