Animal Feed Science and Technology 211 (2016) 227–240

Contents lists available at ScienceDirect

Animal Feed Science and Technology

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

A comparison of methane emissions from beef cattle

measured using methane hoods with those measured using
respiration chambers
Shane M. Troy a,∗ , Carol-Anne Duthie a , Dave W. Ross a , Jimmy J. Hyslop b ,
Rainer Roehe a , Anthony Waterhouse a , John A. Rooke a
a
SRUC, West Mains Road, Edinburgh EH9 3JG, United Kingdom
b
SAC Commercial Ltd., West Mains Road, Edinburgh EH9 3JG, United Kingdom

a r t i c l e i n f o a b s t r a c t

Article history: The methane hood (MH) system is a novel method of quantifying methane (CH4 ) emis-
Received 14 August 2015 sions from cattle during feeding. The MH system measures CH4 concentrations exhausted
Received in revised form 2 December 2015 from beneath a hood designed to partially enclose the volume above a feed bin. To test the
Accepted 9 December 2015
system two experiments were conducted on two groups (n = 84 and 80) of finishing beef
cattle of differing breeds, whereby CH4 measurements taken over 56 days using the MH
Keywords: system were compared to CH4 output measured subsequently on the same animals using
Greenhouse gas mitigation
respiration chambers (RC). The primary objective of this study was to compare the MH and
Beef cattle
RC measurements and to develop an equation to predict CH4 measured in RC from mea-
Methane
Measurement technique surements taken during group feeding. The second objective was to determine whether
Respiration chamber the MH system could detect dietary treatment effects from diets designed to reduce CH4
emissions. In experiment 1, cross-bred Charolais and purebred Luing steers were offered 2
contrasting diets consisting of forage to concentrate ratios (g/kg DM) of 500:500 (Mixed,
M1) and 80:920 (Concentrate, C1). Within each diet there were 3 treatments: (i) Control,
(ii) Nitrate (calcium nitrate with 77% nitrate on a DM basis), and (iii) Rapeseed cake (higher
fat content). Both the MH and RC measurements detected differences in CH4 emissions
between diets M1 and C1 (P < 0.001), and between the Control and Nitrate treatments within
diet M1 (P < 0.05). In experiment 2, cross-bred Aberdeen Angus and cross-bred Limousin
steers received the same Mixed diet (M2) with 4 treatments containing nitrate and/or high
oil in a 2 × 2 factorial arrangement: (i) Control, (ii) Nitrate, (iii) Maize dark grains (higher
fat content) and (iv) Combined (both nitrate and higher fat). Again, both the MH and RC
measurements detected reductions in CH4 emissions from animals receiving treatments
amended with nitrate (P < 0.001). However, only the MH measurement technique detected
differences when animals received treatments with higher fat contents (P < 0.001). Using
the CH4 concentrations measured by the MH system and the dry matter intake (DMI) mea-
sured during the 56 days test period prediction models for individual animal daily CH4
output were developed and subsequently validated. The best prediction model produced a
good correlation between predicted and measured CH4 output (R2 = 0.77, P < 0.001), with a
concordance correlation coefficient of 0.84. We conclude that the MH system can be used
to estimate the effects of dietary mitigation strategies on CH4 emissions. Furthermore, the
predicted CH4 output from MH measurement supports the use of this system as a tool for
the genetic selection of cattle based on CH4 emissions.
© 2015 Elsevier B.V. All rights reserved.

http://dx.doi.org/10.1016/j.anifeedsci.2015.12.005
0377-8401/© 2015 Elsevier B.V. All rights reserved.
228 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240

1. Introduction

Global atmospheric concentrations of methane (CH4 ) are increasing, with a global average concentration of 1774 ppb
CH4 in 2005 compared with a 715 ppb pre-industrial value (IPCC, 2013). To meet the food demands of the increasing world
population (expected to be 10.9 billion by 2100; United Nations, 2013), global meat production in 2050 is projected to be
75% higher than 2005 (Alexandratos and Bruinsma, 2012). Meeting this demand for food, whilst reducing the environmental
impact of livestock, is a significant challenge; without additional policies CH4 emissions are predicted to increase 1.6-fold
from 2005 to 2030 (IPCC, 2013). Globally, emissions from livestock production are estimated at 7.1 Gt CO2 -equivalent per
annum, with beef production responsible for 2.9 Gt CO2 -equivalent (Gerber et al., 2013). Methane emissions account for 44%
of the total GHG emissions from beef production systems and are primarily affected by feed intake and quality. Mitigation
strategies for enteric CH4 emissions fall into 3 broad areas; nutrition (e.g., through feed additives and diet manipulation),
breeding and management (Cottle et al., 2011; Grainger and Beauchemin, 2011). Nutritional strategies to reduce enteric CH4
emissions are the most developed and most likely to be applied in the field (Martin et al., 2010). Accurate measurements of
CH4 emissions are therefore required in order to develop and test the effectiveness of CH4 mitigation strategies.
Various methods of measuring CH4 emissions from ruminants have been developed, each with their own advantages and
disadvantages. Housing animals in respiration chambers (RC) has long been used to precisely measure the emissions from
individual ruminants. However, this method is time consuming and expensive, and the animal’s behaviour may be altered
while confined individually in the RC resulting in emissions values that do not accurately relate to emissions from their
normal environment. There is interest in developing new methods of estimating CH4 emissions from ruminants which have
a larger throughput than RC and can be used without removing the animals from the regular production systems. The sulphur
hexafluoride (SF6 ) technique is used successfully to estimate CH4 emissions from individual ruminants at grazing (Hulshof
et al., 2012; McGeough et al., 2010). However, this technique requires the insertion of a permeation tube containing SF6 gas
into the rumen, the attachment of a gas collection apparatus to each animal. It is also labour intensive and requires daily
handling of the animals. Other methods such as micro-meteorological dispersion methods can be used to estimate emissions
from groups of animals in outdoor conditions (Tomkins et al., 2011). However, these techniques cannot measure emissions
from individual animals, nor can they be used for indoor housed animals. Furthermore, the scale of micro-meteorological
techniques makes their use difficult for testing mitigation options (McAllister et al., 2011).
Recently, a number of studies have attempted to directly measure exhaled CH4 concentrations to estimate CH4 output
(g/d) and yield (g/kg DMI) from dairy and beef cattle. Chagunda et al. (2013) and Ricci et al. (2014) successfully estimated
CH4 emissions using a hand-held laser CH4 detector based on infra-red absorption spectroscopy, targeted at the nostrils of
dairy cows and beef steers, respectively. This method is relatively inexpensive and large groups of animals can be measured
in a short time period. However, it requires a trained person to stand and point the detector at individual animals, and
it can be affected by atmospheric conditions, particularly wind. Recently, a number of studies comparing CH4 emissions
measured using the GreenFeed system (C-Lock Inc. Rapid City, USA) with subsequent RC measurements from the same
animals (Hammond et al., 2015; Velazco et al., 2015) found that the average CH4 output measured using both techniques
were similar. Other studies have exploited the fact that dairy cows are confined a number of times daily during milking
to their advantage (Garnsworthy et al., 2012; Lassen et al., 2012). Garnsworthy et al. (2012) accurately predicted the CH4
yield of dairy cows by measuring CH4 concentrations in breath samples taken regularly in a semi-enclosed headspace during
milking. Supplements are offered to the cows at milking and air samples are taken from close to the animals head. However,
this technique is not directly applicable to beef animals in group housing.
A novel methane hood (MH) technique has been developed to measure CH4 concentrations from group-housed animals
during feeding. The primary objective of this study was to compare CH4 measurements for large groups of finishing beef
cattle across two experiments from this MH system to CH4 output measured using RC, in order to develop a CH4 output
prediction equation from MH measurements. The second objective was to investigate whether differences in CH4 emissions
resulting from dietary CH4 mitigation strategies could be detected by the MH system.

