Introduction
French bean (Phaseolus vulgaris) is an important legume vegetable crop; green tender beans without parchment are used as vegetables and seeds as pulse (Graham and Ranalli, Reference Graham and Ranalli1997; Chaurasia, Reference Chaurasia and Jaiswal2020). French bean is a historic crop in terms of crop evolution and domestication (Debouck et al., Reference Debouck, Toro, Paredes, Johnson and Gepts1993; Gepts, Reference Gepts1998; Bitocchi et al., Reference Bitocchi, Rau, Bellucci, Rodriguez, Murgia, Gioia, Santo, Nanni, Attene and Papa2017), and germplasm was spread to different parts of the world. Enormous genetic variations were observed in French bean germplasm in global collections and efforts were made to conserve germplasm in different parts of the world. The performance of germplasm for different growing seasons is determined by edaphic and weather parameters such as day and night temperature, day length and relative humidity (Chapman et al., Reference Chapman, Cooper, Hammer and Butler2000; Malhotra et al., Reference Malhotra, Singh and Erskine2007; Bulyaba et al., Reference Bulyaba, Winham, Lenssen, Moore, Kelly, Brick, Wright and Ogg2020; Canales et al., Reference Canales, Montilla-Bascón, Gallego-Sánchez, Flores, Rispail and Elena2021). In India, the rabi season (October to February months) is ideal for French bean production due to the cooler day and night temperature (Gepts, Reference Gepts1998; Beebe, Reference Beebe and Janick2012).
French bean is a rabi season crop in the Indian Peninsular, Deccan Indian Plateau and the Gangetic Plain regions in northern India. Off-season cultivation of French bean is generally followed in the North Indian Hill regions (Raghuvanshi and Sharma, Reference Raghuvanshi and Sharma2016; Das et al., Reference Das, Poonia, Jha and Goyal2020), which fetches higher prices to farmers. Bangalore region in India is located in an elevated area of 922 m above the mean sea level (MSL) and bestowed with moderate climatic conditions (Mohapatra, Reference Mohapatra2002; Rajashekara, Reference Rajashekara2019, Reference Rajashekara2020). The maximum temperature during April is 32.65–35.34°C, and the mean temperature in January is 15.27–17.11°C (Rajashekara, Reference Rajashekara2020). The average rainfall in the Bangalore region ranges from 750 from 800 mm and is mostly distributed in the monsoon period from June to September (Mohapatra, Reference Mohapatra2002; Rajashekara, Reference Rajashekara2019). Therefore, Bangalore and adjoining regions are bestowed with moderate winter and summer seasons making an ideal climate for French bean cultivation throughout the year.
Many biometrical approaches such as best linear unbiased prediction (BLUP), yield-related environmental maximum (YREM), additive main effects and multiplicative interactive effect stability model (AMMI) stability analysis and AMMI-based stability parameters were used in plant breeding to understand the genotype × environment interaction (GEI), and helpful in the identification of genotypes and germplasm with stable performance across environments (Yan, Reference Yan1999; Purchase et al., Reference Purchase, Hatting and van Deventer2000; Ashwini et al., Reference Ashwini, Ramesh and Sunitha2021; Mahadevaiah et al., Reference Mahadevaiah, Hapase, Sreenivasa, Hapase, Swamy, Anilkumar, Mohanraj, Hemaprabha and Ram2021). AMMI stability model is a multiplicative model, that predicts the nonlinear multiplicative effects in two main factors viz. environments and genotypes (Gauch, Reference Gauch2013; Elias et al., Reference Elias, Robbins, Doerge and Tuinstra2016). BLUP estimates the true genetic effect of genotypes in multi-environment trials (METs) (Robinson, Reference Robinson1991; Piepho, Reference Piepho1994, Reference Piepho1998a, Reference Piepho1998b; Piepho et al., Reference Piepho, Möhring, Melchinger and Büchse2008), and YREM is a simple, intuitive tool helpful in predicting the unpredictable components of GEI and crossover interactions (Yan, Reference Yan1999). AMMI stability value (ASV) and genotypic selection index (GSI) assist in choosing the genotypes based on interactive environmental principal components (PC) and ranks derived from mean and ASV respectively (Purchase et al., Reference Purchase, Hatting and van Deventer2000; Ashwini et al., Reference Ashwini, Ramesh and Sunitha2021). There are several studies that demonstrated the wide utilization of stability models in French bean related to grain yield (Joshi et al., Reference Joshi, Mehra and Station1993; Mekbib, Reference Mekbib2003; Nimbalkar et al., Reference Nimbalkar, Baviskar and Bajaj2006; Ligarreto-Moreno and Pimentel-Ladino, Reference Ligarreto-Moreno and Pimentel-Ladino2022; Mutari et al., Reference Mutari, Sibiya, Gasura, Kondwakwenda, Matova and Chirwa2022) and a few studies in snap bean (Vilela and Gravina, Reference Vilela and Gravina2011; Traka-Mavrona et al., Reference Traka-Mavrona, Georgakis and Koutsika-Sotiriou2014; Rebecca et al., Reference Rebecca, Oliveira, Gravina, De, Araujo, Araújo, Pureza, Junior, Vivas and Daher2018).
