Disclaimer: All data summaries and exploration presented here are preliminary and may not be indicative of the final data that will be incorporated in the 2025 assessment models

Overview

The document provides a description of data available at the time of the workshop and model specifications being considered for use in the 2025 assessment of Rougheye and Blackspotted rockfishes. These two species, together, are treated as a combined species complex. Rougheye and Blackspotted rockfishes are genetically distinct, but difficult to visually distinguish, thus most data are only available for the Rougheye/Blackspotted rockfish complex, not at the species level. They range from northern California up to and throughout Alaska. They both greatly overlap in latitude and depth, and are generally considered slope rockfish, with an otogentic shift from shallower to deeper, and adults commonly found at 360 m. Rougheye seems to be proportionally more abundant when survey samples are genetically identified, and Blackspotted tend to be found, on average, deeper than Rougheye. They can school and may segregate by size and age. While we treat these species as one assessed stock from this point forward, we recognize and are mindful of the above distinctions as we conduct our analyses.

Summary of the 2013 Assessment

The last assessment of the Rougheye and Blackspotted rockfishes in U.S. West coast waters was conducted in 2013. The estimated stock status for the beginning of 2013 was 47% of the unfished stock size, and the stock size never showed a decline below the target biomass of 40% of the unfished stock size over the period of time assessed (Figures SO_2013 and RSS_2013). This assessment was informed by both length and age compositions, mean weights, and several indices of abundance. Despite the number of different data types, the large amount of ageing error and mostly flat and largely uncertain indices of abundance resulted in wide uncertainty intervals around the derived outputs like stock size and status. The complete 2013 assessment document can be found here.

Bridging the assessment model from Stock Synthesis 3.24 to 3.30

More than 10 years have passed from the last assessment and in that time, the model and the Stock Synthesis 3 (SS3) modelling framework has undergone many changes. While the specific changes in the model can be found in the model change log, here we simply update the model from the older 3.24O version to the newer 3.30.22.1 version. We want to ensure that any update to the newest SS3 model software is not a cause of any changes in model outputs when we hold all data and model specifications to be exactly the same as in 2013. We therefore transferred all the older data and model specifications to the newest version of SS3 and compared the outputs. The status (Figure SO_2013) and scale (Figure RSS_2013) of both models are exactly the same, as are the estimates of within model uncertainty. This allows us to conclude that we can move forward using the latest version of SS3 without concern of inheriting any model difference due solely to the choice of the SS3 version.


Figure SO_2013. Estimates of relative stock size (current spawning output/unfished spawning output) for the Rougheye/Blackspotted rockfish complex in U.S. west coast waters from the 2013 assessment, and compared to the using the same data in the newest version of SS3 (3.330.22.1). Shading denotes 95% confidence intervals.
Figure SO_2013. Estimates of relative stock size (current spawning output/unfished spawning output) for the Rougheye/Blackspotted rockfish complex in U.S. west coast waters from the 2013 assessment, and compared to the using the same data in the newest version of SS3 (3.330.22.1). Shading denotes 95% confidence intervals.
Figure RSS_2013. Estimates of spawning biomass for the Rougheye/Blackspotted rockfish complex in U.S. west coast waters from the 2013 assessment, and compared to the same data in the newest version of SS3 (3.330.22.1). Shading denotes 95% confidence intervals.
Figure RSS_2013. Estimates of spawning biomass for the Rougheye/Blackspotted rockfish complex in U.S. west coast waters from the 2013 assessment, and compared to the same data in the newest version of SS3 (3.330.22.1). Shading denotes 95% confidence intervals.


Unresolved Questions and Issues from the 2013 Assessment

  1. Historical landings and discards
  2. Natural mortality
  3. Maturity and fecundity
  4. Age data and error
  5. Understanding the stock structure and biology of Rougheye and Blackspotted rockfishes

Stock Structure

There are at least two questions to think about when considering stock structure for Rougheye and Blackspotted rockfishes.

  1. Rougheye and Blackspotted rockfishes are two different species– can we separate them as two stocks and conduct separate assessments? These two species remain difficult to differentiate visually in the catch, thus remain reported as a species complex. Frey et al. (in prep.) provided insight into the ability of the Northwest Fisheries Science Center West Coast Groundfish Bottom Trawl Survey biologists to identify the two species, with 90% of genetically identified Rougheye rockfish being correctly identified in the field. When mis-identifications occured, it was usually a Blackspotted rockfish being mis-identified as a Rougheye rockfish. There were few mis-identifications when a fish was identified as a Blackspotted rockfish. While this is promising for potential future species-specific data coming from the survey, it does not alleviate the historical problem of separating fishery data into the two species. Frey et al. (in prep.) therefore also considered whether ecological factors like depth or latitude could help separate samples by species. They found that both species occur within the range of this assessment’s considered areas (California to Washington), and heavily spatially overlap. Interestingly, there seem to be relative hot spots for these species where one species is more common than the other, and in general, Rougheye rockfish seems to be more common than Blackspotted rockfish (however, Blackspotted rockfish may be the more common of the two in parts of Alaska). Overall, there seems to be little ability to separate current or historical fishery data reliably in order to separate these two species into two stocks, so we will maintain a species complex approach, though given absolute presence off the U.S. West coast, this may be considered more of a Rougheye than Blackspotted stock assessment. We also note that throughout the range of these stocks, all assessments have maintained a species complex approach.
  2. Both species range into Canada and Alaska– are they one stock? While genetics studies have focused mostly on identification of the two species, little is known about the population structure of either species. This assessment and the 2013 assessment represent the most southerly range of these species. Comparing the absolute abundance of the 2013 assessment to the most current estimates of the Alaskan stocks, the absolute number in this southerly range is much smaller than in the Gulf of Alaska (GOA), but higher than the Bering Sea/Aleutian Island (BSAI) stock (Figure SO_comp). The two smaller stocks have similar trend of decline and stabilization, whereas the higher biomass GOA stock looks to have not dropped at all over the time period considered (Figure RSS_comp).
Figure SO_comp. Estimates of spawning biomass (current spawning output/unfished spawning output) for the Rougheye/Blackspotted rockfish complex from the two most recent Alaska (Bering Sea/Aleutian Islands (BSAI) and Gulf of Alaska (GOA)) and the 2013 U.S. west coast stock assessment.
Figure SO_comp. Estimates of spawning biomass (current spawning output/unfished spawning output) for the Rougheye/Blackspotted rockfish complex from the two most recent Alaska (Bering Sea/Aleutian Islands (BSAI) and Gulf of Alaska (GOA)) and the 2013 U.S. west coast stock assessment.


