Category: Ecology

Summary: Finding how unevenly nutrient recycling can be distributed across trophic layers before a recycling-stabilized food web loses stability again.

Nutrient recycling can stabilize ecological communities, but it is not just the total amount of recycling that matters. This experiment asks how much heterogeneity in recycling across trophic layers a mismatched yet recycling-stabilized consumer-resource community can tolerate before the leading mode becomes unstable.

The model builds dense non-symmetric Jacobians for trophically coherent food-web communities with fixed stoichiometric mismatch, fixed overall recycling strength, and fixed mean locality. GPU-accelerated eigensolves, disorder sampling, and iterative deepening are then used to bisect the recycling-dispersion parameter and track the stability boundary across larger systems.

This isolates the architecture of recycling from its average strength. The result is a finite-size map of when uneven recycling stops acting as a buffer and instead undermines the same stabilization it was meant to provide.

Method: GPU dense non-symmetric Jacobian eigensolve with iterative deepening and bisection on recycling dispersion at fixed mismatch, recycling strength, and mean locality.

What is measured: Critical recycling-dispersion threshold, leading-eigenvalue stability boundary, localization statistics, and threshold bracket width.