Spatial structure and survey method influence population estimates for endangered Comanche Springs Pupfish

Authors: Matthew Acre, Joshuah Perkin, Megan Bean

Year: 2019

Abstract

Desert fishes in the American southwest are vulnerable to anthropogenic landscape alterations. Nearly 90% of pupfish (family Cyprinodontidae; 38 of 43 species) are listed as imperiled and many persist within freshwater protected areas (FPA). Pupfish population assessments within FPA are infrequent, in part, because of inconsistencies and few comparisons among methods. We developed density estimates for the Comanche Springs Pupfish, Cyprinodon elegans, using mark-recapture and unmarked individuals observed through visual count (VC) and minnow trap (MT) surveys at Balmorhea State Park (BSP), Texas, USA. During Spring 2019, we collected 1,328 C. elegans during three repeated VC and MT surveys at 40 locations within BSP. We fit a Schnabel estimator to the mark-recapture data and N-mixture models to VC and MT count data and assess the influence of habitat covariates and spatial autocorrelation using Akaike’s information criterion. We found N-mixture model estimates were generally greater than mark-recapture model estimates, though models agreed in some habitats. We also found differences in density estimates from VC versus MT methods, including higher densities from MT in shallow water and VC in deep water. Best-fit models included habitat covariates such as depth, water temperature, and fine sediments, though spatial autocorrelation variables were the strongest predictors. We conclude that N-mixture models fit to unmarked individuals from VC and MT methods include trade-offs in their performance depending on habitats surveyed, and that spatial contexts matter. Monitoring other fishes inhabiting other FPA using the methods described here could provide consistent and repeatable abundance indices to benefit conservation biology.