Final changes

documents/pre-Star/21f-.Rmd

@@ -18,7 +18,7 @@ Commercial landings data in Oregon were available from 1892--2022. Historical la
 
 Commercial landings data in California were available from 1916--2022. Historical landings from 1916 to 1980 were obtained following the process described in the southern California vermilion rockfish (*Sebastes miniatus*) assessment [see @dick2021vermillion for complete details]. Briefly, landings 1916--1968 came from a state reconstruction [@ralston_documentation_2010]. Landings from unknown gears in known regions were allocated proportional to landings from known gears within the same region for each year. Landings from unknown locations (Region 0) and unknown gears were allocated proportional to the landings from known gears across all known locations for each year. Landings from 1969--1980 came from \gls{calcom}, which for 1969--1977 incorporates fish ticket data including mixed species categories for rockfish that were assigned to individual species using the earliest species composition samples (from the late 1970s and ealy 1980s). Recent (1981--2022) landings by trawl and non-trawl gears were obtained from \gls{pacfin} (extracted on 03/23/2023).
 
-Additional catches caught off the coast of Oregon or Washington but landed in California from 1948-1968 were included, as these were not incorporated within the Ralston et al. [-@ralston_documentation_2010] reconstruction (E. Dick, personal communication, 3/16/23). These final landings are in the California landings history for consistency with treatment for other states, because historical catches landed in Oregon or Washington but caught in the other state's waters are similarly attributed to the state where they were landed.
+Additional catches caught off the coast of Oregon or Washington but landed in California from 1948-1968 were included, as these were not incorporated within the Ralston et al. [-@ralston_documentation_2010] reconstruction (E. Dick, \gls{swfsc}, personal communication, 3/16/23). These final landings are in the California landings history for consistency with treatment for other states, because historical catches landed in Oregon or Washington but caught in the other state's waters are similarly attributed to the state where they were landed.
 
 #### Commercial Discards
 
@@ -60,15 +60,15 @@ Recreational landings for ocean boat modes from 2001--2022 are available from \g
 
 Recreational landings in California were available from 1928--2022. Historical estimates of recreational landings 1928--1979 were obtained from the previous assessment [@thorson_status_2015], which were in turn were obtained from \gls{calcom}. For years since 1979, recreational landings along with estimates of dead discards came from \gls{recfin} via the \gls{mrfss} (1980--2003) and \gls{crfs} (2005--2022) sampling programs. Landings and dead discards were added together to obtain estimates of total removals. Discards prior to 1980 were assumed to be zero.
 
-A number of years with missing or incomplete estimates for recreational removals were filled in. The removals in 1980 from \gls{mrfss} were not used due to survey quality problems related to 1980 being the first year of \gls{mrfss} [@Karpov1995; @cope_status_2009]. Removals for this year were averaged across removals in 1979 and 1981. No or minimal party/charter (PC) estimates were available from 1993--1995, so PC mode removals for these years were calculated assuming an average ratio of private/rental (PR) to PC modes across all years with data within \gls{mrfss} and added to PR values to obtain total recreational estimates. The \gls{mrfss} sampling program did not sample from 1990--1992, and so estimates during these three years were obtained as averages from neighboring years. Estimates in 1990 were the average from 1987--1989, estimates in 1991 were the average from 1987--1989 and 1993--1995, and estimates in 1992 were the average from 1993-1995. Because of the transition from \gls{mrfss} to \gls{crfs}, no estimate was available in either database in 2004 for this assessment cycle. \Gls{cdfw} provided a value of 10.59 mt for 2004 (J. Budrick, CDFW, personal communication), based on an available previous pull of \gls{mrfss} data for the gopher rockfish assessment. Estimates for California recreational removals from the previous canary rockfish assessment were not used because these appeared to only reflect estimates of landings, and not total removals.
+A number of years with missing or incomplete estimates for recreational removals were filled in. The removals in 1980 from \gls{mrfss} were not used due to survey quality problems related to 1980 being the first year of \gls{mrfss} [@Karpov1995; @cope_status_2009]. Removals for this year were averaged across removals in 1979 and 1981. No or minimal party/charter (PC) estimates were available from 1993--1995, so PC mode removals for these years were calculated assuming an average ratio of private/rental (PR) to PC modes across all years with data within \gls{mrfss} and added to PR values to obtain total recreational estimates. The \gls{mrfss} sampling program did not sample from 1990--1992, and so estimates during these three years were obtained as averages from neighboring years. Estimates in 1990 were the average from 1987--1989, estimates in 1991 were the average from 1987--1989 and 1993--1995, and estimates in 1992 were the average from 1993-1995. Because of the transition from \gls{mrfss} to \gls{crfs}, no estimate was available in either database in 2004 for this assessment cycle. \Gls{cdfw} provided a value of 10.59 mt for 2004 (J. Budrick, CDFW, personal communication, 3/20/23), based on an available previous pull of \gls{mrfss} data for the gopher rockfish assessment. Estimates for California recreational removals from the previous canary rockfish assessment were not used because these appeared to only reflect estimates of landings, and not total removals.
 
