Abstract
Organic soybean production requires genotypes capable of maintaining high performance under variable environmental conditions and without synthetic inputs. This study evaluated 22 soybean genotypes grown in 24 environments across different regions of Brazil over five growing seasons (2019/2020 to 2023/2024) to identify materials with superior productivity, adaptability, and stability. The genotype × environment (G×E) interaction, analyzed using mixed models (REML/BLUP), was the only significant source of variation, indicating a strong environmental influence on genotype performance. Selection indices based on genotypic values (HMGV, RPGV, and HMRPGV) were applied to simultaneously assess yield, stability, and adaptability. Genotype BRSMG534 showed the best overall performance, with high productivity, stability, and broad adaptability. Additional promising genotypes included SOYPLENASG7401, BRS6013, LIN16, and IPR115. Conversely, BRS531, BRS573, and BRS391 exhibited low performance in the organic system. The results demonstrate that biometric modeling and BLUP-based indices are effective in recommending genotypes adapted to organic farming, highlighting that only a limited group of materials combines high yield, phenotypic stability, and environmental responsiveness.
Introduction
The introduction of soybean [Glycine max L. (Merr.)] in Brazil has changed the prospects for agribusiness, currently accounting for a significant portion of the national economy (Carvalho et al., 2016). In the 2024/2025 crop season, the planted area reached 47.35 million hectares, resulting in 171.5 million tons of grain and a mean productivity of 3,621 kg ha⁻¹ (CONAB, 2025). The soybean production chain also plays a relevant role in generating direct and indirect jobs, boosting regional and national development (Toloi et al., 2021).
Currently, three main cultivation systems are employed: conventional, based on the use of chemical fertilizers and pesticides; transgenic, which uses genetically modified seeds; and organic, which dispenses with chemical molecules and depends on ecological management, nutrient cycling, and biological control. Recent studies highlight the advantages of integrated systems for soybean sustainability (Silva et al., 2023) and the viability of the organic system as a less environmentally impactful alternative (Reginatto et al., 2025). The growing global interest in healthier foods has increased the demand for organic products, opening new market opportunities for Brazilian soybeans. However, for this segment, it is necessary to identify genotypes appropriate to the organic system, with high competitiveness, tolerance to pests and diseases, lower nutritional requirements, and stable performance in environments with greater variability (Feledyn-Szewczyk and Jończyk, 2016; Borja Reis et al., 2022).
Recommending genotypes for this system requires specific trials, since synthetic fertilizers and chemical pesticides are not used (Pradebon et al., 2023). Organic production depends directly on cultivars adapted to low-input environments, capable of resisting biotic and abiotic stress, exhibiting vigorous initial growth, and demonstrating greater symbiotic efficiency (Venancio et al., 2024). Thus, it becomes fundamental to conduct multi-environment trials to identify genotypes suitable for such conditions. Studies in different crops, such as cowpea, show that biometric models applied to multi-environment trials allow the selection of genotypes that are simultaneously productive, stable, and adaptable (Sousa et al., 2019).
The use of these models has promoted advances in the positioning of soybean genotypes, allowing the identification of agronomically superior materials in different environments (Carvalho et al., 2016; Sousa et al., 2019; Rother et al., 2019). This approach is crucial because the same cultivar can exhibit contrasting performance depending on the growing environment, due to genetic and environmental effects and their interaction, called genotype × environment interaction (G×E). Recent studies indicate that G×E reduces the association between phenotypic and genotypic values and makes it difficult to recommend superior materials (Happ et al., 2021; Abebe et al., 2024). Thus, methods based on genotypic values, such as those that simultaneously integrate productivity, stability, and adaptability, become essential to guide selection in systems with greater environmental variability, such as organic farming (Colombari Filho et al., 2013).
Based on the above, it is hypothesized that adaptability and stability indices are efficient tools for recommending promising soybean genotypes for organic farming systems. The objective of this work was to evaluate the G×E interaction in an organic system, identifying genotypes with greater adaptability, stability, and productive performance in different environments.
Results and discussion
When comparing the overall mean air temperature among the environments, it is noted that Bela Vista do Norte/PY, Maracaju/MS, and Rio Verde/GO presented values above 25 °C (Fig. 1). Diamante D’Oeste/PR, Dourados/MS, Jaboticabal/SP, and Laguna/MS exhibited temperatures below 25 °C for most of the cycle, while São Francisco de Paula/RS and Major Vieira/SC presented values around 17 °C, being the environments with the mildest temperatures in the study. Thus, air temperature did not appear to be a limiting factor for soybean development, since soybeans require values between 20 °C and 30 °C, with an ideal temperature around 25 °C (Alsajri et al., 2022).
Mean daily rainfall values show a distribution between 3 and 4 mm day⁻¹ for most of the soybean cycle, with values exceeding 5 mm day⁻¹ only in October, with December being the month with the least rainfall (Fig. 1). Faxinal dos Guedes/SC, Capinzal/SC, Passo Fundo/RS, Augusto Pestana/RS, and Tiradentes do Sul/RS were among the environments with daily means exceeding 4.5 mm day⁻¹. Maracaju/MS and Jaboticabal/SP had the lowest averages. Soybeans require a total water volume ranging from 450 to 800 mm (Farias et al., 2007), being relatively tolerant to water deficit during the vegetative period, but very sensitive during flowering and grain filling (Sentelhas et al., 2015).
