Character association and principal component analysis in interspecific hybrids of Saccharum robustum

Abstract

The multiple usage of sugarcane crop necessitates a breeding program that generates gene pool which enables the identification of genotypes for multiple purposes. Once the magnitude and pattern of existing genetic base is determined for traits of interest, the breeding program has to follow an effective selection procedure that utilizes both direct and indirect methods to improve the quantitative traits. Several statistical procedures measures the magnitude and direction of interrelationship between any two casually related variables and models like principal component analysis (PCA) and cluster analysis can be further used for diversity study and for selection of genotypes. Earlier reports have explained the relationship of fibre yield with cane yield and cane quality traits, millable canes and single cane weight with highest direct effect on fibre yield. The present study reports the association of cane yield and quality characters with fibre % in improved S. robustum interspecific hybrids, partitioning of the total variance of original traits into a small number of uncorrelated new variables for visual differentiation among genotypes and cluster analysis to identify diverse combinations for further improvement in clonal population. The study also resulted in identification of high fibre types with S.robustum base that combined yield and quality.The experimental material comprises 36 elite interspecific hybrids: 27 F1 hybrids (improved S. robustum x improved S. officinarum: 12, 9 from improved S. officinarum x improved S. robustum, 6 from improved S. robustum x commercial) and nine back cross hybrids with NMC > 31 and cane yield per clump > 14 kg, were evaluated in RBD with two replications in a plot size of 6m x 0.9m during the crop season 2014 at ICAR-Sugarcane Breeding institute, Coimbatore. Observations on NMC, cane height (cm), cane thickness (cm), SCW (kg), cane yield per clump (kg), brix %, sucrose %, purity %, CCS % and fibre content were recorded at 12 months. Fibre content was calculated (Thangavelu and Rao, 1982) as given below Fibre % = A-B/C x 100, A = Dry weight of bag + bagasse after drying (g), B = Dry weight of bag alone (g), C= Fresh weight of cane (g). 21 genotypes were identified as very high fibre types (>25 %), 9 were high fibre types (20-25%) and 6 were low fibre types (<20%). Correlation coefficient and path analysis was done by standard statistical procedures (Singh and Chaudhary, 1985). PCA and cluster analysis (using euclidean distance) was performed with PAST Statistical software, version 2.17 (Hammer et al., 2001). Fibre % showed high significant positive genotypic association with cane yield/clump (0.716), NMC (0.537), cane height (0.515) and SCW (0.439). The association of fibre % with cane thickness was non-significant and negative (-0.011) at phenotypic level while it was positive and significant (0.172) at the genotypic level (Table 1). Correlation of fibre % with brix % was positive and significant; sucrose % showed negative but non-significant association with fibre %. This indicates the possibility to combine both high fibre and high sucrose content in same genotype since there was no significant negative correlation between fibre % and brix % and sucrose %. Hence selection could be made based on both the traits to raise genotypes with increased sugar and fibre %, thus satisfying the future demands through development of multipurpose cultivars. Genotypic path analysis was done to estimate the direct and indirect effects of different traits on fibre %. Cane yield/clump showed a relatively high positive direct effect on fibre % (0.442). It also exhibited high indirect positive effect through NMC (0.195) and SCW (0.103). These positive indirect effects resulted in higher correlation coefficient of cane yield/clump with fibre content (0.716). NMC recorded the highest positive direct effect on fibre % (0.624). But its negative indirect effects through brix %, sucrose %, cane thickness and juice weight results in reduced total correlation of NMC with fibre % (0.537). Cane height recorded a low positive direct effect (0.284) on fibre yield but the total correlation was significant and of higher magnitude (0.515) due to the positive indirect effects through cane yield/clump (0.177), NMC (0.082) and SCW (0.062). SCW exerted lowest direct effect on fibre % (0.198); but its positive indirect effect through cane height (0.246) resulted in high positive correlation with fibre yield (0.439). Residual effect was 0.59, suggesting the characters studied showed significant influence on fibre %. PCA was performed to explain the combination of variable which accounts for maximum genetic variation in S. robustum interspecific hybrids. The first three principal components (PC) with eigen value >1 contained the majority of variance information with the cumulative variation of 76.40 %, remaining components with eigen value < 1 contributed to 23.60 % of variability. Cane yield, NMC, SCW and cane height were characterized with large factor loadings in PC1 (Table 2). Quality parameters such as Brix %, Sucrose % and purity % were characterized with large scores in PC2. The genotypes that appeared in quadrant (1) and (2) were characterized with high fibre %, cane yield and moderate brix %, while those in quadrant (3) and (4) were characterized with high brix % and fibre % < 20. The eigen value of cane yield, NMC, SCW, cane height, fibre % and brix % in PC1 were positive. The corresponding eigen vectors of PC2 also showed that quality traits (brix % and sucrose %) and cane yield characters (NMC, SCW, cane yield/clump) and fibre % were grouped together. The present results suggests that the traits viz. NMC, SCW and cane height are the primary traits to be considered in clonal selection to improve fibre content followed by brix and sucrose % and thereby to develop genotypes with multipurpose utility. Cluster analysis (euclidean distance) grouped 36 S. robustum interspecific hybrids into 10 distinct clusters (Figure 1), where two clusters contained a single genotype. Cluster I was distinct from cluster X (13.065), VI (13.061), IV (11.070) and II (12.680). The smallest distance was observed between cluster II and III (5.241). The clones, RO 15073 (RxO 1), OR 15066 (OxR 2), RO 15012 (RxO 5), RO 15019 (RxO 10), OR 15081 (OxR 9), BC 15084 (BC 4), OR 15043 (OxR 3), RO 15027 (RxO 12) and BC 15055 (BC 8) were identified as energy canes with more than 20 % fibre and 15 % juice brix; RO 15033 (RxO 2), RO 15029 (RxO 8) and RO 15041 (RxO 3) as energy canes with more than 25 % fibre are recommended for further genetic manipulation and utilization. The study also envisages the evaluation of these high biomass genotypes under larger plot size for their use as energy canes in cogeneration. Crossing between genotypes of clusters with large distance and significantly different for yield and quality traits following the information about indirect effects of traits on fibre % and sucrose will increase efficiency of breeding programs in selection of clones for multiple purposes.