Legume pastures can reduce N2O emissions intensity in subtropicalcereal cropping systems

Massimiliano De Antoni Migliorati a,*, Michael Bell b, Peter R. Grace a, Clemens Scheer a,David W. Rowlings a, Shen Liu c

a Institute for Future Environments, Queensland University of Technology, Brisbane, QLD 4000, AustraliabQueensland Alliance for Agriculture and Food Innovation, University of Queensland, Kingaroy, QLD 4610, Australiac School of Mathematical Sciences, Queensland University of Technology, Brisbane, QLD 4000, Australia

A R T I C L E I N F O

Article history:Received 19 August 2014Received in revised form 5 February 2015Accepted 16 February 2015Available online 25 February 2015

Keywords:Nitrous oxide emissionsLegume pastureCereal cropping systemN fertiliser managementAutomated greenhouse gas measuringsystem

A B S T R A C T

Alternative sources of N are required to bolster subtropical cereal production without increasing N2Oemissions from these agro-ecosystems. The reintroduction of legumes in cereal cropping systems is apossible strategy to reduce synthetic N inputs but elevated N2O losses have sometimes been observedafter the incorporation of legume residues. However, the magnitude of these losses is highly dependenton local conditions and very little data are available for subtropical regions. The aim of this study was toassess whether, under subtropical conditions, the N mineralised from legume residues can substantiallydecrease the synthetic N input required by the subsequent cereal crop and reduce overall N2O emissionsduring the cereal cropping phase. Using a fully automated measuring system, N2O emissions weremonitored in a cereal crop (sorghum) following a legume pasture and compared to the same crop inrotation with a grass pasture. Each crop rotation included a nil and a fertilised treatment to assess the Navailability of the residues. The incorporation of legumes provided enough readily available N toeffectively support crop development but the low labile C left by these residues is likely to have limiteddenitrification and therefore N2O emissions. As a result, N2O emissions intensities (kgN2O-Nyield�1 ha�1) were considerably lower in the legume histories than in the grass. Overall, these findingsindicate that the C supplied by the crop residue can be more important than the soil NO3

� content instimulating denitrification and that introducing a legume pasture in a subtropical cereal cropping systemis a sustainable practice from both environmental and agronomic perspectives.

ã 2015 Elsevier B.V. All rights reserved.

1. Introduction

Mitigating climate change and achieving food security are twoof the key challenges of the twenty-first century. Cereals are by farthe world’s most important food source, contributing on average50% of daily energy intake and up to 70% in some developingcountries (Kearney, 2010). By 2050 the world’s population isforecast to be over a third larger than at present (UNFPA, 2011) andcereal demand is predicted to increase by 60% (FAO, 2009).Pronounced intensification of cereal production is expected to takeplace in subtropical regions (Smith et al., 2007), identifying theneed for more nitrogen (N) to be supplied to these agro-ecosystems.

There is consensus (e.g., Tilman et al., 2002; Crews and Peoples,2004; Jensen et al., 2012) that both the manufacture and use ofsynthetic N fertilisers in crop production generate substantialenvironmental threats, with the emission of significant amounts ofnitrous oxide (N2O) arguably one of the most important. N2O is apotent greenhouse gas (298 CO2-eq over a 100 year time horizon(Myhre et al., 2013) associated alsowith the depletion of the ozonelayer in the stratosphere (Ravishankara et al., 2009). Together,these issues confirm that alternativemeans of intensificationmustbe implemented to avoid the increase of subtropical cerealproduction through an overuse of synthetic N. If not, the resultwould be a net increase in N2O emission rates from these agro-ecosystems.

The reintroduction of legumes in crop rotations is one possiblestrategy to reduce synthetic N inputs whilst sustaining grain yields(Crews and Peoples, 2004; Jensen et al., 2012). Owing to theirability to fix atmospheric N2 in symbiosis with rhizobia bacteria,legumes can reduce N demand of the subsequent crop, and

* Corresponding author. Tel.: +61 7 3138 1360; fax: +61 7 3138 4438.E-mail address: max.deantonimigliorati@qut.edu.au (M. De Antoni Migliorati).

http://dx.doi.org/10.1016/j.agee.2015.02.0070167-8809/ã 2015 Elsevier B.V. All rights reserved.

Agriculture, Ecosystems and Environment 204 (2015) 27–39

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment

journal homepage: www.elsev ier .com/ locate /agee

consequently decrease N2O emissions associated with synthetic Nfertilisers. In an extensive review on the use of legumes tomitigateclimate change, Jensen et al. (2012) concluded that N2O emissionsduring the legume growing season did not differ substantially fromunplanted or unfertilised soils. However, elevated N2O losses weresometimes reported after the termination of a legume crop, whenplant residues were returned to the soil (Gomes et al., 2009; Pappaet al., 2011). The low C:N ratio of legume residues can indeed leadto rapid tissue mineralisation once incorporated into the soil. As aresult, accumulation of mineral N can occur in the soil, increasingthe potential for substantial amounts of N to be lost as N2O vianitrification and denitrification (Jensen et al., 2012).

The magnitude of N2O losses in response to legume residueincorporation is, however, highly dependent on local climate andsoil conditions (Rochette et al., 2004). Subtropical croppingsystems are characterised by intense and frequent rainfall eventsduring the summer months. The warm and moist soil conditionsduring these periods can accelerate legume tissue mineralisationcompared to temperate environments, leading to ideal conditionsfor nitrifying and denitrifying bacteria and therefore magnifyingthe risk of high N2O emissions (Granli and Bøckman, 1995; Skibaet al., 1997).

However, little data is currently available for these agro-ecosystems (Mosier et al., 2004) and, although numerous studieshave investigated N2O emissions after the termination of legumeley pastures prior to a return to cropping, the vast majority wereconducted in temperate climates (Wagner-Riddle and Thurtell,1998; Baggs et al., 2000; Robertson et al., 2000; Rochette et al.,2004; Schwenke et al., 2010).

The overall aims of this study were therefore to assess whether,under subtropical conditions: (i) the N mineralised from legumeresidues can substantially reduce the synthetic N input required bythe subsequent cereal crop; (ii) N2O losses occurring after theincorporation of legume residues can be minimised via synchro-nizing the release of N derived from the residues with the Ndemand of the subsequent crop; (iii) reducing the synthetic Ninput applied to a cereal in rotation with a legume crop cansignificantly decrease overall seasonal N2O during the cerealphase.

Seasonal N2O emissions and yields were monitored in a cerealcrop (sorghum) following a legume (legume–ley pasture) andcompared to the same crop in rotationwith a non-leguminous crop(grass ley pasture). Each rotation included both a nil and a fertilisedtreatment. The N fertiliser applied to sorghum in the legume–cereal rotationwas reduced compared to the grass–cereal to assessthe availability of the N fixed by the legume ley pasture.

This study is the first to use a fully automated greenhouse gasmeasuring system to precisely quantify N2O emissions in asubtropical cereal crop after the termination of a legume leypasture. The results of this study will contribute to definemitigation strategies for the sustainable intensification of sub-tropical agroecosystems.

2. Materials and methods

2.1. Local climate and soil characteristics

This experiment was conducted in the subtropical region ofAustralia at the J. Bjelke Petersen Research Station of theDepartment of Agriculture, Fisheries and Forestry (DAFF). Theresearch site is located in Kingaroy (26�34054,300S, 151�49043.300E,altitude 441m a.s.l), in the southern Burnett region of southeastQueensland, Australia. The subtropical climate (classified as Cfaaccording to the Köppen climate classification) has warm, humidsummers and mild, dry winters. Daily mean maximum andminimum temperatures range from 20.1 �C to 4.0 �C in winter and

from 29.6 �C to 16.5 �C in summer, respectively. Local mean annualprecipitation is 776.2mm and varies from a minimum of 28.6mmin August to amaximumof 114.1mm in January (Australian Bureauof Meteorology, 2015). The soil is a Orthic Ferralsol (FAO, 1998),characterized by relatively slowpermeability and high clay content(50–65% clay). The effective rooting depth is 1.2m and the plantavailable water holding capacity is 100mm. The main physical andchemical soil properties of the field site are highlighted in Table 1.

2.2. Experimental set-up

The experiment was established in a slit plot design with twomain plots (legume and grass ley pastures) and two sub plots(N fertiliser rates) with three replicates. Each main plot was30m�10.8m, with main plots split into two subplots (15m� and10.8m) during the sorghum cropping season. Allowing for bufferrows, the effective subplot area was 12m�7.2m, or 8 crop rowsspaced 0.9m apart.

