Irrigation and tillage effects on soil nitrous oxide emissions in maize

ABSTRACT

Agricultural management practices such as irrigation, stover management, tillage and nitrogen fertilization have an important role on soil greenhouse gas emissions (GHG) and particularly on N2O emissions (Reay et al., 2012;IPCC, 2014).The production of N2O in the soil is the result of the interaction between biotic and abiotic factors.Soil microbial communities throughout the nitrification, denitrification and nitrifier denitrification processes control soil N2O production, which are influenced by the soil water-filled pore space (WFPS) (Bateman and Baggs, 2005;Bremner, 1997;Firestone and Davidson, 1989).Under aerobic conditions, when WFPS range between 35 to 60%, nitrification is the main process involved in the N2O production.However, when WFPS is above 60% up to 80%, denitrification is the principal process responsible of N2O production in the soil due to anaerobic conditions (Linn and Doran, 1984).In addition to biotic factors and WFPS, abiotic factors such as soil nitrate and soil ammonium contents, soil temperature and soil organic carbon are also key factors on the production and dynamics of N2O in the soil (Davidson et al., 2000;Bouwman et al., 2002;Butterbach-Bahl et al., 2013).Trost et al. (2013) in a review of eight studies around the world about the effect of irrigation on soil N2O emissions reported that soil N2O emission increased after irrigation because increased the WFPS.Moreover, in Mediterranean areas, several studies have concluded that irrigation is an important agricultural practice contributing to soil N2O emissions (Aguilera et al., 2003;Cayuela et al., 2017;Sanz-Cobena et al., 2017).
Similarly, Deng et al. (2018) evaluated the impact of the irrigation system on soil N2O emissions for the California cropland using the DNDC model and predicted a reduction of 38% on soil N2O emissions under sprinkler irrigation compared with flood irrigation.
Likewise, other studies reported different effects of tillage on soil N2O emissions.For instance, Ball et al. (1999) and Venterea et al. (2011) observed an increase on soil N2O emission under no-tillage.However, Ussiri et al. (2009) and Omonode et al. (2011) reported lower emissions under no-tillage than under conventional tillage.In rainfed barley (Hordeum vulgare L.) monoculture in NE Spain, Plaza-Bonilla et al. (2018) recently observed a reduction on the N2O emitted per unit of yield under no-tillage systems when compared to conventional tillage.This last work is in agreement with the results obtained by the same authors in a previous work (Plaza-Bonilla et al., 2014), in which they observed lower or similar emissions in no-tillage systems compared to conventional tillage systems, when no-tillage was performed for more than 10 years.
However, in the first years after the implementation of no-tillage, they obtained higher emissions under no-tillage than conventional tillage, similar to the findings of Van Kessel et al. (2013).
Crop stover removal may influence soil microclimate conditions, favouring higher soil temperature and r soil water evaporation (Sauer et al., 1998).Additionally, a decrease on soil organic carbon (SOC) and a degradation of soil physical properties are associated with removing the stover from the field (Blanco-Canqui and Lal, 2008).Then, crop stover management could affect soil N2O emissions, but their effect is no clear.Jin et al. (2017) observed higher N2O emissions in irrigated maize when maize stover was maintained in the field.However, Bent et al. (2016) observed higher soil N2O emissions when maize stover was removed in Ontario, Canada.
In the last decade, several experiments have been carried out to study agronomical aspects of sprinkler irrigation systems regarding to the crop water requirements and crop performance under Mediterranean climatic conditions (Cavero et al., 2003;Robles et al., 2017;Cavero et al., 2018).However, limited studies have been done to assess the impact of different agronomical practices on N2O emissions in Mediterranean climatic conditions under irrigation and mostly focused on nitrogen (N) fertilization management (Álvaro-Fuentes et al., 2016;Maris et al., 2018).Similarly, a limited number of studies have been conducted to compare the effect of different tillage practices on N2O emissions in Mediterranean climatic conditions and all these studies were carried out under rainfed conditions (Plaza-Bonilla et al., 2014;Plaza-Bonilla et al., 2018).
This study was aimed to assess the impact of irrigation system, specifically sprinkler and flood irrigation systems, and the soil tillage system, conventional tillage and notillage systems, on soil N2O emissions under Mediterranean climatic conditions.Since sprinkler irrigation allows to apply lower amounts of water at higher frequency than flood irrigation, we hypothesize that irrigation system would result in different soil water content during the growing season which would affect N2O emissions and maize yields.
Besides, given that tillage has also shown to affect N2O emissions, we also hypothesize that irrigation system would interact with tillage.

