QTL analysis for carbon assimilate translocation-related traits during maturity in rice (Oryza sativa L.)

Abstract

Problems with carbon assimilate translocation from source organs to sink (grains) during ripening cause yield losses in rice (Oryza sativa L.), especially in high-sink-capacity varieties. We conducted a genetic analysis of traits related to such translocation by using recombinant inbred lines. Shoot weight (SW) of T65, a japonica parent, was retained from heading to late maturity, whereas that of DV85, an aus parent, was greater than that of T65 at 5 days after heading (DAH) and then decreased until 20 DAH. This difference was observed clearly under standard-fertilizer but not low-fertilizer conditions. Non-structural carbohydrate (NSC) contents in the parents showed a tendency similar to that for SW. QTL analysis revealed pleiotropic QTLs on chromosomes 5 and 10, probably by associations with heading date QTLs. A QTL associated with harvest index and NSC at 5 DAH was detected on chromosome 1. By considering the temporal changes of the traits, we found a QTL for decrease in SW on chromosome 11; the DV85 allele of this QTL facilitated assimilate translocation and suppressed biomass growth. A suggestive QTL for NSC decrease was located on chromosome 2. These QTLs could represent potential targets for controlling carbon assimilate translocation in breeding programs.

Introduction

The world population reached 7.3 billion in 2015 and is projected to reach 8.5 billion by 2030 (United Nations World Population Prospects; https://population.un.org/wpp/) and 9.6 billion by 2050 (Food and Agriculture Organization; http://www.un.org/en/development/desa/news/population/2015-report.html). This growth occurs mainly in developing regions such as Africa, where demand for rice is increasing, and Asia, where rice is considered as a staple food. As a consequence, these regions are vulnerable to food shortages (Alexandratos and Bruinsma 2012). Thus, rice production must be increased by 70% by 2050 to meet the demand of the population growth and economic development (Godfray et al. 2010). Therefore, breeding new high-yield rice varieties is a pressing need.

As in other grass species, grain yield in rice depends on two sources: carbohydrates accumulated in the shoot (culm, leaf sheath, and leaf blade) as temporary storage before heading and photosynthates produced during maturity (Chen and Wang 2008, Cock and Yoshida 1972, Gendua et al. 2009, Katsura et al. 2007, Laza et al. 2003, Yoshida 1981). The proportion and the speed of assimilate flow from these two sources to sink varies greatly across varieties and environments (Nagata et al. 2001, Samonte et al. 2001). In rice, up to 24%–27% of the carbohydrates in the grain originates from stem starch reserves (Cock and Yoshida 1972). A high concentration of the stem carbohydrate reserve at heading can be a target to select high-yielding varieties (Katsura et al. 2007, Samonte et al. 2001).

Grain filling is affected by the environment. Excess nitrogen and water supply at maturity delays leaf senescence and decreases the amount of stem reserves (Fu et al. 2011, Hirano et al. 2005, Pan et al. 2011). Under nitrogen deficiency, photosynthesis activity is decreased, and carbohydrate metabolism is altered so that the root/shoot biomass ratio increases (Antal et al. 2010, Hermans et al. 2006, Paul and Driscoll 1997). Translocation of dry matter from shoot to grain also compensates for the shortage of assimilate supply for grains under unfavorable conditions during grain filling (Blum et al. 1994, Lemoine et al. 2013, Tian et al. 2016). Under water deficiency, translocation from stem reserves to grains increases, shortening the grain-filling period (Yang et al. 2000).

Many rice varieties have low yield despite having a large source and high sink capacity because of a low grain-filling ratio (Peng et al. 2008). Environmental and developmental changes often reduce the potential of the sink and source (White et al. 2016). Therefore, optimization of the source/sink ratio depending on the environment should be considered for improving yield potential. Identification of QTLs controlling the source/sink distribution of assimilates would enhance our understanding of the complex relationships among these traits and facilitate genetic improvement. A limited number of candidate genes for improving assimilate translocation have been reported (Hirose et al. 2013, Wada et al. 2017, Wang et al. 2017).

In many studies, translocation efficiency was determined on the basis of the amount of non-structural carbohydrates (NSC) in stems at heading and maturity stages (Nagata et al. 2002, Slewinski and Braun 2010, Slewinski 2012). In rice, there are reports showing the locations of QTLs related to carbohydrate assimilation and translocation (Kanbe et al. 2009, Kashiwagi and Ishimaru 2004, Kashiwagi et al. 2006, Nagata et al. 2002). Ishimaru et al. (2007) reported 13 QTLs for carbohydrate production and revealed that there is no correlation between the amount of carbohydrates stored in shoots before heading and grain yield. This finding contradicts previous studies, which showed that higher capacity of the carbohydrate reserve can increase grain yield (Katsura et al. 2007, Samonte et al. 2001); this discrepancy can likely be explained by the complex nature of grain filling in rice.

