Evaluation of Traits Associated With Seed Characteristics in Arkansas Restorer Lines

Rice Grain dimension and weight are two critical factors for marketing and increasing yield capacity. Seed shape is measured by its length, width, thickness, and ratio of length-width. In this study, an experiment was conducted in a controlled condition from fall 2017 to 2020 to identify QTL and candidate genes associated with seed dimension and weight using a bi-parental population resulting from two University of Arkansas developed genotypes: a restorer line 367R and an advanced breeding line RU1501139, in Stuttgart, Arkansas. Five seed dimension traits, including seed length, seed width, seed thickness, seed length-width ratio, and 100-seeds weight, were obtained for QTL detection. The study detected a total of 17 QTL. Four QTL associated with seed length were identied, in which two were positioned on chr. 3, one on chr. 7, and one on chr. 11. Two QTL related to seed length-width ratio were detected on chr. 3 and 7. Whereas a total of three QTL were identied for seed thickness, one each on chr. 5, 6, and 8. Eight QTL associated with seed weight were found, of which four QTL were detected on chr. 12, two each on chr. 1 and 10, and one on chr. 3. Of 17 QTL, four QTL originated from RU1501139, while the origin of the other 13 QTL was 367R. Since multiple genes could control the yield and seed physical characteristics, the detected QTL can play a role in introducing superior parental lines for developing conventional and hybrid rice production. the US. Therefore, the objective of this study was to identify QTL associated with seed characteristics, including SDLG, SDWD, SDTH, L/W, and SWT100. Results of this study could contribute to the improvement of the genetic background of yield-related QTLs through the introduction of each QTL themselves for the advancement of rice’s yield potential. more signicant impact on the phenotypic variations in a 17 QTL QTL associated with seed Annotation ve QTLs It can be concluded that 1) the annotation analysis of the QTL validates our nding via previously reported genes/QTLs associated with traits, and 2) these QTLs can be incorporated into the genomes of new superior genotypes.


