Comprehensive Assessment of the Genotype-Environment Interaction and Yield Stability of Boro Rice Genotypes under Four Environments in Bangladesh Using AMMI Analysis

Yield stability, alongside high yield potential and broad adaptation to various agroclimatic environments, is a key objective for rice breeders aiming to ensure food security. This study aimed to explore the most suitable and stable Boro rice genotypes for Bangladesh. Ten Boro rice genotypes underwent testing in four environments during the 2022 Boro season to investigate genotype-environment interaction (GEI) and yield stability performance. The experiment utilized three replications of a completely randomized block design. Yield stability performance was assessed through combined analysis and the additive main effects and multiplicative interaction (AMMI) model. The combined ANOVA revealed that the environment explained 10.23%, while GEI accounted for 9.17%, and the genotypes captured 80.60% of the variance, significantly impacting grain yield. Significance was observed in the environment, genotype main effects, and GEI. Analysis indicated that BRRI dhan 68 yielded the highest (6,754 kg·ha−1) and BRRI dhan 88 the lowest (5,620 kg·ha−1) among the investigated genotypes. In addition, genotypes BRRI dhan 84, BRRI dhan 81, and BRRI dhan 67 exhibited the highest grain yields. The Rangpur environment demonstrated considerable stability across the four environments with a high mean value of grain yield (7,206 kg·ha−1). Therefore, the AMMI model emerges as a valuable tool for identifying the most suitable and stable Boro rice genotypes with high-yielding potential across various regions in Bangladesh, as well as under diverse conditions.


Introduction
Rice (Oryza sativa L.) is a staple diet for most of the world's population, providing about 20% of the world's dietary energy needs as valuable cereal [1,2].Bangladesh has made tremendous advancements in the production of rice over the past 30 years and is regarded as a nation of rice cultivators and consumers in the world [3].According to research studies, the intensifcation of social, economic, and demographic environmental pressures will demand an additional 112 million metric tons of rice to meet global human intake by 2035 [4].In addition, Bangladesh's population is growing by two million each year and might add 30 million more people in the next 20 years.As the population forecast for Bangladesh is for an increase from the current 169 million to 220 million by 2050, the importance of rice production becomes even clearer.In recent years, however, overall proftability has declined due to rising input prices and increasing labour costs, which make rice cultivation more difcult [3].Reducing income volatility and increasing proftability is an important step towards increasing social welfare and the sustainability of rice production [5].Yield is closely linked to proftability, and increases in yield have helped rice farming to remain proftable, especially after 2005 in Bangladesh [6].Approximately 0.15% of the country's arable land is diverted from agriculture each year due to the demand for additional dwellings and business enterprises to meet the expanding population, and Bangladesh currently faces a tremendous problem in preserving food self-sufciency.To supply the growing demand for food in Bangladesh, it is crucial to utilize genotypic adaptability, current agricultural techniques, and water management [7].Terefore, genotypic adaptability to environmental change is critical for crop yield stability across locations and years.Furthermore, high yielding genotypes are critical to the food security program because they improve output potential and yield stability [8].
Te three seasons of Aus, Aman, and Boro are used to cultivate rice throughout the year in Bangladesh.Boro rice is one of the principle types and accounts for a substantial portion of the nation's overall rice production [9].However, Bangladesh has a great subtropical environment for Boro rice productivity, which is low in comparison to other Asian countries, e.g., Indonesia and Malaysia.According to the Bangladesh Bureau of Statistics [10], the target for cultivating Boro rice is set at 4,872,600 hectares for the fscal year 2021-2022.Until February 27, farmers had brought 27,07,572 hectares (76.61% of the target) under Boro cultivation.Te average Boro rice yield in Bangladesh varies depending on a number of variables, including the rice genotypes, the region, the season, inefcient nutrient management, disease, stresses such as drought, food, salinity, extreme temperature, low soil fertility, and agricultural practices used, all of which reduce rice yields and increase their variability as well as yield stability in Bangladesh [11].Terefore, extensive Boro rice cultivation is being made possible by improving crop management, increasing the application of appropriate fertilizer, pest management, and yield stability over a genotype of environments.
Yield stability is one of the primary objectives of improving genotypes with high potential for yield along with improved and broad adaptation to varied agroclimatic environments and it is the goal of rice breeders in a food security package because multilocation genotype evaluation enables plant breeders to assess a genotype's adaptation to a given environment as well as its stability over a range of habitats [12].A genotype's level of interaction with various growing conditions is usually utilized to assess how well it has adapted to various environments.As a result, a genotype is considered more adaptable or stable if it has a high mean yield with low variation for yield capability when grown in diferent environments or locations.However, in order to generate adaptive and high yield genotypes, an environmental evaluation must be conducted because GEI may afect yield and other performance variation [13].Furthermore, when making decisions about rice cultivation, farmers always consider the expected yields against production costs.However, due to fuctuations in the market price of rice [6], volatility in rice income has become the norm among rice farmers in Bangladesh [14].
For efective policy interventions, it is crucial for the government to analyze the cost, actual yield, proftability, and risk associated with rice cultivation under various growing conditions, such as diferent locations, varieties, input usage, irrigation sources, and planting dates.Consequently, to measure the stability and yield performance of rice genotypes, site-specifc genotypes must be selected through extensive environmental experiments.Multienvironment trials, based on yield performance and genotype stability across locations, can identify superior genotypes and ofer valuable guidance for breeders in selecting optimal genotypes [15].Tus, this study aimed to explore the most suitable and stable Boro rice genotypes for Bangladesh by assessing in four environments using yield stability and AMMI model.

