Genetic Diversity Studies in Drought Tolerant Rice (Oryza sativa L.) Genotypes

Rice (Oryza sativa L.) is the staple food crop for more than half of the global populations. Irrigated low land is the most important ecosystem for rice in Asia which requires substantially more water (i.e. about 5000 liters of water/kg of rice grain). Declining water availability and increasing water scarcity has threatened the sustainability of this system all over the world including India. Aerobic system of rice cultivation is gaining momentum in recent times.  The lack of varieties with stable yield under aerobic system is a major limitation in achieving the maximum yield potential under water limited conditions (Sandhu et al., 2019).To combat this situation, it is essential to focus the research priority towards the development of water efficient ‘aerobic rice’ varieties.

Estimation of genetic divergence is very much essential for the selection of parental lines for hybridization programme. Identifying the diverse parental combinations for creating the segregating progenies with maximum genetic variability for further selection and introgressing the desirable genes from the diverse germplasm into the available genetic base is the prerequisite for the any breeding programme.

Considering the importance of developing aerobic rice varieties, the present investigation was undertaken i) to assess the genetic diversity for identifying genetically diverse parental lines and ii) to find out the nature of the relationship among the yield and yield contributing traits by Association analysis.

The present experiment was conducted with 65 rice genotypes including popular rice varieties of Tamil Nadu, drought tolerant rice varieties, aerobic rice genotypes and land races at Rice Research Station, Tirur, Tiruvallur during kharif, 2015. The genotypes were raised in randomized block design with three replications in 3 m row length (15 rows) with the spacing of 20 x 10 cm  under aerobic condition and irrigated at weekly intervals. Observations were recorded on five randomly selected plants per replication for the characters viz., plant height, no. of productive tillers/plant, panicle length, no. of grains/panicle, Hundred seed weight and panicle harvest index whereas, the characters like plot yield and days to 50% flowering were recorded on plot basis. Then the data were subjected to statistical analysis by following Singh and Chaudhary (1995) for correlation coefficient and Dewey and Lu (1959) for path analysis. Selection based on the detailed Principal Component Analysis was worked out through correlation matrix using PAST 3. Dissimilarity between the genotypes was calculated based on Euclidean distance and the genetic divergence was computed using Wards D minimum method of clustering through R 3.5.1 statistical package.

Principal Component Analysis (PCA)
 
Principal component analysis reflects the importance of larger contributor to the total variation at each axis of differentiation (Sharma, 1998). Eigen values are used to determine the number of major principal components to be retained (Hailegiorgis et al., 2010).

In the present study, it revealed that first five principal components in the PCA contributed to a maximum of 78.67% of the total phenotypic diversity among the 65 rice genotypes (Table 1).The first three principal components have eigen values more than one which showed cumulative variation of 56.86% of the total variation (Fig 1). Girma et al., (2018), Shalini Singh et al., (2021), Mohan et al., (2021) and Yugandhar et al., (2018) reported 64%, 86.7%, 79.12% and 71.4% of total variation explained by first three, four, five and six principal components respectively in various studies.  Shivangi et al., 2021 reported that first eight principal components exhibited eigen value more than 1.0 and 81.84 per cent variability among the 28 traits studied in 67 rice germplasm lines. The predominant characters coming together in a particular principal component by contributing towards variability offer opportunity for its utilization in crop breeding. To assess the contribution of characters which influenced greater the PCA value, it is customary to look for individual loadings. Scree plot graph (Fig 1) obtained by plotting eigen values and principal component numbers explained the percentage of variation associated with each principal components. PC1 showed 22.91% variability with Eigen value 1.83, which then declined gradually.





The contribution of each component to total variation is presented in Table 1. The first principal component (PC1) with an eigen value of 1.83 explained 22.91% of the total variation. PC1 was associated mainly with panicle length (0.561), no. of grains per panicle (0.507), plant height (0.341), plot yield (0.327) and days to 50% flowering (0.321). In PC1, major contribution was from the yield attributing traits viz., panicle length and no. of grains per panicle. The second principal component (PC2) accounted for 19.53% of the total variation and was mainly related to productive tillers/plant (0.593), single plant yield (0.504) and panicle harvest index (0.378). PC3 accounted for 14.42% of the total variation and was characterized by panicle harvest index (0.574) and days to 50% flowering (0.450). The traits panicle length, no. of grains per panicle, plant height, days to 50% flowering and no of productive tillers per plant from the first two principal accounted for major contribution to the total variability.
 
