Species distribution models of ESA-listed corals highlight niche differences among taxa that may guide recovery planning

Abstract

Understanding the biogeography of living marine resources is essential for informed management, particularly in data-limited environments. Here, we use two corals listed under the U.S. Endangered Species Act (ESA) around Tutuila, American Samoa, to evaluate a transferable workflow that combines presence-only occurrence records, open biodiversity repositories, and open-access environmental data to extend existing survey records and support species monitoring. We developed ensemble maximum entropy (MaxEnt) models for two Indo-Pacific corals, Acropora globiceps and Isopora crateriformis , using multi-source occurrence records collected from 2007 to 2025. Dynamic environmental predictors, including thermal stress, water clarity, and wind variability, were derived from open-access satellite datasets and temporally summarized to match field sampling dates, while static predictors represented seafloor structure and land-based stressors. To evaluate one-year-ahead forecast performance, we trained models on records through 2024, generated 2025 relative habitat-suitability forecasts using lagged 2024 environmental summaries, and evaluated them against independent 2025 observations. The 2025 forecasts showed moderate discrimination under temporal validation, with area under the receiver operating characteristic curve (AUC) values of 0.74 for A. globiceps and 0.71 for I. crateriformis . The models revealed contrasting realized niches: A. globiceps was associated with shallower, rugose habitats and specific thermal-history conditions, whereas I. crateriformis showed highest suitability at intermediate depths and appeared more constrained by thermal variability. These results suggest that occurrence-based SDMs can help distinguish species-specific habitat associations and guide monitoring or spatial prioritization, while remaining best interpreted as relative habitat suitability rather than definitive forecasts of occurrence. Because the workflow relies on presence-only records and open-access environmental data, it can be adapted to other taxa, regions, and applied questions where field data are limited.