Materials and Methods
2.1.Planting Material.Four environments, namely, Dinajpur (E 1 ), Rangpur (E 2 ), Mymensing (E 3 ), and Sylhet (E 4 ), were utilized for the investigation between January 2022 and April 2022 to assess Boro rice genotypes for yield-contributing traits over a single year.Geographic details of the four experimental locations, including their altitudes, latitudes, and longitudes, are depicted in Figure 1.Tis study involved 10 rice genotypes developed at the Bangladesh Rice Research Institute (BRRI), Gazipur, with genotype being the primary determinant and environments serving as subfactors.Te experimental design included three replications following a completely randomized block design.Each plot measured 4 meters by 2.5 meters, with a 1-meter gap between replications.Te spacing between each line was 30 cm.Seeds were sown in nursery plots on January 14, 2022, and after 35 days, the seedlings were transplanted to the main feld on February 17, 2022.
Te experimental feld received fertilization with 10 tons of cowdung, 170 kg of triple superphosphate (TSP), 250 kg of urea, 85 kg of muriate of potash (MoP), 150 kg of gypsum, and 10 kg of boric acid per hectare, following the fertilizer recommendation guide of 2018 in Bangladesh.During the fnal stage of land preparation, all fertilizers except urea, MoP, TSP, zinc oxide, gypsum, and boron were applied.In addition, urea was divided into three equal doses: one during the fnal land preparation phase, one-half during the vegetative stage, and the remaining half at maturity.Consistent weeding, thinning, irrigation, pest treatment, and other intercultural practices were implemented in all plots.Troughout the growth season, an efective drainage system was maintained to ensure the prompt fow of rainfall from the experimental plot.Te frst and second weedings were completed 15 and 35 days after planting, respectively.Simultaneously, thinning was performed to maintain a gap of 10 cm between plants.During the vegetative stage, pests such as rice bugs and green plant hoppers were observed in the crop, and chlorpyrifos 20% EC was applied at a concentration of 2 ml per liter of water to control the infestation.Te soil properties for each environment are detailed in Table 1.2(c)).

Statistical Analysis
2.4.1.Analysis of Variance.In the initial stages, a pooled analysis of variance was performed on the data.Table 2 provides the ANOVA's structure.