2.2.1. Cropping historiesN2O emissions and yields were measured in plots planted with

sorghum (Sorghum bicolor L.) following two distinct croppinghistories. One crop rotation (hereafter called legume croppinghistory) included two seasons of alfalfa pasture (Medicago sativa, L.,summers 2009/2010 and 2010/2011), one season of maize (Zeamays, L., summer 2011/2012) and one season of sulla ley pasture(Hedysarum coronarium L., winter 2012) prior to sowing sorghum.The other crop rotation (hereafter called grass cropping history)included two seasons of a mixed rhodes grass (Chloris gayana, K.)and alfalfa pasture (summers 2009/2010 and 2010/2011), oneseason of maize (summer 2011/2012) and one season of wheat(Triticum aestivum L., winter 2012). Although the mixed alfalfapasture was sown in consociation with rhodes grass, the rhodesgrass became rapidly predominant and by the end of the firstseason the pasture was composed almost completely by rhodesgrass. All crops in both rotations were unfertilised. Sulla andwheatwere direct drilled in August 2012 and managed as forage crops.Both crops were terminated 28 November 2012 with all residuesreturned to the soil as mulch before being incorporated with fourshallow cultivations (20 cm). The incorporation of sulla residues(2.3 t dry matter ha�1, 1.57% N) was estimated to supply the soilapproximately 36 kg Nha�1, while wheat residues (1.24 t drymatterha�1, 0.75% N) about 9 kg Nha�1. The entire field trial wasirrigated with 20 mm on 10 December 2012, two days before plotswere planted with sorghum (12 December 2012). Further detailson the management of the two crop rotations can be found in Bellet al. (2012).

2.2.2. Sorghum establishment and managementSorghum (cultivar Pioneer G22) was planted with a plant

density of 7 plantsm�2 and an inter-row space of 90 cm. Two N

Table 1Main soil physical and chemical properties of first 30 cm of soil profile for the twocropping histories (mean� SE, n =3) at the beginning of the sorghum season atKingaroy research station, Queensland, Australia.

Soil property (0–30 cm) Legume Grass

pH (H2O) 5.12�0.03 5.30�0.02DOC (kgCha�1)* 43.04�11.98 56.05�2.97PMN (kgNha�1)** 12.78�1.33 9.25�1.08Bulk density 0–30 cm (g cm�3) 1.18�0.08Texture (USDA) ClayClay (%) 55Silt (%) 14Sand (%) 31

* Dissolved organic carbon.** Potentially mineralisable nitrogen.

28 M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39

fertilisation rateswere tested on each cropping history, resulting ina total of four treatments:

� L0: Sorghum grown in the legume cropping history, no Napplied;

� L70: Sorghum grown in the legume cropping history, 70kgNha�1 applied;

� G0: Sorghum grown in the grass cropping history, no N applied;� G100: Sorghum grown in the grass cropping history, 100 kgNha�1 applied.

Treatments L70 and G100 were base dressed at planting,banding 20 kg Nha�1 as urea. On 15 January 2013 (eight leaf stage)both treatments were inter-row cultivated and side dressed withbanded urea, receiving 50 kg Nha�1 (L70) or 80kg Nha�1 (G100).The N application rate for G100was designed to achievemaximumyield potential and was representative of farming practices of theregion. The synthetic N rate used in L70 was reduced compared toG100 to assess whether the estimated 30kg Nha�1 resulting fromthe mineralisation of the sulla residues would have been availableto sorghum.

To prevent water stress limiting the potential yields, the trialwas irrigated three times over the season (25mm on 18 December2012, 40mm on 4 January 2013 and 40mm on 18 January January)using surface stored damwater and overhead sprinklers. Sorghumwas harvested on 18 June 2013. The trial area was left fallow untilbeing cultivated on 6 August 2013 (offset disc and chisel plough toa depth of 20 cm) and on 19 September 2013 (offset disc to a depthof 20 cm) to prepare the seedbed for the next crop. Irrigations wereconducted on 27 and 29 August 2013 (30 and 40mm, respectively)to assess whether significant amounts of N were still available fornitrification or denitrification after harvest. Details about croprotations and farming operations are displayed in Table 2.

2.3. Measurement of N2 and CO2 emissions

The use of a fully automated greenhouse gas measuring systemenabled a long-term high temporal resolution dataset to beestablished. N2O fluxes were monitored for nine months, fromsorghum planting (12 December 2012) to the final preparation ofthe seedbed for the subsequent crop (19 September 2013) to assessoverall N2O losses over the cropping season as well as the post-harvest period.

N2O emissions were captured using twelve automated sam-pling chambers (one per plot) made of transparent acrylic panels.Each chamber measured 50 cm�50 cm�15 cm and was attachedvia a rubber seal to stainless steel frames inserted 10 cm into the

ground. The chambers were closed airtight with lids operated bypneumatic actuators and connected to a fully automated samplingand analysis system as described in De Antoni Migliorati et al.(2014b) and Scheer et al. (2013).

During a measurement cycle a set of four chambers closed for60min with each chamber sampled 4 times for 3min. A certifiedgas standard of 500ppb N2O (BOC – Munich, Germany- and AirLiquide –Dallas, TX, USA)was pumped into the gas chromatographevery 15min. At the end of the cycle the chambers reopened andthe next set of four chambers closed for sampling. One completecycle of twelve chambers lasted 3h, during which each chamberwas sampled for 1h and remained opened for 2h to restoreambient conditions. This method enabled the determination of upto 8 single fluxes per chamber per day.

The air samples taken from each chamber headspace wereautomatically pumped to the sampling unit. During the 3minsampling period the air sample was continuously analysed for CO2

concentration using a single path infra-red gas analyser (Licor, LI820, St. Joseph, MI, USA). N2O concentrationwas analysed injecting3ml of gas sample into the carrier gas (N2) of a gas chromatograph(Model 8610C, SRI Instruments, USA) equippedwith a 63Ni electroncapture detector (ECD). A column filter containing sodiumhydroxide-coated silica (Ascarite, Sigma–Aldrich, St. Louis, MO,USA) was installed upstream of the ECD to minimize the CO2 andwater vapor interference on N2O measurements. The column filterwas replaced every two weeks.

During the 3h of a complete measuring cycle the system wasautomatically calibrated twelve times by a single point calibrationusing the certified gas standard of 500ppb N2O. Greater accuracywas achieved with a multi-point calibration during the measuringseason using certified gas standards of 500, 980, 5030ppb N2O(BOC – Munich, Germany). The GC response over this range wasdetermined to be linear so no correctionwas necessary to preciselydetermine high fluxes. The detection limit of the system wascalculated using the methodology established by Parkin et al.(2012) and was approximately 0.5 g N2O-N ha�1 day�1 for N2O and1kg CO2-C ha�1 day�1 for CO2, respectively. Throughout the seasonall system components were constantly checked for leaks, makingthe sample dilution due to leakage negligible. The system wasprogrammed to open the chambers if rain events exceeded 5mmor the internal air temperature of the chamber exceeded 55 �C.During irrigation events the system was stopped and all chamberlids were opened to allow water to enter inside the chambers.

The chamber placement strategy was based on the methodolo-gy established by Kusa et al. (2006) and Parkin and Kaspar (2006)in order to measure N2O emissions from both a diffused source(crop residues) and a localized source (banded fertiliser). Two of

Table 2Details of crop rotations and farming operations for the four treatments at Kingaroy research station.

Date L0 L70 G0 G100

Summer 2009/2010 Alfalfa pasture Alfalfa pasture Rhodes grass + alfalfa pasture Rhodes grass + alfalfa pastureSummer 2010/2011 Alfalfa pasture Alfalfa pasture Rhodes grass + alfalfa pasture Rhodes grass + alfalfa pastureSummer 2011/2012 Maize Maize Maize MaizeWinter 2012 Sulla pasture Sulla pasture Wheat Wheat28–30/11/2012 Pastures terminated and entire field trail cultivated four times with offset disc to 20 cm10/12/2012 Entire field trail irrigated with 20mm12/12/2012 Sorghum planted Sorghum planted applied 20 kg Nha�1 Sorghum planted Sorghum planted applied 20 kg Nha�1

18/12/2012 Entire field trail irrigated with 25mm04/01/2013 Entire field trail irrigated with 40mm14/01/2013 Applied 50 kg Nha�1 Applied 80 kg Nha�1

18/01/2013 Entire field trail irrigated with 40mm18/06/2013 Entire field trail harvested06/08/2013 Entire field trail cultivated twice with offset disc down to 20 cm27/08/2013 Entire field trail irrigated with 30mm29/08/2013 Entire field trail irrigated with 40mm19/09/2013 Entire field trail cultivated with offset disc down to 20 cm

M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39 29

the three replicate chambers of each treatment were positionedover the crop row and the third in the inter-row. Chamberspositioned over the crop row included the banded fertiliser (10 cmfrom the plant row) in the chamber area. Sorghumplants inside thechamber positioned over the crop row were cut when exceedingthe chamber headspace, a practice established by Drury et al.(2008),Halvorson et al. (2008) and Hu et al. (2013). As recom-mended by Kusa et al. (2006) and Parkin and Kaspar (2006), theimpact of this practice on belowground C and N dynamics, andtherefore N2O emissions, was minimised by relocating allchambers placed over the crop row to a new section of the croprow every fortnight. This strategy proved to be effective as onlymarginal differences in daily N2O emissions were observed in thecontrol treatments between chambers placed in the inter-row andover the crop row (Table 3, Fig. 5).

