Toshiyuki Tki*, Ptrik Lumngls, Eliz Vie Simon,Yumiko Ari-Snoh,Hidetoshi Asi, Nouy Koyshi,**

aJapan International Research Center for Agricultural Sciences,Tsukuba,Ibaraki 305-8686,Japan

bInternational Rice Research Institute,DAPO Box 7777,Metro Manila,Philippines

cInstitute of Crop Science, National Agriculture and Food Research Organization (NARO),Tsukuba,Ibaraki 305-8518,Japan

Keywords:Adaptability Genotype×environment interaction Rice Sink capacity Yield potential

ABSTRACT Increasing rice yield potential is a continuous challenge posed by world population growth.To increase yield potential, favorable alleles of valuable genes need to be accumulated in promising germplasm. We conducted comparative yield trials for two years in Tsukuba,Japan, in a temperate region and at the International Rice Research Institute (IRRI),Philippines, in a tropical region using five high-yielding rice cultivars: Takanari and Hokuriku193, developed in Japan, and IR64, NSIC Rc158, and YTH183, developed in the Philippines.Genotype plus genotype×environment interaction(GGE)biplot analysis across four environments (two regions × two seasons) classified the five cultivars into four categories: Takanari and YTH183 showed high adaptability to both tropical and temperate regions, Hokuriku193 was suitable for temperate regions, NSIC Rc158 was suitable for the tropics, and IR64 was inferior to the other cultivars in both regions. The high yield and adaptability in Takanari and YTH183 were attributed to their large sink capacity with good grain filling.The plant type for high yield was different,however,between the two cultivars;Takanari was a panicle-weight type, whereas YTH183 was a panicle-number type.Evaluations of F2 progeny of a cross between Takanari and YTH183 showed transgressive segregation for number of panicles per plant as well as number of spikelets per panicle,leading some F2 plants to produce more spikelets per plant (corresponding to larger sink size) than the parental cultivars in both regions. These results suggest the possibility of developing rice cultivars with high yield potential in both temperate and tropical regions by crossing temperate with tropical high-yielding cultivars.

1. Introduction

Rice, first domesticated in the Pearl River basin in southern China [1], is a staple food worldwide. With increasing population growth and limited expansion of arable land, rice production must be boosted through increases in yield potential and optimal cropping practices [2,3].

To continue to increase rice yield potential, favorable alleles of valuable genes for yield-related traits need to be accumulated in promising germplasm in high-yield breeding programs [4]. However, because breeding programs are generally carried out region-specifically, it is not clear whether the new developed cultivars show high yield and adaptation in other regions unless they are evaluated there.Once adaptation is confirmed, the cultivars can be used for breeding programs in those new locations.

In the tropics, in the 1990s and 2000s, the International Rice Research Institute (IRRI) released New Plant Type (NPT)[5]cultivars according to an ideotype breeding strategy,based on knowledge of crop physiology and morphology, aimed at the improvement of yield potential using tropical japonica germplasm. One of the NPT cultivars, NSIC Rc158, outyielded previously developed indica cultivars including IR64 under tropical irrigated conditions [6]. YTH183, also recently developed at IRRI, carries some genomic segments from an NPT cultivar in the genetic background of IR64 [7]. Yield trials in the tropics confirmed the superior performance of YTH183 over IR64 [8].

In temperate Japan,breeders released two indica-dominant high-yielding cultivars, Takanari in 1990 [9] and Hokuriku193 in 2007[10],using both indica and japonica germplasm.Recent yield trials in Japan confirmed the superior performance of Takanari and Hokuriku193 over other japonica-dominant highyielding cultivars,and both cultivars recorded nearly 13 t ha-1of grain yield[11].

