Chickpea (Cicer arietinum L.) 2 n = 2 x = 16, is the third most important food legume belongs to family Fabaceae. It is globally, occupying an area of 14.56 m ha with a production of 14.78 million tons (Mt) (FAOSTAT, 2022). India is the largest producer of chickpea with 72% of area covering about 10.56 million ha and production of 11.17 million tones grain. The present yield level is 1077 kg/ha (Project Coordinator’s Report, All India Coordinated Research Project on Chickpea, 2021), which is far below the potential yield (5000 kg/ha) of the crop. Uttar Pradesh is a traditional state for chickpea cultivation covering an area of 501.0 thousand ha with production of 578.66 thousand tones and productivity of 1155 kg/ha while, the area under chickpea cultivation in Bihar is 58.22 thousand hectares with production of 67.19 thousand tons and productivity of 1154 kg/ha (Project Coordinator’s Report, All India Coordinated Research Project, Chickpea, 2021). This is only 4% of pulse production of the country. Particular reference to chickpea in Bihar, the area of chickpea has declined from 2.45 lakh in 1975 to 0.58 lakh hectares in 2017 although productivity has increased from 550 kg/ha to 1000 kg/ha during the same period. Highest productivity of 1430 kg/ha was recorded in Bihar state during 2012. The study of genotype × environment interaction in its biometrical aspect is thus important not only from the genetically evolutionary point of view but also it is very relevant to production problems of agriculture in general and to plant breeder in particular (Breese, 1969).  The seed yield of chickpea is influenced by many factors including genotype, growing season, geographical site and agronomic practices (Tawaha et al., 2005). The fluctuation in chickpea production may be affected by environmental changes and use of varieties that are not adapted to wide range of environments (Balapure et al., 2015). Genotypes are considered to be more adapted or stable if they show low degree of fluctuation in yielding ability under different environments (Dinkar et al., 2025). The chickpea production in India can be increased and stabilized by two approaches. The first one is stratification of chickpea growing areas followed by development of suitable varieties for target regions and the second one is development of cultivars with wide adaptability for its cultivation in diversified environments. The climatic factors, such as rainfall and temperature change from year to year even in the same region. Therefore, most suitable approach to attain stability would be development of widely adapted varieties with high yield potential.Genotype × environment interaction is the change in relative performance of genotypes across sites (De Lacy et al., 1996; Karimizadeh et al., 2021). G × E interaction should be investigated so that the breeder can decide to restructure the program to minimize the interaction effect, or exploit it to produce varieties with specific adaptation to particular environments (Eisemann et al., 1990). Eberhart and Russell (1966) model consists of three parameters, (a) mean yield over locations or seasons, (b) regression coefficient and (c) deviation from regression. According to this model a stable variety is one with a regression coefficient of unity (b=1) and a minimum deviation from the regression line (S² d=0). Using their definition a breeder would usually desire to develop a variety with high mean yield and satisfying the above requirements for stability (Singh and Narayanan, 2004). Present study was undertaken for estimation of the impact of genotypes environment interaction on grain yield of chickpea and to identify relatively stable genotypes across environments.

The experiment was conducted under two sowing environments, namely normal and late sowing, at SHUATS, Prayagraj and BAU, Sabour during Rabi 2019-20 and Rabi 2020-21. At SHUATS, Prayagraj, normal sowing was undertaken on 06 November 2019 and 16 November 2020, while late sowing was carried out on 15 December 2019 and 10 December 2020. At BAU, Sabour, normal sowing was performed on 12 November 2019 and 17 November 2020, whereas late sowing was done on 13 December 2019 and 08 December 2020 during the respective crop seasons.
       
Data were gathered on days to maturity, plant height (cm), number of primary and secondary branches per plant, number of effective pods per plant, biological yield per plant (g), harvest index (%), 100-Seed weight (g) and seed yield per plant.
       
Five randomly chosen plants were used to estimate all these characters. Three replications of the genotypes were used in a randomized block design. Before performing the combined analysis, a separate analysis of variance for seed yield and its component was conducted for each season. The genetic parameters and broad-sense heritability were estimated as suggested by Burton (1952). The G × E interaction and stability parameters were statistically analyzed using the models developed by Eberhart and Russell (1966).

Coefficient of variability, heritability and genetic advance
 
 A wide range of variability was observed for yield and yield attributing traits. The pooled analysis showed that grain yield per plant (15.38% and 11.84%), biological yield per plant (14.92% and 13.52%), effective pods per plant (20.00% and 16.59%), primary branches/plant (19.93% and 19.19%) and secondary branches per plant (17.78% and 16.70%) had moderate phenotypic and genotypic coefficient of variation, while other traits showed low coefficients of variation (Table 1). These results were in accordance with the result of Khan et al., (2011) and Babbar et al., (2012). Pooled estimations from the current study show that the percentage of heritability was higher for all traits (Table 1). For primary branches per plant, secondary branches per plant, effective pods per plant and biological yield per plant, demonstrated high heritability coupled with high genetic advancement as a percentage of mean. Similar results were also reported by Kumar et al. (2017); Babbar and Tiwari (2018); Saini et al., (2020) and Devi et al. (2021).


