Prediction of Heterosis for Agronomic Traits in Half-Diallel Cross of Rice (Oryza sativa L.) under Drought Stress Using Microsatellite Markers

Rice breeders are encouraged to classify potential F1-hybrids without crossing all viable mixtures by testing depending on genetic distance due to lack of labor and time in yield trials. The goals of this research were to establish heterosis and investigate the relationship between genomic distance and agronomic attributes under drought. Half-diallel mating design, 28 F1′s and 8 parents were evaluated under drought and genotyped using 11 microsatellite markers. In total, 39 alleles were detected. Results indicated that the greatest heterotic effects for grain yield were observed in Sakha 103 × Sakha 104 and GZ7576-10-3-2-1 × Giza 179, which gave 29.32–22.57% heterosis, respectively. Heterosis for grain yield in these crosses occurred as a rise in panicle weight, filled grains per panicle, low sterility and 100-grain weight. Correlations of marker-based genetic distance with mid-parent heterosis were positively and significantly correlated with sterility percentage (r = 0.390 *, p < 0.05). However, better-parent heterosis was positively and significantly correlated with sterility percentage (r = 0.352 *, p < 0.05) and grain yield per plant (r = 0.345 *, p < 0.05). Associations indicate that high grain yield and low sterility of rice crosses can be expected from microsatellite marker-defined distances of parents. This study indicated that genetic distance is very effective for heterosis prediction in breeding programs.


Introduction
Rice (Oryza sativa L.) is the greatest essential nutrition crop in the globe since it is the staple food for nearly 50% of the world population [1,2]. Around 490.27 million metric tons of rice were consumed universally in the 2018/2019 agricultural season, while 437.18 million metric tons were consumed in the 2008/2009 crop season [1]. In 2019, the world's top 10 main rice-producing nations were China, India, Indonesia, Bangladesh, Vietnam, Thailand, Myanmar, Philippines, Japan and Brazil [1]. In Egypt, during the 2019 rice-growing season, the area cultivated with rice was 485,622.54 hectares with an average of 8.37 t/h and total production was 4.8 million tons [1]. Major developments have happened in rice production because of the broad-scale adoption of developed rice varieties. However, the request for rice in low-income nations remains to be increased because of increases in public rice customers and developments in living standards. It is expected that the globe will need to yield 50% additional rice by 2050. To meet this task, high-yielding potential genotypes are needed. Several approaches have been employed for developing new genotypes with superior yielding potential and abiotic stress tolerance, such as population development, ideotype breeding, heterosis, wide crossing, genetic engineering and molecular breeding [3][4][5]. Rice is considered the greatest common and essential field crop in Egypt for numerous reasons as a (i) essential food after wheat for the Egyptian population, (ii) second exporting crop after cotton and (iii) soil reclamation crop for developing the production of the salty soil crops in which all farmers could increase money during its growing season. Rice productivity and production have remarkably improved year after year. Because of adopting the new short-duration rice genotypes, about 30% of the irrigation water use was saved each year [6].
Drought is severe abiotic stress that can cause serious losses of yield and productivity in most crop plants in dry and semi-dry regions [7]. Egypt is self-sufficient in rice, but due to a high population growth rate, the ongoing process of climatic changes and the decline of Egypt's share of the river Nile water, rice production may decline to levels insufficient to maintain self-sufficiency. The development of water stress-tolerant genotypes that maintain excellent yield under drought is a priority area of rice study for sustainable rice production [7].
Hybrid improvement is an expensive and time-consuming procedure. Identifying the parental combinations that produce superior offspring is critical, and one of the extra expensive aspects of the development of excellent hybrids. Previous studies that were carried out on significant heterosis were reported for plant height and grain yield/plant [8]. Other traits, namely panicle length, number of panicles/plant, sterility% and 100-grain weight, were examined [9]. In self-pollinated crops such as rice, the better abrades are to develop a true-breeding homogeneous population with superior agronomic and other desirable traits. The accomplishment of these objectives would depend on the suitable choices of the parental material, nature of gene action controlling characters under consideration and rational choice of breeding method for bringing out quick and maximum genetic improvement. This would imply the basic knowledge of the genetic behavior of the characters under improvement, which is a pre-request for the breeder to manipulate the breeding material to isolate superior lines [10,11].
Diallel analysis has been used in recent years by many breeders to evaluate parental materials before taking any decision regarding the style of breeding system to be used in this concern [10]. Knowledge of the genetic diversity and associations between rice genotypes is necessary for identifying promising combinations for the exploitation of heterosis and the establishment of heterotic collections for use as source resources in rice breeding programs [12][13][14][15]. Increasing evidence indicates that there is a strong relationship between the genetic distance among parents and F 1 performance or heterosis for some traits [16][17][18]. Coefficients of correlation among microsatellite-based genetic distance and heterosis mainly for grain yield have ranged from not significant to significantly strong. However, there is still a lack of knowledge about the association between genetic distance and cross success in rice germplasm; additionally, the factors that contribute to low correlations must be clarified.
The aims of the present study are to (i) assess the heterosis of 28 hybrids under drought stress, (ii) investigate the relationship among microsatellite-based genetic distance and heterosis effects of agronomic trait and yield efficiency and (iii) identify the best cross combinations that can be used for further objectives in rice breeding programs. These research goals will provide reliable information that will aid rice breeders in detecting promising genotypes and verifying whether hybrid prediction based on microsatellite genetic distance is sufficient.