Deviance analysis based on the likelihood ratio test revealed statistically significant differences only for the G×E interaction effect on grain yield (Table 1). The estimated heritability value was 0.07, indicating a high influence of the environment on genotype performance. However, the genotypic correlation across environments (rge) and the coefficient of determination of the G×E interaction (GEIr²) indicated a predominantly simple G×E interaction, with values of 0.77 and 0.71, respectively.
| Variable | Model | logLik | AIC | LRT | Pr(>Chisq) |
|---|---|---|---|---|---|
| GY | G | -445 | 1074 | 2.49 | 1.14 × 10−1 |
| GY | G×E | -529 | 1243 | 171 | 3.69 × 10−39* |
| Parameters | Values | ||||
| 0.65 | |||||
| 0.07 | |||||
| 0.71 | |||||
| 0.80 | |||||
| 0.89 | |||||
| 0.77 | |||||
| 7.29 | |||||
| 12.60 | |||||
| 0.58 | |||||
G: genotype effect; G×E: genotype × environment interaction effect; GY: grain yield (t ha⁻¹); logLik: Log Likelihood; AIC: Akaike Information Criterion; LRT: Likelihood Ratio Test; Pr(>Chisq): probability of the χ2 test; : phenotypic variance; : broad-sense heritability; : coefficient of determination of the G×E interaction; : average genotypic heritability; : accuracy; : correlation coefficient of the G×E interaction; : coefficient of genotypic variation; : coefficient of residual variation; : Proportion between and .
High values for these coefficients suggest a proportional and predictable response from the genotypes (Pupin et al., 2015). In other words, this shows that the performance of a given genotype does not tend to change drastically in the face of the different environments evaluated. The selective accuracy obtained was 0.89, associated with an average genotypic heritability of 0.80. These values indicate high precision in predicting genotypic values and highlight the potential for selecting promising genotypes for the organic system.
The largest GY ranges were observed in the 2022/2023 and 2020/2021 growing seasons (5.62 and 5.14 t ha⁻¹, respectively), with the smallest for the 2019/2020 growing season (3.64 t ha⁻¹). This information can be extracted from the predicted values for grain yield () in Tables 2 and 3. This suggests a greater environmental effect in crops with larger amplitudes. In the 2019/2020 crop season, the best genotypes were BRS284, DF 2353, and BRS523, grown in Jaboticabal/SP, Três Passos/RS, and Diamante D’Oeste/PR, with GY between 3.4 and 5.0 t ha⁻¹ (Table 2). However, the genotypes BRS523 and DF 2353, along with BRS511, also showed the worst performance when grown in General Câmara/RS, which shows that the specific edaphoclimatic conditions of each environment were decisive for the success of organic farming in the 2019/2020 crop season. The genotypes BRS511, IPR122, BRS525, BRS539, and BRS511 evaluated In Augusto Pestana/RS, Três Passos/RS and General Câmara/RS, GY yields ranged between 4.6 and 6.5 t ha⁻¹ in the 2020/2021 crop season, while BRS391 and BRS525 in Entre Rios do Oeste/PR, Salto do Lontra/PR, Major Vieira/SC and Toledo/PR were the worst genotypes.
| 2019/2020 | 2020/2021 | 2020/2021 (continued) | ||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| E | G | g | ge | E | G | g | ge | E | G | g | ge | |||
| E17 | G1 | 0.1 | 1.3 | 5.0 | E2 | G3 | 0.1 | 1.8 | 6.5 | E14 | G2 | -0.3 | 0.3 | 3.3 |
| E17 | G13 | 0.1 | 0.2 | 3.9 | E24 | G20 | 0.1 | 1.9 | 5.8 | E14 | G8 | 0.1 | -0.2 | 3.2 |
| E24 | G1 | 0.1 | 0.7 | 3.8 | E10 | G5 | 0.0 | 0.9 | 5.1 | E24 | G8 | 0.1 | -0.7 | 3.2 |
| E17 | G4 | -0.1 | 0.1 | 3.6 | E10 | G8 | 0.1 | 0.4 | 4.7 | E18 | G21 | 0.2 | -0.4 | 3.2 |
| E6 | G4 | -0.1 | 0.5 | 3.4 | E10 | G3 | 0.1 | 0.3 | 4.6 | E24 | G2 | -0.3 | -0.4 | 3.2 |
| E24 | G13 | 0.1 | 0.2 | 3.4 | E18 | G5 | 0.0 | 1.2 | 4.6 | E18 | G2 | -0.3 | 0.1 | 3.2 |
| E20 | G14 | -0.2 | 0.0 | 3.3 | E3 | G21 | 0.2 | 0.7 | 4.6 | E3 | G14 | -0.2 | -0.4 | 3.1 |
