Do co-occurrence matrices suggest more than usual when comparing vegetation types? A test based on data from Picea abies and Abies alba forests in the Friulian Alps

In this study, we introduce a framework for comparing vegetation types based on species co-occurrence matrices. The comparison relies on two complementary approaches: (1) quantifying the negentropy of the co-occurrence matrix for each vegetation type, and (2) evaluating the evenness of eigenvalues derived from pairwise co-occurrence matrices extracted from the overall species co-occurrence matrix of the vegetation system. The method is illustrated using phytosociological relevés from Picea abies and Abies alba forests of the Friulian Alps (NE Italy). Our results show that vegetation types associated with more mesophilic environmental conditions—according to Landolt’s ecological indicators—exhibit co-occurrence matrices with higher species connectance, as captured by information-based metrics. Furthermore, for the dataset considered, similarity matrices derived from species co-occurrence patterns outperform traditional approaches (frequency vectors, mean cover, presence/absence data, and standard similarity indices) in predicting ecological indicator patterns.