meeting with managers (Apr 2019) - prep musings

[ann-marieH]

Cam asked for ways to make the single vs mixed stock work more relevant to managers. I'm expanding that to general thoughts for April 2019 mtg. If I don't make a home for my thoughts here, I'm going to lose them! Add'l thoughts/Qs welcome.

general thoughts

• keep as generic/non-FRSK as possible (most things applied to FR SK would be done via FRSSI process & model, anyways), so context:
  • we have data for FR SK & have seen how it behaves in other models (so would have an idea when results are odd),
  • FR SK covers the entire gamut of WSP color rainbow,
  • can create data-deficient stocks in the model based on FR SK and see how different MPs would perform on them vs the "real" stocks, whereas we can't do the opposite
• focus meeting on the questions / scenarios - keep model details to a minimum or we will lose them. what if...
  • 20 min overview of model
  • incl. Dept priorities it can help with: e.g. SARA, WSP, climate change, etc.
  • pose the Qs we want answered today
  • have a discussion on priority topics - have we missed any?
  • note: if we have, we might be able to pull together some scoping Qs specific to the new topic for the group over lunch
  • topic # 1:
  • what we think the Q is, what our add'l Qs are
  • add'l model info associated with topic 1, if needed (10 min max)
  • revisit Qs, have discussion, pull out our default suggestions if needed
  • topic # 2: ...repeat

what are the Qs we want answered??? (and what are our default "answers"?)

• confirm main workplan priorities (below headers?)
• what are the questions/issues behind the priorities? i.e. not "what MPs do you want us to test" but "what's the underlying concern behind the priority" and "what are you trying to achieve?"
  • though yes, it would be good to know if there are any specific HCRs or MPs that they are interested in trying out

mixed vs single

• am thinking that a SC Chum based MP might be helpful to show - e.g. 20% in mixed, then each remaining "terminal" area has their own rules ... not that I know what any of them are. Pieter would be a good person to ask.
  • for illustrative purposes, each MU could be a "terminal" area
  • think the existing HCR (singlestk:retro) would work for this illustration:
  • 25% TAC caught in mixed stock; then different "terminal" harvest rules --> focus on effects that different terminal harvest rules have on one MU (e.g. Summers)

harvest levels and rebuilding time

• suggest using BMs from historical time series (e.g. p25 & p50 from escapement time series) instead of Sgen type when showing these results, b/c:
  • I don't know how many stocks have Sgen values calc'd
  • easier to understand/it has meaning to non-biologists
• ditto using fixedERs as opposed to TAMs
• fixed escapement would be an interesting comparison...use Summer MU & fixed escapement "goal" associated with mid-range of TAM?

impacts of reduced productivity

• I have some recursive Bayes Ricker alpha results that may help inform %reductions (i.e. 50% isn't out of the realm of what's suggested occurred in the past on a few stocks, but is on the pessimistic side ... so is probably defensible for a generic "what if" for productivity)

consequences of less accurate or less frequent estimates of abundance

• impact of less accurate and/or less frequent spawner escapement programs I think will be of very high interest. I don't think a S/R time series is created in the model, though?
  • at this stage, would it be worth just doing a retrospective analysis? e.g. take one stock (e.g. Chilko) and put it thru its paces? then explain how we would model it? .... this idea is very prelim!!!

[carrieholt]

Thanks for this Ann-Marie. Yes, that’s a great point to focus on management questions/issues and not MPs or OMs. (among other good points) Maybe we should pull together more concrete ideas on structuring the workshop late Feb/early March. Sound okay Cam?

Carrie

From: Ann-Marie Huang notifications@github.com Sent: February-13-19 10:19 AM To: CamFreshwater/samSim samSim@noreply.github.com Cc: Subscribed subscribed@noreply.github.com Subject: [CamFreshwater/samSim] meeting with managers (Apr 2019) - prep musings (#3)

Cam asked for ways to make the single vs mixed stock work more relevant to managers. I'm expanding that to general thoughts for April 2019 mtg. If I don't make a home for my thoughts here, I'm going to lose them! Add'l thoughts/Qs welcome.

general thoughts

• keep as generic/non-FRSK as possible (most things applied to FR SK would be done via FRSSI process & model, anyways), so context:
  • we have data for FR SK & have seen how it behaves in other models (so would have an idea when results are odd),
  • FR SK covers the entire gamut of WSP color rainbow,
  • can create data-deficient stocks in the model based on FR SK and see how different MPs would perform on them vs the "real" stocks, whereas we can't do the opposite
• focus meeting on the questions / scenarios - keep model details to a minimum or we will lose them. what if...
  • 20 min overview of model
    • incl. Dept priorities it can help with: e.g. SARA, WSP, climate change, etc.
  • pose the Qs we want answered today
  • have a discussion on priority topics - have we missed any?
    • note: if we have, we might be able to pull together some scoping Qs specific to the new topic for the group over lunch
  • topic # 1:
    • what we think the Q is, what our add'l Qs are
    • add'l model info associated with topic 1, if needed (10 min max)
    • revisit Qs, have discussion, pull out our default suggestions if needed
  • topic # 2: ...repeat

what are the Qs we want answered??? (and what are our default "answers"?)