Informativeness of Microsatellite Markers
In total, 11 microsatellite markers were tested for their ability to generate microsatellite banding patterns from DNA corresponding and evaluate the genetic diversity of 8 Egyptian and exotic rice varieties. All microsatellite markers utilized in the current investigation produced polymorphic fragments, producing a polymorphism rate of 100% among the 6 varieties from different origins (Table 1). A total of 39 alleles were identified. The alleles  number per locus varied from 2 for locus RM 162 with a size range of 67-295 bp to 4 for  RM 1, RM 19, RM 144, RM 271, RM 316, RM 338, RM 452 with a mean number of 3.54 alleles  per locus (Table 1).

Gene Diversity
Polymorphic information content (PIC), a measure of gene diversity, for the 11 microsatellites ranged from 0.456 for RM 338 to 0.827 for RM 433 with a mean of 0.658 (Table 1). The PIC value mean across 8 rice genotypes was 0.658, suggesting that the markers were highly informative (PIC > 0.5).

Parental Performance
The performance of heterotic effects can be evaluated in three ways: relative to both better parent (BP) values and mid-parent (MP) values (relative heterosis), or commercial check for yield and its component traits. These effects were examined in all studied attributes, but the level of heterosis indicated a difference from trait to trait and from the cross to cross. Significant and highly significant values of heterosis would be of interest in all traits under investigation from the rice breeder's point of view; in the positive direction values for the traits, i.e., number of panicles per plant, panicle length, panicle weight, filled grains per panicle, 100-grain weight and grain yield per plant. On the contrary, the negative direction values were desirable heterosis towards the traits, i.e., plant height and sterility percentage. In the current analysis, heterosis over BP and MP was computed for eight agronomic, grain yields and its components traits. The data showed a broad range of heterotic patterns for all the attributes (Tables 2 and 3). Heterosis for agronomic, grain yield along with its components traits is a very important consideration in modern plant breeding.

Heterosis over Mid-Parent (MP)
Heterosis effect evaluated with MP values for all agronomic, yield and its component studied traits (Tables 2 and 3). Concerning plant height, 16 crosses out of 28 exhibited significant and highly significant heterosis over the mid-parents (HMP) and 14 of them gave undesirable positive values. The other two crosses were found to be highly significant in the negative direction of heterotic values, namely Sakha 102 × Wab880sg33 (−4.30%) and Sakha 101× Wab880sg33 (−6.25%) ( Table 2). Accordingly, these two crosses could be applied in the hybrid breeding program to enhance plant height. Medium or short plant height is useful to avoid plant losses due to lodging caused by heavy winds. For this reason, negative heterosis for plant height is desirable.

