Song Chen*,Showen Liu1,Xi Zheng,Min YinGung ChuChunmei XuJinxing YnLiping ChenDnying WngXiufu Zhng*

aChina National Rice Research Institute,Chinese Academy of Agricultural Sciences,Hangzhou 310006,Zhejiang,China

bFaculty of Agriculture,Life and Environmental Sciences,Zhejiang University,Hangzhou 310029,Zhejiang,China

Keywords:Rice(Oryza sativa L.)Paddy–upland rotation Nitrogen use efficiency Winter crops

ABSTRACT To evaluate the effects of various rotation systems on rice grain yield and N use efficiency,a paddy–upland cropping experiment(2013–2016)was conducted in southeastern China.The experiment was designed using six different rice––winter crop rotations:rice–fallow(RF),rice–wheat(RW),rice–potato with rice straw mulch(RP),rice–green manure(Chinese milk vetch;RC–G),rice–oilseed rape(RO),and rice–green manure crop(oilseed rape with fresh straw incorporated into soil at flowering;RO–G)and three N rates,N0(0 kg N ha-1),N1(142.5 kg N ha-1),and N2(202.5 kg N ha-1).Average rice yields in the RF(5.93 t ha-1)rotation were significantly lower than those in the rotations with winter crops(7.20–7.48 t ha-1)under the N0 treatment,suggesting that incorporation of straw might be more effective for increasing soil N than winter fallow.The rice yield differences among the rotations varied by year with the N input.In general,the grain yields in the RP and RO–G rotations –were respectively 11.6–28.5%and 14.80–37.19%higher than those in the RF in plots with N applied.Increasing the N rate may have tended to minimize the average yield gap between the RF and the other rotations;the yield gaps were 18.55%,4.14%,and 0.23%in N0,N1,and N2,respectively.However,the N recovery efficiency in the RF was significantly lower than that in other rotations,except for 2015 under both N1 and N2 rates,a finding that implies a large amount of chemical N loss.No significant differences in nitrogen agronomic efficiency(NAE)and physiological efficiency(NPE)were found between the rotations with legume(RC–G)and non–legume(RO and RW)winter crops,a result that may be due partly to straw incorporation.For this reason,we concluded that the return of straw could reduce differences in N use efficiency between rotations with and without legume crops.The degree of synchrony between the crop N demand and the N supply was evaluated by comparison of nitrogen balance degree(NBD)values.The NBD values in the RP and RW were significantly lower than those in the other rotations under both N1 and N2 rates.Thus,in view of the higher grain yield in the RP compared to the RW under the N1 rate,the RP rotation might be a promising practice with comparable grain yield and greater N use efficiency under reduced N input relative to the other rotations.The primary yield components of the RF and RP were identified as number of panicles m-2and numbers of kernels panicle-1,respectively.The NAE and NPE were positively correlated with harvest index,possibly providing a useful indicator for evaluating N use efficiency.

1.Introduction

Riceisaprimarystaplefoodin Asia,accountingfor respectively 47.8%and 38.5%of the production and planting areas of cereal crops in 2011[1].For several decades,farmers in China have generally applied large amounts of chemical fertilizers to achieve high rice yields.For example,farmers in Jiangsuprovincegenerallyapplyasmuchas259.5to 292.5 kg N ha-1for a single planting of rice[2],although the rice plants consume＜20%of the N applied[3].With the increase in the use of chemical N fertilizers,Nitrogen agronomic efficiency(NAE)has decreased from 15 to 20 kg kg--1(1958–1963)to 9.1 kg kg-1(1981–1989)[4].In addition,the overuse of chemical fertilizers wastes resources and results in environmental pollution[5–7].For these reasons,China currently faces the great challenges of achieving agricultural sustainability while ensuring food security and environmental health.

Increasing the diversity of crop rotations is the predominant approach to overcoming the negative aspects of short cropping sequences and monocultures in intensive cropping systems[8].Rotational cultivation by double cropping,which consists of paddy rice(Oryza sativa L.)and upland crops,has been practiced for years as a primary form of land management to maintain rice production in southeastern China[9,10].In a paddy–upland rotation system,paddy rice is generally cultivated in the summer(late June to mid-October)followed by upland crops in the winter(late October to early June of the next year).Various upland crops,such as Chinese milk vetch(Astragalus sinicus L.),oilseed rape(Brassica napus L.),wheat(Triticum aestivum L.),and potato(Solanum tuberosum L.),are cultivated in the drained paddy fields in the winter[4,11–13,40]for various purposes or benefits,such as the maximum use of land[15].However,these crops generally result in changes in soil fertility,owing to the incorporation of crop residue[16],application of chemical fertilizers[17],or N fixation from the atmosphere[12].To be sustainable,intensive agriculture systems based on summer rice need to reach close synchrony between the amount of N available after the winter crop harvest and the rice N requirements,in order to avoid N losses from leaching or denitrification[18].Accordingly,there is a need to evaluate the efficiency of rice N use under different rotation treatments.Such efficiency involves chemical N application,the existing soil N supply,and the N requirements of the plants.