Stability Analysis.
According to Eberhart and Russell [17], the stability analysis (5)was completed.Te fundamental model used is as follows: where Y ij � mean of the i th genotype at j th environment, μ i � mean of the i th genotype over the environments, β i � regression coefcient of i th genotype to varying environmental indices, I i � environmental index, i.e., mean of all genotypes at j th environment minus grand mean, and σ ij � deviation from regression of i th genotype at j th environment.

Joint Regression Analysis.
Te formula for calculating the sum of squares for each source using joint regression analysis is presented in Table 3. Te impact of genotypes, environments, genotypes environments interaction, environment + (genotypes × environments) interaction environment (linear), and genotypes × environments (linear) on variance was assessed in comparison to pooled error.However, in order to test the pooled deviation (equation ( 6)), the pooled error was compared to the pooled deviation using the following formula: where r � number of replications.

Regression coefficient b
where where  j Y ij Ij 2 /  jIj 2 � variance due to regression.
A stable genotype is one that has a deviation (S 2 d ) that is equal to zero and a regression coefcient (b i ) that is equal to one (b i �1.0).Tese stability criteria, along with the mean value of the attributes, are used to determine how desirable a genotype is.Te regression coefcient (b i ) (equation ( 10

Range and Mean Variance Result of Ten Boro Rice
Genotypes.In the present investigation (Table 6), a comparison of the means across genotypes for various traits in four diferent environments revealed that E 1 (Dinajpur) exhibited the highest mean for panicle number per plant (17 no.), harvest index (45.95%),and the lowest mean for plant height (113.50 cm), tiller number per plant (8 no.), leaf area index (1.68%), and unflled grain per plant (240 no.).E 2 (Rangpur) displayed the highest mean for leaf area index (2.10%),flled grain per plant (790 no.), and grain yield (7,205 kg•ha − 1 ), while E 3 (Mymensing) showed the highest mean for unflled grain per plant (275 no.) and the lowest for flled grain per plant (630 no.).Te mean values for plant height, total dry weight per plant, total grain weight per plant, and 1,000-seed weight were 122.00 cm, 43.14 g, 27.33 g, and 43 g, respectively.
Furthermore, the results indicated that only the harvest index exhibited a better mean in the E 1 (Dinajpur) environment, whereas E 2 (Rangpur) and E 4 (Sylhet) exhibited superior mean values for leaf area index, flled grain per plant, grain yield, total dry weight per plant, total grain weight per plant, and 1,000-seed weight.Comparison of ranges (Table 6) across diferent traits in the four environments revealed that E 4 (Sylhet) exhibited the highest range for plant height (117-124 cm), total dry weight (38-45 g), total grain weight (22.95-30.46g), 1,000-seed weight (40-46 g), and harvest index (55-62).On the other hand, E 1 (Dinajpur) had the widest range only for panicle number per plant (15-21 no.).Te second widest range was observed in E 2 (Rangpur) for panicle length (24.40-29.0cm), leaf area index (1.71-2.39%),flled grain per plant (696-904 no.), and grain yield (6,038− 8,797 kg ha − 1 ).E 3 (Mymensing) exhibited the widest range for tiller number per plant (14-16 no.) and unflled grain per plant (214-368 no.).Te results highlighted the fexibility of all environments in terms of yield-contributing traits, with E 4 (Sylhet) and E 2 (Rangpur) showing the widest ranges for most of these traits, thus making them ideal environments for screening genotypes.

Scientifca
Table 3: Joint regression analysis of variance.
In this study, moderate GAM was obtained for plant height (23.81%), panicle length (19.06%), and harvest index (22.11%) in E 3 (Mymensing).Moderate GAM suggests a balanced infuence of both additive and nonadditive genes on the control of these traits.In addition, this study reported low GAM for total grain weight (5.90%) and leaf area index (1.92%) in E 2 (Rangpur).

