Genetic Divergence and Heritability of 42 Coloured Upland Rice Genotypes (Oryzasativa) as Revealed by Microsatellites Marker and Agro-Morphological Traits

Coloured rice genotypes have greater nutritious value and consumer demand for these varieties is now greater than ever. The documentation of these genotypes is important for the improvement of the rice plant. In this study, 42 coloured rice genotypes were selected for determination of their genetic divergence using 25 simple sequence repeat (SSR) primers and 15 agro-morphological traits. Twenty-one out of the 25 SSR primers showed distinct, reproducible polymorphism. A dendrogram constructed using the SSR primers clustered the 42 coloured rice genotypes into 7 groups. Further, principle component analysis showed 75.28% of total variations were explained by the first—three components. All agro-morphological traits showed significant difference at the (p≤0.05) and (p≤0.01) levels. From the dendrogram constructed using the agro-morphological traits, all the genotypes were clustered into four distinct groups. Pearson’s correlation coefficient showed that among the 15 agro-morphological traits, the yield contributing factor had positive correlation with the number of tillers, number of panicles, and panicle length. The heritability of the 15 traits ranged from 17.68 to 99.69%. Yield per plant and harvest index showed the highest value for both heritability and genetic advance. The information on the molecular and agro-morphological traits can be used in rice breeding programmes to improve nutritional value and produce higher yields.


Introduction
Rice (Oryza sativa L.) is the staple food and most important crop in most Asian countries. It belongs to the Poaceae family and is the main source of carbohydrate in these countries. Additionally, rice comprises about 20% of calories consumed worldwide [1]. About 90% of both rice

Experimental design and layout
The selected forty two coloured rice genotypes were germinated in a petri dish after which they were transferred into growing buckets (23× 21 cm) in the glasshouse at the rate of 5 plants/ bucket. The experimental design was randomized complete block design (RCBD) with three replications. Agronomic practice such as weed control was done manually while disease control was through the application of 5g per bucket Furadan (PT Bina Guna Kimia, Indonesia) and 5 ml per 1 L Malathion (Hextar chemicals Sdn. Bhd., Malaysia). The fertilisers urea, muriate of potash (MOP), and triple super phosphate (TSP) were applied 3 times at 5, 25, and 55 days after planting, to provide N, K and P nutrition, at the rate of 160 kg N/ha, 80 kgP 2 O 5 /ha, and 60 kgK 2 O/ha.

Data collection
Fifteen agro-morphological traits were identified by measuring five plants per genotype in each replicate and their means were used for further analysis. These traits include: (i) plant height; (ii) number of tillers per plant; (iii) number of panicles per plant; (iv) percentage of filled grain; (v) 100 grain weight; (vi) harvest index; (vii) days to first flowering; (viii) days to maturity; (ix) grain dimension and shape; (iix) length of flag leaf; (xi) panicle length; (xii) kernel length; (xiii) length breadth ratio;and (xiv) chlorophyll SPAD reading at 40 days and (xv) 60 days.

DNA extraction protocol
DNA was extracted from the seed samples using the modified conventional method [17]. About 100 mg of each seed sample was ground in the mortar using a pestle, and 400 μL extraction buffer (200 mMTris-HCL, 200 mMNacl, 25mM EDTA, 0.5% SDS). Then, the solution was transferred into a 2 ml microcentrifuge tube and 400 μl of CTAB solution (2% CTAB, 100 mMTris-HCL, 20 mM EDTA, 1.4 M NaCl, 1% PVP) was added. Next, 400 μl mixture, ratio of chloroform: isoamyl alcohol: phenol (24:1:5%), was added in the same tube. The mixtures were then well mixed by vortex and centrifuge (14,000 rpm, 5 minutes) at room temperature. The supernatant was then transferred into a new 2 mL microcentrifuge tube and 2/3 volume of isopropanol added. The mixtures were gently mixed by inverting the microcentrifuge tube and then incubated at room temperature for 10 minutes. Following this the mixture was centrifuged again (14,000 rpm, 5 minutes) at room temperature. The supernatant was then discarded and the pellets rinsed with 70% alcohol for a few minutes. Subsequently, pellets were air dried and re-suspended in 50 μL of TE buffer. The quality and quantity of the DNA was measured by NanoDrop ND-1000 spectrophotometer (Thermo Fisher Scientific, USA).

