Projecting U.S. Crop Yields for Comprehensive Bioenergy Assessments: Trends, Variability, and Spatial Clustering.

The aggregate and distributional impacts of U.S. bioenergy mandates on land use change, net GHG emissions, environmental quality, and food prices will be highly influenced by future productivity trends along with inter-annual variations in crop yield.  To date, comprehensive assessments of bioenergy mandates commonly assume that current rates of yield increase will be maintained indefinitely and that yields in any given location are independent from those in other regions (e.g. Searchinger et al, 2008).   Year-to-year productivity changes due to environmental factors and the spatial autocorrelation of these phenomena can be significant and should be accounted for in comprehensive bioenergy assessments.  The objectives of this project were three-fold: 1) quantify contemporary (1970-2008) state-scale yield trends for maize, soybean, wheat, and alfalfa,  2) characterize inter-annual variability in these trends, and 3) explore the spatial structure and covariance among crops.  With a few exceptions (e.g. maize productivity plateaus in AZ and NM), yield trends for maize, wheat, and soybean were positive and linear, although absolute rates of gain varied significantly by crop and by state.  Likewise, substantial differences between states in year-to-year variability were reflected in higher average absolute deviations around the trend-line means (e.g. for maize, SC:  22% versus ID:  5%).   Within-state patterns of temporal variability were mixed with many states evidencing declining trends whereas others, specifically along the eastern seaboard (DE, MD, and PA), experienced increase variability.  As was expected, spatial clustering was evident with states such as AL, GA, FL, and MS having correlation coefficients > 0.6 for annual maize yields (p < 0.000).  Productivity patterns between crop types were also linked, with an especially strong correlation between maize and soybean.  Together, these results provide a realistic basis for projecting future crop yields and are discussed in the context of exploitable yield gaps in different regions of the U.S.