The measuring system was deployed immediately afterplanting and temporarily withdrawn to permit farming operations(side dressing, harvest, post-harvest cultivations). During the ninemonths of this study an average of 2700 valid N2O and CO2 fluxeswere obtained for each treatment.

2.4. Calculation of N2 and CO2 emissions

Hourly N2O fluxes were calculated with the method describedby Barton et al. (2008), determining the slope of the linear increaseor decrease of the four gas concentrations measured during the60min of chamber closure period. In contrast, hourly CO2 fluxeswere computed using the linear increase of six concentrationsmeasured in the first two sampling intervals, a method used toavoid possible saturation of CO2 partial pressure in the chamber.N2O and CO2 fluxes were corrected for the three factors of airtemperature inside the chamber, atmospheric pressure and theratio between chamber volume and soil area using:

F ¼ b� VCH �MW� 60� 106

ACH �MVcorr � 109 (1)

where F is the emission rate (mgm�2 h�1), b is the variation of gasconcentration inside the chamber (ppbmin�1), VCH is the volumeof the chamber (m3), MW is the gas molar weight (28 for N2O-Nand 12 for CO2-C), 60 is the conversion from minutes to hours, 106

converts g tomg,ACH is the surface area of the chamber (m2),MVcorr

is the mole volume (m3mol�1) corrected for pressure andtemperature as presented in Eq. (2), 109 converts ppb to mLm�3.

MVcorr ¼ 0:02241� 273:15þ T273:15

� p0p1

� �(2)

where 0.02241m3 is 22.41 L molar volume, T is the temperature ofthe chamber at the time of the measurement (Kelvin), p0 is the airpressure at sea level and p1 is the air pressure at the study site. Toprovide greater accuracy, air pressure at the site was determinedusing a barometric equation based on the local altitude.

The Pearson correlation was then used to quality-check fluxmeasurements. Fluxes above the detection limit were discarded ifthe regression coefficient (r2) was<0.80 for N2O and<0.90 for CO2,

respectively. Mean daily fluxes for each treatment were calculatedusing weighted averages of hourly data from the three replicates.That is, for each treatment, hourly fluxes from the two chambersover the crop row (covering 50 cm around the crop row) wereaveraged. The obtained mean flux was then averaged with themean of hourly fluxes measured by the chamber in the inter-row(covering 50 cm in the inter-row). This method made it possible toaccurately calculate the average N2O emissions of each treatment,accounting for the spatial variability occurring between two croprows (98 cm). Cumulative N2O fluxes (kgN2O-Nha�1) weredetermined by summing daily N2O fluxes measured during thestudy period.

Emission factors were corrected for background emissions(Kroeze et al., 1997) using the following:

EF% ¼ N2OðFertÞ � N2OðUnfertÞN fertiliser input

� 100 (3)

where EF% is the emission factor reported as a percentage of Nfertiliser input (kgNha�1 season�1) lost as N2O-N, N2O (Fert) andN2O (Unfert) (kgNha�1 season�1) are the cumulative N2O-Nemissions measured in the fertilised and unfertilised treatmentswith the same cropping history, respectively.

Soil CO2 fluxes, considered a proxy data for estimating soilrespiration rates (inclusive of microbial and roots respiration),were calculated using only the chambers placed in the inter-row.Missing daily N2O and CO2 fluxes (due to rare occasional failures ofthe measuring system) were estimatedwith the Amelia II multipleimputation model (Honaker and King, 2010) using daily values ofsoil water-filled pore space (WFPS) (0–10 cm, 10–20 cm and20–30 cm) and mineral N content (0–30 cm).

2.5. Auxiliary measurements

Chamber air temperature and soil temperature were measuredevery 5min using RTD probes (Temperature Controls Pty Ltd.,Australia) buried at 10 cm depth near to a chamber. Four FrequencyDomain Reflectometers (FDR, EnviroScan probes, Sentek Technol-ogies, Australia) measured water dynamics of the soil profilethroughout the field trial. Before the beginning of the experimentall FDR probes were calibrated for the local soil type followingproducer recommendations (Sentek Technologies, 2011). The FDRprobeswere deployed at planting and programmed tomeasure thevolumetric soil water content at three depths (0–10 cm, 10–20 cm,20–30 cm) at 30min intervals. Water filled pore space (WFPS) wascalculated using a particle density of 2.79 g cm�3.

Soil chemical properties at the site were determined atsorghum planting, with soil samples collected from every plotwith a manual open-faced bucket auger (10 cm diameter). Eachplot was sampled at three depths (0–10, 10–20, 20–30 cm) andthen analysed for texture (hydrometer method as described byCarter and Gregorich, 2007), pH, NH4-N, NO3-N, dissolved organicC and potentially mineralisable N. Routine soil sampling was thenconducted at regular intervals during the growing season and soil

Table 3Seasonal N2O average fluxes, cumulative N2O fluxes, N2O intensities (mean� SE, n =3), emission factors and cumulative CO2 fluxes as a function of the four treatments.Meansdenoted by a different letter indicate significant differences between treatments (p<0.05).

Measurement Treatment

L0 L70 G0 G100

Average flux (gN2O-Nha�1 d�1) 0.85�0.08 a 2.41�0.82 a 0.94�0.18 a 5.07�0.58 b

Cumulative N2O flux (kgN2O-Nha�1 season�1) 0.24�0.02 a 0.68�0.23 a 0.27�0.05 a 1.43�0.16 b

N2O intensity (kgN2O-N t-yield�1 ha�1) 0.09�0.0 a 0.13�0.04 a 0.28�0.05 b 0.28�0.03 b

Emission factor (%) – 0.63 – 1.17Cumulative CO2 flux (kg CO2-C ha�1 season�1) 3340.33 3495.97 2922.56 3330.21

30 M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39

samples (0–10,10–20, 20–30 cm)were analysed for NH4-N, NO3-N.Each soil sample consisted of three subsamples taken at 10 cmintervals from the crop row of then mixed in order to ensure itrepresented the banded and non-banded areas of the plot.

Soil NH4-N and NO3-N were extracted by shaking 20 g soil in100ml 1M KC1 solution at room temperature for 60min (Carterand Gregorich, 2007). This solutionwas then filtered and stored ina freezer until analysed colorimetrically for NH4-N and NO3-Nusing an AQ2+ discrete analyser (SEAL Analytical WI, USA).Dissolved organic C was extracted at room temperature by shaking20g soil in 100ml of deionised water for 60min. The suspensionwas then centrifuged for 15min at 10,000 rpm and the supernatantfiltered with a 0.45mm pore diameter cellulose membrane filter(based on Scaglia and Adani, 2009). The samples were analysedusing a supercritical water oxidation technique with the SeiversInnoVox laboratory TOC analyser (General Electric, Boulder, CO,USA).

Potentially mineralisable N was determined by incubating soilsamples at field capacity at 30 �C for 0, 7 and 14 days (Bremner,1965). The samples were taken on 12 December 2012 samplingeach plot at three depths (0–10, 10–20, 20–30 cm). Mineral-Nformed during the incubation was measured by 2M KCl extractionfollowed by automated colorimetric determination. For eachtreatment, the amount of potentially mineralisable N wascalculated as the difference between the mineral N determinedat day 7 and 14 in order to avoid the Birch effect (Birch, 1958).

Total biomass was determined at physiological maturity bycollecting duplicate samples of 1m of crop row in each plot.Samples were oven dried at 60 �C to determine dry weight beforebeing weighed and ground for total N content, which wasmeasured using a C-N analyser after Dumas combustion (LECOTruMac LECO Corporation, St. Joseph, MI, USA). Grain yield wasmeasured in each plot by harvesting duplicate 1.8mwide strips forthe plot length using a plot combine. Grain sampleswere also driedat 60 �C before quantification of yields on a dry weight basis, whilegrain N was determined using a methodology similar to that inbiomass samples.

