Morphological variability of Indonesian rice germplasm and the associated SNP markers

*Corresponding author: Puji Lestari, Indonesian Center for Agricultural Biotechnology and Genetic Resources Research and Development-IAARD, Jl. Tentara Pelajar, No. 3A Bogor 16111, Indonesia. E-mail: plestari129@yahoo.com Received: 27 March 2016; Revised: 18 June 2016; Accepted: 24 June 2016; Published Online: 15 July 2016 Lestari, et al.: Morphological variability of Indonesian rice germp Emir. J. Food Agric ● Vol 28 ● Issue 9 ● 2016 661 characterization at morphological and molecular level will enable rice breeders to exploit genotypic diversity for conservation and further crop improvement (Thomson et al., 2007; Juansa et al., 2012; Rabara et al., 2015). Morphological variations are attributable to classify rice that contains two major subspecies, japonica and indica. The typical grains and environment between the two are distinctive. The other subspecies is broad-grained and rives under tropical conditions and possesses specific morphology, called as tropical japonica. Based on the morphology and molecular analyses, scientists also try to sort rice into different number of cultivar groups namely temperate japonica, aromatic, aus etc. (Garris et al., 2004; Glaszmann, 2004). Morphologically, rice differs in growth stages in respect to vegetative characters, leaf, culm, panicle, and seed properties which are important for guiding selection process in breeding. In addition to phenotypic characters, molecular characterization proves rapid scientific advances and have introduced new techniques for the conservation and protection as well as their broad applications. Potential molecular markers as marker-assisted selection (MAS) gain prominent in rice breeding and offer new opportunities that increase the efficiency of food production (Franco and Hidalgo, 2003). Markers are also currently widely used to provide information of genetic relatedness and linkage/association with important traits and other function. Modern breeding programs will require high-throughput genotyping that can handle large numbers of samples at a low cost (Thomson, 2014). A new generation of markers based on single nucleotide polymorphisms (SNPs) is now rapidly overtaking previous markers. SNPs are potential to greatly increase the speed and reduce the cost of molecular marker genotyping with high throughput in order to make it feasible to ‘mainstream’ MAS into conventional breeding program (Thomson et al., 2010). A number of custom SNP sets from low to high density have been developed using different platforms with automated genotyping (Thomson et al., 2010; Utami et al., 2013; Chen et al., 2011; Singh et al., 2015). These efforts will allow more efficient strategies in genotyping for use in a wide range of genetics and breeding applications. Therefore, this study was carried out to assess the morphological diversity of rice genotypes in Indonesia according to rice descriptor from Bioversity International (2007) and to identify SNP markers significantly associated with the morphological traits using GoldenGate 384-plex SNP genotyping assay. MATERIALS AND METHODS Plant materials Sixty rice genotypes belonged to the gene bank collection of Indonesian Center for Agricultural Biotechnology and Genetic Resources Research and Development (ICABIOGRAD) under the Indonesian Agency for Agricultural Research and Development (IAARD) were selected for genotyping of custom 384 SNP chip in this study. The genotypes were chosen based on their diverse genetic background depending on subspcies. Based on the botanical features, the 60 genotypes comprised of 46 indica, 5 japonica, and 9 tropical japonica. The indica group included improved varieties, upland, lowland, hybrid, and IR group developed by International Rice Research Institute (IRRI). The list of total rice genotypes used in this study is presented in Table 1. Morphological characterization All genotypes were grown in green house and two individuals of each genotype were planted in pot in three replications. Rice cultivation was managed according to the standard recommendation of rice practices. Nineteen quantitative and qualitative morphological traits were observed from vegetative to generative growth stages to the descriptors for rice (Bioversity International, IRRI and Warda, 2007) and the data were recorded. The following quantitative traits were observed such as leaf length (LLNG), leaf width (LWDT), diameter of basal internode (DBITR), panicle length (PLNG), leaf angle (LAGL), flag leaf angle (FLAGL), secondary branching of panicle (SBP), culm angle (CAGL), and ligule length (LGLNG). While the remaining 10 qualitative traits included leaf blade pubescence (LBPBS), collar color (CCLR), auricle color (ACLR), leaf blade color (LBCLR), leaf sheath color (LSCLR), ligule shape (LSHP), ligule color (LGCLR), culm color (CUCLR), attitude of panicle branch (APB), and panicle axis (PAXS). 384-plex SNP genotyping with GoldenGate assay DNA Isolation The healthy leaves of three week-old rice seedling germinated from a single plant of each genotype was harvested, collected and stored in -80°C until use for DNA isolation. Genomic DNA was isolated using the Thermo Scientific King Fisher Plant DNA kits (Qiagen, 2011). The DNA was eluted with 100 μl of TE buffer. The concentration and quality of DNA were measured with NanoDrop1000 (Thermo Scientific Co.) on wavelength of 260/280 and 230/280 (Sambrook and Russel, 2001). Each DNA was concentrated with Speedvac and adjusted to final concentration of 50 ng/μl to meet the required recommendation for GoldenGate assay using iScan array (Illumina, San Diego, CA). GoldenGate assay for 384 SNP chip using iScan array The GoldenGate assay used the customized OPA 384 SNP chip distributed across 12 chromosomes of rice that Lestari, et al.: Morphological variability of Indonesian rice germp 662 Emir. J. Food Agric ● Vol 28 ● Issue 9 ● 2016 was design in the previous study (Utami et al., 2013). The genotyping was conducted in accordance to the manufacture protocol for plant genome as described by Illumina’s BeadArray technology-based GoldenGate (GG) assay (Fan et al., 2003) at the facility of Indonesian Center for Agricultural Biotechnology and Genetic Resources Research and Development (ICABIOGRAD) under the IAARD. This assay was done on a high-resolution confocal scanner (iScan, Illumina) which was able to read the arrays and generated intensity data. Allele calling was performed automatically using Genome Studio v. 2011.1 (Illumina, San Diego, CA). The SNP genotyping scores (Call rates, GeneTrain and GenCall) were exported using GenomeStudio internal tool for subsequent analysis. Genotypes with a GenCall score below 0.25 were assigned as missing. Data analysis Numerical values for all morphological traits were subjected to analysis using XLSTAT program, particularly statistical summary and correlation matrix (Pearson estimate). Principal component analysis (PCA) was generated on ranged data with linear reduction. The PCA was used to view the traits with significant variation in the rice germplasm. Visualizing of PCA cluster was generated using XLSTAT. To fit the PCA cluster visualization, rice genotypes were named following the default instruction in the program instead of the real name of each rice variety/ accession. Consequently, both the rice varieties name and the corresponding new label depicted in the PCA cluster were described (Table 1). Flapjack also facilitated to generate a phylogenetic tree based on genotypic data of 384 SNPs and the 19 morphological traits in total genotypes (Milne et al., 2010) on the basis of similarity matrices. Flapjack program allowed to analyzing both genotypic data containing allele per marker for each genotype and phenotypic information of either numerical or categorical. Association between SNPs and morphological traits were tested with linear mixed model (SAS, 2002).