Heterosis over Better Parent (BP)
Heterosis effect was evaluated to BP values for all agronomic, yield and its component studied attributes (Tables 2 and 3). For plant height, 11 crosses exhibited significant HBP in the negative direction of heterotic values. The highest estimates of heterotic effects for plant height were detected for crosses Sakha 102 × Wab880sg33 (−9.86%), Sakha 101× Wab880sg33 (−8.05%) and Giza 179 × IET 1444 (−6.69%) ( Table 2). Accordingly, these hybrids could be applied in the hybrid breeding program to enhance plant height.

Hybrid Prediction Based on Microsatellite Genetic Distance
GD was determined using Nei's similarity coefficient. The relationships between genetic distance (GD)-based microsatellite marker-based with MP and BP heterosis effects for all attributes are shown in Table 4. The associations of microsatellite marker-based genetic distances with MP heterosis (MPH) were positively and significantly associated with sterility percentage (r = 0.390 *, p < 0.05) under drought stress. These results indicate that the molecular markers might be valuable for predicting the high sterility percentage. However, BP heterosis was positively and significantly associated with sterility percentage (r = 0.352 *, p < 0.05) and grain yield per plant (r = 0.345 *, p < 0.05) under drought stress. These associations indicate that high grain yield and low sterility of rice crosses can be expected from microsatellite marker-defined distances of the parents, even though the association values were not very high.

Discussion
A varied range of allelic variations was detected for each microsatellite locus. A comparable pattern of allelic difference was also identified at other loci. A different number of alleles has been found in rice utilizing microsatellite markers. Brondani et al. [19] stated a mean of 5.2 in 30 elite upland rice genotypes. Giarrocco et al. [20] detected a mean of 8.4 alleles per marker from 26 microsatellite markers in 69 accessions from Argentine. Cheng et al. [21] detected a mean number of 11.5 alleles in rice accession from 40 various origins in Africa, Asia, Europe, South America and Oceania. Cui et al. [22] identified an average of 12.5 alleles in 122 south Asia collections. Ghaley et al. [23] reported an average of 7.7 alleles per marker from 27 microsatellite markers in 352 Bhutan landraces. Das et al. [24] found an average of 7.9 alleles per marker from 23 rice microsatellites in 91 rice accessions. The value obtained here was lower than most earlier results, but it was comparable with results obtained by Meti et al. [25], who found 2.08 alleles per locus in rice.
Nevertheless, these results confirmed the conclusion of Salem and El-Zanaty [26] that the small number of markers is adequate to differentiate strongly related genotypes and carry out phylogenetic studies; consequently, selection of genotypes for the highest genetic diversity might be possible. Rice microsatellite markers revealed a mean of Polymorphic information content (PIC) value 0.658, which proves rice microsatellite markers are greatly informative. Botstein et al. [27] investigated that PIC value > 0.5 is counted as being a highly informative marker while 0.5 > PIC > 0.25 is just an informative marker, while PIC ≤ 0.25 is a slightly informative marker. Gene diversity achieved in the current study was equivalent to former findings on the genetic diversity of rice using a microsatellite study. Giarrocco et al. [20] found an average PIC value of 0.69 in 68 accessions from Argentina. Ghaley et al. [23] reported an average PIC value of 0.61 in 352 Bhutan landraces. Das et al. [24] described an average PIC value of 0.746 in 91 rice accessions. This shows that the markers were highly informative.