| E24 | G3 | 0.1 | 0.2 | 3.3 | E10 | G21 | 0.2 | 0.2 | 4.6 | E20 | G8 | 0.1 | 0.5 | 3.1 |
| E6 | G3 | 0.1 | 0.1 | 3.2 | E2 | G8 | 0.1 | -0.1 | 4.6 | E10 | G4 | -0.1 | -1.0 | 3.1 |
| E22 | G13 | 0.1 | 0.3 | 3.2 | E24 | G21 | 0.2 | 0.4 | 4.4 | E24 | G5 | 0.0 | -0.8 | 3.0 |
| E17 | G14 | -0.2 | -0.3 | 3.1 | E17 | G8 | 0.1 | 0.3 | 4.3 | E14 | G4 | -0.1 | -0.2 | 3.0 |
| E5 | G13 | 0.1 | 0.5 | 2.9 | E2 | G11 | 0.0 | -0.2 | 4.3 | E23 | G13 | 0.1 | 0.2 | 3.0 |
| E24 | G14 | -0.2 | 0.0 | 2.8 | E14 | G1 | 0.1 | 0.9 | 4.3 | E14 | G14 | -0.2 | -0.1 | 3.0 |
| E16 | G4 | -0.1 | 0.7 | 2.8 | E17 | G5 | 0.0 | 0.4 | 4.2 | E18 | G8 | 0.1 | -0.5 | 3.0 |
| E22 | G3 | 0.1 | -0.3 | 2.6 | E3 | G1 | 0.1 | 0.4 | 4.2 | E14 | G5 | 0.0 | -0.4 | 3.0 |
| E6 | G1 | 0.1 | -0.6 | 2.5 | E3 | G5 | 0.0 | 0.4 | 4.2 | E8 | G8 | 0.1 | 0.6 | 2.9 |
| E17 | G3 | 0.1 | -1.3 | 2.4 | E3 | G8 | 0.1 | 0.3 | 4.1 | E17 | G2 | -0.3 | -0.6 | 2.9 |
| E5 | G14 | -0.2 | 0.2 | 2.3 | E24 | G19 | 0.1 | 0.2 | 4.1 | E13 | G1 | 0.1 | 0.1 | 2.9 |
| E16 | G3 | 0.1 | -0.2 | 2.1 | E3 | G13 | 0.1 | 0.2 | 4.0 | E13 | G8 | 0.1 | 0.1 | 2.9 |
| E5 | G3 | 0.1 | -0.3 | 2.1 | E18 | G4 | -0.1 | 0.7 | 4.0 | E18 | G18 | -0.2 | -0.4 | 2.8 |
| E10 | G1 | 0.1 | 0.2 | 1.9 | E2 | G7 | 0.0 | -0.5 | 4.0 | E14 | G3 | 0.1 | -0.7 | 2.8 |
| E5 | G4 | -0.1 | -0.4 | 1.8 | E24 | G14 | -0.2 | 0.3 | 4.0 | E13 | G3 | 0.1 | -0.1 | 2.7 |
| E24 | G18 | -0.2 | -1.1 | 1.7 | E23 | G3 | 0.1 | 1.2 | 4.0 | E5 | G3 | 0.1 | -0.5 | 2.7 |
| E16 | G1 | 0.1 | -0.6 | 1.6 | E24 | G3 | 0.1 | -0.1 | 3.8 | E13 | G4 | -0.1 | 0.1 | 2.7 |
| E10 | G13 | 0.1 | -0.1 | 1.6 | E14 | G13 | 0.1 | 0.4 | 3.8 | E5 | G13 | 0.1 | -0.7 | 2.5 |
| E10 | G3 | 0.1 | -0.1 | 1.6 | E18 | G19 | 0.1 | 0.3 | 3.8 | E13 | G2 | -0.3 | 0.1 | 2.5 |
| E10 | G4 | -0.1 | -0.1 | 1.4 | E5 | G8 | 0.1 | 0.6 | 3.7 | E3 | G2 | -0.3 | -1.1 | 2.3 |
| E18 | G3 | 0.1 | 0.2 | 3.7 | E23 | G14 | -0.2 | -0.2 | 2.3 | |||||
| E17 | G4 | -0.1 | -0.1 | 3.7 | E23 | G4 | -0.1 | -0.4 | 2.2 | |||||
| E24 | G18 | -0.2 | 0.0 | 3.7 | E8 | G13 | 0.1 | -0.1 | 2.2 | |||||
| E20 | G4 | -0.1 | 1.2 | 3.7 | E18 | G20 | 0.1 | -1.4 | 2.0 | |||||
| E3 | G4 | -0.1 | 0.0 | 3.6 | E23 | G1 | 0.1 | -0.8 | 2.0 | |||||
| E5 | G5 | 0.0 | 0.6 | 3.6 | E8 | G14 | -0.2 | 0.0 | 1.9 | |||||
| E18 | G1 | 0.1 | 0.1 | 3.6 | E8 | G21 | 0.2 | -0.4 | 1.9 | |||||
| E13 | G21 | 0.2 | 0.7 | 3.6 | E8 | G1 | 0.1 | -0.4 | 1.9 | |||||
| E24 | G1 | 0.1 | -0.4 | 3.5 | E20 | G3 | 0.1 | -0.8 | 1.8 | |||||
| E10 | G1 | 0.1 | -0.8 | 3.5 | E8 | G3 | 0.1 | -0.4 | 1.8 | |||||
| E24 | G4 | -0.1 | -0.3 | 3.5 | E23 | G2 | -0.3 | -0.6 | 1.8 | |||||
| E23 | G21 | 0.2 | 0.7 | 3.5 | E23 | G5 | 0.0 | -0.9 | 1.7 | |||||
| E3 | G3 | 0.1 | -0.4 | 3.5 | E13 | G5 | 0.0 | -1.1 | 1.6 | |||||
| E23 | G8 | 0.1 | 0.7 | 3.5 | E20 | G5 | 0.0 | -1.0 | 1.5 | |||||
| E2 | G5 | 0.0 | -1.1 | 3.4 | E8 | G2 | -0.3 | -0.5 | 1.3 | |||||
| E8 | G5 | 0.0 | 1.2 | 3.4 | ||||||||||
g: predicted value for the i-th genotype (); ge: predicted value for the interaction of the i-th genotype with the j-th environment (BLUP); E1: Arapoti/PR; E2: Augusto Pestana/RS; E3: Bela Vista do Norte/PY; E4: Cachoeira do Sul/RS; E5: Capinzal/SC; E6: Diamante D'Oeste/PR; E7: Dourados/MS; E8: Entre Rios do Oeste/PR; E9: Faxinal dos Guedes/SC; E10: General Câmara/RS; E11: Jaboticabal/SP; E12: Laguna/MS; E13: Major Vieira/SC; E14: Mangueirinha/PR; E15: Maracaju/MS; E16: Palmital/SP; E17: Palotina/PR; E18: Passo Fundo/RS; E19: Rio Verde/GO; E20: Salto do Lontra/PR; E21: São Francisco de Paula/RS; E22: Tiradentes do Sul/RS; E23: Toledo/PR; E24: Três Passos/RS; G1: BRS284; G2: BRS391; G3: BRS511; G4: BRS523; G5: BRS525; G6: BRS531; G7: BRS537; G8: BRS539; G9: BRS546; G10: BRS573; G11: BRS6013; G12: BRSMG534; G13: DF 2353; G14: GE; G15: IDRBASALTO; G16: IDRPETROVITA; G17: IDRPÉVERMELHO; G18: IPR110; G19: IPR115; G20: IPR122; G21: LIN16; G22: SOYPLENASG7401.
