Novel QTL Associated with Shoot Branching Identified in Doubled Haploid Rice (Oryza sativa L.) under Low Nitrogen Cultivation

Shoot branching is considered as an important trait for the architecture of plants and contributes to their growth and productivity. In cereal crops, such as rice, shoot branching is controlled by many factors, including phytohormones signaling networks, operating either in synergy or antagonizing each other. In rice, shoot branching indicates the ability to produce more tillers that are essential for achieving high productivity and yield potential. In the present study, we evaluated the growth and development, and yield components of a doubled haploid population derived from a cross between 93-11 (P1, indica) and Milyang352 (P2, japonica), grown under normal nitrogen and low nitrogen cultivation open field conditions. The results of the phenotypic evaluation indicated that parental lines 93-11 (P1, a high tillering indica cultivar) and Milyang352 (P2, a low tillering japonica cultivar) showed distinctive phenotypic responses, also reflected in their derived population. In addition, the linkage mapping and quantitative trait locus (QTL) analysis detected three QTLs associated with tiller number on chromosome 2 (qTNN2-1, 130 cM, logarithm of the odds (LOD) 4.14, PVE 14.5%; and qTNL2-1, 134 cM, LOD: 6.05, PVE: 20.5%) and chromosome 4 (qTN4-1, 134 cM, LOD 3.92, PVE 14.5%), with qTNL2-1 having the highest phenotypic variation explained, and the only QTL associated with tiller number under low nitrogen cultivation conditions, using Kompetitive Allele-Specific PCR (KASP) and Fluidigm markers. The additive effect (1.81) of qTNL2-1 indicates that the allele from 93-11 (P1) contributed to the observed phenotypic variation for tiller number under low nitrogen cultivation. The breakthrough is that the majority of the candidate genes harbored by the QTLs qTNL2-1 and qTNN4-1 (here associated with the control of shoot branching under low and normal nitrogen cultivation, respectively), were also proposed to be involved in plant stress signaling or response mechanisms, with regard to their annotations and previous reports. Therefore, put together, these results would suggest that a possible crosstalk exists between the control of plant growth and development and the stress response in rice.


Introduction
The global population is increasing at a relatively high growth rate per annum, and is projected to reach about 9.8 billion people by 2050 [1]. The current status of the food and nutrition security in the world question on how to provide sufficient, healthy and nutritious food, at all times to the growing population. To achieve this, several initiatives have been taken to combat malnutrition and food insecurity, and alleviate poverty and eradicate hunger, in line with the United Nations Sustainable Development Goals (SDGs), through research and development, among others.
Since the domestication of rice, this Poaceae is the second most important cereal crop after corn and remains today the only cereal crop solely cultivated for human consumption, in addition to being the staple food crop for more than half of the world's population, despite the increasing preference for various foods and the diversification of diet. According to the Food and Agriculture Organization of the United Nations (FAO) statistics (http://www.fao.org/faostat/en/#data/QC/visualize, accessed on 5 March 2021), nearly 755.5 million metric tons of paddy rice were harvested in 2019, and around 162.1 million hectares were occupied by rice (upland, wetland, and irrigated) fields across the globe.
To achieve optimum plant growth and development, and realize their potential in terms of productivity and quality, plant crops require nitrogen (N), among other essential macronutrients [2]. The role of nitrogen in the metabolism of plants is important in promoting seed germination, cell division, plant growth and development, formation of amino acids and proteins, and plant productivity [3][4][5][6]. Rice cultivation requires abundant use of nitrogenous fertilizers [7], particularly in high yielding varieties [8]. However, excessive application of nitrogenous fertilizers has been associated with the emission of greenhouse gases (GHG), such as methane (CH 4 ) and nitrous oxide (N 2 O) [9][10][11][12].
In the recent years, the trend in plant breeding revealed a growing interest in developing rice varieties with an optimized and balanced productivity and quality, while exhibiting a high nitrogen use efficiency (NUE), which may help reduce the use of large amounts of fertilizers, particularly the N-rich fertilizers or a combined mineral-organic fertilization regime [13][14][15][16], as part the efforts to reducing GHG emission from agricultural activities, and to promoting the use of ecofriendly agricultural practices. To achieve this, scientists have employed various strategies that include the use of advanced plant breeding techniques and the available and emerging technologies to develop rice varieties with an improved nitrogen use efficiency (NUE), which includes nitrogen uptake, transport, assimilation, and remobilization. It has been shown that an increase in nitrogen (N) fertilizers is associated with a large decrease in NUE [17]. Therefore, there is a pressing need to improve agricultural practices in order to contribute to the reduction of GHG emissions from irrigated, rainfed, and flood-prone rice ecosystems, to predict the future emissions and develop rice technologies that will reduce GHG emissions, and yet will meet the required increase in rice production [18].
