Modeling biological-composition time series in integrated stock assessments: data weighting considerations and impact on estimates of stock status P. R.

Slides:



Advertisements
Similar presentations
Data Mining Methodology 1. Why have a Methodology  Don’t want to learn things that aren’t true May not represent any underlying reality ○ Spurious correlation.
Advertisements

Sheng-Ping Wang 1,2, Mark Maunder 2, and Alexandre Aires-Da-Silva 2 1.National Taiwan Ocean University 2.Inter-American Tropical Tuna Commission.
By, Deepak George Pazhayamadom Emer Rogan (Department of ZEPS, University College Cork) Ciaran Kelly (Fisheries Science Services, Marine Institute) Edward.
1 Ecological and Economic Considerations in Management of the U.S. Pacific sardine Fishery Samuel F. Herrick Jr NOAA Fisheries Southwest Fisheries Science.
An evaluation of alternative binning approaches for composition data in integrated stock assessments Cole Monnahan, Sean Anderson, Felipe Hurtado, Kotaro.
Growth in Age-Structured Stock Assessment Models R.I.C. Chris Francis CAPAM Growth Workshop, La Jolla, November 3-7, 2014.
C3: Estimation of size-transition matrices with and without molt probability for Alaska golden king crab using tag–recapture data M.S.M. Siddeek, J. Zheng,
Model time step and species biology considerations for growth estimation in integrated stock assessments P. R. Crone and J. L. Valero Southwest Fisheries.
Black Sea Bass – Northern Stock Coastal-Pelagic/ASMFC Working Group Review June 15, 2010.
The current status of fisheries stock assessment Mark Maunder Inter-American Tropical Tuna Commission (IATTC) Center for the Advancement of Population.
Impact of plot size on the effect of competition in individual-tree models and their applications Jari Hynynen & Risto Ojansuu Finnish Forest Research.
4-1 Statistical Inference The field of statistical inference consists of those methods used to make decisions or draw conclusions about a population.
Lecture 10 Comparison and Evaluation of Alternative System Designs.
Hui-Hua Lee 1, Kevin R. Piner 1, Mark N. Maunder 2 Evaluation of traditional versus conditional fitting of von Bertalanffy growth functions 1 NOAA Fisheries,
Age-structured assessment of three Aleutian fish stocks with predator-prey interactions Doug Kinzey School of Aquatic and Fishery Sciences University of.
Quantitative Genetics
How many conditional age-at-length data are needed to estimate growth in stock assessment models? Xi He, John C. Field, Donald E. Pearson, and Lyndsey.
LEARNING PROGRAMME Hypothesis testing Intermediate Training in Quantitative Analysis Bangkok November 2007.
Hypothesis Testing in Linear Regression Analysis
Chapter 7 Estimation: Single Population
Developing a statistical-multispecies framework for a predator-prey system in the eastern Bering Sea: Jesús Jurado-Molina University of Washington Jim.
Terms of Reference #5 and #7 Overview/Relation to Management Deirdre Boelke, NEFMC Staff, Scallop PDT Chair March 17-19, 2015.
EVAL 6970: Cost Analysis for Evaluation Dr. Chris L. S. Coryn Nick Saxton Fall 2014.
WP4: Models to predict & test recovery strategies Cefas: Laurence Kell & John Pinnegar Univ. Aberdeen: Tara Marshall & Bruce McAdam.
MANAGEMENT AND ANALYSIS OF WILDLIFE BIOLOGY DATA Bret A. Collier 1 and T. Wayne Schwertner 2 1 Institute of Renewable Natural Resources, Texas A&M University,
Review of Chapters 1- 5 We review some important themes from the first 5 chapters 1.Introduction Statistics- Set of methods for collecting/analyzing data.
Pacific Hake Management Strategy Evaluation Joint Technical Committee Northwest Fisheries Science Center, NOAA Pacific Biological Station, DFO School of.
CJT 765: Structural Equation Modeling Class 7: fitting a model, fit indices, comparingmodels, statistical power.
Slide 1 Estimating Performance Below the National Level Applying Simulation Methods to TIMSS Fourth Annual IES Research Conference Dan Sherman, Ph.D. American.
Pacific Hake Management Strategy Evaluation Joint Technical Committee Northwest Fisheries Science Center, NOAA Pacific Biological Station, DFO School of.
1 Chapter 12 Simple Linear Regression. 2 Chapter Outline  Simple Linear Regression Model  Least Squares Method  Coefficient of Determination  Model.
Evaluation of a practical method to estimate the variance parameter of random effects for time varying selectivity Hui-Hua Lee, Mark Maunder, Alexandre.
