Main Article Content
Traditional rice cultivars in Sri Lanka have diverse characteristics that prove the excellent potential for utilizing them for varietal improvement. Seventy-one Sri Lankan traditional rice cultivars obtained from Plant Genetic Resources Center, Gannoruwa, Sri Lanka were used for diversity analysis. Data were collected in the middle-row-plants of each replicate and altogether 80 plants were evaluated in four replicates of each cultivar. Data were collected on 11 parameters in two consecutive Maha seasons in 2012-2014. According to the standard evaluation system of international rice research institute, 67% rice cultivars were tall, and 66% were low tillering. There was no any rice cultivar with high and good reproductive tillers/plant. Around 55% rice cultivars were partially sterile and 43% were fertile. The shortest days to maturity was recorded in Gonabaru (62 days) and the longest days to maturity was recorded in Vellainellu (105 days).
The variation of agro-morphological characteristics was broad in studied rice cultivars: The recorded highest yield was 22.74 g/plant while the lowest was 1.55 g/plant. The hundred-grain-weight ranged from 3.49-1.27 g and the plant height ranged from 95-192 cm. The recorded highest number of total tillers per plant was 10 and the lowest was 3. Panicle length and panicle weight were ranged from 13-32 cm and 0.36-4.69 g consecutively. The smallest grains were in Herath and the largest grains were in Mudaliwi. The heaviest above ground biomass was achieved by Kallurundoivellai and its harvest index was the least (0.09). The highest filled grain percentage (91.48%) was recorded by Galpa Wee.
The correlation analysis revealed that the yield per plant was significantly correlated with panicle weight, the total number of spikelets/panicle and number of fertile spikelets/panicle. There was no correlation between panicle length and the final yield of rice though panicle length was correlated with plant height and panicle weight. Total number of spikelets/panicle, the number of fertile spikelets/panicle, and plant height were significantly correlated with panicle weight. None of the parameters were correlated with hundred grain weight. Total biomass was correlated with plant height, the total number of spikelets, and fertile spikelets per panicle. There was no significant correlation between the biomass of rice with the final yield, however, a number of fertile tillers were correlated with a number of total tillers.
Eleven parameters were sorted into four principal components that explained 87.44% of total variance and cluster analysis categorized the rice cultivars into six clusters at rescaled cluster distance 10 in Ward's linkage analysis.