GENOTYPE X ENVIRONMENT INTERACTION AND STABILITY ANALYSIS OF LOWLAND RICE GENOTYPES

Twenty-one lowland rice genotypes were evaluated for their stability parameters with respect to grain yield in a multi-locational trial at five different sites of Eastern India viz. Central Rice Research Institute, Cuttack (Orissa); OUAT, Bhubaneswar (Orissa); CRS, Masodha (UP); RAU, Pusa (Bihar) and RARS, North Lakhimpur (Assam). Pooled analysis of variance reflects existence of genotype x environment interactions and contribution of both linear and non-linear components to genotype x environment interactions. Through stability parameter analysis, it was found that Rayda B3, CR 778-95 and CR 661236 were suitable for all environments. The genotypes Sabita and OR 1358-RGA-4 were suitable for better environments. PSR 1209-2-3-2, CR 780-1937, Ambika, OR 877-ST-4-2, NDR 40055-2-1 and CR 662-2211 were identified for poor environments.


Introduction
Rice (Oryza sativa L.) is the most important cereal crop of India.As this crop is grown under a varied range of agro-climatic conditions ranging from upland to lowland and irrigated to rainfed situations, their phenotypic responses vary greatly in accordance with the environment.Rainfed lowland environments are mostly unfavourable and characterized by variable water regimes, occurrence of submergence and water logging.Nearly 38 million hectares in the world are rainfed lowland out of which 35 million ha are in South-east Asia.Water depth in rainfed fields is mostly variable depending on rainfall distribution, pattern and soil topography.The rice growing environments were analyzed and have been reported on decline in rice production that was due to physical environmental constraints (Roy and Panwar, 1994).The major efforts in crop technology, under unfavourable environment should be yield stabilizing, cost reducing, risk minimizing and returns enhancing (Nanda and Tomar, 1981).The genotypes should therefore be high stability cultivars besides high yielding cultivars.Many methods (Finlay and Wilkinson, 1963;Eberhart and Russell, 1966;Perkins and Jinks, 1968;Freeman and Perkins, 1971) are available for assessing the stability of performance of crop varieties.These models are helpful in identification of adaptable genotypes over a wide range of environments; achieving stabilization in crop production over locations; developing phenotypically stable high potential cultivars; effective selection for yield stability and prediction of varietal responses under changing environments.The present investigation was aimed at identifying high yielding and high stability for rainfed unfavourable lowlands of eastern India.

