Multi-Factor Coral Disease Risk: A new product for early warning and management

Area of Science:

• Marine ecology
• Ecological forecasting
• Conservation science

Background:

• Ecological forecasts are vital for conservation but near-real-time systems are scarce.
• Existing systems struggle with ecological complexities and uncertainty.
• Coral reefs face threats from diseases like white syndromes and growth anomalies.

Purpose of the Study:

• Develop a novel coral disease ecological forecasting system.
• Incorporate ecological relationships and uncertainty into forecasts.
• Assess how forecast skill varies with lead time.

Main Methods:

• Developed the Multi-Factor Coral Disease Risk product using quantile regression forests.
• Integrated ecological and marine environmental conditions for risk prediction.
• Generated weekly forecasts for a 6-month window at ~5km resolution.

Main Results:

• Forecasts demonstrated superior skill in predicting disease risk on withheld data (2012-2020).
• Greatest improvements were observed for mid- to high-disease level predictions.
• Prediction uncertainty primarily stems from model uncertainty, not lead time.

Conclusions:

• The system offers near-real-time coral disease risk assessments for the Pacific and Australia.
• Model uncertainty limits accuracy gains with shorter lead times, a common ecological forecasting challenge.
• The tool supports decision-making for coral reef managers and stakeholders, with identified areas for future enhancement.