Abstract
The covariance among a range of 20 network structural properties of food webs plus net primary productivity was assessed for 14 published food webs using principal components analysis. Three primary components explained 84% of the variability in the data sets, suggesting substantial covariance among the properties employed in the literature. The first dimension explained 48% of the variance and could be ascribed to connectance, covarying significantly with the proportion of intermediate species and characteristic path length. The second dimension explained 19% and was related to trophic species richness. The third axis explained 17% and was related to ecosystem net primary productivity. A distinct opposite clustering of connectance, the proportion of intermediate species, and mean trophic level vs. the proportion of top and basal species and path length suggests a dichotomy in food‐web structure. Food webs appear either clustered and highly interconnected or elongated with fewer links.