2. Materials and methods

Respiration chamber data from two experiments were used in this study (Troy et al., 2015 and unpublished) in which
finishing beef steers were fed a range of dietary treatments. In both experiments CH4 measurements were taken over a
period of 8 weeks using the MH system while the animals were housed in groups. Following the MH measurement phase,
individual animal daily CH4 outputs from the same animals were measured using 6 respiration chambers. Throughout both

Abbreviations: C1, high concentrate diet of experiment 1; CH4 , methane; CMB, combined dietary treatment; CTL, control dietary treatment; DMI, dry
matter intake; GHG, greenhouse gas; M1, mixed forage and concentrate diet on experiment 1; M2, mixed forage and concentrate diet on experiment 2;
MDG, maize dark grains dietary treatment; MH, methane hood; MH-MPR, methane production rate measured by the methane hood system; MH-Yld,
methane production rate corrected for dry matter intake; NIT, nitrate dietary treatment; RC, respiration chamber; RSC, rapeseed cake dietary treatment.
∗ Corresponding author. +44 1315353121.
E-mail address: shane.troy@sruc.ac.uk (S.M. Troy).
 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240 229

Table 1
Ingredient and chemical composition (g/kg DM) of the diets fed to the steers during the methane hood and respiration chamber measurement periods of
experiment 1.

Basal diet Mixed (M1) Concentrate (C1)

a
Treatment Control Nitrate RSC Control Nitrate RSC

Ingredient composition, g/kg DM

Wholecrop barley 312 316 315
Grass silage 189 193 192
Barley 340 392 296 739 803 700
Rapeseed meal 128 43 7 146 57 10
Barley straw 84 82 80
b
Minerals 10 9 10 10 10 10
Molasses 20 21 20 21 21 21
Calcinitc 27 26
Rapeseed cake 160 179

Chemical composition, g/kg DM

DM, g/kg 486 481 484 857 855 861
Ash 51 46 52 35 29 37
Crude protein 144 150 146 133 135 136
ADF 221 207 223 137 118 135
NDF 329 314 321 232 210 217
Starch 263 289 238 434 466 411
Oil 26 25 49 27 27 53
GE, MJ/kg DM 18 17 19 18 18 19
a
RSC:rapeseed cake treatment.
b
Contained (mg/kg): Fe, 6036; Mn, 2200; Zn, 2600; Iodine, 200; Co, 90; Cu, 2500; Se 30; (␮g/kg): vitamin E, 2000; vitamin B12, 1000; vitamin A, 151515;
vitamin D, 2500.
c
Contained (g/kg DM): nitrate, 769; Ca, 229.

experiments all steers were offered one of the experimental dietary treatments ad libitum once daily at approximately 1.05
times of actual daily intake.
Both experiments were conducted at Scotland’s Rural College (SRUC) Beef and Sheep Research Centre in Edinburgh. The
experimental protocol was approved by SRUC’s Animal Welfare and Ethical Review Body, the Animal Experiments Committee
and was conducted in accordance with the requirements of the UK Animals (Scientific Procedures) Act, 1986.

2.1. Experiment 1

The experiment (full details in Troy et al., 2015) was a 2 × 2 × 3 factorial design, consisting of 2 breed types (42 cross-bred
Charolais steers and 42 purebred Luing steers), 2 diet types, and 3 dietary treatments (Table 1). Within each breed, the
steers represented 6 sire groups which were balanced across diets and treatment groups. The animals on each of the 6 basal
diet × treatment combinations were loose-housed indoors in 1 pen per diet × treatment group.
All animals were fed one of two complete basal diets consisting (g/kg, DM basis, forage concentrate) of either 500:500
(Mixed [M1]) or 80:920 (Concentrate [C1]), respectively. There were 3 treatments within both diets M1 and C1 consisting
of: a control treatment containing rapeseed meal as the main protein source (M1-CTL and C1-CTL); a treatment in which
rapeseed meal was replaced with calcium nitrate to give a dietary nitrate content of 21.5 g nitrate/kg DM (M1-NIT and
C1-NIT); and a treatment in which rapeseed meal was replaced with rapeseed cake to increase the fat content of the diet
(M1-RSC and C1-RSC). Nitrate was added in the form of calcium nitrate (Calcinit; Yara, Oslo, Norway), while the rapeseed
cake was a by-product of cold-pressing of rapeseed obtained from a local producer.

2.2. Experiment 2

This experiment was a 2 × 2 × 2 factorial design, consisting of two breed types (40 cross-bred Aberdeen Angus steers and
40 cross-bred Limousin steers) and 4 dietary treatments (Table 2), in a 2 × 2 factorial design containing nitrate and/or high
oil. Within each breed, the steers represented 6 sire groups which were balanced across treatment groups. The animals on
each of the dietary treatments were loose-housed indoors in a single pen per treatment.
All animals were fed 1 complete basal diet (Mixed [M2]) consisting of 500:500 g/kg forage concentrate (DM basis). The
4 dietary treatment consisted of; a control treatment containing rapeseed meal as the main protein source (M2-CTL); a
treatment in which rapeseed meal was replaced with calcium nitrate to give a dietary nitrate content of 21.5 g nitrate/kg DM
(M2-NIT); a treatment in which rapeseed meal was replaced by maize distillers dark grains (M2-MDG) to increase dietary fat
concentration, and a treatment combining both nitrate and maize dark grains (M2-CMB). Nitrate was added in the form of
calcium nitrate (Calcinit; Yara, Oslo, Norway), while maize distillers’ dark grains were a by-product of the whisky distillation
industry obtained from a local producer.
230 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240

Table 2
Ingredient and chemical composition (g/kg DM) of the diets fed to the steers during the methane hood and respiration chamber measurement periods of
experiment 2.

Treatment Control Nitrate MDGa Combined

Ingredient composition, g/kg DM

Wholecrop barley 347 347 346 346
Grass silage 210 211 209 210
Barley 336 388 289 263
Rapeseed meal 79
Mineralsb 9 9 9 9
Molasses 19 21 19 19
Calcinitc 25 25
Maize dark grains 128 127
Chemical composition, g/kg DM
DM, g/kg 533 531 533 533
Ash 52 48 51 51
Crude protein 136 141 136 163
ADF 184 166 184 183
NDF 308 195 317 313
Starch 281 308 264 250
Oil 25.0 23.4 36.7 35.9
GE, MJ/kg DM 18.1 17.6 18.5 18.0
a
MDG:maize dark grains treatment.
b
Contained (mg/kg): Fe, 6036; Mn, 2200; Zn, 2600; Iodine, 200; Co, 90; Cu, 2500; Se 30; (␮g/kg): vitamin E, 2000; vitamin B12, 1000; vitamin A, 151515;
vitamin D, 2500.
c
Contained (g/kg DM): nitrate, 769; Ca, 229.