Figure RSS_comp. Estimates of relative stock size (current spawning output/unfished spawning output) relative to 1977 (the common year in all stock assessments compared) for the Rougheye/Blackspotted rockfish complex from the two most recent Alaska (Bering Sea/Aleutian Islands (BSAI) and Gulf of Alaska (GOA)) and the 2013 U.S. west coast stock assessment.
Figure RSS_comp. Estimates of relative stock size (current spawning output/unfished spawning output) relative to 1977 (the common year in all stock assessments compared) for the Rougheye/Blackspotted rockfish complex from the two most recent Alaska (Bering Sea/Aleutian Islands (BSAI) and Gulf of Alaska (GOA)) and the 2013 U.S. west coast stock assessment.


Sex Structure

The 2013 stock assessment used a one sex model configuration. This was based on the observation that the biology of females and males is very similar, thus justifying the simplifying assumption of one sex. The Biology section below demonstrates that females and males do generally have similar growth, though there are differences, but may have different natural mortality values. The current assessment will use a two sex configuration that allows for flexibility to set female and male parameters either equal (i.e., functionally equivalent to a one sex model) and or sex-specific. Figures Sex1vs2_SO and Sex1vs2_Bratio show that using a two sex configuration with the same life history parameters for females and males is equivalent to the one sex model. Note that the one sex model sums up both female and male biomass, thus why it is twice the size as the two sex female-only spawning output (Figure Sex1vs2_Bratio).


Figure Sex1vs2_SO. Comparison of spawning output using the 1 sex and 2 sexes set to equal values models based on the 2013 Rougheye/Blackspotted Rockfishes assessment data. The 1 sex model has double the biomass because it includes both females and males.
Figure Sex1vs2_SO. Comparison of spawning output using the 1 sex and 2 sexes set to equal values models based on the 2013 Rougheye/Blackspotted Rockfishes assessment data. The 1 sex model has double the biomass because it includes both females and males.
Figure Sex1vs2_Bratio. Comparison of spawning output using the 1 sex and 2 sexes set to equal values models based on the 2013 Rougheye/Blackspotted Rockfishes assessment data.
Figure Sex1vs2_Bratio. Comparison of spawning output using the 1 sex and 2 sexes set to equal values models based on the 2013 Rougheye/Blackspotted Rockfishes assessment data.


Fleet Structure

Defining fleets is largely based on differing fleet selectivity (i.e., how the fishery captures fish by length and/or age). In the stock assessment model, selectivity translates into how the removals are taken via length and/or age out of the population. Currently, the following fleet structure is being considered for modelling commercial fishery removals as there is no record of a recreational fishery for this stock complex:

  1. Commercial trawl
  2. Commercial fixed gear (mainly the long-line fishery)
  3. At-sea hake fishery
  4. Dead discard trawl
  5. Dead discard non-trawl

In 2013 assessment, fisheries removals were split among three fleets –trawl, hook-and-line and at-sea hake fishery bycatch. For the first two fleets (trawl and hook-and-line), removals were divided between landings and discards, with selectivity and retention curves estimated within the model.

For this assessment, we plan to treat discards in trawl and non-trawl fisheries as separate fleets from landings fleets. This approach provides several advantages, including:

  • With separate discard fleets, we can easily track relative amounts of landings and discards within a fishery (they are not being combined into the total catch).
  • This approach provides more flexibility to explore different selectivity assumptions for both landed and discarded fish –dome-shaped vs asymptotic, mirroring one to the other, etc.
  • We can easily compare how similar (or different) selection curves for retained and discarded fish (easier than in case of selectivity and retention curves estimated within a single fleet).
  • The biological data for landings and discards are collected independently (port sampling vs on-board observers), using different sampling approaches. Treating landings and discards as separate fleets in the model allows us to weight these data separately as well, to balance the representation of samples.

The change in treating discards as separate fleets does not impact model results (Figures Discard_1 and Discard_2), regardless of the selectivity form being assumed for the discard fleets.

We plan on using a length-based selectivity curve for all fleets for the current stock assessment model (same as was done in the 2013 assessment), as there is no reason to believe significant age-based selectivity is occurring. We will consider both logistic and dome-shaped selectivity options.


