Genotype x environment interaction and stability analysis for yield and its components in lentil

The objective of this investigation is to determine the performance and stability of 24 lentil (Lens culinaris Medik.) genotypes under a wide range of variable environments. The regression model and ecovalence (Wi) were used to analyze the response of the lentil genotypes to variable environmental conditions for yield and some of its components in six experiments in three seasons under two locations. Results indicated that both environmental conditions (E) and studied genotypic accessions (G) influenced significantly on the performance of yield and yield components. Moreover, the performance of genotypes varied highly significantly from environment to another for all traits, except 100 seed weight as proved by significance of G x E. Therefore, further stability analyses were performed for traits that recorded significant G x E. Four genotypes were stable for pods plant -1 either measured by Wi or S 2 d. For this trait, all genotypes were non responsive to environmental conditions except PL81-17 which may behave positively to pod bearing conditions. For seed yield plant -1 only Sinai 1 was significantly unstable measured by Wi and S 2 d, respectively. The significance of b's for seed yield feddan -1 proved that only 3 genotypes were responsive to environments. Two of them (XG88-17 and Giza 51) may behave better under good environments and the third (Giza 4) may be recommended under poor ones. It may be concluded in lentil breeding programs, which the performance of genotypes under each location should be evaluated firstly and those reliable ones will be tested for stability across various environmental conditions prior to recommendations.


INTRODUCTION
Lentil (Lens culinaris Medik.) is one of the most important food legume crops grown in the Mediterranean region and Middle East.According to FAO statistics of 2010, the acreage of lentil in Egypt is 1380 ha and yielded 2178 tons, which represents 2% of the national annual need.
The self-insufficiency of lentil may be attributed to several constraints that limit lentil production and cultivars potentiality.These factors include mainly the sensitivity of cultivars to various biotic and abiotic stresses such as weeds, fungal pathogens and other pests (Sarker et al., 2009), in addition to drought and salinity (Saxena et al., 1993;Hamdi et al., 2004).
In Egypt, several investigations were conducted to develop promising lentil cultivars using local and exotic accessions.The resultant materials were evaluated under variable conditions of Egyptian environments.The outcomes of these investigations were several cultivars, that is, Giza 9, Sinia 1 (Precoz) and Giza 370 which were recommended due to resistance to root rot and wilt, rainfed cultivation and multi-locations stability (Hamdi et al., 2003).
The evaluation of genotypes is the key to utilization in *Corresponding author.E-mail: samirrabie10@yahoo.com.Tel: +201007776331.Fax: +20235723476.
The genotype x environment interactions could be attributed to predictable and non-predictable effects as reported by Allard and Bradshaw (1964).The first one may be due to macro-environmental conditions, but the second is mainly caused by climatic and microenvironmental conditions.
Several methods were proposed to analyze G x E interaction to determine the stability of performance (Becker and Léon, 1988).The most frequently utilized two methods for detecting the stability are partitioning of G x E interaction of evaluated genotypes (Wricke, 1962) and the regression model (Eberhart and Russell, 1966).Eberhart and Russell (1966) considered two parameters for measuring the varietal phenotypic stability: the regression coefficient (b i ) and deviation from regression (S 2 d ).The variety with a (b i ) value not significantly different from unity (b i = 1) and (S 2 d ) value not significantly different from zero would be described as a stable variety.Therefore, the aim of the present investigation is to evaluate different genotypes and selections of lentil for yield and its components a n d to determine the G x E of these lentil genotypes under six variable environments of Upper and Lower Egypt during three seasons.
The experimental plot consisted of 10 rows, each was 30 cm apart and 3 m long (=9 m 2 ).Fertilizers were applied during seed bed preparation at the rate of 30 kg P2O5 and 15 kg N/fed.At harvest, number of pods/plant, number of seeds/pod, 100-seed weight and seed yield/plant were recorded using a random sample of 10 guarded plants from each plot.Seed yield/plot was used to calculate seed yield in ardab/fed (One ardab=160 kg and one fed = 4200 m 2 ).

Statistical analysis
The regular analysis of variance of RCBD as outlined by Gomez and Gomez (1984) was applied on the obtained data of each trial.The Bartelett's test of the homogeneity of error variances adopted indicated the validity of applying combined analysis of variance for the six environments.
To detect the differences among genotypes across all the studied environments (Ei), least significant difference (LSD) test was used.

Ecovalence
Ecovalence parameter (Wi) proposed by Wricke (1962) was used to calculate the ecovalence of each genotype as follows: SS refer to the summation of squares; xij = the mean performance of character on the i th variety in the j th environment; x i. = mean of i th variety across all environments; x.j = mean of j th environment across all varieties; x.. Grand mean and r = No. of replicates df is degrees of freedom; t signifies genotypes and p, environments.

Regression analysis
The performance of an individual genotype was regressed on the environmental index (deviation of the mean yield at that environment from the overall mean yield of all environments) as outlined by Eberhart and Russell (1966).In this model, the regression coefficient, bi and the deviation from regression mean squares (S 2 d) were considered as parameters of response and stability, respectively.