The soil N supply and the rice N demand rely on the internal cycling of N within the system.First,the winter crops chosen for the rotation are the key to synchronizing soil N supply with rice N demand[8,11,19].However,studies on individual sites generally show a wide range of effects on rice N use efficiency.Ockerby et al.[17]investigated fallow,cereal,and legume cropping systems in Australia and reported that prior crops either increased or decreased paddy rice yield and altered its response to N fertilizers.Second,the incorporation of straw also affects crop yield or N use efficiency.Su et al.[20]suggested that rice straw mulching has the potential to increase the productivity of winter oilseed rape.Wang et al.[21]reported some trials with winter wheat in northern China in which straw return reduced the yield by 0.6%–7.1%.Thus,the N available for plant requirements in various cropping systems might be altered by the type,quantity,and quality of residue or straw that is incorporated into the soil[9].Third,the rice physiological response to the rotation types also affects the plant N demand.Higher nitrogen physiological efficiency(NPE)is generally accompanied by more kernels panicle-1[22],higher harvest index(HI)[23],or greater leaf area index(LAI)[24].Thus,the rice yield response to the rotation is also critical to balancing the N supply and demand in the system.In summary,understanding the integrated effects of rotation and straw incorporation on rice grain yield and N use efficiency is highly desirable in subtropical paddy soils under paddy–upland cropping systems.

Since 2002,a series of long-term agroecosystem experiments have been conducted to investigate the effect of rotation types with full straw incorporation on soil properties and rice yield in the paddy–upland cropping system in southeastern China.In 2012,N fertilization treatments with three N levels were applied based on the existing experiment to examine the rotation effect on rice N use efficiency.The objectives of this study were(1)to evaluate the effect of rotations on rice grain yield,N uptake,and N use efficiency under different N levels;(2)to evaluate the response of physiological components to the rotations,and(3)to elucidate the relationship between the physiological traits and the N use efficiency response to the rotations.

2.Materials and methods

2.1.Study site description

Field experiments were conducted from 2012 to 2016 at a longterm field trial that was established in 2003[9]and located at the experimental farm of the China National Rice Research Institute(120.2°E,30.3°N,at an elevation of 11 m above sea level).The soil is a Ferric–Accumulic Stagnic Anthrosol,with properties previously[9]described.The site is located on the middle and lower Yangzte Plain in Hangzhou,Zhejiang province,China.The area is characterized by a subtropical monsoon climatewith annualmean temperaturesof 13–20 °C,ranging from 2 °C in January to 35 °C in July,and a mean annual precipitation of 1200–1600 mm with approximately 80%falling between April and September(Fig.1).

2.2.Experimental design and treatments

From 2012 to 2016,treatments were established annually in the same plots and were arranged as a split plot design with three replicates.The treatment was a paddy–upland crop rotation system as the main– plot and the N rates as the split–plots.A schematic of the field experiment is shown in Fig.2.

The main–plot treatment was the rotation treatment based on the paddy–upland rotation,comprising summer rice and various winter crops.The following six crop rotations were used:continuous monoculture rice-fallow(RF),rice-wheat(RW),rice-potato with rice straw mulched(RP),rice-green manure crop(Chinese milk vetch;RC-G),rice-oilseed rape(RO),and rice-green manure crop(oilseed rape with fresh straw incorporated into soil at flowering;RO-G).The split-plot treatment consisted of three N rates during the rice-planting season.Each primary plot(8 m wide×24 m long)was divided lengthwise to accommodate the three N rate treatments:N0,N1,and N2 with total nitrogen inputs of 0,142.5,and 202.5 kg N ha-1,respectively.Approximately 50%of the N fertilizer was incorporated during the land preparation as basal fertilizer one to three days before transplanting seedlings.The remaining N fertilizer was applied as urea at the rice tillering and booting stages using 25%in each application.A total of 135 kg ha-1of potassium was applied to each plot with 50%as basal dressing and 50%as topdressing at panicle initiation in the form of potassium chloride.A total of 65 kg ha-1phosphorus was applied to each plot with 100%as basal dressing in the form of superphosphate.

