• IFIM;
  • barbel;
  • Fisher's exact test;
  • log-linear models


Under experimental conditions, we tested whether variations in preference curves, weighted usable areas, and electivity profiles for velocity of 0+ juvenile barbel Barbus barbus could be attributed to environmental variables (discharge, water depth, time of day) and/or time of sampling. We elaborated preference curves and weighted usable areas according to the Instream Flow Incremental Methodology, as well as electivity profiles. Variation among preference curves, weighted usable areas and electivity profiles was always observed, regardless of time of day, water depth, or discharge combinations. Our results support published concerns that the shape of preference curves is highly dependent on the time of data collection, and that curves derived from a single field sampling excursion can be viewed as representative for that moment in time only, whereas curves generated as composites from typological or time-series samples may provide an approximation for a study area, but at the expense of between-site and/or between-date variation. Spatial/temporal variation is an important aspect of fish habitat studies and should be incorporated as a (layered) component into descriptive/predictive models. To this end, we illustrate how a PCA-based technique and log-linear modelling can be used to assess temporal patterns in collections of preference curves and electivity profiles, respectively. Copyright © 2004 John Wiley & Sons, Ltd.