Methane production in dairy cows. Individual cow variability in methane production

Abstract

Sources of variation contributing to individual animal differences in methane emissions.

Full text

Methane Production in Dairy Cows
Individual Cow Variability in Methane Production
Edward Hernando Cabezas Garcia
Faculty of Veterinary Medicine and Animal Science
Department of Agricultural Research for Northern Sweden
Umeå
Doctoral Thesis
Swedish University of Agricultural Sciences
Umeå 2017

Acta Universitatis Agriculturae Sueciae
2017:43
ISSN 1652-6880
ISBN (print version) 978-91-576-8859-0
ISBN (electronic version) 978-91-576-8860-6
© 2017 Edward Hernando Cabezas Garcia, Umeå
Print: SLU Service/Repro, Uppsala 2017
Cover: Swedish Red Cows at Röbäcksdalen Research Centre in Umeå
(Photo: Salvatore Maoddi)

Individual Cow Variability in Methane Production
Abstract
Enteric methane (CH4) emissions vary between individual cows, and this variation is
attributed to both animal and dietary factors. In addition, measurement technique of in
vivo CH4 emissions from individual animals still represents a major challenge for
successful emissions mitigation strategies. This thesis investigated the contribution of
different factors to between-animal variation in CH4 production, in order to improve the
current knowledge of its biological basis. In a study comparing on-farm systems for
measuring CH4 emissions from large numbers of animals and the variation between
individual animals, the GreenFeed system was used as the normal set-up (flux method)
or modified to mimic gas analysers systems based on CH4 concentrations (sniffer
system) to measure CH4 emissions. Measurements taken by the GreenFeed system
proved be more reliable than those made by the simulated sniffer method. The
GreenFeed data were consistent with literature values determined in respiration
chambers, while the sniffer method were poorly correlated to flux method values and
were not significantly related to either feed intake or milk yield. Despite GreenFeed
being a spot sampling method, it proved to be a promising tool for ranking cows as
high and low CH4 emitters. A meta-analysis based on an individual cow dataset
investigating the effects of between-cow variation and related animal variables on
predicted CH4 emissions from dairy cows. Between-cow variation in fermentation
pattern are not likely be the major factor influencing predicted in vivo CH4 emissions.
Variation and repeatability for volatile fatty acid concentrations were greater for
ruminal concentrations than molar proportions, indicating strong control by the
individual cow. Digestion kinetics variables were more repeatable than rumen
fermentation or microbial synthesis, as a result of variations in passage rate. In studies
in which late-cut silage and rolled barley were gradually replaced with early-cut silage
in the diet of dairy cows, production responses and in vivo CH4 emissions were studied
in 16 intact lactating cows and possible physiological mechanisms were assessed in
four rumen-cannulated cows. Improvements in forage quality by graded addition of
early-cut silage was an effective strategy to reduce concentrate supplementation,
without compromising performance or increasing CH4 emissions in lactating dairy
cows. Differences in intake between treatments were partly compensated by differences
in silage digestibility.
Keywords: Dairy cattle, methane yield, between-cow variation, repeatability.
Author’s address: Edward Hernando Cabezas Garcia, SLU, Department of Agricultural
Research for Northern Sweden, P.O. Box, 901 83 Umeå, Sweden
E-mail: edward.cabezas.garcia@slu.se; ecabezasg@hotmail.com

Dedication
To my beloved family for your unconditional support during the moments of
uncertainties.
To all of science teachers whom I had the chance to share with since I was a
child for their encouragement, example and guidance through my scientific
journey.
The ruminant animal: “A small fermentation unit which gathers the raw
material, transfers it to the fermentation chamber, and regulates its further
passage, continuously absorbs the fermentation products, and transforms them
into a few valuable substances such as meat and milk. To these advantages
must be added the crowning adaptation: the unit replicates itself”
Robert E. Hungate, 1950