• confirm main workplan priorities (below headers?)
• what are the questions/issues behind the priorities? i.e. not "what MPs do you want us to test" but "what's the underlying concern behind the priority" and "what are you trying to achieve?"
  • though yes, it would be good to know if there are any specific HCRs or MPs that they are interested in trying out

mixed vs single

• am thinking that a SC Chum based MP might be helpful to show - e.g. 20% in mixed, then each remaining "terminal" area has their own rules ... not that I know what any of them are. Pieter would be a good person to ask.
  • for illustrative purposes, each MU could be a "terminal" area

harvest levels and rebuilding time

• suggest using BMs from historical time series (e.g. p25 & p50 from escapement time series) instead of Sgen type when showing these results, b/c:
  • I don't know how many stocks have Sgen values calc'd
  • easier to understand/it has meaning to non-biologists
• ditto using fixedERs as opposed to TAMs
• fixed escapement would be an interesting comparison...use Summer MU & fixed escapement "goal" associated with mid-range of TAM?

impacts of reduced productivity

• I have some recursive Bayes Ricker alpha results that may help inform %reductions (i.e. 50% isn't out of the realm of what's suggested occurred in the past on a few stocks, but is on the pessimistic side ... so is probably defensible for a generic "what if" for productivity)

consequences of less accurate or less frequent estimates of abundance

• impact of less accurate and/or less frequent spawner escapement programs I think will be of very high interest. I don't think a S/R time series is created in the model, though?
  • at this stage, would it be worth just doing a retrospective analysis? e.g. take one stock (e.g. Chilko) and put it thru its paces? then explain how we would model it? .... this idea is very prelim!!!

— You are receiving this because you are subscribed to this thread. Reply to this email directly, view it on GitHubhttps://github.com/CamFreshwater/samSim/issues/3, or mute the threadhttps://github.com/notifications/unsubscribe-auth/AVatNfrwlaSu6fhpRZ4y9dDKqdIJjE-gks5vNFcDgaJpZM4a5_Tv.

[carrieholt]

Thanks for this Ann-Marie. Yes, that’s a great point to focus on management questions/issues and not MPs or OMs. (among other good points) Maybe we should pull together more concrete ideas on structuring the workshop late Feb/early March. Sound okay Cam?

[CamFreshwater]

Yeah this lines up well with my thinking after our call yesterday Ann-Marie.

For now I'm planning to refine the single stock analyses by comparing the following suite of MPs

• Modest fixed exploitation rate (35% ~mean with the TAM rule) with variable allocation
  • Allocation ranges from 0-100% mixed stock
  • Only retrospective single stock HCR
• Ultra-conservative fixed exploitation rate based on the Umsy of the "weakest" stock and no changes in allocation
• Liberal fixed exploitation rate based on either mean Umsy among stocks in an aggregate or some estimate of aggregate Umsy and no changes in allocation

These will all be explored across a few productivity (FWIW in the low alpha I'm using 35% decline based on Kalman filter estimates that Carrie forwarded me) and en route mortality scenarios. As previously discussed outputs will focus on aggregate PMs as well as CU-specific PMs, but constrained to a single MU (likely Summer or Lates).

I'm happy to adjust the harvest control rules to address other questions or refocus my efforts on other aspects of model development (e.g. survey design, rebuilding timelines), but would prefer to discuss those with some of you prior rather than winging it on my own ;). On that note I definitely think a meeting on structuring the workshop would be helpful. If we want that meeting to inform model development it would be great if it could happen sooner rather than later, otherwise anytime before mid-March is fine with me!

[ann-marieH]

Suggest using Summers as example. I'm thinking:

• LStu & Quesnel will probably provide enough contrast for Chilko
• don't have to contend with needing to explain everything thru LShu cycle lens - cycles give non-FR SK people headaches, or so I've been told ... will be more "generic" than Lates
• don't have to contend with the specter of Cultus (which some of the managers are very sensitive about)

[carrieholt]

Thanks both. How about we work out workshop structure at our next biweekly meeting?