Sulaiman Muhammad 201705 PHD

Net Ecosystem Carbon Budget of Dairy Feed Crop Systems
by
Muhammad Firdaus Bin Sulaiman
A Thesis
presented to
The University of Guelph
In partial fulfilment of requirements

for the degree of
Doctor of Philosophy
in
Environmental Sciences
Guelph, Ontario, Canada
c Muhammad Firdaus Bin Sulaiman, April, 2017

ABSTRACT
NET ECOSYSTEM CARBON BUDGET OF DAIRY FEED CROP SYSTEMS
Muhammad Firdaus Bin Sulaiman Advisor:

University of Guelph, 2017 Dr. Claudia Wagner-Riddle
Dairy production emits significant amount of greenhouse gases (GHG) to the atmosphere.
Cultivation of hay and corn for dairy feed have the potential to partially mitigate GHG
emission from dairy production by sequestering plant assimilated carbon in soil as sta-
bilized soil organic matter. The research presented here investigated the net ecosystem
carbon budget (NECB) and greenhouse gas balance (GHGB) of hay and corn over three
years, and compared methods for measuring the gas exchange needed for estimating bud-
get components. Maintaining the sampling height (z) for net ecosystem exchange (NEE)
measurement above the roughness sublayer (RSL) is challenging when measuring over
fast-growing agricultural crops in a multi-plot study, which is often fetch-limited. Net
ecosystem exchange measured over hay and corn using the flux-gradient technique (FG) at
various z were compared to measurements made using the eddy covariance (EC) technique.
Net ecosystem exchange measured using FG were comparable to EC but disagreed when z
was less than 1.4 times the mean canopy height. Soil GHG flux measured using chambers
complements measurement made using micrometeorological methods but chambers lack
in spatial and temporal resolution. Comparison of nitrous oxide (N2 O) and methane (CH4 )
fluxes measured using chambers and micrometeorological methods over corn revealed an
overestimation of cumulative trace gas flux when linear interpolation was used to estimate
daily mean fluxes. Average NECB over three years showed corn was a higher carbon emit-
ter than hay. Components of the NECB were similar on average between hay and corn,
except for the amount of carbon removed from harvest. Emissions of N2 O were higher in
corn than in hay throughout the three years. Higher NECB and N2 O emission ultimately
resulted in corn having higher GHGB than hay. The lower carbon emitted from hay how-
ever came at a cost of lower feed production compared to corn. This research provides new
information on dairy cropping systems that can be used to estimate the carbon sequestration
potential of annual and perennial crops.

ACKNOWLEDGEMENTS
• Thank you to Dr. Claudia Wagner-Riddle for taking me in me as her student, for her
support and guidance and the knowledge that she had passed on to me.
• Thank you to my advisory committee members, Dr. Jon Warland, Dr. Paul Voroney
and Dr. Philippe Rochette for their guidance and advice.
• Thank you to Dr. Shannon Brown for her assistance in processing the raw data,
troubleshooting problems and for the countless favors that I had asked.
• Thank you to the technicians and summer students who have helped me with field
work and instrument maintenance; Jordan Forsyth, Jen Smith, Naomi Steiger, Kieran
Lehan, Taylor Thurtell, Paola Costa, Patricia Braun, Jordan Mayfield, Valerie Free-
mantle and Dr. Eduardo de Figueiredo.
• Thank you to Pedro Machado for volunteering his time and energy helping me out
with field and lab work.
• Thank you to Agricultural Greenhouse Gas Program (AGGP) administered by Agri-

culture and Agri-Food Canada, Dairy Farmers of Canada and NSERC Discovery
Grant for funding this research.
• Thank you to my employer, Universiti Putra Malaysia and the Ministry of Higher
Education Malaysia for giving me the opportunity and financial assistance to further
my studies.
• Thank you to my parents, Dato’ Sri Sulaiman Keling and Datin Sri Zuraini Yaacob
for their love and support.
• Thank you to my children, Arissa Adrianna, Aidan Azril and Azra Annisa for bring-
ing joy to me throughout the stressful journey of completing this thesis.
• Thank you to my wife, Miratul Hada for her love, sacrifice, emotional support and
for standing by me through thick and thin.
iv
TABLE OF CONTENTS
LIST OF TABLES vii
LIST OF FIGURES viii
1 INTRODUCTION 1
1.1 General Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Thesis Format and Objectives . . . . . . . . . . . . . . . . . . . . . . . . . 4
2 MEASUREMENTS OF CO2 FLUX USING FLUX-GRADIENT AND EDDY

COVARIANCE TECHNIQUES 7
2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
2.2 Materials and Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.1 Site description . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
2.2.2 NEE measurement . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.3 Comparison of NEE measured between techniques and statistical
analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.3 Results and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
3 MEASUREMENTS OF N2 O AND CH4 FLUXES USING MICROMETEO-

ROLOGICAL AND CHAMBER TECHNIQUES 30
3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
3.2.2 Chamber method . . . . . . . . . . . . . . . . . . . . . . . . . . . 35
3.2.3 Flux-gradient technique . . . . . . . . . . . . . . . . . . . . . . . 39
3.2.4 Eddy covariance technique . . . . . . . . . . . . . . . . . . . . . . 40
3.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.3.1 N2 O flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.3.2 CH4 flux . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
v
3.3.3 Diurnal trend . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
3.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4 NET ECOSYSTEM CARBON BUDGET AND GREENHOUSE GAS BAL-

ANCE OF ANNUAL AND PERENNIAL DAIRY FEED CROP SYSTEMS 54
4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.2.2 Net ecosystem exchange and N2 O flux measurements . . . . . . . . 61
4.2.3 Net ecosystem carbon budget and greenhouse gas balance . . . . . 66
4.2.4 Plant sampling . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.2.5 Statistical and uncertainty analyses . . . . . . . . . . . . . . . . . 68
4.3 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
4.3.1 Meteorological conditions and soil moisture . . . . . . . . . . . . . 69
4.3.2 Net ecosystem exchange . . . . . . . . . . . . . . . . . . . . . . . 71
4.3.3 Gross primary production and ecosystem respiration . . . . . . . . 75
4.3.4 Yield . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.3.5 Net ecosystem carbon budget and greenhouse gas balance . . . . . 78
4.4 Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
4.4.1 Net ecosystem exchange . . . . . . . . . . . . . . . . . . . . . . . 82
4.4.2 Effects of climate variability . . . . . . . . . . . . . . . . . . . . . 83
4.4.3 Net ecosystem carbon budget . . . . . . . . . . . . . . . . . . . . 85
4.4.4 Greenhouse gas balance . . . . . . . . . . . . . . . . . . . . . . . 90
4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
5 GENERAL CONCLUSION 92
REFERENCES 95
APPENDIX 108
vi
LIST OF TABLES
2.1 Slope of the linear regression and Pearson’s product moment of correlation
coefficient (r) between NEE measured using the EC and FG techniques,
the roughness length at peak growth (z0 ) and maximum mean crop height
(hc ) in corn and hay in 2014 and in corn in 2015. Linear regressions were
performed on the overall, daytime and nighttime NEE. . . . . . . . . . . . 21
2.2 The mean±S.D. enhancement factor (γ) during the growing season of each
crop according to each respective atmospheric stability condition. . . . . . 25
4.1 Soil physical and chemical properties at different depths . . . . . . . . . . 58

4.2 Treatments applied to each plot, manure composition and application rate . 61
4.3 Mean grain corn and hay yield and their respective nitrogen content to av-
erage yield recorded in Ontario (Statistics Canada, 2012). . . . . . . . . . . 78
4.4 Net ecosystem carbon budget±uncertainty of hay and corn from 2012 to
2014 (g C m−2 year−1 ) and its components. Different subscript letters
within each year indicate significant difference between crop systems. . . . 80
4.5 The net greenhouse gas balance±uncertainty and its components. NECB
= net ecosystem exchange + carbonadded + carbonremoved , N2 O = emission
expressed in CO2 equivalent using global warming potential of 298 and
GHGB = NECB + N2 O. Unit = g CO2 -eq m−2 year−1 . . . . . . . . . . . . 81
4.6 Average annual NEE, NECB and carbon input-output from select published
studies (unit: g C m−2 ). Only studies that reported amount of carbon re-
moved were selected for comparison. For multi-crop rotations studies, only
years when corn was planted is included in the table. . . . . . . . . . . . . 87
4.7 Average annual NEE, NECB and carbon input-output from select published
studies (unit: g C m−2 ) made on pasture and grassland. Only studies that
reported NECB and pasture that was not used for grazing were selected for
comparison. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
vii
LIST OF FIGURES
2.1 Aerial view of the study site and schematic of the plot layout and position
of the sonic anemometers, eddy covariance (EC) open path analyzers, flux-
gradient (FG) intakes and sample tubes directing air to the tunable diode
laser trace gas analyzer (TDLTGA). Area bordered by yellow and red lines
are the hay and corn fields respectively. The areas surrounded by white
frames shows the specific area of interest of the study. . . . . . . . . . . . . 12
2.2 Aerial view of the study site and schematic of the plot layout and posi-
tion of the sonic anemometers, eddy covariance (EC) open path analyzers,
flux-gradient (FG) intakes and sample tubes directing air to the tunable
diode laser trace gas analyzer (TDLTGA). Area bordered by red lines are
all planted in corn. The area surrounded by white frames shows the specific
area of interest of the study. . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.3 Scatter plot of NEE measured using the eddy-covariance (EC) and flux-
gradient (FG) techniques in 2014 for hay (A) and corn (B) and corn in
2015 (C). Values are for 30-min periods and inclusive of day and nighttime
NEE. Red line indicate 1:1. . . . . . . . . . . . . . . . . . . . . . . . . . . 19
2.4 Scatter plots of NEE measured using the FG and EC techniques during
daytime in hay (A), corn in 2014 (C), corn in 2015 (E) and nighttime in
hay (B), corn in 2014 (D) and corn in 2015 (F). Red line indicate 1:1. . . . 20
2.5 Frequency of the ratio of the sampling height (z) to mean canopy height
(hc ) in hay (A), corn in 2014 (B) and corn in 2015 (C). . . . . . . . . . . . 23
2.6 Mean canopy height (z) and sampling intake height (hc ) of hay, corn in
2014 and corn in 2015 during the growing season. . . . . . . . . . . . . . . 24
2.7 Scatter plot of NEE measured using the FG and EC techniques over hay in
2014 according to atmospheric stability conditions. . . . . . . . . . . . . . 26
2.8 Scatter plot of NEE measured using the FG and EC techniques over corn
in 2014 according to atmospheric stability conditions. . . . . . . . . . . . . 27
2.9 Scatter plot of NEE measured using the FG and EC techniques over corn
in 2015 according to atmospheric stability conditions. . . . . . . . . . . . . 28
viii
3.1 Schematic of the two study plots (diagram not drawn to scale). Three inter-
rows were selected to the west of the flux–gradient intake of each plot. Four
chambers were deployed in each inter-row. . . . . . . . . . . . . . . . . . . 36
3.2 The frequency of wind speed (m s−1 ) and direction (o ) at the study site
during the hours of 900 to 1200. . . . . . . . . . . . . . . . . . . . . . . . 37
3.3 Mean daily soil N2 O fluxes measured using chamber and the flux-gradient
method in the fall-applied plot (A) and spring-applied plot (B). Bars indi-
cate standard error. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
3.4 Wind rose on day 169 (A) and 176 (B) when high N2 O flux were measured
using chambers. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.5 Mean daily soil CH4 fluxes measured using the eddy covariance method
and mean soil CH4 fluxes measured using chamber. . . . . . . . . . . . . . 46
3.6 Mean diurnal cycle of N2 O fluxes measured in the fall-applied (A) and
spring-applied plot (B) using the FG technique. Bars indicate standard error. 47
3.7 Mean diurnal cycle of CH4 fluxes measured using the EC technique. Bars
indicate standard error. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.8 The flux footprint of the micrometeorological methods during times of
chamber sampling on day 169 in the fall-applied (A) and spring-applied
(C) plots and on day 176 in the fall-applied (B) and spring-applied (D)
plots. Yellow box demarcate the cluster of 12 chambers on each plot. The
outermost contour line indicate 90% of the source area of the measured
flux. Contour lines and flux footprint was generated using the program
Footprint (http://www.geos.ed.ac.uk/homes/rclement/micromet/EdiTools/)
based on the model of Kormann and Meixner (2001). . . . . . . . . . . . . 50
4.1 Aerial view of the study site and schematic of the plot layout and position of
the intakes and sonic anemometers and the tunable diode laser gas analyzer
system (TDLTGAS). Area bordered by yellow and red lines are the hay
and corn fields respectively. The area surrounded by white frames shows
the study plots (Map data: Google, DigitalGlobe). . . . . . . . . . . . . . . 60
4.2 The 30-year normal monthly mean air temperature (dashed line) and monthly
mean air temperature measured during the three study years (A) and the
30-year normal total monthly precipitation (vertical columns) and monthly
total precipitation (B) measured during the three study years. . . . . . . . . 70
4.3 Growing season daily total precipitation and mean volumetric soil water
content (θs ) of the two corn and hay subplots measured at 5 cm during the
three study years. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
4.4 Cumulative NEE of the hay and corn in all study years. . . . . . . . . . . . 73
4.5 Mean monthly diurnal NEE trend for hay (top panels) and corn (bottom
panels) of each study year. . . . . . . . . . . . . . . . . . . . . . . . . . . 74
ix
4.6 Mean daily gross ecosystem production (GEP) and ecosystem respiration
(Re ) of the combined hay (A) and corn (B) plots. Gross ecosystem pro-
duction which is the inverse of gross primary production (GEP = -GPP) is
presented for clarity. Arrows in panel (A) indicate times when hay were
cut. Dashed and solid arrows in panel (B) indicate times when corn was
planted and harvested respectively. . . . . . . . . . . . . . . . . . . . . . . 76
4.7 Cumulative gross primary production (GPP) and ecosystem respiration (Re )
in hay and corn in all study years. . . . . . . . . . . . . . . . . . . . . . . 77
A1 Scatter plots of the modeled vs. actual EC fluxes in corn used for training
the model, testing and validation of the neural-network model . . . . . . . 109
A2 Scatter plot of the modeled vs. EC nighttime fluxes in corn in 2012 . . . . . 110
x
CHAPTER 1
INTRODUCTION
1.1 General Introduction
Dairy production is an important industry in Canada which generated $6.02 billion in farm
cash receipts in 2015 (Canadian Dairy Information Centre, 2015), provided jobs to 215,000
Canadians and contributed $15.9 billion to Canada’s gross domestic product in 2009 (Agri-
culture and Agri-Food Canada, 2012; ÉcoRessources, 2015). Despite its socio-economic
benefits, the dairy industry is a large source of greenhouse gases (GHG) primarily methane
(CH4 ), nitrous oxide (N2 O) and carbon dioxide (CO2 ) (McGeough et al., 2012; Rotz et al.,
2010). One of the proposed strategies to mitigate GHG from dairy production is to switch
from perennial to annual feed crop production as perennials have been suggested to have
larger GHG mitigation potential through higher carbon sequestration in the ecosystem (Co-
nant et al., 2001; Paustian et al., 2016; Soussana et al., 2010).
To investigate if carbon sequestration by perennials is larger than annual crops, a three-
year side-by-side comparison study was conducted to quantify and compare the carbon
sequestration hay (Medicago sativa L. and Phleum pratense L.) to corn (Zea mays L.).
1
Both crops are mainly grown for production of dairy feed in Ontario, Canada. The net
ecosystem carbon budget (NECB) was used as a metric to determine the short-term change
of sequestered carbon in the crop system which was determined as the balance between the
net ecosystem exchange (NEE), carbon gained from manure application and carbon lost
through harvest.
Estimating the NECB of the two crop systems however only describes the carbon se-
questration potential of the crop systems. As N2 O emission from denitrification of nitroge-
nous fertilizer applied to crop fields is a large source of anthropogenic GHG (Hofstra and
Bouwman, 2005; Mulvaney et al., 1997), emissions of N2 O must be incorporated to the
estimated NECB of each crop systems to account for their contribution towards net green-
house gas emission. The net greenhouse gas balance (GHGB) was calculated as the sum of
the NECB and N2 O emission measured in a separate study on the same plots studied here,
converted to CO2 equivalent for each crop system and the GHGB of the two crop systems
were compared.
Apart from fertilizer application, soil tillage i.e. breaking up soil aggregates has been
previously demonstrated to increase denitrification and N2 O emission (Wagner-Riddle et al.,
2007). Hay renovation is commonly practiced to improve production and often requires in-
tense soil tillage (e.g. moldboard plowing). The effect of plowing hay in the final year of
the study on its NEE, NECB and GHGB was determined and compared to the less-intense
soil tillage (i.e. disking) employed in corn.
The eddy covariance (EC) and flux-gradient (FG) techniques are the two main microm-
2
eteorological methods used for measuring NEE over agricultural landscapes (Denmead,
2008). Between the two methods, EC is more widely used as it is a direct measure of NEE
whereas FG is not, but rather relies on parameterizations. The FG however is suitable for
measuring trace gas fluxes for side-by-side comparison of plots as it only requires a single
gas analyzer whereas employing EC for a similar type of study would require an individual
analyzer for each plot, increasing cost and measurement error.
One of the challenges of measuring NEE using FG over multiple plots, which is often
limited by the fetch of each plot, is the rapidly changing mean canopy height (hc ) of agri-
cultural crops. The height of the air intake (z) has to be frequently raised to be well above
the hc to avoid being in the roughness sublayer (RSL) that can cause an underestimation of
flux (Monteith and Unsworth, 2013). In fetch-limited multiple plot studies, maintaining a
narrow z/hc while avoiding sampling in the RSL is important to avoid sampling beyond the
treatment plot (Leclerc and Thurtell, 1990; Mölder et al., 1999).
The FG technique relies on several simplifications and assumptions that may result in
inadequate performance in certain situations. Reference measurements of NEE using the
EC method during two growing seasons were used to assess the quality of FG estimates
over corn and hay, two main dairy feed crops each having distinct physical attributes,
canopy architecture and aerodynamic properties. Measurements made using FG at vari-
ous z/hc were compared to NEE measured using EC. The EC method is a direct measure
of flux and considered more accurate and hence served as a reference method (Denmead,
2008; Loubet et al., 2013).
3
A partial alternative to the micrometeorological methods in measuring fluxes is the
chamber method. The widespread use of chambers is largely due to its simplicity of
operation, low cost, portability, useful in survey measurements, and small measurement
footprint, suitable for small-scale studies. However, chambers are limited spatially and
temporally in measuring fluxes when compared to micrometeorological methods. Due to
logistical reasons, measurements are often limited to one sampling per day during daytime
(Alves et al., 2012) and most measurements are conducted at weekly or bi-weekly intervals.
Deployment of chambers in the field is laborious limiting the number of units available to
account for spatial variability due to soil heterogeneity (Luo and Zhou, 2010).
Several comparisons between micrometeorological methods and chambers have been
previously made with various outcomes. The previous comparison studies however were
conducted over short duration (5-14 days) and/or over short vegetation. In this study, mi-
crometeorological and chamber methods were compared over a period of 210 days in corn,
a tall crop which has different aerodynamic properties than short crops.
1.2 Thesis Format and Objectives
This thesis is presented in paper format. Three studies are presented in Chapters 2, 3,
and 4 whereas a general conclusion is presented Chapter 5. Some repetition especially in
the description of measurement methods and setup will appear between the chapters. The
overall goal of this thesis is to estimate and compare the carbon sequestration and GHG
4
mitigation potential of hay and corn crops and to assess different methods used to estimate
trace gas fluxes.
Chapters 2 and 3 discuss the challenges of obtaining reliable trace gas flux measure-
ments using commonly used measurement techniques. Chapter 2 of this thesis compares
the NEE of hay and corn measured using FG and EC methods and discusses the differ-
ences in the results obtained. It is hypothesized that FG provides similar estimates of NEE
as EC. The objectives of the study were (1) to compare measurements of NEE over agri-
cultural crops made using FG against the EC technique and (2) to investigate the effects of
measuring NEE using FG at various z/hc over two crop types.
Chapter 3 of this thesis presents and compares N2 O and CH4 fluxes measured using
micrometeorological techniques and chamber method in corn during the growing season
of 2014. It is hypothesized that chamber methods do not provide the temporal and spatial
coverage that micrometeorological methods do for accurate estimation of N2 O and CH4
fluxes. The objective of the study was to expand the finding of previous studies by making
comparison between methods over a taller crop (i.e. corn) and longer period (150 days).
Chapter 4 of this thesis presents the NEE, the NECB and the GHGB estimates of corn
and hay crops. Differences in the results obtained for the above mentioned parameters
between the two crop systems are discussed and compared with other previous studies. It
is hypothesized that perennial hay sequesters more carbon than annual corn. The objectives
were (1) to quantify and compare the NEE and NECB of a corn and hay field, (2) to
calculate the GHG balance of a corn and a hay field considering NECB and N2 O emissions
5
from a previously published study conducted in the same fields, and (3) to investigate the
effects of plowing a hay field on its annual NEE, NECB and GHG balance.
The final chapter of this thesis summarizes the findings of Chapters 2 to 4 and recom-
mends future research topics.
6
CHAPTER 2
MEASUREMENTS OF CO2 FLUX