2.3.Crop management

Fig.1–Climatic conditions at the experimental site.

Fig.2–A field overview of the experiment(photographed on the harvest date of the RO–G rotation).

The rice varieties used were Zhongzheyou 1(2013,2014,and 2016)and Chunyou 84(2015).These cultivars are major rice cultivars with high yield for single-season rice in southeastern China.Chunyou 84 was used in 2015 owing to its high yield in 2014.However,because Chunyou 84 is sensitive to false smut disease,Zhongzheyou 1 was used again in 2016.In each year,pre-germinated rice seeds were sown in a seedbed on May 20 and seedlings were transplanted to the field on approximately June 20.Rice seedlings were transplanted at spacings of 30 cm ×16.7 cm in 2013–2015 and 26.7 cm ×13.3 cm in 2016,with two plants hill-1.Oilseed rape(including the treatment in which it was used as a green manure crop)was planted at a spacing of 20 cm×15 cm(30,000 plants ha-1)in 1.2 m-wide rows in early November.Wheat and green manure(Chinese milk vetch)crops were seeded at 150 kg ha-1and 15 kg ha-1,respectively,in late October.Weeds were controlled in rice,wheat,and oilseed rape with both pre-and post-emergent herbicides.Insects and disease were controlled according to local requirements to avoid yield loss.The cropping arrangements with seeding/planting and harvesting dates in the summer and winter seasons are presented in Table S1.

With respect to the managementofstraw andcrop residues,the ricestraw was removedafter harvest following the summer rice season(it was collected manually for the potato system).The residues were retained only with the winter crops under various rotations,except for the RP,in which rice straw was mulchedonthepotato-plantingplotandconsequentlyretained until the following rice season.The straw or residues from the different winter crops,including the mulched rice straw,fresh straw of green manure(Chinese vetchand oilseed rape straw at flowering),and the straw of potato,wheat,and oilseed rape after harvest,were chopped and incorporated into the soil in early May using a micro-tilling machine(Dongfeng 151/121,Changzhou,Jiangsu,China).

2.4.Sampling and analyses

The incorporated straw or green manure(dry matter of Chinese milk vetch,wheat,oilseed rape,and potato with rice straw mulch)was investigated by sampling the wheat straw/oilseed rape residues or green manures from three randomly chosen 1-m2sections of each plot before the rice was transplanted.The collected materials were separated into aboveground biomass and roots and oven-dried at 105°C to constant weight to determine dry weight.The C,N,P,and K concentrations in the plant tissues were determined as described by Lu[25].

The heading stage was identified by the emergence of 90%of the panicles in the plot,and the plant was considered to have reached maturity when 95%of spikelets had turned yellow.Twelve hills with the mean culm(panicle)number from each plot were collected at each sampling for dry weight measurements of the constituent organs(the hills were separated into straw and panicles).The leaf area was determined using a Li 3100 system(LI-3100,LI-COR,Lincoln,NE,USA).The plant tissues were oven-dried at 80°C and weighed.The N concentrations in the plant tissues were determined using the Kjeldahl method after wet digestion[26].The rice yield was determined by sampling the plants from a 5 m2area in each plot at maturity.Unhulled(rough)rice kernels were obtained after reaping,threshing,and winnowing.The weight of the rough rice kernels was adjusted to a moisture content of 13.5%.

The yield components were analyzed as previously described[27].Briefly,12 plants were hand-threshed to count filled and unfilled kernels.The filled kernels were separated by submergence in a NaCl solution with a specific gravity of 1.06 g m-3and then hulled and oven-dried at 105°C to constant weight to determine grain dry weight.The grain setting was calculated as the ratio of filled to total kernels.

2.5.Crop response to N input

2.5.1.Rice N use efficiencies

N use efficiency parameters were calculated based on the grain yield and N accumulation in plots treated at different N rates.The definitions are as follows:

and the degree of synchrony between crop N demand and supply,the so-called nitrogen balance degree(NBD),was calculated as follows[28]:

where GY+Nis the grain yield of the plots that received N fertilizer,GYN0is the grain yield in the N0 plot,FNis the amount of N fertilizer applied,TN+Nis the total N accumulation in the plots that received N fertilizer at maturity,and TNN0is the total N accumulation in the N0 plot at maturity.

2.6.Statistical methods

Statistical analyses were performed using SAS software(Statistical Analysis System,version 9.1,SAS Institute,Cary,NC,USA).Mixed models were used for an analysis of variance with crop rotation and N level as fixed effects and year(block)as a random effect.Pearson correlations between the N use efficiency traits and yield attributes were calculated.

