PHYTOPLANKTON PIGMENT IN THE GULF OF ST.-LAWRENCE, CANADA, AS DETERMINED BY THE COASTAL-ZONE-COLOR-SCANNER .2. MULTIVARIATE-ANALYSIS

We report here on a statistical study of physical-biological interacti ons in the Gulf of St. Lawrence that uses for the first time ocean col or images (phytoplankton pigments) and data on runoff and wind. Based on a Monte Carlo test for statistical significance, we extracted four orthogonal (independent) spatial patterns (Empirical Orthogonal Functi ons, EOF) in pigments that explain 64% of the total variance. We also computed four EOFs from the wind data that explain 90% of the total va riance. Based on multiple correlations among these EOFs and runoff ano malies, we derived two modes of physical-biological variability. The f irst mode is the dual regulation of production by runoff in the Lower St. Lawrence Estuary (LSLE) and the western Gulf and by alongshore win d stress in the northeastern Gulf. The second mode incorporates mesosc ale circulation features (eddies, meanders) that respond to low freque ncy fluctuations in runoff (in the lower estuary) or wind (Gaspe Curre nt, northwestern Gulf, northern Esquiman Channel). The third mode refl ects the impact of seasonal wind regimes on pigment levels in the LSLE , the Gaspe Current, and the southwestern Gulf. The fourth mode is dom inated by one coccolithophore bloom event in August 1979. The analysis also gives some insight into sources of interannual variability in pi gment levels and distributions. The spring bloom does not dominate the seasonal pigment cycle; only the third pigment EOF (5% of the variati on) displays a spring peak stronger than the fall peak. The seasonal c ycle of pigments is in part linked to that of runoff, driven mostly by year to year differences in the spring freshet. However, year to year differences in the summer wind and runoff regimes also play a role. T his suggests that late summer and fail blooms and the physical factors that regulate them deserve more study. (C) 1997 Elsevier Science Ltd.