Title: Optimization hardness constrains ecological transients

Abstract: Living systems operate far from equilibrium, yet few general frameworks provide global bounds on biological transients. Here, we frame equilibration in complex ecosystems as the process of solving an analogue optimization problem. We show that functional redundancies among species in a complex foodweb lead to harder ill-conditioned problems, manifesting as slow modes on a nearly-degenerate solution landscape. Common dimensionality reduction methods precondition ecological dynamics, by isolating fast relaxation from slow solving timescales that exhibit transient chaos for harder problem instances. In evolutionary simulations, we show that selection for steady-state diversity produces ill-conditioning, an effect quantifiable using complexity bounds for constrained optimization. Our results demonstrate the physical toll of computational constraints on biological dynamics.