Site description
A field study was performed during three maize seasons (2015-2017) at the experimental farm of the Experimental Station of Aula Dei, Zaragoza, Spain (41º 42´ N, 0º 49´ W, 225 m altitude).The area is characterized by a Mediterranean semiarid climate with annual mean air temperature of 14.1 ºC, annual precipitation of 298 mm and grass reference crop evapotranspiration (ETo) of 1243 mm and silty loam soils (Table 1).

Experimental design
Previously to the establishment of the experiment, the field had been under cultivation, alternating different cereal crops, mainly irrigated winter wheat (Triticum aestivum L.) and maize under conventional tillage and flood irrigation.The previous crop was winter wheat that was grown during one year (2014).In addition, this experimental field had the possibility to install a hand-move sprinkler irrigation system.Therefore, in 2015, the field was divided in two parts, one irrigated by flood irrigation and the other one by a handmove sprinkler irrigation of 18 m × 18 m sprinkler square spacing and with a sprinkler application rate of 5 mm h −1 .
The experimental layout consisted on a split-block design with two factors and three replicates per treatment.Two irrigation systems (i.e.sprinkler, S, and flood, F) and three soil tillage systems (i.e.conventional tillage, CT; no-tillage maintaining the maize stover, NTr; and no-tillage removing the maize stover, NT) were combined obtaining six different treatments with a 6 x 18 m plot size.
In CT, tillage operations previous to maize sowing consisted in one pass of a subsoiler to 30 cm depth followed by one pass of a disk harrow both performed on December 2014, 2015 Maize daily crop evapotranspiration (ETc) was computed by multiplying the reference evapotranspiration (ETo), obtained by the FAO Penman-Monteith method (Allen et al., 1998), and the crop coefficient (Kc) determined using an equation developed in the same experimental farm based on a function of the thermal time (Kiniry, 1991;Martínez-Cob, 2008).Crop irrigation requirement (CIR) for each week was determined by subtracting the effective precipitation, 75% of the total weekly precipitation (Dastane, 1978), to the weekly ETc considering an irrigation efficiency of 85%.Irrigation water was applied by sprinkler irrigation to all the plots until V6 growth stage to favour plant emergence and to avoid differences in plant density among treatments.
Irrigation frequency depended on the irrigation system.Thereby, during the three growing seasons, sprinkler irrigation events occurred two times per week (Monday and Wednesday), whereas flood irrigation events occurred every 10-14 days.Although the sprinkler irrigation system allows applying an exact irrigation dose, this is not possible with flood irrigation.Thus, the irrigation water applied in the sprinkler system was each year within 2% of the CIR.However, the irrigation water applied in the flood system was 16% to 30% higher than in the sprinkler system (Table 2).Irrigation applied in the sprinkler system was measured with a flowmeter and in the flood system with a Cipolletti weir.All tillage treatments under the same irrigation system received the same amount of irrigation water.-------------------------mm-----------------------------

