Heritability and Selection Using GGE Biplots and the Sustainability Index (SI) of Maize Mutants under Different Cropping Systems in Upland

Abstract

A good maize plant breeding program must consider the effect of genotype-by-environment interactions (GEIs) and the correlation of important traits. The purpose of this study was to evaluate genetic variation, identify and investigate the implications of GEIs on breeding maize mutants in order to determine the ideal genotypes (stable and high yield), identify the best intercropping system for maize mutants, and identify the heritability and relationships of the traits tested that are important in cultivation. This research was carried out in five different intercropping systems in Upland West Java, Indonesia. A randomized block design with two replications was used in the field experiment. The measurement results revealed that the genetic diversity of maize mutants is broad, with six axes having eigenvalues ranging from 8.76 to 1.07 and a cumulative value of 76.64%. The neighbor-joining tree also showed a wide range of variation, yielding five distinct groups. The GEIs had a significant effect on the 14 traits tested, particularly yield. The environment had a significant impact on the variation of all the traits tested. The ideal cropping system for selecting the ideal maize mutant was Environment 4 (maize + rice). Superior maize mutants were successfully selected based on the GGE biplot. Thus, the sustainability index (SI) was used to successfully select maize mutants that were superior by 20.00%. These two methods selected only 15.00% of the maize mutants. One trait has a low heritability value, five traits have a moderate heritability value, and 17 traits have a high heritability value. The yields correlated positively and significantly with nine yield traits. Future maize-breeding programs can be based on data on genetic diversity, heritability, and the relationship of the traits tested. The best intercropping system can be used as an alternative for increasing maize cultivation income. Selected maize mutants can be proposed as new superior genotype candidates.

1. Introduction

Maize (Zea mays L.) is one of the most crucial grain crops in the world, especially in Indonesia. This crop has an important role as a source of food, feed, employment, and profits for farmers [1]. The plant is widely cultivated in tropical countries like Indonesia. Although the research and development on maize have been extensive, maize yields in the tropics still need to be improved. This entails a consistent and continuous selection process to develop crop yields. In addition, strict selections to obtain stable genotypes with high yields in extensive field environments need to be made [2]. Selection is a very important part of genetic improvement in maize.

The genetic improvement of a plant is centered on the strength of genetic diversity within plant species. High variability provides options for origin selection, which is made to enhance the probability of hybridization [3]. In order to accelerate the development of crop improvement, genotypic correlations have been employed as a powerful technique to identify the links among agronomic variables in genetically varied populations [1,4]. Information regarding the relationships among characteristics in plants is vital for effective and rapid selection during plant development [5].

Estimating heritability is a crucial technique in crop genetic development. It aids breeders with allocating the required resources to successfully select desired features and obtain the greatest genetic gain in a relatively short amount of time [1,6]. There are various methods to calculate heritability. This can be estimated in broad or narrow terms, in single plants, individual plots, or averages [4,7]. The most favorable settings for selection are those with high heritability estimates and high genetic advancements [1,8]. Additionally, it shows that the characteristic contains additive genes and further argues that crop improvement can be achieved by selecting for a trait. It is more accurate and significant to estimate heritability by genetic gain than to take individual parameters into account [4,9].

One of the efforts to increase maize production in the tropics is by planting superior varieties produced from plant breeding programs. Mutational breeding in maize is one way to obtain new superior varieties with high yields that are adaptive to a wide range of environments. However, the continuous use of the same plants on the same land can lead to the consumption of high amounts of water and fertilizers [10]. The excessive use of chemical fertilizers can cause soil damage and environmental degradation [11]. Efforts to maintain natural soil quality are needed in order to continue to produce optimal yields. One way of doing so is by intercropping.

Intercropping is an environmental modification that can be carried out to increase farmers’ income and maintain soil fertility. This technique is a form of modification to the natural growing environment through cultivation techniques that have been developed for generations in developing countries [12]. Intercropping can maintain soil fertility and reduce crop failure and the differentiation of nutritional sources [13,14]. The intercropping of maize with other plants prioritizes the concept of modifying natural and integrated cultivation techniques. This technique has been used by many researchers, such as in the intercropping of maize with chili pepper [15] and maize with soybeans [11,14].

Planting maize in various intercropping systems can lead to genotype-by-environment interactions. This interaction between maize and its growing environment must complement each other in order to produce optimal yields [10]. Intercropping between maize and other crops must produce optimal yields for maize as the main crop and the least reduced yields for the secondary crops. Therefore, it is necessary to choose maize that is adaptive for cultivation in an intercropping system.

The continuous improvement of maize is essential for increasing competition in the crop. This can be achieved by selecting suitable parental materials with significant genetic variability. The stability of most maize morphological traits, the expression of information, and the correlation of the different characteristics are beneficial to selecting superior genotypes. In order to improve breeding efficiency for maize traits, it is necessary to recognize the variation in these traits in various environments. The main objective of the multi-environment trial was to observe genotype stability in the whole environment and identify superior genotypes and locations that are the most targeted representative environment for production [12]. Up to this point, several genetic studies have been conducted on several traits in maize [1,16], Pinus elliottii [5], roses [8], cowpea [6], and Mason pine [7]. This study aimed to evaluate genetic variation, identify and investigate the implications of GEIs on maize mutants breeding to determine the ideal genotypes (stable and high yield), identify the best intercropping type for maize mutants, and identify the heritability and relationships of the traits tested, which are important in cultivation.

The article was structured as follows: the first part of the article provides background information and the aims and objectives of the research. The second part provides information on the plant materials used and the methodology of data analysis, which includes the analysis of genetic diversity and heritability, a combined analysis of variance (ANOVA), stability analyses using GGE biplot, as well as the relationships between the traits tested using correlation analysis and GT biplot. The third part contains the resulting findings, including the interpretation and significance of the findings of the analysis. The last part contains the conclusions and recommendations from the analysis, as well as suggestions for further research.

2. Materials and Methods

2.1. Plant Materials and Experimental Design

The plant materials used were 60 maize mutants (

Table 1. The maize mutant materials used in the experiment.

