Corn is planted on about 90 million hectares throughout the United States each year. With all this data, it takes several months after the harvest for government agencies to analyze the total yield and quality of grain. Scientists are working to shorten that time limit, making predictions about the yield at the end of the season by the middle of the season. However, few researchers have dealt with crop quality predictions, especially on a large scale. The study at the University of Illinois begins to fill that void.
The study, published in Agronomy, uses a newly developed algorithm to predict the yield at the end of the season and the composition of the grain – the percentage of starch, oil and protein in the kernel – by analyzing the weather conditions over the three important stages in the development of corn. It is important that the forecasts refer to the whole of the maize crop in the Midwest of the United States, regardless of the genotypes of maize or production practices.
"There are several studies that evaluate factors that influence the quality of specific genotypes or specific locations, but prior to this study, we could not make general predictions on this scale," said Kerry Bucht-Wilmsmeyer, research assistant at the Department of Cultural Sciences in U by myself and co-author of the study.
As maize arrives on elevators throughout the Midwest every season, the US Grain Council takes samples to assess the composition and quality of their annual summary reports used for selling exports. It was this comprehensive database used by Boots-Wilmsmeyer and her colleagues in the development of their new algorithm.
"We used data from 2011 to 2017, which included years of drought, as well as the years that yielded record results and everything in each other," said Julian Sieuber, chief research researcher for the U Department of Crop Science and co-author of the study .
The researchers merged grain quality data with weather data from 2011 to 2017 from the regions that feed on each grain elevator. In order to build their algorithm, they concentrated on time over three critical periods – emergence, shelling, and filling of grain – and found that the strongest indicator of grain yield and composition quality is the availability of water during crushing and filling of grain.
The analysis has deepened, identifying conditions that lead to higher concentrations of oil or proteins – information that is important for cereal buyers.
The proportion of starch, oil and whey protein in wheat is affected by the genotype, the availability of nutrients and time. But the effect of time is not always clear when it comes to protein. In conditions of drought, the emphasized plants deposit less starch in the grain. Therefore, the grain has proportionally more proteins than those of plants that do not face dry stress. Good time can also lead to higher protein concentrations. Many water means more nitrogen is transported to the plant and incorporated into proteins.
In the analyzes, "over-average levels of protein and fat from cereals were favored by less nitrogen depletion during early vegetative growth, but also higher flowering temperatures, while higher protein concentrations of fats result from lower temperatures during flowering and grains, "the authors said in the study.
The ability to better predict protein and petroleum concentrations in cereals could affect global markets, taking into account the growing domestic and international demand for corn with higher protein for the use of feed. The new algorithm should be theoretically possible to make yields at the end of the season and predictions of quality weeks or months before the harvest, simply by looking at the time patterns.
"Other researchers have made predictions about real-time yields, using much more complex data and models. Our relatively simple approach, but we managed to add quality and achieve decent accuracy," says Bots-Wilmsmeyer. "The changing weather conditions that we thought were important in this study could be used in more complex analyzes to achieve even greater precision in predicting gender and quality in the future."