A single attractor governs the population dynamics of marine fishes worldwide

Population fluctuations are now empirically recognized to be nonlinear, but the governing attractors are poorly understood. Here we show that population dynamics of 457 marine fish stocks from 121 species worldwide are governed by a single attractor that partitions into three archetypes reflecting top predators, mesopredators and forage fish, with stock-specific ecological and evolutionary variation. Correlation between attractors was high across all stocks (mean 0.53 to 0.63), greater within archetypes (0.57 to 0.77), and increased with geographic, phylogenetic and life history similarity. Leveraging shared attractors increased prediction skill in ∼90% of cases, raising mean recruitment skill from 0.38 to 0.75 and even enabling moderate prediction for otherwise poorly predictable stocks. These results reveal that marine fish population dynamics are not heterogeneous but are realizations of a single attractor structured by trophic level, with immediate predictive benefits for global fisheries.