Stock Assessment & Setting of Annual Catch Limits in New England. Steve Cadrin, Jake Kritzer SSC, NEFMC Steve Correia and Tom Nies PDT, NEFMC

Transcription

1 Stock Assessment & Setting of Annual Catch Limits in New England Bob O Boyle, O Steve Cadrin, Jake Kritzer SSC, NEFMC Steve Correia and Tom Nies PDT, NEFMC

2 Stock Assessment & Peer Review Stock Assessment Review Committee (SAW/SARC) Transboundary Resources Assessment Committee (TRAC) Groundfish Assessment Review Meeting (GARM) Process: NE Regional Coordination Committee (all NE clients: clients: NEFMC, MAFMC, ASMFC, NERO, NEFSC) determines schedule of assessments & terms of reference NEFSC/Canada DFO/State Agency/University teams produce assessments to address TORs External review (CIE) with SSC chair Plan Development Team post processing (e.g., projections for management alternatives that meet objectives of F<Fmsy, rebuilding) SSC review of status determination & catch advice

3 What is GARM III? Groundfish Assessment Review Meetings November 2007 August groundfish stocks managed under US Northeast Multispecies Fishery Management Plan GARM I - October 2002 Benchmarks GARM II - August 2005 GARM III most comprehensive to date

4 GARM III Stocks Cod GB & GOM Haddock GB & GOM Yellowtail GB, GOM, SNE Witch Flounder American Plaice Winter Flounder Windowpane GB & SNE Pollock White Hake Redfish Ocean Pout Atlantic Halibut GB, GOM, SNE

5 Four Meetings Data Inputs - 29 Oct 2 Nov 2007 Assessment Models Feb 2008 Biological Reference Points - 28 April 2 May 2008 Assessments August 2008

6 Data Inputs Terms of Reference Commercial & survey data Statistical methods Tagging programs for cod & yellowtail flounder Industry-Based Surveys Estimates of precision & bias of data & derived quantities Measures of uncertainty & implications for use in assessment models

7 Models Terms of Reference Index methods, Production Models & Age- or Length- based Models SCAA vs VPA with respect to Retrospective patterns Alternative parameterizations Incorporation of external information e.g. tagging & environmental data Incorporation of prior, external information Potential factors responsible for retrospective patterns Assessment model to determine stock status & productivity until next benchmark

8 BRP Terms of Reference Influence of retrospective patterns (F, SSB, recruitment) on BRPs & forecasting Trends in life history parameters & assess importance for BRPs & rebuilding scenarios Aggregate ecosystem yield of groundfish fishery & compare to aggregate single stock yield projections BRPs for use in stock status determination Models for forecasting & for evaluating rebuilding scenarios

9 Assessors & Reviewers NEFSC Woods Hole (approx 25) Data inputs Crecco, Van-Eeckhaute, Kahn, Needle, Rothschild, Smith, Helge-Vølstad lstad Models De Oliveira, Gavaris, Ianelli, Jiao, Jones, Medley Biological Reference Points Bell, Gavaris, Haist, Reeves, Thompson Assessments Cieri, Mohn, Rosenberg, Sinclair Industry Experts Butterworth, Rademeyer Managers & industry

10 GARM III Stock Status

11 Data Inputs Reliability of total Catch New trip/landings hierarchical algorithm VMS verification Sampling adequacy Estimation of uncertainty in CAA Design (bootstrap) vs model based provided similar results Importance of sampling rates at port level & distribution of sampling effort amongst trips Discard estimation (Observer data) Ratio of Sums method (sum discard wt / sum kept wt) preferred over Avg (Discard wt / Kept wt) per trip More reliable data on total landings rather than total trips

12 Models Range of Models Relative Trends Trends in relative abundance & fishing mortality Production By+1= +1=By +Yrep -Cy Yrep constant ( (Yrep =a), linear ( (Yrep =a +bby) ) or quadratic ( (YrepYrep =rby [1-By /K] Age based VPA & Statistical Catch at Age (SCAA) Length-and and-age-based Age data limited Tendency for model & software to become confounded Focus on model first Then choose software

13 Models Species Stock GARM II GARM III Cod GB VPA VPA Cod GOM VPA VPA Haddock GB VPA VPA Haddock GOM AIM VPA Yellowtail Flounder GB VPA VPA Yellowtail Flounder SNE/MA VPA VPA Yellowtail Flounder CC/GOM VPA VPA American Plaice GB/GOM VPA VPA Witch Flounder VPA VPA Winter Flounder GB ASPIC VPA Winter Flounder GOM VPA VPA & SCALE Winter Flounder SNE/MA VPA VPA Redfish RED SCAA White Hake GB/GOM ASPIC & AIM SCAA Pollock GB/GOM AIM AIM Windowpane Flounder GOM/GB AIM AIM Windowpane Flounder SNE/MA AIM AIM Ocean Pout Index Method Index Method Altantic Halibut None Replacement Yield