Joint Regression Analysis of Variance.
Joint regression analysis for yield and its components has been performed in following the model provided by Eberhart and Russell [17], and the results are shown in Table 10.In the joint regression analysis, the pooled deviation of the mean sum of square was tested against the pooled error mean sum of square, and where it was signifcant, all the other mean square were tested against pooled deviation mean sum of square.If pooled deviation mean sum of square was nonsignifcant, then pooled error was used for testing the statistically signifcant of the mean square.Te joint regression analysis revealed signifcant variance due to genotype for all characters.Nonsignifcant diference among environments

Scientifca
(genotype × environment) was observed for all the characters.Nonsignifcant variance due to environment (linear) was also found for all the characters in the experiment.
Results showing variance due to genotype × environment (linear) was signifcant for plant height, tiller number per plant, panicle number per plant, total dry weight, total grain weight, 1,000-seed weight, harvest index, flled grain per plant, unflled grain per plant, and grain yield, while nonsignifcant for panicle length and leaf area index [19].

Stability Analysis.
Stability performance of a genotype is judged based on linear regression coefcient (b i ) and nonlinear deviation from regression (S 2 d ) for yield performance over the environments [2].According to Finlay and Wilkinson [20], regression coefcient (b i �1) indicated average stability, while greater than unity (b i > 1) indicated below average stability, and less than unity (b i < 1) indicated above average stability.Deviation from regression (S 2 d ) if signifcant, the performances of a genotype for a given environment may be predicted.In this study, the b i ranged from − 0.02 to 1.79.Genotype G 5 (BRRI dhan 67) is considered the most stable due to a b i value closer to unity.

AMMI Analysis.
From the result of AMMI analysis of variance (Table 12), there are highly signifcant diferences in the environment, genotype, and GEI, i.e., genotype environment interaction (GEI).Te combined ANOVA revealed that the environment had a considerable impact on grain yield due to signifcant variance at the 0.1% level, which accounted for 10.23% of the environment's total variation, while the genotypes and GEI accounted for 9.17% and 80.60%, respectively, of the total sum square.Te three main principle components of GEI factors were signifcant and together accounted for 100% of the overall GEI impact variance in grain yield.A total of 66.5% of the variation produced by the interaction was accounted for by the frst  10 Scientifca principle interaction component (IPCA1), followed by 23.7% (IPCA2) and 9.8% (IPCA3).Te frst two bilinear terms jointly accounted for 90.2% of the G × E sum of squares and used 20 of the total 27 degrees of freedom available in the interaction.Terefore, the frst two IPCA indicated the model has a strong relationship between the characters and ft to describe stability.

IPCA Interactions with AMMI Analysis.
Te relationships between the IPCA1 and IPCA2 and the mean of genotypes and environments are visually expressed using the AMMI analysis.A multivariate approach called principle component analysis can identify fgure arrangements as well as similarities and contrasts between the variables that have been established and arranged in a consecration operation of multivariate systems.Table 13 displayed the IPCA1 to IPCA3 scores that describe how a genotype interacts with various environments and how genotypes and environments relate to one another.A genotype that scores highly on the IPCA in a number of contexts must counteract negative interactions in other environments [21].Tese results demonstrate an uneven genotype response to the environment.Nevertheless, signifcant major interactions exist and are persistent for both positive and negative indications, as well as genotypes and environments with high IPCA scores.Te genotypes that do best in these environments and are stable are those with IPCA1 and IPCA2 scores of zero or almost zero.However, these genotypes interact very little across environments.Conversely, genotypes with negative IPCA1 and IPCA2 values did not interact across environments [22].As a result, among the ten genotypes, G 6 (BRRI dhan 68), G 7 (BRRI dhan 81), G 8 (BRRI dhan 84), G 2 (BRRI dhan 28), G 4 (BRRI dhan 100), and G 10 (BRRI dhan 29) were

