Larval dispersal of pearl oysters Pinctada margaritifera in the Gambier Islands (French Polynesia) and exploring options for adult restocking using in situ data and numerical modelling.

Black pearl farming is the second source of French Polynesia income after tourism, and Gambier Islands are the main farming sites. Gambier main lagoon contains several sub-lagoons critical for pearl oyster rearing and spat collecting (SC). The Rikitea lagoon, traditionally had good SC rates in the warm season which ensured steady supplies of oysters for black pearl production. However, since 2018, SC has abruptly decreased. To assess the factors affecting SC, Gambier lagoon hydrodynamics was investigated in 2019-2020 to calibrate a hydrodynamic model and simulate larval dispersal around the SC areas. The model shows the strong wind influence on larval dispersal and accumulation patterns and suggests that windy months in the warm season as it can occur during La Niña episodes can explain recent poor SC. Larval dispersal scenarios also informed on best locations to perform adult oyster restocking, a practice that can also enhance SC on the long term.

A multivariate approach to synthetize large amount of connectivity matrices for management decisions: Application to oyster population restocking in the pearl farming context of Tuamotu Archipelago semi-closed atolls.

In applied ecology, numerical biophysical modelling allows running numerous simulations of spatial connectivity between source and destination locations. To characterize population connectivity, larval dispersal and resulting connectivity matrices can be computed for various forcing conditions of wind, density of spawners, or pelagic larval durations. Here, we investigate a methodology to synthetize meaningfully all numerical experiments performed for three atoll lagoons in the Tuamotu Archipelago pearl farming context. The objective is to identify the best restocking locations that consistently maximize the spread of pearl oyster larval dispersal, considering all forcing conditions. A multivariate generic approach is used to process and synthesize time-series of connectivity matrices and identify afterward with contextual criteria the spawning locations that match a variety of specific connectivity, logistical and ecological criteria. Similar synthesis of large volume of connectivity matrices will likely gain momentum considering the increasing use of numerical models for applied science and population management.