Multivariate Analysis of Extreme Physical, Biological and Chemical Patterns in the Dynamics of Aquatic Ecosystem

Abstract

This study is a part of the research in monitoring systems of environmental impacts in coastal re-gions in order to develop trophic dynamic models to be used in the aquatic systems management. Meteorological influences in the variability of the nutrients, larvae concentration, dissolved oxygen (DO) and chlorophyll a were investigated in a region where upwelling occurs. Extreme seasonal variations of reanalysis, QuikSCAT, and surface stations from the southeast coast of Brazil, as well as, surface seawater data collected in Anjos Bay, Arraial do Cabo city northeast of Rio de Janeiro state, are analyzed. Seasonality and correlations are applied to verify the relationship between them, considering minimum values of sea surface temperature (SST) and sea level variation and maximum values of the other variables. Principal Component Analysis (PCA) and Hierarquical Cluster Analysis (HCA) are applied to verify spatial and temporal variances and to describe more clearly the structure of the local ecosystem. The seasonality of northeasterly extreme wind stress follows the seasonal pattern expected for the study area with peaks during spring. The SST has a well-defined seasonal pattern with maximum peaks from February to July and minimum peaks from September to January. Chlorophyll a presents higher seasonal peak in February, being in accordance with DO; both are related to the maximum primary productivity. Correlations of the physical variables (local and remote) with nutrients and larvae present a relatively similar pattern around 0.5, showing these variables have a reasonable interaction with the meteorological forcing. PCA shows a strong variability in pressure data around 0.9, which may be related to the seasonal variations in South Atlantic subtropical anticyclone (SASA) and consequently to the occurrence of upwelling in the region. HCA shows the twenty-five parameters into two big clusters with predominance of biotic variables in one side and abiotic ones at the other. The degree of refinement of similarities allowed a division into six clusters of samples, giving the most satisfactory results at forming distinct clusters with more accurate regarding physical and biological elements.