Random Forest habitat suitability modeling for Acanthurus spp. in Philippine coral reefs: environmental drivers and climate projections

Surgeonfish of the genus Acanthurus (family Acanthuridae) are ecologically pivotal herbivores on Indo-Pacific coral reefs, yet their habitat requirements and distributional responses to environmental gradients and climate change remain poorly characterized in the Philippines. A Random Forest (RF) machine learning framework was used to model habitat suitability for six commercially and ecologically important species –A. lineatus, A. nigricans, A. olivaceus, A. triostegus, A. xanthopterus, and A. pyroferus– across Philippine coral reef systems. Geo-referenced occurrence records (n = 352-714 per species) from OBIS, GBIF, and ReefBase were paired with ten environmental predictors encompassing reef structure, oceanographic conditions, bathymetry, and physical disturbance. RF models achieved mean AUC of 0.929 and TSS of 0.805, indicating excellent predictive performance. Coral cover, depth, and rugosity were the dominant habitat predictors across species. Current suitable habitat ranged from 6,890 km2 (A. olivaceus) to 11,350 km2 (A. xanthopterus). Under SSP5-8.5 by 2100, projected habitat declined by 25.4-36.1%, driven primarily by thermally induced coral cover loss. Acanthurus triostegus was most vulnerable (-36.1%) and A. xanthopterus most resilient (-25.4%). Multi-species overlap analysis identified Verde Island Passage, Tubbataha Reef, and Apo Island as priority conservation nodes. Gap analysis revealed three unprotected high-overlap zones as candidates for new marine protected area (MPA) establishment. These findings provide a spatial evidence base for climate-adaptive reef fisheries management across the Indo-Pacific.