Recommendations from 2017 assessment review and report

[okenk]

Major uncertainties (STAR Panel report):

• The mechanisms responsible for the apparent differences in natural mortality rates between the two sexes (in both models) also warrant greater investigation. Likelihood profiles indicate significant tension between compositional (length and age) data and index data in the northern assessment. There is little evidence in the likelihood profiles that steepness can be reliably estimated, consequently fixing steepness at the point estimate of the prior was considered to be appropriate.

Research recommendations (STAR panel report):

• The age data available for the northern model are strongly indicative of a higher natural mortality rate for female yellowtail rockfish. Yellowtail are one of several rockfish species that demonstrate this apparent difference in sex ratios at age, and historically there has been concern regarding whether differential natural mortality rates throughout the lifespans of populations are reasonable, whether these differences reflect greater mortality rates of older females, or whether these difference reflect some manifestation of dome-shaped age-based selectivity (such that older females are less vulnerable to fisheries or surveys, presumably as a result of habitat associations). These models did not explore alternative explanations for these differences at the level that has been done for other stock assessments (such as Canary and Black Rockfish, where increasing natural mortality rates for females with age, as well as both size- and age-based dome-shaped selectivity have been explored). Given the richness of age data for yellowtail rockfish in the north, additional investigations that better quantify the phenomena, and evaluate potential mechanisms for the observed discontinuities, should be pursued. [To do]
• The draft northern yellowtail assessment models included indices of relative abundance based on fishery-dependent time series, including a trawl logbook CPUE index and an index of abundance based on yellowtail bycatch in the at-sea Pacific whiting fishery. Upon greater discussion of challenges associated with the development of these indices, particularly with regards to possible differences by state in the resolution of market categories in the logbook data, and how the bycatch rate information was standardized for the whiting fishery bycatch index, the STAT recommended exclusion of both indices in the final base model. However, as the indices were influential with respect to model results, greater exploration of the potential for these data to inform a relative abundance index, particularly for the trawl logbook cpue data, would benefit future assessment efforts. [Worthwhile???]
• As yellowtail rockfish is a semipelagic species, it may not always be reliably sampled in bottom trawl surveys, particularly if the depth distribution is sensitive to environmental conditions (for example, for widow rockfish it has been suggested in the past that El Niño years or other periods of low productivity, individuals may have a more benthic, relative to pelagic, distribution and therefore be more vulnerable to bottom trawl surveys). Consideration of alternative survey methods (e.g, acoustic surveys, midwater trawl surveys) and/or the means to account for changes in catchability that may be associated with environmental factors, could improve the ability of survey indices to track stock abundance. [Didn't happen]
• As northern yellowtail presumably represents a transboundary stock and resource, work towards a combined US/Canadian stock assessment would greatly aid our overall understanding of stock status. [Didn't happen]
• Common documentation of data streams and sources to support fishery independent and fishery dependent indices and compositional data could reduce the burden on assessment analysts to provide details about each data source, and allow reviewers a robust source of information on the most important, common data sources for any given stock assessment cycle. [This is now much improved]

Research recommendations (assessment report):

• A problem common to assessments of all stocks caught in the midwater is the lack of a targeting survey. The STAR panel report accompanying this document suggests several avenues to approach this problem. [Didn't happen]
• Research to determine whether old females of a variety of rockfish species actually have a mortality rate different than that of younger females. Assessments variously treat the discrepancies seen in sex ratios of older fish as either mortality-related or due unavailability to the fishery (e.g., ontogenetic movement offshore, or to rockier habitats). As these assumptions impact model outcomes very differently, resolving this issue would greatly improve confidence in the assessments. [To do (various sensitivities)]
• A hindrance to analysis of the commercial fishery is the inability to distinguish between midwater and trawl gear, particularly in data from the 1980s-1990s. Reliable recording of gear type will ensure that this does not continue to be problematic for future assessments. [Worthwhile?]
• For the next full assessment, we suggest the following:
  • A commercial index in the North. This is by far the largest segment of the fishery, and the introduction of the trawl rationalization program should mean that an index can be developed for the current fishery when the next full assessment is performed. [Worthwhile?]
  • Further investigation into an index for the commercial logbook dataset from earlier periods. [This seems harder?]
  • Further analysis of growth patterns along the Northern coast. The previous full assessment subdivided the Northern stock based on research showing differential growth along the coast, and although data for the assessment is no longer available along the INPFC areas used in that analysis, there may be some evidence of growth variability that would be useful to include in a future assessment. [I have looked into this a little with survey data. It is there for sure. Hard to account for within the model, but could try, for example, only including northern vs southern comp data]