Brian Macpherson Ph.D, Professor of Statistics, University of Manitoba Tom Bingham Statistician, The Boeing Company.
Nick Smith, Kim Iles and Kurt Raynor Partly funded by BC Forest Science Program and Western Forest Products Sector sampling – some statistical properties.
ALADYM (Age-Length Based Dynamic Model): a stochastic simulation tool to predict population dynamics and management scenarios using fishery-independent.
The Stock Synthesis Approach Based on many of the ideas proposed in Fournier and Archibald (1982), Methot developed a stock assessment approach and computer.
P. R. Crone, M. N. Maunder, B. X. Semmens, and J. L. Valero Center for the Advancement of Population Assessment Methodology (CAPAM) Southwest Fisheries.
Limits to Statistical Theory Bootstrap analysis ESM April 2006.
Workshop on Stock Assessment Methods 7-11 November IATTC, La Jolla, CA, USA.
Simulated data sets Extracted from:. The data sets shared a common time period of 30 years and age range from 0 to 16 years. The data were provided to.
M.S.M. Siddeeka*, J. Zhenga, A.E. Puntb, and D. Pengillya
Eurostat Accuracy of Results of Statistical Matching Training Course «Statistical Matching» Rome, 6-8 November 2013 Marcello D’Orazio Dept. National Accounts.
What is the likelihood that your model is wrong? Generalized tests and corrections for overdispersion during model fitting and exploration James Thorson,
The effect of variable sampling efficiency on reliability of the observation error as a measure of uncertainty in abundance indices from scientific surveys.
Extending length-based models for data-limited fisheries into a state-space framework Merrill B. Rudd* and James T. Thorson *PhD Student, School of Aquatic.
SEDAR 42: US Gulf of Mexico Red grouper assessment Review Workshop Introduction SEFSC July , 2015.
PROGRESS IN THE SIMULATION TESTING OF PENGUIN CLOSURE EFFECT RESPONSE An Introduction.
Estimation of growth within stock assessment models: implications when using length composition data Jiangfeng Zhu a, Mark N. Maunder b, Alexandre M. Aires-da-Silva.
Using distributions of likelihoods to diagnose parameter misspecification of integrated stock assessment models Jiangfeng Zhu * Shanghai Ocean University,
Estimation of selectivity in Stock Synthesis: lessons learned from the tuna stock assessment Shigehide Iwata* 1 Toshihde Kitakado* 2 Yukio Takeuchi* 1.
CAN DIAGNOSTIC TESTS HELP IDENTIFY WHAT MODEL STRUCTURE IS MISSPECIFIED? Felipe Carvalho 1, Mark N. Maunder 2,3, Yi-Jay Chang 1, Kevin R. Piner 4, Andre.
Some Insights into Data Weighting in Integrated Stock Assessments André E. Punt 21 October 2015 Index-1 length-4.
MSE Performance Metrics, Tentative Results and Summary Joint Technical Committee Northwest Fisheries Science Center, NOAA Pacific Biological Station, DFO.
Lecture 10 review Spatial sampling design –Systematic sampling is generally better than random sampling if the sampling universe has large-scale structure.
A bit of history Fry 1940s: ”virtual population”, “catch curve”
Accuracy, Reliability, and Validity of Freesurfer Measurements David H. Salat
Data weighting and data conflicts in fishery stock assessments Chris Francis Wellington, New Zealand CAPAM workshop, “ Data conflict and weighting, likelihood.
NWFSC A short course on data weighting and process error in Stock Synthesis Allan Hicks CAPAM workshop October 19, 2015.
Recommended modeling approach Version 2.0. The law of conflicting data Axiom Data is true Implication Conflicting data implies model misspecification.
Sampling and Sampling Distributions. Sampling Distribution Basics Sample statistics (the mean and standard deviation are examples) vary from sample to.
Is down weighting composition data adequate to deal with model misspecification or do we need to fix the model? Sheng-Ping Wang, Mark N. Maunder National.
STA248 week 121 Bootstrap Test for Pairs of Means of a Non-Normal Population – small samples Suppose X 1, …, X n are iid from some distribution independent.
PRINCIPLES OF STOCK ASSESSMENT. Aims of stock assessment The overall aim of fisheries science is to provide information to managers on the state and life.
Nick Smith, Kim Iles and Kurt Raynor
CJT 765: Structural Equation Modeling
How to handle missing data values
Basic Training for Statistical Process Control
Basic Training for Statistical Process Control
Gerald Dyer, Jr., MPH October 20, 2016
MULTIFAN-CL implementation of deterministic and stochastic projections
Presentation transcript:

