A spectral graph theoretic study of predator-prey networks

Abstract

Predator-prey networks originating from different aqueous and terrestrial environments are compared to assess if the difference in environments of these networks produce any significant difference in the structure of such predator-prey networks. Spectral graph theory is used firstly to discriminate between the structure of such predator-prey networks originating from aqueous and terrestrial environments and secondly to establish that the difference observed in the structure of networks originating from these two environments are precisely due to the way edges are oriented in these networks and are not a property of random networks.We use random projections in $\mathbb{R^2}$ and $\mathbb{R^3}$ of weighted spectral distribution (WSD) of the networks belonging to the two classes viz. aqueous and terrestrial to differentiate between the structure of these networks. The spectral theory of graph non-randomness and relative non-randomness is used to establish the deviation of structure of these networks from having a topology similar to random networks.We thus establish the absence of a universal structural pattern across predator-prey networks originating from different environments.