USING FLUX-GRADIENT AND EDDY
COVARIANCE TECHNIQUES
2.1 Introduction
Agricultural production holds the potential for helping in mitigating emission by seques-
tering atmospheric CO2 through enhancing buildup of soil carbon (Johnson et al., 2007;
Vergé et al., 2007; Gregorich et al., 2005). To evaluate the effectiveness of atmospheric
CO2 mitigation strategies in agricultural production, periodic monitoring of soil carbon
changes is necessary. However, significant changes of carbon in agricultural soil can only
be detected over periods of >10 years (Smith, 2004) and thus, other methods of detection
are required for a more rapid assessment of the efficacy of the atmospheric CO2 mitigation
strategies taken. Measuring the net CO2 flux or the net ecosystem exchange (NEE) of an
agricultural ecosystem allows for assessment of sequestered carbon as it directly measures
the net exchanges of CO2 at the soil-vegetation interface with the atmosphere.
The eddy covariance (EC) and flux-gradient (FG) techniques are the two main microm-
7
eteorological methods used in determining NEE over agricultural landscapes (Denmead,
2008). Eddy covariance is a direct measure of NEE where the covariance of vertical wind
speed and CO2 concentration fluctuations is measured whereas measurement of NEE us-
ing FG is based on the product of the turbulent diffusivity and vertical CO2 concentration
gradient (Denmead, 2008). Eddy covariance requires fast-response gas analyzer capable
of sampling at >10 Hz (Wagner-Riddle et al., 2005) whereas FG can be deployed using
gas analyzer that measures gas concentration at a lower sampling frequency (Prueger and
Kustas, 2005). Before the advent of fast-response gas analyzers, FG was the more com-
monly used micrometeorological method for measuring trace gas fluxes but with recent
technological advancement in gas analyzers, EC has been gaining more prominence over
FG (Rannik et al., 2004).
Despite the decline in the number of users and its popularity, FG is still relevant in
measuring trace gas fluxes as it is particularly well suited for side-by-side comparisons
of mid-size plots (1-4 ha in size) (Wagner-Riddle et al., 2005) subjected to different agri-
cultural management (Wagner-Riddle et al., 1997; Warland et al., 2001; Wagner-Riddle
et al., 2007) as air samples can be sequentially transported through long tubes from differ-
ent study plots to be detected for concentration by a single gas analyzer. In contrast, the
high frequency response requirement of EC does not allow for the use of long tubes. To
use EC for measurements over multiple plots requires the use of separate gas analyzers for
each plot which would be very costly and could lead to potential errors from systematic
differences in analyzer precision. Flux-gradient however is not a direct flux measurement
8
method but rather relies on parameterizations. Therefore, to increase confidence in NEE
measurements obtained using FG, results need to be compared to EC (i.e. the direct mea-
surement and standard method). Employing more than one technique when measuring
fluxes is recommended as a quality control process (McGinn, 2006).
One of the challenges of measuring NEE over agricultural landscapes using FG is the
rapidly changing mean canopy height (hc ) of agricultural crops. The height of the air intake
(z) above hc has to be frequently raised to avoid sampling air from the roughness sublayer
(RSL), directly above the canopy. Within the RSL, the Monin-Obhukov similarity theory
(MOST), on which the FG method is based, breaks down due to underestimation of the
transport coefficient (Simpson et al., 1998). Instruments located within the RSL may not
adequately represent the turbulence developed over the ecosystem of interest, but rather
may characterize the local effects or disturbances such as individual leaves or branches
(Burba, 2013).
The effect of sampling in the RSL however is not always constant at a static z to hc
height but is also dependent on atmospheric stability conditions (Foken, 2008). Under un-
stable conditions, turbulence is well developed with intense vertical eddy velocities which
may penetrate below the RSL (Kaimal and Finnigan, 1994). Meanwhile under stable con-
ditions, vertical eddy velocities are dampened due to lack of buoyancy leading to weak and
intermittent turbulence (Mahrt, 2010).
Currently there is no consensus on the exact method for quantifying the RSL height
(Florens et al., 2013) although values between 2 and 5hc estimated by Raupach et al. (1991)
9
have been extensively used (e.g. Rotach (1999); Jiménez (2004)). Subsequent works by
Rotach (2001), Pokrajac et al. (2007) and Cheng and Castro (2002) showed the height of
the RSL to be closer to 2hc (Florens et al., 2013). The height of the RSL was reported to
vary with vegetation type with values of 2hc in corn (Cellier and Brunet, 1992) and 3.0-
8.0hc in forest (Garratt, 1978; Fazu and Schwerdtfeger, 1989; Prueger and Kustas, 2005).
Simpson et al. (1998) found that NEE can be measured with reasonable confidence using
FG at 1.4hc <z<1.6hc over a pine forest but found large discrepancies when measuring at
z<1.4hc .
Another challenge when measuring NEE over agricultural crops using FG, especially
in a multiple plot layout is sampling above the RSL while avoiding measuring beyond
the treatment plot (Leclerc and Thurtell, 1990; Mölder et al., 1999) following the general
rule of thumb of z to fetch ratio of 1:100 (Businger, 1986). Fluxes that did not fit the
fetch criteria would need to be filtered out as the signals would be convoluted with those
originating from surrounding areas of different vegetation make up and treatments. As
such, z has to be kept as close to the canopy as possible to maximize the number of valid
data while avoiding sampling in the RSL.
As FG is still relevant for measuring fluxes over agricultural ecosystems especially
in studies over multi-plot experiment but is vulnerable to sampling in the RSL, a study
measuring NEE over agricultural crops using FG to evaluate the impact of variable z/hc
on FG measurements over contrasting agricultural crops is needed to provide guidance
on intake height placement. The present study was conducted over corn and hay, two
10
main dairy feed crops each having distinct physical attributes, canopy architecture and
aerodynamic properties. Measurement of NEE using FG over corn was conducted over the
growing seasons of 2014 and 2015, and over hay in 2014, for periods with variable z/hc .
The objectives of this study were to compare measurements of NEE over agricultural
crops made using FG and EC techniques and to investigate the effects of air sampling height
on NEE measurements using FG over two crop types.
2.2 Materials and Methods
2.2.1 Site description
This study was conducted at the Elora Research Station, Ontario, Canada (43o 38’27.8”N
80o 24’20.4”W) during the growing season from June to October 2014 and June to October
2015. The soil at the site is a Guelph silt loam (fine loamy, mixed, mesic Glossoboric
Hapludalf ). Detailed description of soil physical and chemical properties are found in
Table 4.1.
In 2014, the experiment was conducted on two 8-ha areas (Fig. 2.1) which were part
of an experiment to measure greenhouse gas emission from different dairy manure man-
agement and dairy feed crop types (see Chapter 4). The southern section of the plot was
planted in corn (Zea mays cv. Syngenta R NK-N21J-3220 E-Z Refuge) at a seeding rate of
74 000 ha−1 on May 26 (day of year (DOY) 146) and harvested on November 4 (DOY 308)
while the northern section of the plot had been planted with hay crop since 2008 but was
11
Figure 2.1: Aerial view of the study site and schematic of the plot layout and position of the
sonic anemometers, eddy covariance (EC) open path analyzers, flux-gradient (FG) intakes
and sample tubes directing air to the tunable diode laser trace gas analyzer (TDLTGA).
Area bordered by yellow and red lines are the hay and corn fields respectively. The areas
surrounded by white frames shows the specific area of interest of the study.
renovated and reseeded with Medicago sativa L. cv. Pioneer R 53Q32 in May 26 (DOY
146) at a seeding rate of 18 kg ha−1 . Hay was cut on July 27 (DOY 208) and August 18
(DOY 230).
In 2015, all measurement instruments were moved towards the south of the 2014 site
for a new study on greenhouse gas emission from corn amended with different types of
inorganic nitrogen fertilizers (Fig. 2.2). The study area was planted in corn (Zea mays cv.
Syngenta R NK-N20Y-3122 E-Z Refuge) at a seeding rate of 74 000 ha−1 on May 13 (DOY
12
Figure 2.2: Aerial view of the study site and schematic of the plot layout and position of the
sonic anemometers, eddy covariance (EC) open path analyzers, flux-gradient (FG) intakes
and sample tubes directing air to the tunable diode laser trace gas analyzer (TDLTGA).
Area bordered by red lines are all planted in corn. The area surrounded by white frames
shows the specific area of interest of the study.
133) and harvested on October 27 (DOY 300).
2.2.2 NEE measurement
Measurement of NEE was made using the flux-gradient (FG) and the eddy-covariance (EC)
techniques in the two study years. Net ecosystem exchange according to using FG (FF G )
13
is given by
∂C
FF G = −K (2.1)
∂z
where K is the eddy diffusivity of CO2 at height z (m) and ∂C/∂z is the CO2 concentration
gradient (µg m−3 ) at height z. Assumption of similarity between sensible heat and CO2
turbulent transport, and integration of Eq. (2.1) between the two heights (z1 and z2 ) result
in:
u∗ κ∆C
FF G = −d (2.2)
[ln( zz12 −d ) − ψh2 + ψh1 ]
where u∗ (m s−1 ) is the friction velocity, κ is the von Karman constant (= 0.41), ∆C
is the CO2 concentration difference between heights z1 and z2 , d (m) is the displacement
height, and ψh2 and ψh1 are the integrated Monin-Obukhov similarity functions for heat
for each sampling height. These functions were calculated using the stability parameter,
ζ = (z-d)/L, where L is the Obukhov length, as derived by Dyer and Hicks (1970) and
Paulson (1970). Sonic anemometers (CSAT3, Campbell Scientific, Logan, UT) were used
to measure sensible heat and u∗ over each crop system (Fig. 2.1).
Two air sampling assemblies were installed in each hay and corn plot in 2014 (Fig. 2.1)
while for the study in 2015, four air sampling assemblies were installed in the corn plot
(Fig. 2.2). The intakes were set at ≈ 1.6hc and adjusted to changes in crop height. The end
of each intake was connected to a two-way solenoid valve (ASCO R 8320 Series, Florham
14
Park, NJ, USA) housed in a gradient valve assembly unit (17883, Campbell Scientific,
Logan, UT, USA) and the outlet of the solenoid valve was connected to a manifold valve
(8×1 Lee Valve, The Lee Company, Westbrook, CT, USA) via 160-180 m of Synflex R
tubing. A relay controller (SDM-CD16AC, Campbell Scientific, Logan, UT, USA) which
was controlled by a datalogger (CR1000, Campbell Scientific, Logan, UT, USA) was used
to actuate the solenoid to alternately switch between sampling the upper and lower intake
every 15 s. The manifold valve was actuated by the relay controller to select the air sample
from one of the plots to be analyzed in sequence for 30 min at a time. Water was removed
from the air sample by a series of two dryers (PD625, Campbell Scientific, Logan, UT,
USA) which housed 50 tubes of Nafion R dryer elements. After the dryers, air was directed
to be analyzed for CO2 concentration by a tunable diode laser trace gas analyzer system
(TDLTGAS) (TGA100, Campbell Scientific, Logan, UT, USA).
Net ecosystem exchange using the eddy-covariance method (FEC ) is calculated using
the covariance between fluctuations in vertical wind velocity and CO2 concentration
FEC = ρ¯d w0 s0 (2.3)
where, w is vertical wind velocity, s is CO2 mixing ratio and ρd is air density. Prime
indicates instantaneous fluctuations about the mean, and the overbar denotes a time average
over 30 min. Fluctuations of CO2 concentration and vertical wind speed were measured
using an open-path infra-red gas analyzer (LI-7500, LI-COR, Lincoln, Nebraska, USA)
and the sonic anemometer, respectively. The open-path analyzer was installed horizontally
15
aligned to the same sonic anemometer used in the flux-gradient technique. For the study in
2014, an open-path analyzer was installed in each corn and hay plot while for the study in
2015, only one open-path analyzer and one sonic anemometer were used.
Raw data from the sonic anemometers and open-path gas analyzers were processed
using the EddyPro R Software (Ver. 6.0, LI-COR, Inc., Lincoln, Nebraska, USA) for calcu-
lation of NEE measured using the eddy-covariance technique. Corrections applied by the
software include the double rotation scheme to compensate for anemometer tilt correction,
application of the WPL-correction to compensate for fluctuations in the density of CO2 and
H2 O (Webb et al., 1980) and low-pass and high-pass filtering correction (Moncrieff et al.,
1997), self-heating corrections (Burba et al., 2008) and statistical tests for raw data screen-
ing (i.e. spike removal, amplitude resolution, drop-outs, absolute limits, and skewness and
kurtosis) (Vickers and Mahrt, 1997).
The calculated half-hourly NEE from the EddyPro R software were later subjected to a
spike removal procedure described in Papale et al. (2006) which uses the position of each
half-hourly NEE with respect to the values just before and after and is applied to blocks
of 13 days. The outlier detection was based on the double-differenced time series, using
the median of absolute deviation about the median (MAD). Night and daytime data were
treated separately with nighttime defined as the time of day between astronomical sunset
and sunrise time while global radiation was <20 Wm−2 (Meeus, 1998).
Fluxes were filtered for periods when weak turbulence (between 16 to 23% throughout
the study) was deemed to have occurred by determining the u* threshold following the
16
protocols of Reichstein et al. (2005) and Papale et al. (2006). The u* thresholds for years
2014 and 2015 were 0.081 and 0.083 m s−1 , respectively.
2.2.3 Comparison of NEE measured between techniques and statisti-
cal analysis
Comparisons of NEE measured using FG and EC for each crop type were based on the
overall NEE measured during the growing season. Measurements of NEE using FG and EC
were separated into daytime and nighttime measurements as NEE has a diurnal variation
due to the distinct processes of CO2 uptake and respiration during the day and respiration
only at night.
Evaluation of the comparison between techniques was made using linear regression
of NEE measured using each technique where the degree of agreement was evaluated by
the slope of the regression (β1 ) and the Pearson’s product moment correlation coefficient
(r). Significance of the regressions was tested at α = 0.05. Strength of the correlation
relationship was classified as very good (r = 0.80 - 1.00), good (r = 0.60 - 0.79), moderate
(r = 0.40 - 0.59), weak (r = 0.20 - 0.39) and very weak (r = 0.00 - 0.19) (Evans, 1996). The
slopes of the regression (β1 ) were compared against β2 = 1 with Ho : β1 = β2 and H1 : β1 6=
β2 . Significance of the comparison was tested at α = 0.05.
The magnitude by which NEE measured using FG systematically underestimated NEE
measured using the EC technique in the RSL was expressed as an enhancement factor, γ
17
calculated as
FEC
γ= (2.4)
FF G
where within the inertial sublayer, γ ≈ 1 whereas within the RSL, γ >1 (Raupach and Legg,
1984). This parameter was also used to assess the discrepancy between measurements.
2.3 Results and Discussion
Overall, NEE measured during the growing season using the two techniques in both years
and crop systems showed very good agreement (Fig. 2.3). Better agreement between
techniques however were observed in hay (r = 0.95) (Fig. 2.3A) and corn in 2015 (r =
0.95) (Fig. 2.3C) than in corn in 2014 (r = 0.90) (Fig. 2.3B). The slopes of all linear
regressions were > 0.89, and significantly different than 1 (Table 2.1), indicating that FG
and EC are highly correlated but FG was slightly biased.
Daytime NEE showed very good agreement between techniques in hay (r = 0.95) (Fig.
2.4A), corn in 2015 (r = 0.91) (Fig. 2.4E) and corn in 2014 (r = 0.89) (Fig. 2.4C). Nighttime
NEE also showed very good agreement between techniques in hay (r = 0.92) (Fig. 2.4B)
and in corn 2015 (r = 0.82) (Fig. 2.4F) while good agreement was observed in corn 2014
(r = 0.71) (Fig. 2.4D). All slopes of the linear regression were >0.81 with the exception
of nighttime NEE in corn in 2014 (slope = 0.44) (Fig. 2.4D). The peculiarly low slope of
linear regression of nighttime NEE in corn in 2014 indicated an underestimation of NEE
18
20
Slope = 0.98
−20
−40 (A)
−40 −20 0 20
FG CO2 flux (µmol m−2 s−1 )
20
Slope = 0.89
−20
−40 (B)
−40 −20 0 20
20
Slope = 1.03
−20
−40 (C)
−40 −20 0 20
EC CO2 flux (µmol m−2 s−1 )
Figure 2.3: Scatter plot of NEE measured using the eddy-covariance (EC) and flux-gradient
(FG) techniques in 2014 for hay (A) and corn (B) and corn in 2015 (C). Values are for 30-
min periods and inclusive of day and nighttime NEE. Red line indicate 1:1.
19
10 14
Slope = 0.99 Slope = 0.81
12
0
10
−10 8
6
−20 4
2
−30 (A) (B)
0
−40 −2
−40 −30 −20 −10 0 10 −2 0 2 4 6 8 10 12 14
10 25
0 20
−10 15
−20 10
−30 5
−40 (C) 0 (D)
−50 −5
−50 −40 −30 −20 −10 0 10 −5 0 5 10 15 20 25
20 20
10
15
0
−10 10
−20 5
−30
0
−40 (E) (F)
−50 −5
−50 −40 −30 −20 −10 0 10 20 −5 0 5 10 15 20
Figure 2.4: Scatter plots of NEE measured using the FG and EC techniques during daytime
in hay (A), corn in 2014 (C), corn in 2015 (E) and nighttime in hay (B), corn in 2014 (D)
and corn in 2015 (F). Red line indicate 1:1.
measured using FG when compared to EC. As with the overall NEE, separated day and
night NEE also showed all of the slopes of the linear regression between EC and FG to be
significantly different than 1 (Table 2.1).
20
Table 2.1: Slope of the linear regression and Pearson’s product moment of correlation coefficient (r) between NEE measured
using the EC and FG techniques, the roughness length at peak growth (z0 ) and maximum mean crop height (hc ) in corn and hay
in 2014 and in corn in 2015. Linear regressions were performed on the overall, daytime and nighttime NEE.
2014
2015
Corn Hay
All Day Night All Day Night All Day Night
21
Slope 0.89 0.81 0.44 0.98 0.99 0.81 1.03 1.03 0.80
r 0.90 0.89 0.71 0.95 0.95 0.92 0.92 0.91 0.82
Max hc (m) 3.05 0.39 2.43
z0 (m) 0.035±0.008 0.14±0.05 0.12±0.04
The differences between FG and EC NEE estimates may be the result of inadequate
air sampling heights for the FG system. From the minimum RSL height estimated by
Raupach et al. (1991); Cheng and Castro (2002); Rotach (2001) each using distinct indi-
vidual estimation methods of 2hc and the estimated RSL height for corn also at 2hc (Cellier
and Brunet, 1992), it is surmised that ideally, z/hc >2. However, as NEE can be measured
with reasonable confidence over a pine forest when 1.4<z/hc <1.6 (Simpson et al., 1998),
1.4<z/hc <2 was considered to be adequate to obtain good measurements of NEE while
z/hc <1.4 was considered to be inadequate.
During the growing season of 2014, the occurrence of ideal z/hc in corn was 2.3%,
while occurrence of inadequate and adequate z/hc were 19.1 and 77.9%, respectively. In
comparison to hay, the occurrence of ideal, adequate and inadequate z/hc were 16.3, 77.0
and 6.7%, respectively (Fig. 2.5A) while in corn in 2015, the occurrence of ideal and
adequate z/hc were 16.2 and 83.8%, respectively (Fig. 2.5C) while no inadequate z/hc was
observed.
The large discrepancy between NEE measured using EC and FG in corn in 2014 at night
could therefore be attributed to the higher occurrence of inadequate z/hc during the growing
season in that year (Fig. 2.5B), which potentially caused the intakes to be sampling in the
RSL. The higher growth rate of corn in 2014 (Fig. 2.6) however impeded placement of
z at ≈ 2hc in tandem with crop growth during the rapid vegetative stage (ca. DOY 170 -
190). The height of the air sample intakes was increased every three to four days during the
vegetative growth stage but this was not sufficient adjustment to accommodate the rapid
22
40
(A)
30
20
10
100
(B)
80
Frequency
60
40
20
60
50 (C)
40
30
20
10
1 1.1 1.2 1.3 1.4 1.5 1.6 1.7 1.8 1.9 2 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3
z/hc
Figure 2.5: Frequency of the ratio of the sampling height (z) to mean canopy height (hc ) in
hay (A), corn in 2014 (B) and corn in 2015 (C).
23
growth of corn in 2014. In contrast, the relatively lower growth rate of hay (Fig. 2.6)
allowed z/hc to be maintained at the adequate and ideal range with the same z adjustment
frequency. The situation in corn in 2015 however was different as vegetation at the site was
more homogeneous resulting in a relativity longer fetch and it was less critical to maintain
a narrow z/hc as required for corn in 2014. Our observations reiterate the challenge of
measuring NEE using FG over fast growing agricultural crops arranged in multiple plots
specifically on adjusting z/hc to maximize the number of valid flux values.
5
hc corn 2014
hc hay
4.5 z corn 2014
z hay
z corn 2015
4
hc corn 2015
3.5
3
Height (m)
2.5
1.5
0.5
0
150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300
Day of year
Figure 2.6: Mean canopy height (z) and sampling intake height (hc ) of hay, corn in 2014
and corn in 2015 during the growing season.
Nevertheless, the discrepancy between techniques in corn in 2014 happened only dur-
ing nighttime whereas daytime NEE estimates showed very good agreement (Figure 2.4C).
24
Scatter plots of the NEE measured using EC and FG classified into stability conditions
(Figs. 2.7–2.9) revealed that the underestimation in corn in 2014 happened almost exclu-
sively under stable condition (Fig. 2.8) indicating a coupled effect between measuring at
inadequate z/hc and stable condition. Other than the slope of regressions, the enhance-
ment factor, γ calculated at each stability conditions also indicated better agreement under
unstable than under stable conditions (Table 2.2).
Table 2.2: The mean±S.D. enhancement factor (γ) during the growing season of each crop
according to each respective atmospheric stability condition.
Unstable Neutral Stable
(ζ ≤ -0.01) (-0.01 <ζ <0.01) (ζ ≥ 0.01)
Hay 0.98±0.03 (1641) 0.94±0.02 (342) 0.96±0.02 (1690)
Corn (2014) 1.14±0.05 (1391) 1.28±0.08 (219) 1.71±0.09 (1381)
Corn (2015) 0.87±0.08 (13) 1.25±0.09 (13) 0.80±0.03 (13)
Previous studies gave varying accounts on the link between atmospheric stability con-
dition and underestimation of fluxes by FG. Cellier and Brunet (1992) and Garratt (1978)
found no influence of atmospheric stability condition on the the discrepancy of fluxes mea-
sured using the two techniques while Raupach (1979) and Fazu and Schwerdtfeger (1989)
found that γ increased to be >1 under unstable and neutral conditions. In contrast, Simpson
et al. (1998) in their post-senescence study and our observation in corn 2014 observed γ
values to be closer to unity under unstable and neutral conditions while γ>1 under stable
conditions.
25
20
10
−10
−20
Stable
Unstable
Neutral
−30 Slope = 0.87
Slope = 1.02
Slope = 0.91
1:1
−40
−40 −30 −20 −10 0 10 20
Figure 2.7: Scatter plot of NEE measured using the FG and EC techniques over hay in
2014 according to atmospheric stability conditions.
26
30
20
10
−10
−20
Stable
−30 Unstable
Neutral
Slope = 0.70
−40 Slope = 0.81
Slope = 0.46
1:1
−50
−50 −40 −30 −20 −10 0 10 20 30
Figure 2.8: Scatter plot of NEE measured using the FG and EC techniques over corn in
27
20
10
−10
−20
−30 Stable
Unstable
Neutral
−40 Slope = 0.82
Slope = 1.03
Slope = 1.11
1:1
−50
−50 −40 −30 −20 −10 0 10 20
Figure 2.9: Scatter plot of NEE measured using the FG and EC techniques over corn in
28
2.4 Conclusion
Our two-year comparison study of NEE measured over agricultural crops using the FG and
EC techniques showed very good agreement between each other. Our observations showed
that NEE measured using FG at 1.4z/hc <z<1.6hc over agricultural crops was comparable
to NEE measured using EC. We also demonstrated that NEE can be measured at z as low as
1.4hc but only under unstable atmospheric conditions when turbulence is well-developed.
When measuring at 1.4z/hc under stable atmospheric conditions, NEE measured using FG
was underestimated when compared to NEE measured using EC. The very good agreement
between techniques showed that FG can be deployed successfully for measuring NEE over
agricultural crops in a multiple-plot layout for side-by-side comparison studies. Frequent
adjustments of z during the rapid growth stages of agricultural crops is critical if fetch
is limited to avoid sampling in the RSL. However, if maintaining z/hc ≈ 2 is not always
possible, NEE measured when 1.4<z/hc under stable atmospheric conditions should be
scrutinized.
29
CHAPTER 3
MEASUREMENTS OF N2O AND CH4

