Bl Shi-ting, LUO Xiang-yu, ZHANG Chen, Ll Peng-fei,2, YU Cai-lian, LlU Zhi-lei,2#, PENG Xian-long,2#

1 College of Agricultural Resources and Environmental Sciences, Northeast Agricultural University, Harbin 150030, P.R.China

2 Key Laboratory of Germplasm Enhancement, Physiology and Ecology of Food Crops in Cold Region of Ministry of Education,Northeast Agricultural University, Harbin 150030, P.R.China

3 The School of Material Science and Chemical Engineering, Harbin University of Science and Technology, Harbin150080, P.R.China

Abstract The relationship between the fate of nitrogen (N) fertilizer and the N application rate in paddy fields in Northeast China is unclear, as is the fate of residual N.To clarify these issues, paddy field and 15N microplot experiments were carried out in 2017 and 2018, with N applications at five levels: 0, 75, 105, 135 and 165 kg N ha–1 (N0, N75, N105, N135 and N165,respectively).15N-labeled urea was applied to the microplots in 2017, and the same amount of unlabeled urea was applied in 2018.Ammonia (NH3) volatilization, leaching, surface runoff, rice yield, the N contents and 15N abundances of both plants and soil were analyzed.The results indicated a linear platform model for rice yield and the application rate of N fertilizer, and the optimal rate was 135 kg N ha–1.N uptake increased with an increasing N rate, and the recovery efficiency of applied N (REN) values of the difference subtraction method were 45.23 and 56.98% on average in 2017 and 2018, respectively.The REN was the highest at the N rate of 135 kg ha–1 in 2017 and it was insignificantly affected by the N application rate in 2018, while the agronomic efficiency of applied N (AEN) and physiological efficiency of applied N (PEN) decreased significantly when excessive N was applied.N loss through NH3 volatilization, leaching and surface runoffwas low in the paddy fields in Northeast China.NH3 volatilization accounted for 0.81 and 2.99% of the total N application in 2017 and 2018, respectively.On average, the leaching and surface runoffrates were 4.45% and less than 1.05%, respectively, but the apparent denitrification loss was approximately 42.63%.The residual N fertilizer in the soil layer (0–40 cm) was 18.37–31.81 kg N ha–1 in 2017, and the residual rate was 19.28–24.50%.Residual 15N from fertilizer in the soil increased significantly with increasing N fertilizer, which was mainly concentrated in the 0–10 cm soil layer, accounting for 58.45–83.54% of the total residual N, and decreased with increasing depth.While the ratio of residual N in the 0–10 cm soil layer to that in the 0–40 cm soil layer was decreased with increasing N application.Furthermore, of the residual N, approximately 5.4% was taken up on average in the following season and 50.2% was lost, but 44.4% remained in the soil.Hence, the amount of applied N fertilizer should be reduced appropriately due to the high residual N in paddy fields in Northeast China.The appropriate N fertilizer rate in the northern fields in China was determined to be 105–135 kg N ha–1 in order to achieve a balance between rice yield and high N fertilizer uptake.

Keywords: fate of N fertilizer, NH3 volatilization, leaching, surface runoff, residual nitrogen, yield

1.lntroduction

As a critical factor for increasing rice production and meeting the dietary needs of the ever-growing population,mineral nitrogen (N) application has been regarded as one of the main techniques for increasing rice yields(Ray and Foley 2013; Yanet al.2014).However, there are differences in N use efficiency (NUE) between China and other regions of the world, as agricultural production has advanced by 25–35 to 52–67% due to high rates of N fertilizer input (Liuet al.2015; Nkebiweet al.2016; Yuet al.2022; Caiet al.2023).Heilongjiang Province is the second largest rice-growing region in China (accounting for up to 13.69% of the total rice production), and it plays an important role in ensuring food security (NBS 2023).Based on a comparison of early rice in southern China(where NUE is 18.7–37.5%), lower rates of N application and relatively higher NUE (25–45%) were achieved in this northeastern province (Lianget al.2021; Liuet al.2022; Xuet al.2022), but even higher values have been reported for regions at similar latitudes in Japan (45–65%)(Olusegunet al.2022).Therefore, finding an effective approach to improve paddy field NUE and reduce N loss in northeastern China remains a crucial challenge.

N fertilizer is always at risk of being lost through NH3volatilization, leaching, denitrification or runoff due to its active chemical properties (Juet al.2009), and close correlations were observed between N loss and the N application rate, fertilization practices and type of fertilizer(Panet al.2016).Up to 60% of N fertilizer is lost by NH3volatilization worldwide, although some studies have shown a negligible amount of NH3loss (Chienet al.2009;Xueet al.2014; Liet al.2017).Therefore, a relatively higher input of N fertilizer would subsequently result in an increase in NH3volatilization (Chenet al.2015).In southern China, 67.88% of N loss in paddy fields (35.3 kg N ha–1on average) occurred through NH3volatilization,which is the main pathway of N loss (Wanget al.2015),while significantly less NH3volatilization was recorded in northern China (less than 15%) (Wanget al.2018).Significant amounts of N account for 1.4–6.4% of N application in China, and they are also lost through other pathways including leaching and runoff(Penget al.2011;Wanget al.2015).Some studies showed that surface N runoffin paddy fields is relatively low (Liet al.2018); while other studies suggest that N runoffaccounts for 5.6–68.2%of the total N application, which is too high to be ignored(Jinet al.2006).Therefore, further systematic studies on N fertilizer loss in the paddy fields of Northeast China are needed.

