NPFMC SSC October 7, 2026 - Day 3

Expanding the R0 likelihood profiles to show all the, like, penalty and prior components, because I'm not sure we're seeing everything that's driving the scale of the population. I asked Megan if she could do a leave one out survey analysis, which she said she would, but I'll keep that recommendation in there. And that's all I have. Thanks.

Excellent. Thanks, Dana. Comments from other SSC members? Chris. Yeah, this just struck me as Franz was talking because we have another stock in the Gulf that's coming up that has a proposed change of tiers in the future, and I just think about, it's like, should we as a group think about some criteria of when we're changing those things so we can communicate that ahead of time to industry so it's not just a shock if we go from a Tier 4 to a Tier 5 or a Tier 5 to a Tier 6.

Just something to be transparent and available because those are relatively large changes and do— making those decisions on the fly always seems a little tenuous and might just be a little— might be useful to think those decisions through ahead of time to have some criteria. So it was just a thought from Franz's comments of that Tier 5 potential. So thanks.

Dana. Yeah, I just had one procedural question. So we're willing to take 25.0 as the base and not 16.4 because it's just like a data change, or what were you thinking there, Franz?

Ja, I thought they did a pretty good job of showing that it had a pretty minor impact and that it was a small enough change from the previous base model that we would accept that. Okay. Thanks.

Kerry. Yeah. I guess I've been thinking a little bit about the future of the stock. So we've heard a lot about the loss of survey data. We've got a 10-year gap in the slope survey and no prospects for deeper sampling until 2031, as I understand it.

And I was looking back a little bit at the history of this assessment to see if there had been exploration of fishery-dependent data for informing this. We heard testimony from FLC indicating that at least up till 2021, they were targeting turbot with longline, potentially again in the future with pots. I know there's trawl survey effort targeting the species, and I wonder if that potentially is a source of information that could be brought in, at least to inform trends across time, or potentially for inclusion in the stock assessment. Perhaps as a recommendation for future consideration.

Okay. Looks like that wraps up our discussion. Uh, to quickly summarize, we'll thank the authors for their work, especially the extensive reevaluation of many aspects of this model. We will note the public testimony that we received, particularly highlighting, um, the lack of deepwater sampling in the trawl survey in recent years. We will note that this is a biennial assessment and that the most recent model that was used was 16.4.

However, we, after noting the changes that were made, we are going to recommend bringing forward 25.0 as the base model, as it's close enough to 16.4. We will have some discussion about the relative merits of the models that were presented, noting that the author and plan team recommended 25.0, 25.3A, and 25.4A. We will differ slightly in our recommendation and suggest dropping 25.3A and instead having 25.0, 25.4A, and 26.3, which is the model that includes a 1977 start date, as our— as 3 potential models for evaluation in December.

We have some rationale for those choices based on the how well selectivity blocks were identified, convergence issues, selectivity questions, and retrospective patterns, which will support those— that suggestion of those models going forward. We will also recommend bringing forward a Tier 5 model in December as a potential interim to deal with the gaps in the ongoing gap in the slope survey and request more information on the scaling of the RIMA models and specifically how the longline survey was included in the scaling for that analysis. I guess at this point, we, we may just note that this is that where we are considering tier changes, we might benefit from a broader discussion about tier changes. And I guess that's something procedurally we'll have to try and figure out how we can fit into our thinking. Probably not in December, as we're pretty well booked for December, but maybe, maybe sometime between now and next year's cycle, we can think about how to have a more general discussion of tier changes to make that more transparent and easy to understand for the public.

We will specifically for this model, so that recommendation may get elevated up above in perhaps into the planning discussion rather than falling under Greenland/Turbot. Specifically for Greenland/Turbot, we're going to request seeing an expanded— expanded likelihood profiles that would include all the likelihood components. As well as a leave-one-out analysis on the various survey or data components. We're going to highlight the critical importance of slope station sampling in general, as well as better age data for this stock assessment. And I think we will suggest at least a future, not for December, but a future evaluation of the potential for fishery-dependent data to provide some additional information in this assessment as well.

Sherry, I realize this is after the summary. Uh, you mentioned future conversation about changes in tier and criteria for that. I almost wonder whether we could say that the SSC had discussion of that, and then if other SSC members are supportive, you know, to let the plan teams know that if they have thoughts about that, that we would welcome— we would welcome any thoughts that the plan team has on that. And that would at least inform our conversations in the future. Okay.

The other one was, um, I just wanted to say this publicly. Generally, in the SSC model following OMB procedures for influential scientific information, we name and we summarize public testimony. So in this case, since it's a model run and it's not influential scientific information, we normally name the presenters but don't summarize. And I would suggest that we continue doing that for consistency. Consistency, but I just wanted the public to know.

Excellent, thanks for clarifying that, Sheri. So we will, we do have assessments here that we are finalizing, which are influential scientific information, including our harvest projections, but the proposed models, we will note we had public testimony, but not try to paraphrase it. Thanks for that. And that will actually include sablefish. Yes, because those are just proposed models.

Thanks for that clarification.

Okay, I think that wraps up Greenland Turbot. We have one more proposed model run to get through, which is EBS Pacific Cod, and we'll go first to Kuri. Yep.

Thank you. I can keep this fairly brief.

This round, the authors explored two aspects of model design, the definition of the plus group. It was noted that that, you know, based on new aging methodology, no individuals older than 14 years had been observed. So the authors had a very nice exploration of alternative plus group definitions from age 20 to age 10. They also considered tradeoffs between sex-aggregated and sex-specific models driven by evidence for some moderate sexual dimorphism evident for this stock. They developed paired single and split-sex models that required a slight modification to the treatment of— or the expansion of age compositions.

As always, I think we can thank the authors for a clear documentation of the model comparisons. I always like having links to the GitHub repository with all the model files and SS3 output figures. It's super helpful.

In terms of results of the analysis of plus group sensitivity, it seems like the change in plus group primarily impacted survey catchability and log mean recruitment. You know, reducing that plus group lowered log mean recruitment and increased survey catchability, which as a result reduced the scale of recruitment and total spawning, total and spawning biomass. As well as the estimate of unfished biomass. In general, status kind of was pretty insensitive to that change.

In terms of the development of a split-sex model, you know, there's a few tradeoffs, right? When you add a second sex to a model, you have additional complexity in terms of parameterization. So when we think about kind of model parsimony, this is always a bit of a balance, and I think that the authors did a good job of sort of thinking through that.

Overall, when comparing Model 26.2 and 26.3, the single and split-sex versions, spawning biomass and recruitment trajectories were very similar.

And in terms of the age composition data themselves, there were some slight differences in model fit as well, but overall the differences were quite minor between the single and split-sex versions.

And at the end of the day, I think the SSC should concur with the author's recommendation that at least at this time, that's not a model that needs to move forward for certainly this assessment cycle, perhaps an area for further exploration in the future. But as demonstrated, differences in the outcomes are not sufficient to warrant the additional complexity that's implied by a two-sex model. So I would recommend that the SSC concur with the author and the Groundfish Plan team in recommending having two single-sex models be brought forward for consideration in December. That's 24.1, that includes the original— this is the base model including the original age 20 definition for the, the plus group, and then the model 26.1 that has the truncated 14-plus group. Um, there were a couple of additional recommendations from the plan team that I think the SSC can support.

Thinking about age-specific natural mortality, and of course, considering implications for selectivity when, if and when that's implemented within the stock assessment itself. Thank you. Thanks, Kerry. Next, we'll go to Fabio. Thanks, Mr. Chairman.

I'd like to thank Dr. Cunningham for your thorough review. You addressed all of my comments, so I have nothing to add. Thank you.

Excellent. Well, I won't reiterate everything that was just said since— well, I should open it up. Is there any— is there any other SSC comment or discussion?

Okay, then I will— I will simply reiterate that we're going to support the author and plan team recommendation to bring forward 24.1 as the base model and 26.1 the model using a plus group at age 14 and will support the other research recommendations that were made as well.

Okay, now we are going to make a transition to our GOA plan team report, but first I think we'll take a 15-minute break. So back at 9:30.

Okay, welcome back, everybody. We're now transitioning into our GOA Groundfish plan team items. And as we did for the Bering Sea, we're going to have a mix of proposed model runs and final specifications. Staff will walk us through and make it clear which are which, but just so that people know, we're going to take public testimony after each of the influential scientific information, each of the final specifications items, and then we'll have one public testimony section at the end for all the proposed models and the GOA Plan Team report in aggregate. If you're not sure where to testify and/or you want to testify on multiple items, feel free to just testify at anything up to the item that you want to comment on, and we'll include that in all subsequent discussion.

So with that, I'll hand over to staff. Welcome, Sarah. Good morning, Mr. Chairman. Uh, for the record, Sarah Kleeburg, council staff. Um, this morning we have Mr. Chris Lunsford and Dr. Ianelli, Dr. Jim Ianelli, uh, online.

I believe we have a few other authors as well to help us go through the Groundfish Plan Team report and the authors' presentations.

For kind of the order of things, we're planning to walk through the assessments for which you'll be recommending final specifications at this meeting first, and that includes harvest projections, and then we'll walk through our proposed models and some survey update information after that. We have, um, Dr. Cole Monahan online to present the POLIC assessment later on, and, uh, Aaron Lambert from ADFNG to present the DSR assessment, or the proposed models for DSR and proposed models for POLIC. Uh, if, if Chris didn't try to run away from this experience, then I will pass it over to him.

Thanks, Sarah. I'm pretty slow, fortunately for you guys. Good morning, everyone. Sarah, I think you're running the slides, so maybe we can go to the next slide. As Sarah's— or next slide after that.

I guess we did want to recognize that Sophia Wasserman was on the plan team for several years, and she's stepping down. So we just want to thank her for all of her efforts and work that she helped us out with, and we're going to be looking for some new appointments soon, I hope. So moving on, as Sarah mentioned, we've got quite a portfolio of things to provide to you. We're going to start out with the final full assessments, which is unusual for October. Those are shown in bold.

We also have 4 final harvest projections to provide, and then we'll move into proposed models for 3 stock assessments, survey update, and and then also some information on BRDs specific to skates. So we can move on. We'll start out with Aerotooth, and through the chair, feel free to stop me at any time. Otherwise, I'll just keep on going and pause at the end of each one. So for Aerotooth flounder, this was a final assessment, and note we've tried to put this as, as boldly as possible and on all our slides just to show everyone what final assessment is.

And we'll try to give some background since last year was a bit confusing coming into this year's process. Um, Kalei provided the overview of both the assessment and the ESP. We'll start with the assessment. The original bridging exercise, just for some memory sakes, was bridging it to our Seattle, and that was completed in 2024 by the authors. Last year they brought forward a full model since the last full model had been done in 2021.

And in September the plan team saw that model and basically included a lot of data improvements for input to the model. This included removing old trawl survey data, some of the older fishery comps that were in there, really looking at fishery length bends, length at age transition matrix, and she also updated the weight of age information. So there was quite a bit, it was mostly data things. The plan team had accepted that model last September You, as the SSC, saw some of that in October. And so for this year, she brought that model forward along with Model 25.1, an alternative model that basically tried to estimate natural mortality.

And so we took a look at both models as part of the plan team, and what the Model 25.1 came out pretty early on to show was that it estimated quite high mortality rates compared to what was used and were fixed in Model 25.0. And it certainly showed that this confounds M with a lot of the other processes and shifts the stock scale, which if we go to the next slide, we can just take a quick look at what was shown. And this basically shows that the total biomass went up and the spawning stock biomass flipped and went way down. This is because of the extreme high M values that were estimated. This basically caused a massive increase in recruitment, basically a lot of population turnover because of the high M rates and fewer older fish that were like making it to the spawning ages.

So both the author and the team reviewed this and felt that, you know, this was not plausible. The m-values were too high that this model was estimating. And so we backed off of Model 25.1 and then focused our attention on Model 25.0. On the next slide.

And basically, the team agreed with the author and recommended 25.0 to come forward in— as the final model, not in November. I almost caught myself there. With a fixed sex-specific natural mortality. This captured a lot of the extensive data updates that we'd seen in 2025. It also— the authors provided a nice diagnostics look at things and definitely supported the model specification and stability.

It showed a slight increase of— for increase of 12% in 2027 ABC. And the team and the author both discussed and pointed out that this is a fairly lightly exploited stock with cached ABC of about 11%. So that was the final recommendation for what model to go with in ABC. And I can pause there because my next slide will deal with the ESP if there's any questions.

Don't see any. Go ahead, Chris. Okay, and if there's no questions, we'll jump into the ESP. For background again, uh, the ESP was asked for years ago. It was— the team was formed and initiated in 2024.

A full draft of the full ESP was provided in 2025. Um, after that, it was put on hold, obviously, and for 2026 They continued to work on the ESP as a whole, but specifically on DSAM. And Clay pointed out that there's a publication in review with Arrowtooth used as an example for how to do the DSAM, and that that work is being incorporated into the ESP. So the idea is to bring forward DSAM within the RCATL framework. Basically, she went over quite a diverse portfolio of all the indicators that are being used, 19 ecosystem, 5 socioeconomic, including 4 predictive indicators that are listed right there.

And that their intention is to integrate this into the RCETL framework in the future. And as you heard in the joint discussion, and you'll hear in the public discussion a little more, There's some overlap across several of our stocks in narrowtooth and Gulf pollock especially on bringing in DSEM to the assessment. So next slide.

So from a team perspective, there was some discussion just in general about the indicators and the ESP. It basically showing that they're expecting very low age zero survival in 2025 based on some of what the indicators are showing.

It was pointed out by the team in terms of DSAM and trying to estimate recruitment that Arrowtooth actually has a really flat time series but has this huge boomer year class in 2000. So there was just some, not formal recommendations, but just suggesting that this might offer some interesting investigation into the relationships, what was going on in 2000 that really helped this large recruitment, because unlike a lot of our other stocks, it's a pretty stable recruitment time series. And then we basically kind of dived into discussion of DSAM applications, and we've captured some of those in our general recommendations as a team, just because I think they span not just Arrowtooth at this point, because they're going to bring it forward, um, in the future I think just the overall general DSAM, I'm guessing the team and the SSE likely have some general DSAM thoughts, and you're going to hear about some more of that in the Pollack assessment coming up too.

So that is my last slide on Arrowtooth. So if there's any questions, we can try to answer them.

Question from Fabio. Thanks, Dr. Lundsfors. So I have a maybe more clarification question about the ESP. So how well does the de-sync predict the recruitment outside the period used to develop the relationship? And if it can explain the unusual recruitment events such as the 2000-year class?

Before that's been incorporated into the operational projections? Thank you. Yeah, through the chair, thank you. I'm trying to rack my brain here. Sarah and Jim, please step in, or Clay, if you're on, you might want to respond to this too.

But I think at this point the DSAM is external to the model for Arrowtooth. They're planning on bringing it into the model. Pollock is in the model currently, so that's where you'll hear that. And I saw a yellow hand, so I'd love to defer the specifics to Clay if you want. Go ahead.

Uh, thank you. Through the chair, thank you for the question, Fabio. Uh, we do have a table in— I think it's Table 7A.6 in the Arrowtooth ESP that shows the percent reduction of unexplained variation, um, and that would give you an idea of how much what the effect is on the recruitment information, the sigma R, for using the structural causal information. So we have this as an 83% or 82% reduction for the structural causal model and 83% for the regression model. We also have a pretty close AIC between the structural and the regression.

So this was where we were talking about, um, perhaps we need to think a little bit more about these indicators and whether the structural causal part is really, um, done well for this model because there wasn't a lot of variability in recruitment. And this was what Chris was talking about, but we did have a reduction for including these indicators, but quite a bit of a reduction. And so if we go to the, um, the integrated version in the next assessment, we can include that within the model itself. And then at that point, we can decide whether we would use that with— for the operational model. At this point, as Chris mentioned, it's external.

So we're just predicting recruitment estimates outside of the model. I hope that helps.

Thank you so much, Dr. Shahdad.

Chris. Yeah, thank— Chris, remind me, we made the transition to BRDs for this and apportionment, is that true? And so we're no longer using kind of ABC or sub-area ABCs when we're talking about apportionment for this stock and/or others, is that true? Yeah, through the chair, Chris says, I'll let Sarah step in if she needs to, but that's correct. And all our Gulf apportionment talk, we're now going to be calling them BRDs rather than subareas.

So I think you'll— we're trying— it's a tough change. It's been tough for me, but I think that's what we're trying to do across the board for all the Gulf talks. Yeah, thanks. I just— there's some in this document. There's definitely some in the tables ABC references, even though I think they're BRDs, so it's just adding confusion to that.

So I'll probably make some recommendations on just for this doc explicitly, but in general to make sure we get those tidied up if we can. Thanks. Yeah, great, appreciate it. We saw some of that too, and it's actually one of the team's general recommendations too, is to really tighten up our BRD references and tables and whatnot for the next cycle. So we get it.

Okay, I can move on to RecSOL in the next slide. This was presented by Kerry McGilliard. Chris, I think we're going to take these in order as we go.

Oh, okay. Yeah, so we're just going to cap off, uh, Arrowtooth. We will check for public testimony, have our discussion, and then we'll jump back to you for Rex as soon as we're done our discussion on Arrowtooth. Great, sounds good. Okay, so I'm checking for public testimony.

We have no signups for Arrowtooth, so we'll go into our discussion. And first up will be Fabio. Thank you so much, Mr. Chairman.

So the SSC received a presentation on the 2026 GOA R2 Flounder Stock Assessment and the first full R2 Flounder ESP. The SSC thanks the authors for their substantial effort in updating the assessment and developing the new ESP. This is an operational full assessment on a 4-year cycle with the previous full assessment conducted in 2021. The SSC appreciates the response to previous SSC and planning team recommendations. There was no public testimony.

The 2026 assessment incorporates updated feature and survey data, revised length-at-age and weight-at-age information, and updated aging error matrix. And it trans— the transition from AD&B to T&B with the RCO framework. Two models were presented: the Model 25.0, the TMB-based RCIO model with fixed sex-specific M of 0.20 for females and 0.35 for males, and Model 25.1, which estimates sex-specific M. The SSC agrees with, uh, GOA, Ground Fish Planting, and the authors that Model 25.0 should be brought forward for management advice. Although the models have similar overall trends, Model 25.1 produced substantially higher M, near 3 times higher than the mean recruitment, change in selectivity, and a substantially different population scale. These results indicate strong confound among M, natural mortality, selectivity, recruitment, and also the survey catchability.

The SSC also could emphasize that the stability of the Model 25.0 does not resolve this structural uncertainty. The evidence for higher mortality of younger RO2 flounder from the adult catch curve and multi-species analysis, together with the difference between Models 25.0 and 25.1, highlights the need for further evaluate the M-selectivity-recruitment relationship.

Uh, the SSC considers survey catchability an important source of uncertainty in the population scale, and the potential influence of bottom temperature and the 2003 survey observations warrant further evaluation. The different trajectories of the AGFNG large mesh and AFSC longline surveys also provide an opportunity to evaluate whether these surveys can provide complementary information on abundance and selectivity. Regarding the new ESP, the SSC supports the de-seam approach for investigating recruitment variability but recommends treating the current estimated ecosystem relationship as research results rather than establish causal relationships for management advice, and two additional indicator refinement, model validation, and updated recruitment estimates are incorporated. The, the SEIM identified relationships involve spring sea surface temperature, spring arroturf larva, summer-fall alphas and arrotouffe biomass consumption, but some estimated relationships have signs that differ from the hypothesized mechanism. The SSC also cares for concerns identifying during the 2025 ESP review regarding indicator availability, model complexity, and unexpected relationships.

A substantial proportion of indicators had a limit, uh, recent availability, and the initial DSIM had too many latent variables and did not converge until the indicator set was reduced. The SSC supports using the 2000-year class as an important test of the DSIM's predictive and biological value. The SSC also supports continued development towards eventual integration of DSIM information into our Seattle and potentially off-cycle projections that provide predictive performance, robustness, and the ability of ecosystem information to reduce recruitment uncertainty that can demonstrate before— that can be demonstrated before operational integration.

The SSC also supports continued development of socioeconomic indicators and clear consideration of how such information may inform ABC versus TAC decisions. The stock is not overfished and overfishing is not occurring. The most recent complete year catch was substantially below the OFL and projected female spawning biomass remain well above B35% in the assessment projections. The SSC supports the author recommended to 2027 OFL of 157,261 metric tons and ABC of 134,083 metric tons under Tier 3A. All 4 risk table categories were rated Level 1, and the planning team supported the maximum permissible ABC with no reduction from max ABC.

Given the favorable, favorable stock status, low utilization, and current risk table assessment, the SSC does not recommend an additional ABC buffer for 2027.

In addition, the SSC recommends the following priorities for future assessment and SP development: continue investigation of MSL activity recruitment, confounding, include evaluating of age-varying M and alternative selective formulations, evaluated alternative treatment and weighting of composition data, including fishery composition data and survey age composition weighting, and access their effects on selectivity, recruitment, population scale, and management quantities, continued DSEEN validation and robustness testing, including evaluation of the 2000-year class indicator availability, alternative indicators definitions and lag structures, alternative causal structures, and predictive performance.

Develop a framework for eventual integration of DC information into our Seattle and off-cycle projections, including evaluation of whether ecosystem information improves recruitment estimation or reduces recruitment uncertainty before operational implementation. And finally, continue development of socioeconomic indicators and clarify their potential application to ABC versus TAC decisions. That's all I have. Thank you.

Great. Thanks, Fabio. Next, Chris. Yeah, thanks, Fabio. That was great.

I don't have much to add on that. The things that I'm going to add are really kind of presentational and thinking about BRD these explicitly for this. Just for reference, I'm mostly looking at, to start with, page 7 of the, of the SAFE document. And one of the goals, I think, of the apportionment is, you know, we heard it earlier, is to reduce the potential of localized depletion. So that means we need a clear way to track that over time.

So I'm just going to suggest a couple of recommendations. One that we see the apportionment back a handful of years so we can track how that has changed over time and not just a year or two, because that doesn't give us enough historical context from that standpoint. I would also suggest— and really the overarching recommendation is a common presentation of spatial apportionment for all the different stocks that are using BRDs, but a couple of explicit things. It'd be really great if the years for apportionment are all in rows. There's a, there's a table in here where they're the years are columns.

It's just hard to track how things change over time. If that's one of our goals as a quick look, that would be super helpful. And then because of the clarification that we got, it would really be great to see all the terminology ABC removed from any of the tables that are in the apportionment unless it's explicitly the total ABC that's explicit in that part of the table, right? So there's some of those, those couple of things I think would make us track this way more efficiently and, and make it cleaner for everybody to move forward from that standpoint. And just a small detail for Kalei, and I think it's on page 32 of the SAFe, there's a typo in one of the BRD sections and it says 8,020, I think, and there should be an extra digit in there.

So it's supposed to be 80,209. So just a— it's a— I don't know that it matters dramatically, but it's a fairly— it's an order of magnitude off, so it'd be nice to just tidy, tidy that up. Those are really my big things for this. This is what struck me, and I know that we're going to be talking about birds kind of all the way through, so it'd be nice to see some focal effort on consistency on how we present those moving forward. So that's my big comment for this.

So thanks, Fabio, again for that summary. And that's all I have.

Thanks, Chris. Other SSC comments?

Sherry? This is actually just sort of a record keeping. So Chris, you mentioned that most of those were general recommendations for BRDs, and we can figure this out as we go along, but I'm just wondering whether we want to put that in a general groundfish comments section, unless you think—. My comments are explicitly to arrowtooth Flounder, because I haven't looked at all the other ones, but I assume they're comparable in all the other assessments. So I assume my comments are more broadly applicable than just to Arrowtooth.

I'd be great if they happened in Arrowtooth, but I think they probably should happen elsewhere as well. Okay, great. The other one was a question for Fabio. Um, you had mentioned that there was no reduction for MAX because of low utilization and a stable model, and the SSC has been on record saying that not needing the TAC or ABC isn't a reason not to reduce it for Max if there is a reason to. And so I'm wondering if we don't consider that, would you— do you still agree with the plant aim and no reduction for Max?

Yes. Great, thanks for that clarification. I actually had a couple clarifications as well. So for a question for Fabio and Chris and others, we're encouraging additional work on the DSIM approach, and— but I'm not entirely clear whether we're asking for more work before they start showing us an integrated version within RCATL or more work by showing us an integrated version in RCATL and then developing from there. So are we asking them to do specifically to do more things before they put it in or after they put it in to better understand it?

Yeah, I would say the latter. I'm not— I think we should see it as it comes, and when they're ready to present it to us, that'd be fine.

Personally, I would agree as well. If doing it inside the model makes more sense than outside, and it's— it can still be a research model. Okay, the only other clarification that I think we should make is that, um, just for the public and everybody in the process, I think we should highlight that we reviewed this assessment in 2025, and we have a two-part review process, which is we see it once and as a proposed model and then again as a final, and we saw it as a proposed model and now we're seeing it again as the final. So we're just to make it really clear that we're following our normal procedure. We just had an extra long lag in between the proposed model and the final model.

