Agriculture, Ecosystems and Environment 206 (2015) 71–83

Contents lists available at ScienceDirect

Agriculture, Ecosystems and Environment

journal homepage: www.elsevier.com/locate/agee

Assessing the effects of agricultural management on nitrous oxide

emissions using flux measurements and the DNDC model
Kingsley Chinyere Uzoma a , Ward Smith a, * , Brian Grant a , Raymond L. Desjardins a ,
Xiaopeng Gao b , Krista Hanis b , Mario Tenuta b , Pietro Goglio a , Changsheng Li c
a
Eastern Cereal and Oilseed Research Centre, Agriculture and Agri-Food Canada, 960 Carling Avenue, Ottawa, ON K1A 0C6, Canada
b
Department of Soil Science, University of Manitoba, Winnipeg, MB R3T 2N2, Canada
c
Institute for the Study of Earth, Oceans, and Space, Complex Systems Research Center, University of New Hampshire, Durham, NH 03824, USA

A R T I C L E I N F O A B S T R A C T

Article history: Biogeochemical models are useful tools for integrating the effects of agricultural management on GHG
Received 13 August 2014 emissions; however, their development is often hampered by the incomplete temporal and spatial
Received in revised form 12 March 2015 representation of measurements. Adding to the problem is that a full complement of ancillary
Accepted 13 March 2015
measurements necessary to understand and validate the soil processes responsible for GHG emissions is
Available online xxx
often not available. This study presents a rare case where continuous N2O emissions, measured over
seven years using a flux gradient technique, along with a robust set of ancillary measurements were used
Keywords:
to assess the ability of the DNDC model for estimating N2O emissions under varying crop-management
DNDC
Modeling
regimes. The analysis revealed that the model estimated soil water content more precisely in the normal
Nitrous oxide and wet years (ARE 3.4%) than during the dry years (ARE 11.5%). This was attributed to the model’s
Agricultural management inability to characterize episodic preferential flow through clay cracks. Soil mineral N across differing
Flux measurements management regimes (ARE 2%) proved to be well estimated by DNDC. The model captured the relative
Soil nitrogen differences in N2O emissions between the annual (measured: 35.5 kg N2O-N ha
1, modeled: 30.1 kg
Validation N2O-N ha
1) and annual-perennial (measured: 26.6 kg N2O-N ha
1, modeled: 21.2 kg N2O-N ha
1)
cropping systems over the 7 year period but overestimated emissions from alfalfa production and
underestimated emissions after spring applied anhydrous ammonia. Model predictions compared well
with the measured total N2O emissions (ARE
11%) while Tier II comparison to measurements (ARE

75%) helped to illustrate the strengths of a mechanistic approach in characterizing the site specific
drivers responsible for N2O emissions. Overall this study demonstrated the benefits of having near
continuous GHG flux measurements coupled with detailed ancillary measurements towards identifying
soil process interactions responsible for regulating GHG emissions.
Crown Copyright ã 2015 Published by Elsevier B.V. All rights reserved.

1. Introduction Several management options have the potential to reduce N2O

The increase in agricultural N2O emissions, which accounts for a
major portion of the total anthropogenic N2O emissions (Kroeze
et al., 1999; Mosier and Kroeze, 2000), has been attributed
primarily to increased reliance on fertilizer nitrogen (N) for crop
production (Saggar, 2010; Scott et al., 1996; Snyder et al., 2009). The
level of N use by crops depends on crop species, fertilizer type, and
the timing and quantity of fertilizer application. Unrecovered N can
be lost to the environment either through runoff, leaching or
gaseous emissions (Brady and Weil, 2002).
* Corresponding author. Tel.: +1 613 759 1334; fax: +1 613 759 1432.
E-mail address: ward.smith@agr.gc.ca (W. Smith).
emissions such as improved fertilizer placement and scheduling,
soil N testing to optimize application rate, the use of N inhibitors,
slow release fertilizers, catch and cover crops. In the IPCC 5th
Assessment Report it was indicated that cropland management
was one of the most cost-effective options to mitigate GHG
(greenhouse gas) emissions in agriculture (Smith et al., 2014).
Several authors (Goglio et al., 2012; Sieling and Kage, 2006) have
suggested that environmental impacts at the field level should be
assessed at the crop system scale. It has been acknowledged that
changes in management practices to produce a given environ-
mental effect may also produce unfavorable environmental
impacts that counterbalance the benefit. Also, the effects of
management on GHG emissions are strongly tied to climate and
soil properties such as soil organic carbon (SOC) content, soil
nitrogen content, and pH, thus there can be contrasting effects of

http://dx.doi.org/10.1016/j.agee.2015.03.014
0167-8809/ Crown Copyright ã 2015 Published by Elsevier B.V. All rights reserved.
72 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83

management both temporally and spatially. Soil temperature and
moisture have significant impacts on all processes (mineralization,
nitrification, and denitrification) affecting N2O emissions (Dalal
et al., 2003; Frolking et al., 1998; Saggar, 2010; Snyder et al., 2009)
and also strongly influence crop production.
The high variability of N2O fluxes makes it difficult to quantify
emissions across contrasting environments. Chambers, automated
chambers and micrometeorological techniques provide crucial
data on a site-basis for a specific period of time, however, models
and tools are required to extrapolate these data to larger spatial
and temporal scales. Micrometeorological techniques, which allow
for continuous field scale monitoring of N2O emissions (Denmead
and Raupach, 1993) are of particular value for understanding
soil–climate–plant processes required to develop and test models
(Del Grosso et al., 2008).
As an alternative to direct measurements, process-based
models enable the integration of the contributing biogeochemical
processes and are increasingly used internationally to simulate
N2O emissions in agroecosystems (Jarecki et al., 2008; Kariyapper-
uma et al., 2011; Parton et al., 1988; Smith et al., 2002; Wolf et al.,
2012). Dynamic process based models are mathematical models
which describe the growth and development of a crop interacting
with a soil–carbon system (Wallach et al., 2006). A major strength
of process-based models is that they consider agriculture from a
systems perspective including productivity, GHG emissions and
soil carbon, all of which are inter-dependent. The models ensure
that a mass balance of C, N and water is maintained.
The development of predictive models is limited both by the
mathematical integration of complex processes and the lack of
understanding of some processes. It is important to test models
rigorously against comprehensive datasets and report findings
regarding which processes work and which processes require
improvement under various environments. Model testing and
validation is of great importance to the scientific community to the
extent that large scale projects such as the Agricultural Model
Inter-comparison and Improvement Project (AgMIP; Rosenzweig
et al., 2013) have been formed to link various communities who
focus on improving agricultural models.
The DeNitrification DeComposition (DNDC) model is a widely
used process-based model which simulates C and N dynamics and
trace gas emissions for a wide range of management practices
(Smith et al., 2013). The model was primarily designed and
parameterized using measurements and cropping systems in the
US (Li et al., 1992) which prompted many researchers to develop
country-specific versions of the model to better enable the
simulation of regional soils, climate events, crop cultivars and
management practices (Brown et al., 2002; Giltrap et al., 2008;
Kröbel et al., 2011; Werner et al., 2007). Over the last few years a
Canadian regionalized version of the DNDC model has been
developed in parallel with the standard DNDC model through close
collaborations with the model author in order to improve model
performance in cool weather conditions (Grant et al., 2015; Kröbel
et al., 2011; Smith et al., 2013).
Because of the dependence of N2O emissions on environmental
factors, which are continuously changing with climatic conditions,
vegetation types and farming systems, there is a need for
continuous testing, improvement and validation of models as
new measurements become available. Most DNDC validation
studies for Canada focused on using chamber N2O measurements
(Grant et al., 2015; Smith et al., 2008, 2002) as continuous
micrometeorological data were rarely available. The objectives of
this study were: (i) to use continuous measurements to evaluate
DNDC from a systems perspective considering soil water,
temperature, N dynamics, and N2O emission mechanisms, (ii) to
evaluate the ability of the DNDC model to estimate the influence of
management practices (tillage, annual vs. perennial crops, timing

Table 1
Agricultural management performed on the four plots (P1, P2, A1 and A2) during the experimental period (2006–2012).

