
7 minute read
The Science Behind the Seed
What's the secret to successful soybean seed development?
By Joli A. Hohenstein
Identifying genes for useful traits helps breeders develop improved varieties. Better tools pave the way for better decisions, and rapidly developing technology delivers streams of insights on the environment and goodness-of-fit. But what is the secret? How do the most successful soybean seeds go from lab to field?
As you might expect when talking about genetics, the answer is rarely straightforward and always complex.
“And it gets more complicated when we look to combine more than one gene or trait,” says Dr. Eliana Monteverde Dominguez, assistant professor and leader of the soybean breeding program at the University of Illinois.
Maria Monteros, Bayer’s North America Varietal Breeding Lead (South) for Asgrow Soybeans, agrees, saying, “We test millions of combinations to narrow it down to the best products. Think of how we create the new varieties like solving a Rubik's Cube puzzle.”
The Process
Although it’s rarely a straight road, soybean breeding maintains a common thread: trial and then retrial and then trial again. Because it must be fieldand yield-viable.
In other words, no matter how a variety performs during testing, if it doesn’t perform in a farmer’s field and produce comparable or greater yield to current varieties, it isn’t a viable commercial choice.
At the University of Illinois, Dr. Monteverde and her team first test varieties in “plant rows,” which are small plots of 20 to 25 plants. They plant 3,000 to 4,000 of those. Each row has a genetically different component.
The next year, they plant in two-row plots. These are the preliminary yield trials and are typically tested in at least two locations, depending on how much seed is available.
Based on this trial, they make yield-based selections, and remove the seed types that didn’t perform. The numbers vary significantly from trial to trial, she says.

The following year, they plant in advanced trials. These are full-row plots, usually three to four locations planted in Urbana and two to three additional locations across Illinois, depending on maturity and seed quantity.
“Now we have data not just from preliminary trials but also different locations,” she explains. “We can make a better decision.”
The team can also test another year in advanced trials or send varieties to USDA Uniform Trials. These are planted in many more locations and can be one or two years long.
“It takes many years, and we do it that way to be sure we develop the best possible varieties,” she says.
Bayer and Asgrow soybeans take a similar approach, testing at multiple locations and in controlled environments with off-season locations that enable the completion of multiple growth cycles through multiple seasons, making progress faster.
“It starts with our customers and the robust testing pipeline,” says Maria. “We’re using prescriptive field evaluations and collecting a global network of data.”
The difference is the technology.
Promising New Technologies
Machine learning and artificial intelligence (AI) are being leveraged by large commercial breeders, such as Bayer, to develop Asgrow soybeans, which has established itself as a leader in this technology.
“Traditional breeding approaches would take five to seven years,” says Maria. “Now, we can cycle through things in about four months. We used to have about 15,000 data points for products that would launch. Now we have collected over 300,000 data points for the new varieties.”
A large amount of data can be extracted from the plants, but managing and analyzing the information presents challenges in terms of cost and interpretation.
The human element remains critical.
Think of it like this: AI is smart, but it is not wise. It can’t yet do what 40 years of human experience can: remove all the noise and make inferences based on background information. And AI itself requires training.
Bayer and Asgrow soybean researchers have expertise in many different specialties for that reason, including breeders, discovery scientists, mathematicians, data scientists, agronomists and even rocket scientists.
“The combined expertise of a talented team allows us to innovate, gain additional insights from the data, generate prediction models and test those predictions in the field,” Maria explains. “Because we have many different testing locations, we’re able to evaluate the performance in a specific soil type, with variable moisture and environmental conditions or the response to multiple pathogens, complemented by phenotyping for a specific pathogen in a controlled environment, to truly understand the genetics driving a plant’s response to those conditions or pathogens.”
Even as AI continues its rollout in agriculture and research, technologies such as GPS and unmanned aerial vehicles (UAVs) have already had a significant impact on plant breeding. “When I was a grad student, I would sit at the back of a planter with envelopes to plant,” says Maria. “Now, we can plant thousands of plots per hour and use GPS and UAVs to track performance.”
Every innovation starts with knowing what the customer thinks. “In order for us to know what we want to improve, we have to know what our customers need,” she explains.
Data automation rolls through many different products at Bayer and other breeders, enabling them to design varieties aligned with grower needs. “Innovation and new technology enable us to understand the genetic profile of each seed,” says Maria. “We make the promise of continuous innovation, but we are customer-centric first.”
Bayer and Asgrow soybeans use Precision Breeding to increase the rate of product improvement and match the new varieties to specific areas. Machine learning and AI may be newer tools, but scientists are working with diverse genetic ingredients like they always have, she says.

“For example, Illinois growers may face challenges with lodging or soybean cyst nematode (SCN). How do we deliver varieties to address those?” she explains. “The difference is we’ve gone from selecting the best to designing the best, utilizing our industry-leading germplasm.
New Approaches
In her position, Dr. Monteverde is working to augment traditional breeding with new genomics, computational and sensor-based approaches to accelerate cultivar development.
“We’re using more molecular markers and genomic selection,” she says. “We select the variety based on their genetic composition and some help from phenomics.”
Essentially, genomics refers to the DNA of the plant – the genetic material inside the plant. Phenomics refers to the plant’s phenotypes – how it looks, how it behaves, how it flowers – its outward appearance and behavior. Scientists use many markers to select for disease resistance, genes for high protein and other traits.
“The idea is to overlay research on breeding,” Dr. Monteverde explains. “I like to know what’s going on inside the plant.”
At the University of Illinois, the team is developing varieties with more genetic sources of resistance.
“Right now, there are varieties with SCN resistance, but only one resistance gene,” she says. “We already have experimental lines that have two different combinations of three genes with resistance to SCN. They are very promising.”
In general, before a variety is ready, its performance is predicted through genetic characterization. At an increasing number of locations, varieties go through multiple years of testing with more replications as they approach commercial launch.
In fact, HT4 is Asgrow’s fourth-generation technology that combines five herbicide tolerances with yield potential and brings more flexibility to protect that yield. “We’re already working with fifth-generation technology to bring in six herbicide tolerances,” says Maria.
The bottom line of breeding for Maria: “It’s about continuous innovation with Precision Breeding and exclusive, elite genetics not found in any other seed bag.”