Because tropical environments are different from temperate ones, NSIC Rc158 and YTH183, developed in tropical regions, are suggested to have different high-yield potential characteristics from Takanari and Hokuriku193, developed in temperate regions.If NSIC Rc158 and YTH183 show high yield and adaptation in temperate environments or Takanari and Hokuriku193 show high yield and adaptation in tropical environments, these cultivars could be highly desirable germplasm for allele accumulation for increasing yield potential in either environment.However,no such yield trials have been conducted using these cultivars, so that limited information about wide environmental adaptability has been available for high-yielding rice cutivars[12].

The objectives of the present study were to (1) investigate yield potential and wide adaptability among the high-yielding cultivars mentioned above by comparative yield trials in irrigated paddy fields in two regions: the tropical Philippines and temperate Japan; (2) identify key traits that determine high yield and adaptability in both tropical and temperate regions;and(3)test these traits in F2populations derived from a cross between the promising cultivars, with the aim of developing further high-yielding cultivars in both tropical and temperate regions by combining the novel traits or genes of both cultivars.

2. Materials and methods

2.1. Field experiments

Field experiments were conducted in the temperate region at the National Institute of Crop Science, Tsukubamirai (hereafter, Tsukuba), Japan (36°01′N, 140°02′E), in 2016 and 2017 and in the tropical region at IRRI, Los Ba?os, Philippines (14°17′N,121°26′E), in the dry seasons (DS) of 2017 and 2018. The soils were a Gleyic Fluvisol at Tsukuba and an Eutric Gleysol in Los Ba?os. Five rice cultivars: Takanari, Hokuriku193, IR64, NSIC Rc158,and YTH183(IR84636-13-12-2-6-3-3-2-2-B),were grown under continuously flooded conditions in four environments(two regions×two years).Takanari and Hokuriku193 are highyielding cultivars developed in Japan, and IR64, NSIC Rc158,and YTH183 are high-yielding cultivars developed by IRRI at Los Ba?os in the Philippines. IR64 showed wide adaptability and at its peak was cultivated in Asia and west Africa on ＞10 Mha[13].

Rice was grown by the conventional method in both Tsukuba and Los Ba?os. In Tsukuba, seeds were sown in seedling nursery boxes on April 22, 2016 and April 24, 2017 and seedlings were transplanted into the experimental paddy field at one seedling per hill on May 16,2016 and May 18,2017,respectively.The planting density was 22.2 hills m-2,with 15 cm between hills and 30 cm between rows. The experimental plots (9 m2each) were arranged in a randomized complete block design with four replicates. Basal fertilizer was applied 3 days before transplanting at 10.0 g N m-2as controlled-release fertilizer (3.3 g LP40 and 6.7 g LP100), 16.0 g P2O5m-2, and 12.0 g K2O m-2. LP40 and LP100 release 80%of their total N content at a uniform rate by 40 and 100 days,respectively,after application at 20-30 °C.At the panicle initiation stage, 5.0 g N m-2was topdressed as LP40. A total of 15.0 g N m-2was applied for the field experiments.

In Los Ba?os, seeds were sown in seedling nursery boxes on January 3,2017 and December 23,2017 and seedlings were transplanted into the experimental paddy field at one seedling per hill on January 24, 2017 and January 13, 2018,respectively. The planting density was 20 hills m-2, with 20 cm between hills and 25 cm between rows. The experimental plots (10.8 m2each) were arranged in a randomized complete block design with four replicates in the 2017 DS and five replicates in the 2018 DS. Basal fertilizer was applied before transplanting at 4.5 g N m-2, 4.5 g P2O5m-2, and 4.5 g K2O m-2as compound fertilizer. Ammonium sulfate was applied as topdressing at 2 weeks after transplanting(3.0 g N m-2), 4 weeks after transplanting (4.5 g N m-2), and on the day when the earliest cultivar reached heading(3.0 g N m-2). A total of 15.0 g N m-2was applied in the field experiments.

2.2. Measurement of biomass production, yield, and yield components

Eight to ten plants were sampled from each plot at panicle initiation stage, heading stage, and maturity. Heading stage was defined as the date when half of the panicles in each plot had emerged. Maturity was defined as the date when 95% of the spikelets had turned from green to yellow. Days to heading and days to maturity were determined as the number of days from sowing to heading and maturity, respectively.After the number of tillers (panicles) from each plant was recorded, soil and roots were removed and the entire shoots were dried at 80 °C for 72 h and weighed.

