A Comparison of Two Ecosystem Models of the Oceanic Gulf of Mexico

There are many marine ecosystems that have been overexploited, urging researchers, government agencies, and international organizations to assess commercially fished populations. Initially, these stock assessments are often based on single-species models, but indirect biotic and abiotic effects in the ecosystem (e.g., changes in prey and predator population sizes, and increases in water temperature) are often ignored. Recently, ecosystem-based fisheries management has been advocated as more data have become available. Multiple ecosystem models exist that operate under different assumptions and algorithms, and with various interfaces, suggesting alternative models may be suited better for diverse monitoring situations. Presented here, two ecosystem models are developed and compared for the oceanic Gulf of Mexico (seaward of the 1000 m isobath): Ecopath with Ecosim (EwE) and Object-oriented Simulator of Marine ecOSystems Exploitation (OSMOSE). Results indicate that EwE has greater power in analyzing network indices (e.g., identifying indicator species, quantifying trophic impacts). However, EwE is constrained by diet matrix inputs, and assumes all individuals within a functional group are identical. OSMOSE simulates organisms throughout an individual’s life, which creates output regarding ontogenetic diet shifts (i.e., changes in diet composition and trophic level). In OSMOSE, predation is simulated using a size-based opportunistic approach, suggesting that highly migratory species with opportunistic diets may be better modeled compared to EwE. OSMOSE also requires coupling with a biogeochemical model (e.g., ROMS-NPZD), introducing spatiotemporal variability among low-trophic level organisms that influences processes at higher trophic levels. Both EwE and OSMOSE are useful ecosystem models that utilize similar inputs, but EwE is better at characterizing the trophic structure and OSMOSE is more advanced in terms of spatiotemporal and intraspecific variability.