In rice, shoot branching is considered as one of the major agronomic traits with a high potential to contribute to the yield and productivity of rice [19,20]. It is well established that phytohormones play fundamental roles in the control of shoot branching and bud outgrowth in plants, operating in a synergetic or antagonistic manner [21][22][23][24][25], in addition to their roles in seed dormancy and germination, plant growth and development, flowering and organogenesis, seed formation and maturation, fruit ripening, and signaling during abiotic or biotic stress occurrence [26]. Several studies have proposed a number of quantitative trait loci (QTLs) controlling tiller number in rice, and mapped to almost all rice chromosomes (10)(11) using different sorts of mapping populations, including doubled haploid (DH) lines [27]. Many genes have been identified; however, only a few genes have been cloned and functionally characterized [28][29][30][31][32][33][34][35][36][37]. A mapping population derived from a cross between indica and japonica would offer the best cross combination to investigate QTLs associated with rice plant architecture, due to the morphological difference between indica and japonica subspecies.
The use of fixed homozygous populations, such as doubled haploid (DH) [38,39] and recombinant inbred lines (RILs) [40] have been favored when investigating QTLs for various agronomic traits in rice. The indica and japonica rice subspecies, are the most cultivated rice, and are said to be independently domesticated from the wild ancestor [41][42][43].
Therefore, from this perspective, this study performed a linkage mapping analysis on a doubled haploid (DH) developed through anther culture of a cross between 93-11 (P1) and Milyang352 (P2), typical indica and japonica rice cultivars, respectively, grown under normal nitrogen application and low nitrogen-level field conditions, to investigate putative QTLs controlling rice tiller number as well as agronomic traits contributing to the yield of rice using Kompetitive Allele-Specific PCR (KASP) and Fluidigm markers. Additionally, this study identified candidate genes associated with tiller number in rice under low nitrogen cultivation, with interesting annotated molecular functions, and suggested to have the potential to regulate shoot branching and panicle number in rice. The experiments were initially conducted during the rice cropping season (June-November) of 2019 and repeated in 2020 to validate the phenotypes prior to performing QTL analysis. One set of the mapping population and parental lines were transplanted on a plot with a normal nitrogen cultivation regime (90 kg ha −1 N, 45 kg ha −1 P 2 O 5 , and 57 kg ha −1 K 2 O), and the other on a plot with a low nitrogen level (45 kg ha −1 N, 22.5 kg ha −1 P 2 O 5 , and 28.5 kg ha −1 K 2 O) maintained over 15 years. Each rice line had at total of 100 seedlings transplanted in four rows, with 25 plants per row and the spacing between and within the lines of 30 cm × 15 cm, respectively. Parental lines were planted after every 20 DH lines, starting from the initial rows. The weather parameters and conditions during the experimental period can be found in Figure S1.

Materials and
The soil chemical analysis was performed following the method described by Park et al. [44] and the basic soil chemical properties are given in Table 1. The phenotypic measurements of the studied traits, which included plant height (PH), tiller number (TN), culm length (CL), panicle length (PL), number of panicles per plant (PN), root length (RL), shoot dry weight (SDW), and root dry weight (RDW), were done on ten individual plants in triplicate randomly pooled from the inside rows, excluding the border rows to avoid the border effects on the traits studied and competition between lines. The plant height and culm length were measured using a 1-m long wooden stick ruler at maximum tillering stage. The panicle length was measured with a regular ruler soon after harvesting. The number of tillers per plant was counted at maximum tillering stage. SDW was measured from an oven-dried (at about 45 • C overnight) separated shoot from the roots at the collar, cooled down, and weighed on a precision scale (CASKOREA Co. Ltd, MW2-6004, 0.02-60 g, Seongnam-si, Korea). Similarly, RDW was measured after washing off any loose soil particles, followed by blotting of samples with a soft paper towel to remove any free surface moisture. Samples were oven-dried, weighed, and the results expressed in grams.

