Nitrous oxide emission from highland winter wheat field after long-term fertilization

Nitrous oxide (N2O) is an important greenhouse gas. N2O emissions from soils vary with fertilization and cropping practices. The response of N 2O emission to fertilization of agricultural soils plays an important role in global N2O emission. The objective of this study was to assess the seasonal pattern of N 2O fluxes and the annual N 2O emissions from a rain-fed winter wheat ( Triticum aestivumL.) field in the Loess Plateau of China. A static flux chamber method was used to measure soil N 2O fluxes from 2006 to 2008. The study included 5 treatments with 3 replications in a randomized complete block design. Prior to initiating N2O measurements the treatments had received the same fertilization for 22 years. The fertilizer treatments were unfertilized control (CK), manure (M), nitrogen (N), nitrogen + phosphorus (NP), and nitrogen + phosphorus + manure (NPM). Soil N2O fluxes in the highland winter wheat field were highly variable temporally and thus were fertilization dependent. The highest fluxes occurred in the warmer and wetter seasons. Relative to CK, m slightly increased N 2O flux while N, NP and NPM treatments significantly increased N2O fluxes. The fertilizer induced increase in N 2O flux occurred mainly in the first 30 days after fertilization. The increases were smaller in the relatively warm and dry year than in the cold and wet year. Combining phosphorous and/or manure with mineral N fertilizer partly offset the nitrogen fertilizer induced increase in N 2O flux. N2O fluxes at the seedling stage were mainly controlled by nitrogen fertilization, while fluxes at other plant growth stages were influenced by plant Correspondence to: M. D. Hao (mdhao@ms.iswc.ac.cn) and environmental conditions. The cumulative N 2O emissions were always higher in the fertilized treatments than in the non-fertilized treatment (CK). Mineral and manure nitrogen fertilizer enhanced N 2O emissions in wetter years compared to dryer years. Phosphorous fertilizer offset 0.50 and 1.26 kg N2O-N ha−1 increases, while manure + phosphorous offset 0.43 and 1.04 kg N 2O-N ha−1 increases by N fertilizer for the two observation years. Our results suggested that the contribution of single N fertilizer on N 2O emission was larger than that of NP and NPM and that manure and phosphorous had important roles in offsetting mineral N fertilizer induced N2O emissions. Relative to agricultural production and N2O emission, manure fertilization (M) should be recommended while single N fertilization (N) should be avoided for the highland winter wheat due to the higher biomass and grain yield and lower N2O flux and annual emission in m than in N.

Correspondence to: M. D. Hao (mdhao@ms.iswc.ac.cn) and environmental conditions.The cumulative N 2 O emissions were always higher in the fertilized treatments than in the non-fertilized treatment (CK).Mineral and manure nitrogen fertilizer enhanced N 2 O emissions in wetter years compared to dryer years.Phosphorous fertilizer offset 0.50 and 1.26 kg N 2 O-N ha −1 increases, while manure + phosphorous offset 0.43 and 1.04 kg N 2 O-N ha −1 increases by N fertilizer for the two observation years.Our results suggested that the contribution of single N fertilizer on N 2 O emission was larger than that of NP and NPM and that manure and phosphorous had important roles in offsetting mineral N fertilizer induced N 2 O emissions.Relative to agricultural production and N 2 O emission, manure fertilization (M) should be recommended while single N fertilization (N) should be avoided for the highland winter wheat due to the higher biomass and grain yield and lower N 2 O flux and annual emission in m than in N.