No.	Code	Mutants	No.	Code	Mutants
1	M1	MDR 1.1.12	31	M31	MDR 10.2.2
2	M2	MDR 1.1.3	32	M32	MDR 12.3.1
3	M3	MDR 1.2.3	33	M33	MDR 12.3.2
4	M4	MDR 1.6.3	34	M34	MDR 14.1.1
5	M5	MDR 3.1.10	35	M35	MDR 14.2.1
6	M6	MDR 3.1.2	36	M36	MDR 14.2.2
7	M7	MDR 3.1.4	37	M37	MDR 14.3.1
8	M8	MDR 3.6.1	38	M38	MDR 14.3.8
9	M9	MDR 4.1.3	39	M39	MDR 14.3.11
10	M10	MDR 4.2.2	40	M40	MDR 14.5.1
11	M11	MDR 4.7.2	41	M41	MDR 16.1.1
12	M12	MDR 4.8.8	42	M42	MDR 16.5.15
13	M13	MDR 5.4.1	43	M43	MDR 16.6.14
14	M14	MDR 5.5.1	44	M44	MDR 16.7.1
15	M15	MDR 7.1.2	45	M45	MDR 18.3.1
16	M16	MDR 7.1.7	46	M46	MDR 18.4.1
17	M17	MDR 7.1.9	47	M47	MDR 18.5.1
18	M18	MDR 7.2.3	48	M48	MDR 18.8.1
19	M19	MDR 7.2.5	49	M49	MBR 153.1.2
20	M20	MDR 7.3.1	50	M50	MBR 153.3.2
21	M21	MDR 7.3.2	51	M51	MBR 153.4.1
22	M22	MDR 7.4.1	52	M52	MBR 153.6.1
23	M23	MDR 7.4.2	53	M53	MBR 153.7.1
24	M24	MDR 8.5.3	54	M54	MBR 153.9.8
25	M25	MDR 8.6.1	55	M55	MBR 153.10.1
26	M26	MDR 8.6.3	56	M56	MBR 153.10.2
27	M27	MDR 8.8.1	57	M57	MBR 153.11.1
28	M28	MDR 9.1.3	58	M58	MBR 153.13.1
29	M29	MDR 9.1.5	59	M59	MBR 153.14.1
30	M30	MDR 9.4.1	60	M60	MBR 153.15.1

). The research was conducted in Arjasari, West Java, Indonesia, in five intercropping types, namely Environment 1 (sole crop), Environment 2 (Maize + Soybean), Environment 3 (maize + sweet potato), Environment 4 (maize + rice), and Environment 5 (maize + chili pepper). The experiment used a randomized block design with two replications.

2.2. Data Collection

Data collection was carried out on vegetative traits, flowers, yields, and yield components. The traits tested were plant height (PH), cob height (CH), stem diameter (RD), number of leaves (NoL), leaf length (LL), leaf width (LW), chlorophyll content 1 (CC1), chlorophyll content 2 (CC2), male flowers (MF), female flowers (FF), cob weight (CW), cob weight without cobs (CWWS), number of cobs per plot (NoCPP), cob length (CL), cob diameter (CD), number of rows seeds per cob (NoRPC), number of cobs (NoSPC), Slender diameter (PD), cobs weight 3 samples (CW3S), seed weight 5 samples (SWPC), seed weight per plot (SWPP), weight 1000 seeds (1000SW), Yield (GY). Data collection was adjusted to the descriptor of maize [17].

2.3. Data Analysis

The characters were compared per genotype across the four environments. A randomized complete block design was employed in the analysis of all the traits inside and across the settings using PBStat [18] in order to estimate the variance components of traits. The means of the genotypes, the genetic and phenotypic variances, and the heritability were calculated for each of the attributes using an analysis of variance (ANOVA).

2.4. Multivariate Analyses

Using principal components analysis (PCA), the trait data were graphically analyzed across the five environments. Mean values were used for this analysis. PCA biplots made it possible to examine the association and opposition between characteristics and genotypic variation on a multivariate scale.

2.5. Genotype by Environment Interactions and Stability Analysis

The following equation represents the combined ANOVA statistical model to calculate GEIs:

Y_ijkl = μ + G_i + E_j + GE_ij + R_k(j) + B_l(k) + ε_ijkl

where Y_ijkl represents the value of mutant i in plot l and the value of each replication in environment j; μ stands for the grand mean yield; G_i stands for the effect of mutant i; E_j stands for the effect of the environment j; GE_ij stands for the effect of genotype by environment interactions on mutant i and environment j; R_k(j) stands for the effect of the replicate k on the location j; B_l(k) stands for the effect of replication k on plot l; and ε_ijkl is the error effects from mutant i in plot l and repeat k of environment j, respectively. Genstat 12th was used to generate the combined ANOVA.

The variation resulting from the genotypes and interaction between the genotypes and environments (GEIs) is explained by the genotype plus the genotype × environment (GGE) calculation. The GGE biplot model, as represented in [19], uses the following formula:

$${\mathrm{Ῡ}}_{mn}-{\mathsf{\mu }}_m={\mathsf{\beta }}_n+{\sum }_{k=1}^t{\mathsf{\lambda }}_o{\alpha }_{mo}{\gamma }_{no}+{\epsilon }_{mn}$$

where Ῡ_mn; μ_m; β_n; k; λ_o; α_mo; γ_no; and ε_mn are the appearance in environment ‘n’ from mutant ‘m’; the whole average yield; the effect of environment ‘n’; the number of primer components; the singular value from primer component ‘o’; the value of mutant ‘m’ and environment ‘n’ for primer component ‘o’; and the error of the mutant ‘m’ in environment ‘n’, respectively. The GGE biplot analysis was carried out using PBstat [18].

The formula below, which was employed by us, was used to estimate the sustainability index (SI) [20]:

$$SI=\left[\frac{\left(Y-\mathsf{\sigma }n\right)}{YM}\right]\times 100$$

where Y is a maize mutant’s mean yield performance, σn is the standard deviation, and YM is its greatest yield performance across all cropping systems. Five categories—very low (up to 20%), low (21–40%), moderate (41–60%), high (61–80%), and very high (beyond 80%)—were arbitrarily chosen to categorize the SI values [21]. SI was computed with Microsoft Excel 2013.

2.6. Heritability Analysis

The following equation was used to compute the broad-sense heritability (H²) from across environments and within each environment:

$$H^2=\frac{\left(S_g^2\right)}{\left[S_g^2+\left(\frac{S_e^2}{r}\right)\right]}$$

Here S_g² and S_e² stand for the genetic and residual variance for each environment, respectively, and r is the number of replicates of each genotype.

2.7. Relationship between Traits

2.7.1. Genotype by Traits (GT) Biplot Analysis

The connection between each attribute studied was determined using the GT biplot technique, and the best maize mutant genotypes were chosen based on each trait [22]:

$$\frac{T_{ij}-T_j}{sj}={\lambda }_1{\zeta }_{i1}{\tau }_{j1}+{\lambda }_2{\zeta }_{i2}{\tau }_{j2}+{\epsilon }_{ij}$$

where T_ij is the mean of the ith genotype for the jth trait, T_j is the mean of the jth characters in all genotypes, and sj is the standard deviation of the jth trait among the genotype averages; ζ_i1 and ζ_i2 were the 1st and 2nd principal component (PC1 and PC2) values for the ith genotype, τ_J1 and τ_J2 were the PC1 and PC2 scores for jth character, respectively, and ε_ij is the model error related to the ith genotype and jth trait.

