Assessment of rapid sampling methodologies for pond-breeding amphibian assemblages

Background/Question/Methods

Amphibian assemblages make up critical ecological components of many wetlands ecosystems as both predators of invertebrates and prey of larger vertebrates, and they often exceed the combined biomass of other terrestrial vertebrates within the system. Additionally, amphibians can serve as useful bioindicators of environmental change because they are sensitive to environmental and habitat alteration. Because of their importance in wetland ecosystems, there is a need to develop rigorous but rapid sampling methods to assess wetland function based on amphibian diversity. In order to determine the most appropriate sampling strategy, we sampled 80 wetlands in east-central Illinois during the 2012 field season (February-July). Each wetland was sampled once every five weeks (over a twenty week period), with three samples during each week (12 total samples). We then evaluated the effect of removing sample periods on species detection estimates and wetland conservation coefficients (an indicator of amphibian diversity). 

Results/Conclusions

We found that decreasing the total number of samples from 12 resulted in a significant increase (p = 0.046) in the standard error values for detection estimates as well as wider (and less precise) confidence intervals. Additionally, as sample number decreased, detection estimates consistently increased. When fewer than 12 samples were used, wetland conservation coefficients were significantly underestimated (p < 0.001). In lieu of a generic metric such as species richness, wetland conservation coefficients provide an adjustment for a species’ tolerance of disturbance, its rarity, and its distribution. Overall, the sampling regime that included 12 samples had conservation coefficient values 8.41% higher than sampling regimes with 9 or fewer samples (range 6.82-10.29%). Our results indicate that rapid assessment methodologies for amphibian assemblages with fewer than 12 samples in a season may overestimate detection values (which can lead to faulty occupancy models) and underestimate wetland conservation coefficients.