2.3. Methane hood measurement phase

Before MH measurement began all animals were adapted to their diets and dietary treatments, and were trained to use
the electronic feed bins. In both experiments feed was offered using 32 electronic feeders (HOKO, Insentec, Marknesse, The
Netherlands), split between the pens. The feed bins which were accessed by the animals through an automatic door opened
using an electronic identification (eid) reader which identified eid ear tags. The HOKO feed system recorded the start and end
times of each visit together with corresponding weight of feed in the feed bin. The dry matter content of each treatment was
calculated daily and average daily individual animal dry matter intake (DMI) during the MH measurement phase (MH-DMI)
was then calculated. The growth and performance parameters of these steers were monitored closely during this phase as
described in Duthie et al. (2015). Animals were bedded on wood fibre and sawdust and had ad libitum access to feed and
water.
In order to measure CH4 concentrations during feeding a polycarbonate hood was built over each electronic feeder (Fig. 1).
The bottom of each hood was positioned approximately 20 mm above the top of each feed bin, so as not to interfere with
operation of the feeder. The hoods enclosed the volume above the feed bin on 3 sides. The fourth side was left open to allow
the animals’ heads to enter the bins through the automatic door. The air below each hood was constantly exhausted by a
fan located in an opening on top of the hood at a rate of approximately 1 m3 /min (Fig. 1). Exhaust air velocity was measured
every second using in-line velocity metres placed in each exhaust air duct. These velocity metres were calibrated weekly
using a UKAS calibrated vane anemometre (Testo 417 Anemometer, Testo Ltd. Hampshire, UK), in order to determine air
flow (m3 /hr) from the velocity measurements. The exhaust air was vented away from the pens and the opening to the hoods.
Wind speed in the group pens was recorded as the average speed over a 1 min interval using a weather station (WMR88
Weather Station, Oregon Scientific, Buckinghamshire, UK).
Methane concentrations were measured using 4 infrared multigas analysers (MGA3000, ADC Ltd. Hoddesdon, UK), each
analyser measuring samples from 8 feeders. Air was drawn continuously from the exhaust air duct above each hood (from
between the fan and outlet) to the inlet of one of the gas analysers, using 4 twin diaphragm pumps (Charles Austen, Surrey,
UK). Using a series of solenoid valves, air from each of the 8 hoods connected to each analyser was directed sequentially
through the analyser. An equilibrium period of 45 s was allowed before the CH4 concentration was recorded. Since there
were 8 channels per analyser the concentration of CH4 from each hood was recorded every 6 min. Methane production
rates were calculated by multiplying gas concentration by volumetric dry air flow and correcting to standard temperature
and pressure (25 ◦ C and 1013 hPa). The analysers were calibrated 3 times per week with calibration gases for zero (99.998%
Nitrogen, BOC Ltd. Surrey, UK) and span (500 ppmv CH4 , BOC Ltd. Surrey, UK). The drift in accuracy between calibration
bouts was found to be negligible.
The MH measurement phase lasted for 8 weeks for each experiment, during which time the hoods and analysers ran
continuously, apart from a period of approximately 1 hour each morning when the hoods were lifted to allow refilling of the
feed bins.
 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240 231

G a s sa m p le ta k e n
fro m e x h a u st d u c t
E x h a u st a ir d ire c te d
F a n to e x h a u st a ir a w a y fro m th e g ro u p p e n
b e n e a th h o o d

P o ly c a rb o n a te h o o d
sits a b o v e th e fe e d b in

E ID
re a d e r

A u to m a tic d o o r a ll o w in g
e n try to fe e d b in

H o k o fe e d b in

Fig. 1. Schematic of the methane hood measurement system used to measure CH4 concentrations during feeding. The automatic door was opened when
a photoelectric sensor recorded the presence of the animals head above the door and the EID tag was read. The door closed when the photoelectric sensor
no longer recorded the animal’s presence.

2.4. Respiration chambers measurement phase

Following the 8 week MH measurement phase in both experiments, the steers were incrementally moved, in groups of 6
per week, from the group pens to the Greencow RC facility on the same site. In experiment 1, 76 animals were subsequently
measured in the RC measurement phase, while in experiment 2, 72 animals were measured. The animals remained in
grouping housing until it was their turn to enter the RC facility, during which time DMI monitoring continued.
Before entering the chambers the steers were housed in training pens, identical in size and shape to the pens inside the
chambers. These training pens were designed so that the animals would become accustomed to being housed individually.
After 6 days in the training pens the steers were moved to 1 of 6 indirect open-circuit RC. The steers were allocated to the
RC using a replicated randomised block design so that allocation was balanced for live-weight, breed and treatment. The
method of measurement in the RC is described in Rooke et al. (2014) and Troy et al. (2015). The results of the RC phase of
experiment 1 have been reported by Troy et al. (2015).
232 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240

2.5. Calculations and statistical analysis

The Hoko feeder entry and exit times for each animal were matched with the analyser CH4 concentration for each MH.
Feeding bouts of less than 1 min in duration were removed as there was insufficient time to allow the analyser to equilibrate.
Feeding bouts greater than 1 min in duration were used if a gas measurement was taken from the relevant feed bin between
1 min after the animal’s head first entered the feed bin and 30 s after the animals head was removed from the feed bin.
This 1 min initial cut-off period was required as it took approximately 15 s to pump the sample from the MH exhaust to
the analyser and the analyser then required 45 s to stabilise. The extra 30 s following the feeding bout was included again
to allow for pumping time to the analysers and also because the CH4 concentrations did not begin to reduce until at least
30 s after the animal had removed its head from the bin. Gas concentrations taken outside these time limits were deemed
unsuccessful measurements and removed.
The average of all CH4 measurements for each steer over the 8 week sampling period was used to calculate an average
MH CH4 production rate for each animal (MH-MPR, g/d). Methane yields corrected for DMI were also computed for each
animal (MH-Yld, g/kg DMI). Weekly, fortnightly and four-weekly means for both MH-MPR and MH-Yld were also computed
to compare the repeatability of the MH technique across sampling periods, and to determine the minimum sampling time
required.
In experiment 1, 10 days of MH measurements were discarded due to problems with the gas analysers, while in experi-
ment 2, 10 days were also lost as the Hoko feed bin and gas analyser times were not synchronised properly. In total, there
were 46 successful measurement days from each experiment. Gas samples taken when the average wind speed was greater
than 1.5 m/s were deemed unsuccessful and discarded, as high winds reduced the concentration of CH4 measured by the
MH. This resulted in less than 3% of the CH4 measurements being discarded due to high winds. During the RC measurement
phase measurements from one steer in experiment 1 were discarded as the animal’s level of feed intake decreased substan-
tially (>40%) whilst being housed in the RC, leaving data from a total of 75 individual steers. In experiment 2 a faulty feed
bin in one RC resulted in the feed intake measurement for 1 animal being discarded, leaving full RC data from a total of 71
individual steers.
All statistical analysis was performed using the Statistical Analyses System (SAS 9.3, SAS Institute Inc. North Carolina,
USA). Statistical analyses of both MH and RC CH4 measurements were conducted using the Mixed procedure in SAS with
the fixed effects of breed, diet, and treatment for experiment 1, and breed, nitrate and fat for experiment 2. For the RC mea-
surements the random effects of week and chamber were included. The interaction effects of breed × diet, diet × treatment
and breed × treatment, or breed × nitrate, breed × fat and nitrate × fat were also included in the model when these effects
proved significant (P < 0.05). Repeatability of weekly, fortnightly and four-weekly average MH-MPR values for individual
steers and for treatments across all 8 sampling weeks were investigated using the repeated measures Mixed procedure in
SAS. Data are reported as means with their standard errors of the mean. Differences between means were tested using a
protected least squared means test with probability values of P < 0.05 deemed to be significant, while probability values
between P = 0.05 and P = 0.1 were deemed to indicate a tendency. The RANK procedure was used to rank the animals based
on weekly, fortnightly and four-weekly MH-MPR measurements.
The data from experiments 1 and 2 were combined to predict and validate CH4 outputs (g/d) using measurements taken
during the MH measurement phase. To achieve this, approximately 60% of individual animal MH measurements from each
breed × diet × treatment combination across both experiments were randomly allocated to the prediction group (n = 88),
while the rest were allocated to the validation group (n = 58). Models for predicted chamber CH4 outputs (g/d) were computed
by linear regression using the REG procedure in SAS. The prediction models were validated by applying them to the MH
measurements for the animals in the validation group, and comparing the level of agreement with RC measurements using
the REG procedure in SAS. The accuracy of the individual animal predicted CH4 emissions compared to those measured by
the RC were estimated by computing the concordance correlation coefficient (Lin, 1989) using SAS IML (Interactive Matrix
Language), with a value of 1 denoting perfect agreement and 0 denoting no agreement. The prediction bias of the best two
prediction models were evaluated by examining residual plots for mean and linear bias (Patra and Lalhtiatpuii, 2016). The
RANK procedure was used to rank the animals based on CH4 measurements for both the MH and RC measurement phases.