In this study, the main objective was to genetically dissect the translocation-related traits in rice. We introduced a simple phenotyping method to identify QTLs for shoot reserve translocation. T65, a japonica rice cultivar, DV85, an aus cultivar, and recombinant inbred lines (RILs) derived from these two parents were used as the mapping population. This study is the first to use the weight of shoots and panicles for QTL analysis. In addition, QTLs for NSC content and individual carbohydrates at three stages after heading were analyzed.

Materials and Methods

Plant materials

A population of 92 RILs (F₁₀ and F₁₁) constructed by a single-seed descent method from a cross between T65 (japonica cultivar) and DV85 (aus cultivar) was kindly supplied by the National Bioresource Project (NBRP). Seeds were pre-germinated in water at room temperature for 3 days. Only germinated seeds were sown on 22 May 2016. Ten 30-day-old seedlings per cultivar or RIL were transplanted in a single row with a spacing of 20 cm between hills and 30 cm between rows. Field experiments were conducted under natural conditions in the paddy fields (Oxyaquic Dystrudept soil) of Togo Field, Nagoya University (35°06′36.5″N, 137°05′06.3″E). The RILs were planted in two plots, one with standard fertilizer (SF) with basal dressing (30 kg N/ha, 25 kg P/ha, 30 kg K/ha) and dressing at the maximum tillering stage (40 kg N/ha, 35 kg K/ha), and the other one with no fertilizer (low fertilizer, LF). Pests and diseases were intensively controlled using chemicals to avoid yield loss.

Sampling and evaluation of phenotypes

Because the shoot carbohydrate reserve can re-accumulate at the final grain-filling stage (Kashiwagi et al. 2006), we took samples at 5, 20, and 35 days after heading (DAH). The heading dates of the plant materials (days to heading, DTH) were monitored for each individual, and the aboveground parts of two plants (including dead leaves but excluding roots) were taken at 5, 20, and 35 DAH for RILs, whereas samples of parental lines were taken every 5 days from 5 to 35 DAH, 2 plants per each sampling time; sampling was performed between 10 am and 2 pm. Ten plants of each RIL were planted in a row. Because the order of the RILs were random, the effect of the locations in the fields were randomized. Second and third plants in a row of an RIL were taken at 5 DAH, fifth and sixth at 20 DAH and eighth and ninth at 35 DAH, to minimize the border effect. Sampling of the parents was done in 5 days intervals using 3 rows, first row for 5, 10, and 15 DAH, second row for 20, 25 and 30 DAH, and third row for 35 DAH. If the target plants were damaged or contained off-types, fourth or seventh plants were used as alternatives. Despite lodging or shattering in some of the samples, we tried to recover the maximum parts of the plants. However, the samples of 35 DAH with visible damages were removed from the analysis: in SF, 35 DAH samples of 11 RILs were not used and in LF, 35 DAH samples of 3 RILs could not be sampled. On the other hand, 25, 30 and 35 DAH samples of DV85 with small damages in panicles were analyzed to clarify the difference between the parents. Samples were washed to remove soil and dried in an oven at 80°C for 48 h to obtain constant weight. All samples were separated into shoots (stems, leaf blades, leaf sheaths) and panicles. Shoot weight (SW) and panicle weight (PW) were measured and used as trait values (referred to as SW5, PW5, SW20, PW20, SW35, and PW35). In particular, the differences in SW and PW between the three sampling time points were used to figure out QTLs for assimilate translocation. Total biomass (BM) and harvest index (HI) were defined as BM = SW + PW, HI (%) = PW35 BM35⁻¹ × 100. PW35 was used as a proxy for yield per plant (harvested product) to calculate HI.

Measurement of NSC contents

NSC contents of the stems were determined as described in Sugiura et al. (2015). Stems were ground to fine powder in a grinder (Sansei SC-01, www.sanshoindy.com). Samples (3–5 mg) were weighed immediately after additional drying at 50°C for 24 h. Soluble sugars were extracted with 80% ethanol at 80°C. Sucrose was hydrolyzed to glucose and fructose by invertase. Starch precipitate was heated in water at 100°C for 1 h and then treated with amyloglucosidase (A-9228, Sigma Aldrich, St. Louis, MO, USA) in 50 mM Na-acetate buffer (pH 4.5) at 55°C for 1 h. The glucose content of each fraction was quantified based on an enzymatic method. The concentrations of these substances were converted to mg/g dry weight of the stems and used as trait values (NSC5, NSC20, and NSC35).

Measurement of chlorophyll (Chl) content

A SPAD 502 Plus Chlorophyll Meter (Konica Minolta, Tokyo, Japan) was used to estimate the Chl content of flag leaves. Heading date of each line was monitored and Chl was measured at 5, 20, and 35 DAH (designated SPAD5, SPAD20, and SPAD35, respectively). Because a linear model fits SPAD values to Chl content (Wakiyama 2016), difference of the SPAD values were used as the indicators for temporal changes.

