Trophic redundancy, not species diversity, predicts food web structural robustness: edge-level bifurcation analysis across 249 ecosystems

Abstract

The structural mechanisms that confer resilience in food webs remain poorly understood despite decades of research linking biodiversity to ecosystem stability. Here we present an edge-level analytical framework that derives, for each trophic interaction, a closed-form bifurcation threshold from local triangle counts. We define a Fragility Index (FI) quantifying the fraction of interactions lacking triangular support, defined as interactions connecting taxa with no shared trophic partner. Applied to 249 trophic food webs from the GlobalWeb database spanning fresh water, marine, and terrestrial ecosystems, we show that trophic redundancy (the fraction of edges embedded in at least one triangular motif) is the strongest structural predictor of network-level fragility among those tested (Pearson r = −0.938, p < 10−4, n = 249), a result that strengthens to r = −0.949 in species-resolved networks (n = 148). By contrast, species richness shows negligible predictive power (r = +0.06). Edge density shows a moderate negative association (r = −0.48), but trophic redundancy explains substantially more variance. Because the Fragility Index and trophic redundancy are both derived from local triangle counts, this strong negative correlation is analytically expected; the empirical contribution lies in the cross-ecosystem variation of triangulation and in the failure of species richness and interaction density to predict structural robustness. Critically, 38% of analysed food webs exhibit FI exceeding 80, suggesting that most natural ecosystems operate near structural fragility thresholds, a pattern compatible with, though not sufficient to establish, a near-critical structural regime. These results suggest that interaction topology, rather than species richness, is the appropriate unit of analysis for structural resilience in ecological networks.