Years Crop Tillage type Planting date fertilizer type Fertilizer Fertilizer timing Harvest date
(day of year) (kg N ha
1) (day of year) (day of year)
P1 and P2 (annual–perennial system)
2006 Corna RT 136 N-P-K-S/urea 109 136 279
2007 Faba beanb RT 131 239
2008 Alfalfac RT 149
2009 Alfalfac RT 197 187, 323
2010 Alfalfac RT 192, 239
2011 Alfalfac RT 210, 217
2012 Corna IT 124 N-P-K/urea 63 124 292

A1 (annual system)
2006 Corna IT 136 N-P-K-S/urea 109 136 279
2007 Faba beanb IT 131 239
2008 Spring Wheatd IT 142 N-P-K 98 142 260
2009 Rapeseede IT 150 Urea 146 158 265
2010 Barleyf IT 142 Urea 98 146
2011 Spring Wheatd IT 161 Anhydrous NH3 100/160 161/308 270
2012 Corna IT 124 292

A2 (annual system)
2006 Corna IT 136 N-P-K-S/urea 109 136 279
2007 Faba beanb IT 131 239
2008 Spring Wheatd IT 142 N-P-K 98 142 260
2009 Rapeseede IT 150 Urea 146 158 265
2010 Barleyf IT 142 Urea/NH3 98/100 146/288
2011 Spring Wheatd IT 161 270
2012 Corna IT 124 Anhydrous NH3 160 124 292
a
Zea mays L.
b
Vicia faba var minor L.
c
Medicago sativa L.
d
Triticum aestivum L.
e
Brassica napus L.
f
Hordeum vulgare L.
 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83
of fertilizer application) on N2O emissions from a poorly drained
clay soil in western Canada and (iii) to identify biogeochemical
processes that should be targeted for improvement in the model.
2. Materials and methods
2.1. Study site
The flux gradient technique was used to estimate the influence
of agricultural management on N2O fluxes at the TGAS-Man study
site (AAFC-AGGP) at the University of Manitoba’s Glenlea Research
Station in Southern Manitoba, Canada (49.64 N, 97.16 W; 235 m
elev.) (Maas et al., 2013). The site was located in the Red River
Valley on glaciolacustrine clay floodplains and has a temperate,
sub-humid climate with a 30-year average (1971–2000) annual
precipitation of 530 mm and a mean annual air temperature of
2.4  C (Environment Canada, 2012). The 30 year normals
(1971–2000) for mean air temperature and precipitation for the
growing season (May 1–October 31) were 14.1  C and 399 mm,
respectively, and 1810 was the average GDD at a 5  C base
temperature. The soil, classified as a Gleyed Humic Vertisol
(Canadian System) or Dystric Vertisol (FAO, 2014), had a surface
(0–0.2 m) bulk density of 1.2 Mg m
3, organic C content of
32 g kg
1, a clay texture (60% clay, 35% silt, 5% sand) and pH (2:1
water:soil mixture ratio) of 6.2. Over the course of the study three
main management treatments were employed: (1) intensive till
(IT) vs. minimum till (RT), (2) annual vs. perennial cropping, and (3)
fall vs. spring applied anhydrous ammonia.
The experiment was initiated in the fall of 2005 on land that
was in fallow that year. In the spring of 2006, treatments were
introduced on a 2 by 2 square grid of plots within a 30-ha field
comprising four 200 m by 200 m plots (Glenn et al., 2012, 2010).
The two westward annual-perennial plots were designated as
P1 and P2 and the two eastward annual plots as A1 and A2
(Table 1). In the spring of 2006, all plots were sown to corn (Zea
mays L.) and fertilized with a broadcast applied urea at 109 kg N
ha
1. The tillage treatments were started in 2006 and ended in the
fall of 2007 and consisted of the intensive tillage (IT) treatment on
A1 and A2, which were tilled with a heavy-duty chisel plow to a
depth of approximately 0.2 m, and the minimum tillage (RT)
treatments on P1 and P2, with tillage kept at a shallow depth of
0.02–0.03 m. Corn above ground residues were mulched and
incorporated in IT treatments while in RT treatments residues
were mulched and left on the soil surface. Faba bean (Vicia faba var
minor L.) was planted on all plots in the spring of 2007. In the spring
of 2008 the annual vs. perennial crop treatments were initiated.
Perennial forage alfalfa (Medicago sativa L.) was planted on P1 and
P2 and remained as alfalfa until the fall of 2011, while A1 and
A2 remained in annual crop rotation in the annual sequence of
spring wheat (Triticum aestivum L.), rapeseed (Brassica napus L.),
barley (Hordeum vulgare L.) and spring wheat (2008–2011). A fall
vs. spring applied anhydrous ammonia treatment was initiated in
the fall of 2010 on A1 and A2 plots. In 2012, the four plots were
tilled with a heavy harrow and straight spring time harrow and
sown with corn. Collectively, P1 and P2 were designated P system,
and A1 and A2 as A system.
Daily maximum (Tmax) and minimum air temperatures (Tmin),
precipitation, solar radiation and wind speed data (2006–2012)
were collected from the on-site weather station at the TGAS-Man
study site. Soil temperature was measured in three replicates at
depths of 0.02, 0.05, 0.1, 0.2 and 0.5 m using Type-T copper-
constantan thermocouples and recorded using a data logger
(CR1000, Campbell Sci. Inc.). Volumetric soil moisture content was
measured using HOBO (Onset) soil moisture smart sensors
installed vertically in the soil down to a depth of 0.1 m. Prolonged
wet or dry conditions occasionally resulted in either soil flooding
73
or cracking, resulting in the damage of the sensors circuits or a
partial loss of contact between the soil moisture probes and the soil
matrix, leading to unusable soil moisture data during these
conditions. Soil NH4+-N and NO3
-N concentrations were deter-
mined on soil samples taken from the 0–0.3 m depth, using the
phenate and copper cadmium reduction to nitrite methods,
respectively (Glenn et al., 2012).
2.2. Micrometeorological N2O flux measurements
The flux gradient (FG) micrometeorological technique
(Wagner-Riddle et al., 2007) was used to determine the net flux
of N2O (FN) between the soil–crop system and the lower
atmosphere over 30-min intervals (Glenn et al., 2012). Specifically
stated as FN =
K  D[N2O]/Dz, where K represents the eddy
diffusivity for N2O, and D[N2O] is the concentration gradient of
N2O over the vertical distance Dz (Glenn et al., 2012). Using a
similarity theory, eddy-covariance-based aerodynamic method
described by Denmead and Raupach (1993) an estimate of K was
derived. Experimental design was conceptualized such that similar
crop management was maintained within and outside of the
experimental plots to ensure K estimates characterized an even
wider area. With the experimental plots being fairly large (4 ha)
along with the micrometeorological towers being positioned in the
centre of the four plots, for all wind directions a fetch to
observation height ratio of approximately 100:1 was maintained
ensuring that most of the flux footprint originated within the
treatment plot (Glenn et al., 2012). It was determined that the flux
detection limit was 0.05 nmol N2O m
2 s
1 (1.21 g N2O-N ha
1
d
1) based on a friction velocity threshold of 0.12 m s
1 and a
minimum gradient detection level of 0.045 nmol mol
1 deter-
mined as the standard deviation of concentration measurements
when the intakes were placed at the same height (Glenn et al.,
2012). Nitrous oxide flux gradients were calculated on 30 min
intervals sequentially across P1, A1, A2, and P2, resulting in
approximately one average 30-min gradient every 2 h per plot,
leading to up to 12 half hourly N2O flux measurements per day per
plot as described by Glenn et al. (2010).
2.3. The DNDC model and regionalized improvements
2.3.1. DNDC model
The DNDC model was developed to simulate daily and seasonal
C and N dynamics in agroecosystems (Li et al., 1992). The model
includes three user input forms, to describe climate/soil/manage-
ment inputs that influence C&N biogeochemical cycles. These
inputs characterize (i) climatic conditions consisting of daily
precipitation, and maximum and minimum daily air temperatures,
CO2 concentration, wind speed, relative humidity and solar
radiation, (ii) soil parameters such as texture, bulk density,
hydraulic conductivity, pH, and total organic C and microbial
activities, and (iii) farm management including crop vs. fallow,
planting and harvesting dates, tillage intensities and dates, and
timing, type and amount of fertilizer N and manure applications,
weeding, irrigation, grazing and cutting. The general model
framework consists of hundreds of interdependent processes
which are aggregated together into sub-models based on their co-
dependencies. These sub-models can be further aggregated and
described as being part of one of two major model components.
The first of these consists of the soil–climate, crop growth, and
decomposition sub-models. The soil–climate sub-model predicts
the daily soil temperature and moisture profiles in one dimension,
soil redox potential (Eh), water flow, and potential evapotranspi-
ration, freeze–thaw activities in the soil and snow insulation. The
crop growth sub-model simulates crop biomass accumulation and
partitioning by tracking photosynthesis, respiration, water and N
74 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83
demand, C allocation, crop yield, and litter production. The
decomposition sub-model partitions soil organic C (SOC) into
four pools: plant residue, microbial biomass, and active and
passive humus, with each pool consisting of one or more sub-pools
with different properties. It calculates decomposition rate,
substrates of NH4+ and dissolved organic C (DOC) and CO2
production. The second component which consists of the
nitrification, denitrification and fermentation sub-models, pre-
dicts the effects of soil and environmental factors on the growth of
nitrifier and denitrifier microbes, and resulting emissions of CO2,
CH4, NO, N2O and N2 from the plant–soil systems. Microbial
populations are considered to be homogenous for both autotrophs
and heterotrophs. Ammonia volatilization is estimated as a
function of NH3 concentration in soil gas phase and soil air-filled
porosity. Nitrogen gas emissions are regulated by an “anaerobic
balloon” concept whereby the redox potential (Eh), determined
using the Nerst equation (Li et al., 2000), regulates the size of the
aerobic (nitrification) and anaerobic (denitrification) fractions. The
nitrification sub-model predicts the growth and death of nitrifiers,
as well as the nitrification rate and N2O production from the
nitrification process as a function of water filled pore space and the
quantity of N nitrified. In the denitrificiation submodel, the
quantity of denitrifier-bacteria is estimated using a multi-nutrient
dependent (Michaelis–Menten) growth function dependent on
temperature, DOC, soil water, Eh, and pH. Denitrification is
calculated as stepwise transformation process as a function of
microbial growth and pH.
2.3.2. Regionalized DNDC improvements
In this study, a regionalized Canadian version of DNDC
(DNDCv.CAN1) has been utilized in order to better represent
crop production, soil C and GHG emissions for cool weather
conditions (Grant et al., 2015; Kröbel et al., 2011; Smith et al.,
2013). Our regionalized version of DNDC has been under
development for the last several years, similar to the family of
other DNDC model versions developed in Germany, New Zealand,
and the UK (Brown et al., 2002; Giltrap et al., 2008; Haas et al.,
2013). In the Canadian model version, the empirical growth
curves, which regulate N and water demand, were modified for
cool weather cultivars of corn, soybeans (Glycine max L.) and
wheat (Grant et al., 2015; Kröbel et al., 2011). Note that the default
growth curves are still available for use and the model can be used
in warmer climates. A deep N pool was conceptualized to
represent N stored below the 50 cm depth (similar to the deep
water pool) and the holding capacity of the deep water pool was
adjusted based on soil properties (Grant et al., 2015). Using a
previously published study in Canada (Yan and Hunt, 1999), the
temperature stress on crops was reformulated to be based on
cardinal temperatures (Smith et al., 2013). The effect of exposure
to high temperature during anthesis on harvest index and new
research regarding the effects of CO2 fertilization on crop growth
was incorporated (Smith et al., 2013). Recently, we reformulated
the transpiration algorithms in DNDC to be based on daily water
demand requirements of crop production and a crop coefficient
approach was employed to adjust potential evapotranspiration
estimated through the Penman–Monteith equation. Our develop-
ment approach is such that all developments made by our group
are shared with the model author so that they can later be
incorporated into future DNDC releases. This regionalized
Canadian version is openly available, upon request.
2.4. Model initialization and calibration
2.4.1. Initialization
Seven years of measurements at the TGAS-Man site were used
to evaluate the model for simulating N2O emissions. The model
simulation was carried out using site specific climate data (Tmin,
Tmax, daily precipitation and solar radiation), soil parameters
(bulk density, texture, organic carbon content, pH), and agronomic
practices employed at the site (Table 1). External to the model we
employed pedotransfer functions described by Saxton and Rawls
(2006) to estimate field capacity and wilting point as measure-
ments were not available. A spin-up period with a 10 year duration
was simulated prior to establishment of the experimental study to
stabilize the partitioning of the C&N pools in DNDC, in agreement
with previous modelling assessments (Smith et al., 2013, 2010,
2008). Soil nitrogen and soil water cannot be initialized to match
the first measurement as the model maintains an annual mass
balance of these components and therefore derives its own
estimate.
2.4.2. Calibration
As DNDC aims to characterize a full mass balance of water and
nutrients as well as describing the complete C&N cycling between
the plant/soil/atmosphere it is often prudent to try to minimize
the influence of the plant interactions when investigating soil
trace gas processes. In doing so estimates of crop production can
only be assessed from the standpoint of their potential impact in
contributing to differences between measured and estimated
nitrogen and water balances and their subsequent effect on
estimates of trace gas emissions. Our recent modeling studies
were focused on model development and parameterization for
soils under cool weather conditions in Canada (Grant et al., 2015;
Kröbel et al., 2011; Sansoulet et al., 2014; Smith et al., 2013). The
model developments from these published studies provided an
independently parameterized model for our current study.
Additionally it should be noted that the trace gas model
framework in DNDCv.CAN1 is identical to the DNDC95 release
version. As previously explained it was deemed necessary to
adjust crop growth rate relevant to cultivars grown at the TGAS-
Man site. This was accomplished by adjusting the potential
biomass for each crop such that average modeled values over the
7 year period were similar to averaged measured values. It was
also required that we lower the GDD0 of corn from 2550 to
2200 to account for the cooler weather cultivar employed in this
study in comparison to the corn that we previously parameter-
ized for eastern Canada. All other crop, soil, management and
climate parameters were not modified. Based on these calibra-
tions only the model’s predictive ability of the effects of
management on soil climate, soil N dynamics and N2O emissions
can be assessed.
2.5. Model validation using statistical analysis
The model accuracy in predicting soil water, soil NH4+, soil NO3