So when we, when we put the narrative together of how the history of when it was done and reviewed, that I think that's important to include. Any other last comments from the SSC? Ian. Just a clarification, Sherri, should we just for minutes just recognize the broader context here and say, see the— wherever we end up putting a broader comment for BRDs, it's like we have comments to improve presentation of BRDs and see further comments somewhere else, or do you want me to go into some of those details here in Arrowtooth?

Um, you mean in Arrowtooth in the report? Um, I think that if you only looked at Arrowtooth and nobody else notices that those issues apply elsewhere, I think we need to, like you said, keep it in Arrowtooth. If other people observe similar things, then it makes sense to either reference or move it to general.

Yep. Great. Okay, um, so I'll just summarize briefly. We're going to thank the authors for their work on the assessment and the ESP. We'll note the history of this assessment from 2021 through the models we reviewed in 2025 to the models that we see in front of us now.

Um, we will provide some background on the, the different models and the results from the ESP. I won't characterize that again. Fabio covered that. Um, we will support, uh, continued evaluation of the DSUM approach by integrating that into the modeling framework in the next round, perhaps as a research model. Importantly, we will support the plan team and author recommended Model 25.0 for the OFL and the ABC and agree with their recommendation for no buffer from the maximum permissible.

We will note that this stock is not overfished and that overfishing is not occurring. We will encourage further research on the confounding between M selectivity and recruitment. More work on data weighting, and additionally, as I mentioned, the further exploration of DSEM to explain recruitment variability. We will also encourage further continued development of socioeconomic information as part of the ESP. And then we will have comments requesting—and this one actually is general, I think—reporting the apportionment results for the last few years.

I think that we can just make that general across stocks, and so that it can be tracked how it changes, and to try and standardize that reporting across the stocks that we see. Specific perhaps to this assessment, to keep the years in the rows and recommend carefully going through all the documentation to make sure that the wording is reflective only of BRDs rather than any lingering sub-area ABC references.

I think that's what we have for this stock.

Okay.

Back to you, Chris, to resume on Rexol. Great. Thanks. So, yes, Rexol was provided by Carrie McGilliard. This is an identical situation to Arrowtooth.

So you saw the proposed model last year. And December was skipped. And so the final model coming forward now, basically the last time this assessment was done as a full assessment was in 2021. Um, last year what the plan team and the SSC saw in the fall was, uh, Model 25.0. This was a model that was prepared for a CIE review that was held in 2025, um, and it was, uh, basically quite similar, I think, in ideas to what was done for Arrowtooth where there was some data cleanup.

Again, dropping some of the older survey biomass observations, revising how fishery age compositions are put together, different things like that. So last fall, both the plan team and the SSC kind of approved those changes and moved on. And then the other model brought forward this year was Model 25.1. Which simply built upon 25.0 by adding aging error matrix that are specific to Rexel. There was quite a bit of documentation provided last year by the author, Kerry, after the CIE review regarding a lot of this.

So it's kind of a mix of what you saw last fall to what we saw now. And we can go to the next slide.

The basic idea was that the Model 25.0 and Model 25.1 gave fairly similar or very similar results, and that Model 25.1 was an improvement because of the addition of the aging error matrix. When the plan team and author were discussing things, it was pretty evident that the most recent couple of survey observations, biomass observations, and then some of the composition fits seemed to be a concern. In discussions, it was definitely that, you know, the models kind of explain the long-term historical dynamics, but the last few years don't seem to match up too well. The general idea and pointing to things were that, you know, this went from a one-area model to a two-area model, which was a huge improvement when that was done, and that there's very likely some movement and growth among or between areas that the model 25.0 and 25.1 aren't capturing very well. So we can go to the next slide.

Just to show you the survey fits as an example, both the 2023 and 2025 surveys, the model misses quite a bit. It was also noted that the 2025 Eastern Gulf survey was the highest and was exceptionally high biomass estimate. We did have some of the survey staff in the room and there was a pretty healthy discussion on what to look for on that. It was pointed out it wasn't thought to be like an effect of the stratification change that happened in 2025. And one of the survey staff thought they remembered a large toe.

And so it kind of centered on that discussion of investigating a little farther to see if it was a large toe that maybe influenced this. The same conversation came up in Pawlik, so we'll touch on that later. But just observation that that was the highest biomass estimate recorded. The model's having a tough time with it. Maybe it can be explained off because of an abnormally large toe.

But anyway, the point being that the two models are doing fairly well, but definitely struggling a little bit with the last couple survey years. So next slide.

The, the team went with the recommendation, which author also recommended, of Model 25.1, which includes aging error. Basically, for this stock, spawning biomass is above B40. It's projected to decline slightly, but all in all, pretty good shape. The final part of this assessment was an appendix that the author along with Matt Chang provided. Carrie and Matt worked on basically bridging exercise to Spork, and it was brought forward now rather than waiting for the next operational assessment simply because that could be 4 years from now.

So both the ability to like pursue some of the research goals that Carrie has on looking at spatial components components and different things between the areas such as growth, Spork will be advantageous for that, and it also gives Carrie and/or the next author a little bit of leeway in terms of waiting for 4 years and trying to find the executable files for SS3, things like that. So I— we've seen some of these bridging examples so far in the last couple years, and the appendix we all felt like did a really good job of showing that they successfully bridged it across, and the team recommended that they, they consider that the bridging exercise and that sport can be used for the next operational assessment. I will note in the appendix they brought forward both 25.0 and 25.1, just not knowing which the team or the SSC would approve. So both of those models were bridged. So I think we can put that one to rest.

I think that is all. Yeah, that's all we have for Rex Soule. So take any questions.

Dana. Hello, Chris.

So the only real change from 2500 to 2501 is the inclusion of the aging error matrix. I dug back into September last year, and this document I couldn't really find any information on how that was calculated. Um, did you guys talk about that at all at the plan team? Yeah, hello Dana. Um, let's see, we, we had— I actually brought that up to the team just to check, and everyone felt it was a fairly standard, um, method that was done.

I thought maybe last year there was a document that was provided by the author in addition to the assessment but I might be wrong on that, into the CIE materials. But, uh, Jim may also know more about that as he's worked quite a bit with Carrie over the years on this. So Jim, do you have any input for that?

Uh, sorry, drawing a blank, uh, Chris. I, I can— Carrie's online, maybe she could, um, be brought forward as a presenter or a panelist.

She's got her hand up, I think. Excellent. Thank you, Carrie. Please help us out.

Yeah, thank you. I, I believe it is in the, um, was presented in September 2025, so it's not in the CIE review materials, but it is in the September 2025 document, which I think was at least at Plant Team, that document was linked online, but it does use the package that André Punt originally developed for calculating age uncertainty. It doesn't have any aging bias, just age uncertainty based on double reads. Does that answer your question, or are you looking for more, more details? That's good for now, thanks.

Other questions for— yep, Martin.

Oh, I don't— maybe you're going to get to this. Um, I was going to ask about the BRDs, Chris. Are you— do you have another slide on that? Uh, no, sorry, we do not. Um, but in terms— we can try to answer.

We got Carrie here too. In terms of using apportionment versus BRDs, or about the specific how they were calculated? Yeah, no, thanks. I, I had actually not even noticed it, um, in, and when I did the review, and I realized that it was probably because all that's reported is the results of the RIMA, uh, RIMA analyses were used for each area, and And it's— that's all we hear. There's no showing of fits or model structures or anything like that.

And I'm just wondering if there's, um, from your perspective, Chris, is, is there sort of a minimum supporting, um, documentation for REMA's to calculate PRDs? Is that part of the terms of reference, or is it just no longer even sort of considered a scientific issue that's worth looking at?

Thanks for the question, Martin. I guess you're asking just about authors providing the documentation versus RIMA. I guess my short answer— maybe I'm not getting your question well enough— is that We've got the— the REMA is definitely documented, the package is out there, we've been doing it for years. So I think the authors are just running it and referring to using the REMA package. So I don't think most authors are probably writing too much more content than that.

Jim, I don't know if you got a different slant on that response. Oh, I see Carrie's got her hand up. Maybe she can speak to it again? Sorry.

Martin, I think that it's something that we could do. We could include more model fits.

I never have. I wasn't aware that other people were for the purpose of BRDs, but more for the purpose of Tier 5 assessments.

Okay, I guess maybe, I mean, there are some settings that you can use, and I mean, you know, there's some configuration details as well as maybe interest in the FIT, and I guess I'm, I don't know if we've even discussed this, I kind of feel like, you know, just the fact that there was Nothing sort of hit me, you know, just a little surprising because, you know, it's— so maybe we— maybe the SSC needs to have a discussion right now. I don't know that I feel strongly either way. If it's really regarded as being just, you know, plug the numbers in and take your output out and, you know, not really report any of the details. Maybe, maybe that's, that's enough. It is, it seems to be a really robust sort of approach.

I mean, um, I, I don't know. I mean, do you think that just reporting the details is, would be enough, or do you, does it seem to you that maybe we should ask for more documentation?

Martin, I don't know if that question is back to me or Kerry or anyone. Either of you, I think. I mean, I'm just probably just sort of thinking it through myself right now, which is kind of awkward. Yeah, I know how that goes. Me too.

Yeah, I think a general recommendation, if the SSC would like that, I think, you know, the Tier 5 authors, just off the cuff, I feel like the Tier 5 authors go into a lot more detail. It's certainly possible to bring that forward, like, in our guidelines for the Tier 3, or just using that for the BRD designations, um, to look at the specifics. I don't know consistency-wise off the top of my head how consistent it is across, like, the Tier 3 stocks that, like, Rex sold here, but that's certainly something we could look into. Thanks, Martin. I would need to get back to you later if you wanted to know more specifics about REXEL.

Okay, it was— that was your last slide for REXEL, Chris? Correct. Great. We'll check for public testimony then, unless there are more questions for Chris or Carrie. Carrie, thanks for being online.

Okay, seeing no more questions, uh, we have no public testimony for REXEL, so we can go right into our discussion. And I— first up, we have Martin.

So, um, this is the final assessment that follows from the initial review of a draft assessment and proposed model runs in October '25. Um, there was also a CIE review of RECSOL in June of 2025 that suggested improvements to the assessment. The final assessment could not be completed in 2005 due to the federal government shutdown. The assessment is a Tier 3, 2-area Eastern and Western Central age-structured model with the fixed proportion of recruitment in each area and with different growth patterns by area. Selectivity to the fishery in the survey are modeled as asymptotic using the double normal selectivity option in stock synthesis.

The SSC notes it's a— there's a good balance between complexity and data availability in the models presented. Two models were recommended in October '25 for the final assessment, 25.0 and 25 25.1, the models differed only in the incorporation of an aging error matrix in Model 25.1. Differences between the two models are relatively minor, although overall model fits are slightly worse with the inclusion of aging error, but this isn't unusual.

Adding an aging error matrix was a request of the SSE in 2017. So the SSC is pleased to see this addition. The, the author noted an increased risk level for assessment-related considerations due to the similarity of the maturity and selectivity curves, which produces sensitivity in reference point calculations, the poor fit to the 2025 survey biomass, and variability in the fits to the conditional agent length data. There were no other elevated risk scores. The plan team recommended 25.1 for harvest specification.

The SSC supports this recommendation. The SSC also endorsed the BRDs calculated using the ARIMA model. And I mean, I think I'll— I don't have it written out yet, but I would think we should also recommend that Just basic documentation should be provided when the REMA model is run, including, I guess, model configuration as well as a figure showing model fix.

It's— new figures were added to the assessment showing, um, RECSOL food web relationships for Eastern and Western population as well as figures showing Rexel diet and predation mortality. The SSC found that these were valuable additions to the assessment. Um, at June 2026, the SSC requested that assessment authors for deepwater, uh, Gulf species examine and describe how historical survey estimation methodology and estimates compare to those under the new survey design. And highlight impacts to the assessment. This was not— request was not completed for Rexol, so the SSC recommends that this request be addressed in a future assessment.

The author does indicate that the fraction of rock soil occurring deeper than 500 meters is usually low, and that survey estimates for rock soil have not been corrected for missing depth strata, so this may be of less of a concern. Um, this assessment author also included a bridging analysis between models 25.0 and 25.1 and comparable models using Sporc. Um, while not under consideration for this assessment cycle, the author wanted to make this transition now since the next RECSOL assessment is in 4 years. And SS3 is no longer being actively supported.

SPORK was considered a suitable replacement because of its ability to model multiple areas. A comparison between an unfitted SPORK model using parameters estimated using SS3 showed identical model output. Comparison of fitted SPORK models and their SS3 equivalent showed slight differences. These differences were considered to be a result of differences in how recruitment deviates were modeled, and, um, in addition, the seasonal timing of survey biomass in SS3 had been misspecified. The SSC considered the bridging analysis to provide a sufficient basis to transition to the SPURC model for the next assessment model.

The SSC recommends exploration of RT&B's tools for modeling process error to evaluate potential time variation in biological processes, with growth being a priority. Temporal variation could also be evaluated for natural mortality, selectivity, and catchability. However, it's unclear whether there's sufficient data to support additional modeling complexity for RECS also. These evaluations should proceed cautiously.

I'm done.

Thanks, Martin. Next, we'll go to Dana. Yeah, thank you, Martin, for the comprehensive summary, and I'll just add a couple extra things. The SSC supports the new aging error matrix in 25.1, but the safe text itself doesn't explain how that was calculated. We got a little feedback from Kerry just now, but I guess I'd like to request that the next safe include a brief technical summary showing the sample sizes, the model used, and the smoothing methods to build the matrix, and maybe a figure of it.

Just so there's something to reference in the future. I would like to see the— an evaluation of that large jump in the GOA survey biomass and whether it was just a couple of toes or could represent some more broad-based eastward movement or maybe large recruitment events. The potential furlough 5-year projections that we requested were in this and in the GOA ATF assessment we just saw. Thankfully, we didn't need them this year, but they appear to use different assumptions for projected catches. These were the only Tier 3 models for this meeting that we're going to see this in, and as we said yesterday, we won't see these for Tier 5 models.

But I just think maybe we should put a recommendation in the general for the rest of the November upcoming Tier 3 models that we're making sure authors are using consistent methods for these projections. That's it. Thanks.

Okay, comments from other SSC members?

All right, I hardly need to summarize this, but just briefly, we will summarize the review and assessment development history. The basic structure of the model, noting that there were two models that were, that were brought forward this year, 25.0 and 25.1. We were pleased to see the addition of aging error in response to our previous comments, and we will support the plan team and author recommended model 25.1 for specifications this year, including the OFL and ABC with no reduction from the maximum ABC. We'll—. And I don't know if we said this out loud, but we will note that the stock is not overfished and overfishing is not occurring.

I think we need that in our minutes.

Uh, we will support for the next stock assessment, we will support the use of 25.1 implemented in SPORK. So we are adopting the bridging analysis that the author and plan team have reviewed and saying, okay, now 25.1 in SPOC can be the base model for the next assessment cycle. We have several recommendations, some of which are going to be probably general across stocks. The first is with regard to providing basic REMA model specifications when the REMA model is used to calculate BRDs. This could— would include model specifications as well as model fit in the form of a figure.

And I think that'll be a general comment across stocks.

We will also, specific to this stock, we will request a description of the methods used to develop the aging error matrix.

We will also request further evaluation of the increase in the Eastern Goa survey biomass. And how that influences the stock assessment. We will have a general comment again to all authors regarding consistent methods for use in making the projections, noting that this is the first time we've seen the 5-year tables, at least in, in the upper part of the document.

We will recommend addressing the— or providing the A response to the request we've made previously to address the treatment of deepwater survey missing strata in the historical time series for deepwater species in the Gulf.

And yes, along with the transition to Spork, we will support exploration— future exploration into evaluation of utilizing the process error capabilities of that framework for potentially time-varying processes. We'll list out a couple examples there. Sherri. I have a question for Martin. Well, or Ian.

In the summary, you mentioned that— how's this— in the chapter, the author did say that how they handled deep stations in the past is they didn't do an analysis because catches in those deep stations were low. So they actually, I think, have answered that question. But Martin, was yours We asked, we asked, um, I think in June, was your question to say how they, how the assessment may be affected by the change in surveys, or was it what Ian had mentioned, just to document what they've done historically?

I, the explicit recommendation that Ian Ian mentioned was something coming out of our June meeting, and there wasn't anything that was sort of— I mean, I sort of pulled some details out of the assessment that suggested maybe it wouldn't have been that big of a deal for Rexall. And so I more or less assembled a response to Ian's query, and maybe that is considered adequate. I, I mean, if it's really not an issue, and I But I don't know that, you know, we've seen, like, the actual estimates of, you know, the proportion in deep water and whether it's, you know, whether that's a consideration.

So the type of analysis that Ian suggested wasn't done.

Great. Thank you.

And so I guess just if we reference specifically our recommendation from last June. Yeah, no, that's exactly what I've done. So, yeah, excellent. Last chance to add anything to our summary for Rex.

Okay, back to you, Chris.

Okay, thank you. So now we're going to move on to skates, and I'll just preface that you thought you said BRD a few times. We're going to say them a few more. We have a section on the stock assessment for skates, and then we have several slides as follow-up particular to BRD. So heads up on questions for that.

So this was provided by Lee Cronin Fine. The last full assessment was done in 2021. There was an update completed in 2023, and this would have been a full— or a full/update last year but got pushed to this year. So Fairly straightforward. This is a Tier 5 REMA-based model.

M is steady at 0.1. I think in regards to BRDs, we thought we should note that the big and long nose skates are assessed by the BRD areas— Western Goa, Central Goa— and then other skates which are part of the complex are assessed Goa-wide. Next slide, Sarah.

We're just going to throw up the survey trends here since this is Tier 5. Specifically for big skates, the survey trends are quite stable. The current catch overall is below ABC and OFL, and we'll point out that a lot of the bulk of the biomass is in the Central Gulf for these this species, but that the central GOA BRD in 2025 was nearly exceeded. I did follow up with Abby John, and a lot of that catch was attributed to a high catch of skates by some of the fixed gear vessels in the central Gulf, namely cod vessels. So that was what was seen in 2025.

I believe when you look at the 2026 catch, it's come down quite a bit. So something to note for later discussion.

Going to the next slide. Longnose skates. This one is fairly important to note that the biomass is definitely centered in the central Gulf, less on each side of it. And in the western Gulf, this— these figures don't do it justice, but quite a bit less in the western Gulf. And basically on a go-wide basis, the stock's been, or the species has been stable since 2003.

But when we look at the 2025 survey biomass estimate, it went down quite a bit. And since this is done on a specific area basis, this will be dropping the ABC quite a bit from 104 tons to 55 tons. Due to that survey estimate in the western area, which is where very little of the biomass is.

Next slide.

So the author presented all that. I didn't put in a figure for the other skates, but we could address any questions if there's any on that. But the team as a whole had some discussions and came out with several recommendations. One was that several team members noted that it was worth checking into the other states' CV and how that was put together, and then also to consider adding the additional observation error option that's in REMA. Basically, that was an attempt to smooth the uncertainty bands when you look at those figures.

So I think it was a pretty good discussion. I think that should be easy to do within the REMA option. And just double-checking that CV and basically seemed pretty tight for the other states. The team recommended and agreed with the author's recommended OFL, ABC, and the BRDs. And then the team did bring up that the author didn't note any recommendations from previous assessments, but there actually were some going back a while.

So from 2023, there was a recommendation for reevaluation of natural mortality as well as putting together a stock structure template. So the team made a point of re-recommending that be done, especially in light of the BRD discussions that have been going on. And I think that would be a good thing to bring forward for the next assessment. So I think that is all I have for skates. If there's questions on the assessment, we do have several slides on BRDs too.

I think we we can save that for after the assessment discussion. Chris, we do have questions. Ali?

Yeah, thanks, Chris. Just a quick question to confirm the assessment schedule for this particular stock, because I do think it might be applicable coming up in BRDs. And I didn't see it in the document anywhere, so I just wanted to confirm that this assessment would be on the schedule for 2027 following our typical Gulf survey schedule. I didn't find it anywhere in the plan team report, so I just wanted to confirm that. Yeah, thanks, Ali.

I think, I think that's correct. I'd have to go back to our stock assessment memo that we put out that has planned dates, but that would be because it was planned for last year. I would expect it to be back on schedule next year in 2027. Definitely worth making a recommendation on that though, if SSE wants that bringing forward so that the center knows to prioritize that.

Thanks, Chris. It looks like Sarah might have something to add. Through the chair.

Uh, thanks. Uh, through the chair, Chris and Ali, it looks like, um, at least on the stock assessment memo, this is scheduled for a catch report in 2027. So, um, if the SSC has a recommendation on something different from that, um, it'd be good to have it in your notes. Thanks, Sarah. I had a 50/50 chance.

That's how it goes. Thank you, both of you.

Do we have other questions on skates? Chris, I have a question for you in terms of procedure here. You said you have more slides on BRDs. Are they relevant to the skate discussion?

Or— because if so, I think we should go ahead and take those rather than try to finish off skate. Let's, let's go ahead and do the BRD discussion too. Yeah, absolutely, that sounds fine. I don't want to drive the bus too much on that, but happy to do those first, and, uh, let's jump into them. So yeah, you bet.

So this slide, um, Sarah put together, and please Sarah, step in at any time on this. This is definitely a group effort trying to put this together. Um, basically it's worth reading this council motion That's basically asking for flexibility in the survey-based apportionments for those specific stocks. And what came to the team without a whole lot of preparation beforehand, so a lot of this we were kind of dealing with it on the fly, but what came to the team was that, you know, small changes from the area apportionments Could that happen and could we also consider last year's or the last several years of what the apportionments were versus this year's? So hopefully everybody read that slide, understands that, and basically the team was specifically asked if using last year's BRDs in some areas would have biological impacts at the stock level.

And so there was some team discussion specific to that, and we actually came back a little bit later when we were able to collect our thoughts a little bit. And if you go to the next slide, Sarah, I can provide some of what that discussion captured. I think, you know, the big thing that was noticed was the decline in the longnose skate biomass in the western Gulf. And it was pretty, it was noted and discussed that this clearly seems to be like the edge of their range when you look at how high the biomass is in the Central Gulf versus the Western Gulf. And apologize for the figures again, but you can see like that, that's quite a bit less there.

So it's not surprising maybe to see that there's a large variation in the survey estimates from year to year. And that we really don't have much knowledge of the stock structure. Part of why we brought that recommendation forward. So just on the fly, it's pretty hard to, um, say much about that.

Um, it was also brought up that these area-specific BRDs for big and long-nosed skates were a legacy from decisions made in 2005, and that was due to concerns the author brought up in disproportionate harvest levels between the management region. So clearly there was some concern back then. The BRDs have occasionally been exceeded or at least approached. There was some recommendations that came out following the 2023 assessment that was specific to the Eastern Gulf when a lot of the fixed gear vessels had pretty high catch rates and went above the BRD. You can look back on the big skates and see that back in those 2000s you saw catch go quite a bit higher than what the BRD was.

So, you know, not repetitive, but definitely has happened in the past. And that, I think, most notably what we go by for setting up these BRDs is the survey distribution of biomass and it's definitely, you know, we recognize that that western Gulf area has quite low biomass compared to the central Gulf, but it was a big decrease in 2025 and was the lowest biomass estimate since the 1990s. So, you know, disproportionate western harvest wasn't expected to affect the core population probably that resides in the central Gulf, but also We always point to localized depletion, and that still remains a concern despite what we're looking at here. And then the next thing, you know, I think we're a science body, so the team naturally went to, well, we can figure out how to do this. And so, you know, I think the obvious thing the team thought about was looking at survey uncertainty and then trying to develop scientifically informed ranges that we, you know, that authors could compute and the teams in the SSE could provide.

I think, you know, for this year being asked at the table to be thinking about it without a whole lot of the ability to prep for any of this, the legacy of the area-specific BRDs, um, leads to the conclusion that there likely wouldn't be an impact at the stock level of distributing harvest. Um, but from a process standpoint, the team really would like to point to better scientific methods for determining what's appropriate and what amount of that deviation could be. So I think right now it's like we still look to the, um, The BRDs as a survey distribution, that's our recommendation. And then the council decisions for determining tack and redistribution can certainly be moved around a little bit, but it's really hard to just point at a value or a number this year for being able to do that. And I'm going to pause there and I'm going to make sure Sarah I guess we have one more slide, don't we?

I think I got ahead of myself. Sorry. So, you know, basically the recommendation that came forward was that we support using uncertainty in survey estimates to develop a range that could be used by the council. We also think that these legacy area-specific BRDs and the necessity of them could be revisited. That was over 20 years ago those were set up.

We haven't seen overages in terms of catch to BRD, so that's something we could definitely look at. And then, you know, there's methods that can be explored if so chosen to figure out a better quantitative way to move amounts between regions. So I'm going to pause there, and I'm going to see if Sarah wants to add in anything on to that.

Thanks, Chris. I think you nailed it, but I'll wait and see if there are questions before I confirm that. Happy to answer any questions.

Chris, I have a question, a skate-specific question, and I'm not, I don't know all the details about the various skate fisheries, but I understand that not all the skates that are included in the catch are retained. And I know that there's a lot of evidence for survival of discarded skates. I'm wondering, did that— was there information presented to the plan team about how much— what fraction of the catch is actually retained? And was there any discussion about whether there's already a buffer built in here for survival of discards? Yeah, great question, Ian, and I should have hit on that.

