Climate vulnerability of Earth's terrestrial biomes.

Area of Science:

• Ecology
• Climate Science
• Machine Learning

Background:

• Sustainable ecosystem management and restoration require predicting plant species' survival under current and future climate conditions.
• Shifting climate envelopes due to environmental stresses can have unpredictable consequences for ecological communities.
• Accurate biome-climate envelopes (BCEs) are essential for understanding these dynamics.

Purpose of the Study:

• To develop and apply machine learning models to predict changes in biome-climate envelopes (BCEs) under various climate change scenarios.
• To identify terrestrial regions likely to experience significant BCE shifts or high uncertainty by 2080.
• To inform climate-smart restoration and ecosystem management strategies.

Main Methods:

• Trained machine learning models on global biome distribution data, incorporating soil and climate variables.
• Validated models to accurately capture contemporary biome-climate envelopes (BCEs) with 90% accuracy.
• Predicted future BCEs under Representative Concentration Pathway (RCP) 4.5 and 8.5 scenarios.

Main Results:

• 11-17% of the terrestrial surface is projected to experience a change in BCE by 2080.
• 16-19% of areas exhibit high uncertainty in their BCE trajectory over the coming decades.
• Poleward shifts in BCE boundaries are observed, with tundra BCE shrinking up to 47% and dryland BCEs expanding under RCP 8.5.

Conclusions:

• Climate change is expected to significantly alter biome-climate envelopes globally, impacting vegetation suitability.
• Identified regions where current vegetation may face stressful conditions due to future climate shifts.
• Highlighted the need for adaptive management strategies to address predicted ecological changes and uncertainties.