journa l homepage: www.e lsev ier .com/ locate /e ja
Modeling nitrous oxide emissions from organic and conventionalcereal-based cropping systems under different management, soil andclimate factors
Jordi Doltraa,∗, Jørgen E. Olesenb, Dolores Báezc, Aránzazu Louroc,Ngonidzashe Chirindad
a Cantabrian Agricultural Research and Training Centre, CIFA, c/Héroes 2 de Mayo 27, 39600 Muriedas, Cantabria, Spainb Aarhus University, Department of Agroecology and Environment, Blichers Allé 20, P.O. Box 50, DK-8830 Tjele, Denmarkc Centro de Investigaciones Agrarias de Mabegondo, Apdo. 10, 15080 A Coruna, Spaind International Centre for Tropical Agriculture, CIAT, Apartado Aéreo 6713 Cali, Colombia
a r t i c l e i n f o
Article history:Received 19 August 2014Received in revised form 29 January 2015Accepted 3 February 2015Available online 18 February 2015
Keywords:Greenhouse gas emissionsNitrogen lossesFASSET process-based modelMitigationCrop management
a b s t r a c t
Mitigation of greenhouse gas emissions from agriculture should be assessed across cropping systems andagroclimatic regions. In this study, we investigate the ability of the FASSET model to analyze differencesin the magnitude of N2O emissions due to soil, climate and management factors in cereal-based croppingsystems. Forage maize was grown in a conventional dairy system at Mabegondo (NW Spain) and wheatand barley in organic and conventional crop rotations at Foulum (NW Denmark). These two Europeansites represent agricultural areas with high and low to moderate emission levels, respectively. Field trialsincluded plots with and without catch crops that were fertilized with either mineral N fertilizer, cattleslurry, pig slurry or digested manure. Non-fertilized treatments were also included. Measurements ofN2O fluxes during the growing cycle of all the crops at both sites were performed with the static chambermethod with more frequent measurements post-fertilization and biweekly measurements when highfluxes were not expected. All cropping systems were simulated with the FASSET version 2.5 simulationmodel. Cumulative soil seasonal N2O emissions were about ten-fold higher at Mabegondo than at Foulumwhen averaged across systems and treatments (8.99 and 0.71 kg N2O-N ha−1, respectively). The averagesimulated cumulative soil N2O emissions were 9.03 and 1.71 kg N2O-N ha−1 at Mabegondo and at Foulum,respectively. Fertilization, catch crops and cropping systems had lower influence on the seasonal soilN2O fluxes than the environmental factors. Overall, in its current version FASSET reproduced the effectsof the different factors investigated on the cumulative seasonal soil N2O emissions but temporally itoverestimated emissions from nitrification and denitrification on particular days when soil operations,ploughing or fertilization, took place. The errors associated with simulated daily soil N2O fluxes increasedwith the magnitude of the emissions. For resolving causes of differences in simulated and measured fluxesmore intensive and temporally detailed measurements of N2O fluxes and soil C and N dynamics wouldbe needed.
Agricultural practices such as soil tillage, nitrogen (N) fertil-ization and residue management, may significantly influence soilnitrous oxide (N2O) emissions during and after the crop growingseason (Mutegi et al., 2010; Van Groenigen et al., 2010; Brozyna
et al., 2013). Moreover, soil properties and climatic factors affectthe processes responsible for both N2O production and emission.In order to mitigate climate change and the associated impacts, itis particularly important to identify and adopt crop managementpractices that most effectively reduce greenhouse gas (GHG) emis-sions from agricultural soils in cropping systems covering majorareas within each agro-climatic region as well as from croppingsystems that, due to climate and/or soil factors, have a high risk ofGHG emissions (IPCC, 2014).
Lesschen et al. (2011) proposed a correction to the default IPCCemission factors (EF) that considered management and environ-mental variables such as precipitation and soil type for different
J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20 9
regions across Europe. Also Li et al. (2001), based on predic-tions from an agro-ecosystem simulation model, found that soilorganic matter (SOM) content was the main driver of variabilityin GHG emissions from fertilizer applications in China. It is there-fore, essential to have reliable tools that can aid understandingand quantification of differences in N2O emissions among differ-ent N management options under specific agro-climatic regions(Butterbach-Bahl et al., 2013). Such tools would allow us to iden-tify those measures that have the highest N use efficiency, providebetter N cycling and reduce environmental impacts from nitrateleaching to groundwater and surface water and climate impactsfrom N2O emissions.
The ability of biogeochemical models to integrate complex pro-cesses controlling N2O production, consumption and transportmake them important tools for evaluating N2O emissions from dif-ferent cropping systems and assessing emissions from agriculturalsystems at both regional and global scales. Examples of the use ofprocess-based models to assess N2O emissions from different crop-ping systems are Li et al. (2004) with DNDC, Del Grosso et al. (2005)with DAYCENT, Chatskikh et al. (2005) with FASSET and Metay et al.(2011) with NOE.
Chirinda et al. (2011) tested the FASSET model against data fromdifferent organic arable farming systems and concluded that themodel was capable of simulating trends of seasonal soil N2O emis-sions, but it showed deficiencies in modeling SOM turnover thataffected the predictions of heterotrophic soil respiration in systemsthat included catch crops. A reformulation of the function used inthe model to represent soil tillage has been implemented in FAS-SET 2.5. This allows for a more realistic distribution of crop residuesin the soil profile after soil operations (tillage and harrowing) tosimulate a non-homogenous distribution of crop residues.
In our study, we investigate the ability of FASSET 2.5 to analyzedifferences in the magnitude of N2O emissions due to (a) manage-ment factors, including N management in organic and conventionalcropping systems with and without catch crops and (b) environ-mental factors (soil and climate) in two cereal-based croppingsystems representing major agricultural areas in each region. For-age maize in the Atlantic North Spain (Galicia, Asturias, Cantabriaand Basque Country) represents 77% of the area dedicated to thiscrop in the country and 94% if only rainfed forage maize is con-sidered (MAGRAMA, 2012). Winter wheat represents about 44%of the total cereal area in Denmark and together with spring bar-ley they constitute about 30% of the area under organic farming(Plantedirektoratet, 2009). According to Lesschen et al. (2011),the cereal-based cropping systems investigated in this study werelocated in an area with high (Galicia, Spain) and low to moderate(Central Jutland, Denmark) risk of soil N2O emissions.
2. Materials and methods
2.1. Forage maize experiments at Mabegondo (Galicia, Spain)
Field trials with forage maize (Zea mays L.) in a conventionaldairy system were performed for two growing seasons (2009 and2010) on a silty loam soil at Mabegondo (Galicia) in North WestSpain. The site has a South Atlantic climate (Iglesias et al., 2009)with an annual average temperature and precipitation of 13.1 ◦Cand 1101 mm, respectively, for the period 1998–2007. The dailypattern of these two climatic variables during the years studied(Apr 09–Oct10) is shown in Fig. 1. Main site and crop managementdetails are described in Table 1. The maize crop in 2009 was pre-ceded by a fallow period in autumn-winter after a crop sequence oftriticale-pea mixture as winter crop on the previous year and maizein summer. In 2010, maize was preceded by a five-year grazedgrassland which was discontinued in spring, two months beforesowing of the maize crop. Crop residues were always returned to
the soil and managed with conventional tillage. Different fertilizerswere tested: mineral fertilizer, cattle slurry and pig slurry. Also, anon-fertilized treatment was used as control. The target applica-tion rate for the organic fertilizers was 200 kg N ha−1 applied 3–4days before sowing. The same amount was applied to the mineralfertilizer treatment split in two applications: N–P–K 15-15-15 (7%nitrate-N; 7% ammonium-N) at a rate of 125 kg N ha−1 at sowingand urea 46% at a rate of 75 kg N ha−1 for top dressing when theplant was 40 cm tall.
