Improving Genetic Attributes of Confectionary Traits in Peanut ( Arachis hypogaea L . ) Using Multivariate Analytical Tools

Peanut is gaining importance more for its confectionery and nutritive values than for its oil content around the world. Improving confectionery qualities is an added advantage for farming community. Hence, in the present study, multivariate analytical tools were used to identify parents with complementary traits for using them in breeding programme. PCA revealed contribution of pod yield, 100-seed weight, oil content, and O/L ratio towards variance. Pod yield was positively associated with 100-seed weight, oil and protein contents. Oil content had weak association with protein content, oleic acid and O/L ratio. UPGMA clustering revealed grouping of cultivars based on origin and its area recommendation. Cultivars superior for yield (GPBD-4, M-28-2 and JL 24) and confectionery traits (S-230 and Dh-8) were identified. Strong positive relation of yield with confectionery traits indicates possibility of breeding high yielding confectionery grade cultivars. Multivariate analytical tools could be used to identify parents for location specific breeding for improvement of Confectionary traits.


Introduction
Peanut (Arachis hypogaea L.) is an important oilseed crop around the world as well as in India.It is ranked as the second most important cultivated grain legume and the fourth largest edible oilseed crop in the world (Shilman et al., 2011).It is grown in more than 100 countries of which China (41 per cent), India (21 per cent), Nigeria (8 per cent), and the United States (7 per cent) are the largest producers (FAO, 2009).The crop is grown primarily for human consumption and it is a rich source of oil (40-50 per cent), proteins (20-50 per cent) and carbohydrates (10-20 per cent) (Belamkar et al., 2011).Wide variation in productivity has been observed among various peanut growing states in India from 529 kg/ha in the state of Bihar to 2286 kg/ha in the state of Pondicherry (DAC, 2010).Though productivity in few states was above the world average (1522 kg/ha (FAO, 2009), but the national average (1063 kg/ha) (DAC, 2010) is very less.This low productivity of India could be attributed to few major peanut growing states like Karnataka and Andhra Pradesh.Hence, there is a need to increase the productivity in these regions at the earliest to improve national average.Though germplasm accessions are capable of raising genetic ceiling on improvement (Kannenberg and Falk, 1995), timescale may be too long (Upadhyaya et al., 2010).In this context, elite breeding lines are considered as best genetic resources because they contain combination of genetic traits that satisfies the marketplace (Troyer, 1990).
Improving the genetic potential of peanut for pod yield and oil content were the major objectives in most peanut improvement programs.Recent estimates revealed that only 49 per cent of its total produce is crushed for oil and the rest is used for sowing, direct consumption and export (SEA, 2010).Improving the confectionery qualities along with yield in low productivity regions may have significant impact on farming community.But, Quality requirement of peanut for confectionery purpose is different from those meant for oil extraction.Characters which are preferred for confectionery grade varieties are high protein and sugar, low oil and aflatoxin contamination, attractive seed size and shape, pink or tan seed colour and ease of blanching and high Oleic/linoleic acid (O/L) ratio (Nigam et al., 1989;Dwivedi and Nigam, 1995).
Hence, with the aim of improving confectionery attributes of the cultivars present study was aimed at understanding pattern of diversity among cultivars, optimum number of clusters involved and to identify suitable donor for further breeding programme using various multivariate analytical tools (UPGMA method of Clustering and PCA analysis).These tools are gaining importance to classify genotypes into different groups and to identify pattern of diversity among genotypes.The PCA reduces multivariate data to few principal axes, produces Eigen vectors for each axis and produces component scores for the characters (Sneath and Sokal, 1973).This study will also help in studying relation between various confectionery attributes which is important for developing confectionery grade cultivars.

