Assessment of water quality using multivariate techniques in River Sosiani, Kenya.

Multivariate techniques can infer intrinsic characteristics of complex data by generating correlation, similarity, dissimilarity, and covariance vector matrix to ascertain their relationships. The study evaluated the effect of anthropogenic activities by analyzing selected physicochemical water quality parameters (WQP) as indicators of pollution in River Sosiani, located in western Kenya, at six stations from August 2012 to February 2013 (Aug-Oct ≡ wet season, Nov-Feb ≡ Dry season). Temperature, pH, Total Dissolved Solids (TDS), conductivity, and Dissolved Oxygen (DO) were measured in situ while Total Phosphorus (TP), Total Organic Nitrogen (TON), and Biologial Oxygen Demand (BOD) were measured in vitro using standard methods. Except for DO and pH, the other variables were increasing in concentration downstream. Cluster analysis grouped stations with municipal discharge, to be the most distant linked to other stations in both seasons. Multidimensional scaling had four categories of stations with similar WQP: before, after, and wet and dry seasons. Principal component analysis with (60.5 and 26.1% for components 1 and 2) evaluated TON and TP as key pollutants in both seasons. Factor analysis with varifactor two at 35.3 and 27.1% variance in wet and dry seasons, respectively, had strong absolute factor loading of BOD (wet 0.878, dry 0.915) and TP (wet 0.839, dry 0.709) inferring sites with organic pollution also had nutrient pollution. Assessment of pollution with the selected WQP identified two major effects: nutrient and organic. Additional variables may identify other pollutants along the river. Multiple pollution effects, changing environment, and intrinsic characteristics of aquatic ecosystems generate complex data which are better assessed with multivariate techniques.