2.7.2. The Study of the Relationship between Each Character

Based on the Pearson correlation, the relationship between each tested attribute was computed. The following equations were used:

$$r_{xy}=\frac{\sum _{i=1}^n{x}_i{y}_i-\sum _{i=1}^n{x}_i\sum _{i=1}^n{y}_i}{\sqrt{\sum _{i=1}^n{x}_i^2-{\left(\sum _{i=1}^n{x}_i\right)}^2}\sqrt{\sum _{i=1}^n{y}_i^2-{\left(\sum _{i=1}^n{y}_i\right)}^2}}$$

where r_xy is the association coefficient. x is the variable (trait) x, and y is the variable (trait) y. The relationship is very strong if the value is 1 or −1. The connection between the two variables is in a state of deterioration if the value is near 0.0. Using the online statistical program PBStat, the correlation between each trait was examined, and Ms. Excel 2013 was used to examine the GT biplot.

3. Results and Discussion

3.1. Genetic Diversity of Maize Mutants Based on Principal Component Analysis (PCA)

Principal component analysis (PCA) was used to identify the traits that affect the genetic diversity of maize genotypes. This method can help reduce the number of variables in the data held to assist the selection process [23]. According to Singh [24], PCA can simplify complex data by converting a large number of variables into a small number of variables, which are called principal components. In the main component (PC) resides an eigenvalue, where the eigenvalue can explain the cumulative factor and diversity, which is more than or equal to one [25,26]. Therefore, the number of PCs used was determined by more than one eigenvalue.

The first six principal components (PCs) with eigenvalues greater than one accounted for 76.64% of the variability amongst the 60 maize mutants (

Table 2. Eigenvalue, percentage, and cumulative of agro-morphologycal traits.

	PC1	PC2	PC3	PC4	PC5	PC6
Eigenvalue	8.76	2.64	2.01	1.73	1.41	1.07
Variance (%)	38.10	11.48	8.72	7.53	6.13	4.66
Cumulative (%)	38.10	49.59	58.31	65.84	71.97	76.64

PC = Principal component.

). The results of the PCA measurements for the 60 maize mutants tested based on agro-morphological traits showed six axes with eigenvalues between 8.76 and 1.07, with a cumulative value of 76.64% (Table 2). The PCA measurements for the genetic diversity of the 60 tested maize mutants showed a variation in the first component (PC1) of 38.10%, with the influential traits being cob height, CC1, cob weight, cob weight without cobs, number of cobs per plot, cob length, cob diameter, number of seeds per cob, cob weight of 3 samples, seed weight of 5 samples, seed weight per plot, weight of 1000 seeds, and yield.

In the second component (PC2), the variation contribution was 11.48%, influenced by the male and female flowers. In the third component (PC3), the contribution of the variation was 8.72%, influenced by the length of the leaf. In the fourth component (PC4), the stem diameter characters, CC1 and CC2, contributed to a variation of 7.53%. In the fifth component (PC5), the variation contribution was 6.13%, with the influential character being the number of cobs per plot, while in the sixth component (PC6), the variation contribution was 4.66%, but no characters contributed to it. Based on the data obtained, it can be seen that the yield and yield components dominate the causes of variation. The trait values that affected the genetic diversity of the 60 maize mutants are presented in

Table 3. Trait values that influenced the diversity of 60 maize mutants.

Traits	Principal Component (PC)
	1	2	3	4	5	6
Plant height (PH)	0.45	−0.27	0.73	−0.14	0.13	0.03
Cob height (CH)	0.58	0.34	0.41	−0.23	0.22	−0.16
Stem diameter (RD)	0.25	0.12	0.49	0.58	−0.18	0.21
Number of leaves (NoL)	0.41	0.14	0.42	−0.13	0.04	−0.44
Leaf length (LL)	0.36	0.35	0.69	0.15	0.04	0.08
Leaf width (LW)	0.46	0.40	0.09	0.40	−0.10	0.30
Chlorophyll content 1 (CC1)	0.52	−0.25	0.14	−0.51	0.38	0.05
Chlorophyll content 2 (CC2)	0.42	−0.03	−0.01	−0.66	−0.08	0.33
Male flowers (MF)	0.10	0.86	−0.23	−0.13	0.22	−0.09
Female flowers (FF)	0.10	0.83	−0.27	−0.12	0.30	−0.12
Cob weight (CW)	0.92	0.10	−0.12	0.01	−0.22	−0.14
Cob weight without cobs (CWWS)	0.90	0.09	−0.17	0.11	−0.20	−0.15
Number of cobs per plot (NoCPP)	0.55	0.18	−0.05	0.01	−0.54	−0.27
Cob length (CL)	0.76	−0.11	−0.01	−0.14	−0.13	0.03
Cob diameter (CD)	0.67	−0.19	−0.13	0.13	0.33	0.27
Number of rows of seeds per ear (NoRPC)	0.40	−0.38	−0.14	0.41	0.47	−0.22
Number of cobs (NoSPC)	0.66	−0.41	−0.14	0.14	0.20	−0.31
Slender diameter (PD)	0.38	0.32	−0.18	0.27	0.38	0.30
Cobs weight: 3 samples (CW3S)	0.83	−0.25	−0.28	0.02	−0.13	0.13
Seed weight: 5 samples (SWPC)	0.83	−0.17	−0.10	0.06	0.12	0.17
Seed weight per plot (SWPP)	0.84	−0.09	−0.13	0.09	0.02	−0.09
weight 1000 seeds (1000SW)	0.71	0.10	−0.09	−0.28	−0.24	0.30
Yield (GY)	0.94	0.10	−0.09	0.04	−0.09	−0.11

Note: Numbers in bold indicate a discriminant of >0.5 or <−0.5 that contributed to variance [27].

.

Based on Table 3, there were traits that made positive and negative contributions. According to Haydar et al. (2007) [25], the positive traits showed that they contributed maximally to diversity, while the negative contributing traits indicated that they contributed but not optimally to diversity. Therefore, all traits in PC1 contributed maximally to diversity. In PC2, the traits that had the maximum contribution were the male and female flowers. In PC3, the trait that had the maximum contribution was leaf length, while in PC4, the trait that had the maximum contribution was stem diameter. Solankey and Singh (2018) [28] used PCA to determine the maximum contribution of traits toward sweet potato diversity, especially those related to yield productivity. In other studies, PCA was also used to classify models of static and dynamic stability [2,29,30], as well as classify and determine the environmental factors that had the most influence on maize yields [31].

The PCA biplots presented include PC1–PC2 (Figure 1a), with a cumulative value of 49.59%, PCA biplots for PC1–PC3 (Figure 1b), with a cumulative value of 46.82%, PCA biplots for PC1–PC4 (Figure 1c), with a cumulative value of 45.63%, and PCA biplots for PC1–PC5 (Figure 1d), with a cumulative value of 44.23%, while PC1–PC6 cannot be represented in a biplot pattern because PC6 does not contain a dominant trait for variation. The PCA biplot uses dots to represent the observed scores for the main components and uses vectors to represent the variable coefficients for the main components [27].

