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

# get the block_breaks vector from this file:
knitr::purl("doc/model-selectivity.Rmd")
source("model-selectivity.R")
usethis::use_data(block_breaks_north, overwrite = TRUE)
usethis::use_data(block_breaks_south, overwrite = TRUE)

Selectivity and retention {#sec-model-selectivity}

Selectivity was modeled using the double-normal function of length (option 24 in Stock Synthesis; Table \ref{tab:output-specifications}) with three of the six possible parameters estimated for each fleet.

The estimated parameters controlled the length at peak selectivity, the slope of the ascending limb and the slope of the descending limb (parameters 1, 3, and 4 in SS). The parameter controlling the width of the top was fixed at a small value (-15) to make a smooth transition from ascending to descending with no flat top. ```{asis, opts.label = 'north', echo = TRUE} Parameters 5 and 6 which control additional scaling of initial and final values were not used (via the -999 code in SS).

```{asis, opts.label = 'south', echo = TRUE}
Parameters 5 and 6 which control additional scaling of initial and final values
were not used (via the -999 code in SS) with the exception of the \gls{s-tri}
selectivity. In that one case, the estimation of parameter 5 controlling the
initial selectivity at for the smallest size bins provided a better fit to the 
length comps for the smallest individuals and resolved a problem of the ascending
slope parameter hitting the lower bound.

Retention was modeled for the commercial trawl and commercial fixed-gear fleets using a logistic function of length with 2 of the possible 4 parameters estimated. The two estimated parameters controlled the length at 50\% retention and the slope of the ascending curve. Asymptotic retention was fixed at 100\% and no sex-specific differences in retention were estimated.

Changes in selectivity and retention over time were modeled to account for management changes and gear changes. For changes in selectivity, all three estimated double-normal parameters were allowed to vary among time blocks as any change in the peak parameter will impact the shape associated with the other two. For changes in retention, the length at 50\% retention was allowed to vary over time, but the slope parameter was assumed to apply to all years after the management measures were imposed in 1998.

Prior to 1998, all selected fish are assumed to be retained in the commercial fisheries in keeping with comments from those familiar with the history of the commercial fishery at the pre-assessment workshop and the Pikitch study in which more than 99\% of the observed r spp were retained (pers. comm., J. Wallace, \gls{nwfsc}).

For commercial fisheries the assumed discard mortality rates are retained from the previous assessment: 50\% for trawl and 7\% for fixed gear.

Selectivity for the recreational fisheries was modeled as representing retained fish only and the retention process is represented within the assessment model only through time-varying selectivity that accounts for changes in minimum size limits. A 7\% mortality rate has been assumed for lingcod discarded in recreational fisheries and these dead discards are accounted for in the total catch, but an exploratory analysis (Figure \@ref(fig:rec-selectivity-illustration)) indicated the the difference between the expected length distributions with or without the dead discards included was small. Therefore, the length compositions from recreational fisheries exclude all discards. Alternative approaches to these assumptions were considered, but the available data do not support them as it would "require 'unscrambling the egg' since B1 is a mixture of retained and discarded fish in the MRFSS data" (pers. comm., E.J. Dick, \gls{swfsc}).

The specific years with time blocks used for each fleet are noted below.

For commercial trawl, the time period of blocking on selectivity parameters were reduced from the the 2017 assessment, removing some blocks from the early years. The remaining points of change in selectivity were:

• 1993 (change in mesh size)
• 1998 (trip limits and size limits)
• 2011 (implementation of the catch shares program)

Additionally, there were retention changes modeled in the following years, retained from the 2017 assessment.

• 1998 (implementation of groundfish regulations)
• 2007 (groundfish regulations)
• 2010 (pre-catch share behavior change and clear reduction in discard rate)
• 2011 (implementation of the catch shares program)

All blocks were applied to the length at 50\% retention, but only the changes in 1998 and 2011 were applied to the parameter controlling the slope of the retention function.

# create empty list
block_breaks <- list()

# first additions of blocks break points
block_breaks[["Comm_Trawl_sel"]] <- c(1993, 1998, 2011)
block_breaks[["Comm_Trawl_ret_infl"]] <- c(1998, 2007, 2010, 2011)
block_breaks[["Comm_Trawl_ret_width"]] <- c(1998, 2011)

For commercial fixed-gear, selectivity was assumed to be constant and retention was assumed to change in the following years:

• 1998 (implementation of groundfish regulations)
• 2011 (implementation of the catch shares program)

The 2017 south model had a additional blocks starting in 2002 and 2003 for the south model to account for closed areas. However, the composition data don't show evidence of a change around that time.

# add block break points to list created above
block_breaks[["Comm_Fix_ret_infl"]] <- c(1998, 2011)
block_breaks[["Comm_Fix_ret_width"]] <- c(1998, 2011)

```{asis, opts.label = 'north', echo = TRUE}

Changes in selectivity of the recreational WA fleet began in

• 1987 (management changes)
• 1995 (management changes)
• 1998 (first year of a series of management changes also made in 1999, 2000, 2001)
• 2007 (management changes)
• 2011 (management changes)
• 2017 (22" minimum size requirement was lifted from July 2017 to current for all catch areas but the change happened in the middle of the year)

```r 
# add block break points to list created above
block_breaks[["Rec_WA_sel"]] <- c(1987, 1995, 1998, 2007, 2011, 2017)

```{asis, opts.label = 'north', echo = TRUE} Changes in selectivity of the recreational OR fleet began in

• 1995 (the first year with a minimum size limit--22 inches)
• 1998 (min size limit was increased to 24 inches)
• 2007 (min size limit was reduced from 24 to 22 inches)

Not accounted for is a max size limit in place in Oregon for one year only (2000) of 34 inches, which is visible in the data, but the absence of larger fish observed in a single year was assumed to have a negligible impact on the model results.

```r 
# add block break points to list created above
block_breaks[["Rec_OR_sel"]] <- c(1995, 1998, 2007)

Changes in selectivity of the recreational CA fleet began in

```{asis, opts.label = 'north', echo = TRUE} - 1999 (min size limit 24 inches) - 2011 (min size limit decreased to 22)

This represents a smaller number of blocks than used in the south model which also included blocks starting in 1983 and 2017 as the data from northern California used in the north model are sparser and correspond to a small fraction of the catch compared to the south model.

```{asis, opts.label = 'south', echo = TRUE}
- 1983 (notable reduction in the number of small fish observed)
- 1999 (min size limit 24 inches)
- 2000 (min size limit increased to 26)
- 2002 (min size limit decreased to 24)
- 2011 (min size limit decreased to 22)
- 2017 (recreational fishery had access to deeper waters)

Not accounted for is a min size limit of 30 inches in 2004 as sample sizes
were low in that year.

# split breaks into separate north and south
block_breaks_north <- block_breaks
block_breaks_south <- block_breaks

# add block break points to list created above
block_breaks_north[["Rec_CA_sel"]] <- c(      1999,             2011)
block_breaks_south[["Rec_CA_sel"]] <- c(1983, 1999, 2000, 2002, 2011, 2017)

A block on selectivity associated for the Triennial survey was explored beginning 1995, the year associated with expanded spatial coverage and shift in timing, but not included in the base model.

# add block break points to list created above
# block_breaks[["Surv_TRI_sel"]] <- c(1995)

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