Contents
List of Publications 7
Abbreviations 9
1 Introduction 11
1.1 The role of ruminants in methane and greenhouse gas emissions 11
1.2 Methane production in ruminants 15
1.2.1 Rumen fermentation 15
1.2.2 Hydrogen production and H+ sinks in the rumen 18
1.3 Strategies to reduce methane emissions 20
1.3.1 Diet manipulation (feeding level and digestibility) 20
1.3.2 Increased level of concentrate 20
1.3.3 Fat supplementation 21
1.3.4 Additives 21
1.3.5 Manipulation of microbes 24
1.4 Techniques to measure methane emissions in vivo 25
1.4.1 Respiration chambers 25
1.4.2 Spot sampling methods 26
1.4.3 Tracer gas methods 28
1.4.4 The laser technique 29
1.4.5 Proxies to measure in vivo methane 30
2 Objectives 31
3 Materials and methods 33
3.1 Paper I 33
3.2 Paper II 35
3.3 Paper III 36
3.4 Paper IV 37
4 Results 39
4.1 Paper I 39
4.2 Paper II 40
4.3 Paper III 41
4.4 Paper IV 42

5 Discussion 45
5.1 Measurement technique 45
5.1.1 Between-cow variability related to methods 45
5.1.2 CH4/CO2 ratio 52
5.2 Animal-related factors 56
5.2.1 Effects of rumen fermentation pattern 57
5.2.2 Variability and repeatability of volatile fatty acids in the rumen 60
5.2.3 Passage rate and associated factors 63
5.3 Dietary factors 68
6 Conclusions 71
7 Future perspectives 73
8 Popular scientific abstract 75
9 Populärvetenskaplig sammanfattning 77
References 79
Acknowledgements 91

7
List of Publications
This thesis is based on the work contained in the following papers, referred to
by Roman numerals in the text:
I Huhtanen, P., E. H. Cabezas-Garcia, S. Utsumi, and S. Zimmerman (2015).
Comparison of methods to determine methane emissions from dairy cows
in farm conditions. Journal of Dairy Science 98, 3394–3409.
II Cabezas-Garcia, E. H., S. J. Krizsan, K. J. Shingfield, and P. Huhtanen.
(2017). Between-cow variation in digestion and rumen fermentation
variables associated with methane production. Journal of Dairy Science
100, 1–16 (In Press).
III Cabezas-Garcia, E. H., S. J. Krizsan, K. J. Shingfield, and P. Huhtanen.
(2017). Effects of replacement of late-harvested grass silage and barley with
early-harvested silage on milk production and methane emissions. Journal
of Dairy Science (In Press).
IV Cabezas-Garcia, E. H., S. J. Krizsan, K. J. Shingfield, X. Dai, and P.
Huhtanen. (2017). Effects of replacement of late-harvested grass silage and
barley with early-harvested silage on ruminal digestion efficiency in dairy
cows (Manuscript).
Papers I-II are reproduced with the permission of the publishers.

8
The contribution of Edward Hernando Cabezas Garcia to the papers included
in this thesis was as follows:
I Planned the research with the first author and performed the field
experiment at Röbäcksdalen research barn including data collection and
statistical analysis. Participation in writing the manuscript jointly with the
rest of co-authors.
II Planned the research jointly with the main supervisor, collected the dataset
in collaboration with the co-authors, analysed the data and wrote the
manuscript.
III Planned the experiment in collaboration with the co-authors, performed the
experiment, and analysed the data together with co-authors and wrote the
manuscript.
IV Planned the experiment in collaboration with the co-authors, performed the
experiment, processed experimental samples, and analysed the data
together with co-authors and wrote the manuscript.