Nitrous oxide fluxes were measured using the closed chambertechnique (Ryden and Rolston, 1983). Two chambers per plot (i.e.,six chambers per treatment) were placed between rows and left inthe same position during the experiment. After the N applications,gas samples were taken three or five times a week for analysis ofN2O. N2O fluxes for dates between samplings were calculated usingthe trapezoidal method (Cardenas et al., 2010; Louro et al., 2013).Cumulative N2O fluxes were calculated by summing daily flux rates.Coinciding with gas samplings, soil samples at 10 cm depth werecollected for the analysis of mineral N contents (NH4-N and NO3-N)and soil moisture. Soil NH4-N and NO3-N were determined col-orimetrically after extracting 100 g of fresh soil with 200 ml 1 MKCl. Soil moisture content was determined gravimetrically afteroven drying the samples at 105 ◦C for 24 h. Porosity was calculatedfrom bulk density (Bd) in each site by assuming a particle densityof 2.65 Mg m−3. Water filled pore space (WFPS) was calculated bydividing the soil moisture content and Bd by the porosity.
2.2. Winter wheat and spring barley in arable cropping systemsat Foulum (Central Jutland, Denmark)
Winter wheat (Triticum aestivum L.) and spring barley (Hordeumvulgare L.) were grown in different organic and conventional arablecrop rotations in 2008 and 2009 on a loamy sand soil at Foulum(Denmark) in Central Jutland. Main soil characteristics, average cli-mate and crop management details are reported in Table 1. Severalcombinations of catch crop, fertilizer and manure managementpractices were selected for our study. These treatments includedthe application of fresh or digested pig slurry, mineral fertilizersand unfertilized control plots, in systems with or without catchcrops. Fertilization was performed in spring for all crops (end ofMarch–mid-June). In the spring barley, grass-clover was under-sown in spring and in the following year the grass-clover was cutand removed in the manure treatment, whereas cuttings were lefton the soil in the treatment without fertilizer application. A detaileddescription of these treatments can be found in Chirinda et al.(2010) and Brozyna et al. (2013). The cereal crops were sown at adepth of 2–4 cm and at a row distance of 12–12.5 cm. Weeds werecontrolled with tine harrowing in the organic systems and withchemical spraying in the conventional ones.
Measurements of soil N2O fluxes during the growing cycle ofall the crops and during the non-growing period were performedusing static chambers described by Chirinda et al. (2010), and with afrequency from every two weeks to few days following fertilizationevents. Each study plot had two replicate chambers (two plots pertreatment). Cumulative fluxes in each treatment were computedby linear interpolation between measurement dates. Every otherweek, concurrent with gas samplings, soil samples were collectedin all the plots to monitor NO3-N and NH4-N contents (0–30 cmdepths). Soil mineral N in the collected samples was determinedaccording to Keeney and Nelson (1982).
2.3. Simulating N2O emissions with FASSET 2.5: modeldescription and equations
The dynamic crop simulation model Farm ASSEsment Tool (FAS-SET) (Olesen et al., 2002; Berntsen et al., 2003) is a deterministic
10J.D
oltraet
al./Europ.J.Agronom
y66
(2015)8–20
Table 1Site description, average climate for the years during which the study was conducted, soil characteristics at 0–25 cm depth and crop management at the two study sites in Spain and Denmark. Sand, silt, clay and SOM (soil organicmatter) in dry soil; Bd, bulk density; P and T, average annual precipitation and temperature; Cs is cattle slurry, Ps pig slurry, Ds digested slurry and Mf mineral fertilization.
Location Mabegondo (Spain) Foulum (Denmark)Coordinates 43◦14′N, 8◦15 W 56◦30′N, 9◦34′E
Crop Maize Winter wheat Spring barley Winter wheat Spring barley Winter wheat Spring barleyCover crop No No No/Yesb Grass-cloverCrop in preceding year Maize–no crop Grassland Potato Winter wheat Potato Winter wheat Potato Winter wheatSowing 22 May 25 May 24 September 22 April 24 September 22 April 26 September 17 AprilCultivar DKC3745 Tommy Mixture (Power,
Simba, Smilla)Tommy Mixture (Power,
Simba, Smilla)Opus Mixture (Power,
Simba, Anakin)Target density (plants m−2) 10 10 400 300 400 300 400 300Tillage 13 April
Irrigation No 27 May (28 mm)08 June (30 mm)05 July (36 mm)
27 May (28 mm)08 June (30 mm)05 July (36 mm)
02 July (32 mm)
Harvest 29 September 29 September 18 August 18 August 14 August 7 AugustCrop residues Added to soil (spring ploughing) Added to soil (spring ploughing)
a Lesschen et al. (2011).b Mixture of N-fixing and non-fixing crops.c Total nitrogen applied in mineral fertilizers or organic manures.
J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20 11
Fig. 1. Daily maximum, minimum, average temperature and precipitation during the study period at Foulum (a and c) and Mabegondo (b and d).
model that dynamically simulates crop growth and yield, as wellas soil N and carbon (C) fluxes in the plant-soil-atmosphere contin-uum. The model has been calibrated and tested for wheat, barleyand other arable crops under the environmental conditions ofDenmark (e.g., Olesen et al., 2002; Berntsen et al., 2006; Doltra et al.,2011, 2014; Chirinda et al., 2011) and it has also been tested in otherenvironments and agroclimatic regions (Palosuo et al., 2011; Rötteret al., 2012; Asseng et al., 2013). GHG emissions have been simu-lated with FASSET in grasslands (Chatskikh et al., 2005) and arablesystems (Chirinda et al., 2011). In this work, we use FASSET 2.5 toevaluate the soil N2O emissions under a range of cereal-based sys-tems as affected by cropping system, N management, soil type andclimate.
Soil N cycling in FASSET includes mineralization from SOMand crop residues, immobilization, ammonia volatilization, gasemissions, nitrification, denitrification and nitrate leaching. Sevenorganic matter pools are represented in the model: slowly andeasily decomposing added organic matter, autochthonous andzymogeneous microbial biomass, soil microbial residue, nativeorganic matter and inert organic matter. The turnover of all poolsfollows a first-order kinetics and it is modified by temperature andsoil water potential (Petersen et al., 2005).
The simulation of soil N2O emissions in the model follows ahole-in-the-pipe approach (Davidson et al., 2000), where N inter-mediate products from nitrification and denitrification are sourcesfor N2O emissions. The algorithms responsible for the quantifica-tion of soil N emissions, including N2 and N2O, are described indetail by Chatskikh et al. (2005). Briefly, nitrification and deni-trification are described in the model by first-order kinetics. Thesimulation of soil N2O fluxes is done in two steps. In a first stepthe model calculates the potential total N2O production (�∗) fromnitrification (�n) and denitrification (�d):
�∗ = knFTQw�n + �d (1)
where kn is a constant, FT is a function of soil temperature and Qw
is WFPS. Nitrification is calculated as a function of the potentialnitrification rate, set to 0.10 d−1 according to Hansen et al. (1990),WFPS, soil temperature and NH4-N concentration. Denitrificationis calculated as a function of SOM, clay content, soil temperatureand NO3-N concentration. In the second step the potential N2Oproduction is divided into N2O and N2 by semi-empirical relationscontrolled by layer depth and gas diffusion. The soil N2O flux (�N2O)is then calculated as:
�N2O = �∗FNT(1 − FQ)FCFD (2)
where FNT is a function of soil temperature, FQ is a function of WFPS,FC is a function that increases with soil clay content, and FD is adecreasing function of soil depth.