FLUXES USING
MICROMETEOROLOGICAL AND
CHAMBER TECHNIQUES
3.1 Introduction
Agricultural activities emit significant amounts of trace gases to the atmosphere that are
responsible for the greenhouse effect (Cole et al., 1997; Johnson et al., 2007; Smith et al.,
2008). Nitrous oxide (N2 O) and methane (CH4 ) are the two main trace gases being emit-
ted from agricultural activities (Smith et al., 2014; Environment Canada, 2015). In crop
fields, emission of N2 O is mainly from the application of nitrogen fertilizer where nitrate
is reduced to N2 O by denitrifying bacteria (Firestone et al., 1980). Methane is emitted by
fermentation from waterlogged soils under anaerobic conditions due to imperfect drainage
of crop fields (Topp and Pattey, 1997). Under aerobic conditions however, soil can act as
CH4 sink where methanotrophic bacteria in soil consume CH4 from the atmosphere as an
energy source (Le Mer and Roger, 2001).
30
Monitoring of N2 O and CH4 emission in agricultural fields is important to gauge the
extent of agronomic activities in emitting N2 O and CH4 and to measure the effectiveness of
mitigation strategies through modification of current agronomic practices. Micrometeoro-
logical methods have the advantage of measuring N2 O and CH4 fluxes over agricultural
ecosystems due to the capability of automatically making quasi-continuous year-round
measurement and are non-intrusive, causing very minimal disturbance to the ecosystem be-
ing studied (Wagner-Riddle et al., 2005). In addition to that, micrometeorological methods
spatially integrate fluxes from the study area resolving the spatial variability that commonly
exist in agricultural fields due to soil heterogeneity (Pattey et al., 2006).
An alternative to the micrometeorological methods in measuring fluxes are chamber
methods. Several types of chambers for measuring trace gas flux in agricultural ecosystem
exist but the non-flow through non-steady state (NFT-NSS) chamber is the most commonly
used (Bouwman et al., 2002; Rochette and Bertrand, 2007). The widespread use of cham-
bers is largely due to its simplicity of operation, consisting in sealing an impermeable
chamber over a known area of soil surface and monitoring the headspace gas concentration
to determine its rate of change (Denmead, 2008). Apart from the simplicity of operation,
chambers are relatively inexpensive compared to the instruments required for micromete-
orological measurements. Other advantages of chambers over micrometeorological meth-
ods are: portability, usefulness in survey measurements, and small measurement footprint,
suitable for small scale studies such as experiments that involve multiple treatments and
experimental units.
31
Nevertheless, despite its simple operation, cost-effectiveness and portability, chambers
suffer several drawbacks in measuring fluxes when compared to micrometeorological meth-
ods. One of the limitations is that chambers only measure trace gas flux directly over the
soil surface whereas micrometeorological methods measure fluxes above the crop canopy,
allowing measurement of trace gas flux emitted from both soil and vegetation. This is par-
ticularly important for CH4 as it is known to also be transported from soil to the atmosphere
through plant vascular tube (Seiler et al., 1983; Cicerone et al., 1983). Nitrous oxide has
also been suggested to be transported through plant vascular tube but only very few studies
(Yu et al., 1997; Yan et al., 2000) have elucidated such transport mechanism.
Another major drawback of chambers, in cases when manually operated chambers (i.e.
NFT-NSS chambers) are used, is their temporal limitation in measuring fluxes. Due to lo-
gistical reasons measurements are often limited to one sampling per day during daytime
taken once a week or every 2-3 days. Depending on the location of the study site, ease of
access to the study site and availability of personnel, frequent sampling (i.e. daily) may not
always be possible. Automatic chambers have the capacity to address issues with temporal
resolution but are limited in the spatial coverage provided. Apart from the temporal lim-
itation, deployment of chambers in the field is laborious and requires multiple personnel
to operate and is also bound to spatial variability due to soil heterogeneity (Luo and Zhou,
2010).
Several comparisons between micrometeorological methods and chambers for N2 O flux
have been made previously with various outcomes. Smith et al. (1994) in their comparison
32
over a grass field, found the mean fluxes measured using chambers were higher than when
measured using micrometeorological methods. On the other hand, comparisons over wheat
stubble (Christensen et al., 1996) and carrot and corn fields (Laville et al., 1999) found
both techniques to agree with each other. All the aforementioned studies however were
only conducted over short periods of time (9-14 days) until recently, a comparison study
by Wang et al. (2013) was conducted for three months. The study by Wang et al. (2013)
made over a cotton field found that chamber fluxes were lower than when measured using
a micrometeorological method.
Previous comparison studies between chambers and micrometeorological techniques
for CH4 were mostly made in natural ecosystems in peatlands and wetlands (Schrier-Uijl
et al., 2010; Yu et al., 2013; Hendriks et al., 2010) and in forest (Wang et al., 2013). To
our knowledge, only Denmead et al. (2010) made a comparison between techniques for
CH4 in an agricultural ecosystem. In all of the aforementioned studies in natural ecosys-
tems, all studies found good agreement between techniques. The study of Denmead et al.
(2010) however found larger CH4 flux measured using the micrometeorological method
due to fluxes originating from drains in the vicinity of the site that were not captured by the
chambers.
Comparison between techniques for N2 O flux showed inconsistent outcomes while
comparisons for CH4 were more consistent. Nevertheless, the consistency of CH4 compar-
isons were confined to natural ecosystems which possess distinct soil water characteristic
than agricultural ecosystem. Most comparison studies for N2 O were conducted over a short
33
duration and thus were not able to demonstrate any temporal variability. The study by Wang
et al. (2013) provided valuable insight of the differences between methods over an extended
period but nevertheless was made over a short crop, which could potentially yield different
outcomes than over tall crops. Comparing chamber and micrometeorological methods is
highly influenced by wind direction, wind velocity and atmospheric stability (Laville et al.,
1999) and crops with varying height and canopy architecture may have distinct effect on
wind velocity (Monteith and Unsworth, 2013).
In the present study, fluxes of N2 O and CH4 measured using chambers were compared
to fluxes measured using micrometeorological methods to expand the works of previous
studies by making comparison between methods over a taller crop (i.e. corn) and ex-
tended period. The present study was made in a corn field grown for dairy feed where
the flux-gradient (FG) and the eddy covariance (EC) techniques were the micrometeoro-
logical methods used.
This study was conducted at the Elora Research Station in Elora, Ontario, Canada. The
soil at the site is a Guelph silt loam (fine loamy, mixed, mesic Glossoboric Hapludalf ).
The study was set up on a 200 × 400 m plot grown in continuous corn (Zea mays L.) at a
seeding rate of 74 000 ha−1 . The corn field was part of a study that investigated the effects
34
of manure application timings (i.e. fall and spring) where during each application, manure
was applied at a target rate of 150 kg N ha−1 . The study plot was split into two subplots of
equal sizes (200 × 200 m) with one plot applied with dairy manure in the fall and the other
applied in spring. Manure application rates and chemical composition are listed in Table
4.2.
3.2.2 Chamber method
Measurement of N2 O and CH4 fluxes were made by using non-flow through non-steady-
state (NFT-NSS) chambers (Rochette and Bertrand, 2007). The chamber comprised of two
components, the chamber collar and the chamber lid which were constructed from acrylic
sheet with dimensions of 0.6 m × 0.6 m × 0.15 m (l × w × h). The lid was covered
with reflective aluminum bubble wrap that insulated the temperature inside the chamber
from being altered by incoming solar radiation. The top of the lid consisted of two ports, a
sampling port from which air samples were drawn from and a vent port which equilibrates
the pressure inside the chamber with atmospheric pressure. Closed cell foam was glued to
the base of the lid to provide an impermeable seal when placed over the collar. Prior to
the measurement campaign, the chamber was tested for leaks by breathing around the lid
while the sampling port was connected to an infra-red CO2 gas analyzer (LI-6262, Li-Cor
Biosciences Inc., Lincoln, Nebraska, USA) and observing for spikes in CO2 concentration
in the chamber which is an indication of leakage.
Twenty four chamber collars were installed in the field with 12 collars installed each
35
in the fall-applied and spring-applied plots (Figure 3.1). In each plot, three corn inter-
rows located west of the flux-gradient air sample intake to be within the path of the W-SW
predominant wind (Figure 3.2) were selected for chamber placement. Within each inter-
row, four collars were inserted ≈ 5 cm in the soil and spaced ≈ 10 m from one another.
Spring Fall
Flux-gradient intake
Eddy covariance and sonic anemometer
Soil chamber
Figure 3.1: Schematic of the two study plots (diagram not drawn to scale). Three inter-
rows were selected to the west of the flux–gradient intake of each plot. Four chambers
were deployed in each inter-row.
Soil N2 O and CH4 fluxes were measured twice weekly for four weeks after corn was
planted and weekly afterwards. During each measurement day, the chambers were de-
ployed for 30 minutes with air sampled from the chamber at 10-min intervals (time = 0, 10,
20 and 30 min). At time 0, the chamber was placed on top of the collar and four building
bricks were placed on top of the chamber at each corner to weigh it down and ensure a
good seal between the collar and the chamber. Air samples were drawn through the sam-
36
WS ≥ 6 Wind Rose
5 ≤ WS < 6 N
4 ≤ WS < 5
3 ≤ WS < 4
2 ≤ WS < 3
1 ≤ WS < 2
0 ≤ WS < 1
W E
Figure 3.2: The frequency of wind speed (m s−1 ) and direction (o ) at the study site during
the hours of 900 to 1200.
37
pling port by using a 20 mL syringe fitted with a hypodermic needle (26G 3/8, Becton
Dickinson, Rutherford, NJ, USA). Prior to drawing the air sample, the syringe plunger was
pumped twice to fully flush the syringe and the dead volume of the port (Rochette and
Bertrand, 2007). The sampled air was then transferred to a pre-evacuated 12 mL glass vial
(Exetainer, Labco Ltd., High Wycombe, UK). The air sampling procedure was repeated
for all soil collars at each time interval. Sampling was made between 0900 to 1200 EST
on each measurement day to provide the best representation of the daily flux (Alves et al.,
2012; Reeves and Wang, 2015). The gas samples were analyzed for N2 O and CH4 con-
centration at the Department of Soil Science at the University of Manitoba, Winnipeg, MB
using a gas chromatograph equipped with an electron capture detector (CP-3800, Varian
Inc., Palo Alto, CA).
Soil N2 O and CH4 fluxes measured using chambers (Fcham (nmol m−2 s−1 )) were cal-
culated using the rate of change of its concentration inside the chamber during deployment
(Hutchinson and Livingston, 1993) as:
∂C V c M
Fcham = × × (3.1)
∂t A Vm
where ∂C/∂t is the rate of change of mixing ratio of the gas of interest (nmol mol−1 ), Vc
(m3 ) is the chamber headspace volume, A (m2 ) is the area covered by the chamber, M (g
mol−1 ) is the molecular mass of the gas of interest and Vm (m3 mol−1 ) is the molecular
volume at chamber temperature and barometric pressure calculated from the ideal gas law.
The change of N2 O and CH4 concentration from each chamber over time (∂C/∂t) was
38
fitted to either a linear and quadratic regression depending on which fit had the highest R2
following the procedure by Rochette and Eriksen-Hamel (2008). The slope of the regres-
sion was tested for significance using a t-test and the ∂C/∂t rate was set to zero when it was
not significantly different than zero. Fluxes made on each measurement day were consid-
ered to represent the daily mean flux. Linear extrapolation was carried out to estimate the
daily mean flux between successive measurements to obtain a complete daily mean flux
dataset throughout the study period for calculation of cumulative flux.
3.2.3 Flux-gradient technique
The N2 O flux (FN2 O ) was measured using the flux-gradient (FG) technique:
∂C
FN2 O = −K (3.2)
∂z
where K is the eddy diffusivity of N2 O at height z (m) and ∂C/∂z is the N2 O concentration
gradient (ng m−3 ) at height z. Assumption of similarity between sensible heat and N2 O
turbulent transport, and integration of Eq. (3.2) between the two heights (z1 and z2 ) results
in:
u∗ κ∆C
FN2 O = −d (3.3)
[ln( zz12 −d ) − ψh2 + ψh1 ]
where u∗ (m s−1 ) is the friction velocity, κ is the von Karman constant (κ = 0.41), ∆C is
the N2 O concentration difference between heights z1 and z2 , d is the displacement height,
39
and ψh2 and ψh1 are the integrated Monin-Obukhov similarity functions for heat for each
sampling height. These functions were calculated using the stability parameter (z-d)/L,
where L is the Obukhov length, as derived by Dyer and Hicks (1970) and Paulson (1970).
Sonic anemometers (CSAT3, Campbell Scientific, Logan, UT) were used to measure
sensible heat flux and u∗ over each crop system (Figure 4.1). The sonic anemometers were
set facing 270o W which was the prevailing wind at the site 2-3 m above mean crop height.
All instruments were connected to a data logger (CR1000, Campbell-Scientific, Logan,
Utah, USA).
Wind direction together with air sample intake heights and position within the plots was
used to select periods when the wind direction allowed for a fetch-to-height ratio of at least
50:1 (horizontal distance to height of measurement ratio) to assure that >80% of the flux
measured originated within the experimental plots (Leclerc and Thurtell, 1990).
3.2.4 Eddy covariance technique
Measurement of CH4 flux using the eddy covariance (EC) technique is calculated as the co-
variance between fluctuations in vertical wind velocity and CH4 concentration. The general
equation for describing CH4 flux (FCH4 ) is written as:
FCH4 = ρ¯d w0 s0 (3.4)
where, w is vertical wind velocity (m s−1 ), s is CH4 mixing ratio (nmol mol−1 ) and ρd
is air density (kg m−3 ). Prime indicates instantaneous fluctuations about the mean, and
40
the overbar denotes a time average. Fluctuations of s and w were measured using closed
path wavelength-scanned cavity ring down spectroscopy analyzer (WS-CRDS) (Picarro
G2311-f Analyzer, Picarro Inc., Santa Clara, California, USA) and a sonic anemometer,
respectively. The air inlet to the WS-CRDS analyzer was installed horizontally aligned to
the same sonic anemometer used in the flux-gradient technique and accordingly, was also
set facing west and its height was adjusted throughout the season according to the height
of the sonic anemometer. Similar to the sampling frequency of the sonic anemometer, the
WS-CRDS analyzer also sampled at 10 Hz.
Data from the sonic anemometer and WS-CRDS analyzer were processed using the
EddyPro R Software (Ver. 6.0, LI-COR, Inc., Lincoln, Nebraska, USA) for calculation of
CH4 flux. Corrections applied by the software include the double rotation scheme to com-
pensate for anemometer tilt correction, application of the WPL-correction to compensate
for fluctuations in the density of CH4 and H2 O (Webb et al., 1980), low-pass and high-pass
filtering correction (Moncrieff et al., 1997) and statistical tests for raw data screening (i.e.
spike removal, amplitude resolution, drop-outs, absolute limits, and skewness and kurtosis)
(Vickers and Mahrt, 1997).
41
3.3 Results
3.3.1 N2 O flux
The average daily mean N2 O flux rates throughout the observation period measured in the
fall-applied plot were 0.48 and 0.14 nmol m−2 s−1 using chambers and FG, respectively
(Figure 3.3A). In the spring-applied treatment, the average daily mean N2 O flux rates were
0.44 and 0.28 nmol m−2 s−1 measured using chambers and FG, respectively (Figure 3.3B).
The highest daily mean flux rate was recorded on day 176 (June 25, 2014) in the fall-
applied treatment by both measurement methods (Figure 3.3A). The flux rate measured
using chambers on day 176 (4.61 nmol m−2 s−1 ) however was higher than the flux rate
measured using the FG technique (2.72 nmol m−2 s−1 ). Similar to the fall-applied plot,
the highest daily mean flux rate measured using chambers in the spring-applied was also
observed on day 176 (3.28 nmol m−2 s−1 ) which was close but still higher when compared
to FG flux rate on the same day of 2.85 nmol m−2 s−1 . The highest flux rate measured using
FG in the spring-applied plot was 3.13 nmol m−2 s−1 on day 169 (June 18, 2014) which
was higher than flux measured using chambers on the same day (1.69 nmol m−2 s−1 ).
Fluxes in both fall and spring-applied plots measured using the two techniques showed
a decline of flux rate after the peak flux on day 176 (Figure 3.3). After day 176 in the
fall-applied plot, N2 O flux rates measured using both methods consistently remained low
(close to 0) until the end of the observation period when N2 O flux measured using the FG
technique fluctuated between -0.4 to 0.7 nmol m−2 s−1 . The flux measured using chambers
42
(A) Flux−gradient Chambers
5
2
Mean daily flux (nmol N 2 O m −2 s −1 )
−1
(B)
5
−1
150 160 170 180 190 200 210 220 230 240 250 260 270 280 290 300
Day of year
Figure 3.3: Mean daily soil N2 O fluxes measured using chamber and the flux-gradient
method in the fall-applied plot (A) and spring-applied plot (B). Bars indicate standard error.
fluctuated between 0 to 0.4 nmol m−2 s−1 . Meanwhile in the spring-applied plot, N2 O
flux measured using the FG technique fluctuated between -1 to 1 nmol m−2 s−1 while flux
measured using chambers fluctuated between 0 to 0.4 nmol m−2 s−1 .
Due to the large extent of the study plot and the limited number of chambers that could
feasibly be deployed, the chambers were clustered over an area upwind from the FG in-
takes based on the predominant wind direction (180 - 270o ) to ensure that fluxes measured
using chambers were within the predominant flux footprint of the FG. Clustering the cham-
bers instead of evenly spreading them out throughout the field was an effort to reduce the
43
variability due to spatial soil heterogeneity. The strategy employed unfortunately did not
work out as intended during the two days of high flux (DOY 169 and 176) where wind
was predominantly from north and northeast instead of west-southwest that was expected
(Figure 3.4). On day 169, wind was blowing from the predominant wind direction only
27% throughout the whole day while on day 176, no wind blowed from the predominant
wind direction.
(A) (B)
Figure 3.4: Wind rose on day 169 (A) and 176 (B) when high N2 O flux were measured
using chambers.
A t-test performed between daily mean flux measured using chambers and FG during
the whole study period, periods of high (DOY 156-176) and low flux (after DOY 177)
showed no significant difference (P>0.05) between them in both the spring and fall-applied
plots.
44
Cumulative N2 O flux throughout the observation period in the spring-applied plot were
1.13 × 10−1 and 9.51 × 10−2 g N2 O-N m−2 as measured using chambers and the FG
technique, respectively. In the fall-applied plot, cumulative N2 O flux was 1.29 × 10−1 and
4.74 × 10−2 g N2 O-N m−2 as measured using chambers and the FG technique, respectively.
3.3.2 CH4 flux
Throughout the observation period, CH4 flux measured using chambers ranged between
-0.013 to 0.001 nmol CH4 m−2 s−1 (Figure 3.5). Half-hourly flux measured using the
EC technique on the other hand ranged between -8.7 to 9.7 nmol CH4 m−2 s−1 while
calculated daily mean ranged between -0.64 to 2.58 nmol CH4 m−2 s−1 . Methane fluxes
measured using EC also showed no discernible temporal trend. Cumulated emission during
the observation period was -5.83×10−5 and 0.045 g CH4 m−2 as measured using chambers
and the EC technique, respectively.
Despite the higher daily mean flux measured using EC, comparison of actual EC mea-
sured daily mean flux (non-linearly interpolated) was not significantly different by t-test
(P<0.05) when compared to fluxes measured using chambers.
3.3.3 Diurnal trend
Half-hourly fluxes of N2 O measured using the FG technique in both the spring and fall-
applied plot showed no discernible diurnal trend (Figure 3.6). A Kruskal-Wallis one-way
45
3
Chamber
CH4 flux (nmol m−2 s−1 )
Eddy covariance
2
−1
−2
185 195 205 215 225 235 245
Day
Figure 3.5: Mean daily soil CH4 fluxes measured using the eddy covariance method and
mean soil CH4 fluxes measured using chamber.
ANOVA on ranks performed on the half-hourly means within each plot showed no signifi-
cant difference (P<0.05) between half-hourly means.
Similar to N2 O flux, diurnal CH4 flux also showed no discernible trend (Figure 3.7)
and a Kruskal-Wallis one-way ANOVA on ranks performed on the half-hourly means also
showed no significant difference (P<0.05) between half-hourly means.
46
0.5
(A)
0.4
0.3
0.2
0.1
N2 O flux (nmol m−2 s−1 )
0
0130 0330 0530 0730 0930 1130 1330 1530 1730 1930 2130 2330
0.5
(B)
0.4
0.3
0.2
0.1
0
0200 0400 0600 0800 1000 1200 1400 1600 1800 2000 2200 2400
Hour of day
Figure 3.6: Mean diurnal cycle of N2 O fluxes measured in the fall-applied (A) and spring-
applied plot (B) using the FG technique. Bars indicate standard error.
47
3
CH4 flux (nmol m−2 s−1 )
−1
−2
0200 0400 0600 0800 1000 1200 1400 1600 1800 2000 2200 2400
Hour of day
Figure 3.7: Mean diurnal cycle of CH4 fluxes measured using the EC technique. Bars
indicate standard error.
3.4 Discussion
The similar temporal trend of N2 O flux observed by both measurement methods, where
episodes of high fluxes were observed soon after corn was planted has been observed in
many N2 O flux studies (Smith et al., 1994; Roy et al., 2014). However, despite the similar
temporal trend, there were differences of flux magnitude during the peak flux in the two
treatment plots. The difference in magnitude could be attributed to the placement of the
chambers within the field and within corn rows. Consequently, daily mean N2 O flux derived
from measurements using FG was essentially from different source area as the chambers
(Figure 3.8). Inherent soil heterogeneity in addition to possible non-uniformity of nitrogen
48
content in each pass of manure applied very likely caused different sections of the field
to emit varying N2 O flux rates. The difficulty of making direct comparisons between mi-
crometeorological and chamber techniques largely reside on the variable wind direction
which affect comparison of measured fluxes between chamber and micrometeorological
techniques (Christensen et al., 1996).
Nevertheless, even if the mean wind direction during the peak flux was from the 180
- 270o sector, chamber placement within corn rows can still affect the flux magnitude be-
tween techniques. Chambers were confined to only being placed on corn inter-rows and
thus, essentially measured N2 O flux over the inter-row away from living roots whereas the
FG technique spatially integrated fluxes from both inter-rows and crop rows. Corn rows
with living roots have indirect effect on soil gas emission by changing water and nutrient
uptake, and by releasing organic exudate (Rochette, 2011). Therefore, measuring N2 O flux
in the inter-row without the presence of roots can potentially yield different flux rate than
when measured in the corn row. There is no consensus in published studies on the mag-
nitude of N2 O fluxes from inter-row vs. crop row. Ruser et al. (1998) found higher N2 O
emission from crop row than from inter-row positions while Kessavalou et al. (1998) and
Cai et al. (2012) found inter-row to emit higher N2 O flux. Haile-Mariam et al. (2008) on
the other hand found no differences between inter-row and crop row.
Apart from the spatial resolution pitfalls, temporal resolution is also another shortfall
of chambers when measuring fluxes. Ideal daily sampling may not be feasible to most re-
searchers and thus, linear interpolation between successive samplings is relied upon to esti-
49
(A) (B)
(C) (D)
Figure 3.8: The flux footprint of the micrometeorological methods during times of chamber
sampling on day 169 in the fall-applied (A) and spring-applied (C) plots and on day 176
in the fall-applied (B) and spring-applied (D) plots. Yellow box demarcate the cluster of
12 chambers on each plot. The outermost contour line indicate 90% of the source area of
the measured flux. Contour lines and flux footprint was generated using the program Foot-
print (http://www.geos.ed.ac.uk/homes/rclement/micromet/EdiTools/) based on the model
of Kormann and Meixner (2001).
50
mate missing N2 O flux as there is no widely acceptable gap-filling algorithm such as those
available for gap-filling CO2 (Mishurov and Kiely, 2011). Such could be the case when
a complete time series is required for calculating cumulative emission. In the case of the
present study, linear interpolation between chamber fluxes made on days 169 and 176 were
shown to largely overestimate the actual flux as measured using FG (Figure 3.3). Between
days 169 and 176, fluxes measured using FG were shown to be low, with values close to 0
nmol m−2 s−1 contrary to the linearly interpolated chamber fluxes. Consequently, cumula-
tive N2 O flux measured using chambers were calculated to be higher than the cumulative
FG flux. The observation by the FG showed better agreement with previous observations
by Snider et al. (2017) in their conducted study at the same site as the present study us-
15
ing N isotope that spring manure application resulted in higher N2 O emission than the
fall-applied plot due to the rapid cycling of soil nitrogen after manure application.
The overall low CH4 flux observed by both techniques demonstrated that the well
drained field of the site in the present study was not suitable for comparing between mea-
surement techniques as peat (Schrier-Uijl et al., 2010) and acid sulphate soils (Denmead
et al., 2010), sites that are strong sources of CH4 . It was initially anticipated that CH4 con-
sumption would be observed but unfortunately, significant CH4 uptake was not detected.
Similar to N2 O flux, the limited temporal resolution of chambers did not capture the
flux fluctuations that was observed by the EC. Nevertheless, unlike N2 O flux that was
more episodic, CH4 fluctuated within a narrow range during the study period and thus,
the cumulative flux ended up being relatively similar between techniques. Unlike N2 O
51
flux comparison however, comparing CH4 flux measured using EC to measured chamber
flux did not reveal any notable difference in flux magnitude as they were not significantly
different by t-test (P>0.05).
The half-hourly flux of both N2 O and CH4 did not show any significant differences
between each other and no significant diurnal trend. Our observations reiterated that rec-
ommendations by Alves et al. (2012) that sampling between 0900 to 1100 is representative
of the daily mean flux.
3.5 Conclusion
Chambers lack the spatial integration that micrometeorological methods possess making
chambers less suitable for measurement of fluxes over large landscapes due to the need for
a large number of chambers to capture the spatial variability of the site. Chambers also
lack the temporal resolution that is needed to capture intermittent flushes of GHG, espe-
cially for fluxes of N2 O which are highly episodic. In this 210 days study, we show that
although daily mean N2 O fluxes obtained with chambers did not differ significantly from
FG daily mean fluxes, the integrated cumulative N2 O fluxes obtained with chambers were
overestimated. Nevertheless, chambers would still remain relevant for measuring GHG
over smaller scales and to distinguish fluxes of individual sources in an ecosystem. More
accurate estimation of cumulative N2 O flux measured using chambers can be achieved with
better gap-filling algorithms for estimation of fluxes between successive chamber measure-
52
ments and more frequent measurements early in the season, but not later in the season.
The possibility of observing differences between chambers and micrometeorological
methods for measuring CH4 fluxes were hampered due to the well-drained field that re-
sulted in a weak CH4 signal.
No diurnal trends were observed in both N2 O and CH4 fluxes measured using microm-
eteorological methods, increasing the confidence that chamber measurements made on a
single time of the day can represent the daily mean flux.
53
CHAPTER 4
NET ECOSYSTEM CARBON