14 Retrospective Pattern Significant part of GARM review Primary potential causes Unrecorded catch Change in natural mortality Change in abundance index catchability (q) Change in fishery selectivity Not possible to identify cause through model diagnostics Adjusting model assumptions (e.g. altering survey q, catches or M) to remove pattern does not guarantee problem resolved Model may continue to be mis-specified specified Need testable hypotheses on cause(s)

15 Georges Bank Cod SSB Retrospective

16 GARM Practice on Retrospective if acceptable adjustment cannot be made Default Terminal estimates in error whilst historical estimates correct Check age-specific retrospective patterns to determine age where magnitude consequential Adjust population numbers for terminal year of VPA (initial year of projection) to account for retrospective pattern Conduct projections using adjusted population numbers

17 Adjustment for Retrospective Georges Bank Atlantic Cod Base rho adj Base Split Final /2 SSB MSY SSB MSY 1.75 Fishing Mortality2007/FMSY F MSY Spawning Stock Biomass 2007 /SSB MSY Rho adjusted Base & Split Survey generally produced similar current status

18 Partial Recruitment PR on older age groups recurring issue Dome PRs may resolve retrospective but generated cryptic biomass Model biomass that has not been observed in either fishery or surveys Burden of proof (default) Demonstrate that fish exist when not observed in fishery & surveys, even if model fit with dome-shaped PR is better Need external data to resolve Gulf of Maine Cod & White Hake Patterson (ICES 2002): In highly parameterized models fishing mortality tended towards zero population numbers tended towards infinity

19 Gulf of Maine Cod Butterworth & Rademeyer used SCAA & dome PR Requires supporting evidence Extend age range in Catch at Age April 2008: 7 + group August 2008: 11+ group Tagging Small Dome fit Raised issues in VPA computation If F declines due to dome PR, expect fraction cod recovered will decline as age of tagging increases 100,000 tagged cod released during (6000 recoveries to date)

20 Gulf of Maine Cod Tagging Predicted Recovery at age assuming Dome Predicted Recovery at age assuming Flat Top More Flat than Domed PR Observed Recovery

21 Survey Qs based on Swept Area Allowed Examination of survey catchability patterns Stock by stock Season by season Species by species Patterns by age versus expectation Survey catchability expected to asymptote or decline at some age Patterns similar across stocks?

22 Biological Reference Points Limit BRPs F lim = Fmsy or proxy B lim = ½ Bmsy or proxy Relies on Stock Recruitment Relationship Default If recruitment & SSB derived from assessments informative, compute FMSY & BMSY using parametric projections If S / R not informative, use F 40%MSP as proxy & B MSY proxy computed using non-parametric projections Non-parametric projections Conduct hindcast model Conduct projections using bootstrapped residuals Sample from empirical S/R using Breakpoint if necessary

23 Hierarchy of Limit RPs No Target RPs as yet

24 Estimation of BRPs Species Stock S_R Model Bmsy or proxy Fmsy or proxy Cod GB Non-parametric SSB/R (40%MSP) F40%MSP Cod GOM Non-parametric SSB/R (40%MSP) F40%MSP Haddock GB Non-parametric SSB/R (40%MSP) F40%MSP Haddock GOM Non-parametric SSB/R (40%MSP F40%MSP Yellowtail Flounder GB Non-parametric SSB/R (40%MSP) F40%MSP Yellowtail Flounder SNE/MA Non-parametric SSB/R (40%MSP) F40%MSP Yellowtail Flounder CC/GOM Non-parametric SSB/R (40%MSP) F40%MSP American Plaice GB/GOM Non-parametric SSB/R (40%MSP) F40%MSP Witch Flounder Non-parametric SSB/R (40%MSP) F40%MSP Winter Flounder GB Non-parametric SSB/R(40%MSP) F40%MSP Winter Flounder GOM Non-parametric SSB/R (40%MSP) F40%MSP Winter Flounder SNE/MA Non-parametric SSB/R (40%MSP) F40%MSP Redfish Non-parametric SSB/R (50%MSP) F50%MSP White Hake GB/GOM Non-parametric SSB/R (40%MSP) F40%MSP Pollock GB/GOM Visual interpretation External Rel F at replacement Windowpane Flounder GOM/GB Visual interpretation External Rel F at replacement Windowpane Flounder SNE/MA Visual interpretation External Rel F at replacement Ocean Pout Visual interpretation External Rel F at replacement Altantic Halibut Implied Internal F0.1 Most stocks used non-parametric approach to BRPs