9
Abbreviations
CH
4
VFA Stoichiometric methane
CH4/DMI Methane yield
CH4/ECM Methane intensity
CV Coefficient of variation
DMI Dry matter intake
ECM Energy corrected milk
GE Gross energy
GEI Gross energy intake
GHG Greenhouse gases
iNDF Indigestible neutral detergent fibre
kd Fractional digestion rate
kp Fractional passage rate
NDF Neutral detergent fibre
OM Organic matter
Rep Repeatability
VFA Volatile fatty acid

10

11
1 Introduction
1.1 The role of ruminants in methane and greenhouse gas
emissions
Population growth is challenging for agricultural systems around the world
because it means producing more food to support an increasing population in a
more efficient manner within the constraints of available natural resources,
without compromising the future of coming generations. The future demand
for livestock products in a scenario where most of the population lives in large
cities, rather than in rural areas, will dictate consumption trends in coming
years (Kearney, 2010; FAO, 2016). The demand for livestock products will
more than double by 2050 compared with 2000 (Steinfeld et al., 2006; FAO,
2011). In developing countries, the demand will increase as a consequence of
increasing population and net income (FAO, 2011, 2016), which are usually
associated with increases in demand for animal products. On the other hand, in
developed countries the demand will increase slower than in developing
countries and socio-cultural values will be more relevant in people’s food
choices (FAO, 2011).
Livestock production has been criticised in recent decades for reasons such
as: use of agriculture land that could be used for human food production, water
consumption, deforestation, environmental pollution, animal welfare, human
health concerns from eating animal products etc. Since feedstuff production is
what links livestock production to land use, both directly via grazing and
indirectly via traded grain or forage, environmental sustainability is an issue of
major importance in the feed industry (Herrero et al., 2013). Livestock supply
13% of the energy in human diets, consume around 50% the world’s
production of grains (Smith et al., 2013) and, at the same time, are responsible
for about 14.5% of total anthropogenic greenhouse gas emissions (7.1 Gt CO2-
equivalents per year) (Gerber et al., 2013).

12
Ruminants have the unique ability of transforming roughages which are not
used by monogastric animals into human food (e.g. milk, meat) and could
therefore reduce the competition for arable land between animal feed
production and food production for humans. Comparing total and human-
edible efficiency for different livestock production systems e.g. in the USA
(Table 1), it can be seen that in terms of conversion of feed resources into
human-edible products, ruminants are more efficient than monogastric animals
despite inefficiencies in total terms, as in the case of beef production. Milk
production is advantageous compared with beef meat production when an
efficiency perspective is considered. Monogastric animals in intensive
production systems are fed high levels of grain with high quality protein
supplements in the diet, whereas forage is the main component of rations for
dairy cattle. In addition, concentrate supplementation can be reduced
significantly in grazing-based systems and even more when agricultural by-
products are included in the diet. Intensive feedlot production systems for beef
cattle usually use very high levels of concentrate in the diet (>90%) and,
despite the advantages in terms of faster production returns, this type of
production system represents direct competition for food with humans. In
extensive production systems (i.e. tropical conditions), despite low animal
productivity per hectare human-edible efficiency should be considered, since
ruminants also represent economic status and wellbeing in developing
countries.
Table 1. Comparative efficiency of different livestock production systems in the USA (adapted
from Gill et al., 2010)
Energy efficiency Protein efficiency
Product
Total1 Human-edible2 Total1 Human-edible2
Milk
0.25 1.07 0.21 2.08
Beef
0.07
0.65
0.08
1.19
Pigs
0.21 0.30 0.19 0.29
Poultry meat
0.19
0.28
0.31
0.62
1Total efficiency calculated as outputs of human-edible energy and protein divided by total energy
and protein inputs.
2Human-edible efficiency calculated as outputs of human-edible energy and protein divided by
human-edible inputs.
Methane (CH4) has 28-fold higher greater global warming potential than
carbon dioxide (CO2) (IPCC, 2007). Methane emissions to the atmosphere
derive from both natural sources, e.g. natural wetlands, termites, ocean and
hydrates, and anthropogenic sources, e.g. rice fields, ruminants, landfills,
biomass burning and fossil fuels (Moss et al., 2000; Aronson et al., 2013). The

13
concentration of CH4 in the atmosphere has increased from 750 ppb during
pre-industrial times to about 1800 ppb today as a consequence of human
activities (IPCC, 2013). In such a scenario, the contribution of the livestock
sector to the total anthropogenic CH4 emissions is important. Within the
livestock sector, it is clear that cattle (beef and dairy) have the highest total
GHG emissions (CO2 equivalents) compared with monogastric animals (Figure
1). Most of the GHG contribution of ruminants per unit of edible product
(>40%) comes from enteric CH4 fermentation (Gerber et al., 2013).
Figure 1. A) Global estimated emissions by species. Emissions are attributed to edible products
and non-edible products. B) Emissions from cattle milk and beef supply chains. Source: GLEAM.
Modified from Gerber et al. (2013).

