So happy to see our paper “Coexistence of #Nestedness and #Modularity in Host-Pathogen Infection #Networks” published in @natecoevo. We uncovered hidden dimensions of infection networks, which challenges established views in #ecology.
nature.com/articles/s4155…
nature.com/articles/s4155…
Our goal was to understand how the environment mediates in the pattern of host-pathogen interactions, i.e., nestedness and modularity. To do so, we gathered empirical data on plant-virus infection networks surveyed at different habitats in central Spain, from 2000 to 2002. 
But standard analyses of bipartite networks reported ambiguous results, and could not explain structural dependence on seasonality and habitat differences. Looking at networks is helpful, but nearly not enough to fully grasp the complexity of natural infections.
The problem is infection networks do not explicitly represent environmental effects. Infections come in threes. Pathogens don’t infect isolated genotypes, but host ecotypes (an ecotype is a variation or breed that has adapted to specific environments).
We model ecotype infections using hypergraphs, a mathematical object encoding the 3-way interactions b/w hosts, pathogens and habitats. From the hypergraph, we can recover ecotype-pathogen and bipartite networks, without information loss.
en.wikipedia.org/wiki/Hypergraph
en.wikipedia.org/wiki/Hypergraph
In network theory, elements of an adjacency matrix indicate whether pairs of nodes are linked or not (top). The hypergraph approach extends this definition to a 3-D adjacency matrix. en.wikipedia.org/wiki/Adjacency…
Our theoretical model combines neutrality and local adaptation to find domains in which networks are, or are not, likely to be useful. Natural selection may impose significant constraints on some features of interactions but negligible constraints on others.
The model shows the emergence of ecotypes is a consequence of increasing the diversity of habitats (here C =4). Theory predicts that nestedness is an universal feature of community assembly, while (habitat) modularity is not and depends on relative speciation rates.
Real ecosystems are predicted to exist in a continuum between nested-modular and nested systems driven by infraspecific competition. Depending on the spatiotemporal scales involved, apparently opposite structures ("shadows") may coexist in hypergraphs (3-D adjacency matrix).
Empirical study reports significant nestedness but low modularity, which invalidates the assumption of isolated habitats. Deviations from neutrality (except Fallow Field) in habitat- or season-specific subsystems expose local host adaptations and sampling effects.
This study was a multi-disciplinary collaboration between plant virologists and network theoreticians. I thank my co-authors @BlaiVidiella @Raul_MontanezM from @IBE_Barcelona, and Auroral Fraile, Soledad Sacristan, and Fernando Garcia-Arenal from @CBGP_Madrid for their work.
And that’s all for now! We look forward to see more applications of #hypergraphs in ecology and #evolution !