INTRODUCTION
Rice (Oryza sativa L.) is widely grown in the world to diet billions people living in the subtropics and tropics.The crop will continue to remain a main source of daily food by growing population in many regions (Sasaki, 2002) and contributes toward food security not only in the world but also in Indonesia.In Indonesia, even though national rice production increases significantly in comparison with that of one decade ago (Tobias et al., 2012), consumption of rice has continued to rise per annum.The high demand of rice leads to put higher efforts since this crop has high nutritional value, providing 20% of calories, 15% of proteins, minerals and fibers, and a part as animal feed.To increase rice production in Indonesia by the use of a wide range of technologies, an optimal utilize of rice germplasm collection should be addressed to support breeding program.
Rice germplasm in Indonesia including improved varieties (such as upland, lowland and swamp cultivation), hybrid, local accessions, and other breeding status is an important source of useful genes and can be used for crossing parents for desired traits (Zeng et al., 2003).The germplasm is a national asset that needs to be preserved for breeding materials.Development of rice varieties with important traits such as productivity, resistance/tolerance to abiotic and biotic stresses needs broad genetic diversity.Consequently, successful breeding program should consider genetic diversity of rice to improve variety with desired traits of economic importance.High yielding rice varieties along with other superior traits such as resistance to pest and disease, early maturity, grain quality and tolerance to drought, salinity and lodging have been infused for surviving in harsh environments in Indonesia (Silitonga, 2010).A well characterized germplasm with genetic different could be useful for their effective utilization.Genetic characterization at morphological and molecular level will enable rice breeders to exploit genotypic diversity for conservation and further crop improvement (Thomson et al., 2007;Juansa et al., 2012;Rabara et al., 2015).
Morphological variations are attributable to classify rice that contains two major subspecies, japonica and indica.The typical grains and environment between the two are distinctive.The other subspecies is broad-grained and rives under tropical conditions and possesses specific morphology, called as tropical japonica.Based on the morphology and molecular analyses, scientists also try to sort rice into different number of cultivar groups namely temperate japonica, aromatic, aus etc. (Garris et al., 2004;Glaszmann, 2004).Morphologically, rice differs in growth stages in respect to vegetative characters, leaf, culm, panicle, and seed properties which are important for guiding selection process in breeding.
In addition to phenotypic characters, molecular characterization proves rapid scientific advances and have introduced new techniques for the conservation and protection as well as their broad applications.Potential molecular markers as marker-assisted selection (MAS) gain prominent in rice breeding and offer new opportunities that increase the efficiency of food production (Franco and Hidalgo, 2003).Markers are also currently widely used to provide information of genetic relatedness and linkage/association with important traits and other function.Modern breeding programs will require high-throughput genotyping that can handle large numbers of samples at a low cost (Thomson, 2014).
A new generation of markers based on single nucleotide polymorphisms (SNPs) is now rapidly overtaking previous markers.SNPs are potential to greatly increase the speed and reduce the cost of molecular marker genotyping with high throughput in order to make it feasible to 'mainstream' MAS into conventional breeding program (Thomson et al., 2010).A number of custom SNP sets from low to high density have been developed using different platforms with automated genotyping (Thomson et al., 2010;Utami et al., 2013;Chen et al., 2011;Singh et al., 2015).These efforts will allow more efficient strategies in genotyping for use in a wide range of genetics and breeding applications.Therefore, this study was carried out to assess the morphological diversity of rice genotypes in Indonesia according to rice descriptor from Bioversity International (2007) and to identify SNP markers significantly associated with the morphological traits using GoldenGate 384-plex SNP genotyping assay.

