Atlantic Herring MSE
5. Herring (January 24-26, 2017) #1
Atlantic Herring MSE Part I ‐ Data and Methods Overview Part II – Preliminary results Dr. Jonathan Deroba, NEFSC NEFMC January 2017 Council Meeting Portsmouth, NH
1
Part I: Data and Methods Overview Multiple operating models represent uncertainty Defined in Workshop #1
Sarah Gaichas
Herring N, B, Wt Herring recruitment (high or low?) Herring natural mortality (high or low?) Herring growth (good or poor?) Herring assessment error/bias (yes or no?)
Min‐Yang Lee
Evaluate ABC control rules for each OM
2
Recruitment and Natural Mortality define Hi production and Lo production operating models
3
Recruitment and Natural Mortality define Hi production and Lo production operating models
Based on longevity and size data, and stock assessment data and fits 4
Uncertainties At the May Workshop we identified uncertainties: Herring recruitment Herring natural mortality Herring growth Herring assessment error/bias
5
Growth good and poor growth operating models
Based on survey weight at age data
6
Uncertainties At the May Workshop we identified uncertainties: Herring recruitment Production Herring natural mortality Hi Lo x Herring growth x Herring assessment error/bias x x
Growth Good Poor x x x x
7
Assessment Error and Bias unbiased and biased operating models
8
Assessment Error and Bias unbiased and biased operating models Biased assessment results in biomass higher than reality
Based on the stock assessment retrospective pattern
9
Uncertainties At the May Workshop we identified uncertainties: Production Growth Herring recruitment Hi Lo Good Poor x x Herring natural mortality x x Herring growth x x x x Herring assessment error/bias x
x x
x x x
x x
Assessment bias On Off x x x x x x x x
Uncertainties combined into 8 different operating models 10
Fishery Selectivity
Based on purse seine and MWT age composition data
11
Predator models Stolen or adapted from presentations By Dr. Sarah Gaichas, NEFSC
12
Predator models
Are
Are not
• Focused on evaluating stock‐ wide herring ABC harvest control rules applied annually • Developed balancing Council/ stakeholder specifications and time constraints of MSE • Based on information from the Northeast US shelf and most recent stock assessments
• Spatial, do not address local scale or seasonal dynamics • New or full stock assessments • Accounting for any impacts on predators other than changes due to herring control rules • Intended to predict actual predator population dynamics 13
Two components of predator modeling Predator population model
Herringpredator relationship
• Delay‐difference dynamics • Information required:
• What about herring…
• • • • •
Stock‐recruitment relationship Natural mortality rate Fishing mortality rate Initial population size Weight at age
• Assessments or observations
• Total abundance? Biomass? • Certain ages or sizes?
• Affects what about the predator • Predator growth • Predator reproduction • Predator survival
• And how? Base on observations 14
Predator population model summary Highly migratory
Seabird
Groundfish
Marine mammal
Stakeholder preferred species
Bluefin tuna
Common tern
Not specified
Not specified
Species modeled
Bluefin tuna (western Atlantic stock)
Common tern (Gulf of Maine colonies as defined by GOM Seabird Working Group)
Spiny dogfish (GOM and GB Atlantic cod stocks also examined)
None, data limited (Minke & humpback whales, harbor porpoise, harbor seal examined)
Stock‐recruitment (or adults, recruits)
Porch and Lauretta 2016, ICCAT 2015
Derived from GOMSWG data
Rago and Sosebee 2010
No time series data for our region
Natural mortality
ICCAT 2015
Nisbet 2002
Rago and Sosebee 2013, 2015
Derivable from Waring et al. 2015?
Fishing mortality
ICCAT 2015
n/a
Rago 2016
Waring et al. 2015?
Initial population
ICCAT 2015
GOMSWG data
Rago 2016
Waring et al. 2015?