Genotyping of RILs by sequencing

DNA was extracted from lyophilized leaf samples by using a modified Dellaporta method, and its quality was checked by electrophoresis on a 0.6% agarose gel in 0.5×TBE buffer. A QuantiFluor dsDNA System and Quantus Fluorometer (Promega, USA) were used for DNA quantification.

A genotyping-by-sequencing library was prepared following the protocol by Poland et al. (2012) and Furuta et al. (2017). Briefly, (1) 200 ng (20 ng × 10 μl) of DNA from each sample was digested with KpnI (“rare cutter”) and MspI (“common-cutter”); (2) digested DNA was ligated to barcode adaptors (with a KpnI site) and a ‘Y-adapter’ (with an MspI site); (3) ligated samples were pooled (multiplexed) and purified using a QIAquick PCR Purification Kit (Qiagen); (4) pooled DNA was amplified with outer primers to add sequences for primer sites and flowcell-binding sequences for Illumina sequencing; (5) amplified pooled DNA was purified using the QIAquick PCR Purification Kit, quantified with the QuantiFluor dsDNA system, and its quality was checked using a MultiNA electrophoresis instrument (Shimadzu); and (6) the library was diluted to 10 pM and used for next generation sequencing on a MiSeq platform (Illumina Inc., San Diego, CA, USA) together with 5% PhiX control (PhiX Control v3, Illumina). Informatics was conducted as described in Furuta et al. (2017) to obtain genotypes.

Statistical analysis and QTL mapping

All of the data were input into Microsoft Excel and R 3.51 (R Core Team 2018). Correlation analysis was conducted by using R. QTL analysis was based on interval mapping with the “hk” option of “scanone” function implemented in R/qtl (Broman et al. 2003). Significant QTLs were identified on the basis of an empirical threshold determined by 1000 permutation tests (Churchill and Doerge 1994).

Results

Parental phenotypes

Frequency distributions and parental phenotypes are shown in Supplemental Fig. 1. Temporal changes in SW and PW of the parents are shown in Fig. 1. In general, SW and PW were higher in the SF plot than in the LF plot. The difference between the parents was more obvious in the SF plot. At 5 DAH, SW was greater in DV85 than in T65. In DV85, SW decreased rapidly from 5 to 20 DAH, and then slightly until 35 DAH, but remained constant over the whole maturity period in T65 (Fig. 1A). The rapid decrease of SW in DV85 corresponded to a rapid increase of PW. On the other hand, PW of T65 increased slowly during maturity. The decrease of PW of DV85 from 25 to 35 DAH can be explained by lodging and shattering due to heavy rain and typhoon during this period. No clear difference between the parents was observed in the LF plot (Fig. 1B).

Fig. 1

Temporal changes in panicle weight (PW) and shoot weight (SW) (mean ± SD) of T65 and DV85 in (A) standard fertilizer (SF) and (B) low fertilizer (LF) plots. DAH, days after heading.

The concentrations of carbohydrate components in the stems of T65 and DV85 under two different nitrogen levels are shown in Fig. 2. NSC was defined as the total of glucose, sucrose, and starch; the latter two were the main NSC components. In general, both parents accumulated significantly more NSC from 5 to 15 DAH in the LF plot than in the SF plot. DV85 accumulated higher NSC concentrations than did T65 from 5 to 10 DAH in both plots (Fig. 2). Sucrose and starch concentrations decreased rapidly from 5 to 20 DAH in DV85 in both plots (Fig. 2C, 2D). T65 in the SF plot (Fig. 2A) retained more constant concentration over the maturity period. Re-accumulation of NSC at 30–35 DAH was observed in both parents in both plots.

Fig. 2

Concentrations of non-structural carbohydrate components (mean ± SD) of (A, B) T65 and (C, D) DV85 in the (A, C) standard fertilizer (SF) and (B, D) low fertilizer (LF) plots.

Trait correlations in RILs

Because the differences between the parents were more apparent in the SF plot, we focused on this plot. Correlations among the traits in RILs are shown in Fig. 3 and Supplemental Fig. 2. Indicators of temporal changes in BM, BM20-5, BM35-20, and BM35-5 were defined as the differences between the BM values at 5, 20, and 35 DAH, respectively; similar traits were defined for SPAD and NSC. In the same manner, shoot weight transfer ratios (SWTR) were defined as follows: SWTR5-20 = (SW5 – SW20) SW5⁻¹; SWTR20-35 = (SW20 – SW35) SW20⁻¹; and SWTR5-35 = (SW5 – SW35) SW5⁻¹. Positive correlation with PW35 (used as a proxy for yield per plant) was observed for HI, BM5, BM20, BM35, BM20-5, BM35-20, BM35-5, SW5, SW20, SWTR20-35, SWTR5-35, NSC20-35, and NSC5-35, whereas negative correlation was observed for NSC35, NSC5-20, and SPAD35. These results imply that higher PW35 (yield) was accompanied by a decrease in SW at 35 DAH and was related to SW changes at a late maturity stage because SWTR5-20 and NSC5-20 were not correlated with PW35.