Plant material and study area
Experimental materials comprised of 17 cultivars of which 10 were location specific released for the state of Karnataka (Spanish improved, and rest had broad genetic base released for cultivation in different states of India (Table 1).The experiment was laid out in randomised complete block design (RCBD) with two replications at Botanical Garden of University of Agricultural Sciences, Dharwad Karnataka, India.The experimental site is located in the Northern Transitional Tract of Karnataka (15˚13' latitude of 75˚07' E longitudes; altitude, 678m above mean sea level).Each cultivar was grown in a single row of 5 m length with a spacing of 30 cm between rows and 10 cm between plants.In each cultivar five competitive plants were randomly selected to record observations on quantitative traits namely number of primary branches, number of secondary branches, plant height (cm), leaf length (cm), leaf width (cm), pod yield per plant (g) and 100-seed weight (g).
Since peanut is gaining importance as a snack food, improving nutritional qualities along with yield will have added benefits and may improve income of farming community.Hence confectionery qualities like oil content, protein content and fatty acid profile (palmitic acid, stearic acid, saturated fatty acid (SFA), oleic acid, linoleic acid and O/L ratio) were also determined.Oil content was estimated by soxhlet method as given by Jambunathan et al., (1985).Five gram of peanut seeds were made into fine powder in a pestle and mortar and peanut meal was extracted with petroleum ether (60-80°C bp) for 5 hours in the soxhlet apparatus and subsequently ether was evaporated.Powder weight before and after extraction were recorded and difference between the two was expressed in terms of oil percentage.Protein content was determined by using defatted peanut meal as described by Lowry et al., (1951).Fatty acid profile of the 17 cultivars was studied using gas liquid chromatography following the method of AOCS (1999) involving two steps namely, preparation of fatty acid methyl esters (FAME) and gas liquid chromatography (GLC).

RAPD assay
The RAPD reactions were performed in 20 μl volumes in 200 μl PCR tubes (Tarson Pvt Ltd, India).The reaction mixture contained 15 ng of template DNA, 1× amplification buffer, 10mM each of dCTP, dGTP, dATP, and dTTP, 5 pM primers and 1U Taq DNA polymerase (Bangalore Genei, Pvt. Ltd., India).The reactions were performed in a Master Cycler Gradient 5331 (Eppendorf version 2.30).The reaction had an initial denaturation step at 94°C for 5 min, followed by 38 cycles of 94°C for 2 min, 36°C for 1 min, 72°C for 2 min.The final extension step was at 72°C for 10 min.The PCR products were mixed with 2 μl of loading dye (bromophenol blue) and was separated on 1.4% agarose gels (Sigma Aldrich, USA) at 5 V/cm in 1× TBE (89 mM Tris-HCl, 89mM boric acid and 2mM EDTA, pH 8.0) buffer, stained with ethidium bromide (0.75μg/ml) and visualized under UV trans-illuminator (UVI Tech, England) and the results were documented.

Statistical analysis
The data of different quantitative character were standardised by subtracting observations of each character by its mean value and subsequently dividing it by its standard deviation.This standardises the values of each character to mean zero and standard deviation of 1.The standardised data were subjected to principal component analysis (PCA) (Pearson, 1901) and then scatter plot was developed for component 1 and 2. The PCs with eigen value >1.0 were considered as inherently more informative than any single variable alone (Kaiser, 1960).To display the genotype by trait two-way data on a bi-plot following formula was used (Yan and Rajcan, 2002) T ij -Tj = (λ 1 ζ i1 ҭ j1 + λ2ζ i2 ҭ j2 +ҭ ij ) s j where Tij is the average value of genotype i for trait j, Tj is the average value of trait j over all cultivars, sj is the standard deviation of trait j among the genotype averages; ζ i1 and ζ i2 are the PC1 and PC2 scores, respectively, for genotype i; ҭ j1 and ҭ j2 are the PC1 and PC2 scores, respectively for trait j and ij is the residual of the model associated with the genotype i in trait j.A genotype by trait (GT) bi-plot is constructed by plotting the PC1 scores against the PC2 scores for each genotype and trait.A dendrogram was also constructed for different quantitative character based on genetic distance matrix using standardised data by unweighed pair-group method using arithmetic averages (UPGMA) algorithm of Sneath and Sokal (1973) with Euclidean similarity measure.Pair-wise genetic similarity (Sij) between cultivars for RAPD assay was estimated using Dice similarity coefficient.Clustering was done using symmetric matrix of similarity coefficient.A dendrogram based on Sij values was constructed using clustering technique of unweighted pair group arithmetic mean (UPGMA) using SHAN module of NTSYSpc version 2.0 (Rohlf, 1998).