14
In a global perspective, CH4 emissions from ruminants are a function of
ruminant population size, productivity, diet composition and associated manure
management systems (Knapp et al., 2014). Methane production also represents
an energy loss to the animal, which can vary from 2 to 12 % of gross energy
intake depending on intake level and diet composition (Johnson and Johnson,
1995). On the other hand, it also could be assumed that by increasing feed
efficiency, the amount of feed and waste material (e.g. manure, CH4 emissions)
produced per unit of product can be reduced. However, in the near future the
major challenge will not only be to increase the feed efficiency of animals, but
also to mitigate the impact of the livestock industry on the environment
(Godfray et al., 2010), especially in developing countries (e.g. Latin America),
where it is predicted that cattle production will continue to expand because
agriculture is still a major source of income (McAllister et al., 2011).
Improvements in animal productivity per hectare by strategic animal feeding
practices using local feeds would contribute significantly to lowering CH4
emissions in those countries. It is worth mentioning that in tropical regions,
ruminant animals raised in extensive conditions are not only used as food, but
also for cultural purposes, draft power and financial security.
Modern dairy cows are not the same animals as the old phenotypes. As a
consequence of genetic selection and improvements in feeding and
management practices, modern dairy cows have increased their production
performance and consequently their dry matter intake (DMI). The productivity
in the US dairy herd has increased considerably (milk yield per cow was 2074
kg/year in 1944, compared with 9193 kg in 2007) and these improvements
have been accompanied by a substantial reduction in the number of cows
(Capper et al., 2009). At individual animal level, this means that the total
energy requirement per kg of milk produced is reduced by decreasing the
energy requirement for maintenance, and hence the cows are more efficient in
feed conversion. Capper et al. (2009) also calculated that a cow (650 kg;
3.69% milk fat) yielding 29 kg/day needs only 4.6 MJ net energy (NE) per kg
of milk, compared with 9.2 MJ NE/kg milk when the production level is only 7
kg/day. Despite this, the carbon footprint per cow increased two-fold between
1944 and 2007, although it decreased when expressed per kg milk (from 3.66
to 1.35 CO2-eq/kg milk). This change was also reflected in substantial
reductions in nitrogen, phosphorus and CH4 emissions from manure per unit of
edible product. The Swedish Board of Agriculture has estimated that enteric
CH4 from ruminants in 2011 constituted one-third of the total emissions (2.6
million tons CO2-equivalents) from Swedish agriculture (Naturvårsverket,
2008).

15
Methane production from enteric fermentation decreased by about 12%
over the period 1990-2011 (Naturvårsverket, 2013), mainly due to a reduction
in the dairy cattle population (from 525,000 to 338,000 head between 1993 and
2014), whereas the total amount of milk delivered to dairies decreased only
marginally (SCB, 2016). At individual animal level, milk production increased
from 9100 kg milk/year/head in 2014, compared with 7060 kg milk/year/head
in 1990.
Global warming potential from ruminants cannot be neglected, but statistics
sometimes exaggerate its contribution. It is also important to standardise the
criteria for measuring GHG emissions, since there is great variation in current
estimates. For instance, in some cases such exaggerated estimates detract from
the major cause of climate change, which is mainly associated with combustion
of fossil fuels (Herrero et al. 2011). Ruminants have the advantage of
transforming carbon from photosynthesis into human-edible food and, from an
ecological point of view, CH4 emissions from ruminants can be considered
recyclable carbon.
1.2 Methane production in ruminants
1.2.1 Rumen fermentation
Enteric CH4 is produced from anaerobic fermentation of feeds, which takes
place mainly in the rumen with a minor contribution from the hindgut. No
single microbial species is responsible for complete degradation of substrate in
the rumen. Instead, a complex succession of organisms takes part in the
cooperative catabolism of substrates and the production of fermentation end-
products. The diversity, size and activity of the microbial population in the
rumen are largely determined by the diet composition (Van Soest, 1994), but
are also influenced by animal-related factors such as saliva production, rumen
volume and rates of intake and passage (Pinares-Patiño et al., 2003; Hegarty,
2004) and management factors such as inclusion of essential oils in the diet
(Patra and Yu, 2012).
Three separate factors that affect CH4 emissions per unit intake can be
identified: the rate of degradation of organic matter, the efficiency of microbial
growth and the type of volatile fatty acids (VFA) produced from the
fermentation of organic matter (Czerkawski, 1986; Van Soest, 1994). The
fermentation of carbohydrates is by far the most important source of energy for
rumen microbes (Ørskov, 1990) and bacteria are the principal organism
fermenting carbohydrates in the rumen (Hungate, 1966).
The main pathways of carbohydrate metabolism in the rumen and relevant
rumen microbes involved are shown in Figure 2. Feed entering the rumen is

16
primarily digested by bacteria, fungi and protozoa (primary fermenters), which
digest feed components to simple monomers (McAllister et al., 1996). The
breakdown of carbohydrates performed by the rumen microbiota can be
divided into two stages (McDonald et al., 2011; Morgavi et al., 2010). The first
stage involves the hydrolysis of complex carbohydrates to glucose equivalents
and is performed by primary fermenters such as Fibrobacter succinogenes for
the cell wall carbohydrates.
Figure 2. Overview of carbohydrate metabolism in the rumen and examples of microbiome
species involved with substrate fermentation and methane (CH4) production. Adapted from Van
Soest (1994) and McAllister et al. (1996).
The hydrolysis of carbohydrates in the rumen is briefly described by Mc
Donald et al. (2011) as follows: Cellulose is decomposed by β-1,4-
glucosidades to cellobiose, which is then converted either to glucose or,
through the action of a phosphorylase, to glucose-1-phosphate. Starch and

17
dextrin are first converted by amylases to maltose and isomaltose and further
converted to glucose or glucose-1-phosphate by maltose phosphorylases and
1,6-glucosidases. Fructans are hydrolysed by enzymes attacking 2,1- and 2,6-
linkages to release fructose units, which may also be produced together with
glucose by the digestion of sucrose. Pentoses are the major product of
hemicellulose hydrolysis, which is brought about by enzymes attacking the β-
1,4 linkages xylose and uronic acids. The most common pathway of hexose
metabolism in the rumen is glycolysis, which produces two equivalents of
pyruvate, ATP and NADH.
The second stage (microbial fermentation) of carbohydrate digestion
involves the conversion of pyruvate, a 3-carbon simple molecule, to different
fermentation end-products through metabolic pathways that produce metabolic
hydrogen and reducing equivalents (Moss et al., 2000; McDonald et al., 2011).
Because the rumen is an anaerobic habitat, substrates are only partially
oxidised and reducing equivalent disposal (e.g. NADH) is a critical feature for
fermentation (Russell, 2002). Primary and secondary fermenters are involved
in the degradation of simple sugars to the main products of rumen
fermentation, such as main VFAs (acetate, propionate and butyrate), hydrogen
gas (H2) and CO2, whereas CH4 is produced in the final stage by methanogens
(e.g. Methanobrevibacter ruminantium), using H2 (80%) or formate (HCOO-;
18 %) together with CO2 as the main substrates (McAllister et al., 1996). When
considering VFA, CO2 and CH4 as sole fermentation end-products using
stoichiometry principles (Wolin, 1960), the fermentation equation for hexoses
is to produce 100 units of VFA in the ratio 60:20:15:
57.5 Hexose → 115 Pyruvate → 65 Acetate + 20 Propionate + 15 Butyrate +
35 CH4 + 60 CO2
The fermentation pattern of the main VFAs in the rumen varies depending
on diet composition and interval since feeding. Commonly, the molar ratio of
acetate to propionate to butyrate is found to vary between 75:15:10 and
40:40:20 (Bergman, 1990). Acetate is produced from pyruvate following the
loss of one carbon as CO2, whereas butyrate is formed by the condensation of
two molecules of acetyl-CoA. The reactions involved in the formation of
acetate and butyrate from pyruvate are interrelated and all proceed through
acetyl-CoA (Bergman, 1990). As a result of acetate formation, re-oxidation of
NADH occurs and H+ is produced, and methanogens use it to reduce CO2 to
CH4 that is subsequently utilised for their maintenance and growth (McAllister
and Newbold, 2008).