Frequency Distribution, Quantile-Quantile Plots, Principal Component Analysis, and Correlation Analysis
The frequency distributions of each phenotypic trait under normal and low nitrogen cultivation and the correlation analysis were performed using GraphPad Prism software (version 7.00, © 1992-2016 GraphPad, San Diego, CA, USA). The normality of the distribution was assessed using the Shapiro-Wilk W-statistic for the test of normality in IciMapping (version 4.1.0.0, 2016, Chinese Academy of Agriculture Sciences, Beijing, China). In addition, the Quantile-Quantile (Q-Q) plots and the pairwise kinship matrix were generated using the GAPIT function: my_GAPIT <-GAPIT(Y = myY, G = myG, Model.selection = TRUE, SNP.MAF = 0.05) in RStudio (version 1.2.5042, © 2009-2020 RStudio, Inc., Boston, MA, USA). The cluster analysis was performed using the R packages ggplot2, tidyverse, cluster, and fviz_cluster R function. The principal component analysis (PCA) was performed using the R package ggplot2 and functions myPr, plot, and biplot [45,46].

Genotyping and Molecular Markers Analysis
For the molecular markers analysis, Kompetitive Allele-Specific PCR (KASP) markers and Fluidigm markers were employed. KASP markers were amplified from genomic DNA and the allelic discrimination was performed using the Nexar system (LGC, Biosearch Technologies, Hoddesdon, UK), while the Fluidigm markers for SNP genotypes were determined using the BioMark TM DH system (Fluidigm, San Francisco, CA, USA) and 96.96 Dynamic Array IFC (96.96 IFC) chip as described earlier [47].

Construction of Linkage Maps and QTL Analysis
A total of 240 markers, including KASP markers earlier reported [48] to be specific for detecting polymorphism between japonica subspecies and Fluidigm markers [49] specific for detecting polymorphism between indica and japonica subspecies, and the phenotypic data of the mapping population consisting of 117 doubled haploid (DH) rice lines and their parental lines (93-11, P1 and Milyang352, P2) were used to perform QTL analysis in order to identify putative QTLs associated with plant growth, tiller number, panicle length, panicle number, root length, and shoot and root dry weights in rice, under low nitrogen cultivation. The genotype and phenotype raw data were initially formatted in RStudio using the fread, pheno.raw, geno.raw, and cbind functions. The QTL analysis was performed and the linkage maps constructed with IciMapping software v.4.1.0.0, for a biparental population using position mapping and Kosambi mapping functions [50]. The permutation (1000 times) parameters, which explains the probability for detecting statistically significant QTLs associated with the target traits, were selected.
The rice genome annotation database publicly available online browser (http://rice. plantbiology.msu.edu/, accessed on 22 February 2021) was used to identify candidate genes pooled from the detected QTL qTNL2-1 (684.4 kbp, Chr2:26706424..27390847) region flanked by KASP markers KJ02_057 (left, closest marker) and id2012773 (right marker) on chromosome 2, associated to the control of the number of tillers under low nitrogen cultivation.

Differential Growth Patterns between Doubled Haploid Lines under Normal and Low Nitrogen Conditions
We evaluated the phenotypic response of the mapping population and their parental lines under normal and low nitrogen cultivation conditions. The results revealed a normal distribution for plant height ( Figure 1A,B, Table S3, under normal and low nitrogen cultivation), number of panicles ( Figure 1I, Table S3, under normal nitrogen cultivation), and root length ( Figure 1K,L, Table S3, under normal and low nitrogen cultivation). However, for culm length ( Figure 1F, under low nitrogen cultivation) a negative skewness was observed. In contrast, a positive skewness was recorded for tiller number ( Figure 1C,D, Table S3), culm length ( Figure 1E, Table S3, under normal nitrogen cultivation), panicle length ( Figure 1G,H, Table S3, under normal and low nitrogen culti-vation), panicle number ( Figure 1J, Table S3, under low nitrogen cultivation), shoot dry weight ( Figure 1M,N, Table S3, under both normal and low nitrogen cultivation), and root dry weight ( Figure 1O,P, Table S3, under normal and low nitrogen cultivation). Similarly, the Quantile-Quantile (Q-Q) plots (observed quantiles plotted against theoretical quantile or z-scores) reflected the same distribution patterns of the traits observed earlier ( Figure S2A-P).