3.Results

3.1.Straw properties

To evaluate the incorporation of straw(including the root residues)in the rotation treatments,the retained amounts of straw dry matter,the content of N,P,K,and C in the treatments were determined(Fig.3 and Table S2).The amounts of straw and its C incorporated into the soils were 98.8%–115.1% and 87.7%–10.4%,respectively,significantly greater in the RP than in the other rotations.It should be noted that the mulched rice straw accounted for approximately 90%of the total straw in the RP treatment.The total retentions of N,P,and C in the RP treatment were respectively 24.7,3.31,and 160.6 g m-2,which were the highest amounts retained among the winter crops.The amounts of straw N ranked in the order RP＞RC-G＞RO-G＞RW/RO.Using the order RO＞RW＞RP＞green manure(RC-G and RO-G),the C/N ratios of the straws of the five winter crops ranged from 18.69 to 70.09.According to the levels of the C/N ratio,the rotation treatments could be classified into two groups:high C/N ratio(RO and RW)and low C/N ratio(RP,RC-G,and RO-G).

3.2.Effect of rotation systems on rice yield

The effect of rotation on the rice grain yield varied across N rates and years(Table 1 and Fig.4).The inclusion of the winter crops in the crop rotation increased the rice yield,except in 2014.The rice yields,averaged across 2013–2016,in the plots without N fertilizer were 5.93,7.38,7.07,7.48,7.23,and 7.20 t ha-1in the RF,RP,RW,RO,RO-G,and RC-G treatments,respectively.However,the magnitude ofthe increase depended on the type of winter crop.Compared to the RF,the RC-G,RO,RO-G,RP,and RW rotations increased grain yield by 9.09%–39.44%,22.71%–31.95%,11.62%–28.49%,14.80%–37.19%,and 12.95%–24.79%,respectively(Fig.S1).

Typically the annual grain yields for the treatments under N1 rate(142.5 kg N ha-1)ranked in the order RP＞RO-G＞RO＞RW＞RC-G＞RF(Fig.4).The rotations of the RF and RP generally produced the lowest and highest grain yields from 2013 to 2016,except during 2014,in which the RO-G produced the lowest grain yield(Fig.S1).The average grain yields in the RP rotations were slightly(respectively 2.3%,6.0%,5.0%,and 5.9%)higher than those of the RO-G,RC-G,RO,and RW,(Fig.4).

For the treatments under the N2 rate(202.5 kg N ha-1),the effects of the rotations on grain yield were inconsistent among years(Fig.4).The average grain yield in the RF across the yearswas8.39 t ha-1,comparable to theaverage(8.40 t ha-1)of the other rotations.The result in the RP was similar to that under the N1 rate,except during 2016.The annual grain yield(excluding 2016)was 5.03%–10.02%greater in the RP than in the other rotations.In addition,abnormal yields were observed in individual years.For example,large yield losses were recorded in the RC-G in 2014 and RO in 2015(Fig.S1),and could be attributed to insect and disease damage.

Fig.3–Straw properties:C/N ratio(A)and straw dry weight(B)in straw under five rotations.Mean±SE(n=36)across years and N treatment are shown.Lowercase letters indicate statistical differences among rotations across years at P=0.05.RF,ricefallow;RW,rice-wheat;RP,rice-potato with rice straw mulched;RC-G,rice-green manure crop(Chinese milk vetch);RO,riceoilseed rape;RO-G,rice-green manure(oilseed rape with fresh straw incorporated into soil at flowering).

Table 1 –ANOVA of rice grain yield and physiological traits under nitrogen rates and rotations across 2013–2016.

3.3.The effect of rotation treatments on plant absorption

The effects of the various rotations and N rates on rice N accumulation and content from 2013 to 2016 are described in Table 2.The effects of interaction between rotation and N on the biomass,N accumulation,and content were not significant in any year(P＞0.05).For biomass accumulation,the grain yields in the RP,RC-G,RO-G,and RO rotations were generally greater than those in the RW and RF,and that of the RF was 6.6%–10.3%lower than those of the other four rotations(Table 3).The application of N fertilizer significantly improved the biomass,but the difference between the N1 and N2 rates was not significant from 2013 to 2016(Table S3).Yearly differences in biomass accumulation were also observed,with the highest observed in 2015,a result that could be attributed to the planting of the japonica/indica hybrid variety Chunyou 84.For N accumulation,the results were similar to that of biomass accumulation.The rotation system that included fallow resulted in a reduction in N accumulation relative to the other five rotations,except during 2014(Table S3).The N accumulation was 7.5%–28.2%,13.2%–26.8%,13.6%–31.3%,and 10.8%–22.8%lower in the RF than in the RP,RO,ROG,and RC-G,respectively.No significant difference was found between the RF and RW.N accumulation increased with N fertilization rate(Table 3).However,the difference between 142.5 and 202.5 kg N ha-1was not statistically significant,except during 2013.The N contents in the RF(0.89%)and RP(0.94%)rotations were significantly lower than those in the RO-G(1.01%),RO(1.00%),and RC-G(1.00%),whereas no statistical difference was found between the RW and RP.On average,the N content ranked in the order RO-G＞RO＞RCG＞RW＞RP＞RF.N content of rice plants increased with N fertilization rate,except during 2016,when there was no difference between the N rates of 142.5 and 202.5 kg N ha-1(Table S3).

