Monitoring terrestrial salamanders: Repeatability and validity of area-constrained cover object searches

Terrestrial salamanders are increasingly being used in long-term amphibian monitoring programs. Issues concerning the validity and repeatability of most relative abundance indices derived from terrestrial salamander studies are poorly resolved, however. We examined the feasibility of using area-constrained searches of natural cover objects for monitoring terrestrial salamanders by comparing repeatability and power for a monitoring design that uses yearly site averages versus single annual searches of individual plots. We also examined the validity of using surface searches as measures of population decline by comparing mean surface catch on 15 plots with a mark-recapture estimate of absolute population size. We conducted annual searches of 54 plots at 18 sites in eastern Tennessee and western North Carolina for 2-3 years. Repeatability was moderate to high for both site and individual plot values (mean r(2) = 0.59-0.96 for different taxonomic groupings), and average repeatability for individual plots was only slightly lower than repeatability based on site averages for three searches per year. For a given annual sampling effort, greater power is achieved by conducting a single annual search of plots rather than repeated seasonal searches (three plots per site) at fewer sites. Surface catch was strongly correlated with the estimate of absolute population size (r(2) = 0.66-0.84; P < 0.003) for the two numerically dominant groups (P. jordani and the D. ochrophaeus complex) and for all salamanders pooled. These relationships suggest that area-constrained searches of natural cover objects generate valid indices for use in monitoring programs.