 
Stability analysis
 
The stability parameters-mean (X), regression coefficient (β₁) and deviation from regression (S²d_i)-were estimated for all traits. The partitioned analysis of variance (Table 2) revealed highly significant differences for genotype, environment (linear), pooled deviations, Environment + (Genotype × Environment) and Genotype × Environment (linear). Most traits exhibited significant deviations from linearity, suggesting strong environmental influence on their expression, consistent with Eberhart and Russell (1966). The ANOVA indicated significant genotypic differences across all traits over environments. Variance due to Environment + (Genotype × Environment) was highly significant for most traits, except secondary branches per plant and harvest index, when tested against pooled error. The Genotype × Environment interaction was also highly significant for most traits, except days to maturity, secondary branches per plant, biological yield per plant, harvest index and grain yield per plant. The mean square due to environment (linear) was highly significant for all traits, indicating that a large proportion of variation was explained by linear regression. Genotype × Environment (linear) effects were significant for most traits, except secondary branches per plant and harvest index, suggesting predictable performance. Pooled deviation mean squares were significant for all traits except primary branches per plant, indicating the importance of non-linear (unpredictable) components in Genotype × Environment interaction.Thus, both linear and non-linear components contribute to stability assessment. These findings are in agreement with earlier reports by Shivani and Sreelakshmi (2015); Sharma et al., (2017); Yadav et al., (2014) and Babbar and Tiwari (2018).


       
The genotypes had regression coefficient lesser than unity coupled with mean values less to grand mean revealed that above average stability of genotypes (Table 3 and 4).




       
Eberhart and Russell (1966) emphasized the need of considering both linear (bi) and non linear (S²d_i) components of G × E interaction in judging the stability of genotypes. An ideal genotype is defined as, one possessing high mean performance, with regression coefficient around unity (bi=1) and deviation from regression (S²d_i) close to zero. The stability parameters for fifty genotypes have been given in the Table 3 and 4. An overall study of stability parameters revealed that not a single genotype was ideally stable for all the characters. The stability parameters for seed yield per plant showed that five genotypes ICCV15112, BRC-1047-33, BRC-1009-84, BRC-1048-15, PBC501, BG3043 and PhuleG13110 were stable over the eight environments. Out of these the genotype ICCV15112 having highest grain yield per plant, bi=1 (unity) and S²d_i=0 and is found the best stable a for grain yield per plant along with other yield contributing traits like number of primary branches per plant and number of secondary branches per plant for over all eight environments. The genotypes BRC-3, BRC-7, BRC-9, PhuleG13110, H12-62, SAKI9516 and PG186 had regression coefficient lesser than unity coupled with mean values less to grand mean revealed above average stability of these genotypes for grain yield per plant. The genotypes BRC 1047, PBC 501, BG3043, GCP 105 and KWR 108 had the regression coefficient above unity and also with very low mean values over the environment indicating below average stability. These genotypes were found stable for grain yield per plant in unfavorable environment. The genotypes namely, Sabour chana-1, BRC1055-155, BRC1058-16, GNG2215, BRC1082-137, BRC1084-127, ICCV15112, GNG469, NDG14-24 and BRC1047-33 had the regression coefficients greater than one coupled with high mean values indicating specific adaptation of these genotypes for exploitation of character for grain yield per plant. Stability in the seed yield was earlier reported by many workers (Arshad et al., 2003; Prakash et al., 2006; Abbas et al., 2008) by using stability analysis identified some stable chickpea genotypes for different environments. This indicated that the efficiency of a breeding program aimed at yield improvement is impaired due to genotype by environment interaction, which complicates the process of crop variety development especially when varieties are selected in one environment and used in others.

High yielding genotypes BRC-3, BRC 1009-84, BRC-7, BRC-9, Phule G13110, SAKI9516 and PG 186 were clearly stable with regard to the majority of the yield-attributing traits, suggesting that the stability of these traits may be the cause of the observed grain yield. The aforementioned findings have made it possible to cultivate chickpeas in double cropping (after rice) under very late planting conditions. Additionally, it was determined that the genotypes ICCV15112, BRC-1047-33, BRC-1009-84, BRC-1048-15, PBC501, BG3043 and PhuleG13110 were the most stable and suited to the various environments. As a result, they could be incorporated into the hybridization program to converge the stability characteristics of grain yield in order to create stable cultivars that are multi-environment adapted.

All authors declare that they have no conflicts of interest.