Smart technology transforms maize breeding

Harvest is approaching across most of the maize trial fields, which span 65 mu (about 4.33 hectares), at Xiangyang Chia Tai Seeds Co., Ltd.'s research facility in Ledong Li Autonomous County, south China's Hainan Province.

The site is divided into more than 16,000 small test plots of different sizes, each planted with distinct breeding materials or varieties.

Over the past winter, Wu Kun, chief biological breeding expert at Xiangyang Chia Tai Seeds Co., Ltd., and his team worked on-site to purify and evaluate more than 11,000 maize parent lines used for hybrid breeding.

They also produced over 5,000 hybrid combinations to prepare seed stock for large-scale field trials across the Huang-Huai-Hai region — China's primary summer maize production region — scheduled for early June.

Wu demonstrated how scanning a plot's barcode with his mobile phone can pull up data through a smart field terminal developed by Biobin Data Sciences Co., Ltd. in Changsha, central China's Hunan Province.

The system records key traits such as plant height, disease resistance and ear development. Once collected, the data is uploaded directly to a digital breeding platform.

In the past, all of this information was recorded by hand on paper or stored in scattered spreadsheets. Today, not only is data entry far more efficient, but the system also links all records into a data chain, making it possible to evaluate breeding materials in a more scientific and evidence-based way.

"Traditional breeding has always relied heavily on the breeder's experience and intuition. Now, data-driven models are helping to make up for the limitations of subjective judgment," Wu explained.

In recent years, Xiangyang Chia Tai Seeds Co., Ltd. has partnered with Biobin Data Sciences Co., Ltd. to build a digital breeding platform. The platform enables the intelligent storage, recording, and processing of germplasm resources and breeding data throughout the entire breeding process.

To develop a single high-quality maize seed, breeders must first sift through vast germplasm resources with highly diverse genetic backgrounds to identify suitable male and female parent lines. They are then used to generate tens of thousands of hybrid combinations for field planting and evaluation.

After harvest, breeders select the top 10 percent based on field performance. The selected materials are then crossed again with chosen parent lines from other varieties or self-pollinated, followed by another round of screening.

This process is repeated cycle after cycle until a hybrid combination emerges that performs consistently well in yield, disease resistance and other key traits.

A data analytics expert from a seed company said that intelligent breeding platforms have helped genome-wide selection technology dramatically improve the efficiency of maize and rice breeding. By modeling and validating across multiple ecological zones, this approach can cut field trial scale by 50 percent while still retaining nearly 80 percent of superior lines, improving overall breeding efficiency by nearly 60 percent.

Xiangyang Chia Tai Seeds Co., Ltd. began developing and refining a maize breeding model incorporating dozens of parameters several years ago.

In June 2025, the company conducted field trials on more than 4,000 hybrid combinations across three representative test sites in the Huang-Huai-Hai region. Wu said all of these combinations were selected by breeders, with more than 10 percent having already been filtered using predictive models.