At maturity, plants covering 2.0 m2and 2.2 m2(40 plants and 48 plants) in Los Ba?os and Tsukuba, respectively, were harvested from each plot to determine yield and its components.The panicles were counted and then threshed to obtain unhulled grain, whose weight and moisture content were measured. Approximately three 40-g subsamples of grain(subsamples) were removed from the sample and counted with an electronic seed counter (WAVER IC-VAi, Aidex Co.Ltd., Aichi, Japan). Spikelet number m-2was calculated by multiplying grain number per unit weight in the subsamples by the total grain weight m-2.Spikelet number per panicle was calculated as spikelet number m-2divided by panicle number m-2. The grain subsamples were submerged in tap water and the grains that sank were separated as filled grains. These were oven-dried at 37 °C to constant weight, counted, and weighed and their moisture content was determined. The filled spikelet percentage was calculated as the number of filled grains divided by the whole number of grains in the subsamples. Single-grain weight was calculated by dividing filled grain weight by the number of filled grains. Grain yield was determined as the product of the components.Grain yield and single-grain weight were adjusted to 14% moisture content. Sink capacity was defined as the product of singlegrain weight and the number of spikelets m-2[11].

2.3. Evaluation of F2 plants derived from a cross between Takanari and YTH183

Takanari was crossed with YTH183 and F1seeds were obtained. F1plants from the F1seeds were grown and selfpollinated, and F2seeds were harvested. ＞100 F2plants (176 and 144 F2plants) derived from the parental cultivars Takanari and YTH183 were grown in paddy fields in Tsukuba and Los Ba?os in 2018 and 2018 wet season(WS),respectively.At maturity,the number of panicles per plant and the number of spikelets per panicle on the main stem were recorded for 100 F2plants and 10 parents except for border plants.

2.4. Statistical analysis

Statistical analyses were performed using a general linear model in SPSS 23.0 software (IBM, Chicago, IL, USA). Analysis of variance (ANOVA) was conducted to test the effects of genotype and environment on yield, its components, and biomass across four environments (two regions and two years). Genotype (G) and environment (E) and genotype by environment(G × E)were treated as fixed effects and replication as a random effect. Significant fixed effects (P ＜0.05)were analyzed using the Tukey-Kramer HSD test. Grain yield was analyzed using the genotype plus genotype × environment interaction (GGE) biplot method in PBTools software(IRRI, Los Ba?os, Philippines) to examine the G × E effect in depth. The broad-sense heritabilityof days to heading,days to maturity, grain yield, and biomass at maturity was calculated using the following equation.

where VGis the variance of G,VGEis the variance of G × E,Veis the variance of residuals, and E and r are the numbers of environments and of replications per environment, respectively. Variance components were estimated using PBTools software.

3. Results

3.1. Climate conditions

Daylength from sowing to heading was longer in Tsukuba(13.0 to 14.5 h)than in Los Ba?os(11.3 to 12.2 h)(Fig.1).Mean temperatures during the growth period were higher in Los Ba?os than in Tsukuba; they were constantly above 25 °C in Los Ba?os(Fig. 1-A),whereas they showed a gradual increase until heading in early August (14 °C to 27 °C) and then a gradual decrease during grain filling until maturity (27 °C to 20 °C) in Tsukuba (Fig. 1-B). Mean solar radiation during the growth period was 16.7 MJ m-2d-1in Tsukuba and 14.9 MJ m--2d-1in Los Ba?os. Solar radiation was higher at the vegetative stage than at the grain-filling stage in Tsukuba but higher at the grain-filling stage than at the vegetative stage in Los Ba?os.

