1 of 48

1

2 of 48

Meeting Goals

Review types of questions salmonMSE can address

Feedback on key assumptions in the operating model (single population)

Feedback on proposed outputs

Provide update on conditioning model to estimate parameters

Identify process to solicit further feedback on model structure and outputs and for reviewing case application for WCVI Chinook

2

3 of 48

Who we are

Technical Advisory Group for salmonMSE

Role: Provide technical advice on model structure and conditioning, model outputs, ease of use, and application to the WCVI Chinook case study

Participants: DFO and First Nations, including SEP, WCVI Area Stock assessment, DFO Core Science, Resource Management, SARA, and PSSI

If prototype successful, a broader TAG will be created in future phases to guide the development and application of the tool more generally

3

4 of 48

4


Introduction/review of last meeting	1:00-1:10
Review of types of questions salmonMSE can address	1:10-1:15
Overview of operating model: Key assumptions Comparison with AHA Example visual outputs	1:15-:145
Discussion	1:45-2:20
Health break (as required)	2:20-2:30
Update on model conditioning	2:30-2:40
Proposal for further feedback and model application with targeted groups	2:40-2:50
Discussion	2:50-3:00

Agenda

5 of 48

5

Pacific salmon life cycle

Drivers of change

6 of 48

6

Habitat

Hatchery

Harvest

Pacific salmon life cycle

Management levers

7 of 48

Current and Emerging needs

Evaluate performance of candidate management actions for rebuilding plans under the Fisheries Act and Recovery Potential Assessments under the Species at Risk Act, including strategic prioritization of harvest, hatchery, and habitat management levers

Support SEP programs and planning: evaluating trade-offs among benefits and risks of hatchery production

Develop transparent, accessible and flexible tools to support co-governance arrangements

Account for risks and uncertainties under DFO’s Precautionary Approach, including those related to genetic introgression from hatcheries
Consider environmental conditions affecting stocks, including those in both marine and freshwater stages
Plan strategically for impacts of climate change

7

8 of 48

Decision support tool for medium-long term strategic planning��Management Strategy Evaluation

8

Harvest, Hatchery and Habitat Management

Fishery and hatchery sub-models

MSE is tool for evaluating candidate management actions in simulation, and can help identify risks and benefits of MSs given uncertainties in underlying dynamics

Management strategy - the sum of all the management measures selected to achieve the objectives of the fishery, including monitoring, assessment, harvest, hatchery, and habitat management actions. When management actions depend on the state of the system (as in this example), policies that include them are called feedback or ‘closed-loop’ policies, and simulations that include them are called ‘closed-loop simulations’,

Operating model- “represents the simulated dynamics of the fishery system and will include components for the population and ecosystem dynamics, fishery resources, and social-economic aspects of the fishing fleet dynamics” (Kaplan et al. 2021)

9 of 48

salmonMSE

this tool can provide a risk-based approach for prioritizing management levers

identify trade-offs in achieving biological and harvest objectives among management levers for medium-to-long term strategic planning

9

Harvest

Hatcheries

Habitat

10 of 48

Last meeting- January 2024

Key messages on scoping

This tool could support strategic planning at medium-long term and would work with broader-scale tools (e.g., RAMS) and finer-scale tools (e.g., environmental niche modelling of Josie Iacarella)
Suggestion to focus on the interaction among levers, and between levers and habitat threats (e.g., as identified in the ‘Follow the Fish’ project, and Climate Vulnerability Assessments), and less on exploring new HCRs as these are fairly hard-wired for Pacific salmon
There is value in working towards a universal understanding of key objectives in any/all salmon systems based on existing policies and legislation. Although this is not the goal of this project, it could contribute to laying these out. Being clear about these objectives will be essential in any application of the tool
Suggestion to have targeted meetings with specific client groups to get feedback on model structure and types of outputs/visualizations. (to follow up on after this meeting)

10

11 of 48

Timeline

11

Project initiation Nov 2023

Technical Advisory Group Meeting

Jan 2024

Initial prototype of (operating) model complete, accessible online
Documentation developed (equations and vignettes)

Phase 1

Phase 2

Phase 3

Funding: PSSI(year 1) PSSI(year 2) PSSI(TBD)

Updated prototype based on feedback from SEP and team leads
Developed initial reporting outputs
Developed initial conditioning model

Review of model PIs and SEP

Technical Advisory Group Meeting

Oct 2024

Review model structure with client groups
Update conditioning model based on feedback
Develop and multi-population component
Application to WCVI Chinook

Small group meetings to review model & application to WCVI Chinook

Technical Advisory Group Meeting

Spring/summer 2025

Apply salmonMSE example management questions
Scope application to other case studies
Explore Gui-based interface
Public release of R package with Tech Report

