Tuning into Tuna: Investigating Spatial Complexity Tradeoffs in Stock Assessment Models Based on a High-Resolution Simulation Experiment of Yellowfin Tuna in the Indian Ocean

Understanding spatial heterogeneity is critical for improving fish stock assessments and sustainable management strategies, especially for highly migratory species such as yellowfin tuna (Thunnus albacares). This study investigates trade-offs associated with incorporating spatial complexity into stock assessment models, using a high-resolution blinded simulation experiment based on yellowfin tuna in the Indian Ocean. We developed a suite of spatially structured assessment models within the Stock Synthesis 3 framework, incorporating recruitment apportionment, movement dynamics, tagging data, and standardized catch-per-unit-effort (CPUE) indices via Bayesian spatiotemporal models. Under the blinded design, a stepwise modeling process was first applied to a reference dataset to identify the preferred spatial configuration. Selected model steps were then evaluated across 100 stochastic replicates to assess their ability to reproduce the true operating model dynamics. Results highlight that simpler models captured population depletion trends better but substantially underestimated absolute stock size (SSB), whereas increasing spatial complexity greatly reduced SSB bias. Our findings emphasize the importance of balancing model complexity, realism, and management needs when developing stock assessments and offer recommendations for incorporating spatial processes.