A better estimator of mortality rate from age-frequency data

作者：Millar Russell B^*

来源：Canadian Journal of Fisheries and Aquatic Sciences, 2015, 72(3): 364-375.

DOI：10.1139/cjfas-2014-0193

The Chapman-Robson and weighted-regression estimators are currently the two preferred methods for estimation of instantaneous mortality, z, from a cross-sectional sample of age-frequency data. They are derived under the assumption of steady-state population dynamics. Here, a new estimator is developed from a population model that explicitly includes annual variability in recruitment. The new estimator is trivial to implement using existing generalized linear mixed model software. It is vastly superior to both the Chapman-Robson and weighted-regression estimators under a wide range of simulation scenarios in which sources of variability include partial recruitment to the fishery, autocorrelated annual recruitment, variability in annual survival, ageing error, and sampling randomness. All estimators produced confidence intervals that had lower actual coverage than their nominal 95% coverage. Nonetheless, the new estimator had the highest actual coverage, and under some scenarios this was achieved with a narrowest confidence interval.

• 出版日期2015-3