An example of a multispecies GADGET model for the Celtic Sea'

1
An example of a multispecies GADGET model for the
Celtic Sea.
John K. Pinnegar, Verena Trenkel Daniel Howell
IFREMER
2
A few definitions
Gadget is the Globally applicable Area
Dis-aggregated General Ecosystem Toolbox. Gadget
has been developed from 2 existing models,
Bormicon and Fleksibest. Bormicon was developed
internally by the Marine Research Institute,
Reykjavik, Iceland, and Fleksibest was developed
internally by the Institute of Marine Research,
Bergen, Norway. The current funding for the
development of Gadget has come from EU project
"Development of Structurally Detailed
Statistically Testable Models of Marine
Populations" QLK5-CT1999-01609 (known as dst2),
3
The Celtic Sea case study

• DST2 Workpackage 5.3
• To implement models for selected species in the
  Celtic Sea that incorporate both spatial and
  temporal information on stocks.
• The Celtic Sea will provide a considerably
  different case study, both in terms of the model,
  with more species and technical interactions but
  also somewhat different data, much of which will
  be in more aggregate form than initial raw
  samples

4
Celtic Sea case study
UK
France
ICES Areas VII e-k (since same as
Southern-Shelf working group)
Model run from 1984-2006
Pre-run period (UK Survey data)
Modelled Period
Predicted Period
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
2000
2001
2002
2003
2004
2005
2006
5
Tentative Celtic Sea Food-web from Nantes
meeting, June 2000
Predator ?
Whiting
Merlangius merlangus
6
Which fish species should be included?
Fish Biomass (t/km2)
Fish Landings (t/y-1)
Cod
Plaice
Whiting
Haddock
Megrim
Megrim
Haddock
Hake
Cod
Sole
Haddock
Cod
Whiting
Hake
Hake
Monkfish
Other Fish
Other Fish
Megrim
83.3
97.4
(Based on ICES 1998 Data)
(Based on CEFAS 1999 Survey Data)
Monkfish
Value of Fish Landings (UK )
Plaice
Cod
Haddock
Hake
Other Fish
Megrim
Sole
50.7
Whiting
Monkfish
Plaice
Sole
Whiting
(Based on CEFAS 1998 Data)
7
Interactions among chosen species
0.711 (99.9)
0.002 (66.7)
0.038 (88.4)
0 (0)
0.001
0.002
0.001
0.001
0.001
0.158 (97.5)
0.001
0.145 (99.3)
0.056 (88.9)
0.574 (96.5)
0.471 (99.6)
0.006
5.271 (99.5)
0.001
0.017
0.002 (100)
0.092 (100)
0.001
0.207 (100)
0.531 (100)
All fluxes in t/km2
0.226 (100)
0.032 (100)
Predation by non-chosen species
Consumption of non-chosen species
8
All fluxes gt0.001 t/km2
Cetaceans
Hake
Whiting
Seals
Horse Mackerel
Megrim
Monk
Fish Feeders
Haddock
Cod
MC Fish
Mackerel
Cephalopods
Blue Whiting
Zooplankton Feeders
Polychaete Feeders
Eggs Larvae
Macro- plankton
Small crust. Feeders
Echinod. Feeders
Euphausids
Plaice
Mollusc Feeders
Sole
Crabs Lobsters
Shrimps
Polychaetes
Zooplankton
Sessile Invertebrates
Micro- plankton
Small Crustaceans
Molluscs
Echinoderms
Meiofauna
Micro-benthos
Pelagic Bacteria
Phytoplankton
Detritus
Algae
9
Other candidates for inclusion in Gadget model
of Fish
of Fish
of


Value
Biomass
Landings
Predation by
Predation on
chosen spp
chosen spp
Blue Whiting
0.92
0.47
-
58.05
0
Mackerel
55.94
14.07
10.24
6.46
0.02
Horse Mackerel
12.26
50.45
3.92
15.44
0
Blue-whiting
Mackerel
Horse-mackerel
10
Celtic Sea Food-web from Reykjavik meeting, 2001
Whiting
Merlangius merlangus
Cod
Gadus morhua
Blue Whiting
Micromesistius poutassou
11
Celtic Sea Cod model (COD)

• ICES Areas VII e-k (since same as
  Southern-Shelf working group)
• 1984-2006, modelled period 1991-1998
• Single species, single-area, single-fleet model
• Ages 1-10
• Lengths 15-120cm
• Model uses UK survey index, French UK
  commercial survey ALK,
• length-distribution, mean length data.

12
Celtic Sea whiting model (WHG)

• ICES Areas VII e-k (since same as
  Southern-Shelf working group)
• 1984-2006, modelled period 1991-1998
• Single species, single-area, single-fleet model
• Ages 1-10
• Lengths 6-68cm
• Model uses UK survey index, French UK
  commercial survey ALK,
• length-distribution, mean length data.