Weight at age
Restrepo et al. 2010 Nisbet 2002
Rago et al. 1998
General literature 15
Predator‐prey relationships: Northeast US Herring vs. Antarctic krill
16
HerringPredator relationship issues/caveats • Predator populations are affected by MANY factors, prey is one • Northeast US predators have MANY prey options, herring is one • Time limitation enforced model simplicity for these complex relationships • Our approach is to use the best‐ supported relationship for each predator based on observations from the Northeast US ecosystem
• Isolating a clear herringpredator relationship from observations is difficult or impossible (e.g. cod) • Even with good observations, the modeled herringpredator relationship may require strong assumptions and not be statistically significant (e.g. terns) • Apparent positive herring predator relationships may not arise from the modeled mechanism (e.g. dogfish) 17
Predator relationships summary Predator and overlap
Modeled herring relationship
• Western Atlantic bluefin tuna
Herring population average weight affects bluefin tuna growth
Forage throughout North Atlantic, seasonally in GOM
• Common terns Forage seasonally near island breeding colonies in GOM
• Spiny dogfish Forage through same range as herring most of the year
• Marine mammals
Herring total biomass affects common tern reproductive success (productivity) Herring total abundance affects dogfish survival Food web model simulations 18
Tuna condition
Herring biomass
0.0230
Herring size, Gulf of Maine
“The decline in bluefin tuna condition, 0.0220 despite high prey biomass in the Gulf of 0.0215 Maine, suggests that managing for high 0.0210 abundance at middle trophic levels does 0.0205 not guarantee the success of all top 0.0200 predators. In fact, it suggests that for 0.0195 0.0190 some upper level predators, the quality 0.0185 of the prey may be more important than 0.0180 the overall abundance.” 0 0.05 0.1 0.15 0.2 0.25 0.3
Growth Intercept
0.0225
Modeled relationship: Tuna grow better with heavier herring overall, and/or with a higher proportion of large herring in the population
Herring Avg Wt
19
Tuna modeling notes • Available data do not support a positive relationship between herring and tuna populations: 1200000
60000
500000
Bluefin Tuna Recruitment
450000 400000 350000
Bluefin Tuna Spawning Stock Biomass
50000
600000
30000
20000
400000
20000
10000
200000
10000
30000
200000 100000 50000 0
0 0
500,000
1,000,000
1,500,000
50000 40000
300000
150000
1000000 800000
40000
250000
60000
BFTrec HerringSSB BFTSSB
0
500,000
1,000,000
Northeast US Herring Spawning Stock Biomass
1,500,000
0 1960
1970
1980
1990
2000
0 2010
Year
• Our models do not address herring/tuna interactions in a specific place or time. Tuna follow herring and likely aggregate around herring while feeding. • We can draw no conclusions from our modeling about predator/prey co‐occurrence at the local scale. • Similarly, without additional observations, we cannot extrapolate local scale co‐occurrence to population level relationships. 20
21
Herring Common tern
1.10 Predator Recruit Multiplier
1.08 1.06 1.04 1.02 1.00 0.98 0.96 0.94 0.92 0.90 0
500000
1000000
1500000
2000000
2500000
Herring Abundance
Modeled relationship: Common tern productivity is improved when herring total biomass above a threshold (400,000 t). Productivity diminishes below this threshold.
22
Top groundfish predators of herring (NEFSC, 1972‐2015)
23
Dogfish, Georges Bank cod and Gulf of Maine cod all ate herring in proportion to herring abundance, 1972‐2015. However, increased herring in diet was positively related to spawning stock biomass only for dogfish.
24
Herring Dogfish
0.100
Predator Annual Natural Mortality
0.090 0.080 0.070 0.060 0.050 0.040 0.030 0.020 0.010 0.000 0
1000000
2000000
3000000
4000000
5000000
6000000
Herring Abundance
• The dogfish relationship assumes herring abundance improves dogfish survival because no clear relationship was found with recruitment or growth. • Increased survival may not be the mechanism for the observed positive influence of herring in diet on the dogfish population. 25
Predator Modeling Summary • Our models are designed for evaluating alternative herring control rules, not predator stock assessment and population prediction. • We caution against generalizing results for these particular predators to other predators, as population parameters and herring relationships differ. • Although we selected predators with high herring diet proportions, observed predator population responses to herring alone do not dominate dynamics, and our herringpredator models reflect that. • Predator responses to aggregate prey dynamics are likely to be much clearer than responses to individual prey in the Northeast US ecosystem given its food web structure. 26
Herring Output Metrics – From Workshop #1 • Spawning Stock Biomass & SSB relative to SSBMSY & SSBunfished • Probability that • SSB =1 • Frequency that tuna weight > average
28
Output Metrics Economic • Yield: output of the herring model • Net revenue = (price*yield) – cost • Stability = the degree to which net revenue was “stable” or “streaky” (i.e., fairly steady over time vs. booms and busts) • Net revenue and stability demonstrate similar tradeoffs as herring yield and variation in yield, and so not presented in detail
29
Pause for Questions on Part I? Data and methods developed for MSE models (herring, predator and economic) 30
Fishing Mortality or Catch
Part II –Analysis of Potential Control Rules
? Biomass or Abundance
31
Six Control Rule Types presented at Workshop #2 1. 2. 3. 4. 5. 6.
