The use of drone-based aerial photogrammetry in population monitoring of Southern Giant Petrels in ASMA 1, King George Island, maritime Antarctica

Abstract

Southern Giant Petrels (SGPs) are surface nesting birds with a circumpolar Southern Hemisphere breeding distribution. The species tends to have no natural enemies on land, but is sensitive to human disturbance. The search for new methods is crucial and may minimize or exclude stress and risk of nest disturbance, related to ground-based research activities. The aim of this study was to conduct a population census of the SGP of the Antarctic Specially Managed Area no. 1 (ASMA no. 1), Admiralty Bay, King George Island, using an unoccupied aerial system (UAS) based on aerial photogrammetry and to determine the optimal parameters of the aerial mission for the identification of SGP adults and chicks on orthophotos while simultaneously not causing behavioural changes. To this end, in a preliminary survey in the 2019/20 season, the locations of all breeding areas for SGPs in ASMA no. 1 were determined, and the presence of 3 colonies, Llano Point/Rescuers Hills (LP/RH), Vaureal (V) and Petrel Hill (PH), was confirmed. Terrain models for two of the colonies (LP/RH and V) were established, and the flight parameters for the next season were determined. In 2020/21, a total of 23 (DJI Inspire 2 with a Zenmuse X5S camera) drone missions were performed at various stages of the breeding period over the LP/RH and V colonies. This assessment yielded estimation of the number of active nests and chicks over the entire ASMA no. 1 area, and included 508 active nests and 380 chicks for the 2020/21 season. To determine the minimum flight altitude at which no SGP behavioural response was observed, an experiment was performed that showed the vertical distance between the potential nest of SGP and the drone should be greater than 21 m given that lowering the altitude yielded statistically significant differences in bird behaviour. Image analyses showed the possibility of identifying adults and chicks at a ground sampling distance of 2.15 cm, which corresponded to an altitude of 130 m based on the equipment used and the terrain characteristics. The proposed method requires several missions during the incubation phase to determine a reliable number of active nests without using correction factors. To obtain a nesting success factor, it is recommended to perform at least one raid in the post-brooding phase of chick rearing (when the chick is not covered by an adult and is visible in orthophotos). The proposed method is not able to replace traditional methods in the context of many ongoing surveys, but we believe that it may provide a less bird-invasive, human-intensive and time-consuming option to replace ground-based census surveys.