Identification of Bean Lines (Phaseolus vulgaris) with Low Genotype–Environment Interactions Under Rainfed in Two Semiarid Sites of North-Central Mexico

Abstract

The seed yield of 28 bean (Phaseolus vulgaris) lines from different crosses and two check cultivars was evaluated under rainfed conditions in two sites in North-Central Mexico. The aim was to identify high-yielding lines with low genotype–environment interactions (GEIs). Trials were conducted under a 6 × 5 square lattice design with four replicates; due to the lack of rainfall in Zacatecas, the trial was helped with supplemental irrigation. Data were analyzed by location and combined to determine the effects of GEIs using the additive main effects and multiplicative interaction model (AMMI) model. The combined analysis showed that 75.45% of the yield variation among lines was due to the effect of the environment, 11.75% was due to genotypes, and 12.79% was due to GEIs. Lines 5 and 10 displayed the highest yield, which slightly surpassed the checks (2.1 and 0.11%, respectively) and showed greater stability than those in the test environments. The AMMI analysis allowed for the selection of stable and high-yielding lines under drought conditions. Data on the weight and yield per hectare of a hundred seeds between and within locations identified lines 5, 10, and 16 as outstanding and capable of being used as a parent in a future hybridization program or as a new cultivar with drought tolerance.

1. Introduction

Dry beans (Phaseolus vulgaris) are a strategic crop in Mexico for the social and economic development of producers and play an important role as a protein source for the population. In 2022, the total area sown with beans nationwide was 1,472,462.29 ha, where the state of Zacatecas established the largest area with 618,225 ha; however, only 3.63% of that is produced under irrigated conditions, and the rainfed bean, grown in the semiarid highlands, had an average yield of 434 kg ha [1]. The low yield is due to the limited and erratic rainfall during the crop cycle, a situation that is aggravated by the low capacity of the soil to retain moisture. Currently, the search for drought-tolerant materials is the main strategy to increase bean yield under rainfed conditions [2].

The “Flor de Junio” commercial bean type is of high interest in the center of the country, where it is produced under irrigated and rainfed conditions in the states of Zacatecas, Guanajuato, and San Luis Potosí [3]. Therefore, the search for materials with drought tolerance is the main strategy to increase bean yield [2]. In addition to this, in recent years, the temperature has increased due to the effects of climate change, and alterations in the physiology of the plant have been observed. Therefore, under rainfed conditions, adaptation, stability, yield per hectare, and resistance to diseases, which occur in various environments, must be considered; likewise, in the selection, it is recommended to test two or more levels of soil moisture and use the geometric mean as another selection criterion [4]. In order to select drought-resistant lines and consider them in genetic improvement programs, germplasm evaluation must be carried out in low rainfall sites to identify the superior genotypes in the genetic pool available for these climatic conditions [5].

Currently, studies under rainfed conditions aim to develop bean genotypes with adaptability and stability in different environments, where both the grain yield per hectare and resistance to diseases that occur naturally in these environments are the objectives. It is important to note that the additive main effects and multiplicative interaction model (AMMI) facilitate the identification of genotypes with resistance to diseases and high yield in various environments [6], even in critical environments with low rainfall, making it a selection tool that should be considered in crop genetic improvement programs [7]. Likewise, the AMMI model allows the identification of genotypes with low genotype–environment interactions (GEIs) and to identify the materials with greater stability in the evaluation environments [8]. The objective of this study was to identify high-yield “Flor de Junio” bean lines under rainfed and supplemental irrigation conditions in two locations in the semiarid highlands of Mexico.

2. Materials and Methods

2.1. Ubication

In the spring–summer cycle of 2023, a yield trial was established under rainfed conditions in two locations, the Bajío Experimental Field (CEBAJ) (20°35′01″ N latitude, 100°49′22″ W longitude, and 1763 masl) and the Zacatecas Experimental Field (CEZAC), located at 22°54′31″ N latitude, and 102°39′34″ W longitude, and 2198 masl. In CEBAJ, the soil is of the pelvic vertisol type with a clay loam texture with 1.8% organic matter, 15.1 mg kg⁻¹ of phosphorus, 18.0 mg kg⁻¹ of nitrogen, 1086 mg kg⁻¹ of potassium and pH = 7.7, while at CEZAC, the soil of the experimental site is of the luvic kastanozem type with a clay loam texture, and organic matter content of 1.91%, 8.5 mg kg⁻¹ of phosphorus, 8.4 mg kg⁻¹ of nitrates and 90 mg kg⁻¹ of K, and pH = 8.5.

