multispecies virtual population analysis summary of model, applications, and advances
DESCRIPTION
Multispecies Virtual Population Analysis Summary of Model, Applications, and Advances Lance Garrison, Jim Ianelli, Megan Tyrrell, Jason Link NEMoW Workshop 28-31 August 2007. Structure of MSVPA Model. Suitability Params. Diet Data. Consumption = Predator BM * %DR. Other Food. - PowerPoint PPT PresentationTRANSCRIPT
Multispecies Virtual Population Analysis
Summary of Model, Applications, and Advances
Lance Garrison, Jim Ianelli, Megan Tyrrell, Jason Link
NEMoW Workshop28-31 August 2007
Structure of MSVPA Model
Other Food
Consumption = Predator BM * %DR
Pprey = (Suitable Biomass)prey / Total Suitable Biomass
Cprey = Consumption * Pprey
M2prey = Cprey / BMprey
M2age
Single Species VPA
BMage BMage BMage BMage BMage
Diet Data
SuitabilityParams.
Other Predators
Applications of MSVPA
North Sea – ICES Working GroupCod, Haddock, Whiting, Pout, Saithe, Herring, Sprat, Mackerel,Plaice, Sand lance
Northeast US – Tsou & CollieCod, Haddock, Dogfish, Hakes, Herring, Mackerel,Sand Lance, Skates, Flounder
Eastern Berring Sea – Livingston & Juardo-MolinaWalleye Pollock, Pacific Cod, Turbot, Yellowfin Sole, Arrowtooth Flounder, Fur Seal, Rock Sole, Pacific Herring
Implementation in the “4M” Package from ICES
A slice of the food web
Model Inputs and Data Requirements
Age-structured catch and biological information for all predator and prey species and associatedtuning indices for VPAs
Diet data including prey size/age information
Consumption parameters: daily rations or temperature dependent evacuation rates
Other food biomasses (and/or other predators)
Known Weaknesses in MSVPA
Overparameterized - not a statistical modelthat fits data and provides uncertainty
4M formulation results in a Type II feeding responsewhich leads to depensatory dynamics at low pop. Sizes
Assumes constant suitability parameters and requiresa comprehensive, large scale diet data set
Data intensive – but then so are all Ecosystem Models
Expanded MSVPA (MSVPA-X)Developed for ASMFC to address interactionsbetween Atlantic Menhaden and its major predators
Explicitly incorporates tuned VPAs in the formof extended survivors analysis
Implements a “weak” Type III feeding response
Decomposes “suitability” into preference, spatialoverlap, and size preference
- increases the ability to assimilate data- results in dynamic suitabilities
Implements a predator growth model
NEUS Application of MSVPA-X
Megan Tyrrell, Jason Link
Five most important predatorsSpiny Dogfish, Winter Skate, White Hake, Northern Goosefish,Georges Bank and South Cod
Porportional Consumption by Top 5 Predators on Major Prey Types
0%10%20%30%40%50%60%70%80%90%
100%
Prey Type
Po
rpo
rtio
nal
co
nsu
mp
tio
n GB&S Cod
Northern Goosefish
White Hake
Winter Skate
Spiny Dogfish
NEUS Application of MSVPA-X
Megan Tyrrell, Jason Link
Total Mortality of Mackerel
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
Year
Mo
rtal
ity
Average F ages 0-7
Average M2 ages 0-7
Total Mortality of Herring
0
0.2
0.4
0.6
0.8
1
1.2
Year
Mo
rtalit
y
Average F ages 0-10
Average M2 ages 0-10
NEUS Application of MSVPA-X
Megan Tyrrell, Jason Link
Multispecies statistical model
Data
Prior
information
Population Dynamics
O bjective function
Predation equations
Posterior D is tribution
Likelihood
Profile
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
Data
Jim Ianelli
MSM Implementation: Eastern Bering Sea
• Species:
– Pollock, Pacific cod and arrowtooth flounder
• Coded in C++ (ADMB)
• Tuned to:
– Fishery catch
– Survey indices
– Age (pollock) and length (arrowtooth flounder, Pacific cod) compositions
Jim Ianelli
MSM system for the Bering Sea
Walleye pollock
Pacific codFishery
Arrowtooth flounder
Pollock abundance (age 3+)
0
5000000
10000000
15000000
20000000
25000000
30000000
35000000
40000000
1975 1980 1985 1990 1995 2000 2005 2010
Year
N3+
(10
00's
)
Pollock recruitment
0.0E+00
2.0E+07
4.0E+07
6.0E+07
8.0E+07
1.0E+08
1.2E+08
1975 1980 1985 1990 1995 2000 2005 2010
MSVPA MSM SSP
Year
Rec
rutm
ent
(100
0's)
Why Use MSVPA or MSM Approaches ?
These are MRM models, so suited for specific questionsor trophic interactions
Data rich situations with age-structured catch and biological data for a few species
Both data and outputs are directly related to SSassessment models. As such, easy to compare to dataand a common “language” for managers
Poised for “tactical” advice