Modeling the building blocks of biodiversity.

ABSTRACT

BACKGROUND: Networks of single interaction types, such as plant-pollinator mutualisms, are biodiversity's "building blocks". Yet, the structure of mutualistic and antagonistic networks differs, leaving no unified modeling framework across biodiversity's component pieces. METHODS/PRINCIPAL FINDINGS: We use a one-dimensional "niche model" to predict antagonistic and mutualistic species interactions, finding that accuracy decreases with the size of the network. We show that properties of the modeled network structure closely approximate empirical properties even where individual interactions are poorly predicted. Further, some aspects of the structure of the niche space were consistently different between network classes. CONCLUSIONS/SIGNIFICANCE: These novel results reveal fundamental differences between the ability to predict ecologically important features of the overall structure of a network and the ability to predict pair-wise species interactions.