Training on use of salmonMSE Winter 2026

&Fall/Winter 2025-6

12 of 48

salmonMSE

Types of management questions that proposed tools can address

Prioritizing among harvest, hatchery, and habitat management levers

Evaluating impacts of:

hatchery impacts on degree of ‘wildness’ of populations
mixed-stock fisheries on status of component populations/CUs
mark selective fisheries
incorporating parental-based tagging PBT into assessment framework (costs vs benefits)
habitat improvement (increase in survival or capacity) on management performance
uncertainty in population productivity on management performance.
Climate-driven changes in survival or capacity (as identified from “Follow the Fish” program on WCVI and/or RAMS)

E.g., What is the maximum hatchery production possible while maintaining ‘wild-integrated’ population status, as defined by proportion natural influence PNI goals by Withler et al. (2018)?

12

13 of 48

What does the model look like?

13

14 of 48

14

Natural-origin spawners(NOS)

Hatchery-origin spawners (HOS)

15 of 48

15

Natural-origin spawners(NOS)

Hatchery-origin spawners (HOS)

Spawning in natural environment, including density-dependent survival

16 of 48

16

Natural-origin spawners(NOS)

Hatchery-origin spawners (HOS)

17 of 48

Model assumptions, comparison with AHA

What is AHA?

Excel tool to evaluate hatchery, harvest, habitat management options developed by the Hatchery Scientific Review Group of the Pacific States Marine Fisheries Commission

salmonMSE is AHA and more..

17

18 of 48

Model assumptions, comparison with AHA

18

AHA	salmonMSE
Life cycle model by life stage Single brood-year return	Annual age structure model of life stages (juveniles, return, escapement, spawner, egg, smolt) Supports multiple brood-year return
Beverton-Holt stock-recruit relationship for density-dependent smolt production	Beverton-Holt or Ricker SRR
Specify starting spawners and project to determine long-term equilibrium properties	Condition starting abundance from estimation model and project forward in time (evaluate short-term vs. long-term dynamics)
Stochastic for marine survival (SAR)	Stochastic natural survival, maturity (by age class), productivity, capacity, starting abundance
Time-varying SAR to incorporate climate effects Habitat effects implicit in specified productivity/capacity parameters	Time-varying (cyclical) natural survival, maturity by age class Differentiate freshwater/marine climate + habitat effects on survival by assigning to age class Change productivity/capacity proportionately to evaluate habitat effects
	Supporting visualizations for DFO objectives

General + Habitat

19 of 48

Model assumptions, comparison with AHA

19

AHA	salmonMSE
Terminal fisheries	Preterminal + terminal fisheries
Specify differential harvest rate for natural/hatchery origin return	Mark-selective fishing Specify mark rate + harvest rate of kept catch of marked fish Realized exploitation rate from release mortality (accounting kept and discarded catch)

Harvest

20 of 48

Model assumptions, comparison with AHA

20

AHA	salmonMSE
Selective broodtake (implied mark rate = 1) to meet pNOB target	Selective broodtake determined by specified mark rate
Maximum brood/escapement ratio of natural origin fish	Max. brood/escapement by natural origin fish (AHA) or total escapement (SEP guidelines)
Hatchery egg survival is density-independent	Hatchery egg survival is density-independent
Sub-yearling/yearling releases alone determine target broodtake (differential hatchery survival between release groups)	AHA assumptions + Density-dependent survival of sub-yearlings (competition with natural origin young)
En-route mortality of escapement to spawning ground, in-river HOS removal from spawning ground	En-route mortality of escapement to spawning ground, in-river HOS removal from spawning ground
Fitness calculations from hatchery fish in natural environment	Fitness calculations from hatchery fish in natural environment

Hatchery

21 of 48

Model assumptions, comparison with AHA

Comparison with AHA and salmonMSE

21

22 of 48

salmonMSE

22

23 of 48

Example question for a hypothetical population?

What is the maximum hatchery production possible while maintaining ‘wild-integrated’ population status, as defined by proportion natural influence PNI goals by Withler et al. (2018)?

What are the trade-offs between PNI and harvest objectives?

How robust is this choice to uncertainty in freshwater and marine survival?

(What are impacts of habitat improvements?)