3. Results

3.1. Repeatability of the methane hood system

In Experiment 1, a greater number of successful MH measurements were taken from animals receiving diet M1 compared
with animals fed diet C1 (P < 0.001). On average 18.4 ± 0.47 successful MH-MPR measurements per day were captured from
each animal on diet M1, while 14.8 ± 0.66 successful measurements per day were taken from animals receiving diet C1.
Treatment had no effect on the number of MH samples captured from each animal (P > 0.05). The rate of sample capture
depended on the amount of time spent eating per day and the length of feeding bouts. For diet M1, the average length of
feeding visit, which provided a useable MH-MPR value, was 8.1 min, while for diet C1 this was 7.5 min. For Experiment 2, on
average 14.3 ± 0.35 MH-MPR measurements per day were captured from each animal on diet M2. Treatment had no effect
on the number of MH samples captured from each animal (P > 0.05).
 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240 233

Table 3
Dry matter intakes (DMI), average CH4 production rate (MH-MPR) and v production rate corrected for DMI (MH-Yld) from the methane hood (MH)
measurement phase, and DMI, CH4 output (g/d) and yield (g/kg DMI) from the respiration chamber (RC) phase for experiment 1.

Basal diet Mixed (M1) Concentrate (C1) Significance

Treatment 1
CTL NIT RSC CTL NIT RSC SEM Diet Treatment Diet × treatment2

MH measurement phase n = 14 n = 14 n = 14 n = 14 n = 14 n = 14
MH-DMI, kg/d 12.0 12.1 11.9 11.0 10.8 11.0 0.34 <0.001 0.96 0.82
MH-MPR, g/d 95.0c 65.6b 92.2c 42.7a 43.0a 43.8a 2.69 <0.001 <0.001 <0.001
MH-Yld, g/kg DMI 7.9c 5.4b 7.8c 3.9a 4.0a 4.0a 0.22 <0.001 <0.001 <0.001
Predicted-CH4 , g/d (Model 6) 232.4 213.1 237.5 159.4 145.4 150.8 5.88 <0.001 <0.05 0.27

RC measurement phase n = 13 n = 12 n = 12 n = 12 n = 13 n = 13
DMI pre-RC3 11.6ab 12.5b 11.5ab 11.4ab 10.1a 11.1ab 0.41 <0.01 0.86 <0.05
RC-DMI, kg/d 9.8 10.4 10.5 10.0 9.1 9.7 0.53 0.05 0.57 0.10
RC-CH4 , g/d 241.2 212.4 241.8 148.8 136.1 149.8 9.97 <0.001 <0.05 0.60
RC-CH4 , g/kg DMI 25.1c 20.6b 23.2c 14.6a 15.3a 15.8a 0.96 <0.001 0.14 <0.05
1
Treatments–CTL:Control; NIT:nitrate; RSC:rapeseed cake.
2
Where there are significant diet × treatment effects, differences are displayed using superscripts: within each row, means without a common superscript
differ (P < 0.05).
3
Average DMI as measured in the group pen environment for the 4 weeks immediately prior to the animal entering the RC facility.

For both experiments 1 and 2 there was differences in individual animal average weekly MH-MPR values (P < 0.001),
average fortnightly MH-MPR values (P < 0.001), and average four-weekly MH-MPR values (P < 0.01 for experiment 1 and
P < 0.05 for experiment 2). For Experiment 1, the ranking of individual animals based on weekly, fortnightly and four-weekly
MH-MPR values was more consistent for animal fed diet M1 than those fed diet C1. For animals fed diet M1 each individual
weekly, fortnightly and four-weekly MH-MPR ranking deviated from the 8-week MH-MPR ranking by an average of 5.2,
3.2 and 2.1 ranking places, respectively. However, for those animals fed on diet C1, this average deviation was 8.0, 6.7 and
4.2 ranking places, for weekly, fortnightly and four-weekly MH-MPR values respectively. Similarly for Experiment 2, the
ranking of animals based on MH-MPR values improved when longer time-periods were used; each individual MH-MPR
ranking deviated from the 8-week MH-MPR ranking by an average of 7.2, 5.5 and 3.5 ranking places, for weekly, fortnightly
and four-weekly rankings, respectively.

3.2. Prediction of individual animal emissions from methane hood system

There was a gap of between 2 and 13 weeks between the final week of MH measurement phase and the individual animal
measurements in the RC phase. Despite this gap, there was good agreement when the individual animal MH-MPR values were
directly compared with their subsequent daily RC-CH4 output (Experiment 1 R2 = 0.64, Experiment 2: R2 = 0.24; P < 0.001).
Similarly, when both measurements were corrected for DMI there was good agreement between individual animal MH-Yld
values and the subsequent RC-CH4 yield (g/kg DMI; Experiment 1: R2 = 0.64, Experiment 2: R2 = 0.31; P < 0.001). Experiment
2 had a smaller range of CH4 emissions than experiment 1 as a result of only having 1 basal diet type. This resulted in poorer
correlations when the individual animal MH values were directly compared with their subsequent RC measurements.
The models used to predict individual animal RC-CH4 emissions (g/d) from MH measurements are given in Table 5. The
model (M6) with the highest correlation used MH-Yld, MH-DMI, DMI-Ratio and Diet factors. The DMI-Ratio factor is used
to correct for the change in individual animal RC-DMI compared to MH-DMI. The best model which did not use the Diet
factor was model M4. Using models M4 and M6 on all the data across both experiments, there is good correlation between
Predicted-CH4 and RC measured CH4 (M4 expt. 1 R2 = 0.71, expt. 2 R2 = 0.63; M6: expt. 1 R2 = 0.74, expt. 2 R2 = 0.68; Fig. 2).
Using M6 only the concordance correlation coefficient between measured and predicted CH4 output was 0.83 for experiment
1, 0.79 for experiment 2, and 0.85 when both experiments were combined. There were no significant mean or linear biases
for models M4 and M6 (Fig. 3; P > 0.05). When all animals measured in the RC phase were ranked by both RC-CH4 (g/d) and
M6 predicted chamber CH4 (g/d), the average deviation of animals predicted CH4 ranking from the measured CH4 ranking
was 8.5 in experiment 1 and 10.8 ranking places and experiment 2.

