Statistical Learning and Uncommon Soil Microbiota Explain Biogeochemical Responses after Wildfire

Area of Science:

• Environmental Science
• Soil Science
• Microbiology

Background:

• Wildfires globally disrupt soil biogeochemical processes, impacting plant growth and water quality.
• Understanding soil responses to fire at various scales remains challenging due to complexity and variability.

Purpose of the Study:

• To predict and explain soil biogeochemical responses to wildfire using statistical learning (SL).
• To investigate the role of soil microbiome, particularly the rare biosphere, in post-fire recovery.

Main Methods:

• Examined two wildfires in Colorado over two post-fire years.
• Applied statistical learning models using soil biogeochemical and DNA sequencing data.
• Developed hybrid models combining microbiome and biogeochemical data.

Main Results:

• SL accurately predicted most soil biogeochemical responses.
• Hybrid microbiome + biogeochemical models best explained 9 out of 13 analytes.
• Uncommon soil microbiota (rare biosphere) were key predictors in hybrid models.

Conclusions:

• Statistical learning, especially when incorporating the rare biosphere, effectively predicts post-wildfire soil biogeochemical changes.
• This approach offers a powerful tool for understanding and managing ecosystems affected by disturbances.