The United States is consistently among the top coffee-consuming nations in the world, largely importing its favorite beans from South American countries such as Brazil and Colombia.
University of Florida researcher Felipe Ferrao says the Sunshine State could become a global hub for coffee research. That’s why he’s leading studies into the caffeinated beverage and the plants that produce it.
“We have the laboratory expertise to determine the full genetic makeup of different coffee varieties,” Ferrao said. “The University of Florida also has an ongoing artificial intelligence initiative, so combine this with our access to the HiPerGator supercomputer, and we could lead the way in improving coffee production and flavor.”
The first project’s scientific question is a very basic one, Ferrao said. For the past year, scientists have grown about 150 coffee plants in a greenhouse on UF’s Gainesville campus, and the plants are about to be exposed to Florida’s elements.
“It’s a simple ‘yes’ or ‘no’ answer we’re seeking in this first phase,” said Ferrao, who splits his time with blueberry research as part of the UF/IFAS Blueberry Breeding Program team. “Determining what will grow is the first step to identifying the plants most adaptable to Florida’s conditions.”
Researchers obtained the Arabica and Robusta seeds from a Puerto Rican collaborator, Puerto Rico Coffee Roasters, because the territory is one of the few U.S. locations that has found success growing the plants. Now, the small plants are destined for three fields: one near Gainesville, another at the UF/IFAS Indian River Research and Education Center (IRREC) in Fort Pierce, and the final at the UF/IFAS Tropical Research and Education Center (TREC) in Homestead. Planting is planned in the coming weeks, weather permitting.
Coffee drinkers prefer Arabica beans for their superior flavor and aroma, Ferrao explained, but the plants are less productive than Robusta varieties. Arabica plants are also more susceptible to the challenges of climate change in the regions where they have long been produced.
Florida’s climate won’t necessarily make for more favorable growing conditions, though. The North Florida plants will grow in containers under high tunnels, in the hopes that the young plants can be protected from any potential freezing temperatures. Plants in the Central and South Florida locations will be in the ground in open fields.
“As an exploratory analysis, we should see some level of success or failure in each field by the end of this year,” Ferrao said. “Our follow-up questions will be to understand the behavior of the plant in these new conditions; then the flavor of what we can grow here in Florida; and finally, maybe most importantly, whether it can be profitable as a crop.”
The project is funded by the UF/IFAS Research SEEDIT Program, which supports research in emerging agricultural enterprises. Ferrao has collaborators in Gainesville and across the state: Jonathan Crane, a horticultural scientist and associate director of TREC; William Hammond, a plant ecophysiologist in the agronomy department; Patricio Munoz, a horticultural scientist who leads the Blueberry Breeding Program, and the program’s lab manager Werner Collante; Márcio Resende, a horticultural scientist who leads the Sweet Corn Genomics and Breeding lab; Lorenzo Rossi, a horticultural scientist who leads the Plant Root Biology Laboratory at IRREC, and postdoctoral researcher Guilherme Locatelli in the same lab; and Ali Sarkosh, a horticultural scientist who primarily studies stone fruits and grapes.
Simultaneously, though, Ferrao is already looking into the flavor of coffee varieties, although not from plants grown in Florida.
A project called “DeepFlavor: Using Artificial Intelligence to predict and understand flavor preferences” combines traditional consumer sensory panels with deep learning technologies. The work is being incorporated into existing breeding programs, like blueberries, to identify the chemical components that match what consumers indicate they enjoy.
Each booth at the UF Sensory Laboratory is equipped with a computer, where panelists are asked to rate the food or beverage on a scale from 1-9, based on individual preference, with directives regarding features like bitterness, sweetness, texture, etc. In Ferrao’s DeepFlavor project, researchers added webcams to each computer for select tests, starting with a coffee panel held in March. The cameras capture the panelists’ expressions in a quick series of photos in the moments after the product is tasted. The images are never seen by a human, but instead go straight to the HiPerGator for analysis, using a model Ferrao trained.
“I don’t want to claim that this is going to be the new way to do sensory panels,” Ferrao said, “but I can see a future that combines the traditional with this new technology.”
Most interesting in the first round of the coffee test is that panelists ranked a Robusta variety highly. Of the nine total and three Robusta options offered to the March panelists, a Robusta finished second overall.
“For years and years, Robusta was considered bad coffee – grows well, high caffeine content, but its wide diversity of flavors was not explored,” Ferrao said. “Well, we can now say that almost 300 consumers on this panel did not agree. Some Robusta growers are focusing on quality, and both specialists and regular consumers are recognizing it. Now, we’re trying to understand the chemical and genetic attributes that make consumers say they enjoy it more.”
This may be important to the coffee-growing research, even though the projects are not directly linked, as Ferrao suspects Robusta varieties may grow most successfully in Florida conditions.
The DeepFlavor project is funded by a UF/IFAS Dean for Research program called “Launching Innovative Faculty Teams in AI,” or LIFT AI. Other scientists involved are Munoz, Resende, microbiologist Raquel Dias, food scientist Charles Sims and horticultural scientist Denise Tieman.
Flavor from brewed coffee beans is complex, Ferrao said, and it’s complicated by the fact that scientists did not consider genetic information in previous breeding efforts. That’s the third project currently underway.
“We know the environment it grows in is super important, how the beans are harvested, how they’re roasted, how they’re brewed,” he said. “But the genetic profile creates the chemicals that determine flavor in the end product.”
Ferrao partnered with scientists at Instituto Capixaba de Pesquisa, Assistência Técnica e Extensão Rural (Incaper) in Brazil – a leader in coffee research – and his lab received leaves and green coffee beans from Arabica and Robusta plants grown in Brazil. In their initial research, recently published in Crop Science Society of America’s journal, “The Plant Genome,” the scientists noted that improvements in Robusta varieties could be accelerated using breeding methods informed by genetic sequencing.
“Coffee production needs to adapt to a changing climate, and knowing the plant’s full genetic profile could hold the keys to that,” he said. “Down the line, we could find the traits that cause the plant to thrive and others that make for a preferred flavor and aroma, and the potential is wide open for breeding.”
As these projects continue, Ferrao predicts each one will inform the next. There could be a Florida-grown coffee tested on the sensory panel, or the development of new genomic tools that can accelerate the selection of favorable traits.
“Coffee is a bit of a social phenomenon in America, and its future is at a critical point,” Ferrao said. “This work is a new frontier.”