Breeding For Flavor

Insights From University of MN Barley Breeder Kevin Smith

Written by Christopher on June 5, 2017

Malting is beer’s flavor frontier. Developing new malting recipes, experimenting with alternative grains, or using grain varieties selected for flavor offers immense opportunity for innovation in the brewhouse. In today’s article, I’m focusing on the the specifics of barley breeding and the possibilities to enhance flavor from the seed itself. Much of the information I’ll cover came from a conversation I had with Kevin Smith, Ph.D, the professor of barley genetics and breeding at the University of Minnesota.

Kevin Smith taking notes in barley trial plots

Image Courtesy of U of M 2016 Field Crop Trials Bulletin

For thousands of years, humans have improved plants. We’ve made seeds bigger, improved disease resistance, and adapted varieties to thrive in specific climates. The work started by neolithic humans continues today with the hobbyist cross-pollinating squash flowers by hand, the farmer selecting the most promising plants from their field, and the university scientist developing statistical models to speed iteration and improve outcomes.

Plant breeding is one of humanity’s longest running projects, and it’s a hard one. Unlike so many other project mediums, plants demand a really slow iteration time. Bread recipes, software, and music production can be iterated on multiple times per day. Brewing can take a month or longer to complete a full learning cycle, but plants take an entire growing season, and can be limited to just one chance a year.

Most farmers iterate on growing barley around 40 times in their life, and a breeder with access to greenhouses and warmer winter climates likely has only 120 growing seasons to find the most promising varieties during their career. This time and season constraint means breeders have to be very intentional about how they conduct and sort through hundreds of thousands of crosses each year. Technological tools make this work possible, and today many breeders use the statistical and genetic based method called Genomic Selection to accomplish their breeding goals.

Kevin’s barley lab uses Genomic Selection to develop cropping systems beneficial to growers, processors, and the environment. They also focus on the improvement of breeding methods, which strengthens the scientific foundation of breeding research and supports students to be better breeders in the future.

Breeding for specific flavors is not currently a central project within Kevin’s program, but it’s an idea he finds exciting and challenging, and which he is collaborating on with Pat Hayes, Ph.D., the professor of barley breeding at Oregon State University. Their research is using Genomic Selection to create a strong foundation for developing new, delicious barley varieties. To explain it properly, I’m going to talk about the past, present, and future of barley breeding.

Past Breeding

Breeding has long been a mixture of art and science, but recent developments have increased the scientific capabilities of breeders to successfully improve plants.

Gartons malting advertisement

Courtesy of Wikipedia

Gregor Mendel published the theory of genetic inheritance in the 1860’s, which laid the foundation for future exploration by the first commercial breeders in the 1890’s. John Garton figured out how to create hybrid cereal grains at that time by artificially crossing different varieties together to create improved varieties, and eventually built Gartons Limited, the British Empire’s largest plant breeding and seed company. We’re growing several of the original oat and barley varieties developed by Garton, including Abundance, an oat variety from 1892 and the first commercially released hybrid cereal.

John Garton

John Garton. Courtesy of Wikipedia

In the century since, barley breeders have taken barley from the original landraces, which were varieties selected over time by farmers, to hybrids bred for a variety of improved traits. Landraces eventually become genetically uniform after continuous selection, necessitating the introduction of new genetic diversity to continue breeding improvement. Breeders can accomplish this either through crossing in promising parental lines, or by inducing mutations. With this new diversity, new crosses can be compared to find varieties that meet the breeder’s goals.

For barley, these goals are for both farmers and processors. The primary agricultural focuses are improved yields, plants with shorter and stiffer straw, better disease resistance, a convenient ripening time, and hardiness in cold climates. Malting characteristics of importance are kernel plumpness, lower protein content, higher extract, rapid germination, adequate enzymes, and lower beta glucans.