And yes, yes, and yes. So the author does provide some information that in some of the tables in the document. It definitely came up in the team, uh, like discussion and conversation, and we were all scratching our heads because we didn't understand. Going over clearly isn't what anyone wants to do, but since there— most, uh, cases— I don't remember the numbers off the top of my head, but I want to say like 10% or something are just, uh, are retained or less— that a lot of these individual skates are being discarded and the fixed gear is responsible for a lot of the catch. So right to your point, I think a lot of these are probably being thrown over and maybe are surviving.

So the team did hit on it. In terms of specific recommendations that came out from talking about that, we didn't have anything, but it was definitely an observation we had.

Thanks, Martin. Yeah, sorry, I might have missed some of your discussion, Chris, but Just the idea of using uncertainty seems a little puzzling to me, and I wondered if you could sort of, could you sort of map out how that might work and why uncertainty is appropriate rather than something like how, you know, how resilient the stock is or what its productivity is or other things that you might consider Yeah, sure. I guess definitely welcome, Jim and Sarah have thought a lot about this too, but to blend it down to a quick answer, I think, you know, this has come up in the past and it's often the minor rockfish species that there's a lot of concern about, and I think you can definitely dive into specific species or stock concerns on such things as vulnerability. That said, the majority of all these stocks, we have such little information, even you know, bringing forward the genetic stock structure for skates, we're not sure how much we're going to gain from seeing that. So I think the idea was just to provide the idea that the survey isn't absolute, and knowing where the distribution is, that we could just point to the uncertainty and survey estimates and tie into the CVs on those and realize, you know, especially for rockfish, we really highly uncertain survey estimates that bounce around from year to year, we definitely could probably deviate from that, that absolute biomass estimate that's generated and use some level of like 10% or, you know, link it direct to the CV for the stock even.

And I guess in general, we were more focused on a simple approach that could be applied across numerous stocks without getting into too much of a specific analysis of stock by stock. But that was just general discussions. Once again, it was kind of on the fly and would be what we thought an easier way to just kind of like say, yes, we'd be comfortable with some deviation from that survey biomass estimate based on survey uncertainty of what percentage. That's something the team and the SSC would have to agree on, I guess. But That was the general gist of our conversation.

If I could just try to support Chris a little bit on this, you know, it gets down to how well we can actually manage given the limited stock structure information, and it's a really small area— I mean, small stock in the region that we're worried about, and it makes sense that the uncertainties inflated there, and that considering pooling it with the western, or at least giving a little bit of flexibility in the BRD for the western Gulf for that stock, might be something to consider in the future. Thanks.

Yeah, and just— oh, I'm going to follow up real quick, if that's okay, on what Jim said, that idea of like combining the Central and Western, uh, you know, the BRDs was brought up after the meeting, so the team didn't discuss it. So we didn't want to like put it too big on our presentation here, but that's, that's something that was thought of after the fact, done in the past, and something that could be on the table for consideration.

Thanks for that. Uh, good question from Sherry.

Sure, I think that my question is when the bullet here that about supporting the idea of using uncertainty and survey estimates, was this specific to skates or was this a suggestion for all species BRDs? Yeah, thanks, Sherri. This, I think, definitely was more of a general suggestion. And again, like I said before, I think the team— the wormhole the team went down into was more of like, what's a simple way we can provide some information to the council that says, yes, you can deviate, and here's, you know, we clearly want— as a science body, you're looking for a quantitative method. And so that's kind of what we, we zoned in on.

But again, this was a suggestion on the fly, so this hasn't been investigated too much.

Sarah, go ahead and join. Thanks. Um, so my understanding too, and Chris and I have gone back and forth on this trying to, uh, disentangle this topic a little bit, but, um, my understanding was specifically for this conversation it was specific to skates, but With the understanding that the team was looking, was asked specifically in this case to look at skates, but in the future if there are stocks that continue to come up that the council's asking about flexibility to deviate the TAC from BRD, that using uncertainty in survey estimates may be applicable across other stocks, but I think process-wise, the discussion of, as Martin suggested, um, this, the status of the stock and any stock structure information that can inform whether or not there should be flexibility in deviating at all would be a good discussion for, um, the plan team members and SSC to have, and then using quantitative method, um, such as some of the options that the plan team talked about after that information has already been provided. Um, that was kind of my, my sense of the conversation is this 2-step method of, of whether or not flexibility, um, should be provided given what we understand about the stock, and then, uh, this quantitative method that, um, is yet to be determined.

Rob. Yeah, thanks for that explanation. And just to kind of follow up on also the comment about potentially combining areas, and, you know, so I think that's an important part of this is in turning— is understanding how this— the different areas and the structure areas were originally set up and what we actually know about stock structure. So like, for example, in this situation with the Longnose Skate, I guess the, the, at least the, um, the discussion to potentially combine Central and Western, is that relative? I mean, is that consideration even brought up because over the long term the Western has had such a low population level relative to central, whereas as opposed to if it was a declining trend or an increasing trend, for example, how would you handle that?

Yeah, thanks. I think that's a question to me, Rob. Um, good point. And I think, you know, just to clarify, this maybe wasn't something the team went into discussions about, but I think it's worth pointing out that in 2005, there was a decision to do that, and I don't recall— I was probably around, but I don't recall all the rationale behind that. We could go back and find it, but that I think is like, it's on the edge of the range.

Should it not be tried to manage at such a small amount, which is what Jim pointed to, that maybe it could just be blended together with the Central Gulf? You know, those are all questions, you know, that could be looked at. And I think the, the stock structure template I hope the results from that, if it comes forward next assessment season, will be able to provide some insight into those things. But I'm not sure, I was trying to pick out the question on that, so please let me know if I didn't address your question. I can try to elaborate a little bit more.

I definitely was around, and the assessment author at the time, This was around the time that skates were pulled out from the other species complex. I don't remember the date for that or when that happened. And it was just over conservation concerns, primarily basically not knowing anything about the life history of these stocks, you know, their fecundity being unique and all of that. And so we went with precautionary approach, and that was very discrete blocks. And there is a figure in the original document, in Lee's document— he's available online here too to answer direct questions on it— but there's a figure that rescales the, the one that Chris had up just a second ago to show the trend, and I think it's relatively flat.

So in terms of trying to diagnose a trend in the Western Gulf. I think that's what Rob's trying to get at there. Thanks.

Sherri.

Sure, I've got a— this goes beyond your minutes, so I don't know whether it was discussed, but I'm just wondering, so, you know, looking at the survey variability, was there any discussion of sort of how the variability would be used to set the amount that it can deviate. So, um, and so that's, I guess, one question. Um, the second one then that goes along with it is, was there any discussion of variability in survey estimates for species that actually are highly dynamic, and it may actually be biomass changes versus some with a life history where it's like rockfish and they're poorly sampled and the variability is from, you know, error in estimation.

Yeah, those are great questions, Sherri. I— the team didn't get into that level of detail because it was pretty evident we weren't putting forward a recommendation on here's the formula to use based on this and this. It was general discussion. And I think your points You know, I think it'd take a bit of a, a lift, but it's doable to separate out those things and use survey uncertainty. But there's other methods too.

I just, I think we tried to capture where the team naturally went and discussed, and we kind of stopped it at some point because it's like, well, we're not putting forward— this was, this was meant to be kind of change a little bit, yes or no, and we had some concerns just because of that Western Gulf. But then we also decrease in biomass, but then we also prefaced with that, you know, these are— it's on the edge of its range potentially. You know, it's a really low biomass in that area. So to answer, like, your question on that, I don't think the team went into any level of detail to try to get answers to that level.

Thanks. And no problem at all. I wasn't able to listen to the discussion, but I was guessing you probably didn't get that deep. Thank you. Yeah, definitely.

It was probably because the air handler was so loud in our meeting, but Jason.

Yeah, I'm getting back to sort of— I appreciate the idea coming up with some sort of simple and tractable method to do this. And thinking back to Martin's question about more information on the stock structure or lack thereof. I'm wondering, did the team have any discussion, or is there information maybe available about whether there's interaction with the Bering Sea, and that might be sort of another sort of flexibility thing to consider with this stock?

Yeah, great question. I think it's worth also prefacing that we had a short discussion when Lee presented his stock assessment, but we did a follow-up and didn't have Lee in the room with us on that, so we really didn't have the expert there to kind of address some of those questions. And so I don't think we got into that level of detail at all.

Okay, I'm not seeing any additional questions. Is that all you had on skates and BRDs? That's all we got.

Okay, I'm going to suggest that we take a 15-minute break and then we'll come back and take public testimony on skates if there is any and have our discussion. So we'll be back at quarter after the hour.

Do you think it's helpful to bifurcate this discussion explicitly and say we have a discussion on skates and we have a discussion on the arguments when we get back to the world themes, or do you think we should really just laser focused on states make a decision.

You know, I think since they mentioned two-step process, all species, I kind of think that we say reference that. We were told in the presentation that this would be a two-step process and that it was, you know, considered across species and then give her comments. I mean, I'm kind of tickled that exactly what we hoped or thought would happen.

Sorry, I didn't mean to yell. Um, I'm saying, someone's mic is on.

That was pretty cool. Look at that. So mine is on, but the light is off.

You guys on the mic?

Testing, 1, 2, 3.

Okay, we're going to come back into session.

We, um, I will reopen for another— any chance to ask any additional questions of, uh, plan team chairs, staff, any follow-ups to this BRD discussion? It looks like we're going to have both a discussion of BRDs with regard specifically with regard to the skate assessment and also likely a slightly broader discussion of BRDs, and that's fine. We can have both of those now. But this is the last opportunity to ask questions, analysts and staff.

Okay, I'm going to open this up for public comment, um, on skates and BRDs.

Seeing no public testimony, we can go into our discussion. And, uh, first up we've got Ali.

Yeah, thanks, Ian. Um, so I think we can start out by thanking the author and the plan team for their work and their comments on this assessment. The skate complex in the Gulf is assessed on a 2-year cycle with the previous full assess— operational assessment in 2023. A catch report is currently scheduled for Gulf skates in 2027. We had no public testimony, and this is setting final specifications for this stock.

So this complex consists of 3 components: big skate, longnose skate, and an other skates category, which comprises a group of 13 other more rare species. All of the components of the stock are assessed in Tier 5, which uses survey biomass and a natural mortality rate to determine harvest specifications. Big and longnose skate have Gulf-wide OFL and ABC, which is then apportioned into 3 regional BRDs for the western, central, and eastern Gulf. The other skates component OFL and ABC are Gulf- wide. In terms of assessment methodology, two REMA models are fit to— for each of BIG and Longnose Skate survey biomass.

One is to the Gulf-wide survey biomass estimates to determine an OFL and ABC, and then a second REMA model is fit to individual region-specific biomass estimates to determine the proportion of biomass in that region, which is then multiplied by the Gulf-wide biomass estimate to get the BRDs. For other states, a single REMA model is fit to those Gulf-wide estimates. Natural mortality for all 3 components is assumed to be 0.10, and the OFL is determined by multiplying that natural mortality rate by the exploitable biomass estimate from those REMA models. And the maximum ABC is 75% of OFL.

So this 2026 assessment is an update assessment. Updated data include updated 2025 and 2026 catches through August of 2026, and then additional biomass estimates from the 2025 Gulf Bottom trawl survey. No changes were made to the assessment methodology, and none of the 3 components were experiencing overfishing and being in Tier 5, an overfished determination cannot be made.

Survey data indicate that in general, Gulf-wide biomass trends for big and longnose skate have been stable in recent years, though there are some relevant deviations at the regional scale. Notably, the 25 big skate estimate in the Eastern Gulf increased by 92% from 2023 And the Western Gulf estimate for longnose skate continues a recent, a recent kind of multi-year decline, though the author notes that this region has consistently lower biomass estimates and higher variability over time. Biomass estimates for other skates increased by 29% in 2025, though still remain at relatively low levels. Following a decline in the mid-2010s. Uh, all of the risk table scores were at Level 1, um, indicating no elevated concerns for this data-limited complex.

So in terms of our harvest specification recommendations, I think our minutes should reflect that the SSC concurs with the Gulf, um, plan team and author recommendations for OFL Max ABCs, for the Gulfscape complex and with the BRDs, for big and long-nosed skates.

Um, the plan team had a number of suggestions for analytical improvements to the assessments, including checking the design-based standard errors for other skates, which seem, uh, artificially tight in some recent years, um, and using an additional error option that's available in RIMA. And I think the SSC should support these suggestions from the plan team. I think we should also reiterate our comments related to the need to reevaluate the natural mortality assumption for this assessment, and that more information is needed on stock structure. Specifically, we recommended in 2023, and I think Chris brought this up in his presentation, specifically we recommended in 2023 that the Gulf and Bering Sea skate authors coordinate to update the stock structure template for Gulf skates. And the plan team has supported these suggestions generally as well.

And as our discussion for BRDs is coming up here shortly, I would suggest that this is a very high priority for this stock.

All right, getting into BRDs and comments specific to skates.

So the, the council requested SSC advice regarding flexibility and spatial apportionment for Gulf skates in October 2025, specifically for big and longnose skates. So as I noted earlier, this is final specifications for this complex. And this will be the first time that we provide feedback to the council on flexibility and spatial apportionment. And so I think our minutes should reflect at the beginning here that this process, um, will be further developed and potentially standardized, uh, as we move forward.

Specifically, the council requested SSE advice on whether using last year's BRDs, um, and By this, I would interpret this as the previous set of BRDs, which in this case is from the 2023 assessment, or small deviations from the current year BRDs would likely to— be likely to result in biological impacts. And I would note that we identified our role in this process is to characterize the associated with exceeding a specific BRD, um, and that this risk is, um, a combination of the risk of local depletion, and that is related to potential stock structure, um, and also uncertainty in apportionments. So I think the plan team had a really excellent discussion of this in their report for long nose SCATE, and Chris went through some highlights of that, so I'm not going to repeat them here. I feel there is little we could add to their discussion, though maybe others around the table have additions. One addition that I thought of was that while the survey estimates in the Western Gulf have been declining over the last few years, estimates in the Central Gulf and the Eastern Gulf have been increasing concurrently.

And the Gulf-wide survey estimate is also slightly higher in general.

The GPT focused their discussion on longnose skate, but I think that our minutes will reflect that for big skates, many of the same considerations apply. For big skates in particular, the regional survey estimates appear to be stable or increasing. Model-based estimates indicate similar trends, although I would note that projected biomass in the western Gulf for Big Skate decreased by 23%. But overall, the Western Gulf estimates for Big Skate do appear to be more variable, similar to Longnose.

Given that, and this was just me looking through the tables in the assessment, given that the regional-specific BRDs were exceeded for Big Skate only once and Longnose Skate only a few times, and over the last 10 years of complete catch data, we do have a relatively infrequent frequency of exceeding the BRDs in recent years. Our survey estimates in general seem to be increasing or stable for both species. The exception, of course, is Western Gulf and Longnose, and that risk table scores for the assessment are all at level 1, which indicate no strong concerns external to the assessment model.

Um, I would suggest that our minutes, um, that the SSC suggests that the risk of a biological impact to these 2 components of the Gulf Skate complex is low overall if the council wishes to use the previous set of apportionments or, um, a minor deviation from the 2026 BRDs. That would be the 2027 BRDs, so the updated BRDs. Sorry, typo in my own notes.

The main impact of using the, the previous year's BRDs would be an increase in the Western Gulf for longnose skate from 55 to 104 metric tons. Chris outlined that in his presentation. I would note that the bottom trawl survey generally samples skates very well, although some of the more rare species in this complex are relatively patchy in the environment and more variable in the survey data.

I think our minutes should note that the risk for big skate appears to be lower than that of longnose skate, given the recent increases in survey biomass across regions for this species in recent years. And I think the information we have at hand suggests a relatively higher risk for longnose skate. That includes the general lack of stock structure information for this component of the complex, declines in that particular survey region over the last few years, and a higher variability in survey biomass in the western Gulf in particular. A lot of this is also brought up in the Gulf Plan Team report. However, there may be some information that might offset this relatively higher risk for longnose skate, such as the overall increase Gulf-wide, the overall increase in estimated biomass in other regions, which, lacking any concrete information from a stock structure template, may suggest some movement within regions, or among regions rather.

The survey generally samples skates well, and retention of long-nosed skate has been kind of hovering around 10% in recent years. And the plan team also suggested that declines in this region alone are unlikely to impact the core population.

I think our minutes should also suggest that if the council does choose to deviate from the recommended BRDs that this approach be reevaluated with new Gulf survey data in 2027. And further, I would also suggest that if the Council does choose to deviate from the recommended BRDs, that we would request that Gulf States be prioritized for an operational assessment in 2027.

All of this kind of leads me to suggest that we do generally concur with the plan team recommendation to exercise some caution for longnose skate in the Western Gulf in particular, and suggest that the risk of exceeding the BRD in that region for that species is higher than in other regions for longnose skate, or overall for big skate. Unfortunately, this is obviously the BRD that is likely to be most limiting for the fleet.

I have some general comments on BRDs, and I would— I'm assuming other folks around the table here will also have comments as well. I would suggest that we include these in our general groundfish comments, as they will be applicable across stocks. We'll be seeing more stocks in December where the Council has requested, uh, advice on flexibility and spatial apportionment. Um, and so, um, I'll kind of kick off this bigger discussion with my contributions. Um, I think the SSE should strongly support the development of quantitative approaches or explorations to inform risk discussions of flexibility, um, and spatial apportionment.

Um, in the past, um, and I believe this was December 2024, um, the SSC has requested that survey coverage and representativeness and uncertainty be explicitly communicated in our apportionment recommendations. I think our minutes should reflect that using survey biomass uncertainty to bound regional apportionments, uh, seems like a good piece of information that is rooted in the spatial distribution of the stock and scientific uncertainty around that. However, I would caveat that by noting it will be less informative for species that our surveys don't sample well or sample infrequently.

I think we can also note that, you know, this is potentially only one piece of information that might inform a more thorough discussion of the of exceeding any particular one BRD.

Um, I think the SSC notes that the plan team asked for— asked, um, uh, two process questions that do seem to be directed at the council, of course, regarding the timeline for providing or reevaluating advice for flexibility and spatial apportionment. Um, I, I think I would just note that The SSE's advice may not change substantially in between assessment cycles, which are typically tied to survey schedules, but obviously we stand ready to provide an assessment of risk of exceeding the BRDs as often as the Council requests.

I would also suggest that our minutes include that a tracking mechanism be developed for stocks where the Council does choose to deviate from BRDs? And, you know, perhaps this could be modified slightly to include Chris's comments earlier on clearly documenting in the assessments how the BRDs have changed over time. And just particularly if the Council chooses to change tax from kind of the default BRDs. And this is very similar to a request we made in December 2024, where we request assessment authors' help with tracking these changes. I think this can be done, at least in the, the existing tables in the skate assessment.

I think it would be very simple to implement this, but it does sound like there are some differences across species. And so perhaps we could get the plan team to kind of help a standardized kind of mechanism here that keeps, keeps track of it for, for every stock where this is applicable.

And last, I think we should reiterate our earlier recommendations that deviations from BRDs should be applied judiciously, as we do still feel the spatial distribution of the stock is best represented by estimates from large-scale fishery independent surveys, and that persistent or long-term deviations from this may negatively impact the stock at hand. And those are my comments.

Great, thanks, Ali. Next, we'll go to Curry. Yeah, thank you very much, Ali. I'll start with my thoughts on the assessment and move to the BRD discussion. I think I concur with everything that you said and wanted to reiterate the point that has been brought up before about the importance of revisiting natural mortality for these stocks.

I noted that that value was originally set in 2003, which seems like a long time ago, and the science has matured in between. And hopefully in the future, there can be more frequent updates to these critical values that for Tier 5 species drive allowable catch directly.

I wanted to highlight that this review of natural mortality should consider the literature on maximum age and age at maturity for skates from Alaskan fisheries and surveys where possible, and if not available, from other regions including British Columbia and the West Coast where there has been perhaps additional research conducted, and also to consider assumptions or estimates of natural mortality from assessments for related skate species in other regions. I'll note that on the Pacific Coast, based on a bit of background digging, I think for big skate, they assume a substantially higher natural mortality rate than what's assumed up here. With that being said, the Pacific Coast assessment does highlight that the maximum age seems to increase for individuals as you go from the Pacific Coast up into British Columbia, so they may be even older up here, and maybe natural mortality 0.1 is appropriate, but I think that thorough review is probably long overdue.

In terms of the recommendations and the buffer, I concur with the plan team recommendations and, and, uh, what you highlighted. Uh, with respect to BRDs and flexibility therein, um, I, I agree we, we should be clear that this is intended to be a rare event and requires careful consideration. Um, and when we think about the, the pieces of information that help us identify whether or not there's a conservation risk in terms of deviating from that, it's of course the, the current stock status and recent trends, The overall level of conservation concern, and I think we can link this to the question of mortality rates and retention rates for these species in particular. Also a consideration of the movement ecology of the species. I note for skates, there's a couple examples.

There's a paper from 2010 that looked at tagging of skates, big skates off of British Columbia, and a more recent review from Matt Siski that sort of show that skates actually don't move that far. But probably something important to consider in the context of BRDs versus a more highly mobile species, like sablefish, for example, if we were to take an example.

I think it's also important to consider the history of tag utilization in neighboring regions. For longnose skate, of course, the western area is the area perhaps of of greatest concern in terms of, of, um, uh, the BRD that's implied. But in the central region, the catch has been, I think, below 50% of the TAC in recent years. So even if there is moderate levels of movement, there's certainly, uh, underexploited, or at least, um, catch below TAC in the neighboring region, which should give us, um, perhaps a little less concern about an overage in the western Yeah. I think in general, the plan team had some interesting discussion and was brought up here that generally speaking, how our level of flexibility with respect to BRDs is in some way proportional to the uncertainty in the biomass indices on which they're based.

I mean, that's just the reality of it. If we have data that are— or if we have a species for which a survey is not particularly well indexed, I think there's a general idea that the flexibility should be proportional to the uncertainty in the regional biomass indices. I'm not sure that there's a direct mathematical way that we can establish that, but that general idea I think holds. Thank you.

Thanks, Curry. Other comments? Sherri?

I guess I'm going to be the devil's advocate on this one, um, and if nothing else, it'll bring some discussion. Um, there's a couple of things, and this is actually— I'll just give a couple of details first, and this is sort of in response to Ali. So the SSC, this has come up. I'm going to try to put the context, uh, timing of this in context. We talked about this extensively in December of 2024.

We actually discussed it under Dusky Rockfish, and that was because this was actually when we first proposed to the council about flexibility. And is there a way, instead of waiting for an assessment to catch up and miss a whole year or two, we were wondering if there was any flexibility that they could have intact that would allow us to make changes on cycle. And so all of that was done before the council made their motion. And so while we did describe our role at that time and what we thought a role would be, I'd be hesitant to cite our December minutes because that was before the council took any action on it at all. Um, and so if we want to describe what our— what we see as our role, I would do that this meeting and have that in our minutes.

So that was one. The other thing was at that time we did discuss survey variability considerably. That was one of our first ideas that we put forward. That actually didn't make our minutes because we talked ourselves out of it, and instead we ended up saying that— oh no, I had to Um, we ended up saying that subarea limits— and again, this is before the council took action— would be informed by the best available science, including stock structure and risk of local depletion, while offering flexibility and tack apportionment among subareas. And we noted that simple methods for specifying appropriate subarea limits could be considered.

This would be one that they made in his example. We just didn't specifically name it like you had you talked about, Ali. Um, and those would be developed, compared, and reviewed by the authors and the plan teams as well as the SSC. So I think one thing that's pretty fantastic is that, that this was a sort of an original idea early on, and I'm— even though that plan team was surprised, I'm really glad that they had a discussion. Um, and I am with you that I think that having metrics would be valuable.

Using metrics in some way. I really appreciated actually Sarah's sort of extrapolation and saying that they talked about a process first looking at stock structure and then moving to it. That wasn't in their minutes though, and so I'm hesitant.

I think when we ask for, you know, further development and consideration, which like Chris said it would help them to have some time to think about this, that maybe we ask for— I think you mentioned it as well, Allie— a more developed process on how best to do this and how to use a metric. And the things that they could consider, obviously, are the very— how to go from the variability to some metric for how much can change, whether that would differ whether that would differ for species that are highly variable but the survey does pretty well estimating them, and for species that don't change much and there's a lot of survey error. So I think that those would be clarifications that would be helpful. So that's general process. As far as skates, I have to say I'm, I'm I'm really quite uncomfortable personally in sort of giving the okay to use the 2026 apportionments and allow and tell the council that there's flexibility within those.

And I think the reason we talked in the meeting today a lot about the flexibility, but in the planning minutes They— and actually, let me pull— there were a number of things that they mentioned. They said the team was concerned about excessive catch in one area could result in localized depletion. That is one thing that we highlighted as being a key consideration, is localized depletion.

