Chemotaxonomy and Size Structure of Phytoplankton Communities to Assess the Water Quality via Non-linear Partial Least Squares Regression

The evolution of water quality status requires the development of environmental screening tools that are rapid, reproducible and easily applied to large scale monitoring programs. In recent years, the High Performance Liquid Chromatography (HPLC) of phytoplankton pigments has been shown to be very informative to the community composition, as it holds the advantage to detect the small-size component, usually underestimated by microscopy (Garmedia et al., 2013). The water quality sensu within the Marine Strategy Framework Directive (2008/56/EC, MSFD) will be explained observing functional groups, chemotaxonomic composition, and cell size of the phytoplankton communities in different environmental conditions of the Adriatic sea, which represents an excellent area to study the spatio-temporal variations of primary production processes, in terms of biomass abundance and size structure. In this paper, we aim to validate and support the study of the water quality sensu. We analyse the size-structure and pigment spectra of the phytoplankton communities based on the HPLC by using a suitable non-parametric and non-linear regression method. Among them, we propose to consider multivariate additive partial least spline regression (Lombardo, Durand and De Veaux, 2009) to assess the water quality as related to the Eutrophication (Descriptor 5) of the MSFD. In particular we investigate the trophic status of the Adriatic sea by studying the dependence of the total chlorophyll a from various predictors like the salinity, the different size classes [micro- (> 20μm), nano- (20-2 μm) and pico- (< 2μm)], the Fp indicator (Mangoni et al., 2013) or the chemotaxonomic composition of phytoplankton assemblages, and the spatial site position.