Pasar al contenido principal

    Casal2 Stock Assessment for Antarctic krill in CCAMLR Subarea 48.1

    Solicitar acceso a documento de reunión
    Número de documento:
    WG-FSA-2023/14
    Autor(es):
    Kinzey, D. and G.M. Watters
    Presentado por:
    Dr Doug Kinzey (Estados Unidos de América)
    Aprobado por:
    Dr George Watters (Estados Unidos de América)
    Punto(s) de la agenda
    Resumen

    The Casal2 model for Antarctic krill initially reported in Kinzey and Watters (2023) now incorporates MCMC estimation of parameter uncertainty in addition to the MPD estimates described in the earlier paper. Four model configurations are explored employing two different methods of estimating annual recruitment multipliers - simplex transformation and non-simplex - and two alternative approaches to modeling future projected catches, either with all model years combined into a single ‘@process Instantaneous_Mortality’ block, or with future catch years separated from past catch years into a second ‘@project future_catch’ block. Differences in AIC scores between the simplex and non-simplex approaches indicated the two non-simplex configurations were much better descriptors of the data than the two configurations using the simplex transformation. The two configurations that combined past and future catches into the same block produced MPD and MCMC outputs that could be processed by the ‘extract.mcmc()’, ‘extract.mpd()’ and ‘extract.tabular()’ functions in the r4Casal2 package so that the CCAMLR decision rules could be applied to the projections.  As implemented here, in the two configurations that separated future catches into a ‘@project future_catch’ block the spawning stock biomasses (SSBs) for the projected years 2022-2041 were not extracted by the r4Casal2 ‘extract()’ functions so the CCAMLR decision rules could not be applied to those. One of these configurations, the non-simplex model configuration using a separate block for future catches, had the best AIC score of the four configurations. The configuration with the second-best AIC (∆AIC = 1.2) was able to be read by the ‘extract()’ functions so those results are reported here. The Casal2 input files, ‘estimation.csl2’, ‘observation.csl2’, ‘population.csl2’, ‘reports.csl2’ and ‘config.csl2’ for the models described in this paper are available at ‘https://github.com/us-amlr/Casal2-krill-model-mcmc’.