Comparison of yield performance between direct-seeded and transplanted double-season rice using ultrashort-duration varieties in central China

Le Xu, Shen Yuan, Xinyu Wang, Zhifeng Chen, Xiaoxiao Li, Jing Cao, Fei Wang, Jianliang Huang,Shaobing Peng

National Key Laboratory of Crop Genetic Improvement, Hubei Hongshan Laboratory, MARA Key Laboratory of Crop Ecophysiology and Farming System in the Middle Reaches of the Yangtze River, College of Plant Science and Technology, Huazhong Agricultural University, Wuhan 430070, Hubei, China

Keywords:Direct seeding Double-season rice Grain yield Nitrogen use efficiency Ultrashort-duration variety

ABSTRACT Labor scarcity requires double-season rice to be planted by direct seeding as an alternative to transplanting. Only ultrashort-duration varieties can be used in direct-seeded, double-season rice (DSD) in central China where thermal time is limited.Whether ultrashort-duration varieties grown in DSD can be as productive and efficient in nitrogen(N)use as transplanted double-season rice(TPD)remains unclear.Field experiments were conducted in Hubei province, central China with two establishment methods (DSD,TPD) and three N rates in the early and late seasons of 2017 and 2018. Nitrogen treatments included zero-N control (N0), total N rate of 60 kg N ha-1 with equal splits at basal, midtillering, and panicle initiation (N1), and weekly N application at 15 kg ha-1 from seeding/transplanting to heading (N2). Both early-and late-season rice under DSD matured within 95 days,on average 9 days shorter than rice under TPD. The grain yield of DSD was comparable to or higher than that of TDP in both seasons, although the daily yield was significantly higher under DSD than under TDP.Before heading,DSD had higher leaf area,stem number, intercepted radiation, and radiation use efficiency than TPD, which compensated for the negative effect of short growth duration on biomass production. Total dry weight and harvest index under DSD were comparable to or higher than those under TDP. In general, the recovery efficiency of fertilizer-N under DSD was higher than that under TPD, but the reverse was true for physiological N use efficiency. Thus, there was no significant difference in agronomic N use efficiency between DSD and TPD. These results suggested that DSD with ultrashort-duration varieties is a promising alternative to TPD in central China for maintaining high grain yield and N fertilizer use efficiency with less labor input.

1. Introduction

Rice is the staple food for more than 65% of the Chinese population [1]. With growing food demand, rice production must increase at the rate of 2 million tons per year to ensure food security[2].Rice production is determined by crop yield and harvested area, the latter depending on existing cropland and harvest frequency (the frequency of crop harvest per year on the same piece of cropland). China is a country with a shortage of cultivated land whose cropland area per capita is only 43% of the world average[3].It is not feasible to expand cropland by clearing natural ecosystems, which presents environmental drawbacks including biodiversity loss and greenhouse gas emissions [4]. However, on-farm rice yield is approaching its yield potential ceiling[5].It will be difficult to further increase rice yield unless there is marked genetic gain in yield potential[6].One promising strategy for meeting food demand is to increase harvest frequency by multiple cropping on existing cropland [4,7].

Double-season rice that allows two harvests per year is an effective measure to ensure food security of China [8]. Double-season rice is commonly grown in central China where early- and lateseason rice crops are manually transplanted into the puddled field[9]. However,about 2.0 million hectares of double-season rice has been changed to single-season rice from 1998 to 2016,leading to a decline in rice harvested area and production [10]. This change occurred because the high labor input reduced the profit of rice farming, discouraging farmers from growing transplanted double-season rice(TPD)[11].Moreover,China is the country with the lowest water availability per capita in Asia, and depletion of water resources also threatens the sustainability of TPD [12]. It is thus vital to develop labor-and water-efficient double-season rice cropping systems.

In TPD,massive labor and water resources are consumed in the process of seedling raising and transplanting twice a year [13].Direct seeding is an alternative establishment method, which requires less water and labor to gain more economic profit compared with transplanting [11,14]. In recent years, there has been a shift from transplanting to direct seeding in many countries worldwide [15]. Many researchers have proposed that labor and water input would be greatly reduced by direct seeding of double-season rice [11]. Since the 1970 s, there have been a few studies on direct-seeded,double-season rice(DSD)in central China[11,16]. However, DSD has seen very limited adoption, because ultrashort-duration varieties with high yield potential are unavailable in central China,where only 190 to 209 days in a year are suitable for rice growth [17]. Because DSD omits two seedbed periods compared to TPD, ultrashort-duration varieties that mature within approximately 95 days are required for both early and late seasons [14].

