27 Physical Geography, 2009, 30, 1, pp. 27���42. Copyright �� 2009 by Bellwether Publishing, Ltd. All rights reserved. DOI: 10.2747/0272-3646.30.1.27 SENSITIVITY OF CORN AND SOYBEAN YIELD IN ILLINOIS TO AIR TEMPERATURE AND PRECIPITATION: THE POTENTIAL IMPACT OF FUTURE CLIMATE CHANGE David Goldblum Department of Geography Northern Illinois University DeKalb, Illinois 60115 Abstract: The potential impact of climate change on agricultural production has fre- quently been evaluated at national and regional scales. This study considers the potential county-scale impact of climate change on corn (Zea mays L.) and soybean (Glycine max [L.] Merr.) yield in the important agricultural state of Illinois, USA. By identifying specific monthly climate variables (mean daily temperature and precipitation) to which corn and soybean yield is sensitive, this study compares monthly regional General Circulation Model (GCM) predictions with the monthly climate variables to which corn and soybean yield is sensitive to predict crop yield under future climate. Corn yield is negatively corre- lated with July and August temperature in much of the state, and positively correlated with precipitation from the previous September (in the central portion of the state) and July and August precipitation in most of northern and southern Illinois, respectively. Soybean yield is negatively correlated with mean monthly temperature in central and southern Illinois during the summer, and positively correlated with July and August precipitation in the same regions. Given the regional GCM prediction for increased summer temperatures and summer drought, both corn and soybean yield will likely decrease under future climate conditions. This is likely to be most pronounced in the central and southern portions of Illinois. Additionally, given higher summer temperatures, the risk of summer drought is more pronounced. [Key words: climate change, agriculture, corn, soybean.] INTRODUCTION Unlike other environmental systems, the potential impact of climate change on agriculture embodies both positive and negative aspects. First, agriculture is vitally important to human existence and any decrease in crop yields literally threatens humanity. Mitigating that danger however, agricultural systems may be highly adap- tive (with human manipulation) to those environmental changes (Helms et al., 1996 Rosenzweig and Hillel, 1998 Evenson, 1999 Smith, 2004), assuming appro- priate measures are taken. Numerous elevated-CO2 enclosure studies suggest that yield for many common C3 crops (rice, soybeans, and wheat) are likely to increase while C4 plants (corn and sorghum) generally do not show increased photosyn- thetic activity (e.g., Kimball, 1983 Cure and Acock, 1986 Long et al., 2004, 2006), potentially resulting in decreased crop yield. However, these results from controlled enclosure studies may not translate to field conditions. Long et al. (2006) recently reported that C3 crops exposed to elevated CO2 levels using free-air concentration enrichment (FACE) technology exhibited dramatically smaller yield increases than was predicted from enclosure studies, and C4 plants showed no significant increase in growth under elevated CO2 levels compared to current ambient CO2 levels.

28 DAVID GOLDBLUM Unfortunately, many of the current crop yield models utilize results from enclosure studies when considering the impact of enhanced CO2, and therefore may overpre- dict the positive benefit of elevated CO2 levels (Kimball, 1983 Long et al., 2006). Cautious predictions based on these models are still useful nonetheless. Globally, climate change might only lead to a small decrease in crop production (Rosenzweig and Parry, 1994). Considering the United States as a whole, some studies have suggested that agricultural production may increase (Reilly et al., 2003) while others suggest minimal increases due to climate change (Smith, 2004 Isik and Devadoss, 2006), while at the local scale and in low-input agricultural sys- tems in poorer regions, crop yield may be adversely affected by climate change (Rosenzweig and Parry, 1994 Fuhrer, 2003). However, other studies show a general decrease in crop yield due to climate change (Brown and Rosenberg, 1999). Addi- tionally, other variables may need to be considered���for example, crop damage has been predicted to increase in response to excess precipitation (Rosenzweig et al., 2002), and pests are likely to be more prevalent (Rosenzweig and Hillel, 1998). Narrowing the geographic range within the United States to consideration of regional or statewide impacts, the question becomes more complex (Mearns, 2003). In general, Lobell and Asner (2003) identified corn and soybean yields from the midwestern United States to be favored by cooler, wetter years with a region in the northern Great Plains favored by hotter, drier years. Both Reilly et al. (2003) and Alexandrov and Hoogenboom (2000) predict that overall crop yields would likely drop in the