Introduction

Maize (Zea mays L.) cultivation holds significant economic and social importance. Its cultivation occurs under a wide range of soil, climate, and technological conditions due to its significant genetic variability and diverse genotypes. This versatility positions it as one of the most vital agricultural species globally, with widespread cultivation across nearly all continents1,2.

This cereal plays a versatile role in various daily scenarios, serving as a staple in both human and animal nutrition. Additionally, it contributes to our energy resources by serving as a valuable source of biofuel. Currently, world maize production is concentrated in the United States, China, and Brazil, with an estimated production for the 2023/2024 harvest of 389.7, 288.8, and 122.0 million tons respectively. In Brazil, maize cultivation holds the second position both in terms of cultivated area and grain production, being surpassed only by soybean cultivation2,3.

As agriculture has undergone modernization over the years and with the pivotal role of maize on the global stage, genetic improvement has emerged as one of the most advantageous strategies for achieving enhancements in agricultural productivity and grain quality4,5. Given its significance, genetic improvement programs for maize have incorporated multi-environment assessments as a pivotal step. This approach has emerged as a key strategy for identifying optimal maize hybrids with high yields tailored to specific regions. The process involves evaluating their adaptability to the diverse edaphoclimatic conditions across cultivation regions in various environments. This comprehensive analysis, coupled with a focus on broad adaptability and stability, aids in the selection of superior maize hybrids6,7,8.

Multi-environment assessment is crucial, as it enables the identification of genotype × environment interactions (GEIs). These interactions can significantly influence the relative performance of genotypes, particularly when they exhibit diverse responses to environmental changes across various settings. Understanding such complexities is essential for making informed selections and recommendations of genotypes, a task that becomes challenging in the face of divergent environments9,10,11,12. Genotype × environment (GxE) interactions can be classified as crossover or non-crossover, depending on changes in the relative ranking of genotypes between environments or merely variations in the magnitude of responses.

GE interactions can manifest in simplicity, where genotype classification remains consistent across variable environments, and the observed interaction stems from variations in response magnitude. Alternatively, interactions can take on complexity, wherein genotype classification (ranking) varies across different environments, underscoring its heightened significance in the context of plant improvement13,14,15. While the investigation of GEIs holds paramount significance in breeding, it falls short of furnishing detailed insights into the specific responses of individual genotypes to environmental variations. Consequently, adaptability and stability analyses are undertaken to address this gap16.

Adaptability can be defined as the ability of a genotype to respond to environmental stimuli, whereas stability is the constancy of performance across environments and can also be expressed as the smallest average variation in the considered environments. Stability is a function of the predictability of a genotype’s response16. Various methodologies, each characterized by distinct concepts and biometric procedures for estimation, can be used in the analysis17. These methodologies enable the identification of genotypes that demonstrate enhanced adaptability, stability, and responsiveness to their respective test environments.

Among the latest methodologies, the analysis utilizing the GGE (genotype main effects + genotype environment interaction) biplot model, as proposed by Yan et al.6, incorporates both the main effects of genotype and the interactions between genotypes and environments. The biplot method was developed by Gabriel18to graphically represent the results of the analysis of principal components17, bringing together information from different variables in a single plot and allowing a clearer visualization of these relationships.

Therefore, the identification of superior genotypes is essential to enable their recommendation. The GGE biplot method has been widely used in various studies to assess the adaptability and stability of different maize cultivars. Several studies have used this methodology for this purpose, recently such as Lima et al.19 and Oliveira et al.20 applied the GGE biplot to identify common maize hybrids that exhibited stability and high yield. Crevelari et al.15 used this approach for silage maize employing the same three environments used in this study, highlighting its efficiency in varied environments. Santos et al.21 applied the GGE biplot in popmaize. The GGE biplot method is particularly useful for identifying the best genotypes in multi-environment trials, allowing researchers to better understand genotype × environment interaction and hybrid stability across different locations and conditions. Thus, this study aimed to evaluate the performance of maize hybrids through GGE biplot analysis, focusing on identifying those that demonstrate greater adaptability and stability across diverse environments in the North/Northwest regions of Rio de Janeiro, in order to select superior genotypes for recommendation. To address a knowledge gap in the existing literature, this study specifically investigates the adaptability and stability of single, double, and interpopulation maize hybrids in the unique environmental conditions of this region, an area where comparative data is scarce. Unlike previous studies, our analysis aims to provide insights into how these hybrids respond to the specific edaphoclimatic factors in the North/Northwest of Rio de Janeiro, an approach that has not been sufficiently explored in the context of maize hybrids in this region.