In previous studies (Chirinda et al., 2011; Doltra et al., 2011) itwas found that the rapid release of N from N-rich plant materialsadded to the soil was not always captured by the model and thisaffected the performance of the simulation of N2O fluxes after addi-tion of green manures and catch crops. A new approach to representsoil tillage aiming for a more realistic SOM distribution throughthe soil profile following soil operations has been implemented inFASSET 2.5. Essentially, this new function represents an unequalredistribution of added organic materials as given by tillage depth,while the original formulation assumed a homogenous distributionapproach. In the new model, the crop residues on the soil surfaceare mixed within the lower half of the plough layer.
12 J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20
Table 2Values for the main soil and crop input parameters used in FASSET for modeling N2O emissions: �sat, �fc and �pwp are water content at saturation, at field capacity and atpermanent wilting point, respectively; kfc is hydraulic conductivity at field capacity; Ts and Tb are the sum of temperatures and the base temperature in each crop phase,respectively, where the subscript indicate 0 the phase from sowing to emergence; 1 from emergence to anthesis; 2 from anthesis to the end of grain filling, and 3 from theend of grain filling to ripeness; � is maximum radiation use efficiency; k is extinction coefficient; LAI DM is maximum ratio between LAI and dry matter of the vegetativeabove-ground biomass; LAI N is maximum ratio between LAI and nitrogen of the vegetative above-ground biomass; Nstorage is minimum and maximum nitrogen concentrationof the storage organs; fill factor is net production fraction after anthesis that goes into grain, and FixN DM is maximum nitrogen fixation per produced dry matter.
Main soil input data, required by FASSET and used in the simu-lations of this study included measured soil properties such as soiltexture, SOM, organic N and Bd reported in Table 1, as well as waterretention properties (i.e., soil water contents at different soil watertensions) and soil hydraulic conductivity that were measured atFoulum or obtained by pedo-transfer functions (Saxton and Rawls,2006) at Mabegondo (Table 2). The model was not calibrated forsoil water, C and N dynamics, and thus, the same default modelcoefficients as well as soil parameters other than soil input datawere used for all treatments.
The main crop input parameters used in the simulations affect-ing grain yield and N uptake are reported in Table 2. This set ofparameters includes those related to the crop phenology, radiationuse efficiency, canopy development and grain N storage. At Foulumthe parameter values were those reported by Doltra et al. (2011)that, for winter wheat, were derived from previous field experi-ments. For spring barley, grass-clover and catch crops the valueswere obtained from previous studies in Denmark or from the lit-erature as explained in Doltra et al. (2011). The rest of the cropparameters at Foulum were default model values. At Mabegondo,the phenology parameters were adjusted to observations of devel-opment stages from several on-going and previous trials withsilage maize in Northern Spain (unpublished data; Salcedo, 2011).Maximum and minimum N in storage organs were adjusted tomeasurements in the same on-going field experiments. Maximumradiation use efficiency, based on photosynthetically active radia-tion, was estimated from reported literature (Sinclair and Muchow,1999), while default model values were taken for all the other maizeparameters in Table 2. The simulation of the maize crop was vali-dated with two end-of-season variables. Aboveground dry matterobservations were taken from 33 observations and abovegroundN yield from 12 observations from different plots and seasons attwo locations in North Spain (Salcedo, 2011; García et al., 2012).The average mean error of the simulation of both variables waslower than the average standard deviation of the measurements,
thus, indicating positive model efficiency (see statistical analysessection).
2.5. Model initialization
The FASSET model requires daily maximum and minimum tem-perature, precipitation, potential evapotranspiration and globalradiation. These climatic data were obtained from meteorologicalstations located close to each experimental site in Denmark andSpain. There were no available measurements of the C distributionamong the different soil organic pools at any of the sites. In thesimulations it was assumed that in the topsoil an initial fraction of95% of organic C was allocated to the soil organic matter contentand the remaining 5% being initially allocated to the added organicmaterials and soil microbial pools. The initialization of model runswas done by simulating six or seven years prior to the start of thetreatment at the two locations using available management infor-mation at the plot level. However, in the case of the 2009 maizeexperiments, only information from two years of previous field dataat the plot level was available.
2.6. Statistical analyses
The performance of FASSET to predict seasonal soil N2O fluxesat a daily time step was assessed using four different statisticalindices. The index of agreement d is a descriptive and boundedindicator of the relative size of the differences between model pre-dictions and observations (Willmott, 1982):
d = 1 −
n∑i
(Oi − Mi)2
n∑i
(|O′i| − |M′
i|)
2(3)
where Oi and Mi are the observed and model simulated valuesand |O′
i| and |M′
i| are the differences with respect to the average
J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20 13
observation, of the observed and model simulated values, respec-tively. A value of 1 indicates that the model is highly accurate,while the lowest accuracy level is represented by 0. The root meansquare error RMSE reflects the differences between observationsand predictions, and the mean bias error MBE shows the systematicdeviations. A negative MBE is an indication of model underestima-tion and, if it is positive, overestimation. The equations proposedby Willmott (1982) were used to calculate these two statistics:
RMSE =
√√√√1n
n∑i
((Oi − Mi)2) (4)
MBE =
n∑i
(Mi − Oi)
n(5)
The accuracy of predictions was also calculated with the modelefficiency EF following Nash and Sutcliffe (1970):
EF = 1 −
n∑i
(Mi − Oi)2
n∑i
(Oi − O)2
(6)
This parameter ranges from −1 to 1, with values lower than zeroindicating that the mean of the measured values is a better esti-mator than model-predictions. The quality of predictions increaseswhen approaching 1.
An ANOVA test of significance (P < 0.01) was performed for thestatistical analysis of the relationship between soil N2O fluxes andNO3-N content, with a log transformation of variables avoiding
negative values for N2O. The same test was applied to the rela-tionships between the statistical indices of model performance andthe average daily soil N2O flux (double-log transformed) in eachtreatment.
3. Results
3.1. Observed and simulated cumulative seasonal N2O fluxes
The observed and simulated cumulative soil N2O emissions inthe different cereal-based cropping systems are shown in Table 3.The cumulative fluxes are from sowing to harvest in maize, whilefor winter wheat and spring barley they may comprise pre-sowingand/or post-harvest dates depending on the systems and years.Cumulative soil seasonal emissions were about ten-fold higher forthe silty loam soil in North Spain in comparison with the loamysand soil in Denmark when averaged across systems and treat-ments (8.99 and 0.71 kg N2O-N ha−1, respectively). The differencesin magnitude of the emissions between the sites were well capturedby FASSET (averages of 9.03 and 1.71 kg N2O-N ha−1 at Mabegondoand Foulum, respectively, Table 3), but the absolute deviationsbetween observations and simulation fluxes were much higher atMabegondo than at Foulum for each particular treatment and year.However, if averages of the two years are considered for each fer-tilization treatment at Mabegondo (with simulations ranging from5.3 to 16.7% within the average cumulative measurements in thedifferent treatments), there was a reasonable agreement betweenobserved and simulated cumulative N2O emissions in both loca-tions as shown in Table 3. These results are reflected in the overallgood d and EF indexes but also in the high RMSE of the cumulativeN2O emissions simulations across systems.
