Field warming experiments shed light on the wheat yield response to temperature in China

Chuang Zhao, Shilong Piao, Yao Huang, Xuhui Wang, Philippe Ciais, Mengtian Huang, Zhenzhong Zeng, Shushi PengNature Communications 2016 · DOI

TL;DR

This paper synthesizes data from field warming experiments, crop models, and statistical models to understand how wheat yield in China responds to rising temperatures. It reveals that while models often predict yield losses, field experiments show a more nuanced picture, with positive yield responses in cooler, wetter regions, suggesting that adaptation strategies need to be region-specific.

What Problem Does This Paper Address?

Predicting the impact of climate change on global food security is a critical challenge. Wheat, a staple crop, is particularly sensitive to temperature. However, there's significant uncertainty about how future warming will affect wheat yields, especially in major breadbaskets like China. Previous studies using different modeling approaches have yielded conflicting results, making it difficult to develop effective adaptation strategies. This paper aims to reconcile these differences by comparing the temperature sensitivity of wheat yield (S<sub>Y,T</sub>) across three distinct methodologies: direct field experiments, local process-based crop models, and statistical models. The goal is to provide a more robust understanding of regional yield responses and to identify ways to improve future crop modeling.

Methods

The authors employed a meta-analysis approach, compiling data from various sources:

  • Field Warming Experiments: They gathered 46 S<sub>Y,T</sub> estimates from field experiments where temperatures were artificially increased. This method provides direct, empirical observations of yield response.
  • Local Process-Based Crop Models: They collected 102 S<sub>Y,T</sub> estimates from models that simulate crop growth and development based on physiological processes, climate, and soil conditions. These models are often calibrated for specific regions.
  • Statistical Models: They also included S<sub>Y,T</sub> estimates derived from statistical relationships between historical crop yields and climate variables. These models capture observed correlations but may struggle with extrapolation beyond current climate conditions.

To compare these diverse data sources, they used a standardized metric: S<sub>Y,T</sub>, representing the percentage change in wheat yield per degree Celsius increase in temperature. They also performed regression analyses to explore how S<sub>Y,T</sub> relates to background climate variables like growing-season temperature, water supply, and diurnal temperature range. Finally, they compared their findings with simulations from global gridded crop models (GGCMs) participating in the Inter-Sectoral Impact Model Intercomparison Project (ISI-MIP).

Datasets

  • Field Warming Experiments: 46 S<sub>Y,T</sub> estimates from studies conducted in China.
  • Local Process-Based Crop Models: 102 S<sub>Y,T</sub> estimates from studies in China.
  • Statistical Models: S<sub>Y,T</sub> estimates derived from statistical analyses of wheat yield and climate data in China.
  • Global Gridded Crop Models (GGCMs): Simulations from the ISI-MIP-Phase 1 project, providing spatially explicit yield projections under different climate scenarios.
  • Climate Data: Gridded climate data (temperature, precipitation, daylight hours, diurnal temperature range) from the Chinese Academy of Sciences' China Meteorological Forcing Dataset (CMFD) and other sources were used for analysis and to drive models.

Key Insights and Ideas

  • Models vs. Reality Discrepancy: A significant finding is the divergence between model predictions and field experiment observations. While local crop models tend to predict negative S<sub>Y,T</sub> (yield loss with warming), field experiments reveal a more complex picture, with some regions showing positive responses.
  • The Importance of Local Climate: The study highlights that S<sub>Y,T</sub> is not uniform across China. Cooler regions with sufficient water supply tend to show positive yield responses to warming, while warmer, drier regions are more susceptible to negative impacts. This underscores the need for region-specific adaptation strategies.
  • Water is Key: Water supply emerges as a critical factor influencing temperature sensitivity. In regions with limited water, warming can exacerbate drought stress, leading to yield losses. Conversely, in wetter areas, warming might be beneficial, potentially by increasing water use efficiency or extending the growing season.
  • GGCMs Show Promise: Global gridded crop models (GGCMs) from the ISI-MIP project appear to better capture the spatial patterns of S<sub>Y,T</sub> observed in field experiments compared to local crop models. This suggests that GGCMs might be more reliable for projecting large-scale climate impacts.
  • Beyond Simple Temperature: The study suggests that factors beyond just average temperature, such as diurnal temperature range and water availability, play crucial roles in determining wheat yield response to warming.