Materials and methods

Plant genetic material

The genetic materials used in this study consisted of hybrids from the maize genetic breeding program at the State University of Northern Rio de Janeiro Darcy Ribeiro (UENF), namely: UENF MSV 2210, UENF MS 2208, UENF 506 − 11, and UENF 506 − 16 (Registration number with MAPA: 56523), and the commercial maize hybrids available on the market, that is, controls, AG 1051, BM 207, LG 6036, and 30F35R.

The hybrids UENF 506 − 11 and UENF 506 − 16 were obtained from the 11th and 16th cycles, respectively, by the reciprocal recurrent selection method of a family of full siblings, as recommended by Hallauer and Ebehart22, with modifications following Berilli et al.23 and Pereira et al.24. In this method, these hybrids were obtained by crossing between the Cimmyt and Piranão populations in their respective cycles of recurrent selection, which belong to distinct heterotic groups. The CIMMYT population belongs to a ‘Flint’ type heterotic group, while the Piranão population belongs to a ‘Dent’ heterotic group23,24. The hybrids UENF MSV 2210 and UENF MS 2208 were obtained by the topcross method from genotypes belonging to the maize collection of the State University of Northern Rio de Janeiro Darcy Ribeiro. These genotypes, all from the ‘dent’ heterotic group, were crossed with a single tester, Piranão 12, which also originated from the Genetics and Plant Breeding program at UENF. Piranão 12 has a broad genetic base and belongs to the same ‘dent’ heterotic group25.

Evaluation trials

The hybrid evaluation trials were conducted in two growing seasons in the 2021 winter season and 2021/2022 summer season in three locations simultaneously: Escola Técnica Estadual Agrícola Antônio Sarlo, in Campos dos Goytacazes – RJ, located 13 m above sea level, with geographic coordinates of 21°45′16″ South and 41°19′28″ West, mean annual precipitation of 1073 mm, and mean annual temperature of 23.6 °C; Instituto Federal Fluminense, Advanced Campus Cambuci – RJ, located 35 m above sea level, with geographic coordinates of 21°34′31″ South and 41°54′40″ West, mean annual precipitation of 1200 mm, and mean annual temperature 23 °C; and Barra do Pomba Island Experimental Station in Itaocara – RJ, located 60 m above sea level, with geographic coordinates of 21°40′09″ South and 42°04′36″ West, mean annual precipitation of 1221 mm, and mean annual temperature of 23 °C. The choice of environments for the evaluation trials was based on the value for cultivation and use (VCU) of the hybrid UENF 506–1624. The regions of Campos dos Goytacazes, Cambuci, and Itaocara in the state of Rio de Janeiro exhibit edaphoclimatic variations, composing the North/Northwest Fluminense regions. These areas have significant differences in terms of soil, precipitation, and temperature, which allowed for the evaluation of the adaptability and stability of hybrids under diverse environmental conditions. Furthermore, these regions are important for local agricultural production, making the results more applicable to producers. The choice was also facilitated by the infrastructure of the institutions where the trials were conducted, ensuring technical support for the research.

The statistical experimental design consisted of randomized blocks with four replications. The experimental unit consisted of four rows 4.0 m long, considering a spacing between rows of 0.70 m and a spacing between plants in the row of 0.20 m, resulting in 20 plants in each row and 80 plants per plot.

Sowing fertilization was performed with the NPK formula 04–14-08 (80 g/m). Subsequently, two top dressing fertilizations were performed: 30 days after planting, with the NPK formula 20-00-20 (30 g/m), and 45 days after planting, with urea (20 g/m). Cultural management was carried out as recommended for the crop26 and the harvest was carried out 120 days after sowing.