In the forage maize systems, N2O emissions from fertilized treat-ments were higher (about 46%) than from the treatment with no
Table 3Observed and model simulated cumulative N2O fluxes (kg N2O-N ha−1) during the growing cycle of the different crops and crop rotations together with the statisticalperformance of the seasonal cumulative simulated emissions including 21 individual systems: index of agreement (d) root mean square error (RMSE), mean bias error (MBE),and model efficiency (EF). Units for RMSE and MBE are kg ha−1. The range of observations represents two and three replicates at Foulum and Mabegondo, respectively.
Location System Crop Year Fertility management Measured Range Simulated
Mab (SP) Dairy Maize 2009 Unfertilized 8.07 6.70–9.02 3.59Dairy Maize 2010 Unfertilized 5.24 5.10–5.34 7.51Dairy Maize 2009 Mineral 11.7 10.10–13.76 8.44Dairy Maize 2010 Mineral 7.83 7.48–8.13 13.74Dairy Maize 2009 Cattle slurry 10.91 10.33–12.03 6.95Dairy Maize 2010 Cattle slurry 8.76 8.49–9.07 11.53Dairy Maize 2009 Pig slurry 10.75 10.64–10.85 6.87Dairy Maize 2010 Pig slurry 8.66 7.89–9.42 13.56Dairy Maize Average Unfertilized 6.66 5.55Dairy Maize Average Mineral 9.77 11.09Dairy Maize Average Cattle slurry 9.84 9.24Dairy Maize Average Pig slurry 9.71 10.22Dairy Maize Average Average of all dairy systems 8.99 9.03
Model performance statisticsd 0.88MBE 0.65RMSE 2.75EF 0.57
14 J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20
Fig. 2. Observed and model simulated soil NO3-N and NH4-N contents in the 0–30 cm soil layer at Foulum during the growing period of winter wheat in three organiccropping systems, digested pig slurry manure and catch crop (a and b), unmanured with catch crop (c and d) and pig slurry without catch crop (e and f), and one conventionaltreatment with mineral fertilizer and without catch crop (g and h). Bars indicate standard error (n = 2).
fertilization, but no clear effect of fertilizer type was observed. Thispattern was consistently reproduced by the model simulations,showing an increase of 53% on average with respect to observedN2O emissions in the unfertilized plots (Table 3). The application ofdigested pig slurry in spring barley and winter wheat crops in thesystems with catch crops did not consistently affect the measuredcumulative soil N2O emissions in comparison with the same unfer-tilized systems. Model simulations, however, showed an increaseof N2O emissions in the system fertilized with digested pig slurryin comparison with the same unfertilized system. Mineral fertil-ization compared to pig slurry increased observed and simulatedsoil N2O emissions in winter wheat (33% and 12%, respectively)
and spring barley (59 and 66%, respectively) systems without catchcrop. The inclusion of a catch crop in winter wheat fertilized withpig slurry showed a small increase of soil N2O emissions (ca 20%)that was also captured although magnified by the model (ca 50%).
3.2. Soil NO3-N and NH4-N dynamics
The performance of FASSET to simulated topsoil mineralN dynamics in different cropping systems at the Foulum andMabegondo sites are shown in Figs. 2 and 3, respectively. The over-all pattern at Foulum shows low levels of soil mineral N in all thesystems. In spite of a general underestimation of soil mineral N
J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20 15
Fig. 3. Observed and simulated soil NO3-N and NH4-N contents in the top 0–10 cm soil layer during the growing period of maize in 2010 at Mabegondo in the cattle slurry(a and b) and unfertilized (c and d) treatments. Bars indicate standard error (n = 3).
peaks observed following the application of manure and fertilizer,soil mineral N was adequately simulated by the model at this site.While the NH4-N dynamics pattern at Mabegondo was similar tothe one at Foulum, that of NO3-N was completely different as shownas an example for the cattle slurry and unfertilized treatments in2010 (Fig. 3). Soil NO3-N contents in the topsoil at Mabegondo weremuch higher than those at Foulum (6–88 kg N ha−1 in the soil top10 cm) and tended to increase during the maize growing season inboth manured and unfertilized treatments (Fig. 3a and c). This is anindication of very high mineralization rates in the Mabegondo soil,which the FASSET model was not able to reproduce. However, theslightly higher soil mineral N observed in the cattle slurry treatmentwas simulated by the model.
Unlike at Foulum, daily soil N2O fluxes at Mabegondo were notlinearly related to daily soil NO3-N content (Fig. 4). Contrary toNO3-N, the topsoil NH4-N evolution during the growing season was
Fig. 4. Linear regression between log transformed daily observed values of soil N2Ofluxes and NO3-N content in the 0–30 cm and 0–10 soil layers at Foulum (n = 81) andat Mabegondo (n = 56), respectively. *P < 0.01.
well simulated in both treatments (Fig. 3b and d). To better under-stand the model performance of soil N dynamics at Mabegondo,WFPS estimated from soil physical properties and moisture con-tent and simulated by the model is shown in Fig. 5. The initialunderestimation of WFPS, when soil water content was close to thesaturation level, could result in an overestimation of water drainageand, hence, of downward movement of soil NO3-N from the top-soil to deeper layers during the first weeks after maize sowing.However, the general pattern of the modeled and simulated WFPSduring the whole growing season clearly shows that the low simu-lated values of soil NO3-N at Mabegondo cannot be explained solelyby an overestimation of water drainage. Moreover, the simulatedNO3-N leaching was fairly low (<10 kg N ha−1, data not shown) butconsistent with expected amounts at the experimental field (Báez,personal comm.).
3.3. Observed and simulated daily soil N2O fluxes
The performance of the FASSET model to predict daily soil N2Ofluxes in all of the systems investigated, except for maize in 2009when daily data was not available, is shown in Fig. 6. Higher system-atic model error and deviations between observed and simulateddaily soil N2O emissions were obtained at the Mabegondo sitewhere emissions were higher (Fig. 6b and c). However, soil andclimate factors, cereal crop, catch crop or fertilizer managementdid not influence model performance in terms of model agreementor model efficiency (Fig. 6a and d).
Measured and simulated daily soil N2O fluxes at Foulum, for thesystems shown in Fig. 2, and at Mabegondo, for all treatments in2010, are shown in Figs. 7 and 8, respectively. FASSET simulationsof the seasonal dynamics of soil N2O emissions were fairly good atFoulum in both conventional and organic systems with and with-out manure and with or without catch crops (Fig. 7). There was,however, overestimation of fluxes on particular days when soiloperations, i.e., ploughing and fertilization, were carried out. Noclear system effects were observed in the seasonal dynamics of soilN2O emissions. The slightly higher daily N2O emissions measured
16 J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20
0
20
40
60
80
100
120
WFPS(m
3m
-3)
measured
simulated
a
0
20
40
60
80
100
120
WFPS(m
3m
-3)
measured
simulated
b
Fig. 5. Observed and simulated water-filled pore space (WFPS) in the top 0–10 cmsoil layer during the growing period of maize in 2010 at Mabegondo in the cattleslurry (a) and unfertilized (b) treatments. Bars indicate standard deviation (n = 3).
in the conventional and organic systems with no catch crops (Fig. 7cand d) were consistently reproduced by the model.
