Development of Multispecies Models of Fish Community Dynamics on Georges Bank William T. Stockhausen Michael J. Fogarty Northeast Fisheries Science Center NOAA Fisheries Woods Hole, MA
Patterns of Fish Abundance on Georges Bank
Approach to Modeling Construct energy flow models –Network analysis –Static picture: assumed equilibrium, linear processes –Underlying dynamic processes recognized Different time periods ↔ Different environmental conditions Develop dynamic models –Represent change in the system –Accommodate nonlinear processes –Focus on components: Take advantage of differences in connectivity within/between trophic levels
Overlapping Data Requirements Dynamic models: time series of –Abundance (biomass) for each species –Catch (biomass) for each species Energy flow models: estimates of –Average production for species groups Prod = Time-averaged biomass x P/B –Averaged consumption for species groups Cons = Time-averaged biomass x C/B –Average total harvest for species groups
Multispecies Production Models for the Fish Community Develop multispecies production models for the fish community Primary objective is to test for species interactions Model incorporates –Competitive interactions –Predator-prey interactions (Type I functional response) Parameter estimation will be done in a Bayesian context (using WinBUGS)
Initial Multispecies Matrix CodHaddock Silver Hake Yellowtail Flounder Winter Flounder Winter Skate Little Skate HerringMackerel Cod Pred Comp CompPred HaddockComp Silver Hake CompPred Yellowtail Flounder Comp Winter Flounder Comp Winter Skate Comp Pred Little Skate Comp HerringPrey Comp MackerelPrey Comp
Estimating Biomass Time Series
Results from stock assessments generally inappropriate –Spatial coverage –Temporal coverage –Not available for all species
Estimating Biomass Time Series: NEFSC Bottom Trawl Survey Data 40 year dataset Semiannual surveys (Spring, Fall) Cape Hatteras to Gulf of Maine Nearshore to continental shelf break Stratified random survey design Standardized gear, procedures
Estimating Biomass Time Series: Survey Strata
Estimating Biomass Time Series: Survey Indices
Estimating Biomass Time Series: Issues with using trawl survey data Unequal catchabilities among species High sampling variability
Adjusting for Catchabilities Species-specific catchability factors based on –Published literature Visual observations and acoustic measurements Comparisons between survey and assessment results Meta-analysis of assessment-based catchabilities –Analysis of trawl survey data Day/night correction factor for individual net tows
Smoothing Survey Biomasses Model observation error as uncorrelated multiplicative noise Assume observed and underlying processes can be characterized by ARIMA models Smoothed results based on
Smoothed Biomass Time Series
Estimates of Production and Consumption For the Energy Flow Models
Summary Multspecies production models will provide dynamic counterpoints to static, equilibrium-based energy flow models Procedures developed for estimating time series of species biomass from NEFSC Bottom Trawl Survey data –provide biomass time series for the production models –provide production/consumption estimates for energy flow models