Quick look
Models built on large-scale data collection throughout a growing season can help crop breeders target the most durable genetic effects and find the fleeting ones difficult to spot.
AMES, Iowa - Though it's largely a hereditary trait, siblings often grow to be different heights. But even if they end up topping out at the same stature, they may take different paths to get there. As they progress through childhood, growth spurts can come at varying ages with varying intensities.
It's essentially the same for crops, though breeders typically pay far more attention to the season's-end destination than the journey. That could change as technology such as drones, robots and sensors make large-scale data collection throughout a growing season feasible, giving crop scientists the capability to discern the timing and duration of genetic effects, said Jianming Yu, a professor of agronomy at Iowa State University.
"With the power of these modern tools, it's time to leverage high-throughput phenotyping. Otherwise, if we just see the final result, we'll never determine what's transient and what's persistent," said Yu, the Pioneer Hi-Bred Distinguished Chair in Maize Breeding and director of the Raymond F. Baker Center for Plant Breeding.
Sorting out which genes have a fleeting impact and which stick around could benefit efforts to improve crops in multiple ways, Yu said. For most purposes, DNA regions that cause persistent effects are higher-value targets with greater stability and potency. But breeders could also harness transient genes - which are otherwise difficult to detect - to enhance seasonal adaptive qualities in crops, such as cold tolerance or drought resistance.
Growth curves uncover more
A research team led by Yu demonstrated the possibilities in a recent paper published in the Journal of Experimental Botany, which described a three-year study of sorghum at an ISU research farm near Boone. Researchers planted 544 variations of sorghum from two sets of populations and flew an automated drone over the plots from three to six times during each growing season.
Using the hundreds of overhead images captured during the drone flights, researchers calculated the heights of individual plots at each stage of measurement. With that data, they modeled growth trajectories for each variant and then analyzed those curves against their corresponding genotypes to look for DNA regions - also known as quantitative trait loci, or QTLs - that are linked with height. The process is called functional mapping.
Researchers confirmed the effects of four persistent QTLs previously known to influence sorghum height as well as the onset and strength of their impact. They also identified several transient QTLs, which typically appeared during a single season and had a temporary effect that would've been missed by genetic studies based on harvest-time heights alone.
"If you're looking with this more comprehensive analytical framework - functional mapping with extensive data gathered by high-throughput phenotyping - you're going to find a lot more QTLs," said the study's first author, Boris Alladassi, a former ISU graduate student with Yu, now a postdoctoral researcher at the University of Illinois Urbana-Champaign.
Better models coming
Alladassi said the timings and durations of persistent and transient QTLs he diagrammed in the study are likely to apply in other contexts.
"We suspect it is probably a general feature of many complex traits. But it was just much easier for us to work on sorghum plant height to illustrate the case," he said.
With data collection expanding rapidly, the framework the researchers used in the study will soon be even more powerful, Yu said. The technology is improving in quality even as costs fall, which will lead to models based on far more measurement stages. Some researchers are even running daily drone flights over field trial plots, Yu said.
"That's what gets me so excited. We're ready for this," he said.
Beyond the advantages gained by growing crops enhanced by better breeding, farmers also stand to benefit from models built on ongoing data collection, Yu said. It could lead to more dynamic management on farms.
"Farmers will have the ability to make earlier predictions that support decision-making because they have growth curves built with data from start to end," he said.