Fig.4–Rotation effects on rice yields under N0,N1(142.5 kg N ha-1),and N2(202.5 kg N ha-1)fertilization rates.Mean±SE(n=12)across years shown.Lowercase letters indicate statistical differences among rotations at P=0.05.RF,rice-fallow;RW,rice-wheat;RP,rice-potato with rice straw mulched;RC-G,rice-green manure(Chinese milk vetch);RO,rice-oilseed rape;RO-G,rice-green manure(oilseed rape with fresh straw incorporated into soil at flowering).

Table 2–ANOVA of N accumulation,concentration,and N use efficiencies under N fertilization rates and rotations across 2013–2016.

3.4.The effect of rotations on rice N use efficiencies

N use efficiency parameters in the plots were calculated based on grain yield,chemical N fertilizer,and N accumulation(Tables 2 and 4).The NAE,NPE,NRE,and NBD values were 4.95–12.85 kg grain kg-1N, 15.22–39.67 kg grain kg-1N,0.35–0.47,and 30.33–78.30,respectively,across rotations,N rates,and years.However,the effect of interaction between yearof rotation and N rate wassignificant(Table2).Accordingly,the effects of rotation and N rate on the N use efficiency parameters are presented by year(Table S4).

In general,the rotations using RF resulted in slightly or significantly higher NAE and NPE than the other rotations across the various N rates from 2013 to 2016;the interaction of N with rotation was not significant in 2013 and 2015(Table S4).The NAE in the RF was 8.4%–86.0%,16.7%–87.6%,25.0%–242.0%,31.0%–92.6%,and 16.7%–232.6%higher than that in the RP,RW,RO,RO-G,and RC-G,respectively,and the NPE was 17.1%–175.3%,33.9%–170.1%,40.4%–345.1%,56.5%–153.3%,and 47.0%–205.4%,respectively.In contrast,the RO and RC-G rotations showed relatively low values of NAE and NPE.The NAE values were respectively 23.5%–80.8%and 30.1%–85.7%lower in the RO and the RC-G than in the RF,and the NPE values were respectively 22.5%–71.3%and 32.7%–68.1%lower(Table 4).The effects of the N rates on the NAE and the NPE were consistent:the values for 142.5 kg N ha-1were significantly higher than for 202.5 kg N ha-1(Table 4S).

The RO-G and RW rotations resulted in increases in NRE of 15.4%–80.0%and 5.5%–31.3%,respectively,relative to the other rotations(Table 4).In contrast,the RF rotation resulted in lower NRE values than in the other rotations,with the exception of 2015.However,the effect of rotation on NRE varied by year,and no significant differences among the rotations were observed in 2014 and 2015.In addition,no marked difference in NRE among the various N rates was observed in any year(Table S4).

The rotation effect on NBD varied with N rate(Table 4).Examining the interaction between rotation and N rate indicated that the NBD values in the RP,RW,and RO-G rotations were significantly lower than those in the RF,RO,and RC-G under the N rate of 142.5 kg N ha-1.The NBD values increased with N rate,except for the RF,in which the NBD decreased by 10.5%for the N rate of 202.5 kg N ha-1compared with that of 142.5 kg N ha-1(Table 4 and 4S).

Table 3 –N content and accumulation in rice plants under rotations and N fertilization rates across 2013–2016(n=12).

Table 4 –NAE,NPE,NRE,and NBD of rice plants under rotations and nitrogen fertilization rates across 2013–2016(n=12).