3.3. Comparison of dietary effects across both methods

3.3.1. Experiment 1
During the MH measurement phase steers offered diet M1 had a greater daily DMI than those offered diet C1 (P < 0.001,
Table 3). This also tended to be the case during the RC measurement phase (P = 0.05). The addition of nitrate or rapeseed cake
to either basal diet did not affect DMI during either measurement phase (P > 0.05). Dry matter intake measured during the
MH measurement phase (11.6 kg/d) was similar to average DMI measured in the group pens in the 4 weeks prior to individual
animals’ RC measurement phase (11.3 kg/day; P = 0.60). However, individual animal DMI during the RC measurement phase
was on average 14% lower than the MH measurement phase, and 12.5% lower than the DMI measured in the 4 weeks pre-RC
measurement (P < 0.001).
234 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240

400
y = 1.0353x - 8.9496
350 R² = 0.7585

300

Predicted CH4 (g/d)

250

200

150
Expt 1 - Diet M1
100 Expt 1 - Diet C1
Expt 2 - Diet M2
50

0
0 50 100 150 200 250 300 350 400

Measured Chamber CH4 (g/d)

400 y = 1.014x - 4.476

R² = 0.7135
350

300
Predicted CH4 (g/d)

250

200

150 Expt 1 - Diet M1

Expt 1 - Diet C1
100
Expt 2 - Diet M2
50

0
0 50 100 150 200 250 300 350 400

Fig. 2. Relationship between individual animal daily CH4 output predicted (from Model M6 (top) and M4 (bottom)) from the methane hood system and
daily CH4 output as measured by the respiration chambers in two experiments (n = 147). Dashed line indicated x = y. Model M6 includes ‘Diet type’ in the
prediction equation while M4 does not.

During the MH measurement phase MH-MPR and MH-Yld values from the animals which received diet M1 were signif-
icantly higher than from the animals fed diet C1 (P < 0.001; Table 3). Similarly, as described in Troy et al. (2015), RC-CH4
emissions were higher from the steers who received diet M1 compared with those steers who received diet C1, when
measured per day and per kg DMI (P < 0.001).
The addition of nitrate to diet M1 resulted in reduced MH-MPR and MH-Yld values compared with M1-CTL (P < 0.001).
However, when nitrate was added to diet C1 there was no reduction in MH-MPR (P = 0.1) or MH-Yld (P = 1) values when
compared with C1-CTL. Similarly, when measured in the RC, the animals who received the M1-NIT treatment had lower
RC-CH4 output (g/d) and yield (g/kg DMI) than the animals who received the M1-CTL (P < 0.01), however the addition of
nitrate caused no reduction in daily RC-CH4 output (P = 0.28) or RC-CH4 yield (P = 0.62) from steers fed the C1-NIT treatment,
compared with C1-CTL. The addition of rapeseed cake to both diets did not reduce MH-MPR (P = 0.95) or MH-Yld (P = 0.1)
values, nor was there a reduction in daily RC-CH4 output (P = 0.90) or yield (P = 0.67), when compared with the Control
treatments. The dietary effects calculated using each individual animals’ Predicted-CH4 emissions are similar those calculated
from the measured RC-CH4 emissions (Table 3).
Fig. 4 shows the average MH-Yld measurement for each dietary treatment compared with the RC-CH4 yield results, with
both measurement methods scaled so that the result for treatment M1-CTL = 1. The MH measurement method showed a
larger percentage reduction in emissions from the diet C1 compared to diet M1 (44%), than was observed from the RC
 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240 235

100

80

60

Residual CH4 (g/day)

40

20

0
0 50 100 150 200 250 300 350
-20

-40

-60

-80
Predicted CH4 (g/day)

120
100
80
Residual CH4 (g/day)

60
40
20
0
0 50 100 150 200 250 300 350
-20
-40
-60
-80

Fig. 3. Plot of residual CH4 production (measured minus predicted CH4 production) versus predicted CH4 production (from Model M6 (top) and M4
(bottom)). Residuals were regressed on the predicted vales where for M6: Y = −9.0(±10.41; P = 0.39) + 0.04X (±0.049; P = 0.47), R2 = 0.0036; for M4:
Y = −4.5(±11.44; P = 0.70) + 0.01X (±0.054; P = 0.79), R2 = 0.0005).

Table 4
Dry matter intakes (DMI), average CH4 production rate (MH-MPR) and CH4 production rate corrected for DMI (MH-Yld) from the methane hood (MH)
measurement phase, and DMI, CH4 output (g/d) and yield (g/kg DMI) from the respiration chamber (RC) phase for experiment 2.

Basal Diet Mixed (M2) Significance

Treatment 1
CTL NIT MDG CMB SEM Nitrate Fat Nitrate × fat2

MH measurement phase n = 20 n = 20 n = 20 n = 20
MH-DMI, kg/d 11.8 11.4 11.8 11.6 0.32 0.76 0.44 0.71
MH-MPR, g/d 121.0c 75.9a 87.5b 69.0a 2.49 <0.001 <0.001 <0.001
MH-Yld, g/kg DMI 10.3c 6.8b 7.4b 6.0a 0.20 <0.001 <0.001 <0.001
Predicted-CH4 , g/d (Model 6) 262.9b 221.3a 232.1a 216.5a 5.79 <0.001 <0.001 <0.01

RC measurement phase n = 18 n = 17 n = 18 n = 18
DMI pre-RC3 11.3 10.9 11.4 11.5 0.28 0.12 0.37 0.18
RC-DMI, kg/d 10.4 9.8 10.2 10.2 0.43 0.48 0.81 0.30
RC-CH4 , g/d 245.5 218.6 238.2 209.9 9.35 <0.001 0.37 0.59
RC-CH4 , g/kg DMI 24.0 22.1 23.4 20.9 0.69 <0.001 0.12 0.82
1
Treatments were a 2 × 2 (nitrate × fat) factorial design–CTL:Control; NIT:nitrate; MDG:maize dark grains; CMB:combined (nitrate + fat).
2
Where there are nitrate × fat interaction effects, differences are displayed using superscripts: within each row, means without a common superscript
differ (P < 0.05).
3
Average DMI as measured in the group pen environment for the 4 weeks immediately prior to the animal entering the RC facility.
236 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240

1.2

0.8

0.6

0.4

0.2

0
C1-CTL C1-NIT C1-RSC M1-CTL M1-NIT M1-RSC
Dietary Treatments
MH-CH4 (g/kg DMI) Chamber CH4 (g/kg DMI)

1.2

0.8

0.6

0.4

0.2

0
0 0 0 0
Dietary Treatments

Fig. 4. Comparison of the mean (±SE) dietary treatment effects as measured by the methane hood and respiration chamber methods in experiment 1 (top)
and experiment 2 (bottom). Scaled to M1-CTL and M2-CTL = 1.

measurements (39%). Similarly, the MH-Yld values recorded from the animals receiving treatment M1-NIT were 31% lower
than those receiving M1-CTL, which is a larger reduction than 18% reduction in RC-CH4 yield between these 2 treatments.
The dietary and treatment effects on MH-MPR values were persistent over time; when the 8 week MH measurement
phase is split into 2 four-week period and analysed separately, similar reductions associated with feeding diet C1 compared
to M1 (P < 0.001), and with the addition of nitrate to diet M1 are found in both periods (P < 0.01). Similarly, neither period
showed a decrease in MH-MPR values when nitrate was added to diet C1 (P = 1), or when fat was added to either diet (P = 0.97
and P = 0.94 for periods 1 and 2, respectively).

