 |
Wednesday, June 6, 2007 - 9:30 AM
168
Relating Fish Assemblages to Environmental Patterns at Three Multistate Scales
Robert M. Hughes1, Alan T. Herlihy1, and Jeannie C. Sifneos2. (1) Fisheries and Wildlife, Oregon State University, 200 SW 35th St., Corvallis, OR 97333, (2) Geosciences, Oregon State University, 200 SW 35th St., Corvallis, OR 97333
Key challenges to studying and managing riverscapes include understanding how factors measured at various spatial-scales influence aquatic biota and developing accurate predictive models where study data are limited. Currently ecoregions and river basins are commonly used for classifying fish faunas. These classifications reduce the apparent variability occurring at a large scale, but also include considerable heterogeneity. We analyzed a 780 site data set obtained from the USEPA’s EMAP western survey. First, we determined fish clusters at three spatial scales in the western USA (all 12 conterminous states, all western mountains, Pacific Northwest mountains). We next determined that the predictor variables for those clusters changed with spatial scale. For example, longitude, dams and temperature were the best predictors for all sites, but latitude, turbidity, and canopy density ranked highest for Pacific Northwest mountains. The best three variable models included site, basin, and ecoregion predictor variables. However, basin, ecoregion, state, and site variables alone only accounted for half of the mean within-group similarity demonstrated by the fish clusters. We conclude that using large quantitative fish assemblage data sets linked with quantitative physical and chemical habitat data and landscape data to predict fish assemblage patterns is preferable to using preexisting landscape classifications.