Kabtih et al. (2025) Risk of successive hot-pluvial extremes on crop yield loss over global breadbasket regions

Identification

• Journal: Communications Earth & Environment
• Year: 2025
• Date: 2025-11-22
• Authors: Abebe K. Kabtih, Cheng Qian
• DOI: 10.1038/s43247-025-02989-5

Research Groups

• State Key Laboratory of Earth System Numerical Modeling and Application, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China
• University of Chinese Academy of Science, Beijing, China
• High Performance Computing & Big Data Analytic Center of Excellence, Addis Ababa Science and Technology University, Addis Ababa, Ethiopia
• Division of Maths-Physics-Statistics, Natural and Applied Science, Addis Ababa Science and Technology University, Addis Ababa, Ethiopia

Short Summary

This study investigates the risk of successive hot-pluvial extremes (SHPEs) on crop yield loss in global breadbasket regions, revealing an increasing trend in SHPE occurrences from 1979 to 2024 and a significant association with synchronized low yields for maize, rice, soybean, and wheat. Using machine learning models, the research quantifies yield sensitivity and predicts yield responses to these compound extreme events.

Objective

• To investigate the risk of synchronized low crop yield associated with successive hot-pluvial extremes (SHPEs) over global breadbasket regions.
• To quantify crop yield sensitivity to SHPE abrupt alternation.
• To model crop yield responses to the emergence of frequent SHPE signals.
• To hypothesize that adverse interactions between SHPE events and crop-physiology responses have a significant synergistic impact on staple food yield.

Study Configuration

• Spatial Scale: Global breadbasket regions. Meteorological data at 1° × 1° resolution. Crop yield data at 0.5° resolution, coarsened to 10° to match meteorological variables.
• Temporal Scale: SHPE analysis: 1979-2024. Crop yield data: 1981-2016. Base period for extreme event thresholds: 1981-2010.

Methodology and Data

• Models used:
  • XGBoost classifier and regression
  • Random Forest (RF) classifier and regression
  • Locally Weighted Regression Smoothing (LOWESS) for detrending yield data
  • Sen’s slope and Mann–Kendall test for trend analysis
  • Composite analysis
• Data sources:
  • Meteorological Data: European Centre for Medium-Range Weather Forecasts (ECMWF) fifth-generation atmospheric reanalysis (ERA5) daily maximum temperature (Tmax) and precipitation data.
  • Crop Yield Data: Global Dataset of Historical Yield (GDHY) v1.2+v1.3.
  • Cropland Area Fraction: Cropland and pasture area in the 2000 Dataset.
  • Crop Calendar Data: From Sacks et al. (2010).

Main Results

• A notable increasing trend in recurrent SHPE occurrences during intra-growing seasons was observed from 1979 to 2024, with the rate of increase accelerating in recent decades (statistically significant at p < 0.01, with adjusted R² values ranging from 0.66 to 0.81 for quadratic trend models).
• Low-latitude maize, rice, and soybean lands (e.g., Africa, Southeast Asia, South America, India, China) exhibit a high frequency of SHPEs. Wheat lands in the eastern United States, Ethiopia, India, Myanmar, and North China are also vulnerable.
• SHPEs show longer durations in the Southern Hemisphere, with maize and rice lands experiencing approximately 0.45% or more of their crop calendar under SHPEs, compared to 0.3% or more for soybean and wheat croplands.
• Composite analysis reveals that pluvial intensity following hot events is increased in 63% of maize, 62% of rice, 57% of soybean, and 54% of wheat lands. Similarly, heatwave intensity preceding pluvial events is intensified.
• SHPE events lead to crop yield reductions of up to -4% compared to expected yields based on long-term trends. Globally, 51-63% of maize, rice, and wheat lands, and 48% of soybean lands, experience negative yield percentage changes.
• Linear regression analysis indicates that a majority of croplands (52.7% maize, 62.4% rice, 47.4% soybean, 51.4% wheat) exhibit a negative sensitivity to recurrent SHPEs, particularly in Brazil (rice), Central Africa (maize, rice), and South America (soybean).
• The XGBoost classifier predicts a negative likelihood of 49% for maize, 50% for rice, 49% for soybean, and 50% for wheat yield loss due to frequent SHPE events.
• XGBoost regression explains approximately 36.2% of maize, 40.9% of rice, 42.1% of soybean, and 37.4% of wheat yield variability globally due to recurrent SHPEs. The Root Mean Square Error (RMSE) values are 0.57 t/ha for maize, 0.56 t/ha for rice, 0.26 t/ha for soybean, and 0.57 t/ha for wheat.
• Hotspot regions for yield loss are predominantly found in marginal communities and non-developed countries (e.g., Bantu region of Africa, South America, South and Southeast Asia), which experience frequent and prolonged SHPE events. Developed countries (e.g., US, Eastern Europe) show a lesser likelihood of SHPE-related yield losses.

Contributions

• This study provides the first global investigation into the risk of successive hot-pluvial extremes (SHPEs) on crop yield loss across major breadbasket regions.
• It quantifies the increasing trends and spatial patterns of SHPE occurrences and their durations during intra-growing seasons from 1979 to 2024.
• The research employs advanced machine learning models (XGBoost classifier and regression) to quantify crop yield sensitivity and model yield responses to SHPEs, explaining a significant portion of yield variability.
• It highlights the synergistic and non-linear impacts of SHPEs on crop physiology and yield, demonstrating that their combined effect can be more detrimental than individual extreme events.
• The findings offer crucial insights for developing climate-smart agricultural practices and tailored adaptation strategies, particularly for vulnerable regions in the Global South facing heightened food insecurity risks.

Funding

• National Key R&D Program of China (grant No. 2023YFF0805504)
• National Natural Science Foundation of China (42175175)
• Jiangsu Collaborative Innovation Center for Climate Change

Citation

@article{Kabtih2025Risk,
  author = {Kabtih, Abebe K. and Qian, Cheng},
  title = {Risk of successive hot-pluvial extremes on crop yield loss over global breadbasket regions},
  journal = {Communications Earth & Environment},
  year = {2025},
  doi = {10.1038/s43247-025-02989-5},
  url = {https://doi.org/10.1038/s43247-025-02989-5}
}