Changes in diurnal temperature range and national cereal yields

David B. LobellAgricultural 2007 · DOI

Introduction

Standard assessments of climate change impacts on agriculture typically focus on shifts in mean surface temperature ($T_{avg}$). However, plants respond to the full diurnal cycle, and a specific increase in $T_{avg}$ can result from various combinations of changes in daytime maximums ($T_{max}$) and nighttime minimums ($T_{min}$). This study investigates the independent influence of the Diurnal Temperature Range (DTR)—defined as the difference between $T_{max}$ and $T_{min}$—on the yields of the world's three primary cereal crops: wheat, rice, and maize.

Methodology

To isolate the impact of climate variability from long-term technological advancements, such as improved cultivars and management practices, the study utilizes a first-difference statistical approach. By analyzing year-to-year changes in national yields and climate variables from 1961 to 2002, the model effectively controls for non-climatic trends.

Climate data from the Climate Research Unit (CRU) was spatially weighted to reflect the actual growing regions of each crop within the top ten producing nations. Furthermore, the study integrated projections from 11 General Circulation Models (GCMs) under the IPCC A2 emissions scenario to evaluate how future changes in both $T_{avg}$ and DTR might affect global food security by the mid-21st century.

Crop Sensitivity to Temperature Fluctuations

The empirical analysis revealed that DTR is a significant predictor of yield for maize and rice, whereas wheat showed less sensitivity at the national scale.

  • Maize and Rice: In 8 of the 30 country-crop combinations studied, increased DTR was associated with significant yield reductions. The negative response suggests that the physiological stress caused by higher daytime peaks—such as increased evaporative demand and potential spikelet sterility—outweighs any benefits from cooler nighttime temperatures.
  • Wheat: The lack of significant DTR sensitivity in wheat may be attributed to its lower optimal development temperature. In some regions, higher daytime temperatures might be compensated by changes in the rate of crop phenology, leading to a more neutral net effect.
  • Regional Variations: Japan provided a notable exception; rice yields there showed a positive correlation with warming, likely because current temperatures in Japanese rice-growing regions remain below the crop's physiological optimum.

Future Projections and "Tail Risk"

A critical finding of the study is the positive correlation between $T_{avg}$ and DTR in many climate model projections. Scenarios that predict the highest average warming also tend to predict the greatest increases in diurnal swings.

For US maize, accounting for DTR shifts the mean projected yield loss from -23% to -25%. More significantly, it dramatically alters the "tail risk" of the distribution: the 5th percentile (worst-case) scenario drops from a -35% loss to a -45% loss when DTR effects are included. Conversely, in India, models project a potential decrease in DTR for rice-growing regions, which may provide a slight buffer against the negative impacts of rising average temperatures.

Conclusion and Limitations

The inclusion of DTR provides a more nuanced and potentially more severe outlook for global crop yields under climate change. However, significant uncertainties remain. Current GCMs often struggle to accurately simulate historical DTR trends, frequently underestimating the magnitude of observed changes. Additionally, the statistical coupling of DTR with precipitation and cloud cover makes it difficult to isolate "pure" temperature effects. Future research must focus on the specific biological mechanisms—such as the impact of $T_{max}$ on reproductive success—to refine the next generation of process-based crop models.

Summary updated

BibTeX
@inproceedings{lobell2007changes,
  author = {David B. Lobell},
  title = {{Changes in diurnal temperature range and national cereal yields}},
  booktitle = {Agricultural},
  year = {2007},
  doi = {10.1016/j.agrformet.2007.05.002},
}
Figure