Assessing the spatial structures and dynamics of marine populations is still a major challenge due to the interlocked scales of oceanic variability and the highly dispersive early-life stages of most marine species. Indeed, larval dispersal and connectivity control several key evolutionary and ecological processes that are fundamental pre-requisites for effective ecosystems protection and management. Our modelling approach, the Lagrangian Flow Networks (LFN), provides a systematic characterization of multi-scale dispersal and connectivity of marine organisms. It consists in subdividing the basin into sub-regions or habitat patches which are interconnected through the transport of larvae by ocean currents. Post-processing of connectivity matrices permits the identification of hydrodynamical provinces, the computation of various connectivity proxies measuring retention (Self-Recruitment, Local Retention) and exchange (Source-Sink) of larvae. Following an “ecosystem-approach” to connectivity, those diagnostics were computed for a large range of parameters to better understand how to design a network of connected marine reserves and how dispersal knowledge helps planning population genetics studies with more adequate sampling strategies and improved gene flow hypotheses. Focusing on the European Hake, a commercially and ecologically important demersal species, we then explore the impact of connectivity processes on population structure at the scale of the Mediterranean basin. We combined estimated spawning areas, ensemble of realistic connectivity matrices and predicted settlement grounds of juveniles to compare putative hake subpopulations against the delimitation of fishery assessment units. Our objective separation of the seascape illustrate how bio-physical constraints control early-life connectivity, which in turn shapes the metapopulation structure. Zooming on the north-west Mediterranean, we showed that the inter-annual variability of recruitment across transnational contiguous management units is well reproduced by an hydroclimatic index and our synthetic connectivity estimates. Self-Recruitment is the most powerful metric as it integrates both local and remote influences and is able to capture repeated circulation patterns that affect recruitment success in each stock. Finally, we combined LFN simulations with otoliths analyses to locate and quantify larval origins for coastal fishes D. sargus and D. vulgaris in the Adriatic Sea. Robust correlations between otolith geochemistry and Lagrangian model simulations allow delineating spawning areas and assessing their relative importance for the larval replenishment of the Apulian coast. Our results suggest that the Tremiti marine reserve contributes largely to the larval replenishment of the Apulian coast. Under constant development, our modelling framework helps characterizing population connectivity and provides relevant information to stakeholders for the conservation of marine ecosystems.