2.2. Germplasm Tested and Experimental Design

Twenty-eight lines from different crosses and two control varieties, “Junio León” and “Dalia”, were tested. This set of lines was derived from crosses made in 2013 and from the F₂ generation onwards for five consecutive generations, which were subjected to individual plant selection, two generations per year, and entered into preliminary yield trials as F₁₀ lines in 2021. These lines have been evaluated for seed yield under rainfall and irrigated conditions at the CEBAJ since the summer of 2021. The characteristic color of the “Flor de Junio” bean type is a cream background with pink stripes. The design used was a rectangular lattice of 5 × 6 with four replicates; the experimental unit consisted of two furrows six meters in length with 0.75 m between furrows and 10 cm between plants.

2.3. Crop Management

Bajío Experimental field (CEBAJ). The trial started with sowing on 7 July 2023; the applied fertilization dose was 30-50-30 N-P-K (nitrogen, phosphorus, and potassium), respectively, which was established at the time of sowing. Two mechanical cultures were carried out: the first at 20 days after sowing (DAS) and the second at 25 days after the first culture [9].

Zacatecas Experimental field (CEZAC). Sowing was performed on dry soil on 1 August 2023, then drip tape was placed on each furrow, and ten hours of irrigation started at 65.8 mm to ensure germination. This was due to the lack of rain for crop establishment; afterward, the experiment was managed under rainfall conditions plus supplementary irrigation applied at the flowering stage, which involved six hours of applying 39.48 mm of rainfall during the cycle, and 121.4 mm was accumulated. Two mechanical cultivations were given, including the first 15 days after sowing, and at the same time, chemical fertilization was applied 30-50-30 for nitrogen, phosphorus, and potassium, respectively; the second cultivation was carried out 25 days after the first. Afterward manual weeding was carried out to keep the crop free of weeds [10].

At both locations, the minimum, mean, and maximum temperatures and precipitation were recorded daily at the station closest to the experimental site. Days to flowering (DF), physiological maturity (DM), the duration in days of the reproductive period (DPR), and water use efficiency (WUE), which relates to the kg of grain to mm of water, and whether rainfall or irrigated water was used, were recorded. When the lines reached physiological maturity, they were harvested to determine the seed weight from each plot and the 100-seed weight (100SW). The yield per plot was converted to kilograms per hectare.

2.4. Statistical Analysis

The statistical analysis used for both locations for the measured variables was a complete random block design (Y_ij = µ + α_i + β_j + ε_ij), where Y_ij = the phenotypic value of the i-th genotype obtained in the j-th repetition, µ= the effect of the general mean, α_i = the effect of the i-th treatment, β_j = the effect of the j-th repetition, and ε_ij = the error of the i-th treatment and the j-th repetition. And the combined form (Y_ijk = µ + L_i + T_i + LT_ij + Br_(i)j + E_ijk) was used to determine the differences between environments (L), genotypes (T), and the location of genotype interaction. The analysis of variance followed a purely additive effects model capable of identifying the effect of the genotype by environmental interaction (IGA). On the other hand, the AMMI principal component (PC) model takes into account the additive model plus the multiplicative model; thus, this method is useful for identifying stable and adapted genotypes. The AMMI model is the most appropriate for distinguishing IGA analysis since it shows greater precision compared to the additive model [11]. The analysis was performed with the SAS 9.4 program [12], and Duncan’s multiple range test was used to compare means with α = 0.05.