3.5.Rice physiological trait responses to rotation treatments and N rates

The response of the various rice physiological attributes to rotation(R)and N rate(N)across years(Y)is presented in Table 1.The interaction effects of N×R and N×R×Y were insignificant for panicles m-2,kernels panicle-1,grain setting,and HI.The interaction effect of Y×R was insignificant for kernels panicle-1,grain setting,and HI.Thus,the effects of rotation on kernels panicle-1,grain setting,and HI were highly similar,regardless of N rate and year.The kernels panicle-1in the RP were respectively 6.7%,6.1%,8.4%,8.0%,and 7.8%higher than in the RF,RW,RO,RO-G,and RO-C rotations(Fig.5-B).However,no significant differences in kernel weight were detected among the rotations,regardless of year and N rate(Fig.5-D).The N effects on kernels panicle-1,grain setting,and HI were significant;however,the results also varied by year(Table 1).The highest kernels panicle-1of 225.8,256.8,and 143.6 were observed under the N1 rate in 2013,2014,and 2016,respectively.However,minimal differences were found between the N1 and N2 rates in 2015.The grain setting was higher in the plants under the N0 and N1 rates than in those under the N2 rate in 2015 and 2016,but the difference was not significant among the various N rates in 2013 and 2014.The HI decreased with N rate,except in 2016,when the difference between the N0 and N1 rates was not significant(Table S5).

The effect of rotation on the number of panicles m-2and grain setting was significant,but the effect varied across the years(Y×R,P＜0.01)(Table 1).The numbers of panicles m-2were 7.9%–14.0%and 5.1%–11.4%lower in the RF rotation than in the other rotations in 2013 and 2016,respectively,whereas minimal differences were observed among the five other rotations regardless of year(Figs.5-A and S2).The number of panicles m-2increased in parallel with N rate,except in 2014 and 2015,when the difference between the N1 and N2 rates did not differ significantly(Figs.5-A and S2).The difference among the rotations was not significant for grain weight,except during 2013,when grain weight was greatest in the RF rotation.Plants subjected to the N0 rate showed the highest grain weight,which was significantly higher than that under the N1 and N2 rates,except for 2016,in which minimal differences were recorded among the various N rates.

The biomass accumulation at the harvest stage was consistent with the grain yield.The biomass of the RP rotation wasgenerallyhigherthanthatoftheothers,whilethatoftheRF was the lowest,exceptin 2013,in which the value in the RF was slightly higher than those in the RW,RO,and RO-G(Table 3).

The physiological status of rice at heading stage is positive correlated with the rice yield and N use efficiency.The biomass accumulation at heading was significantly affected by both rotation and N rate.However,these effects varied over years.Biomass accumulation was higher in the RP than in the other rotations in 2013 and 2014,whereas the differences among the rotations were not significant in 2015 and 2016.The RF rotation produced lower biomass accumulation than the other rotations over the four years.The specific leaf area among the various N rates under the RO,RO-G,and RC-G rotations was within the range of 178.2–180.2 cm2g-1,values significantly higher than those under the RP and RF(170.3–-170.5 cm2g-1).The LAI in the RP,RO,RO-G,and RO-G rotations was generally higher than those in the RF and RW,except during 2016,when the LAI was significantly lower in the RP and RC-G than in the RO(Table S5).

4.Discussion

4.1.Rice grain yield and N uptake

Fig.5–Effect of rotation on panicles m-2(A),kernels panicle-1(B),grain setting(C),and the grain weight(D),across N fertilization rates and years.Mean±SE(n=36)across years and N treatment are shown.Lowercase letters indicate statistical differences among rotations at P=0.05.RF,rice-fallow;RW,rice-wheat;RP,rice-potato mulched with rice straw;RC-G,ricegreen manure(Chinese milk vetch);RO,rice-oilseed rape;RO-G,rice-green manure(oilseed rape with fresh straw incorporated into soil at flowering).

Grain yields in the absence of N fertilizer are functions of available soil N for plant use from net mineralization and have been suggested[29]to be an indicator of field variability in native N supply.The present results indicate that incorporation of straw in the rotation increased rice yield in plots without N application.The average rice yields in the RF(5.93 t ha-1)were significantly lower than those in the other rotations(7.20–7.48 t ha-1;Fig.4).Similar results were found for the total N accumulation at maturity in the N0 plots.The difference between RF and the other rotations in the N0 plot might result from the different potential supplies of N present in the soil[29].The average N accumulation in the RF was 10.56 g m-2,which was 10.95%–31.93%lower than that in the rotations with winter crops(Table 3).The role of fallow in the crop rotation system is still debated.For ecological scientists,a fallow treatment provides primary support for crop production by suppressing weeds,regenerating soil fertility,controlling erosion,and sequestering carbon[30–32].However,the present evidence suggests that the inclusive of winter crops(with straw incorporated)might be more effective for increasing the soil N supply than winter fallow.In addition to its effects on N,previous study also indicated that incorporating straw also led to an increase in pH and a decrease in soil bulk density in the rotations relative to the RF[9],increasing plant growth and yield in the paddy field.

