The genotype and genotype by environment biplot model is an excellent tool for visual multienvironment trials data analysis. In this study we investigated grain yield of six rice genotypes (three tested, one released hybrids and two inbred check varieties) in five environments. The combined analysis of variance for grain yield data indicated that the differences among all sources of variation were highly significant (P<0.001). Environment (E), Genotype (G) and G × E interaction effects accounted for 12.49, 76.51 and 10.21% of the total sum of squares respectively. The first two principal components (PC1 and PC2) were used to display a two-dimensional GGE biplot. Thus, genotypic PC1 scores>0 classified the high yielding genotypes while PC1 scores<0 identified low yielding genotypes. Unlike genotypic PC1, genotypic PC2 scores discriminated the unstable ones. The GGE biplot analysis was useful in identifying stable genotypes with high yield performance. In this study, the polygon view of GGE biplot showed that the vertex genotypes were BRRI1A/BR168R (G1), BRRI10A/BRRI10R (G2) and BRRI dhan28 (G5) having the largest distance from the origin, which was most discriminated genotypes with the unstable ones. These vertex genotypes BRRI1A/BR168R (G1) and BRRI10A/BRRI10R (G2) gave higher yield (PC1 scores>0) while another vertex genotype BRRI dhan28 (G5) produced low yield (PC1 score<0). Hence, the vertex genotype BRRI10A/BRRI10R (G2) was high yielding for all environments and it fell into section 1 following IR58025A/BRRI10R (G3) and BRRI hybrid dhan1 (G4). Mean yield and stability performance over environments of each genotype is explored by using the average environment (tester) coordinate (AEC) methods. These methods show that the genotypes BRRI10A/BRRI10R (G2), IR58025A/BRRI10R (G3) and BRRI hybrid dhan1 (G4) had higher stability as well as higher mean yield while the genotype IR58025A/BRRI10R (G3) had the highest stability out of these three genotypes. The ideal genotype biplot suggests that the closer to 'ideal' genotype was IR58025A/BRRI10R (G3) followed by G2 and G4 being more desirable than the other genotypes. Similarly, the environment Barisal (E3) was 'ideal' environment followed by E1 (Gazipur), E2 (Comilla) and E5 (Satkhira). Hence, the environment Barisal (E3) is more stable and suitable for all genotypes following Satkhira (E5) because it has large PC1 and small PC2 score but Rangpur (E4) is a discriminating environment because it has large PC2 score. The interrelationship among the environments according to the small angles of test environments was highly positively correlated. Gazipur (E1), Comilla (E2), Barisal (E3) and Satkhira (E5) were closely correlated with small angles but Rangpur (E4) had medium long angles. Comparison between two genotypes showed that BRRI10A/BRRI10R (G2) and IR58025A/BRRI10R (G3) were high yielder in test environments. Thus, the difference between G2 and G3 was relatively small in test environments.

Twenty-six restorer line were selected for evaluation for genetic variability, heritability and genetic association among physicochemical, nutritional, yield and yield contributing traits in randomized complete block design (RCBD) with three replications. The analysis showed that restorer line has milling rice percentage, head rice recovery percentage, cooking time, elongation ratio, amylase content and protein content with average value among 26 genotypes (64.60%, 73,00%, 16.67 min, 1.38, 19.26% and 6.36% respectively). Highest grain yield was obtained from the restorer line 3038R (15.99 g/plant), followed by BRRI20R (14.83 g/plant), PR828R (14.10 g/plant), PR312R (13.91 g/plant), BRRI31R (13.61 g/plant) and BRRI19R (13.28 g/plant). The line BRRI20R and BRRI31R had high (25.00%) amylase and protein content, milling rice percentage, head rice recovery percentage. F2277R had less (14.00 min) cooking time. Grain yield exhibited a strong positive correlation with head rice recovery percentage, amylase content and thousand grain weight. Insignificant genotypic correlation was observed among physicochemical, nutritional, yield related traits and grain yield. Most of the studied traits indicating that selection with these traits might be possible without compromising any yield loss. Based on mean, range, genetic parameters, correlation coefficient and path coefficient values, direct selection through five traits i.e., head rice recovery percentage, cooking time, amylase content, protein content and thousand grain weight would significantly improve the grain yield of restorer line. Some of these restorer lines can be used for development of hybrid rice regarding these physicochemical, nutritional quality and yield related traits.

Significant variations were obtained among the component lines for most of the floral traits. BRRI9A exhibited superiority for angle of open florets, stigma length and breadth, anther breadth, filament length, style length and panicle exertion rate while its corresponding maintainer lines BRRI9B showed superiority for duration of open florets, stigma length and breadth, anther breadth and filament length. This indicated BRRI9A and its corresponding maintainer line BRRI9B possessed very good floral traits that influence out crossing. Among the restorer lines BR168R showed superiority for duration of open florets, angle of open florets, duration of bloomed florets/panicle, stigma length, anther length and number of pollen/microscopic focus. This suggested BR168R could be used as promising restorer line with BRRI9A for hybrid seed production in local condition.DOI: http://dx.doi.org/10.3329/bjb.v43i1.19739 Bangladesh J. Bot. 43(1): 1-8, 2014 (June)

Combining ability on grain yield and its components from line × tester analysis of 70 rice hybrids produced by crossing seven CMS lines and ten testers of local and exotic origin were studied. The analysis revealed higher sca variance than gca variance for all the traits indicating the prevalence of non-additive gene action. Testers played important role towards panicle length (48.18%), spikelet fertility (60.46%) and grain yield/plant (55.44%) indicating predominant of paternal influence for these traits. The contribution of interactions (line × tester) were found vital for plant height (58.55%), effective tillers/plant (51.69%), days to 50% flowering (48.95%), days to maturity (48.32%), flag leaf length (68.55%), panicles/m 2 (48.61%) and panicle weight (50.39%). D.ShanA was the only good general combiner among the female parent for earliness and dwarfness. Male parent IR64R was best general combiner for dwarfness. The crosses BRRI9A × BR168R and D.ShanA × BR168R were identified as most promising for yield and desired traits based on sca effects, per se performance and gca effects of parents for grain yield and its components in rice which could be exploited beneficially in future rice breeding program by adopting heterosis breeding strategy.

The desirable characteristics of fourteen genetically diverse CMS lines were assessed for their phenotypic acceptability, plant height (cm), days to 50% flowering. panicles per plant, stability for pollen sterility, panicle exertion rate and out crossing rate. The CMS lines viz. BRRI1A. IR 58025A. BRRI10A. BRRI9A. BRRI3A, IR 75608A, and 1132A have been found to be usable female parents for hybrid rice breeding due to their appreciable phenotypic acceptability, stable pollen sterility, panicle exertion rate and good out crossing rate. Panicle exertion rate of CMS lines had moderate heritability (38.34%) coupled with low genetic advance (2.34%) expected in the subsequent generations, that proved to be vulnerable to environmental fluctuation. The genotypic and phenotypic coefficient of variation was not remarkably varied from each other for most of the characters studied, indicated negligible environmental influence on these traits. Overall genetic correlation coefficients were higher than corresponding phenotypic correlation coefficients indicated strong relation present among the studied characters.