Our data indicate that about 8.5% and 91.5% of the DH lines had relatively short plants (P1-like phenotype) and tall plants (P2-like phenotype), respectively, under normal nitrogen conditions. Whereas, under low nitrogen cultivation, nearly 39.3% and 60.7% had P1-like (relatively short plants) and P2-like (relatively tall plants) plant height, respectively ( Figure 2B). In the same way, about 83.8% of the mapping population exhibited an increase in culm length (P1-like, >58 cm) against 14.5% that showed a P2like culm length (≤58 cm) under normal nitrogen cultivation. However, when the DH lines were grown under low nitrogen conditions, nearly 82.1% and 17.9% showed P1-like (>59 cm) and P2-like (≤59 cm) culm length, respectively ( Figure 2C). Additionally, of the 117 DH lines cultivated under normal nitrogen level, about 37.6% and 62.4% had long panicles (P1-like) and short panicles (P2-like), respectively ( Figure 2D), while under low nitrogen cultivation, 39.3% and 60.7% had long panicles (P1-like) and short panicles (P2like), respectively. Furthermore, from the set of the mapping population grown under normal range of nitrogen level, we observed that about 93.2% exhibited an increase in number of panicles per plant (P1-like), and 6.8% had P2-like number of panicles. Similarly, when grown under low nitrogen cultivation conditions, 98.3% and 1.7% of the DH lines recorded an increase in number of panicles per plant, P1-like and P2-like, respectively ( Figure 2E).
We were also interested to see the changes in the root growth and the biomass dry weight of the mapping population under low nitrogen cultivation compared to the normal range nitrogen supply. Therefore, we measured the root length, shoot dry weight, and root dry weight. The results in Figure 2F indicate that only 3.4% of the DH lines had P1-like root growth pattern (relatively long roots), while nearly 96.6% of the same population grown under normal nitrogen cultivation exhibited a P2-like root growth pattern (relatively short roots). When grown under low nitrogen level, we observed that only 7.7% of the DH lines showed a P1-like root growth pattern against 92.3% with a P2-like root growth pattern ( Figure 2F). Regarding the shoot dry weight of the mapping population cultivated under normal nitrogen conditions, our data indicate that about 96.6% and 3.4% had P1-like (high) and P2-like (low) shoot dry weight (SDW), while about 87.2% and 12.8% recorded a high SDW (P1-like) and low SDW (P2-like) under low nitrogen cultivation conditions ( Figure 2G). As for root dry weight (RDW), about 81.2% and 14.5% recorded a high (P1like) and low (P2-like) RDW under normal nitrogen cultivation, respectively; whereas, about 39.3% and 60.7% had a P1-like and P2-like RDW under low nitrogen cultivation ( Figure 2H).
In the same perspective, the root-shoot ratio (R:S) results indicate that about 46.2% and 38.5% of the DH lines recorded a high R:S (P1-like, R:S > 0.42 and >0.35 based on length ( Figure 2I) and dry weight ( Figure 2J), respectively), while 61.5% and 53.8% (P2-like, R:S ≤ 0.42 and ≤0.35 based on length and dry weight, respectively) showed a low R:S (P2-like). Meanwhile, under low nitrogen conditions, about 19.7% and 62.4% (R:S based on length and dry weight, respectively) had a P1-like R:S (>0.35 and >0.37, based on length and dry weight, respectively) against 37.6% and 59% (R:S ≤ 0.37 and ≤0.35, based on length and dry weight, respectively).    In addition, the phenotypic response of the parental lines (Figure 2A-L) indicates that both 93-11 (P1) and Milyang352 (P2) experienced a reduction in almost all the traits when grown under low nitrogen cultivation, compared to the normal nitrogen supply conditions, except a slight increase in culm length recorded in 93-11 (1.1% increase) and Milyang352 (0.4% increase), and a significant increase in number of panicles per plant observed in 93-11 (P1) against 16.6% reduction recorded in Milyang352 (P2) under low nitrogen cultivation.