The following table shows the progression from older to newer varieties, and the improvements breeders achieved across multiple areas. Notice how flavor isn’t a trait? Klages is one of 2 barley varieties Kevin says he hears brewers regularly ask about. It fell out of production because of the traits in this table, but brewers continue to ask for its flavor.

Klages Harrington Ellice
Relative grain yield (%) 100 105 106
Time to maturity (days) 96 93 93
Lodging (Scale 1-9) 4 3 2
Extract (%) 79.7 80 80.3
Diastatic power 113 120 132
Alpha-Amylase 44 46 41
Protein 12.2 12.1 11.8
Wort beta-glucan 140 75 36
Malts and Malting by Briggs, page 78

Golden Gophers

U of Minnesota

The University of Minnesota has played an important role in this steady march of barley improvement. 19 varieties have been released since 1918, representing the successful discovery of superior varieties from millions of possible progeny. The first variety was developed by Harry Harlan, the namesake of <a href="/harlan/" target="blank" rel="noopener noreferrer">The Harlan Society_, our open-source heritage seed project.

After traveling and collecting 5,000 varieties from around the world, Harlan spent part of his career at the U of MN. He released Manchuria while there, one of the original US varieties that became the genetic ancestor of many modern varieties.

These 19 varieties have had a huge economic impact. <a href="" target="blank" rel="noopener noreferrer">According_ to the U’s website, “a 1992 economic study documented that about two-thirds of all beer produced in the U.S. contained barley developed by U of M Agricultural Experiment Station scientists.” In the time since, disease and other factors have forced production out of Minnesota and into western states. Minnesota grew a million acres of barley in the 1990’s, but it’s down to 100,000 today.

This has been challenging for Rahr, which owns the largest single-site malting facility in the world, just outside of the Twin Cities. Historically they were surrounded by barley producers, but now face rising transportation costs from importing barley from western states and Canada.

Kevin’s primary research focus is developing cropping systems beneficial to growers, processors, and the environment. Perhaps the best example of this is his work to develop winter 2-row barleys hardy enough for the upper midwest. Winter barleys are planted in the fall, go dormant in the winter, and then have a head start once spring comes. They typically have higher yields, and protect soil by keeping fields covered through the winter. He’s working with a colleague at the U of M to simultaneously develop a shorter season soybean crop to attain a double crop system.

There are multiple long-term benefits of a winter barley double crop system that can survive Minnesota winters:

  • Environmental benefits to water and soil systems by holding soil year round.
  • Support the local beer industry and return barley production closer to Rahr, an enormous processor, to reduce transportation emissions and costs.
  • Increase farmer profitability by offering two crops from one growing year.

Modern Breeding Methodology

The U of M uses Genomic Selection to develop new varieties and improve breeding methods. Genomic Selection entails the use of statistical models and genetic testing to predict the traits of a cross. Prediction is a dramatic improvement in efficiency, because it skips the time and money necessary to grow out a new cross, malt it, and brew it. Genomic Selection is a huge step forward in reducing plant breeding iteration time.

Barley has only 7 pairs of chromosomes, but its genome is twice the size of our genome overall. The plant has 32,000 different genes, with hundreds of genes controlling a trait breeders focus on, like yield. This makes it very difficult to try to directly relate specific genes to specific traits. The work is made harder by the need for breeders to improve multiple different traits simultaneously. A cross may show improved yield, but get rejected because of decreased extract.

So instead of focusing on specific genes, they focus on areas of DNA. Within the genome of a specific barley variety they are using in a cross, there are unique sites that act as markers. These markers represent a nearby area of genes, which acts as a positive identifier of a desired trait. The presence or absence of this marker in new varieties provides information on the traits associated with it.

Breeders build a statistical model by identifying 400-500 markers within 300 varieties of barley related to their current breeding program. This model can then look at a newly genotyped (sequenced) variety to determine if, based on the markers, it’s likely to exhibit the traits they are breeding for. These models increase breeding program efficiency dramatically, because genotyping a new cross is a fast and cheap process.