3.3.2. Experiment 2
During the MH and RC measurement phases neither the addition of nitrate nor increasing the dietary fat content of diet
M2 adversely affected DMI (Table 4). Dry matter intake measured during the MH measurement phase (11.6 kg/d) was similar
to average DMI measured in the group pens in the 4 weeks prior to individual animals’ RC measurement phase (11.3 kg/d;
P = 0.33). However, individual animal DMI during the RC measurement phase was on average 12.4% lower than during both
the MH measurement phase, and 9.7% lower than the DMI measured in the 4 weeks pre-RC measurement (P < 0.001).
During the MH measurement phase the addition of nitrate to diet M2 resulted in reduced MH-MPR and MH-Yld values
compared with M2-CTL (P < 0.001). Similarly, during the RC phase steers who received treatments amended with nitrate had
lower daily RC-CH4 output and RC-CH4 yield than the animals who received the treatments containing no nitrate (P < 0.001).
For diet M2 MH-MPR and MH-Yld (P < 0.001) values were both lower for treatments with higher oil content. However, when
measured by the RC increasing the oil content did not result in any reductions in daily RC-CH4 output (P = 0.37) or RC-CH4
yield (P = 0.12). For Predicted-CH4 emissions both nitrate addition (P < 0.001) and increased oil content (P < 0.01) reduced
CH4 emissions, which differs from the results from the measured RC-CH4 emissions (P < 0.001), where only nitrate addition
reduced CH4 emissions (Table 4).
Similar to experiment 1, the MH measurement method showed a larger percentage reduction in MH-Yld values from
animals receiving treatment M2-NIT compared with M2-CTL (34%), than the reduction in RC-CH4 yield between these 2
 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240 237

Table 5
Accuracy of models to predict and validate CH4 (g/d) from respiration chambers (RC) based on measurements taken during the methane hood (MH)
measurement period.

Model Methane prediction models1 R2 (prediction R (validation P (validation Concordance

group)2 group)2 group)3 (validation
group)

M1 107.1 + 10.4 MH-DMI − 71.6 Diet 0.56 0.64 <0.0001 0.79

M2 75.73 + 0.77 MH-MPR + 7.30 MH-DMI – 40.6 Diet 0.63 0.67 <0.0001 0.78
M3 −123.8 + 18.5 MH-DMI + 134.6 DMI-Ratio 0.30 0.39 <0.0001 0.55
M4 −121.4 + 1.34 MH-MPR + 8.23 MH-DMI + 153.4 DMI-Ratio 0.72 0.70 <0.0001 0.82
M5 −20.2 + 11.6 MH-DMI + 129.6 DMI-Ratio – 70.8 Diet 0.67 0.68 <0.0001 0.79
M6 −72.1 + 0.91 MH-MPR + 8.20 MH-DMI + 144.9 DMI-Ratio – 34.3 Diet 0.77 0.75 <0.0001 0.84
M7 27.0 + 1.28 MH-MPR + 7.27 MH-DMI 0.54 0.61 <0.0001 0.74
M8 151.0 + 0.90 MH-MPR – 40.5 Diet 0.60 0.57 <0.0001 0.70
1
MH-MPR is the average CH4 output over the 8 week MH sampling period (g/d), MH-DMI is the average daily DMI during this 8 week sampling period
(kg), DMI-Ratio is the ratio of DMI during the RC measurement phase with DMI during the MH measurement phase (RC-DMI/MH-DMI), and diet type is a
correction for diets (mixed forage and concentrate diet = 0 and high concentrate diet = 1).
2
Prediction group, n = 88; validation group, n = 58.
3
The P-value here tells us whether the relationship between the measured chamber CH4 output and the chamber output as predicted from the model
are significant for the validation group; with the ‘null hypothesis’ being ‘hood values and chamber values are unrelated’.

treatments (8%, Fig. 4). Similarly, the reduction resulting from feeding the M2-MDG and M2-CMB treatments respectively
were greater when measured by the MH system (28 and 42%), than when measured by the RC system (2 and 13%).
Similarly to Experiment 1, the effects of the dietary treatments on MH-MPR were persistent over time. Both the addition
of nitrate and increased fat content resulted in similar reductions in MH-MPR values when they are analysed separately as
2 four-week periods (P < 0.001).

4. Discussion

4.1. Detection of treatment effects using the methane hood system

Over both experiments the MH system detected similar dietary treatment effects to the RC. Both MH and RC measurements
showed reductions in CH4 emissions when the proportion of concentrate feedstuff in a diet was increased. This result agrees
with many previous studies which showed increasing the level of cereals, to change the pattern of fermentable carbohydrates
in the diet, reduces CH4 emissions (Aguerre et al., 2011; Rooke et al., 2014). Similarly, both the MH and RC measurements
demonstrated reductions in CH4 emissions when animals nitrate was included in the Mixed diet. The addition of nitrate
to mixed forage and concentrate diets has previously been shown to reduce CH4 emissions from ruminants (Hulshof et al.,
2012; Van Zijderveld et al., 2011; Troy et al., 2015). Neither measurement method detected a reduction in CH4 emissions
when the high concentrate diet was supplemented with nitrate or increased oil.
In experiment 1, neither measurement method detected a reduction in CH4 emissions when the oil content of the Mixed
diet was increased. However, in experiment 2 the MH system detected a reduction in emissions from animals receiving diets
with increased oil content. This may be a result of the higher power in experiment 2 (n = 20) compared with experiment 1
(n = 14). When the animals from experiment 2 were measured in the RC, the effect of increased oil content on CH4 emissions
was no longer detected. This may be a result of a reduction of the effect of increased oil content over time; by the time the
animals were measured in the RC they had received the higher oil diet for between 10 and 23 weeks.

4.2. Using short term breath samples to estimate daily CH4 output

A number of different studies have compared using CH4 measurement techniques which use consolidated short-term
breath measurements with longer term RC measurements of daily CH4 output (Garnsworthy et al., 2012; Hammond et al.,
2015; Hegarty, 2013; Ricci et al., 2014; Velazco et al., 2015). In two experiments comparing daily CH4 output from the same
cattle measured using both the GreenFeed system (C-Lock Inc. South Dakota, USA) and RC, Velazco et al. (2015) found no
differences in CH4 output or CH4 yield between the two measurement methods. They also found a high correlation between
GreenFeed and RC measures of daily CH4 output (R2 = 0.72, n = 10) and yield (R2 = 0.36). Similarly, in two experiments using
growing heifers Hammond et al. (2015) also found daily CH4 output and yield measured using the GreenFeed system was
similar to that measured using RC. However, Hammond et al. (2015) found that the GreenFeed system did not detect the
same dietary treatment effects to those measured in RC. They attributed this to the variability of measurements caused by
the timing and limited number of short-term measurements obtained by the GreenFeed system. Over the three experiments
reported in that study, the animals averaged between 1.60 and 2.34 visits per day, at 5 min per visit, to the GreenFeed
system, with the GreenFeed system reliant on eructation events occurring during these visitation periods. In comparison,
over the two experiments detailed in the present study an average of between 12 and 15 MH system measurements were
obtained from each animal every day, with feeding visits averaging 8 min. This may explain the greater accuracy of the MH
system to detect dietary treatment effects, compared with those reported by Hammond et al. (2015) using the GreenFeed
238 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240

