Laboratory analysis was performed on 1,752 groundfish stomachs from the eastern Bering Sea and 62 groundfish stomachs from the Gulf of Alaska. During this quarter, 382 stomachs were analyzed during research surveys in the Bering Sea, and 460 stomachs were analyzed during surveys in the Aleutian Islands.
Further stomach collection and shipboard stomach analysis is ongoing for the 2006 field season. A total of 9,849 records were added to the groundfish food habits database.
By Troy Buckley, Geoff Lang, Mei-Sun Yang, and Kerim Aydin
Predator/Prey Interactions
“Food Habits of Groundfishes in the Gulf of Alaska in 1999 and 2001” by Yang et al. 2006 has been published as a NOAA Technical Memorandum (NOAA Tech. Memo. 164;
http://www.afsc.noaa.gov/Publications/AFSC-TM/NOAA-TM-AFSC-164.pdf). A total of 7,899 stomachs from 40 species were analyzed to describe the food habits of the major groundfish species in the Gulf of Alaska in 1999 and 2001. Arrowtooth flounder, Pacific halibut, sablefish, Pacific cod, bigmouth sculpin, big skate, and Bering skate were the main piscivores analyzed in this study. Overall, walleye pollock was the dominant prey fish.
Figure 1. Main food items (percent wet weight) of arrowtooth flounder by predator size, as measured by food habits data from summer trawl surveys in the Gulf of Alaska, 1999 and 2001. (n = number of stomachs with food.)
The main predators that fed on Tanner crabs were halibut, Pacific cod, big skate, longnose skate, and great sculpin. Flathead sole, sharpchin rockfish, rougheye rockfish, longnose skate, and walleye pollock were the main consumers of pandalid shrimp. Sharpchin rockfish, Pacific ocean perch, redbanded rockfish, Atka mackerel, and pollock fed mainly on zooplankton. Southern rock sole, northern rock sole, rex sole, Dover sole, deepsea sole, and darkfin sculpin were benthic feeders; they fed mainly on polychaetes, marine worms, and brittle stars. Internannual variation was also observed, for example, the diets of arrowtooth flounder showed higher consumption of pollock and osmeridae (primarily capelin) in 2001 while, in 1999, zooplankton was more prevalent in arrowtooth diets (Fig. 1 above).
By Mei-Sun Yang and Kerim Aydin
Figure 2. Temporal trend of maximum likelihood estimates of age-1 walleye pollock predation mortality, MSVPA – multispecies virtual population analysis, MSM – multispecies statistical model.
Figure 3. Posterior distribution of the suitability coefficient of Pacific cod as predator (age-8) and walleye pollock as prey (age-1).
Multispecies Modeling
The AFSC is developing a scientific framework for providing ecosystem-based advice for the management of groundfish fisheries. This framework has three main goals for protecting the ecosystem attributes: 1) maintain predator/prey relationships, 2) maintain energy flow and balance, and 3) maintain diversity. The framework includes using multispecies models (biological and technological interactions) for developing statistically rigorous multispecies forecasts. In particular, we have developed a multispecies virtual population analysis (MSVPA) and a multispecies forecasting model for the eastern Bering Sea.
Unfortunately, these models lack statistical assumptions which hamper the inclusion of uncertainty into multispecies model parameter estimation. Therefore, we recently developed a simple version of a multispecies statistical model to show that it is possible to incorporate MSVPA predation equations into a statistical catch-at-age model. In this work we show a more elaborated version of this model set up in the AD Model Builder platform (Otter Research Ltd.).
The multispecies statistical model (MSM) includes two species, walleye pollock and Pacific cod. The MSM estimates abundance, suitability coefficients, and predation mortality based on catch-at-age data (1979-2002) for both species. It incorporates the AFSC’s bottom trawl survey (walleye pollock and Pacific cod), the echo integration trawl (EIT) survey (walleye pollock), survey age composition data, fishery age composition data, and annual total catch of both species (1979-2002), predator ration, and stomach contents data in a statistical framework.
With the MSM configured in the AD Model Builder platform, we were able to estimate for the first time the uncertainty of the predation mortality (Fig. 2) and the suitability coefficients (Fig. 3) and the residual mortality. This new version is more efficient and has more flexibility for the estimation of parameters, indicators, and their uncertainties. It is also an ideal platform to incorporate the technological interactions and create a powerful new analysis tool that will improve our evaluations of broad range of implications of management policies within a multispecies framework.
By Jesús Jurado-Molina, James Ianelli, and Patricia Livingston