Shortening growth duration for planting DSD would negatively affect rice yield because of a substantial reduction in total incident solar radiation during the growing season [18]. For this reason,researchers have focused mostly on breeding and optimizing crop management practices for varieties with medium or long growth duration,rather than for ultrashort-duration varieties.For reducing labor and water inputs in double-season rice production in central China, we have conducted pot and field experiments [14,19] to select suitable ultrashort-duration varieties for DSD and identified plant traits associated with yield.However,a shift from TPD to DSD with ultrashort-duration varieties may raise concerns from farmers and policy makers about the potential risk of yield loss.

Nitrogen(N) is the most essential nutrient element that affects rice growth and yield formation [20]. Ultrashort-duration variety has limited time to absorb N, possibly leading to low grain yield.This disadvantage could be overcome with high N input, but that may result in low N use efficiency (NUE) due to rapid N loss[21,22]. Nitrogen management strategies for achieving both high yield and high NUE simultaneously have been extensively studied for medium- and long-duration varieties [23,24]. The key to optimization of N management is to ensure the synchronization of in-season N supply with N demand of the crop during its growth[25].However,there is limited information on the maximum grain yield and its N requirement for ultrashort-duration varieties.Moreover, no study has been conducted to determine whether the ultrashort-duration varieties grown under DSD can be as productive as that in TPD. Therefore, two-year field experiments were conducted in central China to (1) compare the yield performance between DSD and TPD using ultrashort-duration varieties, and(2) evaluate N uptake and NUE of ultrashort-duration varieties grown under DSD and TPD.

2. Materials and methods

2.1. Experimental sites

Field experiments were conducted in the subtropical environment of Wuxue county, Hubei province, China (29°51′N,115°33′E) in 2017 and 2018. Wuxue county is a typical doubleseason rice growing region in central China. Soil samples were taken from the upper 20 cm layer before the application of basal fertilizers in the early season of 2017.The soil was a clay loam with a pH of 5.13, organic matter of 31.0 g kg-1, total N of 2.39 g kg-1,Olsen phosphorus of 54.3 mg kg-1,and exchangeable potassium of 140.7 mg kg-1.In both years,climate data including daily radiation and air temperature during the growing period were collected from a meteorological station (WS 800, Campbell Scientific, Inc.,Logan,UT,USA),about 1.5 km distance from the experimental site.Annual accumulated effective temperature (≥10 °C) at the experiment site was 3523 °C averaged across the two years, a value within the range typical for central China [26]. Detailed climate information is presented in Fig. S1.

2.2. Experimental design and crop management

Experiments were arranged in a split plot design with establishment methods (DSD and TPD) as the main plot, and N rates as subplot with four replications. The subplot size was 25 m2(5 × 5 m). The ultrashort-duration varieties Xianzhaoxian 6(XZX6) and Zaoxian 615 (ZX615), were used in the early and late seasons, respectively. These two varieties are indica inbreds that were bred for transplanted early-season rice and could also be used in transplanted late-season rice [27,28]. In our previous experiments, these two varieties were identified as suitable for DSD owing to their short growth duration and good yield performance [14]. For both early and late seasons, N treatments included zero-N control (N0), total N rate of 60 kg ha-1with equal splits at basal, midtillering, and panicle initiation (N1),and weekly N application at 15 kg ha-1starting at basal until heading (N2). Basal fertilizer application was done one day before seeding and transplanting for DSD and TPD, respectively. Because optimal N management practice was unknown for ultrashortduration varieties grown under the conditions at the experimental site, weekly N application was included to ensure sufficient N supply for both DSD and TPD. In N2 of DSD, the second N application was omitted because the field was kept aerobic to promote seed germination and N topdressing could not be applied without standing water. Detailed information about N application is presented in Table S1. All N fertilizers were applied in the form of urea. To minimize seepage between plots, all bunds of plots were covered with plastic film installed into a depth of 20 cm below the soil surface.