The average yield differences between the RF and the other rotations decreased with increasing N rate and were 18.55,4.14,and 0.23 under 0,142.5,and 202.5 kg N ha-1,respectively(Fig.4).Increased N may have helped to minimize the yield gap between the RF and the other rotations.However,grain yield response to the increase in N(from the N1 to N2 rate)varied with rotation and year.Higher yield increases in the RF and RW rotations than in the other rotations were observed with medium to high N rates,whereas the grain yields in the RF and RW increased slightly,by 3.23%and 1.93%respectively.However,yield declines were observed in the RO,RO-G,and RC-G rotations(Fig.4).

Several studies[29,33–35]have identified potential risks of crop rotation to rice yield with increasing N rates,particularly when green manure crops are included.The results of the present study are consistent with those results.Yield loss may be caused by unfavorable weather,suboptimal nutrient supply,weed competition,and damage from insects and diseases,highlighting the importance of robust crop growth and yield to increased N use efficiency[34,36].Crop N demand is based on biomass production and physiological N requirements[6,34,37],as well as by crop management practices and climate[23,31].Thus,the high N rate(202.5 kg N ha-1)applied in the RO,RO-G,and RC-G rotations was greater than the demand,and yield decreases from the medium to high N rates resulted from the overapplication of N fertilizer,in turn further decreasing grain yield by increasing susceptibility to lodging and/or damage from pests and diseases[5].

4.2.Rice N use efficiency

This study evaluated the N use efficiency response of rice to the rotations under N rates of 142.5 and 202.5 kg N ha-1by assessing NAE,NPE,NRE,and NBD(Table 4).The NAE on average ranged from 7.76 to 14.70 kg grain kg-1N across rotations,values higher than the 9.1 and 6.4 kg grain kg-1N values reported previously by Lin[4]and Wang et al.[21],respectively.The NAE values in the rotations ranked in the order RF＞RO-G＞RP＞RW＞RC-G＞RO(Table 4S).The average NPE ranged from 22.68 to 45.57 kg grain kg-1N across rotations.The RF rotation showed higher NAE and NPE than the other rotations(Table 4).However,the biomass accumulation and grain yields were lower in the RF rotation than in the other rotations without N application(Table 3 and Fig.4).Thus,the higher NAE and NPE in the RF are likely to be artifacts of lower yield and reduced N uptake without applied N rather than representing improved performance with applied N[38].

Rice yield tends to decrease when non-legume species are in the rotation sequence[39,40].Zhu et al.[40]reported that the adoption of ryegrass(Lolium multiflorum Lam.)in a paddyupland system reduced grain yield and plant N uptake in early rice and double-season rice compared to winter fallow in the Dong Ting Lake Plain,China.A meta-analysis[39]also suggested a yield decrease of 8%in corn rotated with a nonlegume species,compared with legume crops.However,our results showed no significant differences in the NAE and NPE between rotations with legume(RC-G)and non-legume(RO and RW)winter crops,in contrast to the previous reports.One of the reasons for this finding could be the straw incorporation in the present study.In fact,the adoption of a crop rotation generally exerts a residual effect in the following season[20,33].The incorporation of various crop straws in paddy field has long been recognized to improve soil fertility[41].The finding of little difference in grain yield between the RO and RC-G in the plots without N application supports the effect of straw incorporation on the soil N supply.Thus,our results suggest that the adoption of straw incorporation could reduce differences in N use efficiency among the various rotations.However,further research is needed to quantify the contribution of straw incorporation to rice N use efficiency.

The average NRE across the year ranged from 0.22 to 0.45 with a mean of 0.32(Table 4),values consistent with the 0.30–0.40 range in previous reports[3,40].The RO-G and RW rotations resulted in respective increases in NRE by 15.4%–80.0%and 5.5%–31.3%relative to the other rotations.The opposite result was found for the RF rotation,which resulted in lower NRE values than in the other rotations(Table 4).The poor NRE indicated less nitrogen uptake from the chemical N,resulting in a large chemical N loss through ammonia(NH3)volatilization and denitrification [12].Thus,ourresults suggest that high application of N should not be recommended for the rice and fallow rotation system.The NRE of the hybrid japonica/indica variety(Chunyou 8)ranged from 0.49–0.73 with an average of 0.60.The higher biomass in japonica/indica hybrids with higher total N,P,and K accumulation has been attributed[45]to greater heterotic biomass production than in other varieties.These results indicate that the rice variety influences the N balance during the paddyupland crop rotation.Further research is essential to reevaluating soil-crop N balance,particularly when new varieties with high biomass production are introduced into the system.

