Reconstructing Net Primary Productivity in Northern Greater Khingan Range Using Tree Rings.

Area of Science:

• Ecology
• Climate Science
• Forestry

Background:

• Boreal forests are critical global carbon sinks, but their net primary productivity (NPP) is poorly understood due to limited long-term data.
• Understanding NPP dynamics is essential for accurate terrestrial carbon cycle modeling and climate change impact assessments.

Purpose of the Study:

• To reconstruct high-resolution, interannual forest NPP in the Northern Greater Khingan Range from 1968 to 2020.
• To identify the primary climatic drivers influencing NPP and tree radial growth in this region.
• To establish a reliable framework for NPP reconstruction using tree ring data and machine learning.

Main Methods:

• Developed a standardized tree ring chronology for *Larix gmelinii*.
• Applied Pearson correlation coefficients and SHAP values to determine factors driving NPP.
• Integrated XGBoost and Extreme Random Forest (ERF) models for NPP reconstruction.

Main Results:

• Growing season drought was identified as the dominant factor affecting both NPP and tree radial growth.
• The XGBoost model demonstrated higher accuracy in NPP reconstruction compared to the ERF model.
• Reconstructed NPP series correlated strongly with MODIS NPP products (r > 0.6) and exhibited interdecadal cycles (10, 28, 49 years) and shorter periodicities (2-8, 15-18 years).

Conclusions:

• A novel, reliable framework for high-resolution NPP reconstruction was established.
• The study clarifies the response mechanisms of the boreal forest carbon cycle to climate variability, particularly drought.
• Findings highlight the sensitivity of boreal forest NPP to climate change and provide crucial data for carbon cycle research.