Air sampling and N2O analyses
Two polyvinyl chloride (PVC) rings (31.5 cm internal diameter) per plot were inserted 10 cm into the soil on April 2015, before to start the soil gas measured.The rings were only removed at tillage, planting and harvesting operations.Soil N2O emissions were measured with the closed chamber technique (Hutchinson and Moiser, 1981) from April 2015 to September 2017, using PVC chambers (20 cm height) covered with a reflective layer of aluminium film to diminish internal increases in temperature.On the center of the top of the chamber, a chlorobutyle septum was attached as a sampling port.
Soil N2O emissions were measured weekly from planting until mid-August (VT growth stage), every two weeks from mid-August until harvest (late September) and every three weeks during the fallow period (October-March).During tillage operations, soil air samples were taken 24 h before, and 24 and 96 h after the tillage operations.
Throughout the fertilization events, soil air samples were taken 24 h before and 24, 48, 72, 96, 144 and 192 h after fertilization.Finally, soil air sampling frequency was increased during the five days after of each irrigation event over the three growing seasons, in order to characterize the flood irrigation events.
Air samples were collected at 0, 20 and 40 min after chamber closure and 20 mL of air sample were transferred to an evacuated 12-mL Exetainer ® borosilicate glass vial (model 038W, Labco, High Wycombe, UK).Air temperature inside the chamber was measured introducing thermometers in the chamber before the enclosed of the chambers.
Concentration of N2O in the air samples was measured by gas chromatography using an automatically injection system (PAL3 autosampler, Zwingen, Switzerland).The gas chromatography systems (Agilent 7890B, Agilent, Santa Clara, CA, United States) was equipped with an electron capture detector (ECD) and a HP-Plot Q column (15 m long, 320 μm in section and 20 μm thick), using He as a carrier gas at 2 mL min -1 .The injector and the oven temperatures were set to 50 and 35ºC, respectively.The temperature of the ECD was set to 280ºC and a 5% methane in Argon gas mixture at 30 mL min -1 was used as a make-up gas.Ultra-high purity N2O standards (Carburos Metálicos, Barcelona, Spain) was used to calibrate the system.Emission rates (mg N2O-N m -2 day -1 ) were calculated by the linear increase in the N2O concentration during the chamber enclosure time and corrected by the internal air chamber temperature.

Soil, biomass and grain yield sampling and analyses
Soil ammonium (NH4 + ) and nitrate (NO3 − ) contents were quantified on each air sampling date from the 0-5 cm soil layer by extracting 50 g of fresh soil with 100 mL of 1 M KCl.The extracts were frozen and later analysed with a continuous flow autoanalyser (Seal Autoanalyser 3, Seal Analytical, Norderstedt, Germany).Concentration values were transformed to kg N ha −1 using the soil bulk density and corrected by the soil moisture.
Soil temperature and moisture content were measured using a Crison TM 65 probe (Carpi, Italy) and GS3 soil moisture probes (Decagon Devices, Pullman, WA), respectively.
Volumetric soil moisture content and soil bulk density, measured once per month for each plot by the cylinder method (Grossman and Reinsch, 2002), were used to calculate soil water filled pore space (WFPS) assuming a soil particle density of 2.65 Mg m -3 .
Maize grain yield for each plot was determined by weighing the total grain harvested by a commercial combine and corrected to 14% moisture content.A grain subsample from each plot was dried at 60 ºC for 48 h and weighed to determine maize grain moisture.
Afterwards grain subsamples were grinded and analysed to determine the N content by dry combustion (TruSpec CN, LECO, St Joseph, MI, USA).

