Mapping the Genomic Regions Controlling Germination Rate and Early Seedling Growth Parameters in Rice

Seed vigor is the key performance parameter of good quality seed. A panel was prepared by shortlisting genotypes from all the phenotypic groups representing seedling growth parameters from a total of 278 germplasm lines. A wide variation was observed for the traits in the population. The panel was classified into four genetic structure groups. Fixation indices indicated the existence of linkage disequilibrium in the population. A moderate to high level of diversity parameters was assessed using 143 SSR markers. Principal component, coordinate, neighbor-joining tree and cluster analyses showed subpopulations with a fair degree of correspondence with the growth parameters. Marker–trait association analysis detected eight novel QTLs, namely qAGR4.1, qAGR6.1, qAGR6.2 and qAGR8.1 for absolute growth rate (AGR); qRSG6.1, qRSG7.1 and qRSG8.1 for relative shoot growth (RSG); and qRGR11.1 for relative growth rate (RGR), as analyzed by GLM and MLM. The reported QTL for germination rate (GR), qGR4-1, was validated in this population. Additionally, QTLs present on chromosome 6 controlling RSG and AGR at 221 cM and RSG and AGR on chromosome 8 at 27 cM were detected as genetic hotspots for the parameters. The QTLs identified in the study will be useful for improvement of the seed vigor trait in rice.


Introduction
The yield potential of a variety is realized by using quality seeds along with the recommended practices, which are important inputs for rice production. The future demand for the staple food rice is increasing due to a worldwide increase in population size. We need an additional production of about 1-2 MT of rice per year to fulfill our rice requirement by 2030 [1]. However, the higher production need beyond 2030 must

Phenotyping of Germination Rate and Early Seedling Growth Parameters in the Germplasm Lines
Seed physiological characteristics such as rate of shoot growth (RSG), absolute growth rate (AGR), mean germination rate (MGR) and relative growth rate (RGR) were calculated for the association study. A total of 50 seeds of each line in three replications were used for germination following the top of paper method by incubating at 30 • C. The estimate for RSG was performed by measuring the shoot length per day on the 7th and 10th days of germination and expressed in cm day −1 . AGR is the rate of change of seedling length per day, whichwas calculated as per the procedure of Reford [33]. Seedling relative growth rate is the incremental accumulation of dry mass per unit of existing dry mass, which was estimated following the procedure Fisher [34]. Germination time is the time for the seed to germinate. The value of mean germination rate (MGR) is the velocity of germination. It is the inverse of mean germination time (MGT). MGR is expresses as day −1 and ranges from 0 to 1. MGR and MGT were calculated according to the formula of Ellis and Roberts [35], MGT (day) = ΣnD/N, where n is the number of seeds germinated on each day, D is number of days counted from the beginning of germination and N is the total number of seeds germinated. MGR (day −1 ) = 1/MGT. For estimation of the seedling growth parameters, five seedlings from each replication were recorded and the mean value of each germplasm line was estimated.
Analysis of variance (ANOVA) of the four traits was conducted using the software Cropstat 7.0. The other statistical parameters, namely range, mean and coefficient of variation (CV %), were computed using this software. The relationships among the four growth parameters were determined using Pearson's correlation coefficient values based on the average values of the 124 germplasm lines, and a correlation matrix heat map was prepared. The genotypes were grouped into very high, high, medium and low value-carrying germplasm lines based on the seedling growth parameters for this association study.