Modeling biological-composition time series in integrated stock assessments: data weighting considerations and impact on estimates of stock status P. R. Crone Southwest Fisheries Science Center (NOAA) Center for the Advancement of Population Assessment Methodology (CAPAM) 8901 La Jolla Shores Dr., La Jolla, CA 92037, USA Fishery 1 Fishery 2

Study description Results Conclusions Further work Presentation outline

Motivation and expectations o Better understanding of impact that data weighting considerations in typical assessments have on baseline management statistics … contribute to good practices for stock assessment development o Meta-analysis is based on a limited pool of assessments … is able to provide quantitative results for particular statistical comparisons, is not a substitute for simulation-based tests Study description

Assessment archive o Pool of recently conducted fish stock (species) assessments used for management o Assessments for small pelagic (3), large pelagic (7), and groundfish (19) species o Assessments based on the Stock Synthesis model o Majority of assessments conducted in 2015, some Biological-composition time series o Length (‘marginal’, e.g., no./pct. by length bin and time step) o Age (marginal) o Conditional age-at-length (‘random at length’, age-length key format) o Size (marginal, e.g., weight, biological compositions based on different bin structure) o Weight (unfitted empirical weight-at-age data) o Various ways of using/combining biological-composition time series in assessments General Study description

General (continued) Study description Data weighting of biological compositions ‘outside’ the model o Initial (input) sample sizes for biological compositions are assessment/analyst-specific o Sometimes based on actual number of fish (e.g., sport fishery compositions, CAAL) o More often based on number of boat trips, hauls, sets, wells, sample adjustment formula, etc. o Can be based generally on variance estimates determined from sample/survey programs o Can be based generally on variance estimates from simulation analysis (e.g., bootstrap methods) o Often caps (thresholds) are used for input sample sizes (e.g., ) o Input sample size determination was not addressed in this evaluation

General (continued) Study description Data weighting of biological compositions ‘inside’ the model o Variability of biological-composition time series is based initially on input sample size … subsequently, adjusted internally based on comparing observed and expected values from fits to the time series o Various data weighting approaches for composition time series in integrated assessment models … McCallister and Ianelli (1997) and Francis (2011) methods often considered in practice o ‘Effective’ sample size in Stock Synthesis model (McCallister and Ianelli methods) reflects number of random samples (drawn from multinomial distribution) needed to produce fit as precise as model’s predicted fit o Actual weighting values (scalars) for composition data reflect various mean estimates calculated from ratios of effective to input sample sizes (multiplicative based) o Francis method basis is variation of mean length/age of the composition time series, accounts for correlation among length or age groups, results in greater variation surrounding composition time series o In practice, ad hoc caps (thresholds) are implemented for estimated scalars >1 o Internally implemented data weighting methods for composition time series were addressed in this evaluation