Relatedness, Principal Component Analysis, and Correlation between Traits
In Figure 3A, our data show the KASP marker-based kinship matrix, also known as the co-ancestry or half relatedness, revealing the distribution of coefficients of co-ancestry of the mapping population. Additionally, the DH lines were grouped into two distinct clusters with regard to their number of tillers pattern (P1-type or P2-type) under low nitrogen cultivation ( Figure 3B Initially, we were interested in investigating a possible relationship between tiller number and other quantitative traits of interest recorded in the mapping population. For this reason, we performed a linear regression analysis. The results shown in Figure 3C,D suggest a strong positive correlation between tiller number and panicle number (R 2 = 0.603*** and R 2 = 0.560***, under normal and low nitrogen cultivation, respectively). In contrast, plant height showed a negative correlation with tiller number (R 2 = −0.465*** and R 2 = −0.279***, under normal and low nitrogen cultivation, respectively). However, culm length (R 2 = 0.085 ns and R 2 = 0.198 ns ), panicle length (R 2 = −0.199* and R 2 = −0.222 ns ), root length (R 2 = 0.046 ns and R 2 = 0.003 ns ), and root dry weight (R 2 = 0.173 ns and R 2 = 0.288**) showed a nonexistent or very low correlation with tiller number (Tables S1 and S2). In the same perspective, root length and root dry weight had a positive but weak correlation under normal nitrogen cultivation (R 2 = 0.209*, Table S1), while showing a stronger positive correlation under low nitrogen cultivation (R 2 = 0.440***, Table S2). Likewise, shoot dry weight and root dry weight are predicted to have a strong positive correlation. Furthermore, it was interesting to see that shoot dry weight (SDW) of the mapping population showed a strong positive correlation with tiller number (R 2 = 0.583***, Figure 3F, Table S2) under low nitrogen cultivation conditions, while under normal nitrogen supply, a weak positive correlation was observed (R 2 = 0.317***, Figure 3E, Table S1). It is then thought that a unit increase in the number of tillers would cause an increase in the shoot dry weight of rice by the same measure. Tables S1 and S2 provide details information about the Person r correlation between traits. Data in these two tables also revealed that the correlation level between tiller number and other traits differ depending on each nitrogen condition.

Quantitative Trait Loci (QTLs) Associated with Plant Growth and Shoot Branching under Normal and Low Nitrogen Cultivation Were Mapped to Different Chromosomes
We used 240 KASP makers and Fluidigm markers, and the phenotype data of 117 DH lines and parental lines, cultivated under normal nitrogen and low nitrogen levels, to perform linkage mapping and QTL analysis. The results revealed that, in total, 28 QTLs (all traits considered) associated with plant growth and shoot branching were mapped to all rice chromosomes, except chromosomes 6 and 11 (Table 2, Figure 4A     In the case of tiller number, two QTLs (qTNN2-1, LOD: 4.14; and qTNN4-1, LOD: 3.92) were detected in DH lines cultivated under normal nitrogen conditions, accounted for about 14.5% of the phenotypic variance explained (PVE) (Table 2, Figure 4B,D). However, solely one QTL (qTNL2-1, LOD:6.05) mapped to chromosome 2, flanked by KJ02_057 and id2012773, was associated with tiller number in DH lines grown under low nitrogen cultivation conditions, and accounted for about 20.5% of the PVE (Table 2, Figure 4B). The recorded additive effect (1.81) indicates that the allele from 93-11 (P1, indica) contributed to the observed phenotypic variation for tiller number under low nitrogen cultivation.
In the same way, a set of five QTLs (qPLN2-1, LOD: 4.81; qPLN3-1, LOD: 3.92; qPLN3-2, LOD: 8.84; qPLN7-1, LOD: 5.94; and qPLN8-1, LOD: 8.06) associated with panicle length were only detected in the mapping population cultivated under normal nitrogen conditions (Table 2, Figure 4B,C,F,G). The additive effects recorded for both normal and low nitrogen cultivation conditions show that 93-11 and Milyang352, P1 and P2, respectively, contributed to the observed panicle length phenotypic variation. For panicle number trait, a total of four QTLs were detected, of which number three (qPNN2-1, LOD: 3.49; qPNN4-1, LOD: 9.63; and qPNN8-1, LOD: 4.34) were associated with normal nitrogen cultivation regime (Table 2, Figure 4B,D,G), and one QTL (qPNL4-1, LOD: 2.95) with low nitrogen cultivation (Table 2, Figure 4D). The additive effects indicate that, for both normal and low nitrogen cultivation conditions, the alleles from the indica parental line (93-11, P1) contributed to the observed phenotypic variation for panicle number (Table 3).  A successful establishment of plants is expected when they develop a robust rooting system for efficient nutrients and water acquisition. Our results indicate that a single QTL (qRLL4-1, LOD:3.41) associated with root growth under low nitrogen cultivation was detected, and mapped to chromosome 4, and flanked by KJ04_053 and KJ04_057 (closest marker) KASP markers. The additive effect (−1.1) suggests that the allele from Milyang352 (P2, japonica) contributed to the observed phenotypic variance explained (PVE, 11.98%) (Table 2, Figure 4D). Biomass dry weight is affected by many factors, including the acquisition of mineral nutrients and crop productivity [51]. It was also proposed that there is an interdependent relationship between the shoot and roots [52,53]. The QTL analysis results showed that one QTL associated with shoot dry weight (qSDWN5-1, LOD: 3.74) (Table 2, Figure 4E), and two QTLs (qRDW1-1, LOD: 4.06; and qRDW4-1, LOD: 3.22) (Table 2, Figure 4A,D). Although many QTLs associated with different traits analyzed were identified, we opted to unveil candidate genes found within qTNL2-1 (684.4 kbp, Chr2:26706424..27390847) essentially due to the fact that this major QTL was the only one associated with the control of shoot branching under low nitrogen cultivation (Table 3), which may help reduce excessive nitrogen-rich fertilizers application in rice cultivation, and contribute to reducing methane (NH 4 ) emissions from paddy fields. An additional list of candidate genes with interesting annotated functions, pooled from the qTNN4-1 region (1.193 Mbp) and associated with tiller number under normal nitrogen cultivation conditions, is attached as Table S4.