system. Indeed, Hammond et al. (2015) proposed that more frequent GreenFeed measurements were required over a longer
time period to improve the accuracy. The MH system has advantages and disadvantages when compared with other sniffer
techniques such as the GreenFeed system. As the MH system measures CH4 emissions while the animal is assessing the
feed bin, there is no requirement to provide extra feed supplements to entice the animal into the measurement area. This
may make the MH system more appealing for those wishing to investigate dietary effects. However, measuring CH4 during
feeding may result in a daytime bias as animals tend to consume more feed during daylight hours. The MH system is also
affected by high wind speed which may make it unsuitable for some outdoor feedlot systems.
In another study similar to the current one using dairy cows, Garnsworthy et al. (2012) detected similar dietary treatment
effects when CH4 emissions from cows fed 2 different diets were measured using both RC and a novel breath sampling tech-
nique used during milking. This novel method took measurements spanning 3 milking events over 14 days, with between 4
and 15 eructation events per milking. The accuracy of these 3 separate breath sampling techniques may depend on increas-
ing the length and frequency of the breath sampling time. In the current study, there were significant differences between
individual animal weekly, fortnightly and four-weekly MH-MPR concentrations. As the time period analysed increased the
differences between individual animal MH-MPR measurements decreased (i.e. the variation in average weekly measure-
ments for individual animals was greater than the variation in four-weekly measurements). This improved accuracy with
a longer sampling time is a result of the increased number of measurement events which increased the accuracy of the
mean measurement over time. However, CH4 output on subsequent weeks has previously been shown to vary, with the
correlations in CH4 output measured weeks or months apart using RC often shown to be lower than expected (Hegarty et al.,
2013).
In the present study, there was a good correlation between individual animals average weekly MH-MPR concentrations
across each sampling week, suggesting that the ranking of individual animals was similar across sampling weeks, with the
accuracy of the individual animal ranking improving as longer sampling periods were used. In order to rank individual
animals in terms of CH4 emissions, a 4 week sampling period would suffice. However, to predict individual animal CH4
emissions an 8-week sampling period gives improved accuracy. Furthermore, persistency of CH4 mitigation effects can be
investigated by comparing MH measurements over longer sampling periods.
In the current study, the magnitude of the reductions in CH4 emissions as a result of different dietary treatments were
always greater for the MH system when compared with the reductions measured in the RC. This may have been caused by
the MH sampling system which predominately measures CH4 during a feeding event, where displacement of CH4 from the
rumen is occurring. Conversely, when CH4 is measured in an RC all emitted CH4 (respiration, eructation and flatulation) is
measured. The rate of CH4 output at any given moment has been shown to depend on (1) methanogenesis closely associated
with feeding level and time, (2) feed type and (3) short-term variations in rumen gas release (Hegarty et al., 2013). It has been
estimated that on average 87% of emitted CH4 is released from the rumen, with approximately 11% released by respiration
and 2% by flatulation (Murray et al., 1976; Ricci et al., 2014). The MH system cannot measure CH4 released from flatulation.
Furthermore, during a feeding visit the proportion of CH4 released from the rumen will be higher than the daily average,
as a result of the displacement of rumen gas by feed. The MH measurements were concentrated during daylight hours,
particularly in the 3 h post-fresh feed being offered. Previous studies have shown that CH4 emissions are related to feed
intake times, tending to peak just after feeding and then decrease with time after this peak (Ricci et al., 2014; Rooke et al.,
2014). This may explain some of the variation in the dietary induced reductions in CH4 emissions measured by both systems.
Aligned with this, the volume of food consumed, and thus the volume of rumen gas displaced, is lower during a Concentrate
diet feeding visit compared with a Mixed diet feeding visit, due to the higher dry matter and lower density of the Concentrate
diet. As the MH measurement events correspond with feeding events, the greater displacement of rumen gas associated with
the Mixed diet may explain the greater percentage differences between in emissions from the Concentrate diet compared
with the Mixed diet, when measured by the MH system compared with the RC.

4.3. Comparison of individual animal measurements from the methane hoods with those from the respiration chambers

Across both experiments detailed in this study there was a significant reduction in average animal DMI during the RC
phase compared to when the animals were group-housed during the MH phase. Previous unpublished work by the same
authors also found this, while in a study with sheep, Bickell et al. (2014) found a reduction in DMI of between 15 and 25%
when the sheep were moved to their pen in the animal house to a RC. Both the current study and the study reported by
Bickell et al. (2014) acclimatised the animals to the RC prior to recording. The decrease in feed intake may be as a result of
the animal being uncomforTable in the chamber surroundings, the lack of competition as the animal is housed singly, or due
to lower energy requirement due to being confined in a RC.
Despite this reduction in DMI, and the effect this has on CH4 production, there was a significant correlation between MH-
MPR and MH-Yld values and the measurements of daily CH4 output and yield from the RC. This correlation was improved
when the MH-MPR concentrations were combined with DMI and diet to produce models for predicted CH4 output. The
concordance correlation coefficient showed good agreement between the models of predicted CH4 output and CH4 output
measured in the RC. Correlations for the CH4 prediction models are comparable to other prediction models from Garnsworthy
et al. (2012) on dairy cows (R2 = 0.79) and Ricci et al. (2014) on cattle (R2 = 0.39). Including MH-MPR measurements alongside
MH-DMI values in models M2, M4 and M6, increased the accuracy and precision of these models compared with similar
models using MH-DMI alone (models M1, M3 and M5, respectively).
 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240 239

Furthermore, the ranking of individual animals based on their CH4 output predicted from model M6 and measured RC-
CH4 emissions were similar, again the concordance correlation coefficient showed good agreement between predicted CH4
ranking and CH4 ranking based on RC measurements. The MH system may therefore, be used to rank individual animals as
high or low emitters, with the consistency of ranking supporting the use of this system as a tool for the genetic selection of
cattle for reduced CH4 emissions. Although, the RC will always be the most precise method of quantifying CH4 emissions,
the MH system offers a viable alternative. The MH system also has some advantages over the RC as it is cheaper and less
labour intensive. Furthermore, the animals can be measured while housed in a group-pen environment thus allowing the
animals to express normal behaviour. Emissions from a larger number of animals can be measured simultaneously and in a
shorter period of time, compared to the RC where often measurements are restricted to 1 animal per chamber per week.

5. Conclusions

The MH system can be used to estimate CH4 output from individual beef cattle in a group housed environment. The
predicted CH4 output has been shown to be consistent with the CH4 output measured in RC. The consistency of this ranking
of individual animal emissions and the high throughput possible using the MH system supports the use of this system as a
tool for the genetic selection of cattle based on CH4 emissions. Furthermore, the timescale required to gather the MH data
is similar to the timescale required to determine residual feed intake (RFI), and both set of measurements can be taken
simultaneously. The average ranking based on MH measured CH4 concentrations for individual animals were similar for
4 and 8 week sampling times. Therefore, in future a measurement period of 4 weeks may be sufficient to rank animals
based on CH4 emissions. However, an 8-week sampling period is advised to prediction individual animal CH4 yield from MH
measurements.
Furthermore, the MH system can be used to determine the effects of dietary mitigation strategies on CH4 emissions
from beef cattle. There was strong agreement between the treatment effects measured by the MH system and the treatment
effects measured subsequently on the same animals in RC. However, the MH system has an advantage over RC as this method
of measurement is non-intrusive and can take place while animals are housed in group pens. This allows other important
measurements required to test dietary mitigation strategies, such as animal performance and efficiency to be measured at
the same time as CH4 emissions.