Introduction
Nitrous oxide (N 2 O), an important greenhouse gas in the atmosphere, has increased from pre-industrial concentrations of 270 ppb to 319 ppb in 2005 (Forster et al., 2007).Soil is acknowledged as the major source of N 2 O, accounting for about 70% of total emissions.Agricultural soils account for a large proportion (70-81%) of the increase in N 2 O emissions to the atmosphere, with the increase linked to increased N fertilizer use (Bouwman, 1990).A recent calculation showed that 3.3 Tg N 2 O-N yr −1 is emitted globally from fertilized cropland (Stehfest and Bouwman, 2006).However, the linkages between agricultural soil emissions and global X. R. Wei et al.: Nitrous oxide emission from highland winter wheat field emissions are still uncertain (Stehfest and Bouwman, 2006), and further understanding of N 2 O emissions from cropped land are still necessary for accurate global N 2 O emission prediction.
The response of soil N 2 O emissions to N fertilizatioN has been widely studied in different ecosystems.Dittert et al. (2005) and Lampe et al. (2006) showed larger N 2 O emissions from mineral N than from slurry treated grassland plots and demonstrated that the soil N pool was always clearly the major source of N 2 O. Zhang and Han (2008) reported a linear relationship between cumulative N 2 O and N application rate in the semi-arid grassland of northern China, while Mc-Swiney & Robertson (2005) found a nonlinear response of N 2 O flux to incremental fertilizer additions in a continuous maize (Zea mays L.) cropping system in southwest Michigan.Although the response pattern of soil N 2 O emission to N fertilizers was not identical, the increased N 2 O emission associated with mineral N fertilizer application is widely acknowledged (Bouwman et al., 2002).As for manure fertilizer, Davidson (2009) showed that manure has been important for N 2 O emission since 1860, whereas mineral fertilizer became important only during the last half of the twentieth century, suggesting that the contribution of manure fertilizer to N 2 O emission cannot be ignored.However, estimations by Flynn et al. (2005) indicated that manure makes a significantly smaller contribution than mineral N fertilizers to N 2 O emissions from cropped soils.Many other studies also showed that N 2 O emission due to manure were less than that due to mineral N fertilizers (Dambreville et al., 2008;Alluvione et al., 2010).Nevertheless, little research has been performed to study how N 2 O emissions respond to a combined application of manure and mineral N fertilizers.Combinations of manure and mineral N fertilizers are used extensively around the world.
Manure fertilizer has been extensively applied to farmland in China for more than 2000 years.In the past 50 years, with the use of chemical fertilizers, manure and chemical fertilizers were often used together to increase crop yields (Shen, 1998).For example, in the central Loess Plateau, the average manure application rate before 1960 was 22.9 ton/ha.Chemical fertilizers were not used before 1960.Yet after 1960 the manure application rate was 24.9 toN ha −1 and the chemical fertilizer rate was 178 kg ha −1 (Hao et al., 1991).Therefore, even after chemical fertilizer applications became popular, manure still accounted for an important part of the total N applied to farmland soils in the area.Thus, the effects of manure on N 2 O emissions should be investigated.
Wheat is an important contributor to the global food supply.The global wheat area in 2007 was 210 million ha, while the area in China was 23 million ha.N fertilization is fundamental for wheat production.In 1996, a total of 4.97 million tons of N fertilizer were applied for global cereal production, and wheat accounted for approximately 29% of the total (Raun and Johnson, 1999).However, the nitrogen use efficiency of wheat is relatively low and N losses attract much attention from scientists (Xing and Zhu, 2000;Silgram et al., 2001;Bouwman et al., 2005;Sudling et al., 2005;Vitousek et al., 2009).As global wheat production expands, N fertilizer applications for wheat production increase, and the N fertilizer induced N 2 O emissions in wheat fields account for an increasing proportion of global N 2 O emissions.Further studies of N fertilizer effects on wheat growth and N 2 O emissions are needed in support of improving estimations of global N 2 O emissions.
The response of N 2 O emission to fertilization varied greatly with fertilization years (Hall and Matson, 1999), which could be ascribed to the unstable fertilization effects during the first several years of the experiment and the relatively stable effects during the middle or latter periods of the experiment.To avoid inconsistent treatment start-up responses of N 2 O to fertilization, N 2 O emissions under longterm fertilized conditions should be investigated in order to objectively determine the stable fertilization effects on N 2 O emissions.
In this study, a two-year N 2 O flux observation was conducted in a highland winter wheat field after 22 years fertilization in the Loess Plateau of China.The objective was to assess seasonal patterns of N 2 O flux and annual emissions as affected by long-term fertilization practices in rain-fed winter wheat field plots.

