Duration: up to one (1) year

Modality: remote consultancy, with possible travels

Estimated level of effort: Up to 45 working days (indicative)

Maximum budget: EUR 39,999 (VAT excl.), including mission fees (travel + per diem)

1. Objectives of the assignment

The assignment falls under the project “Sustainable Western Indian Ocean Programme – Subcomponent 1.2” (“SWIOP 1.2”), implemented by Expertise France in support to two Regional Fisheries Management Organisations (RFMO), including the IOTC, and contributes to the project’s objective of strengthening the scientific basis for RFMO decision-making. The consultant will support selected activities related to stock assessment and scientific analysis for yellowfin tuna, with a focus on length composition standardisation, the improvement of assessment inputs, and the testing of alternative assessment modelling approaches to support IOTC scientific work ahead of Working Party on Tropical tunas. The two tasks are expected to be implemented in a coherent manner, so that the work on length composition standardisation can inform, where relevant, the comparison between the main Stock Synthesis assessment and the alternative exploratory model.

The consultancy shall also implement an alternative assessment model for yellowfin tuna using an appropriate platform or package. The alternative model is expected to be based on the same types of observational data and key biological parameters as the main yellowfin assessment to facilitate comparison.

2. Scope of the assignment

2.1. Task 1 - Stock assessment length composition standardisation modelling (SWIOP 1.2, Activity 2.2 - Stock assessment and ecosystem-based scientific analysis, Task 2.2.1-Stocks assessment on two IOTC species)

Length composition data are an integral component of the IOTC tropical tuna assessment and provide critical information on selectivity, recruitment, and mortality. Bias in the length composition data can substantially affect the estimation of key management quantities. To date, the length composition data reported by IOTC members (length frequency samples by fleet, time and area strata) are compiled for inclusion into the assessments by aggregating together the raw samples from all sampling/reporting units for a fishery. It is recognized that simple aggregation of nominal length data may introduce bias if the spatial, seasonal, and fleet variabilities in the length samples are not appropriately accounted for. The design approach based on stratified sampling to estimate length composition is widely used in national fisheries. However, for IOTC, the sampling design, coverage, data collection and reporting vary greatly among fleets, which makes analyses of the length compositions using a design approach impractical.  

Model-based methods provide alternative means for standardizing length composition to estimate stock-level length distribution while accounting for appropriate weighting for the underlying data. The application of Model-based standardization of length compositions has become more common recently. Dirichlet-multinomial modelling has been shown to work reasonably well in simulation studies in standardizing ecological count data and was able to reduce sampling bias when adequately structured to attribute overdispersion to known sources of spatial variability. Vector Autoregressive Spatio-Temporal (VAST) models are utilized to standardize fishery length composition data by integrating spatial, temporal, and environmental variables into a comprehensive framework. More recently, a new approach for analysing length composition data based on Poisson-factorised multinomial regression and was successfully applied in a number of stock assessments in WCPFC. These standardisation models account for variations across time and space, allowing for a nuanced understanding of fish population dynamics and distribution.

Furthermore, the length frequency data needs to be better handled within the assessment through the standardization procedure designed to provide specific information to different processes: i.e., one that mirrors the abundance index with constant selectivity, and one that can accommodate variable selectivity for fishery removal. These analyses serve to inform how these observations can be configured within the assessment to reconcile with other observations and better inform the various population and fishing processes. 

To support this effort, the consultancy shall conduct standardization analysis of length composition data for the IOTC yellowfin and bigeye tuna assessments, following a similar framework of the application of the Albacore tuna (Neubauer, Hill-Moana 2024)) and silky shark (hard (Neubauer et al. 2024)) in the WCPFC region. 
The consultant shall specifically follow the below-detailed methodology:

