Multiple survey methods reveal greater abundance of endangered pupfish in restored habitats

Freshwater organisms inhabiting arid ecosystems are imperilled by human alterations to water‐limited landscapes. This is especially true among desert‐dwelling cyprinodontid fishes, 90% of which are imperilled by habitat destruction within limited or shrinking ranges. Constructing habitats that mimic natural habitat form and function may provide a tool for species conservation, especially within freshwater protected areas. However, pupfish population assessments within degraded compared with restored habitats are infrequent, and few comparisons among survey methods exist. Density estimates were developed for Endangered Comanche Springs pupfish Cyprinodon elegans throughout altered and restored habitats in a freshwater protected area by using mark–recapture and N ‐mixture models fitted to data collected using minnow trap and visual count survey methods. This allowed comparison of habitats, survey methods, and statistical methods commonly used to generate population size estimates for imperilled pupfish. Population estimates varied across major habitat types and were largest among habitats constructed to mimic naturally occurring ciénegas. Estimates using visual counts were higher than estimates from minnow traps where water was deeper and where macroalgae cover was highest. N ‐mixture models generally estimated higher abundances than mark–recapture and were not limited by recapture ability. The results provide strong evidence that restored habitats house greater abundances of pupfish, but survey and statistical methods commonly used to detect these differences have trade‐offs in performance according to the habitats surveyed. This work benefits the field of conservation biology by providing guidance for existing and emerging monitoring programmes assessing abundance‐based fish responses to habitat improvements.