Ground-based counting methods underestimate true numbers of a threatened colonial mammal : an evaluation using drone-based thermal surveys as a reference

Context: Accurate and precise monitoring practises are key for effective wildlife conservation management; providing reliable estimates of spatiotemporal changes in species abundance on which sound decision-making can be based. Advancements in drone and satellite technology are providing new standards for survey accuracy and precision and have great potential for enhancing population monitoring of numerous difficult to survey species. Flying-foxes (Pteropus spp.) are large bats that roost in groups of a few hundred to many thousands in the canopies of trees, where they are difficult to census accurately and precisely by human observers. Globally, 35 of the 64 flying-fox species are listed as threatened under the IUCN Red List of Threatened Species, and reliable monitoring methods are needed for the effective management of this ecologically important group. Aims: Recently, we showed that drone-acquired thermal imagery can be used to count flying-foxes in their roost with high accuracy and precision. In the present study, we aimed to assess the accuracy and precision of whole colony counts derived from ground-based counting methods against reference counts derived from drone-acquired thermal imagery. Methods: We evaluated the relationship between ground-based counts by two groups of human observers to highly accurate and precise counts derived from drone-acquired thermal orthomosaics for 25 counts conducted across seven flying-fox roosts throughout the Greater Sydney region, Australia. Key results: We found that ground-based counts by human observers were positively correlated with those obtained from concurrent drone-acquired thermal imagery. However, drone-acquired estimates of colony size were 2.05 and 1.92 times higher than ground-based counts by the experimenter and Australian government counters respectively. When compared against drone-acquired reference counts, the precision (coefficient of variation) of ground-based counts was 26.3% when conducted by a single counter and 55.1% when conducted by multiple counters. Conclusions: Our research indicates that ground-based counting methods underestimate true population sizes by substantial margins and have limited precision. Drone-based monitoring provides highly accurate and precise population estimates, and thus is expected to yield more reliable information on flying-fox abundance and allow for trends to be established over shorter timescales. Implications: Using ground counting methods alone, population trends can only be established with significance after protracted periods of monitoring. Incorporating the use of thermal drones into current monitoring practises would enhance the capacity to detect population trends earlier and more accurately, so that conservation management can more effectively respond