Update on the development of MSE analysis of candidate Management Procedures for Indian Ocean albacore

This document presents the latest iteration in the development of the operating model (OM) for
Indian Ocean albacore tuna. The operating model is developed around the Stock Synthesis (SS3)
stock assessment, conducted by WPTmT in 2016, and considers a number of sources of uncertainty,
as identied by WPTmT and WPM, in the estimation of population trajectories and dynamics.
The complete grid of structural uncertainty covers seven model elements, as follows:
• Natural mortality (M), 5 values.
• Variance in stock-recruitment residuals (sigmaR), 2 values.
• Steepness in the Beverton & Holt stock-recruitment relationship, 3 values.
• Coecient of variation for the t to CPUE data (cpuecv), 4 values.
• Eective sampling size of the length composition data (ess), 3 values.
• Yearly increase in the catchability coecient of the CPUE series (llq), 2 values.
• Functional form of the CPUE selectivity curve (llsel), 2 values.
This creates an initial set with a total of 1,440 SS3 model runs. The results of those runs are then
checked for convergence and a total of 1,095 runs are kept for further processing.
The OM has then been extended to the start of 2018 from the last year of data employed in the
model conguration, 2014. The population model has been projected by aplying the nominal
catches of the 2015-2017 period, as reported to IOTC in September 2018. A large number of model
runs estrimated population abundnaces in 2014 that could not explain the catches observed in the
2015-2017 period, or did so only through extremely large increases in eort (tenfold or larger).
Those runs were thus considered unfeasible, and excluded from the reference case OM. This leads
to a model containing 414 equally weighted population and shery trajectories.
This reference set has been used to nd the Management Procedure (MP) parameter sets able to
lead to the performance values requested by TCMP (IOTC 2018). For albacore tuna the four tuning
objectives are:
• TA1: Average Spawning biomass (SB) over the period 2019-2038 exceeds SB MSY in exactly
50% of the simulations.
• TA2: The stock status is in the Kobe green quadrant over the period 2019-2038 exactly 50%
of the time (averaged over all simulations).
• TA3: The stock status is in the Kobe green quadrant over the period 2019-2038 exactly 60%
of the time (averaged over all simulations).
• TA4: The stock status is in the Kobe green quadrant over the period 2019-2038 exactly 70%
of the time (averaged over all simulations).
Two candidate MPs are being tested so far: a model-free MP that calculates the Total Allowable
Catch (TAC) based on on trends and levels of a single CPUE series, and a model-based one that
estimates stock status based on a Pella-Tomlinson biomass dynamics model t and suggests a TAC
according to target and limit levels for biomass depletion.

The tuned MPs are able to obtain the requested performance values, but in some cases at the
cost of bringing the stock to dangerously low levels. The combined eect of a population that
is on average at a higher level of abundance that management requests, and the computation of
performance indicators for tuning for the whole period of projection, leads to this behaviour. The
eect of dierent performance computation periods and of alternative probability levels, has been
computed.