Looking Inside Grass Flowers

Grasses are some of the most important agriculture crops on the planet. Wheat, maize, rice, and sorghum have been cultivated for thousands of years and provide essential food sources for humans and livestock. Humans primarily eat the seeds of grasses, which in sorghum are grown on branched, flowering structures called panicles. Panicles exhibit an immense amount of variation in size and structure, which can reflect differences in yield, nutritional quality, and disease resistance. Until recently, scientists didn’t fully understand this variation, partly because “sometimes the panicle is so dense, you can’t easily measure inside it,” says Mon-Ray Shao, a postdoctoral researcher at the Donald Danforth Plant Science Center.

Shao, three other scientists at the Danforth Center, and two collaborators at Washington University found a way to look inside grass panicles using 3D imaging technology. Their work, which was published in and on the cover of New Phytologist in May, demonstrates how to measure a wide array of sorghum panicle traits, some of which had never been quantified before. “Everything we knew about sorghum panicles was based on the tools available at the time,” says Christopher Topp, a principal investigator at the Danforth Center and corresponding author of the study. “Now we have tools that let us explore new shapes and measure traits that were previously inaccessible.”

Out with the old, in with the new

Scientists typically use rulers, scales, calipers, and 2D images to measure plant traits. This labor-intensive approach influences which traits can be measured. For example, researchers can measure seed length or area using 2D images, but not seed volume. Because it’s difficult to comprehensively measure plant traits manually, scientists might overlook biologically or agriculturally important traits. Manually measuring plants can also be subjective because the way traits are measured often varies from person to person.

Over the past five years, 3D imaging techniques, such as X-ray computed tomography (X-ray CT), have made it possible to measure plant traits in a more comprehensive and objective way. This approach requires bringing experts together from various scientific fields. “This is the most interdisciplinary study I’ve done,” says Shao, a co-lead author of the study. “It was great to work with people who have so many different perspectives.”

Shao and Topp have expertise using X-ray imaging techniques to explore plant morphology. Mao Li, a new principal investigator at the Danforth Center and co-lead author of the study, is an applied mathematician who develops quantitative methods to measure plant traits. Tao Ju, a professor at Washington University, and a PhD student in his lab, Dan Zeng, are computer scientists that establish computational methods for analyzing plant structures. Finally, Elizabeth “Toby” Kellogg, Robert E. King Distinguished Investigator at the Danforth Center, is a world expert on grasses who was recently elected to the National Academy of Sciences for her scientific achievements. Given her extensive knowledge of grasses, Kellogg helps interpret which traits are key for generating new biological and agricultural insights.

An exploration inside grasses

In this study, the scientists evaluated the diversity of panicle and seed traits of five types of sorghum that each have a unique genetic background. Then, they asked whether the patterns they observed in panicle traits reflected underlying genetic differences. To do this, they collected about 90 sorghum panicles from plants spanning the five major sorghum types and 55 cultivars. They placed each panicle on a spinning platform and created a 3D reconstruction of it using X-ray CT. As opposed to the labor-intensive process of manually measuring traits, “You can computationally dissect plants,” says Mao Li.

The researchers measured 77 panicle and seed traits for each plant. Some of these traits were internal features, such as branch length, which were previously impossible to quantify without destroying the panicle. The researchers also quantified distributional traits that are particularly difficult to measure by hand, like how seeds are arranged along the flowering stalk. Then, they created a computational pipeline to determine which traits differed across the genetic types of sorghum. Using this approach, they found that some traits, such as seed shape, were key features that distinguished among the five sorghum types. However, panicle architecture overall did not reflect underlying genetic patterns. While this result was somewhat surprising, it demonstrates that “nature doesn’t necessarily work in categories,” says Shao.

Agricultural history impacts today’s biology

While it’s difficult to determine why there is so much diversity in panicle size, shape, and structure, the scientists hypothesized that this variation may reflect historical agricultural practices. Specifically, crop breeders often select for sorghum traits such as increased yield, nutritional quality, and disease resistance, which indirectly influence panicle structure. For instance, more compact panicles tend to have higher yields, whereas more open panicles tend to be more resistant to mold and pests.

The relative importance of these traits often varies depending on which environment sorghum is grown in. In wetter regions, crop breeders may prioritize disease resistance in sorghum plants, whereas in drier regions, they may prioritize yield. Even if plants in a particular region exhibit similar panicle traits, there can be extensive diversity of panicle forms throughout the broad geographic range of sorghum.

In addition to potential regional differences in panicle traits, crop breeders also interbreed sorghum types to create new and beneficial trait combinations. Over time, this process can generate an immense amount of variation in panicle architecture that doesn’t always align with underlying genetic patterns. Although it is unclear which processes drove the vast amount of panicle diversity identified in this study, the researchers have already begun another experiment to map finer-scale connections between sorghum genetics and panicle traits.

A roadmap for future research

By bringing together plant biologists, mathematicians, and computer scientists, the researchers established a framework for others interested in using 3D imaging technology to comprehensively and objectively measure plant traits. Even though X-ray CT is used more often now than it was five years ago, there are still challenges with processing and analyzing the massive amounts of data generated using this approach. The mathematical and statistical approaches developed in this study help address these challenges. “X-ray systems will become normal in the not-so-distant future,” says Topp, “Our goal is to reduce barriers for other groups to use this technology in their own plant research.”