Shoot branching has been shown to be regulated at different levels of the plant cellular metabolism. It is well established that phytohormones signaling pathways are key regulators of shoot branching in plants, operating in synergetic or antagonistic manner. Among the identified candidate genes (Table 3), we found two genes encoding GRAS (GIBBERELILIN-INSENSITIVE (GAI), REPRESSOR of GA1-3 (RGA), and SCARECROW (SCR)) transcription factor (TF) (Os02g44360 and Os2g44370). Proteins in the GRAS domain family are known as key players in gibberellin (GA) signaling, which regulates various aspects of plant growth and development, by inhibiting the proliferation and expansion of cell-mediated plant growth [54]. Another gene found in the same region is OsGPX3, a glutathione peroxidase putative expressed earlier reported to be involved shoot and root development in rice and stress response [55], as well as abscisic acid (ABA) response in rice [56]. Similarly, we could also find many other genes with interesting predicted functions, such as OsDhn1 (Os02g44870), earlier suggested to be involved in abiotic stress response in rice [57,58], OsRALF8 (rapid alkalinization factor, Os02g44940) gene, one of which FERONIA (a receptor-like kinase (RLK) member of CrRLK1L, known as a major plant cell growth modulator in distinct tissues) is known to be a receptor, and previously reported to play a critical role in the regulation of plant defense and development (inhibitory or suppression of growth) [59].
Reports indicated that carbohydrates regulate different aspects of plant growth via the modulation of cell division and expansion. For instance, a gene encoding DUF581 domain was reported to be a generic SnRK1 (sucrose nonfermenting-1 (SNF1)-related protein kinase) interaction module and co-expressed with SnRK1 during plant cell signaling [60]. We also identified a DUF encoding gene OsFBDUF13 (F-box-like domain and domain of unknown function (DUF295) containing protein) within the qTNL2-1 region. In the same way, a gene identified as being involved in carbohydrates metabolic process, among other biological processes, was found (OsHUGT1, Os02g44510). The other set of genes encoding the Aquaporin (OsPIP1 (plasma membrane intrinsic protein 1) or OsRWC1 (water channel protein), Os02g44630), the cytochrome P450 (Os02g44654) [61,62], zinc finger RING-type domain (Os02g44700), OsSID1 or OsIDD4 (Os02g45054) [63], and MYB (Myeloblastosis) transcription factor encoding gene (Os02g4580) was found within the detected QTL, among other genes (Table 3).

Tiller Number Positively Correlated with Shoot Dry Weight, While Shoot and Root Dry Weights Showed Positive Relationship
In rice, tillers represent the branches that develop from the leaf axils at each unelongated node of the main shoot or from other tillers during vegetative growth. It is said that tiller number determines the number panicles per plant [39,64,65], which happens to be an important yield component of rice. The number of tillers is influenced by environmental factors [66], agricultural practices, including nitrogen fertilization [67][68][69][70][71][72], or genetic factors [65,[73][74][75][76][77]. It is also expected that an increase or decrease in tiller number per plant could cause significant changes in the shoot biomass. Here, the results of the linear regression analysis showed that shoot dry weight positively correlated with tiller number (R 2 = 0.583*** and R 2 = 0.317***, under low and normal nitrogen cultivation, respectively) and plant height (R 2 = 0.374*** and R 2 = 0.350***, under low and normal nitrogen cultivation, respectively). Therefore, these findings suggest that a unit increase in tiller number would result in the increase in shoot dry weight by the same measure. In the same way, we recorded that panicle number showed a strong positive correlation with tiller number (R 2 = 0.560*** and R 2 = 0.603***, under low and normal nitrogen cultivation, respectively). It could then be said that rice plants with a high tiller number would eventually develop more panicles, which would result in an increased productivity.