Data analysis
Cumulative soil N2O emissions on a mass basis (i.e., kg N ha -1 ) were quantified using the trapezoid rule (Levy et al., 2017).Repeated measures analysis of variance (ANOVA) for logarithm transformed data of N2O fluxes, soil NH4 + and NO3 − content, and WFPS, and soil temperature were performed for sprinkler and flood irrigation and for 2015, 2016 and 2017 growing seasons (i.e.April -October) and for 15-16 and 16-17 fallow period (i.e.November -March) separately with soil tillage system, date of sampling and their interactions as sources of variation using the JMP 10 statistical package (SAS Institute Inc, 2012).
In addition, different ANOVA were performed for 2015, 2016 and 2017 growing seasons for cumulative N2O emissions, grain yield, N uptake by the grain, grain yield N2O scaled emissions and grain N-uptake N2O scaled emissions with irrigation system, soil tillage system and their interactions as sources of variation.When significant, differences between treatment means were evaluated by Tukey test at 5% significance level.The relationships between soil N2O flux and concentration of soil soil NH4 + and NO3 -, WFPS and soil temperature was evaluated by the significance of Pearson coefficients by using JMP 10 statistical package (SAS Institute Inc., 2012).
Soil temperature was significantly affected by the interaction between tillage and sampling date in all measurement periods except during the fallow 15-16 for S irrigation (Table 3, Fig. 2c and 2d).Over the three growing seasons, mean soil temperature was   2. Soil water-filled pore space (WFPS) in the 0-5 cm depth and soil temperature at 5 cm depth for sprinkler (a, c) and flood (b, d) irrigated plots as affected by soil tillage systems: CT (conventional tillage), NTr (no-tillage maintaining maize stover), NT (no-tillage removing maize stover).*Indicates significant differences between treatments within a date at p<0.05.
In both irrigation systems, soil NO3 -content (0-5 cm depth) increased after the preplanting application of fertilizer for a period of 45 days.The maximum values were reached during the top dressing application (V6 growth stage) of nitrogen fertilizer and lasted 4 days, then soil NO3 -content started to decrease (Fig. 3a and 3b).A significant interaction between soil tillage and sampling date was observed for both fallow periods and during 2017 under S irrigation, while under F irrigation the significant interaction between soil tillage and sampling date affected soil NO3 -content in 2016 and fallow 15-16 (Table 3).
Soil temperature (ªC  a strong increase after the fertilizer applications (Fig. 3c and 3d).Under S irrigation a 292 significant interaction between soil tillage and sampling date was observed in 2015, while 293 under F irrigation the significant interaction between soil tillage and sampling date 294 affected soil NH4 + content in 2015 and 2017 (Table 3).295
These relationships were similar in both irrigation systems, showing a quick increase on soil daily N2O fluxes when soil temperature was above 20ºC and when the total available soil inorganic nitrogen content increased due to the N fertilizer application, especially during the top dressing application (June).
In addition to the relationships between daily soil N2O fluxes and soil temperature and total available N, daily soil N2O fluxes showed a significant relationship with the WFPS for both irrigation systems (Fig. 5c).This positive relationship between soil N2O fluxes and WFPS were only observed during the pre-planting and top dressing applications of N fertilizers (i.e.24 h prior and 24, 48 72 and 96 h after N fertilizer application).However, no significant relationship between soil N2O fluxes and WFPS were observed during the rest of the measurement period.Daily soil N2O fluxes showed a strong increase when WFPS values were higher than 60% and reaching the highest fluxes at 80% of WFPS.This large impact of the WFPS on the daily soil N2O fluxes were only observed for F irrigation (black triangles), which resulted in soil N2O peak fluxes three times higher compared with S irrigation (empty circles) (Fig. 5c).Irrigation system significantly affected cumulative soil N2O emissions in 2015, 2016 and 2017 (Table 5).In the three growing seasons, S irrigation resulted in a reduction of cumulative N2O emissions of 34, 51 and 40% for 2015, 2016 and 2017, respectively, compared with F irrigation.Cumulative soil N2O emissions were not affected significantly by the tillage system nor by its interaction with the irrigation system (Table 5).
Effects and levels † Cumulative N2O emissions Grain yield was differently affected by the irrigation and the tillage system depending on the growing season.In 2015, a significant interaction between irrigation and tillage system was observed, increasing grain yield in the order NTr-S>CT-S=CT-F>NTr-F, with values ranged between 10.15 to 14.34 Mg ha -1 .Likewise, the interaction between irrigation and tillage system affected significantly the grain yield in 2016, in which NTr tillage obtained the greatest values under S irrigation (13.26 Mg ha -1 ) compared with F irrigation (10.21 Mg ha -1 ) (Table 6).However, in 2017, irrigation and tillage system, but no their interaction, had a significant impact on the grain yield, obtaining the greatest values under S irrigation and CT tillage, 17.08 and 17.00 Mg ha -1 respectively.S irrigation resulted in higher grain N uptake compared with F irrigation over the three growing season (Table 6).In 2015, the interaction between irrigation and tillage system affected grain N-uptake (Table 6).Under sprinkler irrigation, the NTr treatment produced higher N uptake than the CT treatment, but no differences were found between tillage treatments under flood irrigation.(Table 6).In contrast, in 2016 and 2017, irrigation and tillage systems affected the grain N-uptake but not their interaction.CT tillage showed higher grain N-uptake than NTr and NT tillage in 2016, while in 2017 significant differences were only observed between CT and NT tillage (Table 6).Finally, grain yield N2O scaled emissions (g N2O-N Mg -1 grain) and grain N-uptake N2O scaled emissions (g N2O-N kg -1 N grain) were significantly affected by the irrigation system in the three growing seasons, presenting the lowest values under S irrigation compared with F irrigation (Table 7).In the three growing seasons, S irrigation reported a reduction of the grain yield N2O scaled emissions of 49 (2015), 59 (2016) and 47% (2017) compared with F irrigation.Similarly, the grain N-uptake N2O scaled emissions showed a decrease of 53, 59 and 51% for 2015, 2016 and 2017, respectively, under S irrigation compared with F irrigation (Table 7).
In 2017, the tillage system showed significant differences for grain yield N2O scaled emissions and grain N-uptake N2O scaled emissions, with the greatest values under NTr and NT tillage for both indexes (Table 7).However, in 2015 and 2016, no significant differences were observed neither in grain yield N2O scaled emissions nor in grain Nuptake N2O scaled emissions between tillage systems, even when the scaled emissions were almost two times greater under NTr compared with CT as it was observed in the 2015 maize growing season (Table 7).

