Multivariate statistics applied to phytoplankton data from two Gulf Stream warm core rings’

Multivariate statistical techniques were used on data from two cruises to Gulf Stream warm core rings to reveal patterns and relationships between phytoplankton species abundance and environmental parameters. Principal component analysis (PCA) was performed on each cruise separately and on the combined data to help delineate groups of stations, depths, and species. A ring center station with high numbers of diatoms and a Slope Water station formed separate groups. Deep samples from all the stations grouped together. We consistently observed strong similarities in phytoplankton abundance and species composition between fall samples from the two rings. Inverse analysis was performed using presence or absence data, and the values of the first three principal components were then used in a cluster analysis to determine species co-occurrences. This analysis resulted in separate groupings of a large number of diatom species found at the center of the rings, of deep species, and of the most abundant and the most frequently observed taxa. Simple correlations, multiple regressions, and canonical correlations were performed with the phytoplankton log abundance data and the environmental data. The canonical coefficients and correlations suggest that high numbers of dinoflagellates and other algae are associated with lower salinity in the Slope Water and at the ring edge, and that higher temperatures and low values of silica (dissolved) are reflected in higher numbers of diatoms and coccolithophorids at ring center. Northward meanders of the Gulf Stream can result in warm core, anticyclonic rings whose effects can reach to the sea floor (Weatherly and Kelley 1985). The core, an 18°C isostad, is considered to be Sargasso Sea water, bordered by a higher velocity current composed of Gulf Stream water (Joyce and Wiebe 1983). The entire ring is embedded in Slope Water and usually moves in a southwesterly direction, so that the phytoplankton is carried with the ring out of its original environment into new temperature and nutrient regimes. This bowl-shaped body of water influences and is influenced by the surrounding Slope Water and the indigenous species there in a dynamic way, and phytoplankton changes in days and even hours have been noted (Joyce et al. 1984; Fryxell et al. 1985). Phytoplankton from two Gulf Stream warm core rings of different ages has been compared as to abundance and taxa (Fryxell et al. 1985), and these same two rings are examined here. One ring was in the process of “pinching off” from the Gulf Stream when it was studied in 1982; we call it meander/ ’ This research was supported by NSF grants OCE 8 1-O 1785 and OCE 85-03984 to G.A.F. and by a University Sea Grant Marine Fellowship to R.W.G. WCR 82-H. The other, WCR 8 l-D, was 3 months old when sampled during cruise 110 of Atlantis II in 198 1: Both rings were visited during September-October but in consecutive years. The forming ring showed a great deal of uniformity in distribution of the phytoplankton; the older ring had more stratification of species (possibly vertical patches, not strata: Venrick 1984) with rapid change of the flora through advective processes. As a general statement, dinoflagellates were more abundant in the high velocity region (ring edge) and at the Slope Water station, whereas diatoms and coccolithophorids responded opportunistically to overturns from storms in the ring center. Our major objective here is to examine relationships at the phytoplankton community level by using a multivariate analytical approach. Such an approach can test hypotheses, simplify and clarify structure that is apparent in the data set but is difficult to extract and interpret, or reveal structure that would otherwise be overlooked or misinterpreted. We are attempting to reveal and clarify structure, but some hypothesis testing comes into play in the regression analyses, where explicit models for the dependent variables are required. Although phytoplankton data are commonly orga-