Scientists Map the Genes That Could Finally Make Soybean Thrive in Rwanda’s Highlands

Soybean is one of the world’s most valuable crops, prized for its high protein content in animal feed and its role as a leading source of vegetable oil. Yet in Rwanda, a country where demand for the legume is rising steadily, farmers struggle to obtain yields that exceed a single ton per hectare. The reason is deceptively simple: nearly all of the varieties currently grown were bred for other parts of sub-Saharan Africa and are poorly matched to Rwanda’s unusual combination of near-equatorial day lengths and high-elevation temperatures. A new genome-wide association study, published in Theoretical and Applied Genetics, offers the most comprehensive genetic picture to date of how soybean could be retooled for this challenging environment, identifying hundreds of genomic markers linked to flowering, maturity, plant architecture and seed composition.

The research, led by Doreen Mutoni of the University of Missouri together with colleagues including Felix Fritschi, Trupti Joshi, Kerry Clark, Jason Gillman and Kristin Bilyeu, tackled a problem rooted in soybean’s basic biology. The crop is a facultative short-day plant, domesticated roughly 5,000 years ago in higher-latitude China. Flowering is triggered only when day length drops below a genotype-specific threshold, which means that varieties moved to new latitudes often fail dramatically. Genotypes adapted to long summer days flower prematurely in the tropics, producing spindly plants with little biomass, while low-latitude varieties grown at higher latitudes may never flower at all. Rwanda sits between 1 and 3 degrees south latitude, but its production zones lie at elevations of roughly 1,300 to 1,700 meters, where cooler temperatures complicate the photoperiod equation further.

To dissect the genetics of adaptation, the team drew on the USDA National Plant Germplasm System, obtaining 1,004 genetically diverse soybean accessions spanning maturity groups III through X, with origins in China, Russia, Vietnam, Korea, Japan and the United States. After an initial screening season in 2016 at Bugesera and Rubona, 492 accessions that successfully matured and set seed were advanced into formal trials. Following quality control, 466 accessions were evaluated across six environments in Rwanda during 2017 and 2018, covering the Bugesera, Nyagatare and Rubona locations in randomized complete block designs with two or three replications. The traits measured ranged from days to flowering (R1), beginning pod (R3) and full maturity (R8), to reproductive period lengths, plant height, single-row harvest weight, 100-seed weight and seed protein and oil content determined by near-infrared spectroscopy.

The field results confirmed that genetics and environment interact powerfully in this setting. Genotype, environment and genotype-by-environment effects were highly significant for essentially every trait measured. Strong positive correlations, ranging from 0.89 to 0.99, linked the phenology traits and plant height, while seed protein and oil content were negatively correlated, a well-known trade-off in soybean. Heritability estimates were encouragingly high for many traits: days to flowering reached 0.93 to 0.96, 100-seed weight hovered near 0.95, and protein heritability ranged from 0.86 to 0.90, indicating that breeders can make reliable genetic progress. Notably, accessions in maturity groups IV through VII demonstrated full-season adaptation to Rwanda’s high-elevation environments, a finding that immediately narrows the search space for locally suited germplasm.

The genomic analysis relied on the Illumina SoySNP50K BeadChip, with 34,948 high-quality SNPs retained after filtering for the association mapping. Using the BLINK method within the GAPIT 3 framework, and accounting for population structure with principal components, the researchers applied a Bonferroni significance threshold. The scan yielded 342 quantitative trait nucleotides in total, with 222 unique positions: 30 QTN for days to flowering, 26 for beginning pod, 33 for maturity, 25 and 9 for the two reproductive period windows, 31 for plant height, 30 for 100-seed weight, 29 for harvest weight, 17 for protein and 8 for oil. In a methodologically important twist, the team also ran parallel association analyses using growing degree days, a heat-accumulation metric, rather than calendar days, producing an additional 104 QTN hits across the phenology traits.

Several of the detected signals corresponded to known maturity genes, validating the approach. The E1 locus, the single most influential regulator of soybean flowering, appeared in the flowering and beginning-pod results, with its dominant allele delaying flowering by 3.2 days and adding roughly 20 growing degree days to both the flowering and pod-set stages. E2, E9, Tof9, Tof11 and Tof12 were also recovered, along with Dt1, the gene governing indeterminate stem growth habit, which showed positive effects on maturity, height and harvest weight. Tof12, which encodes the pseudo-response regulator PRR7b, reduced maturity by 1.9 days and accounted for 6.2 percent of phenotypic variation. Interestingly, the J locus, a cornerstone of tropical adaptation elsewhere in Africa and Asia, was absent from the results simply because no accessions carrying that allele were included in the panel, hinting that Rwandan breeding programs may benefit from introducing it.