Fertiliser N recovery in the crop (REfN)was determined applying15N-labelled urea in micro-plots (0.9m�1.5m) located next to themeasuring chambers. The L70 and G100 treatments received 5%excess 15N enriched urea both at planting and at side dressing. The

15N-labelled urea was dissolved in 1 L of deionised water andapplied as a liquid solution in a sub-surface band, minimising inthis way N losses via runoff and NH3 volatilisation. Plants in themicro-plots were sampled at crop harvest by collecting above- andbelow-ground material. The 15N analysis was performed using a20–22 isotope ratio mass spectrometer (Sercon Limited, UK). Forfurther information on experimental settings and main findingssee De Antoni Migliorati et al. (2014a).

2.6. Statistical analysis

Statistical analyses were performed within the MATLAB 2012aenvironment (MathWorks Inc., Natick, MA, US), where thetemporal patterns of daily N2O emissions displayed by the fourtreatments were compared using the autoregressive integratedmoving average (ARIMA) model (Box and Pierce, 1970). This modelwas fitted to each time series using the following formula:

DdYt ¼ f0 þ f1D

dYðt�1Þ þ . . .þ fpDdYðt�pÞ þ at þ u1aðt�1Þ

þ . . .þ uqaðt�qÞ; (4)

where D = (1�B) are the backshift operators. The roots of the ARand MA polynomials satisfied the stationarity and invertibilityconditions, respectively. The values of p, d and q were determinedby the Bayesian information criterion (BIC), and the unknownparameters were estimated by least squares estimators. The

residuals at were computed by FðBÞQ�1ðBÞYt, where FðBÞ and

QðBÞ denote the estimates of the autoregressive and movingaverage parameters, respectively. The bootstrap residual resam-pling method was used to evaluate the variation, while predictionintervals were constructed using the percentile method.

Amixed-design analysis of variance (ANOVA) was performed todetermine the influence of fertilisation rate or cropping history onN2O emissions and grain yields. The Bonferroni post hoc test wasused to compare average and cumulative N2O emissions, N2Ointensities, grain yields, above ground biomass productions andharvest indexes.

As with the method used for the calculation of daily N2O fluxes,standard errors of average and cumulative N2O emissions and N2Ointensities were calculated by assigning different weights to thechamber in the inter-row (0.5) and over the rows (0.25).

[(Fig._1)TD$FIG]

Fig. 1. Minimum and maximum daily air temperatures, soil temperatures (0–30 cm), rainfall and irrigation events at Kingaroy (Queensland, Australia) during the sorghumseason.

M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39 31

3. Results

3.1. Environmental conditions

Over the study period (12 December 2012–19 September 2013)a total of 827mm of rain fell at the study site, including one heavyrainfall event of 234mm during a thunderstorm on 27 January2013. While this total corresponded to 148% of the growing seasonmean for the 108-year period between 1905 and 2013, it isnoteworthy that over 70% of the total rainfall was concentratedbetween 25 January and 3March (Fig.1). Themean air temperaturewas 17.9 �C, with the maximum (38.0 �C) and minimum (�4.4 �C)hourly air temperatures recorded in January and August 2013,respectively. Average soil (0–10 cm) temperatures ranged from amaximum of 29.7 �C (December 2012) to a minimum of 11.9 �C(May 2013).

3.2. Seasonal variability of soil conditions

TheWFPS of the topsoil (0–30 cm) varied in response to rainfall,irrigation events and crop growth. At the beginning of the seasonWFPS values fluctuated between 36% and 47% due to the rapid cropdevelopment and irrigation events (Fig. 2). WFPS ranged from 40%to 66%, with highest values measured between late January andmid-March as a consequence of the intense rainfall events thatoccurred during this period. After March average WFPS valuesstarted gradually to decrease as a result of the declined rainfallregime.

In the 0–30 cm sampling zone, NH4+-N was the dominant form

of soil mineral N. With the possible exception of the samplingevent at sowing, no consistent response to history or fertiliserapplication was observed (Fig. 3a). At sorghum planting, NH4

+-Ncontents in the top 30 cm averaged 40 kgNha�1 and then stabilisedbetween 16 and 33kg Nha�1 for the remainder of the season.Limited response to history or N fertiliser applicationwas observedalso in soil NO3

�-N contents (Fig. 3b), with the exception of thefirst month of the growing season. At this stage the legumehistories (L0 and L70) showed higher NO3

�-N contents than thegrass history equivalents, although differences were lower than20 kg N/ha.

At the beginning of the sorghum season dissolved organiccarbon contents in the grass history tended to be higher than in the

legume. Slightly higher potentially mineralisable N values wereobserved in the legume history treatments compared to the grass(Table 1).

3.3. N2 emissions

N2O emissions varied temporally and spatially in response to Nfertilisation rate and cropping history. In all treatments significantN2O losses occurred betweenmid-December 2012 and mid-March2013, when soil mineral N contents and WFPS values were higher

[(Fig._2)TD$FIG]

Fig. 2. Daily soil N2O fluxes and water-filled pore space (WFPS, 0–30 cm) for the four treatments during the sorghum season in Kingaroy (Queensland, Australia). Arrowsindicate the timing of N fertiliser applications.

[(Fig._3)TD$FIG]

Fig. 3. Soil ammonium (a) and nitrate (b) contents (0–30 cm) for the fourtreatments during the sorghum seasons in Kingaroy (Queensland, Australia).Arrows indicate the timing of N fertiliser applications.

32 M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39

than in the remainder of the study period (Fig. 2). During thisperiod N2O emissions increased shortly after rain or irrigationevents. This trend was particularly evident in the G0 andG100 treatments, where N2O emission rates increased moreabruptly than in L0 and L70 (Fig. 2).

In all except G0 treatments, the highest emission pulse wasobserved after the rainfall event on 27 January, when a total of234mm rainfall fell over 24h. During this event N2O emissions in

treatments L70 and G100 were up to 4 and 6 fold those of theunfertilised treatments, respectively. After this event, N2Oemission in all treatments progressively declined to backgroundlevels, with the only exception being G100, where a substantialN2O emission pulse was measured after another 260mm rainfallfell at the field site between 19 February and 3March (Fig. 2). Aftermid-March daily N2O fluxes in all treatments never exceeded 1g Nha�1 day�1 despite several rain events. Emissions did not increase

[(Fig._4)TD$FIG]

Fig. 4. 95% confidence intervals of N2O fluxes in the different treatments during the period of highest emissions (December 2012–March 2013) in Kingaroy (Queensland,Australia). Confidence intervals are displayed using different scales. Arrows indicate the timing of N fertiliser applications.

M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39 33

even after the two irrigation events on 27 and 29 August 2013 orthe two cultivation events on 6 August and 19 September 2013.

During the period of highest emissions (December 2012 toMarch 2013) the ARIMA model highlighted significant treat-ment effects on the temporal pattern of N2O emissions. Beforeside dressing, the two N2O emission pulses measured inG100 and G0 significantly exceeded those in L70 and L0,respectively (Fig. 4a and b). After side dressing the emission

pulse in G100 was significantly higher than that in L70, whileno substantial differences were observed between G0 and L0(Fig. 4c and d).

The chamber placement highlighted two different patterns inthe spatial variability of N2O flux rates during the period of highemissions. In the two unfertilised treatments and in G100, N2Ofluxes from the crop row (inclusive of the banded fertiliser) did notdiffer significantly from those from the inter-row (Fig. 5b–d). In the

[(Fig._5)TD$FIG]

Fig. 5. Daily soil N2O fluxesmeasured in the row (R) and inter-row (IR) chambers for the L70 (a), L0 (b), G100 (c) and G0 (d) treatments during the period of highest emissions(December 2012–March 2013) in Kingaroy (Queensland, Australia). Arrows indicate the timing of N fertiliser applications. Graphs are in different scales.

34 M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39

L70 treatment instead, average N2O emissions measured in thecrop row exceeded those in the inter-row by a factor of 6 (Fig. 5a).

Only the N2O cumulative losses measured in G100 weresignificantly higher (p<0.05) than those of the other treatments.Cumulative losses in L70 did not display significant differencescompared to the unfertilised treatments (Table 3).

The mixed-design ANOVA analysis indicated that the maineffect regulating N2O emissions was the N fertiliser rate, while thecropping history per se had no significant effect on measured N2Olosses (Table 4).