ICAR-Indian Institute of Horticultural Research (IIHR) Bangalore is bestowed with ideal climatic weather and is suitable for French bean cultivation throughout the season or all around the year. However, there will be differences in genotypic performance across seasons and the identification of germplasm or varieties with stable performance across seasons helps breeders for further utilization in crop improvement programmes. Hence, we evaluated 86 germplasm accessions with determinate growth habits for green pod yield attributing traits in three different seasons and identified better adapted germplasm to specific seasons and across seasons. These accessions are useful for utilization in crop improvement programmes.
Materials and methods
Materials and experimental design
A total of 86 germplasm, which includes accessions of both common bean and string bean with determinate growth habits were used as study material and these germplasm were maintained at ICAR-IIHR, Hesaraghatta Lake Post, Bangalore (online Supplementary Table S1). The germplasm accessions were evaluated in randomized block design with two replications for three seasons during rabi, 2022–2023 (November 2022 to February 2023), summer (March to May 2023) and kharif (June to October 2023) seasons. Meteorological data for rabi, summer and kharif seasons are presented in online Supplementary Figs. S1–S3. The experiments were conducted on the raised beds, each bed measuring 6.0 m in length and 70 cm in width. The recommended basal dosage of NPK fertilizer (25 kg of nitrogen, 75 kg of phosphorous and 30 kg of potassium) was supplemented followed by covering with a layer of soil, drip laterals (16 mm inline laterals with a spacing of 40 cm emitters and 4.0 litre per hour (LPH)) were laid and each bed was covered with 30 μm thick UV-stabilized polyethene mulch sheet. Each accession was sown on one bed in paired rows and 80 seeds were dibbled on raised beds with a spacing of 30 cm between rows and 15 cm between the plants. Topdressing with 25 kg of nitrogen and 30 kg of potassium was applied after 3rd week of sowing. The crop was irrigated regularly as per the requirement and recommended crop management practices were adopted (Reddy et al., Reference Reddy, Nair, Senthilkumar, Hebbar, Aghora, Pichaimuthu, Varalakshmi, Singh, Dhananjaya, Prasanna, Padmini, Rao, Thangama, Rajashankar, Mahadevaiah, Raghu, Ponnam, Hegde, Mishra, Sridhar and Kumar2023).
Recording of observations
Green pod yield attributing traits such as days to flowering (DFL), pod length (PL, cm), pod width (PW, mm), 20 pod weight (TPW, g) and number of pods per plant (PN) were recorded in our studies. DFL for each accession was recorded as the number of days taken by 50% of plants for flowering; PL was measured from pedicel to tip of the pods and PW was measured using callipers in the pod having the maximum width on randomly selected five green tender pods. TPW was measured as the weight of 20 green tender pods and PN was recorded on the numbers of pods at physiological maturity on five randomly selected plants. Pod yield per plant (PY, g) was recorded from five randomly selected plants.
Statistical analysis
Data from each season were analysed individually and also tested for the homogeneity of the error variance using the Levene test (Levene, Reference Levene and Olkin1960) by using the SPSS package (IBM Corp., Released 2016). Non-significance for homogeneity for error variance was observed for traits such as DFL (P value: 0.882), TPW (P value: 0.831), PL (P value: 0.684), PY (P value: 0.134) at both 95 and 99% levels of confidence and, PW (P value: 0.031) and PN (P value: 0.041) at 99% level of confidence indicates the homogeneity of variance across seasons and found ideal for combined statistical analysis across seasons.
