• Assessment of Best-Management Practice Effects on Rangeland Stream Water Quality Using Multivariate Statistical Techniques

      Ellison, Christopher A.; Skinner, Quentin D.; Hicks, Larry S. (Society for Range Management, 2009-07-01)
      Quantifying the effects of watershed improvement efforts is critical to agencies responsible for protecting water resources of the semiarid western United States. A complex water quality data set collected from 1994 to 2004 of upper Muddy Creek Basin was subjected to cluster analysis, discriminant analysis, and canonical correlation analysis to improve understanding of basin fluvial processes and to investigate whether livestock grazing best-management practices (BMPs) improved the water quality of the watershed. Hierarchical agglomerative cluster analysis grouped nine sampling sites into two clusters based on similarity of biological indices, separating the clusters into aquatic communities more and less tolerant of degraded stream conditions. Discriminant analysis yielded strong spatial and temporal distinctions, providing important data reduction by rendering seven key parameters (total dissolved solids [TDS], temperature, elevation, slope, 10-dominant taxa, percent collector-gatherers, and percent Plecoptera) for the spatial variation and four parameters (TDS, dissolved oxygen, total taxa, and community tolerance quotient) for the temporal variation. Canonical correlation analysis identified strong negative relationships among Plecoptera taxa and total taxa with TDS and turbidity in addition to strong positive associations with elevation, slope, and channel substrate weighted embeddedness value. Despite the onset of severe drought midway through the study period, overall reductions of 13% for TDS and a 30% increase in macroinvertebrate total taxa occurred across years, strongly suggesting that improvements in water quality were correlated to BMPs that stabilized stream channels and improved the condition of riparian areas.