Biomass based Biomass based with 3 year block Biomass based with 5 year block Biomass based with 3 year block and 15% restriction Constant catch Conditional constant catch with max F = 0.5Fmsy
32
Control Rules biomass based Three ‘parameters’ with many variants
Fishing Mortality
Upper biomass parameter Max F parameter
Biomass or Abundance
Lower biomass parameter 33
Control Rules biomass based Three ‘parameters’ with many variants
Fishing Mortality
Upper biomass parameter = Lower biomass parameter = 0
Max F parameter
Biomass or Abundance 34
Control Rules biomass based Three ‘parameters’ with many variants
Fishing Mortality
Upper biomass parameter = Lower biomass parameter > 0
Max F parameter
Biomass or Abundance 35
Control Rules biomass based Three ‘parameters’ with many variants Evaluated 16 different values for each biomass threshold ranging from 0 to 4x Bmsy Evaluated 10 different values for maximum F ranging from 0.1 to 1.0x Fmsy 1,360 combinations
36
Control Rules status quo – biomass based with 3 year block
37
Control rules Constant Catch
Catch
One parameter Evaluated 10 different values ranging from 0.1 to 1x MSY
Biomass or Abundance 38
Control Rules Conditional Constant Catch
Biomass or Abundance
0.5Fmsy
Catch
Fishing Mortality
Two parameters Evaluated 10 different values ranging from 0.1 to 1x MSY with max F of 0.5Fmsy
Biomass or Abundance 39
Control Rule Types and Shapes Biomass based Biomass based with 3 year block Biomass based with 5 year block Biomass based with 3 year block and 15% restriction Constant catch Conditional constant catch with max F = 0.5Fmsy
1,360 alternatives 1,360 alternatives 1,360 alternatives 1,360 alternatives 10 alternatives 10 alternatives 5,460 alternatives x 8 operating models 43,680 40
Control Rules For each operating model, each control rule alternative was simulated for 150 years and this was repeated for 100 simulations
41
Prelim Results – Herring Yield vs Stability
42
Frequency Dogfish > 0.5Bmsy vs herring SSB
43
Frequency tern prod > 1.0 vs herring SSB
44
Frequency tuna good condt’n vs herring SSB good herring growth
45
Post Workshop #2 • At workshop: less support for BB with 15% restriction and CC/CCC ‐ Too much yield lost for short‐term stability, poor performance elsewhere ‐ More likely to require short‐term deviations in application
• Post workshop: Herring AP and Committee tasked with: 1) identifying priority metrics and tradeoffs; and 2) identifying a reasonable number of CR alternatives. • Herring PDT prepared 4 example control rule shapes, and evaluated their performance for a handful of possible metrics. 46
Control Rules Biomass based 1,360 alternatives Biomass based with 3 year block 1,360 alternatives Biomass based with 5 year block Biomass based with 3 year block and 1,360 alternatives 1,360 alternatives 15% restriction Constant catch 10 alternatives Conditional constant catch with max F 10 alternatives = 0.5Fmsy 4,080 alternatives x 8 operating models 43,680 32,640 47
Let’s Compare 4 CRs
48
Let’s Compare 4 CRs
49
Let’s Compare 4 CRs
50
Let’s Compare 4 CRs
51
Let’s Compare 4 CRs 1. Examine tradeoffs and uncertainty in tradeoffs 2. Examine effect of assessment bias 3. Examine effect of annual ABC, 3 year blocks, and 5 year blocks
52
Note some CRs more robust than others
1.
Examine tradeoffs and uncertainty in tradeoffs
Let’s Compare 4 CRs all with 3 year block – unbiased assessment Yield/MSY vs. SSB/unfished
Yield/MSY vs. Frequency Overfished
53
Results more certain here
1.
Examine tradeoffs and uncertainty in tradeoffs
Let’s Compare 4 CRs
all with 3 year block – unbiased assessment Yield/MSY vs. Variation in yield
Frequency Overfished vs. Zero ABC (no fishery)
54
Note some CRs more robust than others
1.
