In pbs-assess/gfdata: Data Extraction for DFO PBS Groundfish Stocks

knitr::opts_chunk$set(echo = TRUE)
library(dplyr)
library(gfdata)
library(ggplot2)

Background

We want to estimate historical Rex Sole discards. Our approach is to apply a ratio of modern Rex Sole discards to modern landings of a commonly co-caught species, or group of species, that was recorded well historically.

1. Pull modern catch data for Rex and cocaught species (from gfdata::get_cocaught_species()).

get_cocaught_species

rex <- get_cocaught_species(610, "groundfish trawl")
head(rex)

Here it appears Dover Sole and Pacific Cod are most frequently caught with Rex Sole. I expect Pacific Cod to have better records of historical landings, so will build modern Rex discard ratios relative to Pcod landings (as is done in reconstructing rockfish historic discards).

1. Sum annual catch of Rex and cocaught species (here, Pcod) landings and discards (from groundfish trawl fishery 1996-2006), and calculate annual ratio of target (Rex) discard to cocaught landings.

cocaught_ratios

rex_pcod <- cocaught_ratios(610, 222, start_year = 1996, end_year = 2006, 
  sector = "groundfish trawl")
head(rex_pcod)

1. Next, select trusted reported years (if not 1996-2006), calculate overall mean ratio (mean of annual means, as is done for rockfish), and apply to desired historic catch years.

discard_ratio_pcod <- mean(rex_pcod$`target_discard/cocaught_landed`)
discard_ratio_pcod

rex_discards_pcod <- arrange_catch(222) %>%
  filter(fishery_sector == toupper("groundfish trawl"), between(year,1970,1996)) %>% 
  group_by(year, fishery_sector) %>% 
  summarise(landed_kg = sum(landed_kg)) %>% 
  ungroup() %>% 
  mutate(rex_discards = landed_kg * discard_ratio_pcod,
    rex_discards_t_pcod = rex_discards/1000)
head(rex_discards_pcod)


ggplot(rex_discards_pcod, aes(x = year, y = rex_discards_t_pcod)) +
  geom_col()

Some questions:

Would a different reference species be better for this purpose?

Do we also want to calculate historical Rex landings in this way?

Or, if we trust our historical Rex landing records, do we want to base our discard estimates on those instead of a co-caught species?

Attempt # 2

Trying a reference species group instead of a single species - the top 5 co-caught species with Rex in the trawl fishery since 2008.

rex_group <- cocaught_ratios(610, c("dover sole", "pacific cod", "english sole", "petrale sole", "arrowtooth flounder"), start_year = 1996, end_year = 2006, 
  sector = "groundfish trawl")
head(rex_group)

Or without Arrowtooth Flounder:

rex_group2 <- cocaught_ratios(610, c("dover sole", "pacific cod", "english sole", "petrale sole"), start_year = 1996, end_year = 2006, 
  sector = "groundfish trawl")
head(rex_group2)

Estimate Rex discard ratio from mean of mean annual Rex discards/reference group landings (1996-2006). Here I'm using years from 1996-2006 to calculate the discard ratio because the trawl records are trusted in these years for rockfish reconstructions - is there a reason to restrict this period or use a different time period altogether?

discard_ratio_group <- mean(rex_group$`target_discard/cocaught_landed`)
discard_ratio_group

discard_ratio_group2 <- mean(rex_group2$`target_discard/cocaught_landed`)
discard_ratio_group2

Apply Rex discard ratio to historic time period of interest.

rex_discards_group <- arrange_catch(c("dover sole", "pacific cod", "english sole", "petrale sole", "arrowtooth flounder")) %>%
  filter(fishery_sector == toupper("groundfish trawl"), between(year,1970,1996)) %>% 
  group_by(year, fishery_sector) %>% 
  summarise(landed_kg = sum(landed_kg)) %>% 
  ungroup() %>% 
  mutate(rex_discards = landed_kg * discard_ratio_group,
    rex_discards_t_group = rex_discards/1000)
head(rex_discards_group)

rex_discards_group2 <- arrange_catch(c("dover sole", "pacific cod", "english sole", "petrale sole")) %>%
  filter(fishery_sector == toupper("groundfish trawl"), between(year,1970,1996)) %>% 
  group_by(year, fishery_sector) %>% 
  summarise(landed_kg = sum(landed_kg)) %>% 
  ungroup() %>% 
  mutate(rex_discards = landed_kg * discard_ratio_group2,
    rex_discards_t_group2 = rex_discards/1000)
head(rex_discards_group2)

ggplot(rex_discards_group, aes(x = year, y = rex_discards_t_group)) +
  geom_col()

ggplot(rex_discards_group2, aes(x = year, y = rex_discards_t_group2)) +
  geom_col()

pbs-assess/gfdata documentation built on Feb. 16, 2025, 7:47 a.m.