Figure Discard_1. Comparison of spawning output using retention curves or discard fleets using the 2013 Rougheye/Blackspotted Rockfishes assessment.
Figure Discard_1. Comparison of spawning output using retention curves or discard fleets using the 2013 Rougheye/Blackspotted Rockfishes assessment.
Figure Discard_2. Comparison of relative spawning output using retention curves or discard fleets using the 2013 Rougheye/Blackspotted Rockfishes assessment.
Figure Discard_2. Comparison of relative spawning output using retention curves or discard fleets using the 2013 Rougheye/Blackspotted Rockfishes assessment.


Commercial Landings & Discards

Commercial landings are the main source of removals for Rougheye/Blackspotted rockfishes. They are a mainly slope species, typically found around 365 m (~200 fathoms), but can also be common shallower (down to 275 m). This complex is known to school, thus increasing encounters when in an area and available to fishing gear. Trawl and non-trawl (mostly long-line) are the two major fleets they are taken in, with the at-sea-hake midwater fishery also encountering them in notable numbers over the past two decades. Landings are lower in California (Figure Ct_CA) and increase northward, with the biggest trawl catches in Oregon (Figure Ct_OR), and biggest non-trawl catches in Washington state (Figure Ct_WA). There was also a notable catch of Rougheye/Blackspotted accounted for in the trawl fishery that came from the foreign fleet Pacific Ocean Perch fishery in the 1960s.

There are some updates in landings by state and fishery, but are overall small (Figure Ct_All). One notable change is that non-trawl now includes all non-trawl gears, not just hook-and-line as was used in the 2013 assessment. Landings over the last couple of decades are generally lower than in the preceding decades, although we note that more discards are accounted for in recent years (Figure Ct_fleets). Updating the 2013 stock assessment with these newest catches leads to a small increase in overall spawning stock output (FIgure Figure Ct_compsSO), and essentially no change in the relative stock status (FIgure Figure Ct_compsRSS).

Figure Ct_CA. California state landings for non-trawl and trawl fisheries compared between the 2013 assessment and updated landings for the 2025 stock assessment model.
Figure Ct_CA. California state landings for non-trawl and trawl fisheries compared between the 2013 assessment and updated landings for the 2025 stock assessment model.


Figure Ct_OR. Oregon state landings for non-trawl and trawl fisheries compared between the 2013 assessment and updated landings for the 2025 stock assessment model.
Figure Ct_OR. Oregon state landings for non-trawl and trawl fisheries compared between the 2013 assessment and updated landings for the 2025 stock assessment model.


Figure Ct_WA. Washington state landings for non-trawl and trawl fisheries compared between the 2013 assessment and updated landings for the 2025 stock assessment model.
Figure Ct_WA. Washington state landings for non-trawl and trawl fisheries compared between the 2013 assessment and updated landings for the 2025 stock assessment model.


Figure Ct_All. Landings across all states for non-trawl and trawl fisheries compared between the 2013 assessment and updated landings for the 2025 stock assessment model.
Figure Ct_All. Landings across all states for non-trawl and trawl fisheries compared between the 2013 assessment and updated landings for the 2025 stock assessment model.


Figure Ct_fleets. Landings across all fleets used in the 2025 stock assessment.
Figure Ct_fleets. Landings across all fleets used in the 2025 stock assessment.


Figure Ct_compsSO. Comparison of spawning output using updated catches vs using catches from the 2013 Rougheye/Blackspotted Rockfishes assessment.
Figure Ct_compsSO. Comparison of spawning output using updated catches vs using catches from the 2013 Rougheye/Blackspotted Rockfishes assessment.


Figure Ct_compsRSS. Comparison of relative spawning output using updated catches vs using catches from the 2013 Rougheye/Blackspotted Rockfishes assessment.
Figure Ct_compsRSS. Comparison of relative spawning output using updated catches vs using catches from the 2013 Rougheye/Blackspotted Rockfishes assessment.


Additional Items for Discussion

  • What to use for at-sea-hake 2024 value?
  • What to use for trawl and non-trawl discard 2024 value?


Indices of Abundance

Given Rougheye/Blackspotted are associated with deep, structured habitats, it can be difficult to survey them with trawl gear. Four general fishery-independent bottom trawl surveys were used in the 2013 assessment, and are again included in this assessment:

  • Triennial (every three years) survey (1980-2004; Figure Index_Tri)
  • Alaska Fishery Science Center Slope survey (1997-2001; Figure Index_AFSC_slope)
  • Northwest Fisheries Science Center Slope Center (1999-2001; Figure Index_NWFSC_slope)
  • West Coast Groundfish Bottom Trawl Survey (WCGBTS; 2003-2024; Figure Index_WCGBTS)

Only the WCGBTS has new data for this assessment, but new methods (spatial and spatiotemporal GLMMs with TMB or sdmTMB) to develop an index of abundance were applied to all surveys to update all indices. Two distributions (gamma and lognormal) were considered, as was the case in 2013 model when a non-spatial generalized linear mixed model was used to develop indices of abundance.

Comparing the standardized versions (i.e., Z-scores, which puts all the indices on the same scale for better comparison of trends) of the indices shows very similar trends among each model output (Figure Z_score_Indices), suggesting little difference in choice of model type. The variance in the indices is generally high (0.3-0.5; Figure Sd_log_Indices), suggesting the information content in these indices is low. This is not a surprise given the challenge of sampling these species with trawl gear. Overall, catches densities are highest in northern Oregon and Washington (Figure RE_BS_density).


Figure Index_Tri. Comparison of index values for the Triennial Trawl Survey using the 2013 GLMM approach and the 2025 sdmTMB approaches. Error bars are 95% confidence intervals.
Figure Index_Tri. Comparison of index values for the Triennial Trawl Survey using the 2013 GLMM approach and the 2025 sdmTMB approaches. Error bars are 95% confidence intervals.