Fig. 3

Correlations of phenotypes in the SF plot. Additional information is provided in Supplemental Fig. 2. Color of dots indicates the correlation coefficient values corresponding to the bar on the right.

QTLs detected in RILs

To detect QTLs for the traits analyzed, 3276 single nucleotide polymorphism markers were obtained by genotyping-by-sequencing and used for interval mapping (Supplemental Figs. 3, 4, Table 1).

Table 1 QTLs detected in the SF plot^a

QTL	Chrb	Trait	Marker	Position (Mb)	LOD	AEc	PVE (%)d
qHI1	1	HI	S1_9626817	9.63	4.59	−4.74	23.49
qNSC1	1	NSC5	S1_9626839	9.63	3.76	23.29	20.61
qNSC2	2	NSC5-35	S2_33486011	33.49	3.14e	24.88	20.54
qHI5	5	HI	S5_24022776	24.02	3.80	−3.88	19.87
qDTH5	5	DTH	S5_24022776	24.02	4.30	2.84	19.37
qBM5	5	BM5	S5_24481151	24.48	7.04	8.65	29.71
qBM5	5	BM20	S5_24481151	24.48	3.78	6.73	18.31
qSW5	5	SW5	S5_24481151	24.48	5.47	7.71	27.32
qSW5	5	SW20	S5_24481151	24.48	5.09	6.40	25.67
qSW5	5	SW35	S5_24520974	24.52	4.48	5.86	23.00
qBM5	5	BM35	S5_24621229	24.62	2.67	5.99	14.39
qDTH10	10	DTH	S10_16626134	16.63	9.98	−4.09	39.34
qSW10	10	SW20	S10_16626134	16.63	4.23	−6.27	21.87
qSW10	10	SW35	S10_16626134	16.63	4.70	−6.46	23.95
qHI10	10	HI	S10_16626134	16.63	6.42	5.32	31.21
qNSC10.f	10	NSC35	S10_16626134	16.63	3.98	−30.23	21.17
qBM10	10	BM5	S10_17367149	17.37	3.39	−7.02	15.59
qNSC10.1f	10	NSC20	S10_17472779	17.47	3.62	−28.04	19.01
qNSC10.1f	10	NSC5	S10_17472779	17.47	3.44	−22.62	19.02
qNSC10.2	10	NSC20	S10_21362510	21.36	3.44	−26.36	18.39
qSWTR11	11	SWTR20-35	S11_16893255	16.89	6.76	0.09	33.27
qSWTR11	11	SWTR5-35	S11_16893255	16.89	5.67	0.09	28.75
qBM11	11	BM35-5	S11_16893255	16.89	2.56e	−4.34	13.86

^a  Full information on QTL analysis is provided in Supplemental Fig. 4.

^b  Chromosome.

^c  Additive effects of the marker calculated as ((average of DV85) − (average of T65))/2. Positive values mean that the DV85 allele increased the trait value.

^d  Percentage of variance explained by the QTL.

^e  Non-significant (suggestive) QTLs.

^f  These QTLs were considered to be identical, so the same designation was used.

PW, SW, BM, HI, and DTH

No significant QTLs for PW (Supplemental Fig. 4) were detected. For BM5 and BM20, a major QTL was detected on chromosome 5 (qBM5), and a QTL for BM5 was detected on chromosome 10 (qBM10) (Fig. 4A, Table 1). SW, DTH, and HI shared the same QTLs on chromosomes 5 and 10. At the QTL on chromosome 5, the DV85 allele increased BM and DTH and decreased HI, whereas the one on chromosome 10 showed opposite effects (Fig. 4B–4D, Table 1). For HI, another QTL was detected on chromosome 1 (qHI1); the DV85 allele decreased HI (Fig. 4D, Table 1).

Fig. 4

Interval mapping showing the locations of QTLs detected in the SF plot. Trait names are above each panel. The black, red, and blue lines correspond to 5, 20, and 35 DAH, respectively, in panels A, B, and F, and to shoot weight transfer ratios (SWTR) 5–20, 20–35, and 5–35 in panel E. Asterisk in E indicates a false-positive LOD peak detected by strong segregation distortion. Horizontal dotted lines in all panels except H indicate an empirical threshold at the 5% level.

