An ASPIC Based Assessment of Redfish (S. mentella and S. fasciatus) in NAFO Divisions 3LN (can a surplus production model cope with bumpy survey data?)

Abstract

There are two species of redfish in Divisions 3L and 3N, the deep-sea redfish (Sebastes mentella) and the Acadian redfish (Sebastes fasciatus) that have been commercially fished and reported collectively as redfish in fishery statistics. Redfish in Div. 3LN is regarded as a management unit composed of two Grand Bank populations from those two very similar redfish species. The present ASPIC assessment is based on the logistic form of a non-equilibrium surplus production model (Schaeffer, 1954; Prager, 1994), adjusted to a standardized catch rate series (Power, 1997) and to most of the stratified-random bottom trawl surveys conducted in various years and seasons in Div. 3L and Div. 3N from 1978 onwards. These surveys were framed according to the input formulation previously adopted on the 2nd take of the ASPIC 2008 assessment (Ávila de Melo and Alpoim, 2010a). The assessment was preceded by an exploratory analysis to check the response of the model to the inclusion of the Spanish spring survey series on Div. 3N and to the 2010-2011 update of the remaining three Canadian survey series that are at present the backbone input of this assessment. Each of these series includes recently high points that are well above their overall increasing trends, observed 2002 onwards. The analysis point out that in terms of consistency with previous assessments and the past history of the redfish fishery, as well as performance of the model, the ASPIC 2012 option with the exclusion of the Spanish survey and the removal of the recent outliers from the respective Canadian series represents the better update of the survey data input framework. The chosen input formulation run afterwards with different last year survey results and different starting guesses for key parameters and different random number seeds, in order to test the robustness of ASPIC results to turbulence in the inputs used to initialize the model deterministic run. A 2012 versus 2010 ASPIC comparative assessment (both on FIT and BOT modes) and a 2012-2010 retrospective analysis were also carried out to check the consistency between the two last full assessments and the magnitude of bias on relative biomass and fishing mortality in response to the general increase of the still standing survey series. Regardless the input formulations, the starting guess region, the mode of the ASPIC runs or the retrospective patterns, the 2012 assessment reiterates the main conclusion of the previous ones: the biomass of redfish in Div. 3LN is above Bmsy , while fishing mortality is well below Fmsy . Most recent catches continue to be at a low level on the historical context of this fishery and the answer of the stock to a direct fishery of the magnitude of years between the mid 1960’s and the mid 1980’s is unknown. Projections were made on the short term, in order to select a 2013-2012 catch roof large enough to consolidate the reopening of a redfish direct fishery on divisions 3L and 3N but keeping a high probability of the stock staying above Bmsy. This high probability is defined by the lower 80% confidence limit of the projected relative biomass trajectory being at or above Bmsy in 2013-2015.