In iantaylor-NOAA/Lingcod_2021: Estimate the Status of Lingcod off the US West Coast in 2021

Priors {#sec-priors}

The prior distributions for female and male natural mortality (Figure \@ref(fig:compare-north-vs-south-pars)) were based on the Hamel [-@hamel_method_2015] meta-analytic approach with an assumed maximum age of 18 and 13 years old for females and males, respectively (see Section \@ref(sec-biological-data) for details).

The prior distribution for steepness (Figure \@ref(fig:compare-north-vs-south-pars)) was based on the prior distribution used in the assessment of Pacific Hake, Merluccius productus [@JTC2021], which is based on the 20th, 50th, and 80th percentiles (0.67, 0.79, and 0.87, respectively) from Myers et al. [@myers1999]. The prior has a beta distribution given the following parameters: 9.76 and 2.80, which translates to a mean of 0.777 and a log-standard deviation of 0.113. This prior has been used for Pacific Hake since 2007. The terms of reference for groundfish managed by the \gls{pfmc} suggest that priors on steepness for rockfish should have a mean of 0.72 and a standard deviation of 0.16 but do not specify anything related to r spp. An analysis from FishLife suggested a prior with a mean of 0.67 and a standard deviation of 0.22.

Estimated parameters {#sec-estimated-parameters}

# get parameter names from base model
estimated_pars <- model$parameters %>%
  dplyr::filter(!is.na(Active_Cnt)) %>%
  rownames()

estimated_pars <- estimated_pars[!grepl("ForeRecr", estimated_pars)]
nrecdevs <- sum(grepl("RecrDev", estimated_pars))
first_main <- min(model$recruit$Yr[model$recruit$era=="Main"])
last_main <- max(model$recruit$Yr[model$recruit$era=="Main"])

The base model has a total of r length(estimated_pars) estimated parameters (Table \@ref(tab:table-pars-base)) that can be grouped into the following categories and are described in more detail in the following sections:

• 2 natural mortality parameters (female and male $M$)
• 10 growth parameters, where females and males each had
• 3 von Bertalanffy parameters (length at age 0.5, length at age 14, and $k$)
• 2 parameters controlling variability in growth, the CV in length at age 0.5 and the CV in length at age 14 with a linear ramp in length-at-age
• r 2 + sum(grepl("RecrDev", estimated_pars)) recruitment parameters
• $log(R_0)$ controlling equilibrium recruitment
• $h$ controlling the steepness of the stock-recruit relationship
• r nrow(model$recruit) recruitment deviations parameters covering the range r min(model$recruit$Yr)-r model$endyr, with r first_main-r last_main representing the "main" period modeled as a zero-centered deviation vector
• r sum(sum(grepl("Q_extra", estimated_pars))) extra standard deviation parameters for indices ```{asis, echo = TRUE, eval = ifelse(params$area == "South", TRUE, FALSE)}
• 2 parameters representing the catchability in the two periods of the index for the recreational California fleet (all other catchability values were derived analytically)

- `r sum(grepl("Size_DblN", estimated_pars))` selectivity parameters, of which
`r sum(grepl("Size_DblN.*BLK", estimated_pars))` represented changes over time
- `r sum(sum(grepl("Retain", estimated_pars)))` retention parameters, all of which
represented changes over time because the retention prior to 1998 was assumed to be
high for all selected fish.

#### Fixed parameters {#sec-fixed-parameters}

Commercial fishery retention parameters for the period prior to 1998 were
fixed at values that led to almost 100\% retention as discussed in Section 
\@ref(sec-model-selectivity).

```{asis, echo = TRUE, eval = ifelse(params$area == "North", TRUE, FALSE)}
The extra standard deviation parameters for two Oregon CPUE indices were fixed
at 0 as discussed in Section \@ref(sec-model-selection). 
The parameter controlling the descending slope of the dome-shaped selectivity 
function hit a bound for many combinations of fleet and time-block in the north model.
Model diagnostics indicated that these parameters were causing problems with 
convergence so they were all fixed equal to the bound.

<!-- manual updates needed here if models change -->
Selectivity was estimated as asymptotic in many cases 
(Figure \@ref(fig:selectivity_comm)-\@ref(fig:selectivity-noncomm)),
including the commercial trawl prior to 
1993, the fixed-gear fleet (not time-varying), any many of the time-blocks 
for the recreational fleets. In all these cases, the parameter controlling
the descending slope of the dome-shaped selectivity was fixed at the upper bound. 
One descending slope parameter hit the lower bound, for commercial trawl 
in the period 1993-1997. This parameter was
also fixed at that bound (1.0). The length at 50\% retention for the 
commercial trawl fleet hit the upper bound of 100 cm for the period 1998-2006
and was likewise fixed at that bound.

The following additional parameters were fixed based on standard practices for west coast assessments. ``{asis, echo = TRUE, eval = !any(grepl("sigma", estimated_pars))} The standard deviation of recruitment deviations was fixed atr sprintf("%.2f", model$parameters %>% dplyr::filter(grepl("sigma",Label)) %>% dplyr::pull(Value))`.

<!-- manual updates needed here if models change -->
A tuning algorithm [@methot_adjusting_2011] indicated little change was needed from 
this starting value and sensitivity analyses showed the results to be relatively
insensitive to alternative values.
```{asis, echo = TRUE, eval = !any(grepl("Mat", estimated_pars))}
The maturity parameters were fixed at values based on the analysis described in Section
\@ref(sec-biological-data-maturation-and-fecundity).

{asis, echo = TRUE, eval = !any(grepl("Wtlen", estimated_pars))} The weight-length parameters were fixed at values estimated externally using data with paired observations of length and weight from the \gls{s-wcgbt}, as described in Section \@ref(sec-biological-data-Length-Weight-relationship).

iantaylor-NOAA/Lingcod_2021 documentation built on Oct. 30, 2024, 6:42 p.m.