Examine tradeoffs and uncertainty in tradeoffs
Let’s Compare 4 CRs all with 3 year block – unbiased assessment Yield/MSY vs. Frequency SSB Avg. vs. SSB/unfished
Freq. tern prod. >=1 vs. SSB/unfished
56
1.
Examine tradeoffs and uncertainty in tradeoffs
Let’s Compare 4 CRs all with 3 year block – unbiased assessment Frequency dogfish>0.5Bmsy vs. SSB/unfished
57
1.
Let’s Compare 4 CRs
Examine tradeoffs and uncertainty in tradeoffs
all with 3 year block – unbiased assessment
Herring Yield vs. SSB
58
Let’s Compare 4 CRs • What metrics/tradeoffs do you value most? • For example, if you highly value yield then you likely favor CRs with certainty in high amounts of yield, but do you get “acceptable” performance for other metrics?
Note some CRs more robust than others, esp. Freq. Overfished
2.
Examine effect of assessment bias
Let’s Compare 4 CRs
all with 3 year block – effect of assessment bias Yield/MSY vs. SSB/unfished
Yield/MSY vs. Frequency Overfished
60
Let’s Compare 4 CRs • Relying solely on biased results may duplicate other processes • We make adjustments for bias (e.g., retrospective adjustments) • We have peer review • We have an SSC
• Robustness to bias, which did vary among CRs, desirable
3.
Longer blocks cost yield and SSB
Examine effect of annual ABC, 3 year blocks, and 5 year blocks
Status Quo CR – comparing 1, 3, and 5 years Yield/MSY vs. SSB/unfished
Yield/MSY vs. Frequency Overfished
62
Let’s Compare 4 CRs • Is the short‐term stability of longer blocks worth the cost in: yield, long‐term variation in yield, frequency of overfished, decrease in frequency of desired tern production? • What is industry’s preferred planning horizon?
63
Other Considerations • Other tradeoffs of interest? • Identifying or refining CR alternatives can by achieved by: • Specifying preferred performance for various metrics • Moving CR parameters (Hi and low thresholds and Max‐F); “What if?”
64
Pause for Questions on Part II? Preliminary analyses of potential control rule alternatives 65
Atlantic Herring MSE Part I ‐ Data and Methods Overview Part II – Preliminary results Dr. Jonathan Deroba, NEFSC NEFMC January 2017 Council Meeting Portsmouth, NH
1
Part I: Data and Methods Overview Multiple operating models represent uncertainty Defined in Workshop #1
Sarah Gaichas
Herring N, B, Wt Herring recruitment (high or low?) Herring natural mortality (high or low?) Herring growth (good or poor?) Herring assessment error/bias (yes or no?)
Min‐Yang Lee
Evaluate ABC control rules for each OM
2
Recruitment and Natural Mortality define Hi production and Lo production operating models
3
Recruitment and Natural Mortality define Hi production and Lo production operating models
Based on longevity and size data, and stock assessment data and fits 4
Uncertainties At the May Workshop we identified uncertainties: Herring recruitment Herring natural mortality Herring growth Herring assessment error/bias
5
Growth good and poor growth operating models
Based on survey weight at age data
6
Uncertainties At the May Workshop we identified uncertainties: Herring recruitment Production Herring natural mortality Hi Lo x Herring growth x Herring assessment error/bias x x
Growth Good Poor x x x x
7
Assessment Error and Bias unbiased and biased operating models
8
Assessment Error and Bias unbiased and biased operating models Biased assessment results in biomass higher than reality
Based on the stock assessment retrospective pattern
9
Uncertainties At the May Workshop we identified uncertainties: Production Growth Herring recruitment Hi Lo Good Poor x x Herring natural mortality x x Herring growth x x x x Herring assessment error/bias x
x x
x x x
x x
Assessment bias On Off x x x x x x x x
Uncertainties combined into 8 different operating models 10
Fishery Selectivity
Based on purse seine and MWT age composition data
11
Predator models Stolen or adapted from presentations By Dr. Sarah Gaichas, NEFSC
12
Predator models
Are
Are not
• Focused on evaluating stock‐ wide herring ABC harvest control rules applied annually • Developed balancing Council/ stakeholder specifications and time constraints of MSE • Based on information from the Northeast US shelf and most recent stock assessments
• Spatial, do not address local scale or seasonal dynamics • New or full stock assessments • Accounting for any impacts on predators other than changes due to herring control rules • Intended to predict actual predator population dynamics 13
Two components of predator modeling Predator population model
Herringpredator relationship
• Delay‐difference dynamics • Information required:
• What about herring…
• • • • •
Stock‐recruitment relationship Natural mortality rate Fishing mortality rate Initial population size Weight at age
• Assessments or observations
• Total abundance? Biomass? • Certain ages or sizes?
