real work by adam langley some slide sorting and comments by ian taylor
DESCRIPTION
MULTIFAN-CL and Stock Synthesis A comparison of the 2010 Indian Ocean yellowfin tuna assessment including tagging data. real work by Adam Langley some slide sorting and comments by Ian Taylor. Introduction. - PowerPoint PPT PresentationTRANSCRIPT
MULTIFAN-CL and Stock SynthesisA comparison of the 2010 Indian Ocean yellowfin tuna assessment
including tagging data
real work by Adam Langleysome slide sorting and comments by Ian Taylor
Introduction
• In 2010, spatial model with tag data developed for Indian Ocean yellowfin tuna in MULTIFAN-CL (MFCL)
• In 2011, model was translated into Stock Synthesis (SS)
• Structural assumptions of two models compared
• Model results compared
Spatial structure
Key structural assumptions• Fixed natural mortality, growth, steepness (0.7).• Growth – SS lacks flexibility of MFCL.• Selectivity – modal structure can not be adequately
fitted with double normal. Most problematic for PS FS fisheries. Cubic splines (new feature in SS).
• Initial Fs (fixed SS, estimated MFCL).• Total recruit deviates 1972-2008• Regional recruitment deviates (1977-2006).• Tag reporting rates.• LL CPUE share q among regions.
Assumptions about tags and movement
• MFCL: tag releases input by length bin• SS: each tag group associated with single age• Each model has different levels of flexibility in
parameterization of – reporting rates (grouping in MFCL, not in SS), – tag loss rates,
• MFCL: separate movement rates for each quarter• SS: set up with quarters as years which doesn’t allow
different movement rates by quarter– (seasonal model would allow it, but has other drawbacks)
5 10 15 20 25
050
100
150
Age class
Leng
th c
m
SSMFCL
24
68
1012
14S
td d
ev L
en-a
t-age
cm
Growth
Can not duplicate the MFCL growth pattern in SS.
Selectivity
SS selectivity shifted to younger fish due to different growth (mean length-at-age).SS cubic spline (5 node) for LL and PS FS, others double normal. Need to resolve problems with SS LL selectivity parameterisation.
0 5 10 20
0.0
0.4
0.8
1. GI 1
0 5 10 200.
00.
40.
8
2. HD 1
0 5 10 20
0.0
0.4
0.8
3. LL 1 Post 1972
0 5 10 20
0.0
0.4
0.8
4. OT 1
0 5 10 20
0.0
0.4
0.8
5. BB 2
0 5 10 20
0.0
0.4
0.8
6. PS FS 2 2003-06
0 5 10 20
0.0
0.4
0.8
7. LL 2 Post 1972
0 5 10 20
0.0
0.4
0.8
8. PS LS 2 2003-06
0 5 10 20
0.0
0.4
0.8
9. TR 2
0 5 10 20
0.0
0.4
0.8
10. LL 3 Post 1972
0 5 10 20
0.0
0.4
0.8
11. LL 4 Post 1972
0 5 10 20
0.0
0.4
0.8
12. GI 5
0 5 10 20
0.0
0.4
0.8
13. LL 5 Post 1972
0 5 10 20
0.0
0.4
0.8
14. OT 5
0 5 10 20
0.0
0.4
0.8
15. TR 5
0 5 10 20
0.0
0.4
0.8
16. PS FS 3
0 5 10 20
0.0
0.4
0.8
17. PS LS 3
0 5 10 20
0.0
0.4
0.8
18. TR 3
0 5 10 20
0.0
0.4
0.8
19. PS FS 5
0 5 10 20
0.0
0.4
0.8
20. PS LS 5
0 5 10 20
0.0
0.4
0.8
21. PS FS 2 Pre 2003
0 5 10 20
0.0
0.4
0.8
22. PS LS 2 Pre 2003
0 5 10 20
0.0
0.4
0.8
23. PS FS 2 2007-09
0 5 10 20
0.0
0.4
0.8
24. PS LS 2 2007-09
Pro
porti
on
Movement coefs
Differences in parameterisation – SS not seasonal, recruit at 0 age, ramp (0,1 ages).SS high movement from 3 to 2 and 2 to 3 – consistent with MFCL. Recruitment in region 4 moving to 2 (differs from MFCL).