They also warned against sort of dem— selection against demographic traits. Like different growth, movement behavior, and life characteristics, which are more likely to happen at the end of a range. One thing that they did mention in, in questions and discussion today was that yes, most of the stock is not in the Western Aleutian Islands. And I really appreciated your comment, Kari, that even if there's minor movement, that the fact that they're not fully harvested in the Gulf will hopefully help the Western. And so that's actually— that was one thing that makes me wonder if I should back off.

But the other thing is they didn't— there's not a whole lot known about stock structure, and they didn't— we haven't yet seen a new stock structure template. And so the two things that we named as the biggest considerations in December were looking at stock structure and localized depletion. The minutes for the plan team specifically say it could result in localized depletion, and we don't know much about movement and don't have a stock structure template. So I have to say that I'm having— allowing that big of a change in BRDs, you know, given what we have, makes me uncomfortable. I'm willing to defer if the SSC feels differently.

You know, if we were going to go with— I don't think that I would have concern given the things that you've mentioned in plan team has mentioned, I wouldn't— I would not— I think that there is some flexibility.

The problem is that we haven't had time yet in this process to define what small is and how much might be okay. And while we said to use the variability, we haven't quite figured out how and whether that would be consistent. And the plan team was surprised by this. It wasn't something that they had had time to prepare before this meeting. And so With all of those things in place, my tendency would be to say that personally, I think that there's concerns with using the previous year's BRDs due to the things that are named in the plan team's report.

So we're not— I wouldn't endorse that at this time and say yes, there can be some small changes, but we're not yet developed enough in this process to be able to identify them for this round.

Um, I don't know, but I'm totally open. I wanted to put it out there that there's concern, but I wanted to hopefully get some more discussion from people on the SSC.

Alli. Yeah, thanks, Ian, and thanks, Sherri, for comments. Um, I think, um, maybe we're not so far apart as you, um, might be suggesting here, because if I, if I could maybe clarify my, um, my main points is that I, I do think that the risk overall is relatively low, um, but that the relative risk risk in the western Gulf for the longnose skate is higher.

And I also suggested that we were, we were in agreement with the plan team that the council may want to exercise some caution in that region for that species in particular.

And what's the other thing I wanted to suggest?

I, um, so, so I, I don't think we're that far apart, um, on that, um, piece of it. Um, so you can let me know if you agree with that or not, I guess. Um, I think, I think my other point is, um, you know, that, uh, I guess you describe your general hesitancy to to move forward with this, um, at all for skates because we lack some key information, um, the stock structure template, which we have requested for a couple of different cycles now, um, and, um, and, and that I, I don't disagree with you that that is a really key piece of this discussion, um, but, um, you know, we did We did say in 2024 that we were ready to move forward with this process, um, and, and, and, and kind of have this iterative back and forth, um, potentially for a number of years while we kind of settle into this process. Um, and so I'm really hesitant to say on the first species where the council requests us to provide some advice on this to them, um, that we, that we don't respond to that at all. Um, and, and maybe that's just a matter of, you know, kind of clarifying that, you know, we are really lacking some important information to fully form this discussion for this particular species.

I think we're going to be in different places for each species that we do this for though. Um, and so, uh, hopefully, um, I, yeah, I, I do think it, that it's important to respond to the best of our ability at this point, but note, caveat that with, you know, the lack of information that we clearly have. So hopefully that helps us come together a little bit more. I don't think you're playing devil's advocate. A little bit.

Go ahead, Sherri, if you want to follow up. No, thank you for that. And I definitely do agree that I think that there's a difference between risk in one region and possibly with individuals that may have different growth characteristics or whatever and the overall population. That's the piece that I had a really hard time reconciling. Since we had mentioned localized depletion as one of our concerns, that's why I sort of bumped it up and ended up more concerned.

But I think that there's flexibility in there. You got to make a call somehow.

Totally understand not wanting at the very first species to say, nope, we aren't ready to move at all.

I think that thankfully the council has been— I, I don't know what the response on this issue would be, but I will say there has been understanding as we prepped for this meeting that, um, we might not be able to answer all questions for BRDs this year. Um, we had actually tried to find room on our June schedule because the chairs mentioned that we needed— one thing that would be valuable is to have a BRD discussion on process before the plan team met so that they would have some direction and that we could be prepared for this fall. And then when there wasn't time for that, then we also named that there might be important to have some time on our schedule to discuss BRDs and sort of changing by a small amount. And recognizing our packed schedule due to the shutdown, um, the direction, you know, that we've received through council staff is they fully recognize that we may not have time to talk through and decide what a small amount is this year, and that that's okay to take the step that we can. And so I think that, um, I feel a little— well, you never want to say know, I also want to acknowledge that we haven't had time to develop this yet.

And so I'm not sure— I'm a little concerned that we not be pressured to, to accept it if that's something that we're not comfortable with. So I think we really do have the freedom to make our choice. That doesn't mean our choice is to be where I was on it. It just means I think we do have the freedom, um, to do that. The other thing is that a lot of times interactions and communication go sort of back and forth between the SSC and the Council.

So your comment, Ali, was, you know, we agree that we may need to— that the Council should exercise caution.

The problem is the Council asks us how much caution is safe, and so when when we're actually answering questions to the council, giving SSC reports, the question is for us, how much is safe? And so for us to say you should be cautioned but not tell them how much, I think is gonna be kind of a role reversal to what they're asking for. And so I had, but I had a really hard time being able to come up with an answer for them at this point. And so, I realize it's punting on one species, but that's part of why I sort of got— we're kind of in a pickle because of time, and, and I was willing— I fully would expect that this may change as it develops. So anyway, I don't know if that helps at all.

Are there others that want to weigh in?

Allie, if you want to do a follow-up, that's fine. Yeah, thank you. Yeah, those are some great points, Sherri. I think, I mean, our I, I also got in, in writing these minutes and, um, you know, starting to write our report on this and, and looking at this information, got hung up on the idea of how much is, how much is small. Um, and I guess I did come back to this idea of, um, of, of our role being to present a risk level, um, associated.

And I know that risk is tricky because we have different definitions of risk, and, and some are, some are legal, some are associated with our risk tables, but the risk of exceeding a BRD is what I am specifically referring to. And, and I felt a lot more comfortable using that as a relative framework for the different BRDs for these two species in three regions. So we're really talking about 6 BRDs here, um, I really felt a lot more comfortable using that as a framework, um, to, to, um, characterize our advice to the council. Um, and so I think that, um, you know, even though I say that overall, um, I think the risk of biological impacts to the population if, if one BRD in one year is exceeded, is— I do feel that one is relatively low overall, um, that for this region in particular where the council is most likely to desire a change, um, that the risk is higher there for a number of reasons, including the lack of information on movement and connectivity among the regions and, um, But there's also kind of some flip sides there as well. And so it's this kind of, you know, it's a discussion, it's a qualitative discussion.

We don't have a number that we can say, yes, this is safe. And we're— I mean, are we ever going to be there for all of the species that we manage? Probably not. Maybe we can get there for some species eventually. And I think that, you know, kind of comes back to the importance of this kind of iterative process.

Process and what we learn through it. So hopefully that response helps.

Kari. Yeah, as part of that iterative process, I think hopefully we can provide some clear guidance to authors and to the plan team about the key areas in the event that this arises. Again, acknowledging that it should be a rare event, the key areas of consideration that should be articulated. In describing the risk both at local scales and for the stock as a whole across regions. And I think we can do a pretty good job of that in our minutes.

Shiri.

I really appreciate both of those. And I also, I respect that you are leads and did the biggest digging, and that you also didn't necessarily end up feeling the concern that I do. And I, I think I respect that. And I'm wondering, what, what if we say in our minutes, um, go with your recommendations, and then also include in our minutes that we had discussion and there was actually— that there was, you know, some discomfort on the SSE with doing this, there hasn't been time yet to establish a full process how to do this and how to, you know, I would have been comfortable if there was some way to do some smaller amount, you know, but we just didn't have that. So we hope to have that in the future and just acknowledge that we didn't necessarily have all the tools, you know, at our disposal for there yet.

And then I think that your, your comment, Ali, with, you know, to risk going over one BRD in one year, I think that, and the iterative process, I think that's huge. And what that indicates is that if we had some concern this year, it doesn't mean we don't have to do the same thing next year when we have more of a process. And it also gives the council a heads up that we're working within the process and the time that we have, and, um, and that we're acknowledging that we're saying, okay, for one year, you know, let's go ahead and allow the flexibility, and we can reevaluate when this is further along next year, or the two years, or whenever the next assessment is.

Is that something that sounds reasonable?

Yeah, I personally, I think that there is clearly some degree of flexibility here, and it wouldn't quite be the right signal to say there's none just because we don't know what to do with it. So opening the door that yes, we, we see that even though there's a lot of lack of information, there is definitely some flexibility, particularly in a 1 BRD and 1-year type framework with no guarantees that we would reach the same decision next year or for even a very similar stock. I think for me, that's a comfortable place.

Okay. So just to be clear, and the keywords that I think in our report of whether we say, um, that the— there's a low risk if the council chooses to use a small deviation from the 2027 values, or whether we say there's a low risk if they choose to use the previous BRDs or a small variation in the previous values. That— are we going to name the, the previous BRDs or just say small value for which we're clearly not going to reach a conclusion on? I just want to clarify and make sure that we're all on the same page there.

I mean, I can— maybe I can go first. Um, I mean, the council motion specifically requests advice on either of those, so I'm inclined to include either, um, and just not specify which one we think might be better.

I— that's where I landed originally. Sure. Yeah, I don't think that we evaluated a small amount. We didn't decide what a small amount was. So I— it sounds— that's why I felt like I think what we're doing is saying we're okay with last year's BRDs, but there was concern among the SSC in discussion We would have preferred to have been able to talk about some small amount, but we don't have a process yet.

So for one year, I think what we're saying is we're okay up to the BRDs. Otherwise, they're going to ask what a small amount is and we're not going to have it. Or they have to decide, which is what they're asking us to do. And I think that neither of those are totally helpful.

Okay, I think we're there then.

Um, so I will summarize our discussion. Rob, I just wanted to—. I fully support that discussion, and I just wanted to add to, as far as the, um, you know, that the uncertainty is also related to these boundaries too. So if we can just include some reference to that. Um, and, and I guess my concern here is, like, for example, is, I mean, the reference to localized depletion.

Um, unless we know there's something genetically, biologically, ecologically relevant about a boundary that's already in place, then the spatial extent or the spatial relevance of a local depletion is unknown. And how much movement there is in between. So, um, it just— I guess just to add a spatial component to that too in terms of the boundaries is, is this a summary of that statement?

Thanks for that, Rob.

Okay.

So for skates, we're going to thank the authors and the plan team for their work on the skate assessment this year. We'll note there was no public testimony. This is going to be final specifications for big, long nose, and other skates, all of which are Tier 5, and we will not see these again this year, even though the council won't take them up until December. We'll note that, uh, big and long nose skates have region-specific BRDs, and the other skates are are assessed go-wide. This assessment includes the updated 2025 biomass estimates and updated catches.

We will note that overfishing did not occur and that we don't do a status determination because this is a Tier 5 assessment. Overall, we'll note that the trends are relatively stable with some changes at the regional level. We'll note that there were no elevated risk scores, and we will support the recommended plan team and author OFL ABC and BRDs with no adjustment from the maximum permissible ABC.

For research recommendations, we will support the plan team recommendation regarding the— yes, we will support the plan team recommendation regarding further work or further presentation of the REMA modeling evaluation of that, the reevaluation of the M equals 0.1 assumption, which we'll note should include evaluation of maximum age from Alaska and elsewhere, as well as considering assessments for related skate species from other regions, particularly big skate on the West Coast for comparison. We will also note the ongoing request for updated stock structure templates. With regard to BRDs for this, for skates specifically, we will note that the council requested advice on flexibility for big and long nose skates. We will highlight probably several times that this is our first time through this exercise and that it will certainly develop as we move forward, but that we have identified our role currently as characterizing the risk of exceeding a particular BRD. And this would be a risk of local depletion, and this would be primarily informed by our understanding of stock structure and uncertainty.

We have a laundry list of things that we— that would be considered when undertaking this, including current stock status and recent trends, potential discard and discard mortality rates, movement ecology the species, tack utilization in neighboring regions.

We will note some of the relevant trends for these stocks, that namely that the survey for longnose skate in the Western Goa has declined, but that the Central Goa and Eastern Goa has increased, and that there are similar considerations for big skates. We'll note the low risk table scores. We will highlight for the council that we've had an extensive discussion about how to approached this, that we have at least initially had even some disagreement on exactly how we should proceed and whether there could be flexibility for skates. We'll note that we really have not had time to develop this as a process yet and that we may be in a different place when we repeat this for other species and/or repeat it for this species again next year or over time. However, we will suggest that the risk of biological impact to these skate species is low if the Council chooses to use the previous 2026 BRDs or a small deviation from the 2027 reported values.

While we note that there may be some small changes, we cannot identify the magnitude of what would be considered a small change at this time due to lack of information specifically on stock structure and movement. However, we will emphasize that there is likely a low risk of exceeding a BRD for one year.

We will also highlight that of the various— of the 6 BRDs, we will agree with the plan team that long-nosed skate in the western GOA is probably the one BRD of most concern.

In justification for having some flexibility, we will note that movement is certainly possible among these areas, that there's relatively low retention of skate species. Species, and that declines in one region alone are unlikely to affect the core population, particularly the western Goa, as it's the tail of the population distribution.

We will recommend moving forward that should there be a deviation from the 2027 BRDs, um, that we would recommend reevaluating our, our position on this based on the new survey data in 2027. And that we would request an operational assessment be conducted in 2027 if there is a deviation.

For BRDs in general, we would like to see— we'll obviously a lot of that discussion maps over into our general discussion of BRDs, but we would like to see more evaluation of the process. We'd like to see consideration of survey versus actual population variability.

We heard some description about a two-stage process of looking at stock structure and then at the range of flexibility there might be once some conclusions were made on stock structure, and I think we would encourage further work in that direction.

We will try to provide some guidance to authors and plan teams on many of the factors that we already listed. Which should— which could be described more thoroughly to inform these decisions. And that would obviously come from the author up through the plan team to us. We will note that using survey uncertainty, at least for the range of potential flexibility, makes sense as a primary basis, but this may not work well for poorly sampled species, and it's only one piece of information that would be used to inform the risk of exceeding a BRD.

I think I've already said it, but we will highlight again that our advice may change appreciably between assessment cycles as we— as this process matures. We will ask for, in general across stocks, we will ask for a tracking mechanism to be developed for stocks where there are deviations from BRDs, and I think this actually might be covered to some degree or overlap with our previous recommendation about having a table of recent BRDs.

And we will also note that uncertainty about the risk associated with exceeding a BRD is closely related to the BRD boundaries themselves, and that we will encourage consideration not only of flexibility in the BRDs but also the definition of where these BRDs applied simultaneously. And then I think we will, we will highlight also that reiterating our previous comment that deviations from BRDs should be applied judiciously, that we still feel strongly that surveys represent the best information on stock distribution and that persistent or long-term deviations from the stock distribution may affect the stocks in the long term.

Sherry, I just have one question, and it may be for Ali. Um, we talked about having possibly an operational assessment in 2027. Is there going to be survey data that year to inform it?

There will be. Okay, great, thanks.

Okay, excellent. That concludes our discussion of skates. We will next be taking up harvest projections, but first we will take up an hour of lunch. See you at quarter past 1.

Okay, welcome back, everybody. We are closing in on the end of our agenda and going now to— going now to our GOA harvest projections. You're not supposed to laugh when I say we're getting close.

So back to staff and plant team chairs.

Thanks, Ian. I'm glad someone's still laughing. The next part of our presentation is harvest projections. This is the only slide we provided for the SSC presentation. We saw 4 different HPs, you saw 5 different documents because the northern and southern rock sole was done as part of the shallow water flatfish documents.

I don't want to confuse anyone there. Basically, the team supported maximum permissible ABC for the flathead sole and dusky rockfish stocks, and because of mixed tier levels, we can't say that for the other 2, but we accepted the author's recommendations based on the the harvest projections. And the one note that the team had was that they noted the shallow water flatfish had gone 4 years, and so they were suggesting an operational assessment be done in 2027, which I believe is the plan by AFSC. That's what I have for harvest projections.

Do we have any questions for Chris?

We're going to go ahead and just finish this one off while we're on it. So we will open for public comment, and we do have 2 people signed up for public testimony.

I realize our schedule's been pretty jumbled, so hopefully they're available and have been following along. If not, we'll get— Yes, this is Ben. Can you hear me? Excellent, we can. And so first up we have Ben Lay.

Welcome, Ben. Hey, thank you guys. And yes, we have been following you around and we're scattered here as you guys are, I think, but appreciate the opportunity to speak here today. So I'm just going to highlight the, the letter I submitted first on the behalf of PFC and then touch on another couple topics here. But first and foremost here, we, PFC respectfully requests that the council consider a small interim ad hoc reallocation of Pacific cod to the Western Gulf of Alaska for the A season only.

Specifically, we request that the council consider reallocating 3% to the Western Gulf for the winter fishery as a conservative, conservative temporary measure while stock assessment and management framework continue to evaluate Pacific cod movement between the Bering Sea and Gulf of Alaska. Recent scientific discussion and presentation material from Pete Holsen and Steve Barbeau support the conclusion that some portions of Bering Sea Pacific cod biomass is present in the western Gulf of Alaska during the winter A-season fishery. While the exact magnitude of that movement remains under review, the available information strongly indicates that the fish harvested, harvested in the western Gulf during winter may include biomass that migrates to the Bering Sea during summer survey periods. The requested 3% reallocation is intentionally conservative based on the range of potential migrating biomass discussed in the recent analysis. A 3% adjustment reflects a cautious assumption of approximately 30% migration, even though the actual movement may be closer to 50%.

This request is not intended to overreach or prejudge the outcome of the formal modeling process. Rather, it is intended to recognize on an interim basis the biological reality that some Bering Sea-associated Pacific cod are available to and harvested by the Western Gulf Fleet during a season. From a biological and management perspective, this interim adjustment would more closely align harvest accounting with what is likely occurring on the water if a portion of the winter Western Gulf harvest is composed of fish that would otherwise be reflected in Bering Sea biomass during summer surveys, then a small, a-season-only reallocation is a reasonable bridge while the formal assessment work proceeds. The council's own assessment process recognizes that Pacific cod movement and distribution are important factors in setting sustainable fishery limits and evaluating stock structures. This action would also provide meaningful near-term support for local Alaska fishing communities and small vessel participants in the Western Gulf of Alaska.

A 3% adjustment would be very small change relative to the scale of the Bering Sea fishery, but it would make a substantial distance to the Western Gulf fleet during the winter fisheries. The benefit would be concentrated among the, the local vessels and communities most directly affected by constrained Western Gulf opportunity because the formal model and, and any resulting management change may be a year, at least a year away, if not longer. An interim, a season-only allocation will provide a practical and precautionary way to address the issue now without committing the council to a permanent allocation decision. It would acknowledge that the best available science apply conservative assumption and help ensure that Western Garth harvest opportunity better reflects the seasonal presence of Pacific cod in the area. For these reasons, we respectfully ask you to consider a small interim ad hoc reallocation of 3% to the Western Gulf of Alaska for the A season only.

That kind of sums up that portion of my testimony. And then, so then I've been taking a couple notes as we've— I've been listening to you guys here. First and foremost, I guess I want to clarify when I did hear the Western Gulf catch reports earlier, I think yesterday at some point, but I was just going to further elaborate on that. The cod did not get fully harvested. That was kind of due in fact to the majority of the fleet taking time off to represent themselves at the Board of Fish.

So that's kind of, it wasn't lack of effort, it was kind of the timing of the fish run. And, and kind of not being able to be in two places at one time. And then, and then Pawlik, um, we didn't have processing there. There was an abundance of fish there this winter, and when the guys actually quit fishing, the CPE was very high. It's just we ran out of processing time.

Um, and I guess finally to close, if I got a couple seconds here, I kind of wanted to talk about the, the Stellar Sea Lion restrictions and apportionment a little bit, you know, for the trawl sector, uh, specifically like for cod, we get a B season apportionment that is basically just a rollover into another sector because the, the, the target species is not aggregated to even fish in the fall. In the past, when we've kind of brought this up, they've talked about a Steller sea lion issue, like it being an SSL issue, and it's my understanding that right now there's directive for the North Pacific region to review all the Steller sea lion restrictions. And I see I'm out of time, but I guess I can leave that, leave that there.

Okay, thank you for your testimony.

See, we have one question if you'd be willing to take one. Absolutely. Go ahead. Thank you very much. Just in terms of your last point, you were talking about allocation between A and B season, just so I understand Yes, yes.

And so can you elaborate a bit more? It seems like the allegation—. Yeah, so I mean, in the, in the, in the trawl sector specifically, you know, that, that quota and, and bycatch being as, as, you know, we all know how, how trawling is in the crosshair these days. And, you know, being a combination boat that does, does all different kinds— seining, pot fishing, trawling, longlining. I mean, we kind of do it all.

We don't— we, we're ethical fishermen in that regard. I mean, I can pass the red face test and honestly speak that, that trawling is not perceived from a 58-foot perspective. But bycatch being first and foremost, you know, we can't, we can't go out and catch an aggregated cod in a B season, uh, no way, no how. I mean, we're going to catch a lot of wrong fish and they're just not aggregated in that time, so that that trawl sector allocation in particular is, you know, we, we have our A season, which is usually relatively terrific CPUE and, and fast fishing, and then we get a B season that basically just completely goes by the wayside. It's lost opportunity for the trawl sector.

It gets rolled over into another sector, but that's kind of not as intended. By default it does because it goes unharvested. That's— and, and in the past when you've kind of asked about these apportionment issues or or season allocations, a lot of times we come up to the roadblock of the Steller sea lions, and there's a lot of barriers around that. Even this fall, as we kind of, we were on the hunt for pawlik, and pawlik are kind of hard to find this fall. And firsthand experience, you know, we're cruising around a lot of areas, and I just kind of realized, you know, we're cruising around rings, sea lion rings out in the western Gulf that are, that are vast.

I mean, we cover a lot of fishable ground that is, that is wrung out from stellar sea lions. And I'm just not saying it wasn't a terrific reason to go in there. I just, seeing that mandate and seeing that maybe that now is the time to address, uh, the larger elephant in the room with sea lions. Um, I know I kind of talked on two points there, but, but the both of them are equally important.

Thank you. Thank you. Okay, and thank you for your testimony. Next, we will take Kylie Thompson.

Kylie, if you are on the Zoom meeting, could you raise your hand and be promoted?

I'm trying.

Excellent, we've got you now. Go ahead when you're ready. You hear me? We can, thank you. Thank you, Mr.

Chair, members of the committee. My name is Kylie Thompson. I live in Sandpoint, Alaska. I operate the F/V Decision. We fish cod in the western and central GOA using both pot and trawl gear.

We were the lead vessel used for the tagging project since its inception in 2021. The tagging project was born out of Western Gulf fishermen's observations that cod, the cod stock in the Western Gulf, a season particularly, seemed to be highly migratory. After 5 years of tagging with both satellite and conventional tags, this theory has been proven. How much of the stock moves may be open for discussion, but the fact that these fish exist is proven. After seeing the presentation by both the Bering Sea and GOA stock assessment authors at the recent plan team meeting, I have some serious concerns.

Clearly, a highly migratory stock that spawns primarily in the Western Gulf of Alaska in the winter and migrates to the Bering Sea in the summer has been identified. We have not been accounting for this stock based on the current Fishery Management Plan system. Western Gulf of Alaska fishermen and communities, including myself, depend on a strong A-season cod fishery. Under the current system, it is assumed that the majority of this stock would be surveyed in the Eastern Bering Sea Survey. However, a large portion of this harvest— of the harvest in the Western Gulf A season is clearly on this stock.

This was clearly demonstrated during the most recent Gulf cod stock collapse when, from a fisherman's perspective, cod were still abundant in the Western Gulf A season, yet summer Gulf surveys reflected the decline of the Gulf-based stocks.

I would ask the SSC to recommend to the Council and the the plan team prior to the next meeting to create an ad hoc TAC to be moved immediately from the Bering Sea to the Western Gulf, somewhere between 3 to 7% based on the author's presentation. Now that the stock has been identified, status quo is unacceptable. It seems prudent to protect it as it— as it has not been protected under the current system ever. As model work continues, numbers could continue to be moved up or down depending on the best available science. But right now, the best available science says we have a newly identified stock that needs to be properly accounted and protected from overfishing.

This stock migrates twice yearly through areas of extremely high fishery effort in the Bering Sea, and in the face of another El Niño event, which could create a blob situation similar to 2016 through '18, getting a handle on this situation immediately seems prudent. I would also add several Western Gulf vessels were used in the tagging project over the years, and myself and other fishermen who have been trained in tagging would like to continue the tagging program on a voluntary basis with at minimum conventional tags, which could be deployed either during the season or with a permit immediately post-aseason pot and state water fisheries to reduce recapture. And I would also like to thank AFSC, Allusions Eastborough, and ADF&G staff for all the work on this project over the years. Thank you.

That's all I got. Okay, thank you for your testimony.