Evaluated traits

The following traits were evaluated: mean length of ears without straw (LEWS), obtained by measuring six ears at random from each of the rows in the experimental unit using a digital caliper, with values expressed in millimeters; mean diameter of ears without straw (DEWS) was obtained by averaging the same six ears, measured with a digital caliper and expressed in millimeters; mean weight of ears without straw (WEWS), obtained by weighing all ears of experimental units without straw, with results transformed into kg ha−1; 1000-grain weight (W1000), obtained by weighing a sample of 1000 healthy grains from each row of experimental units using a precision scale, with results expressed in grams; and grain yield (YLD), obtained from weighing the grains of each row of the experimental units after threshing, with values converted into kg ha−1 and grain weights corrected for 13% moisture.

Statistical analysis

The data were initially subjected to an individual analysis of variance for each environment separately and the Hartley27 test for testing the homogeneity of residual variances. Subsequently, a joint analysis of variance was conducted with the three locations in both seasons (totaling six environments), according to the following statistical model:

Yijk = µ + Gi + B/Ejk + Ej + GEij + eijk.

Where, Yijk is the evaluation in the i-th genotype, j-th environment, and k-th block; µ is the general constant; Gi is the fixed effect of the i-th genotype; B/Ejk is the effect of the k-th block in the j-th environment; GEij is the effect of the interaction between the i-th genotype and j-th environment; and eijk is the random error associated with the observation, with NID (0, σ2).

Subsequently, the adaptability and phenotypic stability of the genotypes were studied using the GGE biplot methodology, and the components of variance and genetic parameters were estimated The Genes28 program (version 6.0) was used to perform the analysis of variance and the Tukey29 mean test, while the GGE biplot methodology was performed using by R30 program (version 4.3.1), using the METAN31 package, which allows for a more in-depth visual representation of genotype × environment interactions and is particularly useful in multi-environment trials.

The estimates of the variance components and genetic parameters were calculated through the following calculations:

Genotypic variability Eq. (1):

$$\:{\varTheta\:}_{g}^{2}=\frac{(QMG-QMGE)}{re}$$
(1)

Phenotypic variance Eq. (2):

$$\:{\sigma\:}_{f}^{2}=\:\frac{QMG}{re}$$
(2)

Genotypic determination coeficiente Eq. (3):

$$\:{H}^{2}=\left(\frac{{\varTheta\:}_{g}^{2}}{{\sigma\:}_{f}^{2}}\right)$$
(3)

Results and discussion

Analysis of variance

We initially ensured that the mean squared error ratio (MSE>/MSE< ) for all variables was consistently below seven to conduct the analysis of variance. According to Pimentel Gomes32, this value recommended considering the homogeneities of variances of residuals (Table S1), thus allowing to evaluate them in a joint analysis (Table 1).

Table 1 Summary of the joint analysis of variance with the respective mean squares of the sources of variation (SV) and degrees of freedom (DF), estimates of the components of variance, genetic parameters, and means for six traits evaluated in maize genotypes in Campos dos Goytacazes, Cambuci, and Itaocara, RJ, Brazil.
Full size table

The joint analysis of variance showed that the coefficients of experimental variation (CVe) ranged from 2.34% for the diameter of ears without straw to 16.76% for yield, showing good experimental precision and being within the classifications proposed by Scapim et al.33 and Fritsche-Neto et al.34.The coefficient of genotypic determination (H²) the traits of 1000-grain weight and prolificacy both showed values of 86%, while the other traits had values greater than 80%, ranging from 92% for ear length to 99% for ear diameter. High H² estimates suggest that most of the observed phenotypic variation is due to genetic causes.

In the joint analysis of variance (Table 1), significant effects (P < 0.01) were observed for all variables, except for yield, which showed significance at the 5% level. The sources of variation for genotypes and environment were significant at the 1% level for all traits evaluated. The genotype × environment interactions were also significant at the 1% level for all traits, except for yield, which was significant at the 5% level. These significant interactions between genotypes and environments during the respective seasons emphasize the importance of multi-environment evaluations. The analysis of variance revealed that environmental conditions play an important role in the expression of the evaluated traits, highlighting the need for evaluations across multiple environments to obtain more reliable genotype recommendations.

