You are here:
Spatial distributions of biophysical conditions on the Ohio River
Citation:
BOLGRIEN, D. W., R. MEYER, M. S. PEARSON, T. M. JICHA, T. R. ANGRADI, D. L. TAYLOR, M. F. MOFFETT, AND B. H. HILL. Spatial distributions of biophysical conditions on the Ohio River . River Systems. E. Schweizerbart'sche Verlagsbuchhandlung, Stuttgart, Germany, 19(2):19-34, (2011).
Impact/Purpose:
To document research results.
Description:
Conceptually, landscape and hydrogeomorphic features associated with large floodplain river ecosystems impose spatial organization on river biota, nutrients, and habitat. We examined whether resulting patchiness was evident in basin and riparian landcover, water chemistry, fish assemblages, and habitat of the Ohio River. Spatial autocorrelation (variograms and Moran’s I) was measured for 66 metrics from 120 sites, selected by a probability design and corrected for longitudinally trends. Spatial autocorrelation determines the maximum inter-site distance over which measurements were significantly correlated (or mean patch length). Overall, 20% of the metrics could be considered patchy. Forested, agricultural, and developed landcover in catchments 10-km and 50-km upriver from each site, and the river valley and 400-m wide riparian buffer within 10-km upriver of each site were very patchy with mean patch lengths ranging 97-204 km. Dissolved oxygen, water temperature, Secchi depth had mean patch lengths of 187 km, 32 km, and 97 km, respectively. Mean patch length for chlorophyll a could not be determined although it had a significant positive Moran’s I value. Other water quality and habitat measurements (including total and dissolved nutrients and carbon, and major dissolved ions) did not display significant patchiness. Patch lengths of sand and gravel in the littoral zone could not be measured in our data or in an independent data set with 7 x more spatial density. Both data sets found the same longitudinal gradients. Only 25% of fish assemblage metrics (mostly species richness) were patchy. Patchiness of species richness of a guild including native, intolerant, and non-schooling species and the proportion of individuals with deformities would be consistent with un-measured specific habitat requirements and exposure to stressors, respectively. However, patchiness of non-indigenous fishes and tolerant species was counter-intuitive because of their perceived non-specific habitat requirements. For metrics with strong spatial autocorrelation, Gaussian geostatistical simulation accurately interpolated values for unmeasured sites along the river. It reproduced the population (i.e. river) data frequency distribution of the probability design. Although no causal processes are suggested, spatial autocorrelation reveals that the spatial distributions of some water quality and fish assemblage measurements of the Ohio River is organized by land cover summarized at several scales.