At Mabegondo, daily soil N2O emissions were much higher andvariable than at Foulum and noticeable differences were foundonly between the unfertilized and fertilized treatments (Fig. 8). Thehigh soil N2O emissions at this site were captured by FASSET. Thesimulated dynamics of the N2O fluxes generally agreed with obser-vations, with the exception of some negative fluxes observed onfew days and a clear overestimation peak that occurred early inJune. This peak resulted from more than 100 mm of rainfall duringa four-day period (Fig. 1). The simulated emission peak was highestin the treatments fertilized with mineral fertilizer (Fig. 8d) and low-est in the unfertilized treatment (Fig. 8a) with intermediate peaks inthe slurry treatments. This indicates that the overestimation of soilN2O fluxes in this short interval of days was affected by the inter-action of fertilization and soil moisture dynamics. In particular, itcould be produced by the overestimation of the N2O/N2 ratio due tolower simulated moisture than the measured saturation conditionsat this time (Fig. 5). This clearly influenced the overestimation ofthe cumulative fluxes at the end of the growing season (Table 3) aswell as model performance (Fig. 6). The second mineral fertilizationin mid-July did not cause any additional peak in N2O emissions forthe remaining growing season, when the cumulative precipitationwas 45 mm until harvest, and this was correctly simulated by themodel (Fig. 8d).
4. Discussion
4.1. Influence of soil type and climate on soil N2O emissions incereal-based systems
Simulating the cereal systems at Foulum using soil and climateinputs used for Mabegondo increased the seasonal N2O emis-sions to the levels found at the Spanish site (results not shown).Thus, according to FASSET simulations the differences in the mag-nitude of emissions between regions would be mainly due tosoil characteristics, influenced by historical soil management, andto patterns of temperature and precipitation (Table 1, Fig. 1).While the cumulative seasonal soil N2O fluxes at Foulum where
Fig. 6. Linear regression between average daily soil N2O flux (double-log transformed) in the different cereal systems at Foulum (black circles) and Mabegondo (white circles)and the statistical performance of the simulated daily N2O fluxes (16–40 measurements per system): (a) index of agreement (d), (b) root mean square error (RMSE), (c) modelbias error (MBE) and (d) model efficiency (EF). *P < 0.01.
J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20 17
Fig. 7. Observed and simulated daily soil N2O emissions during the growing period of winter wheat and spring barley at Foulum in three organic cropping systems, digestedpig slurry manure and catch crop (a), unmanured with catch crop (b) and pig slurry without catch crop (c), and one conventional treatment with mineral fertilizer and withoutcatch crop (d). Bars indicate standard error (n = 2).
about 1 kg N2O–N ha−1 or less in the different cereals systems atMabegondo they ranged from 6.7 to 9.7 kg ha−1 when averagingthe two seasons. These figures agree with the potential risk of emis-sions established in both regions by Lesschen et al. (2011) based inmanagement and environmental parameters.
Roelandt et al. (2005) using an empirical model showed thatspring temperature and summer precipitation may explain up to35% of annual N2O emissions from cropland. No differences insummer precipitation among sites in 2009 or higher cumulativesummer precipitation at Foulum in 2010 (Fig. 1) indicate thattemperature would be the main climatic driver influencing thedifferences on seasonal N2O emissions in this study with highertemperatures accelerating the soil biochemical processes respon-sible for the formation of this gas.
Among the soil factors, organic C content, pH and texture havebeen reported to significantly influence N2O emissions (Stehfestand Bouwman, 2006). The levels of soil C contents (>3%) as well asof soil pH (<5.5) at Mabegondo would be most favorable to highN2O emissions levels by stimulating the process of denitrificationand/or increasing the ratio N2O/N2 (Stehfest and Bouwman, 2006;Peoples et al., 2009). A finer soil texture at Mabegondo than atFoulum would also improve soil water retention, increasing theperiod where moisture conditions are conducive for N2O emissions(Davidson et al., 2000). Moreover, a lower C/N ratio at Mabegondothan at Foulum would result in higher N mineralization rates andpotentially higher soil N2O fluxes. The soil environment, therefore,probably contributed by increasing the magnitude of difference inseasonal N2O emissions between the two sites.
Fig. 8. Observed and simulated daily soil N2O emissions during the growing period of maize at Mabegondo in 2010 in a conventional forage system: unfertilized (a), cattleslurry (b) pig slurry (c), and mineral fertilizer (d) treatments. Bars indicate standard error (n = 3).
18 J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20
The simulations performed with the FASSET model capturedthe magnitude of the seasonal N2O emissions in the two differ-ent edaphoclimatic conditions and cropping systems. This indicatesthat the functions implemented in the model to represent soil N2Oemissions (Chatskikh et al., 2005) are sufficiently sensitive to soiland climate differences covered in our study. However, we foundthat the uncertainty associated to the simulated soil N2O emis-sions increased with the magnitude of the emissions, thus, beingmuch higher at Mabegondo than at Foulum (Fig. 6). The agree-ment between modeled and measured soil moisture at Mabegondo(Fig. 5) was close to that at Foulum (Chirinda et al., 2011). Otherfactors possibly explain the higher uncertainty of the modeledemissions at Mabegondo. First, the model, uncalibrated for C andN dynamics, was unable to reproduce the extremely high levels ofsoil NO3-N in the topsoil 10 cm during the whole maize growingseason. Furthermore, although many studies report strong corre-lations between N2O emissions and soil NO3-N (e.g., Ruser et al.,2001; Turner et al., 2008; Laville et al., 2011), here, when pool-ing data from the two sites, this relationship was not observed atMabegondo (Fig. 4). Several years of grazed grassland precedingthe maize crop may have contribute to a large spatial and temporalvariability of emissions and soil NO3-N (Turner et al., 2008). Thismight be explained by preferential areas for animal movement atthe plot and spatial variability of urine and dung deposition at thefield level. All these factors suggest a high uncertainty associatedwith the measured values and propagated to the modeling resultsat Mabegondo.
4.2. Impact of crop and soil fertility management on N2Oemissions
In comparison with the edaphoclimatic factors, fertilizationmanagement and cropping system had lower direct influence onsoil N2O emissions. Increasing N2O emissions have been found tobe associated to increasing levels of N fertilization (e.g., Ruser et al.,2001; Stehfest and Bouwman, 2006; Van Groenigen et al., 2010;Hoben et al., 2011), even under organic farming (Rees et al., 2013).However, the type of fertilization, including comparison betweenorganic and inorganic forms of N, has a much lower effect on N2Oemissions (Stehfest and Bouwman, 2006; Chirinda et al., 2010) asalso found here for two contrasting sites. Meijide et al. (2007) foundthat the use of digested slurry reduced soil N2O fluxes by 25% inan irrigated maize crop compared to untreated pig slurry. In ourexperiment at Foulum, we could not see effects of different manuretypes. Rather, other management factors, such as crop residue man-agement and tillage have been observed to mostly influence N2Oemissions at this site (Mutegi et al., 2010; Brozyna et al., 2013).
The impact of fertilization on cumulative N2O emissions wasgenerally well reproduced by FASSET at both sites (Table 3). More-over, the current version of FASSET was able to reproduce thedynamics of seasonal N2O fluxes in conventional and organic, fertil-ized, manured and unfertilized cereal crops, with or without catchcrops (Figs. 7 and 8), thus, indicating a better ability of the modelwith respect an earlier version (Chirinda et al., 2011) to conductmodel-based studies aiming to reduce soil N2O emissions fromcropping systems.
4.3. Limitations and improvement prospects for modeling N2Oemissions
FASSET is currently unable to reproduce measured negative N2Ofluxes, i.e., soil as N2O sink, as those found at both sites on somedays. However, this restriction had an effect only at Foulum wherethe net cumulative seasonal N2O emissions were, on average, 10%lower than the emissions quantified solely by the positive N2Ofluxes to the atmosphere (data not shown). The difference between
the net seasonal cumulative N2O emissions and the positive N2Ofluxes represented on average less than 1% at Mabegondo. Theinability to reproduce soil N2O uptake at a daily step have alsobeen found in other studies applying process-based models (e.g.,Chirinda et al., 2011; Abdalla et al., 2014; Fitton et al., 2014).