Figure Index_AFSC_slope. Comparison of index values for the Alaska Fishery Science Center Slope Survey using the 2013 GLMM approach and the 2025 sdmTMB approaches. Error bars are 95% confidence intervals.
Figure Index_AFSC_slope. Comparison of index values for the Alaska Fishery Science Center Slope Survey using the 2013 GLMM approach and the 2025 sdmTMB approaches. Error bars are 95% confidence intervals.


Figure Index_NWFSC_slope. Comparison of index values for the NWFSC Slope Survey using the 2013 GLMM approach and the 2025 sdmTMB approaches. Error bars are 95% confidence intervals.
Figure Index_NWFSC_slope. Comparison of index values for the NWFSC Slope Survey using the 2013 GLMM approach and the 2025 sdmTMB approaches. Error bars are 95% confidence intervals.


Figure Index_WCGBTS. Comparison of index values for the West Coast Groundfish Bottom Trawl Survey using the 2013 GLMM approach and the 2025 sdmTMB approaches. Error bars are 95% confidence intervals.
Figure Index_WCGBTS. Comparison of index values for the West Coast Groundfish Bottom Trawl Survey using the 2013 GLMM approach and the 2025 sdmTMB approaches. Error bars are 95% confidence intervals.


Figure Index_Zscore. Comparison of model-based index Z-score values by each survey. Z-scores standardize the scale of the index so different model outputs can be compared.
Figure Index_Zscore. Comparison of model-based index Z-score values by each survey. Z-scores standardize the scale of the index so different model outputs can be compared.


Figure Index_sdlog. Comparison of model-based log-space standard deviation values by each survey. These measures of uncertainty are comparable across models and among surveys.
Figure Index_sdlog. Comparison of model-based log-space standard deviation values by each survey. These measures of uncertainty are comparable across models and among surveys.


Figure RE_BS_density. Catch-per-unit-effort of Rougheye/Blackspotted rockfishes along the U.S. west coast.
Figure RE_BS_density. Catch-per-unit-effort of Rougheye/Blackspotted rockfishes along the U.S. west coast.


Details on the WCBGTS index: the data were truncated to depths less than 875 m prior to modelling given that there were zero positive encounters in depths deeper than 875 m. The prediction grid was also truncated to only include available survey locations in depths between 55-875 m to limit extrapolating beyond the data and edge effects.

The response variable in the model was catch (mt) with an offset of area (km2) to account for differences in effort. Fixed effects were estimated for each year. The following additional covariate was included: pass.

Spatial variation, but not spatiotemporal variation, was included in the encounter probability and the positive catch rate model. Spatial variation was approximated using 200 knots, where more knots led to non-estimable standard errors.

Model bridging with abundance indices

There are two different ways of considering how the new indices affect the model relative to the 2013 stock assessment while keeping all other data the same as in 2013. One is to use the same four indices with the new approach to creating the abundance indices (as shown in the above section). The second is to extend the WCGBTS index to the most recent available year (as it is the only one that can be extended). The comparisons are shown in Figure Index_comp_SO and Figure Index_comp_RSS and demonstrate that the updated and extended versions of the indices do not make significant changes in either the scale or status.


Figure Index_comp_SO. Comparison of spawning output among abundance index treatments from the 2013 and current Rougheye/Blackspotted Rockfishes assessments.
Figure Index_comp_SO. Comparison of spawning output among abundance index treatments from the 2013 and current Rougheye/Blackspotted Rockfishes assessments.


Figure Index_comp_RSS. Comparison of relative spawning output among abundance index treatments from the 2013 and current Rougheye/Blackspotted Rockfishes assessments.
Figure Index_comp_RSS. Comparison of relative spawning output among abundance index treatments from the 2013 and current Rougheye/Blackspotted Rockfishes assessments.


Additional Items for Discussion

  • Are there any other indices we should be considering?


Composition Data

Commercial length compositions

Commercial length data are available for all fleets, which will allow for the estimation of selectivity (Figures Lt_trawl, Lt_Lt_trawl_discard, Lt_nontrawl, Lt_nontrawl_discard, and Lt_ashop). Data go back into the 1990s for the trawl and non-trawl landed fisheries, and start in the 2000s for the at-sea-hake fishery. Dead discard fishery length data begin in the 2010s. Aggregated length compositions (Figure Lt_agg) offers a look at how the composition compare in sampling the population. The trawl fishery gets smaller unsexed fish relative to the non-trawl and at-sea-hake fishery. The trawl dead discards, all unsexed, also show much smaller fish. The dead discards from the non-trawl fishery look to include slightly larger fish than the landings. Comparing the common years in the 2013 to data for the current assessment shows very little to no difference (Figures Lt_trawl_compare, Lt_Lt_trawl_discard_compare, Lt_nontrawl_compare, Lt_nontrawl_discard_compare, and Lt_ashop_compare). One notable difference is that there are more years available for the discard data in the 2013 stock assessment than are currently available. After the 2013 assessments, the West Coast Groundfish Observer Program re-evaluted their data and found that some of the earlier years were not deemed to be sufficiently randomly sampled and were subsequently removed. This explains the lack of those early years’ length data for discarded trawl and non-trawl fisheries.

