Skip to main content

Effects of Surface Ozone and Climate on Historical (1980–2015) Crop Yields in the United States: Implication for Mid-21st Century Projection

Abstract

Surface level ozone pollution imposes significant crop yield damages. However, the quantification has mainly involved chamber experiments, which may not be representative of results in farm fields. Additionally, the relative impacts of ozone under future climate change and their possible interactions remain poorly understood. Here we attempt to empirically fill this gap using historical county-level crop yield, ozone, and climate data in the United States. We explore ozone impacts on corn, soybeans, spring wheat, winter wheat, barley, cotton, peanuts, rice, sorghum, and sunflowers. We also incorporate a variety of climatic variables to investigate potential ozone-climate interactions. The results shed light on future yield consequences of ozone and climate change individually and jointly under a projected climate scenario. Our findings indicate significant negative impacts of ozone exposure for eight of the ten crops we examined, excepting barley and winter wheat. Meanwhile, corn exhibits to be more sensitive to ozone than soybeans. These results differ from those found under chamber experiments. We also find rising temperatures tend to worsen ozone damages while water supplies mitigate that. We find that the average annual historical damages from ozone reached $6.03 billion (in 2015 U.S. dollar) from 1980 to 2015. Finally, our results suggest that the damages caused by climate change-induced ozone elevation are much smaller than the damages caused by the direct effects of climate change itself.

This is a preview of subscription content, access via your institution.

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7

Notes

  1. To clarify, ppm and ppb are ozone concentration units and stand for parts per million and parts per billion, respectively. 1 ppm = 1000 ppb.

  2. See for example studies in the United States (Deschênes and Greenstone 2007; Schlenker and Roberts 2009; Burke and Emerick 2016), studies in China (Chen et al. 2016; Zhang et al. 2017; Chen and Gong 2020), and studies from a global perspective (Lobell et al. 2011; Deryng et al. 2014; Wing et al. 2021).

  3. For instance, the dose–response functions derived from experiments are usually expressed with simple linear functional forms (nonlinear functional forms also exist but are much less widely used). The most widely used linear function takes the form: \(y = a*Ozone + b\). Where y is the relative yield (RY); Ozone indicates ozone measurement; a is a negative coefficient indicating ozone damages; b normally takes a value around 1 so that in scenarios with no ozone, the relative yield will be 1 (i.e., no ozone damages).

  4. Agronomically, ozone enters plants via stomata and damages both the internal structure and physiological function (Fuhrer et al. 1997; Mills et al. 2007; Pleijel et al. 2007; Ainsworth 2008; Ainsworth et al. 2012). Additionally, temperature and water availability also impact stomatal conductance, and thus impact ozone uptake (Tai and Val Martin 2017).

  5. We did this by excluding counties that were labelled “irrigated” in USDA NASS.

  6. It is the monthly average of daily maximum temperature. It measures extreme temperatures, however, not as extreme as the hottest daily temperature in a certain month.

  7. To address potential multi-collinearity issues among climatic variables (FDD, GDD, precipitation, monthly maximum temperature, and SPEI), we report the variance inflation factor (VIFs) for corn, soybeans, spring wheat, and winter wheat in Table A4 in the Online Appendix. VIFs for precipitation, FDD, and SPEI are well below 10 (Wooldridge 2013). VIFs for GDD and monthly maximum temperature are higher but mostly remain below the cutoff, except for the GDD of soybean which has a VIF of 11.3. Nevertheless, we do not feel that VIF with such a margin would be a major threat to our estimation. See more discussion in the Online Appendix.

  8. While a linear relationship between ozone exposure and crop yields is recommended, we are not aware any evidence in the literature indicating a nonlinear relationship (i.e., a quadratic term of ozone).

  9. For instance, Deryugina et al. (2019) instrumented for air pollution using changes in local wind direction. Godzinski and Castillo (2021) collected a novel and large set of altitude-weather data as candidates and then filtered optimal instrument variables (IVs) for air pollutants (i.e., 10 IVs for ozone, 14 IVs for NO2, and 15 IVs for PM2.5).

  10. Nevertheless, we performed two sets of robustness checks with potential factors (i.e., nitrogen fertilizer application and electricity consumption) that could impact yields and ozone formation simultaneously. The estimations on ozone variables barely changed after the inclusion of those factors. See the Online Appendix for details.

  11. These dose–response functions are y = − 0.0036x + 1.02 for corn and y = − 0.0116x + 1.02 for soybean. Here y is the relative yield and x indicates the AOT40 ozone measure.

  12. This is based on the dose response function: y =  − 0.0161x + 0.99 in (Mills et al. 2007).

  13. It should be noted that Mills et al. (2007) was not clear about which type of wheat they considered (i.e., spring wheat or winter wheat). They adopted the wheat-ozone relationship from (Fuhrer et al. 1997) where the authors pooled spring wheat and winter wheat data to estimate the dose–response function.

  14. It should be noted that while we performed robustness checks independently for each of the ten crops, we only report the results for corn, soybean, spring wheat, and winter wheat in the Online Appendix.

  15. The data were obtained from the North American Regional Reanalysis (NARR, https://psl.noaa.gov/data/gridded/data.narr.html) with a spatial resolution of 32 km*32 km (0.3°) and a monthly temporal resolution. We converted the data to county level by weighted averaging over grid cells that overlap each county. The county level monthly variables were averaged across crop-specific growing seasons.

  16. One should be noted that SPEI is an index variable. A unit increase in SPEI could indicate a significant change in water conditions from dry to wet. Therefore, we adopted an increase in 0.1.

  17. These two crops are the most widely planted crops in the United States with a total of 179.2 million planted acres in 2018 (USDA NASS 2019).

References

Download references

Acknowledgements

We would like to thank the Editor Carlo Fezzi and two anonymous referees for helpful comments and advice. All errors remain our own.

Funding

This study was funded by the Department of Agricultural Economics, Texas A&M University.

Author information

Authors and Affiliations

Authors

Corresponding author

Correspondence to Bruce McCarl.

Ethics declarations

Conflict of interest

The authors have no conflicts of interest to declare that are relevant to the content of this article.

Additional information

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Supplementary Information

Below is the link to the electronic supplementary material.

Supplementary file1 (DOCX 1321 kb)

Rights and permissions

Reprints and Permissions

About this article

Verify currency and authenticity via CrossMark

Cite this article

Da, Y., Xu, Y. & McCarl, B. Effects of Surface Ozone and Climate on Historical (1980–2015) Crop Yields in the United States: Implication for Mid-21st Century Projection. Environ Resource Econ 81, 355–378 (2022). https://doi.org/10.1007/s10640-021-00629-y

Download citation

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s10640-021-00629-y

Keywords

  • Crop yields
  • Surface ozone
  • Climate change
  • Economic damages