DISCUSSION
In the irrigated Mediterranean conditions evaluated in this study, it has been assessed that irrigation system impacted soil N2O emission in maize monocropping systems.In Differences between studies could be related to the different air sampling protocol used in both studies since in the Álvaro-Fuentes et al. ( 2016) study did not increase air sampling frequency during the fertilization events.Moreover, these differences in soil N2O fluxes between both studies could be also related with the high temporal and spatial variability of soil N2O emissions associated to different factors, such as soil properties, climate, management and microorganism populations (Leip et al., 2011;Venterea et al., 2012).
In 2016 and 2017 but not in 2015 and for both irrigation systems, soil N2O fluxes were affected by the interaction between soil tillage system and sampling date.This interaction was observed after the fertilizer application events, especially after the top dressing application, when the greatest peak of soil N2O fluxes occurred as other researchers observed previously (Smith et al., 1997;Shen et al., 2018).
The increase in total available soil nitrogen content (ammonium and nitrate) after N fertilizer applications had an impact on soil N2O fluxes in both irrigation systems, as it is demonstrated in a relationship shown (Fig. 5b).This relationship agrees with other studies (McSwiney and Robertson, 2005;Hoben et al., 2011;Zhou et al., 2016) which pointed out the key role of N fertilizer applications on the N2O emitted from the soil (Dobbie and Smith, 2003;Vallejo et al., 2005).Moreover, the soil N2O flux peaks observed during the top dressing application were related not only with the high N fertilizer rates applied (200 kg N ha -1 ) but also with the high soil temperatures measured, similar to the observations reported by other authors (Dobbie and Smith, 2001;Zhou et al., 2016).The warmer soil temperature during the top dressing application (June), could lead in more optimal conditions for the production of N2O by soil microorganisms, favouring a rapidly increase of the soil N2O fluxes, "pulsing effect", since soil temperature is a key factor that control nitrification, denitrification and nitrifier denitrification processes (Bouwman et al., 2002;Sanchéz-Martín et al., 2008;Butterbach-Bahl et al., 2013).Soil N2O peaks measured under F irrigation were 3 to 4 times greater compared to the peaks measured under S irrigation.The difference in soil N2O peaks between irrigation systems were related to the different WFPS found.Maximum N2O peak values (observed during top dressing N fertilizer application) were measured under F irrigation when WFPS were between 70 to 80%, considered as the optimum values for N2O production (Davidson, 1991).However, under S irrigation, WFPS values were always lower than 60%.Therefore, differences in WFPS explained also the higher cumulative N2O emissions under F irrigation found compared with S irrigation.On average of the Furthermore, over the three growing seasons, conventional tillage trended to result in greater grain yield and grain N uptake compared with no-tillage systems.Afzalinia and Zabihi (2014) and Salem et al. (2015) observed similar reductions in crop yields during the first year of implementation of the no-tillage systems for a maize crop under Mediterranean conditions.Several reasons such as waterlogging, poor crop establishment, lower root development by compaction, nutrient deficiencies or time of implementation are pointed out as possible reasons, which would explain the worst crop performance under no-tillage systems (Pittelkow et al., 2015).In our work, the lower mean soil bulk density found in conventional tillage compared with no-tillage systems (1.38, 1.53 and 1.53 for CT, NTr and NT, respectively) would lead to more optimal conditions for the development of maize roots and thus the better crop performance under conventional tillage (Cid et al., 2015).
Grain yield N2O scaled emissions measured were in the range of the values obtained by Omonode et al. (2015) for a maize crop under conventional tillage and no-tillage systems with a nitrogen application rate of 200 kg N ha -1 year -1 .Likewise, grain N uptake N2O scaled emissions presented in this work were in the range of values reported by Álvaro-Fuentes et al. (2016) in the same region.For both N2O scaled emissions, by grain yield and by grain N uptake, irrigation system had a significant impact.The S irrigation system presented lower values compared with the F irrigation system over the three growing seasons studied.The lower cumulative N2O emissions and the higher grain yields and N-uptake by grain obtained under S irrigation system explained the lower N2O yield-scaled emissions by grain yield and by N-uptake by grain found.Moreover, in 2017, no-tillage systems resulted in an increment of the N2O scaled emissions compared with CT tillage mainly due to the decrease in grain yield and grain N-uptake found under notillage systems.