The accuracy of modeling soil N2O emissions at daily time scaleunder field conditions is probably one of the main aspects affectingthe use of process-based models for predicting seasonal N2O emis-sions as a consequence of management practices. This is becausesubstantial deviations on simulated and actual N2O fluxes arereflected on the performance of the modeled cumulative emissionsunless influenced by compensating errors. Thus, even if patterns ofdaily emissions are well reproduced by FASSET (Figs. 7 and 8), achallenge remains to improve simulations of the N2O peaks asso-ciated with soil operations (soil tillage, addition of fertilizers orincorporation of crop residues) for assessing low-impact practices.Similar difficulties have been found in other widely used modelssuch as DNDC, MoBILE-DNDC, and DAYCENT (e.g., Smith et al., 2008;Chirinda et al., 2011; Abdalla et al., 2014). These models have con-trasting differences in the soil profile initialization schemes, e.g.,DNDC, similarly to FASSET, uses input values for SOM pools andDAYCENT is based on the long-term simulation of historical soiluse. However, DNDC and DAYCENT showed similar deviations forsimulating cumulative seasonal N2O (Del Grosso et al., 2005; Smithet al., 2008).
Data on the site history is necessary to adequately simulate theseasonal N2O emissions, since soil C and N dynamics in the differentsoil organic matter pools in FASSET are greatly affected by historicalinputs, in particular with grasslands and livestock farming (Turneret al., 2008). Site specific adjustment of soil processes responsiblefor N2O fluxes has been used as a way to improve the accuracyof the simulations (Scheer et al., 2014), but this cannot be appliedwhen investigating strategies for mitigation. Tillage depth is a verysensitive input parameter in FASSET influencing the dynamics ofN2O peaks. This implies that seasonal changes in soil propertiessuch as bulk density or porosity to a certain depth, which temporalvariation is not dynamically modeled, should be considered as afuture model improvement. More intensive field samplings (i.e.,hourly) during short-periods would lead to a better understandingof temporal dynamics of the N2O fluxes related to field operations,as shown by Laville et al. (2011), and thus, provide basis for modelimprovements to reduce the uncertainty of the simulated effect ofsoil tillage on N2O emissions.
4.4. Role of field-scale modeling for mitigating soil N2O emissions
Different studies have shown that the estimation of direct soilN2O emissions from croplands based on the use of process-basedmodels (e.g., Li et al., 2001; Chatskikh et al., 2005; Del Grosso et al.,2005; Fitton et al., 2014) for national inventories represents anstep forward compared with the IPCC methodology relying on con-stant emission factors derived from a limited number of data points(Bouwman, 1996). This is explained by the inclusion of tempo-ral dynamics in the models as influenced by environmental andmanagement factors as well as its interactions under current andhistorical land use. As an example, reported non-linear relation-ship between N input rates and soil N2O emissions (Van Groenigenet al., 2010) on a given site could, theoretically, be investigated witha model-based analysis, but not if the analysis is solely done on thebasis of constant emission factors. This is a strong reason for whichprocess-based models constitute important tools for the analysesof strategies aiming to reduce N2O emissions in agricultural soilsunder diverse environmental conditions and cropping systems asshown in our study and several other recent works (Abdalla et al.,2014; Abdalla et al., 2014). In the case of the FASSET model, itsability to simulate cereal yields under a range of cropping systems
J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20 19
at the two locations of the study (Doltra et al., 2011; unpublishedat Mabegondo) make it possible to use the model for identifyingmanagement strategies that also consider crop productivity, andthat result in minimum yield-scaled N2O emissions, as proposedby Van Groenigen et al. (2010).
To increase the confidence on N2O model projections, withthe increasing number of modeling studies, it is necessary to bet-ter understand the many sources of uncertainty affecting N2Osimulations. In an analysis performed by Fitton et al. (2014) itwas concluded that among different input variables related tosoil and climate, soil pH was the main driver of uncertainties inN2O simulations using the DailyDaycent model. In our study wefound the highest uncertainty on simulated seasonal N2O fluxes atMabegondo, particularly linked to the high variability of daily soilNO3-N concentrations (Fig. 3) and N2O fluxes (Fig. 8). In its currentversion FASSET has been shown to reproduce the seasonal dynam-ics of N2O fluxes in conventional, organic, manured, unfertilizedand systems with and without catch crops in contrasting environ-ments from low to high risk of soil N2O emissions. This indicates apotential for this model to be used in up-scaling studies at a regionalor national level.
Crop process-based model ensemble studies (e.g., Asseng et al.,2013, Martre et al., 2015) are increasingly being undertaken toimprove the modeling of climate change impacts on crop yields.Analogously, inter-comparisons of biogeochemical models are stillvery rare (De Vries et al., 2010), but they would contribute to iden-tify best modeling approaches at field and regional scales to betterestimate N2O emissions under different environmental conditions,including soils, climates and cropping systems.
5. Conclusions
The environmental factors such as temperature and soil charac-teristics, influenced by historical soil management, appear to be themain drivers for the ten-fold difference in N2O emissions betweenthe Mabegondo and Foulum sites. Fertilizer management, catchcrops and cropping systems had a lower influence on the seasonalsoil N2O emissions compared to edaphoclimatic factors. FASSETversion 2.5 was found to generally reproduce well the effects of thedifferent factors investigated, i.e., soil, climate, fertilization, catchcrop or cropping system, on cumulative seasonal soil N2O emis-sions. However, overestimation of fluxes on particular days wasrelated to field operations, i.e., ploughing and fertilization, beingindicative of a temporary overestimation of nitrification and deni-trification processes. A more intensive field sampling (i.e., hourly)during these short-periods would enable a better understanding ofthe temporal dynamics of the N2O fluxes related to field operationsand provide a basis for model improvement. Finally, the ability ofFASSET to reproduce the effects of fertilization strategies based ondifferent N rates still needs to be evaluated in order to investigatestrategies that minimize yield-scaled N2O emissions.
Acknowledgements
This study was supported by the projects RTA2012-00065-C05-03 funded by the Spanish National Institute for Agricultural andFood Research and Technology (INIA), 10MRU503001PR (Xunta deGalicia), the EU-FP7 LegumeFutures project (grant 245216) and theMACSUR project funded by the Danish Strategic Research Council(contract 0603-00507B). J. Doltra, via TAD/CRP JA00077691 fellow-ship under the OECD Co-operative Research Programme: BiologicalResource Management for Sustainable Agricultural Systems.
References
Abdalla, M.J.W., Hastings, A., Helmy, M., Prescher, A., Osborne, B., Lanigan, G.,Forristal, D., Killi, D., Maratha, P., Williams, M., Rueangritsarakul, K., Smith, P.,Nolan, P., Jones, M.B., 2014. Assessing the combined use of reduced tillage andcover crops for mitigating greenhouse gas emissions from arable ecosystem.Geoderma 223–225, 9–20.