and N2O emissions was evaluated by calculating the root mean
square error (RMSE (%) (Eq. (1)): Wallach et al. (2006)), and the
average relative error (ARE (%) (Eq. (2)): Mayer and Butler, 1993).
The RMSE estimates the scatter between simulated and measured
data, and ARE determines the bias between simulated and
measured values.
" #0:5

1
X
N

1
RMSEð%Þ ¼ N  ðPi
Oi Þ2 O  100 (1)
i¼1
X
N
ðPi
Oi Þ
i¼1
AREð%Þ ¼  100 (2)
NO
 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83 75

Fig. 1. Cumulative precipitation vs. cumulative GDD during the growing seasons.

where O denotes observed values at a point in time i,O is the Soil temperature during the growing season was slightly under-
average of the observed values, P denotes predicted values at a predicted in most years (Table 2). DNDC under-predicted soil
point in time i, and N is the number of observations. temperature by 5% for the P system and by 8% for the A system.
All statistical analyses were performed using actual daily The RMSE or scatter between measured and predicted values was
measurements with no interpolation of the time series. Soil small at about 9% on average for the P system and 11% for the A
temperature and soil moisture were analyzed using continuous system. Smith et al. (2008) found that DNDC89 also under-
and near continuous daily measurements and reported on a mean estimated soil temperature for a clay loam soil by 5–8% but was
annual basis. Measurements that occurred when the soil was more accurate with an ARE of 2% for a lighter textured loam soil. In
frozen or where it was apparent that the probes had lost contact
were disregarded from the statistical assessment for soil moisture.
Soil moisture statistics were also calculated for dry, normal and
wet growing seasons (Fig. 1) with dry and wet conditions defined
as when the cumulative precipitation was either below the 20th or
above the 80th percentile of the 30-year normal (1971–2000)
(Environment Canada, 2012) cumulative growing season precipi-
tation, respectively (Folland, 1999; Utset et al., 2006). Soil NH4+ and
NO3
measurements were infrequent and compared as point in
time measurements for the statistical assessment and aggregated
across the distinct management periods (tillage, annual, perennial
etc.). Nitrous oxide measurements were mostly continuous and
were statistically compared daily for both management and annual
assessment periods. Finally cumulative assessments of nitrous
oxide emissions were compared annually between measurements,
DNDC and Canadian specific IPCC Tier II methodology (Rochette
et al., 2008b) for estimating N2O with above ground and below
ground residue and their respective N contents estimated from
values reported in Janzen et al. (2003).

