Figure 1. The food-web underlying FEAST, showing level of detail for the groups modeled. Lines depict trophic flows, line thickness is proportional to magnitude of flow and color represents pelagic (green) or benthic (blue) routes.
More and more, high resolution end-to-end models have started incorporating fish as one of their components. Such exercises are usually restricted to a few years, and do not include fisheries removals. FEAST is a length-based, spatially explicit bioenergetics model that comprises the fish portion of the vertically integrated model of the Bering Sea Integrated Ecosystem Program (BSIERP).
The vertical model itself contains five modules: 1) climate; 2) oceanography (ROMS); 3) lower trophic levels (NPZ); 4) fish; and 5) fisheries (FAMINE). FEAST models 14 fish species linked to five zooplankton groups (Fig. 1) and 20 fisheries specified by sector, gear, and target species.
Species include walleye pollock, Pacific cod, arrowtooth flounder, Pacific salmon, capelin, Pacific herring, eulachon, sandlance, myctophids, squids, shrimp, crab, epifauna, and amphipods; these have a two-way interaction with six groups from the Nutrient - Phytoplankton - Zooplankton (NPZ) module: small copepods, oceanic/shelf copepods, oceanic/shelf euphausiids, and benthos.
Temperature and advection estimates from the physical oceanography portion (ROMS) are used in the fish bioenergetics and movement components. The model has a spatial resolution of approximately 10 km and will be run both with past climate (1970-2010 hindcast) and three different climate projections stemming from three different climate models. In addition, FEAST is the "real world" model to be used in a management strategy evaluation (MSE) for walleye pollock and Pacific cod, two of the main commercial groundfish in the Bering Sea.
We have made several substantial changes at different levels: improved the temperature estimates in the oceanography module (ROMS), added the length-only and biomass pool species, and added predation as a component of fish movement. Early in April a review of the temperature showed the model was still estimating temperatures about 2 degrees higher on average, which prompted a re-parameterization that lowered the bias by 1 degree. The addition of the rest of the species decreased the predation mortality on pollock (via prey switching) and the fish movement, previously based on maximizing individual potential growth, is now based on maximizing fish biomass. The net improvement has been on better fits of estimated diet composition to stomach content data from the Food Habits Database across different regions of the Bering Sea shelf.
Current work is focused on adjusting for low winter production in the zooplankton module and excessive mortality of the age-0 pollock. The background work to parameterize fish growth, predation, and movement in FEAST is being summarized in three draft manuscripts for the upcoming third special issue of BSIERP in Deep Sea Research Part II.
Working titles are: "A length-based, spatially explicit bioenergetics model for walleye pollock, Pacific cod, and arrowtooth flounder;" "Extension of a three to fourteen species dynamic prey selectivity model based on lengths and prey preferences;" and "Modeling fish movement for walleye pollock, Pacific cod, and arrowtooth flounder in the Bering Sea."
In addition to the parameterization of FEAST, we have also streamlined the supply of key model output to the management strategy evaluation routine, so that all the data included in the stock assessments for Bering Sea pollock, Pacific cod, and arrowtooth flounder can be extracted seamlessly from FEAST and incorporated into the stock assessments, e.g., length frequency, diet composition, and CPUE of each species by station.
Regular updates of FEAST-MSE are presented to the North Pacific Fisheries Management Council for informational as well as feedback purposes. Also, results from FEAST were presented in May at the Second International Symposium on Effects of Climate Change on the World's Oceans in Yeosu, Korea.