Predicting heterosis is practiced since heterogeneity is a dynamic phenomenon. In this research, DNA marker-dependent diversity analyses are utilized to predict heterogeneity by associating the coefficient of molecular polymorphism based on hybrid microsatellite markers with the combination of parental genotype and hybrid ability, high average morphological trait values, heterogeneity analysis on the combination of parental genotype and hybrid ability. For a higher yield/heterosis ratio, some workers stressed differential parents as heterosis is calculated in terms of rice yield. Based on several parameters, various genotypes are chosen, including pedigree, morphological, physiological, biological and DNA markers (SSRs, SNPs or ESTs) [28].
In the current research, a total of eight morphological traits were utilized to analyze the diversity of eight rice genotypes and were grouped into four groups. It has been described that genotype within a group with a high degree of variation will generate more appropriate breeding material to maximize genetic progression concerning the crop itself, provided there is a suitable complement [29]. The effectiveness of plant breeding has improved dramatically due to shortcomings in the study of diversity based on morphological data and developments in molecular techniques [30]. Microsatellites are the most accurate, functional and highly effective molecular markers for detecting polymorphisms [31]. Heterosis, however, is calculated primarily by the yield of rice. Therefore, for predicting heterosis, the use of intragenic markers is more effective. In this analysis, 11 microsatellite pairs were polymorphic.
This suggests that the study of molecular diversity is more accurate compared to the analysis of morphological diversity. Microsatellites are unaffected by the environment and DNA transcription regions, and have a higher transferability rate across species and regions that are more preserved in productivity-related genes. However, a heterosis analysis showed a maximum heterosis level in Sakha 103 × Sakha 104 and GZ7576-10-3-2-1 × Giza 179 hybrids, which gave 29.32 and 22.57% heterosis, respectively. This is because heterosis is a complex phenomenon and other factors may involve a limited range of genetic variation in the genomic regions examined among the selected parents with microsatellite markers, use of fewer markers, poor correlation with heterogeneity between marker and genes/QTLs, and interaction between genes/QTLs and error in the choice of experimental characteristics, materials and experimentation [28].
In this analysis, the highest heterotic effects for grain yield per plant, the typical heterosis of all 28 F 1 hybrids ranged from 22.57% (GZ7576-10-3-2-1 × Giza 179) to 29.32% (Sakha 103 × Sakha 104). Positive values for parental combinations and substantial sterility percentage (r = 0.352 *, p < 0.05) and grain yield per plant (r = 0.345 *, p < 0.05) were shown in the correlation analysis between microsatellite marker-based genetic distances of parents with better-parent heterosis for grain yield and its components traits. Microsatellite-based genetic distance failed to be a dependable predictor in this analysis, as there was no meaningful association between microsatellite-based-GD and heterosis for most studied traits.
In earlier research, this marker failed to predict hybrid efficiency and heterosis, but in certain cases, the microsatellite marker may be a better predictor [32]. The utility of specific microsatellite determinants was also verified by Joshi et al. [33], who showed a significant association with hybrid efficiency. The outcome of the current study also validated the idea of main DNA markers to predict the heterosis suggested by Cho et al. [34]. Zhang et al. [28] predicted heterosis based on productivity-associated functional genes. Several variables, including germplasm type, environmental impact, genome coverage and related marker to QTL, have been proposed in former studies to explain unstable correlation coefficients.