3.2. Growth duration

Days to heading and days to maturity were significantly affected by environment and genotype (Table 1). Rice plants headed 20 to 26 days earlier in Los Ba?os than in Tsukuba and matured 44 to 47 days earlier in Los Ba?os than in Tsukuba. Takanari headed and matured earliest among the five cultivars, whereas NSIC Rc158 was the latest. There was a significant G × E interaction for days to heading and days to maturity. At IRRI in the 2017 DS, 12 days’ variation in days to heading was observed among the five cultivars, whereas there was only six days’ variation in Tsukuba in 2016 (Fig. S1). Theof days to heading and days to maturity were 89.6% and 84.2%, respectively.

3.3. Yield performance

Grain yield was significantly affected by the main effects of environment and genotype and their interaction (Table 1).Grain yields in Tsukuba were significantly higher than those in Los Ba?os. IR64 showed the lowest yield among the five cultivars. Hokuriku193 produced the highest grain yield in Tsukuba in 2017 but the lowest grain yield in Los Ba?os in the 2018 DS (Table 2). In contrast, NSIC Rc158 produced the highest grain yield in Los Ba?os but the lowest grain yield in Tsukuba in 2017. Takanari and YTH183 produced stably high grain yield and IR64 stably low grain yield throughout the experimental sites.Theof grain yield was 52%. To further investigate G × E interaction,the GGE biplot method was used to analyze grain yield. GGE biplot showed that the first principal component(PC1)corresponded to the adaptability of cultivars to the Tsukuba environments and the second component (PC2) to the adaptability of cultivars to the Los Ba?os environments (Fig. 2). Among the five cultivars,Takanari was the most adaptable and stable across the Tsukuba and Los Ba?os environments, followed by YTH183(Fig. 2). Hokuriku193 was suited to Tsukuba but not to Los Ba?os environments, whereas NSIC Rc158 was suited to Los Ba?os but not to Tsukuba environments. IR64 did not show adaptable performance in either environment.

Fig.1- -Daylength,mean temperature, and solar radiation at the experimental stations,Tsukuba,Japan(A),and IRRI,Los Ba?os,Philippines(B).DS,dry season.

Table 1--Days to heading,days to maturity,grain yield,and biomass at heading and maturity for five rice cultivars grown in Tsukuba and Los Ba?os.

Table 2-- Grain yield and biomass at heading and maturity for five cultivars grown at Tsukuba and Los Ba?os.

Biomass at heading and that at maturity were significantly affected by environment and genotype (Table 1). Biomass in Tsukuba was significantly higher than in Los Ba?os. Unlike grain yield, biomass at maturity did not show a significant effect of G × E interaction. Hokuriku193 consistently showed good biomass production at both heading and maturity for all environments(Table 2),scoring first in Tsukuba and second in Los Ba?os,although it adapted well only to Tsukuba and yield was low in Los Ba?os. The h2Bof biomass at heading and maturity were 86%and 83%,respectively.

3.4. Comparisons of yield components

Because GGE biplot analysis indicated the adaptability of the five cultivars to Tsukuba and/or Los Ba?os environments, we compared yield components between Tsukuba and Los Ba?os among the five cultivars. Among yield components, number of spikelets m-2and sink capacity were significantly correlated with grain yield(r ＞0.80)at both sites(Table 3).Takanari produced the largest panicles(the highest number of spikelets per panicle)in both Los Ba?os and Tsukuba,leading to a high number of spikelets m-2(Table 4). In contrast, YTH183 produced the largest number of panicles m-2at both sites,also leading to a high number of spikelets m-2. YTH183 also produced the highest single-grain weight at both sites. Thus,Takanari and YTH183 showed high sink capacity at both sites.Both cultivars reached ＞70% filled spikelets at the two experimental sites.

For all cultivars, sink capacity was markedly lower in Los Ba?os than in Tsukuba, owing mainly to the reduction in number of spikelets per panicle. This large difference in number of spikelets per panicle could be attributed to tiller size (Table 5); biomass per tiller at heading in Los Ba?os was 46%to 68%of that in Tsukuba,probably because of the shorter growth duration in Los Ba?os. These results suggest that the advantage of Takanari, which performs as a high-yielding cultivar via large panicle formation, could be potentially limited in tropical regions.