Interaction of AMMI Model with PCA1 Values
Te IPCA1 and mean of genotypes with environments connection is visually expressed by the AMMI biplot (Figure 3).Te interaction pattern was best explained by the frst interaction principle component axis (AMMI component 1), which was highly signifcant.Te mean yield was displayed against the IPCA1 scores for the genotypes and the environments, respectively.Te relationships between the genotypes and the environments may be easily shown by plotting both the genotypes and the environments on the same graph.In the AMMI analysis, the IPCA scores of genotypes provide a measure of its environmental stability or adaptability.If the IPCA score is a relative number, the higher the score, positive or negative, the more specifcally a genotype has adapted to a certain environment.Te more closely the IPCA scores approximate zero, the more stable or adaptive the genotype was in the investigated environments as a whole.
Te "0" is a perpendicular line in this case.Te illustration demonstrates that genotypes with environments on the right side (both higher and lower) of the perpendicular line always have the greatest mean values of grain production.More high mean grain yield values are found in the upper right quadrant than the lower right quadrant, which has medium mean grain yield values.Te lowest mean grain yield fgures are shown on the left.Using the perpendicular line's performance as a benchmark, genotypes in the E 2 (Rangpur) environment with a high mean value of grain yield (7,206 kg•ha − 1 ) had high mean values and favourable interactions with IPCA1.Genotypes G 2 (BRRI dhan 28) and G 6 (BRRI dhan 68) had high mean values with the E 2 (Rangpur) environment and were stable genotypes with the environment.In addition, genotypes G 8 (BRRI dhan 84), G 7 (BRRI dhan 81), and G 5 (BRRI dhan 67) had high mean values and stability with the E 1 (Dinajpur) environment.Besides, G 10 (BRRI dhan 29), G 3 (BRRI dhan 89), and G 9 (BRRI dhan 88) had stability in E 3 (Mymensing) environment.In contrast, E 4 (Sylhet) is the poorest environment among the four, as shown on the left side of the perpendicular line, and G1 (BRRI dhan 58) had the best-suited genotype for this environment with a yield of 5,801 kg ha − 1 .Te E 1 (Dinajpur) and E 2 (Rangpur) environments are on the right side of the vertical axis, indicating rich environments, whereas the E 3 (Mymensing) environment and E 4 (Sylhet) environment are generally poor environments.Tus, the AMMI biplot shows that the studied genotypes difered from each other not only in their interactive efects but also in their mean grain yield values, as shown in Figure 3.

Interaction Biplot of AMMI Model with PCA1 and PCA2
Values.Te relationships between the mean of genotypes and environments and the IPCA1 and IPCA2 are visually expressed in the AMMI biplot (Figure 4).Table 13 displays the IPCA1 and IPCA2 scores that describe how a genotype interacts with various settings and how genotypes and environments relate to one another.According to Islam et al. [23], a genotype that scores highly on the IPCA in a number of contexts must neutralize negative interactions in other environments.Tese results demonstrate an uneven genotype response to the environment.However, both positive and negative signals, as well as genotypes and environments with high IPCA scores, exhibit strong, stable interactions.Te genotypes that do best in these contexts and are stable are those with IPCA1 and IPCA2 scores of zero or almost zero.However, these genotypes interact little across environments.Tere was no interaction between genotypes with negative IPCA1 and IPCA2 values across environments, and all of them had yields which were below average.Similarly, those genotypes have zero scores on the IPCA1, suggesting that the environment has less of an impact on them.12 Scientifca In addition, because they are used to stable conditions and are often adaptable to all environments, the genotypes indicated above typically provide an IPCA1 score close to zero.For grain yield, the biplot showed G 6 (BRRI dhan 68) had the highest mean value (6,754 kg•ha − 1 ) with E 2 (Rangpur) environment.Te maximum mean value was 7,206 kg•ha − 1 , and the interactions were strong.Te second highest mean value G 5 (BRRI dhan 67) with 6,073 kg•ha − 1 was shown in E 2 (Rangpur) environment.Among the experiments for grain yields G 1 (BRRI dhan81) and G 4 (BRRI dhan 89) with the environment E 4 (Sylhet) were the most unstable and discriminate while G 2 (BRRI dhan 28) with mean grain yield (6,406 kg•ha − 1 ) was the greater yieldperforming genotype in that particular environment.Similar genotypes and environments were represented by related symbols of the IPCA1 score, which suggested a positive association and, as a result, a higher yield of the genotypes in that specifc environment.Te E 1 (Dinajpur) environment had positive IPCA1 and IPCA2 scores and registered above average yields and the stable environments for most of the genotypes.While E 4 (Sylhet) had negative IPCA2 values, thus the Sylhet environment was unfavourable for most of these genotypes.

