Feeding future generations in a shifting climate
Global agriculture faces increasing pressure from population growth, climate variability, and shifting production conditions. Yet staple crops like legumes and sorghum, which are critical for food and nutritional security in many regions, have not seen the same productivity gains as crops like maize.
GEMINI integrates expertise across disciplines to rethink how crop performance is studied and predicted. By linking environmental variation directly with plant biology, the initiative is building a data-driven, predictive breeding framework that can forecast how crops will perform even in environments that have never been tested in the field.
Operational Framework
GEMINI has developed from a conceptual research effort into an operational, globally deployed platform. Rather than treating data collection and modeling as separate steps, the system connects four core components in a continuous feedback loop:
| Field-Based Imaging | AI Trait Extraction | 3D Biophysical Crop Modeling | Genomics |
| High-throughput sensing systems capture detailed phenotypic data directly in the field. | AI and computer vision models identify and measure key plant traits straight from field imagery. | Digital models simulate plant growth and physiology and generate synthetic data to train and refine the AI models. | Genetic tools link observed and simulated traits back to underlying genetic variation. |
By feeding real-world measurements into 3D simulations to improve AI training, this loop drives progressively more accurate genetic predictions across complex genotype × environment interactions.
Global Field Network
GEMINI operates a multi-environment research network spanning the United States and Africa, with strong partnerships in countries including Nigeria, Senegal, Kenya, and Uganda.
Across these sites, field trials evaluate shared genetic material in grain legumes (common bean, tepary bean, and cowpea) and sorghum. Testing identical genotypes across diverse climates enables researchers to map exactly how plants respond to environmental gradients such as temperature, photoperiod, and water availability.
Prediction in New Environments
A central goal of GEMINI is to predict how crops will perform in previously untested genotype × environment combinations. Specifically, the framework is designed to forecast the success of:
- New crop varieties in familiar climates
- Established varieties in new climates
- Entirely new varieties in new environments
Achieving reliable prediction across all three scenarios will transform plant breeding, allowing for the selection and deployment of resilient varieties far more efficiently as climate conditions shift.
Building on Phase I’s technology development, GEMINI II focuses on understanding how genotype, environment, and management interact to shape crop performance, then translating those insights into practical tools for breeding programs. Working with regional breeding programs and trialing networks across Africa and the United States, the team is refining crop models and evaluating predictions for key traits like flowering time and grain yield in cowpea and sorghum. Phase II centers on two priorities:
- Directly Guiding Breeding Strategy: Integrating environmental data, genomic predictions, and image-derived traits to help breeders select promising parents and hybrids, optimize field trial placement, and match candidate varieties to target environments.
- Building Accessible Workflows: Packaging the GEMINI toolkit into repeatable data-processing pipelines, complete with intuitive visualizations and connections to existing breeding data systems, so teams can easily interpret environmental variation, trial coverage, and predicted performance.
Impact and Vision
By combining sensing technologies, biophysical modeling, and genomics, GEMINI is developing a new generation of predictive breeding tools. These tools reduce reliance on manual phenotyping, deepen our understanding of plant-environment interactions, and accelerate the delivery of productive, nutritious, climate-resilient crops, allowing breeders and farmers to make informed decisions about what to grow, where, and under what conditions.