Cluster analysis
Dendrogram obtained for various quantitative characters based on UPGMA (Figure 3) has grouped genotypes into three different clusters.Cluster 1 included six cultivars (JL-24, M-28-2, R-8808, ICGV-86590, Dh-40 and TAG-24) whereas cluster 2 (Spanish improved, Dh 3-30, S 206, KRG 1, K 134 and TMV-2) and cluster 3 (S 230, Dh 43, Dh 8 and GPBD-4) had six and four cultivars respectively.Cultivar R-9251 did not group with any of the clusters.Mean, range and variance values for different qualitative and quantitative traits for all clusters based on UPGMA are presented in Table 2. Cultivars in cluster 3 were superior for oil and protein content, had more number of branches, less saturated fatty acids, high oleic acid and high O/L ratio whereas for cultivars in cluster 1 average pod yield and 100-seed weight was high.

Principal component analysis (PCA)
Results obtained from UPGMA cluster analysis was further corroborated by Principal component analysis (PCA) with correlation matrix.Five principle components accounted for approximately 84% variance.Breakdown of this cumulative variance revealed contribution of 35 % and 19 % for PC 1 and PC 2 respectively.Components 3 and 4 contributed 14% and 9% towards cumulative variance.Discriminating power of principle components as inferred from PCA analysis was high for PC 1(5.9) and low for PC 5 (1.3) (Table 3).First component (PC 1) that explained 35 per cent of variation significantly correlated with (>0.7) oleic acid and O/L ratio in positive direction and with leaf length, leaf width, palmitic acid, linoleic acid and saturated fatty acid in negative direction.Second component (PC 2) significantly correlates to pod yield, other unsaturated fatty acids and oil content.Components 3 and 4 (PC 3 to PC 5) correlated significantly to primary branches (-0.8) and protein content (0.7) respectively.Hence, for developing confectionery grade varieties genotypes with high PC 1 and high PC 2 have to be selected which would intern would contribute complementary traits for developing high yielding confectionery traits.

Molecular diversity
Twenty primers on 17 cultivars generated a total of 145 amplified fragments, out of which 99 were polymorphic (Table 4).The polymorphism for primers ranged from 0 (OPA-15) to 100 per cent (OPC-03) with an overall average of 68.3 per cent.Number of amplified fragments ranged from 1 to 15 for a given primer.On an average, 7.3 bands per primer were amplified out of which 4.9 bands were polymorphic.Banding pattern of seventeen genotypes is presented in Figure 5. High polymorphism observed among the cultivars may be due to use of reported polymorphic primers (Subramanian et al., 2000;Dwivedi et al., 2001;Rajendraprasad, 2003;Vijay, 2003;Nagaraja Reddy et al., 2004;Pattanashetti, 2005).In contrast, Dwivedi et al., 2001 have reported low to moderate level of polymorphism ranging from 8.7 to 33.1 per cent among primers in twenty-six accessions including interspecific derivatives, land races and released cultivars.Based on the Dice Coefficients (Okuno et al., 1998) mean similarity indices for 17 cultivars ranged from 0.67 to 0.97 (mean 0.84) indicating that accessions had 84 per cent of their fragments in common thus indicating low level of diversity among the cultivars studied.The genotypes Dh-8 and ICGV-86590 were most diverse in comparison with other genotypes.The dendrogram, revealed two distinct clusters 'A' and 'B' at Sij of 0.85.Cluster B had R-8808, GPBD-4 and ICGV-86590 (Figure 6).All other varieties were grouped in cluster A. JL-24 and ICGV-86590 was found to span the extremes of the entire dendrogram with the remaining types distributed between them.Cluster B consisted of R-8808, GPBD-4 and ICGV-86590.They had at least one parent, which is not native to India.Cluster A had varieties, which were obtained either from selection of land races.So, the genetic similarity among varieties is high.

