Some Catch-at-Age Analysis Methods and Models Compared on Simulated Data

Thorvaldur Gunnlaugsson

.

DOI: 10.4236/ojms.2012.21003                

Abstract

Estimation of parameters and random effects using true maximum likelihood methods is compared to the commonly used penalized maximum likelihood method. The simulated catch-at-age datasets have all conceivable noise in stock and fishing dynamics in addition to the observation error on the catch. Improvement is modest in simple models but refinements that are only possible with these methods provide additional precision. Unbiased estimation of natural mortality is made possible with these methods, but precision is low unless variation in fishing effort between years is large and other variation small, in particular the uncertainty in the recruitment. Relatively unbiased estimates of all other in-put parameters and variances were obtained. Alternately the stock may be updated with the catches directly, rather than through the fishing mortality. This can be done exactly and such that bias in the final stock is small. Such a model will test a different error structure and may also be more appealing for presentation of results as the catches are in better agreement with the changes in the estimated stock.

Keywords

Stock Assessment; Simulation; AD Model Builder; Integrated Likelihood; Natural Mortality;Random Effects; VPA

Conflicts of Interest

The authors declare no conflicts of interest.

References

[1]	J. G. Pope, “An Investigation of the Accuracy of Virtual Population Analysis,” International Commission for the Northwest Atlantic Fisheries Research Bulletin, Vol. 9, 1972, pp. 65-74.
[2]	A. D. MacCall, “Virtual Population Analysis (VPA) Equations for Nonhomogeneous Populations, and a Family of Approximations Including Improvements on Pope’s Cohort Analysis,” Canadian Journal of Fisheries and Aquatic Sciences, Vol. 43, No. 12, 1986, pp. 2406-2409. doi:10.1139/f86-298
[3]	T. A. Branch, “Differences in Predicted Catch Composition between Two Widely Used Catch Equation Formulations,” Canadian Journal of Fisheries and Aquatic Sciences, Vol. 66, No. 1, 2009, pp. 126-131. doi:10.1139/F08-196
[4]	A. C. Harvey, “Forecasting Structural Time Series Models and the Kalman Filter,” Cambridge University Press, Cambridge, 1989.
[5]	G. Guemundsson, “Time Series Analysis of Catch-at-Age Observations,” Applied Statistics, Vol. 43, No. 1, 1994, pp. 117-126. doi:10.1016/0167-6687(94)90696-3
[6]	P. de Valpine and R. Hilborn, “State-Space Likelihoods for Nonlinear Fisheries Time-Series,” Canadian Journal of Fisheries and Aquatic Sciences, Vol. 62, No. 9, 2005, pp. 1937-1952.
[7]	H. J. Skaug, “Automatic Differentiation to Facilitate Maximum Likelihood Estimation in Nonlinear Random Effects Models,” Journal of Computational and Graphical Statistics, Vol. 11, No. 2, 2002, pp. 458-470. doi:10.1198/106186002760180617
[8]	H. J. Skaug and D. A. Fournier, “Automatic Approximation of the Marginal Likelihood in Non-Gaussian Hierarchical Models,” Computational Statistics & Data Analysis, Vol. 51, No. 2, 2006, pp. 699-709. doi:10.1016/j.csda.2006.03.005
[9]	ICES, NWWG Report, 2011. http://www.ices.dk/reports/ACOM/2011/NWWG/Sec10Icelandichaddock.pdf
[10]	S. Aanes, S. Engen, B. S?ther and R. Aanes, “Estimation of the Parameters of Fish Stock Dynamics from Catch-at- Age Data and Indices of Abundance: Can Natural and Fishing Mortality Be Separated?” Canadian Journal of Fisheries and Aquatic Sciences, Vol. 64, No. 8, 2007, pp. 113-1142. doi:10.1139/f07-074