A fast-track approach

Predictive breeding: data to help you see into the future

In order to make quick decisions about breeding, the team uses statistical modeling for forecasting. Their approach, termed “predictive breeding,” uses data from conventional field trials from the years 2016 to 2019. Colleagues in the UK have measured the phenotypic trait yield for several thousand individuals for which molecular marker profiles are also available. “This data is now used as a training set.” Simulations have shown that known relationships between the marker profile and the phenotype are suitable for predicting new genetic combinations, at least for two to three breeding cycles.

Valentin Wimmer is now carrying out the computationally intensive task of comparing the data from the field trials of previous years with the marker profiles derived from the leaf samples to forecast the yield. “We make our decision on which plants to select based on the prediction.” Only progenies that have good prospects of delivering greater yields are crossed. Three colleagues apart from Valentin Wimmer are involved in the entire process: Rowena Haardt (crossing and genotyping), Parastoo Hoseinzadeh (marker data analysis), and Gabriella Everett (population development and phenotyping).

Prediction: more yield after one year

Two cycles have now been completed and the figures are highly promising. According to the statistical model, predictive breeding has increased the yield of the crossed winter barley by some percentage points in one year. By way of comparison: In conventional breeding, a cycle up to the next crossing usually lasts five years and is expected to improve yields by one to two percent per annum.

Litmus test in the field

However, the winter barley from the greenhouse in Thriplow still has to pass a validation experiment. The team has selected plants from the RapidGS project that are to be cultivated further in the field beginning in January 2021 and harvested in the summer of 2021. “We’ll then proceed with conventional trials with that seed in the fall of 2021 and harvest the crop in July 2022. By the time we’ve crossed the plants from January 2020 five more times, we’ll know for the first time how good they really are.”

The team’s goal in the pilot project is to find out how much data is required to enable a breeding decision to be made and how well breeding works with a small plant that is not even homozygous. For now, the researchers are mainly focusing on the yield of feed barley. However, predictive breeding of resistant and tolerant varieties is possible as well, given appropriate data. The team is now establishing an additional training population with the focus on biotic stress traits.

This shows that “conventional field trials will still be the key tool for breeders in the future.” Valentin Wimmer surmises that there will merely be a shift in focus: “The purpose of field trials to date has been to make breeding decisions. In the future, we’ll use them more and more to create training populations.”

Obtaining a finished variety in conventional field trials

Nevertheless, field trials will still be needed to obtain a finished variety. “Predictive breeding enables us to speed up the crossing cycle for population improvement. As a result, we can generate and evaluate new genetic combinations faster. However, multiplying seeds and the creation of homozygous lines will be carried out in accordance with traditional breeding.”

That is also the objective for winter barley in Thriplow: “We ideally intend to obtain a variety for KWS. Or an interesting crossing parent for conventional breeding.” |