In pbs-assess/gfsynopsis: Data Synopsis Reports for PBS Groundfish

Summary of changes to the MSSM survey and implications for groundfish indices {#sec:summary-of-changes}

To highlight changes that may affect interpretation of the groundfish data we used data extracted from the GFBio database for 1975--2022 [@databases2023]. A publicly available data for the Multispecies Small Mesh survey is also available on the Ocean Biodiversity Information System (OBIS) [@mssm2022obis].

Survey timing and area

The Multi-Species Small Mesh (MSSM) survey is typically conducted in the last week of April to the first weeks of May with the exception of late surveys in 1985 and 1987 (Table \@ref(tab:change-table)). In 1977 and 1978, additional surveys were conducted in September on commercial vessels, but these have been excluded from this report. The consistency in survey timing means that timing is unlikely to affect abundance indices.

The MSSM survey is a fixed station survey with transects run in an east-west direction [@rutherfordinprep]. Over the years, the extent of the survey has decreased (Figure \@ref(fig:historical-grid)). In part, this is likely because in early years, when shrimp were caught near the east or west boundaries of the survey area, additional trawls would be carried out further in this direction until shrimp stopped being caught (Personal communication: D.T. Rutherford).

In 2022, sampling was extended eastward and northward to cover the original extent better than any year since 1975 (Figure \@ref(fig:spatial-shift)). Changes in spatial coverage over time mean that an index standardized using a model-based approach may be better suited for calculating an index over the full time-series than a design-based index.

knitr::include_graphics(here::here("report/mssm-tech-report/figure/grid-historical-nav-changes_no-points.png"), dpi = NA)

The navigation system used in the MSSM survey has changed twice: once in 1979 from Loran A to Loran C and again in 1998 from Loran C to the Global Positioning System (GPS) (Figure \@ref(fig:historical-grid)) [@rutherfordinprep]. The improved accuracy of GPS means that early survey locations may therefore be more variable. As a consequence of more accurate positioning, @sinclair2001pcod found that there was a trend in distance travelled over time with shorter distances travelled for the 30 minute tows in the MSSM survey over time.

Between 1975 and 1977, there was a noticeable shift in sampling locations relative to subsequent years (Figure \@ref(fig:spatial-shift)). It is possible that this reflects surveys used to establish the extent of the shrimping grounds (Personal communication: D.T. Rutherford). This could also be the result of using the less accurate Loran A navigation system.

knitr::include_graphics(here::here("report/mssm-tech-report/figure/grid-spatial-sampling-changes.png"), dpi = NA)

Gear changes

Gear changes occurred in 1977, 2006, 2007, and 2011 (Table \@ref(tab:change-table)). In 1977, the net, foot rope, and doors were changed. A series of calibration tows were performed in 1976. The 70 foot (21.3 m) semi-balloon shrimp net was replaced with the National Marine Fisheries Science (NMFS) 61 foot (18.6 m) shrimp trawl net [@rutherfordinprep]. No calibration tows were performed, but @boutillier1976cruiserep found that the NMFS net was 1.4 more efficient than the 70 foot semi-balloon net, meaning that catch weights in the first years of the survey are likely to be biased low relative to 1977 onward.

In 2006, the NMFS net was replaced with a 61 foot American shrimp trawl net (Table \@ref(tab:change-table)). However, only two comparable sets of tows were collected [see Table 2 and 3 in @rutherfordinprep]. In these two tows, there were three species not caught in the NMFS net and seven species not caught in the American net (Figure \@ref(fig:net-comp)). The mean catch of these species was less than 10 kg in the net that did catch them. The lack of sufficient calibration tows means that it is challenging to account for any changes in catchability or selectivity in a modeled index. Furthermore, the visual comparison of only two comparative tows does not provide a clear indication of how the gear change affected capture rates (Figure \@ref(fig:net-comp)).

knitr::include_graphics(here::here("report/mssm-tech-report/figure/net-comp.png"), dpi = NA)

Catch sorting and sampling

post_2003_spp <- readRDS(here::here("report/mssm-tech-report/data-outputs/post-2003-spp.rds"))

The most notable change to catch sorting and sampling occurred in 2003; from 2003 to present, all species have been sorted and identified to the species or lowest taxonomic level [@rutherfordinprep]. Prior to 2003, r length(post_2003_spp) species had zero catch weights and catch counts. Of these, six species are caught regularly: Big Skate, Sandpaper Skate, Longnose Skate, Blackbelly Eelpout, Stripetail Rockfish, and Slim Sculpin (Figure \@ref(fig:mean-catch-pre-post-2003)).

knitr::include_graphics(here::here("report/mssm-tech-report/figure/sampling-2003.png"), dpi = NA)

To check if some species were identified at higher taxonomic levels prior to 2003, we extracted data from GFBio for higher order taxonomies of species that have been identified in the MSSM survey since 2003 (Figure \@ref(fig:spp-aggregate-plot)). Sculpins and skates are the species that were counted at the family level prior to 2003. Whereas prior to 2003, flatfishes and rockfishes appear to be identified to species if they were counted/weighed. The mean annual catch of all flatfishes identified to species was higher after 2001, which is the year that the catch was sorted below deck on a conveyor belt (Figure \@ref(fig:spp-aggregate-plot)). This suggests that flatfish counts before 2003 are likely underestimated because the catch was not fully sorted.

In 2002, abnormally high catches of 'Eelpouts' and 'Skates' were recorded relative to years before 2002 and when we aggregated species catches for post-2003 years (Figure \@ref(fig:spp-aggregate-plot)). These outliers do not appear to reflect a decimal recording error, duplicated data (there were three high 'Skate' sets in 2002), nor is it likely that it is due to a misidentification since these are easily distinguished from other species groups. It is possible that these species aggregates were data entry errors (e.g., species coding errors).

knitr::include_graphics(here::here("report/mssm-tech-report/figure/aggregated-id-level-plot.png"), dpi = NA)

Species that were possibly misidentified prior to 2003 include (Personal communications: D.T. Rutherford, M. Wyeth, S. Hardy):

• Stripetail Rockfish as Sharpchin Rockfish
• Yellowmouth as Pacific Ocean Perch
• Small Yelloweye may have been mixed in with Canary Rockfish
• Small Pacific Cod as Pacific Tomcod
• Petrale Sole as Flathead Sole (at all sizes that Flatheads are observed)

For example, recorded abundances of Pacific Tomcod decrease after 2003, and it is possible that this is because of misidentified Pacific Cod prior to 2003 (Figure \@ref(fig:cod-comparison)). However, these abundances were relatively small, with annual catch sizes of Pacific Tomcod typically less than 5 kg/year.

knitr::include_graphics(here::here("report/mssm-tech-report/figure/sampling-cod.png"), dpi = NA)

\clearpage

pbs-assess/gfsynopsis documentation built on March 26, 2024, 7:30 p.m.