• Affects what about the predator • Predator growth • Predator reproduction • Predator survival
• And how? Base on observations 14
Predator population model summary Highly migratory
Seabird
Groundfish
Marine mammal
Stakeholder preferred species
Bluefin tuna
Common tern
Not specified
Not specified
Species modeled
Bluefin tuna (western Atlantic stock)
Common tern (Gulf of Maine colonies as defined by GOM Seabird Working Group)
Spiny dogfish (GOM and GB Atlantic cod stocks also examined)
None, data limited (Minke & humpback whales, harbor porpoise, harbor seal examined)
Stock‐recruitment (or adults, recruits)
Porch and Lauretta 2016, ICCAT 2015
Derived from GOMSWG data
Rago and Sosebee 2010
No time series data for our region
Natural mortality
ICCAT 2015
Nisbet 2002
Rago and Sosebee 2013, 2015
Derivable from Waring et al. 2015?
Fishing mortality
ICCAT 2015
n/a
Rago 2016
Waring et al. 2015?
Initial population
ICCAT 2015
GOMSWG data
Rago 2016
Waring et al. 2015?
Weight at age
Restrepo et al. 2010 Nisbet 2002
Rago et al. 1998
General literature 15
Predator‐prey relationships: Northeast US Herring vs. Antarctic krill
16
HerringPredator relationship issues/caveats • Predator populations are affected by MANY factors, prey is one • Northeast US predators have MANY prey options, herring is one • Time limitation enforced model simplicity for these complex relationships • Our approach is to use the best‐ supported relationship for each predator based on observations from the Northeast US ecosystem
• Isolating a clear herringpredator relationship from observations is difficult or impossible (e.g. cod) • Even with good observations, the modeled herringpredator relationship may require strong assumptions and not be statistically significant (e.g. terns) • Apparent positive herring predator relationships may not arise from the modeled mechanism (e.g. dogfish) 17
Predator relationships summary Predator and overlap
Modeled herring relationship
• Western Atlantic bluefin tuna
Herring population average weight affects bluefin tuna growth
Forage throughout North Atlantic, seasonally in GOM
• Common terns Forage seasonally near island breeding colonies in GOM
• Spiny dogfish Forage through same range as herring most of the year
• Marine mammals
Herring total biomass affects common tern reproductive success (productivity) Herring total abundance affects dogfish survival Food web model simulations 18
Tuna condition
Herring biomass
0.0230
Herring size, Gulf of Maine
“The decline in bluefin tuna condition, 0.0220 despite high prey biomass in the Gulf of 0.0215 Maine, suggests that managing for high 0.0210 abundance at middle trophic levels does 0.0205 not guarantee the success of all top 0.0200 predators. In fact, it suggests that for 0.0195 0.0190 some upper level predators, the quality 0.0185 of the prey may be more important than 0.0180 the overall abundance.” 0 0.05 0.1 0.15 0.2 0.25 0.3
Growth Intercept
0.0225
Modeled relationship: Tuna grow better with heavier herring overall, and/or with a higher proportion of large herring in the population
Herring Avg Wt
19
Tuna modeling notes • Available data do not support a positive relationship between herring and tuna populations: 1200000
60000
500000
Bluefin Tuna Recruitment
450000 400000 350000
Bluefin Tuna Spawning Stock Biomass
50000
600000
30000
20000
400000
20000
10000
200000
10000
30000
200000 100000 50000 0
0 0
500,000
1,000,000
1,500,000
50000 40000
300000
150000
1000000 800000
40000
250000
60000
BFTrec HerringSSB BFTSSB
0
500,000
1,000,000
Northeast US Herring Spawning Stock Biomass
1,500,000
0 1960
1970
1980
1990
2000
0 2010
Year
• Our models do not address herring/tuna interactions in a specific place or time. Tuna follow herring and likely aggregate around herring while feeding. • We can draw no conclusions from our modeling about predator/prey co‐occurrence at the local scale. • Similarly, without additional observations, we cannot extrapolate local scale co‐occurrence to population level relationships. 20
21
Herring Common tern
1.10 Predator Recruit Multiplier
1.08 1.06 1.04 1.02 1.00 0.98 0.96 0.94 0.92 0.90 0
500000
1000000
1500000
2000000
2500000
Herring Abundance
Modeled relationship: Common tern productivity is improved when herring total biomass above a threshold (400,000 t). Productivity diminishes below this threshold.
