Crop yield projections for biofuels fall short
In 2005 the US Department of the Environment and the US Department of Agriculture released an influential report claiming that by 2030 US farms could produce almost a billion tons of biomass for conversion to biofuel each year. More than half would come from annual crops, and the rest from perennial crops and manure.
A key assumption of the report is that grain yields will increase by 50% by 2030. Established yield trends suggest that we won’t reach that target.
Three annual crops dominate America’s farms: Corn, soybean, and wheat.
Corn yields doubled between 1950 and 1977, and almost doubled again by 2005. Although yields continue to climb, the annual rate of increase is slowing. If established trends continue, corn yield will reach 182 bushels per acre by 2030 — 28% more than in 2005.
Soybean yields doubled between 1950 and 2005, and the annual rate of increase is climbing. We can expect a 33% increase by 2030, to 54 bushels per acre. That’s impressive, but it’s still well short of the target required to achieve the billion ton vision.
Wheat yields are levelling off. We can expect to harvest about 42 bushels per acre by 2030, only 2% more than in 2005.
On average, yields for the big three annual crops should be about 22% higher in 2030 than they were in 2005 — not even half the increase needed to reach the target.
The fourth major crop is hay, a perennial mixture. Since the billion ton vision involves large-scale conversion of agricultural land to perennial bioenergy crops, it’s interesting to look at established yield trends for hay to make future yield projections for similar low-input perennial crops.
It doesn’t look good. Hay yields appear to have peaked around the turn of the millennium, and be on their way down. If the trend continues they’ll be 22% lower in 2030 than they were in 2005. A similar decline in hay yields has been observed for even longer in Canada, where researchers pin the bulk of the blame on rising fertilizer prices, and speculate that climate change may also play a part.
And what about other major crops grown in the US? I conducted the same analysis for barley, canola, flax, oats, rice, rye, sorghum, sweetpotato, sugarbeet, sugarcane, and sunflower (Click on the thumbnails below to see larger images). There was considerable variability between these crops, but the average yield increase between 2005 and 2030 was just 5% for the crops not already discussed.
Here are a few observations:
- Only flax showed any promise to achieve the 50% increase target.
- Although yields have increased for most crops over the past 58 years, the annual rate of increase is declining in most cases. Meeting the 50% increase target between 2005 and 2030 would require that each year’s yield increase be bigger than the year before. That’s not happening for most crops.
- Sugarcane — an intensively farmed perennial — has seen pronounced yield decline. The same phenomenon has been observed in Australia, where researchers blame soil degradation associated with long-term monoculture production. There may be a lesson here to temper the unbridled enthusiasm for long-term monocultures of dedicated energy crops like switchgrass and miscanthus.
- Like wheat and hay, yields of oats, rye, and sorghum appear to be levelling off or declining. The steep declines predicted by the projected lines should be treated with skepticism, though: The line fitting technique used here will always predict a steep decline if a plateau follows a rapid rise. I would expect an extended plateau, or gentle decline, to be more realistic.
- Canola and sunflower are relatively new large-scale crops. Canola and sunflower yield trends shown here are based on a relatively short history, and show a relatively poor fit to the data. Projections are less robust for these crops than for others.
My geekier readers may appreciate a slightly more technical description of how I conducted these analyses. Historical yield data came from the USDA. I used data from 1950 to 2008 when available, but records for canola and sunflower don’t go back that far. I projected the best fit second-order polynomial equation (y=ax2 + bx + c) for the historical data out to 2030, where y is yield and x is year (0 for 1950 to 80 for 2030). Terms of each equation are shown below, along with R2 values to show the strength of the fit. I was impressed with how closely the line matched historical trends in most cases, but the fit was poor for canola and sunflower.
I presented these projections on Tuesday at the Biofuels for Aviation Summit in Arlington, VA, as part of a talk called Can We Fly and Eat Too? I suggested that yield assumptions in the 2005 report were among the unrealistic assumptions used to claim that biofuels have greater potential than they do.