Ever search for something, not find it, but wonder if you simply over-looked it? We investigated this problem for amphibian and reptile surveying. Declines in amphibian and reptile (herp) populations resulting from human activities have been reported from many regions of the world. Development of effective inventory and monitoring programs is of fundamental importance in determining the status of species and documenting changes in their abundance and geographic distribution, especially in response to rapid climate change. However, data on the effectiveness of various survey methods are scarce. We tested the effectiveness of several herp monitoring methods in the Lake Michigan Basin, performing intensive surveys and over-sampling to develop detection probabilities and minimum sampling requirements for each method and species for use in proportion of area occupied modeling. This methodology allows the use of data from existing, less robust, monitoring programs for regional analyses (such as calling frog surveys), and identifies overlaps in methods for detecting species to guide monitoring program development.
We detected up to 20 species per sampling area. Call surveys, aquatic funnel traps, and casual observations detected the most species. Method success varied with the species present, and some species require specialized methods. Detection probabilities were > 0.3 (recommended for percent area occupancy modeling) for 8 anurans by call surveys, 11 amphibians and 1 snake by aquatic funnel traps, 6 anurans by casual observations, 3 snakes by cover object surveys, and 2 turtles by baited hoop nets. The minimum number of samples required for 95% confidence in detection was calculated for each species and method. For call surveys, one survey detected only the spring peeper at this confidence level, whereas two surveys detected an additional 3 anurans (5 more at 90% confidence). For aquatic funnel traps, five trap nights detected 3 salamanders, 5 anurans, and 1 snake, while eight trap nights detected an additional 3 anurans. For baited hoop nets, eight trap days detected 2 turtles. Ten cover object sampling days detected 4 snakes. In our study area, sampling recommendations include: call surveys for anurans (two samples per calling period), aquatic funnel traps for salamanders (5 trap nights minimum), and cover objects for snakes (10 sampling days minimum). Obtaining detection probabilities, understanding variance in detection probabilities among sites, species, and over time, and correcting for false negatives in data analyses where species are under-sampled, is recommended for any inventory and monitoring programs.