1 to 1 1 to 2 1 to 3 1 to 4 1 to 5 2 to 1 2 to 2 2 to 3 2 to 4 2 to 5 3 to 1 3 to 2 3 to 3 3 to 4 3 to 5 4 to 1 4 to 2 4 to 3 4 to 4 4 to 5 5 to 1 5 to 2 5 to 3 5 to 4 5 to 5
0.0
0.2
0.4
0.6
0.8
SS movement rate estimates
MFCL movement rate estimates
Movement - MFCL
LL CPUE indices
MFCL has seasonal catchability deviates. Incorporated seasonal catchability in SS model by splitting LL CPUE index by season.
1980 1990 2000 2010
0.00
0.05
0.10
0.15
0.20
0.25
0.30
0.35
Region 1
1980 1990 2000 2010
0.0
0.2
0.4
0.6
0.8
1.0
1.2
Region 2
1980 1990 2000 2010
0.0
0.5
1.0
1.5
Region 3
1980 1990 2000 2010
0.0
0.1
0.2
0.3
0.4
Region 4
1980 1990 2000 2010
0.0
0.2
0.4
0.6
0.8
1.0
Region 5
Quarter 1Quarter 2Quarter 3Quarter 4
SS Length data (aggregated by fishery)
50 100 150 200
0.0
1.0
2.0
3.0
1. GI 1
50 100 150 200
0.0
1.0
2.0
2. HD 1
50 100 150 200
01
23
45
3. LL 1 Post 1972
50 100 150 200
0.00
0.10
0.20
4. OT 1
50 100 150 200
02
46
8
5. BB 2
50 100 150 200
0.0
0.4
0.8
6. PS FS 2 2003-06
50 100 150 200
02
46
7. LL 2 Post 1972
50 100 150 200
0.0
0.5
1.0
1.5
8. PS LS 2 2003-06
50 100 150 200
02
46
10. LL 3 Post 1972
50 100 150 200
02
46
8
11. LL 4 Post 1972
50 100 150 200
02
46
8
12. GI 5
50 100 150 200
02
46
13. LL 5 Post 1972
50 100 150 200
01
23
4
14. OT 5
50 100 150 200
01
23
45
6
15. TR 5
50 100 150 200
0.0
1.0
2.0
3.0
16. PS FS 3
50 100 150 200
02
46
17. PS LS 3
50 100 150 200
0.0
0.4
0.8
19. PS FS 5
50 100 150 200
0.0
1.0
2.0
20. PS LS 5
50 100 150 200
0.0
1.0
2.0
21. PS FS 2 Pre 2003
50 100 150 200
02
46
22. PS LS 2 Pre 2003
50 100 150 200
0.0
0.2
0.4
0.6
23. PS FS 2 2007-09
50 100 150 200
0.0
0.4
0.8
1.2
24. PS LS 2 2007-09
Pro
porti
on
MFCL Length data (aggregated by fishery)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Fishery
Rep
ortin
g R
ate
0.0
0.2
0.4
0.6
0.8
SSMFCL
Tag reporting rate (SS vs MFCL)
PS Fisheries 23 and 24 are the only fisheries with considerable numbers of recoveries. For MFCL, these fisheries share a reporting rate (equivalent to mean of SS reporting rates for the two fisheries).