BUDGET AND GREENHOUSE GAS
BALANCE OF ANNUAL AND
PERENNIAL DAIRY FEED CROP
SYSTEMS
The following Chapter was submitted to the Agriculture, Ecosystems and Environment on
September 7, 2016 and has been accepted for publication on May 1, 2017. I am the primary
author of the paper, with Dr. Claudia Wagner-Riddle, Dr. Paul Voroney, Dr. Jon Warland
and Dr. Philippe Rochette as co-authors on the journal article.
4.1 Introduction
The dairy industry is a large source of greenhouse gases (GHG) emitting 1.97 Gt of CO2 -eq
in 2007 corresponding to 4% of the global GHG emission (FAO, 2010). The major sources
of emission from dairy production at the farm scale are enteric fermentation (a source of
methane, CH4 ), manure storage (a source of CH4 and nitrous oxide, N2 O), manure and
fertilizer use in growing feed crop (N2 O) and use of energy for farm machinery and drying
54
of grain feed (CO2 ) (McGeough et al., 2012; Rotz et al., 2010).
In an effort to mitigate GHG emission from the dairy sector, one of the mitigation strate-
gies proposed is to reduce growing annual crops for animal feed and alternatively increase
the farm land in perennial crops (Conant et al., 2001; Paustian et al., 2016; Soussana et al.,
2010). Perennials cropping systems have been shown to have higher potential over annuals
in sequestering carbon due to their large belowground biomass production and turnover,
and the less frequent use of tillage compared to annuals (Soussana et al., 2004). The re-
duced tillage frequency not only avoids breaking up soil aggregates exposing physically
protected soil organic carbon (SOC), but also reduces CO2 emission from the consumption
of fossil fuel for tillage operation (Lal, 2004).
Conant et al. (2001) observed that conversion of an existing annual cropland to peren-
nial grassland yielded the highest rate of SOC increase among the other management tech-
niques evaluated to increase SOC by increasing forage production. Hawkins et al. (2015)
used an optimization model that considered a range of rations used in intensive milk pro-
duction and suggested that replacing annual corn silage with perennial alfalfa hay as the
primary roughage component can lead to significant decline in GHG emissions from milk
production. However, the carbon sequestration potential of corn and perennial hay needs
to be quantified in field studies.
The net ecosystem carbon budget (NECB) was used to determine the short-term change
of sequestered carbon in an ecosystem as significant changes in SOC can not be detected in
a timescale of between 5-10 years (Smith, 2004). The NECB is the balance between gains
55
in carbon to the system through plant CO2 accumulation and manure application, and what
is lost through net ecosystem exchange (NEE) and plant removal from harvest. As soil
type and climate conditions can greatly affect NEE (Law et al., 2002), it is important for
comparison between different crop types to be conducted in close proximity to each other.
Numerous studies on quantifying NECB have been previously conducted on hay and
corn but with variable conclusions within each respective crop system. Studies by Hirata
et al. (2013) and Taylor et al. (2013) for example showed hay as a carbon sink while that of
Alberti et al. (2010) and Skinner (2007) observed hay was a carbon source. Corn systems
studied by Verma et al. (2005) and Loubet et al. (2011) were carbon sinks while studies
by Alberti et al. (2010), Ceschia et al. (2010), Glenn et al. (2010) and Jans et al. (2010)
showed corn was a carbon source. The contradicting findings of the previous studies can
be largely related to management practices, specifically the amount of biomass harvested
and amount of carbon returned as manure as almost all of the studies showed the annual
cumulated NEE to be negative (CO2 sink).
To our knowledge, few studies have been previously made on comparing the NECB
of annual crop versus perennial hay from side-by-side grown crop fields. One of those
studies was by Alberti et al. (2010) who studied the effects of switching from planting
corn to alfalfa in Northern Italy for two years. They observed that an annual to perennial
crop conversion resulted in the perennial alfalfa in being a larger carbon source compared
to corn, which contradicted the general notion that perennial hay has greater potential to
sequester carbon compared to annual crop. In contrast, several perennial grass species
56
grown for biofuel production acted as carbon sinks whereas a corn-soybean rotation was
a carbon source (Zeri et al., 2011). Taylor et al. (2013) studied the effects of switching
crops from perennial hay to annual oat and canola found that the perennial hay that was
maintained and not converted to annuals was a carbon sink at the end of their study while
the converted plot became a carbon source.
Renovating hay fields is a common practice carried out to increase hay production by
removing weeds and pests, improving drainage, aeration and contour of the field and re-
placing low-producing hay (Bruns, 2012; Stevens et al., 2007). One of the more intense
renovating method practiced is the complete turnover of the soil by plowing followed by
reseeding which is often preferred over less intense methods such as direct seed drilling for
the benefits of controlling weeds and improving aeration (Rutledge et al., 2014). Tillage
can possibly increase CO2 emission from the breaking up of soil aggregates exposing phys-
ically protected SOC. Effects of tillage on N2 O however is inconsistent with some studies
finding increased emission (Wagner-Riddle et al., 2007), reduced emission (Beheydt et al.,
2008) and also inconsistent result between years (Gregorich et al., 2008) or moment of
the year (Rochette and Angers, 1999). Accordingly, the effect of plowing hay on its NEE,
NECB and GHG balance (GHGB i.e. sum of NECB and N2 O emission expressed in CO2 -
equivalent) was also determined in the present study by plowing down the hay field in the
final year of the study.
The objectives of this study were (1) to quantify and compare the NEE and NECB of
a corn and hay field, (2) to calculate the GHG balance of a corn and hay field considering
57
NECB, and N2 O emissions from a previously published study conducted in the same fields,
and (3) to investigate the effects of plowing a hay field on its annual NEE, NECB and GHG
balance.
This study was conducted at the Elora Research Station, Ontario, Canada (43o 38’27.8”N
80o 24’20.4”W). The soil at the site is a Guelph silt loam (fine loamy, mixed, mesic Glos-
soboric Hapludalf ). Select soil physical and chemical properties are listed in Table 4.1.
Table 4.1: Soil physical and chemical properties at different depths