Introduction
Rice (Oryza sativa L.) is one of the major crops for food and income resources for almost half of the world's population. With the rapid increase in the world's population, rice production must continuously increase as well. To satisfy the demand, an increase of 30% in rice production by 2050 is necessary (World Bank, 2013; Feng et al., 2014). In order to speed up the improvement of rice yield, yield components must be improved. In particular, the number of seeds per panicle, panicle number, and seed weight should be further studied (Weng et al., 2008. Seed width (SDWD hereafter) is controlled by multiple genes and several quantitative trait loci (QTLs) and has a great impact on improving yield (Weng et al., 2008;. Additionally, increasing seed dimensions are key breeding objectives for higher yield. Other Seed dimension characteristics that affect the yield potential include seed length (SDLG hereafter), seed width (SDWD hereafter), seed thickness (SDTH hereafter), and seed length-width ratio (L/W hereafter) Qiu et al., 2012. Based on L/W, rice is classi ed into three market classes long-grain, medium-grain, and short-grain (Hardke et al., 2018& Qiu et al., 2017. Seed length and SDWD and their ratio determine the kernel size where the ratio is between 3.0 or greater in long-grain rice and 2.0 to 3.0 in medium-grain and 2.0 or smaller for short-grain rice. Researchers have reported several QTLs related to yield and seed sizes (Fan et al., 2006. Che et al. (2015) conducted a QTL study on an F 2 population created by crossing two indica rice lines (RW11 x BobaiB) that were signi cantly distinct (about 37%) from each other in terms of their SDLG s. They identi ed a QTL identi ed on chr. 2 as GL2 from the backcross with RW11. The GL2 improved the seed dimension around 24% for SDLG, 16% for SDWD, and about 27% more in 1000 seed weight. Qiu et al. (2020) conducted a two-year (2015-2016) genetic mapping study to clarify the QTLs associated with seed dimension. Qiu et al. (2020) used 1016 accessions in ve populations: indica, japonica, aus, basmati, and admixture from the 3K Rice Genome Project (accessions collected from China, India, Philippines, Bangladesh, Japan, and other Asian countries). Seventy QTL were identi ed for seed dimension (SDLG, SDWD, L/W) on all 12 chromosomes. Twenty-four QTLs were identi ed on chr.s 1-7and 9-11 for L/W, and the phenotypic effect was between 1-30%. Twenty-one QTLs were identi ed on all chromosomes, excluding chr.s 10 and 12 for SDWD, and the phenotypic variation changed between 1 and 42%. They detected 25 QTLs for L/W on chr.s 1-8, 11, and 12 with between about 1 and 28% phenotypic variation (Qiu et al., 2020). Eizenga et al. (2018) identi ed a total of 27 QTLs for yield-related traits. A recombinant inbred line (RIL) population was developed using two tropical japonica lines, 'Estrela' and 'NSFTV199.' F 1 seeds were advanced to F 7 , producing a nal population size of 256 RILs. Seed characteristics studied include SDLG, SDWD, L/W, and 100-Seed weight (SWT100hereafter). The research detected seven QTLs, including a major QTL 'qHULGRLG3', associated with grain length explaining around 40% of the phenotypic variation on chr.3. Six QTL for SWT100 were identi ed with the major QTL 'qHULGRWD5', explaining 38% of the phenotypic variation. Eight QTL were detected for L/W, which were at the same locations as QTL 'qHULGRLG3 and qHULGRWD5', explaining 32.6% and 38.9% phenotypic variations, respectively. Six QTL were identi ed for SWT100. University of Arkansas (UA hereafter) Rice Research and Extension Center rice breeding program located in Arkansas, US, has developed a number of rice cultivars used in most US Southeast rice-growing regions, some of which are used by the public rice breeding institutions for developing new rice cultivars. However, the knowledge of seed dimension traits in UA's rice germplasm is limited.