The 2013 assessment considered blocking of selectivity as well as dome-shaped selectivity for all fleets and surveys. Dome-shaped selectivity is an appropriate choice given that natural refugia in the habitats of Rougheye and Blackspotted rockfishes (based on depth and rugged habitat preference) are difficult to access and sample with the two major gear types and may explain why older, larger individuals are less present in the data. We note, though, that large and old individuals are encountered in both fisheries and surveys, so there is some selection occurring for those sizes and ages.

Figure Lt_trawl. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the trawl fishery. Red is female; blue is male.
Figure Lt_trawl. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the trawl fishery. Red is female; blue is male.


Figure Lt_trawl_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the trawl survey.
Figure Lt_trawl_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the trawl survey.


Figure Lt_trawl_discard. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on dead discarded fish in the trawl fishery.
Figure Lt_trawl_discard. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on dead discarded fish in the trawl fishery.


Figure Lt_trawl_discard_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on dead discarded fish sampled in the trawl survey.
Figure Lt_trawl_discard_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on dead discarded fish sampled in the trawl survey.


Figure Lt_nontrawl. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the non-trawl fishery. Red is female; blue is male.
Figure Lt_nontrawl. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the non-trawl fishery. Red is female; blue is male.


Figure Lt_nontrawl_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the non-trawl survey.
Figure Lt_nontrawl_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the non-trawl survey.


Figure Lt_nontrawl_discard. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on dead discarded fish in the non-trawl fishery.
Figure Lt_nontrawl_discard. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on dead discarded fish in the non-trawl fishery.


Figure Lt_nontrawl_discard_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on dead discarded fish sampled in the non-trawl survey.
Figure Lt_nontrawl_discard_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on dead discarded fish sampled in the non-trawl survey.


Figure Lt_ashop. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the at-sea hake fishery. Red is female; blue is male.
Figure Lt_ashop. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the at-sea hake fishery. Red is female; blue is male.


Figure Lt_ashop_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the at-sea hake fishery.
Figure Lt_ashop_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the at-sea hake fishery.


Figure Lt_agg. Aggregated length composition data over all years by year, sex, and fleet for the Rougheye/Blackspotted rockfish complex.
Figure Lt_agg. Aggregated length composition data over all years by year, sex, and fleet for the Rougheye/Blackspotted rockfish complex.


Survey length compositions

Three of the four surveys have length composition data that allow for estimation of selectivity (Figures Lt_tri, Lt_AKslope, Lt_WCGBTS). The NWFSC slope survey shares (or mirrors) the estimated selectivity from the AFSC slope survey. Looking at the aggregate compositions (Figure Lt_agg), the WCGBTS clearly samples the largest range in lengths, including the smaller individuals. The Triennial and WCGBTS and the trawl discard, all show an interesting double mode, with a common valley in lengths around 35-40 cm. Length composition in the current assessment are very similar to those used in the 2013 assessment for shared years (Figure Lt_WCGBTS_compare).


Figure Lt_tri. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the Triennial bottom trawl survey. Red is female; blue is male.
Figure Lt_tri. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the Triennial bottom trawl survey. Red is female; blue is male.


Figure Lt_tri_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the Triennial bottom trawl survey.
Figure Lt_tri_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the Triennial bottom trawl survey.


Figure Lt_AKslope. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the Alaska Fishery Science Center slope survey. Red is female; blue is male.
Figure Lt_AKslope. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the Alaska Fishery Science Center slope survey. Red is female; blue is male.


Figure Lt_AKslope_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the Alaska Fishery Science Center slope survey.
Figure Lt_AKslope_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the Alaska Fishery Science Center slope survey.


Figure Lt_WCGBTS. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the West Coast Groundfish Bottom Trawl Survey. Red is female; blue is male.
Figure Lt_WCGBTS. Length composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the West Coast Groundfish Bottom Trawl Survey. Red is female; blue is male.


Figure Lt_WCGBTS_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the West Coast Groundfish Bottom Trawl Survey.
Figure Lt_WCGBTS_compare. Length composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the West Coast Groundfish Bottom Trawl Survey.


Effective Sample Sizes

Effective input sample sizes for lengths are an indicator of how much weight the model will put on fitting the length composition data. A higher effective sample size, the more the model will attempt to fit that data. The effective sample size is not usually the number of fish that are sampled, but a measure of independent samples. This means that the fish caught are random samples from the population. When fish are caught together, they could be of similar size or age, and therefore not necessarily a random sample of the population. Some combination of number of fish and number of independent sampling units like hauls or sets are often used.

For the 2025 stock assessment, we use the equation based on the number of trips and fish to calculate the effective sample size by year, fleet and sex for commercial trawl and non-trawl fisheries. These equations are currently the standard ones used across many groundfish stock assessments.


If Nfish/Ntrips is < 44: \(N_{eff} = N_{trips} + 0.138 * N_{fish}\)

If Nfish/Ntrips is ≥ 44: \(N_{eff} = 7.06 * N_{trips}\)


The at-sea-Hake fishery samples hauls instead of trips, so there is not an equation devised for it. For this reason, it has been determined that hauls are the best available metric to represent independent sampling units of fish.

For surveys, unique tows are recorded and multiplied by 2.43 to calculate the input effective sample size. This equation is commonly used for West Coast slope groundfish and based on the work of Stewart and Hamel (2014).

Input effective sample sizes for each fishing (Figure Neff_FD) and survey (Figure Neff_FI) fleet are given below. Higher effective sample sizes tend to be in the most recent years, and most years are below or around 100. Discard fishery samples are unsexed, while all other sources of lengths are largely sexed. Sample sizes for common years between the 2013 and current assessment are typically higher for the current stock assessment.