3. Results and discussion

3.1. Measured and simulated temporal soil conditions

3.1.1. Soil temperature

Fluctuations in measured soil temperature were better
simulated by the DNDC model during the growing season
(Fig. 2) than in the fall and winter. Although the timing of
freeze and thaw events was generally well simulated, the model
tended to under-predict cold periods indicating that discrep-
ancies in estimated snow thickness or snow insulation effects Fig. 2. Measured and estimated soil temperature at the 0.1 m depth in P1: (a) 2006,
existed. (b) 2009.
76 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83

Table 2 it was 22% for the Smith et al. (2008) study in humid eastern
Measured and estimated soil temperature ( C) statistics at 0.1 m depth during the
Canada where preferential flow due to cracks in clay soil was likely
growing season (May 1–October 31).
not a factor. The RMSE in wet and normal years for this study was
2006 2007 2008 2009 2010 2011 2012 on average 20%.
P system
Mean measured ( C) 18.3 14.7 13.9 13.2 13.2 14.4 15.7 3.1.3. Soil nitrogen
Mean modelled ( C) 16.7 13.8 12.6 12.6 13.6 14.0 14.5
Measurements indicated that over the seven year period
ARE (%)
9.0
6.3
8.9
4.4 2.6
2.3
7.7
RMSE (%) 11.2 9.2 12.1 11.0 8.5 9.4 1.8
(Figs. 3 and 4 Table 4) there was on average similar levels of
total mineral N occurring in P and A systems, however, there were
A system temporal differences in N levels in response to fertilizer application
Mean measured ( C) 17.9 15.3 13.0 13.9 14.3 15.8 16.7 in the A system, particularly in 2011 when anhydrous ammonia
Mean modelled ( C) 16.4 14.3 12.2 12.6 13.7 14.0 14.9
was applied during both spring and fall to A1 and not to A2. Under
ARE (%)
8.6
6.9
6.8
9.2
4.1
10.9
10.7
RMSE (%) 10.9 9.6 11.4 13.8 13.9 15.1 2.3 alfalfa production soil NH4+ tended to be higher than soil NO3–
levels which is not surprising considering that NH3 is produced in
nodules which can contribute to the NH4+ pool. These results are in
DNDC the thermal conductivity and heat capacity of the soil are not agreement with reports of Rochette et al. (2004) in which increases
currently dependent on soil texture. Another consideration is that in soil NH4+ was observed under alfalfa production in comparison
soil water content for this site was generally over-predicted by to timothy production. In P1 there was low soil N occurring in
DNDC and this could dampen soil temperature in the summer 2011 likely due to low mineralization rates and low rate of N
months. In DNDC, changes in soil temperature are estimated as a fixation during very dry conditions. Also of interest is that soil N did
function of the change in heat flux from layer to layer divided by not appear to increase substantially immediately after plow down
the soil volumetric heat capacity. of alfalfa (Fig. 3).
The statistical analyses using all measurements over the seven
3.1.2. Soil water content year period indicated that the model overestimated soil NH4+ by
The clay soil at this site was subject to cracking under dry 20%, underestimated soil NO3– by 9% and overestimated total
conditions which imposed technical challenges for the soil mineral N by 2% (Table 4). Since the model underestimated soil
moisture probes used for measurements. The probes appeared temperature, NH3 volatilization and the rate of nitrification from
to sometimes lose contact with the soil resulting in zero readings NH4+ to NO3
could have been underestimated. Owing to the
during prolonged dry periods. Readings smaller than 0.05 cm3/cm3 NH4+/NH3 equilibrium, lower NH3 volatilization estimates would
were not used in the statistical comparisons. For all periods where result in higher estimates of soil NH4+ compared to the measured.
prolonged dry conditions were observed, the model over-predicted The magnitude of NH4+, NO3
and total mineral soil N were all
soil moisture contents (Table 3) mainly attributed as a result of not very well simulated under intensive tillage in 2006 and 2007,
characterizing soil cracking induced preferential flow. During whereas under minimum tillage NH4+ was over-predicted, NO3

wetter years (Table 3) and periods with a higher frequency of was under-predicted but total N was accurately predicted (Table 4).
precipitation events, there was a better comparison between Note that the total mineral soil N for both measured and modeled
simulated and measured values. were greater in the minimum tillage than under intensive tillage.
On average for all plots, the over-prediction of soil water While DNDC predicted a higher gross N mineralization under
content by DNDC was 11.5% in dry seasons, 5.4% in normal seasons intensive tillage than under minimum tillage (200 vs. 183 kg N
and 1.3% in wet seasons. In 2011, a year with a dry growing season, ha
1 y
1) this was offset by higher rates of N immobilization under
DNDC over-predicted soil water content by 15 and 21% in the P and intensive tillage (139 kg N ha
1) than under minimum tillage
A systems, respectively. There was clear indication in that year that (110 kg N ha
1).
preferential flow through clay cracks occurred whereby water There was some discrepancy between the model estimates and
content dropped below 10% and there was little response to rainfall soil N measurements in the perennial system. Soil N was under-
events. We investigated adding a routine in DNDC whereby clay predicted in the first 3 years of alfalfa production and over-
cracking would occur at low water contents resulting in bi-modal estimated in the last year in 2011 (Fig. 3). In 2011 measurements
flow similar to the concept used by Fredlund et al. (2010) but it was were low perhaps because there was less mineralization occurring
deemed that there was not enough field data on the timing and in the only dry year that alfalfa was grown. DNDC overestimated
extent of clay cracking to formulate an appropriate algorithm. The soil water content in 2011 by 15% and thus the simulated
average RMSE of soil water content for this study was 27% whereas mineralization rate was likely too high. The gross N mineralization

Table 3
Measured and predicted volumetric soil water content statistics at 0.05 m depth during the growing season (May 1–October 31).

2007 2008 2009 2010 2011 2012 Dry Normal Wet

(dry) (normal) (normal) (wet) (dry) (dry)
P system
Mean measured VMC (m3 m
3) 0.29 0.34 0.37 0.40 0.28 0.32 0.30 0.36 0.40
Mean modelled VMC (m3 m
3) 0.33 0.37 0.38 0.38 0.33 0.35 0.34 0.37 0.38
ARE (%) 13.8 7.7 0.9
4.5 15.4 7.6 12.3 4.3
4.5
RMSE (%) 25.9 23.7 20.7 16.6 44.3 35.6 35.3 22.2 16.6

A system
Mean measured VMC (m3 m
3) 0.34 0.35 0.32 0.37 0.29 0.30 0.31 0.34 0.37
Mean modelled VMC (m3 m
3) 0.35 0.35 0.36 0.40 0.35 0.33 0.34 0.36 0.40
ARE (%) 0.54 0.97 12.1 7.2 21.2 10.6 10.8 6.5 7.2
RMSE (%) 17.9 12.6 27.5 22.3 42.9 39.5 33.4 20.0 22.3
 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83
Fig. 3. Comparison of observed (Obs) and DNDC estimated soil ammonium (NH4+), nitrate (NO3
) and total inorganic soil nitrogen (NH4+ + NO3
) in the 0–0.3 m soil layer for
P1 and P2: (A) NH4+, (B) NO3
, and (C) total mineral N. Error bars show standard errors of the measured mean. N fertilizer application timings and amounts are denoted by
vertical arrows and numbers, respectively (kg N ha
1).
simulated by DNDC was on average at 92 kg N ha
1 y
1 for the first
three alfalfa years (2008–2010) but increased to 153 kg N ha
1 y
1
in 2011.
The largest overestimation of soil NH4+ of 86% occurred after fall
application of 160 kg N ha
1 of anhydrous ammonia, however,
during this period soil NO3 was underestimated by 12% and the
total simulated mineral N was overestimated by 10% (Table 4;
Fig. 4). In the spring, only 100 kg N ha
1 was applied and
simulations were more accurate but soil NH4+ was still over-
estimated by 20%.
Modeled soil N was most under-predicted in the annual crop
system, likely caused by higher rates of N leaching which may not
have occurred in the field. Leaching as estimated by DNDC was on
average 24 kg N ha
1 greater under the annual cropping than the
perennial alfalfa from 2008 to 2011.
Overall, soil N was well predicted in this study, showing
sizeable improvements over other Canadian studies for which
DNDC under-predicted soil N at sites under semi-arid, sub-humid
and humid climate (Grant et al., 2015; Kröbel et al., 2011; Smith
et al., 2008). This was primarily the result of modifying the passive
humus fraction from 90 to 80% to better reflect the partitioning
for soils in Canada, a value which was used in the development of
the Campbell empirical soil organic carbon model (Campbell et al.,
2001). Although the lower passive fraction resulted in an increase
in mineralization rate, soil carbon levels remained stable for the
duration of our study.
3.2. Measured and simulated N2O fluxes
Predicted and measured daily N2O fluxes for all the treatments
during the study years (2006–2012) along with precipitation are
77
shown in Fig. 5 (P1 and P2) and Fig. 6 (A1 and A2). As expected, N2O
emission events occurred after fertilizer application, and often
coincided with the timing of rainfall events when sufficient soil
substrates were available. Note that the measured detection limit
was low at 1.21 g N2O-N ha
1 d
1. When aggregated over the entire
seven year experimental period, and by only considering days with
measurements from all plots without any interpolation of the
time-series, the DNDC model estimated a cumulative N2O
emission of 21.3 kg N2O-N ha
1 on average for the P system
(P1 and P2) and 30.0 kg N2O-N ha
1 for the A system (A1 and A2),
whereas measured values were 26.6 kg N2O-N ha
1 for P system
and 32.5 kg N2O-N ha
1 for the A system. Simulated total N2O
emissions from A1 and A2 were close to measured values (
1 and

