Mutation breeding for quantitative traits in the era of genomic selection
This study demonstrates that while genomic selection cannot directly predict private mutations in individual mutant lines, it effectively captures induced polygenic variation once these mutations are shared among relatives through crossing, thereby offering a viable strategy to integrate mutation breeding into pre-breeding programs.
Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer
Farmers and scientists have long relied on two distinct paths to improve crops. The first is selective breeding, a centuries-old practice of choosing the best plants from a field and crossing them to combine their desirable traits. This method works by shuffling existing genetic cards that are already present in the population. The second path is mutation breeding, a more modern technique that uses chemicals or radiation to deliberately scramble the DNA of seeds, hoping to create entirely new genetic cards that nature never produced on its own. While selective breeding has fed the world for generations, mutation breeding has often been a separate, slower process, relying on researchers to manually hunt for rare, beneficial changes among thousands of damaged plants. The question facing modern agriculture is whether these two worlds can finally merge: can the powerful computer models used to predict the success of selective breeding also help breeders harness the chaotic potential of induced mutations?
A team of researchers at Scotland's Rural College set out to answer this by testing a method called genomic selection on two groups of advanced mutant plants: 190 lines of wheat and 74 lines of tomato. Genomic selection is a tool that uses a plant's genetic code to predict its future performance, such as how much grain it will produce or how sweet its fruit will be, without waiting for the plant to fully grow. The researchers wanted to see if this tool could work on mutant lines that had never been crossed with each other, and if it could help breeders combine multiple small, beneficial mutations into a single superior crop. They also ran computer simulations to see how these mutant lines would behave if they were crossed with wild plants over several generations, testing whether the computer models could successfully track and accumulate these new mutations.
The results from the real-world plant panels revealed a surprising limitation. When the researchers tried to use genomic selection to predict the value of the mutant wheat and tomato lines based solely on the mutations each line carried, the model failed. The computer could not "see" the unique, induced mutations because they were private to each individual line; no two plants shared the same new genetic change. Instead, the model's predictions came entirely from the background genetic similarities the plants shared with their original parent varieties. In the wheat group, for instance, the predictive signal came from large chunks of chromosomes that were still mixed from the parent plant, not from the new mutations. In the tomato group, where the mutations were even more unique, the model found no predictive power at all in the private mutations. This finding suggests that simply gathering a collection of independent mutant lines is not enough to use these advanced prediction tools; the mutations must first be shared among relatives through crossing.
However, the story changed when the researchers looked at what happens after those mutants are crossed. In their computer simulations, they created populations where mutant lines were crossed with wild plants, allowing the new mutations to mix and spread through the offspring. In these scenarios, genomic selection proved to be a powerful engine for improvement, but only under specific conditions. When the mutations were numerous and included a balanced mix of helpful and harmful changes, the computer-guided selection outperformed traditional methods by a wide margin. After just two cycles of selection, the genomic approach produced plants with a genetic value of 2.84, compared to just 0.45 for the traditional method, which focuses on finding and keeping only the single largest beneficial mutation. The computer model succeeded because it could simultaneously select for many small, beneficial changes while filtering out the harmful ones, a task too complex for manual selection.
The simulations also showed that the value of this approach depends heavily on the genetic diversity of the starting population. In scenarios where the natural genetic variation was low, introducing mutant founders boosted the genetic gain by nearly seven times over five cycles of selection. In these cases, the new mutations provided the only source of improvement, and the computer model was able to capture almost all of that gain. Conversely, when the natural variation was already high, the added benefit of the mutations was minimal, increasing the gain by only about five percent. This indicates that mutation breeding is most valuable when a crop has run out of natural diversity to work with, and the new mutations provide a fresh reservoir of traits.
Ultimately, the study clarifies a critical path forward for crop improvement. It demonstrates that genomic selection cannot predict the value of a mutant line in isolation because the unique mutations are too rare to be learned from other lines. But once those mutants are crossed and their new genetic changes begin to segregate among relatives, the computer models can effectively capture and accumulate those changes. This offers a practical route to integrate mutation breeding into modern pre-breeding programs: use mutagenesis to create new diversity, cross the mutants to share the changes, and then use genomic selection to rapidly stack the beneficial traits. This approach could help breeders develop crops that are more resilient to climate change and capable of feeding a growing global population, turning the chaotic potential of random mutations into a predictable and powerful tool for agriculture.
Drowning in papers in your field?
Get daily digests of the most novel papers matching your research keywords — with technical summaries, in your language.