The N use efficiencies can be improved by both indigenous and applied chemical N[34].Achieving closer synchrony between crop N demand and N supply(N balance degree)from all sources throughout the growing season is critical to achieving high levels of N use efficiency in crop-production systems by reducing N loss[42].The uptake efficiency of the available N derived from organic N sources in preceding crops by incorporation of legume green manure,cover crops,and straw is positively correlated with the amount of N fertilizer applied[43].Closer synchrony between supply and demand is indicated by smaller NBD values[34,42].In the present study,the NBD values in the RP and RW were significantly lower than those in the other rotations under both N1 and N2 rates(Table 4).The higher grain yield in the RP than in the RW under the N1 rate(Fig.4)suggests that use of the RP rotation could be a promising practice affording grain yield and N use efficiency under reduced N input comparable to that with other rotations.

4.3.Rice yield components and physiological traits

Aside from the optimal use of applied and indigenous N,an increase in N use efficiency may also be attributed to crop health,insect and disease management,multi-nutrient supply,the use of adapted varieties for more efficient uptake of available N,and increased conversion of plant N to grain yield[11,44].The agronomic and physiological traits of rice plants also represent responses to rotations affected by the amount of applied N[37,45–47].The increase observed in the number of kernels panicle-1in the RP rotation relative to the other rotations(Fig.5-B)indicates greater N uptake during panicle initiation and/or heading when most yield components are determined[48].The greater biomass accumulation observed in the RP is also indicative of high growth rates in the rice plants,resulting in part from a decrease in the soil bulk density and an increase in soil organic matter content and nutrients other than N[9].Loose and fertile soil also supports root growth[49],relieves transplant shock[50],accelerates seedling establishment[51],and ultimately,more effectively exploits available soil resources[19].Above benefits also partly explain the observed higher yield and lower NBD in the RP rotation than in the other rotations.In contrast to the RP,the RF rotation resulted in a lower number of panicles m-2than the other rotations(Fig.5-A),a finding that implies limited tillering development during the early stage.This limit may have resulted from both reduced native N(Table 3)and a decrease in soil bulk density[9].Compact soil may cause poor or delayed seed germination and seedling development and affect the final number of panicles[33].However,further studies are needed to characterize the relationship between soilproperties and yield components underdifferent rotations.

Fig.6–Biomass accumulation and grain yield as functions of N uptake.

Increased N uptake was a primary contributor to heavier aboveground biomass accumulation,although the relationship between grain yield and crop N content was quadratic(Fig.6).This finding suggests that the N efficiency which converts biomass to grain yield decreased with higher N uptake,which generally controls crop physiological N requirements[35].Mae et al.[22]suggested that the higher physiological N use efficiency in rice plants is attributable to a higher proportion of dry matter partitioning in the panicles.In another study,NPE was positively correlated with HI[23],as found in our study.We also found a positive correlation between HI and NAE(P＜0.001,Table 5).However,NPE and NAE were negatively correlated with SLA and LAI at the heading stage(Table 5).Thus,HI could be a useful indicator for evaluating the NAE and NPE differences caused by rotations or other treatments.

5.Conclusions

The inclusion of winter crops in rotations increased the soil N supply available for the subsequent rice crop.Average rice yields in rotations that included winter crops were greater than those in a rice and winter fallow rotation.Increasing the N fertilization rate may have helped to minimize the yield gap between the RF and the other rotations,but the decreased NRE in the RF suggests large N loss from the chemical fertilizer.Incorporating straw resulted in no significant differences in NAE and NPE between rotations with legume(RC-G)and nonlegume(RO and RW)winter crops,suggesting that the adoption of straw incorporation could reduce the differences in N use efficiency between rotations with and without legume crops.Thus,adopting winter crops rotated with summer rice with straw return may be a practice that maintains rice yield with less N input.The primary yieldattributes of the RF and RP rotations were respectively the number of panicles m-2and number of kernels panicle-1.The negative correlation of NPE with N accumulation in kernels could be a useful indicator for evaluating the N use efficiencies of rotations.

Table 5–Pearson correlations between N use efficiencies and yield attributes(n=144).

Acknowledgments

The National Key Research and Development Program of China(2016YFD0300108,2016YFD0300208-02),the National Natural Science Foundation of China(31671638),the China Agriculture Research System(CARS-01-04A),and the Special Fund for Agro-scientific Research in the Public Interest(201203096)partly supported this study.

Appendix A.Supplementary data

Supplementary data for this article can be found online at https://doi.org/10.1016/j.cj.2018.07.007.