Species Association Networks of Tropical Trees Have a Non-neutral Structure

An important question in community ecology is how species interactions influence the structure of ecological communities. Here, we ask if patterns of species associations within a diverse tropical forest in Costa Rica are consistent with neutral models of community organization or consistent with niche based models. We quantified species associations using the extent of crown overlap between trees. Crown overlap is proportional to the amount of light and soil resources shared between individuals. We studied the effects of niche structuring by representing the community as a species association network, a graph whose nodes are species. Two species are connected by an edge in this graph if the crowns of at least two individual trees in the two species overlap. In order to assess network structure predicted by neutral vs. niche structured models, we compared the structure of species association networks with those of randomized communities. Two indicators of network structure showed highly significant differences between empirically observed and randomized communities. These differences became more pronounced for large trees with a high extent of crown overlap. This indicates that factors associated with niche structuring (such as competition, facilitation, and abiotic filtering) influence the structure of the community. The effects were most striking in older, large trees. Our results also show that how we choose to represent and analyze community structure is critical for our ability to reject a neutral hypothesis.