We have one question if you're willing to take it. Yes, sir. Martin, go ahead. Okay, yeah, um, I Um, I'm— what— so, so you're saying that the COD tagging is no longer ongoing, and it's kind of a bit of a concern because the models that people are working on are really dependent on having good information, and tagging is going to be an important component to that. So is that the tagging is sort of an ongoing project, or is it, or is it sort of Not, not, not still going on.

It's not going on currently. Charlotte Levy at Lucian's Eastborough was kind of the lead on this entire project, and I've spoke with her and she was going to try to get us this, try to get this conventional tagging program moving. But I mean, I think an SSC recommendation would be great. I think the satellite tagging provided a lot of information, but the funding ran out for that in 2025. So that's kind of where we're at with that.

But yeah, it is necessary. I mean, the only really— there doesn't seem to be a genetic marker, so tagging is the only way to get information on this stock. Thanks.

Okay, thanks again for your testimony. Thank you.

Okay, we will go now into our discussion specific to, uh, GOA harvest projections and the slide we were just looking at. Uh, we'll go first to Rob.

Okay, uh, thanks, Ian. And just procedurally, I'm, um, the public presentation wasn't on this topic, so Am I just skipping over that for now, or—. Yes, I think we can— this— we can note this public testimony under our general planting report rather than under the specific harvest projections since we don't have a specific harvest projection for cod. Okay, great, thank you. Um, okay, yes, thank you.

And, um, I'm gonna try not to make this too much longer than Chris's presentation of his slide, but so I'll go over just a little bit of information for each of the stocks, and if we can add more to the, as far as what's in the trends, etc., in our minutes if we care to. So starting off with deepwater flatfish, which includes Dover sole, Greenland turbot, Kamchatka flounder, and deep-sea sole. The last full assessment was in 2023. And it's on a 4-year assessment cycle. Catch is considerably below TAC, even though in 2026, at least as of September 8th, it's above the 2025 level.

So it's up a bit, but still within the TAC. So the Gulf of Alaska Dover Sole stock is not being subjected to overfishing and is not— is neither overfished nor approaching an overfished condition. Information is insufficient to determine the stock status relative to overfish criteria for Greenland turbot, camp jacket flounder, and deep-sea sole. For shallowwater flatfish, which includes Alaska plaice, butter sole, English sole, and sand sole, starry flounder, yellowfin sole, northern rockfish, and southern rock sole, this is a 4-year— this assessment's on a 4-year cycle, last conducted in 2021, as was mentioned. And I think the SSC should support the plan team's recommendation and expectation of an operational assessment in 2027.

Catch for this complex also remains below the TAC and below levels where overfishing would be a concern. So information though was insufficient to determine stock status relative to overfish criteria for the entire complex, however. For rock sole, the projection model indicates that they are not overfished, nor are they approaching an overfished condition. For Flathead Sole, this last assessment occurred in 2022. Next assessment will be in 2027.

There's a stable biomass. In summary, this stock is not being subjected to overfishing and is neither overfished nor approaching an overfished condition. For Dusky Rockfish, this one's actually on a 2-year assessment cycle. Last full assessment was conducted in 2024. The projection model was run— this projection model, the projections that we're seeing were run using an updated catches.

Catches were higher than the past 2 years but still under ABC, so at this point the stock is not being subjected to overfishing, is not currently overfished, nor is it approaching an overfished condition. And I just recommend that SCSE supports the team and the authors' recommendations for the presented OFLs, ABCs, and BRDs for all 4 stocks.

Excellent. Thanks, Rob. Kaylan?

Yeah, thanks a lot, Rob. I don't have anything to add to Rob's summary here.

Great. Thanks. I won't re-summarize then. That was very clear. We're accepting all of the harvest projections across the board, and there's no— nothing to report with regard to overfishing or overfished status where that's determined.

Okay, back to staff for, I believe, DSR proposed models.

Great. Thanks, Ian. This is Chris. I believe this is where we're going to ask Aaron Lambert, who's the new author for DSR, to give you a short presentation on the work he did. And then I have a follow-up, just 2 slides afterwards with team recommendations.

So Aaron, if you're on, feel free to share and take it away.

Excellent, thank you, Chris. Through the chair, my name is Aaron Lambert and I'm a biometrician with the Alaska Department of Fish and Game. In September, I presented on the GOA demersal shelf rockfish to the plan team, primarily on the Tier 5 Southeast Outside yelloweye rockfish component of the stock.

Oops, it's not advancing. Excellent.

Oh, there we go. Okay, so before I jump into everything, there's a couple pertinent changes to the input data that'll be relevant to everything I talk about.

So the yelloweye rockfish is assessed with REMA model that has two indices and an extra observation error. I'm not proposing any changes to that structure. But for the IPHC survey data, which is one of the indices that goes into that model, we did update the data through 2025. And I also found that 13 of those stations that were previously included in the SEO were found to be outside of that boundary, so I removed those. More importantly, there hasn't been an ADF&G ROV biomass survey since 2023 in eastern Yakutat.

We did purchase a new ROV. There were mechanical issues. It's a long story, so I won't go into it, but there wasn't a survey this past summer like we planned. There are some possible solutions, one of which is pretty promising. I've had some more discussions about this since the ground fishing plan team, but we did put in a request for funding for fiscal year 2028 for a statewide ROV.

But more promising, we've been talking about either contracting or building a semi-autonomous towed camera system, which I think could be used effectively to resume the biomass survey.

All right, so the proposed model changes for 2026 is just changing the way that we standardize the IPHC yelloweye CPUE, and I'm proposing a spatiotemporal standardization. This was motivated by comments in 2024 from the GOA groundfish plan team, and because there was reduced survey coverage in the SCO in 2024 and 2025, which I've highlighted in this table on the right in the blue box. All 4 management units were surveyed well below average. There was no eastern Yakutat in 2024 and only 3 stations in 2025, all of which were zero catch. And then there was no NSEO survey in 2025.

Of course, the benefits to using a spatiotemporal standardization is that we're able to share information across space and time and hopefully produce a more stable index of abundance, given the reduced survey by the IPHC in that area.

All right. So to standardize the index, I used SDM TMB. I fit 7 candidate models, all with the Tweedy error distribution consistent with the 2024 GAM standardization method. Each model was fit with either an AR1 or an IID spatial-temporal structure. I also fit one model with no spatial-temporal component for comparison.

I would argue the AR1 structure is more ecologically consistent with yelloweye ecology, where they don't move around a whole bunch from year to year. All 7 models were fit with a combination of depth and/or soak time smooths, or neither. And then the model was fit to the entire SEO area as one, ignoring management areas, since those aren't really biologically informed delineations.

And then for model selection, I used cross-validation and AIC, and the model selected is circled in green on the top right there, which was model M2, which included an AR1 spatial-temporal temporal component and the depth and soak smooths. Then to get management-specific indices, I predicted to each management area separately and then calculated CV from that.

So here I have plotted in blue the standardized index using the spatial-temporal method for each management region, and then in yellow, the standardized index using the GAM method from 2024. 3 Of the 4 areas correlate pretty highly with each other. There is less correlation in the NSEO area, which is the smallest area. There's only 7 or 8 stations surveyed there each year. So using that spatial-temporal model and fitting it to the entire SEO area and then predicting to that one area, I think really smooths smooth things out with that spatial temporal and field.

And of course, I'm recommending the SDM TMB standardization for the 2026 assessment.

Next, I conducted a bridging analysis, which was requested in 2024 by the SSC. This is all in in regards to changes to how the IPHC index was treated before being put into the REMA model. In 2024, there were a whole bunch of updates, and so I'll step through each one of the steps in this bridging analysis real quickly. Step number 1 would be the 2022 accepted method for the IPHC index. All these models and the bridging analysis, by the way, are using the exact same biomass indices and the same REMA model structure.

But for the first step, the 2022 model only included IPHC stations that had catch, so all nonzero stations. CPU units were in numbers per hook, and there was no hook adjustment factor. And then the way we calculated which stations were in eastern Yakutat, the boundary was off a bit, and so it wasn't including a whole bunch of the Eastern Yakutat stations. So that first step is reflecting all that. Next, I added in the hook adjustment factor.

In step 3, I corrected that longitude boundary to include all the stations in Eastern Yakutat. In 4, I converted the numbers per hook to kilograms per hook using commercial port sampling data. And then number 5 would be the 2024 accepted model, which is the GAM standardized index, which using the Tweedy error distribution, which allowed us to use all stations in the SEO that were less than or equal to 250 fathoms. Following that, I basically did the same step as number 4, but with those corrected stations from the SEO where I removed 13 of those stations that weren't actually in the boundaries. Step 7 is the same as step 5, but removing those stations.

And then step 8 is the, the method for 2026 that I am recommending.

And so here I've plotted out the total SEO biomass estimates from each of those model steps. The green line is the first step there with the 2022 model. The largest change from any of these— the largest increase in biomass from any of these changes was from adding the hook adjustment factor and switching to kilograms per hook, which bumped it from about 23,500 tons up to 29,000. The station correction that I did for 2026 did slightly lower each of those biomass estimates, probably because it's removing some more productive stations in those coastal rocky areas. And then the difference between all these methods was most pronounced in about 2022 to 2025, which encapsulates that period of reduced SEO sampling by the IPHC.

Following that, When I did the SDM TMB standardization, I noticed that the CVs were a little bit smaller than all the other methods. And that gave me a little bit of concern about how that affected the total biomass estimates.

It is thought that when you use a spatial-temporal standardization, you might have an increase in index precision. But I was just curious if this is actually an appropriate proxy for observation error. I do believe it is, but to be thorough and check that, I did scale all the SDM TMBs by, or CVs, by multipliers 1 through 5, and also used a mean CV of all the non-spatial-temporal methods to see how that would affect the biomass estimates in the remote model since they're essentially weights for that index.

And so here's a table of those results. And you can see that the larger the CV, the smaller the biomass estimate becomes. At the plan team, we did talk about potentially using the bootstrap method for getting a CV that was used in the 2022 method where they were using nominal CPUE because it'll more reflect the change in station coverage over time, and that would probably be closest to the mean of the other indices. But it's something I was seeking feedback from the plant team and now the SSC on opinions on how to go about making those decisions.

And then lastly, I presented on some questions I have about whether or not Tier 5 is appropriate in future assessments. I am certainly recommending it for this year, but my worry is that there wouldn't be any more biomass surveys in the future, which I think I'm feeling more positive about now because I think we do have a more clear path forward on that. And then if the IPHC continues to reduce its footprint in SEO, um, in the future, how, how that would affect things. So I was looking for feedback from the plan team and the SSC on, um, potential criteria that would help, uh, guide that decision, uh, for future assessments. And that's all the slides I have.

We have one question here. Jason, go ahead. Thanks, Aaron, for the presentation. I just got one question. It's in regard to the base model and the GAMS standardization.

I noted that the East Yak region went to zero. I guess I wanted to verify I think there were stations sampled in there that maybe just didn't have any yelloweye. I guess I just wanted to verify that. And I see it on your slide now, so never mind. Maybe I've answered it.

I guess I— we had requested last time around about combining areas as we start getting down to these situations where we don't have many stations or areas are unsampled. I guess I was wondering if you had any thoughts on that. Yeah, thank you for the question, Jason, through the chair. So for this comparison, I left in those 3 stations in eastern Yakutat that all had zero yelloweye catch. So the model is, you know, assuming there's no fish there, which is clearly not the truth.

For the bridging analysis, I removed those because it also causes some funky issues with the REMA model output. Regarding the recommendation to combine all the areas, it's not something that I think is feasible because of the way they do the ROV biomass survey is only in one region, one or two regions per year. And I think you need that overlap to get those Q scalars to work with the IPHC survey. I do think fitting the index with the spatial coral standardization across the whole area of SEO instead of by management unit or having a management unit coefficient allows you to share that information across the whole area more and kind of ameliorates some of that. So that's kind of my hope to partially address that comment, but I don't think it's possible in the way that we're conducting the assessment now to combine areas like that.

Yeah, thanks. I don't mean to belabor this. I guess maybe one follow-up question with that is, would it be— so I can understand having to change the structure with the RIMA and then also how it's fitting to the ROV, but I guess I'm wondering, you know, for other stocks where we have multiple areas, we apportion on the back end. I guess I'm just wondering, is that something you considered? Yes, through the chair, thanks for the question.

It is something I considered, but I think it's even before an apportionment issue that you're— in the REMA model you have the two indices and the biomass indices, which the IPHC indices getting scaled up to that with that Q-scaler. We're there aren't years where they have the whole year surveying the entire SEO. So it seems to me that if you combine them all, it'd be tough to get a single Q-scaler to use both indices in the REMA model. It's something I could continue to look at, and it is my first year conducting this assessment, but it seems to me at this point that it's not feasible using this exact Okay, I'm not seeing any further questions. Thanks for the presentation.

I guess we'll now go back to the plan team comments.

Thank you. Thanks, Ian. We'll let Sarah pull it up real quick.

Next slide. So I put this slide in, in the off chance Aaron ran for it and didn't present. So I think this pretty much repeats what he did, so we can just consider it documentation for what the authors did. And, uh, we can go to the next slide, which has our discussion recommendations. So this, um, is what the team discussed and provided for recommendations for bringing forward in November.

First off, we all definitely noted the author and his team did a really nice job, we thought, of looking at what he just showed you and with the bridging analysis and the appendices that were provided. It's like pretty, pretty thorough job. So the team thought there should be several models brought forward.

The previously accepted model, which is just for the continuity and the bridging after effort, and then the team agreed with the author-recommended SDM-TMB model to be brought forward. And then the team and the author had a pretty good discussion about the CVs. Aaron mentioned it, provided that sensitivity analysis, which is basically the concern with the low CVs generated by the spatial-temporal model.

The overall theme of the discussion was there's no set idea of what variance inflations would or should be. We discussed some other examples such as the Gulf of Alaska northern rockfish kind of doubles that variance. None of us had the background on why and when it was done. But basically what came up was the trying to bootstrap the data by resampling the IPHC survey sampling units, and then refit to the spatial-temporal model.

We put that in as a recommendation, but we also know that that may be problematic with convergence issues because there are so few stations. So we think it's worth trying, and we put it on there as a recommendation. And then we also suggested the play with turning on the extra observation error parameter in the regional model. So several, I think, healthy discussion topics that Aaron is, I think, going to try to do for November. But those were our general recommendations in terms of bringing model runs forward.

And then he had that final slide in his presentation about the Tier 6 considerations. And when we started discussing it, you know, it's using the IPHC survey, the survey effort effort is going— has been going down for a couple years. It was evident that the ROV survey future is probably also in question, but certainly delayed a couple more years before they might get it back. The IPH survey in general was just providing the first information from the first 20 hooks for Yelloweye. It's certainly a concern, so I think the team overall felt that You know, in terms of defining reliable biomass, we're kind of on the edge of a cliff.

If he keeps losing data and doesn't replace some of what's been lost, we may not actually be able to compute reliable biomass as the base for Tier 5 calculations. So I think the idea was to start chewing on this because it's— this isn't an incidental stock without directed fishing. And so we have to come up with some alternative things. So some things that were discussed were evaluating time periods and rather than basing it on catch, maybe basing it on ABC, average ABC over a defined period. And then also noting that the fishery is managed by the state of Alaska and has historically been managed below ABC.

So some definite things to think about. I think for now we recommend the Tier 5, but I think it was good to get some discussion on the Tier 6. And as you saw from Aaron's slide, he's also looking for some advice on any other ideas in regards to that in case we get to where we don't have reliable survey data in the future. And that was the extent of our team discussion —and recommendations. Thanks, Chris.

We have a question here, Franz. Yeah, thanks, Chris and Aaron. The question is actually for Aaron. I'm a little slow. But in the—and I apologize if that's in the documents—but in the time series for each region, and that's true for both the GAM model and the SDMTMB model, there's a pretty strong kind of even-odd year pattern over portions of the time series, although it's a little out of sync between the GAM and SDMTMB in some time periods.

But I was just wondering if that's, you know, if that is real, or is there some kind of autoregressive parameter on the annual intercept itself that's estimated to be negative, or do you have any thoughts on that?

Through the chair, thanks for the question, Franz. I, um, it's not something I've specifically looked at, um, but I, I will look into that more, and if I find anything, I'll present that at the next planning meeting and SSC meeting.

Uh, yeah, thank you. Can I follow up now? Um, so in the model itself, uh, I know there's a AR structure for the spatial temporal structure, um, does the annual intercept itself also have a— is that IID, or does it also have an autoregressive coefficient?

Um, through the chair, um, I believe the only AR portion of the models in regards to the spatial temporal field. So I believe it's IID.

Okay, thanks.

Okay, thanks. Now that we're in the proposed models, I think we're just going to continue on through the presentations, and we will have one more round of public testimony. Our all of our discussions at the end, so we can go ahead on to the next stock.

That sounds great. Thanks, Ian, and thanks, Aaron. You guys got a lot better presentation from him than you would have gotten from me, so appreciate the authors helping out. Next up is Thornyhead Rockfish. Just for some background, this was planned to be an update assessment.

Luckily, Kevin Swicky, the lead author, realized that with the 2025 survey re-stratification that occurred in the GOA trawl survey, new depth groupings needed to be put together, and therefore, new models had to be developed, which turned this basically into a full model, just based on the process of how we're doing the modeling. So Kevin presented several choices to go from 3 to 2 depth strata. I put the model choices in there. He provided a pretty nice stepwise approach to looking at it. If we go to the next slide, we can see that the imputation— I guess I should back up a little bit, too.

The one choice was imputation, filling in the values, and another of the top choices was dropping the early missing depth years. And the figure on the left pretty much shows that the results are that both of those are pretty similar results and estimates. So it's kind of more of a process of which one to pick versus performance. And then if you look on the right-hand plot, he showed that there is quite a bit of decrease in the biomass estimate. So it's just good to point out that the methods to— including 2025, that data isn't reflected in just the change in method, how he's doing it.

It's actually a change in what the survey biomass was. And so from that, the team had a couple recommendations. Next slide.

So they basically— the team chose to recommend Model 24.2F, which is omitting the early years with the missing 500 to 700-meter data. The 3 bullets under there was the root of the discussion of why choosing that. One, the years are at the beginning of the time series, so there's less influence when applying the REMA model. Second, those years actually represented some extremely low biomass estimates that Kevin had, like, looked into and tried to explain and really couldn't. So in a way, it was kind of removing some unusually low biomasses.

And then finally, not that we have the crystal ball, but we all felt that future surveys are unlikely to go back to sampling greater than 700 meters and that we should probably expect that to persist for some time. So the overall choice was to omit the early years rather than imputing data. No real strong yes or no reason, just based on those rationale, that's what we put forward. And then finally, we had some discussion. I think we landed on somewhere that we need to bring the models from 2024, um, forward in November as continuity to represent the base model.

And I need to qualify here that in the plan team report, it's— the recommendation is to bring models '23 rather than '24. That is a typo. I'm going to blame that on the week of COVID that several of us brought home from plan team. But that's just a typo. It's from the models 2024.

So I want to clarify that. And then the other thing was not to underemphasize that the 2025 Central Gulf Survey biomass was the lowest in the time series. And from that previous figure that was shown, it's going to decrease thornyhead biomass and resulting ABC when the final assessment comes forward in November. And with that, I think that's a summary of the team discussion and recommendations. Question for Martin.

Um, just a pretty simple question. I'm, I'm wondering about the logic for Um, and by the way, I did notice that it wasn't 23 and looked quite a while in various assessment documents trying to find that version. So mistakes matter. But, um, the, um, I'm wondering about the logic for including, you know, all 6 of those runs. Wouldn't it be sufficient to just bring forward 24A as being most consistent with the previous approach and not bother with all the other ones?

I mean, the first 3 are 3 area models that can't even be used in the future, and the other 2 are, are options that you're, you're not recommending.

I am trying to keep up with you, Martin, and I'm sorry we put it in there and gave you a whole bunch of extra time, but now I got to come up with an answer for you. So you're asking why we didn't just put forward 24.2a? Yeah, yeah, it seems like when—. Don't you only need one continuity model and wouldn't that be a good one?

So that one is how REMA package parameterizes the observation error and then using the 3 depth strata. So yeah, I think you could. And Jim, and I don't know if we have Kevin on, I'm scrambling here to provide a good answer. Kevin is Kevin, is it— could we elevate Kevin? I think he's probably got the best response.

Thanks, Kevin. Go ahead. Can you guys hear me? We can. Thanks for joining us.

Yeah, um, at plan team, that's exactly what, what Martin's suggesting is exactly what got discussed. Um, they wanted to see a A through F for the last closest to the base and F, which was the recommended model. That other piece, the A through F, was what they put together in their notes. And I did follow up with Chris at the time and he was—. My understanding was it was just to kind of track the bridging so that the—.

That would be sort of attached to the assessment. So essentially just kind of using the September document as an appendix to have it available. But it, yeah, it's only going to have F in the actual assessment because A can't be updated with the newest data. That A through F will be in an appendix, but we can do whatever. That's just kind of what came to me from the plan team.

Okay, thanks. Yeah, I, I putting it the other runs in the appendix seems fine, but I'm not really seeing a compelling reason not to just use 26— 24A as the continuity model, even though it's not using the most recent data.

Gotcha, Martin. That now I'm understanding that makes sense. And that was purely probably driven by me to make sure we kept the SSC happy. Knowing that we couldn't repeat what was done before. So any recommendations you want to do on that, I'm sure Kevin would appreciate, and it's really fine to do that.

We just were worried about the continuity aspect of it not being able to repeat the base model.

Rob?

Yeah, thanks, Ian. I had a question about the survey, how the survey data are used. So there was a comment during the plan team meeting about the depth strata samples from the bottom trawl survey has changed. That was part of all this effort and that nothing has changed in the longline survey, though the longline survey is changing to an every other year in the golf and it hadn't been in the past. I guess, does that have any effect on the way the longline survey has been used or it'll be used in as alternate years regardless, you know, previously too when matching up with the alternate year of the trawl survey?

I guess, in summary, does that change affect how you use the longline data in the model? Good question, Rob. I think I know the answer, but we got Kevin, so I'll let respond to that. I think he's got that better answer. Sure, thanks for the question.

Uh, the—. Yeah, it's just going to be every other year rather than annual, and so the, the REMA model will just smooth through it. But yeah, it doesn't actually change anything other than the, um, the cycle goes from 1 year to 2.

And I'll just follow up to Rob on that question that the core survey stations for the longline survey didn't change, so it is just the other, every other year, which we will just move through, so shouldn't be an issue.

Okay, thank you.

Okay, I'm not seeing any further questions on thornyheads. Let's continue.

Sorry, there is one more. Martin. Yeah, since we've got Kevin online, I just had one nagging question when I was reviewing the document, and so maybe I'll ask it. You're using the longline survey as a supporting survey for the bottom trawl survey and the REMA models, and it I'm wondering, I mean, it seems like you're using all, for each of the RIMA models for the different depth strata. And I'm wondering if there's, if that makes sense or whether the longline survey could be stratified into the appropriate depth category for the, depth strata that you're using the bottom trawl survey data in?

Yeah, thanks for the question. Unfortunately, yeah, the depth strata that the two surveys use do not line up exactly. And so they do at some of the shallower depths, but the longline survey is like 400 to 600, 600 to 800, whereas the trawl survey was at 500 and 700. And now with the new stratification, those, those depth stratas are not consistent anyways. And so I could possibly look at— yeah, it just doesn't line up, so I couldn't do it exactly.

And so it's the now two-strata method seems to handle it pretty well and should be able to handle the trawl survey in the future.

Hopefully that answers your question. Yeah, thanks.

Okay, I guess we're done with thornyheads, moving on.

Okay, this is an informational slide and it was regarding the May survey of the winter, winter shelikov. Straight survey. The main message was that 11 of the southern transects were missed. The survey timing seemed to be adequate based on what they observed, and basically because of those 11 transects, some adjustments are going to be made to the estimates here, and we'll see those in November. So pretty short and sweet on this one.

We have a question, Franz. Yeah, thanks, Chris. Um, on that, uh, survey here, the, uh, 2026 estimate has fairly tight confidence intervals. So does that not yet reflect sort of the additional uncertainty associated with some missed transect portions or transects?

Yeah, thanks, Franz. I, I believe that's correct. I don't think captures those missed surveys. So that's probably why it's so tight. And we could probably ask Cole for further information if you needed it during the Pollock one.

But that's my understanding as well. Okay, thanks.

And with that, I'd like to introduce Cole Monahan, who is the GOA Pollock author and is going to walk us through some models options as well as some discussion on DSAM. So we're gonna let Cole share and go from there.

Thanks, Chris. Can you hear me okay? We can. Welcome, Cole.

If you want me to share locally— oh, thanks. Yeah, we just Hopefully you can see the slides. Yeah, Cole Monahan. I've been doing the POC assessment for maybe 4 or so years now. I was asked to keep this to 10 minutes, so it's hopefully pretty short and quick.

So yeah, just high-level summary, moving to RSeattle. Is what we're proposing. I think it has a lot of clear advantages. You guys have seen that many times. I'm proposing two models this year.