CONCLUSIONS
In the Mediterranean conditions studied the irrigation system is an important strategy to reduce soil N2O emissions.Throughout three maize seasons, the sprinkler irrigation system reduced soil N2O emissions, and grain yield and grain N uptake N2O scaled emissions, compared to the flood irrigation system.Sprinkler irrigation is a win-win system for irrigated maize: more grain yield and lower soil N2O emissions.The soil tillage system affected daily soil N2O fluxes, especially after the fertilization events, but it had not effect on the seasonal mean soil N2O emissions.However, no-tillage systems showed a trend to increase the grain yield and grain N uptake N2O scaled emissions compared to conventional tillage systems when the same amount of water was applied.More information about the performance of no-tillage in irrigated maize monoculture systems is needed to consider no-tillage systems as a mitigation strategy of N2O emissions under Mediterranean conditions.This work pointed out the importance of an appropriate selection of irrigation and tillage system to minimize soil N2O emissions in Mediterranean agroecosystems.

Figure 4 .
Figure 4. Soil N2O flux for sprinkler (a) and flood (b) irrigation systems as affected by soil tillage systems: CT (conventional tillage), NTr (no-tillage maintaining maize stover), NT (no-tillage removing maize stover).*Indicates significant differences between treatments within a date at p<0.05.Black triangles indicate fertilizer applications.White triangles indicate flood irrigation events.

Figure 5 .
Figure 5. Regression analysis for sprinkler (empty circles) and flood (black triangles) irrigation systems

similar
Spanish conditions, Sanchéz-Martín et al. (2008) obtained soil N2O fluxes for a furrow-irrigated maize similar to the measured in this work under F irrigation.The higher soil N2O fluxes found with F irrigation compared to S irrigation in our work were closer to those measured by Omonode et al. (2011) in a non-irrigated maize in Indiana.Comparing with sprinkler irrigated maize studies, soil N2O fluxes measured in this work were in the range of values reported by Liu et al. (2005) in Colorado.Under Mediterranean conditions, Sanz-Cobena et al. (2012) reported similar soil N2O fluxes when they used an irrigation scheduling similar to our work.In contrast, Álvaro-Fuentes et al. (2016), in the same study area, observed soil N2O fluxes three and ten times lower than the values measured in our study (for S and F irrigation systems, respectively).

Table 1 .
Soil characteristics of the experimental field.

Table 2
. Calculated crop evapotranspiration (ETc), crop irrigation requirement (CIR) and irrigation water applied in both irrigation systems (sprinkler and flood) applied in the maize growing season of2015, 2016     and 2017.

Table 3
ANOVA (p-value)for soil water-filled pore space (WPFS) (0-5 cm), soil temperature (5 cm depth), and soil nitrate and ammonium content (0-5 cm) for sprinkler and flood irrigation as affected by tillage, date and their interaction over the different measurement periods.

Table 4 .
ANOVA (p-value)for the daily soil N2O flux for sprinkler and flood irrigation over the different measurement periods as affected by soil tillage, sampling date and their interaction.
For each effect and growing season values followed by different letters are significantly different according to a Tukey test at p=0.05 level.NS, non-significant.p-values are given when significant. † For each effect, growing season and variable the values followed by different letters are significantly different according to Tukey test at p=0.05 level.NS, non-significant.p- † For each effect, growing season, and variable the values followed by different letters are significantly different according to Tukey test at p=0.05 level.NS, non-significant.pvalues are given when significant. †