Spatial analysis of hydrological and phytoplanktonic data of the Bay of Tunis. Multivariate cartography

Abstract

A method of cartography originally used in geology was adapted to generate regionalization and to obtain 2-D maps of multivariate marine data. The ecological purpose of the method is to divide the studied area through homogeneous regions presenting common multivariate characteristics. Firstly, transformation was applied to the original matrix of hydrological parameters in order to satisfy the condition of multinormality. Then, associative analysis was used in order to produce an easy to interpret partition of sites. The level of heterogeneity between each station and the properties of each group was assessed by measuring the Bayesian probabilities. These conditional probabilities measure the chance that each site has of belonging to a predefined group of sites. Based on the geographical positions of the stations, the probability values for each group of stations were mapped using kriging interpolation algorithm. The obtained maps of iso-probabilities for the different groups of stations were used to define homogenous zones on a single map. Including the phytoplanktonic dataset afterwards, the indicator species were identified for each zone.

This multivariate analysis was applied to a hydrological and phytoplanktonic dataset of the Bay of Tunis. Measures at surface were made at 17 stations, sampled monthly over 2 years. The results illustrated a partition of the bay considering four groups, two coastal and two central groups of stations. The importance of the inshore influence was demonstrated in the setting up of such a regionalization through the inflow of alluvium and other products of coastal activities. The significant presence of the toxic phytoplanktonic community in the bay suggests the need to institute a monitoring program.