QTLs for temporal changes during maturity

To detect translocation-related QTLs, we analyzed temporal changes of the traits. Among them, SWTR20-35 and SWTR5-35 shared the same QTL on chromosome 11 (Fig. 4E, Table 1). At this QTL, the DV85 allele showed higher SWTR values and thus enhanced the loss of shoot weight during the late maturity stage. This QTL was not observed in the LF plot (data not shown).

QTLs for NSC and its temporal changes

On chromosome 1, a QTL for NSC5 was detected at the same position as qHI1, where the DV85 allele increased NSC5. Two QTLs for NSC20 and NSC35 (qNSC10.1 and qNSC10.2) were detected on chromosome 10 (Fig. 4F, Table 1). The position of qNSC10.1 was the same as that of qDTH10. For temporal changes in NSC, no significant QTLs were detected, but one suggestive QTL for NSC5-35 (qNSC2, LOD = 3.14, i.e. slightly below the 5% threshold = 3.28) was detected on chromosome 2; at this locus, the DV85 allele enhanced the decrease of NSC (Fig. 4G, Table 1).

Discussion

Comparison of the characteristics of carbohydrate translocation in the parents

During carbohydrate translocation, photoassimilates are distributed from the source or temporary storage (stem) organs to sink (grains). Previous studies often quantified carbohydrate translocation by measuring stem NSC concentration at heading and maturity (Hirose et al. 2013, Nagata et al. 2001, Wang et al. 2017, Zhang et al. 2017). However, these studies did not analyze translocation of biomass to panicles. In the present study, we found that T65 (temperate japonica) and DV85 (aus) have different translocation characteristics: T65 retained more shoot biomass during maturity and depended more on newly synthesized assimilates, whereas DV85 accumulated more shoot biomass at the heading stage and the biomass was quickly transferred to panicles. We observed a similar tendency for NSC contents in stems (Figs. 1, 2).

Fertilizer conditions greatly affected these traits (Figs. 1, 2). The LF condition facilitated accumulation of NSC in both parents, suggesting that plants retain more NSC under severe nutrient deficiency. This indicates that starved rice plants prioritize survival over growth, in line with the increased root/shoot biomass ratio (Antal et al. 2010, Hermans et al. 2006, Paul and Driscoll 1997). Because the differences between translocation-related traits were more apparent under SF than LF conditions, we considered the SF conditions to be more suitable to detect QTLs that characterize the parents.

Characterization of translocation-related QTLs

SW (SW5, SW20, and SW35) and PW (PW5, PW20, and PW35) showed correlations among different stages (Fig. 3). QTLs for SW were detected on chromosomes 5 and 10, and their positions were the same as those of QTLs for DTH, indicating that the duration of vegetative growth affects these traits. We consider the QTL on chromosome 10 to be Ehd1 (Doi et al. 2004). However, no QTLs for PW or its temporal changes were detected (Supplemental Fig. 4), suggesting that the yield was controlled by many genes that could not be detected because of a low LOD score of each locus. On the other hand, by using the temporal changes in SW, we detected a novel QTL on chromosome 11 (qSWTR11), and the DV85 allele caused a rapid loss of SW at maturity. In addition, a suggestive QTL for BM35-5 was detected on chromosome 11 (LOD = 2.56, Fig. 4H, Table 1), where the DV85 allele suppressed biomass production at maturity (5 to 35 DAH). Taken together, these data indicate that the QTL on chromosome 11 regulates both the allocation of photoassimilates and retainability of photosynthesis activity.

Although the difference between the parents in SW decrease during maturity was observed until 20 DAH (Fig. 1), this QTL was detected at the late maturity stage (SWTR20-35 and SWTR5-35), probably because the photosynthesis activity was retained until 20 DAH and translocation was not very active, when all of the RILs were averaged for QTL analysis. Thus, from 20 to 35 DAH, the decrease in photosynthesis activity because of leaf senescence and assimilate translocation acted as the main force of grain filling, which allowed the detection of these QTLs for SWTR.

Characterization of QTLs for NSC and SPAD

Because NSC and SPAD are physiological traits related to translocation and photosynthesis, we chose them as the target traits. QTLs for NSC at each stage were detected on chromosomes 1 (NSC5) and 10 (NSC5, NSC20, and NSC35) (Fig. 4G). qNSC10.1 was probably Ehd1 (Doi et al. 2004) and affected pleiotropically by heading date. qNSC10.2 was detected as a peak separate from qNSC10.1. The position of qNSC1 was the same as that of qHI1. The DV85 allele at this QTL increased accumulation of NSC and decreased HI. Therefore, we consider the DV85 allele to increase stem NSC reserves at the early maturity stage. This QTL can be a target for selection to breed a high-yielding variety (Katsura et al. 2007, Samonte et al. 2001).

