Dispersal can spread management benefits: Insights from a modeled Fijian coral reef network

The combined effects of coral and macroalgal propagule dispersal, local bistability dynamics and pressures that span the land-sea interface are not well understood, and consequently, are not well accounted for in coral reef management planning. In particular, fishing and sedimentation from nearby watersheds can tip reefs from coral-dominated stable states to macroalgal-dominated stable states. To address these knowledge gaps, we developed a mathematical model of the benthic cover dynamics of a 75-Reef network in Fiji to compare the effectiveness of three different management intervention types: extending the area of periodic fishery closures to encompass more reefs (modeled by increasing herbivore grazing rates; managing a sea-based pressure), improving water quality across Fiji (modeled by decreasing coral mortality rates; managing a land-based pressure) and the two interventions combined (managing land and sea-based pressures simultaneously). We ran the model with three grazing scenarios (low, medium, high) to account for uncertainty in actual herbivore grazing rates among reefs, as well as to represent regimes of macroalgal-dominated, bistable and coral-dominated dynamics in isolated reefs. Our results indicate that the presence of connectivity in the model stabilized the dynamics, with the final benthic cover and management effects exhibiting almost no sensitivity to initial conditions under the medium grazing scenario. The model predicts that the integrated land-sea management is the most effective management intervention for ensuring high coral cover (>30%). We also find that fishery closure management that improves the grazing rate in less than half of the reef network can lead to increases in coral cover across the entire reef network. This result suggests that, as long as a few reefs in the network have high grazing, reefs across the network may trend to high coral cover as long as environmental conditions do not change. Based on an expected value of perfect information analysis, we find that the effectiveness of the integrated land-sea management intervention is robust to the three grazing scenarios and suggests that this model can inform management decisions even with uncertainty. These findings advance our understanding of how a network of ecosystem patches with local bistability could behave and informs their management.