North American vegetation model for land-use planning in a changing climate: a solution to large classification...

Area of Science:

• Ecology
• Climate Change Biology
• Biogeography

Background:

• Understanding biome distribution is crucial for predicting ecological responses to climate change.
• Accurate modeling of biome shifts requires robust statistical methods and consideration of future climate scenarios.

Purpose of the Study:

• To predict the current and future geographic distribution of 46 North American biomes based on climate variables.
• To assess the model's ability to identify novel climate conditions outside the contemporary biome range.
• To project biome shifts under different greenhouse gas emission scenarios and time periods.

Main Methods:

• Utilized intensive sampling data from 46 North American biomes.
• Employed Random Forests classification tree for predictive modeling.
• Incorporated techniques to handle numerous classes and predict future, non-analog climates.

Main Results:

• Prediction errors averaged 3.7%, with individual biome errors ranging from 0% to 21.5%.
• The model identified climates without analogs for 78% of tested extra-North American locations and 81% of Caribbean land area.
• Future projections indicate expansion of tropical dry deciduous forests and desertscrub, stability in eastern USA forests, and northward shifts for temperate forests and grasslands, impacting tundra and taiga.

Conclusions:

• Climate change is projected to significantly alter North American biome distributions.
• Identifying areas with uncertain projections or novel climates is critical for effective land management.
• Focusing resources on areas with more certain projections can enhance land management success.