Abstract Objective Recovery of the Razorback Sucker Xyrauchen texanus has included augmentation of wild populations (i.e., stocking), and managers need to understand rates of natural recruitment across lake and river populations in the Colorado River basin, USA. In augmented populations, a physical marker (e.g., tags)—or lack thereof—is insufficient to determine the provenance of wild individuals. Our study evaluated the reliability of fin ray microchemical signatures for discriminating among possible natal sources of wild Razorback Suckers in the San Juan River. Methods We used a microchemical tracing technique to characterize possible natal sources (i.e., hatchery versus wild) of Razorback Suckers in the San Juan River. We validated three classification methods (linear discriminant analysis, quadratic discriminant analysis, and random forest) using specimens with known natal sources (n = 63) representing four hatchery facilities and the wild. Four possible natal sources were separated based on distinct fin ray microchemical signatures (Sr/Ca, Ba/Ca, Mn/Ca, and 87Sr/86Sr). Results All three models performed with high classification accuracy (98%) using the training data set. Validated classification models were used to predict natal sources of unmarked, unknown-origin Razorback Suckers that were collected in the San Juan River during 2015–2016 (n = 95). Predicted natal sources showed strong agreement across classification methods, with 86.3% of samples assigned to the same sources. One hatchery signature was indistinguishable from the wild signature; we estimated ages based on fin ray growth increments (annuli) to inferentially separate these sources in relation to a distinct change in stock management history. All unknown-origin Razorback Suckers were predicted to be of hatchery origin. Conclusions We demonstrated the use of a nonlethal, specialized analytical technique to reliably predict natal origins of individuals from an imperiled freshwater fish population that is subject to intensive conservation management. Natural recruits were not present in our sample. However, this technique is viable for detecting future natural recruitment.
Barkalow et al. (Tue,) studied this question.