Depth Bulk Density pH C N P K Mg Texture
−3
(cm) (g cm ) (%) (%) (ppm)
0-15 1.26 7.5 2.57 0.25 26.9 180.0 521.9 Silt loam
15-30 1.49 7.6 1.50 0.14 9.8 101.3 462.5 Silt loam
30-45 1.72 7.8 0.86 0.12 3.3 75.8 343.1 Loam
The study was set up on four 4-ha areas (Figure 4.1). Two plots were cropped con-
tinuously to corn (Zea mays) from 2012 to 2014 planted at a seeding rate of 74 000 ha−1 .
Corn was planted on May 10, May 9 and May 26 and harvested on October 2, October
22 and November 4 in years 2012, 2013 and 2014, respectively with cultivars Pioneer R
N19D-3110, DKC38-03 RIB Brand Blend R and Pioneer R N21J-3220 E-Z Refuge planted
in years 2012, 2013 and 2014, respectively. Annual crops were grown in the two plots >10
58
years (soybean, winter-wheat, or corn). The two plots comprising the corn area were part
of a 30-ha (to southeast) area planted with the same crop (Figure 4.1).
A hay crop was established in 2008 in the other two plots and consisted of a grass
(Phleum pratense L.) and legume (Medicago sativa L.) mixture planted at a rate of 6 and
10 kg seed ha−1 , respectively. Hay was cut once in 2012 on May 30 and was cut twice
in 2013 on June 14 and August 28. In October 16, 2013 hay was plowed down (20-cm
depth) and reseeded with Medicago sativa L. cv. Pioneer R 53Q32 in May 26, 2014 at a
seeding rate of 18 kg ha−1 . Hay in 2014 was cut twice on July 27 and August 18. The area
surrounding the hay plots to the northeast and southwest was also grown in hay to ensure
the area was aerodynamically homogeneous (Figure 4.1).
The two 4-ha fields in each crop system were further divided into two 200 × 200 m
subplots where each subplot received different liquid dairy manure application treatments.
In hay, manure was broadcasted in one plot (Hay 1) using a splash-plate system while in
the other plot (Hay 2), manure was placed 10 cm under the soil surface using a coulter
till. Both hay plots were applied with manure on the same day each year in the spring on
April 17, May 7 and May 23 in years 2012, 2013 and 2014, respectively with the exception
of plot Hay 1 in 2014. In corn, one of the plots (Corn 3) was applied with manure in the
fall after harvest each year on October 19, November 13 and November 28 in years 2012,
2013 and 2014, respectively. Plot Corn 4 received manure in the spring before planting on
the same dates that manure was applied in the hay plots. In both corn subplots, manure
59
Figure 4.1: Aerial view of the study site and schematic of the plot layout and position of
the intakes and sonic anemometers and the tunable diode laser gas analyzer system (TDLT-
GAS). Area bordered by yellow and red lines are the hay and corn fields respectively. The
area surrounded by white frames shows the study plots (Map data: Google, DigitalGlobe).
60
was immediately incorporated by disking after application and conventional tillage was
practiced by plowing to 15-20 cm depth after harvest every year. Manure was applied at a
target rate of 80-150 kg N ha−1 and 75-90 kg N ha−1 to the corn and hay plots respectively
(Table 4.2).
Table 4.2: Treatments applied to each plot, manure composition and application rate
Year Plot Treatment D.M. Volume Total N Total C
(%) (m3 ha−1 ) (kg ha−1 )
2012 Hay 1 Broadcast 2.71 56,124 72 554
Hay 2 Subsurface 5.34 37,416 97 728
Corn 3 Fall applied 2.02 74,142 82 546
Corn 4 Spring applied 3.40 125,846 157 1557
2013 Hay 1 Broadcast 3.09 61,785 77 726
Hay 2 Subsurface 3.09 61,785 77 726
2014 Hay 1 No manure n.a n.a n.a n.a
Hay 2 Incorporated 1.79 61,654 91 203
4.2.2 Net ecosystem exchange and N2 O flux measurements
The net CO2 flux or the net ecosystem exchange (NEE) and N2 O flux from each plot were
measured simultaneously using the flux gradient method. A detailed description of daily
N2 O fluxes and its controlling factors as well as data processing is given in Abalos et al.
(2016). This chapter is focused on the NEE results and the integration of annual N2 O
61
emission with NECB to derive a GHGB. The flux-gradient defines the gas flux as:
∂C
F = −K (4.1)
∂z
where K is the eddy diffusivity at height z and ∂C/∂z is the concentration gradient of the
gas of interest at height z. Assumption of similarity between sensible heat, CO2 and N2 O,
and integration of Eq. (4.1) between the two heights (z1 and z2 ) results in:
u∗ κ∆C
F = (4.2)
[ln( zz21 −d
−d
) − ψh2 + ψh1 ]
where u∗ (m s−1 ) is the friction velocity, κ is the von Karman constant (= 0.41), ∆C is
the gas concentration difference between heights z1 and z2 , d is the displacement height,
and ψh2 and ψh1 are the integrated Monin-Obukhov similarity functions for heat for each
sampling height. These functions were calculated using the stability parameter (z-d)/L,
where L is the Obukhov length, as derived by Dyer and Hicks (1970) and Paulson (1970).
Sonic anemometers (CSAT3, Campbell Scientific, Logan, UT) were used to measure
sensible heat flux and u∗ over each crop system (Figure 4.1). The sonic anemometers
were set facing 270o W which was the prevailing wind at the site 2-3 m above mean crop
height (hc ). All instruments were connected to a data logger (CR1000, Campbell-Scientific,
Logan, Utah, USA).
An air sampling assembly was installed in each plot which consisted of two air sample
intakes (17882, Campbell Scientific, Logan, UT, USA) attached to a galvanized steel pipe
62
vertically separated by 50 cm. The bottom intake was set to be between 1.6 to 2.0hc . The
height of the intakes were accordingly adjusted to accommodate the changes in height of
the measured surface.
The end of each intake was connected to a two-way solenoid valve (ASCO R 8320
Series, Florham Park, NJ, USA) housed in a gradient valve assembly unit (17883, Camp-
bell Scientific, Logan, UT, USA) and the outlet of the solenoid valve was connected to a
manifold valve (8×1 Lee Valve, The Lee Company, Westbrook, CT, USA) via 160-180 m
of Synflex R tubing. A relay controller (SDM-CD16AC, Campbell Scientific, Logan, UT,
USA) which was controlled by a datalogger (CR1000, Campbell Scientific, Logan, UT,
USA) was used to actuate the solenoid to alternately switch between sampling the upper
and lower intake every 15 s. The manifold valve was actuated by the relay controller to
select the air sample from one of the plots to be analyzed in sequence for 30 min at a
time. Water was removed from the air sample by a series of two dryers (PD625, Camp-
bell Scientific, Logan, UT, USA) which housed 50 tubes of Nafion R dryer elements. After
the dryers, air was directed to be analyzed for gas concentration by a tunable diode laser
trace gas analyzer system (TDLTGAS) (TGA100, Campbell Scientific, Logan, UT, USA).
The TDLTGAS used in this study was set up in a dual-ramp mode where it simultaneously
measured concentrations of CO2 and N2 O in the same air sample.
The sampling routine was rotated through all four plots in succession resulting in each
plot being analyzed every two hours and each plot having a maximum of 12 half-hourly
gas concentration measurements during the course of one day. The following sign con-
63
vention was used: negative fluxes indicate a net transfer of gas from the atmosphere to the
ecosystem (i.e. ecosystem is a sink) and a positive flux indicates the ecosystem is a source.
Ancillary measurements of air temperature, wind speed and direction and precipita-
tion were obtained from an Environment Canada Reference Climate Station station located
<100 m away from the corn plot. A net radiometer (CNR1, Kipp & Zonen, Deflt, The
Netherlands) was installed on each plot for measurements of global incoming and outgo-
ing radiation and incoming and outgoing longwave radiation. Soil volumetric water content
(θs ) was measured at 5 cm depth using eight time domain reflectometers (CS616, Campbell
Scientific, Logan, UT, USA) where two reflectometers were located on each plot.
Wind direction together with air sample intake heights and position within the plots was
used to select periods when the wind direction allowed for a fetch-to-height ratio of at least
50:1 (horizontal distance to height of measurement ratio) to assure that >80% of the flux
measured originated within the experimental plots (Leclerc and Thurtell, 1990). During
peak corn height of more than 2 m (August-September), the fetch-to-height ratio criterion
was relaxed to extend beyond the corn plot area for winds from the southeast direction,
which consisted of a corn area expected to have similar fluxes.
Missing flux data due to instrumentation failure during inclement weather (i.e. ice
build up, heavy rainfall), instrument removal from site for agronomic activity, and fluxes
that were removed due to data filtering were gap-filled to obtain a complete annual dataset.
The removed fluxes when z<1.4hc were estimated using a model described in Appendix
1. Missing NEE were gap-filled using relationships of measured fluxes with air temper-
64
ature (Ta) and incoming shortwave radiation (Kdown ). Empirical models were derived to
estimate ecosystem respiration fluxes (Re ), gross primary production (GPP) and NEE fol-
lowing procedure described by Barr et al. (2004). The first step of the procedure separated
the measured fluxes into components of Re and GPP where Re is equal to the daily mea-
sured NEE during the non-growing season and nighttime NEE during the growing season.
Nighttime was defined as the time of day between astronomical sunset and sunrise time
(Meeus, 1998) while global radiation was <20 Wm−2 . Gross primary production was de-
fined as the daytime NEE fluxes during the growing season.
The relationship between Re and air temperature (Ta) was then derived by fitting all
the measured Re and corresponding Ta to a logistic function (Eqn. 4.3) while the relation-
ship between GPP and shortwave radiation (Kdown ) was fitted to a rectangular hyperbolic
function (Eqn. 4.4)
r1
Re = (4.3)
(1 + exp[r2 (r3 − Ta )])
α · Kdown · Px
GP P = (4.4)
α · Kdown + Px
where r1 , r2 , r3 , α and Px are fitted parameters.
An additional time-varying parameter was introduced to both functions to take into ac-
count the changing phenological characteristics throughout the growing season according
to Barr et al. (2004). The time varying parameters, rw (t) and Px (t) for the Re logistical
65
equation and the GPP rectangular hyperbolic equation, respectively, were estimated us-
ing the linear regression of modeled Re versus the measured Re , and measured GPP ver-
sus modeled GPP forced through zero within a 100-point moving window resulting in a
new time-varying Re estimation model (Eqn. 4.5) and a new time-varying GPP estimation
model (Eqn. 4.6).
rw (t) · r1
Re = f (Ta , t) = (4.5)
(1 + exp[r2 (r3 − Ta )])
px (t) · α · Kdown · Px
GP P = f (Kdown , t) = (4.6)
α · Kdown + Px
Missing Re and GPP were estimated using Eqn. 4.5 and Eqn. 4.6, respectively. Missing
NEE fluxes were estimated from Eqn. 4.7.
N EE = Re − GP P (4.7)
4.2.3 Net ecosystem carbon budget and greenhouse gas balance
The net ecosystem carbon budget (NECB) was used as a metric to evaluate and compare the
annual net emission or sequestration of carbon from each plot. The NECB was calculated
as the NEE plus carbon exported from each plot through crop harvest (Cremoved ) and carbon
66
imported through manure application (Cadded ) (Eqn. 4.8).
N ECB = N EE + Cremoved + Cadded (4.8)
The Cremoved was estimated from destructive plant sampling described in section 4.2.4
whereas Cadded was calculated from the volume and organic carbon content of the manure
applied. As Cadded is carbon transferred to the ecosystem, a negative sign was assigned.
Similar to the sign convention used for NEE, negative NECB indicates that the cropping
system is a carbon sink, while positive NECB indicates that the cropping system is a carbon
source. The greenhouse gas balance was calculated as the sum of the annual NECB and
the annual N2 O flux reported in Abalos et al. (2016) converted to CO2 equivalent (CO2 -eq)
using the global warming potential of 298 for N2 O.
4.2.4 Plant sampling
Corn plant samples were taken at physiological maturity every year. The corn plots were
subdivided into nine sections, a 5-m row from each of the nine sections was randomly cho-
sen and the number of plants within that 5-m row were counted for plant density calcula-
tion. From the same 5-m row, a 1-m row was selected from which corn cobs were sampled
for grain dry weight determination. For hay harvest, the plots were also subdivided into
nine sections and aboveground plant samples were taken using a 1 x 1 m quadrat randomly
placed within each of the nine subdivisions. Hay samples were taken prior to each cut and
were weighed in the field and oven-dried at 60o C until constant weight for plant moisture
67
and biomass dry matter determination. All plant samples (corn stalk, grain and hay) were
ground to pass through a 1-mm sieve and analyzed for total carbon and nitrogen.
4.2.5 Statistical and uncertainty analyses
To investigate main and interactive effects on CO2 emissions of the crop systems, manure
treatments and inter-annual variation, a two-way repeated measure ANOVA was conducted
on the daily mean of the measured CO2 fluxes of each plot. The Holm-Sidak method was
used for the post-hoc multiple comparison test using SigmaPlot (ver. 11 Systat Software,
San Jose, CA).
Uncertainties from the gap-filling procedure were determined using the bootstrapping
technique following the protocols of Moffat et al. (2007) where 50 different series of NEE,
Re and GPP fluxes in each year were created with artificial gaps of varying gap sizes in-
troduced in each of the series. Each of the 50 time series were then gap-filled using the
procedures described in section 4.2.2. The standard deviation of the 50 gap-filled time
series was then calculated and is presented as the uncertainty of the gap-filling routine.
Details on calculating the gap-filling uncertainties can be found as a supplemental material
in Eichelmann et al. (2016). Uncertainties from gap-filling missing half-hourly N2 O fluxes
by linear interpolation in Abalos et al. (2016) were quantified using similar bootstrapping
technique used for missing NEE.
To determine the statistical significance in the NECB between crop systems and years,
the Gaussian error propagation rule was used to determine the sum of uncertainties from
68
the gap-filling procedure and from destructive plant sampling. The uncertainty from plant
sampling was determined as the standard deviation from the 18 subsamples taken in each
crop system during each destructive plant sampling. Uncertainty from manure application
and chemical analyses were omitted as uncertainty from rate of manure applied in each pass
was unable to be determined while uncertainty from manure analysis was less than 1 g C
m−2 . Values for Re , GPP, NEE, and NECB were considered significantly different between
the years if the numbers including the uncertainty interval did not overlap (Eichelmann
et al., 2016). For N2 O flux, only uncertainty from gap-filling missing fluxes by linear
interpolation was taken into account.
4.3 Results
4.3.1 Meteorological conditions and soil moisture
Each of the three years of this study experienced contrasting weather and climate conditions
(Figure 4.2) with 2012 being between 3 to 6o C warmer than the 30-year normal temperature
from January to March. During the growing season (May to October) of 2012, temperature
was slightly warmer by between 0.2 to 2.7o C than the 30-year normal. Temperature in
2013 was very similar to the 30-year normal while in 2014, temperature was between 3.8
to 5.0o C lower than normal from January to March and 2.6o C lower than normal in July.
Total annual precipitation was below normal (916 mm) in 2012 (679 mm) and 2014
(777 mm) but higher in 2013 (1094 mm). In general, 2012 was a very dry and warm year,
69
2013 was a wet year but with normal air temperature while 2014 was dry with very cold
winter and cool summer.
2012 2013 2014

Air Temperature (o C)
20
15
10
5
0
−5
−10
140
Precipitation (mm)
120
100
80
60
40
20
0
Jan Feb Mar Apr May Jun Jul Aug Sep Oct Nov Dec
Month
Figure 4.2: The 30-year normal monthly mean air temperature (dashed line) and monthly
mean air temperature measured during the three study years (A) and the 30-year normal
total monthly precipitation (vertical columns) and monthly total precipitation (B) measured
during the three study years.
The lowest soil volumetric water content (θs ) recorded in 2012 was 0.23 and 0.10 m3
m−3 in July for the hay and corn plots, respectively. Years 2013 and 2014 did not show
70
severe soil moisture stress as in 2012 and had similar average growing season θs in both
years of 0.4 and 0.3 m3 m−3 in the hay and corn plots, respectively. Soil moisture in hay
was persistently higher than corn (Figure 4.3). Due to the very low precipitation, 2012 was
classified as having a moderate drought (D1) (Agriculture and Agri-Food Canada, 2013)
which was largely due to the absence of rain for one month from June 21 to July 22.
50
Precipitation (mm)
40
30
20
10
0.5
Θs (m3 m-3)
0.3
0.2
0.1
Figure 4.3: Growing season daily total precipitation and mean volumetric soil water content
(θs ) of the two corn and hay subplots measured at 5 cm during the three study years.
4.3.2 Net ecosystem exchange
Mean NEE during the growing season ranged between -2.56 to -5.08 µmol CO2 m−2 s−1
and -5.55 to -8.70 in the hay and corn plots respectively while the non-growing season mean
NEE ranged between 0.33 to 1.12 and 0.54 to 0.97 µmol CO2 m−2 s−1 in the hay and corn
71
plots respectively. Percentage of missing NEE from filtering and removal of equipment
for agricultural operations annually ranged between 38.2% to 47.3% for hay and 43.2% to
56.5% for corn.
Analysis of variance on the NEE in hay and corn showed that there were no year ×
manure treatment interaction effect within hay (P = 0.99) and corn (P = 0.80). No signif-
icant differences were found between the manure treatments within each crop system i.e.
injected vs. broadcast for hay (P = 0.95) and spring vs. fall application for corn (P = 0.65)
in all three years. Accordingly, NEE measurements from the two subplots of each crop
system were pooled together, gap-filled and used for the analysis.
Net carbon uptake (cumulative NEE <0) in hay started early in the year in 2012 (May
6) and 2013 (May 15) (Figure 4.4). Cumulative NEE in hay in 2014, however, was dissim-
ilar to the preceding two years due to the plowing of hay in the fall of 2013 and had net
carbon uptake only after September 6. Total NEE in 2012 (-232 g C m−2 ) was the highest
of the three years in hay whereas cumulative NEE in the two subsequent years were sim-
ilar to each other, -122 and -103 g C m−2 for 2013 and 2014, respectively. The pattern of
cumulative NEE in corn was consistent across all years with corn being a net emitter (cu-
mulative NEE>0) until late May to early June, followed by increasing uptake afterwards
(Figure 4.4). Cumulative NEE was highest in 2012 at -400 g C m−2 while cumulative NEE
in 2013 and 2014 were -64 and -100 g C m−2 respectively.
Corn diurnal NEE was consistently negative (CO2 uptake) in July, August and Septem-
ber of all years, with an early planting in 2012 resulting in significant CO2 uptake in June
72
2012 2013 2014
300
200 Hay
100
0
Cumulative NEE (g C m−2 )
−100
−200
−300
−400
200 Corn
100
0
−100
−200
−300
−400
−500
Feb Apr Jun Aug Oct Dec
Figure 4.4: Cumulative NEE of the hay and corn in all study years.
73
(Figure 4.5). By October, diurnal NEE was zero across all years. Hay diurnal NEE demon-
strated CO2 uptake in May 2012 and 2013 (Figure 4.5). Following ploughing in fall 2013,
hay plots behaved similarly to corn in spring but not in fall. The duration of diurnal NEE<0
for hay was six, five and four months for 2012, 2013 and 2014, respectively while for corn,
the duration of diurnal NEE<0 was four months for 2012 and three months for 2013 and
2014. Months when diurnal NEE rate peaked were also relatively consistent in corn, with
maximum in July 2012 and 2013 and August 2014. Monthly diurnal NEE in hay, on the
other hand, peaked in May, August and September for 2012, 2013 and 2014 respectively.
May Jun Jul Aug Sep Oct

15 15 15
10 10 10
5 5 5
Mean NEE (µmol CO 2 m −2 s −1 )
0 0 0
−5 −5 −5
−10 −10 −10
−15 −15 −15
−20 −20 −20
−25 −25 −25
−30 Hay (2012) −30 Hay (2013) −30 Hay (2014)
−35 −35 −35
0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24
15 15 15
10 10 10
5 5 5
0 0 0
−5 −5 −5
−10 −10 −10
−15 −15 −15
−20 −20 −20
−25 −25 −25
−30 Corn (2012) −30 Corn (2013) −30 Corn (2014)
−35 −35 −35
0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24 0 2 4 6 8 10 12 14 16 18 20 22 24
Hour of day
Figure 4.5: Mean monthly diurnal NEE trend for hay (top panels) and corn (bottom panels)
of each study year.
74
4.3.3 Gross primary production and ecosystem respiration
Daily GPP in hay averaged 0.35±0.27, 0.31±0.25, 0.23±0.31 g C m−2 day−1 and 0.40±0.25,
0.23±0.30, 0.31±0.29 g C m−2 day−1 for corn in 2012, 2013 and 2014, respectively (Fig-
ure 4.6). Hay persistently had longer growing days than corn with 134, 96 and 87 days, as
opposed to 81, 89 and 77 days in 2012, 2013 and 2014, respectively (Figure 4.6). Growing
days in hay is defined as the number of days when mean daily temperature >5o C (Sous-
sana et al., 2007) while in corn, growing days is defined as the duration between sowing to
harvest. Annual mean ± standard deviation of daily Re were 0.069±0.056, 0.084±0.085,
0.053±0.049 g C m−2 day−1 for hay and 0.057±0.082, 0.055±0.077, 0.063±0.074 g C
m−2 day−1 for corn in 2012, 2013 and 2014, respectively (Figure 4.6).
Annual cumulative GPP (Figure 4.7) was consistently higher than annual cumulative
Re in all crop systems and years. The GPP to Re ratios for years 2012, 2013 and 2014 were
1.19, 1.08 and 1.13 in hay, and 1.39, 1.08 and 1.12 in corn.
4.3.4 Yield
Dry matter biomass exported from the system was consistently higher for corn than hay in
all years (Table 4.3). The highest yield for both crop systems was recorded in 2013 when
exported corn grain and hay were 1129 and 813 g m−2 respectively. Within crop system,
the ANOVA of yield between years indicates that there were significant difference for both
corn and hay (P<0.05). Between crop systems, t-test of corn grain and hay yield in each
75
R GEP
(A)
0.4
0.2
Daily flux (g C m −2 day −1 )
−0.2
−0.4
(B)
0.4
0.2
−0.2
−0.4
r−1
2 12 12 12 r−1
3 13 13 13 r−1
4 14 14 14
a u n− e p− e c− a u n− e p− e c− a u n− e p− e c−
M J S D M J S D M J S D
Figure 4.6: Mean daily gross ecosystem production (GEP) and ecosystem respiration (Re )
of the combined hay (A) and corn (B) plots. Gross ecosystem production which is the
inverse of gross primary production (GEP = -GPP) is presented for clarity. Arrows in panel
(A) indicate times when hay were cut. Dashed and solid arrows in panel (B) indicate times
when corn was planted and harvested respectively.
76
2012 2013 2014
1600 Hay (R) 1600 Corn (R)