Roots and shoot are said to be positionally and developmentally related in various ways [78,79]. Therefore, the recorded positive correlation between shoot dry weight and root dry weight (R 2 = 0.491*** and R 2 = 0.309***, under low and normal nitrogen cultivation, respectively), coupled with the respective phenotypes of the mapping population concerning SDW and RDW ( Figure 2G,H: 96.6% and 87.2% of the DH lines showed a high SDW and RDW, respectively, similar to 93-11 (P1, indica and high tillering cultivar)), would suggest that a robust rooting system may contribute to achieving an optimum shoot growth and development in rice. In addition, it is thought that the observed difference of the correlation level between tiller number and other traits under normal and low nitrogen conditions would also give insights into the recorded N-effects on the phenotypic traits analyzed, and on the genetic loci difference controlling tiller number.

Quantiative Trait Loci (QTLs) Associated with Shoot Branching under Low Nitrogen Cultivation Conditions Are Identified
Several agronomic traits of rice have been suggested to be controlled by multiple quantitative trait loci (QTLs) [27,80,81]. A number of studies have proposed a large number of putative QTLs controlling tiller number and panicle number in rice. Most of the reported QTLs related to tiller number or panicle number are mapped across the 10-11 chromosomes of rice, using various methods, marker-types on different sorts of rice populations, at various developmental stages [27,39,82]. In this study, of the 28 QTLs identified for all the growth related and productivity related traits, only three were associated with tiller number. Of this number, two (qTNN2-1 and qTNN4-1) were detected in the set of DH lines grown under normal nitrogen conditions, and one (qTNL2-1) in that grown under low nitrogen conditions, were mapped on chromosomes 2 and 4. Meanwhile, four QTLs related to panicle number were detected (Table 3), with qPNL4-1 being the only QTL associated with panicle number identified in DH lines grown under low nitrogen conditions, and mapped on chromosome 4. A study conducted by Liu et al. [82] identified 12 QTLs associated with tiller number mapped on chromosomes 2 (two QTLs: 12.5 cM and 23.1 cM), 3 (three QTLs: 6.0 cM, 17.2 cM, and 20.4 cM) and 6-9 (three QTLs on chromosome 6: 8.5 cM, 8.8 cM, and 28 cM; one on chromosome 7: 24.5 cM; one on chromosome 8: 8.3 cM; one on chromosome 9: 17.0 cM, and one on chromosome 12: 2.6 cM) using rice microsatellites (RM) markers on single segment substitution lines (SSSLs). In a similar perspective, Yan et al. [39] reported 15 QTLs associated with tiller number at different growth stages, and mapped on chromosomes 1-4, 6-8, and 12. In a converse approach, Liu et al. [27] identified 27 QTLs associated with tiller number at different development stages in rice, and mapped on chromosomes 1-12. However, Zhu and his colleagues reported a QTL associated with panicle number in rice, mapped on chromosome 1 (qPN1) [83]. None of the previously reported QTLs associated with tiller number or panicle number overlapped with the identified QTLs by this study. However, some of them are mapped on similar chromosomes but different positions.

QTLs qTNL2-1 and qTN4-1 Harbor Candidate Genes Proposed to Be Involved in the Control of Shoot Branching in Rice under Low and Regular Nitrogen Cultivation
In the perspective of unveiling the identity of candidate genes located within the genetic region covered by the QTL qTNL2-1 associated with the control of shoot branching under low nitrogen cultivation, we used the available maker positions to explore this region. Interestingly, a substantial number of genes were proposed to have transcription factor binding activity. This is the case for two GRAS transcription factor (TF) encoding genes (Os02g44360 and Os2g44370), which have a sequence similarity of about 86.6%. Proteins in the GRAS domain family are known as key players in gibberellin (GA) signaling [84,85], which regulates various aspects of plant growth and development, by inhibiting the proliferation and expansion of cell-mediated plant growth [54]. Similarly, OsHMG1 (Os02g44930), OsSID1 (Os02g45054), also known as OsIDD4 (a Zinc finger C 2 H 2 -type domain containing protein), and MYB TF encoding gene (Os02g45080), are proposed to code for transcription factors.