Figure Neff_FD. Input effective sample sizes by year and fishing fleet. Shape color indicates whether samples were unsexed or sexed. Bars show the input samples sizes for the 2025 assessment; blue dots are the 2013 input sample sizes.
Figure Neff_FD. Input effective sample sizes by year and fishing fleet. Shape color indicates whether samples were unsexed or sexed. Bars show the input samples sizes for the 2025 assessment; blue dots are the 2013 input sample sizes.


Figure Neff_FI. Input effective sample sizes by year and survey. All survey samples were sexed. Bars show the input samples sizes for the 2025 assessment; blue dots are the 2013 input sample sizes.
Figure Neff_FI. Input effective sample sizes by year and survey. All survey samples were sexed. Bars show the input samples sizes for the 2025 assessment; blue dots are the 2013 input sample sizes.


Age compositions

Age composition data can provide very valuable insights into growth, mortality, and recruitment. Additionally, it may offer improved resolution over lengths in stock status, if the ages are randomly sampled and include the oldest individuals in the population.

Age compositions are included in the stock assessment as conditioned on lengths (conditional age-at-length). This means that there is a composition of ages associated for a given length bin instead of an age composition across age bins (which is termed a marginal age composition, similar to how length compositions are normally set up across length bins). Conditioning the ages on lengths allows for the use of length and age data that come from the same individual, preserving the important assumption of independence in lengths and age samples. It also provides an explicit way to model growth.

Age samples are available from four sources (3 fishery-dependent and 1 fishery-independent):

  1. Trawl fishery
  2. Non-trawl fishery
  3. At-sea-hake fishery
  4. WCGBTS

Age (x-axis) by length bins (y-axis) samples for each data source are shown in Figures AgeC_trawlnon, AgeC_nontrawl, AgeC_ashop, and AgeC_wcgbts. Comparison of age composition data between the 2013 and current stock assessment are shown in Figures Age_trawl_compare, Age_ashop_compare, and Age_WCGBTS_compare for the WCGBTS samples. We expect these age sample to change in the next couple months as more ages become available, but the below comparisons are what are currently available.


Figure AgeC_trawlnon. Age composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the trawl (left panel) and non-trawl (right panel) fishery. Bubble size indicates composition.


Figure AgeC_ashop. Age composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the at-sea hake fishery. Bubble size indicates composition.


Figure Age_trawl_compare. Age composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the trawl fishery.
Figure Age_trawl_compare. Age composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the trawl fishery.


Figure Age_ashop_compare. Age composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the at-sea-hake fishery.
Figure Age_ashop_compare. Age composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the at-sea-hake fishery.


Figure AgeC_wcgbts. Age composition data by sex and year for the Rougheye/Blackspotted rockfish complex based on fish sampled in the WCGBTS. Bubble size indicates composition.


Figure Age_WCGBTS_compare. Age composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the West Coast Groundfish Bottom Trawl Survey.
Figure Age_WCGBTS_compare. Age composition data by year between the 2013 and 2025 assessments for the Rougheye/Blackspotted rockfish complex based on fish sampled in the West Coast Groundfish Bottom Trawl Survey.


Ageing error

Fish do not directly tell us their age (and would we believe them anyway?), so an ageing structure, something that records the age of a fish, is needed to allow us to measure how long a fish has lived. Otoliths and other ageing structures are used, but are not always easy to read the age, leading to uncertainty in the age. In order to measure this uncertainty, one reader can read an ageing structure multiple times or different readers may age them. The difference in readers is also important when understanding how to interpret ages from different labs or individuals. This ageing error matrices offers a way to help interpret the uncertainty in ages that are used in the stock assessment.

We calculate the ageing error, exploring bias and imprecision, for our age samples, by applying the ageing error code found in the pfmc-assessment GitHub repository. We do not currently have all of the age reading comparisons needed to calculate these matrices, so this analysis will come later in the process of preparing data for the assessment.


Additional Items for Discussion

  • Why do we see the multi-modal length distribution in the two surveys and one discard fleet, especially given it is at a similar size?


Biology

Natural Mortality

Natural mortality is a highly influential parameter in age-structured stock assessments. It defines the rate of natural death by age, and thus establishes a stable age-structure and expectation of longevity, and interacts with growth and reproduction to determine stock productivity. It is a very difficult parameter to directly measure, thus empirical relationships based on life history parameters are often used to indirectly determine its value or build prior distributions in belief of what it is in the event we do attempt to estimate it in the model (Cope and Hamel 2022; Maunder et al. 2023). If length and age data are available, it may be possible to estimate it in the model.

An estimate of maximum age tends to be the most reliable life history parameter related to natural mortality to inform its estimation. Cope and Hamel (2022) provide the most up-to-date examination of the relationship between maximum age and natural mortality, with a M = 5.4/tmax (using the Natural Mortality Tool), and this is the equation typically used to estimate a natural mortality point estimate, but is underpinned by the choice of the value of tmax. This equation assumes that the proportion of the stable population at this maximum age is 0.4517%. If we take humans as an example, the longest lived human is 122 years. This is not the maximum age, but the oldest ever recorded age. The maximum age that corresponds to 0.4517% of the population is around 100 years. For Rougheye/Blackspotted, the oldest ever aged individual is 205 years with unknown ageing error. We did not consider this as a realistic maximum age.