Study sites and experimental design
The long-term field experiment was initiated in September 1984 at the Agro-ecological Experiment Station of the Chinese Academy of Science, Changwu County, Shaanxi Province, China (35 • 12 N, 107 • 40 E).The average annual temperature of this site is 9.1 • C and annual precipitation is 585 mm.The soil is a Hei Lu soil according to the Chinese classification system, which corresponds to a Calcarid Regosol according to the FAO/UNESCO classification system (FAO/Unesco, 1988).
The cropping system was continuously cropped winter wheat (Triticum aestivum L.).The fertilizer treatments were unfertilized control (CK), manure (M), nitrogen (N), nitrogen + phosphorus (NP), and nitrogen + phosphorus + manure (NPM).Urea and superphosphate were used as the source of N and P. The manure came from cattle.In all of the fertilizer treatments, the N rate was 120 kg ha −1 , the P rate was 26.2 kg ha −1 , and the m rate was 75 ton ha −1 .The mean total N content of the manure was 1.97 g kg −1 and the available N was 91 mg kg −1 .
Fertilization treatments were replicated three times in a randomized complete block statistical design.Each plot was 10.3 m by 6.5 m.Routine crop management practices for this region were used.Prior to seeding, fertilizers were broadcast on the soil surface and then the land was plowed two times with a cattle-drawn plow to a depth of about 10-cm.Wheat was sown in rows 20 cm apart.After seeding, the soil was raked to cover the seed.Weeds were removed by hand in all of the treatments.When the wheat reached maturity, it was harvested at the ground-level, the straw and grain were removed, and then the soil was plowed two times to a depth of about 15 cm.

Measurement of N 2 O flux
The N 2 O fluxes were measured from September 2006 to September 2008.The soil fertility conditions in each treatment before N 2 O flux measurement are shown in Table 1.
A static closed-chamber technique was used to investigate N 2 O flux.The chambers, with base area of 20 cm×25 cm and height of 20 cm, made of polyvinyl chloride, were inserted 10 cm into the ground in each plot.The chambers remained open (uncovered) between each measurement period.Chambers were closed (covered) to begin a measurement period.Four samples, at intervals of 15 min each, were collected from enclosed interspaces with a 25 mL glass gas-tight syringe.Tang et al. (2006) showed that greenhouse gases fluxes, including N 2 O, measured at 9:00 a.m. were close to daily means, so we collected gas samples near 9:00 a.m.every 10 days in the winter and every 4-6 days in the other seasons.
A gas chromatograph (Shimadzu GC-14B), fitted with a 4 mm×3 m stainless steel column packed with Porapack Q (80-100 mesh) and an 63 Ni electron capture detector (ECD), was used to measure N 2 O concentrations.Highpurity Ar 2 /CH 4 (95% Ar 2 + 5% CH 4 ) was used as the carrier gas, and the flow rate was maintained at 40 mL min −1 .The column and the detector temperatures were set at 65 and 300 • C, respectively.The standard N 2 O gas was supplied by the Japanese National Institute of Agro-Environmental Sciences.

Climate and soil environment measurements
The climate data were determined at a weather station at the Agro-ecological Experiment Station, part of the Chinese Ecosystem Research Network (CERN).Hourly values of rainfall and air temperature at a 1.2 m height were mea-sured automatically at the weather station.Daily values, from 10:00 a.m. to 10:00 a.m. the following day, of rainfall and air temperature were determined from hourly values.
Soil temperature and moisture (volumetric water content) at a depth of 5 cm were monitored by a Delta-T Profile Probe (ML2X) and HH2 reader (Delta-T Devices Ltd, UK) in each plot at the time when gas samples were collected.Waterfilled pore space (WFPS) was calculated using the measured soil bulk density data and an assumed particle density value of 2.65 g cm −3 .

Statistical analysis
The frequency distributions of observed N 2 O fluxes were tested.The N 2 O flux data in this study showed log normal or highly skewed distributions.Therefore, the original data were log-transformed to meet the normality criteria before performing ANOVA analysis.The repeated measures ANOVA was used to test the effects of fertilization and growth stage on soil N 2 O flux.A correlation analysis was conducted to assess the relationships between N 2 O flux and soil temperature and WFPS for each fertilization treatment.All statistical analyses were performed using SAS software (SAS Institute, 1999).

Overview of environmental conditions
In this study, an observation year included a winter wheat season (from September to June in the following year) and a fallow season (June to September).The mean daily air temperatures at the study site for the 2006-2007 and the 2007-2008 observation years were 10.8 • C and 9.6 • C, respectively.In these 2 observation years, the winter wheat seasons were characterized by relatively low air temperatures, which were 12.7 • C and 13.9 • C lower than those in the respective fallow seasons.Generally, the lowest daily air temperatures were observed in December and January (within the wheat season) while the highest daily air temperatures were observed in July (within the fallow season) (Fig. 1a).The rainfalls for the 2 observation years were 358 www.biogeosciences.net/7/3301/2010/Biogeosciences, 7, 3301-3310, 2010 and 509 mm, of which 55 and 53% fell during the periods of winter wheat growth (Fig. 1a).The average daily rainfalls were 0.72 and 0.99 mm for the wheat seasons, and 1.74 and 2.6 mm during the fallow seasons, respectively.Therefore, the study site was characterized by dry and cold wheat seasons and wet and warm fallow seasons.Soil temperature (5 cm depth) varied seasonally in response to air temperature (Fig. 1b).The largest soil temperature (observed in CK) occurred in August, while the smallest soil temperature occurred (observed in NPM) in February.Soil temperatures in CK were 0-2.4 • C and 0-2.0 • C larger than in the fertilized treatment for 2006-2007 and 2007-2008 observation years depending on the sampling date.Nevertheless, the repeated measures ANOVA showed that fertilization did not significantly influence soil temperature.
Relatively large WFPS of surface soil was observed during the study period, and it varied temporally in response to rainfall and wheat growth (Fig. 1c).For the 2 observation years, WFPS was highest during the wet season and decreased during wheat growth.Soil WFPS was significantly influenced by fertilization.Generally, CK, N and NP had WFPS ranging from 37.3 to 77.1%, 34.5 to 81.5% and 35.1 to 82.8%, while m and NPM had WFPS ranging from 33.4 to 79.1% and 33.4 to 75.8%, respectively.The WFPS in CK, N and NP were 3 to 20% higher than in m and NPM.

Fertilization effects on N 2 O fluxes
Soil N 2 O fluxes were temporally variable and fertilization dependent (Fig. 2).Significant increases in N 2 O flux due to fertilization occurred mainly in the first 30 days after fertilization (Table 2).For example, the average N 2 O flux in the first 30 days after fertilization were 1057%, 681% and 614% larger in the N, NP and NPM treatments than in the CK treatment during the 2 observation years (2006)(2007)(2008), while the average flux in the following periods (31-315 days) were 42%, 26% and 44% larger than in CK, respectively.Additionally, the effects of fertilization on N 2 O fluxes in the first 30 days were Biogeosciences, 7, 3301-3310, 2010 www.biogeosciences.net/7/3301/2010/2007 2007-2008 2006-2008 2006-2007 2007-2008 2006-2008 2006-2007 2007-2008 2006-2008 Precipitation Relative to CK, the N treatment had the largest increases in N 2 O flux with an increase range of 0-605 µg N 2 O-N m −2 h −1 during the 2 experimental years, while the m treatment had the lowest increase in flux, ranging from 0 to 38 µg N 2 O-N m −2 h −1 .The increases by the NP and NPM treatments ranged from 0 to 426 µg N 2 O-N m −2 h −1 and 0 to 438 µg N 2 O-N m −2 h −1 , respectively, indicating that phosphorous or manure could partly offset the single mineral N fertilizer induced increases in N 2 O flux.In this study, the phosphorous and manure + phosphorous offset 36% and 35% of the increased fluxes associated with by mineral N fertilizer alone.

Seasonal patterns of N 2 O fluxes as influenced by fertilization
The seasonal patterns of soil N 2 O fluxes were mainly influenced by fertilization, wheat growth, and environmental conditions (Fig. 2 and   2006-2007 and 2007-2008.However, the fluxes in the fertilized treatments were not significantly different in the later growing stages and the fallow seasons.These results indicated that N 2 O fluxes at the seedling stage were mainly controlled by nitrogen fertilization, while fluxes at the other growing stages were mainly influenced by plant and environmental conditions.

Overall emissions
The overall N 2 O emissions were generally always larger in the fertilized treatments than in the CK treatment for the growing season and the fallow season (Fig. 3).Significant differences in emissions among the five treatments occurred in the seedling, tillering, jointing stages and the fallow season during 2006-2007, and  curred in M, and the largest increases occurred in the N treatment.Although the NP and NPM treatments also resulted in increases in N 2 O emissions, the increases were less than the single mineral N fertilizer (N treatment) increase, suggesting that phosphorous and manure fertilizers have the potential to inhibit the mineral N fertilizer increased N 2 O emission in the studied area.
For the observation year 2007-2008, the emissions for all treatments were larger from seedling to booting and lower in heading to maturity and in the fallow season compared with 2006-2007.Totally, 191 mm and 130 mm of rain occurred from seedling to booting in 2007-2008 and 2006-2007, in-dicating that rainfall during this period may have had more influence on the N 2 O emissions more than did the fertilizers.The effects of fertilization on emissions varied with rainfall.The annual emissions in CK and m were 10.7% and 5.1% less in the wetter 2007-2008 than in the dryer 2006-2007, while for N, NP and NPM, emissions were 41. 3%, 22.1% and 26.6% larger in 2007-2008 than in 2006-2007, demonstrating that nitrogen fertilizer enhanced N 2 O emissions more in the wetter year than in the dryer year, in the seedling to booting stages.Because the major aim of fertilization is to increase crop yields, expressing N 2 O emissions on a grain yield basis provides a useful option for evaluating fertilizer impacts.The biomass and grain yields in the NP, m and NPM treatments were significantly larger than those in the N and CK treatments for both 2006-2007and 2007-2008 (Fig. 4 (Fig. 4).The crop grain yield-based and grain N-yield-based N 2 O emissions were significantly larger in the N treatment than in the CK treatment, while those in the NP, M, and NPM treatments were lower than that in the CK treatment (except for the grain yield-based emissions in the NPM treatment in 2006-2007, which was similar to CK) (Fig. 5).Therefore, the combination of N fertilizer with P or manure improved the production and the environmental effects compared to single N fertilization alone.

Fertilization effects on N 2 O fluxes
In our experiment, N 2 O flux peaks just after nitrogen fertilizer applications.The annual average fluxes due to fertilization were relatively large compared to reported fluxes from other ecosystems.Generally, the N 2 O fluxes in forest soils and grassland are typically low, and increases associated with N fertilizer applications are also low compared with farmland (Davidson andKingerlee, 1997, Hall andMatson, 1999;Venterea et al., 2003;Du et al., 2006;Zhang and Han, 2008).The limited effects of N fertilizer for forest and grassland soil might be due to the relatively low N application levels, while the relatively large input of N fertilizer to crop soil often results in a large N 2 O flux (Wagner-Riddle et al., 2007;Barton et al., 2008;Scheer et al., 2008).Irrigation in cropped soil also N 2 O flux (Scheer et al., 2008).The variation of soil N 2 O flux with ecosystem type and fertilization practices can be explained by changes in soil C/N ratio which is negatively related with N 2 O flux (Klemedtsson et al., 2005).Generally, top soils in forests and grasslands have larger C/N ratios than do farmland soils, and the soils with large C/N ratios have lower N 2 O fluxes.The application of mineral N fertilizer to soil reduces the C/N ratio, and thus increases N 2 O flux.Soil C/N ratios of the farmland in our site are larger than those observed by Scheer et al. (2008) and smaller than those by Wagner-Riddle et al. (2007) and Barton et al. (2008), and our fluxes in the CK and mineral N fertilized treatments are smaller than those reported by Scheer et al. (2008) and larger than those reported by Wagner-Riddle et al. (2007) and Barton et al. (2008).The application of manure fertilizer often increases the C/N ratio and consequently decreases the N 2 O flux.The top soil C/N ratios followed the order of M>NPM>N while the fluxes followed the order of M<NPM<N.
The application of manure fertilizer often improves soil structure (Bronick and Lal, 2005), increases soil porosity, and decreases WFPS, which reduces the denitrification rate and thus decreases N 2 O emissions.Liu et al. (2000) and Huo et al. (2008) reported that manure fertilizer has a large potential to increase soil porosity and aggregation in the Loess Plateau, which improves soil aeration.In our study, the WFPS in CK, N and NP were 3 to 20% higher than in m and NPM, which partly explains the manure effects on N 2 O emissions.
Although several studies demonstrated that P promotes N 2 O flux (Minami and Fukushi, 1983;Falkiner et al., 1993;Zhang and Han, 2008), we observed 24% and 27% lower fluxes in the NP treatment than in the N treatment in 2006-2007 and 2007-2008, indicating that P could alleviate single N fertilizer increases of N 2 O flux.We assume that this effect is due to the fact that the NP treatment accelerated the uptake of soil mineral N which decreased soil nitrate and ammonium levels compared with the N treatment.The lower the soil N, www.biogeosciences.net/7/3301/2010/Biogeosciences, 7, 3301-3310, 2010

Seasonal patterns
The fertilizations significantly affected the seasonal patterns of soil N 2 O fluxes, which were characterized by a peak flux within 30 days following fertilization, consistent with other findings (Hall and Matson, 1999).For the whole wheat season and fallow season, the flux was significantly related with soil temperature for each treatment.Except for the seedling period, the seasonal patterns of N 2 O flux in all treatments were closely related with WFPS (Table 4), indicating that N 2 O fluxes in the winter wheat fields of the study area were somewhat temperature and water dependent.The dependence of N 2 O production on soil temperature has been reported in many ecosystems (Dobbie and Smith, 2003a;Flynn et al., 2005;Wagner-Riddle et al., 2007;Zhang and Han, 2008).A threshold of 5 • C for N 2 O productioN has been reported (Dobbie and Smith, 2003b).According to this threshold, there should be little N 2 O flux in the littering stage.However, winter and spring thaw could magnify N 2 O fluxes because freeze/thaw cycles have been related to enhanced microbial activity due to increased available carbon from freezing lysis (Christensen and Tiedje, 1990), and disintegrating aggregates (van Bochove et al., 2000), combined with sufficient soil water content during thawing (Lemke et al., 1998).Combined, these conditions can result in a relatively large flux in the littering stage.Wagner-Riddle et al. ( 2007) also observed increased N 2 O in cold seasons, agreeing with our measurements.Additionally, large available N in soils during the stage following the seedling stage (when the fertilizer was applied) might be another reason for the higher fluxes in the nitrogen fertilized treatment.
In our study, 93% of the observed WFPS values were lower than 70%, which is the threshold for denitrification dominance (Davidson, 1992), suggesting that nitrification is likely the main source of N 2 O in our study.The seasonal patterns of winter wheat growth and WFPS differed significantly with fertilization, indicating that the effects of fertilization on temporal N 2 O fluxes in the growing season were mainly associated with wheat root activities and WFPS changes, which alter soil microorganism activities in N 2 O production together with soil C and N conditions.The effects in the fallow season were dominated by the WFPS and soil C and N conditions.

Implications for N 2 O emission and fertilization
At the global scale, annual N 2 O emissions from cropped mineral soils ranged from 0.3 to 10.7 kg N 2 O-N ha −1 yr −1 (Stehfest and Bouwman, 2006).For wheat soils, the emission ranged from 0.5 to 3.7 kg N 2 O-N ha −1 yr −1 regardless of irrigation or not, while more than 75% of these observations were less than 1.6 kg N 2 O-N ha −1 yr −1 (Roelandt et al., 2005).The background emissions (CK treatment) in our rain-fed winter wheat field were 1.7 and 1.9 kg N 2 O-N ha −1 yr −1 , while our fertilized emissions were 2.1-3.4 and 3.2-7.5 kg N 2 O-N ha −1 yr −1 for the 2006-2007 and 2007-2008 observation years.These were relatively large emission values.
The annual N 2 O emissions from fertilized treatments were in the following order N>NPM and NP>>M, suggesting that the contribution of single N fertilizer alone was larger than that of NP and NPM.This finding agrees with other reports (Flynn et al., 2005;Dambreville et al., 2008;Alluvione et al., 2010).This result implies that the present assessment might be larger because manure and phosphorous fertilizers are widely used in agriculture together with mineral N fertilizer while its role in offsetting mineral N fertilizers induced N 2 O emissions are usually neglected.Barton et al. (2008) also suggested that the observed emission factor in semiarid cropped soil is 60 times less than the values suggested by IPCC.Many estimations separate fertilizer N as mineral N and organic N (from manure fertilizer) (Flynn et al., 2005, Davidson, 2009) without considering the fact that organic N and mineral N and mineral N and P are often used together.These estimations often ignore the offsetting role of manure when it is applied together with mineral N fertilizers.
Our results further show that the m treatment had relatively large biomass and grain yield and relatively low N 2 O fluxes and annual emissions.From the point of agricultural production and N 2 O emission, M is recommended while single N fertilization alone is not recommended for highland winter wheat, when fertilizers are applied at the time of planting.