These results highlight that choosing the best genotype or the most assertive growing environment for the organic system in just a few harvests can lead to erroneous deductions. In this study, for example, the environment of General Câmara/RS went from the worst position to the top of the ranking due to a change in the influence of the harvest effect. The G×E interaction constitutes one of the main challenges faced by breeders when identifying, selecting, and recommending genotypes, since it can lead to inconsistencies in the classification of materials between environments and reduce the association between phenotypic and genotypic values. Recent studies have reinforced this complexity, demonstrating that environmental variation substantially modifies the relative performance of genotypes and demands robust analytical methodologies (Abebe et al., 2024; Happ et al., 2021).
In the 2021/2022 growing season, the genotypes DF 2353, BRS546, BRS539, BRS284, and BRS525 showed GY greater than 4.0 t ha⁻¹ when cultivated in São Francisco de Paula/RS, Passo Fundo/RS, Jaboticabal/SP, and São Francisco de Paula/RS, respectively (Table 3). The worst performance was attributed to the genotypes BRS525, BRS523, and BRS531 in Laguna/MS, with GY between 0.6 and 0.8 t ha⁻¹, being 87 to 82% lower than the GY of the best genotype in this growing season, highlighting the importance of evaluating the stability of the genotypes. Similarly, the highest predicted values for the 2022/2023 crop occurred in Três Passos/RS, where BRS546, BRS539, and BRS511 showed superior performance (6.6; 6.4; 5.6 t ha⁻¹). However, these same genotypes exhibited a decrease in GY when compared to other environments, e.g., the BRS546 genotype, which led in Três Passos/RS (6.6 t ha⁻¹), fell to 2.9 t ha⁻¹ in Palotina/PR and reached 1.6 t ha⁻¹ in Augusto Pestana/RS.
| 2021/2022 | 2021/2022 (continued) | 2022/2023 | 2023/2024 | ||||||||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| E | G | g | ge | E | G | g | ge | E | G | g | ge | E | G | g | ge | ||||
| E21 | G13 | 0.1 | 0.9 | 4.6 | E15 | G14 | -0.2 | 0 | 2.1 | E24 | G9 | 0.2 | 0.4 | 6.6 | E2 | G17 | -0.1 | 1.3 | 6.1 |
| E18 | G9 | 0.2 | 0.7 | 4.4 | E7 | G13 | 0.1 | 0.1 | 2 | E24 | G8 | 0.1 | 0.2 | 6.4 | E2 | G3 | 0.1 | 0.9 | 5.9 |
| E11 | G8 | 0.1 | 0.6 | 4.3 | E7 | G14 | -0.2 | 0.4 | 2 | E24 | G3 | 0.1 | -0.5 | 5.6 | E2 | G5 | 0 | 0.4 | 5.3 |
| E21 | G1 | 0.1 | 0.6 | 4.3 | E12 | G8 | 0.1 | 0.6 | 1.9 | E15 | G5 | 0 | 0.9 | 5.4 | E2 | G15 | -0.1 | 0 | 4.8 |
| E21 | G5 | 0 | 0.4 | 4 | E15 | G4 | -0.1 | -0.2 | 1.9 | E19 | G12 | 0.1 | 0.7 | 5 | E2 | G9 | 0.2 | -0.5 | 4.6 |
| E18 | G5 | 0 | 0.4 | 3.9 | E24 | G3 | 0.1 | 0.3 | 1.9 | E15 | G3 | 0.1 | 0.2 | 4.9 | E2 | G16 | 0 | -0.6 | 4.3 |
| E11 | G7 | 0 | 0.3 | 3.8 | E9 | G4 | -0.1 | -0.9 | 1.7 | E15 | G9 | 0.2 | -0.1 | 4.6 | E1 | G3 | 0.1 | 0.2 | 3.6 |
| E11 | G5 | 0 | 0.3 | 3.8 | E7 | G4 | -0.1 | 0 | 1.7 | E19 | G22 | 0 | 0 | 4.4 | E2 | G8 | 0.1 | -1.4 | 3.6 |
| E10 | G9 | 0.2 | 1 | 3.8 | E7 | G3 | 0.1 | -0.2 | 1.7 | E19 | G9 | 0.2 | -0.2 | 4.3 | E1 | G9 | 0.2 | 0.1 | 3.6 |
| E11 | G3 | 0.1 | 0.1 | 3.8 | E12 | G1 | 0.1 | 0.4 | 1.7 | E19 | G10 | -0.2 | -0.5 | 3.7 | E24 | G9 | 0.2 | 0.6 | 3.5 |
| E18 | G3 | 0.1 | 0.2 | 3.8 | E15 | G5 | 0 | -0.6 | 1.7 | E15 | G8 | 0.1 | -1 | 3.6 | E18 | G5 | 0 | 0.3 | 3.4 |
| E18 | G4 | -0.1 | 0.3 | 3.7 | E7 | G5 | 0 | -0.1 | 1.7 | E18 | G9 | 0.2 | 0 | 3.6 | E18 | G15 | -0.1 | 0.3 | 3.4 |
| E2 | G1 | 0.1 | 0 | 3.6 | E7 | G1 | 0.1 | -0.2 | 1.6 | E17 | G9 | 0.2 | 0.4 | 2.9 | E18 | G3 | 0.1 | 0.1 | 3.3 |
| E9 | G5 | 0 | 0.9 | 3.6 | E24 | G9 | 0.2 | -0.1 | 1.6 | E4 | G9 | 0.2 | 0.6 | 2.1 | E18 | G8 | 0.1 | 0 | 3.2 |
