Structure and spatial self-organization of semi-arid communities through plant-plant co-occurrence networks

In a variety of areas in ecology, a network approach has proven fruitful in studying the interactions among the components of systems. We propose a novel network approach for examining plant species interactions in plant communities. We constructed various networks based on the plant species found on 12 point-intercept transects in a semi arid community, developing one network for each transect. Species were identified as nodes and were considered linked if they occurred at the same intercept point. To understand the general organization of the networks, we calculated the cumulative distribution of the number of links per species in each network, the average nearest neighbour degree of each species, and the self-organization of the most-linked species. The networks behaved as mutualistic ecological systems and displayed a typical truncated power-law distribution for the cumulative distribution of the number of links per species. The results suggest that most of the species were preferentially associated with a few others. In all but one highly competitive species (Stipa tenacissima), the number of links and abundance were positively correlated. Link density was positively correlated with the spatial self-organization of the entomophyllous and facilitated species, while generalist species did not present any relation. For S. tenacissima, the link density and spatial self-organization were positively correlated, as other species were able to establish in the edge of S. tenacissima patches when it was less dominant. Our results agree with past works in networks, and it seems that in a semi-arid environment, at least, our network approach would provide additional information about the nature of interactions that each species develops in the community.