Gulf of Maine Atlantic cod
Groundfish Operational Assessment MeetingSeptember 11-15, 2017Northeast Fisheries Science CenterWoods Hole, MA
Michael PalmerNEFSC/Population Dynamics Branch
2017 Assessment Update: Presentation Overview
• Presentation orientation
• Overview of the previous stock assessment (2015)
• Major signals in the fishery and survey data• Model independent evaluation of stock condition
• Data assembly, modelling, reference point and projection approach
• Assessment data inputs and changes from the 2015 assessment
• Model diagnostics and results
• Updated biological reference points and stock status
• Preliminary short-term (2018-2020) catch and biomass projections
Presentation orientation
• Random things correlated with Gulf of Maine cod SSB
Table X, Figure Y
0
1
2
3
4
5
6
7
0
5000
10000
15000
20000
25000
1982
1985
1988
1991
1994
1997
2000
2003
2006
2009
2012
2015
US pe
r cap
ita avocado
consum
ption (lb
)
GOM cod
SSB
(mt)
Year
SSB Avocado consumption
Time series n Pearson r p-valueUS population (1982-2016) 35 -0.77 <0.001Registered vehicles on US roads (1990-2015) 26 -0.57 0.002US unemployment rate (1982-2016) 35 0.26 0.13US per capita avocado consumption (2000-2014) 15 -0.90 <0.001Worldwide Netflix subscribers (2007-2016) 10 -0.79 0.006Scientific articles referencing Gulf of Maine cod (1982-2016) 35 -0.73 <0.001
2015 Assessment Update
• Stock Status
• Issues and uncertainties:• Two accepted models (M=0.2, M-ramp), overall uncertainty in current
natural mortality• M=0.2 model exhibited major retrospective patter, M-ramp had minor
retrospective pattern• Recent low recruitment compromises rebuilding potential of the stock
Adapted from Table 39
Assessment Proxy reference points M=0.2 M-rampFfull, 2014 0.956 (0.664 - 1.387) 0.932 (0.654 - 1.304)FMSY 0.185 0.187Ffull, 2014/FMSY 5.11 4.98Overfishing Yes YesSSB2014 (mt) 2,225 (1,713 - 2,892) 2,536 (1,942 - 3,298)SSBMSY (mt) 40,187 (27,551 - 58,228) 59,045 (44,976 - 76,525)SSB2014/SSBMSY 0.06 0.04Overfished Yes YesMSY (mt) 6,797 (4,608 - 9,990) 10,043 (7,560 - 13,130)Median age1 recruitment (000s) 4,406 (1,458 - 14,450) 8,965 (2,489 - 15,908)
2015 update
Overview of Fishery and Survey Data
• Fishery: catch source
Figure 5
Overview of Fishery and Survey Data
• Fishery: commercial length distributions• Length frequency distributions of the ‘large’ market category
Adapted from Figure 11
Overview of Fishery and Survey Data
• Fishery: recreational harvest (landings) length distributions
Figure 12
Overview of Fishery and Survey Data
• Fishery: commercial and recreational age structure:
Figures 17, 19
Overview of Fishery and Survey Data
• Fishery: mean age and catch-curve estimated total mortality (Z)
Figures 22, 23
Overview of Fishery and Survey Data
• Fishery: spatial distribution
Figures 13, 14
Overview of Fishery and Survey Data
• Fishery: spatial distribution (commercial landings)
Adapted from Figures 15, 16
Overview of Fishery and Survey Data
• Survey: aggregate indices
Figure 26, not in supplemental info
Overview of Fishery and Survey Data
• Survey: indices-at-age
Figures 28-30
Overview of Fishery and Survey Data
• Survey: mean age and catch-curve estimated total mortality (Z)
Figures 34, 35
Overview of Fishery and Survey Data
• Survey: spatial distribution
Figures 36, 39
2017 Assessment Update Overview
• Incorporate two additional years of data (2015-16) into the assessment time series• Minor changes:
• Transition to use of automated multinomial procedure to fill in length bins in survey age-length keys missing age information (Gerritsen et al. 2006).
• Update the SAW/SARC 55 ASAP models through 2016
• Update FMSY proxies (F40%)• Use recent 3-year average weights
• Update SSBMSY proxies
• Update short-term projections (2018-2020)
Model Inputs
• Biology• Maturity-at-age (1982-2016 time series average)• Assumed natural mortality (M=0.2 or M-ramp from 0.20.4)• Stock weights-at-age
• Fishery removals• Commercial landings and discards (ages 1-9+)
• Discard mortality rate varies by gear• Recreational landings and discards (ages 1-9+)
• Discard mortality rate = 15%• Catch weights-at-age
• Surveys• NEFSC spring and fall (ages 1-9+)• MADMF spring (ages 1-6)
Model Inputs
• Data changes:• Transition to use of
automated multinomial procedure to fill in length bins in survey age-length keys missing age information (Gerritsen et al.2006).