22
Top groundfish predators of herring (NEFSC, 1972‐2015)
23
Dogfish, Georges Bank cod and Gulf of Maine cod all ate herring in proportion to herring abundance, 1972‐2015. However, increased herring in diet was positively related to spawning stock biomass only for dogfish.
24
Herring Dogfish
0.100
Predator Annual Natural Mortality
0.090 0.080 0.070 0.060 0.050 0.040 0.030 0.020 0.010 0.000 0
1000000
2000000
3000000
4000000
5000000
6000000
Herring Abundance
• The dogfish relationship assumes herring abundance improves dogfish survival because no clear relationship was found with recruitment or growth. • Increased survival may not be the mechanism for the observed positive influence of herring in diet on the dogfish population. 25
Predator Modeling Summary • Our models are designed for evaluating alternative herring control rules, not predator stock assessment and population prediction. • We caution against generalizing results for these particular predators to other predators, as population parameters and herring relationships differ. • Although we selected predators with high herring diet proportions, observed predator population responses to herring alone do not dominate dynamics, and our herringpredator models reflect that. • Predator responses to aggregate prey dynamics are likely to be much clearer than responses to individual prey in the Northeast US ecosystem given its food web structure. 26
Herring Output Metrics – From Workshop #1 • Spawning Stock Biomass & SSB relative to SSBMSY & SSBunfished • Probability that • SSB =1 • Frequency that tuna weight > average
28
Output Metrics Economic • Yield: output of the herring model • Net revenue = (price*yield) – cost • Stability = the degree to which net revenue was “stable” or “streaky” (i.e., fairly steady over time vs. booms and busts) • Net revenue and stability demonstrate similar tradeoffs as herring yield and variation in yield, and so not presented in detail
29
Pause for Questions on Part I? Data and methods developed for MSE models (herring, predator and economic) 30
Fishing Mortality or Catch
Part II –Analysis of Potential Control Rules
? Biomass or Abundance
31
Six Control Rule Types presented at Workshop #2 1. 2. 3. 4. 5. 6.
Biomass based Biomass based with 3 year block Biomass based with 5 year block Biomass based with 3 year block and 15% restriction Constant catch Conditional constant catch with max F = 0.5Fmsy
32
Control Rules biomass based Three ‘parameters’ with many variants
Fishing Mortality
Upper biomass parameter Max F parameter
Biomass or Abundance
Lower biomass parameter 33
Control Rules biomass based Three ‘parameters’ with many variants
Fishing Mortality
Upper biomass parameter = Lower biomass parameter = 0
Max F parameter
Biomass or Abundance 34
Control Rules biomass based Three ‘parameters’ with many variants
Fishing Mortality
Upper biomass parameter = Lower biomass parameter > 0
Max F parameter
Biomass or Abundance 35
Control Rules biomass based Three ‘parameters’ with many variants Evaluated 16 different values for each biomass threshold ranging from 0 to 4x Bmsy Evaluated 10 different values for maximum F ranging from 0.1 to 1.0x Fmsy 1,360 combinations
36
Control Rules status quo – biomass based with 3 year block
37
Control rules Constant Catch
Catch
One parameter Evaluated 10 different values ranging from 0.1 to 1x MSY
Biomass or Abundance 38
Control Rules Conditional Constant Catch
Biomass or Abundance
0.5Fmsy
Catch
Fishing Mortality
Two parameters Evaluated 10 different values ranging from 0.1 to 1x MSY with max F of 0.5Fmsy
Biomass or Abundance 39
Control Rule Types and Shapes Biomass based Biomass based with 3 year block Biomass based with 5 year block Biomass based with 3 year block and 15% restriction Constant catch Conditional constant catch with max F = 0.5Fmsy
1,360 alternatives 1,360 alternatives 1,360 alternatives 1,360 alternatives 10 alternatives 10 alternatives 5,460 alternatives x 8 operating models 43,680 40
