Stability of yield in relation to components traits in rice under rainfed low land condition

: Twenty rainfed low land rice genotypes were evaluated for their stability to grain yield and its component traits under three predictable environments created by changing the method of sowing and date of transplanting viz. direct sowing, normal transplanting and delayed transplanting. The pooled analysis of variance indicated that the mean differences due to genotypes and genotype x environment interaction were significant indicating that the genotypes differ in their adaptability and stability. The pooled deviations were also significant for all the characters studied, suggested that these genotypes differed in their deviation from linearity. Most of the characters were influenced by non-linear components of G x E interaction, but magnitude of linear components were greater than non-linear types. On the basis of linear components characters, the genotypes RAU 1401-12-2, Satyam, Kishori expressed average response and relative stability under different environment for yield and yield components. Stability for grain yield and its components were also exhibited by the genotypes TTB 517-16-SBIR 70149-33 and Satyam under favourable environment whereas, genotypes RAU 1305-6-3-2-3 had higher stability under poor environmental condition. The genotypes RAU 1401-12-2 and RAU 1314-3-3-3 were found to be stable under average environment for grain yield per plant. These genotypes are suitable for the rainfed low land condition which will also serve as good parents for the development of high yielding stable lines.


Introduction
The ultimate aim of any Plant breeding programme is to evolve the cultivars with high yield potential, with consistent performance over diverse environments.Productivity of a population is the function of its adaptability while, the later is a compromised of fitness (stability) and flexibility.Stability of a genotype depends on the ability to retain certain morphological and physiological characters steadily and allowing others to vary, resulting in predictable G x E interaction for yield.A population that can adjust its genotypic and phenotypic state in response to environmental fluctuation in such a way to give high and stable yield is termed "well buffered".In the present study an attempt has been made to isolate genotypes with high grain yield adaptability and stability under rainfed low land condition.

Materials and Methods
The experimental material comprised of 20 promising rice genotypes.The elite lines/ rice genotypes were grown in three predictable environments created through method of sowing and date of transplanting viz.direct sowing (E I ), normal transplanting (E II ) and delayed transplanting (E III ).The field trial was conducted in randomized block design with three replications in Kharif 2002 at experimental research farm, RAU, Pusa.The plot size was 5.0 x 1.5m 2 with a row spacing of 20cm both under direct seeding and transplanting.In the latter case 25 day-old seedlings were transplanted 15cm apart with one seedling/hill.Recommended package of practices were followed to raise the crop.
Observations on five random competitive plants from each plot were recorded for plant height,  and Russel (1966) model.

Results and Discussion
Pooled analysis of variance for stability parameters (Table 1) showed that the mean differences between the genotypes were highly, significant for all the seven characters studied except number of tillers per hill.The environments and interaction (G x E), environments (linear) were also significant in respect of all the seven characters with the exception to the environmental effects for panicle length.Highly significant G x E interaction component in all the characters indicated differential reaction of the genotypes to the environment.The pooled deviation (non-linear) in the present study were significant for all the characters and suggested that the genotypes differed in their deviation from linearity.The genotype x environment (linear) components was highly significant for the characters, days to 50 per cent flowering, fertile spiklets per panicle, 1000grain weight and grain yield per plant when tested against pooled deviation.The G x E (non-linear) components were also significant for all the characters studied, when tested against pooled error.High magnitude of environmental (linear) effect was observed for all the characters in comparison to G x E (linear) effect which may be responsible for high adaptation in relation to yield attributing traits in rice.The pooled analysis of variance revealed that both linear and non-linear components of G x E interaction were significant for days to 50 per cent flowering, number of fertile spiklets per panicle and grain yield per plant indicated the predominance of G x E interaction for those traits was of linear type.These findings * and ** : Significant at 5% and 1%, respectively.+ and ++ : Significant against deviation from unity at 5% and 1%, respectively.
are in corroboration with the result of Vivekanandan and Subramanian (1994), Singh et al. (1995), Singh and Payasi (1999), Kulkarni et al. (2000) and Senapati et al. (2002).Finlay and Wilkinson (1963) reported linear regression as a quantitative measure of phenotypic stability to denote varietal adaptability over a range of environments.But, Eberhart and Russel (1966) suggested that both linear (bi) and non-linear components (S 2 di) of the genotypes environment interaction should be considered while, evaluating the phenotypic stability of a genotype.According to Langer et al. (1979), the regression coefficient is a measure of response to varying environments and the mean square for deviation from linear regression is a true measure of productive stability.A genotype with high mean, unit regression coefficient and least deviation from regression was considered as the ideal and widely adapted stable genotype.All three parameters of stability the mean (X), regression coefficient (bi) and deviation from regression (S 2 di) for all the characters are presented in Table 2.According to Eberhart and Russel (1966), an ideal genotype must have high performance unit regression on coefficient (b=l) and the deviation not significantly different from zero (S 2 di).They further pointed out that the varieties exhibiting high regression coefficient (bi > 1) could be considered as below average in stability.Such varieties will perform better only in favourable environment while, their performance will be poor in unfavourable environments.The varieties with low regression coefficient (bi < 1) are adapted specifically to poor environment.
In the present investigation, normal June sowing and July transplanting with recommended doses of NPK fertilizers were considered as the most favourable environment for maximum yield, whereas, the June sowing and August transplanting (delayed transplanting) and direct seeding in April were considered as less favourable environments.Adaptability of twenty rice Thus, the results of the present study revealed that the genotypes RAU 1401-12-2 and RAU 1314-3-3-3 had relatively higher stability under varying environments for yield and yield components followed by Kishori and Satyam.Stability for grain yield and its components were also shown.
The genotypes TTB 517-16-SBIR 70149-33 and Satyam were found stable for grain yield and its components under favourable environments, the genotype RAU 1305-6-3-2-3 under poor environmental conditions.The exploitation of these genotypes in breeding programmes will help in improving the productivity of semidry areas.

Table 1 .
Pooled analysis of variance for seven characters in rice under rainfed low land condition Source

Table 2 .
Stability parameters for seven component traits of 20 genotypes in rice under rainfed low land condition

Table 3 .
Adaptability of rice genotypes under diverse environments