Asseng, S., Ewert, F., Rosenzweig, C., Jones, J.W., Hatfield, J.L., Ruane, A.C., Boote,K.J., Thorburn, P.J., Rötter, R.P., Cammarano, D., Brisson, N., Basso, B., Martre, P.,Aggarwal, P.K., Angulo, C., Bertuzzi, P., Biernath, C., Challinor, A.J., Doltra, J.,Gayler, S., Goldberg, R., Grant, R., Heng, L., Hooker, J., Hunt, L.A., Ingwersen, J.,Izaurralde, R.C., Kersebaum, K.C., Müller, C., Naresh Kumar, S., Nendel, C.,O’Leary, G., Olesen, J.E., Osborne, T.M., Palosuo, T., Priesack, E., Ripoche, D.,Semenov, M.A., Shcherbak, I., Steduto, P., Stöckle, C., Stratonovitch, P., Streck,T., Supit, I., Tao, F., Travasso, M., Waha, K., Wallach, D., White, J.W., Williams,J.R., Wolf, J., 2013. Uncertainty in simulating wheat yields under climatechange. Nat. Clim. Change 3, 827–832.
Berntsen, J., Petersen, B.M., Jacobsen, B.H., Olesen, J.E., Hutchings, N.J., 2003.Evaluating nitrogen taxation scenarios using the dynamic whole farmsimulation model FASSET. Agric. Syst. 76, 817–839.
Bouwman, A.F., 1996. Direct emissions of nitrous oxide from agricultural soils.Nutr. Cycl. Agroecosyst. 46, 53–70.
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.
Butterbach-Bahl, K., Baggs, E.M., Dannenmann, M., Kiese, R.,Zechmeister-Boltenstern, S., 2013. Nitrous oxide emissions from soils: howwell do we understand the processes and their controls? Phil. Trans. R. Soc. B368, http://dx.doi.org/10.1098/rstb.2013.012, 20130122.
Cardenas, L.M., Thorman, R., Ashlee, N., Buttler, M., Chadwick, D., Chambers, B.,Cuttle, S., Donovan, N., Kingston, H., Lane, S., Dhanoa, M.S., Scholefield, D., 2010.Quantifying annual N2O emissions fluxes from grazed grasslands under a rangeof inorganic fertilizer nitrogen inputs. Agric. Ecosyst. Environ. 136, 218–226.
Chatskikh, D., Olesen, J.E., Berntsen, J., Regina, K., Yamulki, S., 2005. Simulation ofeffects of soils: climate and management on N2O emission from grasslands.Biogeochem. 76, 395–419.
Chirinda, N., Carter, M.S., Albert, K.R., Ambus, P., Olesen, J.E., Porter, J.R., Petersen,S.O., 2010. Emissions of nitrous oxide from arable organic and conventionalcropping systems on two soil types. Agric. Ecosyst. Environ. 136, 199–208.
Chirinda, N., Kracher, D., Laegdsmand, M., Porter, J.R., Olesen, J.E., Petersen, B.M.,Doltra, J., Kiese, R., Butterbach-Bahl, K., 2011. Simulating soil N2O emissionsand heterotrophic CO2 respiration in arable systems using FASSET andMoBiLE-DNDC. Plant Soil 343, 139–160.
Davidson, E.A., Keller, M., Erickson, H.E., Verchot, L.V., Veldkamp, E., 2000. Testing aconceptual model of soil emissions of nitrous and nitric oxides. BioScience 50,667–680.
De Vries, W., Lesschen, J.P., Oudendag, D.A., Kros, J., Voogd, J.C., Stehfest, E.,Bouwman, A.F., 2010. Impacts of model structure and data aggregation onEuropean wide predictions of nitrogen and greenhouse gas fluxes in responseto changes in livestock, land cover, and land management. J. Integr. Environ.Sci. 7, 145–157.
Del Grosso, S.J., Mosier, A.R., Parton, W.J., Ojima, D.S., 2005. DAYCENT modelanalysis of past and contemporary soil N2O and net greenhouse gas flux formajor crops in the USA. Soil Tillage Res. 83, 9–24.
Doltra, J., Lægdsmand, M., Olesen, J.E., 2011. Cereal yield and quality as affected bynitrogen availability in organic and conventional arable crop rotations Acombined modeling and experimental approach. Eur. J. Argon. 34, 83–95.
Doltra, J., Lægdsmand, M., Olesen, J.E., 2014. Impacts of projected climate changeon productivity and nitrogen leaching of crop rotations in arable and pigfarming systems in Denmark. J. Agric. Sci. 152, 75–92.
Fitton, N., Datta, A., Smith, K., Williams, J.R., Hastings, A., Kuhnert Topp, C.F.E.,Smith, P., 2014. Assessing the sensitivity of modeled estimates of N2Oemissions and yield to input uncertainty at a UK cropland experimental siteusing the DailyDayCent model. Nutr. Cycl. Agroecosyst. 99, 119–133.
García, M.I., Báez, D., Louro, A., Castro, J., 2012. Influence of different nitrogenfertilizers on forage maize yield and quality. In: Richards, K.G., Fenton, O.,Watson, C.J. (Eds.), Proceedings of the 17th Nitrogen Workshop?Innovationsfor sustainable use of nitrogen resources. 26–29th June 2012. Wexford, Ireland,pp. 417–418.
Hansen, S., Jensen, H.E., Nielsen, N.E., Svendsen, H., 1990. Daisy-Soil PlantAtmosphere System Model, Npo Research Programme Report A10. TheNational Agency of Environmental Protection, Copenhagen.
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. Glob. Change Biol. 17, 1140–1152.
Iglesias, A., Garrote, L., Quiroga, S., Moneo, M., 2009. Impacts of Climate Change inAgriculture in Europe. Peseta—agriculture Study. European Commission JointResearch Centre Institute for Prospective Technological Studies, pp. 1–59(report).
IPCC, Working Group II. Climate change 2014: impacts, adaptation, andvulnerability Intergovernmental Panel on Climate Change, 2014.
Keeney, D.R., Nelson, D.W., 1982. Nitrogen inorganic forms. In: Page, A.L., Miller,R.H., Keeney, D.R. (Eds.), Methods of Soil Analysis Part II. ASA, Madison, WI, pp.643–698.
20 J. Doltra et al. / Europ. J. Agronomy 66 (2015) 8–20
Laville, P., Lehuger, S., Loubet, B., Chaumartin, F., Cellier, P., 2011. Effect ofmanagement: climate and soil conditions on N2O and NO emissions from anarable crop rotation using high temporal resolution measurements. Agric.Forest Meteorol. 151, 228–240.
Lesschen, J.P., Velthof, G.L., de Vries, W., Kross, J., 2011. Differentiation of nitrousoxide emission factors for agricultural soils. Environ. Pollut. 159, 3215–3222.
Li, C., Zhuang, Y., Cao, M., Crill, P., Dai, Z., Frolking, S., Moore, B., Salas, W., Song, W.,Wang, X., 2001. Comparing process-based agroecosystem model to the IPCCmethodology for developing a national inventory of N2O emissions from arablelands in China. Nutr. Cycl. Agroecosyst. 60, 159–175.
Li, C., Mosier, A., Wassmann, R., Cai, Z., Zheng, X., Huang, Y., Tsuruta, H., Boonjawat,J., Lantin, R., 2004. Modeling greenhouse gas emissions from rice-basedproduction systems: sensitivity and upscaling. Glob. Biogeochem. Cycles 18,GB1043.
Louro, A., Sawamoto, T., Chadwick, D., Pezzolla, D., Bol, R., Báez, D., Cárdenas, L.,2013. Effect of slurry and ammonium nitrate application on greenhouse gasfluxes of a grassland soil under atypical south west England weatherconditions. Agric. Ecosyst. Environ. 181, 1–11.
MAGRAMA, 2012. Anuario de Estadística Agraria 2012. Capítulo 13. Superficies yproducciones de cultivos: Cultivos forrajeros. Ministerio de Agricultura,Alimentación y Medio Ambiente.