Baseline (Final) o Assessment model for advising management Unweighted (UW) o Final model that includes no (internally) weighted composition time series o All scalars (‘weighting values, variance adjustments, lambdas’) = 1 McCallister-Ianelli (AM) o Scalar estimate reflects arithmetic mean from model fits to composition time series (based on ratios of effective sample size to input sample sizes) McCallister-Ianelli (HM) o Scalar estimate reflects harmonic mean from model fits to composition time series (based on ratios of effective sample size to input sample sizes) Francis (F0) o Assessments that included only length- and/or age-composition time series and no CAAL time series (based on FA) Francis-Method A (FA) o Assessments that included CAAL time series along with length and/or age-composition time series (mean estimates indexed by year) Francis-Method B (FB) o Assessments that included CAAL time series along with length and/or age-composition time series (mean estimates indexed by year/length bin) Study description Assessment models Data weighting methods

Model development/estimation o For each species, final assessment model re-configured according to recommended scalars from respective data weighting method (cap=100 and single iteration) o For a species, from 3-5 alternative models were developed for overall study, depending on the biological compositions, SS version, convergence issues o Data weighting addressed only biological compositions included in the model, i.e., no weighting applicable to other input data (e.g., index of abundance time series) or parameter assumptions (e.g., σ R of stock-recruit relationship) o Data weighting methods described in McCallister and Ianelli (1997), Francis (2011), Methot and Wetzel (2013), Punt (in press) Output o Management quantities of interest: MSY, F MSY, B current, Depletion (SSB current / SSB 0 ) o Comparisons based on means/CVs and medians/REs Study description General (continued)

Data weighting methods – Example (SS effective sample size)

Mean length (cm) Year 5 Data weighting methods – Example (FA/FB/F0 diagnostic plot)

Species Baseline Final Assessment model (Data weighting method) Assessment model (Final) P. sardine Study description Analysis flow chart McCallister/Ianelli (Harmonic mean) HM McCallister/Ianelli (Arithmetic mean) AM Unweighted (All scalars=1) UW Francis (No CAAL) F0 Francis (CAAL, Method A) FA Francis (CAAL, Method B) FB Output (Management quantities) MSY, F MSY, B current, DEP. N = 29 species

Results

Data Weighting Methods Scalar Ranges by Biological Data Type

Assessment (species) examples Mean and CV

Data Weighting Methods ‘Within Assessment’ Variability MSY

Data Weighting Methods ‘Within Assessment’ Variability F MSY

Data Weighting Methods ‘Within Assessment’ Variability B current

Data Weighting Methods ‘Within Assessment’ Variability Depletion

Data Weighting Methods ‘Between Management Quantity’ Variability MSYB current F MSY Depletion

Data weighting method MSY Species (no. of assessments) Models (no. of replicates) Sample size limit implemented (no. of species) Convergence issues (no. of species) Unplotted models (pct. extreme positive outliers)4%2% 4%1% ‘Within Assessment’ Variability (Relative to Data Weighting Method) Relative error

F MSY Species (no. of assessments) Models (no. of replicates) Sample size limit implemented (no. of species) Convergence issues (no. of species) Unplotted models (pct. extreme positive outliers)1%0%2%7%0% ‘Within Assessment’ Variability (Relative to Data Weighting Method) Relative error Data weighting method

B current Species (no. of assessments) Models (no. of replicates) Sample size limit implemented (no. of species) Convergence issues (no. of species) Unplotted models (pct. extreme positive outliers)6%11%4%12%9%5% ‘Within Assessment’ Variability (Relative to Data Weighting Method) Relative error Data weighting method

Depletion Species (no. of assessments) Models (no. of replicates) Sample size limit implemented (no. of species) Convergence issues (no. of species) Unplotted models (pct. extreme positive outliers)5%10%2%12%1%3% ‘Within Assessment’ Variability (Relative to Data Weighting Method) Relative error Data weighting method