We used CYMMIT developed software such as META-R (Alvarado et al., Reference Alvarado, Rodrígueza, Pachecoa, Burgueñoa, Crossaa, Vargasa, Pérez-Rodríguezd and Lopez-Cruze2020) and GEA-R (Pacheco-Gil et al., Reference Pacheco-Gil, Vargas, Alvarado, Rodríguez, Crossa and Juan2015) for the estimation of BLUP and AMMI stability analysis, respectively (CYMMIT Software Repository). ASV was derived by using the following formula (Purchase et al., Reference Purchase, Hatting and van Deventer2000):
where ${\rm S}{\rm S}_{{\rm IPC}{\rm A}_ 1}$
and ${\rm S}{\rm S}_{{\rm IPC}{\rm A}_ 2}$
are first and second interaction principal component effects of GSI, IPCA1 and IPCA2 are interactive principal component analysis (IPCA) scores for each genotype for the first and second components of GSI. The genotypic selection indices (GSIi) help to select the stable genotypes based on rank of trait mean and ASV (Purchase et al., Reference Purchase, Hatting and van Deventer2000; Ashwini et al., Reference Ashwini, Ramesh and Sunitha2021) was estimated using the following formula:
where rYi and rASVi are the rank of trait mean and ASV of the ith genotype for a given trait. The YREM helps to identify the stable genotype by comparing the yield loss due to the crossover interaction of seasons and genotypes (Yan, Reference Yan1999). The replication-wise YREM of green pod yield and attributing traits was estimated using the following formula:
where Yij is YREM averaged across the seasons for the ith genotype, Xij and MAXj are replication-wise trait mean and maxima for ith genotype and jth season.
Results
Genetic variability studies
The genetic variability parameters such as genotypic variance, genotype × season interaction variance, broad sense heritability and correlation were estimated for each season and also across seasons. The genetic variability for pod quality traits observed in germplasm is presented in Fig. 1. The estimates of heritability were highly influenced by seasons. The broad-sense heritability ranged from 0.58 for DFL in rabi and summer seasons to 0.96 for DFL in kharif seasons (online Supplementary Table S2) and for across seasons, it ranged from 0.17 for PN to 0.87 for PL. The genetic variance was significant for all the traits in all seasons, whereas genotype–season interaction variances for DFL, TPW, PL and PW were significant at 1% level of significance.
Diversity for pod quality-related traits in germplasm of French bean.
The phenotypic and genotypic correlation was significant for PY with TPW (0.924 and 0.669) and PL (0.806 and 0.506) in all seasons and across seasons. However, the genotypic correlation of PY has a significance with PN (0.558) across seasons at 1% level of significance (online Supplementary Table S3). PY showed significant genetic correlation with PN also in rabi (0.600), summer (0.705) and kharif (0.705) seasons. Besides, high phenotypic correlation was observed between PN and DFL (0.999), TPW and PL (0.802), TPW and PN (−0.532), PW and PL (−0.290), PL and PN (−0.379) and PW and PN (−0.575) at 1% level of significance.
The environmental correlation helps in understanding the impact of environments or seasons on the stability of trait expression. In our studies, we found that the genotype–environment correlation between seasons was non-significant (online Supplementary Table S4). The phenotypic correlation between kharif and summer seasons was highly significant for PN (0.296), PW (0.798), PL (0.758), TPW (0.465) and DFL (0.364). Similarly, the phenotypic correlation between rabi and summer seasons for DFL (0.284), PL (0.620), PW (0.625) and rabi season with kharif season for TPW (0.545), PL (0.753) and PW (0.619) were highly significant.
Mean performance
The mean performances of 76 determinate accessions and 10 standard varieties showed very significant differences for pod yield and its attributing traits. The best-performing top 10 accessions and 10 standard varieties are presented in online Supplementary Table S5. The highest PY of 165.66 g/plant was recorded by the accession IIHR-11 followed by IIHR-229 (163.71 g/plant) and IIHR-101-2 (161.75 g/plant) as compared to standard varieties Amber (141.88 g/plant) and Arka Bold (141.65 g/plant). With regards to the number of pods/plant, the accession IIHR-11 recorded the highest number of pods of 26.95 pods/plant followed by IIHR-101-2 (25.48 g/plant) and found on par with standards such as Amber (26.64 pods/plant) and Arka Komal (20.52 pods/plant). The PL ranged from 7.96 to 16.28 cm with a mean of 12.64 cm. The mean DFL across seasons was 33.69 d with a range of 30.50–42.50 d.
