Candy giant Mars taps biotech firm for hardier cacao tree
With threats from disease and climate change, producers try to protect chocolate supply
CREVE COEUR, MO.— The world loves chocolate. But keeping pace with global cravings can be a tall order for chocolate producers — a challenge made even tougher as climate change and disease threaten the world’s cocoa supply.
The threat has prompted major industry players like the candy bar giant, Mars — which makes Snickers, M&M’s and Dove chocolate, among other products — to find a solution in biotechnology.
This month, the company announced that it hired Benson Hill Biosystems, a Creve Coeur, Missouri-based biotech firm, to outfit it with computing tools to help develop more resilient cacao trees, which produce the beans used to make chocolate.
“Cacao is a pretty fragile crop, increasingly affected by climate change and disease pressure,” said Howard Shapiro, the chief agricultural officer for Mars. “Forty per cent of the crop is lost each year due to fungal, viral and pest problems. This is a huge problem for manufacturers like Mars.”
“If it’s hotter and it’s more moist, there are vectors for disease that can thrive more in that environment,” said Matt Crisp, Benson Hill’s president and CEO.
“In some cases with crops like this, they are deforesting higher elevations to move the crop into altitudes where ... it’s cooler, and is resemblant of the environment that this plant might have grown in 20 or 30 years ago, when it was really thriving or more stably being produced.” Through the Mars-Benson Hill deal, for an undisclosed price, Benson Hill will equip the company’s cacao experts — who have bred the plant for 20 years — with a software platform that uses data on plant genetics and traits to speed up and streamline the breeding process.
“What our platform is working to do is tap into that natural genetic diversity of the cacao tree,” Crisp said.
“We can use Benson Hill’s CropOS analytics engine to understand how we can more rapidly tap into that genetic diversity and find lines of genetic variance that make that plant less susceptible to these types of conditions.”
Developed by Benson Hill over the past two-and-a-half years, the computational breeding technique uses machine learning and artificial intelligence to couple plants’ ge- nomic information with records of their physical traits, often gleaned through years of observations and field studies.
The platform “basically accesses a lot of the information that these companies have collected over the years,” said Dylan Kesler, a lead data scientist for Benson Hill and a former assistant professor at the University of Missouri.
The approach can shave years off the development process, while also narrowing the plant breeding options that companies like Mars have to weigh for field trials.
“Say instead of taking 10 years, we can take 5 years. Instead of taking 10,000 options, let’s start with the best 500,” Kesler said.
“We simulate those offspring with machine learning to predict the quality of those offspring. And then they can select and say, you know, ‘We want Pairing A and Pairing C, to really give us the best bet.’”