 ###### CA Recreational Data Impacted by COVID-19 Pandemic
 
-The COVID-19 pandemic impacted recreational port sampling in 2020 and 2021. No sampling occurred at all in April-June, 2020. \Gls{cdfw} provided proxy values for these months (M. Parker, CDFW, personal communication). Total proxy values for canary rockfish weight were summed across district and added to the existing estimate for 2020 from \gls{recfin}. In addition, California recreational total mortality estimates in the "rockfish genus" were inflated due to \gls{crfs} samplers being unable to closely examine catch and identify catch to species. This was a problem for both PR and PC modes in 2020 and primarily for the PC mode in 2021. An effort was made to allocate some of the rockfish genus mortality to other rockfish species for these modes and years (J. Coates, CDFW, personal communication). An expected value of rockfish genus mortality in 2020 and 2021 was generated by mode and year according to the average proportion to the total rockfish mortality that this category represented in 2018 and 2019, when regulations were consistent with 2020 and 2021. Mortality above this expected value was attributed to the other species also based on proportions each species represented to the total from 2018 and 2019. Calculations were made by year, mode, and district. The shore-based modes were grouped in with the PR mode. Calculations were initially made in numbers of fish because rockfish genus mortality is only recorded this way. Numbers of fish by species were then converted to weight in kilograms based on average weights of fish recorded by the \gls{crfs} program by district in 2019. Total reallocated values for canary rockfish weight were summed across modes and districts and added to existing estimates for 2020 from the sum of the proxy value and \gls{recfin} and for 2021 from \gls{recfin}.
+The COVID-19 pandemic impacted recreational port sampling in 2020 and 2021. No sampling occurred at all in April-June, 2020. \Gls{cdfw} provided proxy values for these months (M. Parker, CDFW, personal communication, 2/22/23). Total proxy values for canary rockfish weight were summed across district and added to the existing estimate for 2020 from \gls{recfin}. In addition, California recreational total mortality estimates in the "rockfish genus" were inflated due to \gls{crfs} samplers being unable to closely examine catch and identify catch to species. This was a problem for both PR and PC modes in 2020 and primarily for the PC mode in 2021. An effort was made to allocate some of the rockfish genus mortality to other rockfish species for these modes and years (J. Coates, CDFW, personal communication, 2/10/23). An expected value of rockfish genus mortality in 2020 and 2021 was generated by mode and year according to the average proportion to the total rockfish mortality that this category represented in 2018 and 2019, when regulations were consistent with 2020 and 2021. Mortality above this expected value was attributed to the other species also based on proportions each species represented to the total from 2018 and 2019. Calculations were made by year, mode, and district. The shore-based modes were grouped in with the PR mode. Calculations were initially made in numbers of fish because rockfish genus mortality is only recorded this way. Numbers of fish by species were then converted to weight in kilograms based on average weights of fish recorded by the \gls{crfs} program by district in 2019. Total reallocated values for canary rockfish weight were summed across modes and districts and added to existing estimates for 2020 from the sum of the proxy value and \gls{recfin} and for 2021 from \gls{recfin}.
 