Plant materials
Sixty rice genotypes belonged to the gene bank collection of Indonesian Center for Agricultural Biotechnology and Genetic Resources Research and Development (ICABIOGRAD) under the Indonesian Agency for Agricultural Research and Development (IAARD) were selected for genotyping of custom 384 SNP chip in this study.The genotypes were chosen based on their diverse genetic background depending on subspcies.Based on the botanical features, the 60 genotypes comprised of 46 indica, 5 japonica, and 9 tropical japonica.The indica group included improved varieties, upland, lowland, hybrid, and IR group developed by International Rice Research Institute (IRRI).The list of total rice genotypes used in this study is presented in Table 1.

384-plex SNP genotyping with GoldenGate assay DNA Isolation
The healthy leaves of three week-old rice seedling germinated from a single plant of each genotype was harvested, collected and stored in -80°C until use for DNA isolation.Genomic DNA was isolated using the Thermo Scientific King Fisher Plant DNA kits (Qiagen, 2011).The DNA was eluted with 100 µl of TE buffer.The concentration and quality of DNA were measured with NanoDrop1000 (Thermo Scientific Co.) on wavelength of 260/280 and 230/280 (Sambrook and Russel, 2001).Each DNA was concentrated with Speedvac and adjusted to final concentration of 50 ng/µl to meet the required recommendation for GoldenGate assay using iScan array (Illumina, San Diego, CA).