In this study, the dots represent the 60 maize mutants tested, and the vectors represent the 24 traits measured. Those maize mutants that are close to each other or coincide with each other have a high similarity, and vice versa (Figure 1). The value of PC1–PC2 has a higher total variation compared to PC1–PC3, PC1–PC4, and PC1–PC5. Some researchers reported that the plots of the first two PCs revealed an interesting structure in the data, where these two PCs have the greatest contribution to genetic variation [26,32]. This shows that the PC1–PC2 biplot is more representative of the genetic variation in the 60 maize mutants tested, where the two PCs represent a total variation of 49.59%. The strong similarity between the PCs for the correlation and covariance matrices is due to the almost identical variances for all the variables tested. The first component is interpreted as an overall rating index, although it has a negative coefficient [27]. The shape of the PC1 can be predicted from the correlation matrix. In addition, a positive PC1 coefficient is expected because it has a very large contribution to genetic variation. The second PC (PC2) also has the largest coefficients for the MF and FF properties, but the other coefficients cannot be ignored either. However, this contradicts the implicit assumption that ‘n’ in the observations is identically distributed with a common mean and covariance matrix. Most of the ‘structure’ in the data suggests that the different observations have different means and that PCA seeks the main direction of variation between the means rather than the main direction of the variation in the general distribution [27].

In other studies, the PCA graph was used to identify the correlation between traits from the position of the vectors. The angle formed between the vectors helped to estimate the correlation among the traits [33]. Trait vectors that form an acute angle (closer to 0°) indicate that these traits are strongly and positively correlated, while those that form an obtuse angle (closer to 180°) indicate that these traits are strongly and negatively correlated [34]. The position of the trait vectors and the mutant points indicate the size of the traits in the tested mutants. When a mutant is close to a certain trait, this indicates that the mutant has a high value for that trait [27].

3.2. Genetic Diversity Based on Neighbor-Joining Tree

The distribution of the unrooted group in the maize mutants is shown in Figure 2. The phylogenetic signal value for this grouping is g = 2, indicating the proximity information between the mutants tested [35]. There were five small groups, each consisting of several maize mutants. The group consisted of at least five mutants, and they were M24, M12, M4, M9, and M40. The most numerous group consisted of 20 mutants, namely M34, M36, M32, M42, M60, M37, M53, M57, M11, M58, M35, M44, M43, M26, M38, M45, M59, M2, M55, and M14.

The output provided in the form of a visualization for an unrooted tree is associated with a single output from the neighbor-joining tree algorithm. Our study involved a morphological dataset consisting of 24 traits (vegetative, floral, yield, and yield components). In this case, a pairwise distance matrix was used to determine the stable and unstable sub-trees when many alternatives were considered to explain the data used [36]. In other cases, the conclusions were drawn based on the empirical limits of the total tree size [37,38]. The single output of the resulting neighbor-joining tree algorithm shows the genetic variation of all the traits tested.

Several factors, such as the genetic similarity of some traits, can cause this grouping. This grouping can also depict the relations among species or varieties and infer each gene’s lineage [39]. Based on the picture, the connection between the mutant and its parents will be found in each cluster, possibly developing characters in the same group. Information about genetic diversity plays an important role in the plant breeding process. If the genetic diversity is wide, then the progress of selection will be successfully attained. This will enable the achievement of sustainable agriculture.

3.3. Genotype-by-Environment Interactions on Vegetative Traits, Flowering, Yield Components, and Yield of Maize Mutants in Five Different Environments

In order to ascertain the impact of the genotype, environment, and their interactions on the traits studied, a combined analysis of variance (ANOVA) was conducted on the vegetative traits, flowers, yield components, and yield. The outcomes of the combined ANOVA for the 24 traits assessed indicated that the environmental factor had a substantial impact on all of the traits studied, while the genotypic factor had a significant impact on almost all of the traits studied except for stem diameter (

Table 4. Combined ANOVA for vegetative, flowering, yield attributed, and yield trait in five intercropping systems in West Java, Indonesia.

Traits	E	G	GEIs	Error	Total	CV (%)
PH	227,730.17 **	73,720.12 **	99,655.43 **	84,994.95	503,471.35	8.59
CH	8578.74 **	63,740.08 **	14,826.41	21,941.97	143,860.47	10.55
RD	35,713.70 **	603.31	2022.99	2241.46	40,772.33	15.45
NoL	65.38 **	98.34 **	91.49	114.64	385.23	6.11
LL	14,227.56 **	7778.96 **	8382.64 **	5347.45	36,806.44	4.55
LW	314.89 **	149.06 **	128.69	144.95	766.40	6.97
CC1	21,327.66 **	9677.18 **	9367.22 *	8818.12	50,551.11	15.15
CC2	27,913.78 **	15,360.59 **	13,946.53	15,240.88	74,976.79	17.93
MF	269.14 **	1619.59 **	1106.46	1166.18	4281.19	3.20
FF	2720.02 **	1660.27 **	1286.18	1436.58	7160.47	3.49
CW	213.53 **	121.71 **	215.87 **	86.23	641.30	20.52
CWWS	135.44 **	86.01 **	147.15 **	55.61	425.53	21.74
NoCPP	9470.22 **	2110.50 **	5539.08 **	3823.09	21,275.59	20.34
CL	313.42 **	547.90 **	1080.85 **	925.89	2973.95	10.72
CD	520.29 **	2864.71 **	7202.16 *	7024.73	18,127.46	11.87
NoRPC	37.38 *	501.77 **	1172.24 **	1067.07	2799.79	13.47
NoSPC	316,425.09 **	1,186,271.32 **	1,949,120.20 **	1,047,978.16	4,596,530.07	14.43
PD	2834.76 **	1284.75 *	3980.15	4598.86	12,912.49	15.21
CW3S	7.05 **	2.95 **	4.91 **	3.81	19.24	25.89
SWPC	0.29 **	0.98 **	1.90 **	1.61	4.86	25.81
SWPP	6.27 **	19.18 **	34.19 **	29.87	92.36	30.97
1000SW	0.73 **	0.26 **	0.53	0.58	2.16	15.20
GY	104.48 **	534.49 **	818.81 **	438.03	1906.34	29.19

E = environment; G = genotype; GEIs = genotype-by-environment interactions; CV = coefficient of variation; * = p < 0.05; ** = p < 0.01; For the trait codes, see Table 2.

).