Data Weighting Methods Relative to HM (‘correctly specified’ model) Relative error Data weighting method MSY Species (no. of assessments) Models (no. of replicates) Sample size limit implemented (no. of species)02011 Convergence issues (no. of species)02000 Unplotted models (pct. extreme positive outliers)0% 13%

Data Weighting Methods Relative to HM (‘correctly specified’ model) Relative error Data weighting method Species (no. of assessments) Models (no. of replicates) Sample size limit implemented (no. of species)02011 Convergence issues (no. of species)02000 Unplotted models (pct. extreme positive outliers)3%0% F MSY

Data Weighting Methods Relative to HM (‘correctly specified’ model) Relative error Data weighting method Species (no. of assessments) Models (no. of replicates) Sample size limit implemented (no. of species)02011 Convergence issues (no. of species)02000 Unplotted models (pct. extreme positive outliers)3%0% 13% B current

Data Weighting Methods Relative to HM (‘correctly specified’ model) Relative error Species (no. of assessments) Models (no. of replicates) Sample size limit implemented (no. of species)02011 Convergence issues (no. of species)02000 Unplotted models (pct. extreme positive outliers)0% 7% Depletion Data weighting method

Conclusions Data weighting methods impact on management quantities o Terminal biomass estimates most uncertain in most cases (mean CV=35%), depletion and MSY less so (20%), and F MSY most precise (<10%) o Positively-skewed, median-unbiased relative error distributions o The harmonic mean-based McCallister-Ianelli method (HM) resulted in precise and unbiased estimates in most cases, but … o Unweighted method (UW) also relatively precise and robust in many comparisons o Frances methods (F0, FA, FB) produced generally unbiased estimates, but typically less precise than HM; more similar for MSY-related quantities o FA less bias (equally precise) than FB in many comparisons o For correctly-specified assessment based on HM, better off not weighting (UW) than implementing an alternative data weighting method

Study benefits and further work Replicates (assessments) in meta-analysis are realistic o Replicates associated with typical simulations are unrealistic, i.e., much too similar to one another … increase number/variety of assessments o However, study (experimental) population based on real assessments provides limited cause-and-effect information, given the many data/parameter inconsistencies across replicates Meta-analysis provides baseline information for more focused simulation studies o Contrast between quality of derived management metrics o Fold into MSEs addressing small pelagic species’ fisheries on the USA Pacific coast for basing (much needed) new and improved harvest control rules Information useful for analysts charged with developing ongoing assessments for management purposes o Data weighting approaches in actual assessments are evolving presently, research needed to inform good practices

References Crone, P.R., D.B. Sampson Evaluation of assumed error structure in stock assessment models that use sample estimates of age composition. Pages in Fishery Stock Assessment Models. Alaska Sea Grant College Program Report No. AK-SG-98-01, University of Alaska, Fairbanks, Alaska. Fournier, D., C.P. Archibald A general theory for analyzing catch at age data. Can. J. Fish. Aquat. Sci. 39: Francis, R.I.C.C Data weighting in statistical fisheries stock assessment models. Can. J. Fish. Aquat. Sci. 68: McAllister, M.K., J.N. Ianelli Bayesian stock assessment using catch-age data and the sampling- importance resampling algorithm. Can. J. Fish. Aquat. Sci. 54(2): 284–300. Methot, R.D., C.R. Wetzel Stock Synthesis: a biological and statistical frame-work for fish stock assessment and fishery management. Fish. Res. 142:86–99. Pennington, M., L.-M. Burmeister, V. Hjellvik Assessing the precision of frequency distributions estimated from trawl survey samples. Fish Bull. 100:74–80. Punt, A.E. in press. Some insights into data weighting in integrated stock assessments. Fish. Res. Stewart, I.J., O.S. Hamel Boostrapping of sample sizes for length- or age-composition data used in stock assessments. Can. J. Fish. Aquat. Sci. 671: