Jiangsu Key Laboratory of Crop Genetics and Physiology,Co-Innovation Center for Modern Production Technology of Grain Crops,Yangzhou University,Yangzhou 225009,P.R.China

Abstract It is of great significance to study the root characteristics of rice to improve water and nitrogen (N)use efficiency and reduce environmental pollution.This study investigated whether root traits and architecture of rice influence grain yield,as well as water and N utilization efficiency.An experiment was conducted using the upland rice cultivar Zhonghan 3 (a japonica cultivar)and paddy rice cultivar Huaidao 5 (also a japonica cultivar)using three N levels,namely,2 g urea/pot (low amount,LN),3 g urea/pot (normal amount,NN),and 4 g urea/pot (high amount,HN),and three soil water potentials (SWPs,namely,well-watered (0 kPa),mildly dried (-20 kPa)and severely dried (-40 kPa).The results showed that with decreasing SWP,the percentage of upland rice roots increased in the 0-5 cm tillage layer,and decreased in the 5-10 and 10-20 cm tillage layers,whereas paddy rice roots showed the opposite trend.With increasing amounts of N,the yield of upland and paddy rice increased,and the percentage of root volume ratios of the two rice cultivars in the 0-5 and 5-10 cm tillage layers increased,whereas that in the 10-20 cm tillage layer decreased.The roots of upland rice are mainly distributed in the 10-20 cm tillage layer,whereas most paddy rice roots are in the 0-5 cm tillage layer.These results indicate that the combination of-20 kPa SWP and NN in upland rice and 0 kPa SWP and LN in paddy rice promotes the growth of the root system during the middle and late stages,which in turn may decrease the requirements for water and N fertilizer and increase rice yield.

Keywords:upland rice,paddy rice,root traits,root architecture,soil water potential,nitrogen

1.lntroduction

Rice (OryzasativaL.)is one of the most important crops in the world and is the foremost staple food in Asia.Rice is also the greatest consumer of water among all crops and consumes approximately 80% of the total irrigated fresh water resources in Asia (Bouman and Tuong 2001).Besides water,nitrogen (N)is another key factor that determines crop yield (Juet al.2009).In the last 50 years,global rice yield has continuously increased,partly because of the increase in fertilizer nutrient input,especially N fertilizer (Penget al.2009,2010).However,the use of N fertilizer is generally inefficient,and the apparent recovery efficiency of N fertilizer is only 33% on average (Garnettet al.2009).

Root growth is a dynamic process,root architecture may change in response to alterations in the immediate environment (Bonseret al.1996;Tranet al.2015).Previous studies have shown that both moisture and nutrient conditions are closely associated with the root architecture and morphology in rice (OryzasativaL.),thereby affecting the yield.A number of reports have provided evidence that root traits are directly related to the drought resistance and nutrient uptake capacity of rice plants (Wanget al.2009;Chenet al.2012;Comaset al.2013;Matsunamiet al.2013;Ugaet al.2013;Wang Pet al.2015),particularly N absorption capacity (Linget al.1990;Wang X Bet al.2015).Among the required mineral nutrients,the supply of N,followed by phosphorus,has the most substantial effect on root growth,morphology,and distribution (Marschneret al.1996;Liaoet al.2001).

The effects of water and N on rice roots have been extensively studied (Wuet al.1992;Wagner and Beck 1993;Tranet al.2015).Wuet al.(1992)earlier reported that the roots in paddy fields are mainly distributed in the upper layer of the soil and densely integrate into nets,while much fewer roots are found in the upper layer of soil in upland fields or fields cultivated by irrigation.Wagner and Beck (1993)showed that compared with the low nutrient conditions,high nutrient supply can promote the growth of aboveground and underground parts in a certain range,but it can also result in the decrease of root/shoot ratio,the thinning of root system and the increase of root surface area,which may also cause environmental problems.The above studies are mostly about paddy rice,and studies involving upland rice are very rare,even though upland rice is becoming a partial replacement for standard rice cultivars (i.e.,paddy rice).Field practice has shown that upland rice can save 70-80%of water compared to paddy rice.Lu and Zhou (1994)showed that adventitious root diameter,adventitious root vessel number,diameter and area of upland rice were better than those of paddy rice.The root anatomical structure of upland rice showed water-saving and drought resistance.

