360 likes | 504 Views
Gulf of Maine cod SCAA/ASPM vs ADAPT-VPA Doug Butterworth and Rebecca Rademeyer. __________________________________________________________________ Marine Resource Assessment and Management Group, University of Cape Town. METHODS.
E N D
Gulf of Maine codSCAA/ASPM vs ADAPT-VPADoug Butterworth and Rebecca Rademeyer __________________________________________________________________ Marine Resource Assessment and Management Group, University of Cape Town
METHODS ADAPT-VPA- Backward calculation- Oldest-age F assumption- Catches-at-age exact SCAA/ASPM- Based on separability: Fy,a = SaFy- Forward estimation, typically likelihood based- Errors in proportions-at-age SCAA - Statistical treatment of age data errors May include S/R relationshipASPM - Age-structured dynamics Must include S/R relationship Age data not necessary
Ricker-like not Beverton-Holt S/R relationship ASPM methodology updates • M fixed (0.2) (not estimated) - All Sa’s estimated (no functional form imposed) - Model to 11+, though fit to 7+ data • MSY calculated internally - Fully Bayesian PI computations
- Alt-VPA (a=est): estimates a where VPA methodology • RC-VPA: as Mayo and Col (2006) – asymp. flat selectivity - Alt-VPA (a=1): removes inconsistency in plus-group eqns Fit either to ages 1-6 or 1-7+MSY calculated externally
BUTScaled to get Data used Mayo and Col (2006)
ResultsRC-ASPM(M=0.2, Ricker) Shading shows 95% PIs
Selectivity estimatesRC-ASPM(M=0.2, Ricker) Commercial, pre-1992 Commercial, post-1991
Selectivity estimatesRC-ASPM(M=0.2, Ricker) Commercial, pre-1992 Commercial, post-1991 NEFSC-Spring NEFSC-Autumn
ASPMStock-recruit relationships Ricker -lnL = -46.3
ASPMStock-recruit relationships Ricker Beverton-Holt -lnL = -46.3 -lnL = -39.9 If estimate g parameter (g =1 for Ricker): 1.05 [0.54; 1.15]
-lnL = -11.4 Asymp. flat selectivity ASPMSelectivity dome -lnL = -46.3 S7/S6 =0.52 [0.41; 0.64]
Per-recruit contribution to cohort biomass ASPMPast catch selectivity 1992-2004: Estimated 1981-1991: Estimated Pre-1982: Set equal to 1982-1991 Second order effect (Important assumption is unchanging survey selectivity)
ADAPT-VPA comparisons VPA as Mayo and Col (2006)
ADAPT-VPA comparisons VPA as Mayo and Col (2006) RC-VPA
ADAPT-VPA comparisons VPA as Mayo and Col (2006) RC-VPA Alt-VPA, a=1
ADAPT-VPA comparisons VPA as Mayo and Col (2006) RC-VPA Alt-VPA, a=1 Alt-VPA, a=est
ADAPT-VPA comparisons VPA as Mayo and Col (2006) RC-VPA Alt-VPA, a=1 Alt-VPA, a=est ASPM – Data from 1982
ADAPT-VPA comparisons VPA as Mayo and Col (2006) RC-VPA Alt-VPA, a=1 Alt-VPA, a=est ASPM – Data from 1982 RC – ASPM
ADAPT-VPA comparisons VPA as Mayo and Col (2006) RC-VPA Alt-VPA, a=1 Alt-VPA, a=est ASPM – Data from 1982 RC – ASPM
ADAPT-VPASelectivities RC-VPA
ADAPT-VPASelectivities RC-VPA Alt-VPA, a=1
ADAPT-VPASelectivities RC-VPA Alt-VPA, a=1 Alt-VPA, a=est
ADAPT-VPASelectivities RC-VPA Alt-VPA, a=1 Alt-VPA, a=est ASPM – Data from 1982
ADAPT-VPASelectivities RC-VPA Alt-VPA, a=1 Alt-VPA, a=est ASPM – Data from 1982 RC – ASPM
ConclusionsGeneral • Careful treatment of plus-group • Fully flexible Sa parametrization • Take care with use of Beverton-Holt S/R relationship
ConclusionsGulf of Maine cod • close to and • 2003 difference as ADAPT-VPA imposed asymp. flat selectivity • Asymp. flat selectivity statistically inconsistent with data Independent evidence of mechanisms to give dome shape ? Robust to sensitivities • SCAA to be preferred Enables use of pre-1982 data Greater range of Bsp Better precision of estimates M-S Act focus on MSY