18
The metabolism of oxaloacetate to succinate is the main route used by
rumen organisms to produce propionate, but the pathway through lactate and
acrylate is favoured in the rumen of animals fed a high-concentrate diet (Van
Soest, 1994). Propionate is the only VFA that makes a significant contribution
to glucose synthesis, and is quantitatively the most important single precursor
of glucose. Additional VFAs are also formed in smaller quantities by the
deamination of amino acids such as isobutyrate from valine, isovalerate from
leucine and 2-methyl butyrate from isoleucine. Their production is important,
since they are growth factors for many cellulolytic organisms. The majority of
the VFAs produced are rapidly absorbed through the rumen wall into the
bloodstream and serve as major energy and carbon sources for the animal. In
ruminants such as sheep and cattle, the contribution of VFAs to the energy
requirement can be as high as 70% (Bergman, 1990), whereas CH4 production
represents an energy loss to the animal.
1.2.2 Hydrogen production and H+ sinks in the rumen
As a consequence of the lack of oxygen and the excess of reduced cofactors in
the rumen, it is necessary to have sinks for disposing of H2 produced during
microbial fermentation. Hydrogen is produced during enzymatic oxidation of
the NADH formed during glycolysis to NAD+ (Czerkawski, 1986).
Accumulated metabolic H2 has to be removed, since otherwise it inhibits the
re-oxidation of NADH, microbial growth and fibre degradation (Wolin et al.,
1997; Joblin, 1999; McAllister and Newbold, 2008). Syntrophic (cross-
feeding) interspecies H2 transfer occurs when some microbes produce H2 that
is then further used by other microbes (Krause et al., 2013). The amount of
specific VFAs produced is the major determinant of the amount of H2
produced in the rumen (Table 2).
Compounds with negative oxidation values act as H2 sinks. According to
stoichiometric principles, CH4 has the lowest (-2) possible oxidation state per
unit of carbon compared with VFAs and CO2 (highest +2). Therefore, the
conversion of H+ and CO2 to CH4 and H2O is the most important H2 sink in the
rumen (8H) compared with other pathways such as CH4 conversion from
formic acid (6H) or propionate production (2H) (Wolin, 1960; Van Soest,
1994). In addition to methanogenesis performed by archaea, other H2 sinks can
also be promoted under special conditions such as the addition of nitrates to the
diet. An overview of different H2 sinks in rumen conditions is presented in
Figure 3.