Discussion
In the present study, ten Boro rice genotypes underwent evaluation.Te partitioning of total variation revealed signifcant genetic variability and marked diferences across the tested traits for all genotypes.Consequently, identifying superior genotypes based on specifc morphological attributes positively correlated with grain yield is feasible.Field studies were conducted in geographically and temporally distinct environments, resulting in substantial variance attributed to genotype-environment interaction (GEI).Hence, understanding the sources of variation in genotype performance across diferent environments was deemed crucial.Akter et al. [24], Sharif et al. [25], Islam et al. [23], and Kanfany et al. [26] also noted that signifcant variation in GEI implied the presence of multiple mega-environments, where various genotypes excel in yield performance.Given the diverse grain yield performances of diferent genotypes across environments, mean comparison analysis sufced for identifying genotype diferences [11,27].Comparing means and ranges across environments provides insights into environmental suitability for trait expression.However, the E 3 (Mymensing) environment exhibited inferior mean performance across traits compared to other environments, making it a less favourable option for achieving higher yields.Conversely, E 2 (Rangpur) and E 4 (Sylhet) environments showed promising prospects for cultivation yield.Te results highlighted the fexibility of all environments in accommodating a range of yield-contributing traits.Notably, E 4 (Sylhet) exhibited the widest range across most traits, while E 2 (Rangpur) demonstrated the widest range for four of the yield-contributing traits.Terefore, these two environments were identifed as ideal for genotype screening.
Simple measurements of variability, particularly the genotypic and phenotypic coefcients of variation, are commonly utilized for genotype evaluation [28].Tese relative values ofer insights into the extent of genetic diversity within a population.Consequently, the variance components, including genotypic and phenotypic variances, were computed.Te phenotypic variance was notably greater than the genotypic variance, indicating the infuence of both genetic factors and environmental conditions on the expression of the trait under scrutiny.Tewachewu et al. [29] and Hasan et al. [28] reported low variance components for panicle length.Upon analyzing the data above, it became evident that genetic parameters of variance were estimated for all traits across each environment.Tis observation highlighted high values of σ 2 g and σ 2 p for plant height, panicle number per plant, leaf area index, flled grains per plant, unflled grains per plant, and grain yield.Previous studies by Amegan et al. [30] for grain yield, Limbani et al. [31] for flled grains per plant, and Asante et al. [32] for plant height corroborate these fndings.Similarly, Islam et al. [23] recorded analogous results for upland rice genotypes.
Te genotypes demonstrated their genetic potential as GCV was slightly lower than the PCV, indicating that genetic factors predominantly infuenced the observed characteristics over environmental infuences.Traits such as tiller number per plant, harvest index, flled grains per plant, and grain yield consistently exhibited high levels of both GCV and PCV.Comparable fndings have been reported by Demeke et al. [33] and Nikhitha et al. [34] for tiller number per plant, and by Fathima et al. [35] and Bhargava et al. [36] for flled grains per plant.Conversely, traits such as total grain weight, panicle length per plant, and unflled grains per plant exhibited low levels of both GCV and PCV, suggesting limited genetic variation and thus rendering them less suitable as selection criteria for yield improvement.Introduction of diversity through methods such as mutation and recombination, followed by selection, may, therefore, be necessary for enhancing traits such as panicle length [33].Traits such as plant height, panicle number per plant, total dry weight, 1,000-seed weight, and leaf area index displayed moderate levels of both GCV and PCV.Tese moderate values suggest that the genotypes possess adequate genetic diversity to facilitate improvement through selection.Similar results were also reported by Lakshmi et al. [37], Singh et al. [38], and Pradhan et al. [39].