Significance of mean squares
The mean squares and their significance of each environment and combined across environments are presented in Table 2. Variances due to genotypes (G) were highly significant in all environments for all studied traits, except for seed/pod in 2009/2010 and 2010/2011 seasons at El-Gemmeiza and in 2010/2011 and 2011/2012 seasons at El-Mataana location.However, lentil genotypes varied highly significantly across all environments for all studied traits.
Environments as a source of variation in combined analysis affected highly significantly all the investigated lentil traits.
These results indicated that both environmental conditions and studied genotypic accessions influenced significantly on the performance of yield and yield components of lentil.Moreover, the performance of ( ) genotypes varied highly significantly from one environment to another for all traits except 100 seed weight, as proved by the significance of G x E interaction.From combined analysis, the magnitudes of variances due to environments were larger compared to those of genotypes, particularly for pods/plants (17 fold), seeds/pod and seed yield/plant (10 fold) and seed yield/fed (43 fold).Therefore, further stability analysis could be performed for traits that recorded significant G x E interaction.

Mean effects of environments
The mean performance and environmental index (I) of the investigated lentil genotypes in each environment are presented in Table 3.The environmental index used in this table is the deviation of each environment from the grand mean of all environments.
The prevailing conditions of 2009/2010 at El-Gemmeiza (1 st Env.) and 2011/2012 at El-Mataana (6 th Env.) recorded significantly the highest number of pods/plant, which was reflected in the highest positive environmental indices.Seeds per pod under 1 st , 3 rd and 5 th environments were significantly higher corresponded to the positive I's compared to those negative of other three environments.
Seed index for all seasons under El-Gemmeiza location was lighter with negative environmental indices, whereas those at El-Mataana were heavier with positive I's.
The plant seed yield was performed similar to number of pods per plant, which was only positive under 1 st and 6 th environments.Regarding the outcome product as seed yield per feddan in the 1st, 3rd, 4th and 6th environments was higher product ones with positive environmental indexes.The 1st and 6th environments showed these effects may be due to the positive effects of pods and seed yield per plant.
However, such positive effects on the seed yield per feddan due to other two environments (3 rd and 4 th ), may be due to the compensation of other yield components, that is, seeds per pod and seed index.Similar environmental effects on the performance of lentil genotypes and the G x E interactions were previously detected by several authors (Hamdi and Rabeia, 1991;Hamdi et al., 1995;Selim, 2000;Hamdi et al., 2002;Dehghani et al., 2008).Means in the same column followed by the same letters are not significantly different.

Stability analysis
The stability parameters and mean performance is the investigated lentil genotypes are presented in Table 4.The analysis of stability was performed only for pods and seed yield/plant and per fed using ecovalence and regression analysis was performed.The ecovalence (Wi) as a parameter of stability proposed by Wricke (1962) as outlined by Becker and Léon (1988), measures the G x E of each genotype.The significance of Wi means instability of performance across targeted environments and/or the stable genotype records Wi = 0.0 or possesses high ecovalence.The regression model of stability proposed by Ebarhart and Russell (1966), considered that b as a parameter of response and S2d indicates instability due to the deviation from zero.However, the significance of the coefficient of regression (b) means responsiveness either to favorable environment (b is more than unity) or poor ones (b is less than unity).
Accordingly, 4 genotypes (FLIP 96-19L, FLIP 200-18L, PL81-17 and XG88-6-1) were stable for pods/plant measured by Wi and S2d.For this trait, all genotypes were non-responsive to environmental conditions, except PL81-17 which may behave positively to pod bearing conditions.For seed yield per plant, only FLIP 95-51L and XG88-17 accessions were significantly unstable measured by Wi and S2d, respectively.Another two genotypes  and one genotype (XG88-1-1) may perform better for yield per plant under favorable and less favorable environments, respectively.This is due to the significant estimated b which is lesser than unity in the first case and more than one in the second situation.
Regarding the seed yield per feddan, all genotypes recorded significant W i (except FLIP 95-67L) and S 2 d .This indicates that out of 24 accessions, 23 were instable across the studied environments either measured by W i or S 2 d for seed yield per feddan.The significance of b's for seed yield per feddan proved that only 3 genotypes were responsive to environments.Two of them (XG88 17 and Giza 51) behaved better under good environments and the remaining one (Giza 4) may be recommended under poor ones.
To elucidate the interrelationship between mean performance and estimated stability parameters (W i and S 2 d ), the rank correlation was calculated for all studied traits.The obtained rank correlation coefficients were small in magnitude and insignificant (-0.10 0.13 for pods/plant, 0.19 and 0.31 for seed yield per plant and 0.34 and 0.38 for seed yield per feddan between mean and each of W i and S 2 d , respectively.Such insignificant correlations proved that the performance of the tested lentil genotypes was not related to the extent of stability measured either by W i or S 2 d .Thus, it may be concluded in lentil breeding programs that the performance of genotypes under each location should be evaluated firstly and those showing reliable means will be tested for stability across environmental conditions prior to recommendations.

Table 1 .
Code and pedigree of the studied lentil genotypes.
Source: Food Legume Crops Res.Dep., FCRI, ARC, Egypt.ICARDA = International Center for Agricultural Research in the Dry Areas.

Table 2 .
Mean squares due to different sources of variation of each individual environment and combined across environments for seed yield and its components of 24 genotypes under El-Gemmeiza and El-Mataana locations during 2009/2010 through 2011/2012 season.
ns , * and ** indicate insignificant, significant at 0.05 and 0.01 levels of probability, respectively.

Table 3 .
Mean performance and environmental index (I) of investigated lentil genotypes in each environment for studied characters from combined analysis.

Table 4 .
Mean performance and stability parameters of studied 24 lentil genotypes across the six environments.