Connor J. Hughes
Of the turtles native to the United States, few are as rare or as cryptic as the Bog Turtle (Glyptemys muhlenbergii).
The federally threatened Bog Turtle is endemic to the Eastern United States and is the country’s smallest and rarest freshwater turtle. It has one of the most restricted habitat requirements of any reptile, inhabiting the muck of small and isolated spring-fed wetlands across its range. These wetlands are rare and scattered, and the factors which drive Bog Turtle presence in suitable wetlands are poorly understood. To make matters even more difficult, Bog Turtles have very low detectability even in occupied wetlands, making them difficult to find in already hard to find habitat (Somers and Mansfield-Jones 2008). The Bog Turtle’s cryptic nature, rarity on the landscape, low detectability, combined with incomplete knowledge about the species biology and distribution, make researching and managing Bog Turtle populations challenging.
As such, many states recognize expanding their knowledge of the distribution of potential Bog Turtle habitat and surveying to assess site occupancy as top conservation priorities. Manual surveys for habitat are often biased towards areas that are close to roads and easily accessible, leaving large gaps in coverage. This is further complicated by the fact that much of the land that would support these rare wetlands in the Southeast is privately owned, adding the challenges of obtaining permission to access properties from multiple individual landowners, and making manual searches for this species and its habitat particularly difficult.
The solution is a high-quality habitat model for the species.
Habitat modeling is a powerful tool in ecological research that works by analyzing environmental variables to identify patterns and factors influencing species occupancy across the landscape, and allows us to predict and map other potentially suitable habitats for a given organism (Guisan and Zimmermann 2000). This provides valuable insights into species distribution and environmental preferences and contributes to informed conservation and management strategies. However, one challenge that comes with building large-scale habitat models is picking biologically relevant variables. This requires both an understanding of what aspects of habitat are important to the natural history of the species, and the ability to model those aspects using widely available data. Too often these critical aspects of habitat cannot be modeled on a large scale using broadly available data. For a Bog Turtle important habitat characteristics could include nesting habitat comprised of appropriate microtopography and vegetative community, or soils and hydrology suitable for overwintering and daily life. It is then necessary to find robust proxy variables which are available at that scale, usually in the form of remotely sensed data.
We wanted to start our habitat modeling process by really understanding what aspects of the environment were important to this species’ natural history and how to model them. After an in-depth literature review and consideration of previous work done in the lab, wetland hydrology was deemed a high priority. Previous work on the species had found that there were differences in water table depths between wetlands that supported Bog Turtle populations and those that were either only transiently occupied or unoccupied by the species (Feaga et al. 2012). Further research found that soil water saturation was an important thermal buffer in these wetlands, with model hibernacula constructed in unsaturated soils dipping below freezing and potentially posing a danger to overwintering turtles, and hibernacula constructed in saturated soils typically staying above freezing even during the cold days of winter (Feaga and Haas 2015). These findings present an exciting potential way to differentiate between suitable and unsuitable wetlands but have thus far been limited in application due to the challenges of measuring the groundwater table of many wetlands at a large scale. But if some more broadly available proxy variable correlated with subsurface wetland saturation, it would be a way to include this biologically relevant aspect in a habitat model and help us better understand where this species could occur on the landscape.
Installing groundwater monitoring wells at sites in December 2023 – Connor Hughes
As part of my research, we are expanding on those original studies in an attempt to find a way to use remote sensing as a proxy for seasonal variation in groundwater saturation across different types of wetlands. We know there’s a difference between sites, but is remote sensing powerful enough to pick it up? Remote sensing incorporates a variety of tools, and we are currently looking at several different indices related to moisture or wetness to understand seasonal change across these wetlands. One exciting quality of remote sensing is that it gives us the ability to look into the past, and we can actually match our remotely sensed variables with the timeline of when the data were being collected, including groundwater data from past studies. For my work, we are calculating remotely sensed metrics of interest on a monthly basis at 8 sites from the original study to try to link remotely sensed metrics with ground-truthed data. We’re also adding 10 previously unused sites to the study, to more than double our sample size. Since groundwater level can vary substantially even within a wetland, we have fit each new site with a minimum of 3 groundwater monitoring wells to get a more holistic view of the wetland’s hydrology and mimic the study design of the original work.
Sites are categorized as either ‘occupied’ or ‘unoccupied,’ with the latter emphasizing areas that look visually like they could be appropriate habitat, but on the ground appear to be too firm or lack appropriate hydrology for Bog Turtles. The ability to deem sites like these unsuitable during the modeling process would increase model specificity and survey efficency. Getting access to privately owned wetlands can sometimes be a long and complicated process, and site suitability can sometimes only be apparent after getting into a wetland. Filtering out unsuitable sites from computer-based search efforts before going through the process of landowner permission would save time and money, but we need a strong link between a remotely sensed metric and ground-truthed data and the confidence that it will not filter out actually suitable sites before we put these metrics into application. Our preliminary results seem promising, especially during the critical winter months.
There is still significant work to be done on this project, and we are in the process of determining the role these variables will play in future habitat models, but we are excited about the promising preliminary results that show a correlation between groundwater depth and these remotely sensed variables. Our new wells were just installed and will be collecting data through 2024. We hope to establish these remotely sensed variables as legitimate proxies for Bog Turtle habitat.
Longterm, assuming the observed trends remain consistent, we plan to integrate these variables into future habitat models for the species. By combining this variable with other well-established factors, we aim to enhance the specificity of our models. This will contribute to a better understanding of both the spatial and temporal dynamics of the species’ habitat, addressing both long-term trends and seasonal variation.
Funding for this project was partially provided by the Virginia Department of Wildlife Resources through the U.S. Fish and Wildlife Service’s State Wildlife Grant program, The Orianne Society, and Virginia Tech.
Feaga, J. B., and C. A. Haas. 2015. Seasonal Thermal Ecology of Bog Turtles (Glyptemys muhlenbergii ) in Southwestern Virginia. Journal of Herpetology 49:264–275.
Feaga, J. B., C. A. Haas, and J. A. Burger. 2012. Water Table Depth, Surface Saturation, and Drought Response in Bog Turtle (Glyptemys muhlenbergii) Wetlands. Wetlands 32:1011–1021.
Guisan, A., and N. E. Zimmermann. 2000. Predictive habitat distribution models in ecology. Ecological Modelling 135:147–186.
Somers, A. B., and J. Mansfield-Jones. 2008. Role of Trapping in Detection of a Small Bog Turtle (Glyptemys muhlenbergii) Population. Chelonian Conservation and Biology 7:149–155.i