26A is a very close bridge from kind of the last accepted model, but bridged over to RSeattle and then turning on estimation of σr. And then 26B is the same as 26A except a DSEM module on recruitment. And yeah, bringing forward Model 26B for operational use. We're going to talk a little bit about poor fit to the NIMS Bottom Trawl Survey Index, some of the ideas we have to investigate that. And then just, yeah, there's kind of some familiar issues with scale and data conflict, which I think most people are familiar with.

So just diving straight in, there's not a lot of change in kind of results, management results here from the different models. Here's the 3 kind of spawning biomass overlaid on top of each other, the colors on the bottom there. So they're all very similar, certainly within the uncertainty bounds for SSB. Recruitment's a little bit different story. This is in log— the y-axis is in log space here.

And so you can see some differences in recruitment, particularly in the last year or two. When we have less information about cohort strength, the environmental covariates used in the model 26B are more impactful. And I'll go into more details of 26B later in this. This is the kind of overall fit to the age composition data, which I think is reasonably good. There are a couple issues that I notice.

There seems to be, for both the ADF&G bottom trawl and NIMS bottom trawl, the two left columns, there seems to be misfit in the last age bin, age 10, which you can see on the top plot there. I assume folks have been, have seen these new plots, so, but if not, I can walk through them. But I think overall, there's no huge issues in terms of fitting to the age compositions. And just as a reminder, these are Dirichlet multinomial, so the overdispersion is estimated.

Here's what the index fits look like. So again, the models are very similar in fit. So really what I want to draw your attention to is that red box. This is the NIMS bottom trawl. That's the summer GAP AFSC survey.

And it's not fitting well at all, as you can see. And so I thought I would just dig into that a little bit. This is the story we would have heard last year if we had had a November plan team and SSC meeting. So this is— yeah, the 2025 data point kind of really solidified that something seems wrong here. So I proposed in the September document and presentation, two hypotheses which I'll briefly talk about.

So one of them is maybe Bering Sea fish are entering the Gulf.

I think that seems plausible because this really large estimate is driven to a large degree by one and, you know, basically one, two, or three tows right in 610, right by Unimak Pass. So just kind of started thinking that maybe, you know, maybe fish are coming over and that could be causing an increase in the apparent size of the population in the Gulf. You know, this obviously raises really big questions about stock structure and connectivity. I have plans, so Aging Growth, AGP is Aging Growth Program at the AFSC. They have a couple of really, really appealing ideas on how to test this.

Using data that's at hand and I think methods and models that are familiar to that program. So the idea would be to take otolith morphology, for instance, from existing otoliths and feed them into a machine learning model, a classification model. So basically, I could give an otolith to this model and it would tell me if that's a Bering fish or a Gulf fish, And then we would pass a bunch of those otoliths near kind of 610 in this kind of area to see if the model thinks that some of those are bearing fish. And if otolith morphology doesn't work, the same thing could be done with the FT-NIR spectra. So we hope to be able to make some progress on that and have results in 2027.

I will note that this is not a really great hypothesis. I consider it a fairly weak hypothesis because you know, it doesn't— there's like, it doesn't explain some of the other patterns we see in the indices. So why would the— why would these Bering fish come over but not be seen in the acoustic survey in the summer, for instance, in the same region? So I think there's a lot to unpack here. I think it's worth looking at, but I don't think it's going to be a silver bullet.

The second hypothesis that I talked about is shifts in vertical availability of the bottom trawl.

So in the assessment, we assume constant catchability, and that implies a consistent vertical availability to the gear. So any fish, it's like the proportion of the stock that are kind of within reach of the net, or the effective fishing height, as it's often called, is consistent. And so, like, if the fish are trending down over time, kind of on average across the Gulf, then that would manifest as an apparent increase in the bottom trawl biomass, when in reality it's really a change in availability. This seems like a reasonable hypothesis because we tested that in the Bering Sea and found that there in fact are quite substantial changes in availability from the distribution in the vertical water column over time. That analysis has not been done in the Gulf.

I know that the SSC and plan team have asked for that in the past. There have been some technological advancements, and so we are revisiting that analysis in the Bering Sea and plan to extend it to the Gulf in 2027. And this will be our best, I think, way of understanding vertical availability in the Gulf, which could be important if there are large changes in availability to either or both of these surveys. But one immediate problem we run into with this is that this analysis is— requires basically the bottom trawl and acoustic trawl to occur in the same year. And we have 6 years now from the summer acoustic survey, but that was not done in 2025.

It was suspended and is currently not planned for 2027. So I just want to highlight that that you know, survey has been lost and just highlight the value of it, especially in terms of understanding vertical availability of the stock to both surveys.

Okay, so overall, I think the poor fit to the NIMS bottom trawl is kind of the big story for this year. But like I said, I don't think there's a simple answer. There's not like one— neither one of these hypotheses, if true, would it fully explain the patterns we're seeing across all surveys. But I think it's a place to start. We, of course, have continued issues with scale without that bottom trawl survey and data conflict.

From the COMP data, it seems like maybe the plus group is too low. There's kind of that consistent constant misfit for both bottom trawl surveys, so I will explore that next year. Some issues in the composition fits, you know, I, so selectivity is constant for all the surveys, but the stock undergoes a lot of variation in mean size at age over time, a lot of it. And so if the gear is size selective sizes changing over time, that implies variation in age-based selectivity. So something to think about is, should I be using age-based selectivity for the surveys caused by changes in growth?

And then kind of one final thing I wanted to highlight is there was a planning request to try to include the additional surveyed areas for the winter survey, the winter acoustic survey. So these are areas like Marmot and Chiricoff, which which are not formally used in the assessment. And we haven't— we kind of have an analytical plan in place to do that, some preliminary models, and we expect to present or at least have something to review next year. I think that would be a nice step forward to formally include those into the operational assessment, including, if all goes well, bringing the age 1 and 2 fish back into the assessment, which should be really valuable.

So the second half, I don't— maybe I should pause here and see if there's questions on model fits. The second half is just kind of laying out the DSEM module and where we are with that. So I will proceed on if there's no questions.

Okay.

Okay, so yeah, kind of just backing up a little bit for this DSym approach, like why are we doing this? Why is this in front of you now? So incorporating ecosystem environmental climate information is like a key NIMS goal for fisheries management. You know, it's in this climate science strategy, the eBFM roadmap, kind of NextGen Stock Assessment Enterprise. These are things that are are called for by NIMPS, and it's also a long-term goal of the ESP, which we're all familiar with.

And Kalei mentions in her paper that integrating this into models is kind of where we want to go, and she's presented that to you guys as well. But the reality is, based on this paper Mosa Karp led in 2024, this is just really not happening either in Alaska nor around country. So only about 8 out of, you know, 232 stocks directly link environmental or other covariate into the assessment. So we're just kind of not meeting that mandate. And they identified some barriers.

They proposed some barriers to meeting these mandates. Kind of the top here, limited mechanistic understanding of the system, limited data model capabilities and agency capacity to do the work, and then institutional inertia. And so we are proposing DSAM as kind of— this is kind of a proposal to try to move us forward on that. And I think of this as having 3 components. So structural causal modeling is a scientific framework to build mechanistic understanding of our systems from observational data, so formalizing that our process.

It's kind of step one. And then the ESP is the framework to, you know, collate data and very importantly build a diverse science team to develop these mechanistic hypotheses. Right? We want oceanographers, we want ecologists, we want field staff, we want statisticians and stock assessment scientists all working together to build these hypotheses. And we really try to emphasize the team team approach with that respect.

And then finally, DSAM is kind of a statistical tool to implement the first two points and either doing it externally or integrated into the stock assessment, which has some, I think, key notable advantages and can be used directly for operational fisheries management. So that's kind of why we're here, why we're trying to do this, we're trying to push forward on this. So here's what it looks like The assessment for POLIC Model 26B has basically gone through this process. It's now published in this paper I've— I'm showing here. So we incorporate 8 ESP variables directly into the assessment through recruitment.

You have seen previous versions of this. It's not much changed by intention. And I just want to highlight that we plan to continue using mean recruitment for the calculation of ABC for the management calculations.

This is what it looks like. So what we did is I— we did this paper on our bespoke TMB model. Essentially, Martin Dorn's old model was ported to TMB, and we modified it. We've now moved to RSeattle and added the DSEM module internally to RSeattle so that it can be applied to any stock. With AFSC.

And so that's good. We were able to replicate the findings of the paper in terms of kind of which models were selected, statistical performance, those kinds of things. And we went with— I decided— we decided to use— bring forward the— what we call the moderate complexity DAG from the paper, which is off to the right here.

This one had kind of strong statistical performance. And yeah, just kind of had good statistical performance and made sense to the ESP team that developed these mechanistic hypotheses represented as a DAG here on the right side of the screen. Just as a quick note, it's come up before that DSSEM previously was limited to linear effects and then now is able to do nonlinear and non-stationary effects. That's published and incorporated into the package and incorporated into RSeattle as of now. But I'm not using that feature currently.

It's quite data-rich, and I'm happy to talk about— or data-hungry, I should say. I'm happy to talk about that more if people want to hear more.

So that's kind of where I am with the DSym integrated model.

I think this is all quite useful. New to everyone. And so I think one of the big challenges for us to use this tool is to kind of develop an appropriate workflow. So in the September document, I— and at the plan team, I asked for feedback, and I would repeat that here for this body. There's— I kind of had 6 questions.

The top 3, I would love feedback on this year, and I feel like the bottom 3 are probably slightly longer-term things. But I'd love to, yeah, have a discussion about this, get feedback and advice from this review body. So I'm happy to stop here now, or we could zoom through, basically through the slides, but just kind of a, just kind of showing the table, the differences. Again, there's not really big management implications between the three models. And then, yeah, just a acknowledge DSAM.

This kind of work is— it's a real big team effort. So she highlighting Jim Thorson, who's kind of the intellectual driver of this stuff, and Clay for the ESP, and Grant especially for the integration into RSeattle, and then Jane and Juliette for kind of PIing and leading the paper, helping lead the paper. And of course, thanks to the ESP team. So that's all the prepared slides I had. I thought maybe I could maybe just end here on the discussion, but I'm happy to take questions or anything on any aspect of the material.

Thank you.

Thanks for that presentation, Cole. I have one question. So this would be the first DSAM, fully integrated DSAM in an operational traditional model that we've seen. And I— so we're kind of new to it, and I, I guess I wanted to walk past you what the way I think this process sort of ideally unfolds, the various steps. And then I have one question about the last step.

So as I understand it, the way this would work ideally would be that you would assemble the process understanding and expertise that's out there, you would construct a DAG that reflects that, perhaps several, and in this case you did several of different complexities. Then you'd fit the DSEM model within your stock assessment and assess whether it worked, basically. And then the step that I'm— maybe you could speak to is I think to close the loop, you would then go back and assess the linkages that you estimate in the DAG against your original hypotheses. And so my question is, do the linkages in this this moderate complexity DAG, do those correspond to what you would have expected based on all the process studies before you fit the whole thing?

Yes, great question, Ian. The answer is yes, except for one of them. And this is part of what we call scientific validation as part of the process of building these mechanistic hypotheses represented as DAGs is to ask the team to predict the sign of the relationship. And so— and then post hoc, we can check that against the estimates from the statistical model, and they should match. And they do except for one, and this is— we've actually discussed this, I guess it was in 2024, in that when you have a mismatch there, when you think, you know, when the science thinks that there should be a positive effect, but it's negative, for instance, you can do two things.

You can dig into that more. That can be caused by, for instance, a kind of a misspecified DAG. So you might be missing a variable. This is known in the structural causal literature. Like if you're missing something called a confounder, you can get the wrong sign on other relationships.

So that might be indicating that, you know, you need to revisit your DAG, go back to the science table and say, like, are we missing a variable or, you know, what's gone wrong? I think the other option is you can just think of this as a structural model. And so it's structural causal. So on one end you have— this is how I think about it— structural is it's just a fancy regression. With indirect effects.

It's a model to explain the covariation of all variables at the same time. But you don't make any causal claims about what you get out of the model. On the other end, you have like a fully causal model where you, you know, you do this process and then you have these effect sizes basically like you can see in the circles. These are— those numbers are basically regression coefficients. And you would make strong counterfactual statements about that.

So like we have temperature in the upper left and we could make a counterfactual statement like if temperature had gone up 1 degree, this is what would have happened to recruitment. That counterfactual statement is— and it comes— that's kind of a fully causal representation. So I think of it as a spectrum. You kind of start as a structural model and as you build evidence and confidence that you have the right kind of process captured in your DAG. You can build confidence in causality and then start to make those claims.

But you, of course, can fit a structural model and explain recruitment really well and not care at all in particular about— yeah, about like particular links in here and like learning something about the ecology of the system. So I think it depends a lot on depends on the goal. Like if all you care about is predictive performance, you just stay in the structural realm. If we want to learn about mechanisms in our ecosystem, then we need to do more rigorous validation and things like that to try to build confidence in the kind of causal interpretation of the results. So I think that to a large degree kind of depends on what you guys want to see as a review body and how we want to use this tool.

Thank you for that answer. Uh, next question will be from Curry.

Thank you, Cole. I appreciate you walking us through this stepwise process that you and the team have taken. I think it's important to highlight the steps in that process you have done that have led us to where we are today and the possible inclusion in this assessment model.

My question really relates to the selection amongst the alternative causal maps. And so in the table you have here, right, we have reduction in unexplained variance, AIC. Have you considered sort of out-of-sample predictive performance as an alternative? And if so, you know, is that a viable basis for selecting amongst alternative maps?

Yes, I have. We did that in the paper. And I think all three— and this actually gets at one of my questions in the discussion points. But yes, there's different metrics for kind of statistical performance. We have— I have— I'm showing 2 here.

Delta AIC is out of, you know, approximate out-of-sample predictive power for all variables at the same time, not just recruitment. It's important to note that distinction. The percent reduction in unexplained variance, that far right column, that's just for recruitment, obviously. And so there's no complexity penalty here. And so as you add complexity, to the model, you're always going to explain more variation.

And then the third one, which is in the manuscript, it's in the paper, but I didn't put it in the planning presentation and so didn't add it here, is kind of what we were calling skill testing or leave future out, where we asked the model during the retrospective peeling, we asked the model to project recruitment forward and then compare that and basically ask, can this model do better at predicting next year's recruitment? And in the paper, we showed that yes, it can. I replicated that this week. Grant and I worked through that in our Seattle and replicated it. And yeah, it does about a 30% job better, better job at predicting recruitment by the metric that we used.

So all three of these metrics tell us different things. Like if we're trying to use this for operational management and using projected recruitments to calculate ABC, then probably we care the most about performance of skill testing. If we care about understanding the ecological mechanisms in the system, delta AIC is probably the best tool. If we want to just know how well we're doing with recruitment, kind of the percent reduction in unexplained variation is the best tool. So this is an open question of like which statistical metrics do we use to gauge the performance and help us select from the models.

And I think it will always come back to like what's the goal of the model. And since I'm not yet recommending using the projection capability of DSAM just for recruitment for ABC, I've kind of reverted back to AIC and the and the variance reduction metrics. But I think that's an open question.

Yeah, thank you. It's helpful to understand your thought process for why delta AIC would be the basis given that you're not projecting, at least not at present, project—. Using this to project recruitment forward into the future.

Yeah, and I guess I would say from the manuscript, this moderate causal kind of performed the best across the three metrics, kind of was the— and that's why we selected it in the manuscript and then here bringing that forward. So in this case, I'm kind of balancing the three. Certainly if you— if SSC wants us to pick one metric, then we can do that, and that would be how we would select models among the candidate DAGs or causal maps, whatever you want to call them.

Next question from France. Yeah, thanks, Cole, for laying that out so nicely in the document and here for us.

So, yeah, I mean, you asked in the document, you talked about the projections and that there are, you know, probably some hurdles before you can implement the projections, but kind of thinking ahead, obviously we, we probably wanted to, you know, use it in the, in the projections ultimately. Um, and in that case, um, have you guys thought or had discussions around what happens when, you know, obviously temperature is an important, uh, variable in the model here, and we expect temperatures to probably go outside the range of what we've seen Have you guys thought about, you know, what that might mean and/or how to deal with that? And maybe related to that, have you done targeted sort of analyses where you say, you know, as part of the retrospective peel where you say fit the model to a period prior to the marine heatwave and then it gets confronted with sort of new conditions, warmer conditions, and how it does across that sort of change in the temperature range? Thanks. Yeah, I think it's a great question.

The answer is yes and no. We have done kind of the stability— we've tested the stability, the estimates, using the retrospective function. So basically, do— are kind of the effect sizes consistent as you add years of data? For this stock in this case, the answer is yes, they're quite stable, which gives me confidence that, you know, next year we'll get another data point and that the relationships won't totally change. So there's definitely some stability in the system given the data we have at the moment.

The point about going outside of the range is not one that we've directly investigated or researched.

And I think Pawlik doesn't quite have the same kind of event like, like with Peacock in the Gulf, for instance. And so I think, I think part of our due diligence is to, you know, apply this to more case studies, test that on real data to see how well it would have done kind of detecting that. And then there's also obviously a lot of simulation work that could be done. And I think that would— I recommend it, you know, and plan to work on that kind of line of analysis through both simulation work and just more case studies.

But yeah, I think it's an open question. So if temperature keeps going up, is this going to hold? And I will note that we can do, because of the new features we can do thresholds, we can do other types of nonlinearities, and we can do non-stationary. And non-stationary means that, for instance, the beta— so each— sometimes we call these effects betas, so like the circles, kind of like the effect sizes or regression coefficients, if you will. Those are constant across time, right?

So, okay, temperature to young of year condition is negative 0.31. That's an estimate assuming it's constant for all years. The package can now have that term vary by year essentially. So if you had really strong data and really strong signal, you could use that to detect like a regime change or something like that. So I think there are tools with— statistical tools within the software now that we could used to try to test and explore effects of kind of like nonlinear responses, whether it's a threshold or some other quadratic or something like that.

But we just haven't quite gotten there. But I do think that the vision, long-term, longer-term vision, which I think Kalei has talked about before, is that at the top of these DAGs, pretty much we're always going to have oceanographic physics. Physics causes biology. And so if we have, you know, reliable, accurate MOM6 output, we can do projections by sticking in temperature or wind or whatever you need and, you know, doing a 5-year projection based on what those MOM6 models are doing. And then that information about the the change in temperature and wind and propagates through the entire kind of causal structure and then gives us a more sophisticated understanding of how recruitment would change in the near future.

So certainly we have a lot of flexibility to test and explore these kinds of things, and I think we're at a place where we just need more time, we need to apply it to more case studies, and some guidance from review bodies like the to see on what are the most relevant things to explore for operational management.

Thanks. Yeah, thanks. I have one more question about the DAG maybe, and then I have a couple actually about the kind of the model structure going back a little bit. I was a little slow. And this is— the question on the DAG here is going back to, um, Ian's earlier question about, you know, whether these are consistent with expectations.

I think we commented on that last year, that the relationship between younger year condition and recruitment is kind of counterintuitive. And I was just wondering if there had been some thought put into that, what that might mean, or if anybody was able to make sense of that. Or are you— is that just one of the examples where you just sort of relying on, well, it, it works, so I'm not going to worry about the mechanisms. Yeah, um, I do remember, Frant, your, um, suggestion. I think it was a good one, and I, I wrote about this in the September document, but I intentionally did not explore that because I think that one of the challenges here kind of, um, the second bullet point, like how often should this set of DAGs be revisited and the preferred model changed.

I don't want us to be in a position where every year I'm changing the DAG, um, people are coming to me and saying, oh, I have a dataset, you should put it in, you should put this in, you should do that, you should do that. I think we want to be really careful to have, um, kind of stable and consistent DAGs that, um, and be thoughtful about when we update, when and how we update those. So while the suggestion was good and I made the conscious decision not to update it this year because I don't want to set a precedent that people can just say, oh, try that, try that, try that, and then every year we're kind of, you know, messing with these things because that's one of the main down— big downsides of all this flexibility is that, um, you know, that you can just do so much And so I think we should establish a workflow and some guardrails about when and how to update the DAGs. And so I do think we should revisit that one, but I think it should be done at an appropriate time. And I think I want feedback from you guys about when that time is.

Yeah, thanks. We might get to that in our discussions, I guess. Um, uh, I have a couple questions about earlier portions here. One is actually going all the way back to the MACE acoustic survey, and I had asked Chris about that. Yes, you don't have to go back to it, but, um, you know, my assumption is, I guess, that MACE will adjust the estimate and the CVs for '26 and that you just use that, or is Is there any adjustment that you have to do in the model to deal with the extra uncertainty?

Yeah.

I think at a minimum we're going to do a sensitivity to it. Mace is fairly confident that there were not a lot of fish in that missed portion, but I think we could do a sensitivity where we bracket and say, you know, take the estimate as is, or kind of assume that, you know, a kind of a historically large portion of the stock was missed, kind of a worst-case scenario. What was the worst case of what was missed? And then just run that through the assessment and see if it's really sensitive or not. Hopefully it's not really sensitive.

If it is, then we'll probably have to come up with some sort of middle ground. But I agree, I think it would be probably make the most sense to do it on the data processing side, not on the assessment model side. Like, for instance, you know, adding— having the model, you know, have an inflate— inflate the— try to estimate the kind of the missed proportion or something like that. I think it would make more sense to do it on the data side.

Yeah, thanks. Yeah, we may or may not comment on that, I guess.

Last question, at this point anyway, is around your hypotheses for what might have caused the mismatch between the bottom trawl survey and the model estimates. And I was just wondering if you can maybe remind us or tell us from your paper on the Eastern Bering Sea changes in pollock availability to the bottom trawl survey Was that tied to any kind of environmental conditions or other kind of known conditions, age structure or whatever, that could be used to tie to sort of availability and could be tried in the Gulf? Or was that strictly just looking at how it varied across space and time?

Yeah, so in the Bering Sea, the paper, we mostly just looked at how it varied in space and time. We did test age structure as a kind of a potential driver, with the hypothesis being that, you know, a younger— because the model was an aggregate, sorry, but kind of like, you know, younger fish being higher in the water column. We found no meaningful relationship between that kind of age diversity and the kind of like the percent portion of the stock available to, for instance, the bottom trawl. We did not explicitly incorporate environmental covariates in, but I think that's something we've discussed and would be feasible in the current modeling framework. And so something like turbidity would, you know, maybe make a lot of sense And one of the disadvantages of the 2021 paper is that the model took about a month to run, and so we were really limited in the amount of kind of sensitivities and exploration that we could do.

There's been a good technological advancement on the TMB side, which brings that down to like 20 minutes. And so we're at a place now for both the Bering and Gulf where we can do a lot more exploration and testing of hypotheses, including things like covariates. So I think that's a good idea and something that would be very worthwhile to explore.

All right. Thank you. Actually, I do have one follow-up on that, if I may. Last one.

So for the Gulf, given that there were some summer acoustic trawl surveys And so there is— and that might limit the— or reduce the runtime if there's not much data. But there is at least a few years of concurrent acoustic and bottom trawl survey data that you could use to do a similar analysis. Is that correct? Yes, we have 6 years in the Gulf, 2013 to 2027. I think 2021 maybe, or sorry, yeah, 2011 maybe.

Yeah, I kind of forget the years, but yeah, there are 6 years of overlap, which should be more than enough to kind of get a pilot analysis working and get a glimpse of if we think there's large variation in availability. In the— like, in the Bering Sea, I think it— off the top of my head, I think, you know, it can range, say, the I think the bottom range from like 40 to 80%. So they're pretty big shifts in availability to a gear type.

When we did that initial bearing seed study, we kind of, we talked about the Gulf, and I think the hypothesis from the survey team was that it would be less of an issue in the Gulf, but we haven't tested that, and there's some evidence now that, well, I have a figure in the September plan team document that shows kind of the acoustic estimate of biomass that's unavailable to the bottom trawl, and it's a lot. So kind of the more we look at this and think about it, I think it is an important aspect of the stock that hasn't been investigated enough. So fortunately, I think we'll be able to do it. We have the data processed, we have a model going, I just have to work through it. So I'm confident by next year we'll have a pretty good understanding of how that's varying across those 6 years, at least.

They're biennial too, so that makes it a little harder than in the Bering Sea. All right, thank you. I do recall that figure now. Thanks. Next question from Kristin.

Yeah, thanks, Cole. I'll try to keep this to the point. I've got 2 questions. I think one's on the misfit of the trawl survey, and I'm just curious if you had any thoughts or discussion about, because the, there were at least I think just 3 stations that were a huge amount of the catch that was in the trawl survey, about truncating the data set spatially along the peninsula to see if it's really a peninsula problem. And/or I think about the fact that those are all super close to Unimak Pass, and there was ice in Bristol Bay through April this year, and how the environmental drivers of ice and/or cold pool or whatever is potentially influencing either into the Bering Sea on non-cold pool years or into the Gulf when there's lots of— and how that influences the fit to that, and if it's really a spatial thing where all that misfit is really driven by stations in and around the peninsula.

So I'll stop there for now.

Yeah, I guess I haven't looked at it in that level of— that deeply. I think that would be worth looking at. I think for me, first step is to try to either just test whether those are bearing fish or not. If it's really conclusive that they're not, then there's probably some other process going on. But yeah, that's the idea.

