Garlic Genotypes Prove Unstable Across Ethiopian Environments in Multi-Site Yield Trial

Garlic may be one of the world’s most familiar kitchen staples, but for the scientists who breed it, the crop remains stubbornly unpredictable. A new multi-environment field study conducted across central Ethiopia has revealed just how dramatically the performance of garlic varieties can shift from one location and season to the next, offering both a warning and a roadmap for breeders hoping to lift yields in one of Africa’s most important garlic-producing nations. The research, published in BMC Agriculture, evaluated thirteen garlic genotypes alongside a standard check variety at six distinct environments and found that where a garlic plant grows matters far more than which garlic plant it is.

The numbers behind that conclusion are striking. When the researchers partitioned the total variation in marketable bulb yield, the environment alone accounted for 85.68 percent of the treatment sum of squares. The genotype main effect contributed just 5.03 percent, while the genotype-by-environment interaction, the statistical term describing how varieties respond differently to different conditions, explained 9.30 percent, nearly double the contribution of genetics itself. In practical terms, this means that a farmer’s field, its rainfall, temperature, and soil, exerts an overwhelming influence on how much marketable garlic a crop will produce, and that no single variety can be confidently recommended on the basis of yield trials conducted in one place alone.

The trial was carried out during the main rainy seasons of 2020 and 2021 at four sites in the Oromia and Amhara regions: Chefe Donsa, Debre Zeit, Debre Birhan, and Kulumsa. Chefe Donsa and Debre Zeit hosted trials in both years, while Debre Birhan and Kulumsa were included only in 2021, producing six environments in total. The thirteen genotypes, drawn from collections made by the Debre Zeit Agricultural Research Center across different parts of Ethiopia, were planted in a randomized complete block design with three replications at each site. The recently released variety Chefe served as the standard check. Plots were managed with uniform agronomic practices, including fungicide-treated seed cloves, standardized fertilization with NPS and urea, and disease control against garlic rust.

Across the six environments, marketable bulb yield ranged from 5.31 tonnes per hectare for genotype G-063/06 to 6.68 tonnes per hectare for G-091/06. Yet the rankings were anything but consistent. G-091/06 topped the yield tables at Chefe Donsa in 2020, Debre Birhan in 2021, and Kulumsa in 2021, but slumped to eleventh place at Debre Zeit in 2021. This kind of flip in fortune is the classic signature of genotype-by-environment interaction, and it is precisely the phenomenon that has frustrated Ethiopian garlic breeders, whose previously released varieties have shown inconsistent bulb yield performance when tested across multiple locations.

To untangle the interaction, the team deployed a battery of statistical tools that have become the workhorses of modern plant breeding. The Eberhart and Russell joint regression model estimated each genotype’s linear response to environmental change through a regression coefficient and measured the consistency of that response through deviations from regression. The Additive Main Effects and Multiplicative Interaction, or AMMI, model partitioned the interaction sum of squares into principal component axes, with the first two interaction principal component axes explaining 48.26 percent and 28.68 percent of the interaction variation respectively, a combined 76.94 percent. Finally, GGE biplots, constructed from the first two principal components of genotype plus genotype-by-environment variation, allowed the researchers to visualize which varieties won where, and which test environments best discriminated among the genotypes.

The biplots told a coherent story. Debre Birhan in 2021 and Kulumsa in 2021 emerged as the most favorable and discriminating environments, both characterized by cooler temperatures during the growing season, while the four remaining environments proved less productive. Kulumsa 2021, sitting closest to the average environment coordinate, was judged the most representative and effectively the ideal environment for selecting high-yielding, stable genotypes. Chefe Donsa 2020, by contrast, lay near the origin of the AMMI 2 biplot and contributed little to the interaction, making it the least discriminating site. The GGE analysis grouped the six environments into three mega-environments, with G-091/06 winning in the first, G-025/04 in the second, and Debre Zeit 2020 forming a third on its own.

When the stability indices were tallied, a clear set of winners emerged. The AMMI stability value, which ranks genotypes by their contribution to the interaction, and the genotype selection index, which combines yield rank with stability rank into a single score, both pointed to the same trio: the check variety Chefe, G-129/06, and G-009/06. These three genotypes combined above-average marketable bulb yields of 6.29, 6.08, and 6.26 tonnes per hectare respectively with regression coefficients close to unity and minimal deviation from regression, the statistical hallmarks of wide adaptation. Their low genotype selection index scores of 3, 5, and 8 confirmed their status as the most stable high yielders in the trial, and on the GGE ranking biplot, G-129/06 and G-009/06 sat closest to the ideal genotype.

Not every high yielder earned that endorsement. G-091/06, despite posting the highest overall mean yield, carried a regression coefficient significantly greater than one, indicating below-average stability and a specific adaptation to favorable, cooler environments. Several genotypes, including G-127/06, G-063/06, G-041/04, G-052/02, and G-030/04, showed regression coefficients below one, marking them as suited to unfavorable conditions rather than broad cultivation. At the other end of the spectrum, G-045/06 and G-127/06 recorded the highest genotype selection index values of 22, flagging them as unstable and poorly performing, while G-063/06, G-084/06, and G-045/06 occupied vertex positions in the GGE polygon without winning a single environment, a sign that they were among the poorest performers across most sites.

The interaction extended beyond yield itself. The combined analysis of variance revealed significant genotype-by-environment interaction for six of twelve measured traits, including shaft thickness, leaf width, leaf length, clove length, clove diameter, and yield per plant. Environment was again the dominant force, explaining more than 90 percent of the variation in days to maturity, shaft length, shaft thickness, leaf length, and clove diameter. For clove length, the interaction term accounted for a substantial 23.31 percent of the treatment sum of squares, suggesting that even the architecture of the garlic bulb’s individual cloves responds unpredictably to shifting growing conditions. Traits such as plant height, leaf width, leaf length, clove length, and clove diameter showed no significant differences among genotypes, indicating limited genetic variability for those characters in this material.

The study’s authors conclude that G-009/06 and G-129/06 should advance to variety verification trials for potential release as new, widely adapted Ethiopian garlic varieties, while G-091/06 may find a niche in favorable highland environments such as Debre Birhan and Kulumsa. More broadly, the work reinforces a lesson that resonates across crop science: when genotype-by-environment interaction is significant, breeding programs must evaluate their material across multiple locations and seasons, and tools like AMMI and GGE biplots are indispensable for visualizing those trials. For a crop propagated vegetatively, entirely sterile, and grown from cloves rather than seed, garlic breeding leaves little room for genetic reshuffling, making the careful matching of stable genotypes to their target environments all the more critical for the farmers who depend on this pungent bulb.

Subject of Research: Genotype-by-environment interaction and yield stability of garlic genotypes in Ethiopia

Article Title: Genotype x environment interaction and stability of garlic (Allium sativum L.) genotypes in Ethiopia

Article References: Girma, N., Mohammed, H., Gebretensay, F., Atinafu, G., & Wegayehu, G. (2025). Genotype x environment interaction and stability of garlic (Allium sativum L.) genotypes in Ethiopia. BMC Agriculture, 1(1), Article 12. https://doi.org/10.1186/s44399-025-00015-9

Image Credits: AI Generated

DOI: 10.1186/s44399-025-00015-9

Keywords: garlic, genotype-environment interaction, yield stability, AMMI, GGE biplot, Ethiopia, plant breeding, Allium sativum, multi-environment trials, crop improvement, agronomy, Genotype