For DSD,germinated seeds were manually sown in each plot at a rate of 9 g seeds m-2in the early season and 7 g seeds m-2in the late season.Seeding was done on April 12 and July 21 in 2017,and on April 8 and July 19 in 2018 for the early and late seasons,respectively. For TPD, pre-germinated seeds were sown in a seedbed with the sowing date of March 25 and July 3 in 2017,and March 25 and July 1 in 2018 for the early and late seasons,respectively. In both years, 25- and 20-day-old seedlings were transplanted for the early and late seasons, respectively at a hill spacing of 13.3 × 20 cm with three seedlings per hill. The field and the treatment arrangement within the field were the same for the two seasons in both years. Phosphorus and potassium fertilizer management were the same for all treatments in both seasons and years. Phosphorus as single superphosphate (31 kg P ha-1)was applied at the basal application.Potassium as potassium chloride was applied at the basal application (37 kg K ha-1) and panicle initiation (56 kg K ha-1). The plots for DSD were flooded after the three-leaf stage, and a floodwater depth of 3–5 cm was maintained until one week before maturity except a drainage at maximum tillering stage to reduce unproductive tillers. For TPD,the plots were kept flooded from transplanting to one week before maturity.Weeds,pests,and diseases were intensively controlled to avoid yield loss.

2.3. Observations and measurements

Dates of sowing, transplanting, panicle initiation, heading, and maturity were recorded for determining growth duration. Total growth duration refers to the period from seeding to maturity.Thirty seedlings for TPD were sampled at transplanting to measure seedling biomass.Twelve hills for TPD and 0.5 m2plants for DSD in each plot were sampled for growth analysis at panicle initiation,heading,and maturity.After stem number(main stems plus tillers)and panicles (when present) were recorded, the plant samples were separated into leaves,stems(culm plus sheath)and panicles.Green leaf area was measured with a leaf area meter (LI-3000, LICOR Inc., Lincoln, NE, USA) to calculate leaf area index (LAI). Leaf area growth rate (LAGR) was calculated based on changes in leaf area per unit ground area during vegetative stage.The dry weights of each organ were recorded after oven-drying at 70°C to constant weight, and then aboveground total dry weight (TDW) was calculated.

Canopy light interception was measured during the growing seasons in both 2017 and 2018. Measurements were made between 1100 and 1300 with an interval of 7–12 days using a line ceptometer (AccuPAR LP-80, Decagon Devices Inc., Pullman, WA,USA). Incoming light intensity was recorded by placing the light bar above the canopy. Immediately afterward, the light bar was placed slightly above the water surface to record light intensity inside the canopy. Canopy light interception was calculated as the percentage of incoming light intensity that was intercepted by the canopy. Intercepted radiation during a growth period was calculated using the mean canopy light interception and cumulative incoming solar radiation during the same growth period [(-canopy light interception at the beginning of the growth period + canopy light interception at the end of the growth period)/2×cumulative incoming radiation during the growth period]. Radiation use efficiency (RUE) was calculated as the ratio of TDW to intercepted radiation during a growth period.

At maturity, the panicles were hand-threshed, and filled spikelets were separated from unfilled spikelets by submerging them into tap water.Empty spikelets were separated from half-filled spikelets by winnowing. Three subsamples of filled (30 g), half-filled(6 g) and empty (3 g) spikelets were taken to count spikelets.The dry weights of the rachis and of filled, half-filled and empty spikelets were measured after oven drying at 70 °C to constant weight. The TDW at maturity was the sum of the dry weights of the straw (leaves plus stems), rachis, and filled, half-filled, and empty spikelets. Spikelets per panicle (spikelets m-2/panicles m-2), grain filling percentage (100 × filled spikelets m-2/spikelets m m-2), and harvest index (100 × filled spikelet weight/TDW)were calculated.The grain yield was determined from a 5 m2area in the center of each plot and adjusted to 14% moisture content.The grain moisture content was determined by a digital moisture tester (DMC-700, Seedburo, Chicago, IL, USA). Daily yield was calculated as grain yield divided by total growth duration in days.