One challenge with genomic selection is that models don’t work well with varieties they weren’t trained to work with. For example, the U of M models are not accurate for the varieties being developed at North Dakota State University in Fargo. This means there isn’t one model to rule them all, and specific models for specific breeding programs need to be developed.


Finally we arrive at flavor. The barley flavor research project led by Pat Hayes at OSU, which Kevin is collaborating on, is using 2 varieties known for flavor, Golden Promise and Full Pint (developed at OSU), to examine several questions around flavor:

  • Does variety impact flavor?
  • Is there a genetic basis to flavor?
  • Does geographical growing location impact flavor?

They started by growing out 200 offspring from a cross between Golden Promise and Full Pint, selected 34 promising progeny, and grew these out in addition to the parents and a control. The varieties had genetic markers, which were correlated to flavors after running small scale malting, brewing, and tasting tests (only 60 mL of beer!).

The conclusion was positive for all 3 questions: different varieties have different tastes, genes impact flavor, and the growing environment can impact flavor. Golden Promise has fruity and floral flavors, while Full Pint is malty and toasty. The progeny from the cross had a diverse mix of these flavors, plus others.

The research will be officially published in the future, and the project continued to connect specific flavors and genetic markers, map more of the genes impacting flavor, and explore flavors from diverse, non-traditional barleys, like those we’re growing out in the Harlan Society.

Fun poster for Full Pint

OSU developed Full Pint, and has some fun art for the variety.


This initial research is really exciting, as it implies existing breeding methodology could be used to breed for specific, improved flavors. However, there are two important challenges going forward.

The first is the volume of this work. Genomic selection was used to demonstrate the genetic connection to flavor, but the extra work of malting, brewing, and tasting analysis was necessary to demonstrate the connection. The hard work long term is to build a model to predict positive flavors from genetic markers without malting, brewing, and tasting tests. This will require a lot more training data for the model, which requires all the more time and money to run the real world tests. In turn, this would be applicable for one breeding program, but wouldn’t necessarily work upon introducing a new, different variety from a seed bank.

Second, flavor is subjective. The researchers worked with highly trained experts from renowned breweries and maltsters sponsoring the project: Bells, Deschutes, Firestone-Walker, Russian River, New Glarus, Sierra Nevada, Summit, Mecca Grade Estate Malt, and Rahr. These people know beer flavor, but flavor is still a subjective quality. We have measurements for extract, or hop IBU, but there isn’t a metric for “malty” or “fruity”. What’s good flavor? Bad flavor? This is a whole other area methods and metrics will need to be developed, and where there won’t always be agreement.

The Future

Despite these challenges, this initial research into breeding for flavor is incredibly promising. Flavor is based in a variety’s genetics, and breeders are working towards developing more efficient methods to screen for it before running expensive malting, brewing, and tasting tests. Perhaps one day we’ll be at the point of efficiently screening thousands of barley varieties and crosses for novel flavors, allowing this essential trait currently left out from the breeding process to make its way into our beer.

Breeders need sustained support to pursue flavor research, and the more brewers advocating for this type of work, the better. The Brewers Association has done a great job of providing grants for researchers exploring flavor, and I hope other funding sources follow suit in the future. This type of work takes time, even as technology improves breeding iteration capabilities.

Meeting with Kevin and talking about his work reminded me how important it is to support scientific enquiry. Work done by scientists at public institutions has very real benefits for farmers, processors, beer drinkers, and the environment. Supporting it, even when it can take a while to come to fruition, just makes sense.

As we wait for breeders to develop their flavor-driven breeding models, we’ll continue to explore the existing flavors waiting patiently in seed banks through our work with The Harlan Society. Landrace varieties weren’t bred scientifically for flavor, but taste had to have been a factor in many historical selections. If communities continuously selected a variety, it must have had value to them, and we’d like to taste what that was.

Cheers to flavor, old and new!