Analysis of variance for AMMI stability analysis
Analysis of variance of AMMI stability model showed significant genetic variation for seasons, genotypes and season–genotype interaction (SGI) for pod yield and its attributing traits (Table 1). The phenotypic variation due to the component ‘seasons’ explained 61.86, 50.16, 45.61 and 53.41% of variation for PY, PN, TPW and DFL and the component ‘genotypes’ accounted for 56.67 and 73.77% of variation for PL and PW, respectively. IPCA1 and IPCA2 were significant for all the traits at 5% level of significance. The IPCA1 contributed more than 60% of SGI, ranging from 62.16% for TPW to 74.20% for PN and the contribution of IPCA2 ranged from 25.80% for PN to 37.84% for TPW. The genetic variation due to SGI was further partitioned into SGISignal and SGINoise. The strong GSI signal was observed for PY, PN, TPW and DFL accounting for 90.46, 90.90, 84.69 and 72.22%, respectively (Table 2). Similarly, for PL and PW, noise captured 52.45 and 64.07% as compared to signal accounting to 47.55 and 35.93%, respectively.
Analysis of variance for green pod yield and attributing traits using AMMI stability models in French bean germplasm
SGI, season–genotype interaction; IPCA, interactive principal component analysis; MSS, mean sum of square; PV, per cent phenotypic variation; df, degree of freedom; PY, pod yield per plant (g); PN, number of pods per plant; TPW, 20 pod weight (g); PL, pod length (cm); PW, pod width (mm); DFL, days to flowering.
* Significant at 5% level, **significant at 1% level.
Analysis of variance and per cent variation due to signal and noises for green pod yield and attributing traits in French bean germplasm
SS, sum of the square; PV, per cent phenotypic variation; PY, pod yield per plant (g); PN, number of pods per plant; TPW, 20 pod weight (g); PL, pod length (cm); PW, pod width (mm); DFL, days to flowering.
AMMI stability value
ASV quantifies the stability and ranks the genotypes based on their stability, and the lowest ASV signifies the most stable genotype (Purchase et al., Reference Purchase, Hatting and van Deventer2000). In our studies (online Supplementary Table S6), the lowest ASV for PY was recorded by IIHR-50 (0.058) followed by IIHR-101 (0.061) and IIHR 223 (0.098), ranked 1st, 2nd and 3rd, respectively as compared to standard varieties such as Kanchan (0.121, ranked 5th) and Arka Suvidha (0.269, ranked 30th). For PN, the lowest ASV recorded by IIHR-223 (0.016), ranked first followed by IIHR-79 (0.028, ranked 2nd) and IIHR-117 (0.048, ranked 3rd) as compared to the standard variety Arka Anoop (0.050, ranked 4th), Kanchan (0.131, ranked 12th). With regards to TPW, standard variety Kanchan (0.242, ranked 18th) and Arka Komal (0.250, ranked 21st) recorded the lowest ASV and among the germplasm, the accession IIHR-7 recorded the lowest ASV of 0.084 ranked first followed by GC773/EC931873 (0.112, ranked 2nd) and the accession IIHR-101 (0.127, ranked 3rd). Similarly, IIHR-94-2 (0.072) recorded the lowest ASV for PW, IIHR-21 (0.086) for PL and three accessions GC787/EC931857, IIHR-111 and IIHR-131 for DFL (0.003). The estimates of ASV for rabi, kharif and summer seasons for pod yield and yield attributing traits show that the summer season is the most discriminating season for PY (0.988), PN (0.865), PW (1.002), DFL (0.901) and kharif season for TPW (0.980) and PL (0.946).
Genotypic selection indices
GSIs are derived from the sum of the ranks of ASV and trait mean. The genotype with the lowest GSI is most desirable as they are combined with a high mean value and the stability indices (Farshadfar et al., Reference Farshadfar, Mahmodi and Yaghotipoor2011). In our studies, the estimates of GSIs and their trait mean values for green pod yield and attributing traits are presented in online Supplementary Table S7. The lowest GSI was recorded by IIHR-155 (18) for PY, 30 for PN and 56 for TPW with mean pod yield of 143.63 g/plant across seasons and 211.70, 68.73 and 150.45 g/plant during rabi, summer and kharif seasons, respectively. The second lowest GSI (28) is recorded by IIHR-228 with a mean pod yield of 137.59 g/plant across seasons and pod yield of 203.08, 79.85 and 129.83 g/plant in rabi, summer and kharif seasons, respectively. Among the standards, the lowest GSI value was recorded by Arka Suvidha (55) and Kanchan (60). The lowest GSI recorded for PN were IIHR-79 (22), IIHR-155 (30), IIHR-76 (33), IIHR-229 (35) and significantly lower as compared to the standards such as Arka Anoop (36) and Amber (54). Similarly for TPW, the lowest GSI recorded by IIHR-7 (3), IIHR-32 (18), IIHR-8-2 (20) and IIHR-13 (23) as compared to the standards such as Kanchan (95) and Arka Komal (106).