GT Bi-plot and trait relations
Genotype by trait (GT) bi-plot can be used to (i) identify traits contributing to diversity, (ii) identify superior cultivars for different traits and (iii) to study interrelationships between traits.GT bi-plot drawn using PC 1 and PC 2 scores explained 53% of total variation (Figure 4).Yan and Rajcan, 2002 had used GT bi-plot to identify superior cultivars after analyzing various yield components in soybean for six years and reported 52-63% of the total variation of the data among 28 soybean cultivars.They attributed this low proportion of variation to complex relationships among different traits similar to the ones observed in the present study.Further to study the relationships between different traits, a vector was drawn from origin to trait.The largest variation was explained by pod yield and other unsaturated fatty acids.
There was a strong positive relationship of pod yield with 100-seed weight, oil and protein contents as indicated by acute angle between vectors (Figure 4).Such positive association was observed when 504 accessions of peanut core collection for Asia (Upadhyaya and Ortiz, 2001) were analyzed for 21 different traits at two different locations by Mallikarjuna swamy et al., (2003).These 4 traits had negative association with SFA as indicated by obtuse angle between them.Interrelationships among the above traits are very important to breed confectionery grade peanut cultivars.Few important relationships exhibited by GT bi-plot in the present study were: (i) negative relation of oleic acid and O/L ratio with linoleic acid and SFA.Such a negative relation between oleic acid with linoleic acid and O/L ratio with linoleic acid have already been reported in peanut (Dwivedi et al., 1993;Anderson et al., 1998;Singkham et al., 2010), (ii) the vectors of Oleic acid and O/L ratio were nearly perpendicular to that of oil content indicating a weak correlation.Thus it is possible to have genotypes with low oil and high O/L ratio The above associations among different confectionery traits and also with yield indicates a possibility of selecting genotypes with complementary traits for breeding high yielding confectionery grade cultivars.
GT bi-plot can also be used to compare cultivars on the basis of multiple traits and to identify potential ones for using them in breeding programme.The Vertex cultivars were GPBD-4, S-230, M-28-2, JL-24, TMV-2, Dh-3-30.These cultivars were the best or the poorest for some or all of the traits, because they had the longest distance from the origin of the biplot (Yan et al., 2007).Cultivar S-230 was superior for oleic acid and O/L ratio whereas GPBD-4, M-28-2 and JL-24 were high yielders with high oil content.

Morphological vs. molecular diversity
UPGMA algorithm has been used to construct dendrogram of various qualitative and quantitative traits (Chowdhury et al., 2002;Ali et al., 2010;Seetharam et al., 2009).Hence in the present study UPGMA was used for grouping cultivars using various quantitative traits and molecular data.Grouping of cultivars into different clusters for quantitative traits followed a specific pattern wherein, for example cultivars included in cluster 2 (Table 5 and 1) were recommended for one or more states other than Karnataka.In cluster 3 most of the cultivars had either blood of Spanish improved or JL-24 in their pedigree (Badigannavar et al., 2002;Murty et al., 2004) whereas cultivars grouped in cluster 1 were mainly selections from germplasm accessions (Table 1).Varieties that were derived from hybridisation were more diverse than varieties, which were derived as result of selection from land races.Among the cross derivatives, varieties which were obtained as a result of use of exotic germplasm were more diverse than those derived from crossing between local genotypes.
Comparison of cultivar groupings by different methods are presented in Table 5.Interestingly, for quantitative traits groupings were almost similar by both PCA and UPGMA clustering with few exceptions.For example, cultivars K-134, KRG-1, Dh 3-30, S-206 and Spanish improved were clustered together in all three methods and were derived as selections (except K-134) from local cultivars.Cultivars ICGV-86590 and R-8808 were clustered together for quantitative traits and molecular data and they are derived through hybridization involving exotic germplasm as one of their parents.Based on quantitative data GPBD-4 was grouped with M-28-2.However, this grouping did not corroborate with grouping based on molecular data.Clustering by UPGMA method with RAPD data grouped three cultivars namely R-8808, GPBD-4 and ICGV-86590 into one group and the rest into other group.These three cultivars were derived through hybridization using germplasm lines or introduced material (Chico) as one of their parents.Cultivars S 230 and Spanish Improved derived as selections from germplasm were in different group as per quantitative traits, but at genetic level they were similar.As observed in the present study also very low diversity among cultivars at genetic level have been reported by using RAPD (Subramanian et al., 2000 ;Dwivedi et al., 2001), AFLP (Herselman, 2003) and SSR (Raina et al., 2001;Krishna et al., 2004) markers.
Multivariate analytical tools used in the present study were able to classify cultivars into different groups and also identified the clusters which could be used in breeding programme.It also identified traits which explained maximum variation like pod yield, saturated fatty acid, Oil content, Oleic acid and O/L ratio.Few desirable associations were also observed between yield, oil content, protein content, 100-seed weight and saturated fatty acids.These associations can be utilized in breeding for location specific varieties with good confectionery qualities.

Figure
Figure 1.Per

Table 1 .
Information on pedigree of location specific and broad based peanut genotypes of India

Table 3 .
PCA components for different quantitative characters in 17 peanut genotypes