Genomic DNA Isolation, PCR Analysis and Marker Selection
Rice seedlings of 15 days old were used for genomic DNA extraction by adopting the CTAB method [36]. A total of 143 SSR (simple sequence repeat) markers available in the public domain were used for the study (Supplementary Table S2). The DNA fragments isolated were quantified by resolving the fragments through agarose gel electrophoresis. PCR amplification was carried out by using the 143 selected SSR markers, which covered all the chromosomes and positions for identifying polymorphic loci and estimating the diversity of the 124 rice genotypes (Table 1). PCR reaction was performed by adopting the standard protocol of a denaturation step (4 min, 95 • C), followed by 35 cycles of denaturation (30 s, 94 • C) and annealing/extension (30 s, 55 • C), extension (1.5 min, 72 • C), final extension (10 min, 72 • C) and storage at 4 • C (infinity). Electrophoresis of the PCR products was performed using agarose gel (2.5%) containing 0.80 g mL −1 ethidium bromide. The sizes of the amplicons were estimated by comparing the known 50 bp DNA ladder in the electrophoresis. An electric current of 2.5 Vcm −1 was connected to run the gel for 4 h, and a photograph was captured in a Gel Documentation System (SynGene). The procedures adopted in earlier publications were followed in this isolation and electrophoresis work [37][38][39].

Molecular Data Analysis
Scoring of genotyping data was carried out for the absence or presence of amplified products on the basis of primer-genotype combination. The data in binary form wereprepared as discrete variables for the result data of the marker-genotype entries. The diversity parameters, namely major allele frequency (A), polymorphic information content (PIC), number of alleles (N), gene diversity (GD) and observed heterozygosity (H), were determined by using the software Power Marker Ver3.25 for each SSR locus [40]. The population structure was generated by the software STRUCTURE 2.3.6 [41]. The ideal number of groups (K) was obtained by analysis with the software, which was run with K that varied from 1 to 10 and 10 iterations for each K value. The burn-in period for the high throughput was at 150,000, followed by 150,000 Markov Chain Monte Carlo (MCMC) replications adopted in the run. The value of ∆K of a subpopulation was obtained from the Evanno table, and the peak value was used. The peak value of L(K) was taken from the number of subpopulations. The most probable value of K was ∆K, which is a second-order change of the log probability value for the number of clusters detected by STRUCTURE [42]. The ∆K-value was obtained from the structure harvester, which is a function of K giving a clear peak for the optimal K-value [43]. The NEI coefficient dissimilarity index using an unweighted neighbor-joining unrooted tree with a bootstrap value of 1000 and the principal coordinate analysis for all the germplasm lines were performed by using DARwin5 software [44]. The F IT , F IS , F ST for the presence of molecular variance across the population and within the population, including between the subpopulations, were estimated. Analysis of molecular variance (AMOVA) was performed using the GenAlEx 6.5 software [45].
The "TASSEL 5.0" software was used to detect the marker-trait associations for the four growth parameters. GLM and MLM were adopted for the association study of molecular markers and the growth parameters [46]. A significant p-value and r 2 value were considered for markers with trait associations detected. The marker-trait associations were also checked by the quantile-quantile curve derived by the TASSEL software. An LD decay graph was generated by plotting the measured r 2 value of a marker pair with the distance between the marker pair. The adjusted p-values (q-values) of the false discovery rate (FDR) were computed to check the accuracy of the marker-trait association using R software as described in previous publications [23,28].

Phenotyping for Germination Rate and Early Seedling Growth Parameters in the Target Population
The mean values of 278 germplasm lines for germination rate and three seedling growth parameters were estimated during the wet season in 2019 (Supplementary Table S1). The germination rate and early seedling growth parameters such as rate of shoot growth, relative growth rate and absolute growth rate were estimated from the original population. Significant differences in the mean estimates of the four traits were noted among the germplasm lines. The frequency distributions of the 278 germplasm lines based on the mean phenotypic values were broadly classified into 5 groups for each parameter ( Figure 1). The genotypes were distributed into different groups and further categorized into subpopulations ( Figure 1). The panel population with 124 germplasm lines was prepared  (Table 1; Figure 2). The studied traits in the panel population also exhibited significant variation among the genotypes ( Table 1). The estimates of RSG present in the panel showed very high values in the germplasm lines AC. 9030, AC. 9035, AC. 9038, Kapanthi and Pk-21 ( Figure 2). In addition, a very high relative growth rate was recorded from the germplasm lines AC. 5993 and Kanakchampa ( Table 1). The germplasm lines AC. 43660, AC. 43669 and AC. 43675 showed a high value of >3.0 for AGR (Table 1). In addition, the germplasm lines with a high mean germination rate were estimated from the germplasm lines AC. 43663