13%, respectively) (Fig. 6), whereas simulated emissions from
P2 were also close to measured (+5%), however, emissions from
P1 were under-estimated by 33% attributed largely to the under-
prediction of one very large emission event in 2012 (Fig. 5A). The
difference in P1 and P2 measurements was explained as a result of
micro topography and poor drainage for P1 (Glenn et al., 2012). The
DNDC model only simulates to 50 cm soil depth thus did not
characterize soil hydrology in the deeper profile.
For comparison, we also estimated N2O emissions for the seven
year study using Tier II IPCC methodology for Canada (Rochette
et al., 2008b). To compare to annual Tier II estimates we present the
total estimated annual N2O emissions from DNDC and scaled total
measured emissions based on the ratio of total annual modeled
emissions to modeled emissions at time of measurements (Fig. 7).
The N2O emissions estimated using Tier II were found to be low at
only 9.8 kg N2O-N ha
1 over seven years for the A systems and
8.1 kg N2O-N ha
1 for the P systems. The methodology under-
estimated N2O emissions by 75% at this site whereas DNDC
78 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83

Fig. 4. Comparison of observed (Obs) and DNDC estimated soil ammonium (NH4+), nitrate (NO3
) and total inorganic soil nitrogen (NH4++ NO3
) in the 0–0.3 m soil layer for
A1 and A2: (A) NH4+, (B) NO3
, and (C) total mineral N. Error bars show standard errors of the measured mean.

Table 4
Effects of management on measured and predicted soil NH4+, NO3
and total soil N concentrations.

n (days) ARE (%) RMSE (g N ha
1) RMSE (%) Mean measured Mean modelled
(kg N ha
1) (kg N ha
1)
NH4+
Intensive tillage 23 3 28 86 32 33
Minimum tillage 23 50 38 127 30 45
Annuals 52
10 38 114 33 30
Perennials 50
17 45 139 33 27
Fall application 19 86 36 177 20 38
Spring application 19 20 23 99 23 28
A system 84 3 32 106 31 31
P system 84 37 54 173 31 43
All management 170 20 44 143 31 37

NO3

Intensive tillage 23 0 41 85 48 48
Minimum tillage 23
18 45 85 53 44
Annuals 52
30 33 87 38 27
Perennials 50 1 24 92 26 26
Fall application 19
12 70 99 71 62
Spring application 19
1 54 79 68 67
A system 86
8 41 81 51 47
P system 88
11 35 74 47 41
All management 176
9 38 78 49 45

Total N
Intensive tillage 23 2 55 67 82 83
Minimum tillage 23 6 54 64 85 90
Annuals 50
23 46 64 72 56
Perennials 50
8 57 96 59 54
Fall application 19 10 58 64 91 100
Spring application 19 4 60 66 91 95
A system 82
5 53 64 83 79
P system 84 8 59 74 79 85
All management 168 2 56 69 81 83
 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83
Fig. 5. Precipitation, measured and DNDC estimated N2O fluxes over the seven year study: (A) P1, (B) P2, and (C) precipitation. N fertilizer application timings and amounts are
denoted by vertical arrows and numbers, respectively (kg N ha
1).
underestimated emissions by 11%. Tier II methodology was
designed to perform well on averaged conditions across large
areas, however, discrepancies at the field scale can be expected.
Although the current methodology does include climate effects, it
does not include spring thaw emissions and the influence of soil
texture. Nitrous oxide emissions can be very high from clay soils
that are high in organic carbon levels. Rochette et al. (2008a) found
that emissions from a heavy clay soil in Quebec were 12–45 kg
N2O-N ha
1 during a 3 year study and concluded that such
emissions were likely a result of denitrification from decomposition
of large quantities of organic matter rather than being associated
with N inputs. According to Li et al. (1996, 2001),), SOC is the most
sensitive factor for simulating N2O emissions using DNDC. The
model predicts greater rates of decomposition in clay soil where
SOC is high resulting in a high rate of decomposition and release of
greater quantities of substrates that enhance denitrification.
3.2.1. Seasonal emissions of N2O
For both the perennial and annual rotations (2008–2011) DNDC
estimates agreed very well with the proportion of total emissions
occurring during the spring thaw period, the growing season, and
the year-end period (Table 5). The average overall measured
emissions in the spring (January–April), growing season (May–
October), and year end (November–December) period were 18%,
80% and 1%, respectively, whereas the corresponding modeled
emissions were 17%, 81% and 2%, respectively. Both measured and
modeled N2O emissions were very small in the months of
November and December even though water content was high,
soil N was available and soil temperature was often at a similar
level as during the spring thaw period. This is an indicator that a
79
freeze period followed by a thaw does promote N2O emissions. The
spring thaw component of DNDC93 was reformulated by
Kariyapperuma et al. (2011) to increase emissions during thaw
events from newly produced denitrification in the surface layer
due to availability of microbial populations and substrates that
remained dormant during the freeze period. This routine, which is
also utilized by DNDCv.CAN1 performed well for the TGAS-Man
site. When averaging across years (2008–2011) for perennial and
annual rotations, N2O emissions during the spring period were
well simulated on a monthly basis (January–April; Table 5).
Average percent of spring thaw emissions using DNDC in a national
study was reported to be 30% (Smith et al., 2004), however, a
higher proportion of spring thaw emissions generally occurs in
eastern Canada (Rochette et al., 2008b; Wagner-Riddle et al., 2007,
1997; Wagner-Riddle and Thurtell, 1998). Our current version of
the DNDC model matched the magnitude and timing of the two
largest spring thaw events observed on April 18, 2007 (8–11 days
after soil thawing) and March 16th 2012 (1–3 days after thawing
(Figs. 5 and 6). Several authors (Abdalla et al., 2009; Parkin and
Kaspar, 2006; Wolf et al., 2010) found that peak emission events
can account for approximately 50–90% of the yearly emissions and
focus should be placed on accurately simulating peak emissions.
During the growing season the model predicted that the highest
monthly N2O emissions occurred in June, however, measurements
indicated that half of the yearly emissions occurred in June,
followed by small emissions during July and August, presumably
when soil moisture was low due in part to preferential flow
through clay cracks (Table 5). As a result the model sometimes
simulated more frequent N2O emissions peaks during the growing
season than was observed.
 80
K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83
Fig. 6. Precipitation, measured and estimated N2O fluxes over the seven year study: (A) A1, (B) A2, and (C) precipitation.
 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83 81

tillage due to wetter soil conditions resulting in higher rates of

denitrification (Helgason et al., 2005; Kariyapperuma et al., 2011;
Rochette et al., 2008b; Smith et al., 2010). The emissions at our
TGAS-Man site were similar between tillage treatments, which
might be expected since precipitation levels were moderate.