A suggestive QTL for NSC5-35 was detected on chromosome 2 (Fig. 4G). The DV85 allele at this QTL decreased NSC during maturity and resulted in a rapid decrease of NSC in DV85. This QTL was not detected for other traits and thus appears to control the NSC concentration at the late maturity stage independently of other QTLs.

The SPAD values were analyzed as an indicator of temporal changes in photosynthesis. No QTLs were detected for the temporal changes in SPAD, but two QTLs for SPAD5 (chromosomes 1 and 7), one for SPAD20 (chromosome 1), and one for SPAD35 (chromosome 3) were detected (Supplemental Fig. 4). However, none of the SPAD-related traits was correlated with SWTR or shared a QTL with this trait. This indicated that SPAD values were not appropriate for monitoring actual photosynthesis activity.

Potential for yield improvement

Because the traits analyzed in the present study greatly fluctuate by environment (Figs. 1, 2), it should be noted that the information on the QTL in the present study should be regarded as an example in a specific condition of single year. However, QTLs with relatively high LOD scores are expected to be the target for further analysis. In this study, novel QTLs for SWTR and NSC were detected. A simple method using a temporal change in biomass extracted a QTL for SWTR on chromosome 11. One QTL for NSC was associated with HI (qNSC1), and a suggestive QTL (qNSC2) for a temporal change in NSC independent from other traits was identified. Of the two QTLs identified on chromosome 10, one was probably identical to Ehd1 (qNSC10.1) (Doi et al. 2004) and the other one (qNSC10.2) was linked to qNSC10.1 and needs further confirmation and characterization. The OsGWD, OsHXK6, and ISA2 genes are reported to affect grain filling in rice (Hirose et al. 2013, Wada et al. 2017, Wang et al. 2017). However, the positions of the QTLs in the present study are different from those of these genes. Thus, cloning of the genes underlying the QTLs such as qSWTR11 will elucidate the mechanism of assimilate translocation in rice. We will characterize the QTLs detected in the present study and also other QTLs for yield or other physiological traits by using near-isogenic lines or chromosome segment substitution lines in a uniform genetic background.

To achieve maximum yield, virtually all of the NSC should be translocated to grains. However, the loss of activity of the shoot results in lodging before harvesting (Kashiwagi et al. 2008). In addition, the QTLs detected in this study did not contribute to PW and hence yield, and a different set of QTLs was detected under LF conditions (data not shown). Growth modeling is expected to help optimization of these complicated factors (Li et al. 2017). Currently, ecophysiological models are independent from the genetic characteristics of each variety and DNA genotypes. Because the genetic mechanisms underlying rice yield are complicated and interact with the environment, a different way of phenotyping such as that used in the present study will help to construct a better growth model that will include genotypes, and hence help improve the yield of rice.

Acknowledgments

Part of this research was supported by the Ministry of Education and Training and Ministry of Rural Development-Vietnam to HDP. The authors are also grateful to the Cross-ministerial Strategic Innovation Promotion Program (SIP) for support to KD, DS, and MK; to the National Bioresource Project (NBRP) for support to KD; and to the SATREPS project for support to KD and DM. The authors appreciate kind support by Dr. Katsuya Yano, who provided technical advice.