3. Results

The locations involved in the evaluations are two contrasting sites in the Mexican plateau; Calera, Zacatecas (CEZAC), located in the north, with an annual precipitation of 406 mm per year, an altitude of 2198 masl, alkaline soil (8.5), with 1.9% of soil organic matter (SOM) and low fertility (8.5, 8.4 and 90 mg kg⁻¹, phosphorus, nitrogen and potassium, respectively) and a clay loam texture and soil shallow depth; in contrast, Celaya, Guanajuato, (CEBAJ) located in Central-North Mexico, has higher annual precipitation (625 mm), with a slightly lower altitude of 1763 masl, lower soil pH (7.7), SOM (1.8%), a clay loam texture and low fertility (15.1, 18.0 and 1086 mg kg⁻¹, phosphorus, nitrogen, and potassium, respectively), as well as greater soil depth, which allows greater moisture storage. These differences make this site slightly more favorable for the development of bean cultivation.

In the first five days of July, 56.39 mm of precipitation occurred, which was used to establish the crop on the seventh day; during the crop cycle, 374.80 mm of rainfall occurred, of which 88.01% precipitation occurred in the vegetative stage and the rest occurred in the flowering and grain lining stage and poorly distributed rainfall was recorded at CEBAJ (Figure 1), with the last important instance of rainfall occurring in early September. After this, rainfall was limited; that is, a terminal drought occurred. Periods of little rainfall and maximum temperatures of 30 °C allowed a severe attack by leafhoppers (Empoasca kraemeri), but the insect was controlled with biological and chemical commercial products.

At CEBAJ, the variation in days to flowering and yield (

Table 1. Agronomic characteristics of 30 bean genotypes of the “Flor de Junio” type under rainfed conditions at CEBAJ, Celaya Guanajuato, México, in the spring–summer cycle, 2023.

Líne	Genealogy	DF	DM	DPR	100SW	Seed Yield (kg ha−1)	WUE
					(g)		(kg mm−1)
1	GPO82 82 FJ22-1	46	83	37	26	1484	3.97
2	GPO82 82 FJ22-2	48	81	33	26	1352	3.67
3	GPO82 82 FJ22-3	49	82	33	26.1	1071	3.07
4	GPO82 82 FJ22-4	48	82	34	25.8	1168	3.36
5	GPO82 82 FJ22-6	48	82	34	27.3	1444	3.98
6	GPO61 36	47	84	37	27.8	1128	3.24
7	GPO59 67	50	83	33	25.5	1087	3.18
8	GPO82 NO82-1 FJ35-2	49	85	36	26.3	1042	3.31
9	GPO82 NO82-1 FJ35-1	49	82	33	25.1	820	2.28
10	GPO61 36	48	83	35	27.3	1347	3.67
11	GPO61 36	47	84	37	27	1178	3.46
12	GPO61 36	47	85	38	24.9	1115	3.28
13	GPO61 36	47	85	38	27.9	1190	3.59
14	GPO61 36	47	84	37	26.3	895	2.39
15	FJD121-6	47	85	38	28.3	1074	2.66
16	FJD121-5	47	86	39	26.9	1189	3.71
17	FJD121-4	49	86	37	86.6	1127	2.92
18	FJD121-3	50	84	34	28.3	977	2.67
19	FJD121-2	48	84	36	25.1	954	2.43
20	FJD121-1	49	84	35	26.5	678	2.01
21	FJD121-7	50	86	36	27.8	997	2.77
22	FJD121-8	49	87	38	27.1	1069	2.99
23	FJD121-9	49	87	38	27.3	764	1.99
24	FJD121-10	50	86	36	26.9	1143	3.07
25	FJDGPO61 36	49	85	36	27.4	1023	2.44
26	Junio León (C)	49	86	37	26	1295	3.96
27	Dalia (C)	47	86	39	25.9	1137	3.19
28	FJD124-3	46	85	39	24.2	894	2.13
29	FJD124-2	48	83	35	25.2	971	2.58
30	FJD124-1	48	82	34	26.4	958	2.07
	Mean	48	84	36	28.5	1086	3.00

CEBAJ PV 2023. Abbreviations: 1 DF and DM = days after sowing to flowering and maturity. DPR = duration in days of the reproductive period. 100SW = weight of one hundred seeds and C = control. WUE = water use efficiency.