Figure 32 (example)
Model Formulation
• Model formulation:• Years: 1982-2016• Fishery:
• Single fleet (combined commercial and recreational)• CV = 0.05• Selectivity modelled with three selectivity blocks (all single logistic)
• 1982-1988, 1989-2004, 2005-2016• Surveys:
• CVs from surveys with re-weighting• NMFS spring:+0.2, NMFS fall:+0.1, MADMF spring:+0.3
• Selectivity freely estimated at-age – fixed 6+ (NEFSC) and 1 (MADMF)• Catchability does not vary over time
• Recruitment modelled as deviations from the mean• Lambda=0.2, CV=0.5
• Two models: M=0.2, M-ramp• M=0.2:1982-1988, ramp: 1989-2002, M=0.4: 2003-2016
Model Diagnostics
• Selectivity and survey catchability
Table 31
Value CV Value CVA50 2.33 0.05 2.33 0.05
slope up 0.46 0.09 0.45 0.09A50 3.33 0.02 3.36 0.02
slope up 0.55 0.05 0.52 0.05A50 3.66 0.03 3.71 0.03
slope up 0.57 0.05 0.54 0.05Age1 0.04 0.23 0.03 0.23Age2 0.14 0.19 0.12 0.19Age3 0.30 0.18 0.27 0.18Age4 0.54 0.18 0.52 0.18Age5 0.74 0.19 0.73 0.18Age6 1.00 1.00Age7 1.00 1.00Age8 1.00 1.00
Age9+ 1.00 1.00Age1 0.15 0.25 0.12 0.25Age2 0.35 0.24 0.31 0.24Age3 0.56 0.24 0.54 0.24Age4 0.80 0.24 0.79 0.24Age5 0.93 0.26 0.95 0.26Age6 1.00 1.00Age7 1.00 1.00Age8 1.00 1.00
Age9+ 1.00 1.00Age1 1.00 1.00Age2 0.69 0.15 0.78 0.15Age3 0.69 0.17 0.85 0.17Age4 0.72 0.21 0.93 0.20Age5 0.74 0.29 0.99 0.29Age6 0.55 0.56 0.75 0.56
NEFSC spring 1.08 0.16 0.94 0.17NEFSC fall 0.64 0.22 0.55 0.23MADMF spring 0.21 0.12 0.14 0.12
Survey catchability (q)
M=0.2 M-ramp
Block 3 (2005-2016)
Block 2 (1989 - 2004)
Block 1 (1982-1988)
NEFSC spring
NEFSC fall
MADMF spring
ParameterBlock/survey
Model Diagnostics
• Model fits to total catch and aggregate survey indices
Figure 45
Model Diagnostics
• Model fits to catch-at-age
Figure 47
Model Diagnostics
• Model fits to survey indices-at-age
Adapted from Figures 48-50
Model Results
• Model results• 2017 update results are
consistent with the 2015 update results
• SSB2016• M=0.2: 3,046 mt• M-ramp: 3,262 mt
• F2016• M=0.2: 0.228• M-ramp: 0.237
• Recruitment• Continues to be poor
under both models
Figure 53
Model Results
• Recruitment (000s fish)• M=0.2
• 1982-2014 median: 4,390• 2012-2016 geomean: 1,174
• M-ramp• 1982-2014 median: 8,508• 2012-2016 geomean: 2,067
• Differences in scale between two models is primarily a function of the increase in natural mortality between 1989 and 2016
Figures 56, 57
Model Results
• Retrospective error• Mohn’s rho:
• SSB: M=0.2 (0.53), M-ramp (0.30)• F: M=0.2 (-0.31), M-ramp (-0.17)
Figures 58, 59
Model Results
• Retrospective error• M=0.2: major retrospective error• M-ramp: minor retrospective error
• Similar results to the 2015 update – no retrospective adjustment made for the M=0.2 model.