3.2.2.2. N2O emissions from perennial alfalfa and annual

Fig. 7. Cumulative N2O emissions at the Glenlea site based on observations, DNDC,
and IPCC Tier II methodology (2006–2012).
In several studies, it has been found that the DNDC model did
well in simulating the overall magnitude of N2O emissions but did
not do well in simulating the timing of individual events (Saggar
et al., 2004; Smith et al., 2008). In a previous study in Canada using
DNDC89 (Smith et al., 2008), the RMSE of N2O emissions ranged
from 28 to 121 g N2O-N ha
1 d
1 whereas, in this study DNDC
achieved a lower RMSE ranging from 26 to 89 g N2O-N ha
1 d
1.
Nitrous oxide emissions are known to be highly variable in space,
time and magnitude and this presents challenges in simulating
emissions (Smith et al., 2008). This is particularly true when
simulating with a 1-D model that does not include lateral flow of
water and nutrients, assumes a heterogeneous profile and assumes
a homogenous microbial population each for autotrophs and
heterotrophs.
cropping. Measurements indicated that N2O emissions were
more than 4 times higher under annual cropping than under
alfalfa production (Table 6). Modeled emissions were also
considerably higher under annual than perennial (81% higher);
however, DNDC over-predicted emissions from alfalfa.
Interestingly, in a study in eastern Canada, Rochette et al.
(2004) found that soil N was up to 10 times greater under
alfalfa than under timothy grass (Phleum pretense L.), however, soil
N was not closely related to N2O emissions and emissions were
greater under alfalfa in only 6 out of 10 field comparisons. It is
possible that in an established stand of alfalfa, interception of
rainfall by the canopy and uptake by a dense rooting system may
limit soil water content to levels where significant production of
N2O through denitrification could not occur (Brady and Weil,
2002). In this study soil moisture was over predicted by 5% on
average across the four alfalfa years (Table 4). Interception of
precipitation may need to be better characterized in DNDC.
Both measured and modeled results demonstrated agreement
in how the plow down of alfalfa did not result in any significant
emission events in either the fall or winter; however, large
emissions occurred the following year under corn production in
2012 with only moderate amounts of fertilizer application. A
similar situation was observed in eastern Canada where Wagner-
Riddle et al. (1997) measured high N2O emissions in the spring
following the plow down of alfalfa the previous autumn.

3.2.2. Effect of management on N2O emissions
3.2.2.1. N2O emissions from intensive and minimum tilled
systems. The magnitude of measured N2O emissions was
similar between the intensive and minimum tillage treatments
in 2006 and 2007 (Table 6). Studies in Canada indicate that more
N2O emissions usually occur under conventional tillage than
reduced tillage in dryer semi-arid and sub-humid soils due to
higher rates of nitrification under a more aerated soil, and in
eastern humid soils more emissions usually occur under reduced
3.2.2.3. N2O emissions from fall and spring applied anhydrous
ammonia. For the spring vs. fall anhydrous ammonia
experiment modeled emissions agreed reasonably well with
measured emissions in the fall (ARE
17%), however, DNDC
underestimated spring emissions by 55% (Table 6). This was
primarily because DNDC underestimated the large emission event
that occurred in June of 2012 during a period of high rainfall shortly
after fertilizer application (Fig. 6b; A2). The DNDC model, having a
cascade flow routine which drains the profile to field capacity,
likely underestimated soil water content and denitrification events

Table 5
Percentage of total N2O emissions per month and season during the experimental period (2006–2012).

P system A system Average precipitation (mm) Average temperature ( C)

Percentage of total N2O emission Percentage of total N2O emission

Month Observed DNDC Observed DNDC

January 0.3 0.7 0.2 0.0 5.7
15.4
February 0.5 0.6 0.3 0.0 10.1
15.5
March 6.4 8.3 7.7 3.5 22.5
5.4
April 9.1 8.6 12.2 11.9 19.3 5.4
May 27.2 24.0 16.0 25.2 78.0 10.8
June 49.3 26.6 52.4 35.0 73.1 16.6
July 3.2 4.9 5.5 3.6 37.5 19.5
August 3.7 10.5 2.6 11.0 68.9 18.3
September
0.8 8.5 0.7 6.0 42.5 13.4
October
0.1 4.6 0.9 2.7 36.2 5.6
November 1.1 2.1 0.9 1.1 22.5
3.6
December 0.3 0.6 0.6 0.0 8.6
14.1