19
Table 2. Theoretical stoichiometric carbon-hydrogen balance equations describing conversion of
glucose in the rumen
glucose → 2 acetate + 2 CO
2
+ 8 H
glucose → butyrate + 2 CO2 + 4 H
glucose + 4 H → 2 propionate
CO2 + 8 H → CH4 + 2 H2O
Net 3 glucose → 2 acetate + butyrate + 2 propionate + 3 CO
2
+ CH
4
+ 2 H
2
O
Acetate production increases threefold and propionate and butyrate
are unchanged:
3 glucose → 6 acetate + 6 CO
2
+ 24 H
glucose → butyrate + 2 CO2 + 4 H
glucose + 4 H → 2 propionate
3 CO2 + 24 H → 3 CH4 + 6 H2O
Net 5 glucose + 6 acetate + butyrate + 2 propionate + 5 CO
2
+ 3 CH
4
+ 6 H
2
O
Note: In Case 1, the acetate-to-propionate ratio is 1:1 and the methane-to-glucose ratio is 1:3; in
Case 2, acetate-to-propionate ratio is 3:1, and methane-to-glucose is 3:5 (Source: Van Soest,
1994).
Figure 3. Schematic microbial fermentation of feed polysaccharides and H2 reduction pathways in
the rumen (Morgavi et al., 2010).

20
1.3 Strategies to reduce methane emissions
Different strategies have been proposed to mitigate CH4 emissions from
ruminants and thus reduce their impact on climate change. The effectiveness of
a particular strategy depends upon the level at which it is evaluated (e.g.
individual animal, farm, country) and its impact not only in the short term, but
also in the long term. Effective mitigation strategies have to consider two
major issues: i) improving rumen fermentation efficiency and ii) increasing
animal productivity. In practice, profitability is often the most important
decision-making factor in cattle production systems and it determines the
adoption of a particular mitigation strategy at farm level, since farmers are
unlikely to adopt practices that have no economic benefit or are not mandatory
or supported by governmental subsidies (Hristov et al., 2013).
Many additives which have showed promising results in reducing CH4
production in vitro have failed to produce similar results in vivo (McAllister,
2011). When in vivo reductions in CH4 production have been observed, they
have often been accompanied by a decrease in feed intake, digestibility and/or
productivity. Therefore in vitro results have to be interpreted with caution if
they are not tested in in vivo conditions.
1.3.1 Diet manipulation (feeding level and digestibility)
Intake and diet composition are the main factors affecting enteric CH4
production in ruminants. Higher feed intake is associated with shorter retention
time of feed particles in the rumen and thus rumen microorganisms have less
time to digest the available substrate. In addition, the efficiency of microbial
cell synthesis increases with increased passage rate, which partitions less
fermented carbon to gases and volatile fatty acids. Conversely, higher
digestibility usually increases CH4 emissions per unit intake (Blaxter and
Clapperton, 1965), but emissions per unit digested intake decrease and
emissions per unit of product most likely decrease with improved diet
digestibility. Digested neutral detergent fibre (NDF) produces more CH4 than
digested neutral detergent solubles, due to changes in fermentation pattern in
the rumen (Jentsch et al., 2007).
1.3.2 Increased level of concentrate
The reported effects of the level of concentrate feeding on CH4 emissions are
not consistent. In feedlot-type diets (>90% concentrate on DM basis), it is clear
that increased addition of starch in the diet promotes propionate fermentation
in the rumen (Johnson and Johnson, 1995). Sauvant and Giger-Reverdin (2009)
reported a quadratic effect of the proportion of concentrate on CH4 emissions,
with a maximum at 35% concentrates on a dry matter (DM) basis. Starch