Therefore, this study focuses on identifying QTLs associated with seed characteristics that can be found in the rice breeding program in the State of Arkansas and other rice breeding programs in the US. Therefore, the objective of this study was to identify QTL associated with seed characteristics, including SDLG, SDWD, SDTH, L/W, and SWT100. Results of this study could contribute to the improvement of the genetic background of yield-related QTLs through the introduction of each QTL themselves for the advancement of rice's yield potential.

Plant Materials
A bi-parental population resulting from a cross between the restorer line '367R' and a non-restorer line 'RU1501139' was developed for this study. Restorer line 367R is medium-grain rice and was developed at the UA's Rice Research and Extension Center (RREC), Stuttgart by Yan et al. (2012). It is derived from Katy/IR30//IR140 (PI-458443)/Jasmine-85(PI-595927) crosses. Non-restorer line RU1501139 is a long-grain advanced line developed by the RREC long-grain rice breeding program. A total of 300 F 2 plants from this population were grown in three replications in a greenhouse using a completely randomized design (CRD) to evaluate traits associated with seed characteristics. The F 2:3 seeds were harvested and used for phenotypic evaluation. Each replication consisted of three plants. Three panicles for each plant were randomly collected in the greenhouse. The panicles were dried (15% moisture) and threshed in Stuttgart, Arkansas. In order to analyze the seed dimension, 30 seeds from each line were randomly selected, cleaned, and evaluated via Mettler Toledo® balance and Winseedle® Pro to measure the seed dimension's signi cance level. According to the JMP Pro 14 software (SAS Institute Inc., Cary, NC), an ANOVA analysis followed by Student's T-test had signi cant results regarding SDLG, SDWD, L/W, SDTH, and SWT100 Phenotyping The F 2:3 seeds from the 367R x RU1501139 population were harvested in the greenhouse and measured for L/W, SDWD, SDTH, L/W, and SWT100 in the Spring of 2018 at RREC in Stuttgart, Arkansas. One hundred seeds from each F 2:3 line were measured to evaluate seed dimension and SWT100 via Winseedle® Pro and the Mettler Toledo® balance, respectively. Three replications were obtained for each F 2:3 line.
Each replication consisted of 100 seeds. The mean for each F 2:3 line was calculated from average value of three replications in an Excel® le..
One-way ANOVA analysis followed by Student's T-test signi cance results of the seed dimension (SDLG, SDWD, L/W, SDTH, and SWT100) ( Table 1). In addition, multivariate analysis was run to understand the correlations between traits using JMP Pro 14 software. composite interval mapping (ICI) software using genotypic data from F 2 and phenotypic data from F 2:3 seeds while making QTLs related to seed dimension. The ICI Mapping was used with the Kosambi function for linkage mapping, and SNP markers were ordered for linkage mapping. Identifying and detecting the QTLs, 2.5 LOD score, was considered a threshold level for a major QTL. The Oryzabase database was used to detect any co-localized QTLs. The distribution of the seed dimension and detected QTLs were analyzed using JMP Pro 14 software (Fig. 1). Oryzabase, a comprehensive rice data source, was used to identify candidate genes.