Principal Component and Association Analyses
The first two PCs (principal components) were used to generate a genotype-by-trait biplot diagram of the 4 studied physiological growth parameters estimated from the panel containing 124 germplasm lines ( Figure 3A). Variations of 56.07% and 25.47% were exhibited by the first and second principal components, respectively. The scattering pattern revealed that the genotypes containing high estimates of growth parameters were placed in the 1st quadrant, which formed a cluster and accommodated 23 genotypes. Germplasm lines containing higher value for the MGR and growth parameters are depicted in a circle in the figure ( Figure 3A). The top right (1st quadrant) and bottom left (3rd quadrant) accommodated 65 germplasm lines, forming the biggest group. The majority of the germplasm lines are of a moderate type for the early seedling growth parameters and germination rate with inclusion of a few good types. The 2nd (bottom right) quadrant kept most of the germplasm lines containing high RSG and MGR ( Figure 3A).

Principal Component and Association Analyses
The first two PCs (principal components) were used to generate a genotype-by-trait biplot diagram of the 4 studied physiological growth parameters estimated from the panel containing 124 germplasm lines ( Figure 3A). Variations of 56.07% and 25.47% were exhibited by the first and second principal components, respectively. The scattering pattern revealed that the genotypes containing high estimates of growth parameters were placed in the 1st quadrant, which formed a cluster and accommodated 23 genotypes. Germplasm lines containing higher value for the MGR and growth parameters are depicted in a circle in the figure ( Figure 3A). The top right (1st quadrant) and bottom left (3rd quadrant) accommodated 65 germplasm lines, forming the biggest group. The majority of the germplasm lines are of a moderate type for the early seedling growth parameters and germination rate with inclusion of a few good types. The 2nd (bottom right) quadrant kept most of the germplasm lines containing high RSG and MGR ( Figure 3A).

Genetic Diversity Parameters' Analysis
The representative population comprising 124 germplasm lines shortlisted from the original population showed wide variation for the four studied physiological growth parameters based on the analysis using 143 SSR markers. Gene diversity and other diversityrelated parameters estimated for the marker loci are presented in Table 2. A total of 522 markers alleles were detected, showing 3.65 alleles per locus as an average value. The marker loci varied from two to seven alleles per locus, and the highest numbers of alleles were obtained by the marker RM493for the growth parameters. The estimated mean major allele frequency for the growth parameters associated with the markers was 0.582. The major allele frequency ranged from 0.274 (RM493) to 0.976 (RM14960) ( Table 2). The variation for PIC value ranged from 0.133 (RM6054) to 0.792 (RM495) with an average value of 0.474. The estimated mean Ho (heterozygosity) in the panel population was 0.109, which ranged from 0.00 to 0.963. A zero heterozygosity value was estimated for 19 markers namely RM328, RM1812, RM6947, RM14978, RM22034, RM258, RM1347, RM3423, RM405, RM421, RM6091, RM209, RM245, RM3351, RM471, RM461, RM8007518, RM274 and RM452.
The range for gene diversity (He) was from 0.150 (RM22034) to 0.817 (RM493), showing an average value of 0.530 for the panel.