These results suggest that the genotypes exhibited different behaviors in the studied environments, justifying the need for a more in-depth study of genotype behavior to understand the magnitude of their interactions with the environments35.

Evaluations across multiple environments are essential to understanding genotype performance under diverse ecological conditions, as traits can vary with climate, soil, and other factors. In this study, assessments in regions with distinct environmental characteristics, such as Campos dos Goytacazes, Cambuci, and Itaocara, highlighted how these conditions affect hybrid traits.This multi-environment approach enables the identification of hybrids that combine high productivity with stability under varying conditions. Such analysis is particularly valuable for the North and Northwest of Rio de Janeiro, where climatic and soil variability significantly influence crop performance, supporting the development of tailored cultivation strategies.

Thus, this reinforces the importance of studying the adaptability and stability of the genotypes, to guide recommendations, especially for producers in the North/Northwest regions of the state of Rio de Janeiro.

Adaptability and stability analyses

The GGE biplot method was used to study the adaptability and stability of genotypes. The first two principal components (PC1 and PC2) were related to the main effect of the genotype and the fraction of yield that is associated with the effects of the interaction (G) + (GE). The first principal component (PC1) indicated the genotype adaptability, that is, it is highly correlated to yield. The second principal component (PC2) indicated phenotypic stability, being correlated to environments. The genotypes with the PC2 closest to zero are the most stable6.

According to Yan (2000), both principal components must explain a total variation (G + GE) higher than 60% to indicate the correct application of the GGE biplot methodology. Figure 1 follows this approach, representing all variables, with values of PC1 + PC2 = 86.01% for the length of ears without straw, 97.7% for the diameter of ears without straw, 86.21% for prolificness, 92.09% for the weight of ears without straw, 80.85% for 1000-grain weight, and 91.04% for yield, thus showing high reliability in the results. Crevelari et al.15 Lima et al.19 and Oliveira et al.20 evaluated silage and common maize hybrids and obtained values of 83.68%, 62.74%, and 60.01% for yield, respectively.

GGE biplot provides several tools through graphical plotting to visualize the response of genotypes at different seasons and environments. The polygon called which-won-where (Fig. 1) divides the plot into sectors using dotted blue lines starting from the center of the biplot, grouping the environments and representing those that are most similar36. The formed polygon, represented by the blue solid line, shows the presence of genotypes that behave within the polygon and at its vertices.

Fig. 1
figure 1

GGE biplot representing which-won-where of the main effects of the genotype and the genotype × environment interaction of eight maize hybrids for (a) length of ears without straw (mm), (b) diameter of ears without straw (mm), (c) prolificness (ear/plant), (d) weight of ears without straw (kg ha−1), (e) 1000-grain weight (g), and (f) grain yield (kg ha−1) in six environments. sum: summer crop season (21/22) and win: winter crop season (21/21).

Full size image

The genotypes allocated within the polygon present the lowest mean performance for the evaluated traits. Those positioned at the vertices, but not allocated to any sector that has an environment, are considered unfavorable to the tested environments, reproducing low responsiveness and performance. In contrast, the genotypes positioned at the vertices, which are further away from the center of origin and are contained in a sector that has environments, showed higher responsiveness to stimuli within this group of environments36,37,38.

Thus, the genotype UENF 506 − 11 presented the greatest value for the variable length of ears without straw, followed by the genotype BM 207. The genotype UENF 506 − 11 is positioned at the vertex of the polygon in the sector that contains the environments Campos dos Goytacazes – summer, considered the best environment for this genotype, Itaocara – winter and summer, and Cambuci – winter. In contrast, the genotype BM 207 is located at the other vertex, which covers the other environments (Fig. 1a).