GoldenGate assay for 384 SNP chip using iScan array
The GoldenGate assay used the customized OPA 384 SNP chip distributed across 12 chromosomes of rice that was design in the previous study (Utami et al., 2013).The genotyping was conducted in accordance to the manufacture protocol for plant genome as described by Illumina's BeadArray technology-based GoldenGate (GG) assay (Fan et al., 2003) at the facility of Indonesian Center for Agricultural Biotechnology and Genetic Resources Research and Development (ICABIOGRAD) under the IAARD.This assay was done on a high-resolution confocal scanner (iScan, Illumina) which was able to read the arrays and generated intensity data.Allele calling was performed automatically using Genome Studio v. 2011.1 (Illumina, San Diego, CA).The SNP genotyping scores (Call rates, GeneTrain and GenCall) were exported using GenomeStudio internal tool for subsequent analysis.Genotypes with a GenCall score below 0.25 were assigned as missing.

Data analysis
Numerical values for all morphological traits were subjected to analysis using XLSTAT program, particularly statistical summary and correlation matrix (Pearson estimate).Principal component analysis (PCA) was generated on ranged data with linear reduction.The PCA was used to view the traits with significant variation in the rice germplasm.Visualizing of PCA cluster was generated using XLSTAT.To fit the PCA cluster visualization, rice genotypes were named following the default instruction in the program instead of the real name of each rice variety/ accession.Consequently, both the rice varieties name and the corresponding new label depicted in the PCA cluster were described (Table 1).Flapjack also facilitated to generate a phylogenetic tree based on genotypic data of 384 SNPs and the 19 morphological traits in total genotypes (Milne et al., 2010) on the basis of similarity matrices.Flapjack program allowed to analyzing both genotypic data containing allele per marker for each genotype and phenotypic information of either numerical or categorical.Association between SNPs and morphological traits were tested with linear mixed model (SAS, 2002).

Morphological diversity of rice germplasm
Morphological assessment displayed by 60 rice genotypes in this study revealed variations on qualitative and quantitative characters.An example of morphological characters used Based on the first 2 principal components (PCs) with eigenvalues greater than 2 and 1.9, principal component analysis (PCA) for the 19 morphological traits explained 24.60% of total phenotypic variation in the collection.
The result from PCA analysis with the first two PCs estimates for the 19 morphological traits to classify the 60 rice genotypes is presented in Table 3. Particularly, PC1 with eigenvalue greater than 2.0 contributed the most of variation on seven traits in the germplasm.The seven traits comprised four qualitative traits (leaf blade color, leaf sheath color, ligule color, and collar color) and three quantitative traits (flag leaf angle, ligule length and diameter of basal internode).Of these characters, negative characters were from flag leaf angle and ligule length.There was a contrast between the diameter of basal internode with a positive loaded value in PC1 but a negative loaded value in PC2.Thus, morphological traits related to leaf could be major descriptor contributing in differentiation of rice germplasm collection (Toshinori, 1991;Shikari et al., 2009).
Relationship between 60 accessions/varieties of rice revealed by cluster analysis is presented in Fig. 2. The total genotypes occurred in three main clusters (G1, G2, and G3), in particular, three indica (Isn23-136, Hipa13 and Inpari12) were to be separately from other groups.While the G1 grouped varieties/accessions belonging to indica/ japonica/tropical japonica which accommodated overlap characters.The G2 seemed to be distinct group containing indica varieties/accessions (Inpari 30, BMIP19/IPBM-32-1-3-3 and IR84047-24-3-3-3) and japonica reference cultivar, Nipponbare.These varieties in this group were characterized    varieties developed by IRRI tended to be grouped in the same clade (such as IR30, 64, 42, 65 and 74), similarly to upland rice (Inpago 4, 5 and 8), and two red rice accessions ("beras merah D1" and "beras merah D3").Tropical japonica genotypes also preferentially grouped in the same sub-clade, such as Ketan bubarak, Repong and Bintang, meanwhile Pandanwangi dan Adan Kelabit were on other sub-clade.While, Inpari varieties seemed to be grouped in different sub-clades according to its individual genetic without considering the lowland varieties type.

