ADJUSTING MERCURY CONCENTRATION FOR FISH-SIZE COVARIATION - A MULTIVARIATE ALTERNATIVE TO BIVARIATE REGRESSION

Regression-based methods like analysis of covariance (ANCOVA) are freq uently used to adjust one variable for the correlated influence of a s econd less interesting variable (e.g., mercury concentration and fish size). However, the influence of the covariate (i.e., fish size) is no t removed unequivocally when regression slopes are not parallel. Using data on tissue-mercury concentration and fish size from 30 population s of lake trout (Salvelinus namaycush), we show that data adjusted to a common size with bivariate regression can retain information associa ted with the original size differences. As an alternative, we use univ ariate and bivariate summary statistics from each population as raw da ta in a multivariate analysis to search for differences among populati ons. Ordination axes resulting from this analysis exhibited both small - and large-scale spatial autocorrelation. Localized spatial patterns probably reflect similar geochemical features of the watersheds of nei ghbouring lakes in small geographic areas. In contrast, regional spati al autocorrelation suggested broad-scale patterns that may implicate a tmospheric inputs of mercury. As an extension of this multivariate app roach, both regional and local patterns could be compared with environ mental variables to reveal correlations that may suggest new cause-and -effect hypotheses.