So this map, those 3 red dots, those are kind of the largest toad— those are the 3 largest toads for the survey in 2025. And they're large, but there are, you know, they're not unprecedented. I think I have a CPE figure in the full document. So it's not like they're completely unprecedented, but it's striking.

And so I think some more analysis on my end to look at things spatially would be warranted. And yeah, we plan to we plan to poke at this a lot more in the coming year. And I think that would be, yeah, a really, really good idea to try.

Okay. My next question is kind of a bigger picture question. There's really no one in the audience here, so I'm trying to focus on what kind of industry folks would think about the DSIM and kind of incorporating that into the model. And it's like thinking about the missing misfit of the trawl survey and not having a summer acoustic survey. It's like, where is the context and the priority of adding this level of complexity into the stock assessment model when there's not a huge amount of difference in the reference results and the actionable things that we're doing?

And where do we play, where do we spend our time and prioritize this super interesting, cool modeling to add environmental stuff in there when we have other major concerns, especially in the light that they don't make a huge difference in the outcomes. Thanks.

Yeah, I mean, I suppose my take on it is that my job is— I'm mandated to do this. And, you know, it does, it would, it does affect kind of more recent recruitment estimates, but I think this approach will be most useful and most impactful when we start using the projections of recruitment for management advice. And so I'm not proposing to do that. It just felt like too much to kind of to change at once. So I'm trying to think of this as an intermediate step, a first step, but it's not the last step in the process.

And Kalei, I think, has walked you guys through that, kind of our vision for how that— kind of what success looks like. And I just think there's a lot of big questions about how to do reference points calculations and ABC calculations and so forth with this module. So yeah, I guess I see it as my job to try to fulfill the mandates.

In terms of balancing time, yeah, I don't know. I guess I would just rely on this body to suggest what priorities are. Certainly I admit that the misfit to the trawl survey is more, is a bigger issue. But I don't think there's any—. I don't think there's any issue for me to try to work on them concurrently.

And I would also note, you know, this DSIM approach, we kind of picked the stock that it would be least impactful on because the age data are so strongly informative of cohort strength that the environmental driver, you know, we can kind of— sorry, we can kind of see that— maybe I should go back here.

You know, we can kind of see that, like, look at the historical time series. Recruitment estimates are almost identical. If I delete 5 or 10 years of age data from the model, then the DSEM module is predicting much different recruitment estimates. We're just doing a stock that has age data in every year since 1973, I believe. So it's really pinned down.

The environmental covariates are not moving. They're not competing against the age data. They're only impactful when we have no age data, which is in the recent years, and then of course the projections. But that won't be true for all stocks. And so if we want to move forward with these environmentally linked Integrated Assessments, we're going to see bigger differences in other stocks by nature of having, you know, less informative data and just less data in general.

Okay. It looks like that wraps up our questions. Thanks for the presentation, Cole. Let's go back to the plan team presentation for the additional slides from the plan team perspective.

Yeah, thanks, Ian. I've just got a couple. If Sarah can bring them up and we'll go through them.

So, back— way back to the model work that Cole did, the team did recommend doing the transitioning to our Seattle from Model 23E. Once again, we've seen quite a bit of bridging been done in the last few years, so I think that was pretty straightforward.

The team also recommended that both models, 26A and 26B, with the DSAM built into it be brought forward for November consideration. And then just On side notes, basically Cole covered all this, but, and there's several questions that address that, so I don't think I need to go into detail, but there's still some concern with that trawl survey. We heard Cole's plans to look at the vertical availability and the odolith morphology and definitely support that work to try to figure out why the bottom trawl survey misfits going on. That's a pretty high priority. So basically just endorsing his thoughts and ideas.

We did not come up with anything new in that regard. And then on the next slide, um, just in terms of DSUM, uh, basically there was a lengthy discussion on all of this. Once again, it kind of ties back into what was heard in in the joint team and also in the arrowtooth flounder team or arrowtooth flounder assessment. I think the big thing in trying to address Cole's questions, we also repeated his questions in our plan team report just because they are pretty important and I don't think our team discussion got anywhere near answering all of them for him. But basically some of the discussion that came out of it was, you know, the ESP teams have done a really good job over the last several years to bring forward the ESPs and linking the DSAM to those and, you know, going to where Pollock is now, which is bringing it into the actual assessment model.

But I guess the team had some caution just about being too prescriptive of defining that process. Maybe it just kind of needs to get established and see what else comes up because there's certainly stocks that don't have ESPs, but you might have authors that have developed really good DAGs. And so I think not, you know, not to like say anybody should go rogue, but the team just kind of felt like, you know, have the ESP teams lead a lot of these questions that Cole asked, um, and, you know, kind of develop guidelines that as review bodies, plan team can follow with. And if we see any major things we disagree with, we can bring it up versus actually leading the charge on providing the recommendations because our ESP team's doing the work, know it best. The team didn't address really any of the questions pertaining to risk table use or any of the larger management issues such as using it for ABC OFL calculations or modification of the harvest control rule, just in the sense of like that maybe, you know, that kind of needs to be an offshoot effort maybe versus having a team try to do that at a meeting when they're doing a whole bunch of other reviews.

I'm sure that SSC can relate to some of that. And then I think what the team kind of focused on was trying to discuss and provide some additional diagnostic type evaluations Those are listed in the slide contributing to the assessment and things like that. So I'm going to pause there. I think that was most of what we had for Pollock, and then we just have one general team recommendations I can provide after Pollock's done.

Question from Franz. Ja, thanks, Chris. Did, did you have more— I didn't, I wasn't able to listen to all of it, but did you have more of a discussion around just what it would take to do the projections with the DSM module integrated? Yeah, I don't think we discussed it too much, and Cole should probably jump on if he's still on too, because I think it's pretty straightforward. Forward.

So go ahead and Cole and respond, but I think I'll let you go and then I can follow up. Yeah, thanks Chris.

Um, DSAM, to do a projection in DSAM is trivial. Um, I think the challenge, there's, there's one small challenge on our side. We use this kind of SPM or PROJ software and it's not currently configured to accept projected recruitment, but we, that's something, you know, AFSC staff could modify. So I think it would be rather straightforward for us to do the projections for ABC.

So yeah, no technological hurdles there.

Okay, thanks. That's good to know. And Curry? Yeah, just a quick follow-up. So all of the covariates that are in the current DSIM model are available on a suitable timeframe to be able to do projections for it, or would you have to come up with projections for temperature or other covariates?

Ooh, off the top of my head, I think most of them are. I might defer to Kalei on that, but yeah, they should be. And it's okay if we— DSIM handles missing data, so if one of the variables comes in late or something, that doesn't interrupt the process. So I feel no anxiety about kind of interruptions to the workflow of working up the assessment and calculating the ABC. That should not— I don't think that should be a problem.

I think a bigger problem would be is if we lost a variable permanently or something like that. That would would be kind of the harder situation to deal with. So no, I don't see any hurdles in terms of getting like the in-year ESP data to put into the model by November. Thank you.

Okay, seeing no further questions, why don't you go ahead and finish, Chris? Fantastic, I'd love to. So we moved some of these DSIM thoughts to just general recommendation that basically, you know, there was a lengthy discussion in the Arrowtooth one, I believe, and it was just to like consider different multiple supported models and looking at ecological hypotheses. So I think, you know, it was understood in our conversation and discussion with the folks involved, and I think it's just something ESP teams and DSAM folks can take and look at. The second main recommendation there, which is what Chris had brought up and others afterwards, was the BRD.

We definitely need to work on having the wording consistently in place of sub-area ABCs. And I think if there's, you know, just from a side point of view, if there's any preferred thing like adding in apportionment, having that captured in your minutes, or reaching out. I'll just go off to my AFC— AFSC hat here— reaching out after this meeting on specifics. We'd be happy to do it, just need to, need to make it happen. And then finally, it was, uh, discussed in the joint, um, meeting, but the team agrees with the Halibut DMR Working Group recommendations.

And I don't know, Sarah has a BRD slide she wanted to talk to. This might be a good time, Sarah.

Yes, thank you. This is an excellent time, Chris.

So this slide is mostly just a heads up for the SSC in December that similar to states, the Council is interested in SSC advice as to whether there's flexibility in the BRDs for shortraker, ruffye blackspotted, and thornyhead in the Gulf as we spoke about earlier. Again, from, from a staff perspective, I, I see this kind of as a two-part exercise. Um, first, the council asking whether there is any flexibility in the apportionments recommended by the SSC, uh, based on what is known about the stock and genetic information, stock structure. And then if so, what what that level of flexibility could be in order to prevent any biological impacts at the stock level. And our plan, Chris mentioned this, but our plan is for council staff and Science Center leadership, they already, we have some language in the guidelines for the stock assessment authors to make sure that we are updating our tables and language appropriately, but we'll continue to work with everyone there to both do that as well as essentially compile and bring forward any, already existing information for these 3 stocks in November and December.

Um, so our, our intent isn't any to, uh, to bring forward any new information, but what is our— what is currently out there, make sure that it is in front of the, um, plan teams and the SSC, even if stock structure template from, from an old, a previous year assessment, for example.

The last thing I wanted to mention here was noting timing constraints on developing kind of a more quantitative process to deal with uncertainty or handling the small change language. From a, from a planning perspective, we welcome feedback on whether the type of information that we provided in our planning report for skates is helpful for the SSC, and if that's kind of where we should be focusing our attention at our upcoming November meeting for these 3 stocks or stock complexes, so that we can help the SSC with your discussion when it comes time in November for those other stocks. So welcome any feedback there.

Sherry. Sure, thank you for that. Um, I hadn't really thought of this as a two-part exercise, so if the first step is to say, is there flexibility and what level of flexibility is in there? And if we're not at the point of saying small changes, who knows what we're going to come up with with the stocks. If there is a stock that we Is the goal of that so that if for some reason there's a stock that we thought could handle small amounts but not, say, last year's, and is the goal so that you could identify?

Yes, there is flexibility. No, we haven't had a chance to figure out what— we haven't figured out a process yet for how much if it's not equal to last year's? Is, is that the goal? Or, um, yeah, I guess it just is— is that the goal of the two-part process? Through the chair, so I'm hoping that in your SSC report, um, to the council at this meeting, we can hopefully try to glean some more information if I'm not interpreting this question from the council correctly, because when I read this, um, I read whether there is flexibility in those, and then the second part of— here's useful information.

Um, and while I don't think we've established, and we aren't sure when we'll be able to establish what small changes means, um, I guess I guess throughout, throughout this whole thing, we've kind of— there's two different perspectives that I think we're trying to blend. The first is plan team and SSC are obviously more comfortable with kind of trying to provide a quantitative method that will be a little bit more black and white in terms of this would be too much and this, this change is acceptable. Acceptable from a biological perspective. Um, I think if I think from a, from a council perspective, the, the information, the discussion that you've all had here and that the plan team had, if I use skates as an example, um, is helpful when thinking about if I were to set a TAC level, um, would it be okay if there's a strong change in apportionment from one year to the next, and I'm trying to make sure that, uh, that a TAC is not necessarily exceeded, um, based on kind of the qualitative SSC and plan team discussion. If I'm going— if I'm looking to last year's BRDs, um, I think you can kind of get to the to the gist of what level that TAC would be appropriate at based on a lot of the conversation that you all have had here.

Um, so I know that's not really a cut-and-dry answer, but there at the plan team, we had several cases where we had plan team members saying, yes, we have all these concerns about skates, but would any of this really be likely to result in biological impact? Impacts at the stock level. And, and I think that is really the, the crux of what the council information is trying to get at. Um, because otherwise, you know, if there were certain impacts that were, um, I guess the council, the council has the ability to set their tax at whatever level that they want to. Our council, and I guess I guess our council has historically always been very reliant and dependent upon the SSC, and they have to balance national standard too when they are thinking about their tax setting process.

And so I guess it's— I'm not sure entirely how to answer your question because I don't want to be speaking for the council. Council, especially if my interpretation of, of their motion is not what they're intending. But that's, that's what I'm trying to go off of.

So I'm struggling a little because we definitely do come to council staff and you interact with the council to find out what their interpretation is. So I— if we're named— how's this— the questions that we've gotten before, and my understanding was what I told Ali, and I don't want to— I didn't want to share with the SSC information or direction that's not appropriate. What I thought— we've gotten questions if we said in the past and this has been a few years, but we've said yes, there's some flexibility. I mean, not— this wasn't a BRD issue.

We've been asked how much. And so I thought that these were completely tied, that the council wanted to know, us to determine actually how much. And so I actually had suggested that we not say yes, you know, you can do some, it should be small, or be conservative, but not state it because I thought that was asking the council what they had for us. But if you have a different understanding, I definitely want to follow yours because you've had— I usually get my understanding from council staff. Yeah.

Yeah, I'm sorry. I, I have tried to communicate back and forth between, between the bodies on this one, and, um, I think when the council put this motion forward, um, They were trying to be as explicit as possible, and it seems like we, we do need further guidance on, on what a small change is, um, because we tried to bring forward at the plan team, uh, an example. I think in the past we've talked about talking to in-season management, bringing forward here is the level at which if you set the TAC, um, if you set the TAC at the same level as you did last year, we, we might not end up having a constraint. And we brought that forward as kind of a simplified method so that the advisory bodies wouldn't need to recommend a specific number, but it sounds like process-wise that's an uncomfortable way to go about doing this.

If I can clarify, I'm not certain, I'm not positive that that's what I meant. I thought by splitting this up, is there flexibility? What I was trying to ask you is, is it appropriate to tell the council, yes, there's some flexibility, and leave it at that and not give any other information? That's what I got from your talk, but that's not what I've gotten from the Council in the past. So I just wanted to know whether even qualitative wording works.

Through the Chair, my understanding based on, based on what I've kind of gleaned from all the discussions between SSC and Council is that yes, that information is helpful.

And it, it's hard for me to know exactly how the council's interpretation of this may or may not change in the future. Um, I guess I can only go based off of what we've heard from our, our current council members and those that have been involved with this BRD process.

Yes. Great. Okay. Thanks for answering those questions.

Did you want to show us the overall—. Yes, specs tables? Thanks. Sure. These are just, um, so this is the plan team recommendations for the specific stocks that are finalized at this meeting instead of this upcoming November and December.

I'll note there is one typo in Rexol that, that we will fix, and that's just an order of magnitude too big.

So we will keep these, or the SSC's recommendations for, for these final stocks in our back pocket to include in your December tables. We don't need to do anything with those at the APR Council at this meeting. And then these are just the proposed harvest specifications. These are the rollovers from the final rule earlier this March, and you'll notice that these are the updated tables to include the formatting for the BRD change. And hopefully what is easier about this is that anything in gray does not need a number.

So hopefully that simplifies things just a tiny bit.

Rob? Yeah, thank you. These tables are super helpful and the gray boxes especially too. I just— a couple questions and just wondering for explanation or if they are potentially typos. For ruffeye and black-spotted Rockfish, why the BRDs are slightly different numbers than the other values in their rows, respective rows.

And then also for other Rockfish, the BRD is equivalent to the— for SEO is not the TAC but the same, but the ABC.

Through the chair, Dr. Suriyan, could you just say that one more time? So, and this is for proposed specs, or this— yeah, this isn't the final stocks rough-eye black-spotted is on the proposed specifications one. Yep, so yeah, so lower right for rough-eye and black-spotted rockfish, the values for the BRDs. Are different than any of the other values to— for OFL. Do you see that?

The 224 versus 229, 359 versus 385.

I was just curious why that is like that.

Through the chair, are you looking at the 2026 numbers?

Or the plan team recommended for proposed specifications for 2027 and 2028. I want to make sure I'm looking at the right Table. Um, well, if I could use a pointer, I could show you.

Yeah, it's the, um, yeah, those.

I know, I just have a different version which has The—. I think—. I'll tell you what, here's—. We're going to take our 15-minute afternoon break, give staff a chance to look over the numbers. Numbers.

Maybe, Rob, you could— okay, make sure you're looking at the same ones, and we'll come back, we'll reconcile this, and we'll go right into our discussion. So we'll come back at 40 past, and we will take public testimony if there is any.

Longevity as well. Okay, we're going to come back into session. Something like that. And I will turn back to staff to see if we can answer any of these questions on the specs tables.

Through the chair, um, I don't even know if— I don't know if we need to pull those. Numbers up on the screen or not, but the confusion lies in a combination of things. One, I have about 15 different tables on my screen, so that was confusion between what was on the screen and what was on my screen. The other problem is just that we're trying to not confuse our proposed specifications numbers with what we're proposing for our final specification. So for the stocks that we have both of those at this meeting.

So the numbers that I believe the SSC— the numbers that are coming from the SSC recommendations for proposed specifications, we have a table that is a rollover for those stocks with no changes between the 2026 values and the 2020 2027 values except for the stocks that were assessed either this January or last December. I think we had a one harvest projection in the Gulf, so those numbers are taken care of and the numbers that are posted in the proposed rule, um, plan team recommendations, we can update that table because those are just, I think, uh, typos or transposed incorrectly.

Great, thanks for clarifying that. So just so everybody knows, the final specifications numbers are basically going to be put on the shelf where we are clearing the proposed specs. We're also cleared those final numbers, but they'll just sit on the shelf till December when they'll go into the slipstream as they normally would. Yes, that's correct. So the only point of confusion is that there'll be new proposed specs coming out of here for the same species that we just did final specs on.

Absolutely. Clear as mud. It makes so much sense. Okay, was there anything else you wanted to touch on? Nope.

Excellent. Well, we will close out the plan team report then. Thanks to the plan team chairs and staff for guiding us and authors for guiding us through that. So, what we have left is our discussion items. These are going to be divided into 4 sections.

Before we start our discussion, I'll just check and make sure there's no public testimony. And I don't see any.

So, we, we have 3 proposed models plus the plan team report, and I think we'll start with the general GOA plan team report discussion, and we'll go to Rob first. Okay, yes, thank you very much. And, uh, yeah, I appreciate all the efforts of the plan team to present all the amazing amount of information that they went through during their, their plan team meeting, and also in addressing all our questions too. We also like to thank, uh, Sophia Wasserman for all the efforts, uh, contributed to the plan team, and definitely will miss those contributions. As far as just general comments for harvest projections in general, there was some discussion and within the— and mentioned within the plant team report of AFSC proposing to use projection, more standardized projection scenarios.

So there's a variety of projection scenarios they most frequently are using 2, 3, 6, and 7, which range from author recommendations to 5-year averages and max ABCs for 2 years. It looks like there's some goal, and I think the SSC should support the authors' move toward a standardized set of harvest projection methods. And then also, if, if for any reason there, um, there's a need or request or desire to do something different because of issues that come up, um, by that stakeholders, plan team, SSC, or others, um, have, then those could be addressed. Um, for the winter, um, Shelikhov acoustic trawl survey, uh, there was— we received a brief, um, overview on that during the, um, plan team presentation, and I won't— there's a lot of more details in the plan team report. I won't go through all those, but just at least to mention regarding this survey was successful except for the weather problems that they had.

And what another important part of that survey is the timing, and a couple factors given the— based on the fact that then there were no females greater than 40 centimeters spawning, there is no— and that, or if they were found either spawning or spent condition, It just indicates that the survey timing remained appropriate. So all for all, there's been some adjustments to that survey and some concerns over the timing of that survey. It sounds like they, they hit it just right this, this past year.

And that just the team will be revisiting the missing Shelikov's data and methods for estimating biomass. They used some interpretation— interpolation already. Some general recommendations Just in— as far as the DSIMs, we're very much encouraged by the application of the DSIM in ESPs, and especially the ability to include nonlinear threshold and non-stationary relationships, and the potential benefits as shown for both the Haro Tooth Flounder and the Gulf of Alaska pollock assessments. The possibility that multiple DSIM models could receive substantial support was an issue brought up by the plan team discussion, but also here at the, um, during our, um, our presentations. And I guess the SSC recommending that they support the plan team's recommendation that the DSIM developers consider methods for accommodating multiple models and hypotheses, and appreciate that they're seeking broad input on this, this uncharted effort.

Overall goal will determine model selection for sure, just like with any other statistical analysis or scientific endeavor. But even if the predictive capability is a priority, I guess I would recommend that we still place some importance on considering principles of parsimony. So while a standard AIC or a BSE, whatever model selection -based approaches used. That's a good start, um, but we also agree it's important to consider alternative models that have different ecological interpretations or important considerations that can be effectively justified by competing models on different purposes. For example, for predictive values or intents.

For timing, we briefly mentioned this during our joint plan team comments, but would support the recommendations to valued expertise of the ESP teams since they're experts on this and their input, and in particular to Cole's comment about not changing approved or what— however we're going to classify the DAGs that are, um, are, are being used or have been selected for use, that that's not, not something we want to address annually. And that's something we noted in our joint plan team report too. The team recommended The plan team recommended incorporating more socioeconomic information into the DSIM framework, and the SSC looks forward to future presentations and reports detailing the process for this integration.

For halibut DMRs, we note in the joint plan team, we mentioned the changes were mostly from Gulf of Alaska, but we didn't clearly make the statement, which I guess I should make now, that we, the SSC supports the proposed, the recommendations that were proposed, element DMRs for 2027-2028 for the Gulf of Alaska. We had a variety of comments on BRDs, but I think they were adequately captured in, but with our, the additional discussions that we had already associated with those. In specific stocks. Okay, thank you.

Great, thanks, Rob. Next, we'll go to Kaylin.

Thanks, Ian, and thanks, Rob, for the summary. I don't have anything to add.

Excellent, thanks, Kaylin. Other comments from the SSC?

Martin.

I think this is where it goes, um, concerns the BRDs. Um, I think our minutes should say that for stocks identified by the Council at the October 2025 meeting, um, as to whether there is flexibility in the survey-based spatial apportionments different than BRDs, assessment authors are requested to include: 1, a table showing regional BRD /Regional ABCs for the last 10 years, their associated CV if available, and the catch within that management area. 2, A stock— whether a stock structure template has been completed for the stock, and if so, a link to the document. And 3, any other factors deemed relevant to spatial allocation. I guess, which might include longevity.

I'll send it to you, Rod.

Great. Thanks, Martin. That will probably get added to our general BRD discussion rather than under just the Gulf, but that's great. We can take the wording just as is. Other comments from the SSC?

Sherry, and then we'll go to Franz. I should probably ask this clarifying question. So, and we're naming that for the 3 stocks that we'll be seeing in December?

Yeah, I mean, that's the list that we got, and I, and so, and they're all in the Gulf, so I'm just not I'm not quite sure if we're going to have a detailed— I don't know where the detailed BRD discussion is going to go and who's writing that. But, okay, so I'll give it to Ali.

So I guess I have a further follow-up to that, which is that in theory, this is a two-part process where We get information in October and the council has the opportunity to flag stocks before December. So there's the potential that a flag— a stock could get flagged at this meeting that would then be added to that list of 3. And I don't know that that will happen.

Seems like a big lift to ask all authors to go through this exercise. So maybe we just stick to what— stick to these 3 that are on the list. Uh, but I do think we should be aware that it is possible that another one could come up. Sherry, what if we— what if we say we're asking for 3 stocks unless there's other guidance given by the council in at their meeting? Uh, and I realize as far as, uh, requested stocks, and then hopefully, I mean, the plan team then could check a motion.

Hopefully it would be in a motion that they can check. Um, I don't want it to sound like those 3 don't bring it back, but it's, it is pending whatever the council says at this meeting.

Um, that seems like a good solution to me. Bronze, you had another point, I think. Yeah, I was actually going through my public minutes because I had some general comments on the DSAM in there that I I can probably bring up now since we're commenting on the DSAM here. I think Rob mostly captured that, actually. I just wanted to maybe reiterate that, you know, I think these types of models and model structures should be evaluated and updated periodically, but, you know, based on whoops, considerations of any new hypotheses and/or new evidence through the ESP process.

And I would also add that the considerations should take into account the availability of the data streams needed to kind of routinely update the relevant indices that go into it. And I apologize if Rob had already captured that.

The other thing, and I know Rob talked about the model selection. I think the author of the Pollak assessment also pointed out that the, you know, AIC should be used cautiously because it's known to kind of choose overly complex models. And if others agree, I would say that we might want to suggest that the primary emphasis should be placed on the ability of the model to project future recruitments based on, you know, some kind of retrospective prospectively future-out analysis like Cole suggested.

Hopefully that's consistent with what you had.

Yeah, I guess it is, but there— the Pollack is a good example of where you get a very similar projection from like 3 different models or 3 different— yeah, so which How do you deal with that then? I mean, they chose to use the AIC approach, so I guess to go with the most— well, I mean, if you're going to use a— take the most parsimonious model or the model that is most— seems most real.

Yeah, I guess that's what I was suggesting, the combination of the two. I don't think you can do one without the other. Go ahead, Franz. Um, yeah, and I think that's consistent with what I was suggesting. I just, like, if we were to rank them, I would put priority on, um, the projections.

And if you have to sort it out using others, you can do that.

On the leave future out analysis, it's not really a— I guess it is a projection.

