Spectral radius is a better metric than weighted NODF to detect network nestedness: Linking species coexistence to network structure using a plant – larval sawfly bipartite

Abstract

• 1.

  Network nestedness describes an interaction pattern, wherein specialist species interact with a subset of partner species. Antagonistic networks are predicted to not be nested, because nestedness indicates a high intensity of interspecific competition, which compromises species coexistence. However, network nestedness is commonly observed in antagonistic networks, and the discrepancy between prediction and observation has not been fully resolved.

• 2.

  One of underlying factors explaining this discrepancy is the imperfection of metrics to detect network nestedness. However, studies comparing network metrics often fail to resolve which metric works best, presumably because they lack specific criteria.

• 3.

  We compared the results of the most commonly used metrics (weighted NODF) and a later proposed metric (spectral radius) to measure the nestedness of a quantitative plant - larval sawfly bipartite (including 8 sawfly species and 66 plant species, identified by gut DNA metabacoding of larvae). We also determined whether the sawfly species can coexist in terms of their dietary differences. Because nested structure is not likely to be compatible with species coexistence, we assumed that the metric identifying a non-nested structure is superior to the other.

• 4.

  The two metrics led to contrasting nestedness levels. Both observational and preference networks were found to be nested using weighted NODF, but was not nested using the spectral radius approach.

• 5.

  The dietary differences were significant among each sawfly species pair for both observational and preference networks, indicating low interspecific competitiveness and a high potential for species coexistence.

• 6.

  These results indicate that the spectral radius metric is superior to weighted NODF to detecting network nestedness and should be used in future network studies.