The genotype 30F35R had the highest value for the diameter of ears without straw (Fig. 1b) and is located at the vertex of the polygon, within the sector that includes the environments of Campos dos Goytacazes and Cambuci in both seasons and Itaocara – winter. The genotype AG 1051 is located in this same sector, with the third highest mean. The environment Itaocara – winter is located at the other vertex in another sector, where the genotype LG 6036 is found, showing the second highest mean.

The highest prolificness mean was observed for the genotype UENF 506 − 16, which is found at the vertex of the polygon, in the sector that encompasses all environments, except Itaocara – summer (Fig. 1c). The genotypes 30F35R and LG 6036 had the highest means for the weight of ears without straw, but they are in opposite vertices and sectors. The genotype 30F35R is found in the sector that has most of the environments, together with the genotype AG 1051, which had the third highest mean. The genotype LG 6036, which has the second highest mean, is located in the environment Cambuci – winter, considered the best environment for this genotype, and Campos dos Goytacazes – summer (Fig. 1d).

The genotypes UENF MS 2208 and UENF MSV 2210, located at the vertices of the polygon, presented the highest means for the variable 1000-grain weight (Fig. 1e). These genotypes are in the sector that encompasses almost all environments, except Campos dos Goytacazes – summer. The genotype LG6036 demonstrated the best yield (Fig. 1f) and is located at the vertex of the polygon, which contains the environments of Cambuci – winter and Campos dos Goytacazes – summer, showing a higher response to the stimuli from this set. The genotypes 30F35R and AG 1051 were positioned at the vertices in the sector that contains the other environments, that is, Campos do Goytacazes – winter, Cambuci – summer, and Itaocara – winter and summer. These genotypes were the best within this vertex, where most environments are concentrated.

However, the genotypes found at the vertices of the polygon but do not cluster in any of the studied environments were not considered responsive to these environments for all variables.

The visualization of the mean performance and stability of genotypes is obtained by arranging the two-dimensional graph by drawing an average environment coordinate (AEC) on the biplot (Fig. 2). Firstly, an average environment, represented by the small circle, is defined by the averages of the PC1 and PC2 components of the environments, and the arrow found on the line that passes through the origin of the biplot indicates the highest average performance of the genotypes.

The AEC ordinate is represented by the vertical line that passes through the origin of the biplot and is perpendicular to the abscissa, which corresponds to the AEC horizontal line (represented by solid green lines). Therefore, if the AEC abscissa represents (G), the AEC ordinate approximates the GEI associated with each genotype, which is a measure of variability or instability of the genotypes39. Thus, the projections relative to the AEC abscissa are represented by dotted blue lines on the graph, in which a higher projection means higher instability of the genotypes regardless of the direction, and those that have a smaller projection relative to the AEC abscissa are the most stable40, indicating that the genotypes were more or less affected by the genotype x environment interaction, respectively.

Looking at the graph, based on the arrowhead and the circle of the average environment, we can analyze the means which are located on the abscissa passing through the origin of the biplot, indicating the best average performance of the genotypes. Hence, it is crucial to ascertain the direction of the arrow to discern whether the genotypes fall below or exceed the overall average (AEC ordinate).

Fig. 2
figure 2

GGE biplot representing “average x stability” of the main effects of the genotype and the genotype × environment interaction of eight maize hybrids for (a) length of ears without straw (mm), (b) diameter of ears without straw (mm), (c) prolificness (ear/plant), (d) weight of ears without straw (kg ha−1), (e) 1000-grain weight (g), and (f) grain yield (kg ha−1) in six environments. sum: summer crop season (21/22) and win: winter crop season (21/21).

Full size image

Based on the previous descriptions, Fig. 2a shows that the genotypes UENF 506 − 11 and BM 207 presented the highest means for the variable length of ears without straw, with values of 167.98 and 167.29 mm, respectively. The genotype AG 1051 had the third-highest mean (164.06 mm). However, both UENF 506 − 11 and AG 1051 are considered the least stable due to their higher projections relative to the horizontal line despite being above average. On the other hand, the hybrids UENF MSV 22,010 and 30F35R also showed means above the overall average and are more stable, as their projections are smaller and closer to the horizontal line.