Data analysis
The presence and absence of alleles were scored using the binary system '1' and '0' respectively. The observed number of alleles, effective number of alleles and Shannon's Information index were determined using the Popgen software. Polymorphism information content (PIC) and expected heterozygosity (He) were also calculated using the PIC calculator (http://www.liv.ac. uk/~kempsj/pic.html).
Genetic similarity was calculated using the dice coefficient [18]; the sequential agglomerative hierarchal and nested (SAHN) clustering was performed based on genetic similarity and unweighted paired group method with arithmetic averages (UPGMA); the principal component analysis (PCA) was performed on the matrix of the genetic similarity coefficients; and data analysis was carried out using the NTSYS version 2.1.
The agro-morphological traits collected were subjected to Analysis of Variance (ANOVA) using SAS version 9.2. The Pearson correlation coefficient among all agro-morphological traits calculated using SAS version 9.2. A dendrogram was then constructed based on UPGMA and PCA based on agro-morphological data using NTSYS version 2.1. The broad sense heritability, genetic advance and other variance estimates were calculated using the method of Allard [19]. The following formulas were used to calculate the genetic parameters: Genotypic variance (σ² g ) = (MS 2 -MS 3 )/b Error variance (σ² e ) = MS 3 Phenotypic variance (σ² p ) = σ² g + σ² e Genotypic coefficient of variation ðGCVÞ ¼ ffiffiffiffiffiffiffi Phenotypic coefficient of variation ðPCVÞ ¼ ffiffiffiffiffiffiffi s 2 p p = X Â 100 MS 2 = mean square of populations MS 3 = mean square of error b = number of blocks X = mean of the trait Heritability (h 2 B ) = σ² g/ σ² p Expected genetic advance (GA) and genetic gain (GG) (as percentage of the mean) were calculated using the method of Allard [15] where selection intensity (K) was assumed to be 5% Expected genetic advance ffiffiffiffiffiffiffiffi ffi K is a constant which represents the selection intensity, when K is 5% the value is 2.06, ffiffiffiffiffiffiffi s 2 p p represents the phenotypic standard deviation, h B 2 is the heritability while X represents the mean of the characteristic being evaluated, using the formulae suggested by Burton [20].  (Table 3).

Cluster analysis from SSR markers
The dendrogram based on the UPGMA method grouped the 42 selected coloured upland genotypes into seven groups (Fig 2) at the coefficient of 0.62. Dice coefficient ranged from 0.50 to 1.00. Group 1 comprised the landrace cultivars from Thailand, India and Vietnam. Group 2 comprised all breeding and inbred line genotypes from Philippines and one from Myanmar whereas Group 5 comprised just one genotype from the Ivory Coast.

Principal component analysis from SSR markers
The first three components from the PCA analysis explained about 75.28% of the total variation present in these genotypes. About 5 distinct groups were obtained from the three dimensional PCA (Fig 3). Group 1 comprised landraces and traditional cultivars from India, Thailand and Vietnam. The clustering patterns found in these genotypes group via PCA was similar to that found in the dendrogram where all the breeding and inbred lines clustered to form Groups II and group IV, respectively.

Genetic divergence and morphometric variability in the genotypes
All the genotypes showed higher significant differences at p 0.05 and p 0.01 based on the quantitative traits, as revealed by ANOVA (Table 4). Genotype (V22) IR 5533-15-1-1 showed the highest number of tillers and panicles. Plant height ranged from 82.14 to 170.20 cm for all the genotypes with 124.72 cm being the average (Table 5). Genotype V5(C) recorded the highest panicle length at 30.35 cm. while genotype V33 (chirikata 2) showed the highest percentage of filled grain. Additionally, wide variability was observed in grain yield per plant among all the genotypes ranging from 0.57 V13 (Bibili al) to 6.62 g V5 (C) and showed the lowest and highest grain yield per plant. Genotype V7 (Chokoto 14) showed the highest rice kernel length. Length breadth ratio for all genotypes ranged from 2.77 to 4.85, while genotype V4 (Bi-e-gaw) recorded the earliest days for 50% flowering and days to maturity, which was 49 and 114 days, respectively.