Heritability serves as a valuable indicator of how traits are inherited from parents to ofspring.Assessments of heritability play a crucial role in determining genetic gains and predicting the accuracy of phenotypic value as a breeding value [33].A higher percentage of heritability indicates greater variability that is amenable to manipulation.Tis increased fexibility in variability allows for the potential alteration of trait expression through methods such as mass selection and progeny testing.Previous studies have consistently reported high heritability estimates for most of the studied traits.Ray et al. [40] documented high heritability for panicle length per plant and plant height, while Bhargava et al. [36] and Fathima et al. [35] reported high heritability for 1,000-seed weight, tiller number per plant, and flled grains per plant.Similarly, Nikhitha et al. [34] and Singh et al. [38] found high heritability for tillers per plant, flled grains per plant, and grain yield.Notably, panicle length exhibited high heritability (69.49%) and moderate GAM (12.88%), consistent with fndings reported by Demeke et al. [33], Bhargava et al. [36], and Singh et al. [38].
Genetic advance serves as a pivotal indicator of the potential development achievable through the selection of a suitable population [23].High GAM results signify the capacity for genetic improvement through selection and the maximum control exerted over traits via additive gene action.Consistent fndings have been reported by Akabari et al. [41], Abebe et al. [42], Kumari et al. [43], Tripathi et al. [44], and Ray et al. [40].Across various parameters examined in this research, such as tiller number per plant, 1,000-seed weight, flled grains per plant, unflled grains per plant, and grain yield, both high heritability and high GAM were observed.Tis indicates that nonadditive gene efects have minimal infuence on these traits, while the infuence of additive genes remains strong.Consequently, these attributes may be readily improved through simple selection processes.Tese fndings corroborate the strong heritability and high genetic advance as a percentage of the mean observed for grain yield, as reported by Lakshmi et al. [37].In contrast, moderately high heritability and moderate GAM were observed for plant height and panicle length.Similar results have been documented in rice by Nikhitha et al. [34], Bhargava et al. [36], and Singh et al. [38].Low heritability coupled with low genetic advance as a percentage of the mean were noted for leaf area index and total grain weight.Te b i value, being less susceptible to environmental changes, demonstrates greater adaptability when its value is substantially lower.However, if the b i value is negative, the genotype may only thrive in poor environments.
Te frst two IPCAs in the AMMI analysis indicated a signifcant association between the studied attributes and the model's capability to describe stability [21,45].Consequently, biplots generated by combining genotypic and environmental scores of the AMMI 1 components provide breeders with a comprehensive understanding of how genotypes, environments, and their interaction (G x E) behave.Tis study demonstrated that genotypeenvironment interaction (GEI) signifcantly contributed to rice yield variance, with AMMI biplots efectively illustrating how genotypes responded to diferent environments [46].Te key fndings, based on AMMI analysis and yield stability, indicated that four of the tested genotypes (G 5 , G 6 , G 7 , and G 8 ) performed best in environments E 1 (Dinajpur) and E 2 (Rangpur).Consequently, these genotypes should be considered for the Boro rice production area and as a parental base for breeding programs in Bangladesh.