The hybrids 30F35R, LG6036, AG 1051, and BM 207 presented values for the diameter of ears without straw (Fig. 2b) above the overall average. However, they are considered the least stable for this trait. UENF hybrids were below the overall average but are the most stable for this variable. The genotypes UENF 506 − 16, UENF 506 − 11, BM 207, and UENF MS 2208 are above the overall average for prolificness, with UENF 506 − 11 being the most stable, followed by UENF 506 − 16 (Fig. 2c).

The weight of ears without straw stood out for the hybrids 30F35R and LG 6036, followed by BM 207 and AG 1051. These hybrids are above average but are considered the least stable in terms of the weight of ears without straw (Fig. 2d). On the other hand, we have a final group of hybrids that are below the overall average in terms of weight of ears without straw but have good stability. The hybrids UENF MS 2208, UENF MSV 2210, UENF 506 − 11, and LG 6036 presented the highest means for 1000-grain weight, being above the overall average, and seemed to have good permanence for this trait (Fig. 2e).

The genotypes LG6036 and 30F35R, which are single hybrids, had the highest yield means, with values of 9,743 and 9,644 kg ha−1, respectively, and are also the least stable due to their longer distance of projection relative to the horizontal line. The genotypes AG 1051 and BM 207, which are double hybrids, with means of 8,318 and 8,221 kg ha−1, respectively, were also considered unstable although they presented means higher than the overall average. The UENF genotypes UENF MSV 2210, UENF 506 − 16, UENF 506 − 11, and UENF MS 2208, which are interpopulation hybrids, presented means of 7,047, 6,948, 6,956, and 6,408 kg ha−1 and were the most stable, as they had the smallest projections relative to the AEC abscissa (Fig. 2f).

The genotype considered “ideal” is defined as the one that presents the largest vector in the principal components PC1 and PC2, without projection, represented by the arrow in the center of the concentric circles, thus presenting the best average performance and stability in all tested environments36,41,42. The other concentric circles assist in visualizing the distance between the genotypes.

Based on this definition, the ideotype presents high performance and high stability. Nevertheless, the ideotype under discussion is purely hypothetical, intended solely as a reference for comparing other genotypes. Therefore, the analysis of Fig. 3 allows identifying the ideotype as being one that contains a long vector and low G × E interaction, represented by the arrow inside the smallest concentric circle in the graph area.

Thus, the hybrid BM 207, located in the first concentric circle, is considered the closest to the ideal when interpreting the graph for the variable length of ears without straw, showing higher production stability (Fig. 3a). Similarly, the hybrid 30F35R is the closest to the ideotype relative to the variable diameter of ears without straw, located in the first concentric circle and demonstrating good stability (Fig. 3b).

The genotype UENF 506 − 16, positioned in the second concentric circle, stood out in terms of prolificness (Fig. 3c), with the best productive stability. The hybrids 30F35R and LG6036 are located in the second concentric circle, being closer to the ideal genotype for the variable weight of ears without straw (Fig. 3d). The genotypes UENF MS2208 and UENF MSV2210 are considered ideal for 1000-grain weight, being located in the first and second concentric circles, respectively (Fig. 3e).

The hybrids 30F35R and LG6036, positioned in the second concentric circle, closely approximate the ideal genotype for the variable yield. The hybrid LG6036 has a higher yield, not being so stable, thus giving 30F35R the genotype that would be ideal in terms of average and stability (Fig. 3f).

Fig. 3
figure 3

GGE biplot representing the “ideal genotype” of the main effects of the genotype and the genotype × environment interaction of eight maize hybrids for (a) length of ears without straw (mm), (b) diameter of ears without straw (mm), (c) prolificness (ear/plant), (d) weight of ears without straw (kg ha−1), (e) 1000-grain weight (g), and (f) grain yield (kg ha−1) in six environments. sum: summer crop season (21/22) and win: winter crop season (21/21).

Full size image

The validity of all previously discussed results was affirmed through the use of the mean test (Table 2).