3.4. CO2 emissions

Soil CO2 fluxes showed little variation between treatments andexhibited a temporal pattern influenced by soil temperatures andWFPS (Fig. 6). Average soil CO2 emissions peaked during thewarmest months (early January to late March 2013, average of23 kg CO2-C ha�1 day�1) before decreasing to <10kg CO2-Cha�1 day�1 during the colder and drier period from April to lateAugust 2013.During the fallow period CO2 fluxes remained below5kg CO2-C ha�1 day�1. Emissions in all treatments did not increaseafter the tillage event of 6 August 2013 but rose to an average of17 kg CO2-C ha�1 day�1 after the two irrigation events of 27 and29 August.

In contrast to N2O emissions, CO2 fluxes tended to not to riseuntil several days after a rainfall/irrigation event. This wasparticularly evident with the rainstorm on 27 January, when

CO2 emission did not start to increase until seven days after theevent. Overall, cumulative CO2 emissions measured in the inter-rows showed little variations between treatments (Table 3) and nosignificant differences in the pattern of daily CO2 emissions wasdetected by the ARIMA model.

3.5. Crop biomass, grain production and N uptake

Sorghum biomass production and yield were substantiallyaffected by cropping history and N fertiliser rate. Both biomass andyield in the unfertilised sorghum following the legume ley pasture(L0) were significantly higher (p<0.05) than those in thecorresponding treatment following the grass ley pasture (G0).Grain and biomass production in L70 were comparable to those inG100 and both were significantly higher than those in theunfertilised treatments (Table 5). The harvest index (kg grainha�1/kg total biomassha�1) of L0 was significantly higher than inG0, but comparable to that of both L70 and G100.

The mixed-design ANOVA analysis showed that both the Nfertiliser rate and the cropping history had significant effects ongrainyield. However, the cropping history F valuewas substantiallylower than that of the fertiliser rate, indicating that the fertiliserrate had greater influence on yields (Table 4).

Soil N availability in the legume cropping history was higherthan in the grass one and N uptake values measured in L0 andL70 exceeded those of G0 and G100, respectively. Though the Nfertiliser rate in G100 was 30kg Nha�1 higher than in L70, thefraction of fertiliser N taken by the crop was greater in L70,exhibiting a significantly higher REfN compared to G100 (Table 5).

4. Discussion

4.1. N2 emissions from cropped soils after termination of a pasturephase

To date little research has been undertaken on N2O emissionsfollowing the termination of ley pastures, specifically in terms ofhow management, subtropical climatic conditions and chemicalcomposition of the residues influence N2O losses during

Table 4Significance of treatment effect (applied fertiliser rate and cropping history) on N2Oemissions and grain yields during the sorghum season.

Measurement Factor p-value F statistics

N2O Fertiliser rate ** 17.37Cropping history NS 2.68

Grain yield Fertiliser rate *** 255.79Cropping history ** 14.41

NS: not significant.** Probability significant at 0.01 level.*** Probability significant at 0.001 level.

[(Fig._6)TD$FIG]

Fig. 6. Daily soil CO2 fluxes and water-filled pore space (WFPS, 0–30 cm) for the four treatments during the sorghum season in Kingaroy (Queensland, Australia). Arrowsindicate the timing of N fertiliser applications.

M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39 35

subsequent cropping seasons. This is the first study to investigatethe role of these factors in two subtropical ley pasture–cereal croprotations using a fully automatedmeasuring systemproviding hightemporal resolution data on N2O emissions.

In a review on the role of legumes in mitigating greenhouse gasemissions from agriculture, Jensen et al. (2012) concluded there isa risk of elevated N2O emissions after the incorporation of legumeresidues. However, many of the cited studies referred to theincorporation of grain legume residues, while little is reportedabout incorporation of legume pasture residues. Legume pastureresidues have lower C:N ratio than the senesced vegetative stubblethat typically remains after grain harvest of legume crops (Kumarand Goh, 1999; Fillery, 2001; Peoples et al., 2009). The decompo-sition of legume pasture residues can therefore provide a source ofeasily decomposable N, leading to higher N2O losses compared tograin legume residues.

The outcomes of the few studies conducted on incorporation oflegume pasture residues are however contradictory. Extremelyhigh N2O emissions following the plough-down of an alfalfapasture were measured in winter and early spring in Canada byWagner-Riddle et al. (1997), who reported a total loss of 5.38kgN2O-Nha�1 over a seven-month period. Significantly, in the sameexperiment high emissions (2.84 kg N2O-Nha�1) were observedalso in the control treatment, where no incorporation of alfalfaresidues was implemented. Similar amounts of N2O emissionswere reported in another Canadian study by Wagner-Riddle andThurtell (1998), where 2.63 and 3.79 kg N2O-Nha�1 were emittedduring freezing and thawing periods by a bare soil and a soil afterthe incorporation of alfalfa residues, respectively. Substantial N2Oemission rates (up to 45g N2O-N ha�1 day�1) lasted for over twomonths after the plough-down of a clover pasture in early spring inScotland (Pappa et al., 2011).

In contrast, limited N2O emissions following the incorporationof legume pastures were reported in temperate climates byBrozyna et al. (2013) (approximately 0.25 kg N2O-Nha�1) andBaggs et al. (2000) (0.24kg N2O-Nha�1) and, in a subtropical aclimate, by Gomes et al. (2009) (0.36 kg N2O-Nha�1). UnderMediterranean climatic conditions Sanz-Cobena et al. (2014)observed no significant differences in N2O emissions from maizefollowing a vetch pasture (Vicia villosa) or a fallow period.

In the present study, N2O losses from the unfertilised sorghumfollowing the legume ley pasture (L0)were similar to those reportedby Brozyna et al. (2013) and Baggs et al. (2000), and did not differsignificantly from those measured in the unfertilised sorghumfollowing the grass ley pasture (G0). These findings indicate that theincorporation of legume residues per se is not sufficient to triggerelevated N2O emissions. Conversely, enhanced N2O emissionsappear to be the product of several concurrent factors.

4.2. Factors influencing N2 emissions and yields

In this study cumulative N2O emissions were primarily afunction of the N fertiliser rate applied, while cropping history hadno significant effect. On the other hand, crop biomass and grain

production showed a clear response to increased N availability inthe legume history, irrespective of the N fertiliser input. Whileincorporation of legume residues provided an additional 20–22kgNha�1 to the sorghum crop in both L0 and L70 compared to G0 andG100, this additional N release did not significantly enhance thenitrification or denitrification processed compared to the incorpo-ration of grass residues. Moreover, the temporal pattern of dailyN2O emissions was substantially affected by the cropping historyand in the first 10 weeks after sorghum establishment N2Oemissions pulses in G0 and G100 were significantly higher than inL0 and L70, respectively.

We propose that this apparent contradiction can be explainedby considering three interacting factors: the fertiliser N rateapplied, the cropping history and the synchrony between soil Navailability and crop N uptake.

4.2.1. Fertilisation ratesThe application of synthetic N fertiliser was the main factor

responsible for enhancedN2Oemissions and thehighest cumulativeN2O losses were measured in the two fertilised treatments. InL70 and G100 the highest N2O emission rates were observed afterside dressing, when the majority of the N fertiliser was applied andWFPS exceeded 60% (Fig. 2). The abrupt emission pulses afterfertilisation canbeexplainedby considering the release dynamics ofsynthetic fertilisers (Crews and Peoples, 2005). Under humid soilconditionsurea is rapidly hydrolysed, leading to a fast releaseofhighamounts of mineral N in the soil. At side dressing sorghum plantswere still at an early stage of physiological development and wereable to take up only a fraction of the mineral N applied. ThereforesignificantamountsofmineralNaccumulated inthesoil andbecameavailable to nitrifying and denitrifying microorganisms. Indeed,sufficient surplusNmusthave still beenpresent in thehighestN ratetreatment (G100), where a secondary emission pulse was observedbetween mid-February and mid-March, one month after sidedressing.

Increased N2O emissions following the fertilisation events inthis study did not correspond to elevated soil mineral N contents inthe first 30 cm, due probably to the large rainfall event that fell overthe trial shortly after side dressing. After this rain event a largefraction of the applied N is likely to have leached deeper into thesoil profile, leaving little mineral N in the first 30 cm to be detectedat the following sampling events (Fig. 3).