southeastern United States due to climate change, but Alexandrov and Hoogenboom (2000) suggest that increased CO2 levels would ameliorate that neg- ative effect and lead to an increased yield for some crops (soybeans and peanuts) but still decreased yield in others (corn and winter wheat). Models run by Izaurralde et al. (2003) predict increased yield for dryland corn in the Great Lakes, Corn Belt, and Northeastern regions, while irrigated corn yield would increase universally. Similarly, artificial summer watering of the Illinois corn crop increased yield com- pared to drier control plots (Changnon and Hollinger, 2003), although temperature was not manipulated. For soybeans, Izaurralde et al. (2003) predict that yields will decrease in all regions except the Great Lakes and Northeastern region, a finding echoed by Southworth et al. (2002b). Generally, if changes in agricultural systems do occur, it will likely be more in response to changes in temperature and precipi- tation patterns than increases in CO2 levels (Fuhrer, 2003). Most studies completed to date rely on results from crop response modeling (i.e., DSSAT and EPIC). To assess the degree to which CO2 increases will affect agricul- ture, these models rely on results from enclosure studies (Long et al., 2006). These models use daily temperature, solar radiation, and precipitation values to predict growth rates, yield, and photosynthesis rates. To evaluate the impact of climate change, the models are run with simulated temperature, precipitation, and poten- tially CO2 values. A smaller number of studies utilize correlation or yield response models that consider plant growth and specific climate variables (Dixon et al., 1994). These studies use monthly (or seasonal) climate variables to generate crop yield response functions or correlations between monthly climate variables and crop yield. This approach differs from the crop response modeling approach in that it looks for

CLIMATE CHANGE AND ILLINOIS CROPS 29 statistical correlations between crop yield and climate variables based on historical climate and yield records. Field or greenhouse responses to experimental alteration in temperature, moisture, or CO2 levels are not considered in this approach. Although global or regional assessments of the impact of climate change on agri- culture are important, statewide or county-level assessments are potentially more useful to agricultural practitioners, and thus are the focus of this study. Within the state of Illinois, approximately 11 million hectares are under agricultural produc- tion, comprised of 73,000 farms, covering 80% of the state���s land area (USDA, 2004). The products of these farms generate approximately $9 billion per annum (~75% by corn and soybeans alone) out of a total state economy of approximately $500 billion (USDA, 2006). Building upon the research previously conducted on this topic, this paper first identifies the statistical correlation of mean monthly air temperature and monthly precipitation with crop yield for both soybeans and corn at the county scale within the state of Illinois. Second, regional general circulation model (GCM) predictions for monthly temperature and precipitation values (in the 2050s���2060s) within the state of Illinois are then compared to the climate variables determined to control crop yield with the aim of predicting the likely impact of modeled future climate on corn and soybean yield in Illinois. METHOD Climate Data Monthly precipitation and mean temperature data were collected through the United States Historical Climatology Network (USHCN http://cdiac.ornl.gov/ epubs/ndp/ushcn/newushcn.html). The USHCN has approximately 40 weather sta- tions spread throughout Illinois with most stations collecting from the 1890s to the present. Occasionally a weather station in an adjacent state was the nearest to the center of an Illinois county in that case the out-of-state station was used. Crop Yield Data Crop yield (bushels/acre) for soybeans and corn are available for each county in Illinois (1927���present) through the United States Department of Agriculture���s National Agricultural Statistics Service (NASS www.nass.usda.gov). Given improvements in genetic and fertilizer applications over this time period (Thompson, 1969 Specht et al., 1999 Hu and Buyanovsky, 2003), it was necessary to statistically detrend the time-series of crop yield to remove these factors, and iso- late the role of climate. Soybean yield was detrended with linear regression. Mean (��sd) r2 for soybean regression equations was .84 (��.04), and all regressions were highly significant (p .0001). Corn yield was detrended with a quadratic regression equation mean (��sd) r2 for corn regression equations was .85 (��.03), and all regres- sions were highly significant (p .0001). For all detrending regressions, no first- order autocorrelation was detected (Durbin-Watson D) and visual inspection of residual patterns indicated no evidence of heteroscedasticity.