Machine learning tool helps create drought-resistant cropsSiobhan Treacy | August 26, 2021
Researchers from the University of Illinois at Urbana-Champaign created a series of imaging and machine learning (ML) tools to study plant characteristics. The goal of this research was to help farmers and plant researchers breed or engineer crops that can conserve water without sacrificing yield. This would allow farmers to grow food even in drought conditions.
Plants take in carbon dioxide through stomata, or pores on their leaves. Carbon dioxide drives photosynthesis and contributes to growth. The stomata allow moisture to escape in the form of water vapor. The amount of water vapor and carbon dioxide exchanged depends on the number of stomata, their size and how quickly the stomata open or close in response to the environment. The traditional method for measuring plant traits is slow and laborious and often some characteristics end up not getting measured because it takes up too much time.
The team analyzed how stomata on leaves influence the plant’s water-use efficiency. The study focused on corn, sorghum and grasses of the genus Setaria.
The team's machine learning tool was originally developed to help self-driving cars navigate complex environments. They converted the tool so it could quickly identify and measure thousands of cells and cell features in a leaf sample. They also used sophisticated statistical approaches to identify regions of the genome and lists of genes that likely control variation in patterning of the stomata. Thermal cameras were used in the field and lab experiments to quickly assess temperature to see how water loss was cooling the leaves.
The results revealed key links between changes in microscopic anatomy and the physiological and functional performance of plants. They compared leaf characteristics with the plant’s water use efficiency in the field experiments. They found that the size and shape of stomata in corn appeared to be more important than previously thought. This discovery will inform future efforts to breed and genetically engineer crop plants that use water more efficiently.
This study was published in Plant Physiology.