Overall, N2O emissions displayed a significant correlation withN fertilisation, rising in nonlinear patterns at increasing N fertiliserrates (0, 70 and 100kg Nha�1). As observed by several authors(McSwiney and Robertson, 2005; Hoben et al., 2011; Shcherbaket al., 2014), the fast release of mineral N after fertilisation canexceed the plant uptake capability when N fertiliser is applied athigh rates and the resulting temporary surplus of mineral N canpromote elevated nitrification and denitrification rates if theappropriate soil water conditions are met. These findings indicatethat the best fertiliser management practices aimed at reducingN2O losses coincide with those designed to achieve high levels ofagronomic efficiency. N rates and fertilisation techniques should

Table 5Sorghum grain yield (expressed as dry weight), above ground biomass (expressed as dry weight), harvest index, total N uptake (mean� SE, n =3) and recovery efficiency offertiliser N in the crop (REfN) as a function of the four treatments. Means denoted by a different letter indicate significant differences between treatments (p<0.05).

Measurement Treatment

L0 L70 G0 G100

Grain yield (t ha�1) 2.52�0.22 b 5.29�0.22 c 0.94�0.12 a 5.20�0.11 c

Above ground biomass (t ha�1) 8.38�0.78 b 14.96�0.73 c 4.48�0.34 a 13.68�0.53 c

Harvest index 0.31�0.02 b 0.36�0.01 bc 0.21�0.03 a 0.38�0.01 c

Total N uptake (kg Nha�1) 46.91�4.77 118.94�7.98 25.11�1.53 98.52�6.88REfN (%) – 70.9�2.1 52.8�6.1*

*Probability significant at 0.01 level.

36 M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39

therefore be aimed at maximizing the crop uptake of appliedsynthetic N.

4.2.2. Cropping historyAlthough it did not have a statistically significant effect on

cumulative N2O emissions, the cropping history substantiallyinfluenced the temporal and spatial patterns of N2O fluxes. In thefirst part of the season for example, daily N2O fluxes in G100 andG0 tended to rise immediately after rain events and on theseoccasions emissions were consistently higher than in L70 and L0,respectively (Fig. 4a, b). Moreover, the grass cropping historytreatments constituted a more diffused source of N2O emissions,with G100 and G0 displaying high N2O losses also from the inter-row, while N2O fluxes from the inter-row in L70 and L0 neverexceeded 12g N ha�1 day�1 (Fig. 5).

It is here hypothesized that the sharp increases of N2Oemissions following increments in WFPS, as well as the highfluxes measured across the whole field in G100 and G0, are to beattributed to higher labile C in the soil of the grass cropping history.Enhanced N2O emissions from soils with high DOC contents havebeen reported by numerous studies (Elmi et al., 2003; Yao et al.,2009; Barton et al., 2011). This positive correlation originates fromthe coupled biogeochemical cycles of C and N. The degradation ofplant material provides soil microbes with substantial amounts ofC, which under anaerobic conditions is oxidized by denitrifyingmicroorganisms via reducing NO3

� to N2O (Conrad, 1996).TheDOCvalues observedat the beginningof the sorghumseason

tended to be higher in the treatments following the termination ofthegrasspasture,whereasubstantiallyhigheramountoffinefibrousplant residues was present at sorghum planting. Whilst notquantitatively documented, the presence of undecomposed rootsandcrowns in thegrasshistorywouldhavecontinuedtosupplementthe labileC pool. Thiswould haveprovidedauniformly distributedCsource to support microbial activity and therefore the potential fordenitrification in both row and inter-row areas (Fig. 5). While thisenhanced potential microbial activity in G0 became increasingly N-limited by the end of December, the provision of the N side dressingin G100 allowed that activity to continue.When combinedwith theverywet soil conditions, thishighnutrientavailability resulted in thesignificantN2O emissionspulses observed inG100 from late January(Fig. 4).

Conversely, the low C:N ratio of the sulla plants residues wouldlikely have contributed to a more rapid degradation of residues inthe L0 and L70 treatments, leaving less labile C to supportcontinued microbial activity. This may have changed during theseason in the vicinity of the sorghum rows, where increasing rootdensity would have contributed to labile C stores. The markedcontrast in N2O emissions between the rows and inter-rows in bothL0 and L70 (Fig. 5) are consistent with this hypothesis. Similarresults to this study were reported by Sanz-Cobena et al. (2014),who observed higher N2O emissions from maize after theincorporation of barley compared to the same crop after theincorporation of a vetch pasture.

In contrast withN2O emissions and DOC values at the beginningof the experiment, CO2 emissions did not show significantdifferences between treatments in terms of cumulative emissionsor temporal patterns. The two reasons were probably the low soilmoisture content at the beginning of the sorghum season and therelatively high respiration rate of sorghum roots in the later stagesof the experiment. The early stages of the sorghum season(December 2012–mid-January 2013) experienced relatively dryconditions, with dry surface soil probably limiting microbialactivity. From mid-January onward the trial received a substantialamount of rain, but by that stage the roots of the sorghum plantswould have expanded into the inter-row, possibly overriding therelatively low rates of background soil microbial respiration.

Although the relatively low labile C content of soils with thelegume cropping history would have limited the activity of the soilmicrobial pool, the legumes were able to provide a significantenhancement of the soil labile N pool. While the initial mineral Ncontents in the top 30 cm of the profile were quantitatively similarbetween the legume and grass histories (Fig. 3), there was a higherproportion of that N in the form of NO3-N in the legume (33%) thanthe grass (20%) histories. This data, combined with the increases inPMN (Table 1), suggest greater mineralisable N reserves in thelegume histories which is also reflected in sorghum growth, grainyield and N accumulation. Moreover, while the N supply from theresidue mineralisation was not substantial in G100 and G0, theheavy rain events in January are likely to have promotedsubstantial N2 losses (Schwenke et al., 2013), further reducingtheN supply to the plants of these treatments. The highN losses vialeaching would have severely limited the efficiency of the sidedressing, resulting therefore in the lower REfN values measured inG100 compared to L70. As a consequence, crop biomass and grainyields in L70 were comparable to those of G100 despite a 30%reduction in fertiliser rate, while yields in L0 were approximatelydouble those in G0.

These findings highlight the importance of the soil labile C poolin regulating N2O losses. Specifically, denitrifying microorganismscan be more competitive than plants in using even small amountsof NO3

� when soil labile C content is sufficiently high to sustainelevated microbial activity in anaerobic conditions. This wasevident in the unfertilised treatments, where dry matter and grainyield in G0 were severely limited by N availability but N2Oemissions were almost identical. Under these circumstances theamount and chemical availability of C supplied by the crop residuecan be more important than the soil NO3

� content in stimulatingdenitrification.

4.2.3. Synchrony of N supplySynchrony is a critical aspect in reducing N2O losses after the

termination of a legume pasture (Crews and Peoples, 2005; Jensenet al., 2012), which means matching the N release resulting fromthe degradation of the legume residues with the N uptake of thesubsequent crop. The high emissions reported after the termina-tion of a legume pasture are oftenmeasuredwhen the field site hasbeen left fallow for long periods (Wagner-Riddle et al., 1997;Wagner-Riddle and Thurtell, 1998; Pappa et al., 2011). Conditionscan be highly conducive for elevated N2O emissions when a soil isleft fallow after the incorporation of fresh legume residues since inthe absence of a crop following the pasture, all the readilymineralisable C from the legume residues becomes available tosupport the denitrification of large amounts of NO3

� accumulatedin the soil.

In this study sorghumwas planted 13 days after the terminationof the pasture phase. During the fallow prior to sorghum plantingonly 6mmof rainfall had fallen, limiting therefore the possibility oforganic matter decomposition, mineral N accumulation orgeneration of significant N2O emissions. Typically, the highest Nmineralisation rates from legume residues are reported to occurafter about six weeks from the termination of the pasture (Foxet al., 1990; Becker and Ladha, 1997; Robertson, 1997; Park et al.,2010). In the present study this would have coincided with themoment of maximum N uptake of sorghum, supplying in this wayapproximately an extra 20–22kg Nha�1 to the plants in L70 and L0.The good synchrony between N release from the legume residuesand N uptake of the sorghum plant is confirmed by the high REfNmeasured in L70 (Table 5).

Overall, planting sorghum shortly after pasture terminationproved an effective strategy to reduce N2O losses due to thedecomposition of legume residues. This practice also resultedsuccessful in supplying an extra source of N to sorghum, increasing

M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39 37

significantly the yields in both the fertilised (L70) and unfertilised(L0) treatments.