Best linear unbiased prediction
BLUP refers to the estimation of random effects in mixed linear models and predicts the true genetic effects averaged over all genotypes (Piepho et al., Reference Piepho, Möhring, Melchinger and Büchse2008; Molenaar et al., Reference Molenaar, Boehm and Piepho2018) and is presented in online Supplementary Table S8. The highest BLUP for PY was recorded by IIHR-11 with 127.97 g/plant followed by IIHR-229 (127.39 g/plant) and IIHR-101-2 (126.81 g/plant), significantly higher than standards such as Amber (120.90 g/plant) and Arka Bold (120.83 g/plant). With regards to PN, BLUP ranged from 18.15 to 20.44 pods per plant and the maximum BLUP was recorded in the accession IIHR-10 (20.44 pods/plant) followed by IIHR-11 (20.33 pods/plant) and GC896 (20.32 pods/plant) as compared to the standards Amber recorded the BLUP with 20.28 pods/plant followed by Arka Bold (18.99 pods/plant) and Arka Suvidha (18.82 pods/plant). With regards to PWT, the estimates of BLUP ranged from 36.69 to 71.91 g, the highest BLUP was recorded by IIHR-170 (71.91 g/plant), IIHR-66-1 (68.52 g/plant) and IIHR-96 (67.07 g/plant) as compared to the standards Arka Bold (64.49 g/plant), Arka Suvidha (62.03 g/plant) and Arka Anoop (59.74 g/plant). The estimates of BLUP for PL ranged from 8.54 to 15.83 cm and the accession IIHR-170 recorded the highest BLUP of 15.83 cm followed by IIHR-56 (15.46 cm) and IIHR-150 (15.01 cm). Among the standards, Arka Komal recorded the maximum BLUP with 13.53 cm followed by Arka Sharath (14.17 cm), Arka Anoop (13.38 cm) and Triloki (13.29 cm). The estimates of BLUP for PW and DFL ranged from 6.94 to 14.79 mm and 32.10 to 38.06 d.
YREM for green pod yield and its attributing traits
YREM is a simple statistical analysis that assists the breeders in selecting the genotypes with the lowest unpredictable components of GEI. The genotypes with YREM nearing unity have the lowest or lesser crossover interactions and genotypes with the lowest YREM or nearing zero indicate having the maximum crossover interactions (Yan, Reference Yan1999). In our studies, the highest YREM was observed for IIHR-101-2 (0.7775) followed by IIHR-11 (0.7553), IIHR-170 (0.7327) for PY, IIHR-11 (0.7385), IIHR-10 (0.7186) and IIHR-101-2 (0.6795) for PN and IIHR-170 (0.8907), IIHR-96 (0.8369) and IIHR-66-1 (0.8213) for TPW (online Supplementary Table S9). Among the standards, Amber recorded the highest YREM of 0.5934 followed by Arka Bold (0.5820), Arka Komal (0.5624), Arka Anoop (0.5594) and Arka Suvidha (0.5345) for PY, Amber (0.6671) and Arka Komal (0.4793) for PN and Arka Bold (0.7656), Arka Suvidha (0.7159) and Arka Suvidha (0.7159) for TPW.