R (g C m −2 )
R (g C m −2 )
1200 1200
800 800
400 400
0 0
1600 Hay (GPP) 1600 Corn (GPP)
GPP (g C m −2 )
GPP (g C m −2 )
1200 1200
800 800
400 400
0 0
Feb Apr Jun Aug Oct Dec Feb Apr Jun Aug Oct Dec
Figure 4.7: Cumulative gross primary production (GPP) and ecosystem respiration (Re ) in
hay and corn in all study years.
year indicated that corn grain yield was significantly higher than hay in all years (P<0.05).
Yields expressed in nitrogen content varied between years which was dependent on
yield amount and nitrogen content in yield (Table 4.3). Nitrogen content in corn was 1.5%
in 2012 which subsequently declined to 1.10 and 1.03% in 2013 and 2014, respectively.
Nitrogen content in hay was 1.83, 1.48 and 3.83% in 2012, 2013 and 2014, respectively.
Hay in 2014 was predominantly alfalfa hence the sharp increase of hay nitrogen content in
that year. Low corn yield coupled with low nitrogen content resulted in the low nitrogen
expressed in yield for 2014. The low nitrogen expressed in hay yield in 2012 was largely
due to the low yield from the single cut that year.
77
Table 4.3: Mean grain corn and hay yield and their respective nitrogen content to average
yield recorded in Ontario (Statistics Canada, 2012).
Ontario
This study
average yield
Year Crop
Yield Nitrogen
content
(t ha−1 ) (kg N ha−1 ) (t ha−1 )
2012 Corn 9.46 146.31 9.60
Hay 2.00 36.69 4.59
2013 Corn 11.30 123.74 10.10
Hay 8.14 120.72 5.60
2014 Corn 6.75 69.39 10.10
Hay 3.92 150.09 5.72
4.3.5 Net ecosystem carbon budget and greenhouse gas balance
The first year was the only year when both crop systems had a negative NECB, i.e. both
crop systems were carbon sinks (Table 4.4). Hay had a lower NECB compared to corn in
2012 despite its smaller NEE. In the subsequent two years, both crops had positive NECB
values i.e. both crop systems were carbon sources. In 2013, corn was a larger source than
hay which was due to the lower NEE and larger amount of carbon removed compared to
hay. Net ecosystem carbon budget and its components were similar between crop types in
the third year.
Carbon added from manure constituted a small fraction of the NECB relative to carbon
removed with the exception of hay in 2012 when carbon added in both systems was between
6 to 25% of carbon removed through hay and grain harvest. The single hay cut in 2012
skewed the percentage of carbon removed to carbon added to 78%. Amount of carbon
added from manure varied between years due to the difference in the amount of manure
78
applied which depended on the variable nitrogen content in manure.
Hay N2 O emission was relatively small compared to corn during the first two years
(Abalos et al., 2016), which was 17 and 15% of corn N2 O emission in 2012 and 2013 re-
spectively. Hay N2 O emission in 2014 was 8 times higher than the average N2 O emissions
in 2012 and 2013, which increased sharply after the plow down in the preceding fall. Hay
N2 O emission in 2014 was 68% of corn N2 O emission in that year. Between years, corn
greenhouse gas balance (GHGB) was consistently higher than hay and consequently, over
three years, the average GHGB of corn was higher compared to hay (Table 4.5).
79
Table 4.4: Net ecosystem carbon budget±uncertainty of hay and corn from 2012 to 2014 (g C m−2 year−1 ) and its components.
Different subscript letters within each year indicate significant difference between crop systems.
2012 2013 2014 Average
Hay Corn Hay Corn Hay Corn Hay Corn
NEE -232±14 -400±41 -122±36 -64±56 -103±28 -100±35 -152±28 -188±45
80
GPP 1450±17 1403±62 1602±32 1006±43 1019±17 1156±49 1357±30 1189±54
Re 1218±16 1008±82 1479±45 944±66 916±25 1060±25 1205±25 1004±60
Cremoved 82±18 421±61 373±52 512±77 229±50 305±53 208±43 412±65
Cadded -64 -105 -72 -67 -10 -39 -49 -70
NECB -214±23a -93±73b 176±64a 380±96b 57±57a 173±64a 7±51a 154±79b
Table 4.5: The net greenhouse gas balance±uncertainty and its components. NECB = net ecosystem exchange + carbonadded
+ carbonremoved , N2 O = emission expressed in CO2 equivalent using global warming potential of 298 and GHGB = NECB +
N2 O. Unit = g CO2 -eq m−2 year−1 .
2012 2013 2014 Average
81
Hay Corn Hay Corn Hay Corn Hay Corn
NECB -785±84 -388±268 646±235 1358±352 210±209 670±235 24±187 546±290
N2 O 28±8 169±12 33±8 215±9 248±7 365±9 103±8 250±10
GHGB -757±60 -220±189 679±166 1573±249 458±148 1035±166 127±132 796±205
4.4 Discussion
4.4.1 Net ecosystem exchange
Between crop systems, only annual NEE in 2012 showed corn to be significantly higher
than hay while in the subsequent years, annual NEE between crop systems were not sig-
nificantly different from each other. When averaged over the three study years, annual
NEE in corn and hay were not significantly different. Despite corn being a C4 plant that
has higher photosynthetic uptake rate than the C3 hay (Ehleringer et al., 1997; Pearcy and
Ehleringer, 1984), the annual NEE of both crop systems were not significantly different due
to the longer growing period of hay. Previous studies by Alberti et al. (2010) and Zeri et al.
(2011) who compared NEE of corn to perennial legumes and grasses observed annual NEE
in the perennials to be higher than corn. Our results also contradict Taylor et al. (2013) who
compared NEE of hay to oat and canola where they observed hay to have higher annual
NEE than the annual crops over their 30-month study.
Plow down of the hay at the end of the growing season in 2013 did not affect the annual
NEE as it was similar to the annual NEE of the previous year, indicating that renovation of
hay late in the growing season did not turn the field into a net source of CO2 . The trend of
cumulative NEE in 2014 was similar between corn and hay where cumulative NEE peaked
at ≈200 g C m−2 on day 180 before declining after that (Figure 4.4). Beyond the peak
emission however, the rate of CO2 uptake can be seen to be higher in corn than hay from
the steeper slope of the cumulative NEE and also from the monthly diurnal rate (Figure
82
4.5), again highlighting the superiority of photosynthetic uptake of corn over hay. Our
observations indicated that CO2 emission after plow down of hay up until emergence of the
newly seeded crop is similar to emission of a conventionally tilled corn where cumulated
CO2 emission before crop emergence is offset during crop growth. Our observation for hay
concurred with that of Maas et al. (2013) in their comparison of NEE of a newly established
perennial hay and annual rapeseed and wheat. They found that plowing for soil preparation
did not result in the hay being a net source of CO2 . A contrasting observation however
was made by Fraser and Amiro (2013) who reported that plowing of perennial hay was a
net source of CO2 which they attributed to the low CO2 uptake of the newly established
crop. Interestingly, they did not find a significant increase of CO2 emission after plowing
compared to their undisturbed perennial control plot, which contrasted to our findings.
Nevertheless, the present and the aforementioned studies highlight the importance of the
succeeding crop established after plow down to offset CO2 emitted until crop emergence.
4.4.2 Effects of climate variability
The climate conditions in 2012 are an anomaly when compared with the subsequent two
study years and the 30-year average. The peculiar climate condition was also experienced
by neighboring United States which encountered the most severe drought since the Dust
Bowl period, along with the warmest spring on record (Wolf et al., 2016). The climatic con-
ditions in 2012 had induced an earlier and warmer spring followed by a moderate drought
in the summer and a warmer fall and winter which remarkably led to the highest annual
83
NEE of the three study years.
The mode by which the anomalous climatic conditions in 2012 resulted in the high
NEE was different in hay and corn. In hay, the earlier and warmer spring resulted in CO2
uptake (i.e. GPP) as early as March in that year and continued until the first cut at the end of
May. The onset of the drought coincided with the first cut which consequently impede hay
regrowth and sharply decreased GPP. Reduction of GPP following hay cut during drought
was also reported to have occurred in a lowland grassland site in Switzerland (Wolf et al.,
2013).
The rate of the reduced GPP remained constant throughout the growing season but due
to the warmer fall and late onset of winter, CO2 uptake continued until the end of the year.
The combination of the early spring and late winter (i.e. longer growing period) eventu-
ally compensated for the reduction of GPP due to the summer drought. Our observation
however did not concur with Piao et al. (2008) in their modeling study using satellite ob-
servation data. Piao et al. (2008) suggested that while earlier spring increased GPP, the
warmer fall which prolonged the growing period resulted in higher Re than GPP, ultimately
leading to an annual net carbon loss. Angert et al. (2005) also suggested that increased
GPP due to earlier spring would eventually be canceled out by respiration that occurs un-
der drought in the summer. The analysis of Angert et al. (2005) however only considered
spring and summer fluxes.
Unlike hay, the effect of the anomalous climate conditions on higher NEE in corn in
2012 was not due to prolonged growing period i.e. longer CO2 uptake. In fact, the growing
84
period for corn in 2012 was the shortest (133 d) when compared to 2013 (166 d) and 2014
(162 d). The warmer temperature in 2012 had induced high GPP rates in June and July,
which were highest of the three years while not increasing Re (Figure 4.5). The GPP of
C4 plants is known to increase in proportion with temperature up to 40o C (Pearcy and
Ehleringer, 1984). Although low soil water content was observed in the soil surface layer
(0-5 cm), corn GPP was not affected as no reduction during the drought period (June-
August) was shown by the monthly diurnal NEE rates when compared across the three
years.
4.4.3 Net ecosystem carbon budget
Despite both crop systems having NEE <0 in all years, when taking into account the
amount of carbon removed as grain and hay, the net ecosystem carbon budget (NECB)
in both crop systems was negative (net carbon sink) only in 2012 while the subsequent two
years had positive NECB (net carbon source). Averaged over the three study years, the
mean NECB of hay was significantly lower than corn indicating that hay is a small carbon
source or carbon neutral compared to corn.
The main factor leading to the differences between the two crop systems was the greater
amount of carbon removed in corn than in hay in every study year. Carbon removed was
also the dominant factor in determining the NECB in previous studies on corn (Table 4.6)
and hay (Table 4.7); almost all of them showing to be CO2 sinks based on NEE (NEE <0)
but nearly half of the studies in each crop system being carbon sources due to the high
85
amount of carbon removed. In studies on silage corn (Ceschia et al., 2010; Jans et al.,
2010; Loubet et al., 2011) where all aboveground biomass was removed, the NECB was
remarkably larger than studies on grain corn which had negative NECB (Loubet et al.,
2011; Verma et al., 2005). However, in a highly productive irrigated grain corn (Alberti
et al., 2010; Verma et al., 2005), the high removal of carbon shifted the NECB to a carbon
source. The corn yield in our study was within the range of highly productive corn as our
grain yield is on par with Ontario average (Table 4.3) and higher than the average grain
yield reported in the above-mentioned studies. It should be noted that of the three studies
listed in Table 4.6 that were carbon sinks, two studies in corn (grain, Loubet et al. (2011)
and rain-fed, Verma et al. (2005)) had lower yield than the present study which nevertheless
demonstrated that there is potential for corn fields to be carbon sinks but at the cost of lower
grain output.
High biomass carbon removal does not only affect the NECB of corn but can also shift
the NECB of hay towards being a carbon source when harvest rates are high. In the case
of Alberti et al. (2010) for example where hay was cut five to six times per year, the NECB
indicated fields to be a carbon source despite having double the amount of NEE than corn
measured in the same year. Skinner (2008) also highlighted the inability of pastures to
become effective carbon sinks due to the high removal of biomass. In the case of Skinner
(2008), the hay field that was cut three to four times per year eventually was a carbon
source. Hirata et al. (2013) studied a highly productive hay field which was cut three to
86
Table 4.6: Average annual NEE, NECB and carbon input-output from select published
studies (unit: g C m−2 ). Only studies that reported amount of carbon removed were selected
for comparison. For multi-crop rotations studies, only years when corn was planted is
included in the table.
Source Year NEE Cin Cout NECB Remarks
Alberti et al. (2010) 2007– -408 456 48 Grain
2008
Ceschia et al. (2010) 2005– -141 151 705 413 Silage
2007
Glenn et al. (2010) 2007 -72 123 51 Grain
Jans et al. (2010) 2007 -332 -51 750 367 Silage
Loubet et al. (2011) 2005 -397 297 -100 Grain
Loubet et al. (2011) 2008 -380 600 220 Silage
Verma et al. (2005) 2001– -454 316 -138 Rainfed

2005 corn-soybean
Verma et al. (2005) 2001– -551 528 -23 Irrigated
2004 corn-soybean
Verma et al. (2005) 2003– -441 498 57 Irrigated
2004 continuous corn
Average
Other studies (grain) -387 370 -17
Other studies (silage) -284 685 333
This study -188 412 154
87
Table 4.7: Average annual NEE, NECB and carbon input-output from select published
studies (unit: g C m−2 ) made on pasture and grassland. Only studies that reported NECB
and pasture that was not used for grazing were selected for comparison.
Source Year NEE Cin Cout NECB Remarks
Alberti et al. (2010)2007– -549 726 177 Alfalfa
2008
Ammann et al. (2007) 2002– -254 311 57 Extensively
2004 managed
Ammann et al. (2007) 2002– -467 -47 368 -146 Intensively
2004 managed
Hirata et al. (2013) 2005– -189 -291 392 -89 South
2007 Japan
Hirata et al. (2013) 2005– -111 -490 368 -321 North
2007 Japan
Hirata et al. (2013) 2005– -134 -310 446 2 Central
2007 Japan
Skinner (2008) 2003– -65 -13 126 48 Alfalfa
2006
Skinner (2008) 2003– 27 -40 126 113 Grass
2006
Taylor et al. (2013) 2009– -183 143 -40
2011
Average
Other studies -162 -247 293 -47
This study -152 -49 208 7
88
four times per year and had it not been for the high amount of carbon returned to the field
as manure, the NECB of their sites would also have indicated a carbon source. We consider
the hay field in our study to be a low productive hay as it was cut only twice per year but
still, it was a carbon source in two out of the three study years. Rutledge et al. (2015)
argued that grazed hay pastures have the potential of being carbon sinks while providing
forage for dairy cattle. Their argument might be valid for grazed pasture but the NECB of
grazed pasture is also variable and dependent on stocking rate. For a highly productive hay
cut for fodder (>3 cuts per year) however, it may not be possible to fulfill the dual role of
providing high volume of fodder and being a carbon sink at the same time.
From our observation in the present study and the findings of previous studies, we
conclude that for dairy feed crop production systems to be carbon sinks and mitigate GHG
emission from dairy production, feed crop production has to be reduced to minimize the
amount of carbon removed from the crop system. This will be a predicament for dairy
farmers where their ultimate goal is to maximize feed production be it hay or corn to sustain
their dairy herd in times when arable land is becoming less available and the expansion of
land for feed crop is highly unlikely especially in North America and Europe. It appears
that the potential for mitigating GHG through increased planting of perennials exists as
has been demonstrated by previous studies but only in undisturbed perennials which are
not meant for feed production. We suggest that future research be done on the carrying
capacity of feed crop production systems with reduced harvesting rates to investigate the
viability of reducing yield to meet the purpose of crop feed systems especially perennial
89
hay in mitigating GHG emission from dairy production.
4.4.4 Greenhouse gas balance
Nitrous oxide emission constituted between 2.2 to 27.1% and 16.8 to 34.4% of the global
warming potential of the cumulated net CO2 emitted from hay and corn respectively. Ni-
trous oxide emission in hay can be considered to be minimal in relation to the cumulated
net CO2 emitted, with the exception of the emission in 2014 after the plow down, which
increased the availability of organic C and mineral N for nitrification and denitrification
hence increasing N2 O emission in that year (Abalos et al., 2016). Nitrous oxide emission
from corn was 1.5 to 6.5 times higher than hay throughout the three study years. The higher
amount of biomass removed which tilted the NECB of corn to be lower than hay combined
with the higher N2 O emission, ultimately resulting in the different 3-year GHGB average
between the two crop systems.
4.5 Conclusion
In this three-year study we determined the net ecosystem carbon budget (NECB) of corn
and hay fields by measuring the net ecosystem exchange (NEE), and the carbon input and
output. On average, NECB for hay varied from being a carbon sink to source based on the
uncertainties of measurement whereas corn was a net source of carbon. We also quantified
the greenhouse gas balance (GHGB) which is the sum of the NECB and N2 O and found
90
that averaged over the three study years, the GHGB of corn was six times higher than hay.
We conclude that between the two crop systems, hay was carbon neutral whereas corn was
a net carbon emitter hence reiterating the notion that planting perennial crop over annual
crop can potentially mitigate GHG emission. It should be emphasized however that the
carbon neutrality and GHG mitigation potential of hay came at the expense of a lower
feed production per land area compared to corn which in practice, may not be a viable
option for dairy farmers. Future studies on the carrying capacity of feed crop systems with
reduced production are needed to investigate the viability of reducing yield to meet the
purpose of crop feed systems especially perennial hay in mitigating GHG emission from
dairy production.
91
CHAPTER 5
GENERAL CONCLUSION
Studies presented in Chapters 2 and 3 compared the results obtained when measuring trace
gas flux using different methods. Chapter 2 presented the comparison made between net
ecosystem exchange (NEE) measured over agricultural crops using the flux-gradient (FG)
and eddy covariance (EC) techniques during the growing season over two years. The FG
deployed in multiple plot studies are often constrained by having to set the air sample
intake height (z) close to the mean canopy height (hc ) to avoid having a sampling fetch
beyond the study plot while maintaining z to be above the roughness sublayer (RSL). Our
observations showed that NEE measured using FG at 1.4hc <z<2.0hc was comparable to
NEE measured using EC. Net ecosystem exchange can be measured at z as low as 1.4hc
but only under unstable atmospheric conditions as underestimation of NEE measured using
FG under stable atmospheric condition has been demonstrated.
Chapter 3 presented a comparison study of N2 O and CH4 fluxes measured using manual
92
static chambers and micrometeorological methods. The differences between chambers and
micrometeorological methods for measuring CH4 are indistinguishable due to the well-
drained field that was not a strong source of CH4 . Future chamber-micrometeorological
comparison for CH4 should be conducted over agricultural sites identified as potentially
emitting high CH4 rates (e.g. rice paddies) to clearly observe for differences, if any, be-
tween techniques.
While daily mean N2 O fluxes were not significantly different between the two methods,
cumulative N2 O fluxes measured using chambers were found to be higher than cumulative
N2 O fluxes measured using FG. The discrepancies of the integrated flux highlighted the
drawback of chambers in terms of its limited temporal resolution to capture intermittent
flushes of N2 O which are highly episodic. Improvements in gap-filling algorithms for es-
timation of fluxes between successive chamber measurements and more frequent measure-
ments early in the season can alleviate the temporal resolution issues with chambers. Apart
from that, chambers lack the spatial integration that micrometeorological methods possess
making chambers less suitable for measurement of fluxes over large landscapes due to the
need for a large number of chambers to capture the spatial variability of the site. Neverthe-
less, chambers would still remain relevant for measuring GHG over smaller scales and to
distinguish fluxes of individual sources in an ecosystem. Future comparison between the
two techniques for N2 O flux would require making intensive (i.e. more frequent) chamber
measurement during the early part of the growing season until the flux peak was observed.
The core content of this thesis which was presented in Chapter 4, is the estimation
93
and comparison of the net ecosystem exchange (NEE), the net ecosystem carbon budget
(NECB) and the greenhouse gas balance (GHGB) of corn and hay fields, the two main
crops grown for dairy feed in Ontario, Canada. The main motivation was to verify the
claim that perennial crops sequestered more carbon than annual crops. The three-year study
demonstrated that hay varied from being a carbon sink to source based on the uncertainties
of measurement whereas corn was a net source of carbon, concurring with the claim that
perennial crops sequestered more carbon than annual crops. After taking into account
emissions of N2 O from manure application to both crop systems and moldboard plowing
of hay, on average hay was a lower GHG emitter compared to corn. Planting perennial crop
over annual crop can potentially mitigate GHG emission albeit at the expense of a lower
feed production per land area compared to corn which in practice, may not be a viable
option for dairy farmers.
Future studies on the carrying capacity of feed crop production systems with reduced
production intensity are needed to investigate the viability of reducing yield to meet the
purpose of crop feed systems especially perennial hay in mitigating GHG emission from
dairy production.
94
REFERENCES
Abalos, D., Brown, S. E., Vanderzaag, A. C., Gordon, R. J., Dunfield, K. E. and Wagner-
Riddle, C. (2016), ‘Micrometeorological measurements over 3 years reveal differ-
ences in N2 O emissions between annual and perennial crops’, Global Change Biology
22(3), 1244–1255.
Agriculture and Agri-Food Canada (2012), Statistics of the Canadian Dairy Industry, Tech-
nical report, Government of Canada, Ottawa, ON.
Agriculture and Agri-Food Canada (2013), ‘Canadian Drought Monitor (accessed 10-10-
2015)’.
URL: http://www5.agr.gc.ca/eng/?id=1430766471261
Alberti, G., Delle Vedove, G., Zuliani, M., Peressotti, A., Castaldi, S. and Zerbi, G. (2010),
‘Changes in CO2 emissions after crop conversion from continuous maize to alfalfa’,
Agriculture Ecosystems and Environment 136(1-2), 139–147.
Alves, B. J. R., Smith, K. A., Flores, R. A., Cardoso, A. S., Oliveira, W. R. D., Jantalia,
C. P., Urquiaga, S. and Boddey, R. M. (2012), ‘Selection of the most suitable sampling
time for static chambers for the estimation of daily mean N2 O flux from soils’, Soil
Biology and Biochemistry 46, 129–135.
Ammann, C., Flechard, C., Leifeld, J., Neftel, A. and Fuhrer, J. (2007), ‘The carbon budget
of newly established temperate grassland depends on management intensity’, Agricul-
ture, Ecosystems and Environment 121(1-2), 5–20.
Angert, A., Biraud, S., Bonfils, C., Henning, C. C., Buermann, W., Pinzon, J., Tucker, C. J.
and Fung, I. (2005), ‘Drier summers cancel out the CO2 uptake enhancement induced by
warmer springs.’, Proceedings of the National Academy of Sciences of the United States
of America 102(31), 10823–10827.
Barr, A. G., Black, T. A., Hogg, E. H., Kljun, N., Morgenstern, K. and Nesic, Z. (2004),
‘Inter-annual variability in the leaf area index of a boreal aspen-hazelnut forest in relation
to net ecosystem production’, Agricultural and Forest Meteorology 126(3-4), 237–255.
Beheydt, D., Boeckx, P., Ahmed, H. P. and Van Cleemput, O. (2008), ‘N2 O emission from
conventional and minimum-tilled soils’, Biology and Fertility of Soils 44(6), 863–873.
95
Bouwman, A. F., Boumans, L. J. M. and Batjes, N. H. (2002), ‘Emissions of N2 O and NO
from fertilized fields: Summary of available measurement data’, Global Biogeochemical
Cycles 16(4).
Bruns, H. A. (2012), Concepts in Crop Rotations, in G. Aflakpui, ed., ‘Agricultural Sci-

ence’, In Tech, pp. 25–48.
Burba, G. (2013), Part Three: Implementing An Eddy Covariance Experiment, in ‘Eddy

