'Phytoplankton events' in French coastal waters during 1987-1997

This study aims to propose a tool to describe the long-term (10 years) vari ability of phytoplanktonic assemblages monitored by Rephy (French monitorin g programme for phytoplankon and phycotoxins) in the English Channel, the B ay of Biscay and the Mediterranean French coastal waters. According to the sampling strategy (systematic survey, with a time-step of I or 2 weeks, and a between-sampling station distance ranging from less than one to several kilometres), the information content of the data is mainly relevant to the characterization of seasonal variability of populations at the mesoscale. F or any given area, and for each of the 56 taxinomic units considered here, the data are thus processed in order to recognize the temporal succession o f 'phytoplankton events: an event is defined by retaining qualitative infor mation only. It encompasses the phases of sudden growth, high level of abun dance and decline of a population. Times at which an event begins or ends a re detected by using a time-series segmentation method, which allows to sum marize the whole data set as a multivariate set of event occurrences. Categ orizing observations in such a way also provides an efficient tool for the identification of different patterns of variability over long-term time sca les, for instance: 'recurrent events' (e.g. populations generating events i n a periodic-like manner), or 'anomalies' (e.g. of climatic origin). On an univariate basis, an 'average' event can be defined for each taxum, charact erized by its within-year position, its duration, its magnitude and the ass ociated deviations. On a multivariate basis, graphical representation of ev ent successions could also allow between-year comparisons. A simple multiva riate analysis was also used to describe the seasonal pattern of some taxa. (C) 2001 Ifremer/CNRS/IRD/Editions scientifiques et medicales Elsevier SAS .