Clustering analysis from agro-morphological traits data
The similarity coefficient as shown in the dendrogram varied from 0.15 to 1.44 (Fig 4). All the genotypes were grouped into 4 groups (Fig 4) at 0.79 coefficient. Group1 comprises of red rice genotypes from several geographical origins. This group had the highest mean value of agromorphological traits, such as plant height (148.78 cm), length of flag leaf (53.18 cm), percentage of filled grain (66.45%), panicle length (25.46%), kernel length (6.05 mm), and harvest index (0.77). Three genotypes were found in group 4, and this cluster had the lowest mean for traits such as number of tillers (2.04), number of panicles (1.55), percentage filled grain (31.51%), harvest index (0.09), grain yield/plant (0.65g), kernel length (5.59 mm), and kernel length/breadth ratio (3.19).

Principal component analysis from agro-morphological traits data
Using the PCA all the rice genotypes were grouped into 4 distinct groups which were similar to the cluster analysis grouping (Fig 5). All agro-morphological traits showed the highest cumulative percentage (!70%) except for length breath ratio (69%), chlorophyll SPAD value at 40 days (58%) and 60 days (56%) ( Table 6).

Correlation among the agro-morphological traits data
Yield per plant had positive correlation with number of tillers, number of panicles, percentage filled and unfilled grain, harvest index, panicle length, and chlorophyll SPAD reading at 60 days (Table 7). Chlorophyll SPAD reading at 40 days showed significant negative assosication with plant height, flag leaf and panicle lenght.

Genetic parameters
The genetic parameters calculated, such as genotypic variance (σ 2 g), phenotypic variance (σ 2 p), heritability (h 2 B ), and genetic advance (GA), are presented in Table 8. The broad sense heritability for 15 agro-morphological traits ranged from 17.68% (chlorophyll SPAD value at 40 days) to 99.69% (kernel length), respectively. Yield per plant showed the highest value of genetic advance (126%) among all the traits. Broad sense heritability was high for virtually all the yield component traits with the lowest (percentage filled grain) being approximately 62%.