Prediction of stable genotypes based on BLUP, AMMI and YREM
The stable performing genotypes are having the lowest ASV and located nearer to the origin in AMMI biplots of PC1 versus PC2 (Purchase et al., Reference Purchase, Hatting and van Deventer2000; Gauch, Reference Gauch2013), but they may not be high-yielding accessions. Hence, stable accessions that are located near the origin in the AMMI biplots (PC1 versus PC2) with lowest values of ASV, high trait mean and BLUP and YREM help to predict the most stable genotypes. In our studies, the AMMI biplots (PC1 versus PC2 and PC1 versus train mean) were presented in online Supplementary Fig. S4 and data related to accessions with high mean and BLUP are given in Table 3. For pod yield two genotypes such as IIHR-155 and IIHR-228 located near the origin in the biplot (PC1 versus PC2), high pod yield, BLUP and yield loss due to crossover interaction as indicated by YREM as compared to the standard Arka Suvidha. Similarly, among the five stable accessions (Table 3), four accessions (IIHR-155, IIHR-76, IIHR-131 and IIHR-229) recorded the higher pod number and pod yield as compared to the standard cultivar Arka Komal. With regards to TPW, two accessions (IIHR-155 and IIHR-32) were located near the centre of origin of biplots with higher TPW combined with higher PY. Among all three traits, IIHR-155 (Fig. 2) was recorded as the lowest ASV, located near the origin of the biplots combined with higher PY and trait mean. Besides, the accession IIHR-11 recorded the highest pod yield (165.66 g/plant), the highest BLUP (127.97 g/plant), 38th for ASV (0.319), ranked 6th for GSI (39) and 2nd for YREM for pod yield shall be also considered as the most promising and stable genotype for PY and PN. The accession IIHR-170 recorded the highest mean (79.72 g), the highest BLUP (71.91 g), ranked 43rd for ASV (0.342), 9th for GSI (47) and 1st for YREM (0.8907) is considered as the promising genotype for TPW.
Best performing accessions identified based on AMMI stability parameters, BLUP and YREM in French bean germplasm
BLUP, best linear unbiased prediction; DFL, days to flowering; PWT, pod weight (g); PL, pod length (cm); PW, pod width (mm); PN, pod number per plant; PY, pod yield per plant (g).
Plant growth habit and pod quality parameters in French bean accessions IIHR-11 and IIHR-155.
Discussion
The salient findings in our studies revealed the significant genotypic variability and stability for pod yield and its attributing traits in kharif, rabi and summer seasons among the determinate growth type French bean germplasm. The rigorous stability analysis using various stability models such as AMMI, ASV, GSI, YREM and BLUP helped to identify the most stable accessions across seasons and better performing season-specific accessions. Most of the stability analyses in French bean were related to seed yield and its attributing traits (Joshi et al., Reference Joshi, Mehra and Station1993; Nimbalkar et al., Reference Nimbalkar, Baviskar and Bajaj2006; Ligarreto-Moreno and Pimentel-Ladino, Reference Ligarreto-Moreno and Pimentel-Ladino2022; Mutari et al., Reference Mutari, Sibiya, Gasura, Kondwakwenda, Matova and Chirwa2022) and also for micronutrients such as zinc and iron contents in seeds (Mutari et al., Reference Mutari, Sibiya, Gasura, Kondwakwenda, Matova and Chirwa2022), and a few studies in snap bean (Vilela and Gravina, Reference Vilela and Gravina2011; Traka-Mavrona et al., Reference Traka-Mavrona, Georgakis and Koutsika-Sotiriou2014; Rebecca et al., Reference Rebecca, Oliveira, Gravina, De, Araujo, Araújo, Pureza, Junior, Vivas and Daher2018).
AMMI stability model partitions the total GEI into signals and noises, and to dissect the appropriateness of its application in MET data analysis (Gauch, Reference Gauch2013) and these parameters were used for characterization germplasm in many crops (Ali et al., Reference Ali, Elsadek and Salem2018; Ashwini et al., Reference Ashwini, Ramesh and Sunitha2021; Shimray et al., Reference Shimray, Bharadwaj, Patil, Sankar, Kumar, Priya, Reddy, Singhal, Hegde, Parida, Roorkiwal, Varshney and Verma2022; Verma et al., Reference Verma, Kumar, Rymbai, Talang, Devi, Baiswar and Hazarika2023). In our studies, traits such as DFL, TPW, PN and PY explained by more than 50% of SGI signals elucidate the worthiness of the application of the AMMI model in the present study. Similar studies related the assessing of the appropriateness of AMMI in the analysis of MET data and signal-rich GEI were reported in French bean (Salegua et al., Reference Salegua, Melis, Fourie, Sibiya and Musvosvi2020; Gelete et al., Reference Gelete, Negash, Tsegaye and Teshome2022) and in other related vegetable and legume crops (Shimray et al., Reference Shimray, Bharadwaj, Patil, Sankar, Kumar, Priya, Reddy, Singhal, Hegde, Parida, Roorkiwal, Varshney and Verma2022; Verma et al., Reference Verma, Kumar, Rymbai, Talang, Devi, Baiswar and Hazarika2023).