The environment had a significant effect on all the traits tested (Table 4). Genotype had a significant effect on the 22 traits tested, except for the stem diameter trait (Table 4). GEIs had a significant effect on the 14 traits tested (Table 4). There were variances in the makeup and genetic potential of each studied maize mutant, as evidenced by a very strong genotypic effect on several of the phenotypes [40]. The genotypes of the maize produced by the mutations offered a fantastic chance to obtain genetic variation within quantitative parameters. According to Ruswandi et al. (2017) [40], variations in the parents’ regions of origin can lead to genetic variations in maize. Geographical limitations and physical distance can also have an impact on how genetic variation is distributed. As a result, this variation may have an impact on each genotype’s potential for yield and yield attributes, which may change how each genotype reacts to various environmental factors.

The environmental factors showed significant differences in all the traits tested. Genotypes grown in different environmental conditions will have different responses. This is because each genotype is the result of a mutation that has a random change in gene arrangement such that it will have a different response to the conditions of the testing location and season. According to Wicaksana et al. (2022) [29], maize responded differently to various settings and growing seasons in Indonesia. The varieties of plants employed as intercrops can have an impact on environmental conditions, causing them to respond differently to each studied maize genotype. As a result, the various planting environments can result in various reactions from each genotype studied in each environment. GEIs will affect the appearance of quantitative traits, especially yield and yield components. The data in Table 4 show that GEIs had a significant effect on the 14 characters tested. This indicates a different response between the tested maize mutants and the five intercropping environments. Several researchers have reported that differences in the planting environment can cause GEIs on yield and yield components in maize [2,29]. Plant breeding initiatives lose effectiveness as a result of the GEI effect. This is due to the fact that stability evaluations must be performed during multiple location tests and the plant selection procedure must be carried out in various conditions or seasons. [30]. This aims to evaluate maize genotypes that have broad adaptability to environmental changes and have high yields. The development of GEIs, according to Ruswandi et al. (2022) [2], has hampered the breeding program for maize. Several researchers noted the same thing, citing black soybeans [41,42], stevia [43], butterfly pea [44], and sweet potatoes [30]. In another study, Maulana et al. (2020) [45] conducted tests at different locations and tested yield stability in sweet potato plants in West Java when GEIs appeared. This was carried out to obtain high-yielding and stable genotypes in various habitats. In order to determine the effects of GEIs on this characteristic, it is important to assess the stability of the yield trait under five distinct conditions. The evaluation using the GGE biplot was only carried out on crop yield. This is because yield is the main character in the selection of superior genotypes. Additionally, yield is one of the standards for farmers and academics involved in large-scale development.

3.4. Representative Environment and Yield Stability of Maize Mutants in Five Environments Based on GGE Biplot

The following is in regard to the yield stability visualization of maize mutants using GGE biplot analysis. The results of the GGE biplot analysis of 60 maize mutants showed that PC1 and PC2 contributed 58.5% and 22.4% to the total variation in the yield of the maize mutants, respectively (Figure 3, Figure 4 and Figure 5). According to the ‘representative versus discriminative’ display of the GGE biplot (Figure 3), the test environment can be classified into three types, namely the type I environment, which has a short vector and provides little information about the genotype being tested, so it should not be used as a test environment. Superior genotypes can be chosen in type II environments because they have an average abscissa environment and a lengthy small-angle vector. Since type III environments have mean abscissa neighborhoods and long, large-angle vectors, they are useful for choosing unstable (adaptive) genotypes rather than ideal genotypes [46].

Figure 3 shows the findings from the maize mutant population, which shows that Environment 5 is a type I environment and should not be used as a testing environment. Because of its strong discriminatory and representative nature, small angles for the abscissa, and high representativeness, Environment 4 is a suitable place (type II) to choose superior genotypes. As a type III environment, Environment 2 should not be used to select superior genotypes, but it can be used to select genotypes that are adaptable or particular to a given place. The ‘mean vs. stability’ biplot is presented in Figure 4. Each genotype’s average yield is displayed on the X-axis, while the consistency of the results is displayed on the Y-axis. In this investigation, the yields of the genotypes to the left of the Y-axis were higher than the average yield across all genotypes. The genotypes to the right of the Y-axis, however, had yields that were lower than the average yield across all genotypes. The genotype is unstable if it is far from the X-axis, and vice versa. [30].

In the maize mutant population, there are 24 genotypes located on the left side of the Y-axis, while the other 36 genotypes have smaller yields (below the overall mean). The genotypes M5, M18, M27, and M54 are the most agronomically stable (high yield) because they have the shortest distance to the X-axis and approach the ideal point (arrows in small circles). This genotype was capable of high yields in both marginal and optimal environments. Ruswandi et al. (2022) [2] also reported that the selection of maize in Indonesia, when using a GGE biplot, succeeded in selecting stable genotypes. In this study, the four genotypes were able to produce maximum production in the five growing environments.

The ‘which won where’ graph shows that the five environments had eight sectors with different peak genotypes (Figure 5). There are two sectors that consist of more than one environment so as to form a mega-environment. The first mega-environment consists of Environments 1 and 2, with M20 and M25 as the peak genotypes. The second mega-environment consists of Environments 3, 4, and 5, with M5 as the peak genotype. The other peak genotypes that did not have an environment in each sector were M3 in sector 2, M39 in sector 3, M2 in sector 4, M59 in sector 5, M37 in sector 6, and M33 in sector 7. The genotypes that were in the peaks for each sector indicated that these genotypes had high yields in the environment in that sector [2,29,30]. The genotypes that are in a sector containing more than one environment or mega-environment are the ideal genotypes [29,47]. Conversely, the genotypes that are at the top of the sector but do not have an environment show low yields [29]. Therefore, in this test, the ideal genotypes were M5, M20, and M25.

The results of the GGE biplot analysis identified several genotypes that were close to the center of the axis. They are the M11, M23, M40, M4, M46, M49, and M17 genotypes. Several researchers have revealed that those genotypes that are close to the central axis (0.00) are stable genotypes [48,49]. However, these genotypes still need to be evaluated for their performance (high or low) because this stability only determines the performance of each genotype in all environments. This means that stable genotypes may have low or high yields in all test environments. Therefore, in this study, the selection of the best (ideal) genotypes was emphasized for genotypes that are in the sectors that contain more than one environment (mega-environment), namely M1, M5, M8, M13, M16, M18, M19, M20, M22, M25, M27, M28, M29, M31, M41, M48, M49, M51, and M54.

3.5. Yield Stability Analysis Based on Sustainability Index (SI)

The evaluation of yield stability based on the sustainability index (SI) is presented in

Table 5. Sustainability index of maize mutants under different cropping systems in Upland.