Abstract.
Nitrous oxide (N 2 O) is an important greenhouse gas.N 2 O emissions from soils vary with fertilization and cropping practices.The response of N 2 O emission to fertilization of agricultural soils plays an important role in global N 2 O emission.The objective of this study was to assess the seasonal pattern of N 2 O fluxes and the annual N 2 O emissions from a rain-fed winter wheat (Triticum aestivum L.) field in the Loess Plateau of China.A static flux chamber method was used to measure soil N 2 O fluxes from 2006 to 2008.The study included 5 treatments with 3 replications in a randomized complete block design.Prior to initiating N 2 O measurements the treatments had received the same fertilization for 22 years.The fertilizer treatments were unfertilized control (CK), manure (M), nitrogen (N), nitrogen + phosphorus (NP), and nitrogen + phosphorus + manure (NPM).Soil N 2 O fluxes in the highland winter wheat field were highly variable temporally and thus were fertilization dependent.The highest fluxes occurred in the warmer and wetter seasons.Relative to CK, m slightly increased N 2 O flux while N, NP and NPM treatments significantly increased N 2 O fluxes.The fertilizer induced increase in N 2 O flux occurred mainly in the first 30 days after fertilization.The increases were smaller in the relatively warm and dry year than in the cold and wet year.Combining phosphorous and/or manure with mineral N fertilizer partly offset the nitrogen fertilizer induced increase in N 2 O flux.N 2 O fluxes at the seedling stage were mainly controlled by nitrogen fertilization, while fluxes at other plant growth stages were influenced by plant

Fig. 1 .
Figure 1 514 Precipitation and air temperature (a), 5 cm depth soil temperature (b) and WFPS (c) for different 515 fertilization treatments from September 2006 until September 2008 516

Fig. 2 .
Figure 2 517 N2O fluxes for different fertilization treatments from September 2006 until September 2008 518 For the CK treatment, N 2 O fluxes ranged from 3 to 63 µg N 2 O-N m −2 h −1 , with average fluxes of 21 and 18 µg N 2 O-N m −2 h −1 for 2006-2007 and 2007-2008 observation years.The largest N 2 O fluxes occurred in warm and wet seasons.Application of manure slightly increased the N 2 O flux.The average fluxes in the m treatment were 25 and 35% higher than those in CK for the 2 observation years, respectively.The application of N, NP and NPM significantly increased N 2 O fluxes.For the observation period from 2006 to 2008, the fluxes in N, NP and NPM ranged from 4.9 to 624, 6.1 to 445, and 7.6 to 454 µg N 2 O-N m −2 h −1 , with average fluxes of 64, 48 and 48 µg N 2 O-N m −2 h −1 , which were 3.3, 2.5 and 2.5 times the average flux in CK.Generally, the largest increase in N 2 O flux due to fertilization was observed in the 2007-2008 observation year.The N 2 O fluxes for the N and NP treatments in 2007-2008 were 1.1 times greater than those in 2006-2007.

Fig. 3 .
Figure 3 519 Changes in N2O emissions with winter wheat growth for different fertilization treatments from 520 September 2006 until September 2008 521

Fig. 4 .
Fig. 4. Above-ground biomass and grain yields of winter wheat for different fertilization treatments from September 2006 until September 2008.

29Fig. 5 .
Fig. 5. Grain yield-based and grain N-yield-based N 2 O-N emissions for different fertilization treatments from September 2006 until September 2008.

Table 2 .
Mean N 2 O fluxes during the first 30 days after fertilization and during the most of the year (the later period).

Table 3 .
N 2 O fluxes in growing seasons and fallow season as affected by fertilization.

Table 4 .
Co-relationships between N 2 O fluxes and soil temperature and water-filled pore space (WFPS).