| E11 | G4 | -0.1 | -0.1 | 3.4 | E12 | G9 | 0.2 | 0.1 | 1.5 | E17 | G8 | 0.1 | -0.4 | 2 | E23 | G3 | 0.1 | 0.9 | 3.2 |
| E21 | G4 | -0.1 | -0.2 | 3.3 | E12 | G7 | 0 | 0.3 | 1.5 | E8 | G4 | -0.1 | 0.3 | 1.7 | E1 | G8 | 0.1 | -0.3 | 3.2 |
| E15 | G3 | 0.1 | 0.9 | 3.2 | E24 | G1 | 0.1 | -0.1 | 1.4 | E2 | G8 | 0.1 | 0.2 | 1.7 | E24 | G16 | 0 | 0.5 | 3.1 |
| E10 | G3 | 0.1 | 0.5 | 3.2 | E12 | G13 | 0.1 | 0.1 | 1.4 | E8 | G3 | 0.1 | 0 | 1.6 | E23 | G9 | 0.2 | 0.7 | 3.1 |
| E9 | G8 | 0.1 | 0.3 | 3.1 | E12 | G3 | 0.1 | -0.1 | 1.3 | E8 | G8 | 0.1 | 0 | 1.6 | E18 | G16 | 0 | 0.1 | 3.1 |
| E18 | G1 | 0.1 | -0.6 | 3 | E24 | G5 | 0 | -0.3 | 1.2 | E2 | G9 | 0.2 | 0.1 | 1.6 | E18 | G1 | 0.1 | -0.1 | 3.1 |
| E10 | G8 | 0.1 | 0.2 | 2.9 | E10 | G5 | 0 | -1.5 | 1.2 | E8 | G5 | 0 | 0 | 1.6 | E24 | G3 | 0.1 | 0.2 | 3 |
| E11 | G6 | -0.1 | -0.5 | 2.9 | E15 | G10 | -0.2 | -1 | 1.1 | E4 | G3 | 0.1 | -0.1 | 1.4 | E24 | G1 | 0.1 | 0.3 | 3 |
| E11 | G1 | 0.1 | -0.7 | 2.9 | E12 | G14 | -0.2 | -0.1 | 1 | E2 | G5 | 0 | 0 | 1.3 | E18 | G9 | 0.2 | -0.3 | 3 |
| E21 | G14 | -0.2 | -0.6 | 2.8 | E24 | G4 | -0.1 | -0.4 | 1 | E2 | G1 | 0.1 | 0 | 1.3 | E24 | G8 | 0.1 | 0.1 | 2.9 |
| E15 | G8 | 0.1 | 0.5 | 2.8 | E12 | G6 | -0.1 | -0.4 | 0.8 | E2 | G3 | 0.1 | -0.4 | 1 | E17 | G1 | 0.1 | 0.8 | 2.7 |
| E9 | G9 | 0.2 | -0.1 | 2.8 | E12 | G4 | -0.1 | -0.4 | 0.7 | E4 | G8 | 0.1 | -0.5 | 1 | E18 | G17 | -0.1 | -0.4 | 2.6 |
| E9 | G14 | -0.2 | 0.2 | 2.7 | E12 | G5 | 0 | -0.6 | 0.6 | E8 | G2 | -0.3 | -0.2 | 1 | E15 | G9 | 0.2 | 0.2 | 2.2 |
| E15 | G9 | 0.2 | 0.2 | 2.7 | - | - | - | - | - | - | - | - | - | - | E23 | G15 | -0.1 | -0.1 | 2 |
| E15 | G1 | 0.1 | 0.3 | 2.7 | - | - | - | - | - | - | - | - | - | - | E15 | G8 | 0.1 | 0 | 1.9 |
| E21 | G9 | 0.2 | -1.2 | 2.6 | - | - | - | - | - | - | - | - | - | - | E23 | G8 | 0.1 | -0.4 | 1.9 |
| E10 | G1 | 0.1 | -0.1 | 2.6 | - | - | - | - | - | - | - | - | - | - | E24 | G15 | -0.1 | -0.7 | 1.8 |
| E10 | G13 | 0.1 | -0.1 | 2.6 | - | - | - | - | - | - | - | - | - | - | E17 | G3 | 0.1 | -0.1 | 1.8 |
| E10 | G4 | -0.1 | 0 | 2.5 | - | - | - | - | - | - | - | - | - | - | E23 | G16 | 0 | -0.4 | 1.8 |
| E9 | G13 | 0.1 | -0.4 | 2.4 | - | - | - | - | - | - | - | - | - | - | E17 | G8 | 0.1 | -0.2 | 1.7 |
| E15 | G6 | -0.1 | 0.1 | 2.3 | - | - | - | - | - | - | - | - | - | - | E15 | G3 | 0.1 | -0.2 | 1.7 |
| E18 | G14 | -0.2 | -1 | 2.3 | - | - | - | - | - | - | - | - | - | - | E17 | G9 | 0.2 | -0.5 | 1.5 |
| E24 | G8 | 0.1 | 0.6 | 2.2 | - | - | - | - | - | - | - | - | - | - | E24 | G17 | -0.1 | -1.1 | 1.5 |
| E15 | G13 | 0.1 | -0.2 | 2.1 | - | - | - | - | - | - | - | - | - | - | E23 | G17 | -0.1 | -0.7 | 1.4 |
g: predicted value for the i-th genotype (); ge: predicted value for the interaction of the i-th genotype with the j-th environment (); E1: Arapoti/PR; E2: Augusto Pestana/RS; E3: Bela Vista do Norte/PY; E4: Cachoeira do Sul/RS; E5: Capinzal/SC; E6: Diamante D'Oeste/PR; E7: Dourados/MS; E8: Entre Rios do Oeste/PR; E9: Faxinal dos Guedes/SC; E10: General Câmara/RS; E11: Jaboticabal/SP; E12: Laguna/MS; E13: Major Vieira/SC; E14: Mangueirinha/PR; E15: Maracaju/MS; E16: Palmital/SP; E17: Palotina/PR; E18: Passo Fundo/RS; E19: Rio Verde/GO; E20: Salto do Lontra/PR; E21: São Francisco de Paula/RS; E22: Tiradentes do Sul/RS; E23: Toledo/PR; E24: Três Passos/RS; G1: BRS284; G2: BRS391; G3: BRS511; G4: BRS523; G5: BRS525; G6: BRS531; G7: BRS537; G8: BRS539; G9: BRS546; G10: BRS573; G11: BRS6013; G12: BRSMG534; G13: DF 2353; G14: GE; G15: IDRBASALTO; G16: IDRPETROVITA; G17: IDRPÉVERMELHO; G18: IPR110; G19: IPR115; G20: IPR122; G21: LIN16; G22: SOYPLENASG7401.