Cluster analysis
 
Cluster analysis was used to determine the genetic relationship among the genotypes and find out the suitable genotypes for future breeding programme. Cluster analysis based on eight quantitative traits for 65 genotypes grouped them into six discrete and well defined clusters. Cluster III was the largest consisting of 19 genotypes (Table 2 and Fig 2) followed by the clusters I with 17 genotypes and the cluster II with 16 genotypes whereas, the smallest was cluster VI with three genotypes.





Presence of variability in the 65 genotypes was also reflected in the cluster means for the eight traits evaluated (Table 3). Cluster I comprised of seventeen genotypes characterized by genotypes with maximum plant height (117.7 cm), longer panicle length (26.46 cm) with more number of grains per panicle (118.38). Cluster II comprised of sixteen genotypes with semi dwarf plant types (100.7 cm) and higher panicle harvest Index (94%). Cluster III comprised of 19 genotypes predominantly early maturing, having short panicles (22.75 cm) with lesser number of grains per panicle (104). Cluster IV comprised of four genotypes which were characterized by late in attaining fifty per cent flowering (97 days), less number of productive tillers per plant (7.50) and with lesser hundred seed weight (2.67g). Cluster V comprised of six genotypes which were predominantly characterized by tall genotypes (116.87 cm) having high hundred seed weight (3.79g) with least plot yield. The cluster VI having only three genotypes viz., CB-06-551, CB-13-14, CB-14-112 which are early in attaining fifty per cent flowering with maximum number of productive tillers per plant which leads to higher single plant yield. Clear difference in grain yield was observed in the six clusters. Genotypes found in cluster VI recorded highest grain yield per plot (2.49 g) where as the cluster V comprised of six genotypes registered minimum single plant yield (0.45 g). Crossing between promising genotypes of cluster VI with the genotypes in V may provide superior desirable segregates for developing high yielding genotypes.


 
Association analysis
 
As genotypic correlation coefficient would give the realistic picture of relationships existing among the characters in the absence of environmental influence, the genotypic correlation coefficient was utilized in the present study. Significant and positive associations with plot yield were exhibited by the yield contributing traits viz., no. of productive tillers plant (0.455), panicle length (0.243) and hundred seed weight (0.219) (Table 4).  Similar results were reported earlier by Mohanty et al., (2012), Manohara and Singh (2015), Rukmani Devi et al., (2017) and Sahu et al., (2019) for panicle length, Sathish Chandra et al., 2009 and Kishore et al., 2015 for productive tillers/plant. Significant and positive association of these traits indicates that selection based on these traits would ultimately improve grain yield under aerobic conditions. The inter correlation among the  component traits is also important in order to decide upon which traits to be given due weightage in exercising selection. The inter correlation between the yield contributing traits viz., days to fifty per cent flowering with panicle length (0.272) and no. of grains per panicle (0.272); plant height with panicle length (0.219);  no. of  productive tillers/plant with panicle harvest index (0.401); panicle length with no. of grains per panicle (0.454) and hundred seed weight with panicle harvest index (0.279) were positive and significant indicating the possibility of simultaneous improvement of these traits by concentrating on any one or two of these traits.



Information obtained from correlation study could not indicate whether the association of the yield-related traits with yield is due to their direct effect on yield or is a consequence of their indirect effect via some other traits.  Path coefficient analysis is useful to understand the direct and indirect effects of each character on grain yield and the application of selection pressure in a better way for yield improvement, partitioning of correlation coefficient into direct and indirect effects. In the present study, path analysis showed that the maximum positive direct effect contributing to plot yield was exhibited by no. of productive tillers / plant (0.455) followed by panicle length (0.243) and hundred Seed weight (0.219) which implies that selection for these traits would improve the plot yield (Table 5). Similar results have been reported by Nanda et al., (2019) for the traits number of productive tillers/plant and hundred seed weight and Bhadru et al., 2012 for panicle length. Panicle harvest Index exhibited highest indirect effect on yield via no. of productive tillers per plant (0.198) followed by no. of grains per panicle via panicle length (0.124). This shows that selection for the traits no. of productive tillers per plant and panicle length would indirectly help in increasing the grain yield. A residual effect of 0.815 implies that the casual traits explained only about 18.5% of the variability, leaving 81.5% of the variability unexplained.



It can be concluded that the traits panicle length, no. of grains per panicle, plant height, days to 50% flowering and no of productive tillers per plant from the first two principal accounted for major contribution to the total variability.  The predominant characters coming together in a particular principal component by contributing towards variability offer opportunity for its utilization in the breeding programme.  The traits no. of productive tillers/plant, panicle length and hundred seed weight had positive association with higher direct effect on plot yield which could be used as selection criteria for developing new rice varieties with high productivity under aerobic condition.