AMMI stability model is a robust statistical tool (Zobel et al., Reference Zobel, Wright and Gauch1988; Gauch and Zobel, Reference Gauch and Zobel1997) used to assess the stability in performance of germplasm across seasons and locations in many crops (Liu et al., Reference Liu, Fan, Huang, Yang, Zheng, Wang and Qiu2017; Ashwini et al., Reference Ashwini, Ramesh and Sunitha2021; Shimray et al., Reference Shimray, Bharadwaj, Patil, Sankar, Kumar, Priya, Reddy, Singhal, Hegde, Parida, Roorkiwal, Varshney and Verma2022; Verma et al., Reference Verma, Kumar, Rymbai, Talang, Devi, Baiswar and Hazarika2023). The interactive principal components such as IPCA1 and IPCA2 captured > 99% phenotypic variability (Gauch and Zobel, Reference Gauch and Zobel1997) and IPCA scores were used to assess the stability of germplasm (Alake, Reference Alake2018; Roselló et al., Reference Roselló, Villegas, Álvaro, Soriano, Lopes, Nazco and Royo2019; Balakrishnan et al., Reference Balakrishnan, Kumar, Raj, Verma, Touthang, Kumar, Rai, Da and Mishra2024). In our studies, 64.55% of total variability for PY was captured by IPCA1, 74.20% for PN, 62.16% for TPW, 68.28% for PW, 63.945% for PL and 70.98% for DFL. Similar results of the IPCA1 score accounting for the maximum variability of GEI were reported for different traits in French bean (Nimbalkar et al., Reference Nimbalkar, Baviskar and Bajaj2006; Dinsa et al., Reference Dinsa, Balcha and Tadesse2022; Ligarreto-Moreno and Pimentel-Ladino, Reference Ligarreto-Moreno and Pimentel-Ladino2022; Mutari et al., Reference Mutari, Sibiya, Gasura, Kondwakwenda, Matova and Chirwa2022) and related crops (Ali et al., Reference Ali, Elsadek and Salem2018; Shimray et al., Reference Shimray, Bharadwaj, Patil, Sankar, Kumar, Priya, Reddy, Singhal, Hegde, Parida, Roorkiwal, Varshney and Verma2022).
AMMI biplots are better judging criteria for identification of the most stable accessions (Gauch and Zobel, Reference Gauch and Zobel1997) and identify germplasm with general and specific adaption (Alake, Reference Alake2018; Roselló et al., Reference Roselló, Villegas, Álvaro, Soriano, Lopes, Nazco and Royo2019; Bomma et al., Reference Bomma, Shruthi, Soregaon, Anil, Suma, Pranati, Lohithaswa, Patil, Kumar, Sandeep, Anilkumar and Gangashetty2024). The relative position of genotypes on biplots of IPCA1 and IPCA2 scores represent the relative stability of genotypes (Gauch and Zobel, Reference Gauch and Zobel1997), and the biplots of IPCA1 and trait mean capture the main effect of trait mean and interaction effect with the environment (Gauch and Zobel, Reference Gauch and Zobel1997). The sum of squares of IPCA1 and IPCA2 were used for the estimation of ASV (Purchase et al., Reference Purchase, Hatting and van Deventer2000) and genotypes located in the origin of biplots have the lowest ASV values and show general adaption (Farshadfar et al., Reference Farshadfar, Mahmodi and Yaghotipoor2011). These parameters were used in the characterization and identification of the most stable germplasm in many crops (Ashwini et al., Reference Ashwini, Ramesh and Sunitha2021; Shimray et al., Reference Shimray, Bharadwaj, Patil, Sankar, Kumar, Priya, Reddy, Singhal, Hegde, Parida, Roorkiwal, Varshney and Verma2022; Verma et al., Reference Verma, Kumar, Rymbai, Talang, Devi, Baiswar and Hazarika2023; Balakrishnan et al., Reference Balakrishnan, Kumar, Raj, Verma, Touthang, Kumar, Rai, Da and Mishra2024; Bomma et al., Reference Bomma, Shruthi, Soregaon, Anil, Suma, Pranati, Lohithaswa, Patil, Kumar, Sandeep, Anilkumar and Gangashetty2024). In our studies, IIHR-155 located near to the origin of the biplot for PY, PN and IIHR-32 for TPW was deduced as germplasm with stable performance across seasons.
