Extraction of Unique Plant Species Communities from the Sub-Humid Humid Bioclimate of Martinique

Abstract

Biodiversity in forest ecosystems is crucial for regulating ecological processes and delivering essential ecosystem services. In this study, we investigate how specific plant species communities in secondary forest formations (FSS) reflect particular bioclimatic conditions and successional stages in a Sub-Humid Humid (SHH) environment. Our dataset comprises four survey stations (S51, S103, S119, S120) where minimal sampling areas were determined (ranging from 500 to 1000 m 2 ), yielding 29 to 45 recorded species per station. Environmental variables such as altitude, total biomass, and total basal area were also collected. Using the ECLAT algorithm to identify frequent species sets, followed by hierarchical clustering (CAH) and principal component analysis (PCA), we were able to highlight recurring species assemblages and singular, ecologically significant species. Further bivariate analyses of the distribution index (Id) against basal area (St_Totale) confirmed that the most frequent species often exhibit higher distribution and larger basal area. These findings underscore the potential of combining data mining techniques with conventional statistical methods to unravel complex patterns in forest ecosystem dynamics. Our results provide a valuable foundation for scaling up to more extensive datasets to better understand the ecological and environmental drivers shaping secondary forest communities under changing climate conditions.