Drivers of larval connectivity variability among coral reefs in Southeast Sulawesi

Coral reef patches are connected via dispersal of larvae, i.e., larval connectivity, that varies across space and time. Larval connectivity supports gene flow, sustains fisheries, and stabilizes larval supply. Connectivity also enhances the effectiveness of marine reserves by facilitating valuable conservation processes. How larval connectivity responds to environmental factors, excluding oceanographic factors, is largely unknown, with little
information available on the influence of external factors on connectivity patterns. I address this knowledge gap by identifying how environmental, climate, and habitat factors drive variability in larval connectivity. I correlate graph-theoretic proxies of larval connectivity with sea surface temperature (SST) and climate variables using Generalized Additive Models (GAM) and recursive partitioning with regression trees to assess each factor’s effect on connectivity between 487 reefs in Southeast Sulawesi, Indonesia over a 20-year period. I further simulate how coral reef habitat degradation over that period may change patterns of larval connectivity. There is a significant effect of El Nino, Pacific Decadal Oscillation (PDO), and Sea Surface Temperature (SST) on larval connectivity. SST above 28°C decreased out-degree and in-degree by an average of 0.65 standard deviations and increased self-recruitment by an average of 0.74 standard deviations. This result means that as SST increases above 28°C, there is a decrease in both incoming and outgoing connections between reefs, and more larvae remaining within their source reef. Generalized Additive Model (GAM) analysis of the effect of SST on connectivity metrics shows higher explanation of variance at higher SST. This result supports the existence of an SST threshold at which connectivity for fish species in the region will predictably decline. Spatial analysis using spectral clustering shows a larger effect of reef location (spatial cluster) on connectivity metrics compared to SST. Generally, two out of six clusters have high self-recruitment while the remaining four clusters have high out-degree and in-degree. Habitat degradation decreases cumulative flow of larvae by 73 percent when comparing flow matrices before and after habitat degradation. Additionally, habitat degradation reduces variance of cumulative flow for coral trout and rabbitfish species. Further, these trends are predicted to continue under future habitat degradation values. These results allow us to predict how connectivity will change as SST and habitat degradation increase due to climate change.