Martre, P., Wallach, D., Asseng, A., Ewert, F., Jones, J.W., Rötter, R.P., Boote, K.J.,Ruane, A.C., Thorburn, P.J., Cammarano, D., Hatfield, J.L., Rosenzweig, C.,Aggarwal, P.K., Angulo, C., Basso, B., Bertuzzi, B., Biernath, P., Brisson, C.,Challinor, N., Doltra, A.J., Gayler, J., Goldberg, S., Grant, R., Heng, R.F., Hooker, L.,Hunt, J., Ingwersen, L.A., Izaurralde, J.C., Kersebaum, R.C., Müller, K.C., Kumar,C., Nendel, S.N., O’Leary, C., Olesen, G.J., Osborne, J.E., Palosuo, T.M., Priesack, T.,Ripoche, E., Semenov, D., Shcherbak, M.A., Steduto, I., Stöckle, P., Stratonovitch,C.O., Streck, P., Supit, T., Tao, I., Travasso, F., Waha, M., White, K., Wolf, J., 2015.Multimodel ensembles of wheat growth: many models are better than one.Glob. Change Biol. 21, 911–925.
Meijide, A., Díez, J.A., Sánchez-Martín, L., López-Fernández, S., Vallejo, A., 2007.Nitrogen oxide emissions from an irrigated maize crop amended with treatedpig slurries and composts in a Mediterranean climate. Agric. Ecosyst. Environ.121, 383–394.
Metay, A., Chapuis-Lardy, L., Findeling, A., Oliver, R., Alves Moreira, J.A., Feller, C.,2011. Simulating N2O fluxes from a Brazilian cropped soil withcontrastedtillage practices. Agric. Ecosyst. Environ. 140, 255–263.
Mutegi, J.K., Munkholm, L.J., Petersen, B.M., Hansen, E.M., Petersen, S.O., 2010.Nitrous oxide emissions and controls as influenced by tillage and crop residuemanagement strategy. Soil Biol. Biochem. 42, 1701–1711.
Nash, J.E., Sutcliffe, J.V., 1970. River flow forecasting through conceptual models,Part I–A discussion of principles. J. Hydrol. 10, 282–290.
Olesen, J.E., Petersen, B.M., Berntsen, J., Hansen, S., Jamieson, P.D., Thomsen, A.G.,2002. Comparison of methods for simulating effects of nitrogen on green areaindex and dry matter growth in winter wheat. Field Crops Res. 74, 131–149.
Palosuo, T., Kersebaum, K.C., Angulo, C., Hlavinka, P., Moriondo, M., Patil, R., Ruget,F., Rumbaur, C., Takác, J., Trnka, M., Bindi, M., Caldag, B., Ewert, F., Ferrise, R.,Mirschel, W., Olesen, J., Saylan, L., Siska, B., Rötter, R., 2011. Simulation ofwinter wheat yield and its variability in different climates of Europe: acomparison of eight crop growth models. Eur. J. Argon. 35, 103–114.
Peoples, M.B., Hauggaard-Nielsen, H., Jensen, E.S., 2009. The potentialenvironmental benefits and risks derived from legumes in rotations. Nitrogenfixation in crop production. In: Emerich, D.W., Krishnan, H.B. (Eds.), Agronomy
Monograph 52. Nitrogen Fixation in Crop Production. Am. Soc. Agron., Crop Sci.Soc. Am., and Soil Sci. Soc Am., Madison, WI, USA, pp. 349–385.
Petersen, B.M., Berntsen, J., Hansen, S., Jensen, L.S., 2005. CN-SIM: a model for theturnover of soil organic matter: II. Short-term carbon and nitrogendevelopment. Soil Biol. Biochem. 37, 375–393.
Plantedirektoratet, 2009. Statistik over økologiske jordbrugsbedrifter 2008.autorisation og produktion. Plantedirektoratet, Lyngby, Denmark.
Rees, R.M., Augustin, J., Alberti, G., Ball, B.C., Boeckx, P., Canterel, A., Castaldi, S.,Chirinda, N., Chojnicki, B., Giebels, M., Gordon, H., Grosz, B., Horvath, R.,Juszczak, R., Klemedtsson, Å.K., Klemedtsson, L., Medinets, S., Machon, A.,Mapanda, F., Nyamangara, J., Olesen, J.E., Reay, D., Sanchez, L., Sanz-Cobena, A.,Smith, K.A., Sowerby, A., Sommer, M., Soussana, J.F., Stenberg, M., Topp, C.F.E.,van Cleemput, O., Vallejo, A., Watson, C.A., Wuta, M., 2013. Nitrous oxideemissions from European agriculture: an analysis of variability and drivers ofemissions from field experiments. Biogeosciences 10, 2671–2682.
Roelandt, C., Van Wesemael, B., Rousenvell, M., 2005. Estimating annual N2Oemissions from agricultural soils in temperate climates. Glob. Change Biol. 11,1701–1711.
Rötter, R.P., Palosuo, T., Kersebaum, K.C., Angulo, C., Bindi, M., Ewert, F., Ferrise, R.,Hravlinka, P., Moriondo, M., Nendel, C., Olesen, J.E., Patil, R., Ruget, F., Takac, J.,Trnka, M., 2012. Simulation of spring barley yield in different climatic zones ofNorthern and Central Europe: a comparison of nine crop models. Field CropsRes. 133, 23–36.
Ruser, R., Flessa, H., Schilling, R., Beese, F., Much, J.C., 2001. Effect of crop-specificfield management and N fertilization on N2O emissions from a fine-loamy soil.Nutr. Cycl. Agroecosyst. 59, 177–191.
Ryden, J.C., Rolston, D.E., 1983. The measurement of denitrification. In: Freney, J.R.,Simpson, J.R. (Eds.), Gaseous Loss of Nitrogen from Plant Soil Systems.Developments in Plant and Soil Sciences, vol. 9. Springer, The Hague,Netherlands, pp. 91–132.
Salcedo, G., 2011. Minimización y aprovechamiento del purín en origen de lasexplotaciones lecheras de Cantabria. Centro de Investigación del MedioAmbiente (CIMA), Gobierno de Cantabria.
Saxton, K.E., Rawls, W.J., 2006. Soil water characteristic estimates by texture andorganic matter for hydrologic solultions. Soil Sci. Soc. Am. J. 70, 1569–1578.
Sinclair, T.R., Muchow, R.C., 1999. Radiation use efficiency. In: Sparks, D.L. (Ed.),Advances in Agronomy, 215–265. Academic Press, p. 65.
Smith, W.N., Grant, B.B., Desjardins, R.L., Rochette, P., Drury, C.F., Li, C., 2008.Evaluation of two process-based models to estimate soil N2O emissions inEastern Canada. Can. J. Soil Sci. 88 (2), 251–260.
Stehfest, E., Bouwman, L., 2006. N2O and NO emissions from agricultural fields andsoils under natural vegetation: summarizing available measurement data andmodeling of global annual emissions. Nutr. Cycl. Agroecosyst. 74, 207–228.
Turner, D.A., Chen, D., Galbally, I.E., Leuning, R., Edis, R.B., Li, Y., Kelly, K., Phillips, F.,2008. Spatial variability of nitrous oxide emissions from an Australian irrigateddairy pasture. Plant Soil 309, 77–88.
Van Groenigen, J.W., Velthof, G.L., Oenema, O., Van Groenigen, K.G., Van Kessel, C.,2010. Towards and agronomic assessment of N2O emissions: a case study forarable crops. Eur. J Soil Sci. 61, 903–913.
Willmott, C.J., 1982. Some comments on the evaluation of model performance.Bull. Am. Met. Soc. 63 (11), 1309–1313.