 #### At Sea Hake Fishery
 
-Catches of canary rockfish are monitored aboard the vessel by observers in the \gls{ashop} and were provided for the years 1975--2022 (V. Tuttle, NOAA, personal communication, 4/3/2023). Observers use a spatial sample design, based on weight, to randomly choose a portion of the haul to sample for species composition. For the last decade, this is typically 30-50% of the total weight. The total weight of the sample is determined by all catch passing over a flow scale. All species other than hake are removed and weighed, by species, on a motion compensated flatbed scale. Observers record the weights of all non-hake species. Non-hake species total weights are expanded in the database by using the proportion of the haul sampled to the total weight of the haul. The catches of non-hake species in unsampled hauls is determined using bycatch rates determined from sampled hauls. Since 2001, more than 97% of the hauls have been observed and sampled.
+Catches of canary rockfish are monitored aboard the vessel by observers in the \gls{ashop} and were provided for the years 1975--2022 (V. Tuttle, NOAA, personal communication, 4/3/23). Observers use a spatial sample design, based on weight, to randomly choose a portion of the haul to sample for species composition. For the last decade, this is typically 30-50% of the total weight. The total weight of the sample is determined by all catch passing over a flow scale. All species other than hake are removed and weighed, by species, on a motion compensated flatbed scale. Observers record the weights of all non-hake species. Non-hake species total weights are expanded in the database by using the proportion of the haul sampled to the total weight of the haul. The catches of non-hake species in unsampled hauls is determined using bycatch rates determined from sampled hauls. Since 2001, more than 97% of the hauls have been observed and sampled.
 
 Values for percentages of the catch that was retained were available in the data starting in 1997. When not provided, we assumed percent retained was 100 percent. We also assumed that any canary rockfish not retained were dead. We therefore added the retained landings to the amount of landings not retained to obtained total removals.
 

documents/pre-Star/21f-comp.Rmd

@@ -58,7 +58,7 @@ Between 1987-1989 and 1993-1998 there were recreational length data for the CPFV
 
 Biological length data were aggregated across party/charter and private/rental mode based on similarity in the overall length distributions, and consistent with past canary rockfish assessments. There was a mix of biological length and age samples with and without sex information. The majority of length samples for Washington and Oregon, and all length samples for California did not include information on sex. The majority of age samples included information on sex, so were modeled as sex-specific age distributions. Only lengths from retained fish were used and only samples from ocean areas. In addition, \gls{odfw} provided \gls{mrfss} length samples with the addition of a column that flagged length values imputed from weights to allow for selection of directly measured values. A sensitivty to using lengths from both retained and released fish was explored and described in Section \@ref(sensitivities). Altogether there were very limited number of samples from inland/estuary areas and from which direct measurements were not available so no sensitivity of inclusion of these samples was done.    
 
-For each fleet, the raw observations were aggregated within a year. A catch-weighted approach was explored for Oregon and Washington observations from \gls{recfin} because these states have a formalized appraoch (E.J. Dick, personal communication, \gls{swfsc}) and a separate data table exists within \gls{recfin} (J. Edwards, personal communication, \gls{psmfc}) that includes the weightings. The catch-weighted distributions were very similar to the disributions based on raw observations, so raw distributions were used for simplicity.
+For each fleet, the raw observations were aggregated within a year. A catch-weighted approach was explored for Oregon and Washington observations from \gls{recfin} because these states have a formalized appraoch (E.J. Dick, \gls{swfsc}, personal communication, 3/28/23) and a separate data table exists within \gls{recfin} (J. Edwards, \gls{psmfc}, personal communication, 3/30/23) that includes the weightings. The catch-weighted distributions were very similar to the distributions based on raw observations, so raw distributions were used for simplicity.
 
 The approach to determine the input sample sizes for the recreational length and age data varied by data source and data type. Some data sources had unique trip numbers within the data such as the Deb Wilson-Vandenberg data and the Oregon CPFV releases. Other data sources that lacked a clear trip identifier used combinations of multiple fields to attempt to estimate unique combinations that represented the number of trips sampled. In general the number of trips for both age and length composition data was estimated based on a mix of time, location, and fishery type. The number of trips for length and age compositions from Washington sport biodata was estimated using a combination of fish sample date (or sample date if blank), fish (or sample) punch card area, and fishing mode. The number of trips for length compositions from \gls{orbs} sampling was estimated using a field (ANGLER_ID) that encompasses time, area, and fishing mode. The number of trips for age composition from \gls{orbs} sampling was estimated using sampling date, sampling site (PORT_NAME), and fishing mode. The number of trips for length compositions from \gls{mrfss} sampling was estimated using year and id code, \gls{mrfss} area code and sampling location (INTSITE for California or ORBSport for Oregon), and fishery mode. The number of trips for length compositions from \gls{crfs} sampling was estimated using sampling date, sampling site (COUNTY_NUMBER), and fishing area and mode. Collectively, the estimates for the number of trips are meant to represent a reasonable starting point that generally reflects the degree of similarity of information from sampling a given number of likely similar fish within any sampling events.
 