Mutants	Y	σn	YM	SI (%)	Criteria	Mutants	Y	σn	YM	SI (%)	Criteria
M1	5.84	1.97	7.89	49.05	Moderate	M31	4.77	0.81	5.81	68.05	High
M2	2.22	0.99	3.21	38.22	Low	M32	3.52	1.05	5.04	48.90	Moderate
M3	4.50	1.47	6.93	43.68	Moderate	M33	5.28	2.18	7.63	40.70	Moderate
M4	3.90	0.27	4.41	82.22	Very high	M34	3.69	1.02	5.04	52.96	Moderate
M5	6.46	1.97	9.93	45.21	Moderate	M35	3.57	1.22	5.04	46.63	Moderate
M6	4.46	1.53	6.93	42.33	Moderate	M36	3.78	1.43	5.55	42.51	Moderate
M7	3.89	1.06	5.29	53.51	Moderate	M37	3.68	1.91	6.80	26.00	Low
M8	5.32	1.31	6.55	61.17	High	M38	2.41	0.85	3.40	45.78	Moderate
M9	3.22	0.30	3.78	77.32	High	M39	3.40	1.60	5.67	31.68	Low
M10	3.68	1.40	6.05	37.75	Low	M40	3.87	0.65	5.04	63.69	High
M11	3.87	0.37	4.41	79.34	High	M41	5.34	1.28	7.06	57.41	Moderate
M12	4.08	1.49	6.05	42.81	Moderate	M42	3.44	0.82	4.29	61.10	High
M13	4.67	0.70	5.42	73.22	High	M43	3.31	1.18	5.04	42.22	Moderate
M14	3.16	0.67	3.78	65.88	High	M44	3.22	1.54	5.67	29.70	Low
M15	4.37	0.85	5.67	62.07	High	M45	2.93	1.14	4.16	43.05	Moderate
M16	4.38	0.91	5.55	62.41	High	M46	3.97	0.55	4.72	72.44	High
M17	4.25	1.07	5.17	61.47	High	M47	4.01	1.58	6.93	35.01	Low
M18	5.93	2.08	9.80	39.32	Low	M48	4.68	1.20	6.30	55.18	Moderate
M19	5.40	1.67	7.56	49.25	Moderate	M49	4.50	0.85	6.05	60.37	High
M20	5.61	0.81	6.56	73.23	High	M50	4.24	1.23	6.18	48.76	Moderate
M21	3.96	0.58	4.79	70.38	High	M51	4.89	1.12	6.30	59.79	Moderate
M22	5.49	1.55	7.69	51.25	Moderate	M52	4.01	1.67	5.67	41.29	Moderate
M23	4.14	0.70	5.04	68.20	High	M53	3.58	0.90	5.04	53.04	Moderate
M24	3.71	1.19	5.04	50.13	Moderate	M54	5.42	0.87	6.91	65.81	High
M25	5.61	1.19	7.56	58.48	Moderate	M55	3.46	1.69	5.93	30.01	Low
M26	2.84	1.06	4.29	41.42	Moderate	M56	3.71	0.75	5.04	58.71	Moderate
M27	6.18	1.36	7.56	63.75	High	M57	3.55	1.33	5.17	42.86	Moderate
M28	4.54	0.56	5.42	73.39	High	M58	3.88	1.23	5.04	52.57	Moderate
M29	5.29	1.13	6.65	62.52	High	M59	2.69	1.58	5.29	20.96	Low
M30	3.69	0.99	5.04	53.68	Moderate	M60	3.10	1.31	5.04	35.49	Low

Note: For mutant codes, see Table 1.

. In this study, the SI criteria obtained were divided into four groups: low, medium, high, and very high (Table 5). Several researchers revealed that a high SI estimate indicates a level of stability for certain genotypes [2,44]. The SI estimations for yields in the maize mutant ranged from 20.96% (low) to 82.22% (very high). The wide range of the SI is due to the genetic background of the planting material resulting from a mutation, where mutations can change the genetic ability of a genotype randomly.

An SI with a very high category was only indicated by M4 (82.22%), but it has yields below the overall average yield (3.90 t.ha⁻¹). Therefore, this mutant was not included in the selected category. There were 12 maize mutants that had a high SI category (61–80%) (Table 5). However, only 12 mutants had yields above the overall average yield (>4.18 t.ha⁻¹), and they were M8, M13, M15, M16, M17, M20, M27, M28, M29, M31, M49, and M54 (Table 5). They were included in the ideal group because they had high yields and were stable in the five different environments. Several researchers have analyzed a similar strategy for selecting superior genotypes (high and stable yields) by using SI for Asiatic cotton [50], maize [2,29], and Butterfly pea [44]. Therefore, the group of maize mutants with high SI values and yields (above the overall average yield) represents the priority in terms of the selection of superior maize mutants.

An SI in the moderate category (40–60%) was obtained for 28 maize mutants (Table 5). An SI in the low category (20–40%) was obtained for 10 maize mutants (Table 5). The low SI measurement results for these 10 maize mutants indicated an unstable yield. Where M2 and M59 have the lowest average yields, these results are similar to the previous GGE biplot analysis, as presented in Figure 5. These two maize mutants are in the peak sector with no environment (cropping system). Several researchers have reported that peak genotypes lacking environmental vectors showed lower yields [29,30]. Thus, these two maize mutants were not included in the selected genotypes for both the broad and specifically adapted genotypes.

In order to select the ideal maize mutants, two stability measurements (GGE biplot and SI) were applied. This was also carried out by the authors of [44]. Information on the selected mutants from each measurement is presented in

Table 6. Selected maize mutants based on GGE biplot and sustainability index (SI) measurements.

Measurements	Selected Mutants	Percent (%)
GGE biplot	M1, M5, M8, M13, M16, M18, M19, M20, M22, M25, M27, M28, M29, M31, M41, M48, M49, M51, M54.	31.67
Sustainability Index (SI)	M8, M13, M15, M16, M17, M20, M27, M28, M29, M31, M49, dan M54	20.00
Slice	M8, M13, M20, M27, M28, M29, M31, M49, M54	15.00

Note: For mutant codes, see Table 1.

. The GGE biplot identified 31.67% of the stable mutants, while the SI identified 20.00%. However, only 15.00% of the mutants were selected from the two measurements, namely M8, M13, M20, M27, M28, M29, M31, M49, and M54. The nine maize mutants can be declared the superior maize candidates within various intercropping environments. This will be very helpful in breeding maize hybrids for sustainable agriculture programs.

3.6. Heritability of Vegetative, Flowering, Yield Components, and Yield Traits

The results of this study indicated that the value of genetic diversity in maize mutants is broad (Figure 1 and Figure 2). Selection will be effective if the population has wide genetic diversity [4,9]. The extent of the resulting diversity, both phenotypic and genetic diversity, indicates that there is a great opportunity to select the desired traits. The wide diversity of the genotypes and phenotypes is due to the fact that the seeds used are the result of mutations from selected superior lines. Wide diversity can also occur due to the seeds having different genetic characteristics.

Heritability determines the success of a selection because it can provide clues as to the genetic factors or environmental factors that are more influential on a trait. The estimates of high heritability values are shown by the 17 traits tested (

Table 7. Mean value, standard deviation, genetic variance (σ²g), phenotypic variance (σ²p), and heritability (H²) on agro-morphological traits of 60 maize mutants.

