Chapter 10 Five area model permutations

  • “Init (ALK)” The initial model, assumes multinomial likelihoods for composition. AFs were estimated using the Age length key. Tag recovery data was assumed to be Negative Binomial. Annual recruitment deviations are shared across the regions
  • “Direct Ageing” The same as “Init (ALK)” except using direct ageing estimators AF
  • “D-M” The same “Init (ALK)” but assumes compositional data is Dirichlet Multinomial
  • “Regional Recruitment” same as “D-M” but with regional specific annual recruitment deviations
  • “Recruit constrained” same as “Regional Recruitment” but constrains the annual recruitment deviations to sum = 0
Estimated SSBs across the different five region model

Figure 10.1: Estimated SSBs across the different five region model

Estimated Depletion across the different five region model

Figure 10.2: Estimated Depletion across the different five region model

Estimated Recruitment across the different five region model

Figure 10.3: Estimated Recruitment across the different five region model

Estimated Fishing mortalities for the fixed gear fishery

Figure 10.4: Estimated Fishing mortalities for the fixed gear fishery

Estimated Fishing mortalities for the Trawl fishery

Figure 10.5: Estimated Fishing mortalities for the Trawl fishery

Estimated selectivities

Figure 10.6: Estimated selectivities

Notes about this model

I need to set the delta selectivity parameters for the survey to be the same between male and females so that I get converged models. This is also done in the current assessment.

The trawl selectivity parameters can be problematic. I am not scaling the selectivities by the max which is done in the assessment, this can result in selectivities not have max values close to one. When estimating separate male and female trawl selectivities, the female selectivity frequently is estimated at an usual shape. To stabilize the model I have forced male and females to have the same selectivity for the trawl fishery.