) between genotypes was not significant (p > 0.05), particularly for yield, which is attributed to the strong pressure from leafhoppers and the lack of rainfall during the grain-filling stage. The number of days to the maturity of the materials in the trial was shortened by the terminal drought, taking around 84 days on average, which indicates a reduced cycle for this type of material since it usually matures in 100 days in this location. In this trial, lines 1 and 5 surpassed the check “Junio León” by 10% (189 kg) in yield (Table 1). These lines were also three days earlier to mature than “Junio León”, which was the second to mature with the highest weight per hundred seeds. These superior lines will have to undergo verification tests in environments like the test location to confirm their superiority and possible future registration as new, improved cultivars. It also coincides that lines 5 and 1 present the highest values in water use efficiency (WUE), which shows the close relationship between precocity in maturity and the use of water resources (Table 1).

For the establishment of the crop at CEZAC, irrigation was supplemented to ensure germination, and during the cycle, there were a few rainy events (Figure 2), with the total rainfall in the cycle reaching 121.4 mm. It is important to note that even with the low rainfall recorded, three events with precipitation greater than 25 mm contributed to sustaining the crop: the first event occurred in the vegetative stage and the other two events occurred in the reproductive stage during grain filling. However, there was a 40-day period in which only 31.4 mm of rainfall was recorded, and that included the pre-flowering phase and the beginning of grain filling.

In the analysis of variance for days to flowering, no significant differences (p > 0.05) were found between the bean lines; however, lines 13, 5, 28, and 16 had the longest vegetative stage with 41.5 days after sowing (DAS) to flowering and line 10 was the earliest to flower (

Table 2. Agronomic characteristics of 30 “Flor de Junio” bean genotypes under rainfed conditions at CEZAC, Calera Zacatecas, México, in the spring–summer cycle, 2023.

Líne	Genealogy	DF	DM	DPR	100SW	Seed Yield		WUE
					(g)	(kg ha−1)		(kg mm−1)
1	GPO82 82 FJ22-1	33	75	42	27	352	bcd	1.43
2	GPO82 82 FJ22-2	37	75	38	29	355	bcd	1.44
3	GPO82 82 FJ22-3	35	76	41	29	323	bcd	1.34
4	GPO82 82 FJ22-4	31	76	45	29	406	bcd	1.38
5	GPO82 82 FJ22-6	42	74	32	26	414	bcd	1.26
6	GPO61 36	41	75	34	29	435	bcd	1.70
7	GPO59 67	40	73	33	32	332	bcd	1.32
8	GPO82 NO82-1 FJ35-2	34	76	42	29	374	bcd	1.50
9	GPO82 NO82-1 FJ35-1	31	76	45	29	307	bcd	1.09
10	GPO61 36	35	76	41	31	468	bcd	1.37
11	GPO61 36	38	72	34	32	356	bcd	1.35
12	GPO61 36	33	74	41	30	344	bcd	1.37
13	GPO61 36	42	75	33	27	500	bcd	1.87
14	GPO61 36	37	74	37	30	367	bcd	1.45
15	FJD121-6	34	76	42	29	168	cd	0.68
16	FJD121-5	42	76	34	30	448	bcd	1.99
17	FJD121-4	36	74	38	31	551	b	2.21
18	FJD121-3	38	74	36	35	1243	a	4.76
19	FJD121-2	32	74	42	34	345	bcd	1.46
20	FJD121-1	30	75	45	33	133	d	0.55
21	FJD121-7	41	75	34	32	463	bcd	1.90
22	FJD121-8	37	76	39	31	476	bcd	2.02
23	FJD121-9	41	72	31	29	474	bcd	1.89
24	FJD121-10	37	76	39	30	504	bcd	1.94
25	FJDGPO61 36	31	76	45	30	140	d	0.58
26	Junio León (C)	36	75	39	29	351	bcd	1.28
27	Junio Dalia (C)	37	74	37	28	379	bcd	1.43
28	FJD124-3	42	75	33	27	521	bc	2.08
29	FJD124-2	34	75	41	30	677	b	2.55
30	FJD124-1	34	70	36	30	427	bcd	1.84
	Mean	36	75	38	30	421		1.63

CEZAC. PV 2023. Note: means with the same letter are statistically equal. DF = days after sowing to flowering. DPR = duration in days of the reproductive period. DM = days after sowing to maturity. 100SW = weight of one hundred seeds. C = control. WUE = water use efficiency.