Association between SNP markers and morphological characters
The 384 SNP panel desained across 12 chromosomes of rice (Zhao et al., 2010;Utami et al., 2013) was used for association analysis with morphological traits.Total 154 SNPs with diverse base substitutions were significantly associated with 15 morphological traits at P-value threshold of 0.01.Chromosome 7 contributed greatly with the highest SNP number (21 SNPs) associated with morphological traits and those in chromosome 2 (16 SNPs) and 3 (17 SNPs) were comparable.48 out of total 154 SNPs were identified to be highly significant associated with 10 morphological traits including leaf blade pubescence, flag leaf angle, leaf length, auricle color, culm angle, culm color, diameter of basal internode, secondary of branching panicle, panicle length and attitude of panicle branch.The number of SNPs significantly associated with morphological characters of rice germplasm is presented in Table 4.
Some SNP markers were found to have common association for several traits.TBGI367899 had associations with leaf blade pubescence, leaf angle and panicle length, and TBGI112858 was highly associated with leaf blade pubescence, auricle color and panicle length.In contrast, specific SNPs were only associated with one morphological traits, such as id10005370 with leaf blade pubescence, and TBGI000824 with culm angle.Among such markers, SNPs had the strongest association with corresponding traits.Eleven of total SNPs had strong association with leaf blade pubescence, of which TBGI367899 with A/G base substitution in chromosome 9 was the strongest (P-value=5.91E-09,r 2 =0.49).Two markers of TBGI129684 and TBGI272460 were strongly associated with culm color and explained high phenotypic variation with the same r 2 (0.49).Nine of total 33 significant SNP markers were strongly associated with secondary of branching panicle, and SNP id1009867 located in chromosome 1 had the strongest association (P-value=6.56E-05,r 2 =0.29).In addition, a number of markers associated with morphological traits were detected to be specific to a variety/accession in respect to leaf, culm of panicle, indicating that there are morphological characters of rice plants supported by molecular characterization that could be used as genetic differentiator.The list of significant SNP markers with their detailed information is shown in Table 5.