Population Genetic Structure Analysis
The STRUCTURE 2.3.6 software was used to assess the genetic structure in the representative population by adopting probable subpopulations (K) and selecting a higher delta K-value for the studied physiological growth parameters. This grouped the germplasm lines into 2 subpopulations showing a high ∆K peak value of 349.8 at K = 2 (Supplementary Figure S1). Subpopulation 1 showed an inferred ancestry value of 0.728, while 0.272 was obtained for subpopulation 2. Similarity was noted among the members of each subpopulation, but many deviations were observed in the inclusion of the germplasm lines for the studied growth parameters. Therefore, the next highest peak of the ∆K peak was taken, and the population was classified into four subpopulations, which showed better correspondence with the studied parameters and structure group compared to the two subpopulations at K = 2 ( Figure 4A  The four seedling growth parameters exhibited a fair degree of correspondence among the structure subpopulation members present in the population and the studied trait at K = 4 compared to K = 2. The majority of the germplasm lines in subpopulation 1 showed the presence of a moderate value of growth parameters. Subpopulation 2 mainly accommodated genotypes showing high values, particularly for RSG, AGR and MGR traits. Subpopulation 3 accommodated germplasm lines carrying high values for one or  Table 1. The four seedling growth parameters exhibited a fair degree of correspondence among the structure subpopulation members present in the population and the studied trait at K = 4 compared to K = 2. The majority of the germplasm lines in subpopulation 1 showed the presence of a moderate value of growth parameters. Subpopulation 2 mainly accommodated genotypes showing high values, particularly for RSG, AGR and MGR traits. Subpopulation 3 accommodated germplasm lines carrying high values for one or more traits. Subpopulation 4 showed the presence of poor to moderate growth parameters carrying genotypes. The population also showed a low α value (α = 0.0683) estimated by the structure software at K = 4. The mean α-value had a positively leptokurtic distribution, while Fst 1 , Fst 2 , Fst 3 and Fst 4 showed almost symmetrical distributions with clear differentiation in the groups on the basis of distribution among the estimated Fst values (Supplementary Figure S2).

Analysis of Molecular Variance (AMOVA) and LD Decay Plot
Plants presenting populations related with each other based on the studied growth parameters in the representative population were clustered together and different clusters formed different subpopulations. The estimated AMOVA showed the existence of genetic variations within and between the representative population at K = 4 ( Table 3). The estimated genetic differentiation within and between the subpopulation at K = 4 was found to be 67% among individuals, 20% within individuals and 14% among the populations in the panel population. Wright's F statistics estimates calculated for the four traits showed deviation from Hardy-Weinberg's prediction. The parameters such as uniformity of individual within a subpopulation (F IS ) and individual within the total population (F IT ) were computed to determine the differentiation in the population. The F IT and F IS values based on the genotyping of 143 marker loci were 0.804 and 0.773, whereas F ST was 0.138 among the subpopulations. The population differentiation is measured by the F ST values or the subpopulations within the total population. Each subpopulation showed different F ST values, and the distribution pattern of the genotypes also showed clear cut differentiation of the population into three subpopulations (Supplemental Figure S2). Table 3. Analysis of molecular variance (AMOVA) of the panel population for germination rate and early seedling growth parameters in rice. The association of markers with different traits was successfully used to map the traits through a disequilibrium study. The continuance of LD decay is an important factor for obtaining the disequilibrium in the population. The LD decay rate indicates the markers associated with the growth parameters that will be useful in the discovery of allelic variants or new genes regulating these studied traits. The syntenic r 2 value was utilized to draw the LD decay plot of the population against the physical distance of the two markers in million base pairs ( Figure 5A). The linked markers showed a decrease in r 2 for an increase in the linkage distance. It was noted that the LD decay declined sharply for the linked markers at 1-1.5 mega base pairs and then the decay was gradual, with a very slow decay rate noted. Therefore, it is clear from the graph that LD decay is continuing for the four growth parameters in the population. The germplasm lines showing the admixture type may have originated in the evolution due to the existence of LD decay of these four traits. A similar trend is also observed in the marker R 2 , along with the marker 'F' versus marker 'P' plot ( Figure 5B,C). This study indicated the strength of the marker-trait association with the associated markers for the studied traits.

Sources of Variation
s 2023, 14, x FOR PEER REVIEW 17 of decay rate noted. Therefore, it is clear from the graph that LD decay is continuing for t four growth parameters in the population. The germplasm lines showing the admixtu type may have originated in the evolution due to the existence of LD decay of these fo traits. A similar trend is also observed in the marker R 2 , along with the marker F' vers marker P' plot ( Figure 5B,C). This study indicated the strength of the marker-trait as ciation with the associated markers for the studied traits.

