date_stamp <- format(Sys.Date(), "%d %B %Y")Benchmark outcomes for MSE conditioning
SPRFMO
South Pacific Regional Fisheries Management Organisation
Jack Mackerel Working Group
SCW17/Doc01
Benchmark outcomes for MSE conditioning
10 June 2026
NotePurpose and status
This note is the bridge between the SCW16 jack mackerel benchmark workshop (Lima, May 2026) and the Wageningen MSE workshop (15–19 June 2026). It records what the benchmark settled, what it left as diagnostic, and how each settled outcome is carried into operating-model (OM) conditioning. It is the document walked through on Monday morning of the workshop (09:30–10:30) and underpins the TG-05 hand-off decisions.
Draft for TG-05 review. Items marked [confirm] are inferred from the existing SCW17 materials (Paper-01a, Paper-02, Paper-04, the Wageningen schedule, and the TG-05 briefing pack) and should be checked against the SCW16 benchmark report before circulation.
1 How to read this document
The benchmark and the MSE answer different questions. The benchmark asked what is the best assessment of stock status and productivity? The MSE asks which management procedure performs acceptably across the uncertainty the benchmark identified? The transition is therefore a translation, not a continuation: each benchmark outcome becomes either a fixed input to the OM, an axis of uncertainty in the OM grid, or a diagnostic that informs but does not constrain the MSE.
Every outcome below is tagged with one of three dispositions:
- Settled → fixed: the benchmark closed it; it enters the OM as a single agreed value or structure.
- Settled as uncertain → OM axis: the benchmark concluded that the uncertainty is real and irreducible; it becomes a dimension of the reference or robustness set.
- Diagnostic: retained for interpretation or sensitivity, not as a constraint on the candidate-MP evaluation.
2 What the benchmark settled
2.1 Assessment model structure
The benchmark confirmed the Joint Jack Mackerel Model (JJM, compiled ADMB) as the assessment engine carried into the MSE, with the jjmR / FLjjm stack converting JJM output into FLR/mse operating models (see Paper-04). The assessment structure, likelihood components, and biological inputs documented in the SC13 technical annex and SCW16 working papers are not re-opened for the MSE. Disposition: settled → fixed.
2.2 Stock-hypothesis structure (h1 / h2)
The benchmark retained two stock hypotheses: a single-stock configuration (h1) and a two-stock configuration (h2, with Stock 2 representing the Peruvian zone). The benchmark did not collapse these into one preferred hypothesis — that is precisely the choice escalated to TG-05 (how h1/h2 enter advice: single, both, or weighted). Disposition: settled as uncertain → OM axis. The advice-weighting decision is a TG-05 close-out item.
2.3 Productivity grid
The benchmark established an explicit four-cell productivity grid on the h1_0.16 (and matching h2_0.16) control-file family, crossing steepness and stock-recruit fitting period (Paper-02):
| Productivity case | Steepness | SR fitting period |
|---|---|---|
ll — low steepness, long SR |
h = 0.65 | 1970–2022 |
ls — low steepness, short SR |
h = 0.65 | 2001–2015 |
hl — high steepness, long SR |
h = 0.85 | 1970–2022 |
hs — high steepness, short SR |
h = 0.85 | 2001–2015 |
The benchmark concluded that steepness drives more of the FMSY range than the SR fitting period, and that this productivity uncertainty is real rather than resolvable from the data. Disposition: settled as uncertain → OM axis (productivity is a primary axis of the reference set).
2.4 Reference-point framing
The benchmark and subsequent JMWG discussion established that MSY-based reference points (FMSY, BMSY) are sensitive to the productivity assumption and move across the grid, whereas SPR-based proxies (F35%) and dynamic-Bzero depletion are comparatively stable because they do not depend on the stock-recruit relationship (Paper-02). The MSE should therefore evaluate status against a productivity-robust convention rather than a single estimated BMSY that changes by case. [confirm] the exact reference-point convention adopted for MSE status (e.g. depletion relative to dynamic Bzero, with F35% as the F proxy) — this is a TG-05 item. Disposition: settled framing → fixed convention pending TG-05 sign-off.
