A comparative study of observation-error estimators and state-space production models in fisheries assessment and management

Abstract State-space production models are increasingly being used in fisheries stock assessment as they provide the ability to account for observation and process errors. However, model performance when the population dynamics specified differs from the true biological process requires evaluation. We compared the estimation performance of a standard observation-error approach with a state-space production model for various simulated levels of model, process, and observation errors. We found that the state-space production model was generally superior to the observation-error estimator. However, the advantage of the state-space production model in parameter estimation diminished with increased model errors. The observation-error estimator outperformed the state-space production model when model error exceeded a certain level. A significant number of small process and observation error estimates (

[1]  André E. Punt,et al.  Fitting Surplus Production Models: Comparing Methods and Measuring Uncertainty , 1993 .

[2]  Perry de Valpine,et al.  Review of methods for fitting time-series models with process and observation error and likelihood calculations for nonlinear, non-Gaussian state-space models , 2002 .

[3]  P. de Valpine.  Monte Carlo State-Space Likelihoods by Weighted Posterior Kernel Density Estimation , 2004 .

[4]  W. Reed.  Analyzing Catch–Effort Data Allowing for Randomness in the Catching Process , 1986 .

[5]  Daniel K. Kimura,et al.  Kalman filtering the delay-difference equation: practical approaches and simulations , 1996 .

[6]  Kotaro Ono,et al.  A Bayesian approach to identifying and compensating for model misspecification in population models. , 2014, Ecology.

[7]  André E. Punt,et al.  Model performance analysis for Bayesian biomass dynamics models using bias, precision and reliability metrics , 2012 .

[8]  Cody S. Szuwalski,et al.  How well can FMSY and BMSY be estimated using empirical measures of surplus production , 2012 .

[9]  James T. Thorson,et al.  Uniform, uninformed or misinformed?: The lingering challenge of minimally informative priors in data-limited Bayesian stock assessments , 2017 .

[10]  André E. Punt,et al.  Extending production models to include process error in the population dynamics , 2003 .

[11]  Anders Nielsen,et al.  TMB: Automatic Differentiation and Laplace Approximation , 2015, 1509.00660.

[12]  Alan Hastings,et al.  FITTING POPULATION MODELS INCORPORATING PROCESS NOISE AND OBSERVATION ERROR , 2002 .

[13]  Mark A. Lewis,et al.  State-space models’ dirty little secrets: even simple linear Gaussian models can have estimation problems , 2015, Scientific Reports.

[14]  C. Walters,et al.  Quantitative fisheries stock assessment: Choice, dynamics and uncertainty , 2004, Reviews in Fish Biology and Fisheries.

[15]  R. Hilborn,et al.  State-space likelihoods for nonlinear fisheries time-series , 2005 .

[16]  C. Walters,et al.  Measurement Errors and Uncertainty in Parameter Estimates for Stock and Recruitment , 1981 .

[17]  Yong Chen,et al.  Parameter estimation in modelling the dynamics of fish stock biomass: Are currently used observation-error estimators reliable? , 1998 .

[18]  Noel G. Cadigan,et al.  A state-space stock assessment model for northern cod, including under-reported catches and variable natural mortality rates , 2016 .

[19]  Russell B. Millar,et al.  Non‐linear state space modelling of fisheries biomass dynamics by using Metropolis‐Hastings within‐Gibbs sampling , 2000 .

[20]  Milner B. Schaefer.  Some aspects of the dynamics of populations important to the management of the commercial marine fisheries , 1991 .

[21]  J. Thorson,et al.  Spawning biomass reference points for exploited marine fishes, incorporating taxonomic and body size information , 2012 .