It has been hypothesized that a synergistic interaction between soil moisture and N fertilizer occurs during rice growth to increase rice yield and water and N use efficiency(Liuet al.2013;Yanget al.2015).However,evidence is scarce,and investigations involving root traits and architecture and yield between upland rice and paddy rice in relation to water-N synergism remain rare.

The objective of the present study was to evaluate grain yield,root traits,root architecture,and their relations to yield of upland rice cultivar Zhonghan 3 (japonica)and paddy rice cultivar Huaidao 5 (japonica)under different water and nutrition supply conditions.It provides theoretical and practical basis for saving water,saving fertilizer and optimizing rice cultivation,and reducing environmental pollution.

2.Materials and methods

2.1.Plant materials and site description

Experiments were conducted at the research farm of Yangzhou University,Jiangsu Province,China (32°30′N(xiāo),119°25′E)during the rice growing season (May-October)in 2015 and 2016.Two rice cultivars were used,namely,upland rice cultivar Zhonghan 3 (ajaponicacultivar)and paddy rice cultivar Huaidao 5 (also ajaponicacultivar).Seeds were sown in the field on May 11-12.Thirty-dayold seedlings were then transplanted to porcelain pots(30 cm high,25 cm in diameter,14.72 L in volume).Each pot was filled with 20 kg of sandy loam (Typic fluvaquents,Entisols (U.S.Taxonomy)),and the soil had 2.43% organic matter content,101.4 mg kg-1available N,22.8 mg kg-1available phosphorus,and 90.5 mg kg-1available potassium.Phosphorus (0.5 g single superphosphate)and potassium(0.3 g KCl)were applied and mixed into the soil in each pot before transplanting (June 10-11).Three holes were planted per pot and two seedlings were planted per hole for the two cultivars.Weeds,insects,and diseases were controlled as necessary to avoid yield loss.Both cultivars headed on August 8-13 (50% of plants),and were harvested on October 6-12.

2.2.Treatments

The experiment was a 2×3×3 (two cultivars,three levels of N,and three levels of soil water potentials (SWPs))factorial design with 18 treatment combinations.Each treatment was replicated in 20 pots,of which 12 were used for destructive sampling during the experiment and eight for harvest.Three N concentrations using urea were used,namely,2,3 and 4 g/pot,and were represented as low(LN),normal (NN)and high N (HN),respectively.The LN,NN,and HN pots were top dressed with 1.0 g (LN),1.5 g(NN)and 2.0 g (HN)urea per pot at pre-transplant (1 day before transplanting);with 0.4 g (LN),0.6 g (NN),and 0.8 g (HN)urea per pot at mid-tillering;and with 0.6 g (LN),0.9 g (NN)and 1.2 g (HN)urea per pot at elongation and panicle initiation,respectively.Ten days after transplanting to maturity,three SWPs were added to the LN,NN,and HN treatments.Three SWP treatments were conducted,namely,well-watered,wherein pots were manually loaded with 1-2 cm tap water,and SWP was maintained at 0 kPa;moderate soil-drought,SWP maintained at 0-20 kPa (soil moisture content:0.166 g g-1);and severe soil-drought,SWP maintained at 0-40 kPa (soil moisture content:0.144 g g-1).For the drought treatments,SWP was monitored at a soil depth of 15-20 cm using a tension meter (Institute of Soil Science,Chinese Academy of Sciences,Nanjing)consisting of a 5 cm long sensor inserted into each pot.Tension meter readings were recorded every 3 h from 6 a.m.to 6 p.m.when readings dropped to the desired value,0.35 and 0.15 L of tap water per pot was added to the -20 kPa and -40 kPa treatments,respectively.The pots were placed in the field and sheltered from rain using a removable polyethylene shelter when necessary.During the grain-filling period,the temperature was normal,wherein there were no abnormally high and low temperatures that could hinder grain growth.

2.3.Sampling and measurements

Eighteen plants from each treatment were observed in terms of tiller number.These 18 plants were observed at transplanting,mid tillering,panicle initiation,heading time(50% of plant headed),and maturity.

To ensure representativeness of the sampling,the number of stems in all pots was counted at each treatment,and then three representative pots of plants with a mean stem number were assessed in terms of root and shoot biomass and root morphological indices at the jointing stage.The roots were carefully rinsed using a hydropneumatic elutriation device (Gillison’s Variety Fabrications,Benzonia,MI,USA)and detached from their nodal bases.Portions of each root sample were used for assessment of adventitious roots number,longest root length,root diameter,and root volume,and after measuring,portions of each root sample and the rest of the roots were dried in a 70°C oven to a constant weight and then weighed.To measure root volume,the roots were arranged and floated on shallow water in a glass tray (30 cm×30 cm),scanned using a scanner (Epson Expression 1680 Scanner,Seiko Epson Corp.,Tokyo,Japan),and then analyzed using WinRHIZO Root Analyzer System (Regent Instruments Inc.,Quebec,Canada).The number of adventitious roots in each hole was determined by counting,and the longest root length in each hole was measured using a ruler.The diameter of the adventitious roots was measured with a slide gauge as follows:10 adventitious roots from each hole constituted a group and were placed side by side and subsequently measured.

To determine root architecture (longitudinal distribution of root in the soil),three representative pots of plants were taken during their full heading time.The molded soil from each pot was divided into three segments (0-5,5-10,and 10-20 cm)from top to bottom.The roots in these sections were washed and passed through a 1-mm sieve.The root volume was measured by water displacement,and then the root sections were dried (fixed in 105°C for 30 min and dried at 70°C for 72 h in oven)to a constant weight and weighed.Shoot biomass was also measured for the aboveground parts.Root configuration (longitudinal distribution of root in the soil)was 0-5,5-10,and 10-20 cm root volume as a percentage of total root volume.

2.4.Leaf water potential

Leaf water potentials of the upmost fully-expanded leaves on stems were measured at clear midday (11:30 a.m.)at 46 and 50 days after transplanting upland rice and 43 and 52 days after transplanting paddy rice when SWPs were approximately -20 kPa (D1)and -40 kPa (D2),and at 47 and 51 days after transplanting upland rice and 45 and 53 days after transplanting paddy rice when treatments were rewatered (SWP of both treatments was 0 kPa,indicated by W1 and W2,respectively).Three pressure chambers (Model 3000,Soil Moisture Equipment Corp.,Santa Barbara,CA,USA)were used for leaf water potential measurement,with six leaves for each treatment.

2.5.Final harvesting

Grain yield of all plants was determined using eight representative pots and adjusted to a moisture content of 0.14 g H2O g-1fresh weight.Aboveground biomass and yield components (i.e.,the number of panicles per pot and percentage of filled grains and grain weight)were determined from the eight representative pots.The percentage of filled grains was defined (specific gravity≥1.06 g cm-3)as a percentage of the total number of spikelets.The number of spikelets per panicle was calculated from the grain yield,grain weight (14% moisture content),and percentage of filled grains.

Number of spikelets per panicle=Grain yield per square meter/(Number of panicles per square meter×1 000-Grain weight×Percentage of filled grains)

2.6.Statistical analysis

Microsoft Excel was used to process all data.ANOVA was performed using SAS/STAT statistical analysis package(version 9.2,SAS Institute,Cary,NC,USA).Protected least significant difference (PLSD)was used for multiple comparison.SigmaPlot 10.0 was used to plot graphs.The statistical model used included sources of variation due to replication,year,SWP,N level,and the interactions of year×SWP,year×N level,and SWP×N level.Data from each sampling date were analyzed separately.Means were tested by least significant difference atP＜0.05 (LSD0.05).ANOVA indicated that there was no significant difference between years and no interaction between the year and the treatment.Therefore,the data were presented as the average across the two study years.

3.Results

3.1.Effects of treatments

Table 1 shows the ANOVA results (F-values)of grain yield and root traits/architecture of upland and paddy rice between/among years,SWPs,and N levels.Significant differences among three SWPs,three N levels and SWP×N level were observed (P＜0.05).The difference across years was not significant (P＞0.05).The interactions of year×SWP and year×N level showed no significant contribution to variations in grain yield and root traits/architecture (Table 1).Therefore,data from the two study years were averaged.

Table 1 ANOVA results (F-values)for grain yield and root characteristics of rice between/among years,soil water potentials(SWPs)and N levels

3.2.Soil and leaf water potentials

It took 5 to 7 days to reach SWP of -20 kPa and 8 to 12 days to reach SWP of -40 kPa (Fig.1-A and B).Water application from 10 days after transplanting to maturity was 390 mm to the mildly dried (-20 kPa)treatment and 225 mm to the severely dried (-40 kPa)treatment,which was only 54 and 31%,respectively,of that (720 mm)applied to the well-watered (0 kPa)treatment during the treatment period.

Fig.1 Soil water potential of the upland rice (japonica cultivar;A)and paddy rice (japonica cultivar;B)under various soil water potentials and N levels.0,-20,and -40 kPa represent well-watered,mildly dried,and severely dried treatments in pots after transplanting,respectively.LN,NN,and HN denote low amount,normal amount,and high amount of nitrogen application,respectively.Vertical bars represent standard error of the mean (n=6).

Fig.2 Leaf water potential of the upland rice (japonica cultivar;A,C,E and G)and paddy rice(japonica cultivar;B,D,F and H)under various soil water potentials (SWPs)and N levels.0,-20,and -40 kPa represent well-watered,mildly dried,and severely dried treatments in pots after transplanting,respectively.LN,NN,and HN denote low amount,normal amount,and high amount of nitrogen application,respectively.Measurements were made on the upmost fully expanded leaves at midday (11:30 a.m.)when SWPs were approximately -20 and -40 kPa (D1 and D2)and when plants were rewatered (W1 and W2).Vertical bars represent standard error of the mean (n=12).Different letters indicate significanc e of the same variety at the P=0.05 level,NS indicates no significance at the P=0.05 level.

The midday leaf water potentials of the two cultivars were measured when SWPs were approximately -20 kPa (D1)and -40 kPa (D2),and when treatments were rewatered(SWP of both treatments was 0 kPa,indicated by W1 and W2,respectively).During the soil drying period (D1 and D2),leaf water potentials were -0.39 to -1.09 MPa in the mildly dried (-20 kPa)treatment and -0.48 to -1.31 MPa in the severely dried (-40 kPa)treatment and were significantly lower than those -0.43 to -0.64 MPa in the well-watered (0 kPa)treatment (Fig.2-A,B,E and F).When the plants were rewatered (W1 and W2),leaf water potentials showed no significant difference among the SWPs or among the three N levels (Fig.2-C,D,G,and H).In the same SWP,leaf water potentials slightly increased with increasing N levels,but did not significantly differ among the three N levels,and the leaf water potential of upland rice was slightly higher than that of paddy rice.

3.3.Root traits

The number of adventitious roots in upland rice in various SWP conditions could be arranged in decreasing order as 0 kPa＞-20 kPa＞-40 kPa,which is similar to that observed in paddy rice (Table 2).At LN,NN,and HN levels,the number of adventitious roots decreased when the SWP was lower in both cultivars,but the extent of reduction was smaller for upland rice than for paddy rice.The average number of adventitious roots in upland rice was the highest with NN treatment and the lowest with LN treatment (i.e.,NN＞HN＞LN),whereas for paddy rice,there was no significant difference detected among the three N treatments.Changes in N concentrations had a stronger effect on upland rice than paddy rice,whose number of adventitious roots was not significantly altered by different N concentrations.Upland rice had 18.5% fewer adventitious roots than paddy rice.

Under different N levels,the average root diameter of upland rice was -20 kPa＞0 kPa＞-40 kPa;and the average root diameter of paddy rice was 0 kPa＞-20 kPa＞-40 kPa (Table 2).The average root diameter of upland rice was NN＞HN＞LN,whereas paddy rice had no significant difference in average root diameter among the three N levels.The effect of N level on the root diameter of upland rice was greater than that of paddy rice.The root diameter of upland rice was 33.5% larger than that of paddy rice(Table 2).

