Out of thin air: surveying tropical bat roosts through air sampling of eDNA

Abstract

Understanding roosting behaviour is essential to bat conservation and biomonitoring, often providing the most accurate methods of assessing population size and health. However, roosts can be challenging to survey. Roosts can be physically impossible to access or present risks for researchers and disturbance during monitoring can disrupt natural bat behaviour and present material risks to the population e.g. disrupting hibernation cycles. One solution to this is the use of non-invasive monitoring approaches. Environmental (e)DNA has proven especially effective at detecting rare and elusive species particularly in hard-to-reach locations. It has recently been demonstrated that eDNA is carried in air and, when collected in semi-confined spaces can provide remarkably accurate profiles of biodiversity, even in complex tropical communities. In this study we deploy novel airborne eDNA collection for air for the first time in a natural setting and use this approach to survey difficult to access potential roosts in the neotropics. Using airborne eDNA we confirmed the presence of bats in 9 out of 12 roosts. The identified species matched previous historical records of roost use obtained from photographic and live capture methods demonstrating the utility of this approach. We also detected the presence of the white-winged vampire bat (Diaemus youngi) which had never been confirmed in the area but was long suspected. In addition to the bats, we also detected several non-bat vertebrates, including the big-eared climbing rat (Ototylomys phyllotis), which has previously been observed in and around bat roosts. We also detected eDNA from other local species known to be in the vicinity. Using airborne eDNA to detect new roosts and monitor known populations, particularly when species turnover is rapid, could maximize efficiency for surveyors while minimizing disturbance to the animals. This study presents the first applied use of airborne eDNA collection for ecological analysis and demonstrates a clear utility for this technology in the wild.