Reef-scale variation in larval supply and settlement: validating Lagrangian dispersal predictions with observations of coral larvae

Population persistence and recovery in marine systems is driven by larval dispersal in the water column, generating ecological connectivity between the natal and settlement locations. Connectivity modelling is commonly utilised for the spatial planning of marine protected areas and, more recently, for the prioritisation of restoration interventions. Here, we conducted field experiments to validate the spatial patterns and rates of larval arrival as simulated by a high-resolution connectivity model (∼300 m resolution), using complementary spatial and temporal sampling of coral larvae and newly settled recruits around a cluster of offshore coral reefs. At the within-reef scale, Lagrangian dispersal modelling demonstrated only a fair performance at predicting observed spatial patterns of larval arrival and settlement, at best. However, at the reef cluster level, hydrodynamically-driven interannual variations in larval supply were well correlated with observed interannual variations. Combined, the model results resolve empirical observations for the temporal (inter-annual) and spatial scales relevant to meta-population dynamics (1–10s of km's). At the finer spatial scales of resolution (>1/10 ha to <10 ha), relevant to current restoration interventions, skill at predicting larval density is poor whilst skill at predicting larval settlement is fair. Overall, our findings identify the need for a model validation framework that considers the scales of physical processes resolved by the hydrodynamic modelling, spatial-temporal variability in the propagule populations being measured, the error tolerance for how the outputs of model simulations are being utilised (theoretical versus operational), and the complementary use of modelling and field sampling for different scales of application.