Evaluation of drone surveys for ungulates in southwestern rangelands
Jesse Blum,Aaron M. Foley,Randy W. DeYoung,David G. Hewitt,Jeremy Baumgardt,Mickey W. Hellickson,Humberto L. Perotto‐Baldivieso
DOI: https://doi.org/10.1002/wsb.1515
2024-03-21
Wildlife Society Bulletin
Abstract:Drones equipped with thermal cameras are being used to survey wildlife, but reliability of population estimates are untested. Repeated drone surveys of white‐tailed deer yielded repeatable and precise estimates that were comparable with other estimators. Future work is needed to evaluate the role of habitat characteristics influencing detectability of wildlife. Drone platforms are increasingly used for aerial wildlife surveys, but the validity of population counts has not been fully evaluated in all environments. Aerial surveys generally undercount the true population size and one must estimate detection probability (p) to correct for missed individuals. Detection probability for visual observers is influenced by vegetation and terrain characteristics, but the use of thermal cameras as the observer may also introduce additional factors influencing detection probability. We conducted diurnal, thermal‐based drone surveys for ungulates during February–April 2020 in South Texas, USA, on sites with varying degrees of woody cover and terrain. We examined histograms of georeferenced perpendicular distances to determine the effect of habitat on detection probabilities. We also examined precision in population estimates and variation in repeated surveys. Finally, we compared drone population estimates to independent estimates derived from helicopter, spotlight, and trail‐camera surveys. Distributions of perpendicular distances from the transect were affected by habitat characteristics. A flat grassland site had relatively few detections near the transect because thermal (solar) reflectance in the center of the footage obscured detections. A hilly brushy site had a uniform p (1.00) but resulted in a severe undercount due to the inability of the drone to follow contours of hills. The flat brushy site had a lower p (0.63), indicating visibility bias. Two additional flat, brushy sites were surveyed repeatedly (n = 5 and 9 surveys, respectively). All combinations of up to 3 repeated surveys were pooled to meet the recommended minimum number of detections for distance sampling analyses (≥60 detections). Population estimates had acceptable precision (CV ≤ 20%) 80.7% of the time and variation among repeated surveys was acceptable (CV = 9% and 23%, respectively). Drone‐based population estimates were comparable with estimates generated from corrected helicopter, spotlight, and trail‐camera surveys. Overall, diurnal drone surveys can generate population estimates for large ungulates on southwestern rangelands after accounting for visibility bias, but may be limited by terrain and thermal conditions.
biodiversity conservation