Covariance Method for Scientific, Industrial, Agricultural and Regulatory Applications:
A Field Book on Measuring Ecosystem Gas Exchange and Areal Emission Rates’, LI-
COR Biosciences, Lincoln, Nebraska, USA, p. 151.
Burba, G. G., McDermitt, D. K., Grelle, A., Anderson, D. J. and Xu, L. (2008), ‘Addressing
the influence of instrument surface heat exchange on the measurements of CO2 flux from
open-path gas analyzers’, Global Change Biology 14(8), 1854–1876.
Businger, J. A. (1986), ‘Evaluation of the Accuracy with Which Dry Deposition Can Be
Measured with Current Micrometeorological Techniques’, Journal of Climate and Ap-
plied Meteorology 25(8), 1100–1124.
Cai, Y., Ding, W. and Luo, J. (2012), ‘Spatial variation of nitrous oxide emission between
interrow soil and interrow plus row soil in a long-term maize cultivated sandy loam soil’,
Geoderma 181-182, 2–10.
Canadian Dairy Information Centre (2015), ‘Farming Revenue and Expenses. (accessed
03-05-2017)’.
URL: http://www.dairyinfo.gc.ca/index e.php?s1=dff-fcil&s2=farm–ferme&s3=rev–
dep
Cellier, P. and Brunet, Y. (1992), ‘Flux-gradient relationships above tall plant canopies’,
Agricultural and Forest Meteorology 58(1-2), 93–117.
Ceschia, E., Béziat, P., Dejoux, J. F., Aubinet, M., Bernhofer, C., Bodson, B., Buchmann,
N., Carrara, A., Cellier, P., Di Tommasi, P., Elbers, J. A., Eugster, W., Gr??nwald, T.,
Jacobs, C. M. J., Jans, W. W. P., Jones, M., Kutsch, W., Lanigan, G., Magliulo, E.,
Marloie, O., Moors, E. J., Moureaux, C., Olioso, A., Osborne, B., Sanz, M. J., Saunders,
M., Smith, P., Soegaard, H. and Wattenbach, M. (2010), ‘Management effects on net
ecosystem carbon and GHG budgets at European crop sites’, Agriculture, Ecosystems
and Environment 139(3), 363–383.
Cheng, H. and Castro, I. P. (2002), ‘Near wall flow over urban-like roughness’, Boundary-
Layer Meteorology 104(2), 229–259.
Christensen, S., Ambus, P., Arah, J. R. M., Clayton, H., Galle, B., Griffith, D. W. T., Har-
greaves, K. J., Klemedtsson, L., Lind, A. M., Maag, M., Scott, A., Skiba, U., Smith,
96
K. A., Welling, M. and Wienhold, F. G. (1996), ‘Nitrous oxide emission from an agricul-
tural field: Comparison between measurements by flux chamber and micrometerological
techniques’, Atmospheric Environment 30(24), 4183–4190.
Cicerone, R. J., Shetter, J. D. and Delwiche, C. C. (1983), ‘Seasonal variation of

methane flux from a California rice paddy’, Journal of Geophysical Research: Oceans
88(C15), 11022–11024.
Cole, C. V., Duxbury, J., Freney, J., Heinemeyer, O., Minami, K., Mosier, A., Paustian,
K., Rosenberg, N., Sampson, N., Sauerbeck, D. and Zhao, Q. (1997), ‘Global estimates
of potential mitigation of greenhouse gas emissions by agriculture’, Nutrient Cycling in
Agroecosystems 49, 221–228.
Conant, R. T., Paustian, K. and Elliott, E. T. (2001), ‘Grassland management and conver-
sion into grassland: Effects on soil carbon’, Ecological Applications 11(2), 343–355.
Denmead, O. T. (2008), ‘Approaches to measuring fluxes of methane and nitrous oxide

between landscapes and the atmosphere’, Plant and Soil 309(1-2), 5–24.
Denmead, O. T., Macdonald, B. C. T., Bryant, G., Naylor, T., Wilson, S., Griffith, D. W. T.,
Wang, W. J., Salter, B., White, I. and Moody, P. W. (2010), ‘Emissions of methane
and nitrous oxide from Australian sugarcane soils’, Agricultural and Forest Meteorology
150(6), 748–756.
Dyer, A. J. and Hicks, B. B. (1970), ‘Flux-gradient relationships in the constant flux layer’,
Quarterly Journal of the Royal Meteorological Society 96(410), 715–721.
ÉcoRessources (2015), The Economic Impacts of the Canadian Dairy Industry in 2013,
Technical report, Dairy Farmers of Canada, Quebec City, QC.
Ehleringer, J. R., Cerling, T. E. and Helliker, B. R. (1997), ‘C4 photosynthesis, atmospheric

CO2 , and climate’.
Eichelmann, E., Wagner-Riddle, C., Warland, J., Deen, B. and Voroney, P. (2016), ‘Carbon
dioxide exchange dynamics over a mature switchgrass stand’, GCB Bioenergy 8(2), 428–
442.
Environment Canada (2015), National Inventory Report 1990-2013: Greenhouse Gas

Sources and Sinks in Canada, Technical report, Government of Canada.
Evans, J. D. (1996), Straightforward Statistics for the Behavioral Sciences, Brooks/Cole

Publishing Company.
FAO (2010), Greenhouse Gas Emissions from the Dairy Sector - A Life Cycle Assessment,
Technical report, Food and Agriculture Organization of the United Nations, Rome, Italy.
97
Fazu, C. and Schwerdtfeger, P. (1989), ‘Flux-gradient relationships for momentum and heat
over a rough natural surface’, Quarterly Journal of the Royal Meteorological Society
115(486), 335–352.
Firestone, M. K., Firestone, R. B. and Tiedje, J. M. (1980), ‘Nitrous Oxide from Soil
Denitrification: Factors Controlling Its Biological Production’, Science 208(4445), 749
LP – 751.
Florens, E., Eiff, O. and Moulin, F. (2013), ‘Defining the roughness sublayer and its turbu-
lence statistics’, Experiments in Fluids 54(4), 1500.
Foken, T. (2008), Specifics of the Near-Surface Turbulence, in ‘Micrometeorology’,

Springer-Verlag, Berlin Heidelberg, DE, p. 306.
Fraser, T. J. and Amiro, B. D. (2013), ‘Initial carbon dynamics of perennial grassland con-
version for annual cropping in Manitoba’, Canadian Journal of Soil Science 93(3), 379–
391.
Garratt, J. R. (1978), ‘Flux profile relations above tall vegetation’, Quarterly Journal of the
Royal Meteorological Society 104(439), 199–211.
Glenn, A. J., Amiro, B. D., Tenuta, M., Stewart, S. E. and Wagner-Riddle, C. (2010),
‘Carbon dioxide exchange in a northern Prairie cropping system over three years’, Agri-
cultural and Forest Meteorology 150(7-8), 908–918.
Gregorich, E. G., Rochette, P., St-Georges, P., McKim, U. F. and Chan, C. (2008), ‘Tillage
effects on N2 O emission from soils under corn and soybeans in Eastern Canada’, Cana-
dian Journal of Soil Science 88(2), 153–161.
Gregorich, E. G., Rochette, P., VandenBygaart, A. J. and Angers, D. A. (2005), ‘Green-

house gas contributions of agricultural soils and potential mitigation practices in Eastern
Canada’, Soil and Tillage Research 83(1), 53–72.
Haile-Mariam, S., Collins, H. P. and Higgins, S. S. (2008), ‘Greenhouse gas fluxes from an
irrigated sweet corn (Zea mays L.)-potato (Solanum tuberosum L.) rotation’, Journal of
Environmental Quality 37, 759–771.
Hawkins, J., Weersink, A., Wagner-Riddle, C. and Fox, G. (2015), ‘Optimizing ration
formulation as a strategy for greenhouse gas mitigation in intensive dairy production
systems’, Agricultural Systems 137, 1–11.
Hendriks, D. M. D., van Huissteden, J. and Dolman, A. J. (2010), ‘Multi-technique assess-

ment of spatial and temporal variability of methane fluxes in a peat meadow’, Agricul-
tural and Forest Meteorology 150(6), 757–774.
98
Hirata, R., Miyata, A., Mano, M., Shimizu, M., Arita, T., Kouda, Y., Matsuura, S., Niimi,
M., Saigusa, T., Mori, A., Hojito, M., Kawamura, O. and Hatano, R. (2013), ‘Carbon
dioxide exchange at four intensively managed grassland sites across different climate
zones of Japan and the influence of manure application on ecosystem carbon and green-
house gas budgets’, Agricultural and Forest Meteorology 177, 57–68.
Hofstra, N. and Bouwman, A. F. (2005), ‘Denitrification in Agricultural Soils: Summa-

rizing Published Data and Estimating Global Annual Rates’, Nutrient Cycling in Agroe-
cosystems 72(3), 267–278.
Hutchinson, G. L. and Livingston, G. P. (1993), Use of Chamber Systems to Measure

Trace Gas Fluxes, in ‘Agricultural Ecosystem Effects on Trace Gases and Global Cli-
mate Change’, ASA Special Publication SV - 55, American Society of Agronomy, Crop
Science Society of America, and Soil Science Society of America, Madison, WI, pp. 63–
78.
Jans, W. W. P., Jacobs, C. M. J., Kruijt, B., Elbers, J. A., Barendse, S. and Moors, E. J.
(2010), ‘Carbon exchange of a maize (Zea mays L.) crop: Influence of phenology’,
Agriculture Ecosystems and Environment 139(3), 316–324.
Jiménez, J. (2004), ‘Turbulent Flows Over Rough Walls’, Annu. Rev. Fluid Mech
36(1991), 173–196.
Johnson, J. M.-F., Franzluebbers, A. J., Weyers, S. L. and Reicosky, D. C. (2007), ‘Agri-

cultural opportunities to mitigate greenhouse gas emissions’, Environmental pollution
150(1), 107–124.
Kaimal, J. C. and Finnigan, J. J. (1994), Flow Over Plant Canopies, in ‘Atmospheric

Boundary Layer Flows: Their Structure and Measurement’, Oxford University Press,
New York, NY, USA, pp. 72–73.
Kessavalou, A., Mosier, A. R., Doran, J. W., Drijber, R. A., Lyon, D. J. and Heinemeyer, O.
(1998), ‘Fluxes of Carbon Dioxide, Nitrous Oxide, and Methane in Grass Sod and Winter
Wheat-Fallow Tillage Management’, Journal of Environmental Quality 27, 1094–1104.
Kormann, R. and Meixner, F. X. (2001), ‘An Analytical Footprint Model For Non-Neutral
Stratification’, Boundary-Layer Meteorology 99(2), 207–224.
Lal, R. (2004), ‘Soil carbon sequestration to mitigate climate change’, Geoderma 123(1-
2), 1–22.
Laville, P., Jambert, C., Cellier, P. and Delmas, R. (1999), ‘Nitrous oxide fluxes from a
fertilised maize crop using micrometeorological and chamber methods’, Agricultural
and Forest Meteorology 96(1-3), 19–38.
99
Law, B., Falge, E., Gu, L., Baldocchi, D., Bakwin, P., Berbigier, P., Davis, K., Dolman,
A., Falk, M., Fuentes, J., Goldstein, A., Granier, A., Grelle, A., Hollinger, D., Janssens,
I., Jarvis, P., Jensen, N., Katul, G., Mahli, Y., Matteucci, G., Meyers, T., Monson, R.,
Munger, W., Oechel, W., Olson, R., Pilegaard, K., Paw U, K., Thorgeirsson, H., Valen-
tini, R., Verma, S., Vesala, T., Wilson, K. and Wofsy, S. (2002), ‘Environmental controls
over carbon dioxide and water vapor exchange of terrestrial vegetation’, Agricultural
and Forest Meteorology 113(1-4), 97–120.
Le Mer, J. and Roger, P. (2001), ‘Production, oxidation, emission and consumption of

methane by soils: A review’, European Journal of Soil Biology 37(1), 25–50.
Leclerc, M. Y. and Thurtell, G. W. (1990), ‘Footprint prediction of scalar fluxes using a

Markovian analysis’, Boundary-Layer Meteorology 52(3), 247–258.
Levenberg, K. (1944), ‘A method for the solution of certain non-linear problems in least
squares’, Quarterly of Applied Mathematics 2(2), 164–168.
Loubet, B., Cellier, P., Fléchard, C., Zurfluh, O., Irvine, M., Lamaud, E., Stella, P., Roche,
R., Durand, B., Flura, D. and Others (2013), ‘Investigating discrepancies in heat, CO2
fluxes and O3 deposition velocity over maize as measured by the eddy-covariance and
the aerodynamic gradient methods’, Agricultural and Forest Meteorology 169, 35–50.
Loubet, B., Laville, P., Lehuger, S., Larmanou, E., Fléchard, C., Mascher, N., Genermont,
S., Roche, R., Ferrara, R. M., Stella, P., Personne, E., Durand, B., Decuq, C., Flura,
D., Masson, S., Fanucci, O., Rampon, J. N., Siemens, J., Kindler, R., Gabrielle, B.,
Schrumpf, M. and Cellier, P. (2011), ‘Carbon, nitrogen and Greenhouse gases budgets
over a four years crop rotation in northern France’, Plant and Soil 343(1-2), 109–137.
Luo, Y. and Zhou, X. (2010), Temporal and Spatial Variations in Soil Respiration, in ‘Soil
Respiration and the Environment’, Academic Press, Burlington, pp. 107–131.
Maas, S. E., Glenn, A. J., Tenuta, M. and Amiro, B. D. (2013), ‘Net CO2 and N2 O exchange
during perennial forage establishment in an annual crop rotation in the Red River Valley,
Manitoba’, Canadian Journal of Soil Science 93(5), 639–652.
Mahrt, L. (2010), ‘Variability and maintenance of turbulence in the very stable boundary
layer’, Boundary-Layer Meteorology 135(1), 1–18.
Marquardt, D. W. (1963), ‘An Algorithm for Least-Squares Estimation of Nonlinear Pa-

rameters’, Journal of the Society for Industrial and Applied Mathematics 11(2), 431–
441.
McGeough, E. J., Little, S. M., Janzen, H. H., McAllister, T. A., McGinn, S. M., Beau-
chemin, K. A., Mc Geough, E. J., Little, S. M., Janzen, H. H., McAllister, T. A., McGinn,
S. M., Beauchemin, K. A., McGeough, E. J., Little, S. M., Janzen, H. H., McAllister,
100
T. A., McGinn, S. M. and Beauchemin, K. A. (2012), ‘Life-cycle assessment of green-
house gas emissions from dairy production in Eastern Canada: A case study’, Journal of
Dairy Science 95(9), 5164–5175.
McGinn, S. M. (2006), ‘Measuring greenhouse gas emissions from point sources in agri-
culture’, Canadian Journal of Soil Science 86(3), 355–371.
Meeus, J. H. (1998), Astronomical Algorithms, Willmann-Bell Inc., Richmond, VA.
Mishurov, M. and Kiely, G. (2011), ‘Gap-filling techniques for the annual sums of nitrous
oxide fluxes’, Agricultural and Forest Meteorology 151(12), 1763–1767.
Moffat, A. M., Papale, D., Reichstein, M., Hollinger, D. Y., Richardson, A. D., Barr,
A. G., Beckstein, C., Braswell, B. H., Churkina, G., Desai, A. R., Falge, E., Gove,
J. H., Heimann, M., Hui, D., Jarvis, A. J., Kattge, J., Noormets, A. and Stauch, V. J.
(2007), ‘Comprehensive comparison of gap-filling techniques for eddy covariance net
carbon fluxes’, Agricultural and Forest Meteorology 147(3-4), 209–232.
Mölder, M., Grelle, A., Lindroth, A. and Halldin, S. (1999), ‘Flux-profile relationships over
a boreal forest - Roughness sublayer corrections’, Agricultural and Forest Meteorology
98-99, 645–658.
Moncrieff, J. B., Massheder, J. M., De Bruin, H., Elbers, J., Friborg, T., Heusinkveld,
B., Kabat, P., Scott, S., Soegaard, H. and Verhoef, A. (1997), ‘A system to measure
surface fluxes of momentum, sensible heat, water vapour and carbon dioxide’, Journal
of Hydrology 188-189(1-4), 589–611.
Monteith, J. L. and Unsworth, M. H. (2013), Chapter 16 - Micrometeorology: (i) Turbulent
Transfer, Profiles, and Fluxes, in ‘Principles of Environmental Physics (Fourth Edition)’,
Academic Press, Boston, pp. 289–320.
Mulvaney, R. L., Khan, S. A. and Mulvaney, C. S. (1997), ‘Nitrogen fertilizers promote
denitrification’, Biology and Fertility of Soils 24(2), 211–220.
Papale, D., Reichstein, M., Aubinet, M., Canfora, E., Bernhofer, C., Kutsch, W., Longdoz,
B., Rambal, S., Valentini, R., Vesala, T. and Yakir, D. (2006), ‘Towards a standardized
processing of Net Ecosystem Exchange measured with eddy covariance technique: al-
gorithms and uncertainty estimation’, Biogeosciences 3(4), 571–583.
Pattey, E., Edwards, G., Strachan, I. B., Desjardins, R. L., Kaharabata, S. and Wagner Rid-
dle, C. (2006), ‘Towards standards for measuring greenhouse gas fluxes from agricultural
fields using instrumented towers’, Canadian Journal of Soil Science 86(3), 373–400.
Paulson, C. A. (1970), ‘The Mathematical Representation of Wind Speed and Temperature
Profiles in the Unstable Atmospheric Surface Layer’, Journal of Applied Meteorology
9(6), 857–861.
101
Paustian, K., Lehmann, J., Ogle, S., Reay, D., Robertson, G. P. and Smith, P. (2016),
‘Climate-smart soils’, Nature 532(7597), 49–57.
Pearcy, R. W. and Ehleringer, J. (1984), ‘Comparative ecophysiology of C3 and C4 plants’,

Plant, Cell and Environment Environment 7(1), 1–13.
Piao, S., Ciais, P., Friedlingstein, P., Peylin, P., Reichstein, M., Luyssaert, S., Margolis,
H., Fang, J., Barr, A., Chen, A., Grelle, A., Hollinger, D. Y., Laurila, T., Lindroth,
A., Richardson, A. D. and Vesala, T. (2008), ‘Net carbon dioxide losses of northern
ecosystems in response to autumn warming’, Nature 451(7174), 49–52.
Pokrajac, D., Campbell, L. J., Nikora, V., Manes, C. and McEwan, I. (2007), ‘Quadrant
analysis of persistent spatial velocity perturbations over square-bar roughness’, Experi-
ments in Fluids 42(3), 413–423.
Prueger, J. H. and Kustas, W. P. (2005), Aerodynamic Methods for Estimating Turbulent

