Green Salamander Distribution, Abundance, and Physiology in the Southern Blue Ridge

Green salamanders, Aneides aeneus, are a priority species throughout their range and have been negatively affected by habitat loss, climate change, disease, and over-collection. Many historical locations for this species in the Blue Ridge Escarpment have not been visited for ~25 years and thus were in need of a status update. I constructed both small-scale and large scale distribution models for green salamanders. For the small-scale distribution model, I conducted visual encounter surveys across three counties in South Carolina using a headlamp to search rock outcrops and binoculars to search trees. I detected green salamanders at 30 of the 61 (49.2%) surveyed sites and collected a variety of habitat variables and compared a suite of N-mixture models using an AIC framework. Time of day emerged as the most important predictors for salamander detection, while aspect, habitat size, and elevation influenced salamander abundance. It appears that there may have been a range contraction as well as local extinctions in South Carolina for this species, although low detection probability and a lack of access to some sites makes conclusions on this issue difficult to state with certainty. For the large-scale distribution models, I compared the predictions generated by a correlative-only model to those from a model with mechanistic data added to the correlative framework focusing on Green Salamanders in their disjunct range (North Carolina, South Carolina, and Georgia). I conducted a laboratory study to measure resistance to water loss (ri) and metabolism (VO2) under a range of environmental conditions. The distribution model under current climatic conditions was similar for both the correlative and correlative + mechanistic approaches. Under two different climate change scenarios, models incorporating mechanism predicted less suitable habitat than correlative-only models. Because future climate projections may include non-analog climates (a lack of appropriate training data), incorporating mechanism may be useful for forecasting climate vulnerability