Season
January–April 16.2 18.2 20.5 15.5
Growing season 82.4 79.1 78.0 83.5
November–December 1.4 2.7 1.5 1.1
82 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83
Table 6
Comparison of the effect management on measured versus predicted soil N2O emissions.
Management Period n (days)
Intensive tillage 27 April, 2006–16 May, 2008 1065
Minimum tillage 27 April, 2006–16 May, 2008 1021
Annuals 21 May, 2008–2 May, 2012 2010
Perennials 28 May, 2008–2 May, 2012 2040
Fall application 7 October, 2010–18 October, 2012 640
Spring application 7 October, 2010–18 October, 2012 643
A system 1 January, 2006–31 December, 2012 3660
P system 1 January, 2006–31 December, 2012 3621
during this period. A more detailed water flow model could be
considered in DNDC; however, this would increase input
requirements.
It is of importance to consider the contribution of all reactive N
species when evaluating the effectiveness of a specific manage-
ment option towards reducing environmental impact. The DNDC
model predicted 63% more N leaching for fall applied anhydrous
ammonia as compared to spring application and 101% more NH3
emissions for spring application relative to fall application. To
facilitate model development it would be useful if additional
experimental evidence was gathered across varying climatic
conditions including more reactive N species.
4. Conclusions
In this study, the DNDC model slightly underestimated soil
temperature, primarily during winter months, in both the
perennial and annual systems. Conversely, the model generally
overestimated soil water content across the management systems,
with the greatest overestimation occurring during dry conditions.
The over-prediction was attributed to the inability of DNDC to
characterize preferential flow through clay cracks. In general,
DNDC was able to emulate the temporal trends of the measured
soil N across differing management regimes, however, some
discrepancy in the predicted mineralization rate and nitrification
rate were apparent in the comparisons. These discrepancies could
have been caused by inaccuracies in estimating soil environmental
conditions.
The DNDC model performed well in simulating the overall
magnitude of N2O emissions from the A and P systems (ARE
8%
and
20%, respectively) but the model overestimated emissions
from alfalfa production and underestimated emissions after spring
applied anhydrous ammonia. The model tended to underestimate
emissions during long periods of episodic rainfall due to the
inability of a cascade flow model to simulate water content above
field capacity effectively. The model was able to predict the
proportional seasonal contribution of N2O emissions towards the
annual flux. Overall the DNDC proved effective in characterizing
general influences of management on nitrous oxide emissions,
however, improvements to the model’s soil hydrology framework
and also to the mechanisms describing thermal insulation of snow
and heat transfer through the soil profile should be considered for
further studies.
References
Abdalla, M., Jones, M., Smith, P., Williams, M., 2009. Nitrous oxide fluxes and
denitrification sensitivity to temperature in Irish pasture soils. Soil Use Manage.
25, 376–388. doi:http://dx.doi.org/10.1111/j. 1475-2743.2009.00237.x.
Brady, N., Weil, R., 2002. The Nature and Properties of Soils, 13th ed. Prentice Hall,
Upper Saddle River, New Jersey, USA.
Brown, L., Syed, B., Jarvis, S.C., Sneath, R.W., Phillips, V.R., Goulding, K.W.T., Li, C.,
2002. Development and application of a mechanistic model to estimate
Measured Modeled ARE (%) RMSE
(kg N2O-N ha
1) (kg N2O-N ha
1) (g N2O-N ha
1 d
1)
6.1 6.7 11 43
6.0 6.2 4 56
15.3 17.6 15 59
3.6 9.7 169 26
12.5 10.3
17 52
18.8 8.5
55 89
32.5 30.1
8 61
26.6 21.2
20 74
emission of nitrous oxide from UK agriculture. Atmos. Environ. 36, 917–928.
doi:http://dx.doi.org/10.1016/S1352-2310(01)00512-X.
Campbell, C.A., Selles, F., Lafond, G.P., Zentner, R.P., 2001. Adopting zero tillage
management: impact on soil C and N under long-term crop rotations in a thin
Black Chernozem. Can. J. Soil Sci. 81, 139–148.
Dalal, R.C., Wang, W., Robertson, G.P., Parton, W.J., 2003. Nitrous oxide emission
from Australian agricultural lands and mitigation options: a review. Soil Res. 41,
165–195.
Del Grosso, S., Halvorson, A., Parton, W., 2008. Testing DAYCENT model simulations
of corn yields and nitrous oxide emissions in irrigated tillage systems in
Colorado. J. Environ. Qual. 37, 1383–1389. doi:http://dx.doi.org/10.2134/
jeq2007.0292.
Denmead, O., Raupach, M., et al., 1993. Methods for measuring atmospheric gas
transport in agricultural and forest systems. In: Harper, L.A. (Ed.), Agricultural
Ecosystem Effects on Trace Gases and Global Climate Change. ASA, CSSA and
SSSA, Madison, Wisconsin, USA, pp. 19–44.
Environment Canada, 2012. Canada’s National Climate Archive. http://www.
climate.weatheroffice.ec.gc.ca/.
FAO, 2014. World Reference Base for Soil Resources 2014 International Soil
Classification System for Naming Soils and Creating Legends for Soil Maps. FAO,
Food and Agriculture Organization, Rome.
Folland, 1999. Workshop on indices and indicators for climate extremes, Asheville,
NC, USA, 3–6 June 1997 breakout group C: temperature indices for climate
extremes. Clim. Change 42, 31–43.
Fredlund, D.G., Houston, S.L., Nguyen, Q., Fredlund, M.D., 2010. Moisture movement
through cracked clay soil profiles. Geotech. Geol. Eng. 28, 865–888. doi:http://
dx.doi.org/10.1007/s10706-010-9349-x.
Frolking, S.E., Mosier, A.R., Ojima, D.S., Li, C., Parton, W.J., Potter, C.S., Priesack, E.,
Stenger, R., Haberbosch, C., Dörsch, P., Flessa, H., Smith, K.A., 1998. Comparison
of N2O emissions from soils at three temperate agricultural sites: simulations of
year-round measurements by four models. Nutr. Cycl. Agroecosyst. 52, 77–105.
doi:http://dx.doi.org/10.1023/A.1009780109748.
Giltrap, D., Saggar, S., Li, C., Wilde, H., 2008. Using the NZ–DNDC model to estimate
agricultural N2O emissions in the Manawatu–Wanganui region. Plant Soil 309,
191–209. doi:http://dx.doi.org/10.1007/s11104-007-9527-7.
Glenn, A.J., Amiro, B.D., Tenuta, M., Stewart, S.E., Wagner-Riddle, C., 2010. Carbon
dioxide exchange in a Northern Prairie cropping system over three years. Agric.
For. Meteorol. 150, 908–918. doi:http://dx.doi.org/10.1016/j.
agrformet.2010.02.010.
Glenn, A.J., Tenuta, M., Amiro, B.D., Maas, S.E., Wagner-Riddle, C., 2012. Nitrous oxide
emissions from an annual crop rotation on poorly drained soil on the Canadian
Prairies. Agric. For. Meteorol. 166 41–49. doi:http://dx.doi.org/10.1016/j.
agrformet.2012.06.015.
Goglio, P., Bonari, E., Mazzoncini, M., 2012. LCA of cropping systems with different
external input levels for energetic purposes. Biomass Bioenergy 42, 33–42. doi:
http://dx.doi.org/10.1016/j.biombioe.2012.03.021.
Grant, B.B., Smith, W.N., Campbell, C., Desjardins, R.L., Lemke, R.L., Kroebel, R.,
McConkey, B.M., Smith, E.G., Lafond, G.P., 2015. Comparison of DayCent and
DNDC models: Case studies using data from long-term experiments on the
Canadian prairies. In: Del Grosso, S., Parton, B., Lajpat, A. (Eds.), Advances in
Modeling Agricultural Systems: Trans-disciplinary Research, Synthesize,
Modeling, and Applications. ASA-SSSA-CSSA In press.
Haas, E., Klatt, S., Fröhlich, A., Kraft, P., Werner, C., Kiese, R., Grote, R., Breuer, L.,
Butterbach-Bahl, K., 2013. LandscapeDNDC: a process model for simulation of
biosphere–atmosphere–hydrosphere exchange processes at site and regional
scale. Landsc. Ecol. 28, 615–636. doi:http://dx.doi.org/10.1007/s10980-012-
9772-x.
Helgason, B.L., Janzen, H.H., Chantigny, M.H., Drury, C.F., Ellert, B.H., Gregorich, E.G.,
Lemke, R.L., Pattey, E., Rochette, P., Wagner-Riddle, C., 2005. Toward improved
coefficients for predicting direct N2O emissions from soil in canadian
agroecosystems. Nutr. Cycl. Agroecosyst. 72, 87–99. doi:http://dx.doi.org/
10.1007/s10705-004-7358-y.
Janzen, H.H., Beauchemin, K.A., Bruinsma, Y., Campbell, C.A., Desjardins, R.L., Ellert,
B.H., Smith, E.G., 2003. The fate of nitrogen in agroecosystems: an illustration
using Canadian estimates. Nutr. Cycl. Agroecosyst. 67, 85–102. doi:http://dx.doi.
org/10.1023/A:1025195826663.
Jarecki, M.K., Parkin, T.B., Chan, A.S.K., Hatfield, J.L., Jones, R., 2008. Comparison
of DAYCENT-simulated and measured nitrous oxide emissions from a corn
 K.C. Uzoma et al. / Agriculture, Ecosystems and Environment 206 (2015) 71–83
field. J. Environ. Qual. 37, 1685–1690. doi:http://dx.doi.org/10.2134/
jeq2007.0614.
Kariyapperuma, K., Wagner-Riddle, C., Furon, A.C., Li, C., 2011. Assessing spring thaw
nitrous oxide fluxes simulated by the DNDC model for agricultural soils. J. Soil
Sci. Soc. Am. 75, 678–690.
Kröbel, R., Smith, W., Grant, B., Desjardins, R., Campbell, C., Tremblay, N., Li, C.,
Zentner, R., McConkey, B., 2011. Development and evaluation of a new Canadian
spring wheat sub-model for DNDC. Can. J. Soil Sci. 91, 503–520. doi:http://dx.
doi.org/10.4141/cjss2010-059.
Kroeze, C., Mosier, A., Bouwman, L., 1999. Closing the global N2O budget: a
