ABSTRACT

Q-mode factor and R-mode principal components analyses were used on heavy metal (Mn, Co, Cr, Cu, Zn, Cd, Ag, Ni) concentration gradients, grain size and organic carbon measurements to delineate time-spatial patterns and general principles in complex data. Three Q-mode factors account for 99.5% of the variance in 39 surface samples. Factor I accounts for 77.1% of the total variation and is related to shallow-water samples with low metal concentrations. Factor H is associated with samples of high metal loading and profundal water; this factor accounts for 19.6% of the variance. Factor III accounts for only 2.7% of the variance and is related to samples intermediate to factors I and II. The sediment strata (i.e., slump and varve-like features) were grouped into geochemical clusters relat d to sediment type by Q-mode cluster analysis. R-mode principal components analysis of the chemical and physical parameters within each piston core are highly correlated ( = 0.05). Normalized heavy metal concentrations (weighted to grain size) show upward increasing metal concentrations in homogeneous slump units at depth suggestive of noncultural accumulations and metal migration. Simultaneous analysis of 10-cm surface cores show that Mn, Cd and Zn concentrations correlate the highest with depth of sediment, suggestive of upward migration and accumulation. The average metal concentrations, on a time-spatial basis, are grain size dependent ( = 0.05), but heavy metal distributions in the upper 10 cm are caused by geochemical and/or physical factors resulting in the chemical gradients. The geochemical factor accounts for 11% of the variance, while the physical factor accounts for 75% of the total variance as determined by principal component analysis.