Another set of interesting genes, such as OsGPX3, was earlier reported to be involved in the development of the roots and shoot in rice, in the stress response mechanisms [55], and abscisic acid (ABA) response in rice [56]; OsDhn1, earlier suggested to be involved in abiotic stress response in rice [57,58]; OsRALF8 [86], previously reported to play a critical role in the regulation of plant defense and development (inhibition or suppression of growth) [59], could play important roles in the control of shoot branching, while maintaining a balanced plant growth and development under low nitrogen cultivation in rice. Nietzsche et al. [60] supported that carbohydrates regulate different aspects of plant growth via the modulation of cell division and expansion. Here, a DUF encoding gene OsFBDUF1 or OsMAIF23 found within the qTNL2-1 region, and OsHUGT1 known as UDP-glucose glycoprotein glucosyltransferase 1 precursor, predicted as being involved in carbohydrates metabolic process would contribute to the observed tiller number phenotype under low nitrogen cultivation.
In the same way, Aquaporin (OsPIP1 or OsRWC1), and OsDhn1 or OsLip9 [57,58], associated with cytochrome P450 (earlier suggested to be involved in auxin signaling by Chaban et al. [87] or encoded by OsMAX1, a strigolactone biosynthesis genes according to Marek et al. [61,88] or encoded by OsSLR1/OsGAI (SLENDER RICE 1 encodes a DELLA protein belonging to GRAS [89]), which modulate gibberellin response in rice [62]), are believed to actively contribute to modulating water movements in and outside the cell to facilitate transport of nutrients, including nitrogen, as well as plant growth and development under low nitrogen conditions. From another perspective, Zou et al. [90,91] identified and characterized the OsHTD1 (HIGH-TILLERING DWARF1, Os04g46470) gene encoding a carotenoid cleavage dioxygenase (CCD7 or D7 ortholog of Arabidopsis MAX3, involved in the strigolactone biosynthesis). The OsHTD1 was reported to be required for negative regulation of the outgrowth of axillary buds. Similarly, Wang et al. [29] demonstrated that a partial loss-of-function allele of OsHTD1/CCD7 gene increased tiller number and improved rice grain yield. Here, our finding revealed that the OsHTD1 gene (Chr4:27567824..27570926) is located within the QTL qPNL4-1 mapped on chromosome 4, identified by this study as being associated with the control of panicle number under low nitrogen cultivation. The qPNL4-1 is flanked by ad04009559 (left marker, 26443283 kbp) and ah04001252 (right marker, 28788134 kbp).
The breakthrough is that many of the candidate genes found either in qTNL2-1 or qTNN4-1 are also proposed to be involved in plant stress signaling or response mechanisms. A recent study investigating the transcriptional regulation of hormonal biosynthesis or signaling pathways genes under drought stress conditions, proposed that shoot branching related genes could be involved in the adaptive response mechanisms of plants towards abiotic stress tolerance in Arabidopsis [92]. We could speculate that under normal growth conditions, plants utilize their resources for their growth and development. However, under a reduced level of nitrogen, which could also be perceived as a stress, plants would activate abiotic stress responsive genes and/or growth related genes, in order to either escape or avoid the stress or alleviate the adverse effects that would be caused by the stress, while tending to maintain a balanced growth. Owing to the above, a possible crosstalk between plant growth and development and defense mechanisms could be suggested.

Conclusions
Shoot branching is an important agronomic trait contributing to the yield and productivity of cereal food crops, such as rice. In plants, shoot branching is controlled by a complex hormonal signaling network, operating either in synergetic or antagonistic manner. It is well established that most of the genes involved in the regulation of shoot branching, through promotion or inhibition, are also involved in the adaptive response mechanisms of plants towards stress tolerance or resistance (abiotic or biotic stress). In this study, we identified QTLs proposed to be associated with tiller number and panicle number under low nitrogen cultivation, mapped to chromosomes 2 (qTNL2-1) and chromosome 4 (qPNL4-1). Another QTL associated with tiller number was detected in the doubled haploid population grown under normal nitrogen cultivation conditions (qTNN4-1) and mapped on chromosome 4. These QTLs harbor candidate genes with interesting annotated functions, and proposed for playing important roles in the adaptive response mechanisms towards stress (abiotic and biotic) in rice, in addition to being associated with shoot branching. Candidate genes encoding GRAS protein, cytochrome P450, OsRALF8, Aquaporin and OsDhn1, among others, would play important roles in controlling shoot branching under low nitrogen cultivation with regard to their functions and their similarity with previously reported genes. Therefore, this study suggests a possible crosstalk between the regulation of plant growth and development, and the stress response mechanisms. Functional analysis would help identify major genes and elucidate their function.