So I guess I would offer a small friendly amendment, which would be that I think they were also presenting us with the, with the amount of the residual variance that was explained. And I think that's also a key component. So, to my eye, it sort of brings us back to what Cole was suggesting, which is that it's really a combination of these things should be all of those factors should be used when evaluating. Which unfortunately doesn't give us an easy, clear winner every time. But I would be reluctant to drop that from our list of things that ought to be looked at.

Yeah, I would completely support that too. I mean, it's just like any other analysis. There's generally not a clear picture on what the best option is. So having multiple metrics to evaluate, I think, is relevant, and leave it up to the, the, uh, those conducting the analysis to justify which direction they have chosen.

Curry? Yeah, I think not to beat a dead horse, but I mean, I think we should explicitly request that authors who are looking to justify their model choice should bring forward all three metrics.

Okay, any other comments, general plan team comments? Sheri? Yeah, I'm trying to get this together. I— Sarah, at the very end, asked about feedback to the Groundfish Plan Team on whether their discussion about the BRDs is what we were looking for, and so I was trying to come up with something kind of quickly.

I think that the SSE can say that we appreciated the qualitative discussion of stock structure and localized and possibility or opportunity for localized depletion. Um, I think that, um, we can appreciate their comment about possibly using survey variability for a metric. But ask for further— ask for further description. And I would say not in November.

Well, I don't know about timing. I just— I can't see that they're going to have time at their November meeting to dig into this hugely. I'll just go through. So qualitative discussion was great. In the future, I think that, you know, some more description of how to use the survey variability would be helpful.

You know, are they gonna— do they suggest using a percent confidence limit, same for all species? Should it differ by life history? Should it differ by how well they're sampled by a survey?

Things like that.

Um, those were the only two pieces of feedback that, um, that I had in this quick time.

Okay, seeing no further comments, uh, I will just briefly summarize. I won't go through everything Rob did because it Most of that was really clear. Just hitting the main points, we'll thank the authors and also Sophia Wasserman for her efforts on the plan team. We will support moving towards standardized projection scenarios for all assessments with any differences to be addressed on a stock-by-stock basis. We'll have some comments about the Shilohkoff survey with regard to timing, missing transects, and how that's going to be resolved.

We'll show our general support for the application of DSAM in both the Arrowtooth and Pollock models, and we'll have a number of comments on the use therein.

We will be supportive in the— we're supporting in the future continued consideration of alternative models and alternative complexities. We will note that we also agree that the DAGs don't need to be addressed annually. We'll encourage further introduction of socioeconomic information into the DSIM analyses, and we'll note the value of the expertise on the broad ESP teams. Then following our subsequent discussion on that, we will suggests that revisiting the DAGs should occur on a periodic basis, and it should occur through the ESP process rather than just directly for the author, and that not only should the performance of the approach be evaluated but also the data availability.

We will support a multi-pronged approach to model selection including AIC, Also definitely emphasizing the ability of the model to produce future recruitments, but also including the performance in explaining recruitment variability, and ask the author to go through all of those metrics.

For— we will also support the recommended Halibut DMRs. For BRDs, we will ask that for the 3 stocks identified by the Council already for potential BRD flexibility and any others should they arise subsequent to this meeting, we would request that in November and December we see a table with regional BRDs for the last 10 years, CVs if available, and catch for each of those regions, and a summary of stock structure template for that stock and any other factors factors that would deem to be relevant to the topic.

And then finally, some feedback for the Groundfish Plan Team on BRDs. We appreciated the qualitative discussion of stock structure and the potential for localized depletion, and we recognize there's some merit in further exploring survey variability as a metric, but we would encourage more description of how that would work and more development before we actually try to use it. So basically not this cycle.

Sherri? Yeah, I don't know that I mentioned this. The other thing in future cycles is, you know, we're all going to have to work on what is small. I'll think of a way to I don't know how to say that, I don't, but that's not for this year. So.

Okay, that concludes the GOA plan team section. We'll now go to proposed models for Demersal Shelf Rockfish, and we'll start with Jason.

Yeah, just a quick overview of the DSR complex in the Central GOA, Congo and West Yakutat, assessed under the Tier 6, as are all DSR species in the Southeast Outside Area except yelloweye rockfish. Yelloweye rockfish in the Southeast Outside Area is assessed under Tier 5 that under the approved assessment in December 2024 used a spatially stratified two-survey REMA model fit to the ADFNG ROV survey biomass estimates and a GAM model fit to CPUE estimates of yelloweye rockfish from the IPHC longline survey with additional observation error. I think the SSC certainly appreciates the author's substantial work to address our previous comments and explore additional data inputs. The author evaluated several different changes in the data inputs, including dropping IPHC survey stations that were outside the management area, updating the Southeast outside fishery average weights, expressing IPHC catch per unit of effort as weight per hook rather than number per hook, incorporating IPHC hook adjustment factor to account for competition, and updating the catch information through October 2026.

In addition, the authors evaluated new spatial-temporal standardizations of the IPHC survey data. Spatial correlation among the survey stations was modeled, included blocking for land to avoid correlation across land, and the author evaluated 7 candidate models with Tweedie error distribution and compared AR1 and IID spatial-temporal structure, also with one model with no spatial-temporal component and different combinations of depth and soak time covariates. Uh, SSC certainly appreciates the thorough evaluation as well as the bridging, stepwise additional bridging of potential covariates and spatial correlation structures, and then also the associated statistical measures of performance. The proposed methodology implements the SSC recommendations to standardize the IPHC CPE using a spatial-temporal model in the SDM-TMB, and so it's responsive, and SSC certainly appreciates that work.

I think we should still note that The IPH survey has limitations because it's designed primarily for halibut, and the sampling protocol, of course, is based on the first 25 hooks and may not be ideal for species such as yelloweye rockfish. But I think we should note that it's been used in the past, and it provides the most up-to-date available information on the yelloweye rockfish abundance for this assessment. I also highlight that the survey has been incorporated into the West Coast Yelloweye Rockfish Assessment, so it provides additional evidence that the survey can provide useful information as an indices.

One issue that was evaluated and highlighted in the plan team minutes was the CVs associated with the standardized IPHC index are smaller than those produced by previous approaches, which were bootstrap and GAM. So the author explored a range of CVs to evaluate the sensitivity of these model fits. And the authors noted that above-average recent CPE in the central southeast, outside, and southeast SSCO contributed to the increase in estimated biomass, while the smaller CVs assigned greater relative weight to the IPHC index. In the overall REMA model. So this potentially amplified the influence of the recent IPHC survey, on the estimation— estimated biomass.

So the plan team had extensive discussion about this, and I just highlight that discussion and note that the plan team, recommended bringing forward the previously implemented bootstrap approach for comparison to the current model-based CV. So I think the SSC should concur with this approach and notes that the issue is particularly important for areas with lower sampling effort where greater observation error increases the importance of specifying those CVs. Um, so in summary, I think we should agree with the planning recommendations to bring forward the previous, previously accepted base model for all stocks, which was Model 22.2. The author recommended SDM/TMB model for Tier 5 Yellow Eye, which is Model M2 in the assessment, and the, the SDM/TMB Model M2 for the Tier 5 Yellow Eye, but also with a bootstrapped IPHC or bootstrap CVs as an input to the REMA model. The SSC at this time I don't think should recommend bringing forward a Tier 6 assessment, um, for this cycle.

However, this may be something we have to consider in the future if the survey continues to be sampled at very low levels or there's multi-year gaps.

Um, Along those lines, I think we should note that we remain concerned about the lack of information for this stock and hope that the ROV survey can occur in the future. Also note that the number of IPHC stations sampled has declined in '24 and '25 with no sampling conducted in the East Yak and northern Southeast Outside Areas. And then just note that the SSC appreciated the author's response to its 2024 recommendation to consider combining those small areas with the larger central southeast area, and that we consider the current approach of modeling the spatial and temporal correlation to be more analytically appropriate than combining areas.

And then this is more of maybe housekeeping. I think we should note that the author included specification recommendations in the assessment. However, This is a draft assessment, not final recommendations for the SAFE, and SSC just recommends the author consult the guidelines regarding what should be reported in draft SAFE documents versus final documents.

Thanks, Jason. Next, we'll go to Brad. Thanks, Ian. That was an excellent summary, Jason. I just want to I concur with all of your comments.

Appreciate the team's work. This illustrates a really unfortunate and perhaps a common— more common scenario we'll see going forward where substantial work was done to refine the model, and now we're looking at degraded information to then put into the model. And so the author is battling both of those at the same time and has done, I think, an excellent job to very carefully carefully explore and document that. I want to just reiterate the SSC's sort of request for clarification in the document about how uncertainty in the annual area-level IPHC indices are calculated, and more importantly, how they're propagated into RIMA, noting the variability among the predicted grid cells alone doesn't really quantify that uncertainty. That's a component of it, and that we encourage just continued work.

It sounded both in the discussion that we heard from plan team as well as from the, the analysts that they're well aware of these issues and have a good plan going forward. So, um, I, so I support Jason's recommendation.

Thanks, Brad. Other comments on DSR?

Sherry?

Yeah, this I just wanted to note Aaron is a first-time author. I am— I forgot in my associations, I'm his second-level supervisor and Chris is his third-level supervisor, but neither of us contributed to the discussion.

Thanks for that, Sherri.

I do think we can welcome Aaron as a first-time author and commend him on his work, especially responding to a bunch of previous requests.

That was a very clear summary by both Jason and Brad, and I don't think we had any disagreement or substantive discussion, so I'm going to leave it at that as the, the summary.

We'll go next to our discussion on Thorny Heads, and first to Martin.

Thanks. Um, uh, okay, here we go. Um, GUA thornyheads, including short-spine thornyhead and long-spine thornyhead, are managed as a complex, but only short-spine thornyhead are assessed. The last assessment was in 2004, and they are assessed on a 2-year cycle.

Thornyheads are a Tier 5 stock. And are assessed using random— the random effects model.

The objective at this meeting was to identify one or more preferred runs for the December SSC meeting, where a model will be adopted for harvest specification.

In previous thornyhead assessments, REMA models were conducted by 3 depth strata, 0 to 500 meters, 500 to 700, and 700 to 2,000, for each management area in the Gulf of Alaska, Western, Central, and Eastern. This approach was adopted due to, due to the need to form consistent time series from the bottom trawl survey data, which surveyed different depths, and in different years. This approach could not be utilized this year because the redesigned survey in 2005 no longer uses a 500 to 700 depth strata. Instead, a new approach using two depth strata, 0 to 700 and 700 to 1,000, was evaluated. Various alternatives of treating the historical 500 to 700 depth strata were explored, including interpolation, uh, disregarding and dropping the surveys that did not survey this stratum.

Since these surveys occurred early in the survey time series in, um, 1990, '93, '96, and 2001, dropping them did not affect current trends, or management advice.

This model, Model 24.2F, was recommended by the author and the plan team to carry forward as the final assessment model. The SSC agrees with this recommendation. To reduce confusion, the SSC recommends that models with the new two-stratum structure be renumbered to 26 X. For a continuation model, the SSC recommends using only Model 24.2A, which updates the previously accepted model to the latest version of the REMA model but does not incorporate the 2005 survey data. This differs from the plan team recommendation, which included all models 24F, uh, A to F as a, as continue, um, continuity models.

Uh, the SSC notes that the 700 to 1,000 depth strata REMA model is based on 5 surveys that surveyed to these depths, and the last survey was in 2015. There are no plans to survey this depth stratum in the future, so consequently the biomass estimates for this depth stratum will become increasingly uncertain.

Short-spined are considered much less abundant— oh, sorry, long-spined thornyheads are considered much less abundant than short-spines in Alaskan waters, but this is not well documented in the assessment. The SSC requests that the relative abundance of longspine and shortspine thorny has been reported for the bottom trawl survey and from observer catch sampling.

That's it.

Thanks, Martin. Next, we'll go to Rob. Yeah, thanks, Martin. And yeah, thanks to the author on the, um, on this. Yeah, the review of this is the assessment and the changes that need to be made moving forward.

I appreciate all that. And, um, I guess the only comment I have is just, I guess, a request to the author to consider, uh, what the, um, if, if there's a way or if there's some value in considering the impact of the the changes to the longline survey with the alternate year sampling. It sounds like it was well under control in the way— the response to my question. But I mean, it's also a bigger question, right, in general, is there was a lot of discussion about the impact of this survey switching to alternate years, and is it going to be insignificant to some species or some stocks and not others? But it seems like that would be relevant to to understand.

But for this particular stock, I would just— that was my only recommendation. Thank you. Thanks, Rob. Other comments on thornyheads?

Martin?

Yeah, just a question for Rob. Um, is the cycle— will it line up with the bottom trail survey, or is it going to be on a different You're—. Yeah, the cycle lines up with the Bering Sea and Gulf of Alaska surveys. Okay, so then there's—. Yeah, so the longline and bottom trawl surveys will be on the same cycle as for the Aleutian Islands, Bering Sea, and the Gulf of Alaska trawl surveys.

So they're not—. They're not gonna— it's not gonna be a comprehensive survey because the Aleutian Islands survey is offset from the Gulf survey. Survey? Yes. Okay.

So the Aleutian Islands will be different from the Gulf or alternate years? For the longline survey. That's what I understand. Okay. Yeah.

Yeah. Okay. Same as for the trawl survey. So intrinsically, it shouldn't, I mean, it shouldn't be a problem because they're only using the index to line up with the REMA model, which seems like that's still going to be possible.

Okay. Yeah, that's fine if it's— if it's— if there is no— I guess, yeah, maybe my question wasn't answered fully, um, and— but it seems like you're answering it, whereas there— this is really no change at all to the model before because they used alternate years prior anyways.

Uh, no, historically they had Gulf—. They had longline surveys every year, and then a NOAA trawl survey every other year in recent decades. Yeah, that's right.

Yeah. Okay.

Okay, I have two clarifications. One, um, Martin, in your earlier summary, part of your summary, you mentioned the random effects model, but I think you meant the REMA model, not a different thing.

Yeah. No, I— yeah, that's fine. Just wanted to clarify.

And then For this last point, just to clarify, we're talking about the Sablefish Longline Survey, not the IPHC Longline Survey that we were just talking about in the previous item. And are we lining up with still wanting some description of how the interaction and survey design is gonna change or not?

Yeah, that's a good question, Ian. Maybe I am confusing something here, and please correct me if I am. It appears— yeah, I think Dana, go ahead, since Chris can't. I don't think there's too much to resolve here. I mean, we're going to have a GOA trawl survey and a GOA longline survey in the same year every other year, but there's— so we'll just be fitting the REMA model in the same way.

It'll fill in the gap years. We just won't have the annual GOA, but it's— we've already explored sort of the things. Kevin's already looked at that, so I don't think there's anything more we can do here.

Okay, that's what I'm looking for. Thanks.

Other comments on Thorny Heads?

Okay, I won't go through the whole summary for Thorny Heads, but just to note, um, that there have been some changes in the treatment of depth strata. Um, we're going to support the recommended model by the author and the plan team. With some renumbering to reflect that it's a 26 series model. We're going to recommend— I believe, Martin, correct me if I'm wrong, you recommended 24.2A as the continuity model. Yeah, great.

And we're going to request some additional documentation about the relative abundance in the survey and from observer sampling of longline or long line— long spine versus short spines.

And that's what we're going to request.

Okay.

We will then go to our discussion for Gulf Pollock. And first up, we've got Franz.

Thank you, Mian. So we received a presentation from author on this, on the proposed model runs. There was no public testimony specific to this stock. Uh, Goa Power is a Tier 3 assessment on an annual schedule, and the SEC appreciates the thorough documentation and clear presentation of proposed models for this assessment cycle. Um, for the next couple paragraphs here, please bear with me on some model numbers and discussion as there's been 2 years of work and model development that we're considering here, so I want to work through that kind of stepwise.

Um, the last full assessment was completed in 2024, Model 23D, technically the base model. A modified version of that accepted model, Model 23E, was endorsed by the SC in October '25, but of course the assessment wasn't completed because of the shutdown. The main structural change proposed for 2026 then is a transition from the bespoke TMB model to RSeattle. So the authors did a bridging, a very careful bridging analysis um, uh, and to do that they modified 23E to 23F to accommodate several differences in the implementation between platforms. And then the corresponding R-Seattle model, which was renamed model 26, reproduced, uh, model 23F essentially to machine precision.

Suggesting that the transition to RSeattle doesn't materially affect the assessment results. So, uh, we might just want to note that Model 26 is several steps removed from the 2024 accepted base model, and we appreciate the careful stepwise bridging analysis to a new modeling platform. Uh, and with that, I, I would suggest that we support the use of the new model platform platform for the 2026 assessment, because again, R Seattle allows kind of more rapid and flexible exploration of new features, including like we saw the integration of environmental covariates in a structured framework.

So with that, the author and plan team proposed two model variants for this assessment cycle. Model 26A that estimates recruitment as random effects, which resulted in some shrinkage of recruitment estimates towards the mean, as would be expected, but with minimal impacts on the SSB estimates. So this removes the model one more step from the previously accepted base model, but again, I suggest we accept model 26a as the new base model for this site. Cycle. So that's kind of a lot, but I think they did a really, you know, careful analysis with the bridging and comparisons of the models.

You know, essential figures were in the document and reviewed by the plan team. So I think, uh, I'm, I'm okay with taking these multiple steps to get us to a new base model for, for this cycle.

Uh, then the second Model A proposing 26B extends, uh, 26A by incorporating the, uh, DSUM linking recruitment to a set of ESP variables based on a published study, uh, this moderate causal— based on the moderate causal directed acyclic graph, or DAG. The SSC appreciates the author's thoughtful approach to considering the use of environmental drivers of recruitment in the assessment as supported by the, uh, Champagnat et al. Paper.

Um, I suggest that we should agree with the author and plan team to bring forward Models 26A and 26B in December for catch specifications. Uh, thus the main decision point in December for us may be whether the DSM that's incorporated in 26 is sufficiently mature for inclusion in the operational assessment.

In general, I think the SSC supports the continued development of the DSM approach and considers integration of ecosystem information directly into the assessment, as is done here for the, for the first time, really, a promising avenue for improving recruitment estimation and prediction for this stock.

Uh, while the DSUM substantially reduces, uh, unexplained recruitment variability and has been documented to produce better short-term recruitment projections, um, there may be some operational constraints on incorporating, um, dynamic recruitment projections into management advice at this time, and therefore, uh, I think we should agree with the use of average recruitment and projections for this assessment cycle.

However, if the author can include projections this year, I think we should leave that to the discretion of the authors, and at minimum, the SSC might request that the authors and planning team provide a brief description this November of the requirements and steps needed for using the DSEM projections of future recruits for calculating OFL and ABC in the future.

Um, the SSC also supports the plan team request for additional evaluation of the DSEM before December, including retrospective estimates of the DAC parameters, sensitivities to removal of individual data streams, evaluation of predictive skill for terminal recruitment, and quantification of the contribution of terminal recruitment to the projected biomass and spawning biomass.

The— we might want to say the SSC considers stability of the estimated relationships and sensitivity to individual ESP data streams important considerations before adopting the DSEM operationally.

A couple other DSEM considerations and discussions that are relevant to this particular model. The plan team also discussed the need to consider, you know, multiple plausible DAGs and competing ecological hypotheses rather than treating a single selected causal structure as known. And in general, the SSC notes For this assessment, the authors replicated the analysis of Champagnat et al. In RSeattle to compare and evaluate a range of approaches for including environmental covariates, and the results were consistent with the paper suggesting that the moderate causal model here provides a suitable basis for evaluation in this assessment.

Um, I think the next point here may have been covered in our overall, our general comments on BSAM. Um, uh, the risk of repeating that, I didn't catch all that earlier. Uh, the SSC also reiterates that any risks associated with environmental conditions that are incorporated into the assessment model should no will no longer be considered in the risk table unless they capture risks that are independent of the mechanisms in the model. For example, there's effects on growth when only effects on recruitment are considered in the model.

Um, so that's everything on DSAM. Uh, on the bottom trawl survey, the SSC shares concerns about the persistent poor fit to NIMS bottom trawl survey, which now extends across 5, I believe, biennial surveys beginning in 2017. And that issue is quite important because, because of the informative prior on the bottom trawl survey catchability, that's critical to kind of anchoring the absolute scale of the stock.

The SSC appreciates the author's consideration of alternative hypotheses hypotheses for what may cause this discrepancy and looks forward to further evaluations of the biological hypotheses regarding vertical availability of walleye pollock to the bottom trawl survey and pollock migrations between the Bering Sea and Western Gulf.

Um, and then I had a comment on the May survey, but I think Rob captured that one as well, so So I'll skip that. And that is, that is it.

Thanks, Franz. Next, we'll go to Chris. Yeah, thanks. Thanks, Franz. That was great.

Just a couple of things. One correction, just because I didn't recognize, it's like, this is a biennial survey. So there wasn't a lot of sea ice in that last survey year because the last survey year was 2025 and not this past year, 2026, right? So that's just a thing to think about, just kind of thinking about those big toasts that were previous year, not this current year. So that was just a misread on my standpoint.

The only other things to— I will just reiterate is my general concern of the poor fits for the trawl survey over the last 5 iterations. I mean, that trawl survey incorporates the spatial extent from Southeast Alaska all the way up through Prince William Sound to the Alaska Peninsula. The most comprehensive and consistent of the surveys. So that is a, a pretty large concern from my standpoint, and getting things as correct as we can. I think it's a non-trivial thing that the fits over those last 5 surveys, and I would just from my standpoint suggest some prioritization of how to deal with that is, has a bigger potential downstream impact on how we move forward and with a fishery.

So that's just one thing to reiterate there. And I think really, I think those are my main comments in addition to what Franz has already presented. So thank you.

Great, thanks, Chris. Other SSC comments on Gulf Pollack?

Okay, I will just briefly summarize a couple of the most salient points. I think the big one will note the chain of models leading up to the 26 series. The key recommendation, I think, for us is that we are supporting Model 26A as the new base model for this stock, and in agreement with the plan team and the authors, We'd like to see both 26A and 26B, which incorporates DSAM, to be brought forward for the rest of this cycle in November and December.

We will have a number of comments about— positive comments about the development of DSAM and the implementation.

We will agree that at this time We would recommend that Model 26.B, when we see it in December, does not use the DSAM to make recruitment projections but sticks with the average. But we would like to see a description of the steps that would be needed to achieve that in the future. We have a list of things we will ask in terms of additional evaluation of DSAM, including data streams, predictive skill, quantification of the projected effects on recruitment and spawning biomass. We'd also like to see sensitivity analysis to the individual data streams that are included in the, the DAG.

We will highlight, and I don't think we did cover this earlier, so we will capture it here, we will just note that risks associated with environmental conditions that are included in the DSAM analysis don't— shouldn't be also included in the risk table if they are in the DSAM, if we're using a model that is including DSAM in December. We will highlight the concern over the continued lack of fit to the bottom trawl survey, noting that it's a comprehensive survey and that it's an important scaling component in this model, and we will suggest that this evaluation of this should be a priority moving forward in the stock assessment. And we will— we appreciated the work that has been done on alternative hypotheses, including vertical availability and migration, and encourage further work on both of those avenues.

Okay, I think that brings us to the end of our discussion of Golf Pollack, last chance we've missed anything.

Okay. With that, I am happy then to pause for a second and look to Ali to just remind us of our deadlines, which I know are going to feel pretty tight now. Yes, they are. And I know that folks are traveling tonight. So, yes, if we can have draft draft sections into the main report by 10:00 AM tomorrow morning.

I know that's tight. If you are really having trouble, just send us a note and we'll see what we can do. Please don't forget slides. We will need those by 10:00 AM as well, although obviously they are a bit lower priority. We're going to have the main report open for full SSE comments throughout the day.

Tomorrow, and we are currently planning on closing that at 6:00 PM Alaska time tomorrow evening so that we can be ready to present to council the following day.

And hopefully everyone's access issues have been resolved, but if not, shoot me an email. Thank you. Yeah. Do you do slides for proposed models? Yes.

You do. Okay. I mean, why would the Commission care? There's a— yes, so in the presentation, there's a template ready for you on model runs. It's a single slide, so hopefully it shouldn't be too onerous.

But yes, we do need to have model run slides. And so you're looking for really succinct sort of, these are the models and—. Yes, but there's room for providing additional recommendations if you need that. So add slides as you need. It's much easier for us to remove extraneous things than to add them ourselves later.

So you can err on the side of giving us more, and that's, that's easier for, for us to— it's much easier for us to cut later if we need to. Thank you. Thanks, Allie. Sherri. Yeah, if it, if it helps, and the presentations that we usually do for model runs, the things that I think that the council may be most interested are, I mean, yes, name the model numbers and what we're going to do, but also, I mean, how's this, my high-level summary often is, you know, these are the alternative models.

This is the base model from whatever year, and then also just name, you know, This one explores natural mortality. This one explores selectivity. So broad concepts so that if, um, if council members are interested, they can dig more, but not tons.

And last comment on minute writing. If you're on several items, which most of you are, and you're not sure which one to start first, start— do the hard ones first. Please, because then we have more time to work on the minutes on those and save the easy ones for later. And I'm sure you know by now which ones of those are.