•    Perform the standardization modelling of length composition data for the IOTC yellowfin and bigeye tuna assessment, using the state-of-the-art modelling approaches (e.g., Poisson-factorized multinomial regression). The modelling should adequately consider key sources of variability, including fleet, spatial and temporal variations, and environmental and oceanographic conditions where feasible. Effort should be prioritised towards the yellowfin tuna assessment, where the analytical methods should be developed first. The bigeye tuna length composition data, which has a very similar structure, can then be standardised using the same modelling approach. The analysis should focus on key fisheries such as longline and purse seine, which have the largest impact on the assessments. Other fisheries can be analysed where sample sizes are adequate and appropriate. The analysis should also identify, source, and process relevant oceanographic variables to be considered in the standardisation model where feasible. 
•    Where feasible, the composition for catch fisheries and index fisheries (i.e., fisheries which the CPUE indices are associated with) shall be standardized separately. The analysis will produce population length compositions for the index fishery, where selectivity can be assumed constant, and catch length compositions for the capture fisheries, where selectivity may vary over time. This separation will facilitate their incorporation into the assessment model, providing distinct insights into the selectivity of the catch and abundance index.
•    The analysis will concentrate on the main fisheries and fleets relevant to stock assessment, including the deep-freezing longline fishery, purse seine fishery (distinguishing between free schools and associated schools), pole and line fisheries, and gillnet fisheries.
•    Characterize standardized length frequency distributions for the main fisheries and recommend an appropriate fleet structure for incorporating these data into the stock assessment model. 
•    Support the IOTC Stock Assessment Expert in evaluating the impact of length frequency standardization on the yellowfin assessment.

The primary focus of the analysis will be yellowfin tuna. However, the analytical scripts will be designed to be reusable, enabling standardization of length composition data for other key tuna species, including skipjack, bigeye, and albacore.

2.2. Task 2 - Alternative Stock assessment modelling of yellowfin tuna (SWIOP 1.2, Activity 2.2 - Stock assessment and ecosystem-based scientific analysis, Task 2.2.1-Stocks assessment on two IOTC species)

The expert peer review of the yellowfin assessment has recommended the use of alternative model structures to explore issues identified in the assessment, particularly in view of recent advances in new assessment platforms. For example, the Fisheries Integrated Modeling System (FIMS), co-developed by NOAA Fisheries and academic partners, uses Template Model Builder (TMB) as its primary statistical inference tool and represents the next-generation stock assessment platform intended to replace Stock Synthesis. The WHAM (Woods Hole Assessment Model) and SAM (State-space Assessment Model) are premier state-space stock assessment frameworks capable of estimating time-specific deviations and explicitly handling random effects, which is highly pertinent to tuna species where many parameters are more appropriately treated as time-varying random effects rather than assumed to be static. Additionally, a new package (OPAL) developed for WCPFC assessments, initially designed as a simple framework for data-moderate species (such as sharks and billfish), can be extended towards more substantial models. The OPAL model is also in the process of implementing a length partition to allow for fully length-and-age-based models. This is considered potentially very useful for tuna assessment, as the majority of data are length-based and many biological processes are likely related to the length of the fish.

The consultant should specifically follow the below-detailed methodology:

•    Propose a suitable platform and package for the alternative assessment model of yellowfin tuna and evaluate its suitability for modelling the yellowfin tuna stock.
•    Implement the stock assessment of yellowfin tuna using the proposed platform and develop a fully functional exploratory diagnostic model that supports a comparison with the main Stock Synthesis assessment. The exploratory assessment model is expected to use the same data (catch, length composition, and key biological parameters) as the main assessment. The provision and compilation of data will be assisted by the IOTC Secretariat.

The assessment model and associated analytical scripts should be modular and reusable, enabling potential future application to other IOTC species, including tropical tunas, billfish, and sharks.

3.  Explicit exclusions from the assignment

To avoid any ambiguity of roles, the consultant shall not:

•    Represent Expertise France or IOTC vis-à-vis the European Union, partner institutions, States or stakeholders;
•    Coordinate, supervise or manage project activities or teams;
•    Assume decision-making, validation or arbitration authority;
•    Manage budgets, contracts, procurement or administrative processes;
•    Substitute for the Project Coordinator or any member of the Project Management Unit.

All recommendations provided by the consultant will be non-binding and subject to validation by Expertise France and IOTC. 

4. Expected deliverables 

The consultancy will be structured around a limited number of clearly identified deliverable categories, with an indicative level of effort (LoE) and an indicative delivery timeline, to be confirmed and adjusted at the beginning of the service. 

The table below presents the expected main deliverables, their estimated workload and indicative timing, for an overall ceiling of up to 45 working days over a maximum period of one (1) year.

All deliverables shall be submitted in English and in editable format. Technical deliverables will be reviewed by the IOTC Secretariat, with Expertise France ensuring contractual follow-up and formal validation in accordance with the contract.