Relatedness among the Germplasm Lines through Principal Coordinates and Cluster Analyses
The principal coordinate analysis (PCoA) for the two dimension diagrams is draw on the basis of the genotyping results obtained by using the 143 SSR markers that group the genotypes for genetic relatedness among the germplasm lines ( Figure 6). The iner

Relatedness among the Germplasm Lines through Principal Coordinates and Cluster Analyses
The principal coordinate analysis (PCoA) for the two dimension diagrams is drawn on the basis of the genotyping results obtained by using the 143 SSR markers that grouped the genotypes for genetic relatedness among the germplasm lines ( Figure 6). The inertia for component 1 and component 2 were 11.6% and 7.27% of total inertia, respectively. The germplasm lines were placed on the four quadrants at different spots, which formed four major groups (Figure 7). A total of 13, 13, 18 and 80 germplasm lines were allocated into the 1st, 2nd, 3rd and 4th quadrant, respectively. The germplasm lines of each subpopulation are clustered in different quadrants. The 4th quadrant showed a single major group accommodating 80 germplasm lines. The 2nd major group present in the 1st and 2nd quadrants accommodated 23 germplasm lines. Genotypes from subpopulation 1 and subpopulation 2 are placed together in this group. The majority of the germplasm lines of this group are good for seedling growth parameters. The admix genotypes are depicted in brick color ( Figure 6). The majority of the members of subpopulation 3 are in quadrant 4 and depicted in green color. Eight subpopulation 4 members along with seven admix types, two subpopulation 3 members and one subpopulation 1 member are allotted in the 3rd quadrant.    Table 1. The numbers are colored on the basis of subpopulations obtained from the structure analysis at K = 4 (SP1: blue; SP2: pink; SP3: green; SP4: violet; admix: red).
The germplasms UPGMA tree constructed based on the results of genotyping of 143 SSR markers in the representative population grouped the panel's genotypes into 4 groups as in the case of the PCoA plot. The colors of the four subpopulations depicted in the tree are blue for SP1; pink for SP2; green for SP3; violet for SP4; and red for admix type ( Figure 7A). The unweighted neighbor-joining tree classified the panel population into four different subpopulations including the admix types. The tree discriminated the germplasm lines into different clusters on the basis of the genotyping results using 143 SSR markers that corresponded with the germination rate and 3 early seedling growth parameters. The cluster accommodating subpopulation 3 was differentiated from SP2 by the presence of lines carrying high values for MGR and RSG, while moderate to high genotypes were seen for all four parameters in subpopulation 1. The admix types of germplasm lines in the population shown in the neighbor-joining tree are in red ( Figure 7A). A phylogenetic tree was also constructed using the unrooted tree. This tree lacks a common ancestor or node. Here, the distances of each germplasm line are depicted in the diagram ( Figure 7B). Both the trees are useful for finding the relationship between germplasm lines irrespective of the evolutionary time. The germplasms UPGMA tree constructed based on the results of genotyping of 143 SSR markers in the representative population grouped the panel's genotypes into 4 groups as in the case of the PCoA plot. The colors of the four subpopulations depicted in the tree are blue for SP1; pink for SP2; green for SP3; violet for SP4; and red for admix type ( Figure 7A). The unweighted neighbor-joining tree classified the panel population into four different subpopulations including the admix types. The tree discriminated the germplasm lines into different clusters on the basis of the genotyping results using 143 SSR markers that corresponded with the germination rate and 3 early seedling growth The panel population is broadly grouped into two clusters based on the phenotypic values of the four growth parameters of the germplasm lines. Again, this cluster is divided into two clusters. Finally, 10 different subgroups were seen in the dendrogram based on the values of the four seedling growth parameters (Figure 8). Subcluster I was the largest group, which accommodated 48 genotypes, while subcluster III was the smallest one, with only 4 germplasm lines. Cluster I mainly accommodated the germplasm lines with high AGR and RSG. Subclusters III and IV of cluster II separated out the genotypes carrying high estimates of three or four parameters. Subclusters I and II of cluster II accommodated genotypes mainly with high RSG and MGR. Admix genotypes were observed in cluster I, while no admix was seen in cluster II.

