Semester

Summer

Date of Graduation

2022

Document Type

Thesis

Degree Type

MS

College

Davis College of Agriculture, Natural Resources and Design

Department

Wildlife and Fisheries Resources

Committee Chair

Donald Brown

Committee Co-Chair

Kevin Oxenrider

Committee Member

Kevin Oxenrider

Committee Member

Michael Strager

Committee Member

Petra Wood

Abstract

Turtles are one of the most globally threatened vertebrate groups, largely due to habitat loss, commercial exploitation, climate change, disease, and invasive species. In the United States, the spotted turtle (Clemmys guttata) is among over 50 amphibian and reptile species recently petitioned to be listed under the Endangered Species Act of 1973 (ESA). The U.S. Fish and Wildlife Service subsequently determined the spotted turtle to be a candidate species for listing. Historically, spotted turtles were known to occur at a few locations in the eastern panhandle region of West Virginia. However, previous attempts to determine the distributional extent of spotted turtles within West Virginia did not use standardized site selection or field sampling methodologies. Furthermore, previous efforts did not benefit from using range wide distributional knowledge and high-resolution satellite imagery to guide survey site selection.

Given the spotted turtle is currently a candidate species for listing under the ESA, there is a strong need to update our distributional knowledge for the species within West Virginia. The purpose of my thesis was to conduct field surveys to validate historically known spotted turtle localities and identify new populations, and to develop a wetland-level habitat suitability model to help guide future survey efforts. In addition, I developed a model that estimates depth of isolated wetlands in West Virginia to assist researchers and managers with classifying wetlands as potential habitat for wildlife species of concern.

In Chapter 1, I provide background information on wetland water depth, methods of estimating depth, and its importance to spotted turtle habitat modeling. I also provide information on the biology and ecology of spotted turtles, and a brief summary of the known history of the species within West Virginia. Lastly, I state the goals of this thesis research and give a summary of chapter topics.

In Chapter 2, I estimated mean monthly water depth across all mapped isolated wetlands in West Virginia to allow for a more refined classification of individual wetlands as potential habitat for focal wildlife species (e.g. spotted turtle). I found that watershed scale hydrological modeling is capable of providing reasonable estimates of water depth, indicating that it is possible to remotely estimate ecologically relevant wetland hydrological conditions across large extents.

In Chapter 3, I summarize my field survey efforts to define the contemporary spotted turtle distribution in West Virginia , and provide details for a spotted turtle survey site rapid assessment form. I surveyed 62 sites across 41 unique wetlands within 10 counties in the eastern panhandle, north-central, and northern panhandle regions. Eighty unique spotted turtles were captured across 6 wetlands in 3 counties (Hampshire, Hardy, and Jefferson), all within the eastern panhandle. In October 2021, a photo of a hand captured spotted turtle near a surveyed site in Hardy county was sent to the West Virginia Division of Natural Resources, validating occupancy in Hardy County. There are currently a total of 7 known-occupied wetlands in the state. No spotted turtles were captured or encountered in the north-central or northern panhandle regions, suggesting that spotted turtle occupancy is restricted to the wider valleys of the Ridge and Valley and Blue Ridge physiographic ecoregions within West Virginia where habitat is present.

In Chapter 4, I created wetland-level habitat suitability models using presence-absence data collected from surveys conducted in Chapter 3 and presence data provided by the Virginia Department of Wildlife Resources. I tested two statistical approaches, including a regression-based approach (i.e. logistic regression) and a decision tree-based approach (i.e. random forest). I found that the logistic regression model had very little predictive power and poor overall performance compared to the random forest model. I projected habitat suitability across 17,724 wetlands within the potential distribution of spotted turtles in West Virginia using the random forest model. The model classified 16,703 (94%) of wetlands as not suitable, 634 (4%) wetlands as low suitability, and 387 (2%) wetlands as high suitability. The random forest model indicated that areas of low topographic roughness and local topographic depressions with high wetland richness were positive indicators of spotted turtle habitat suitability. Furthermore, the model suggested that presence of non-suitable landcover types was important for the classification of suitable habitat, which agreed with our survey data, where we typically encountered spotted turtles in rural landscape surrounded by exurban development or agricultural lands.

Download

Share

COinS