Tag recoveries – main release groups
244 246 248 250 252
050
100
150
Quarter
Num
ber o
f tag
s (o
bs, p
red)
240 244 248 252
050
100
150
Quarter
Num
ber o
f tag
s (o
bs, p
red)
240 244 248 252
050
100
150
Quarter
Num
ber o
f tag
s (o
bs, p
red)
238 242 246 250
050
100
150
Quarter
Num
ber o
f tag
s (o
bs, p
red)
244 246 248 250 252
050
100
150
200
Num
ber
of ta
gs (o
bs, p
red)
240 244 248 252
050
100
150
200
Num
ber
of ta
gs (o
bs, p
red)
240 244 248 252
050
100
150
200
250
Num
ber
of ta
gs (o
bs, p
red)
238 242 246 250
010
020
030
040
050
0
Num
ber
of ta
gs (o
bs, p
red)
Tag recoveries (F23 and F24)
235 240 245 250
050
010
0015
00
Quarter
Num
ber o
f tag
s (o
bs2,
pre
d2)
Fisheries were not operating before quarter 240 (= 1st Q 2007).
Recruitment (total)
1980 1990 2000 2010
050
000
1000
0015
0000
2000
00
Rec
ruitm
ent
SS Age0MFCL Age1
1980 1990 2000 2010
0e+0
04e
+04
8e+0
4
Rec
ruitm
ent
Region 1
1980 1990 2000 2010
020
000
4000
060
000
Rec
ruitm
ent
Region 2
1980 1990 2000 2010
020
000
6000
010
0000
1400
00
Rec
ruitm
ent
Region 3
1980 1990 2000 2010
010
000
2000
030
000
4000
0
Rec
ruitm
ent
Region 4
1980 1990 2000 2010
010
000
3000
050
000
Rec
ruitm
ent
Region 5
Rec
ruitm
ent (
1000
s of
fish
)
SS Age0MFCL Age1
Spawning biomass
SS Age specific spawning OGIVE shifted to younger fish to account for faster initial growth.
1980 1990 2000 2010
020
0040
0060
0080
0010
000
Adu
lt bi
omas
s 10
00s
mt
SSMFCL
1980 1990 2000 2010
010
020
030
040
050
060
0
Adul
t bio
mas
s 10
00s
mt
Region 1
1980 1990 2000 2010
050
015
0025
0035
00
Adul
t bio
mas
s 10
00s
mt
Region 2
1980 1990 2000 2010
010
0020
0030
00
Adul
t bio
mas
s 10
00s
mt
Region 3
1980 1990 2000 2010
010
020
030
040
0
Adul
t bio
mas
s 10
00s
mt
Region 4
1980 1990 2000 2010
010
0020
0030
0040
00
Adul
t bio
mas
s 10
00s
mt
Region 5
Adu
lt bi
omas
s 10
00s
mt
SSMFCL
Terminal F-at-age
Differences in F-at-age for younger age classes mainly attributable to differences in assumed growth.
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0.18
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27
SS
MFCL
SS3 performance
• Seven phases. 772 parameters (rec_dev * region = 600 parameters).
• 4GB RAM machine. 3-4 hours w/o hessian.• Convergence (grad = 0.00065).• Hessian, covariance matrix computed.• Key parameters fixed (e.g. Growth) as per
MFCL.• Obj fnt 19972.1 (cf MFCL 290504.6). Obj fnts
not comparable – e.g. MFCL effort deviates.
To do/issues
• Initial F (in 1972).• Selectivity – cubic splines for fisheries with bimodal
structure.• Seasonal LL catchability for CPUE indices.• Recruitment deviation period – constrain last year.• Tag recoveries – aggregation of PS recoveries
(associated and unassociated).• Assignment of tags to age class at release (refinement).• Growth patterns (region specific).• Definition of reference points.
Conclusions• Some differences in dynamics.• Assumptions regarding growth influential.• But, overall (very) similar results from the two
platforms. Derivation of MSY based reference points differ between MFCL and SS.
• Pros and cons of both platforms. SS time varying parameterization; less flexible wrt growth.
• Useful exercise to routinely compare and contrast results – more rigor when considering structural assumptions.