Association of Marker Alleles with Germination Rate and Early Seedling Growth Parameters in Rice
Marker-trait associations for four physiological traits were computed by the TASSEL 5 software using both GLM (generalized linear model) and MLM (mixed linear model)/K+Q models. The estimates of the associations were subjected to filtration at <1% error, i.e., 99% confidence (p < 0.01). Three growth parameters, namely RSG, RGR and MGR, showed significant associations with markers using both GLM and MLM analyses at p < 0.01. However, all four traits were significantly associated with markers analyzed by GLM and MLM separately. A total of 58 and 25 significant marker-trait associations were obtained when analyzed by GLM and MLM, respectively, at p < 0.01. The estimated marker R 2 value when using GLM analysis varied from 0.027 to 0.109, while the value varied from 0.049 to 0.116 with MLM analysis (Supplementary Tables S3 and S4). A total of nine significant marker-trait associations were obtained for RSG, RGR, AGR and MGR by both the models at >0.05 markers R 2 and p < 0.01. AGR showed a significant association

Association of Marker Alleles with Germination Rate and Early Seedling Growth Parameters in Rice
Marker-trait associations for four physiological traits were computed by the TAS-SEL 5 software using both GLM (generalized linear model) and MLM (mixed linear model)/K+Q models. The estimates of the associations were subjected to filtration at <1% error, i.e., 99% confidence (p < 0.01). Three growth parameters, namely RSG, RGR and MGR, showed significant associations with markers using both GLM and MLM analyses at p < 0.01. However, all four traits were significantly associated with markers analyzed by GLM and MLM separately. A total of 58 and 25 significant marker-trait associations were obtained when analyzed by GLM and MLM, respectively, at p < 0.01. The estimated marker R 2 value when using GLM analysis varied from 0.027 to 0.109, while the value varied from 0.049 to 0.116 with MLM analysis (Supplementary Tables S3 and S4). A total of nine significant marker-trait associations were obtained for RSG, RGR, AGR and MGR by both the models at >0.05 markers R 2 and p < 0.01. AGR showed a significant association with four markers; RSG with three markers; and RGR and MGR with one marker each by both the models at p < 0.01 (Table 4). The Q-Q plot also confirmed these marker-trait associations for the seedling stage physiological parameter traits in rice (Figure 9). High r 2 values > 0.1 were detected from the marker-trait association of markers RM337 and RM494 for the trait RSG. The three markers associated with the parameter RSG, namely RM337, RM22034 and RM494, are present at 27, 56 and 221 cM positions on chromosomes 8, 7 and 6, respectively (Table 4). Five markers were significantly associated with AGR, as detected by both generalized linear model and mixed linear model analyses at p < 0.01 and r 2 value > 0.05. The chromosomal regions governing the trait AGR were detected on chromosomes 8, 6 and 4. Among the four markers, RM16686 showed the highest marker R 2 value of 0.076 analyzed by GLM and 0.095 by MLM. The strongly associated marker is located on chromosome 4 at the 300 cM position. RM337, RM7179 and RM494 were significantly associated with the trait at 27, 159 and 221 cM positions on chromosomes 8, 6 and 6, respectively. In addition, two markers, namely RM1812 and RM3735, were significantly associated with the growth parameters RGR and MGR at 44 and 80 cM positions on chromosomes 11 and 4, respectively. The Q-Q plot also confirmed these marker-trait associations for the four physiological growth parameters in rice ( Figure 9).
The common markers were detected to be associated with more than one seedling stage growth parameter in rice. Two markers, RM494 and RM337, exhibited significant associations with two physiological growth parameters, namely RSG and AGR, by both the models at <1% error ( Table 4). The markers are present at 221 and 27 cM positions on chromosomes 6 and 8, respectively. with four markers; RSG with three markers; and RGR and MGR with one marker each by both the models at p < 0.01 (Table 4). The Q-Q plot also confirmed these marker-trait associations for the seedling stage physiological parameter traits in rice (Figure 9).

