Examination of the Feasibility of N-Mixture Abundance Models for Monitoring Cave Aquatic Biota via Direct Observation

Philippi, Tom

Aquatic biota in cave streams occur at low densities, and can be difficult to detect. The logistics of sampling in cave streams far from the entrances can be daunting, and limit the possible sampling effort. However, cave aquatic biota are important enough indicators of cave ecosystem status that they are a priority for monitoring. Low abundance makes indices correlated to abundance very noisy, weakening the correlation between an index and absolute abundance, and increasing the need for detection-corrected estimates of absolute abundance. Hierarchical models can simultaneously estimate the detection probabilities and site occupancy or mean abundance from observations collected with imperfect detection. Such designs use removal sampling, double (or multiple) observers, or multiple within-season visits to provide information on the detection probabilities, then fit a hierarchical model for the abundance plus detection processes. CUPN proposes to use multiple within-season visits to transects in streams to estimate abundances of several cave aquatic species, and repeat that sampling across years to estimate trends in those abundances. The goals of this analysis are to determine the power or uncertainty of estimates of status and temporal trends in abundance, and to determine the optimal allocation of effort into number of transects, transect length, and number of revisits. This is a complement to and extension of Chris Sutherland’s report (Sutherland 2016), and reflects a best-case scenario where individuals don’t move in and out of transects between visits, and his other points are addressed, too.