Premium
A Comparison of Composite Habitat Suitability Indices and Generalized Additive Models of Invertebrate Abundance and Fish Presence–Habitat Availability
Author(s) -
Jowett Ian G.,
Davey Andrew J. H.
Publication year - 2007
Publication title -
transactions of the american fisheries society
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.696
H-Index - 86
eISSN - 1548-8659
pISSN - 0002-8487
DOI - 10.1577/t06-104.1
Subject(s) - habitat , salmo , abundance (ecology) , generalized additive model , brown trout , rainbow trout , invertebrate , trout , taxon , environmental science , fish <actinopterygii> , benthic zone , ecology , selection (genetic algorithm) , fishery , biology , statistics , mathematics , computer science , artificial intelligence
Generalized additive models (GAMs) offer an alternative approach to developing habitat suitability functions; these models may resolve some of the criticisms that have been made of conventional habitat suitability criteria and the associated composite suitability index (CSI). The potential advantages of GAMs include the ability to (1) account for correlation among habitat variables, (2) include interactions among variables, (3) make quantitative predictions of abundance or probability of occurrence at given flows, and (4) identify sharp thresholds in habitat selection. We developed CSIs and GAMs for two data sets—abundance of benthic invertebrates (mayflies Deleatidium spp. and caddisflies Aoteapsyche spp.) and habitat selection of large brown trout Salmo trutta and rainbow trout Oncorhynchus mykiss —and applied them in an instream habitat analysis. The GAMs performed only slightly better than CSIs. The GAMs for two of the four taxa examined ( Deleatidium and brown trout) had strong negative velocity–depth interaction terms, although their inclusion improved the model fit only marginally. The main effect of negative velocity–depth interaction terms was to constrain model predictions in deep, fast‐flowing water, and these GAMs gave more realistic results when applied to conditions beyond those from which they were developed (i.e., larger rivers or higher flows). The GAMs hold a number of potential advantages over conventional CSIs and offer an opportunity to develop more defensible habitat suitability models; ultimately, however, the performance of any model will be limited by the quality of the calibration data.