5. Implications for managing N2 emissions from a cropfollowing a legume pasture

Introducing a legume ley pasture in a cereal-based croppingsystem enabled the reduction of the synthetic N fertiliser appliedto the following cereal crop and significantly reduced the N2Oemission factor for this crop compared to a grass ley pasture. Theemission factor of L70 (0.63%) was almost half of G100 (1.17%) andwas considerably lower than the 1% recommended by the IPCCmethodology for fertilised cropping systems (De Klein et al., 2006).Both Tier 1 and Tier 2 approaches consider an emission factor of 1%for N derived from the mineralisation of crop residues. Accordingto this method, in L70 the total N2O emissions resulting from themineralisation of approximately 30kg Nha�1 contained in thelegume residues and combined with the application of 70kgNha�1 would have amounted to about 1 kg N2O-Nha�1. Similarly,N2O losses from the two unfertilised treatments should havediffered substantially, resulting in 0.3 and 0.1 kg N2O-Nha�1 fromL0 and G0, respectively. However, the different dynamics observedin this study suggest that the amount of N in the soil per se is notsufficient to correctly estimate N2O emission factors, and that thequantity and availability of soil C should also be considered.

The importance of soil labile C is reinforced when the N2Oemissions intensity (kgN2O-N yield�1 ha�1) of the legume andgrass cropping histories are considered. This measure effectivelyquantifies the efficiency of agronomic practices in maximizinggrain yields while minimizing N2O emissions. Despite the broadrange of grain yields, N2O emissions intensities were consistentamong treatmentswith the same cropping history, with intensitiessignificantly lower in the legume compared to the grass history(Table 3).

The introduction of a legume pasture phase in a cereal-basedcrop rotation seems to offer multiple environmental andagronomic advantages. In the fertilised treatments it resulted ina 50% reduction of the N2O-N emitted compared to introducing agrass pasture, proving to be an effective mitigation strategy toreduce the contribution of cereal cropping systems to greenhousegas emissions. A pasture phase can also contribute to increasingthe soil organic matter, aggregate stability, soil microbial pool andorganic N content (Giller and Cadisch,1995; Rochester et al., 2001),benefiting the overall soil chemical and physical fertility. Theseresults overall indicate that introducing a legume pasture in asubtropical cereal cropping system is a sustainable practice fromboth the environmental and agronomic perspective.

Acknowledgements

The authors would like to thank Gary Harch for his valuablesupport during the monitoring campaign and the QueenslandDepartment of Agriculture, Fisheries and Forestry (QDAFF) forproviding the field site, themachinery and the staff for the farmingoperations. The authors also appreciate the contribution of DrPhilip Moody and Mitchell De Bruyn for their assistance in thechemical analyses. This research was conducted as part of theNational Agricultural Nitrous Oxide Research Program (NANORP)funded by the Federal Department of Agriculture and theAustralian Grains Research and Development Corporation (GRDC).

References

Australian Bureau of Meteorology, http://www.bom.gov.au/ 2015.Baggs, E.M., Rees, R.M., Smith, K.A., Vinten, A.J.A., 2000. Nitrous oxide emission from

soils after incorporating crop residues. Soil Use Manage. 16, 82–87.

Barton, L., Butterbach-Bahl, K., Kiese, R., Murphy, D.V., 2011. Nitrous oxide fluxesfrom a grain–legume crop (narrow-leafed lupin) grown in a semiarid climate.Global Change Biol. 17, 1153–1166.

Barton, L., Kiese, R., Gatter, D., Butterbach-Bahl Klaus, B., Renee Hinz, C., Murphy, D.V., 2008. Nitrous oxide emissions from a cropped soil in a semi-arid climate.Global Change Biol. 14, 177–192.

Becker, M., Ladha, J., 1997. Synchronizing residue N mineralization with rice Ndemand in flooded conditions. Driven by nature: plant litter quality anddecomposition 1997.

Bell, M., Harch, G., Moody, P., 2012. Diversification from cropping into mixed crop-livestock systems – the sustainability risks posed by hay removal from pastureor forage blocks. Red 18, 75.

Birch, H.F., 1958. The effect of soil drying on humus decomposition and nitrogenavailability. Plant Soil 10, 9–31.

Boox, G.E.P., Pierce, D.A., 1970. Distribution of residual autocorrelations inautoregressive-integrated moving average time series models. J. Am. Stat.Assoc. 65, 1509–1526.

Bremner, J.M.,1965. Nitrogen availability indexes. In: Norman, A.G. (Ed.), Methods ofSoil Analysis. Part 2. Chemical andMicrobiological Properties. American Societyof Agronomy, Soil Science Society of America, pp. 1324–1345.

Brozyna, M.A., Petersen, S.O., Chirinda, N., Olesen, J.E., 2013. Effects of grass-clovermanagement and cover crops on nitrogen cycling and nitrous oxide emissionsin a stockless organic crop rotation. Agric. Ecosyst. Environ. 181, 115–126.

Carter, M.R., Gregorich, E.G., 2007. Soil Sampling and Methods of Analysis, Seconded. CRC Press.

Conrad, R., 1996. Soil microorganisms as controllers of atmospheric trace gases (H2,CO, CH4, OCS, N2O, and NO). Microbiol. Rev. 60, 609–640.

Crews, T.E., Peoples, M.B., 2004. Legume versus fertilizer sources of nitrogen:ecological tradeoffs and human needs. Agric. Ecosyst. Environ. 102, 279–297.

Crews, T.E., Peoples, M.B., 2005. Can the synchrony of nitrogen supply and cropdemand be improved in legume and fertilizer-based agroecosystems? A review.Nutr. Cycl. Agroecosyst. 72, 101–120.

De Antoni Migliorati, M., Bell, M., Grace, P., Rowlings, D., Scheer, C., Strazzabosco, A.,2014a. Assessing agronomic and environmental implications of different Nfertilisation strategies in subtropical grain cropping systems on Oxisols. Nutr.Cycl. Agroecosyst. 100, 369–382.

De Antoni Migliorati, M., Scheer, C., Grace, P.R., Rowlings, D.W., Bell, M., McGree, J.,2014b. Influence of different nitrogen rates and DMPP nitrification inhibitor onannual N2O emissions from a subtropical wheat–maize cropping system.Agriculture. Ecosyst. Environ. 186, 33–43.

De Klein, C., Novoa, R.S.A., Ogle, S., Smith, K.A., Rochette, P., Wirth, T.W., 2006. N2Oemissions from managed soils, and CO2 emissions from lime and ureaapplication. In: Eggleston, H.S., Buendia, L., Miwa, K., Ngara, T., Tanabe, K. (Eds.),IPCC Guidelines for National Greenhouse Gas Inventories, Volume 4,Agriculture, Forestry and Other Land Use. IGES, Japan.

Drury, C.F., Yang, X.M., Reynolds, W.D., McLaughlin, N.B., 2008. Nitrous oxide andcarbon dioxide emissions from monoculture and rotational cropping of corn,soybean and winter wheat. Can. J. Soil Sci. 88, 163–174.

Elmi, A.A., Madramootoo, C., Hamel, C., Liu, A., 2003. Denitrification and nitrousoxide to nitrous oxide plus dinitrogen ratios in the soil profile under three tillagesystems. Biol. Fertil. Soils 38, 340–348.

World reference base for soil resources. World soil resources reports 84, 21–22.FAO, 2009. How to feed the World in 2050. Proceedings of the Expert Meeting on

How to Feed the World in 2050, 24–26 FAO Headquarters, Rome June 2009.Fillery, I.R.P., 2001. The fate of biologically fixed nitrogen in legume-based dryland

farming systems: a review. Anim. Prod. Sci. 41, 361–381.Fox, R.H., Myers, R.J.K., Vallis, I., 1990. The nitrogen mineralization rate of legume

residues in soil as influenced by their polyphenol, lignin, and nitrogen contents.Plant Soil 129, 251–259.

Giller, K.E., Cadisch, G., 1995. Future benefits from biological nitrogen fixation: anecological approach to agriculture. Plant Soil 174, 255–277.

Gomes, J., Bayer, C., de Souza Costa, F., de Cássia Piccolo, M., Zanatta, J.A., Vieira, F.C.B., Six, J., 2009. Soil nitrous oxide emissions in long-term cover crops-basedrotations under subtropical climate. Soil Till. Res. 106, 36–44.

Granli, T., Bøckman, O.C., 1995. Nitrous oxide (N2O) emissions from soils in warmclimates. Nutr. Cycl. Agroecosyst. 42, 159–163.

Halvorson, A.D., Del Grosso, S.J., Reule, C.A., 2008. Nitrogen, tillage, and crop rotationeffects on nitrous oxide emissions from irrigated cropping systems. J. Environ.Qual. 37, 1337–1344.

Hoben, J.P., Gehl, R.J., Millar, N., Grace, P.R., Robertson, G.P., 2011. Nonlinear nitrousoxide (N2O) response to nitrogen fertilizer in on-farm corn crops of the USMidwest. Global Change Biol. 17, 1140–1152.