Conclusions
To identify high-yielding Boro rice genotypes adaptable to diverse regions in Bangladesh, the AMMI statistical model proves invaluable.Our study revealed that genotype G 6 (BRRI dhan 68) achieved the highest mean yield (6,754 kg ha − 1 ), contrasting with G 9 (BRRI dhan 88) as the lowest (5,620 kg ha − 1 ).Notably, G 8 (BRRI dhan 84), G 7 (BRRI dhan 81), and G 5 (BRRI dhan 67) consistently exhibited top grain yields.Environment E2 (Rangpur) displayed exceptional stability, yielding 7,206 kg•ha − 1 on average.Most genotypes exhibited specifcity to certain environments, as depicted in biplots showcasing genotypeenvironment interaction.Te AMMI model categorized environments into four sections, highlighting optimal choices-G 5 , G 6 , G 7 , and G 8 -for areas such as Dinajpur (E 1 ) and Rangpur (E 2 ).Tese genotypes hold promise for Boro rice production and future breeding programs in Bangladesh.For future research, additional variables such as fertilizer application, plant diseases, and management systems should be considered.

2 Scientifca 2 . 2 .
Test Environment with Soil Characteristics.In the multienvironment experiment, the study was conducted across four distinct environments, and genotypes were evaluated for one season in four locations, accounting for variations in crop quantity and quality due to factors such as animal presence, pests, and diseases.Diferentiating factors in the study included soil type, site conditions, planting time, and season.Genotypic responses were assessed based on parameters such as temperature, relative humidity (RH), sunlight exposure, and rainfall.Each combination of planting site and season was considered a unique environment for evaluating genotypic responses.

Figure 2 :
Figure 2: Maximum (a) and minimum (b) temperature, and rainfall (c) ranges for four environments in Bangladesh.

Figure 3 :
Figure 3: AMMI 1 biplot using IPCA1 and mean grain yield data for ten Boro rice genotypes in four environments.

Figure 4 :
Figure 4: AMMI 2 biplot using IPCA1 and IPCA2 scores data for grain yield with ten Boro rice genotypes in four environments.

Table 1 :
Diferent environments with soil physiological properties.

Table 2 :
ANOVA when genotypes were raised in diferent environments for one year.

Table 4 :
Combined analysis of pooled variance for various morphological traits in ten Boro rice genotypes.

Table 5 :
Grain yield of ten Boro rice genotypes among diferent environments.
* * � signifcant at p < 0.01.Diferent superscript letters indicate signifcant diference within column.Bold values are the highest grain yield from each environments.

Table 7 :
Variance for genotypes and variance for phenotype for yield and component traits in ten Boro rice genotypes in each environment.
Note.E 1 � Dinajpur, E 2 � Rangpur, E 3 � Mymensing, and E 4 � Sylhet.Bold values are the highest variance for genotype and phenotype among four environments.

Table 8 :
Genotypic and phenotypic coefcient variance for yield and component traits in ten Boro rice genotypes in each environment.Note.E 1 � Dinajpur, E 2 � Rangpur, E 3 � Mymensing, and E 4 � Sylhet.Bold values are the highest genotypic and phenotypic coefcent variance among four environments.

Table 9 :
Broad sense heritability, genetic advance, genetic advance as a percentage of mean for yield, and component traits in ten Boro rice genotypes in each environment.Note.E 1 � Dinajpur, E 2 � Rangpur, E 3 � Mymensing, and E 4 � Sylhet.Bold values are the highest broad sense heritability, genetic advance, and genetic advance as a percentage of mean among four environments.

Table 10 :
Joint regression analysis over diferent environments of yield characters.

Table 11 :
Stability parameters (b i and S 2 d ) of ten Boro rice genotypes for plant height, tiller number per plant, panicle number per plant, panicle length, leaf area index, and harvest index.

Table 12 :
AMMI analysis of variance including the partitioning of GEI of grain yield.

Table 13 :
AMMI analysis showing means with IPCA1, IPCA2, and IPCA3 scores of grain yield for ten Boro rice genotypes grown in four environments.