Long, J.M. and W.L. Fisher. 2006. Analysis of environmental variation in a Great Plains reservoir using principal components analysis and geographic information systems. Lake and Reservoir Management 22:132-140.
Abstract
We present a method for spatial interpretation of environmental variation in a reservoir that integrates principal components analysis
(PCA) of environmental data with geographic information systems (GIS). To illustrate our method, we used data from a Great Plains
reservoir (Skiatook Lake, Oklahoma) with longitudinal variation in physicochemical conditions. We measured 18 physicochemical
features, mapped them using GIS, and then calculated and interpreted four principal components. Principal component 1 (PC1)
was readily interpreted as longitudinal variation in water chemistry, but the other principal components (PC2-4) were difficult to
interpret. Site scores for PC1-4 were calculated in GIS by summing weighted overlays of the 18 measured environmental variables,
with the factor loadings from the PCA as the weights. PC1-4 were then ordered into a landscape hierarchy, an emergent property of
this technique, which enabled their interpretation. PC1 was interpreted as a reservoir scale change in water chemistry, PC2 was a
microhabitat variable of rip-rap substrate, PC3 identified coves/embayments and PC4 consisted of shoreline microhabitats related
to slope. The use of GIS improved our ability to interpret the more obscure principal components (PC2-4), which made the spatial
variability of the reservoir environment more apparent. This method is applicable to a variety of aquatic systems, can be accomplished
using commercially available software programs, and allows for improved interpretation of the geographic environmental variability of a system compared to using typical PCA plots.