Riparian forests in northwest Ohio watersheds: Relations among landscape structure, land use/land cover, and water quality in streams.

摘要

Landscape structure within agricultural watersheds reflects impacts from human activities, which in turn, directly impact water quality in streams. I studied linkages among riparian landscape structure, land cover, and water quality within nine NW Ohio watersheds that drain into Lake Erie. Specifically, I investigated whether: (1) landscape structure metrics of riparian forests may be used as effective indicators of water quality in rivers, (2) relationships between water quality and proportions of riparian land cover types changed with riparian corridors of different widths, and (3) width of riparian vegetation was related to adjacent land cover. Multiple regression analyses of landscape structure metrics, computed in GIS for 120-meter wide riparian corridors and nonriparian areas, with water quality annual yields showed that land cover proportions in non-riparian areas and watershed factors (soils, topography, impervious surfaces) predicted (p0.05) water quality for suspended sediments (SS) (89%), total phosphorus (TP)(81%), nitrate+nitrite (NO 23) (91%), and total dissolved solids (TDS) (52%), better than riparian landscape metrics for forests or combined natural cover types. Selected landscape metrics served as water quality indicators for these watersheds. These metrics included the proportion of natural vegetation or row crops in non-riparian areas for SS, the proportion of natural vegetation in non-riparian areas for TP, and the proportions of row crops within entire watersheds, pasture/hay within non-riparian areas, or pasture/hay in riparian corridors for NO 23. Proportions of row crop and pasture/hay cover continuously increased while natural vegetation decreased across incrementally wider (30, 60, 90, and 120 meters) riparian corridors. Such patterns explained increasingly higher percentages (up to 26%, 46%, and 64% for SS, TP, and NO23, respectively) of water quality variability as determined by simple regressions. Furthermore, Chi-square tests and one-way ANOVAs determined that width of riparian vegetation was Significantly (p0.01) associated with land cover adjacent to riparian areas and significantly (p0.01) different among land covers. My results indicated that natural vegetation was highly fragmented or lacking in riparian areas and that reduction of pollutant yields in streams will depend primarily on improving land management in nonriparian portions of watersheds.