A multispecies population assessment model for the Gulf of Alaska

1
A multi-species population assessment model for
the Gulf of Alaska
Kray F. Van Kirk, SFOS, UAF, Juneau Terrance J.
Quinn II, SFOS, UAF, Juneau Jeremy S. Collie,
GSO, URI, Narragansett ftkv_at_uaf.edu
2
Sea Grant

• Jointly funded by AK and RI Sea Grant
• Also collaboration with North Sea colleagues
  Lewy, Vinther (Denmark)
• Project started February, 2004
• Species groups
• Gulf of Alaska
• Georges Bank
• North Sea

3
Justification

• Crisis in Fisheries Management
• Single species analysis does not answer questions
  of multi-species interactions
• Ecosystem approaches desired
• Natural mortality is not constant!
• Harvest strategies cannot evolve until models do

4
Previous MSVPA Models

• Predation mortality estimated from gut studies
• Single predation coefficient
• Works backwards from oldest ages
• Catch-at-age is measured without error
• Applied to North, Baltic and Bering Seas, Georges
  Bank, and a few tropical areas

5
Model Overview

• Catch-age measured with error
• Forward time progression from
• age 1 recruitment
• Cohort abundance subject to fishing,
• flexible predation,
• and residual natural
• mortality

6
Age-structured Analysis
7
Modeling Predation Mortality

• Annual Ingestion of Predator Age (grams)
• consumption of modeled prey of interest
• consumption of non-modeled other food.
• Where annual ingestion rate is given by

Growth Increment
Hall et al. in press
Growth Efficiency
8
Components of Modeled Prey (1)

• Predator j of size b preference for prey of size
  a.
• (from Ursin 1971, where size-preference is a
  function of predator-prey weight-at-age ratio)
• Predator j preference for prey of species i.
• (from AFSC data, as a ratio of prey i consumed
  relative to total prey consumed changes over
  time)

9
Components of Modeled Prey (2)

• Suitability of prey i,a to predator j,b
• size/age preference prey preference
• Availability of prey i,a to predator j,b
• prey suitability prey abundance

10
Total Annual Consumption

• Consumption of prey i,a by predator j,b

Prey Availability
Predator Abundance
Predator Ingestion
Total Available Food (Modeled Prey And Other
Food)
11
Total Available Food
Modeled prey, species i
All other prey
T.A.F. Modeled prey all other prey How to
quantify other prey? - Very little data Use
factor to multiply modeled prey (10x, 50x) to
obtain estimated biomass of Other Prey Set
multiplier as parameter to be estimated
12
Predation Mortality P
Sum of ALL predation on species i age a by ALL
ages of predator j

• Total P for prey i,a by predator j

Total predation on species i,a (grams)
Total biomass of species i,a (grams)
13
Cohort Abundance
Residual Natural Mortality
Fishing Mortality
Predation Mortality
Traditional Natural Mortality M
14
Initial Model Set-up

• Pacific Cod (Gadus macrocephalus)

Arrowtooth Flounder (Atheresthes stomias)
Walleye Pollock (Theragra chalcogramma)
15
Initialization Parameters

• Abundance at age for year 1(SAFE)
• Fixed overall F (Mean 1981 2001)
• Fixed R (0.1)
• Spawner-Recruit alpha and beta
• Prey Preference (Mean 1981 2001)

16
Changes in Prey Preference
17
Population Trends (SAFE)
Arrowtooth Flounder
Walleye Pollock
Pacific Cod
18
Initial Results

Arrowtooth Flounder
Pacific Cod
Walleye Pollock

• Fishing pressure on pollock too great
• Insufficient food available to pollock

19
Increase Other Food factor from 10x to 50x
Walleye Pollock
Pacific Cod
Arrowtooth Flounder
BASE CASE SCENARIO
20
Predation Mortality from Base Case
Pollock
Pollock
ATF
Cod
ATF
Cod
ATF
21
Full Recruitment F for pollock at 1
ATF
Cod
Pollock
Predation on ATF 1981 2001 Base Case
Predation on ATF 1981 2001 F 1
Cod
Cod
ATF
22
Full Recruitment F for pollock at 1
Other Food factor 50x
Other Food factor 100x
23
Full Recruitment F for pollock at 0
Pollock
Cod
ATF
Predation on Pollock 1981-2001 Base Case
Predation on Pollock 1981-2001 F 0
Pollock
Pollock
Cod
ATF
Cod
ATF
24
Full Recruitment F for pollock at 0
Pollock
Cod
ATF
Predation on ATF 1981 2001 Base Case
Predation on ATF 1981-2001 F 0
Cod
ATF
ATF
Cod
25
Model Sensitivities

• Prey Abundance
• Predator Abundance
• Prey preference and Other Food factor
• Fishing Pressure

26
Next Steps

• Parameter Estimation
• Rho (species preference)
• Elements of size-preference function
• Other food multiplier
• Objective Functions
• Fishery catch at age
• Survey catch at age
• Stomach contents (predation rates)
• Addition of Pacific halibut and Steller Sea Lion

27
Areas for Exploration

• Predation between age classes and connection with
  fishing pressure
• Complexity of multi-species model vs. robustness
  of single species models
• Estimation of recruitment parameters once other
  parameters are set