Bioregions in the pelagic ecosystem are frequently established on the basis of remotely sensed properties of the sea surface, such as sea surface temperature or sea surface chlorophyll concentration. Those works dealing with the regionalization of the marine ecosystem by means of the use of properties of the water column are less frequent, and even less those that obtain the data from periodic in situ monitoring programs, which are scarce. In this work we use time series of micro, nano and pico-phytoplanktonic abundances in the upper 100 m of the continental shelves of the Gulf of Cadiz and the Alboran Sea from the projects STOCA and RADMED (southern coast of Spain, Western Mediterranean). The use of times series allows us to estimate the median phytoplanktonic abundances of several phytoplanktonic groups along the water column. These statistics differ substantially from those abundances obtained for one particular campaign, reflecting the large seasonal and inter-annual variability of phytoplanktonic communities. These median profiles, estimated for the four seasons of the year and for several phytoplanktonic groups characterize each of the locations sampled in the aforementioned monitoring programs and are used for establishing the similarity between them. Then, these locations are grouped using a cluster analysis. Using some simulations from numerical experiments we determine which metrics and methods of analysis are the more suitable ones for the regionalization of the area of study. A bootstrap method is also used to determine which differences among bioregions can be considered as statistically significant. Despite the existence of a fast current that connects the Gulf of Cadiz and the Alboran Sea, our results show that the outer part of the Gulf of Cadiz shelf, and that of the Alboran Sea, can be considered as two differentiated bioregions. The latter region shows a higher productivity with a higher abundance of large cells such as diatoms, and the dominance of Synechococcus bacteria over Prochlorococcus ones.
Read full abstract