Weitz Group @ Georgia Tech Theoretical Ecology and Quantitative Biology



Automated phenotyping analysis of rice root system architecture (in Plant Physiology)

Posted by jsweitz
Teqing image

Root systems are vital to plant survival, through their function in nutrient and water acquisition as well as plant anchorage in the soil. However, knowledge regarding the genes regulating RSA is severely limited, particularly for crop species. This limited knowledge is due to three issues: the difficulty in non-destructively viewing belowground structures, the lack of tools to automatically and accurately phenotype the complex root systems of many crop species, and the need to couple large sets of phenotypic data with a statistical method to determine which traits can best be utilized to distinguish varieties, and hence are candidates for mapping and cloning genes. In this manuscript, we help address these issues by developing an integrated suite of methods to non-destructively image, phenotype, and analyze root systems. Our manuscript presents the following principal results:

1. We present a novel platform to non-destructively observe RSA from any angle over several days and weeks.
2. With our platform, we identify novel traits that were previously difficult to discern, but are likely important for resource uptake.
3. We develop an analysis pipeline to automatically and comprehensively phenotype 16 traits in root systems of 12 rice genotypes – in total, 2297 images from 118 individuals.
4. We combine our automated phenotyping with a supervised learning method known as support vector machine (SVM) analysis to identify which traits contribute most significantly in differentiating two varieties.

This manuscript represents joint work with the Philip Benfey lab and with John Harer.