Table 2 shows that for upland rice,with -40 kPa SWP,but not -20 kPa,caused a significant decrease in the longest root length (LRL).In contrast,these low potentials caused a significant decrease in the LRL value of paddy rice(10.6% under -20 kPa conditions and 14.6% under -40 kPa conditions).Under LN conditions,the LRL value of uplandrice was the largest under -20 kPa condition,and the paddy rice value decreased with decreasing water potential.Both upland rice and paddy rice had the longest LRL at the LN level.The average LRL of upland rice was significantly larger than that of paddy rice by 25.4%.

Table 2 Root traits of the rice cultivars upland rice (japonica cultivar)and paddy rice (japonica cultivar)under various soil water potentials (SWPs)and N levels

The average root dry weight of upland rice under various water potential conditions was ranked in decreasing order as follows:-20 kPa＞0 kPa＞-40 kPa.For paddy rice,the rank was 0 kPa＞-20 kPa＞-40 kPa (Table 2).With both LN and NN treatments,upland rice showed the largest root dry weight under -20 kPa conditions,whereas root dry weight of upland rice with HN treatment and paddy rice with the three N levels were negatively correlated with SWP.Overall,the average root dry weight of upland rice was the highest at NN,and for paddy rice,the LN data were the highest.Compared to paddy rice,the root dry weight value of upland rice was 43.9% higher.

3.4.Root architecture

The 0-5 cm root volume ratios of upland rice were negatively correlated with SWP,whereas both 5-10 and 10-20 cm root volume ratios were positively correlated with the potential.These correlations were reversed in paddy rice,suggesting a major difference in the drought resistance between the two cultivars (Table 3).Root volume ratios of two cultivars showed similar trends for the 0-5 and 5-10 cm (but not the 10-20 cm),which were both positively correlated with the N level (Table 3).The 0-5,5-10,and 10-20 cm root volume ratios of upland rice were 30.5,24.6,and 44.9%,respectively,so the rank in decreasing order was 10-20 cm＞0-5 cm＞5-10 cm.For paddy rice,the ratios(0-5,5-10,and 10-20 cm)were 49.4,21.9,and 28.7%,respectively,and the rank in decreasing order was 0-5 cm＞10-20 cm＞5-10 cm (Table 3).

Table 3 Root architecture of the upland rice (japonica cultivar)and paddy rice (japonica cultivar)under various soil water potentials(SWPs)and N levels

3.5.Dry matter weight of aboveground part

Under 0 kPa condition,aboveground dry matter weight of the two cultivars at mature stage increased with the increase of nitrogen application level.Under -20 and -40 kPa conditions,dry matter weight of upland rice under HN was significantly higher than under NN and LN,but there was no significant difference between NN and LN.Dry matter weight of paddy rice under HN and NN was significantly higher than under LN,but there was no significant difference between HN and NN.For both cultivars,the effects of SWPs on average dry matter weight were ranked in decreasing order as follows:0 kPa＞-20 kPa＞-40 kPa (Fig.3-A and B).

3.6.Grain yield

Under 0 kPa condition,LN,NN,and HN yield data were significantly different for both cultivars:higher N level always induced the yield of upland rice,but the yield of paddy rice peaked with NN treatment.Under -20 kPa and -40 kPa conditions,the yield of both cultivars was the highest with NN treatment (Fig.3-C and D).For both cultivars,the effects of SWPs on yield were ranked in decreasing order as follows:0 kPa＞-20 kPa＞-40 kPa,but drought stress reduced the yield of upland rice plants to a smaller extent.The yield decreased as water potential was reduced under all N conditions for both cultivars,except that the yield of upland rice was the highest under -20 kPa conditions with LN treatment.For upland rice,the effects of SWPs on the yield in decreasing order were as follows:HN＞NN＞LN.For paddy rice,NN treatment induced the yield to the maximum (Fig.3-C and D).

