Industries / Food, farming and forestry / Agriculture and farming / Seed production
Breeding
Making the varieties a seed company sells, and the pure parent lines that hybrid seed is made from.
6
Milestones achieved, from 1856 to 2013.
2
To watch for: announced, not yet confirmed outside.
2
Open deliverables. Nobody has shipped them.
The road
AchievedWatch forOpen
Achieved
1856
Plants judged by their offspring
Louis de Vilmorin, of the Paris seed firm, published a note in 1856 on creating a new race of beet and on heredity in plants. His method, which breeders came to call the "Principle of Vilmorin", chose parent plants by testing their offspring (a progeny test) rather than by their own looks; with it he selected beets for size, shape and sugar content, by the account of a 2002 review of plant breeding milestones.
Credit
Louis de Vilmorin
1943
Sterile onions make hybrids
In 1925 H. A. Jones found a plant in the onion variety Italian Red that made no working pollen. The trait is carried in the cytoplasm, the part of the cell outside the nucleus, and passes only through the mother; this is called cytoplasmic male sterility (CMS). Jones reported it with S. L. Emsweller in 1936, and in 1943 Jones and A. Clarke showed that a single nuclear gene restores fertility and set out how to use such plants as the seed parent, so hybrid seed could be made without opening each flower by hand. A 2015 Korean paper says CMS has been used for hybrid seed in many crops "since Jones and Clarke (1943) first devised the scheme", and a 1994 Japanese paper says this onion source gave rise to "nearly all the CMS lines" then used by breeders in Japan and the United States.
Credit
H. A. Jones, S. L. Emsweller and A. Clarke
1959
A line that makes haploids
A haploid plant has only one set of chromosomes, and doubling them gives a fully inbred line, called a doubled haploid, in one step instead of several generations of self-pollination. In 1959 E. H. Coe described Stock 6, a corn line whose pollen produced haploid offspring at 1–3%, the highest rate then known. A 2019 review by scientists at CIMMYT (the international maize and wheat center) and the University of Hohenheim says doubled haploid technology, using inducer lines bred from Stock 6, has become "an integral part of many commercial maize breeding programs" and has been adopted in Europe, North America and China; it gives pure lines in one year instead of three to four.
Credit
E. H. Coe (Stock 6); CIMMYT and the University of Hohenheim, for adapting it to tropical programs
1976
Hybrid rice reaches farms
Rice pollinates itself, so hybrid seed needs a female line with no working pollen. In 1970 Yuan Longping's team, in work led by Li Bihu, found male-sterile plants in common wild rice for the first time, and by 1973 they had matched a sterile line, a line that keeps it going (the maintainer) and a line that restores fertility in the hybrid (the restorer); this is the "three-line" system. Yuan wrote for FAO that the first hybrids were released commercially in 1976, yielded about 30% more than conventional varieties, and covered 15.7 million ha, half of China's rice area, in 1994.
Credit
Yuan Longping and Li Bihu
1987
Weather-switched sterile rice
Nongken 58S, a rice line found in 1973, sheds no working pollen under long days and high temperatures but is almost normally fertile under short days and moderate temperatures, so it can multiply itself one season and serve as a female parent the next. From 1987 a national Chinese project used such lines to make "two-line" hybrids. Yuan reported for FAO that no maintainer line is needed, that more than 95% of rice varieties can serve as the male parent against about 5% in the three-line system, and that two-line hybrids yielded 10% more than three-line ones. Two-line hybrids are now "applied widely", but if the temperature in the sensitive period falls below a critical level the female line pollinates itself and "seed production will fail", as Zhang and colleagues wrote in 2023.
LessonA sterility switch set by weather moves the risk from the breeder's lab to the seed field's forecast.
Credit
Shi (1981), who described Nongken 58S; China's national two-line project
2013
Markers carry flood tolerance
Marker-assisted selection uses DNA tests to follow a chosen gene through each cross, so the gene can be moved into a popular variety without losing that variety's other traits. Breeders used it to move the SUB1 flood-tolerance gene into Swarna, which the authors call India's most popular rice variety; the result, Swarna-Sub1, "maintains all the properties of Swarna". In a randomized trial in 128 villages of Orissa (now Odisha), Swarna-Sub1 yielded about 45% more than the usual variety when fields were under water for 10 days, with no yield penalty when they were not flooded.