23

24 of 48

24

Hatchery <- new(

"Hatchery",

n_yearling = 10000,

n_subyearling = 0,

s_prespawn = 1,

s_egg_smolt = 0.92,

s_egg_subyearling = 1,

Mjuv_HOS = Bio@Mjuv_NOS,

gamma = 0.8,

m = 1,

pmax_esc = 1,

pmax_NOB = 0.7,

ptarget_NOB = 0.51,

phatchery = 0.8,

premove_HOS = 0,

theta = c(100, 80),

rel_loss = c(0.5, 0.4, 0.1),

fec_brood = c(0, 0, 5040),

fitness_type = c("Ford", "none"),

zbar_start = c(93.1, 92),

fitness_variance = 10,

selection_strength = 3,

heritability = 0.5,

fitness_floor = 0.5)

Production targets

https://docs.salmonmse.com/articles/example.html

Marking targets

Fitness parameters

25 of 48

25

Harvest <- new(

"Harvest",

u_preterminal = 0,

u_terminal = 0.203,

MSF = FALSE,

release_mort = c(0.1, 0.1),

vulPT = c(0, 0, 0),

vulT = c(1, 1, 1)

)

Harvest rate and mark-selective fishery options

https://docs.salmonmse.com/articles/example.html

26 of 48

26

Habitat <- new(

"Habitat",

capacity_smolt_improve = 1,

kappa_improve = 1

)

Habitat improvement targets (scalars on capacity and productivity)

Additional levers elsewhere (e.g., changes in maturity, juvenile mortality, egg mortality schedules)

https://docs.salmonmse.com/articles/example.html

Identifying mechanisms that translate habitat changes (e.g., increased scouring) to population-level impacts (e.g., change in smolt production, time-varying survival, fecundity) would be valuable to parameterize this model

27 of 48

Simple example

A risk-based approach evaluates management options when the true state of nature is uncertain.

What we can calculate with a stochastic model

Sample biological parameter (productivity, recruits/spawner) from a distribution
Project model to obtain distribution of state variables (PNI)
Calculate probability of meeting targets, e.g., how likely PNI > 0.8 (threshold for ‘wild-integrated’ category) at end of projection

27

https://docs.salmonmse.com/articles/example.html

28 of 48

28

Run simulation of random Monte Carlo trials accounting for:

Genetic risks of introgression
Uncertainty in underlying population and fitness parameters
Age structure
Life-stage specific threats/drivers

Probability PNI > 0.8 = 0.13

29 of 48

Simulation analysis

Decision-making context:

Explore four hatchery production options (0, 5, 10, 15 thousand releases) against three states of nature (mean productivity of 3, 6, 9 recruits/spawner)

Performance metrics:

Probability PNI > 0.8, Probability Catch > 60

Other details:

Three year life cycle, single brood-year return
Smolt carrying capacity = 17,250
Terminal fishery exploitation rate of 0.5
Non-selective broodtake, max. brood/escapement = 0.5
Start off with 1000 NOS and HOS, project 50 years to identify equilibrium values (AHA approach)

29

https://docs.salmonmse.com/articles/decision-table.html

What is the maximum hatchery production possible while maintaining PNI>0.8?

What are the trade-offs between PNI and harvest objectives?

How robust is this choice to uncertainty in productivity?

Management Q

30 of 48

Time series figures

30

https://docs.salmonmse.com/articles/decision-table.html

Visualize outcomes from simulation, for example, composition and abundance of spawners (annual median across simulations)

Marginal effects:

Higher productivity → more spawners
More hatchery production → increase pHOS (implied low PNI)

31 of 48

Time series figures

31

https://docs.salmonmse.com/articles/decision-table.html

Visualize outcomes from simulation, for example: PNI, pHOSeff, pWILD, fitness

Marginal effects:

More hatchery production → increase pHOSeff and decrease in PNI
Especially at low productivity

32 of 48

Decision tables

Steep dropoff in PNI_80 at medium/high production option (10, 15 thousand)

Exclude options that perform unacceptably low, for example, PNI_80 < 0.50

Robustness = obtaining good performance for a selected management option (column) across states of nature

32

Management option

State of nature

https://docs.salmonmse.com/articles/decision-table.html

33 of 48

Decision tables

At specified harvest rate, likely need hatchery production to meet catch target

33

Management option

State of nature

https://docs.salmonmse.com/articles/decision-table.html

34 of 48

Tradeoff figure

Best management option in top-right corner (high catch & PNI)

Tradeoff seen when the management options align along “off-diagonal”

(top-left & bottom-right)

High hatchery production gives more catch at the cost of PNI

Advice is presented in terms of options and characterizes tradeoffs (after excluding unacceptable choices)

34

https://docs.salmonmse.com/articles/decision-table.html

35 of 48

Results from simple example

What is the maximum hatchery production possible while maintaining ‘wild-integrated’ population status, as defined by proportion natural influence PNI goals by Withler et al. (2018)?

What are the trade-offs between PNI and harvest objectives?

How robust is this choice to uncertainty in freshwater and marine survival?

