Freshwater species are declining rapidly but more complete data are needed for determining the extent and cause(s) of population declines and extirpations. Integrating newer survey techniques, freely available data, and traditional field work may allow for more effective assessment of population decline. We used detailed historical species records and environmental DNA (eDNA) survey methods to identify changes in population distribution of a long‐lived, imperiled stream salamander, the eastern hellbender (Cryptobranchus alleganiensis alleganiensis: Cryptobranchidae). We used logistic regression with Bayesian inference to test whether selected environmental variables may be good predictors of hellbender population persistence and extirpation. Hellbenders persisted in only 42% of the 24 historical record sites. The best fit model indicated electrical conductivity (EC) was the strongest predictor of hellbender population persistence (EC < 278 μS/cm) and extirpation. Conductivity was strongly negatively correlated with canopy cover within the total watershed (r = −0.83, n = 21, p < 0.001) and riparian buffer of the watershed (r = −0.77, n = 21, p < 0.001). Electrical conductivity tends to increase following deforestation, and may inhibit sperm motility and thus limit recruitment of hellbenders and other aquatic vertebrate species with external fertilisation. By integrating historical data, eDNA, field data, and freely available high resolution remote sensing data, our study design allowed for rapid assessment of predictors of and changes in hellbender distribution over a relatively broad geographic area. This cost‐ and time‐effective approach may be used for evaluating other rare aquatic species.

No abstract

Efficiently and effectively identifying and assessing potential wildlife habitat and important ecological resources is essential as rapid anthropogenic land use change alters and detrimentally affects terrestrial and aquatic habitats. Accuracy assessment of remotely sensed data supports ecological planning and management decisions, and is especially important when using freely available, coarse‐resolution spatial datasets, such as the National Land Cover Dataset (NLCD). A popular dataset designed for application at larger regional to national spatial scales, the NLCD has been used in finer scale studies outside of its intended use, often without the imperative support of field‐based accuracy assessment. We ground‐truthed stratified random sampling points to assess the accuracy of the 2016 NLCD at a fine spatial scale relevant to stream‐based ecological research. Our results demonstrated an overall accuracy of <65%, less than the United States Geological Survey recommended accuracy of ≥85%. Results indicated that the NLCD may not be effective as a tool for stream‐level studies and could provide erroneous results for fine‐scale habitat assessment and planning when used as the only reference dataset. When conducting ecological research, it is important to consider the appropriate scale, resolution, and limitations of available datasets to achieve the most accurate results.