Credit
Dar, de Janvry, Emerick, Raitzer and Sadoulet (trial, Scientific Reports, 2013); Xu and colleagues (SUB1A gene, Nature, 2006)
Watch for
Hybrid wheat on sale
Wheat pollinates itself, and a 2019 review describes past failures to make hybrid wheat pay, along with some hybrids marketed in a few European countries. Syngenta says its X-Terra hybrid wheat is "Europe's first commercially scalable hybrid wheat". Two varieties, SY Sphynx and SY Xanthis, were registered in France for the 2026 sowing season, with the UK and Germany named as next. The release gives no yield figures and does not say how the hybrid seed is made.
Achieved whenofficial variety trials, or farm results published by someone other than Syngenta, show the hybrids' yield and seed cost against conventional wheat.
Claimed by
Syngenta, February 2026
Hybrid rice that breeds true
Synthetic apomixis makes a hybrid plant form seed that is a genetic copy of itself, so a hybrid could be multiplied from its own seed instead of being cross-pollinated anew each year. A China National Rice Research Institute team reported new apomictic hybrid rice lines in February 2026. A Hainan University team grew ten lines with high rates of clonal seed for three generations at two sites: seven stayed stable, one declined, and two rose and fell with the site, which the authors link to heat at flowering.
Achieved whenan apomictic hybrid keeps a high clonal seed rate and normal seed set over several seasons in field trials run by an independent party, or passes official variety registration.
Claimed by
Wei and colleagues (China National Rice Research Institute) and Wang and colleagues (Hainan University), February 2026
Open
01
Census of sterile cytoplasms
A yearly public count of the cytoplasm types (normal, C, S or T) in the commercial corn hybrids sold in one large market, read from seed with published PCR tests (a lab method that copies a stretch of DNA so it can be seen), with every hybrid, company and seed lot listed.
The briefCredit · unclaimed02
Ploidy-checked haploid sorting
A public test set of corn kernels from haploid-induction crosses, each photographed and then grown and checked for its true chromosome number, and a scored comparison of how often trained human sorters and image models call haploids wrongly.
The briefCredit · unclaimed
Open 01 · the brief
Census of sterile cytoplasms
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A yearly public count of the cytoplasm types (normal, C, S or T) in the commercial corn hybrids sold in one large market, read from seed with published PCR tests (a lab method that copies a stretch of DNA so it can be seen), with every hybrid, company and seed lot listed.
Why it is still open
The 1970 blight (see Growing seed crops) showed the risk of one cytoplasm in most fields. The last US-wide figure found is an industry estimate for seed sold in 1987, of 66.1% normal, 22.1% C and 11.5% S cytoplasm, quoted in a 2024 Romanian paper that also reports breeders' renewed interest in male-sterile cytoplasm and the old worry that maize "may again be of the same cms-S or cms-C type". The closest recent survey typed seven forage hybrids in Iran (Hemmati-Kamerh and colleagues, 2026).
Built with what exists
Seed bought at retail from bags whose license terms allow analysis, or seed donated by companies; a DNA extraction kit; published primer sets for C, S and T cytoplasm; a university lab's PCR machine and gel; a spreadsheet of hybrid names and lot numbers.
First step
Gather 20 hybrids, extract DNA from five seeds of each, and run the C, S and T tests alongside known control lines.
Done looks like
At least 100 hybrids from at least five companies typed, each from two seed lots, with 10% retyped blind by a second person, and the share of each cytoplasm published with the full hybrid list.
Open 02 · the brief
Ploidy-checked haploid sorting
The first to publish is credited on this line, and the deliverable moves from Open to Achieved.
What to build
A public test set of corn kernels from haploid-induction crosses, each photographed and then grown and checked for its true chromosome number, and a scored comparison of how often trained human sorters and image models call haploids wrongly.
Why it is still open
A 2019 review says imaging sorters had been built for the purple-crown marker (R1-nj) but "reliable estimates of misclassification by these methods are currently not available", and that the marker itself gives false positives and false negatives and is blocked in some tropical and flint corn. Altuntaş, Cömert and Kocamaz (2019) reached 94% accuracy with image models, but their labels came from the same visual marker. De La Fuente and colleagues at Iowa State (2017) did confirm ploidy with DNA markers, on seed from six inbred lines, and found that multispectral imaging identified haploids with over 50% accuracy for most of them; no public image set with checked ploidy was found.
Built with what exists
A phone or USB camera over a light box; PyTorch with an off-the-shelf image model; flow cytometry, a lab instrument that measures the DNA content of cells, on seedling leaves to confirm ploidy; induced seed from a doubled-haploid program.
First step
Ask a university or CIMMYT doubled-haploid program for 500 kernels from one induction cross, photograph each one, and sow them in labeled trays.
Done looks like
2,000 kernels from at least four source populations, including tropical germplasm, photographed and ploidy-checked, with false-positive and false-negative rates for trained human sorters and for at least one image model, and the images and labels published.