Not in supplemental info, Figure 60
Assessment Model SSB F Age-1M=0.2 0.47 -0.32 0.84M-ramp -0.01 0.04 0.20M=0.2 0.53 -0.33 1.05M-ramp 0.17 -0.05 0.42M=0.2 0.54 -0.31 1.02M-ramp 0.20 -0.08 0.42M=0.2 0.53 -0.31 0.88M-ramp 0.30 -0.17 0.41
2014 update
2015 update
2017 update
SARC 55
Biological Reference Points
• Update F40% FMSY proxies• Natural mortality assumed equal to 0.2• Time series average maturity ogive• 3-year average of weights (2014-2016)• Last selectivity block (2004-2016)
Table 37, Figure 62
AgeNatural
mortalityFraction mature
Jan1/SSB weights
(kg)
Mid-year weights
(kg)
Catch weights
(kg)
Fishery selectivity (M = 0.2)
Fishery selectivity (M -ramp)
1 0.200 0.087 0.105 0.253 0.251 0.009 0.0072 0.200 0.318 0.486 0.720 0.840 0.052 0.0413 0.200 0.697 1.028 1.379 1.643 0.240 0.2134 0.200 0.919 1.749 2.065 2.709 0.646 0.6325 0.200 0.982 2.422 2.971 3.519 0.913 0.9166 0.200 0.996 3.416 3.837 4.473 0.984 0.9867 0.200 0.999 3.944 4.429 4.803 0.997 0.9988 0.200 1.000 5.644 7.006 7.160 1.000 1.000
9+ 0.200 1.000 10.895 10.894 10.894 1.000 1.000
Biological Reference Points
• Update SSBMSY proxies• Based on 100 year projections run at the FMSY proxy• Projection model samples from CDF of recruitment from 1982-2014
• When SSB drops below a hinge point recruitment declines linearly to zero• M=0.2: 6,300 mt, M-ramp: 7,900 mt
Figure 63
Stock Status
• Updated reference points and stock status• Overfished and overfishing is occurring
Adapted from Table 39
Assessment Proxy reference points M=0.2 M-rampFfull, 2016 0.228 (0.165 - 0.316) 0.237 (0.169 - 0.331)FMSY 0.174 0.177Ffull, 2016/FMSY 1.31 1.34Overfishing Yes YesSSB2016 (mt) 3,046 (2,301 - 4,025) 3,262 (2,464 - 4,270)SSBMSY (mt) 40,604 (27,631 - 58,553) 59,714 (44,732 - 77,611)SSB2016/SSBMSY 0.08 0.05Overfished Yes YesMSY (mt) 7,049 (4,699 - 10,380) 10,502 (7,734 - 13,822)Median age1 recruitment (000s) 4,377 (1,161 - 14,434) 8,464 (2,353 - 15,934)
2017 update
Short-term Projections
• Updated short-term projections (2018-2020)• M=0.2: base projection and retro-adjusted projection• M-ramp: Two assumptions of natural mortality in the short-term
• M=0.2 or M=0.4
• All projections run at FMSY• Assumed 2017 catch of 428 mt (NEFMC PDT)
Table 40
Catch (mt)
Spawning stock
biomass (mt)
Ffull Catch (mt)
Spawning stock
biomass (mt)
Ffull Catch (mt)
Spawning stock
biomass (mt)
FfullCatch (mt)
Spawning stock
biomass (mt)
Ffull
2016 Model result 599 3,046 0.228 599 1,997 0.332 599 3,262 0.237 599 3,262 0.2372017 Assumed catch 428 4,648 0.092 428 3,041 0.142 428 4,470 0.100 428 4,245 0.1102018 Projection 1,084 5,574 0.174 693 3,569 0.174 1,066 5,558 0.177 791 4,312 0.1772019 Projection 1,181 6,553 0.174 758 4,214 0.174 1,215 7,070 0.177 741 4,572 0.1772020 Projection 1,326 8,401 0.174 855 5,426 0.174 1,505 10,046 0.177 769 5,529 0.177
Year Input
Retrospective adjustment M=0.2M=0.2 model
No retro adjustment M=0.4M-ramp model
Scale and trend
• Results from MA DMF Atlantic cod Industry Based Survey (Dean et al. 2017) provide independent estimates of resource trend and scale
• Swept area biomass estimates from 2003-07 and 2016-17 provide similar perspective of the resource• Catchability q estimated at ~0.97 (less for fish < 40 cm)• Uncertainty in degree of herding between doors• Coverage of population biomass ~74%
IBS results from Dean et al. (2017)
Uncertainty
• Natural mortality• Less important in context of current scale of the population and level of
fishing mortality (similar between models)• More important in understanding future productivity of the resource and
rebuilding targets
• Stock structure• Ongoing process to investigate stock structure
• Model diagnostics• Survey residuals (e.g., tension between NEFSC spring and MA DMF spring)• Retrospective error
Uncertainty
• Veracity of fishery removal estimates• Recreational
• Private mode effort statistics and MRIP recalibration• Commercial
• Landings• Vessel stock-area reporting errors (Palmer 2017)• Dealer misreporting
• Discards• Reports of observer effects and unknown impacts on discard
estimates
Adapted from Palmer (2017)
Summary
• Consistent signals across data sources, models and approaches
• Fishery and survey data continue to show few large/old fish and few incoming recruits
• Survey indices and percent occurrence remains low
• Model results• F2016: 0.228/0.237• SSB2016: 3,046/3,262 mt• Recruitment continues to be poor
• Biological reference points (proxy)• FMSY: 0.174/0.177• SSBMSY: 40,604/59,714 mt
• Stock remains overfished and overfishing is occurring
Questions?