The 2013 U.S. west coast stock assessment used a prior built around a mean of 0.034 (corresponding to a maximum age of 163), but estimated natural mortality at 0.042 (maximum age between 128-129 years; Figure M). The 2023 Gulf of Alaska assessment built a prior conditional on a estimate of natural mortality from their 5 oldest aged individuals that ranged from 126-135 years. This resulted in a mean value of 0.042, similar to the 2013 U.S. west coast stock assessment. The 2023 Bering Sea/Aleutian Islands assessment used M = 0.05 (assumed longevity of 108), and the recent Canadian assessments considered a range of M values from 0.03 to 0.055 (assumed maximum ages of 180 to 98 years; Figure M).

We will attempt to estimate natural mortality, as was done in the 2013 U.S. West coast assessment. Examining the available age data, the oldest 10 individuals range from 139 to 165 and were all males. For females, the 10 oldest individuals range from 130 to 121 years. If those oldest ages were used in the Hamel and Cope (2022) longevity estimator, these ages would correspond to a range of natural mortality values of 0.033 to 0.039 for males, which include the mean of the prior used in the 2013 assessment. For females, it corresponds to natural mortality values of 0.039 to 0.045. All these assume that the sampled population has enough of an age structure still available for sampling, as opposed to having some level of age truncation from the theoretical unfished stable age distribution.

Related to this issue of possible age truncation, applying a catch curve analysis (taking the log of the abundance of numbers of samples in available age classes) on the aggregated ages across all age sources by sex, the total mortality (Natural + Fishing mortality= Total mortality) is 0.046 for females and 0.035 for males, which may indicate the natural mortality could be lower than that used in the 2013 assessment, but within the range of values considered in other areas (Figure CC_Z). This also indicates the possibility of estimating sex-specific natural mortality, as natural mortality may differ by sex. The two sex model allows for this type of model specification exploration. Further exploration was done my truncating the upper ages considered, with the assumption that the older ages may also not be sampled fully (i.e., dome-shaped selectivity). We considered both 100 (Figure CC_Z_100) and 80 (Figure CC_Z_80) as upper age cut-offs. The less older individuals included, the higher the estimate of total mortality, and this a higher natural mortality. But we can see a general overestimate of how many older individuals are expected using these higher Z values, thus dome-shapeness does not see to explain the sampling of these older individuals.

One challenge to estimating natural mortality within the model is the interaction of estimating dome-shaped selectivity with estimating natural mortality. If all fleets assume some level of dome-shaped selectivity, it is difficult to determine if the unseen larger, older individuals are due to natural death or fishing mortality. Typically, at least one major fleet needs to achieve full selectivity for the larger, older individuals. The 2013 assessment suggested some dome-shaped selectivity in the two major fleets, thus any natural mortality estimates will be evaluated depending on the ultimate forms of fleet selectivity.


Figure M. Natural mortality curves by age in years for values of natural mortality used in various Rougheye/Blackspotted Rockfish stock assessments. Dots indicate the range of assumed maximum ages using the equation from Hamel and Cope 2022.
Figure M. Natural mortality curves by age in years for values of natural mortality used in various Rougheye/Blackspotted Rockfish stock assessments. Dots indicate the range of assumed maximum ages using the equation from Hamel and Cope 2022.


Figure CC_Z. Catch curve (log abundance by age) analysis on aggregated ages over all age sources by sex (black points). The peak selected age was 21 for both sexes, so the linear model was run from age 21 until the oldest age (red points). The slope of the linear model is equal to the estimate of an aggregate total mortality (Z).
Figure CC_Z. Catch curve (log abundance by age) analysis on aggregated ages over all age sources by sex (black points). The peak selected age was 21 for both sexes, so the linear model was run from age 21 until the oldest age (red points). The slope of the linear model is equal to the estimate of an aggregate total mortality (Z).


Figure CC_Z_100. Catch curve (log abundance by age) analysis on aggregated ages over all age sources by sex (black points). The peak selected age was 21 for both sexes with a max age of 100, so the linear model was run from age 21 until age 100 (red points). The slope of the linear model is equal to the estimate of an aggregate total mortality (Z).
Figure CC_Z_100. Catch curve (log abundance by age) analysis on aggregated ages over all age sources by sex (black points). The peak selected age was 21 for both sexes with a max age of 100, so the linear model was run from age 21 until age 100 (red points). The slope of the linear model is equal to the estimate of an aggregate total mortality (Z).


Figure CC_Z_80. Catch curve (log abundance by age) analysis on aggregated ages over all age sources by sex (black points). The peak selected age was 21 for both sexes with a max age of 80, so the linear model was run from age 21 until age 80 (red points). The slope of the linear model is equal to the estimate of an aggregate total mortality (Z).
Figure CC_Z_80. Catch curve (log abundance by age) analysis on aggregated ages over all age sources by sex (black points). The peak selected age was 21 for both sexes with a max age of 80, so the linear model was run from age 21 until age 80 (red points). The slope of the linear model is equal to the estimate of an aggregate total mortality (Z).


Age and Growth

Age and length data are used to estimate important growth parameters. Figure AL_1 has the currently available age and length data, with additional age data expected to be provided in the next month. Female and male sample sizes are very similar. Estimated growth curves are also presented in Figure AL_1 and the parameters are provided in Table AL_1. The West Coast Groundfish Bottom Trawl Survey clearly and importantly samples the smallest, youngest individuals compared to the other two data sources. This allows for a better estimate of the age at size 0 (t0) and growth coefficient (k). The female asymptotic size (L\(\infty\)) is estimated notably higher from the PacFIN data, though male estimates of Linf are similar across the data sets.