Control Rules For each operating model, each control rule alternative was simulated for 150 years and this was repeated for 100 simulations
41
Prelim Results – Herring Yield vs Stability
42
Frequency Dogfish > 0.5Bmsy vs herring SSB
43
Frequency tern prod > 1.0 vs herring SSB
44
Frequency tuna good condt’n vs herring SSB good herring growth
45
Post Workshop #2 • At workshop: less support for BB with 15% restriction and CC/CCC ‐ Too much yield lost for short‐term stability, poor performance elsewhere ‐ More likely to require short‐term deviations in application
• Post workshop: Herring AP and Committee tasked with: 1) identifying priority metrics and tradeoffs; and 2) identifying a reasonable number of CR alternatives. • Herring PDT prepared 4 example control rule shapes, and evaluated their performance for a handful of possible metrics. 46
Control Rules Biomass based 1,360 alternatives Biomass based with 3 year block 1,360 alternatives Biomass based with 5 year block Biomass based with 3 year block and 1,360 alternatives 1,360 alternatives 15% restriction Constant catch 10 alternatives Conditional constant catch with max F 10 alternatives = 0.5Fmsy 4,080 alternatives x 8 operating models 43,680 32,640 47
Let’s Compare 4 CRs
48
Let’s Compare 4 CRs
49
Let’s Compare 4 CRs
50
Let’s Compare 4 CRs
51
Let’s Compare 4 CRs 1. Examine tradeoffs and uncertainty in tradeoffs 2. Examine effect of assessment bias 3. Examine effect of annual ABC, 3 year blocks, and 5 year blocks
52
Note some CRs more robust than others
1.
Examine tradeoffs and uncertainty in tradeoffs
Let’s Compare 4 CRs all with 3 year block – unbiased assessment Yield/MSY vs. SSB/unfished
Yield/MSY vs. Frequency Overfished
53
Results more certain here
1.
Examine tradeoffs and uncertainty in tradeoffs
Let’s Compare 4 CRs
all with 3 year block – unbiased assessment Yield/MSY vs. Variation in yield
Frequency Overfished vs. Zero ABC (no fishery)
54
Note some CRs more robust than others
1.
Examine tradeoffs and uncertainty in tradeoffs
Let’s Compare 4 CRs all with 3 year block – unbiased assessment Yield/MSY vs. Frequency SSB Avg. vs. SSB/unfished
Freq. tern prod. >=1 vs. SSB/unfished
56
1.
Examine tradeoffs and uncertainty in tradeoffs
Let’s Compare 4 CRs all with 3 year block – unbiased assessment Frequency dogfish>0.5Bmsy vs. SSB/unfished
57
1.
Let’s Compare 4 CRs
Examine tradeoffs and uncertainty in tradeoffs
all with 3 year block – unbiased assessment
Herring Yield vs. SSB
58
Let’s Compare 4 CRs • What metrics/tradeoffs do you value most? • For example, if you highly value yield then you likely favor CRs with certainty in high amounts of yield, but do you get “acceptable” performance for other metrics?
Note some CRs more robust than others, esp. Freq. Overfished
2.
Examine effect of assessment bias
Let’s Compare 4 CRs
all with 3 year block – effect of assessment bias Yield/MSY vs. SSB/unfished
Yield/MSY vs. Frequency Overfished
60
Let’s Compare 4 CRs • Relying solely on biased results may duplicate other processes • We make adjustments for bias (e.g., retrospective adjustments) • We have peer review • We have an SSC
• Robustness to bias, which did vary among CRs, desirable
3.
Longer blocks cost yield and SSB
Examine effect of annual ABC, 3 year blocks, and 5 year blocks
Status Quo CR – comparing 1, 3, and 5 years Yield/MSY vs. SSB/unfished
Yield/MSY vs. Frequency Overfished
62
Let’s Compare 4 CRs • Is the short‐term stability of longer blocks worth the cost in: yield, long‐term variation in yield, frequency of overfished, decrease in frequency of desired tern production? • What is industry’s preferred planning horizon?
63
Other Considerations • Other tradeoffs of interest? • Identifying or refining CR alternatives can by achieved by: • Specifying preferred performance for various metrics • Moving CR parameters (Hi and low thresholds and Max‐F); “What if?”
64
Pause for Questions on Part II? Preliminary analyses of potential control rule alternatives 65