Literature Cited

•  Alexandratos,  N. and  J.  Bruinsma (2012) World agriculture towards 2030/2050: The 2012 Revision. http://www.fao.org/docrep/016/ap106e/ap106e.pdf
•  Antal,  T.,  H.  Mattila,  M.  Hakala-Yatkin,  T.  Tyystjärvi and  E.  Tyystjärvi (2010) Acclimation of photosynthesis to nitrogen deficiency in Phaseolus vulgaris. Planta 232: 887–898.
•  Blum,  A.,  B.  Sinmenaa,  J.  Mayer,  G.  Golana and  L.  Shpiiler (1994) Stem reserve mobilisation supports wheat-grain filling under heat stress. Aust. J. Plant Physiol. 21: 771–781.
•  Broman,  K.W.,  H.  Wu,  Ś.  Sen and  G.A.  Churchill (2003) R/qtl: QTL mapping in experimental crosses. Bioinformatics 19: 889–890.
•  Chen,  H.J. and  S.J.  Wang (2008) Molecular regulation of sink–source transition in rice leaf sheaths during the heading period. Acta Physiol. Plant. 30: 639–649.
•  Churchill,  G.A. and  R.W.  Doerge (1994) Empirical threshold values for quantitative trait mapping. Genetics 138: 963–971.
•  Cock,  J.H. and  S.  Yoshida (1972) Accumulation of ¹⁴C-labelled carbohydrate before flowering and its subsequent redistribution and respiration in the rice plant. Jpn. J. Crop Sci. 41: 226–234.
•  Doi,  K.,  T.  Izawa,  T.  Fuse,  U.  Yamanouchi,  T.  Kubo,  Z.  Shimatani,  M.  Yano and  A.  Yoshimura (2004) Ehd1, a B-type response regulator in rice, confers short-day promotion of flowering and controls FT-like gene expression independently of Hd1. Genes Dev. 18: 926–936.
•  Fu,  J.,  Z.  Huang,  Z.  Wang,  J.  Yang and  J.  Zhang (2011) Pre-anthesis non-structural carbohydrate reserve in the stem enhances the sink strength of inferior spikelets during grain filling of rice. Field Crops Res. 123: 170–182.
•  Furuta,  T.,  M.  Ashikari,  K.K.  Jena,  K.  Doi and  S.  Reuscher (2017) Adapting genotyping-by-sequencing for rice F₂ populations. G3 (Bethesda) 7: 881–893.
•  Gendua,  P.A.,  Y.  Yamamoto,  A.  Miyazaki,  T.  Yoshida and  Y.  Wang (2009) Responses of yielding ability, sink size and percentage of filled grains to the cultivation practices in a Chinese large-panicle-type rice cultivar, Yangdao 4. Plant Prod. Sci. 12: 243–256.
•  Godfray,  H.C.J.,  J.R.  Beddington,  I.R.  Crute,  L.  Haddad,  D.  Lawrence,  J.F.  Muir,  J.  Pretty,  S.  Robinson,  S.M.  Thomas and  C.  Toulmin (2010) Food security: The challenge of feeding 9 billion people. Science 327: 812–818.
•  Hermans,  C.,  J.P.  Hammond,  P.J.  White and  N.  Verbruggen (2006) How do plants respond to nutrient shortage by biomass allocation? Trends Plant Sci. 11: 610–617.
•  Hirano,  T.,  Y.  Saito,  H.  Ushimaru and  H.  Michiyama (2005) The effect of the amount of nitrogen fertilizer on starch metabolism in leaf sheath of japonica and indica rice varieties during the heading period. Plant Prod. Sci. 8: 122–130.
•  Hirose,  T.,  N.  Aoki,  Y.  Harada,  M.  Okamura,  Y.  Hashida,  R.  Ohsugi,  A.  Miyao,  H.  Hirochika and  T.  Terao (2013) Disruption of a rice gene for α-glucan water dikinase, OsGWD1, leads to hyperaccumulation of starch in leaves but exhibits limited effects on growth. Front. Plant Sci. 4: 1–9.
•  Ishimaru,  K.,  N.  Hirotsu,  Y.  Madoka and  T.  Kashiwagi (2007) Quantitative trait loci for sucrose, starch, and hexose accumulation before heading in rice. Plant Physiol. Biochem. 45: 799–804.
•  Kanbe,  T.,  H.  Sasaki,  N.  Aoki,  T.  Yamagishi and  R.  Ohsugi (2009) The QTL analysis of RuBisCO in flag leaves and non-structural carbohydrates in leaf sheaths of rice using chromosome segment substitution lines and backcross progeny F₂ Populations. Plant Prod. Sci. 12: 224–232.
•  Kashiwagi,  T. and  K.  Ishimaru (2004) Identification and functional analysis of a locus for improvement of lodging resistance in rice. Plant Physiol. 134: 676–683.
•  Kashiwagi,  T.,  Y.  Madoka,  N.  Hirotsu and  K.  Ishimaru (2006) Locus prl5 improves lodging resistance of rice by delaying senescence and increasing carbohydrate reaccumulation. Plant Physiol. Biochem. 44: 152–157.
•  Kashiwagi,  T.,  E.  Togawa,  N.  Hirotsu and  K.  Ishimaru (2008) Improvement of lodging resistance with QTLs for stem diameter in rice (Oryza sativa L.). Theor. Appl. Genet. 117: 749–757.
•  Katsura,  K.,  S.  Maeda,  T.  Horie and  T.  Shiraiwa (2007) Analysis of yield attributes and crop physiological traits of Liangyoupeijiu, a hybrid rice recently bred in China. Field Crops Res. 103: 170–177.
•  Laza,  M.R.,  S.  Peng,  S.  Akita and  H.  Saka (2003) Contribution of biomass partitioning and translocation to grain yield under sub-optimum growing conditions in irrigated rice. Plant Prod. Sci. 6: 28–35.