Preliminary Study
The ANOVA study showed that there are signi cant differences between 367R and RU1501139 on SDLG, L/W (p-value < 0.001), and SWT100, SDWD (p-value < 0.05). However, there was no difference in SDTH between these two lines ( Table 1).

Analysis of F 2:3 Population
SEED LENGTH: The distribution of F 2:3 for SDLG followed a normal distribution (Fig. 1). SDLG had a mean of 9.7 mm with a range from 8.2 to 11 mm. The trait had a standard deviation (SD) of 0.48 and a standard error (SE) of 0.03. These two values explained the signi cance of seed length with a p-value < 0.001 for the population (Table 1).
SEED WIDTH: The distribution of F 2:3 for SDWD followed a normal distribution (Fig. 1). There was no difference in seed width with a mean of 2.5mm and ranges from 2.3 to 2.7mm. The seed width had a 0.15 SD and SE of 0.06. While the trait was not signi cant at a p-value of 0.001, it had signi cance with a p-value < 0.05 (Table 1).
SEED LENGTH-WIDTH RATIO: The distribution of F 2:3 for L/W followed a normal distribution (Fig. 1). There was no difference for seed width; however, the seed length-width ratio had a signi cant difference with a mean of 3.74 mm and a range from 3 to 4.5 mm and an SD of 0.29 and an SE of 0.018. In addition, a signi cant difference between parents 367R and RU1501139 was expressed with a p-value < 0.001 for the population (Table 1).
Seed Thickness: The distribution of F 2:3 for SDTH followed a normal distribution (Fig. 1). For SDTH, the mean of SDTH in the population was 2.11 mm and ranged from 1.7 to 2.11 mm. The SD for SDTH was 0.09, and SE was 0.012. The difference between parents 367R and RU1501139 was not signi cant with a value p-value > 0.05 (Table 1).
100-seed weight: The distribution showed the majority of the F 2:3 lines ranged between 1.25 to 1.5gr (Fig. 1). For SWT100, the mean was 2.5gr, ranging from 2.3 to 2.7gr. The trait had a 0.15 SD and SE of 0.06 in the population. The difference between parents in the population expressed a p-value < 0.05 (Table 1).
Genotypic study A total of 17 major QTL were identi ed in the bi-parental population of 367R x RU1501139. For SDLG, four QTL were detected, including two QTL, qSDLG 3 − 1 and qSDLG 3 − 2 on chr. 3, and one QTL qSDLG 7 − 1 qSDLG 11 − 1 each on chr.s 7 and 11, respectively (Fig. 3). The detected QTL were linked to RU1501139, inferred increasing seed yield, and explained 5.1 to 8.4% of phenotypic variation (PVE) on the population ( Table 2). No major QTL for SDWD was detected; however, 12 minor QTL were found, including eight minor QTL with (2 < LOD < 3): two QTL on chr. 2, and three QTL each on chr.s 7 and 10, respectively. Two major QTL, qL/W3-1, qL/W7-1 were detected on chr.s 3 and 7 for L/W. These two QTL were co-localized with the QTL, qSDLG − 2 and qSDLG 7 − 1, identi ed for SDLG. The detected QTLs were linked to RU1501139 and explained 5.5 to 11.1% of phenotypic variation (PVE) in the population. The qL/W3-1 and qL/W-1 were co-localized with other detected QTL, qSDLG 3 − 2, and qSDLG 7 − 1, respectively (Table 3). Eight QTL were identi ed for SWT100, including two QTL on each of chr. 1, 2, 10, and 12. Seven of these QTLs were co-localized with previously reported QTLs, AQEI043, AQBA011, AQAP004, AQCI003, AQCS003, AQAE008, and AQF014, respectively (Table 3). Furthermore, all eight QTL originated from 367R. Three QTLs were identi ed on chr.s 5, 6, and 8 associated with SDTH. The QTL qSDTH5-1 on chr. 5 co-localized with a previously reported QTL AQFU013 (Table 3) for seed thickness. The detected QTLs linked to the 367R had a range of 4.6 to 7.5% phenotypic variation.   * qSDLG3, QTL associated with seed length; qL/W, QTL associated with Seed Length/ Width ratio; qSWT100, QTL associated with 100 seed weight; qSDTH, QTL associated with seed thickness † The list of genes and QTL and information regarding them can be found in the website Gramene at https://archive.gramene.org/ Detection of Candidate Genes for Major QTL A total of ve candidate genes were identi ed via rice genomic annotation using the online rice database of Oryzabase (https://shigen.nig.ac.jp/rice/oryzabase/), including four for SDLG and two candidate genes for SWT100 (Table 3). Two candidate genes of the GL-7 and OsGASR9 are identi ed within a detected QTL qSDLG 7 − 1 (2.3×10 6 ..2.3×10 6 ) associated with SDLG. GL-7 is a previously reported gene regulating seed length by increasing the length and starch structure in the endosperm . OsGASR9 is a gene associated with plant growth that can be detected in all parts of plant, specially panicle. The OsGASR9 is associated with plant growth and development and increases SDLG and SWT100 by increasing the e ciency of gibberellic acid . It is worth noting that qSDLG 7 − 1 is co-localized with another detected QTL qL/W7-1 associated with L/W.
Two candidate genes were identi ed on the detected QTL qSDLG 11 − 1 (16.28 ×10 6 .. 17.69 ×10 6 ) associated with SDLG on chr. 11, including Rice Big Seed-1 (RBG1) and Flower and Leaf Color Aberrant (FLA). The RBG1 gene is responsible for seed development, abiotic stress tolerance, and the gene improves root development by enhancing the plant's auxin level (Lo et al., 2020). The RBG1 is 948 bp, and its four allelic genes are located near the RBG1gene, 5 kb to M37341, ~ 27 kb to M37342, and M82594l, 46 kb to M44256 (Lo et al., 2020). The FLA gene is a ubiquitously expressed gene and a key factor for ower and chloroplast development. The FLA improves seed length and rice yield. The FLA is located between the marker M11-3 and S6 with 56 kb on the long arm of chr. 11 (Ma et al., 2019). One gene (HAP5L) is located within a detected QTL, qSWT100-10-1(6.64 × 10 6 ..9.26 ×10 6 ) associated with SWT100. The HAP5L is an endosperm-speci c gene-regulating starch accumulation and protein concentration (Xiong et al., 2019). The accumulation of starch increases the width, but any decrease in HAP5L causes sharp reductions in seed weight (Xiong et al., 2019).