In the 2023/2024 growing season, Augusto Pestana/RS provided the best performance, especially for IDRPÉVERMELHO, BRS511, and BRS525 (Table 3). However, these same genotypes showed significant drops in other environments, where IDRPÉVERMELHO reduced GY from 6.1 t ha⁻¹ to 1.5 t ha⁻¹ in Três Passos/RS, and G3 from 5.9 t ha⁻¹ (Augusto Pestana/RS) to 1.7 and 1.8 t ha⁻¹ in Maracaju/MS and Palotina/PR, respectively. This fluctuation reinforces the magnitude of the G×E interaction, with few genotypes maintaining consistent performance across environments.
The BRSMG534 genotype showed the best overall performance for mean GY considering all environments, exceeding 5 t ha⁻¹ (Fig. 2a). Next, the SOYPLENASG7401 and BRS6013 genotypes stood out, with 4.34 and 4.28 t ha⁻¹, respectively. BRS531 and BRS573 were the least productive, with 1.94 and 2.28 t ha⁻¹, suggesting poor adaptation to the organic system. The study by Casarotto et al. (2025), also conducted in an organic system, identified the LIN 16, BRS 539, and IPR 115 genotypes as those with the highest GY values, while BRS 531 and BRS 573 also showed inferior performance, as reported in our study.
Adaptability and stability indices were evaluated by ranking the genotypes based on the obtained values. Thus, it was observed that the genotypes BRSMG534, SOYPLENASG7401, and BRS6013 demonstrated high stability, with HMGV values of 5.05 and 4.38 t ha⁻¹, respectively, while BRS531 and BRS573 were less stable, with HMGV of 1.94 and 2.28 t ha⁻¹ (Fig. 2b). These values are weighted by the genotype's stability in the different environments. From the point of view of adaptability only (RPGV), which indicates the relative performance of genotypic values in relation to the environmental mean, BRSMG534, LIN16, BRS546, and IPR115 stood out, with respective RPGVs of 1.15, 1.13, 1.13, and 1.09 (Fig. 2c).
It can be inferred that these were the genotypes with the best response to environmental variations. In contrast, BRS573, BRS391, and IPR110 showed low adaptability (0.66, 0.78, and 0.79, respectively). Values greater than 1 indicate performance above the overall mean, while values below 1 show performance below the overall mean and low adaptability (Colombari Filho et al., 2013).
The harmonic mean of relative genotypic performance (HMRPGV) identifies genotypes with high genetic value, high stability, and good adaptability, simultaneously penalizing variations between environments (Fig. 2d). The BRSMG534 genotype again showed superiority, with an expected mean performance of 3.42 t ha⁻¹. LIN16 and IPR115 also stood out, with 3.27 and 3.21 t ha⁻¹, respectively, while BRS573, IDRPÉVERMELHO, and BRS391 obtained the worst performance in the organic system. For these genotypes, a mean performance of only 1.79, 2.17, and 2.23 t ha⁻¹, respectively, is expected.
The results show that only a limited set of genotypes combines high yield, stability, and good adaptability to the organic system, with BRSMG534 consistently standing out. This pattern is consistent with recent results by Pradebon et al. (2023), who demonstrated that the selection of genotypes adapted to the organic system depends on characteristics associated with physiological robustness, such as vigorous initial growth, efficient architecture, and good competitive ability. Similarly, Mushoriwa et al. (2022) and Gayosso Barragán et al. (2025) show that low-input environments intensify the G×E interaction, favoring genotypes with greater phenotypic plasticity and productive stability. Furthermore, the dependence on biological nitrogen fixation in the organic system reinforces the importance of genotypes with high nodulation efficiency and a positive response to inoculation (Borja Reis et al., 2022). In this context, the poor performance of BRS531, BRS573, and BRS391 can be attributed to low physiological robustness and reduced resource utilization efficiency under organic system conditions.
Materials and methods
Plant material, experimental design and environments
The study was conducted with soybean genotypes in an organic farming system (i.e., without the addition of synthetic molecules) over five growing seasons (2019/2020, 2020/2021, 2021/2022, 2022/2023, and 2023/2024). The experimental design used was a randomized block design organized in an incomplete factorial scheme with four replications, comprising 24 environments and 22 conventional genotypes (Table 4). The geographic distribution of the experimental environments is shown in Fig. 3.
| Code | Environment | Code | Genotype |
|---|---|---|---|
| E1 | Arapoti/PR | G1 | BRS284 |
| E2 | Augusto Pestana/RS | G2 | BRS391 |
| E3 | Bela Vista do Norte/PY | G3 | BRS511 |
| E4 | Cachoeira do Sul/RS | G4 | BRS523 |
| E5 | Capinzal/SC | G5 | BRS525 |
| E6 | Diamante D'Oeste/PR | G6 | BRS531 |
| E7 | Dourados/MS | G7 | BRS537 |
| E8 | Entre Rios do Oeste/PR | G8 | BRS539 |
| E9 | Faxinal dos Guedes/SC | G9 | BRS546 |
| E10 | General Câmara/RS | G10 | BRS573 |
| E11 | Jaboticabal/SP | G11 | BRS6013 |
| E12 | Laguna/MS | G12 | BRSMG534 |
| E13 | Major Vieira/SC | G13 | DF2353 |
| E14 | Mangueirinha/PR | G14 | GE |
| E15 | Maracaju/MS | G15 | IDRBASALTO |
| E16 | Palmital/SP | G16 | IDRPETROVITA |
| E17 | Palotina/PR | G17 | IDRPÉVERMELHO |
| E18 | Passo Fundo/RS | G18 | IPR110 |
| E19 | Rio Verde/GO | G19 | IPR115 |
| E20 | Salto do Lontra/PR | G20 | IPR122 |
| E21 | São Francisco de Paula/RS | G21 | LIN16 |
| E22 | Tiradentes do Sul/RS | G22 | SOYPLENASG7401 |
| E23 | Toledo/PR | ||
| E24 | Três Passos/RS |
Crop management, trait measurement and environmental data
The experimental units were 15 meters long by 6 meters wide, with rows spaced 0.45 meters apart, totaling 90 m². Sowing was carried out in the first half of October of the respective growing seasons, with a density of 10 seeds per linear meter. Grain yield (GY, t ha⁻¹) was measured from the harvest of the two central rows of each experimental unit when the plants were at physiological maturity, with grain moisture corrected to 13%. The meteorological variables of mean air temperature (in °C) and rainfall (in mm) were extracted to aid in understanding the results obtained, with the help of the NASA Power platform (Sparks, 2018).