•  Lemoine,  R.,  S.  La Camera,  R.  Atanassova,  F.  Dédaldéchamp,  T.  Allario,  N.  Pourtau,  J.L.  Bonnemain,  M.  Laloi,  P.  Coutos-Thévenot,  L.  Maurousset et al. (2013) Source-to-sink transport of sugar and regulation by environmental factors. Front. Plant Sci. 4: 272.
•  Li,  T.,  O.  Angeles,  M.  Marcaida III,  E.  Manalo,  M.P.  Manalili,  A.  Radanielson and  S.  Mohanty (2017) From ORYZA2000 to ORYZA (v3): An improved simulation model for rice in drought and nitrogen-deficient environments. Agric. For. Meteorol. 237–238: 246–256.
•  Nagata,  K.,  S.  Yoshinaga,  J.  Takanashi and  T.  Terao (2001) Effects of dry matter production, translocation of nonstructural carbohydrates and nitrogen application on grain filling in rice cultivar Takanari, a cultivar bearing a large number of spikelets. Plant Prod. Sci. 4: 173–183.
•  Nagata,  K.,  Y.  Fukuta,  H.  Shimizu,  T.  Yagi and  T.  Terao (2002) Quantitative trait loci for sink size and ripening traits in rice (Oryza sativa L.). Breed. Sci. 52: 259–273.
•  Pan,  J.,  K.  Cui,  D.  Wei,  J.  Huang,  J.  Xiang and  L.  Nie (2011) Relationships of non-structural carbohydrates accumulation and translocation with yield formation in rice recombinant inbred lines under two nitrogen levels. Physiol. Plant. 141: 321–331.
•  Paul,  M.J. and  S.P.  Driscoll (1997) Sugar repression of photosynthesis: The role of carbohydrates in signaling nitrogen deficiency through source: sink imbalance. Plant Cell Environ. 20: 110–116.
•  Peng,  S.,  G.S.  Khush,  P.  Virk,  Q.  Tang and  Y.  Zou (2008) Progress in ideotype breeding to increase rice yield potential. Field Crops Res. 108: 32–38.
•  Poland,  J.A.,  P.J.  Brown,  M.E.  Sorrells and  J.L.  Jannink (2012) Development of high-density genetic maps for barley and wheat using a novel two-enzyme genotyping-by-sequencing approach. PLoS ONE 7: e32253.
• R Core Team (2018) R: A Language and Environment for Statistical Computing. https://www.R-project.org
•  Samonte,  S.O.P.B.,  L.T.  Wilson,  A.M.  McClung and  L.  Tarpley (2001) Seasonal dynamics of nonstructural carbohydrate partitioning in 15 diverse rice genotypes. Crop Sci. 41: 902–909.
•  Slewinski,  T.L. and  D.M.  Braun (2010) The psychedelic genes of maize redundantly promote carbohydrate export from leaves. Genetics 185: 221–232.
•  Slewinski,  T.L. (2012) Non-structural carbohydrate partitioning in grass stems: a target to increase yield stability, stress tolerance, and biofuel production. J. Exp. Bot. 63: 4647–4670.
•  Sugiura,  D.,  E.  Betsuyaku and  I.  Terashima (2015) Manipulation of the hypocotyl sink activity by reciprocal grafting of two Raphanus sativus varieties: its effects on morphological and physiological traits of source leaves and whole-plant growth. Plant Cell Environ. 38: 2629–2640.
•  Tian,  Z.,  J.  Li,  X.  Jia,  F.  Yang and  Z.  Wang (2016) Assimilation and translocation of dry matter and phosphorus in rice genotypes affected by salt-alkaline stress. Sustainability 8: 1–14.
•  Wada,  H.,  C.  Masumoto-Kubo,  K.  Tsutsumi,  H.  Nonami,  F.  Tanaka,  H.  Okada,  R.  Erra-Balsells,  K.  Hiraoka,  T.  Nakashima,  M.  Hakata et al. (2017) Turgor-responsive starch phosphorylation in Oryza sativa stems: a primary event of starch degradation associated with grain-filling ability. PLoS ONE 12: e0181272.
•  Wakiyama,  Y. (2016) The relationship between SPAD values and leaf blade chlorophyll content throughout the rice development cycle. JARQ 50: 329–334.
•  Wang,  D.R.,  R.  Han,  E.J.  Wolfrum and  S.R.  McCouch (2017) The buffering capacity of stems: genetic architecture of nonstructural carbohydrates in cultivated Asian rice, Oryza Sativa. New Phytol. 215: 658–671.
•  White,  A.C.,  A.  Rogers,  M.  Rees and  C.P.  Osborne (2016) How can we make plants grow faster? A source–sink perspective on growth rate. J. Exp. Bot. 67: 31–45.
•  Yang,  J.,  J.  Zhang,  Z.  Huang,  Q.  Zhu and  L.  Wang (2000) Remobilization of carbon reserves is improved by controlled soil-drying during grain filling of wheat. Crop Sci. 40: 1645–1655.
•  Yoshida,  S. (1981) Fundamentals of rice crop science. International Rice Research Institute, Los Baños, p. 269.
•  Zhang,  J.,  G.  Li,  Q.  Huang,  Z.  Liu,  C.  Ding,  S.  Tang,  L.  Chen,  S.  Wang,  Y.  Ding and  W.  Zhang (2017) Effects of culm carbohydrate partitioning on basal stem strength in a high-yielding rice population. Crop J. 5: 478–487.