Fig.3 Aboveground dry matter weight at mature stage and grain yield of the upland rice (japonica cultivar;A and C)and paddy rice(japonica cultivar;B and D)under various soil water potentials (SWPs)and N levels.0,-20,and -40 kPa represent well-watered,mildly dried,and severely dried treatments in pots after transplanting,respectively.LN,NN,and HN denote low amount,normal amount,and high amount of nitrogen application,respectively.Vertical bars represent standard error of the mean (n=6).Different letters indicate the statistical significance of the same variety at the level of P=0.05.

3.7.Correlation between root traits/architecture and yield/its components

These data indicate that for upland rice,if they generate more adventitious roots,then the yield is also high,as well as the number of grains per spike;if the root volume of 0-5 cm is large,then the number of spikes is significantly high,higher root volumes of 5-10 and 10-20 cm do not affect the number of spikes or seed setting rate,but are associated with higher numbers of grains per panicle,greater 1 000-grain weight,and higher yield (Table 4).Some interpretations of these data are as follows:for paddy rice,if these have more adventitious roots with higher weight,then seed setting rate,1 000-grain weight,and yield are higher.If the root volume of the 5-10 cm soil layer is higher,then the 1 000-grain weight and yield are higher (Table 4).

4.Discussion

Root growth of plants is closely related to soil environmental factors such as water,oxygen,temperature and fertility,among which water and fertility are the dominant factors,and they are interrelated and interact with each other (Aulakh and Malhi 2005;Wang X Bet al.2015).A large number of studies have demonstrated that during the adaptation process to the drought conditions,plant roots have evolved a stronger water uptake capacity (Langet al.2003;Wanget al.2009;Zhaiet al.2010;Comaset al.2013;Matsunamiet al.2013;Ugaet al.2013;Iijimaet al.2017;Pasolonet al.2017;Alouet al.2018;Fuhrmannet al.2018).Namucoet al.(1993)reported that the longer the main root of rice plants and the more branch roots they produce,the stronger their capacities to absorb and transport water,thereby exhibiting more resistance to drought stress.Some detailed studies further revealed that several root indices are correlated with the drought resistance in plants (Lu and Zhou 1994;Zhanget al.2006;Chenet al.2012;Suraltaet al.2018).Indices such as root diameter,root volume,and root length are positively correlated with the resistance,whereas root number is negatively correlated.This is not surprising because longer roots enable plants to absorb water in the deep layers of the soil;thus,this stronger water absorption capacity increases their resistance to drought stress.

The results of this study showed that for most root indices,including root diameter,LRL,root dry mass and root volume,the values of upland rice are larger than those of paddyrice,except for the number of adventitious roots.One novel finding of the present study is that moderate drought stress induced upland plants to produce stronger roots,as indicated by indices such as root diameter,LRL,and root dry weight,whereas for paddy rice plants,0 kPa seemed to be optimal for root development.With increasing drought stress,changes in upland rice root traits and yield were less significant compared to paddy rice,suggesting that upland rice plants have stronger water absorption capacity and drought resistance.These findings are consistent with some previous reports (Zhanget al.2006;Ameliaet al.2011).

Table 4 Correlation between root traits/architecture and yield/its components of upland rice (japonica cultivar)and paddy rice(japonica cultivar)under various soil water potentials and N levels (n=9)