Honaker, J., King, G., 2010. What to do about missing values in time-series cross-section data. Am. J. Political Sci. 54, 561–581.

Hu, X.-K., Su, F., Ju, X.-T., Gao, B., Oenema, O., Christie, P., Huang, B.-X., Jiang, R.-F.,Zhang, F.-S., 2013. Greenhouse gas emissions from a wheat–maize doublecropping system with different nitrogen fertilization regimes. Environ. Pollut.176, 198–207.

Jensen, E., Peoples, M., Boddey, R., Gresshoff, P., Hauggaard-Nielsen, H., Alves, B.,Morrison, M., 2012. Legumes for mitigation of climate change and the provisionof feedstock for biofuels and biorefineries. A Rev. Agron. Sustain. Dev. 32,329–364.

Kearney, J., 2010. Food consumption trends and drivers. Philos. Trans. R. Soc. B: Biol.Sci. 365, 2793–2807.

Kroeze, C., Mosier, A.R., Nevison, C., Oenema, O., Seitzinger, S., van Cleemput, O.,1997. Agriculture .

38 M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39

Kusa, K., Hu, R., Sawamoto, T., Hatano, R., 2006. Three years of nitrous oxide andnitric oxide emissions from silandic andosols cultivated with maize inHokkaido, Japan. Soil Sci. Plant Nutr. 52, 103–113.

McSwiney, C.P., Robertson, G.P., 2005. Nonlinear response of N2Oflux to incrementalfertilizer addition in a continuous maize (Zea mays L.) cropping system. GlobalChange Biol. 11, 1712–1719.

Mosier, A.R., Wassmann, R., Verchot, L., King, J., Palm, C., 2004. Methane andnitrogen oxide fluxes in tropical agricultural soils: sources sinks andmechanisms. Environ. Dev. Sustain. 6, 11–49.

Myhre, G., Shindell, D., Bréon, F.-M., Collins, W., Fuglestvedt, J., Huang, J., Koch, D.,Lamarque, J.-F., Lee, D., Mendoza, B., Nakajima, T., Robock, A., Stephens, G.T.T.,Zhang, H., 2013. Anthropogenic and natural radiative forcing. Climate Change2013. In: Stocker, T.F., Qin, T.F.D., Plattner, G.-K., Tignor, M., Allen, S.K., Boschung,J., Nauels, A., Xia, Y., Bex, V., Midgley, P.M. (Eds.), The Physical Science Basis.Contribution of Working Group I to the Fifth Assessment Report of theIntergovernmental Panel on Climate Change. Cambridge University Press,Cambridge, United Kingdom and New York, NY, USA.

Pappa, V.A., Rees, R.M., Walker, R.L., Baddeley, J.A., Watson, C.A., 2011. Nitrous oxideemissions and nitrate leaching in an arable rotation resulting from the presenceof an intercrop. Agriculture. Ecosyst. Environ. 141, 153–161.

Park, S.E., Webster, T.J., Horan, H.L., James, A.T., Thorburn, P.J., 2010. A legumerotation crop lessens the need for nitrogen fertiliser throughout the sugarcanecropping cycle. Field Crops Res. 119, 331–341.

Parkin, T.B., Kaspar, T.C., 2006. Nitrous oxide emissions from corn–soybean systemsin the Midwest. J. Environ. Qual. 35, 1496–1506.

Parkin, T.B., Venterea, R.T., Hargreaves, S.K., 2012. Calculating the detection limits ofchamber-based soil greenhouse gas flux measurements. J. Environ. Qual. 41,705–715.

Peoples, M.B., Hauggaard-Nielsen, H., Jensen, E.S., 2009. The potentialenvironmental benefits and risks derived from legumes in rotations. Agron.Monogr. 52, 349–385.

Ravishankara, A., Daniel, J.S., Portmann, R.W., 2009. Nitrous oxide (N2O): thedominant ozone-depleting substance emitted in the 21st century. Science 326,123–125.

Robertson, G.P., 1997. Nitrogen use efficiency in row-crop agriculture: crop nitrogenuse and soil nitrogen loss. Ecol. Agric. 347–365.

Robertson, G.P., Paul, E.A., Harwood, R.R., 2000. Greenhouse gases in intensiveagriculture: contributions of individual gases to the radiative forcing of theatmosphere. Science 289, 1922–1925.

Rochester, I.J., Peoples, M.B., Hulugalle, N.R., Gault, R.R., Constable, G.A., 2001. Usinglegumes to enhance nitrogen fertility and improve soil condition in cottoncropping systems. Field Crops Res. 70, 27–41.

Rochette, P., Angers, D.A., Bélanger, G., Chantigny, M.H., Prévost, D., Lévesque, G.,2004. Emissions of N2O from alfalfa and soybean crops in Eastern Canada. SoilSci. Soc. Am. J. 68, 493–506.

Sanz-Cobena, A., García-Marco, S., Quemada, M., Gabriel, J.L., Almendros, P., Vallejo,A., 2014. Do cover crops enhance N2O, CO2 or CH4 emissions from soil inMediterranean arable systems? Sci. Total Environ. 466-467, 164–174.

Scaglia, B., Adani, F., 2009. Biodegradability of soil water soluble organic carbonextracted from seven different soils. J. Environ. Sci. 21, 641–646.

Scheer, C., Grace, P.R., Rowlings, D.W., Payero, J., 2013. Soil N2O and CO2 emissionsfrom cotton in Australia under varying irrigation management. Nutr. Cycl.Agroecosyst. 95, 43–56.

Schwenke, G., Grace, P.R., Bell, M.J., 2013. Nitrogen Use Efficiency. Timing andApplication Method to Minimise Loss and Optimise N Use Efficiency. NewResearch on Nitrous Oxide Emissions, Volatilisation and Denitrification.Australian Grains Research & Development Corporation Grower GrainsResearch Update.

Schwenke, G., Haigh, B., McMullen, G., Herridge, D.F., 2010. Soil nitrous oxideemissions under dryland N-fertilised canola and N2-fixing chickpea in thenorthern grains region, Australia. Proc.19thWorld Congress of Soil Science, SoilSolutions for a Changing World, Brisbane, Australia, pp. 228–231.

Sentek Technologies, 2011. Calibration manual for Sentek soil moisture sensorsVersion 2.0.

Shcherbak, I., Millar, N., Robertson, G.P., 2014. Global metaanalysis of the nonlinearresponse of soil nitrous oxide (N2O) emissions to fertilizer nitrogen. Proc. Natl.Acad. Sci..

Skiba, U., Fowler, D., Smith, K.A., 1997. Nitric oxide emissions from agricultural soilsin temperate and tropical climates: Sources, controls and mitigation options.Nutr. Cycl. Agroecosyst. 48, 139–153.

Smith, P., Martino, D., Cai, Z., Gwary, D., Janzen, H., Kumar, P., McCarl, B., Ogle, S.,O'Mara, F., Rice, C., Scholes, B., Sirotenko, O., 2007. Agriculture. In: Metz, B.,Davidson, O.R., Bosch, P.R., Dave, R., Meyer, L.A. (Eds.), Climate Change 2007:Mitigation. Contribution of Working Group III to the Fourth Assessment Reportof the Intergovernmental Panel on Climate Change. Cambridge University Press,Cambridge, United Kingdom and New York, NY, USA.

Tilman, D., Cassman, K.G., Matson, P.A., Naylor, R., Polasky, S., 2002. Agriculturalsustainability and intensive production practices. Nature 418, 671–677.

UNFPA, 2011. State of the World Population 2011. United Nations Population Fund,New York, New York, USA.

Wagner-Riddle, C., Thurtell, G.W., 1998. Nitrous oxide emissions from agriculturalfields duringwinter and spring thawas affected bymanagement practices. Nutr.Cycl. Agroecosyst. 52, 151–163.

Wagner-Riddle, C., Thurtell, G.W., Kidd, G.K., Beauchamp, E.G., Sweetman, R., 1997.Estimates of nitrous oxide emissions from agricultural fields over 28 months.Can. J. Soil Sci. 77, 135–144.

Yao, Z., Zheng, X., Xie, B., Mei, B., Wang, R., Butterbach-Bahl, K., Zhu, J., Yin, R., 2009.Tillage and crop residuemanagement significantly affects N-trace gas emissionsduring the non-rice season of a subtropical rice–wheat rotation. Soil Biol.Biochem. 41, 2131–2140.

M. De Antoni Migliorati et al. / Agriculture, Ecosystems and Environment 204 (2015) 27–39 39