INMAN, Richard D.*; NUSSEAR, Kenneth E.; TRACY, C. Richard; Univ, Reno; USGS; Univ, Reno: Elusive species: a novel approach to measuring activity and sampling availability

Many species present difficulties for studying behavior and/or abundance due to their elusiveness. Desert tortoises (Gopherus agassizii) are elusive because they remain underground in burrows for approximately 90% of the time where they can be hidden from sight. We introduce the concept of the �elusiveness filter� as a heuristic to illustrate how this cryptic behavior can obscure, and sometimes bias, studies and surveys. For example, methods for monitoring status and trends of desert tortoise populations have resulted in contentious debates about the efficacy of different methods. It appears that all methods are highly variable in the precision and accuracy of density estimates they can produce. Distance Sampling (and alternative approaches to calculate density) requires precise estimates of the availability of tortoises to be sampled (G₀). G₀ is currently estimated from samples of tortoises (focal animals), which are located periodically with telemetry throughout the sampling season. For the purpose of estimating population density, the entire sampling season is assumed to have a constant value of G₀, and as this is certainly not correct, this assumption introduces additional imprecision. Here, we describe a method for correcting for this elusiveness filter when estimating densities of desert tortoises by using time-specific estimates of G₀. Our method involves estimating tortoise availability (and activity) from models built from environmental data (relative humidity, temperature and light intensity) collected by small data loggers affixed to focal animals. Additionally, we show how tortoise availability (and activity) can be modeled as a function of a suite of climatic variables using artificial neural networks and other multivariate clustering algorithms.