Previous studies (Kondoet al.2000;Liaoet al.2001;Zhaoet al.2004)have shown that under drought stress,to compensate the water loss due to transpiration of the aboveground parts,rice roots elongate and are more distributed in the deeper cultivation layer to better absorb water.Based on these observations,moderate drought stress may be applied to promote root elongation,thereby increasing the proportion of deep roots (Kondoet al.2000;Liaoet al.2001;Zhaoet al.2004).In our study,we showed that with decreasing SWP,upland rice root volume ratio in the 0-5 cm tillage layer increased,whereas in 5-10 and 10-20 cm tillage layers,it decreased.In contrast,the paddy rice root volume ratio in the 0-5 cm tillage layer decreased,whereas in the 5-10 and 10-20 cm tillage layers,it increased.Our paddy rice data are consistent with some previous reports (Wuet al.1992;Zhaoet al.2004),yet the upland rice data were discordant (Wuet al.1992;Zhaoet al.2004).Furthermore,in our experiment,the root ratios in the tillage layer of upland and paddy rice significantly differed.Under no drought-stress (0 kPa)condition,the root volume ratios in the 0-5,5-10 and 10-20 cm tillage layers of upland rice decreased by 58%,increased by 63 and 80% compared with paddy rice,respectively.Under drought stresses (-20 kPa),in the 0-5,5-10,and 10-20 cm tillage layers,the root volume ratios of upland rice showed a 39% decrease,7% increase,and 60% increase compared to paddy rice,respectively.The percentage of upper layer of root system for upland rice was significantly smaller and middle and lower layer was larger than that of paddy rice.The results of this study indicate that the fluctuating SWP(0-20 and 0-40 kPa)and the persistent SWP (0 kPa)have great effects on the root structure and growth of the tow cultivars.On the other hand,the responses of root spatial configuration to soil water treatments were significantly different between paddy rice and upland rice.We propose that through breeding selection,upland rice roots feature a drought-resistant architecture,and thus when subjected to drought stress,the induced root growth is not as dramatic as that of paddy rice roots,we even observed a decrease in the root volume ratio of upland rice in the upper tillage layer and an increase in the lower tillage layer.While these results are not perfectly consistent with some previous studies for paddy rice,it is not unexpected because different cultivars were used in the current investigation.To fully understand the mechanism,comparative genetic studies using different upland cultivars should be conducted.

Paliwalet al.(1997)suggested that N considerably promotes the yield of upland rice,and an increased application of N fertilizer also helps their uptake of N,P,K and Ca.Studies conducted by Fageriaet al.(2014)and Zhanget al.(2005)showed that N utilization in upland rice grains is more efficient than paddy rice grains,and therefore a major difference was observed in terms of their N uptake rate when grown in dry land.Fanet al.(2007)measured several root indices of different N concentrationtreated paddy rice plants.They found that most indices,e.g.,each adventitious root length,each adventitious root diameter,and each adventitious root weight decreased with higher N concentrations,yet only root number per plant,root dry weight,and total root length increased.A lower increase rate was noticed when N concentrations reached a threshold.Similar results were observed by Schiefelbein and Benfey (1991).In their study,they also observed that within a certain range,increasing the N supply promotes root growth,but also induces the plants to develop thinner roots,resulting in a larger root surface.The results of this study show that with the increase in N application level,the number of adventitious roots,root diameter,root dry mass,and root volume for both cultivars peaked with NN treatment,whereas under different nitrogen levels,the root differences between the two varieties were different.There were no significant differences between NN treatment and HN treatment in upland rice and between LN treatment and HN treatment in paddy rice.

While our results are largely consistent with previous reports,some novel aspects are also observed.A similar finding is that as N supply concentration increases,the adventitious roots become thinner and shorter (Zhanget al.2005);however,we showed that the number of adventitious roots and root dry weight with NN treatment were the largest,which is contrary to the reported phenomenon that these indices are always correlated with N concentration (Fanet al.2007).We believe that one of the possible reasons is the use of different experimental materials (i.e.,rice cultivars).Notably,our results also revealed that for both cultivars,root volume ratios increased in the 0-5 and 5-10 cm tillage layers but not in the 10-20 cm tillage layer.Here,the upland rice showed a similar trend to the paddy rice,although the degree was higher,particularly in the 0-5 cm data.These changes in the root system indicate that upland rice might be more sensitive to N fertilizers than paddy rice plants.This is possibly because of the differences in their genotypes,as nutrient-insensitive cultivars show minimal changes in root responses to changes in nutrient supply (Turner 2010).

Correlation analysis showed that for upland rice,the yield was correlated with indices such as the number of adventitious roots and the root volumes in the 5-10 and 10-20 cm tillage layers (r=0.717＊-0.958＊＊),whereas for paddy rice plants,yield was correlated with the number of adventitious roots,root dry weight,and the root volume in 0-5 cm tillage layer (r=0.770＊＊-0.858＊＊).The results showed that the yield of upland rice increased with the increase of the number of adventitious roots,root volume and root amount in the middle and lower layers,and the yield of paddy rice increased with the increase of the number of adventitious roots and root dry weight in the upper layer.