documents/pre-Star/22biology.Rmd

@@ -32,15 +32,14 @@ fishery age data well, and was a better biological explanation for the
 disappearance of older age females than the alternative hypothesis that
 the older females were not vulnerable to the fishery. The pattern of
 male-skewed sex ratio at older ages has been observed along the U.S.
-West Coast in fishery data off Oregon (Rasmussen pers. comm.
-\gls{odfw}), Canada [@dfo2023canary], and Alaska and in black rockfish
+West Coast in fishery data off Oregon (L. Rasmuson, \gls{odfw}), personal communication, 1/23/23), Canada [@dfo2023canary], and Alaska and in black rockfish
 in addition to canary and yellowtail rockfishes. Although it has been
 found in these three semi-pelagic species, the pattern has not been
 observed for bocaccio (*Sebastes paucispinis*), chilipepper rockfish
 (*Sebastes goodei*), shortbelly rockfish (*Sebastes jordani*), or dusky
 rockfishes (*Sebastes ciliatus*, in Alaska), which are also semi-pelagic.
 However, starting in the 2010's the male-skewed sex ratio has evened out
-in fishery data off Oregon (Rasmussen pers. comm. ODFW). The pattern has
+in fishery data off Oregon (L. Rasmuson \gls{odfw}), personal communication, 1/23/23). The pattern has
 remained within the \gls{s-wcgbt} and has not evened off in recent
 years.
 
@@ -115,8 +114,8 @@ Fecundity was fixed at the parameter values from the meta-analysis in
 @dick_meta-analysis_2017. Existing species specific fecundity
 information for canary rockfish is limited and only 9 samples of canary
 rockfish exist within the rockfish fecundity database housed at the
-\gls{sefsc} (S. Beyer, personal communication, May 22, 2023, University
-of Washington). @dick_meta-analysis_2017 performed their analysis on
+\gls{sefsc} (S. Beyer, University
+of Washington, personal communication, 5/22/23). @dick_meta-analysis_2017 performed their analysis on
 subgenera with at least three species, which excluded canary rockfish,
 so the relationship in @dick_meta-analysis_2017 for general *Sebastes*
 species was used for this assessment. Fecundity was assumed as a power
@@ -250,7 +249,7 @@ a positive bias relative to CAPS break-and-burn reads [+3.76 yrs. at age
 A qualitative examination of the new break and burn double reads showed
 no evidence that the ageing estimation had changed over time relative to
 the samples included in the previous analyses. Furthermore, a new
-TMB-based ageing error software has been developed (A. Punt, University of Washington, personal communication)
+TMB-based ageing error software has been developed (A. Punt, University of Washington, personal communication, 2023)
 but has not yet been fully explored or documented. Therefore, revision
 of the ageing error matrices has been left as a research project to be
 completed in time for the next full assessment.

documents/pre-Star/31summary.Rmd

@@ -70,7 +70,7 @@ The 2015 STAR panel and SSC did not list any major deficiencies for the previous
 
 **Recommendation**: Basic life history research may help to resolve assessment uncertainties regarding appropriate values for natural mortality and steepness, and how to best account for the apparent loss of older females in the population.
 
-**Response**: This assessment includes fecundity and maturity curves based on new data made available since the last benchmark assessment. While efforts have been undertaken to understand the apparent loss of older females, this has not resulted in any clear resolution. A research study sampling in non-trawlable habitat along the U.S. West Coast found a similar lack of old females in the data [@Brooks_thesis_2021]. Examination of data from the Oregon trawl fleet indicates that sex ratios have been more constant across ages in the last decade than they had been previously (L. Rasmuson, pers. comm.).
+**Response**: This assessment includes fecundity and maturity curves based on new data made available since the last benchmark assessment. While efforts have been undertaken to understand the apparent loss of older females, this has not resulted in any clear resolution. A research study sampling in non-trawlable habitat along the U.S. West Coast found a similar lack of old females in the data [@Brooks_thesis_2021]. Examination of data from the Oregon trawl fleet indicates that sex ratios have been more constant across ages in the last decade than they had been previously (L. Rasmuson, \gls{odfw}), personal communication, 1/23/23).
 
 ## Response to Groundfish Subcommittee Requests