Georgios Kerametsidis, Centre océanographique des Baléares (COB) – Institut espagnol d’océanographie (IEO), Conseil supérieur de la recherche scientifique (CSIC), a soutenu sa thèse le vendredi 5 juin 2026 à l’Université des Îles Baléares (UIB) sur le sujet suivant : Establishing an interdisciplinary basis for ecological connectivity to inform spatial stock assessment modeling: an application to a red mullet metapopulation in the Mediterranean Sea.
Il a obtenu la mention « cum laude ».
Membres du Jury
Président : Dra. Alexandra Silva, Fisheries Resources Dept., Portuguese Sea and Atmosphere Institute (IPMA), Portugal
Suffragant : Dr Evangelos Tzanatos, Université de Patras, Grèce
Secrétaire : Dr Ignacio Alberto Catalán Alemany, Investigación Marina en Sistemas Ecológicos y Sociales (IMSES), Instituto Mediterráneo de Estudios Avanzados (IMEDEA)
Co-encadrants :
- Manuel Hidalgo, Centro Oceanográfico de Baleares – Instituto Español de Oceanografía (COB-IEO), Consejo Superior de Investigaciones Científicas (CSIC)
- Vincent Rossi, Institut Méditerranéen d’Océanologie (MIO) – Centre National de la Recherche Scientifique (CNRS), Aix Marseille Université.
Abstract
Effective sustainable management of marine resources and biologically realistic assessments rely on accurately identifying stock structure, including not only its external boundaries but also the spatial complexity within. However, most commercially harvested marine species are assessed under the long-standing assumption of a “stock unit” within prescribed boundaries known as assessment units.
This approach persists despite growing evidence of passive and active dispersal documented in a plethora of species across multiple spatiotemporal scales, within and across stock boundaries. The main challenges in developing spatially explicit stock assessment models are:
- describing the most biologically plausible spatial stock structure, and
- obtaining robust, meaningful connectivity estimates, then determining how to effectively integrate these data to model spatial dynamics.
While tagging is a valuable method for quantifying connectivity in adult stages, it has clear limitations for understanding early-life stages dispersal, despite the demonstrated controls it exerts on recruitment and spatial structure. In this PhD thesis, we applied an interdisciplinary approach by integrating a spatiotemporal model, oceanographic simulations, and chemical and morphological markers to investigate the demographic connectivity and population spatial structure of one of the most important demersal species for Mediterranean fisheries, the red mullet Mullus barbatus, across multiple scales and its entire life in the Western Mediterranean Sea.
First, we hypothesized that the spatiotemporal variability of density in two adjacent stock units (i.e. GSA06 and GSA07) is associated with spatially structured environmental processes across multiple spatiotemporal scales. To test this, we applied a generalized empirical orthogonal function and dynamic factor analysis to fishery-independent and -dependent data of red mullet, a highly commercial species, in the Western Mediterranean Sea. Areas with persistent and dynamic high aggregations were detected for both stock units. A large-scale climatic index and local open-ocean convection were associated with both stocks, while other variables exhibited stock-specific effects. We also revealed spatially structured density dynamics within the examined management units.
Later on, we combined simulated larval dispersal modelling with tissue stable isotope analysis to explore whether early life dispersal connectivity persists into adulthood (i.e. is reflected in the stable isotopes signatures of the adult individuals). We observed misalignment between the present “closed” assessment units and the identified biologically informed units. The identified metapopulation system comprises northern and southern subpopulations with dissimilar demographic roles. This principally stems from the Northern Current mediated larval transport from high-density persistent spawning grounds and, secondarily, likely from the regional differences in the isotopic baseline associated to different productivity regimes.
Finally, we integrated otolith trace element analysis—specifically focusing on elements whose concentrations are directly influenced by ambient water chemistry—with wavelet-based contour analysis across three spatial scales. A high (>70%) classification success was achieved for specimens grouped following the spatial scale of the current assessment framework, and a moderate classification (up to 60%) was achieved when specimens were grouped following the more biologically realistic spatial ecoregions. Unsupervised classification based on trace elements revealed two distinct natal sources, while wavelet-based analysis identified seven morphotypes within the metapopulation system. Our findings are primarily explained by the high dispersal during early life stages, the overall spatial uniformity of environmentally influenced chemical markers, and the presence of distinct, environmental- and diet-driven spatial units.
After obtaining the necessary connectivity information, we parametrized a spatially explicit simulation-estimation framework for the red mullet metapopulation system in the Northwestern Mediterranean Sea, which was conditioned on connectivity information (principally) from simulated larval dispersal, a primary driver of spatial dynamics for the species. The metapopulation system consisted of four interconnected subpopulations spanning two assessment units—GSA07 (Gulf of Lion) and GSA06 (Catalan Coast, Gulf of Valencia, Gulf of Alicante)—which are currently considered unconnected under the existing framework.
Various assessment (estimation) models—each representing different connectivity and spatial stock structure assumptions—were fitted to simulated “observed” pseudo-data from the operating model under a range of underlying true spatial dynamics. The estimated parameters were subsequently compared to the true values used in the operating model, allowing us to evaluate the accuracy and potential biases of different assessment models. Across different model scenarios, recruitment estimates exhibited substantial bias, when larval movement was ignored, or stock structure was simplified compared to the OM. In the Gulf of Lion, recruitment was consistently underestimated by more than 50%, while in contrast, the Catalan Coast showed an equally high overestimation of recruitment.
Our research highlights the importance of spatial stock assessment frameworks and early life history spatial dynamics, which remain largely underrepresented in operational fisheries management.