The results also showed that excessive nitrogen input,such as under -20kPa and -40kPa and HN treatment,did not produce high yields for both cultivars.One possible reason is that this study was carried out under pot culture and water stress conditions,the water uptake by upland rice roots was limited by pot volume,especially upland rice roots could not give full play to its advantages,for example,it could not give full play to its advantages of water absorption as upland rice roots did in the field.The results of this study also showed that the interaction between SWP and nitrogen level on the yield of the two cultivars reached a very significant level (Table 1).Because of water stress treatment,even under HN conditions,both cultivars did not achieve high yield.Secondly,under HN conditions,both upland rice and paddy rice may have redundant growth.For example,there is a lot of redundancy in biomass,especially under the treatment of 0 kPa and HN for paddy rice,this redundancy becomes a waste and burden of rice plants,which is obviously unfavorable to high yield.However,under moderate water stress (-20 kPa)and NN levels for upland rice,as well as sufficient water (0 kPa)and LN conditions for paddy rice,both cultivars could reduce excess vegetative growth,ineffective tillers and excess transpiration water,thus reducing water and nitrogen used in production,and improving yield and water and fertilizer utilization efficiency.Therefore,proper input of fertilizer and water is beneficial to control redundant growth of rice and increase yield.

This study was carried out under potted conditions,and the main difference between potted rice and field rice is that:first,potted soil is generally ploughed,screened and blended tillage-layer soil.Potted rice can only absorb nutrients from the tillage layer soil,while field rice can absorb nutrients not only from the tillage layer,but also from the bottom soil.And the bottom of the soil in pot culture is completely closed,which makes the oxidized zone of rice rhizosphere small and the soil reducible.Some elements which are difficult to dissolve in water and easy to deposit on the root surface can inhibit the development of rice root system,and then affect the growth and yield of rice.Secondly,the root system shape of potted rice is cylindrical in the soil,while that of field rice is umbrella in the soil,the space of potted soil hinders the divergent growth of rice roots and cannot make them better absorb nutrients in the soil.Zhanget al.(2016)showed that the yield of rice planted in large fields of the same variety in the same area was about 1.5 times higher than that of potted rice.The author speculated that the optimum combination of water and nitrogen in pot of this study could increase rice yield by more than 0.5 times (Zhanget al.2016)if it was used in rice field management.In order to prevent the growth of potted rice from the influence of soil impermeability and soil space size,water infiltration device should be set at the bottom of the pot to expand the size of the pot bottom,so as to improve the growth environment of potted rice,make it closer to the field test conditions,and make the research results have guiding significance for the field rice production.

Based on our results,to grow rice plants in the field,it is necessary to first consider some important properties of the rice cultivar such as drought resistance,fertilizer tolerance,and root distribution.Targeted water management and fertilization based on these findings can dramatically increase the yield while using less resources.We also propose a general guideline for rice cultivation:under sufficient water or moderate drought conditions,paddy rice cultivars are preferable;however,if the land is dry and less fertile,then upland rice cultivars might be the better choice.Rice cultivars of different genotypes respond differently to water and nutrient supply,and the mechanisms require further investigation.

5.Conclusion

The roots of upland rice are mainly distributed in the 10-20 cm tillage layer,whereas most paddy rice roots are in the 0-5 cm tillage layer.Drought stress had a less impact on upland rice than on paddy rice,while N fertilizer could significantly promote the yield of upland rice than paddy rice.These results indicate that the combination of -20 kPa SWP and NN in upland rice and 0 kPa SWP and LN in paddy rice promotes the growth of the root system during the middle and late stages,which in turn may decrease the requirements for water and N fertilizer and increase rice yield.The findings of the present study provide useful information that may be employed in achieving higher grain yield and high resource use efficiency and provide insights into understanding the interaction between water and N on upland and paddy rice growth.

Acknowledgements

We are grateful for grants from the National Natural Science Foundation of China (31671617),the National Key Research and Development Program of China (2016YFD0300502,2016YFD0300206 and 2018YFD0301306),the Priority Academic Program Development of Jiangsu Higher Education Institutions (PAPD),China.