Traits	Mean ± SD	σ2g	σ2p	σ2ge	σ2e	H2 (%)	Criteria
Plant height (PH)	197.58 ± 28.97	82.72	124.95	67.08	288.12	66.20	High
Cob height (CH)	81.78 ± 15.48	101.75	109.19	0.00	74.38	93.19	High
Stem diameter (RD)	17.84 ± 8.24	0.17	1.02	0.49	7.60	16.18	Low
Number of leaves (NoL)	10.20 ± 0.80	0.13	0.17	0.00	0.39	76.74	High
Leaf length (LL)	93.56 ± 7.83	9.63	13.18	8.70	18.13	73.06	High
Leaf width (LW)	10.06 ± 1.13	0.20	0.25	0.03	0.49	78.41	High
CC1	36.10 ± 9.18	12.43	16.40	4.90	29.89	75.80	High
CC2	40.09 ± 11.18	20.13	26.04	3.72	51.66	77.30	High
Male flowers (MF)	62.04 ± 2.67	2.28	2.75	0.37	3.95	82.92	High
Female flowers (FF)	63.24 ± 3.45	2.27	2.81	0.29	4.87	80.63	High
Cob weight (CW)	2.63 ± 1.03	0.11	0.21	0.31	0.29	55.66	High
Cob weight without cobs (CWWS)	2.00 ± 0.84	0.08	0.15	0.22	0.19	57.23	High
Number of cobs per plot (NoCPP)	17.70 ± 5.95	1.23	3.58	5.26	12.96	34.39	Moderate
Cob length (CL)	16.53 ± 2.23	0.47	0.93	0.72	3.14	50.68	High
Cob diameter (CD)	41.12 ± 5.50	1.80	4.86	3.35	23.81	37.15	Moderate
Number of rows of seeds per cob (NoRPC)	14.12 ± 2.16	0.35	0.85	0.67	3.62	41.59	Moderate
Number of cobs (NoSPC)	413.17 ± 87.53	1184.73	2010.62	2353.25	3552.46	58.92	High
Slender diameter (PD)	25.95 ± 4.64	0.49	2.18	0.64	15.59	22.55	Moderate
Cobs weight: 3 samples (CW3S)	0.44 ± 0.18	0.00	0.00	0.00	0.01	58.49	High
Seed weight: 5 samples (SWPC)	0.29 ± 0.09	0.00	0.00	0.00	0.01	52.11	High
Seed weight per plot (SWPP)	1.03 ± 0.39	0.02	0.03	0.02	0.10	55.41	High
weight 1000 seeds (1000SW)	0.29 ± 0.06	0.00	0.00	0.00	0.00	48.07	Moderate
Yield (GY)	4.18 ± 1.78	0.56	0.91	0.99	1.49	61.72	High

). Five traits show moderate heritability values and one trait shows low values (Table 7). A high heritability predictive value indicates that genetic factors play a greater role in controlling a trait than environmental factors [4,51]. Trait selection with high heritability can be achieved in the early generations. Conversely, if the heritability value is low, then the character must be selected within the next generation [8]. Although the heritability values for the characters of flowering age, harvesting age, plant height, number of pods per plant, and seed weight per plant are high, it should be noted that these heritability values are broad in their sense. The value of heritability in a broad sense includes the influence of additive gene action, dominance, and epistasis. If dominant gene action and epistasis play a greater role in controlling the character in question, then selection cannot be carried out within the early generations. So, it must be selected for in the next generation.

A high heritability value also indicates that genetic factors support the character’s appearance, such that it can complement the progress of selection [5]. The results of this study indicate that the vegetative characters, yields, and yield components have a wide variety of values and high heritability predictive values. Thus, selection to obtain superior mutants can be applied to these characters. High heritability values for key traits, such as yields, play an important role in sustainable agriculture programs. Currently, sustainable agriculture is an important factor in the advancement of agricultural crops [52]. High heritability indicates that these traits can be passed on to the next generation, which can support economic improvement [53]. One of the characteristics of sustainable agriculture is increasing economic profits. Therefore, a high heritability value is highly expected in a sustainable agriculture process.

3.7. The Relationship between the Vegetative Traits, Flowering, Yield Components, and Yields of Maize Mutants via Pearson Correlation and Genotype-by-Trait (GT) Biplots

The relationship between each tested trait is presented in

Table 8. The relationship between each trait tested on 60 maize mutants.

Traits	PH	CH	RD	NoL	LL	LW	CC1	CC2	MF	FF	CW	CWWS	NoCPP	CL	CD	NoRPC	NoSPC	PD	CW3S	SWPC	SWPP	1000SW
CH	0.58 *
RD	0.26	0.13
NoL	0.31	0.43	0.21
LL	0.54 *	0.48	0.48	0.35
LW	0.15	0.26	0.39	0.14	0.38
CC1	0.48	0.39	−0.09	0.33	0.10	−0.03
CC2	0.23	0.23	−0.11	0.18	0.08	0.01	0.50 *
MF	−0.27	0.33	−0.11	0.05	0.20	0.31	−0.08	0.05
FF	−0.30	0.28	−0.08	0.07	0.14	0.23	0.02	0.02	0.89 **
CW	0.25	0.51 *	0.20	0.38	0.25	0.40	0.35	0.31	0.13	0.15
CWWS	0.21	0.47	0.19	0.37	0.22	0.44	0.27	0.25	0.14	0.12	0.96 **
NoCPP	0.12	0.26	0.17	0.21	0.23	0.29	0.09	0.25	0.10	0.10	0.68 **	0.61 **
CL	0.37	0.36	0.15	0.26	0.19	0.32	0.44	0.37	0.01	0.03	0.69 **	0.65 **	0.38
CD	0.27	0.30	0.10	0.02	0.21	0.27	0.43	0.24	−0.05	−0.02	0.54 *	0.53 *	0.20	0.42
NoRPC	0.17	0.05	0.09	0.13	0.02	0.11	0.24	−0.14	−0.13	−0.12	0.23	0.32	0.00	0.23	0.45
NoSPC	0.35	0.22	0.03	0.19	0.05	0.13	0.37	0.17	−0.15	−0.12	0.55 *	0.56 *	0.33	0.54 *	0.53 *	0.59 *
PD	−0.07	0.24	0.22	0.12	0.14	0.32	0.13	0.13	0.23	0.33	0.33	0.34	0.09	0.10	0.41	0.18	0.12
CW3S	0.24	0.22	0.09	0.14	0.03	0.35	0.43	0.33	−0.10	−0.09	0.78 **	0.78 **	0.43	0.70 **	0.58 *	0.42	0.59 *	0.24
SWPC	0.35	0.35	0.20	0.25	0.18	0.38	0.42	0.34	0.01	−0.01	0.69 **	0.67 **	0.19	0.64 **	0.58 *	0.39	0.66 **	0.32	0.77 **
SWPP	0.27	0.42	0.23	0.26	0.19	0.25	0.38	0.29	0.06	0.03	0.74 **	0.73 **	0.47	0.54 *	0.54 *	0.44	0.63 **	0.32	0.66 **	0.76 **
1000SW	0.25	0.39	0.07	0.21	0.21	0.33	0.36	0.49	0.17	0.08	0.64 **	0.64 **	0.38	0.55 *	0.41	0.09	0.16	0.15	0.69 **	0.61 **	0.58 *
GY	0.28	0.55 *	0.22	0.44	0.30	0.42	0.37	0.35	0.15	0.15	0.93 **	0.95 **	0.57 *	0.63 **	0.58 *	0.33	0.58 *	0.39	0.74 **	0.74 **	0.82 **	0.67 **