Five or ten thousand was the highest evaluated option where PNI_80 > 0.50

35

https://docs.salmonmse.com/articles/decision-table.html

Clear tradeoff between long-term mean in PNI and catch in all states of nature. Likely need hatchery production to meet catch target (60)

Ten thousand meets target only in the high productivity scenario. Five thousand more likely to reach PNI goals averaged across states of nature

36 of 48

Do visual outputs support provision of Science advice to support decision making? Suggestions or additions?
Collaborate with the FSAR Working Group for Pacific salmon to ensure figures are consistent with template requirements for CSAS FSAR documentation and/or Rebuilding Plans

36

Discussion questions

37 of 48

Given the goal of providing medium-long term strategic advice on trade-offs among management levers, are key assumptions appropriate?

37

Discussion questions

38 of 48

Given the goal of providing medium-long term strategic advice on trade-offs among management levers, are key assumptions appropriate?

38

Discussion questions

Annual age structure model of life stages (juveniles, return, escapement, spawner, egg, smolt), supports multiple brood-year return	Preterminal + terminal fisheries	Selective broodtake determined by specified mark rate
Beverton-Holt or Ricker SRR	Mark-selective fishing: - Specify harvest rate of kept catch (marked fish) - Realized exploitation rate from mark rate and release mortality (accounting kept and discarded catch)	Max. brood/escapement ratio by natural origin fish (AHA) or total escapement (SEP guidelines)
Condition starting abundance from estimation model and project forward in time (evaluate short-term vs. long-term dynamics)		Hatchery egg survival is density-independent
Stochastic natural survival, maturity (by age class), productivity, capacity, starting abundance		Density-dependent survival of sub-yearlings and natural origin young
Time-varying (cyclical, persistent, etc.) natural survival, maturity by age class Change productivity/capacity proportionately to evaluate habitat effects Differentiate freshwater/marine climate + habitat effects on survival by assigning to age class		En-route mortality of escapement to spawning ground, in-river HOS removal from spawning ground
		Fitness calculations from hatchery fish in natural environment

39 of 48

Conditioning model�

Modification of Walters/Korman ECVI Chinook model for population reconstruction from CWT and escapement estimates, covariates on survival

39

40 of 48

Future development (Winter 2025)�

Application of conditioning model to WCVI case study (estimate parameters from data, borrowing parameters)

Multi-population model incorporates mixed-stock fishing and straying (large systems can impact fitness in small systems)

40

41 of 48

Targeted meetings with self-selected groups (e.g., SARP, SEP, Nations, Stock Assessment, FM) winter 2024/5 to solicit feedback on model prototype
Targeted meetings with WCVI staff winter 2024/5 to review conditioning model and identify example management questions for application of the tool
TAG meeting spring/summer 2025

41

Proposal for further feedback

42 of 48

Further review with targeted groups
Update of operating model and conditioning model based on feedback
Development of multi-population extension
Conditioning of model to WCVI Chinook
Application of model to WCVI Chinook example management questions

42

Next steps

43 of 48

Thank you

43

44 of 48

Extra Slides

44

45 of 48

Emerging needs for WCVI Chinook

WCVI Rebuilding Plan under the Fisheries Act
RPAs for 2 (of 3) DUs on WCVI
Local, First Nations Rebuilding Plans that include decisions about hatchery enhancement considering genetic risks and trade-offs among harvest and biological objectives for abundance ‘wild’ salmon

45

How to prioritize among management levers in medium-to-long-term strategic planning?

46 of 48

Does model conditioning cover types of data typically available?
When data are not available, to what extent can we pull parameters from neighbouring potations or meta-analyses?

At what spatial scale do you see this tool as being most relevant and useful? (single population, multi-population with straying and mixed-stock fisheries)

46

Preliminary feedback/Qs

47 of 48

Time series figures

47

https://docs.salmonmse.com/articles/decision-table.html

48 of 48

salmonMSE

Stochastic model that explicitly accounts for risk and uncertainty, including

Age-structured population dynamics model (samSim)
Harvest, hatchery, and habitat management levers (AHA)
‘Integrated’ hatchery explicitly modeled (release strategy, survival, brood take) (AHA, Withler et al)
Genetic impacts on fitness from hatcheries (AHA, Withler et al)
Age-specific exploitation (as an option), as used for ECVI RPA model (ECVI RPA model)
Marine and freshwater life stages (AHA)
Threats/drivers at various life-stages (identified by RAMS)

Features:

Modular allowing flexibility for case applications
Covering spectrum of data availability data-rich to data-poor
Documented, transparent, reproducible, and open-access
Accessible to decision makers without technical/coding expertise with a visual interface

48