). Likewise, no significant differences were found in the evaluation of physiological maturity (p > 0.05). Lines 24, 25, 15, 16, 8, and 3 were late maturing, and line 30 was the earliest to mature with 70 DAS. No significant differences were found between the lines for 100-seed weight, which is attributed to the stress at the flowering stage and grain filling. Line 18, with 35 g per 100 seeds, had the highest value, while line 28 had the lowest value; it is worth noting that 23 lines were 3.4 to 20% higher for the 100-seed weight than the controls. For seed yield, highly significant differences were found (p > 0.05) among the lines. Line 18 obtained the highest yield with 1243 kg ha⁻¹, followed by line 29 with 677 kg ha⁻¹, whereas line 20 had the lowest yield with 133 kg ha⁻¹. Also, the WUE values were related to this three-line behavior and produced the highest and the lowest values, respectively (Table 2). As mentioned in ref. [6], naturally occurring environmental stress allows for the selection of lines with drought tolerance.

The combined analysis of variance for the 100-seed weight showed significant differences between environments, genotypes, and the interaction genotype by environment (p ≤ 0.0001). Taking the average from the environments, line 18 (FJD121-3) had the highest 100-seed weight, followed by line 21 (FJD121-7), which surpassed the controls; on the contrary, line 28 (FJD124-3) had the lowest 100-seed weight with 25.4 g (

Table 3. A comparison of the averages for the weight of one hundred seeds from the combined analysis of the 30 lines of the “Flor de Junio” bean type established under temporary conditions in two locations in the P-V-2023 cycle.

Líne	100SW (g)	Líne	100SW (g)	Líne	100SW (g)	Líne	100SW (g)	Líne	100SW (g)
18	31.63a	17	29bcde	16	28.20bcde	3	27.56bcdef	13	27.19bcdef
21	29.89ab	10	28.91bcde	24	28.19bcde	2	27.5bcdef	9	27.05cdef
19	29.56abc	15	28.63bcde	6	28.16bcde	12	27.45bcdef	27 *	26.95cdef
20	29.5abc	7	28.5bcde	23	28.13bcde	4	27.38bcdef	1	26.5def
11	29.25abcd	25	28.44bcde	14	27.91bcdef	29	27.35bcdef	5	26.39ef
22	29.06bcd	30	28.21bcde	8	27.63bcdef	26 *	27.26bcdef	28	25.36f

C.V. 8.03. Note: means with the same letter are statistically equal. * Lines 26 León and 27 for Dalia were used as controls.

). Drought during the reproductive stage considerably affected grain size, which agrees with refs. [3,13], which recorded a 100-seed weight of “Junio León” at 33.7 g, while in this trial, it was 31.6 g, which is attributed to the lack of rainfall during grain filling.

In the combined analysis for yield per hectare, highly significant differences were observed between the evaluation environments and between genotypes (p ≤ 0.0001). Likewise, significant differences were observed in the interaction between genotype and environment (p ≤ 0.003). Even with the low rainfall recorded in the locations, three lines were identified with higher yields compared to the “Junio León” control variety (

Table 4. A comparison of the averages for yield per hectare of the combined analysis of the 30 lines of the June flower bean type established under rainfed conditions in two locations in the P-V-2023 cycle.

Líne	Yield (kg ha−1)	Líne	Yield (kg ha−1)	Líne	Yield (kg ha−1)	Líne	Yield (kg ha−1)	Líne	Yield (kg ha−1)
18	1.110a	11	0.866abc	12	0.8091abc	21	0.7291bcd	14	0.630bcd
5	0.928ab	2	0.853abc	8	0.8069abc	7	0.7085bcd	15	0.621bcd
1	0.918abc	13	0.844abc	4	0.7863abc	28	0.7069bcd	23	0.619bcd
26 *	0.908abc	17	0.838abc	6	0.7811abc	3	0.6965bcd	25	0.581bcd
10	0.907abc	24	0.823abc	22	0.7721abc	30	0.6915bcd	9	0.563cd
16	0.897abc	29	0.823abc	27*	0.7571abc	19	0.6493bcd	20	0.405d

C.V. 8.03. Note: averages with the same letter are statistically equal. * Line 26 for June León and line 27 for Dalia were used as controls.