Discussion
The genotypes present in the population were significantly different from each other for the four studied seedling growth parameters ( Table 1). The studied physiological growth parameters at the seedling stage also showed significant correlation among themselves. The panel population showed higher genetic variations and higher correlation coefficients for the growth parameters, indicating the usefulness of the population for improvement of RSG, RGR, AGR and MGR in rice (Table 1; Figure 3B). Reports of usefulness of high variations for many traits in crop improvement programs were previously published [47][48][49][50]. Phenotypic variations for the 4 traits and the existence of diversity estimated based on the genotyping results using 143 markers confirmed the differentiations of the whole population into subpopulations ( Table 2). Existence of more alleles and moderate to high PIC in the population indicated that markers are informative, which may be useful in seedling growth parameter improvement programs. The germplasms used in this study were collected from the states where rich rice genetic diversity exists. In the present experiment, germplasm lines from the Jeypur tract of Odisha were used, which is the secondary center of origin of rice. The germplasm lines AC. 9006, AC. 9021, AC. 9028, AC. 9030, AC. 9035, AC. 9038, AC. 9043, AC. 9044, AC. 9058 and AC. 9063 were high in the content of the four studied traits. These germplasm lines will be useful as potential donors for the seedling-stage growth parameters in breeding programs (Table 1). Therefore, it is expected that the identified donor lines from this population will be useful in improvement of seed vigor and its related traits in rice. The availability of genetic diversity in rice germplasms has been reported by many earlier researchers [51][52][53][54][55][56][57][58]. The presence of different Fst values and four structure groups supported the existence of various LD groups in the population. Existence of admix-type landraces along with a low α value in the population revealed that these traits were originated from a single source and formed many admix races with different ancestry values in the evolutionary process. Earlier workers have supported the correspondence of structure and growth parameters in rice [31,[59][60][61][62]. Additionally, many publications on the correlation of phenotypes of various traits with structure subpopulations have been published [59,60,63].
All four physiological growth parameters were analyzed by both GLM and MLM approaches and found to be associated with nine SSR markers ( Table 4). The markers found to be associated with traits at p < 0.01 with a low 'p' value and detected by both the models are considered as robust and very useful for breeding programs. The markers, namely RM337, RM22034, RM494, RM1812, RM7179, RM16686 and RM3735, will be useful for improvement of seedling growth parameters through molecular breeding for enhancing vigor in rice ( Table 4). The quantile-quantile (Q-Q) plot also confirmed the detected associated markers with the growth parameters in rice ( Figure 9). Marker-trait associations have been reported previously by many workers in rice [24,30,64].
QTLs for the germination rate in rice were reported in previous studies [11][12][13][14][15]65]. QTLs controlling the germination rate located on chromosome 4 were reported by Wang et al. [12] and Yang et al. [15]. Wang et al. reported QTL in the marker interval of RM252-RM317, which is at 102.2-117.5 cM position. As per the report of Yang et al. [15], the QTL is located from 27.3 MB to 28.4 MB. We detected the association for the trait on the chromosome at 80 cM, which is nearer to the QTL reported by Wang et al. [12]. In addition, we detected at a physical distance of 26.2 MB of the marker. Therefore, our detected QTL may be qGP-4 as per Wang et al. [12] or qGR4-1 by Yang et al. [15]. We detected four significant associations of markers with the trait absolute growth rate. Different reports on genetic analysis of growth parameters at the seedling stage were published [65,66]. The growth parameter relative shoot length showed a significant association with the markers RM337, RM22034 and RM494 detected by both GLM and MLM analyses. The bi-parental mapping study of Han et al. located the QTL on chromosome 1 [67]. The team [67] reported a QTL on chromosome 12 at 101-107 cM position. We detected the associations at 27, 56 and 221 cM positions on the chromosomes 8, 7 and 6, respectively. Abe et al. [18], Dang et al. [68] and Anandan et al. [22] also reported the locus controlling the trait but on different chromosomes. The QTLs detected by us are designated as qRSG8.1, qRSG7.1 and qRSG6.1 on the chromosomes 8, 7 and 6, respectively ( Figure 9). Kato et al. [69] mapped the QTL for relative growth rate on chromosome 4 within the marker interval of RM8213-RM335 and the other QTL on chromosome 7 within RM8249-RM5120. The identified loci in this experiment were present on chromosome 11, which was not reported in earlier studies. Hence, the locus we detected was a new QTL, designated as qRGR11.1.
The QTLs controlling RSG and AGR, namely qRSG6.1 with qAGR6.2 on chromosome 6 at 221 cM position, were detected to be co-inherited. Similarly, the QTLs qRSG8.1 and qAGR8.1 on chromosome 8 at 27 cM position for both the traits were detected to be colocalized with each other. In earlier mapping studies, there were reports of co-localization of genes/QTLs controlling traits such as protein, Fe, Zn and antioxidant contents in grains and different growth parameters in rice [24,28,60,62]. It is very easy to improve the traits controlled by the QTL hotspots located on the chromosomes.