Discussion
Genetic divergence as revealed by the molecular markers and agro-morphological traits is important for breeding and improvement of existing rice genotypes to suit consumer demands. Genetic divergence helps in breeding resistance and tolerance to various biotic and environmental stresses and also as a tool to investigate the effects of climate change, in order to select the genotypes with higher potential for use in breeding programmes. Among the 25 SSR primers used in this experiment, only 21 show polymorphism among genotypes. The marker used must be informative in order to reveal the genetic divergence among the genotypes. The mean polymorphic information content (PIC) recorded was 0.41 for all the SSR markers tested. The marker is informative if the PIC value is higher than 0.5 [21]. For instance, Ravi et al. [22] reported the mean PIC value of 0.578 for genetic diversity analysis of rice cultivars using SSR data. Cluster analysis of SSR primers data grouped all the genotypes based on geographical origin and status of the genotypes. Additionally, only one genotype was placed in Group 5. This genotype was an advanced cultivar from the Ivory Coast. The cophenatic correlation value (r) calculated from the dendrogram was 0.83 which shows the good fit of the data for diversity analysis if (r>0.8). Based on the grouping in the PCA, some groups could not be separated using clustering analysis. Groups 3 and 5 (Fig 2), for example, consisted of mixed groupings of genotypes and this cannot be resolved by PCA. Thus, this analysis can be informative for differentiation among the major groups and it will help the breeder to select from diverse breeding lines.
All agro-morphological traits differed significantly. Three genotypes showed the lowest values for yield per plant and percentage of filled grain. The breeding line showed good agro-morphological traits, such as lower plant height, high filled grain, and higher grain yield per plant compared with the landrace or traditional cultivars. Thus, cross breeding between traditional cultivars and breeding lines can be done to produce plants with good yield characteristics.
The dendrogram for agro-morphological traits was constructed based on the matrix of average of taxonomic distance using the UPGMA method. The dendrogram showed that Group 1 consisted of all the landrace and traditional cultivars. These genotypes displayed similar agromorphological characteristics, such as plant height, panicle length and length of flag leaf. Group 3 showed higher mean value of grain yield per plant (3.9g) as it was from the breeding lines from Philippines.
From the dendrogram constructed based on SSR marker and agro-morphological traits, there are differences in grouping of the genotypes. Agro-morphological traits were influenced    Genetic Divergence and Heritability of Coloured Rice Genotypes by environmental factors such as light intensity, disease and humidity. Additionally, clustering based on SSR markers was more accurate. This is because their usage is not influenced by environmental factor thus it will reflect the actual level of genetic difference existing among the genotypes [23]. In addition, SSR markers can detect slight differences of DNA structure. From PCA analysis, the first four components explained about 83.76% of variation [24]. Since the variation is high (! 25%), this analysis can be used along with cluster analysis to show the relatedness among the genotypes [25]. From the results of the first three components, it was seen that days to first flowering, number of panicles, number of tillers and percentage of filled grain played an important role in explaining the variation. This is because positive eigenvalue is shown for each of the 3 components.
Plant height showed a negative correlation with number of tillers and number of panicles. Thus, lower plant height is a good characteristic because it results to higher tiller and panicle numbers. From the result (Table 5), traditional varieties showed the highest plant height when compared with breeding line. In modern rice breeding, gene sd-1 is one of the most important genes controlling dwarfisim in rice plant. It ressesive character helps to improve lodging resistance as a result of shortened culm [26]. Lodging is the most common problem affecting upland rice in the field. Thus reducing the plant height is one of the means of resolving the lodging problem. Most of the modern rice varieties have short plant height and this is the focus of selection by breeders for improvement of rice plant. Thus, plant height is one of the most important traits which needs to be considered in production of high yield variety. In addition, yield per plant was positively correlated with number of tillers, number of panicles, percentage   of filled grain, harvest index and panicle length. This information is valuable to know the agromorphological traits that contribute to the to yield of the plant. From the present study, higher phenotypic variance was recorded than genotypic variance for all traits evaluated. The difference in value of genotypic variance and phenotypic variance was due to environmental factor. Our data shows that environmental factors have pronounced effect on each of those traits compared to the genotypic factors. Broad sense heritability was found to be high for plant height, panicle length, 100 grain weight, harvest index, yield per plant, days to flowering, days to maturity, kernel length and length breadth ratio. The heritability results must be combined with expected genetic advance to achieve reliable results [27]. Yield per plant and harvest index showed the highest value for both heritability and genetic advance. These traits need to be considered for yield improvement of rice plant genotypes through breeding programmes. High heritability and genetic advance for grain yield per plant and harvest index has also been reported by Bisne et al. [28].
From the analysis it is shown that all the 42 coloured upland rice genotypes can be selected for breeding programmes based on the SSR primers and agro-morphological traits evaluation. Genotypes such as C, Chirikata 2, Ble Chu Cau, and IR 9669-PP 836-1 have high potentials for selection. This is because, these genotypes have good agro-morphological traits such as, high percentage of filled grain, high harvest index and high grain yield per plant. Furthermore, the selected genotypes are clustered in different groups based on the SSR marker analysis which indicates that they are genetically divergent and thus breeding from these genotypes can produce good progenies with superior traits.

Conclusions
Cluster analysis from the SSR markers grouped all the genotypes into 7 groups according to geographical origin and status of the cultivar. All agro-morphological data showed significant differences at p 0.05 and p 0.01 for all traits which shows the presence of diversification among the 42 coloured upland rice genotypes. Four groups were constructed using agro-morphological data clustering analysis. Yield contribution factors traits, which are yield per plant and harvest index, showed the highest value of heritability and genetic advance. Selection based on these traits can be done for further breeding programmes. Potential genotypes such as C, Chirikata 2, Ble Chu Cau, and IR 9669-PP 836-1, are recommended for selection and further evaluation in future breeding programmes. This is based on their molecular and agromorphological information. These genotypes, selected from groups 2, 3, 4 and 5 from the SSR primer grouping, also have high grain per yield and harvest index values. For further studies, the evaluation of nutritional value of coloured upland rice can be done since red and purple bran may contain many phytochemical and neutraceutical functional foods. Thus, it will give high impact on rice breeding programme for development high yield and functional rice.