Wang et al. (2025) Drivers of agricultural ecosystem resilience under compound atmospheric-soil drought in Northeast China

Identification

• Journal: Ecological Indicators
• Year: 2025
• Date: 2025-12-11
• Authors: Luchen Wang, Youzhu Zhao, Muhammad Abrar Faiz, Shehakk Muneer Baluch, Min Xu, Dongqi Liu, Mo Li
• DOI: 10.1016/j.ecolind.2025.114525

Research Groups

• School of Water Conservancy and Civil Engineering, Northeast Agricultural University, Harbin, Heilongjiang, China
• Heilongjiang Province Key Laboratory of Smart Water Network, Northeast Agricultural University, Harbin, Heilongjiang, China
• National Key Laboratory of Smart Farm Technology and System, Harbin, Heilongjiang, China
• International Cooperation Joint Laboratory of Health in Cold Region Black Soil Habitat of the Ministry of Education, Harbin, Heilongjiang, China

Short Summary

This study quantitatively evaluates agricultural ecosystem resilience in Northeast China under compound atmospheric-soil drought, revealing significant spatial heterogeneity and an overall improvement trend in agricultural areas compared to non-agricultural regions, driven by precipitation and large-scale cropland with nonlinear effects from compound drought severity on resilience evolution.

Objective

• To quantitatively evaluate the resilience of agricultural ecosystems in Northeast China under compound atmospheric-soil drought using Gross Primary Productivity (GPP) data.
• To identify key environmental and agricultural management drivers influencing agricultural ecosystem resilience.
• To understand the nonlinear response mechanisms and temporal evolution of resilience, particularly concerning compound drought and agricultural intensification.

Study Configuration

• Spatial Scale: Northeast China, covering Heilongjiang, Jilin, Liaoning provinces, and the eastern part of Inner Mongolia.
• Temporal Scale: 2000 to 2023 for most datasets, with human footprint data from 2000 to 2020.

Methodology and Data

• Models used:
  • First-order autoregressive coefficient (AR1) for quantifying ecosystem resilience.
  • Mann-Kendall nonparametric trend test for AR1 temporal trends.
  • eXtreme Gradient Boosting (XGBoost) algorithm for resilience prediction and driver analysis.
  • SHapley Additive exPlanations (SHAP) framework for model interpretability and nonlinear response analysis.
  • Standardized Euclidean Distance (SED) for quantifying compound drought intensity.
• Data sources:
  • Gross Primary Productivity (GPP): Monthly GOSIF GPP v2 dataset (0.05° spatial resolution, 2000-2023); MODIS GPP product (500 m spatial resolution, 8-day temporal resolution) for cross-validation.
  • Vegetation Greenness: Normalized Difference Vegetation Index (NDVI) from MOD13A3 (1 km spatial resolution, monthly, 2000-2023).
  • Crop Distribution: Spatial Production Allocation Model (SPAM) products (5-arcminute grid, 2020); 2020 China cropland dataset (30 m spatial resolution).
  • Climate Variables: TerraClimate (4 km spatial resolution, monthly Vapor Pressure Deficit (VPD), soil moisture, temperature, precipitation, 2000-2023).
  • Human Activity Intensity: Annual Human Footprint maps (1000 m spatial resolution, 2000-2020).
  • Topographic Data: Shuttle Radar Topography Mission Digital Elevation Model (DEM).
  • Soil Nutrient Data: Soil organic carbon and total nitrogen dataset (1 km spatial resolution).
  • All datasets were resampled to a common 5-arcminute grid.

Main Results

• Agricultural ecosystem resilience in Northeast China exhibits significant spatial heterogeneity: 41.82 % of the area shows no significant change, 33.31 % shows declining resilience, and 24.87 % shows increasing resilience.
• Agricultural ecosystems in Northeast China display an overall trend of resilience improvement compared to non-agricultural regions.
• Compound drought events occur infrequently (annual mean frequency of 1.95 %), while single VPD and soil moisture droughts are more widespread (average frequencies around 9 %).
• Precipitation and the proportion of large-scale cropland are the most influential drivers of static agricultural ecosystem resilience (AR1).
• The proportion of large-scale cropland exhibits a nonlinear threshold effect on resilience: low proportions reduce resilience, while high proportions (above approximately 50–70 %) enhance it.
• Crop diversity shows context-dependent effects; in areas with increasing resilience, excessively high crop diversity may reduce system resilience.
• Compound drought severity has a relatively minor impact on static resilience but plays a significant and nonlinear role in the evolution of resilience trends, particularly in areas with increasing resilience.
• In resilience-increasing areas, a negative-to-positive switching effect of compound drought severity is observed: mild drought accelerates resilience enhancement, while extreme drought slows it down.

Contributions

• Provides the first quantitative evaluation of agricultural ecosystem resilience under compound atmospheric-soil drought in Northeast China, addressing a gap in previous studies focused on single drought indicators or natural ecosystems.
• Identifies complex regulatory mechanisms and nonlinear threshold effects of key drivers (e.g., large-scale cropland proportion, crop diversity, compound drought severity) on agricultural ecosystem resilience.
• Utilizes advanced machine learning (XGBoost-SHAP) to provide pixel-level interpretability of driver contributions, quantifying both magnitude and direction, which is a significant improvement over traditional regression models.
• Distinguishes between drivers of static resilience patterns and dynamic resilience evolution trends, offering a more comprehensive understanding of system responses to environmental stressors.
• Offers scientific support for developing differentiated agricultural management strategies, optimizing crop configuration, and enhancing drought adaptation to promote sustainable agricultural development in Northeast China.

Funding

• National Natural Science Foundation of China Grants (52409011, 52479035)

Citation

@article{Wang2025Drivers,
  author = {Wang, Luchen and Zhao, Youzhu and Faiz, Muhammad Abrar and Baluch, Shehakk Muneer and Xu, Min and Liu, Dongqi and Li, Mo},
  title = {Drivers of agricultural ecosystem resilience under compound atmospheric-soil drought in Northeast China},
  journal = {Ecological Indicators},
  year = {2025},
  doi = {10.1016/j.ecolind.2025.114525},
  url = {https://doi.org/10.1016/j.ecolind.2025.114525}
}