Conclusions
Seed quality and vigor can be improved by improving the seed germination rate and the early seedling growth parameters in rice. The seedling growth parameters such as germination rate, RSG, RGR and AGR showed wide variations in a panel of 124 genotypes that represented 278 genotypes. High values of the growth parameters were identified in the landraces such as AC. 9006, AC. 9021, AC. 9028, AC. 9030, AC. 9035, AC. 9038, AC. 9043, AC. 9044, AC. 9063 and AC. 9058. Based on the fixation indices values computed from the subpopulations, the presence of LD was confirmed in the panel population. Existence of moderate to high PIC values, gene diversity and related parameters were obtained in the population by genotyping with a set of 143 markers. The population in the panel was categorized into few subpopulations and subclusters, which had good correspondence with the members for the four studied parameters. A total of eight novel QTLs controlling the traits, namely qAGR4.1, qAGR6.1, qAGR6.2 and qAGR8.1 for AGR; qRSG6.1, qRSG7.1 and qRSG8.1 for RSG; and qRGR11.1 for RGR, were detected from the mapping population. The reported QTL for germination rate, qGR4-1, was validated in this mapping population and will be useful in marker-assisted breeding. Additionally, QTLs present on chromosome 6 controlling RSG and AGR at 221 cM and on chromosome8 at 27 cM for RSG and AGR were detected as genetic hotspots for the parameters. The QTLs identified in the study will be useful for crop improvement programs for seed-vigor-related traits in rice.

Supplementary Materials:
The following supporting information can be downloaded at: https: //www.mdpi.com/article/10.3390/genes14040902/s1, Figure S1: (A) Plot of ∆K value vs. change in the log probability of data between successive K values; (B)population genetic structure obtained for 124 genotypes present in the panel on the basis of membership proportions of individual lines at K = 2. The germplasm lines showing an ancestry value ≥80% were assigned as subgroups and others as admixture groups. The serial number of the genotype in the diagram is as per the serial in Table 1. Figure Table S1: Mean estimates from 278 rice germplasm lines. Table S2. Simple sequence repeat markers used for association mapping of germination rate and early seedling growth parameters in a panel population of 124 rice germplasm lines. Table S3. Significant marker-trait associations detected for germination rate and early seedling growth parameters by GLM approach at p < 0.01 using 143 SSR markers. Table S4. Significant marker-trait associations detected for germination rate and early seedling growth parameters by MLM approach at p < 0.01 using 143 SSR markers.