Among the most striking discoveries were two major pleiotropic regions on chromosome 2 that influenced flowering, pod set, maturity and plant height simultaneously. The SNP ss715581055_T/G alone explained up to 18.9 percent of the variation in days to maturity and 17.4 percent of variation in plant height, with accessions carrying the Williams 82 reference allele flowering about three and a half days earlier, maturing roughly five days earlier and standing about eight centimeters shorter. This region lies near GmAP1d, a MADS-box flowering gene recently implicated in photoperiod response, and near FT2c, a duplicated florigen whose domesticated allele carries a large transposon insertion that suppresses its expression. A third pleiotropic hotspot on chromosome 7 sits roughly 100 kilobases from GmWRKY75, a transcription factor gene whose family members are known to promote flowering. These candidate associations suggest that variation beyond the classical E-gene repertoire shapes adaptation in tropical highland conditions.

The growing degree day analysis carried particular practical significance. In temperate regions, soybean typically requires 2,200 to 2,400 accumulated growing degree days to reach full maturity, but the Rwandan trials recorded significantly lower values, below 2,170 GDD, reflecting the cooler high-elevation climate. Several QTN were detected only in the GDD analyses and not in their calendar-day counterparts, or vice versa, indicating that temperature response and photoperiod response are genetically separable in this panel. The authors argue that adopting the growing degree day metric, already standard for maize in North America and for soybean in northeastern China, would give Rwandan breeders and farmers a more precise language for describing cultivar heat requirements and for matching varieties to the country’s distinct microclimates, especially the cooler Rubona environment.

The study also delivers immediately actionable breeding tools. The authors recommend selecting for the Williams 82 allele at the chromosome 2 SNP ss715581055_T/G in future Rwandan breeding efforts, with the neighboring ss715583777_T/C marker as a secondary option, and suggest that the chromosome 7 and chromosome 17 markers could further improve harvest weight, since accessions carrying those reference alleles produced the heaviest single-row plots despite shorter stature. Seed composition results flagged the well-characterized POWR1 locus on chromosome 20 for oil content, reinforcing its role as a pleiotropic regulator of seed quality and weight. Because the single-row plot design could not measure true grain yield, the harvest weight data serve as a baseline rather than a definitive yield ranking, and the authors stress that well-designed multi-row yield trials are the necessary next step.

Beyond its immediate application to Rwanda, the work carries a broader message about the geography of crop adaptation. Most soybean maturity research has focused on extending the crop poleward into long-day environments, while the genetic logic of tropical high-elevation production, where short days combine with cool temperatures, remains poorly charted. By demonstrating that reference-genome-derived markers can be validated under genuine East African field conditions, and by quantifying how known maturity genes behave when latitude and altitude pull in different directions, the study provides a template for breeding programs across the tropical highlands of Africa and beyond. As demand for plant protein grows and climate variability intensifies, unlocking the underutilized variation in global gene banks may prove one of the most consequential tools available for building resilient food systems.

Subject of Research: Genetic basis of soybean adaptation to high-elevation, low-latitude environments in Rwanda

Article Title: Genome-wide association study of diverse soybean [Glycine max (L.) Merrill] accessions for agronomic and seed composition traits in Rwanda

Article References: Mutoni, D., Fritschi, F. B., Joshi, T., Clark, K., Gillman, J., & Bilyeu, K. (2026). Genome-wide association study of diverse soybean [Glycine max (L.) Merrill] accessions for agronomic and seed composition traits in Rwanda. Theoretical and Applied Genetics, 139(10), Article 281. https://doi.org/10.1007/s00122-026-05383-7

Image Credits: AI Generated

DOI: 10.1007/s00122-026-05383-7

Keywords: soybean, genome-wide association study, Rwanda, photoperiod adaptation, maturity genes, flowering time, growing degree days, seed composition, plant breeding, quantitative trait nucleotides, high-elevation agriculture, Glycine max