Nitrogen concentration of each organ at maturity was determined by Elementar Vario MAX CNS/CN (Elementar Trading Co.,Ltd., Germany). Nitrogen content of each organ was calculated as the product of N concentration and dry weight. Nitrogen uptake at maturity was the sum of N contents for each organ.Fertilizer N recovery use efficiency (REN), physiological use efficiency(PEN), and agronomic use efficiency (AEN) were calculated as follows:

where TN+Nand TN-Nare total N uptake in plots receiving N fertilizer and in the zero-N control, respectively; FN is the amount of N fertilizer applied; GY+Nand GY-Nare grain yields in plots receiving N fertilizer and in the zero-N control, respectively.

2.4. Statistical analysis

Statistical data analysis was performed using analysis of variance (Statistix 8.0, Analytical Software, Tallahassee, FL, USA), and the means of treatments were compared based on the least significant difference (LSD) test at the 0.05 probability level.

3. Results

The mean daily solar radiation and temperature during doubleseason rice growing period were 14.7 and 14.2 MJ m-2d-1, 24.2 and 24.8 °C in 2017 and 2018, respectively. Daily mean temperature showed an increasing trend in the early seasons but a decreasing trend in the late seasons from planting to maturity (Fig. S1).The temperature during vegetative stage in the late season was higher than that in the early season,whereas the temperature during ripening stage in the late season was lower than that in the early season. The temperature during reproductive stage differed relatively little between the two seasons.

The crop under DSD matured within 95 days,whereas the total growth duration of TPD ranged from 95 to 110 days across seasons and years (Table 1). On average, the total growth duration of DSD was 9 days less than that of TPD as a consequence of shorter vegetative and ripening stages. No consistent difference in total growth duration was observed between the early and late season.On average, early-season rice had a 16-days-longer growth duration in vegetative stage but a 19-days-shorter growth duration in ripening stage than did late-season rice. A small difference in the duration of reproductive stage was observed between the two seasons.

The grain yield of DSD was significantly higher than that of TDP in 2017, but there was no significant difference in grain yield between the establishment methods in 2018 (Tables 2, S2, and S3). Grain yield in 2018 was 0.22 and 0.95 t ha-1higher than that in 2017 for the early-and late-season rice,respectively.Daily yield of DSD was significantly higher than that of TPD in both years except for N1 in 2018. High N application rates significantly and consistently increased yield, daily yield, and annual yield. The maximum annual yield of 15.7 t ha-1was produced by DSD under N2 in 2018.For yield components,DSD significantly increased panicle number,but reduced spikelets per panicle compared with TPD(Tables 3, S2 and S3). No consistent difference in grain filling percentage and grain weight were observed between the establishment methods. High N application rate increased panicle number and spikelets per panicle,but did not affect grain filling percentage and grain weight.

For most cases, DSD produced significantly higher TDW than TPD except for the late season of 2018(Table 3).The harvest index of DSD was higher than that of TPD, but the difference was significant in only 3 of 12 comparisons. High N application rates increased TDW, but had no significant effect on harvest index(Tables S2 and S3). The variation in grain yield was attributed mainly to TDW rather than to harvest index (Fig. 1). The higher biomass accumulation before heading was responsible for the higher TDW of DSD compared to TPD (Fig. 2). As shown in Fig. 3,the biomass accumulation from sowing to panicle initiation was closely related to LAGR,and the LAGR of DSD was higher than that of TPD in both early and late seasons.In general,DSD increased LAIand stem number at panicle initiation and heading compared to TPD (Table S4). As a consequence, canopy intercepted radiation before heading of DSD was on average 34% higher than that of TPD (Table 4). RUE before heading under DSD also increased by 16.6% compared to TDP. High canopy intercepted radiation and RUE were responsible for higher biomass accumulation of DSD over TDP before heading. After heading, intercepted radiation of DSD was comparable to or higher than that of TDP, and there was no consistent difference in RUE between DSD and TDP. Over the entire growth period, DSD consistently increased total intercepted radiation compared to TDP, but the establishment method showed an inconsistent effect on RUE.

Table 1 Growth duration (days) of N treatments and establishment methods in the early (ES) and late (LS) seasons of 2017 and 2018.

Fig.1. Relationship of grain yield with aboveground total dry weight(TDW)(a)and harvest index(HI)(b)at maturity.Data were pooled from experiments conducted in the early and late seasons of 2017 and 2018.