DISCUSSION
Diverse rice germplasm needs a rational use with better knowledge of its characteristic.The presented morphological characters of rice genotypes in Indonesia in our study demonstrated diversity within and among rice sub-species and their genetic relatedness.Similar certain morphology was found among genotypes in the same group such as IR-group, glutinous rice (called as "Ketan"), red rice, and improved varieties etc. Genotypes chosen from different subspecies, ecological area (lowland and upland), glutinous rice and rice were more inherently diverse than they were from a similar environment (Thomson et al., 2009).Quantitative morphological of total 60 genotypes proved their diversity even they possessed similar qualitative characters (Table 2).This rice germplasm which encompassed a wide range of genetic background and ecologically distinct regions probably contained greater genetic diversity as a result of divergent selection pressure.The fact of some improved varieties from indica with given name such as Inpago, Inpari and IR-groups derived from selection by human via breeding or adaptability evaluation represented more genetically identical than subspecies of japonica and tropical japonica.Thus, this study demonstrated the common and specific morphological differences observed among rice collection which is in good agreement with previous study based on morphological traits of wild rice species (Suhartini, 2010), local accessions and improved varieties (Saadah et al., 2013).
Correlation matrix performed among morphological characters could easily assume the relationship between characters.In our study, the most prominent correlations were found on flag leaf angle with leaf angle, leaf sheat pubescence with panicle length, and leaf sheat color with culm color.Such correlation between traits allows for simultaneous selection and utilization of related traits interchangeably in rice.Morphological characterization could be highly considered in routine rice germplasm management for breeding purpose.
PCA is able to measure the importance and contribution of each trait to total variance.In our study, the first 2 component in PCA with eigenvalues of arbitrary set above 0.2 explained the most variations including leaf characters in this rice germplasm.This morphological traits related to leaf could be major descriptor contributing in differentiation of rice germplasm collection (Toshinori, 1991;Shikari et al., 2009).In addition, the contrast between the characters with a positive loaded value in PC1 but a negative loaded value in PC2 suggest that These predominant traits observed in our study like leaf, panicle and culm characters were also found to be important in other Asian countries (Shikari et al., 2009;Sinha and Mishra, 2013;Wijayawardhana et al., 2015).
Based on PCA clustering pattern, the diverse indica varieties revealed distribution of them across clusters and grouped preferentially according to its genetic (Fig. 2).Moreover, group or sub-gene pool within indica rice in tropical and temperate regions were still overlap, indicating seed exchange has been existed for many varieties/ accessions, in which diversified rice types were used for various agronomic and cultural practices.Importantly, the morphological changes probably were achieved by the successive rice improvement program (Wijayawardhana et al., 2015).Similar morphological characteristics that are expressed phenotypically could show dissimilarity among accessions in their genomes (Cole-Rodgers et al., 1997).
Consequently, genotypes with genetic nearness may be different in the phenotypic characters; therefore genomewide genotyping with custom SNPs was applied to support morphological characterization in our study.
Assorted rice performance in Indonesia has been progressively studied based on phenotypic and molecular characterization, but limited for the use of advantage of SNP markers.Our results were complementary previous reports on desired morpho-agronomical traits in Indonesian rice breeding (Setyawati, 2003;Suhartini, 2010;Rohaeni and Hastini, 2015) that was supported with molecular markers approaches (Thomson et al., 2007;2009;Utami et al., 2011).384-plex SNP used in our study has been designed based on previous version of 1536 SNP chip which enabled to identify genetic relationship of 284 accessions into japonica/ tropical japonica and dominated indica group (Utami et al., 2013).The 384 SNPs variation along with morphological data confirmed a clear separation of rice collection into two main groups (Fig. 4), indica/tropical japonica (clade I) and indica/japonica/tropical japonica (clade II).Genomewide diversity, decay of linkage disequilibrium and traits of interest possibly affect the genetic relatedness among varieties within and between subspecies (Thomson et al., 2009).Most of indica varieties were distributed into all major clades and sub-clade, reflecting their similar genetic tendency.These morphological and molecular characters observed in these 60 Indonesian rice genotypes could be relevant to previous reports (Rakshit et al., 2007;McNally et al., 2009;Gustavo and Martinez, 2014).
The presence of tropical japonica accessions closer to indica or japonica groups, suggesting their behavior at molecular level which could be explained by genetic flow in a zone transition between japonica and indica.These rice subspecies seemed to share alleles on the domestication selected morphological characters could be addressed for characterization and conservation of rice germplasm.
process.Japonica and tropical japonica are considered as outcomes of selection for adaptation to different ecology and climate (Garris et al., 2004).Furthermore, genetic make-up and environment could affect the morphology of rice plant (McCouch et al., 1998;Rajapakse et al., 2000).This clustering based on combined morphology and 384 SNPs is relatively comparable with previous studies using 1536 SNPs (Zhao et al., 2010;Utami et al., 2013) and the other 384 SNPs designed somewhere else observed in rice (Tung et al., 2010;Chen et al., 2011;Parida et al., 2012).
Overall, the diversity of total rice genotypes used was adequate to provide a valuable resource for association analysis.
The high degree of association between the 154 SNPs with 15 morphological traits demonstrated a high power of these SNPs (Table 5).Particularly, this study highly focused on the vegetative characters of the rice plant, can be used effectively in order to capture a considerable morphological variability associated with Indonesian rice germplasm.The typical morphological characters on leaf, stem and panicle appeared in a certain accessions/varieties, reflecting the existence of its genetic control.This typical morphology along with molecular character could serve as differentiator of rice variety, leading to guide rice breeders that can effectively select morphologically more distinct individuals.
The high significant SNP markers were detected, such as TBGI367899 locus in chromosome 9 associated with leaf blade pubescence, id2000096 with flag leaf angle, TBGI118590 with leaf length, and both TBGI272460 and id1004591 with auricle color etc. Importantly, TBGI204002 (chromosome 4) highly associated with panicle length was located in the region of QTL for panicle length (qPL4), similarly to id1009867 that was in the regions of QTL for culm thickness (AQAW036).Moreover, id1004591 was identified to be near with POLD3-putative DNA polymerase delta complex and a QTL was related to culm thickness (AQDZ006) (Ratho et al., 1991;gramene.org).The SNPs located near or within genes/QTLs of interest need further validation for their useful application.Notably, this high throughput genotyping using 384 SNPs in our study proved their powerful and robustness to scan the genome of Indonesian rice germplasm and finally we were able to obtain potential SNP markers for desired traits.Therefore, the identification of morphological characters which are important to change the rice crop architecture has a greater implication.Since the power association analysis is dependent on both the density of SNPs and the quality of phenotypic data available for analysis (Tung et al., 2010), this result would give a clue of our germplasm panel for genome-wide studies to explore the molecular level of diverse morphological traits in rice (Yang et al., 2015) in Indonesia.