13
Celtic Sea blue-whiting model (BWG)
Very limited data available!

• ICES Areas VII e-k (although ICES stock area much
  larger)
• 1984-2006, modelled period 1991-1998
• Single species, single-area, single-fleet model
• Ages 1-10
• Lengths 9-41cm
• Model uses UK survey index, Spanish ALK
  (1995-1998 only),
• ICES length-distribution (1991-1998), ICES
  catches for area VIIe-k.

14
?
Problems encountered so far.

• Not only the French and UK fish in the Celtic Sea
  also Irish, Spanish and other EU countries. No
  quarterly catch data for all countries.
• Very little data collected by UK or France on
  blue-whiting, only Spain Portugal collect age
  data (for Bay of Biscay).
• Stock areas for pelagic species (blue whiting,
  mackerel, horse-mackerel) often very different to
  stock areas for demersal species.
• Very time consuming to prepare data files in
  required format (now automatic extraction process
  available)
• Very data hungry ! (Requires lots of data,
  including stomach content data)

15
The blue-whiting problem!
Celtic Sea
16
Blue Whiting Model 28-05-03
17
Blue-whiting single-species model
GADGET BWG Model
VPA BWG Stock
3500
120000
3000
100000
2500
80000
2000
Stock Numbers (millions)
Stock Numbers (millions)
60000
1500
40000
1000
500
20000
0
0
1991
1993
1995
1997
1999
1991
1993
1995
1997
1999
Year
Year
Year
18
Blue-whiting recruitment
VPA BWG Stock
GADGET BWG Model
45000
1200
40000
1000
35000
30000
800
25000
Numbers 1 year olds (millions)
Numbers 1 year olds (millions)
600
20000
15000
400
10000
200
5000
0
0
2000
1984
1988
1992
1996
2000
1984
1988
1992
1996
2000
19
Cod-Whiting Multispecies Model
Cod Recruitment
18000
16000
14000
12000
10000
XSA/VPA
Numbers 1 year olds (thousands)
8000
GADGET
6000
4000
2000
Model period
Run-up period
0
1984
1986
1988
1990
1992
1994
1996
1998
2000
20
Cod-Whiting Multispecies Model
Whiting Recruitment
12000
VPA/XSA
10000
8000
6000
4000
2000
0
Numbers 2 year olds (thousands)
600000
GADGET
500000
400000
300000
200000
100000
0
1984
1986
1988
1990
1992
1994
1996
1998
2000
21
GADGET Cod-Whiting Multispecies Model
25000000
1000000000
Cod
Whiting
900000000
800000000
20000000
700000000
600000000
15000000
500000000
400000000
10000000
300000000
200000000
5000000
100000000
0
0
1984
1985
1986
1987
1988
1989
1990
1991
1992
1993
1994
1995
1996
1997
1998
1999
22
Linking single-species models together.

• 26,000 stomach content records have been entered
  for the Celtic Sea.
• Data span 1977-1994
• 1691 Cod stomachs, 1749 whiting stomachs, 179
  blue-whiting stomachs
• French data (1977-1988) give numbers and weight
  of prey consumed. No data on empty stomachs or
  unknown prey
• UK data (1990-1994) give numbers of prey consumed
  and number of empty stomachs
• Some data on prey (fish) lengths

23
Cod, Gadus morhua

• 1691 Stomachs examined
• 3 Stomachs empty
• 7424 Prey items

100
90
80
70
22
60
Percentage
50
40
30
20
10
0
73
73
Predator Length (cm)

24
Predator length vs prey length - cod
40
35
30
25
Prey Length (cm)
20
15
10
5
0
0
20
40
60
80
100
120
Predator Length (cm)
y 0.24x 3.44
25
Whiting, Merlangius merlangus

• 1749 Stomachs examined
• 13 Stomachs empty
• 2072 Prey items

100
90
80
70
2
60
16
Percentage
50
40
30
20
10
0
15-
20-
25-
30-
35-
40-
45-
50-
55-
60-
65-
19
24
29
34
39
44
49
54
59
64
69
Predator Length (cm)

81
26
Predator length vs prey length - whiting
30
25
20
Prey Length (cm)
15
10
5
0
0
10
20
30
40
50
60
70
Predator Length (cm)
y 0.21x 4.21
27
Conclusions summary

• All necessary data has now been entered into DST2
  data warehouse
• Single-species models for cod, whiting and
  blue-whiting are giving sensible values and
  seem to be at some sort of optimum.
• Cod and whiting models have been joined together
  (in terms of predation), but blue-whiting has not
  yet been added
• This is the first attempt at a multi-species
  GADGET model outside of Icelandic waters.
• Next species - hake, megrim or mackerel (using
  automatic extraction process)?