Plant Materials
This research was conducted at the research farm of the Rice Research and Training Center (RRTC), Sakha, Kafr El-Sheikh, Egypt, during two rice-growing seasons (2014 and 2015). Eight rice genotypes were utilized in this study, i.e., GZ7576-10-3-2-1, Sakha 102, Sakha 103, Giza 179, Sakha 101 and Sakha 104 as local genotypes from Agriculture Research Center (ARC), Giza, Egypt, and two exotic genotypes namely, i.e., IET1444 and WAB880-SG33, obtained from the genetic stock of Rice Department, Field Crops Research Institute, Agriculture Research Center (ARC), Egypt. Name, pedigree, origin and main characteristics of the eight parent genotypes are illustrated in (Table 5).

Field Experiments
Eight rice genotypes were used in this research. They were grown in the field at fifteenday intervals to defeat the difference in heading date between them (during rice-growing season 2014). Thirty days after sowing, seedlings of each genotype were individually transplanted into the permanent field in three rows, (5 m long) and 20 × 20 cm apart among plants and rows. A half diallel cross was carried out between these eight rice parents in the first growing season (without reciprocals) Griffing method 2 [35]. The bulk emasculation method was practiced by utilizing the hot water procedure according to Jodon [36].

Drought Test
A total of 28 hybrids were made and the F 1 seeds were cultivated as F 1 plants in the third week of May (during the rice-growing season of 2015), and plants were transplanted individually after 30 days and adopted a spacing of 20 × 20 cm. The experimental plots were arranged in a randomized complete block design (RCBD), with three replications; each replicate consisted of one row for each parent, and three rows for F 1 . Every row was 5 m lengthy and included 25 single plants. A field experiment was irrigated only every 12 days just flushing (water deficit) (parent and hybrid genotypes cultivated), and flash irrigation was used without standing water (two weeks from transplanting till harvest). All suggested farming procedures were applied as usual for the normal rice field.

Phenotypic Data Collection
The data were scored on a single plant base for parents and F 1 s . Ten random plants except border plants were chosen for phenotyping of studied characteristics according to International Rice Research Institute (IRRI) [37]. The following characters were phenotyped for parents and hybrids, i.e., plant height (cm), number of panicles per plant, panicle length (cm), panicle weight (g), filled grains per panicle, sterility percentage, 100-grain weight (g) and grain yield per plant (g).

Genomic DNA Extraction
Total genomic DNA was isolated from leaf tissue for every genotype. Fresh leaves from 8-week-old plants were cut as tissue samples for DNA isolation. DNA was extracted from these genotypes according to Salem [7].

Microsatellite Markers
In total, eleven rice microsatellite (RM) markers were selected from the Rice Genes Database (http://gramene.org (accessed on 23 June 2014)) for genotyping based on their known chromosomal location to give a uniform coverage of the rice genome. The selected RM included one RM locus from each rice chromosome. Microsatellite name, chromosomal location, sequence, motif, annealing temperature (Tm • C) and fragment size (bp) were presented in Table 6. Every 20 µL PCR mixture contained 2 µL of 50 ng µL −1 genomic DNA, 0.2 µL of 0.2 mM dNTPs, 1.6 µL of 1.5 mM MgCl 2 , 2 µL of 0.250 mM primers, 4 µL of 1× Taq buffer, 0.2 µL of 1 U/µL Taq polymerase and 10 µL double distilled water. The following thermal profile was used: 3 min of pre-denaturation at 94 • C, 1 min of denaturation at 94 • C, 1 min of annealing (annealing temperature is relative to the SSR primer, see annealing temperature in Table 6), 1 min of elongation at 72 • C, 45 cycles of repeated steps from denaturation to elongation and 8 min of final elongation at 72 • C. The samples were moved to 4 • C or −20 • C after the final extension (polymerization) step if they were not going to be used right away. After electrophoresis, the amplification products were isolated on 1.5% agarose gels. The bands were detected using a UV-transilluminator and photographed using a gel documentation system [7].

Microsatellite Marker Analysis
As stated by Akagi et al. [38] and Temnykh et al. [39], microsatellite amplification and fragment detection for rice microsatellite markers were carried out.

Cluster Analysis
Each locus of rice microsatellite (RM) bands was manually scored and analyzed as multi-allelic data. The association between eight parental lines was reconstructed using a pairwise matrix of Nei's DA genetic distance (GD), Nei et al. [42] based on the unweighted pair-group process for the arithmetic mean (UPGMA) cluster assessment [43].

Correlation Coefficient
Each locus of rice microsatellite (RM) bands was manually scored and analyzed as multi-allelic data. To assess the association between GD with MPH and BPH for studied attributes and yields, the Pearson linear correlation coefficients were calculated [44]. The next statistical software was applied in the result assessment: (1) SPSS (statistical package for social sciences) version 15 [45] for LSD's analysis and Pearson correlation coefficient and (2) UPGMA cluster analysis by NTSYS-pc version 2.1 software [46].

Conclusions
Results showed that the best cross combination for grain yield was observed in Sakha 103 × Sakha 104 and GZ7576-10-3-2-1 × Giza 179, each yielding 29.32-22.57% heterosis. As a result of the poor correlation, microsatellite-based genetic distance failed to predict hybrid results and heterosis. The bad prediction was thought to be caused by a lack of dominance impact and a comparatively high genetic gap between parental lines. This study showed that for the prediction of heterosis for yield and yield-contributing characteristics in rice, a high correlation between significant microsatellite marker-based genetic distances of parents with MP and BP heterosis can be used. Data Availability Statement: All data generated or analyzed during this study already exist in this published article.