Acknowledgements

The authors are grateful to the technical staff at the Beef Research Centre for their skilled assistance during these
experiments. This research was funded by EBLEX (A Division of the Agriculture and Horticulture Development Board),
the Scottish Government and by DEFRA and the devolved administrations through the UK Agricultural Greenhouse Gas
Inventory Research Platform.

References

Aguerre, M.J., Wattiaux, M.A., Powell, J.M., Broderick, G.A., Arndt, C., 2011. Effect of forage-to-concentrate ratio in dairy cow diets on emission of methane,
carbon dioxide, and ammonia, lactation performance, and manure excretion. J. Dairy Sci. 94, 3081–3093.
Alexandratos, N., Bruinsma, J. 2012. World agriculture towards 2030/2050: the 2012 revision (No. ESA 12), 03. Rome, Italy.
Bickell, S.L., Revell, D.K., Toovey, A.F., Vercoe, P.E., 2014. Feed intake of sheep when allowed ad libitum access to feed in methane respiration chambers. J.
Anim. Sci. 92, 2259–2264.
Chagunda, M.G.G., Ross, D., Rooke, J., Yan, T., Douglas, J.L., Poret, L., McEwan, N.R., Teeranavattanakul, P., Roberts, D.J., 2013. Measurement of enteric
methane from ruminants using a hand-held laser methane detector. Acta Agric. Scand. Sect. A—Anim. Sci. 63, 68–75.
Cottle, D.J., Nolan, J.V., Wiedemann, S.G., 2011. Ruminant enteric methane mitigation: a review. Anim. Prod. Sci. 51, 491–514.
Duthie, C.-A., Rooke, J., Troy, S., Hyslop, J., Ross, D., Waterhouse, A., Roehe, R., 2015. Animal performance and efficiency of two divergent breeds of finishing
steers offered either a concentrate-straw or a silage based diet ad libitum with either nitrate or increased dietary oil. Adv. Anim. Biosci. 6, 176.
Garnsworthy, P.C., Craigon, J., Hernandez-Medrano, J.H., Saunders, N., 2012. On-farm methane measurements during milking correlate with total
methane production by individual dairy cows. J. Dairy Sci. 95, 3166–3180, http://dx.doi.org/10.3168/jds.2011-4605.
Gerber, P.J., Steinfeld, H., Henderson, B., Mottet, A., Opio, C., Dijkman, J., Falcucci, A., Tempio, G., 2013. Tackling climate change through livestock–A global
assessment of emissions and mitigation opportunities. Rome, Italy.
Grainger, C., Beauchemin, K.A., 2011. Can enteric methane emissions from ruminants be lowered without lowering their production? Anim. Feed Sci.
Technol. 166–167, 308–320.
Hammond, K.J., Humphries, D.J., Crompton, L.A., Green, C., Reynolds, C.K., 2015. Methane emissions from cattle: estimates from short-term measurements
using a GreenFeed system compared with measurements obtained using respiration chambers or sulphur hexafluoride tracer. Anim. Feed Sci.
Technol. 203, 41–52.
Hegarty, R., 2013. Aplicability of short-term emission measurmeents for on-farm quantification of enteric methane. Animal 7, 401–408.
Hulshof, R.B., Berndt, A., Gerrits, W.J.J., Dijkstra, J., van Zijderveld, S.M., Newbold, J.R., Perdok, H.B., 2012. Dietary nitrate supplementation reduces
methane emission in beef cattle fed sugercane-based diets. J. Anim. Sci. 90, 2317–2323, http://dx.doi.org/10.2527/jas2011-4209.
IPCC, 2013. Summary for Policymakers. In: Stocker, T.F., Qin, D., Plattner, G.K., Tignor, M., Allen, S.K., Boschung, J., Nauels, A., Xia, Y., Bex, V., Midgley, P.M.
(Eds.), Climate Change 2013: The Physical Science Basis. Contribution of Working Group I to the Fifth Assessment Report of the Intergovernmental
Panel on Climate Change. Cambridge University Press Cambridge, United Kingdom and New York,USA.
Lassen, J., Løvendahl, P., Madsen, J., 2012. Accuracy of noninvasive breath methane measurements using Fourier transform infrared methods on individual
cows. J. Dairy Sci. 95, 890–898.
Lin, L.-K., 1989. A concordance correlation coefficient to evaluate reproducibility. Biometrics 45, 255–268.
Martin, C., Morgavi, D.P., Doreau, M., 2010. Methane mitigation in ruminants: from microbe to the farm scale. Animal 4, 351–365.
McAllister, T.A., Beauchemin, K.A., McGinn, S.M., Hao, X., Robinson, P.H., 2011. Greenhouse gases in animal agriculture—finding a balance between food
production and emissions. Anim. Feed Sci. Technol. 166–167, 1–6.
240 S.M. Troy et al. / Animal Feed Science and Technology 211 (2016) 227–240

McGeough, E.J., O’Kiely, P., Foley, P.A., Hart, K.J., Boland, T.M., Kenny, D.A., 2010. Methane emissions, feed intake, and performance of finishing beef cattle
offered maize silages harvested at 4 different stages of maturity. J. Anim. Sci. 88, 1479–1491.
Murray, R., Bryant, A., Leng, R., 1976. Rates of production of methane in the rumen and large intestine of sheep. Br. J. Nutr. 36, 1–14.
Patra, A.K., Lalhtiatpuii, M., 2016. Development of statistical models for prediction of enteric methane emission from goats using nutrient composition
and intake variables. Agric. Ecosyst. Environ. 215, 89–99.
Ricci, P., Chagunda, M.G.G., Rooke, J.A., Houdijk, J.G.M., Duthie, C.-A., Huslop, J., Roehe, R., Waterhouse, A., 2014. Evaluation of the laser methane detector
to estimate methane emissions from ewes and steers. J. Anim. Sci. 92, 5239–5250.
Rooke, J.A., Wallace, R.J., Duthie, C.-A., McKain, N., de Souza, S.M., Hyslop, J.J., Ross, D.W., Waterhouse, T., Roehe, R., 2014. Hydrogen and methane
emissions from beef cattle and their rumen microbial community vary with diet, time after feeding and genotype. Br. J. Nutr., 398–407.
Tomkins, N.W., McGinn, S.M., Turner, D.A., Charmley, E., 2011. Comparison of open-circuit respiration chambers with a micrometeorological method for
determining methane emissions from beef cattle grazing a tropical pasture. Anim. Feed Sci. Technol. 166–167, 240–247.
Troy, S.M., Duthie, C.-A., Hyslop, J.J., Roehe, R., Ross, D.W., Wallace, R.J., Waterhouse, A., Rooke, J.A., 2015. Effectiveness of nitrate addition and increased oil
content as methane mitigation strategies for beef cattle fed two contrasting basal diets. J. Anim. Sci. 93, 1815–1823.
United Nations, 2013, United Nations, 2013 United Nations, World Population Prospects. The 2012 Revision Volume I: Comprehensive Table s
ST/ESA/SER.A/336., United Nations, Department of Economic and Social Affairs, popilation Division (2013). World Population Prospects. The 2012
Revision Volume I: Comprehensive Table s ST/ESA/SER.A/336.
Van Zijderveld, S.M., Gerrits, W.J.J., Dijkstra, J., Newbold, J.R., Hulshof, R.B.A., Perdok, H.B., 2011. Persistency of methane mitigation by dietary nitrate
supplementation in dairy cows. J. Dairy Sci. 94, 4028–4038.
Velazco, J.I., Mayer, D.G., Zimmerman, S., Hegarty, R., 2015. Use of short-term breath measures to estimate daily methane production by cattle. Animal 12,
1–9.