A comparison of uncrewed aerial vehicle and on-ground surveys of penguin populations on the Antarctic Peninsula

Monitoring changes in wildlife populations over time requires regular assessments using robust methodologies. For many species of seabirds, including Antarctic penguins, population assessments have traditionally involved on-ground methods using standardised approaches. Recently, uncrewed aerial vehicles (UAVs) have emerged as a complementary tool to address challenges with traditional methods. To understand any potential impacts of this methodological shift on long-term datasets, we compared on-ground counts with two independent counts from UAV-derived imagery from 12 sites across the Antarctic Peninsula. We also investigated changes to breeding phenology at one site on the Peninsula, to help to contextualise other factors that influence population sizes across different temporal scales, as UAV surveys are usually conducted opportunistically and so may not capture peak populations. Whilst on-ground and UAV counts generally showed close comparability, only half of the counts fell within the generally accepted 5% repeatability threshold. UAV counts tended to be higher and more consistent with each other than on-ground counts, suggesting the UAV counts were more accurate, and may have captured parts of colonies missed during on-ground surveys. We observed significant variation in breeding phenology, and a negative correlation between survey timing and counts of breeding success. Through comparing our counts with historic data, our findings confirm gentoo penguin (Pygoscelis papua) population growth across the Peninsula, and showed mixed results for chinstrap (Pygoscelis antarctica) populations. We discuss potential sources of discrepancy between the methods and considerations for using UAVs in penguin surveys, particularly for informing management decisions.