The coefficient of variation (CV) of length by age and sex are shown in Figure AL_2. This is a measure of the variation in length for a given age class. Sample sizes are highest from the youngest ages up to around 70 (females) to 80 (males) years. The smoothed line shows the average response, and indicates similar CVs values for females and males, with the highest at the youngest ages, but generally 0.1. The amount and range of age samples, along with repeated length samples within an age class, offers hope that the growth parameters (L\(\infty\), k, t0, and CVs at age) can be estimated in the model. This will require using the ages as conditioned on lengths.

We note that the growth values being estimated in our data are notably different than those used in Alaska. For instance, the growth paramters for the BSAI stock is L\(\infty\) = 51.43, k = 0.06 and t0 = -3.30 and L\(\infty\) = 54.2 cm, k = 0.07, t0= -1.5 for the GOA population (both sex combined). These growth parameters shows a larger size and faster growth of the West Coast stock complex versus those in Alaska, though the West Coast stock complex is more similar to the GOA complex.

Figure AL_1. Age and length data, with fitted von Bertalanffy growth curves, by sex and data source for the Rougheye/Blackspotted rockfish complex. Sample sizes (N) are also provided.
Figure AL_1. Age and length data, with fitted von Bertalanffy growth curves, by sex and data source for the Rougheye/Blackspotted rockfish complex. Sample sizes (N) are also provided.


Table AL_1. Estimated von Bertalanffy growth parameters (and standard deviations) from the currently available data based on different data sources.

#>        Run.name Linf.mean Linf.sd K.mean  K.sd t0.mean t0.sd
#> 1    Female_All     58.81    0.17   0.08 0.001   -1.19  0.15
#> 2  Female_WCGBT     59.07    0.50   0.08 0.003   -1.43  0.20
#> 3 Female_PacFIN     63.22    0.93   0.04 0.004  -14.86  2.30
#> 4  Female_ASHOP     60.10    0.33   0.05 0.003  -14.49  1.79
#> 5      Male_All     57.13    0.15   0.09 0.001   -1.26  0.14
#> 6    Male_WCGBT     58.03    0.43   0.08 0.002   -1.75  0.21
#> 7   Male_PacFIN     58.25    0.45   0.05 0.004  -13.14  1.81
#> 8    Male_ASHOP     58.34    0.26   0.06 0.004  -12.34  1.77
#> 9        FM_All     57.86    0.11   0.08 0.001   -1.17  0.10


Figure AL_2. Coefficient of variation by age and sex for all sources of Rougheye/Blackspotted rockfishes ages. Sample sizes (N) are also indicated by size of the point. The line is a smoothed loess (polynomial) line that gives a moving average of CV by age and sex.
Figure AL_2. Coefficient of variation by age and sex for all sources of Rougheye/Blackspotted rockfishes ages. Sample sizes (N) are also indicated by size of the point. The line is a smoothed loess (polynomial) line that gives a moving average of CV by age and sex.


Maturity and Fecundity

Maturity: We are currently updating the maturity analysis for the Rougheye/Blackspotted rockfish complex with additional samples to estimate the length at which 50% of females in the population are mature (L50). Biological maturity identifies females that are physiologically capable of spawning. Functional maturity identifies females that are physiologically capable of spawning and will likely spawn in a given year. The most recent L50 estimate (not yet updated) of biological maturity is 43.84 cm and the most recent L50 estimate (not yet updated) of functional maturity is 48.44 cm.

Fecundity: The 2013 U.S. west coast stock assessment assumed that fecundity was proportional to weight. Dick et al. (2017) provided a study on rockfishes showing that rockfishes routinely have a non-proportional relationship of fecundity to weight, with larger individuals producing more eggs than expected only by weight. Neither Rougheye or Blackspotted rockfishes have a species- of subfamily-specific estimate for this relationship, so this stock assessment uses the unobserved Genus Sebastes values of a = 6.538e-06 and b = 4.043 using the F=aL^b relationship.


Length-Weight

Female and male length-weight relationships were determined using data from the PacFIN database, West Coast Groundfish Bottom Trawl Survey, and ASHOP samples. Samples size by sex were: female (N=13839), males (13625), and unknown sex (53). Each of the data sources estimated very similar length-weight relationships (Figure LW1).


Figure LW1. Length and weight samples by sex and data source. Lines are the power function fits by data source.
Figure LW1. Length and weight samples by sex and data source. Lines are the power function fits by data source.


The resultant sex-specific length-weight relationships are given in Figure LW2, with the following individual values:

  • Females: W = 0.000008L^3.15
  • Males: W = 0.000012L^3.07

These values are very similar to the previous assessment that used a combine sex value of a=0.0000096 and b=3.12000 (Figure LW2).


Figure LW2. Realized length and weight relationships for female and male Rougheye/Blackspotted rockfishes.
Figure LW2. Realized length and weight relationships for female and male Rougheye/Blackspotted rockfishes.


Stock-recruitment relationship

The Beverton-Holt stock recruit relationship is assumed, as it was in the 2013 assessment, to describe the relationship between spawning biomass and recruitment. The steepness parameter may be considered for estimation, but it is notoriously difficult to estimate in assessment models. The 2013 stock assessment used the previous rockfish steepness mean value of 0.77, but this has subsequently been updated to 0.72, to a value that represents a stock with somewhat lower recruitment compensation. Natural variation in recruitment (i.e., not deterministically taken from the stock-recruit curve) is apparent in the length and age data (as notable length or age classes growing/ageing over time), so deviations in recruitment will be estimated.


Major life history considerations

  • Maximum age of Rougheye/Blackspotted rockfishes