Statistical analysis, genotypic prediction and stability indices
The resulting data matrix was subjected to analysis of the normality of errors and homogeneity of variances using the Shapiro-Wilk and Bartlett tests, respectively. The variance components for the variable GY were obtained from the following model
where is the response variable observed in the k-th block of the i-th genotype in the j-th environment; is the general mean; is the effect of the i-th genotype, considered random; is the effect of the j-th environment, considered fixed; is the effect of the interaction between the i-th genotype and the j-th environment, considered random; is the effect of the k-th block within the j-th environment; and is the random error. Significance was obtained through deviance analysis at a 5% probability level, with the results supported by the test of χ2.
The system of equations was solved using REML Expectation-Maximization (Henderson, 1975). This procedure allowed the estimation of the following parameters: phenotypic variance (), heritability in the broad sense (), genotypic mean heritability (), coefficient of determination of the interaction G×E (), correlation coefficient of the interaction G×E (), genotypic coefficient of variation (), residual coefficient of variation (), and proportion between and ().
Predicted values () of the i-th genotype in the j-th environment were obtained by adding the BLUPs of the genotypic and genotype × environment interaction effects to the estimated mean of the j-th environment, as follows:
where is the estimated mean of the j-th environment (fixed effect); is the BLUP of the genotypic effect of the i-th genotype; and is the BLUP of the interaction effect between the i-th genotype and the j-th environment.
Subsequently, BLUP-based indices described by Colombari Filho et al. (2013) were used to select genotypes with performance and stability. The Harmonic Mean of Genotypic Values was calculated with:
The Relative Performance of Genotypic Values was calculated with:
The Harmonic Mean of the Relative Performance of Genotypic Values was calculated with:
where is the number of test environments for the i-th genotype; is the genotypic value of the i-th genotype in the j-th environment; and is the overall mean for each environment . For the preparation of statistical analyses, the metan (Olivoto and Lúcio, 2020) and ggplot2 (Wickham, 2016) packages were used in the R software (R Core Team, 2026).
Conclusion
The HMGV, RPGV, and HMRPGV indices proved efficient in identifying genotypes with superior performance and responsiveness to environmental variations and the challenges of the organic farming system.
BRSMG534 showed the highest productivity and stability. SOYPLENASG7401, BRS6013, LIN16, and IPR115 also demonstrated good performance, with high adaptability and stability in different environments.
Author contributions
AA, MAC and LCP were responsible for preparing the field data, designing the experiments and conducting the evaluations.
LAG, IRC and WJAB provided methodological support, assisted with the organization of the experimental design and contributed to the technical interpretation of the data.
PB, MGM and CMB performed the statistical analyses, manuscript formatting, and technical and grammatical review.
References
- Abebe AT, Adewumi AS, Adebayo MA, Shaahu A, Mushoriwa H, Alabi T, Derera J, Agbona A, Chigeza G (2024) Genotype x environment interaction and yield stability of soybean (Glycine max L.) genotypes in multi-environment trials in Nigeria. Heliyon. 10(19):e38097. https://doi.org/10.1016/j.heliyon.2024.e38097
- Alsajri FA, Wijewardana C, Bheemanahalli R, Irby JT, Krutz J, Golden B, Reddy VR, Reddy KR (2022) Morpho-physiological, yield, and transgenerational seed germination responses of soybean to temperature. Front Plant Sci. 13:839270. https://doi.org/10.3389/fpls.2022.839270
- Borja Reis AF, Rosso LHM, Adee E, Davidson D, Kovács P, Purcell LC, Below FE, Casteel SN, Knott C, Kandel H, Naeve SL, Singh MP, Archontoulis S, Ciampitti IA (2022) Seed inoculation with Azospirillum brasilense in the U.S. soybean systems. Field Crops Res. 283:108537. https://doi.org/10.1016/j.fcr.2022.108537
- Carvalho IR, Nardino M, Demari GH, Bahry CA, Szareski VJ, Pelissari G, Ferrari M, Pelegrin AJ, Oliveira AC, Maia LC, Souza VQ (2016) Bi-segmented regression, factor analysis and AMMI applied to the analysis of adaptability and stability of soybean. Aust J Crop Sci. 10(10):1410–1416. https://doi.org/10.21475/ajcs.2016.10.10.pne63
- Casarotto RW, Carvalho IR, Pradebon LC, Loro MV, Bandeira WJA, Dalla Roza JP, Magano DA, Silva JAG, Alban AA, Challiol MA (2025) Use of geographic and edaphoclimatic information for the selection of soybeans for organic environments in current and future scenarios. Rev Ceres. 72:e72033. https://doi.org/10.71252/2177-34912025720033
- Colombari Filho JM, Resende MDV, Morais OP, Castro AP, Guimarães EP, Pereira JA, Utumi MM, Breseghello F (2013) Upland rice breeding in Brazil: A simultaneous genotypic evaluation of stability, adaptability and grain yield. Euphytica. 192(1):117–129. https://doi.org/10.1007/s10681-013-0922-2
- CONAB (2025) Acompanhamento da safra brasileira de grãos: safra 2024/25, 12º levantamento, setembro 2025. Companhia Nacional de Abastecimento, Brasília. https://www.gov.br/conab/pt-br/atuacao/informacoes-agropecuarias/safras/safra-de-graos/boletim-da-safra-de-graos/12o-levantamento-safra-2024-25/e-book_boletim-de-safras-12o-levantamento_2025.pdf
- Farias JRB, Nepomuceno AL, Neumaier N (2007) Ecofisiologia da soja. Circular Técnica 48. Embrapa Soja, Londrina. https://www.infoteca.cnptia.embrapa.br/handle/doc/470308/
- Feledyn-Szewczyk B, Jończyk K (2016) Assessment of the suitability of oat varieties (Avena sativa L.) for cultivation in organic system. J Res Appl Agric Eng. 61(3):82–87.