Fig. 2--GGE biplot environmental view of grain yield for five rice cultivars in four environments.DS:dry season.

Table 3--Correlation coefficients(Pearson's r)between grain yield and its components in Tsukuba and Los Ba?os.

3.5. Performance of F2 progeny of Takanari and YTH183

Because Takanari and YTH183 showed different characteristics of yield components but displayed high yield and adaptability in both Tsukuba and Los Ba?os,we developed F2progeny from crosses between Takanari and YTH183 and evaluated them in both regions.In the F2populations,number of panicles per plant and number of spikelets per panicle showed continuous distributions with transgressive segregation in both Tsukuba and Los Ba?os (Fig. 3). In Tsukuba,number of panicles per plant ranged from 10 to 21 and number of spikelets per panicle ranged from 111 to 253, so that some F2plants produced ＞3000 spikelets per plant while the parents produced fewer than 3000(Fig.3-A).In Los Ba?os,the number of panicles per plant ranged from 3 to 21 and the number of spikelets per panicle ranged from 81 to 231,so that some F2plants produced ＞2000 spikelets per plant while the parents produced fewer than 2000(Fig.3-B).

4. Discussion

Grain yield in Tsukuba, in the temperate region, was significantly higher than in Los Ba?os, in the tropics (Table 1).The difference in yield performance between Tsukuba and Los Ba?os could be attributed mainly to growth duration(vegetative period).Longer growth duration resulted in higher biomass production in Tsukuba than in Los Ba?os, probably because of the long daylength and the low temperature in Tsukuba (Fig. 1). High biomass production is recognized as a key trait for increasing yield potential in rice[14-16].However,high biomass production at maturity did not always result in high yield; grain yield in Tsukuba was lower in 2017 than in 2016 despite higher biomass at maturity in 2017 than in 2016.This inconsistency could be explained partly by the fact that biomass at heading in each cultivar in Tsukuba in 2016 was higher than that in 2017 (Tables 1 and 2). Recent studies[17-19] indicate the importance of crop growth two weeks before heading or booting stage to attaining high sink size as well as high non-structural carbohydrate(NSC)accumulation.Although NSC accumulation was not recorded in this study,sink size expressed as number of spikelets m-2in Tsukuba in 2016 tended to be high compared with that in 2017(Table S1).Another possible explanation of the inconsistency is that environmental factors such as low temperature caused low spikelet fertility in some cultivars during grain filling. This speculation is supported by the relatively low temperature between heading and maturity in Tsukuba in 2017 compared with that in 2016 (Fig. 1) and lower spikelet fertility in 2017 than in 2016 (Table S1). Our results indicate that both high biomass production at heading and sufficient grain filling are essential for increasing yield potential. Interestingly,Hokuriku193 had the highest biomass production among the five cultivars at both heading and maturity.This stable trait in Hokuriku193 may be useful for future high-yield breeding programs in tropical regions where short growth duration restricts biomass production,although its yield potential was low owing to its small sink capacity.