Fluxes, in J. M. Baker and J. L. Hatfield, eds, ‘Micrometeorology in Agricultural Sys-
tems’, American Society of Agronomy, Crop Science Society of America, and Soil Sci-
ence Society of America, Madison, WI, pp. 407–436.
Rannik, Ü., Keronen, P., Hari, P. and Vesala, T. (2004), ‘Estimation of forest–atmosphere
CO2 exchange by eddy covariance and profile techniques’, Agricultural and Forest Me-
teorology 126(1), 141–155.
Raupach, M. R. (1979), ‘Anomalies in Flux-Gradient Relationships Over Forest’,

Boundary-Layer Meteorology 16(3), 467–486.
Raupach, M. R., Antonia, R. A. and Rajagopalan, S. (1991), ‘Rough-Wall Turbulent

Boundary Layers’, Appl. Mech. Rev. 44(1), 1–25.
Raupach, M. R. and Legg, B. J. (1984), ‘The uses and limitations of flux-gradient relation-
ships in micrometeorology’, Agricultural Water Management 8(1-3), 119–131.
Reeves, S. and Wang, W. (2015), ‘Optimum sampling time and frequency for measuring
N2 O emissions from a rain-fed cereal cropping system’, Science of The Total Environ-
ment 530–531, 219–226.
Reichstein, M., Falge, E., Baldocchi, D., Papale, D., Aubinet, M., Berbigier, P., Bernhofer,
C., Buchmann, N., Gilmanov, T., Granier, A., Grünwald, T., Havránková, K., Ilvesniemi,
H., Janous, D., Knohl, A., Laurila, T., Lohila, A., Loustau, D., Matteucci, G., Meyers,
T., Miglietta, F., Ourcival, J. M., Pumpanen, J., Rambal, S., Rotenberg, E., Sanz, M.,
Tenhunen, J., Seufert, G., Vaccari, F., Vesala, T., Yakir, D. and Valentini, R. (2005), ‘On
the separation of net ecosystem exchange into assimilation and ecosystem respiration:
Review and improved algorithm’, Global Change Biology 11(9), 1424–1439.
102
Rochette, P. (2011), ‘Towards a standard non-steady-state chamber methodology for mea-
suring soil N2 O emissions’, Animal Feed Science and Technology 166-167, 141–146.
Rochette, P. and Angers, D. A. (1999), ‘Soil Surface Carbon Dioxide Fluxes Induced by
Spring, Summer, and Fall Moldboard Plowing in a Sandy Loam’, Soil Science Society of
America Journal 63(3), 621–628.
Rochette, P. and Bertrand, N. (2007), Soil-Surface Gas Emissions, in ‘Soil Sampling and
Methods of Analysis, Second Edition’, CRC Press, Boca Raton, FL, USA, pp. 851–861.
Rochette, P. and Eriksen-Hamel, N. S. (2008), ‘Chamber Measurements of Soil Nitrous

Oxide Flux: Are Absolute Values Reliable?’, Soil Science Society of America Journal
72, 331–342.
Rotach, M. W. (1999), ‘On the influence of the urban roughness sublayer on turbulence and
dispersion’, Atmospheric Environment 33(24), 4001–4008.
Rotach, M. W. (2001), ‘Simulation Of Urban-Scale Dispersion Using A Lagrangian

Stochastic Dispersion Model’, Boundary-Layer Meteorology 99(3), 379–410.
Rotz, C. A., Montes, F. and Chianese, D. S. (2010), ‘The carbon footprint of dairy
production systems through partial life cycle assessment.’, Journal of Dairy Science
93(3), 1266–1282.
Roy, A. K., Wagner-Riddle, C., Deen, B., Lauzon, J. and Bruulsema, T. (2014), ‘Nitrogen
application rate, timing and history effects on nitrous oxide emissions from corn (Zea
mays L.)’, Canadian Journal of Soil Science 94(4), 563–573.
Ruser, R., Flessa, H., Schilling, R., Steindl, H. and Beese, F. (1998), ‘Soil compaction and
fertilization effects on nitrous oxide and methane fluxes in potato fields’, Soil Science
Society of America Journal 62(6), 1587–1595.
Rutledge, S., Mudge, P. L., Campbell, D. I., Woodward, S. L., Goodrich, J. P., Wall, A. M.,
Kirschbaum, M. U. F. and Schipper, L. A. (2015), ‘Carbon balance of an intensively
grazed temperate dairy pasture over four years’, Agriculture, Ecosystems and Environ-
ment 206, 10–20.
Rutledge, S., Mudge, P. L., Wallace, D. F., Campbell, D. I., Woodward, S. L., Wall, A. M.
and Schipper, L. A. (2014), ‘CO2 emissions following cultivation of a temperate perma-
nent pasture’, Agriculture, Ecosystems and Environment 184, 21–33.
Schrier-Uijl, A., Kroon, P., Hensen, A., Leffelaar, P., Berendse, F. and Veenendaal, E.
(2010), ‘Comparison of chamber and eddy covariance-based CO2 and CH4 emission
estimates in a heterogeneous grass ecosystem on peat’, Agricultural and Forest Meteo-
rology 150(6), 825–831.
103
Seiler, W., Holzapfel-Pschorn, A., Conrad, R. and Scharffe, D. (1983), ‘Methane emission
from rice paddies’, Journal of Atmospheric Chemistry 1(3), 241–268.
Simpson, I. J., Thurtell, G. W., Neumann, H. H., Den Hartog, G. and Edwards, G. C.
(1998), ‘The validity of similarity theory in the roughness sublayer above forests’,
Boundary-Layer Meteorology 87(1), 69–99.
Skinner, H. (2007), ‘Winter carbon dioxide fluxes in humid-temperate pastures’, Agricul-

tural and Forest Meteorology 144(1-2), 32–43.
Skinner, R. H. (2008), ‘High biomass removal limits carbon sequestration potential of ma-
ture temperate pastures.’, Journal of environmental quality 37(4), 1319–26.
Smith, K. A., Clayton, H., Arab, J. R. M., Christensen, S., Ambus, P., Fowler, D., Harg-
reaves, K. J., Skiba, U., Harris, G. W., Wienhold, F. G., Klemedtsson, L. and Galle, B.
(1994), ‘Micrometeorological and chamber methods for measurement of nitrous oxide
fluxes between soils and the atmosphere: Overview and conclusions’, Journal of Geo-
physical Research 99(D8), 16541.
Smith, P. (2004), ‘How long before a change in soil organic carbon can be detected?’,
Global Change Biology 10(11), 1878–1883.
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, C., Romanovskaya, A., Sperling, F. and Tubiell, F. (2014), Agri-
culture, Forestry and Other Land Use (AFOLU). In: Climate Change 2014: Mitigation
of Climate Change. Contribution of Working Group III to the Fifth Assessment Report
of the Intergovernmental Panel on Climate Change, in O. Edenhofer, R. Pichs-Madruga,
Y. Sokona, E. Farahani, S. Kadner, K. Seyboth, A. Adler, I. Baum, S. Brunner, P. Eick-
emeier, B. Kriemann, J. Savolainen, S. Schlömer, C. von Stechow, T. Zwickel and J. C.
Minx, eds, ‘Climate Change 2014: Mitigation of Climate Change. Contribution of Work-
ing 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, pp. 811–922.
Smith, P., Martino, D., Cai, Z., Gwary, D., Janzen, H., Kumar, P., McCarl, B., Ogle, S.,
O’Mara, F., Rice, C., Scholes, B., Sirotenko, O., Howden, M., McAllister, T., Pan, G.,
Romanenkov, V., Schneider, U., Towprayoon, S., Wattenbach, M. and Smith, J. (2008),
‘Greenhouse gas mitigation in agriculture.’, Philosophical transactions of the Royal So-
ciety of London. Series B, Biological sciences 363(1492), 789–813.
Snider, D. M., Wagner-Riddle, C. and Spoelstra, J. (2017), ‘Stable Isotopes Reveal Rapid
Cycling of Soil Nitrogen after Manure Application’, Journal of Environmental Quality
46(2), 261–271.
104
Soussana, J. F., Allard, V., Pilegaard, K., Ambus, P., Amman, C., Campbell, C., Ceschia,
E., Clifton-Brown, J., Czobel, S., Domingues, R., Flechard, C., Fuhrer, J., Hensen, A.,
Horvath, L., Jones, M., Kasper, G., Martin, C., Nagy, Z., Neftel, A., Raschi, A., Baronti,
S., Rees, R. M., Skiba, U., Stefani, P., Manca, G., Sutton, M., Tuba, Z. and Valentini,
R. (2007), ‘Full accounting of the greenhouse gas (CO2 , N2 O, CH4 ) budget of nine
European grassland sites’, Agriculture, Ecosystems and Environment 121(1-2), 121–134.
Soussana, J. F., Loiseau, P., Vuichard, N., Ceschia, E., Balesdent, J., Chevallier, T. and Ar-
rouays, D. (2004), ‘Carbon cycling and sequestration opportunities in temperate grass-
lands’, Soil Use and Management 20(2), 219–230.
Soussana, J. F., Tallec, T. and Blanfort, V. (2010), ‘Mitigating the greenhouse gas bal-
ance of ruminant production systems through carbon sequestration in grasslands’, Ani-
mal 4(3), 334–350.
Statistics Canada (2012), ‘Census of Agriculture, 2011. Statistics Canada Catalogue no.
95-640-XWE. Ottawa, Ontario. May 10.’.
Stevens, D., Walcroft, J., Casey, M., Burggraaf, V., Fraser, T., Webby, R., Smeaton, D. and
Lambert, G. (2007), ‘Benefit analysis: Literature review and modelling outcomes. Liter-
ature review on effects of pasture renewal in New Zealand and simulation modelling of
those effects on whole farm profitability.’, Report prepared for the NZ Pasture Renewal
Charitable Trust .
Taylor, A. M., Amiro, B. D. and Fraser, T. J. (2013), ‘Net CO2 exchange and carbon budgets
of a three-year crop rotation following conversion of perennial lands to annual cropping
in Manitoba, Canada’, Agricultural and Forest Meteorology 182, 67–75.
Topp, E. and Pattey, E. (1997), ‘Soils as sources and sinks for atmospheric methane’, Cana-
dian Journal of Soil Science 77(2), 167–177.
Vergé, X. P. C., De Kimpe, C. and Desjardins, R. L. (2007), ‘Agricultural production,

greenhouse gas emissions and mitigation potential’, Agricultural and Forest Meteorology
142(2), 255–269.
Verma, S. B., Dobermann, A., Cassman, K. G., Walters, D. T., Knops, J. M., Arkebauer,
T. J., Suyker, A. E., Burba, G. G., Amos, B., Yang, H. S., Ginting, D., Hubbard, K. G.,
Gitelson, A. A. and Walter-Shea, E. A. (2005), ‘Annual carbon dioxide exchange in
irrigated and rainfed maize-based agroecosystems’, Agricultural and Forest Meteorology
131(1-2), 77–96.
Vickers, D. and Mahrt, L. (1997), ‘Quality control and flux sampling problems for tower
and aircraft data’, Journal of Atmospheric and Oceanic Technology 14(3), 512–526.
105
Wagner-Riddle, C., Furon, A., Mclaughlin, N. L., Lee, I., Barbeau, J., Jayasundara, S.,
Parkin, G., von Bertoldi, P. and 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 Biology 13(8), 1722–1736.
Wagner-Riddle, C., Thurtell, G. W. and Edwards, G. C. (2005), Trace gas concentration

measurements for micrometeorological flux quantification, in J. M. Baker and J. L. Hat-
field, eds, ‘Micrometeorology in Agricultural Systems’, American Society of Agronomy,
Crop Science Society of America, and Soil Science Society of America, Madison, WI,
USA, pp. 321–343.
Wagner-Riddle, C., Thurtell, G. W., Kidd, G. K., Beauchamp, E. G. and Sweetman, R.

(1997), ‘Estimates of nitrous oxide emissions from agricultural fields over 28 months’,
Canadian Journal of Soil Science 77(2), 135–144.
Wang, J. M., Murphy, J. G., Geddes, J. A., Winsborough, C. L., Basiliko, N. and Thomas,
S. C. (2013), ‘Methane fluxes measured by eddy covariance and static chamber tech-
niques at a temperate forest in central Ontario, Canada’, Biogeosciences 10(6), 4371–
4382.
Warland, J. S., Dias, G. M. and Thurtell, G. W. (2001), ‘A tunable diode laser system for
ammonia flux measurements over multiple plots’, Environmental Pollution 114(2), 215–
221.
Webb, E. K., Pearman, G. I. and Leuning, R. (1980), ‘Correction of flux measurements for
density effects due to heat and water vapour transfer.’, Quarterly Journal Royal Meteo-
rological Society 106(447), 85–100.
Wolf, S., Eugster, W., Ammann, C., Häni, M., Zielis, S., Hiller, R., Stieger, J., Imer, D.,
Merbold, L. and Buchmann, N. (2013), ‘Contrasting response of grassland versus for-
est carbon and water fluxes to spring drought in Switzerland’, Environmental Research
Letters 8(3), 035007.
Wolf, S., Keenan, T. F., Fisher, J. B., Baldocchi, D. D., Desai, A. R., Richardson, A. D.,
Scott, R. L., Law, B. E., Litvak, M. E., Brunsell, N. A., Peters, W. and van der Laan-
Luijkx, I. T. (2016), ‘Warm spring reduced carbon cycle impact of the 2012 US sum-
mer drought.’, Proceedings of the National Academy of Sciences of the United States of
America pp. 5880–5885.
Yan, X., Shi, S., Du, L. and Xing, G. (2000), ‘Pathways of N2 O emission from rice paddy
soil’, Soil Biology and Biochemistry 32(3), 437–440.
Yu, K. W., Wang, Z. P. and Chen, G. X. (1997), ‘Nitrous oxide and methane transport
through rice plants’, Biology and Fertility of Soils 24(3), 341–343.
106
Yu, L., Wang, H., Wang, G., Song, W., Huang, Y., Li, S. G., Liang, N., Tang, Y. and He,
J. S. (2013), ‘A comparison of methane emission measurements using eddy covariance
and manual and automated chamber-based techniques in Tibetan Plateau alpine wetland’,
Environmental Pollution 181, 81–90.
Zeri, M., Anderson-Teixeira, K., Hickman, G., Masters, M., DeLucia, E. and Bernacchi,
C. J. (2011), ‘Carbon exchange by establishing biofuel crops in Central Illinois’, Agri-
culture, Ecosystems and Environment 144(1), 319–329.
107
Appendix
Appendix 1: Modeling of Growing Season Nighttime Flux
To adjust for the underestimated growing season nighttime flux measured by the FG
technique, an artificial neural network modeling tool (MATLAB and Neural Network Tool-
box Release 2014a, The MathWorks, Inc., Natick, MA) was used to estimate the presumed
flux based on actual flux measured using the EC technique. The modeling tool uses the
Levenberg-Marquardt algorithm (Levenberg, 1944; Marquardt, 1963) to train the neural
network model using half-hourly air temperature and crop height as inputs and CO2 fluxes
measured using the EC technique as outputs. Air temperature was used as one of the input
as temperature is a main driver for plant and soil respiration while crop height was used to
capture the physiological dynamics of corn during the growing season.
The performance of the trained model is shown in Figure A1 with very good correlation
(r = 0.83, P <0.001) between modeled and actual flux measured using the EC technique.
In 2012, the EC instruments was set up in corn for testing purposes for 12 weeks. The
recorded flux during that period was compared to flux modeled using air temperature and
crop height during that period and showed good correlation (r = 0.84, P <0.05) between
them (Fig. A2).
108
Training: R = 0.84 Validation: R = 0.80
20 25
20
15
15
10
10
5
5
0
0
−5 −5
0 10 20 0 10 20
Output
Test: R = 0.82 All: R=0.83

25
25
20 20
15 15
10 10
5 5
0
0
−5
−5
0 10 20 0 10 20
Target
Figure A1: Scatter plots of the modeled vs. actual EC fluxes in corn used for training the
model, testing and validation of the neural-network model
109
18
16 1:1
14
R2 = 0.71
EC flux (µmol CO2 m−2 s−1 )
12
10
−2
−2 0 2 4 6 8 10 12 14 16 18
Modeled flux (µmol CO2 m−2 s−1 )
Figure A2: Scatter plot of the modeled vs. EC nighttime fluxes in corn in 2012
110

Sulaiman Muhammad 201705 PHD

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sulaiman Muhammad 201705 PHD

Uploaded by

Copyright:

Available Formats

Net Ecosystem Carbon Budget of Dairy Feed Crop Systems

Muhammad Firdaus Bin Sulaiman

In partial fulfilment of requirements

Guelph, Ontario, Canada

c Muhammad Firdaus Bin Sulaiman, April, 2017

NET ECOSYSTEM CARBON BUDGET OF DAIRY FEED CROP SYSTEMS

Muhammad Firdaus Bin Sulaiman Advisor:

fast-growing agricultural crops in a multi-plot study, which is often fetch-limited. Net

complements measurement made using micrometeorological methods but chambers lack

potential of annual and perennial crops.

• Thank you to Agricultural Greenhouse Gas Program (AGGP) administered by Agri-

LIST OF TABLES vii

LIST OF FIGURES viii

2 MEASUREMENTS OF CO2 FLUX USING FLUX-GRADIENT AND EDDY

3 MEASUREMENTS OF N2 O AND CH4 FLUXES USING MICROMETEO-

4 NET ECOSYSTEM CARBON BUDGET AND GREENHOUSE GAS BAL-

4.1 Soil physical and chemical properties at different depths . . . . . . . . . . 58

1.1 General Introduction

nant et al., 2001; Paustian et al., 2016; Soussana et al., 2010).

To investigate if carbon sequestration by perennials is larger than annual crops, a three-

questration potential of the crop systems. As N2 O emission from denitrification of nitroge-

Bouwman, 2005; Mulvaney et al., 1997), emissions of N2 O must be incorporated to the

previously demonstrated to increase denitrification and N2 O emission (Wagner-Riddle et al.,

soil tillage (i.e. disking) employed in corn.

analyzer for each plot, increasing cost and measurement error.

treatment plot (Leclerc and Thurtell, 1990; Mölder et al., 1999).

inadequate performance in certain situations. Reference measurements of NEE using the

canopy architecture and aerodynamic properties. Measurements made using FG at vari-

2008; Loubet et al., 2013).

temporally in measuring fluxes when compared to micrometeorological methods. Due to

Several comparisons between micrometeorological methods and chambers have been

1.2 Thesis Format and Objectives

and 4 whereas a general conclusion is presented Chapter 5. Some repetition especially in

trace gas fluxes.

measuring NEE using FG at various z/hc over two crop types.

coverage that micrometeorological methods do for accurate estimation of N2 O and CH4

mends future research topics.

MEASUREMENTS OF CO2 FLUX

CO2 mitigation strategies in agricultural production, periodic monitoring of soil carbon

agricultural ecosystem allows for assessment of sequestered carbon as it directly measures

of sampling at >10 Hz (Wagner-Riddle et al., 2005) whereas FG can be deployed using

FG (Rannik et al., 2004).

cultural management (Wagner-Riddle et al., 1997; Warland et al., 2001; Wagner-Riddle

differences in analyzer precision. Flux-gradient however is not a direct flux measurement

fluxes is recommended as a quality control process (McGinn, 2006).

intermittent turbulence (Mahrt, 2010).

originating from surrounding areas of different vegetation make up and treatments. As

data while avoiding sampling in the RSL.

As FG is still relevant for measuring fluxes over agricultural ecosystems especially

on FG measurements over contrasting agricultural crops is needed to provide guidance

on NEE measurements using FG over two crop types.

2.2 Materials and Methods

2.2.1 Site description

133) and harvested on October 27 (DOY 300).

2.2.2 NEE measurement

(TDLTGAS) (TGA100, Campbell Scientific, Logan, UT, USA).

FEC = ρ¯d w0 s0 (2.3)

kurtosis) (Vickers and Mahrt, 1997).

2014 and 2015 were 0.081 and 0.083 m s−1 , respectively.

2.2.3 Comparison of NEE measured between techniques and statisti-

slopes of the regression (β1 ) were compared against β2 = 1 with Ho : β1 = β2 and H1 : β1 6=

β2 . Significance of the comparison was tested at α = 0.05.

The magnitude by which NEE measured using FG systematically underestimated NEE