- Gayosso Barragán O, Acosta Gallegos JA, Alcalá Rico JSGJ, Jiménez Hernández Y, Chávez Aguilar G, Chávez Díaz IF, Aranda Lara U (2025) Genotype × environment interaction and yield stability of “Pinto” bean (Phaseolus vulgaris L.) lines in a semi-arid region of Mexico. Agriculture. 15(20):2150. https://doi.org/10.3390/agriculture15202150
- Happ MM, Graef GL, Wang H, Howard R, Posadas L, Hyten DL (2021) Comparing a mixed model approach to traditional stability estimators for mapping genotype by environment interactions and yield stability in soybean (Glycine max (L.) Merr.). Front Plant Sci. 12:630175. https://doi.org/10.3389/fpls.2021.630175
- Henderson CR (1975) Best linear unbiased estimation and prediction under a selection model. Biometrics. 31(2):423–447. https://doi.org/10.2307/2529430
- Mushoriwa H, Mathew I, Gwata ET, Tongoona P, Derera J (2022) Grain yield potential and stability of soybean genotypes of different ages across diverse environments in Southern Africa. Agronomy. 12(5):1147. https://doi.org/10.3390/agronomy12051147
- Olivoto T, Lúcio AD (2020) metan: An R package for multi-environment trial analysis. Methods Ecol Evol. 11(6):783–789. https://doi.org/10.1111/2041-210X.13384
- Pradebon LC, Carvalho IR, Loro MV, Port ED, Bonfada B, Sfalcin IC, Santos IJ, Bruinsma GMW, Silva JAG, Alban AA, Challiol MA (2023) Soybean adaptability and stability analyzes to the organic system through AMMI, GGE biplot and mixed models methodologies. Cienc Rural. 53(9):e20220262. https://doi.org/10.1590/0103-8478cr20220262
- Pupin S, Santos AVA, Zaruma DUG, Miranda AC, Silva PHM, Marino CL, Sebbenn AN, Moraes MLT (2015) Produtividade, estabilidade e adaptabilidade em progênies de polinização aberta de Eucalyptus urophylla S.T. Blake. Sci For. 43(105):127–134.
- R Core Team (2026) R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. https://doi.org/10.32614/R.manuals
- Reginatto M, Conceição PC, Mandrik GB (2025) Organic no-tillage system: An alternative to minimize the environmental impact and strengthen organic grain production. Rev Bras Agroecol. 20(3):337–357. https://doi.org/10.33240/rba.v20i3.57560
- Rother V, Verdi CA, Thurow LB, Carvalho IR, Oliveira VF, Maia LC, Venske E, Pegoraro C, Oliveira AC (2019) Uni- and multivariate methods applied to the study of adaptability and stability of white oat. Pesq Agropec Bras. 54:e00656. https://doi.org/10.1590/S1678-3921.pab2019.v54.00656
- Sentelhas PC, Battisti R, Câmara GMS, Farias JRB, Hampf AC, Nendel C (2015) The soybean yield gap in Brazil: Magnitude, causes and possible solutions for sustainable production. J Agric Sci. 153(8):1394–1411. https://doi.org/10.1017/S0021859615000313
- Silva JAG, Costa KAP, Silva LM, Severiano EC, Silva FG, Habermann E, Martinez CA, Vilela L, Silva AG, Costa AC, Costa JVCP, Oliveira KJ (2023) Integrated systems improve the sustainability of soybean cultivation in the tropical region. Front Sustain Food Syst. 7:1224530. https://doi.org/10.3389/fsufs.2023.1224530
- Sousa TJF, Rocha MM, Damasceno-Silva KJ, Bertini CHCM, Silveira LM, Sousa RR, Sousa JLM (2019) Simultaneous selection for yield, adaptability and genotypic stability in immature cowpea using REML/BLUP. Pesq Agropec Bras. 54:e01234. https://doi.org/10.1590/S1678-3921.pab2019.v54.01234
- Sparks A (2018) nasapower: A NASA POWER global meteorology, surface solar energy and climatology data client for R. J Open Source Softw. 3(30):1035. https://doi.org/10.21105/joss.01035
- Toloi MNV, Bonilla SH, Toloi RC, Silva HRO, Nääs IA (2021) Development indicators and soybean production in Brazil. Agriculture. 11(11):1164. https://doi.org/10.3390/agriculture11111164
- Venancio WS, Dallago EG, Slusarz GM, Boratto IV, Boratto VNM, Leite AC, Cuenca R, Ratuchne A (2024) Impact of pre-inoculation of soybean seeds with Bradyrhizobium spp. applied 60 days before sowing. Aust J Crop Sci. 18(2):116–124. https://doi.org/10.21475/ajcs.24.18.02.PNE4106
- Wickham H (2016) ggplot2: Elegant graphics for data analysis. Springer, Cham. https://doi.org/10.1007/978-3-319-24277-4
Cite this article
Gollo LA et al. (2026) Adaptability, stability, and productivity of soybeans in an organic multi-environment system. Aust J Crop Sci. 20(11):948–957. https://doi.org/10.21475/ajcs.26.20.11.pne214