Results and Discussion
The analysis of variance for grain yield revealed significant differences among the genotypes and environments (Table 1).Highly significant mean squares due to genotype × environment (G×E) interaction revealed that the genotypes interacted considerably with environmental conditions.Both linear and non-linear components of G×E interaction were found to be significant for grain yield as indicated by highly significant mean squares due to G×E (linear) interaction and pooled deviation.Environmental index (Ij) directly reflects the poor or rich environment in terms of negative or positive values of Ij respectively.Hence, CRRI (Cuttack), OUAT (Bhubaneswar) and CRS (Masodha) represent poor environment (stress conditions), RARS (North Lakhimpur) and RAU (Pusa) represent better environment (Table 2).The yield of different genotypes under study varied from 0.39 t/ ha (NDR 40001-1-2) to 6.75 t/ha (Hanseswari) at CRRI, Cuttack; from 0.70 t/ha (Panikekoa) to 6.00 t/ha (OR 1358-RGA-4) at OUAT, Bhubaneswar; from 1.00 t/ha (LPR 106, RAU 1306-2-2 and Sabita) to 5.80 t/ha (CR 662-2211) at CRS, Masodha; from 1.20 t/ha (CR 682-162) to 6.50 t/ha (OR 1358-RGA-4) at RARS, North Lakhimpur and from 1.35 t/ha (Borjohingia) to 5.50 t/ha (CR 780-1937) at Rau, Pusa (Table 2).Eleven genotypes viz.Rayda B 3 , Sabita, Purnendu, Hanseswari, Ambika, CR 778-95, CR 662-2211, CR 661-236, CR 780-1937, OR 1358-RGA-4 and OR 877 ST-4-2 recorded better yield than the average, other two genotypes (PSR 1209-2-3-2 and OR 1334-16 recorded average yield (Table 2).The general mean over locations revealed that CR 661-236 was the best with the highest grain yield (5.29 t/ ha) followed by CR 780-1937 (5.08  *, ** significant against pooled error at 1% and 5 %. The grain yield of rice fluctuates considerably with the change in environmental conditions.Hence, a variety possessing reasonable stability for yield is desirable for minimizing the risk of yield loss in harsh environments of unfavourable low land situation.Partitioning of genotype x environment interaction revealed highly significant difference in case of Var.x Env.(Lin) and non-linear (pooled deviation) interaction.These highly significant differences are very important for determining G x E interaction.Relatively higher value of the linear component as compared to non-linear one suggested the possibility of prediction of performance for seed yield over the environments.Therefore, linear (bi) and nonlinear (S 2 di) component of G x E interactions were considered while judging the phenotypic stability of a genotype (Finlay and Wilkinson, 1963;Eberhart and Russell, 1966).They further suggested that an ideal variety should have high mean with linear regression co-efficient equal to unity and S 2 di as small as possible.They have emphasized the use of deviation from regression as a measure of stability, whereas the linear regression could be treated as a measure of varietal response to environments (Beese, 1969;Paroda and Hayes, 1971;Jatsara and Paroda, 1980;Ravindra et al., 2012).Accordingly, the mean and the deviation from regression of each genotype were considered for stability and linear regression was used for testing the varietal response.
In the present study, the genotypes Rayada B 3 , CR 778-95 were found to be suitable for a general adaptation, i.e. suitable for all environmental conditions as their bi (linear response) was around 1.0 with least deviation from linearity and above average mean (Table 3).Though the genotype NDR 40001-1-2 had bi value around 1.0, it was rejected because of its low mean performance.The highest yielding genotype CR 661-236 was found to be slightly less responsive as its bi value (0.626) was less than 1.0 (Table 3).The second highest yielding genotype CR 780-1937 was found to be least responsive because of its negative bi value (-0.338) (Table 3).The other varieties with negative bi values are Purnendu and Hanseswari.All these genotypes are suitable for poor environments.The genotypes like Sabita and OR 1358-RGA-4 were found to be suitable for better environments as their bi values are significantly higher than 1.0.The analysis of variance revealed significant difference among genotypes and environments (Table 1).
A significant G×E interaction indicated a considerable interaction of genotypes with environmental conditions that existed at different locations of eastern India.These results were in conformity with the earlier reports on rice (Maurya and Singh, 1977;Ganesh and Soundarapandian, 1988;Kulkarni et al., 1988;Reddy and Chaudhury, 1991;Roy and Panwar, 1994;Sreedhar et al., 2011).The linear contribution of environmental effects on the performance of genotypes was reflected by highly significant mean square due to environment (linear).The significant mean square due to genotype x environment (linear) interaction indicated that a considerable proportion of genotype x environment interaction was contributed by the linear component (Table 1).Therefore, prediction for most of the genotypes appeared to be feasible for yield.Highly significant mean squares due to pooled deviation for yield revealed the importance of a non-linear component accounting for the total genotype x environment interaction.Therefore, the genotypes differed considerably with respect to their stability for yield.Similar results were obtained in rice by Kulkarni et al. (1988).Hence, it is obvious that both linear and non-linear components contribute to the G x E interaction for grain yield indicating the importance of both regression co-efficient (bi) and deviation from regression (S 2 di) in determining the stability of grain yield.These results suggested that wide differences existed among the genotypes over the range of environments and it may be possible to classify these genotypes for their adaptation behaviour.In the present study, only three genotypes Rayda B 3 , CR 778-95 and CR 661-236 fulfilled the conditions for an ideal variety with high mean, linear response and least deviation from linear regression.Hence, these genotypes were identified as suitable for general adaptation i.e. suitable for growing over all the environments under study.The entries, PSR 1209-2-3-2, CR 780-1937, Ambika, Purnendu and Hanseswari are identified for poor environment as they exhibited high mean and negative bi (Table 3).Therefore, it means that genotypes reflect negligible response to the environmental changes i.e. remain steady under poor conditions but cannot exploit the positive improvement in the environment.Among these PSR 1209-2-3-2 and CR 780-1937 are steady although with low Sd 2 value, whereas Ambika deviates moderately with a moderate Sd 2 value.Purnendu and Hanseswari fluctuate considerably with high Sd 2 .Hence only PSR 1209-2-3-2, CR 780-1937 and to a certain extent Ambika can be recommended for the poor environments studied.OR 877-ST-4-2 and CR 662-2211also recorded high yield, low response to environment and moderately fluctuated from linearity (Table 3).Hence, these two varieties may also be recommended for poor environment.Sabita, OR 1334-16 and OR 1358-RGA-4 exhibited high mean, high bi and high Sd 2 (Table 3).These varieties are highly sensitive to environment responding 2-3 times for a unit change in the environmental milieu.Under intensive agriculture, when inputs are not limitations, such varieties can yield maximum, whereas in poor conditions they fail miserably.Hence, these varieties can be recommended for rich environments studied.Therefore, the identified varieties are recommended for their suitability under fragile environmental conditions (Table 4).

Table 2 .
Genotype mean for grain yield (t/ ha) over five different locations.Central Rice Research Institute, OUAT: Orissa University of Agriculture and Technology, NDUAT: Narendra Deva University of Agriculture and Technology, RARS: Regional Agricultural Research Station, RAU: Rajendra Agricultural University.

Table 3 .
Mean performance and stability parameters for grain yield.

Table 4 .
Genotypes recommended for various environments.