Can a Vietnamese maize program reach a better hybrid in fewer years, and show at each step that it did?
Vietnam's average maize yield rose by 0.13 t/ha a year from 1991 to 2010 and by 0.065 t/ha a year from 2011 to 2024 (FAO data). Precision breeding, as used here, joins four tools in one program: markers for the genes behind kernel type and eating quality, doubled haploids for fast inbred lines, genomic prediction to rank lines and crosses before field testing, and image-based phenotyping with multi-trait indices to decide what goes forward. Each section below gives the published work a tool rests on so far and the step still to be built for food maize.
- Germplasm and crossesLocal and exotic specialty maize
- Doubled-haploid linesInbred lines in one year
- Marker setsh2, su1, se1, wx and kernel color
- Genomic predictionLines and crosses ranked before field tests
- Field trialsMulti-location tests; image traits to be added
- Selection decisionMGIDI and combining ability
Markers for kernel quality
SSR markers linked to sh2 and su1 identify the sweetness genes in sweet corn lines, so parents can be chosen by marker rather than by test cross alone.
Kernel type in sweet and waxy maize is set by a few endosperm genes (sh2, su1, se1 and wx), and kernel color by others. The next step is one marker set for all of them, used on every line before it enters a cross.
2023 · Vietnam Journal of Agricultural Sciences · In Vietnamese with English abstract
SSR markers linked to the sweetness genes sh2 and su1
Markers linked to the sh2 and su1 genes were identified in sweet corn lines, for use in selecting parents by marker rather than by test cross alone.
Doubled haploids
Doubled haploids give homozygous inbred lines in one year, against three to four years of recurrent selfing with off-season nurseries.
A VNUA team tested the natural haploid inducer UH400 in 2014–2015 on yellow dent maize, with Nguyen Trung Duc as co-author. Haploids were induced, but few plants survived chromosome doubling. Raising that survival and inducing haploids in sweet and waxy germplasm are the steps still ahead. Review: Chaikam et al. 2019 →
2016 · Vietnam Journal of Agricultural Sciences · In Vietnamese with English abstract
Haploid induction with UH400 in northern Vietnam
Crossed onto 20 yellow dent hybrids in autumn–winter 2014 at Gia Lam, UH400 induced haploids at a mean rate of 7.05%. After colchicine doubling, survival was low (14.4%) and 13 doubled-haploid lines were obtained; the SSR marker phi065 confirmed haploid seeds in five of six crosses.
Genomic prediction
Genomic prediction ranks lines and crosses from their markers before they are field-tested; a dosage-aware pipeline for hexaploid sweetpotato has been written and tested on simulated data.
For food maize, the training population would come from the VNUA specialty inbred lines and their testcrosses, which already have field records from line × tester and testcross trials. The sweetpotato pipeline code is open. Software →
2026 · Ecological Genetics and Genomics
An order for adopting AI and genomic selection in a low-data crop
Genomic selection and genome editing remain rare in hexaploid, clonal sweetpotato; the review proposes staged adoption, starting from multivariate indices such as MGIDI and MTSI, which rank genotypes without genomic data.
2024 · Ecological Genetics and Genomics
Testcross data on tropical sweet corn parents
Clustering sorted 42 inbred lines into four groups, and MGIDI at 20% selection intensity chose eight; in 2022 testcrosses, N39 showed high general combining ability for marketable yield and N28 for soluble solids.
Phenotyping and selection decisions
Image traits predicted rice biomass and nitrogen use efficiency with R² of 0.95 to 0.98, and a multi-trait index chose VNUA191, later recognized for circulation.
MGIDI and combining-ability estimates turn field, image and, later, genomic data into one ranking. The breeding digital twin proposed in a short review adds a prospective audit: each decision is recorded with the prediction behind it and checked against later results.
2023 · Vietnam Journal of Science and Technology · In Vietnamese with English abstract
MGIDI selects two purple waxy corn hybrids, including VNUA191
Among 25 crosses tested in spring and winter 2021 in Hanoi, the index selected THL01 (VNUA261) and THL02 (VNUA191); VNUA191 gave 12.2–13.6 t/ha of fresh ears, with thin pericarp and high anthocyanin, and was later recognized for circulation.
2025 · Plant Stress
Image-based prediction of nitrogen use efficiency in 300 rice genotypes
Plants grown under sufficient and deficient nitrogen were imaged three times with RGB, IR and NIR sensors. Image-based NUEb agreed closely with the destructive measurement (R² 0.98), and three donors were selected: IC463705, Suweon and Cauvery.
2026 · Ecological Genetics and Genomics
A proposed definition of a breeding digital twin
A short review that frames predictive plant improvement as a breeding digital twin with five elements (a breeding referent, a state vector, an action space, a utility function and a prospective audit), using sweet corn as the worked example.
Why now
The yearly gain in Vietnam’s average maize yield has halved since 2010 (FAO data, on the home page).
National policy now names these tools. Decision 808/QĐ-TTg of 6 May 2026 gives the Ministry of Agriculture and Environment a 2026–2030 task to master genetic, gene, cell and gene-editing technologies, precision breeding and genomic selection, to create crop, livestock and aquaculture varieties with high yield, good quality, disease resistance and adaptation to climate change. The plan for biotechnology in agriculture and the environment approved by Decision 1466/QĐ-TTg of 31 July 2026 expects 10–15 new-generation varieties created by cell technology, gene editing and precision breeding over 2026–2035. Both decisions build on Decision 21/2026/QĐ-TTg, which lists advanced biotechnology among the strategic technologies (Annex I, item 4) and new-generation crop, livestock and aquaculture varieties among the strategic technology products (Annex II, group 1, item 16).
Markers, doubled haploids from in vivo haploid induction, and genomic prediction add no foreign genes, so hybrids bred with them follow the existing route of variety testing and recognition.
How progress will be counted
Each step will be reported with a measure that can be checked against trial records:
- doubled-haploid lines produced per year, and the share of haploid plants that survive chromosome doubling;
- the number of lines and testcrosses with both genotype and field data, which form the training population;
- prediction accuracy on lines and seasons left out of model training;
- years from the first cross to entry into official variety testing;
- the share of selected lines and hybrids that keep their rank in later trials.
Working together
The aim is to join the four tools in one breeding pipeline for food maize at VNUA. Collaboration is welcome with seed companies on testing and seed production of new hybrids, with genotyping service providers, and with groups that run doubled-haploid or genomic-prediction programs in tropical maize. Email ntduc@vnua.edu.vn.