<p> Amphibian species have experienced global declines since the 1970s and plethodontid salamanders are no exception. The green salamander, <i> Aneides aeneus,</i> is a plethodontid salamander that has experienced declines throughout its range in the Blue Ridge Escarpment. </p><p> Species distribution models are algorithms that predict occurrences of a species across a landscape and can be used to determine conservation priority areas. However, there are commonly only presence locations without corresponding absence locations available to a researcher. These presence-only datasets can present a challenge when trying to depict reliable distributions for a species of concern. Maximum Entropy (MaxEnt) is an algorithm empirically tested to model species distributions given presence-only datatsets. </p><p> I used landscape-level species distribution models including MaxEnt and logistic regression to model the occurrence of green salamanders across the Blue Ridge Escarpment of North Carolina. These models were used to assess particular features associated with <i>A. aeneus</i> presence as well used to search for new localities. </p><p> MaxEnt models outperformed logistic regressions for all methods of evaluation. MaxEnt models had fairly low omission (false negative) and commission (false positive) rates whereas my logistic regression had extremely high error rates for both. "Area Under the Receiver Operator Curve" evaluation scores were excellent (0.96) and good (0.81) for the top Maxent model and logistic regression, respectively. </p><p> <i>Aneides aeneus</i> is known to be associated with habitat that includes rock outcroppings with thin, deep crevices. My models indicated that forested areas, intermediate elevations, and shallow soils of particular types are desirable landscape features for <i>A. aeneus.</i> Soil was the most important variable in all models, accounting for almost half of the variation in MaxEnt models. Elevation accounted for most of the remaining variation. Percent canopy cover accounted for 4-6.5% of the variation in Maxent models. While these models did not specifically predict presence of outcrops, they were extremely helpful in identifying habitat with conditions supportive for <i>A. aeneus</i> if a rock outcrop was present. With the help of these models I discovered one previously unknown locality for <i> A. aeneus</i> and am confident addition locations can be found.</p>
| Identifer | oai:union.ndltd.org:PROQUEST/oai:pqdtoai.proquest.com:1570497 |
| Date | 31 December 2014 |
| Creators | Hardman, Rebecca Hale |
| Publisher | Western Carolina University |
| Source Sets | ProQuest.com |
| Language | English |
| Detected Language | English |
| Type | thesis |
Page generated in 0.0082 seconds