25 Changes relative to BRPs Number of biomass RPs declined Reduced Productivity

26 Average Weight (kg) at Age Georges Bank Haddock Loess smooth (Weight)

27 Single species vs Aggregate MSY Surplus Production Model fit to Aggregate Demersal Catch & NMFS Spring & Fall Surveys MSY kt versus 145 kt from GARM Bmsy kt vs 950 ky from GARM System wide Fmsy = 0.15

28 Ecosystem Considerations GARM = 4.3 of 14.6 t /km 2 Target biomass for all NE demersal fish species Just below avg productivity of many ecosystems

29 Ecosystem Conclusions 2nd Tier quota could be considered during recovery & for long-term maintenance System recovery predicated on low fishing mortality (FMSY = 0.15) Several stocks constrain recovery of complex (i.e. halibut, GB cod, GB Yt,, white hake, SNE Yt) Unless stocks can be targeted independently, much lower fishing effort required Could estimate M if model nodes stocks & not guilds

30 Where is NEFMC in ACL Implementation? July Groundfish Plan Development Team (PDT) developed approach SSC consideration of approach August - October GARM assessments of 19 Groundfish stocks Comments on draft National Standards 1 guidelines November Further development to consider guidance from this workshop Approach not to be detailed in FMP Amendment 16 Changes through experience without formal management action

31 NS 1 Guidelines Proposal Setting ACLs Process SSC Council OFL Science- Management feedback loop Scientific Uncertainty ABC ACL ACL ABC Rationale for ABC ACL Science- Management feedback loop Management Uncertainty ACT Difference?

32 NE Multispecies FMP Definitions Acronym OFL ABC Definition Catch at F MSY Catch at F control rule or F rebuild Considerations Point estimates of F MSY, stock size (3 years) OFL ABC Incorporates Biological uncertainty ACL ACT optional for Councils who wish to use them <=ABC ABC ACL Incorporates Management Uncertainty Linked to Accountability Measures

33 NEFMC Groundfish ACL Process ACLs element of existing adjustment process SAFE every year & management revisions every two years ACL recommendations for 3 years NMFS Science August peer review of assessments PDT Drafts recommended ABCs for SSC Interacts with SSC on ABC recommendations prior to September ACL Recommendations to Council by September SSC ABCs to Council by September Council Decision on PDT recommendations prior to December ACLs to NMFS prior to mid December

34 NEFMC Stock Assessment / Advisory Process NMFS US / Canada Under Development ACL NMFS Science PDT ACL Council ACL TMGC TRAC ABC SSC ABC Comment on Biological Uncertainty ACL: check for consistency with ABC

35 Role of NEFMC SSC Interface between science & Council add value to process Interpret peer reviewed science for Council process Define peer review as needed Avoid unnecessary duplication of accepted peer review Complement tactical decisions from PDTs with strategic thinking Strategically placed to undertake MSEs

36 Implementation Schedule Groundfish ACLs rarely based on data more recent than two years old Some ACLs based on data five years old SSC Consider Control Rules informed by data available on more frequent basis (e.g. trawl surveys) Process used on South African Hake Validated by MSE

37 ACL Setting Framework Informed by Rosenberg et al (2007) Framework applicable to data poor & data rich stocks Determine vulnerability (risk to impact) of stock Productivity / Susceptibility Analysis (PSA) Relate risk to uncertainty to determine buffer sizes Biological buffer: OFL > ABC Management buffer: ABC >= ACL Design ABC & ACL Control Rules Default buffer size for given level of risk

38 Risk = impact x P (impact) Susceptibility Productivity HIGH RISK Impact LOW RISK P (Impact)

39 Attributes of Vulnerability Productivity Maximum age Age at maturity Size at maturity Annual fecundity Maximum size Reproductive strategy Trophic level If data missing, assume high risk Susceptibility Availability Overlap with fishery Global distribution Encounterability Water column position Depth range overlap Adult Habitat Selectivity Size at Maturity Total records (+/-) ) (TEP, DI, TA/BP) Post-capture mortality Fate on discarding

40 Productivity Scoring Illustrative from Rosenberg Average Rank = 1.43

41 Susceptibility Scoring Illustrative from Rosenberg Average Rank = 1.75

42 Vulnerability Score Illustrative from Rosenberg

43 Determination of ACL As risk to overfishing increases, buffer should increase Higher risk = larger buffer

44 Determination of ACL Aspects of vulnerability already incorporated into BRPs Risk to resource of not conforming to BRPs varies with vulnerability Risk function of biological & management uncertainty

45 Sources of Uncertainty Biological Model Process Observation Assessment Related Management Implementation Management Related

