Principal Investigators: Patricia A. Livingston and Sarah Hinckley
Goals and Objectives
This project proposes to investigate the hypothesis that spatial processes which affect the overlap and availability of juvenile pollock to predators, particularly adult pollock, are important determinants of juvenile pollock survival rates. In order to investigate this hypothesis we will develop a spatially resolved model of upper trophic level predators on juvenile walleye pollock, with particular emphasis on cannibalism. The model will serve as a focus for synthesizing information on juvenile and adult pollock and other upper trophic level predators on juvenile pollock.
by Patricia A. Livingston and Jesús Jurado-Molina
The ultimate goal of our research is to develop a forward projection model of upper trophic level predators on juvenile walleye pollock. Multispecies virtual population analysis (MSVPA), however, is a retrospective method of analyzing catch-at-age data in conjunction with information on diet linkages between species to obtain more realistic estimates of numbers-at-age and natural mortality for juvenile fish, particularly walleye pollock. Juvenile pollock abundance in the forward projection model will be derived, at least initially, from MSVPA outputs. We have completed parameterizing the MSVPA for the eastern Bering Sea and are now conducting a sensitivity analysis of the model.
The MSVPA model, as currently parameterized for the eastern Bering Sea, includes the following species as predators: walleye pollock, Pacific cod, Greenland turbot, yellowfin sole, arrowtooth flounder, and northern fur seal. Arrowtooth flounder and northern fur seals are entered as "other predators", which means that VPA's are not performed for these species. Instead, inputs on their consumption rates, diet, and population abundance are input so that their predation on VPA prey species in the model can be calculated. Prey species are walleye pollock, Pacific cod, Greenland turbot, yellowfin sole, rock sole, and Pacific herring. The modelled time period is 1979 to 1995 and we have included an extensive amount of diet data from 70 predator/year/quarter combinations derived from close to 40,000 stomach samples collected during that time period. The last few months have been spent adding arrowtooth flounder and northern fur seals to the model, adding substantial amounts of diet data, and updating and re-tuning the VPA inputs to reflect the most recent assessment data. The single-species VPA's have been tuned to fit the outputs of the more complex stock assessment models that are presently being used for most groundfish species in the eastern Bering Sea.
Model results show that most predation mortality for the prey species in the model (walleye pollock, Pacific cod, Greenland turbot, yellowfin sole, rock sole, and Pacific herring) occurs on juveniles that have not yet recruited to the fishery. Model estimates of population abundance for exploited ages of each prey species are similar to those provided by single species models. However, abundance estimates of juveniles, particularly walleye pollock, are substantially larger than estimates from single-species model (Figure 1). Walleye pollock was the main prey species consumed by MSVPA predators, and cannibalism constituted the majority of the predation mortality of age-0 fish. The dominant predators on age-1 pollock included adult pollock, Pacific cod, arrowtooth flounder, and northern fur seals. In some years, Pacific cod consumed the largest biomass of walleye pollock prey relative to other predators (Figure 2). However, most of the biomass of pollock consumed by cod tended to be from older pollock.
Future plans
A sensitivity analysis of the MSVPA model is currently underway. Further explorations with the model include: testing the stability of suitability estimates of the model and performing multispecies forecasts using various management scenarios.
The focus for the portion of the project that is sponsored by the Coastal Ocean Program will move towards further defining and parameterizing the spatially explicit model of the eastern Bering Sea. We plan to use the BORMICON model structure created at the Marine Research Institute in Reykjavik, Iceland. The code has already been obtained and compiled locally on a UNIX workstation. The next step will be to precisely define the spatial compartments and time period to be used for model inputs. We anticipate a large effort will be expended on estimating the migration matrices for moving adult pollock between model regions.