retrospective analysis 1500–1994. Global Biogeochem. Cycles 13, 1–8. doi:
http://dx.doi.org/10.1029/1998GB900020.
Li, C., Aber, J., Stange, F., Butterbach-Bahl, K., Papen, H., 2000. A process-oriented
model of N2O and NO emissions from forest soils: 1. Model development. J.
Geophys. Res. Atmos. 105, 4369–4384. doi:http://dx.doi.org/10.1029/
1999JD900949.
Li, C., Frolking, S., Frolking, T.A., 1992. A model of nitrous oxide evolution from soil
driven by rainfall events: 1. Model structure and sensitivity. J. Geophys. Res.
Atmos. 97, 9759–9776. doi:http://dx.doi.org/10.1029/92JD00509.
Li, C., Narayanan, V., Harriss, R.C., 1996. Model estimates of nitrous oxide emissions
from agricultural lands in the United States. Global Biogeochem. Cycles 10,
297–306. doi:http://dx.doi.org/10.1029/96GB00470.
Li, C., Zhuang, Y., Cao, M., Crill, P., Dai, Z., Frolking, S., Iii, B.M., Salas, W., Song, W.,
Wang, X., 2001. Comparing a process-based agro-ecosystem model to the IPCC
methodology for developing a national inventory of N2O emissions from arable
lands in China. Nutr. Cycl. Agroecosyst. 60, 159–175. doi:http://dx.doi.org/
10.1023/A:1012642201910.
Maas, S.E., Glenn, A.J., Tenuta, M., Amiro, B.D., 2013. Net CO2 and N2O exchange
during perennial forage establishment in an annual crop rotation in the Red
River Valley, Manitoba. Can. J. Soil Sci . 93 (5), 639–652. http://dx.doi.org/
10.4141/cjss2013-25.
Mayer, D.G., Butler, D.G., 1993. Statistical validation. Ecol. Model. 68, 21–32. doi:
http://dx.doi.org/10.1016/0304-3800(93)90105-2.
Mosier, A., Kroeze, C., 2000. Potential impact on the global atmospheric N2O budget
of the increased nitrogen input required to meet future global food demands.
Chemosphere – Global Change Sci. 2, 465–473. doi:http://dx.doi.org/10.1016/
S1465-9972(00)0039-8.
Parkin, T.B., Kaspar, T.C., 2006. Nitrous oxide emissions from corn–soybean systems
in the midwest. J. Environ. Qual. 35, 1496. doi:http://dx.doi.org/10.2134/
jeq2005.0183.
Parton, W.J., Stewart, J.W.B., Cole, C.V., 1988. Dynamics of C, N, P and S in grassland
soils: a model. Biogeochemistry 5, 109–131.
Rochette, P., Angers, D.A., Bélanger, G., Chantigny, M.H., Prévost, D., Lévesque, G.,
2004. Emissions of NO from alfalfa and soybean crops in Eastern Canada. Soil
Sci. Soc. Am. J. 68, 493. doi:http://dx.doi.org/10.2136/sssaj2004.4930.
Rochette, P., Angers, D.A., Chantigny, M.H., Bertrand, N., 2008a. Nitrous oxide
emissions respond differently to no-till in a loam and a heavy clay soil. Soil Sci.
Soc. Am. J. 72, 1363. doi:http://dx.doi.org/10.2136/sssaj2007.0371.
Rochette, P., Worth, D.E., Lemke, R.L., McConkey, B.G., Pennock, D.J., Wagner-Riddle,
C., Desjardins, R.J., 2008b. Estimation of N2O emissions from agricultural soils in
Canada. I. Development of a country-specific methodology. Can. J. Soil Sci. 88,
641–654. doi:http://dx.doi.org/10.4141/CJSS07025.
Rosenzweig, C., Jones, J.W., Hatfield, J.L., Ruane, A.C., Boote, K.J., Thorburn, P., Antle,
J.M., Nelson, G.C., Porter, C., Janssen, S., Asseng, S., Basso, B., Ewert, F., Wallach,
D., Baigorria, G., Winter, J.M., 2013. The agricultural model intercomparison
and improvement project (AgMIP): protocols and pilot studies.
Agric. For. Meteorol. 170, 166–182. doi:http://dx.doi.org/10.1016/j.
agrformet.2012.09.011.
Saggar, S., 2010. Estimation of nitrous oxide emission from ecosystems and its
mitigation technologies. Agric. Ecosyst. Environ. 136, 189–191. doi:http://dx.doi.
org/10.1016/j.agee.2010.01.007.
Saggar, S., Andrew, R.M., Tate, K.R., Hedley, C.B., Rodda, N.J., Townsend, J.A., 2004.
Modelling nitrous oxide emissions from dairy-grazed pastures. Nutr. Cycl
Agroecosyst. 68, 243–255. doi:http://dx.doi.org/10.1023/B:
FRES.0000019463.92440.a3.
Sansoulet, J., Pattey, E., Kröbel, R., Grant, B., Smith, W., Jégo, G., Desjardins, R.L.,
Tremblay, N., Tremblay, G., 2014. Comparing the performance of the STICS,
DNDC, and DayCent models for predicting N uptake and biomass of spring
wheat in Eastern Canada. Field Crops Res. 156, 135–150. doi:http://dx.doi.org/
10.1016/j.fcr.2013.11.010.
Saxton, K.E., Rawls, W.J., 2006. Soil water characteristic estimates by texture and
organic matter for hydrologic solutions. Soil Sci. Soc. Am. J. 70, 1569. doi:http://
dx.doi.org/10.2136/sssaj2005.0117.
83
Scott, N.A., Cole, C.V., Elliott, E.T., Huffman, S.A., 1996. Soil textural control on
decomposition and soil organic matter dynamics. Soil Sci. Soc. Am. J. 60, 1102.
doi:http://dx.doi.org/10.2136/sssaj1996.03615995006000040020x.
Sieling, K., Kage, H., 2006. N balance as an indicator of N leaching in an oilseed rape –
winter wheat – winter barley rotation. Agric. Ecosyst. Environ. 115, 261–269.
doi:http://dx.doi.org/10.1016/j.agee.2006.01.011.
Smith, P., Bustamante, M., Ahammad, H., Clark, H., Dong, H., Elsiddig, E.A., Haberl, H.,
Harper, R., House, J., Jafari, M., Masera, O., Mbow, C., Ravindranath, N.H., Rice, C.
W., Robledo Abad, A., Romanoskaya, A., Sperling, F., Tubiello, F., 2014.
Agriculture, forestry and other land use (AFOLU). In: Edenhofer, O., Pichs-
Madruga, R., Sokona, Y., Farahani, E., Kadner, S., Seyboth, K., Adler, A., Baum, I.,
Brunner, S., Eickemeier, P., Kriemann, B., Savolainen, J., Schlömer, S., von
Stechow, C., Zwickel, T., Minx, J.C. (Eds.), Climate Change 2014: Mitigation of
Climate Change. Contribution of Working Group III to the Fifth Assessment
Report of the Intergovernmental Panel on Climate Change. Cambridge
University Press, Cambridge, United Kingdom and New York, NY, USA.
Smith, W.N., Desjardins, R.L., Grant, B., Li, C., Lemke, R., Rochette, P., Corre, M.D.,
Pennock, D., 2002. Testing the DNDC model using N2O emissions at two
experimental sites in Canada. Can. J. Soil Sci. 82, 365–374. doi:http://dx.doi.org/
10.4141/S01-048.
Smith, W.N., Grant, B.B., Desjardins, R.L., Kroebel, R., Li, C., Qian, B., Worth, D.E.,
McConkey, B.G., Drury, C.F., 2013. Assessing the effects of climate change on crop
production and GHG emissions in Canada. Agric. Ecosyst. Environ. 179, 139–150.
doi:http://dx.doi.org/10.1016/j.agee.2013.08.015.
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 in
Eastern Canada. Can. J. Soil Sci. 88, 251–260.
Smith, W.N., Grant, B.B., Desjardins, R.L., Worth, D., Li, C., Boles, S.H., Huffman, E.C.,
2010. A tool to link agricultural activity data with the DNDC model to estimate
GHG emission factors in Canada. Agric. Ecosyst. Environ. 136, 301–309. doi:
http://dx.doi.org/10.1016/j.agee.2009.12.008.
Smith, W.N., Grant, B., Desjardins, R.L., Lemke, R., Li, C., 2004. Estimates of the
interannual variations of N2O emissions from agricultural soils in Canada. Nutr.
Cycl. Agroecosyst. 68, 37–45. doi:http://dx.doi.org/10.1023/B:
FRES.0000012230.40684.c2.
Snyder, C.S., Bruulsema, T.W., Jensen, T.L., Fixen, P.E., 2009. Review of greenhouse gas
emissions from crop production systems and fertilizer management effects.
Agric. Ecosyst. Environ. 133, 247–266. doi:http://dx.doi.org/10.1016/j.
agee.2009.04.021.
Utset, A., Martínez-Cob, A., Farré, I., Cavero, J., 2006. Simulating the effects of
extreme dry and wet years on the water use of flooding-irrigated maize in a
Mediterranean landplane. Agric. Water Manage. 85, 77–84. doi:http://dx.doi.
org/10.1016/j.agwat.2006.03.010.
Wagner-Riddle, C., Furon, A., Mclaughlin, N.L., Lee, I., Barbeau, J., Jayasundara, S.,
Parkin, G., Von Bertoldi, P., Warland, J., 2007. Intensive measurement of nitrous
oxide emissions from a corn–soybean–wheat rotation under two contrasting
management systems over 5 years. Global Change Biol. 13, 1722–1736. doi:
http://dx.doi.org/10.1111/j.1365-2486.2007.01388.x.
Wagner-Riddle, C., Thurtell, G.W., 1998. Nitrous oxide emissions from agricultural
fields during winter and spring thaw as affected by management practices. Nutr.
Cycl. Agroecosyst. 52, 151–163. doi:http://dx.doi.org/10.1023/
A:1009788411566.
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. doi:http://dx.doi.org/10.4141/S96-103.
Wallach, D., Makowski, D., Jones, J.W., 2006. Working with Dynamic Crop Models:
Evaluation, Analysis, Parameterization, and Applications, first ed. Elsevier,
Amsterdam, Oxford.
Werner, C., Butterbach-Bahl, K., Haas, E., Hickler, T., Kiese, R., 2007. A global
inventory of N2O emissions from tropical rainforest soils using a detailed
biogeochemical model. Global Biogeochem. Cycles 21, GB3010. doi:http://dx.
doi.org/10.1029/2006GB002909.
Wolf, B., Kiese, R., Chen, W., Grote, R., Zheng, X., Butterbach-Bahl, K., 2012. Modeling
N2O emissions from steppe in Inner Mongolia, China, with consideration of
spring thaw and grazing intensity. Plant Soil 350, 297–310. doi:http://dx.doi.
org/10.1007/s11104-011-0908-6.
Wolf, B., Zheng, X., Brüggemann, N., Chen, W., Dannenmann, M., Han, X., Sutton, M.
A., Wu, H., Yao, Z., Butterbach-Bahl, K., 2010. Grazing-induced reduction of
natural nitrous oxide release from continental steppe. Nature 464, 881–884.
doi:http://dx.doi.org/10.1038/nature08931.
Yan, W., Hunt, L.A., 1999. An equation for modelling the temperature response of
plants using only the cardinal temperatures. Ann. Bot. 84, 607–614. doi:http://
dx.doi.org/10.1006/anbo.1999.0955.