46 Buffer Estimation Rosenberg et al (2007) envisioned two tables biological & management uncertainty RISK High Med Low Simple catch only largest >[right] and < [above] >[right] and < [above] e.g. Biological Uncertainty Production 1 Index <[left] and >[below] buffer=moderate >[right] and < [above] Full Assessment <[left] and >[below] <[left] and >[below] buffer=smallest Initially choose default buffer size by cell Adjustment of buffer size in each cell informed by analyses of data rich stocks

47 Setting ABC: Biological Uncertainty (OFL ABC) Risk = f (Productivity but not susceptibility); being revisited ABC Control Rule = f (Biological Uncertainty) ACL Control Rule = f (Management Uncertainty)

48 Productivity (draft) Factor SSB/SSB MSY High Risk Low Risk MSY Less than Greater than 1 Recent recruitment Below median Median average Above average Weights At age Declining Stable Increasing Population age structure Narrow truncated Mixed Dispersed expanded Geographic distribution??? Equivalent to PSA chart

49 Biological Uncertainty (draft) High Risk Low Risk Factor Historical Assessment Performance Poor Estimates of F or B highly variable; predicted catch does not match realized F Fair Good Consistent estimates over time; predicted catch matches realized Fs Retrospective pattern Strongly underestimates terminal year B; overestimates terminal year F; Persistent Pattern variable or minor Minimal or no pattern % projected catch based on assumed recruitment Over 25% 10% - 25% Less than 10%

50 ABC Control Rule 10% F control rule 25% F control rule 50% F control rule Low Risk or or or 10% Freb 25% Freb Median catch at Freb Productivity Factor (low to High) 10% Freb ABC set at x% ile of stock projections (AGEPRO) 25% F control rule or 25% Freb 1% F control rule 10% F control rule 10% F control rule or or or High Risk 1% Freb 10% Freb Biological Uncertainty Factor (More to Less) 10% Freb

51 Setting ACL: Management Uncertainty (ABC ACL) Similar approach for setting ABC Productivity Factor table Management uncertainty table ACL Control Rule still needs to be developed

52 Element Enforceability Low (difficult to enforce due to limited /ambiguous compliance criteria & little or no enforcement capacity) Moderate High (easily enforced with clear compliance criteria) Accountability Low (limited ability to revise effort controls or affect within or between fishing years) Moderate High (fishing effort/catch regulated through in- season/post-season measures) Latent Effort High (latent effort exists) Moderate Low (latent effort eliminated or regulated) Other Fishery Catch Low (catch by other fisheries, including state, exempted & recreational fisheries not restricted) Moderate High (actively manage catch by other fisheries, including state, exempted & recreational fisheries) Effectiveness Low (measures have not consistently achieved mortality targets or insufficient information exists to determine effectiveness - e.g., gear performance, reopened closure areas, etc.) Moderate High (measures have consistently achieved mortality targets within acceptable level of accuracy) Precision Low (management measures both exceed & under-achieve mortality/catch targets) Moderate High (measures consistently achieve, exceed, or under- achieve mortality/catch targets)

53 ACL Sub-Components Xs to be determined Important to have ACLs for all subcomponents due to link to AMs

54 Further Developments Relationship between buffer size & uncertainty (biological & management) requires exploration Apply approach to past data to explore relationship & performance using different rules Data rich situations can inform buffer sizes in data poor situations Management Strategy Evaluations (MSE) to explore buffer sizes

55 Management Strategy Evaluation Evaluate impacts of uncertainty on achievement of objectives (biological, economical, social) Assessment Benchmarks & annual fishery & survey indices Strategies Rebuilding schedules & implications for stock & fishery Implementation Seasons, closed areas, gear regulations

56 Management Strategy Evaluation Managers & industry set goals & propose strategies Scientists assist in modeling Economists assist in evaluation of costs & benefits of options (e.g. GOM cod) cost of 50% to 25% prob of exceeding overfishing 27% reduction in landings Negligible difference in achieving rebuilding by 2014 Valuation of Risk Analysis

57 Concerns with NS 1 Guidelines ABC & ACL should be sufficient ACT not stipulated by MSA Keep optional rather than essential Overfishing definition Redefining in terms of catch inconsistent with MSA Keep as fishing mortality AMs & in-season adjustment Adjustment of ACLs & implementation of AMs should be based on fishery performance data In-season adjustment unrealistic

58

59 Summary NEFMC SSC, PDT & Council working on ACL setting framework Groundfish initial focus of work to date PSA framework given broad applicability Need to define OFL ABC & ABC ACL buffer sizes based upon further work

60 THANK YOU!

61 Issues PSA part II of three step ERA process Triage to determine where in depth quantitative analysis required Rosenberg et al (2007) recognized need to learn from data rich situations Guide to ACL setting: not intended to replace assessments Need to ensure that risk is not double counted Assessments present current status based upon adjustment for retrospective patterns Biological reference points incorporate information on productivity & susceptibility How best to incorporate risk into control rules