Discussion
In this study, we aimed to identify the genetic sources associated with seed characteristics in rice. The preliminary study on two genotypes (367R and RU1501139) determined signi cant differences between the two genotypes for four seed characteristics of SDLG, SDWD, L/W, and SWT100. Restorer line 367R is a medium-grain rice cultivar, while RU1501139 is a long-grain breeding line. The seed length-width ratio is an essential measurement for the classi cation of rice cultivars. The results showed a positive correlation between L/W and SDLG, But a negative correlation between L/W with SDWD. In addition, the data showed a positive correlation between SWT100 with SDLG. Although there was no signi cant correlation between SWT100 and SDWD, the data showed a weak negative correlation between these two traits. Furthermore, results revealed that there was a positive correlation between SDWD and SDTH. Therefore, it can be assumed that longer and thicker seeds are heavier than shorter and wider seeds.
Enhancing seed yield, milling, and eating quality of rice can be achieved by developing superior cultivars by incorporating several agronomic traits, such as seed dimension and seed weight. The majority of these traits are classi ed as quantitative traits and are controlled by several QTL located in different parts of the rice genome. Each QTL has a different impact on the phenotypic variation. A breeder considers only those QTL that have the more signi cant impact on the phenotypic variations in a breeding program. In this study, we identi ed 17 major QTL and several minor QTL associated with seed characteristics. Annotation analysis revealed that ve detected QTL contain genes associated with seed characteristics, and 11 were co-localized with previously reported QTLs ( For example, one important detected QTL is qSDLG 7 − 1 on chr. 7 associated with SDLG. This QTL is co-localized with qL/W-1 associated with L/W. Further investigation identi ed two candidate genes, GL7 and OsGASR9, in this genomic region. Another important detected QTL qSDLG 3 − 2 on chr. 3 is associated with SDLG and co-localized with qL/W3-1, associated with L/W. On chr. 11, one QTL qSDLG 11 − 1 was detected for SDLG. Two candidate genes, RBG1 and FLA were identi ed on chr. 11 for SDLG. The RBG1 gene is associated with seed, root development, and stress tolerance by enhancing cell division and auxin levels; thus, it helps to improve root development and stress tolerance, which are essential factors for having a greater yield. (Lo et al., 2020). The second candidate gene, FLA, is a cell membrane protein that belongs to the Ubiquitin-speci c proteases. The FLA is a common amino acid for eukaryotic cells. The FLA improves seed length and yield by regulating chloroplast and ower development (Ma et al., 2019). Thus, we can summarize that the QTLs qSDLG 7 − 1 and q SDLG 11 − 1 contain several candidate genes associated with seed length and signi cantly impact the phenotypic variations; thus, these two QTL can be integrated into a new generation of long-seed rice cultivars.
Although the ANOVA analysis showed the signi cance of SDWD in this population, no major QTL were identi ed on chr.s. However, a total of 12 minor QTL were detected with an LOD from 2 to 3 LOD scores. Therefore, it can be assumed that SDWD is controlled by several minor QTLs that, overall, signi cantly enhance SDWD.
The ANOVA analysis showed there was no difference between 367R and RU1501139 for the SDTH trait, but the genotypic analysis identi ed three major QTLs associated with the SDTH trait. Furthermore, the genotypic analysis showed that the two QTL of qSDTH5-1 and qSDTH6-1 originated from 367R, while qSDTH8-1 originated from RU1501139. Therefore, there is a biological signi cance between these two genotypes despite no statistical signi cance due to these detected QTLs.
Rice is one of the major crops in the world, with vast marketing all around the world. The rice breeders' goals are to address farmers' and consumers' expectations by improving seed yield and seed characteristics. Further study is needed to identify major genes associated with these characteristics and developing molecular markers via a multi-location/year study. The results of this study can be used for markerassisted selection in breeding programs. 1) This material is the authors' own original work, which has not been previously published elsewhere.
2) The paper is not currently being considered for publication elsewhere.
3) The paper re ects the authors' own research and analysis in a truthful and complete manner.
4) The paper properly credits the meaningful contributions of co-authors and co-researchers.

5)
The results are appropriately placed in the context of prior and existing research.
6) All sources used are properly disclosed (correct citation).
7) All authors have been personally and actively involved in substantial work leading to the paper, and will take public responsibility for its content.

8)
We have no con icts of interest to disclose.

9)
We did not conduct any harmful activates on humans or animals, nor used any blood product or microorganisms.