BLUP estimates the true genetic values or genetic merits or random effects of genotypes and is used in MET to identify the stable cultivars (Robinson, Reference Robinson1991; Piepho, Reference Piepho1998a; Piepho et al., Reference Piepho, Möhring, Melchinger and Büchse2008) and BLUP is statistically on par or superior over AMMI models in the identification of stable genotypes (Piepho, Reference Piepho1994). Many research scholars used BLUP in MET to identify the stable genotypes and germplasm characterization (Ashwini et al., Reference Ashwini, Ramesh and Sunitha2021; Anuradha et al., Reference Anuradha, Patro, Singamsetti, Rani, Triveni, Kumari, Govanakoppa, Pathy and Tonapi2022; Ajay et al., Reference Ajay, Kumar, Kona, Gangadhar, Rani, Rajanna and Bera2023). In our studies, accessions such as IIHR-11, IIHR-229, IIHR-101-2, IIHR-170 and IIHR-27 recorded higher BLUP for pod yield across three seasons as compared to the standards. YREM is a superior index in stability analysis and the deviation of YREM from 1.0 indicates the loss of trait value due to the unpredictable crossover interaction in MET (Yan, Reference Yan1999). YREM was used in germplasm characterization to identify stable accession in many crops (Ashwini et al., Reference Ashwini, Ramesh and Sunitha2021; Bomma et al., Reference Bomma, Shruthi, Soregaon, Anil, Suma, Pranati, Lohithaswa, Patil, Kumar, Sandeep, Anilkumar and Gangashetty2024). In our studies, accessions such as IIHR-101-2, IIHR-11, IIHR-170 and IIHR-229 recorded the highest YREM for PY indicating minimal crossover interactions as compared to other genotypes.
The prediction of several parameters such as trait mean, position of the accession on biplot, the lowest ASV, GSI and BLUP data are better criteria to identify the stable germplasm as compared to single parameters (Purchase et al., Reference Purchase, Hatting and van Deventer2000; Piepho et al., Reference Piepho, Möhring, Melchinger and Büchse2008; Farshadfar et al., Reference Farshadfar, Mahmodi and Yaghotipoor2011; Molenaar et al., Reference Molenaar, Boehm and Piepho2018). In our studies, IIHR-155 and IIHR-228 were positioned nearer to origin on the biplot, higher mean and higher BLUP for PY, IIHR-155, IIHR-76, IIHR-131 and IIHR-229 for PN and two accessions such as IIHR-155 and IIHR-32 for TPW are deduced as the most stable germplasm accessions as they are located nearer to the origin in the AMMI biplot (PC1 versus PC2), lowest ASV and GSI ranking with higher PY combined with high BLUP and trait mean. Among all the accessions, IIHR-155 was inferred as the most stable for all three traits. Apart from the above, the accession IIHR-11 ranked first for trait mean and BLUP, 6th ranking for GSI and 38th ranking for ASV for PY also considered the most stable germplasm accession.
Conclusion and future line of work
French bean is a cool season crop and mostly cultivated as a rabi season crop in Indian plain regions and in kharif season North Indian hilly regions. Bangalore is a moderately high-altitude region located at 922 m above the MSL and moderate climatic condition makes it suitable for vegetable cultivation throughout all three seasons. To identify the most stable genotypes across three seasons such as kharif, rabi and summer seasons, we evaluated the 86 germplasm and we identified IIHR-155 as the most stable genotype across seasons based on AMMI stability indices, GSIs and IIHR-11 based on estimates of BLUP values. However, these data collected one a time in each season and the results obtained from the present study are preliminary only. Further studies are required to evaluate across seasons in different locations for accurate assessment and further identified genotypes shall be utilized in the crop improvement programmes for the development of stable genotypes suitable for cultivation in all three seasons.
Supplementary material
The supplementary material for this article can be found at https://doi.org/10.1017/S1479262124000637.
Acknowledgements
We thankfully acknowledge the ICAR-Indian Institute of Horticultural Research, Hesaraghatta Lake Post, Bengaluru, India for logistic and financial support for the study. We also thankfully acknowledge the contribution of Dr T.S. Aghora, Principal Scientist (Retired) for germplasm maintenance, Dr C. Anilkumar, Scientist, ICAR-National Rice Research Institute, Cuttack and Dr B.C. Ajay, Senior Scientist, ICAR-Directorate of Groundnut Research Regional Station, Anantapur for technical guidance in data analysis, Mr H.S. Suresh and Mr Devaraj for technical support, and Mrs Puttamma, Mrs Sunitha, Mrs Radha and Mrs Varalakshmi for their contribution to the field operation in the experimental field.
Author contributions
Conceptualization, funding acquisition, investigation, methodology, data curation, formal analysis, visualization, writing – original draft, writing – review and editing: C. M. and M. V. D. Both the authors read the manuscript and approved it.
Competing interests
None.
