Using whole-genome sequencing and genotyping-by-sequencing approaches, Cornell University's Ed Buckler and his colleagues are investigating the breeding potential of what he calls "the world's most diverse crop" — maize.
Speaking at the Department of Energy Joint Genome Institute's sixth annual User Meeting, held in Walnut Creek, Calif., in March, Buckler said that while they may not have realized it at the time, maize breeders in Thailand, Mexico, and elsewhere have, since the turn of the 20th century, captured a great deal more genomic diversity with their land practices than those in the US have. Outside of the US, "there's lot of very good, useful variation sitting in these diverse, improved lines that we can really start to harness and use in breeding," Buckler said.
Now, using modern genomics technologies and mathematical modeling approaches to predict phenotypes from genotypes, Buckler said that researchers have found themselves "at this extremely exciting point in breeding." While, in general, maize genotyping costs $19 per sample at 96-plex, he expects to see the price dip below the $10 mark in the near future.
"At $10 per sample you can do things that were inconceivable [even] two or three years ago," he said. But perhaps the most exciting possibility, he continued, is that of accelerated crop breeding.
While Buckler acknowledged that the traditional maize breeding cycle — create crosses, inbreed those lines, perform small-scale trials to select the top-tier seeds, and then conduct larger-scale trials to see how those seeds perform in variable environments — has been largely successful over the last century, having increased yield nearly eight-fold in the US, Buckler said the process takes an average of five to seven years. "And in terms of evolution, you're only making a gain every time you make a cross you select," he added.
Buckler said that, by using genomic selection, researchers "can now make crosses and genotype them every four months," such that the underlying "engine of evolution can now tick at a rate that's more than 10 times faster." However, he added, fitting a mathematical model has been the rate-limiting step. "The whole key is whether or not we can make that mathematical model, that's based on genotype and phenotype mapping, good enough to predict which seeds to keep out of those crosses," he said. And even then, that "doesn't mean that four months later you're ready to go sell the seeds to a farmer," he added. Factoring in the need to create inbred lines, and potential statistical inefficiencies, he said, researchers say they can currently accelerate the maize breeding cycle three- to four-fold using genotypic information. Right now, he said, "the hardest part is training the model — that's expensive."
Going forward, Buckler expects that researchers will be able to "make each [cycle] tick as fast as we can make crosses," and that this approach will be applicable to other crops. Using genotyping-by-sequencing and other genomics-based approaches, Buckler said that during the next few decades researchers should be able to double crop yields using the same amounts of fertilizer and water. In addition, "I think we should be perennializing our crops to save our soils," Buckler said. "We also have a great opportunity to biofortify our crops for improving nutrition around the world."