Plant N uptake at maturity in fertilized plots was increased significantly by DSD compared to TPD in 7 of 8 comparisons(Table 5).But no significant difference in plant N uptake was observed between establishment methods under N0. As consequence, RENof DSD was generally higher than that of TDP except for the late season of 2017. On average across N rates, seasons, and years,RENof DSD was 15.6% higher than that of TDP. In contrast, DSD showed lower or comparable PENcompared to TDP. There was no significant difference in AENbetween the establishment methods except for N2 in the early season of 2018. High N application rates increased total N uptake, but had an inconsistent effect on REN, PEN, and AEN.

4. Discussion

Our experimental results demonstrated that the change of establishment method from transplanting to direct-seeding did not lead to any yield penalty for ultrashort-duration varieties grown in the double-season rice system. Overall, the yield performance of DSD was comparable to or even better than that of TPD.Many researchers have demonstrated that direct-seeded rice can be as productive as transplanted rice when proper crop management is practiced including optimum weed and water management [29,30]. When DSD outyielded TDP, the yield difference was associated with panicle number. Liu [31] reported that the increased panicle number per m2was mainly responsible for the similar or even higher yield of direct-seeded rice than transplanted rice.A maximum grain yield of 7.88 t ha-1was produced in DSD at high N rate(Table 2).This yield is comparable with previous studies in which double-season rice yield in the same region ranged from 7 to 9 t ha-1under optimum crop management and using medium-duration varieties [24,32].

Fig.2. Dynamics of aboveground total dry weight(TDW)of N treatments and establishment methods in the early and late seasons of 2017 and 2018.N0,zero N control;N1,60 kg N ha-1;N2,weekly 15 kg N ha-1 applied from direct seeding or transplanting to heading;DSD,direct-seeded,double-season rice;TPD,transplanted double-season rice;SS-PI,from sowing to panicle initiation;PI-HD,from panicle initiation to heading;HD-PM,from heading to maturity.Vertical bars represent±SD of the mean.Different lower cases above columns indicate significant differences between DSD and TPD within the same N treatment according to the LSD0.05.

Fig.3. Relationship between leaf area growth rate(LAGR)and biomass accumulation from sowing to panicle initiation(SS-PI)in the early(a)and late seasons(b)of 2017 and 2018. DSD, direct-seeded, double-season rice; TPD, transplanted double-season rice.

Table 2 Yield, daily yield, and annual yield of N treatments and establishment methods in the early (ES) and late (LS) seasons of 2017 and 2018.

Rice grain yield is also the product of TDW at maturity and harvest index [18]. The yield variation in this study was mainly explained by TDW but not harvest index (Fig. 1). Harvest index of ultrashort-duration varieties generally exceeded 0.50, especially under DSD. Although the ultrashort-duration varieties under DSD produced more dry matter in stems at heading, there was inconsistent difference in the translocation of dry weight from straw to grain between DSD and TPD (Table S5). These results support the observation of insignificant difference in harvest index between the two crop establishment methods.It is difficult to further improve harvest index owing to biological limitations [33]. Khush [34] stated that there was little scope to further increase rice harvest index after the Green Revolution,and the improvement of rice yield in recent decades have depended mainly on increasing biomass production.Total growth duration of DSD was 9 days less than that of TPD. Reduction in total growth duration would reduce the total amount of incident solar radiation for producing biomass through canopy photosynthesis [35]. However, the biomass production of DSD was compensated by rapid leaf area growth rate at early vegetative stage and high stem number and LAI before heading,which contributed to higher intercepted radiation before heading compared to TPD(Fig. 3; Tables 4 and S4). Besides, DSD had 16% higher RUE than TDP before heading. Therefore, DSD had higher biomass production before heading and higher TDW at maturity than TPD(Fig.2).The importance of early vigor characters on TDW was also reported by Sinclair and Laza [36,37]. Considering the limited availability of ultrashort-duration varieties at present, it is suggested that rice breeders pay close attention to these early vigor characters for developing ultrashort-duration varieties with improved the yield performance and N use efficiency.

Grain yield was generally higher in 2018 than in 2017 especially in the late season. The variation in grain yield between two years was largely attributed to the variation in climatic conditions. Grain yield of late-season rice in 2017 was 14.1% lower than in 2018 owing to the reduction in grain filling percentage(Table 3). During the grain filling period, daily solar radiation in 2017 was 2.5 MJ m-2lower than in 2018. Wang and Yin [38,39]found that low solar radiation during grain filling period could result in a sharp decrease in percentage of ripened grains.