). It was observed that line 18 (FJD121-3) had the highest yield with 1110 kg ha⁻¹, followed by line 5 (GPO82 82 FJ22-6) with 928 kg ha⁻¹. The stress to which the genotypes were subjected allowed the identification of outstanding materials.

The multivariate analysis of genotype-by-environment interactions (AMMIs) for the weight of one hundred seeds indicated that 41.28% of the variation was explained by the environment, 31.05% was explained by the genotype, and 27.66% was explained by the interaction (Figure 3). This indicates that seed weight is of intermediate to high heritability, as noted by ref. [14]. In addition, ref. [15] points out that the ideal genotype is statistically identified by having the longest vector length in the CP analysis and a value close to zero. According to ref. [16], the selection of germplasm using the model allows for the identification of genotypes with a pattern related to interaction, adaptation, and stability and is useful for eliminating materials with less stability. The lines that are aligned to the vertical axis have a similar weight per hundred seeds, and those on the same horizontal level present a similar genotype by environmental interaction (Figure 3). The results indicate that the lines present a wide difference in their weight per hundred seeds, and there were also lines with the same level of interaction. The lines with the highest weight of 100 seeds were 18 (FJD121-3), 21 (FJD121-7), and 22 (FJD121-8); however, due to their vector length and their distance from the origin, they are not considered stable. Lines 10 (GPO61 36), 25 (FJDGPO61 36), and 24 (FJD121-10) are the best responders due to their stability and low-level interaction with the environment.

In Figure 4, it can be observed that the lines in vertical projections show similar yield, while those located in the same horizontal position show a similar interaction. Line 5 (GPO82 82 FJ22-6) is identified as the one with the longest length of the vector for yield and with the value closest to zero, followed by lines 10 (GPO61 36) and 16 (FJD121-5). The lines with the greatest stability were 13 (GPO61 36), 4 (GPO82 82 FJ22-4) and 6 (GPO61 36). The lowest-performing line was 20 (FJD121-1), followed by 9 (GPO82 NO82-1 FJ35-1) and 25 (FJDGPO61 36).

Despite the results above, stable lines with higher-than-average yields were identified. According to ref. [7], the use of the AMMI model allows the identification of outstanding germplasm even in critical environments, which can be used in future crop improvement projects.

4. Discussion

The Flor de Junio bean type belongs to the Jalisco race [17]; as such, the landraces of this type are, in general, late-maturing bushes of indeterminate type three growth habit, all of which are photoperiod-sensitive and display phenological and morphological plasticity, such as flowering in flushes and the enhancement of flowering and maturity, which occur as sowing is delayed. This delay causes the crop to enter the reproductive stage as a small plant with low possibilities for high yield. The improved, tested lines were derived from biparental crosses within the same racial genetic pool; they have a medium growth cycle, are disease-resistant, and display some plasticity. Despite these defensive traits, the genotypes tested displayed important GEIs for the 100-seed weight (38%), whereas for seed yield, the main effect was due to the environment (>88%). The environments encountered in the semiarid highlands of Mexico for rainfed dry beans are of short duration with multiple stresses during the growth cycle, and for this reason, the environmental effect recorded was large. Similar results were found by Ashango [18] in Ethiopia when evaluating 16 bean lines for seed yield performance across five locations for two years. Both the main and interaction effects were highly significant (p < 0.01), and the effects of the environment, line, and GEIs explained 81.06, 3.21, and 15.73% of the observed variation, respectively; the larger influence of the environment was highlighted alongside the need for the simultaneous consideration of performance and stability. In contrast, the results of Ligarreto [19] from trials carried out in the subhumid tropics of Colombia showed that the main effects of their yield trials were due to the genotypes (53.14%) followed by the effects of the GEI (30.99%) and a lower effect was found for the test environments. In their trials, they included two bush determinate genotypes, most probably from the Nueva Granada race and five climbers; this strong contrast between the genotypes tested and more favorable environments may explain the large effect of the genotypes on the total variation in the trials.

