UC Davis Awarded $6.5 Million to Develop AI Breeding Tool for Crops

A team of researchers from University of California, Davis, has been awarded a $6.5 million grant to use 3-D modeling, artificial intelligence and crop genetics to develop a tool to improve and accelerate breeding pipelines for legumes and sorghum.

Funding for the project, known as GEMINI*, comes from the Bill & Melinda Gates Foundation.

Research is centered on the common bean, cowpea and sorghum, which are key crops for food and income for smallholder farmers in sub-Saharan Africa. Breeders need to rapidly develop new crop varieties as climate change brings about increased temperatures, droughts and other challenges. These new varieties will be grown under conditions not currently experienced in those areas, thus driving the need for novel, advanced prediction frameworks to power public breeding programs.

Robust data, on-the-ground observations and models will be key in this effort. The GEMINI tool will incorporate that information to predict crop yield, nutritional quality and desirable traits needed for climate adaptability. It will also map crop suitability zones, including areas for potential new varieties. 

Collaborative Partnerships

A key part of the research involves working with breeding programs based in Nigeria, Senegal, Tanzania and Uganda, said Christine Diepenbrock, an assistant professor of plant sciences.

Those on-site breeders will select the traits and plant material to create new crop varieties and that information will be incorporated into the GEMINI tool. Traits that increase yield and nutritional quality will be critical to help ensure an adequate and nutritious food supply, she said.

Diepenbrock, Department of Biological and Agricultural Engineering Assistant Professor Mason Earles and Department of Plant Sciences Assistant Professor Brian Bailey are leading the project. Diepenbrock will focus on breeding and genetics, Earles on artificial intelligence-enabled sensing and Bailey on 3-D crop modeling.

The UC Davis team will conduct field evaluations of how sensors may be incorporated into a low-cost, scalable system that can be deployed in breeding programs for small-scale producers, Bailey said. And climate projections will be incorporated to reflect changing temperatures, water availability and other factors.

Artificial Intelligence Key to Work

Three-dimensional models will not only provide simulated data to help the AI-based models “learn” but also help to predict and evaluate traits that will be advantageous under different scenarios.

“We have to be able to measure some of the things we want to model,” Earles said. “AI can really help accelerate and improve these measurements.” 

Researchers and breeders from several institutions are also participating in the GEMINI work including: International Institute of Tropical Agriculture, Alliance of Bioversity International and the International Center for Tropical Agriculture, Regional Study Centre for Resistance and Adaptation to Drought, UC Riverside and the USDA Agricultural Research Service.

*GEMINI stands for GxExM Innovation in Intelligence for climate adaptation, where “G” stands for genotype, “x” represents by, “E” for environment and “M” for management.

Media Resources