* p < 0.05; ** p < 0.01; Traits code see data collection in Section 2.

. A trait that has a correlation with other traits is notated as “*”. If a trait has a correlation coefficient value with one “*” notation, then the trait has a correlation strength of 5% (p < 0.05). However, if a trait has a correlation coefficient value with the notation of “**”, then the trait has a correlation strength of 1% (p < 0.01). The relationship between a single trait and the other traits in a plant has important meaning in plant-breeding programs. Traits that are positively and significantly correlated show a high relationship between these traits [4]. This result will greatly support the selection process (through these characters) in the next generation. The correlation information between crop yields and other traits is very important in determining genotypic selection. If the value of the correlation coefficient is high, then the selection will be more effective because the characteristics of one trait influence the other [5]. Yield (GY) had a positive correlation coefficient with all the traits tested, but only 11 traits showed a significant value, namely: CH, CW, CWWS, NoCPP, CL, CD, NoSPC, CW3S, SWPC, SWPP, and 1000SW (p < 0.01). Thus, these 11 traits have a very close relationship with maize yields under the five environmental conditions.

GT biplots were used to identify the superior genotypes for certain traits and the relationship between each tested [22]. The results of the GT biplot show that the graph provides very strong evidence because it represents 71.8% (PC1 and PC2) of the total variation (Figure 6, Figure 7 and Figure 8). Figure 6 shows the GT biplot for the maize population based on data from the five growing environments. Nine sectors were generated in Figure 6, with genotypes and traits tested in the different sectors. The genotypes in the same sector as the traits tested had a strong relationship. M5 and M27 were top in sector 1 and correlated with the NoRPC, PH, CL, CD, 1000SW, NoSPC, CW3S, SWPP, SWPC, CC1, and CC2 traits. M60 excelled in sector 3 but did not correlate with any of the other traits. Likewise, in sectors 4 and sector 5, M2 and M59 excelled in sector 4 and M45 in sector 5, but there were no traits in that sector, so the genotypes in that sector did not have a strong relationship with those traits tested. M23 excelled in sector 6 and correlated with the FF and MF traits. M54 excels in sector 8 and correlated with the NoCPP, LW, BD, and LL traits. In sector 9, there was no dominant genotype for the traits. This shows that the genotypes in sector 9 tend to have the same strength for the GY, CWWS, CW, CH, PD, and NoL traits. In this study, M2, M45, and M59 were identified as being far apart and opposite regarding all the traits measured. Several researchers have reported that genotypes that are far apart and opposite from the trait being tested showed the smallest value for these traits [4,54]. This shows that the three genotypes have relatively small values for all the traits measured, so they are not included in the expected genotypes.

Figure 7 shows the relationship between the traits based on the GT biplot for the maize mutant population. The traits that have an acute angle (<90°) are stated to be positively correlated (have a close relationship) and vice versa [4,22]. In the population tested, yield (GY) had a very significant and positive correlation with CW, CH, CWWS, NoCPP, CL, PD, NoL, LL, MF, FF, RD, and LW and had no significant correlation with the CC1 and CC2 traits (Figure 7). This can also be seen in the correlation value of each trait (Table 6). In Table 6, it is shown that several other traits had a positive and significant correlation (p < 0.05), namely PH to CH and LL. CC1 had a positive and significant correlation with CC2; this can also be seen from the angle generated on the GT biplot graph (Figure 7). In this study, there were several traits that were shown to have a strong relationship according to the GT biplot (having an angle <90°) but did not show a significant correlation coefficient value, including GY with PD, NoL, LL, MF, FF, RD, and LW. Relationships between traits using GT biplots have also been carried out for white lupin [54], Barley [55], and stevia [4].

Figure 8 shows the stability axis (horizontal) and the average axis (vertical) based on the values of the traits tested. The genotype to the left of the vertical line (in the direction of the arrow in the small circle) is the expected genotype [56]. On the other hand, the genotypes that are close to the horizontal line show stability towards the traits being tested and vice versa. The genotypes M1, M5, M18, M22, M20, M13, M16, M41, and M12 were quite stable regarding all the traits tested and were most preferred because they were located close to the horizontal line and to the left of the vertical line (Figure 5). Meanwhile, M2 was close to the horizontal line but far from the arrow on the small circle, so its score was lower than the overall average.

4. Conclusions

The measurement results showed that, based on principal component analysis (PCA), the genetic diversity of the maize mutants was broad, with six axes, which had an eigenvalue between 8.76–1.07 and a cumulative value of 76.64%. The neighbor-joining tree also showed wide variation by producing five different groups. The GEIs showed a significant effect on the 14 characters tested, especially on yield. The environment had a significant influence on the variation of all the properties tested. Environment 4 was the ideal environment for selecting the ideal maize genotype. The GGE biplot successfully selected maize mutants that were 31.67% superior, and the sustainability index (SI) successfully selected maize mutants that were 20.00% superior. Only 15.00% of the maize mutants were selected by these two methods; they are M8, M13, M20, M27, M28, M29, M31, M49, and M54. The estimations for the heritability value for plant height, cob height, number of leaves, leaf length, leaf width, CC1, CC2, male flower, female flower, cob weight, cob weight without cob, cob length, number of cob seeds, cob weight of 3 samples, seed weight of 5 samples, and yield were high, while the characteristics of the number of cobs per plot, ear diameter, number of rows of seeds per cob, comb diameter, and 1000 seed weight were moderate, and the stem diameter characteristic showed low heritability. Yield had a positive and significant correlation with CH, CW, CWWS, NoCPP, CL, CD, NoSPC, CW3S, SWPC, SWPP, and 1000SW (p < 0.01). Information about the genetic diversity, heritability, and relationships of the traits tested can be used as the basis for future maize-breeding programs. The best intercropping system can be used as an alternative for increasing the income from maize cultivation. Selected maize mutants can be proposed as candidates for new superior genotypes.