There was a significant G × E interaction for grain yield in this study, and the GGE biplot method classified the five cultivars into four categories: Takanari and YTH183 showedhigh yield performance and high adaptability in both tropical and temperate regions,Hokuriku193 was suitable for temperate regions, NSIC Rc158 was suitable for the tropics, and IR64 was inferior to the other cultivars in both regions (Fig. 2).These results indicate that the recently developed modern cultivars are superior to older cultivars such as IR64 with respect to yield potential. It is interesting that Takanari and YTH183 showed high grain yield and adaptability even though they showed different characteristics in yield components.Takanari produced a higher number of spikelets per panicle than the other cultivars(and thus was a panicle-weight type)whereas YTH183 produced a higher number of panicles m-2than the other cultivars(and thus was a panicle-number type)(Table 4). Previous studies revealed that Takanari produced large panicles because of the favorable alleles of two quantitative trait loci (QTL), GN1a and APO1 [20], as well as producing high leaf photosynthesis via GPS expression[21].In previous studies, qSW5/GW5 increased grain size [8] and qRL6.4 elongated roots in YTH183 [22]. Given that YTH183 showed deeper rooting in a previous study [8], YTH183 may also carry the functional allele of DRO1 that controls root angle[23].These different genetic factors may account for the different plant types of Takanari and YTH183. Still, both cultivars attained large sink capacity with good grain filling and thus high grain yield with high stability and adaptability.The importance of increased sink capacity for increasing yield potential has been recognized even in the recently developed super hybrid rice [24-26]. In this context, Hokuriku193 may not have adapted to the tropics and NSIC Rc158 may not have adapted to the temperate region because sufficient sink capacity was not produced in each cultivar in each environment.

Table 4--Mean yield components and sink capacity for five rice cultivars in two years in two environments,Los Ba?os and Tsukuba.

Table 5--Mean biomass per tiller at heading for five rice cultivars in two years in two environments, Los Ba?os and Tsukuba.

However, even though Takanari produced the largest panicles among the five cultivars, it inevitably showed a reduced number of spikelets per panicle in the tropical relative to those in the temperate region (Table 4), probably because of the short vegetative period, which limited the growth of tillers and thus resulted in low biomass per tiller(Table 5). For this reason, we attempted to develop cultivars with larger sink size than Takanari and YTH183 in both tropical and temperate regions using the F2population derived from the cross between Takanari and YTH183. Some F2plants produced more spikelets per panicle and/or more panicles per plant than the parental cultivars,leading to more spikelets per plant (larger sink size) than the parents even in the tropical location (Fig. 3). These results suggest that the accumulation of known and unknown genetic factors from Takanari and YTH183 may enable us to develop cultivars that produce expanded sink capacity in both tropical and temperate regions. However, it should be noted that there are limitations in testing F2populations because data is collected from single plants, which can be tested in only one season.Additional studies using advanced-generation lines such as recombinant inbred lines, whose genotypes are fixed, are necessary to verify the results of the present study. Our results also suggest that a further combination of indica and japonica genomes is desirable for enhancing the yield potential of rice plants, given that both Takanari and YTH183 carry indica and japonica genomic material derived from different sources.

Fig. 3 --Relationship between number of panicles per plant and number of spikelets per panicle in Takanari, YTH183, and 100 F2 plants derived from crosses between Takanari and YTH183 in Tsukuba in 2018 (A) and in Los Ba?os in the 2018 wet season (WS) (B).

This study indicates the possibility of developing highyielding rice cultivars in both tropical and temperate regions by enhancing sink capacity.Because large sink capacity often results in a reduction in grain filling, the genetic factors for high biomass production in Hokuriku193 may be useful to maintain sufficient grain filling. We are currently advancing the generation of the breeding population toward the selection of lines with good grain filling as well as large sink capacity.

5. Conclusions

Although rice breeding is conducted worldwide, few trials have been conducted to compare tropical rice with temperate rice in both tropical and temperate regions. This study revealed high grain yield with high adaptability in a temperate rice cultivar,Takanari,and a tropical rice cultivar,YTH183.Although the two cultivars had different plant type, both showed large sink capacity with sufficient grain filling and thus high grain yield. F2plants derived from a cross between the two cultivars produced larger sink sizes than the parents.Future breeding programs using these rice materials should advance the development of high-yielding rice cultivars adapted to both tropical and temperate environments.

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2019.06.004.

Acknowledgments

We thank the research technicians and contract workers for their research support at International Rice Research Institute(IRRI) and National Agriculture and Food Research Organization (NARO). This study was financially supported by the Japan International Research Center for Agricultural Sciences-International Rice Research Institute (JIRCAS-IRRI) collaborative breeding project and by a grant from the Institute of Crop Science,NARO,Japan.