The combined analysis for 100-seed weight and yield per hectare showed highly significant differences between locations and for the genotype–environment interaction, indicating the genotype’s differential responses in the test environments. For this reason, the multivariate analysis of the genotype-by-environment interaction (AMMI) model was used since this model allows the identification of outstanding materials with respect to the additive model [9]. This analysis allowed us to select genotypes with low genotype-by-environment interactions, and we identified lines with stability in the tested environments [8]. In this sense, the stability analyses conducted by ref. [18] including the AMMI test, allowed for the identification of some broadly adapted high-seed-yielding lines. The GEI indicates differences in adaptation and can be used to identify lines with specific adaptation if the trend of a specific adaptation in the genotypes is repeatable over years [14]; however, in their study, the specific adaptability trend was not repeated over years as different environments were grouped differently in the two years of study. Therefore, the GEI could not be of use and was ignored; instead, they selected the best lines with broad adaptation.

In the process of selecting genotypes under rainfed conditions, adaptability, stability, and performance in various environments (years and locations) are important and must be taken into consideration for the selection of new cultivars [6], particularly in drought-prone areas. In this study, lines 4 (GPO82 82 FJ22-4), 6 (GPO61 36), and Dalia (control) had intermediate performance in comparison to the best lines and no interaction with the environment; therefore, they were considered stable across locations. In a study carried out in Uganda [20], despite the high heritability value, the effect of the environment (seasons and locations) had a significant effect (p < 0.05) on the 100-seed weight. In this study, the significance of the GEI for this trait was caused by the different stresses encountered by the crop at both test locations during the reproductive phase. The trial at Celaya was subject to terminal moisture stress and a severe attack of leafhoppers (Empoasca kraemeri) during the whole growth cycle. At Zacatecas, the trial was under intermittent moisture stress. Despite the above, stable lines with higher-than-average yields were identified. According to ref. [7], the use of the AMMI model allows for the identification of outstanding germplasms, even in critical environments, which can be used in future crop improvement projects.

According to Darkwa et al. [21], one of the main challenges in bean cultivation in rainfall environments is the accurate identification of drought-tolerant materials, followed by hybridization programs to obtain new materials with enhanced drought tolerance. Schneider et al. [22] and Tosquy et al. [6] pointed out that in the selection of bean germplasm with drought tolerance, the average of the population should be taken into consideration, and then the selection of lines should be based on seed yield. The low bean seed yield observed in drought-prone areas is attributed to what was pointed out by Rosales et al. [23], who indicated that if the bean plant is exposed to water stress in the reproductive stage, flowers and small pods will abort, and a reduction in yield is generated, sometimes accompanied by a reduction in seed size.

To improve dry bean yield and 100-seed weight, the presence and nature of GEIs must be estimated in a reliable way to identify key factors that can affect production. Cultivars regionally adapted can be improved genetically by the identification of crossover interactions that imply the sufficient variability of high heritable traits such as 100-seed weight and disease resistance [4]. In tropical areas where moisture availability is not a problem, multiple disease resistance is key for wide adaptation and yield stability [4,6]. A better understanding of how to manage GEIs requires the consideration of the effects of habit growth and adaptation on the expression of seed yield. According to ref. [24], breeding common beans for increased yield can only be accomplished successfully within the framework of very specific constraints of habit growth, seed size, maturity, and gene pools. In the semiarid highlands of Mexico, the only growth habit used is the indeterminate type three, which is a type that can, if conditions permit, cover the soil and diminish direct moisture evaporation.

Considering the results observed for 100-seed weight and yield at the two environments, three lines were outstanding: 5, 10, and 16; these must be further tested across the region to select one to be registered as a new drought-tolerant cultivar for the “Flor de Junio” bean type.

5. Conclusions

The stress caused by lack of rainfall has been the norm for three out of five years in the semiarid highlands of central Mexico. The drought stress observed during the conduction of the trials in both test locations, terminal at CEBAJ and intermittent at Zacatecas, limited phenology by accelerating maturity and reduced 100-seed weight and the yield of the tested lines. In addition, at CEBAJ, there was a severe attack by leafhoppers during the whole cycle that needed constant control. Despite all these stresses, lines 5, 10, and 16 were outstanding in their results. In particular, considering 100-seed weight and yield at both locations, line 10 was outstanding and could be used as a parent in a future hybridization program or developed into a new cultivar with drought tolerance. The AMMI model allowed the identification of stable lines for the 100-seed weight and yield based on the genotype’s level of interaction with the tested environment.