Table 2 Overall averages of eight genotypes for all evaluated traits with their respective Tukey’s mean tests in three locations in both seasons (six environments).
Full size table

The behavior of the genotypes in the graphs relative to grain yield, which is considered the main trait in maize cultivation, showed the existence of three distinct groups, according to the mean test presented in Table 2. The first group is composed of two commercial genotypes, characterized by higher means, consisting of single hybrids that are least stable. The second group is formed by the other two commercial genotypes, which consist of double hybrids with intermediate means and stability. Finally, the third group is represented by UENF hybrids, which are interpopulation hybrids, with lower means but more stable. This behavior can also be observed in variables highly correlated with yield, such as the diameter and weight of ears without straw.

Single hybrids come from crossing two pure strains and are indicated for production systems that use high technology due to their higher production potential and higher uniformity of plants and ears43. However, these hybrids tend to have a higher cost due to their origin from the crossing of pure strains.

Pure lines are developed through successive self-fertilization and selected over several generations, thus being composed of individuals descending from the same plant and characterized by homozygosity, that is, they have basically the same genetic characteristics44. This genetic uniformity can lead to inbreeding depression. However, hybrid vigor is recovered when crossing two distinct pure lines, maximizing heterosis and giving them high productivity45, resistance to diseases and pests, and adaptability to different climate conditions, among other desirable characteristics.

Importantly, the favorable genetic uniformity of single hybrids in a given environment can become disadvantageous when they are grown in a series of contrasting environments or under biotic or abiotic stress conditions45.

Double hybrids come from crossing two single hybrids and are characterized by combining increased productivity with reduced production costs. Furthermore, these hybrids present higher heterogeneity (genetic variability), which results in higher production stability43. Double hybrids partially capitalize on heterosis, resulting in a medium yield, being recommended for producers with intermediate agricultural technology, seeking a balance between increased productivity and lower production costs.

Interpopulation hybrids allowed for offering alternative technologies for producing hybrid seeds with good genetic potential but at more affordable costs. These hybrids result from the crossing between two maize populations that have a broad genetic base and partially capitalize on heterosis. These populations have genetic variability, and hybrids tend to capitalize on a smaller proportion of heterosis.

These hybrids exhibit greater stability, demonstrating reduced production variability over the years and across specific geographic areas. They consistently respond predictably to environmental enhancements in more favorable years, while also avoiding significant production declines during unfavorable years43. This production stability is a valuable attribute for farmers, as it allows them to plan and predict the results of their harvests more reliably.

These results corroborate the findings of Crevelari et al.15, which demonstrated that the tested UENF hybrids exhibit greater stability. This consistency in performance reinforces the reliability of UENF hybrids under different cultivation conditions.

The hybrids developed by the State University of Northern Fluminense Darcy Ribeiro (UENF) have proven to be a promising alternative for maize cultivation, enhancing productivity among local producers. According to Pereira et al.24, these hybrids not only contribute to the region’s economy but also exhibit competitive characteristics in the national market. Since 1996, UENF has been developing hybrids specifically for the North and Northwest Fluminense regions through its genetic breeding program, addressing local production needs. The evaluation of adaptability and stability of hybrids is an essential step in the selection of superior cultivars in the context of plant genetic improvement. Among the innovations introduced, interpopulation hybrids present a technological alternative for producing seeds with good genetic potential at reduced costs. The applied breeding method requires fewer financial resources and less time, representing a viable option for local producers.

In this context, this manuscript presents research aimed at evaluating the performance of maize cultivars and recommending new hybrids for use in genetic breeding programs, aiming to simultaneously combine high yield and stability in different environments in the state of Rio de Janeiro. The findings of this research extend beyond scientific progress, directly impacting local agriculture and influencing future research directions. These innovations potentially open new avenues for the development of agricultural technologies, benefiting both producers and the community at large.

Conclusion

The application of the GGE biplot analysis allowed for the identification of the best maize hybrids for the state of Rio de Janeiro. UENF hybrids stand out for their stability, making them a viable option for producers in the North and Northwest Fluminense regions, as they can be grown under various conditions without significant fluctuations in productivity.