The overuse of N fertilizer for pursuing high yield has been a major problem in rice production that caused widespread environmental damage in China over the past decades [22,40].Improving RENfrom applied N has been proposed as an important approach to reduce N loss and protect the environment [21]. In this study, RENwas generally above 50% in fertilized treatments(Table 5), which is relatively high compared with typical values of 30%–50% for irrigated rice in Asia [21,41]. More importantly,DSD increased RENby 15.6% compared to TDP. A similar finding was also reported by De Datta [42] that RENof direct-seeded rice was significantly higher than that of transplanted rice. Nitrogen fertilizer loss from rice field occurs primarily during early vegetative stage when high N rate is applied as basal fertilizer and the immature root system has limited ability to absorb N [43]. The high RENin DSD might be attributed to its early vigor characters expressed as rapid leaf area expansion and tiller production compared to TDP (Fig. 3; Table S4).

?

?

Grain yields of DSD with ultrashort-duration varieties were over 7.0 t ha-1in the weekly N application treatment where total N input was 105 to 135 kg N ha-1(Tables 2 and S1). This N rate was much lower than the actual N input practiced by farmers for double-season rice in central China[22].In Hunan and Hubei,it was reported that farmers applied 160 kg N ha-1and produced about 7.0 t ha-1grain yield in transplanted double-season rice with medium-duration varieties [24,44]. The grain yield under weekly N application approached the yield potential of ultrashort-duration varieties in this study.This finding is in agreement with the maximum yield of 7.7 t ha-1in ultrashort-duration varieties in our previous N experiment [45], in which N rates ranged from 0 to 220 kg ha-1. The soil organic matter and total N in our experimental field were 31.0 g kg-1and 2.39 g kg-1, respectively. Xu [46] took 2707 measurements for soil property of double-season rice fields across central China and reported 36.9 g kg-1of average organic matter and 2.85 g kg-1of average total N. These values suggested that our experimental field is representative for most rice paddies in central China in terms of indigenous soil N. Thus, the potential for reduction of N fertilizer input and achieving high yield is feasible for DSD using ultrashort-duration rice varieties.For DSD,reducing N input is also an important management practice to mitigate the risk of lodging,pest, and disease under high plant density. Although weekly N application can ensure the expression of potential yield in DSD with high NUE (Tables 2 and 5), this N management increases the complexity in practice and is not easy to be adopted by farmers because of the high labor requirement[47].Slow-released N fertilizers, such as sulfur- and resin-coated urea, might be alternatives to improve N use efficiency and minimize the N losses[48].Further research on DSD should be conducted to optimize N management for achieving high grain yield and NUE with less labor input.

Averaged across the two years, an annual yield of 15.1 t ha-1under DSD was achieved within 188 days using ultrashortduration rice varieties, which was 9.5% higher than that under TDP(Tables 1 and 2).Moreover,this yield level is much higher than that of single-season rice in the same region, implying the important role of double-season rice in the national food security [49].Given the limitation in thermal time, DSD using ultrashortduration varieties can be practiced safely in most rice growing regions of central China [17]. By practicing DSD, labor cost and water input can be greatly reduced,which would revive the interest of farmers in planting double-season rice.Therefore,our results suggested that DSD with ultrashort-duration varieties is a promising alternative to TPD in central China for maintaining high grain yield and N fertilizer use efficiency with less labor input.

Table 5 Total nitrogen uptake,recovery use efficiency(REN),physiological use efficiency(PEN),and agronomic use efficiency(AEN)of N treatments and establishment methods in the early(ES) and late (LS) seasons of 2017 and 2018.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Acknowledgments

This work was supported by the National Natural Science Foundation of China(31971845,32061143038), the China Postdoctoral Science Foundation (2021 M691179), the China Agriculture Research System(CARS-01-20),the Program of Introducing Talents of Discipline to Universities in China(the 111 Project no.B14032),the Program for Changjiang Scholars and Innovative Research Team in University of China (IRT1247), and a grant from the Bill and Melinda Gates Foundation (OPP51587).

Appendix A. Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2021.07.003.