CONCLUSIONS
Morphological accompanied with high throughput SNP genotyping is more reliable in the evaluation of rice germplasm.The morphology of these rice germplasm could be a performance basis for selection of potential genotypes for further use in rice breeding in Indonesia.48 SNPs highly significant associated with selected quantitative and qualitative morphological characters are valuable and need a validation.Potential SNP markers in the regions of genes/QTLs of interest namely id1004591, id1009867, TBGI204002 were able to be detected.The implications of these results in our study for rice improvement are based on taking advantage of the genetic variability that exists in the Indonesian rice germplasms.

Fig 4 .
Fig 4. Genetic relationship of 60 rice accessions generated by a combined characterization with morphological traits and 384 SNPs observed.I and II are two major clades, and 1, 2 and 3 are three sub-clades belonged to the clade II generated from the Flapjack software.

Tabel 1: List of 60 rice genotypes used in this study
Indicaas descriptors which were observed in total rice genotypes in this study is depicted in Fig.1.Collar color, ligule shape, and panicle angle were uniform in total genotypes, showing green, 2 cleft and >90°, respectively.More than 80% of the genotypes had intermediate of leaf blade pubescence, <45° of flag leaf angle, white of auricle, purple lines of leaf sheath color, <30° of culm angle, heavy of secondary branching of panicle, and green color of culm.However, a greater variety of this germplasm was identified on leaf angle, leaf blade color, ligule color, and attitude of panicle branch.The mean, maximum and minimum values of five quantitative traits among this germplasm also varied.The mean values of leaf length, leaf width, and panicle length were 40.15, 0.66 and 24.18 cm, respectively.These genotypes possessed wide ranges of basal internode and ligule length, from 0.50 to 0.80 cm and 1.25-2.60cm,respectively.The mean and range values of selected five quantitative morphological traits are shown in Table2.
Dancey and Reidy (2004)among traits revealed that only few morphological traits had either significant positive or negative correlations.Based on the correlation category fromDancey and Reidy (2004), correlation coefficient of Pearson (Pearson's r) explained weak to moderate (0.3<r<0.7).Culm angle with leaf blade color (r= -0.31), and leaf width with diameter of basal internode had moderate and weak negative correlations (r= -0.28), respectively.Between leaf sheath pubescence and panicle length (r= 0.38), leaf sheath color and culm color (r= 0.38), and flag leaf angle and with leaf angle (r= 0.41) were moderate positive correlation.

Table 3 : Eigenvalues of the first two principal component axes (PC) for the 19 morphological traits used to classify the rice germplasm Characters Principal component axes PC1 a PC2 b
a,b : Eigenvalues of PC1 and PC2 are 2.0 and 1.9, respectively

Table 5 : List of selected 48 SNPs highly significant associated with morphological traits observed in rice germplasm in this study Traits SNP Marker Chr P-value** SNP
**Significant level at P<0.01, Chr: Chromosome, SNP: Single nucleotide polymorphism