Chinese agricultural scientists have launched a generative AI breeding platform for soybeans, which play a central role in China’s agricultural trade.
The Fengshu 2.0 platform marked “a shift in China’s soybean breeding towards data-assisted breeding and precision decision-making”, Xinhua reported recently, contrasting it with traditional methods that relied on experience and a trial-and-error approach.
The artificial intelligence platform, which is expected to optimise breeding plans by making the processing of research data more efficient, is China’s latest effort to improve home-grown soybean varieties as it seeks to reduce its reliance on imports.
The past five years have seen Chinese scientists develop a batch of breakthrough high-yield, high-oil varieties of the seed plant to replace imports, which currently account for over 80 per cent of total consumption, according to official numbers, and are seen by Beijing as a vulnerability amid heightened geopolitical risks.
Yet most new varieties remained a long way from large-scale commercial planting and had yet to meaningfully curb the country’s reliance on imports, researchers said, even as growing demand for meat from the nation’s 1.4 billion people kept pushing up the consumption of soybean meal – a key animal feed.
Since China launched a seed industry revitalisation plan in 2021, the development of high-yield, high-oil soybean varieties had transitioned from “quantitative accumulation” to “qualitative breakthrough”, said Wang Wenshen, an analyst at bulk commodity consultancy Sublime China Information.
The plan, which aims for self-reliance in seed technology and independent, controllable germplasm resources, pledged to achieve “visible outcomes within five years, and major breakthroughs within a decade”.
Wang said one of the outcomes was “Maiyu 511”, a genetically modified soybean variety approved by agricultural authorities in 2023. He said it boasted an oil content of 23.69 per cent – surpassing the 22 per cent average of imported soybeans – and had achieved yields more than 8 per cent higher than a control group in regional trials.
However, commercial application needed to speed up for meaningful changes to be made, he said, because research and development was still led by government-backed institutions instead of businesses.
“China approves a large number of soybean varieties each year, but very few of them achieve large-scale commercial planting,” Wang said. “And many varieties remain at the research stage, and the technology transfer from research institutions to enterprises is hindered by poor coordination.”
China imports most of the soybeans it uses from Brazil and the United States, but the implementation of current policies, including rehabilitating salt-affected land, approving genetically modified soybeans and promoting healthy dietary transition, could increase its self-sufficiency rate to 74 per cent by 2030 and 80 per cent by 2050, a team led by scientists from China Agricultural University said in a paper published in Nature Food last month.
China’s planting area for high-oil, high-yield soybeans, including genetically modified varieties, reached over 2.53 million hectares last year, an increase of about 65 per cent from 2020, the Beijing-based Economic Daily reported.
However, that was only about a quarter of the total area planted with soybean and remained far from meeting China’s annual demand – which was 115.7 million tonnes (127.5 million short tons) last year – according to data from the China Soybean Industry Association.
Chinese officials have told state media that average US yields are two-thirds higher than those in China.
Wang said most of the breakthrough varieties were concentrated in northeastern China and were not well adapted to cultivation in the south of the country, which imposed constraints on the expansion of production.
Professor Hu Ruifa, from Peking University’s Institute of Advanced Agricultural Sciences, said that if the government truly wanted to unleash the vitality of China’s seed breeding industry, it needed to step back from commercial breeding and make way for enterprises.
Although the government was offering companies incentives to invest in research and development, firms still preferred to buy ready-made varieties directly from government institutions, as the cost of in-house development was far higher than purchasing existing varieties developed by the public sector, he said.
“If the government does not withdraw from commercial breeding, no amount of planning will make a difference,” Hu said. -- SOUTH CHINA MORNING POST
