Abstract
Effective marine protected area (MPA) networks require accurate knowledge of habitat distribution, yet in remote oceanic archipelagos, the logistical and financial costs of comprehensive surveys render this difficult to obtain. Species distribution models (SDMs) offer a practical alternative, however their application to MPA placement evaluation in data-deficient oceanic archipelagos remains limited. The Azores archipelago has an MPA network currently under consideration for expansion, yet site suitability remains unknown and knowledge of coastal reef-fish habitat distribution is incomplete, with extensive surveys being impractical. To address this, we used underwater visual census data collected between 2010 and 2023 across the archipelago, along with seabed and oceanographic variables, to construct SDMs predicting the distributions of six coastal reef-fish species using a random forest approach. These predictions identified areas of high biomass or probability of presence as proxies for high-suitability habitat, and the proportion covered by the present MPA network was assessed. Results suggest Azorean MPAs are generally placed in areas of predicted high-suitability habitat, although due to their small size, no-take reserves protect a minute proportion of this. MPAs with more lenient regulations cover a larger area of predicted high-suitability habitat, although as these permit certain extractive activities, the protection provided is questionable. Given the large proportion of shelf areas assigned legal protection that we suggest to be appropriately placed, we argue MPA policy should prioritise tighter restrictions and stronger enforcement over spatial expansion. We conclude that SDMs are a valuable tool for evaluating reef-fish habitat protection in remote, data-deficient oceanic archipelagos.