Refining sampling protocols for cavefishes and cave crayfishes to account for environmental variation

Subterranean habitats support a diverse array of organisms and represent imperative habitats in many conservation strategies; however, subterranean habitats are one of the most difficult environments to study. Accounting for variable sampling detection is necessary to properly evaluate conservation options for rare species such as karst and other groundwater organisms. New sampling methods, such as environmental DNA, show promise to improve stygobiont detection; however, sources of sampling bias are poorly understood. Therefore, our objective was to determine factors affecting detection probability of both visual and environmental DNA (eDNA) surveys for cavefishes and cave crayfishes. We sampled 40 sites across the Ozark Highlands ecoregion in Arkansas, Missouri, and Oklahoma, USA using visual and eDNA surveys. We used occupancy modeling to estimate the detection probability of the two taxa using both survey methods under varying environmental conditions. Overall, eDNA sampling resulted in higher detection probability for cavefishes when compared to visual surveys, whereas visual surveys typically had higher detection probability for cave crayfishes. Greater water volume at the time of sampling was related to lower detection using visual surveys for both taxa, but there was no relationship between eDNA detection and water volume. Detection probability of both cavefishes and crayfishes was higher using visual surveys when sampling units were classified by coarse rather than fine substrate, whereas detection of cave crayfishes surveyed using eDNA was higher in coarse substrate environments. Detection of cavefishes and cave crayfishes was higher via eDNA sampling when water was flowing, but similar sampling conditions resulted in lower detection using visual surveys. Our results indicate detection should be considered when sampling stygobionts even if using traditional visual surveys. Environmental DNA is a useful tool; however, the limitations we identified indicate eDNA for these taxa currently are not adequate to replace traditional surveys in subterranean environments.