Laffitte et al. (2026) Interpretable Machine Learning Reveals Global Nonlinear Responses of Soil Respiration to Climate

⚠️ Warning: This summary was generated from the abstract only, as the full text was not available.

Identification

• Journal: Geophysical Research Letters
• Year: 2026
• Date: 2026-10-16
• Authors: Benjamin Laffitte, Zhihan Yang, Tao Zhou, Xiangjun Pei, Xiaolu Tang
• DOI: 10.1029/2026gl124051

Research Groups

Not explicitly mentioned in the abstract.

Short Summary

This study used a SHAP-interpreted Random Forest model on 8,111 observations to identify global response patterns of soil respiration (Rs) to key environmental drivers. It revealed a peaked temperature association with a maximum at 27.7°C, plateauing precipitation contributions, and linearly increasing leaf area index contributions, highlighting nonlinearities often overlooked by monotonic assumptions in Earth system models.

Objective

• To identify global response patterns for key environmental drivers of soil respiration (Rs).

Study Configuration

• Spatial Scale: Global
• Temporal Scale: Not explicitly mentioned, but observations likely span various time points to capture responses to seasonal climate extremes.

Methodology and Data

• Models used: SHAP-interpreted Random Forest model
• Data sources: 8,111 observations (specific types not detailed in abstract)

Main Results

• Temperature contributions to Rs showed a peaked association with a descriptive maximum of 27.7°C.
• Precipitation contributions to Rs plateaued at high values.
• Leaf area index contributions to Rs increased approximately linearly.
• These response patterns (shapes and supported ranges) vary among different climate zones.
• Monotonic assumptions in Earth system models may overlook these identified nonlinearities, potentially biasing estimates of terrestrial carbon flux responses to seasonal climate extremes.

Contributions

• Provides a global, data-driven characterization of peaked temperature and saturating precipitation associations for soil respiration.
• Demonstrates how the shapes and supported ranges of these associations vary among climate zones.
• Highlights a critical limitation in current Earth system models, suggesting that monotonic assumptions may lead to biased estimates of carbon flux responses.
• Proposes interpretable machine learning as a transparent framework for evaluating and improving the representation of soil carbon dynamics under climate change.

Funding

Not explicitly mentioned in the abstract.

Citation

@article{Laffitte2026Interpretable,
  author = {Laffitte, Benjamin and Yang, Zhihan and Zhou, Tao and Pei, Xiangjun and Tang, Xiaolu},
  title = {Interpretable Machine Learning Reveals Global Nonlinear Responses of Soil Respiration to Climate},
  journal = {Geophysical Research Letters},
  year = {2026},
  doi = {10.1029/2026gl124051},
  url = {https://doi.org/10.1029/2026gl124051}
}