Certain amounts of applied N remain in the soil,which are regarded as residual N, while the remaining N fertilizer is lost or taken up by rice plants (Kadiyalaet al.2015).The complex influences of multiple factors have resulted in a lack of consensus about the availability of residual N (Xuet al.2015).The uptake of residual N prior to soil N by successive crops in rotation systems was attributed to the significantly higher availability of residual N compared to soil N (Broadbent 1980).A fairly high percentage of NH4+-N was detected in the soil after rice harvest, while the dominance of NO3--N was observed after upland crop production (Buet al.2012).Duet al.(2022) showed that the residual N fertilizer was organic N,and the range of total residual N was 35–55 kg N ha–1with a soil recovery rate of 30–40%.In Myanmar paddy fields,the soil recovery rate was 6–22% (Eldridgeet al.2022).The residual N fertilizer varies significantly between different types of soil.According to Liet al.(2023), the majority of N applied to sandy soil was taken up by plants or subjected to environmental loss, and only a small percentage remained in the soil.In the following year,only 2.3–3.2% of the residual N was taken up by plants,while this percentage reached 5.6–9.8% in loamy soils,and very significant differences were observed among soil types (Liet al.2023).Studies on the fate of residual N have generally focused on dryland or rice field systems in warm regions (Quanet al.2020), whereas few studies have been conducted on residual N in paddy fields in cold regions.

Crop N uptake is closely related to soil and climatic conditions (Yang Tet al.2019; Yanget al.2021).Rice fields in China vary greatly in latitude from the north to the south, and the soil and climatic conditions are quite different as well (Zhanget al.2021).Soil microbial activity and rice N uptake are directly affected by temperature(Makinoet al.2021; Eldridgeet al.2022; Zhanget al.2022).A higher temperature increases the mineralization and release of organic N from the soil while promoting the hydrolysis of urea (Maet al.2020; Xuet al.2021),which also increases N losses.Precipitation can rapidly reduce the ammonium N content and pH of field surface water, and these factors as well as the temperature can inhibit NH3volatilization (Shaet al.2019).Large amounts of precipitation can also increase the risk of runoffdepending on the amount of precipitation and the timing of its occurrence (Tianet al.2007).The soil type determines nutrient release and fertilizer sorption in the soil, causing large variations in microbial communities and physicochemical properties in different soil types(Kanthilankaet al.2021).Consequently, considering the geographical characteristics of the northeastern rice region, especially in Heilongjiang Province, the soil and climatic factors might be one of the reasons for the differences in the N fertilizer fate between Northeast China and other rice-growing regions.

In light of the issues mentioned above, one reasonable hypothesis is that with lower N fertilizer loss and higher N uptake by rice plants, as well as the retention of residues in the soil, a higher NUE will be observed in the paddy fields of Northeast China.To test this hypothesis, two years of field experiments with stable isotope15N tracer technology were conducted to study the N fertilizer uptake and fate in paddy fields in Heilongjinag Province,Northeast China.The objectives were: (1) to clarify the relationship between the N amount and N fate; (2) to determine the occurrence of NH3volatilization, leaching and surface runoffloss characteristics in order to estimate the risk of N fertilizer loss in paddy fields in cold regions;and (3) to determine the crop uptake of residual N fertilizer in the following year.The results of this study can provide theoretical support for increasing NUE in paddy fields.

2.Materials and methods

2.1.Site description

Two years of field experiments were performed in Huihuang Village, Longfengshan Town, Wuchang City, Heilongjiang Province, China (44°53′1.1′′N,127°31′35.4′′E), which is located in a temperate continental climatic zone.The average precipitation, daily temperature and rainfall during the growth seasons are shown in Fig.1.The field soil on the site was under an alluvial soil that developed from clay ice water sediment(Wrb 2014).The natural abundance of15N in the soil was 0.365%.In April 2017, soil samples from the 0–20 cm layer were collected before field preparation to obtain measurements of organic matter (39.0 g kg–1), total N (1.74 g kg–1), Olsen-P (69.8 mg kg–1),exchangeable K (9 7.1 m g k g–1), and pH (6.17;soil:water=1:2.5) (Sparkset al.1996).

Fig.1 Mean daily temperature and precipitation during the 2017 and 2018 growing seasons.

2.2.Experimental design

Five N application rate treatments were set as N0 (0 kg N ha–1), N75 (75 kg N ha–1), N105 (105 kg N ha–1), N135(135 kg N ha–1), and N165 (165 kg N ha–1).The N application rate was set based on the method of Penget al.(2015, 2019), and the recommended N rate was 105 kg N ha–1.In this study, 165 and 75 kg N ha–1were chosen as the respective upper and lower limits of fertilizer application because the corresponding average amount and the lowest value used in production are approximately 140 and 90 kg N ha–1(Penget al.2019).Plots with an area of 80 m2(8 m×10 m) were arranged in a complete randomized block design with four replicates per treatment, resulting in a total of 20 plots.The plots could be irrigated and drained off individually with an exclusive water gate system.

15N microplot experiments were carried out in each corresponding treatment plot to determine the apparent N loss, which was calculated as the total15N input minus the recovery, and included the uptake by rice (aboveground parts of plant, i.e., stems, leaves and seeds) and the residual N in the soil layer (0–40 cm).The microplot device was framed by a PVC tube with an area of 0.19 m2(49 cm in diameter, 50 cm in depth).The device was buried in the soil to a 30 cm depth, leaving 20 cm aboveground to prevent the loss of15N with water and the entry of external N.The drainage of the microplot was achieved by allowing flooded water to naturally percolate downward to the underground after the15N fertilizer was applied.The microplot was cultivated under the same water management practices as the corresponding field plot experiments.The amount of15N-labeled fertilizer was calculated for each N application rate treatment based on the area ratio between the plot and microplot to ensure that the same amount of N was applied per unit area.In 2017,15N-labeled urea tracer (15.24% abundance) was applied in the microplot, and in 2018, the same dose of regular urea was applied.

For the N treatment, 40% of the urea amount was basal fertilizer, 30% was the first top-dressing fertilizer, and the final 30% was the second top-dressing fertilizer.The total P and K rates of each treatment were maintained consistently according to nutrient balance.All P (50 kg P2O5ha–1with superphosphate) and 50% of the K (45 kg K2O ha–1with potassium chloride) were applied during the basal fertilizer period, and the final 50% of the K (45 kg K2O ha–1) was applied with the second top-dressing urea application.All fertilizer was sprinkled by hand.

A high-yielding rice cultivar (OryzasativaL.),Songjing 3, was selected for this experiment.The spacing of rice seedlings (at approximately the three-leaf stage) was 30 cm×13 cm, and consistency was observed between the treatments.There were four hills in each microplot with three seedlings per hill.Herbicides were used to remove weeds, and insecticides and fungicides were used to control pests and diseases as needed.Table 1 shows the specific timing of these operations.

2.3.Sampling and measurements

A precipitation collector (ZJC-V, Zhejiang Hengda Inc.,China) was mounted near the test site to collect the precipitation (Fig.1).The collector was equipped with eight sampling containers mounted 150 cm from the ground.After each rain event, the water volume was recorded, and then 500 mL of rainwater was sampled.Air, water, and soil (7 cm) temperatures were recorded using temperature probes (Onset HOBO Inc., USA).

The air flow enclosure method was used to measure the daily flux of NH3volatilization (Tianet al.1998).The diameter of the volatility chamber was 200 mm, and the height of the chamber above the ground was adjusted according to the depth of the soil.The air flow ratewas set to 15–20 chamber volumes per minute using a vacuum pump.NH3volatilization was measured for 10 consecutive days starting on the day after fertilization from 8:00 to 10:00 a.m.The air was continuously pumped for 2 h through a NH3absorber (0.3% H2SO4).The chambers were moved to avoid any disturbance caused by environmental conditions after the measurement.

Table 1 Dates of cultivation and management in 2017 and 2018

N loss due to NH3volatilization was calculated as:

whereAis NH3volatilization (kg N ha–1);nis the number of days after fertilization;cis the concentration of NH4+in the absorption solution (mg N L–1);vis the volume of the absorption solution (mL);sis the area of the closed chamber (m2); and 12×10–5is the conversion coefficient.

A vacuum hand pump was used to collect the leached samples with a time interval of approximately 10 days during the growing season, and one additional sampling was conducted after each fertilization event.The water that remained in the porous pipe before sampling was removed to avoid contamination.Samples were collected after 24 h.Leached water samples were frozen at–20°C.The NH4+and NO3-concentrations of the leached water samples were analyzed using an Auto Analyzer 3(Bran+Luebbe, Germany).The leached N was calculated as:

whereLis the cumulative amount of leached N (kg N ha–1); the time-interval weighted N concentration was measured at 60 cm soil depth in 2017 and 100 cm soil depth in 2018 (mg N L–1); andVis the total amount of leached water (m3ha–1).

After heavy rainfall events, runoffwater samples were collected immediately, and the volume of runoff was calculated based on the amount of precipitation.Then after the overall volume was recorded, 250 mL of runoffwater was sampled and frozen, and the NH4+-N and NO3--N concentrations of the samples were analyzed using an Auto Analyzer 3.The N loss due to runoffwas calculated as:

whereRis the N loss through runoff(kg N ha–1);Ciis the mineral N concentration in the runoffwater samples (mg N L–1);Sis the area of the plot (m2);Pis precipitation amount (mm); and ΔHis the water level difference of the paddy field (mm).

The total N loss rate was calculated as:

where the N loss rate is the ratio of N loss to the N application rate (%);Ltis N loss in the N fertilizer treatment (kg N ha–1);L0is N loss in the treatment without fertilizer (kg N ha–1); andFtis the fertilizer applied in the N application treatment (kg N ha–1).

The weight of rice at a calculated moisture content of 14.5% in an area of 5 m2in the plot was used to determine the rice yield.The rice plants were collected manually in each microplot for biomass and N content determinations at physiological maturity.Rice samples were divided into panicles and stems plus leaves, which were dried at 80°C to constant weight.The N concentrations in tissues were determined with a continuous-flow analyzer after micro-Kjeldahl digestion (Sparkset al.1996), and the total aboveground N uptake was calculated.Simultaneously,soil samples (0–40 cm depth) were collected at three points in each microplot, and each sample was divided into 0–10, 10–20, 20–30 and 30–40 cm layers.Then, all subsamples from the same depth were mixed, air-dried,and sieved to determine the15N abundance and total N content.To avoid cross-contamination, the sequence of all operational procedures was carried out from low to high atom%15N.

The soil and plant sample15N abundance was measured by an isotope mass spectrometer (MAT-251,USA) with analytical error ±0.02%.All data relevant to15N abundance are denoted as the atom percent excess with background abundance correction (i.e., 0.366%).The amount and percentage of N derived from fertilizer(Ndff) were calculated based on Hauck and Bremner(1976).The15N fertilizer use efficiency (FNUE; %) and the residual N fertilizer in the soil (15N residual rate) were calculated as follows:

where Ndffis the percentage of plant N or soil N derived from15N fertilizer (%);cis the plant or soil15N abundance(%);ais the abundance of fertilizer15N (%); andbis the natural abundance of15N in untreated plants or soil (%).

where15Nsoilis the residual N fertilizer at different soil depths (kg N ha–1); and residual N fertilizer was calculated as the sum of residual N in the soil at different depths from 0–40 cm.

The fate of residual N fertilizer was calculated as follows:

2.4.Statistical analyses

Statistical analyses were performed using SAS 8.1.Twoway ANOVA was used to analyze the interaction between the N rate and year.Significant differences between individual means were determined using Fisher’s least significant difference (LSD) and Duncan’s test at the 5% confidence level based on four biological replicates for each treatment.The values are presented as the mean±SE.

3.Results

3.1.Rice yield, N uptake and NUE under different N application rates

The interaction between the N application rate and year was significant (Table 2).The N application rate significantly affected the rice yield.Compared with N0,the rice yield increased by 47.83% (30.69–58.01%)and 73.72% (57.38–87.91%) on average in 2017 and 2018, respectively, in the other groups.The dosage of N fertilizer increased the rice yield, while the rice yield increase was nonsignificant when the N application rate was higher than 135 kg ha–1, so the relationship between the rice yield and N rate can be described as linear plus platform.

Compared with N0, the uptake and accumulation of N were enhanced by 88.93% (47.21–124.81%) and 127.04% (82.59–170.75%) on average under the N application treatments in 2017 and 2018, respectively(Table 2).As the N application rate increased, the N accumulation increased, with the N application rate of 165 kg N ha–1yielding the highest N accumulation.The15N tracer experiment revealed similar results: as the N application increased, the uptake of15N fertilizer also increased significantly.However, the average accumulation of15N was only 32.89 kg N ha–1, which is significantly lower than the 117.57 kg N ha–1observed in the experiments without15N application in 2017.In 2018,the rice took up the15N fertilizer that remained in the soil from 2017, and the highest N accumulation was also observed at 165 kg N ha–1.

The average RENvalues were 45.23 and 56.98% in2017 and 2018, respectively.The RENfirst increased and then decreased with increasing N application rate in 2017.Rice RENwas the highest at 135 kg N ha–1but the difference was insignificant at 165 kg N ha–1, while the N application rate had little effect on RENin 2018.The RENwas lower in the15N tracer experiment, and the effect of the amount of N applied on the15N recovery rate was not significant.

Table 2 Grain yield and recovery of N fertilizer in the aboveground parts of plants1)

The average AENvalues were 24.12 and 35.18 kg kg–1in 2017 and 2018, respectively.With the addition of N,AENfirst increased and then decreased, with the highest AENat N application rate of 105 kg ha–1in 2017, and a significant decrease was observed in 2018.For PENand PFPN, negative effects were found under higher N rates;i.e., the more N fertilizer applied, the lower the PENand PFPNin both years.Moreover, the N harvest index also decreased with increasing N application, and the values were 66.64 and 63.26% in the N application treatments in 2017 and 2018, respectively.

3.2.Residual N fate and uptake

Distribution of residual N in 2017 Total residual N in the 0–40 cm soil layer was 18.37–31.81 kg N ha–1, for a residual rate of 19.28–24.50%, in 2017.The residual N fertilizer was mainly concentrated in the 0–10 cm soil layer, accounting for 58.45–83.54% of the total residual N (Fig.2).As the fertilizer dose increased, the residual15N fertilizer in the soil increased significantly in both years, with the highest value measured at 165 kg N ha–1.However, the ratio of residual15N to total N application decreased as the N application rate increased in 2017, and the residual rate of N165 was only 19.28%.Furthermore, the residual15N of fertilizer in the soil decreased with increasing soil depth, and a higher N application rate still resulted in a higher residual N content in each soil layer.Interestingly, the ratio of residual N in the 0–10 cm soil layer to that in the 0–40 cm soil layer decreased with increasing N application, and the minimum ratio was observed under N165.In contrast, the ratio of residual N below 10 cm to that in 0–40 cm showed a trend that was the opposite of that above 10 cm, and higher N application, especially 165 kg N ha–1, resulted in the maximum residual rate.In other words, the residual N was mainly concentrated in the 0–10 cm soil layer under a lower N supply.

Fig.2 Variations in residual N at different soil depths during the 2017 and 2018 growing seasons.N75, N105, N135 and N165 represent N rates of 75, 105, 135 and 165 kg ha–1, respectively.Values are mean±SE (n=4).

Fate of soil residual N in 2017 After N application in 2017, 7.53–15.48 kg N ha–1was available as residual15N fertilizer in 2018, which was approximately 8.23–10.03%of the total applied N (Table 3).In 2018, the N application rate of 165 kg ha–1still had the maximum residual N among all the N application treatments.The trends of residual N in each soil layer were similar to those in 2017, in which the 0–10 cm residual N was the maximum among all the soil layers and it decreased with increasing soil depth.There was no significant relationship between the ratios of residual N in each soil layer to the whole soil layer and the N application rate.In 2018, approximately 5.43% of the residual N fertilizer from 2017 was used in rice cultivation, and the RENimproved with increasing N application, as the peak RENvalue was acquired under the 165 kg N ha–1application.Residual N from 2017 was mainly lost in 2018, at 12.73 kg N ha–1on average,accounting for 50.18% of the residual N content from 2017.Moreover, approximately 44.41% of the residual N was still available in the soil in the next rice season.However, the correlation between N loss and residual N among the N application rate treatments remained nonsignificant.

3.3.N loss

NH3volatilization A significant interaction between the N rate and NH3volatilization in the two years (P<0.01) was shown in this study (Fig.3).The total NH3volatilization in paddy fields in Heilongjiang Province was low.The average NH3volatilization levels after N application were 0.99 and 3.58 kg N ha–1in 2017 and 2018, accounting for 0.81 and 2.99% of the total N application, respectively.LossesviaNH3volatilization in the N application treatments were 3.31 and 3.64 times higher than those in N0 in 2017 and 2018, respectively, and NH3volatilization increased significantly in response to an increase in the N rate.However, no significant difference was found between the ratios of NH3volatilization to the N rate among the different treatments in the two years.Inaddition, the NH3volatilization losses from the basal and first top-dressing fertilization were significantly higher than that from second top-dressing fertilization.

Table 3 Uptake, residual and loss of residual N in 20171)

Fig.3 NH3 volatilization from paddy soil under different N application rates during 2017 and 2018.N0, N75, N105, N135 and N165 indicate N rates of 0, 75, 105, 135 and 165 kg ha–1, respectively; Y, year; N, N rate; Y×T, interaction between year and N rate.Values are the mean±SE (n=4).Different letters represent significant differences (P<0.05) based on Duncan’s test.＊＊ represents significance at P<0.01.

N leaching The average N losses by leaching in this study were 5.41 and 5.05 kg ha–1in 2017 and 2018,respectively, accounting for 0.86–2.38% of total N input on average (Fig.4).Compared with paddy fields without N application, the amounts of N lost through leaching were 36.45 and 71.88% higher on average under N application in 2017 and 2018, respectively.The trends of N leaching loss were similar in 2017 and 2018, as N loss increased with an increase in the N rate, but there was no significant difference between the N leaching loss rates among the relatively high N rates.

Surface runoff N lossviasurface runoffwas observed

Fig.4 N leaching losses in northeastern China paddy fields in 2017 and 2018.N0, N75, N105, N135 and N165 represent N rates of 0, 75, 105, 135 and 165 kg ha–1, respectively; Y, year;N, N rate; Y×T, interaction between year and N rate.Values are the mean±SE (n=4).Different letters represent significant differences (P<0.05) based on Duncan’s test.＊＊ represents significance at P<0.01; ns represents not significant.

in both years (Table 4).Two surface runoff events took place in 2017, and only one occurred in 2018.In 2017, the average N lossesviasurface runoff for the N application treatments and N0 were 3.10 and 3.04 kg N ha–1, respectively, and no significant difference was found between the treatments or N concentrations and N loss in surface water among the different N rates in the two surface runoffevents.In 2018, N lossviasurface runofffor the N application treatments was 1.41 kg N ha–1on average, which was 9.59 times higher than that under N0.The N concentration of surface runoff water and N lossviasurface runoffwere both significantly enhanced by an increasing N rate.The ratios of N lossviasurface runoffto the N rate were less than 0.1 and 1.05% on average in 2017 and 2018, respectively, with no significant difference among the N application treatments.

Apparent denitrification N lossviaapparent denitrification in 2017 was 27.25–74.79 kg N ha–1, and it also increased with an increase in the N rate (Fig.5).The average apparent denitrification was 52.25 kg N ha–1, and the average apparent denitrification rate was 42.63%.The apparent denitrification N loss under N165 was 2.74 times higher than that under N75 (P<0.05).The apparent denitrification N loss rate first increased and then decreased with an increasing N fertilizer rate, with no significant difference observed among the N rates of 105–165 kg ha–1.

4.Discussion

4.1.Effects of different N application rates on grain yield and N use efficiency

In this study, the average RENobtained by the subtractionmethod was approximately 45% in 2017, while a significantly lower result of less than 30% was obtained from the tracer method (Table 2), similar to the findings of other studies.The presence of the priming effect, the phenomenon of increasing soil N release with increasing fertilizer application, might be one of the reasons why a lower NUE was obtained from the tracer method than from the subtraction method (Yanget al.2021).

Table 4 Inorganic N concentration in the runoffsolution, N runoffloss, and the ratio of N runoffloss to applied N for three runoffevents

Fig.5 Apparent denitrification and denitrification rates under different N application rates in 2017.N0, N75, N105, N135 and N165 represent N rates of 0, 75, 105, 135 and 165 kg ha–1,respectively.Values are the mean±SE (n=4).Capital letters and lowercase letters represent significant differences (P<0.05)in the apparent denitrification and apparent denitrification rates,respectively, based on Duncan’s test.

Rice variety, fertilizer application and the growing environment have significant impacts on rice yield (Tianet al.2022; Tuet al.2022).In this study, the rice yield and N rate were fitted with a linear plus platform model.That is, rice yield did not decrease when the N rate exceeded a certain amount (135–165 kg ha–1).However,N uptake was significantly promoted by a higher N rate(Table 2), which is similar to the results of previous studies(Chenet al.2019; Zhuet al.2022).The main reason for these results is that most high-yielding rice varieties are highly resistant to lodging; therefore, lodging seldom occurs under excessive N application, so the rice yield is still maintained at a high level (Zhanget al.2014, 2016).In contrast, the yields of rice varieties with lower lodging resistance would be greatly reduced due to lodging and diseases caused by N excess (Penget al.2021; Zhouet al.2022).Additionally, the N harvest index decreased with an increasing N rate since more N was stored in the stem; hence, there was high luxury N uptake by the high-yielding rice varieties (Table 2).Although RENwas not significantly affected by the N rate, the AENand PENwere both significantly reduced.Consequently, excessive N application in high-yielding rice varieties, as well as increased N loss and environmental stress, are very likely to occur if RENis the only factor considered when formulating N fertilizer recommendations.In this case,although high-yielding rice varieties can reduce the risk of yield reduction, the environmental risks are higher due to luxury N uptake.Considering the environmental aspects is vital when determining the total N fertilizer amount and optimizing the N rate for N-sensitive rice varieties.Hence, the appropriate N rate for high-yielding varieties in Heilongjiang Province is 105–135 kg ha–1.However,finding the suitable N rate for other rice varieties in this region still needs further exploration.

4.2.Effects of different N application rates on residual N

In addition to being taken up by crops and lostviaNH3volatilization, leaching and surface runoff, a large portion of applied N fertilizer remains in the soil.In general, N fertilizer residues after harvest usually account for 15–30% of the N applied (Chenet al.2019).In this study, the total residual N in the 0–40 cm soil layer in Heilongjiang Province was 18.37–31.81 kg N ha–1, for a residual rate of 19.28–24.50% (Fig.2).In the double cropping rice region of southern China, the residual N rate was reported to be approximately 22–23% (Wanget al.2015), which is similar to our study results.While it was mainly concentrated in the 0–10 cm soil layer, residual N increased constantly with an increasing N rate, but the lower the N rate, the higher the residual ratio (Fig.2).Under a N rate of 75 kg ha–1, nearly 84% of the overall residual N was maintained in the 0–10 cm soil layer, while the residual N amount was only 58.45% under a N rate of 165 kg ha–1.The residual N content significantly decreased with increasing soil depth, with 8.96 kg ha–1of residual N remaining in the 40–60 cm soil layer under a N rate of 135 kg ha–1, implying that more N would migrate into the lower soil layer with irrigation and wet–dry cycling if excessive N was applied (Chenet al.2019).The plow pan of paddy soil in Heilongjiang generally exists at a depth of 13–18 cm,indicating that the residual N fertilizer in the 40–60 cm soil layer is seldomly utilized by rice; therefore, the risk of N loss by leaching is greater (Wanget al.2017).Increasing the residual N fertilizer in the 0–20 cm soil layer can reduce the N fertilizer loss and replenish the soil N pool to a certain extent, eventually contributing to the protection and restoration of paddy soil fertility in the black soil region (Zhanget al.2018).

The uptake of N fertilizer by rice in Heilongjiang occurred mainly in the first year, while the amount of residual N from the previous year taken up in the second year was only 5.43%, on average (Table 3).The residual N in 2018 still increased with a higher N rate, indicating that the highest residual N amount was obtained under 165 kg ha–1of N application.In this experiment, 50%of the residual N was lost in the following season.In general, soil residual N can be divided into fixed ammonium, microbial fixation and inorganic N, among which the inorganic N in the soil is more easily taken up and used by plants (Kadiyalaet al.2015; Liuet al.2019).However, inorganic N in paddy fields is easily lost because of the obvious alternating wet and dry processes,which might be the reason for the low uptake rate of residual N and large losses in paddy fields.However, the loss mechanisms and proportions of residual N remain unclear and need further in-depth study.

The existing studies on N in rice fields seldom consider residual N fertilizer when calculating NUE.Although approximately 50% of residual N was lost, there was still 44% of the residual N in the soil in Heilongjiang Province on average.Hence, by taking the residual N into account,the NUE should be about 1–2% higher.Considering cost control and environmental protection, it is possible to reduce the N application by taking full advantage of the residual soil N uptake from the previous rice rotation in order to increase NUE, grain yield and the protection of the environment (Macdonaldet al.2002).

4.3.Effects of different N application rates on N loss

A linear relationship between NH3volatilization and the N rate was observed in this research (Fig.3).A higher N rate resulted in greater N lossviaNH3volatilization, while the loss due to NH3volatilization in this experiment was only 0.81–2.99% of the overall N input, which is similar to the results of other studies on rice in northern China (Wanget al.2018; Zhang Det al.2018).NH3volatilization loss is influenced by multiple factors, such as temperature, soil pH and wind.The higher the temperature and pH, the higher the NH3volatilization (Zhanget al.2011; Yang Yet al.2013, 2019).The NH3volatilization losses of the basal fertilizer and the first top-dressing fertilizer were significantly higher than that of the second top-dressing fertilizer due to the weak uptake of N from the fertilizer by rice roots and the sparse rice canopy that intercepted less solar radiation (Yaoet al.2018).However, Tianet al.(2023) showed that the capture of NH3volatilization by the rice canopy was much higher after fertilization at heading than at the tillering stage.In this study, less rainfall and higher temperatures occurred in 2018 compared with 2017, and the fertilizer at heading was applied at an average temperature of 24.2°C in 2018 compared with 19.0°C in 2017 (Fig.1).Hence, NH3volatilization after N application at heading in 2018 was greater than in 2017.

N loss through leaching was low in the paddy field in Heilongjiang Province.The net N leaching loss rates were 0.80–2.80 and 0.19–3.56 kg N ha–1in 2017 and 2018, respectively (Fig.4).The two-year N leaching losses accounted for 0.86–1.70% and 0.25–2.38% of the applied N rates, respectively, and a linear correlation was shown between increased N loss by leaching and an increase in the N rate.However, the interannual variation in the N leaching loss among the treatments was very limited.The loss of N due to leaching is known to be affected by the N rate and rainfall; hence, relatively little N was lost by leaching in Heilongjiang due to the low N application rate and low rainfall.However, N loss through surface runoff varied greatly, which was mainly affected by the timing and scale of the surface runoff events,and the N application rate was also taken into account(Tianet al.2007; Caoet al.2017).The average surface runoffwas approximately 3.10 kg N ha–1in 2017, and no significant difference was observed among the different N treatments (Table 4).This lack of a difference could be explained by the surface runoffevents occurring nearly 30 days after the last fertilization event, so the surface runoffN loss was only minimally affected by N application.The surface runoffin 2018 was positively correlated with the amount of N applied (R2=0.9494) since the surface runoffevent occurred shortly after fertilization; however, the N loss through surface runoffwas lower in 2018, which was attributed to the reduced amount of rainfall.

The N lossesviaNH3volatilization, leaching and surface runoff were less than 3% of the overall N application amount (Fig.6), suggesting that these three pathways are not the main factor of N loss in paddy fields in northern China.This study showed that the apparent denitrification loss was high, approximately 50.98% (Fig.5), which may be the main cause of N loss in Heilongjiang paddy fields.Previous studies confirmed that the loss of N2O in the northern rice region was approximately 2%, and it may be lost in the form of N2, which incurs less environmental stress (Tanget al.2018).Consequently, since northern rice fields produce less hazardous gases than southern rice fields, such as NH3and N2O, the pressure on the environment is lower than that caused by the other rice fields.Considering the balance between reducing N loss and achieving high yield, a fertilizer rate of 105–135 kg N ha–1is more suitable for the northern paddy fields in China.

4.4.Fate of N fertilizer in paddy fields in cold regions and other regions in China

Fig.6 Schematic of the fate of N fertilizer in Northeast China.

The PFPNand PENwere approximately 36.1–42.8 and 38.8–53.4 kg kg–1, respectively, in the southern rice region(Caoet al.2023; Tianet al.2023), while the corresponding values in the northeastern regions were 55.55–103.39 and 44.55–73.15 kg kg–1(Table 2).Previous studies showed that at equivalent yield levels, the N rates recommended for rice production in southern and northern China were approximately 226 and 100–110 kg N ha–1(Penget al.2015), and the N requirements per 100 kg of rice were 2.4 and 1.4 kg, respectively (Penget al.2019).Therefore,the PFPNin northeastern paddy fields is much higher than in other rice regions.At the same time, relatively higher N application could increase rice N uptake and accumulation, and the amount of N accumulation was approximately 200–225 kg ha–1in southern rice regions(Caoet al.2023; Tianet al.2023) but less than 145 kg ha–1in Northeast China (Table 2).Hence, a higher PENwas obtained in northeastern paddy fields, which is one of the main reasons why NUE in northeastern paddy fields is higher than in other rice regions in China (Caoet al.2023;Zhaoet al.2023).However, the underlying mechanisms,such as the possible involvement of microorganisms and N conversion in soil, still need further study.

NH3volatilization is the primary N loss pathway in paddy fields, especially in southern China (Zhaoet al.2012).The average NH3volatilization in southern China was 35.2 kg N ha–1, accounting for 17.2% of the applied N fertilizer.In late rice in southern China, the N loss rate reached 35.2% (Wanget al.2015).Therefore, the amount of NH3volatilization in paddy fields in Heilongjiang Province was far less than that in southern China, which could be explained by the N rate, soil properties and climate.Additionally, paddy soil in Heilongjiang is weakly acidic, and the average temperatures within 10 days after the three fertilization events at the experimental site in this study were 17.46, 22.32 and 25.43°C, respectively, which are significantly lower than the temperatures in southern China.Hence, due to the lower N application dosage, soil pH and temperature, NH3volatilization is much lower in Northeast China than in South China.

The N leaching loss in southern China was reported to be 5.4 kg N ha–1, with a loss rate of 3.3% on average.For single rice in the Yangtze River Basin, the N leaching loss can reach 6.4 kg N ha–1, accounting for 3.8% of the overall N application (Wanget al.2015).In addition,compared with the single rice cultivation region in the Yangtze River Basin (where surface N runoff is up to(5.3±5.9)% with a loss rate of (10.4±11.6) kg N ha–1)(Wanget al.2015), the surface runoffof N in Heilongjiang was negligible, with a value of approximately 0.1–1.05%of the total N rate on average.The rainfall in Heilongjiang is lower than in southern China (Liet al.2022), which might be the reason for the lower N leaching and surface runoff in Heilongjiang.The higher N application rate,227 kg N ha–1on average (Wanget al.2015), is also one of the main reasons for these results.In addition, a stronger priming effect of N fertilizer was present in the paddy soil in Northeast China than in other regions, as shown by the low amounts of mineralized N in other soils without N fertilizer application and the rapid increase in mineralization after fertilization (Yanget al.2021).The low temperatures during the fallow season in Heilongjiang Province limit the activity of microorganisms related to N mineralization, resulting in a significant reduction in the organic N released into the soil (Zhanget al.2022).Together with the absence of downward water movement which greatly reduces the loss of organic N, more residual N fertilizer from the previous season remains in the soil,which could be utilized by crops in the coming season(Zhang Zet al.2018).This factor also reduces N fertilizer losses in Heilongjiang paddy soil to a greater extent.

5.Conclusion

In northeastern paddy fields, the residual N rate accounts for approximately 20%, the absorption is less than 30%,and approximately 45% of the N fertilizer is lost through various pathways.As the rate of N increases, the residual N increases, and NH3volatilization and leaching losses increase proportionally.However, NH3volatilization,leaching and surface runoff losses can be ignored.N2loss, which is less harmful to the environment, may be the main pathway of N loss.The RENof N fertilizer was high in paddy fields and did not decrease significantly with an increasing N rate, while the AENand PENvalues decreased significantly when excessive N was applied.The rice grain yield and N fertilizer rate were fitted with a linear platform model.If the balance between rice yield and environmental benefits is taken into account, the application rate of N in the northeastern paddy fields of China should be 105–135 kg ha–1.

Acknowledgements

This research was supported by the Strategic Priority Research Program of the Chinese Academy of Sciences(XDA28100302), the earmarked fund for China Agriculture Research System (CARS-01-29), the National Key Research and Development Program of China(2017YFD0200104), the Fifth (2019) of ‘Young Talents’Project of Northeast Agricultural University, China, and the Open Program of Key Laboratory of Germplasm Enhancement, Physiology and Ecology of Food Crops in Cold Region, Ministry of Education, Northeast Agricultural University (CXSTOP2021009).

Declaration of competing interest

The authors declare that they have no conflict of interest.