Key Results

  • Average S<sub>Y,T</sub>:
    • Field warming experiments: -0.7 ± 7.8% per °C
    • Local process-based models: -5.7 ± 6.5% per °C
    • Statistical models: 0.4 ± 4.4% per °C
  • Regional Differences: Warming experiments showed positive S<sub>Y,T</sub> in cooler, non-limiting water regions (e.g., Southeast China) and negative values elsewhere. Models generally predicted negative S<sub>Y,T</sub> across most regions.
  • Climate Correlations: Field warming experiments revealed that S<sub>Y,T</sub> is negatively correlated with growing-season mean temperature (T<sub>GS</sub>) and positively correlated with water supply (W<sub>GS</sub>). The relationship with T<sub>GS</sub> crosses zero at approximately 9.3°C, suggesting that above this threshold, warming is likely to reduce yields.
  • GGCM Agreement: GGCM simulations showed better agreement with the spatial patterns of S<sub>Y,T</sub> derived from field warming experiments than local crop models.

Limitations and Future Work

  • Experimental Limitations: The authors acknowledge that field warming experiments have inherent limitations, including variations in experimental methods, potential atmospheric feedbacks not captured by models, and the difficulty in perfectly controlling all environmental factors.
  • Model Calibration: Local crop models are often calibrated for specific regions, which can limit their applicability and lead to discrepancies when compared to broader datasets or different regions.
  • Data Scarcity: While this study compiles a significant number of observations, the authors note that only 46 field warming experiments were available, and the spatial coverage of modeling sites could be improved.
  • Future Work: The study suggests that the insights gained from field experiments can be used to improve crop models. Further research is needed to better understand the physiological mechanisms behind the observed responses, particularly the positive impacts of warming in certain regions. More robust year-to-year comparisons between modeled and field-measured sensitivities are also recommended. The authors emphasize the need for continued research into crop physiology and the development of region-specific adaptation strategies to mitigate the risks to food security posed by climate change.

Summary updated

Abstract

Wheat growth is sensitive to temperature, but the effect of future warming on yield is uncertain. Here, focusing on China, we compiled 46 observations of the sensitivity of wheat yield to temperature change (S<sub>Y,T</sub>, yield change per °C) from field warming experiments and 102 S<sub>Y,T</sub> estimates from local process-based and statistical models. The average S<sub>Y,T</sub> from field warming experiments, local process-based models and statistical models is -0.7±7.8(±s.d.)% per °C, -5.7±6.5% per °C and 0.4±4.4% per °C, respectively. Moreover, S<sub>Y,T</sub> is different across regions and warming experiments indicate positive S<sub>Y,T</sub> values in regions where growing-season mean temperature is low, and water supply is not limiting, and negative values elsewhere. Gridded crop model simulations from the Inter-Sectoral Impact Model Intercomparison Project appear to capture the spatial pattern of S<sub>Y,T</sub> deduced from warming observations. These results from local manipulative experiments could be used to improve crop models in the future.
BibTeX
@inproceedings{zhao2016field,
  author = {Chuang Zhao and Shilong Piao and Yao Huang and Xuhui Wang and Philippe Ciais and Mengtian Huang and Zhenzhong Zeng and Shushi Peng},
  title = {{Field warming experiments shed light on the wheat yield response to temperature in China}},
  booktitle = {Nature Communications},
  year = {2016},
  doi = {10.1038/ncomms13530},
}
Figure