Comments on Feyrer et al. “Modeling the Effects of Future Outflow on the Abiotic Habitat of an Imperiled Estuarine Fish”

Abstract

Feyrer et al. (Estuaries and Coasts 34:120–128, 2011) constructed a habitat index for delta smelt (Hypomesus transpacificus) as a function of abiotic covariates (specific conductance, Secchi depth, and temperature) to evaluate how future hydrologic conditions in the San Francisco Estuary might affect the habitat of delta smelt. In this article, we identify three methodological issues that pertain to the results of Feyrer et al.: (1) the use of an independent abundance estimate, (2) the detection of spatial bias in the Feyrer et al. habitat index, and (3) the procedure used to link the habitat index to estuarine outflow. Like Feyrer et al. (Estuaries and Coasts 34:120–128, 2011), we fit general additive models (GAM) to presence of delta smelt data; however, our models included a region factor. We found that the amount of variability in the presence of delta smelt explained by the conductivity and Secchi terms was reduced relative to Feyrer et al.; conductance dropped from 12.2 to 2.5 % and Secchi dropped from 8.2 to 2.1 %. Furthermore, we found that an annual habitat index based solely on estuarine flow had low predictive ability, but the two-stage process of GAM analysis and subsequent regression modeling on GAM analysis output may mask the detection of low predictive performance. We agree with Feyrer et al. that defining a habitat index for delta smelt is an important contribution to understanding the ecology of the species and to facilitating its recovery. Given our results, the delta smelt habitat index could be improved by including static regional effects, dynamic salinity and turbidity effects, and an independent abundance index.