2.5 Far-north selectivity
Paper-01a found that adding FarNorth selectivity flexibility changed fitted dynamics mainly through penalty terms rather than improving the direct length-frequency or CPUE fits, and that the Stock 2 SR period is best held common across the four h2 productivity cases to stabilise estimates. The benchmark therefore did not adopt extra far-north selectivity flexibility as a structural change. Disposition: diagnostic — retained as a sensitivity, not a reference-set axis. [confirm] whether the working group wants this carried as a robustness scenario rather than dropped.
2.6 Fishery and index structure
The benchmark fixed the fishery partition used for catch allocation in projections — N_Chile, SC_Chile_PS, FarNorth, Offshore_Trawl — and the index set available to candidate MPs (Chile_AcousCS, Chile_AcousN, Offshore_CPUE, Chile_CPUE, Peru_CPUE). These carry directly into the observation-error model and the split.is implementation. Disposition: settled → fixed.
3 What remains diagnostic
These benchmark products are carried into the workshop for interpretation but do not constrain the candidate-MP evaluation unless TG-05 decides otherwise:
- Far-north selectivity flexibility — sensitivity only (see §2.6). [confirm]
- Terminal-year selectivity and weight-at-age effects on BMSY — the reason the MSE prefers productivity-robust status measures; retained as explanatory diagnostics, not axes.
- Individual CPUE index reliability — informs which indices feed which MPs, but index choice is an MP-design question handled in the MP freeze, not an OM axis.
- [confirm] any benchmark retrospective / hindcast diagnostics the group agreed to carry as acceptance checks rather than scenarios.
4 How benchmark outcomes feed OM conditioning
This is the traceability table for the Monday-morning walkthrough — each benchmark outcome mapped to its MSE disposition and the document where it lands.
| Benchmark outcome | Disposition | Enters the MSE as | Lands in |
|---|---|---|---|
| JJM assessment structure | Fixed | Assessment engine; conditioning via conditionJJOM() → FLjjm → OM bundle |
Doc02 (OM conditioning) |
| Stock hypotheses h1 / h2 | OM axis | Single-stock vs. two-stock structure; advice-weighting per TG-05 | Doc01 §2.2, OM grid |
| Productivity grid (ll/ls/hl/hs) | OM axis (reference) | Four productivity cases crossing steepness × SR period | Doc02 reference set |
| MSY vs. SPR vs. dynamic-Bzero framing | Fixed convention [confirm] | Productivity-robust status metric for performance scoring | Doc03 metrics, Paper-02 |
| Far-north selectivity | Diagnostic / robustness? [confirm] | Sensitivity run, not a reference axis | Doc04 robustness |
| Fishery partition (4 fleets) | Fixed | Catch split via split.is (catch_props(om)$last5) |
Doc02/Doc03 |
| Index set (5 indices) | Fixed | Observation-error model; MP data streams | Doc02/Doc03 |
| Recruitment deviation history | Fixed inputs → scenarios | srdevs for projection; regime treatment per Tue 16 Jun |
Doc02 recruitment |
5 Open items handed to TG-05
The benchmark deliberately left the following for the Pre-Wageningen call to close. Each appears as a TG-05 agenda item and as a row in the carryover tracker:
- Stock-hypothesis treatment for advice — single h1, two-stock h2, or weighted. (§2.2)
- Reference-point convention for MSE status — confirm the productivity-robust measure. (§2.4) [confirm]
- Whether far-north selectivity enters as a robustness scenario. (§2.6) [confirm]
- Recruitment-regime treatment in projection — far-north handling, σR, steepness priors. [confirm]
- Reference-set vs. robustness-set boundary — which productivity and structural axes are reference versus robustness.
6 One-line summary for the workshop
The benchmark settled the world (assessment structure, productivity uncertainty, fishery and index structure) and identified the uncertainty that matters (productivity and stock-structure). The MSE’s job is to find management procedures that perform acceptably across that uncertainty — so the transition fixes everything the benchmark closed, turns the irreducible uncertainty into OM axes, and keeps the rest as diagnostics. TG-05 closes the handful of conditioning choices the benchmark left open so Wageningen can run and decide.
Sources within this repository: Paper-01a — Far-North model selectivity, Paper-02 — Reference points, Paper-03 — MP inventory, Paper-04 — MSE flow, Wageningen schedule, TG-05 briefing pack. Benchmark-specific claims marked [confirm] require the SCW16 report.