Application of time-series and multiple regression techniques to assess population variability in two populations of the endangered Barton Springs Salamander (Eurycea sosorum)

Background/Question/Methods

Time-series and multiple regression techniques can be very useful tools for understanding dynamics of animal populations and identifying possible mechanisms of population variability. Despite concern over global amphibian decline, and the applicability of these techniques to evaluate amphibian populations, there is still a scarcity of long-term time-series datasets for these taxa. Many amphibian time-series estimate abundance at too low a temporal resolution to observe important intra-annual fluctuations (i.e. yearly) and/or are too short for appropriate application of time-series analysis. In this study, I examined a seven-year time-series (2004-2010) of monthly counts of juvenile (<25.4mm), young adult (25.4-50.8mm) and adult (>50.8mm) salamanders from two populations of the endangered Barton Springs Salamander (Eurycea sosorum): Eliza Spring and Barton Springs Pool (BSP). This monitoring data is collected by the City of Austin, Texas, as required by its federal permit to operate BSP as a public swimming facility, and is one of the most high-resolution (monthly) and comprehensive datasets known for an amphibian population. Time-series and stepwise multiple regression methods were used to quantify relationships between salamander abundance and climatic, abiotic and biotic habitat variables. Modeling results and knowledge of basic karst hydrogeology were used to explore possible mechanisms by which these variables could affect salamander abundance and make inferences to about the life history and ecology of E. sosorum.

Results/Conclusions

Median densities of all size classes of salamander from the Eliza Spring population (0.85, 1.67 and 1.22 individuals/m² for juveniles, young adults and adults, respectively) were more than an order of magnitude greater than those at BSP (0.05, 0.07 and 0.02 individuals/m²) throughout the study period. Time-series of salamander abundance at Eliza Spring were more variable than BSP, but densities of each size class were significantly synchronized between sites, suggesting that similar environmental conditions at the local scale contribute considerably to population variability (i.e. Moran effect) in these populations. Studies of reproductive biology in closely-related species of Eurycea reveal that larval survival is much greater at lower water temperatures, and that larvae remain in cooler subterranean habitat for up to 6 months.  Long 10-12 month time lags between periods of high rainfall (which subsequently increase flow rates and decrease water temperature) and peaks in salamander abundance suggest a similar pattern may be operating in populations of E. sosorum. Finally, results are used to evaluate monitoring protocols, guide restoration efforts at Barton Springs and improve adaptive management for E. sosorum.