HR: 1330h
AN: B33E-04    [Abstracts]
TI: Explanatory Power of Multi-scale Physical Descriptors in Modeling Benthic Indices Across Nested Ecoregions of the Pacific Northwest
AU: * Holburn, E R
EM: Elaina.Holburn@ColoState.EDU
AF: Department of Civil Engineering, Colorado State University, Campus Delivery 1372, Fort Collins, CO 80523
AU: Bledsoe, B P
EM: Brian.Bledsoe@ColoState.EDU
AF: Department of Civil Engineering, Colorado State University, Campus Delivery 1372, Fort Collins, CO 80523
AU: Poff, N L
EM: LeRoy.Poff@ColoState.EDU
AF: Department of Biology, Colorado State University, Campus Delivery 1878, Fort Collins, CO 80523
AU: Cuhaciyan, C O
EM: Christopher.Cuhaciyan@ColoState.EDU
AF: Department of Civil Engineering, Colorado State University, Campus Delivery 1372, Fort Collins, CO 80523
AB: Using over 300 R/EMAP sites in OR and WA, we examine the relative explanatory power of watershed, valley, and reach scale descriptors in modeling variation in benthic macroinvertebrate indices. Innovative metrics describing flow regime, geomorphic processes, and hydrologic-distance weighted watershed and valley characteristics are used in multiple regression and regression tree modeling to predict EPT richness, % EPT, EPT/C, and % Plecoptera. A nested design using seven ecoregions is employed to evaluate the influence of geographic scale and environmental heterogeneity on the explanatory power of individual and combined scales. Regression tree models are constructed to explain variability while identifying threshold responses and interactions. Cross-validated models demonstrate differences in the explanatory power associated with single-scale and multi-scale models as environmental heterogeneity is varied. Models explaining the greatest variability in biological indices result from multi-scale combinations of physical descriptors. Results also indicate that substantial variation in benthic macroinvertebrate response can be explained with process-based watershed and valley scale metrics derived exclusively from common geospatial data. This study outlines a general framework for identifying key processes driving macroinvertebrate assemblages across a range of scales and establishing the geographic extent at which various levels of physical description best explain biological variability. Such information can guide process-based stratification to avoid spurious comparison of dissimilar stream types in bioassessments and ensure that key environmental gradients are adequately represented in sampling designs.
DE: 0400 Biogeosciences
SC: Biogeosciences [B]
MN: 2005 Joint Assembly