The Georgetown Salamander (Eurycea naufragia) is a species of conservation concern primarily due to its limited distribution and vulnerability to habitat loss. This species is permanently aquatic and only lives in springs and caves in central Williamson County, Texas, which is one of the fastest growing counties in the country. Land use changes in the catchment areas of streams can threaten the health of spring and stream ecosystems.  Several studies have shown an association between increasing urbanization and impervious cover with declining stream health. Additional studies have shown that indicator species, such as salamanders, decline in response to the same variables associated with stream health. This study attempts to use the strong association between impervious cover and stream abiotic and biotic health to analyze the habitat vulnerability of the Georgetown salamander (Eurycea naufragia), by using imagery data and GIS software (ESRI ArcMap) to examine the past and current land use in catchment areas surrounding springs in Williamson County with known populations of salamanders. Using GIS software has become more common in conservation projects due to the ability to conduct spatial analyses and gain information without needing travel to specific locations. Since the proposed endangered listing of the Georgetown salamander is a highly controversial issue, visits to all of the known localities were not feasible, necessitating the use of GIS software to begin vulnerability analyses for the habitats.

To look at the potential impacts of land use change on the Georgetown Salamander, I used GIS software and LiDAR elevation data to model the catchment areas for 12 springs with known populations of salamanders. I then used normalized difference vegetation index (NDVI) models and image classification software (Maximum Likelihood Classification tool) to look at past and present land use and impervious cover. I used imagery from 1996 and 2010 to conduct the NDVI models, and imagery from 1996, 2004, 2008, and 2012 for the image classification software. The percentages of impervious cover were calculated for each modeled catchment area.

Impervious cover is commonly used as a measure of urbanization and has been shown to have negative effects on stream health at as little at 10% impervious cover. Of the 12 springs I modeled, by 2012, 5 springs have already surpassed this 10% threshold and a few others are in highly modified landscapes. Several of these springs have recently undergone changes to land use in the catchment area of the springs, as both models showed only 2 springs having greater than 10% impervious cover in 1996.  This would suggest that almost half of the springs modeled in this study may already be experiencing negative impacts of urbanization. Unfortunately, long-term water quality data does not exist for any of these springs, making it difficult to evaluate this hypothesis.  Additionally, water quality of springs is influenced by groundwater dynamics including the local geology, size of the recharge zone, and outflow rates. Imagery data does not include these groundwater dynamics, which remain largely unknown for the area. Groundwater is affected by the entire recharge zone, which likely includes a much larger area than the surface drainage alone. Predicting the effects of impervious surfaces on groundwater would require groundwater modeling to determine the recharge zone, but this information remains unavailable for this area. Although this information on groundwater remains unknown, runoff from rain events does influence the stream in the immediate vicinity of the spring, making catchment areas relevant to the spring and stream health.

The studies on stream health and similar salamander species have indicated that increased urbanization and impervious cover tends to result in lower habitat quality. The results of these studies can be applied to predict some of the possible implications of land use changes to the salamander survival.  By using the correlations between urbanization and salamander abundance, impervious cover data can then be used to predict potential health of salamander populations.  Unfortunately, very little information exists about Eurycea naufragiaand long term monitoring data does not exist to evaluate the hypothesized effects of observed land use changes on the survival of the salamanders. Additionally, without site-specific information, it seems unreasonable to generalize a 10% impervious threshold for all springs. Furthermore, most of the studies on impervious cover impacts on stream health have not occurred locally. The local geology may change the threshold for seen effects of impervious cover. Very few (if any) studies about stream health and urbanization have occurred in arid or semi-arid regions such as Central Texas. Because no data exists specifically for the area, it remains impossible to make any concrete conclusions as to the impacts the observed changes might have on the stream health and the salamanders. However, we can draw some likely possibilities of the vulnerabilities of the habitat. This study suggests that half of the populations at modeled springs (6) face high risk of extirpation, 4 appear to be moderately at risk, and 2 are at low risk of extirpation. These risk suggestions are based on the rate of land use change within and surrounding the catchments of the springs, predicted land use changes, and any available information on salamander presence or abundance. For the purposes of a viable population, lower salamander detection suggested a higher threat.

This study demonstrated some of the benefits to using two types of land classification. In some cases the NDVI model seemed to match the imagery well; whereas, in other cases, image classification appeared to fit the imagery better. Using both methodologies allowed for some built in corrections as well as greater confidence in the results. Conducting NDVI models for all of the years used in image classification might be helpful in directly comparing the two techniques. Each spring has unique features (such as steepness and flow rate) that may make them either more resistant or more vulnerable to upstream changes. This also means that management strategies may differ on a spring-by-spring basis. Hydrological studies of each spring would be useful for more accurate evaluations of each spring. Future studies should include surveys of habitat variables and if possible salamander abundance at all of the known locations. Furthermore, several non-surveyed springs exist in the area, which could be analyzed as well. Whether or not salamanders occur in these springs may help elucidate the habitat requirements for the species.

Acknowledging the limitations of using imagery data and impervious cover to estimate health of salamander populations and based on the observed land use changes, many of the known locations for the salamanders may already be experiencing some of the degrading effects of increased urbanization. After all, we do know that the salamanders cannot survive if their habitat is destroyed. In other areas, increasing urbanization severely degrades habitats. Specifically,Eurycea naufragia needs interstitial spaces and cobble to hide and forage (presumably though the motivations behind habitat selection for the species are unknown). Increased run-off and sedimentation rates can fill the interstitial spaces with fine sediment.  Such sediment can clog the salamanders’ gills and restrict movement and foraging ability. Additionally, increased run-off can increase the magnitude of stream flow pulses following rain, which may wash cobble and other cover objects downstream.

The rate of urbanization in Williamson County poses a serious concern for preservation of the species. The amount of impervious cover in watershed generally increases with expanding human populations and development.  Ultimately, the salamanders likely are very vulnerable to habitat changes and several springs have seen increased impervious cover over the past sixteen years, four to over 10%.  More than the increased impervious surfaces, the majority of the springs have experienced significant land use change over the same time period. Clearing land of all vegetation can be as or more harmful than impervious cover alone due to increased sedimentation rates. Because the levels of impervious cover remain somewhat subjective to the accuracy of the model and the threshold percentage may realistically differ between springs, overall land use rates may be more informative. Furthermore, the percentage of impervious cover depends on the size of the catchment area, whereas the location of the impervious surfaces may be more important than the overall percentages.  Adding any groundwater data, local hydrology studies, and additional salamander surveys would enhance the accuracy and potential implications of this study.  Fundamentally, E. naufragia’s limited range in an increasingly urban county makes it very vulnerable to extirpation. Conservation decisions should consider the vulnerabilities of each site as proposed in this paper, as well as any additional hydrological and salamander population size data as it becomes available for each spring.

Read More Thesis Projects >