21
supplementation tends to increase butyrate, rather than propionate, in dairy
cattle diets based on grass silage containing up to 70-75% concentrates on a
DM basis (Jaakkola and Huhtanen, 1993). In addition, high levels of starch in
the diet can compromise animal performance by decreasing fibre digestibility
and increasing the incidence of acidosis.
1.3.3 Fat supplementation
Ruminant diets are low in dietary lipids due to the low contents in forages (Van
Soest, 1994). Fat clearly decreases CH4 emissions, as shown in in vivo studies
(Jentsch et al., 2007; Beauchemin et al., 2008), and also in meta-analysis
approaches (Ramin and Huhtanen, 2013). In addition, and especially at high
level of supplementation, fat reduces feed intake and diet digestibility (Jenkins,
1993). At least four mechanisms are involved in the inhibitory effect of fat on
CH4 production: i) Fat is not a fermentable substrate in the rumen, ii) the bio-
hydrogenation of fatty acids acts as an alternative H2 sink in the rumen, iii) fat
promotes increases in propionate concentrations in the rumen and iv) fat may
decrease protozoal numbers (Van Soest, 1994).
However, fat supplementation in dairy cows above economic optimum
increases feed costs and can reduce milk protein content (NRC, 2001). Fat
supplementation has also been suggested to decrease fibre digestibility
(Jenkins, 1993) but, according to a recent meta-analysis (Weld and Armentano,
2017), these effects are observed only with medium-chain and unsaturated fatty
acids. In dairy cows diets, maximum recommended inclusion rate in ruminant
diets is 6 to 7% (total fat) of dietary DM (Hristov et al., 2013).
1.3.4 Additives
The main objective of additives as a mitigation strategy is to improve rumen
fermentation efficiency (Hristov et al., 2013). McAllister and Newbold (2008)
defined two general mechanisms by which these substances act to reduce CH4
emissions: i) by reducing the supply of metabolic H+ for methanogens (e.g.
defaunation, acetogenesis) and ii) by direct inhibition of methanogens (e.g.
plant extracts). A summary of the main additives used to mitigate CH4
emissions is provided below.
Inhibitors
Methane inhibition alters the microbial community, H2 production and
fermentation response in the rumen of cattle (Martinez-Fernandez, 2016).
Different CH4 inhibitors have been studied due to their specific inhibitory
effect on rumen archaea. These include: bromochloromethane (BCM), 2-
bromoethane sulfonate, chloroform and cyclodextrin. However, while these

22
compounds have been found to be effective in reducing CH4 emissions (by up
to 50%; Hristov et al., 2013), they have a harmful effect on the animal
(McAllister and Newbold, 2008). Adaptation of the rumen ecosystem
compromises the effectiveness of using BCM as a mitigation strategy, since it
is transitory (McAllister and Newbold, 2008). Recently, 3-nitrooxypropanol
has been suggested as a promising compound in mitigating CH4 emissions
from ruminants, since it is not toxic for the animal and it has minor effects on
dry matter intake. In dairy cows fed a diet containing 3-nitrooxypropanol,
reductions of up to 30% in CH4 emissions have been reported, without negative
effects on feed intake or milk production, in a long-term (12-week) study
(Hristov et al., 2015a).
Ionophores
Ionophores are highly lipophilic ion carriers that modify ion transport through
biological membranes. Monensin is the most studied ionophore and was
originally marketed as a coccidiostat (anti-protozoan) for chickens. Nowadays,
it is routinely used in North America, but banned in the European Union for
use as a feed additive. Monensin acts on the cell wall of the Gram-positive
bacteria that produce H+ and interferes with ion flux, which results in
decreasing acetate to propionate ratio in the rumen and thus a reduction in
enteric CH4 emissions. Hristov et al. (2013) concluded that ionophores are
likely to have a moderate CH4 mitigating effect, but their effect appears to be
inconsistent. Some studies suggest a stronger anti-methanogenic effect in beef
steers than in dairy cows (mostly fed forage-based diets), e.g. Guan et al.
(2008) reported up to 30% reduction in enteric CH4 production in beef cattle
and up to 9% reduction in dairy cattle Van Vugt et al. (2005). The effects in
dairy cows can be improved by dietary modifications and increasing monensin
dose, as reported by Appuhamy et al. (2013).
Electron receptors
Nitrates/nitrites have shown promising results in decreasing CH4 production
(Van Zijderveld et al., 2010). An important issue as regards using nitrates in
the diet is the potential for increased ammonia production and potential toxicity
from intermediate products (e.g. nitrite; Leng, 2008). The best option to reduce
CH4 emissions using nitrates is to replace urea as a non-protein nitrogen source
in the diet to meet the microbial requirements for rumen-degradable N.
However, when the supply of degradable N is sufficient, nitrate
supplementation will increase nitrogen losses to the environment.
Adding sulphate to the diet of sheep has been found to reduce CH4
production and, when both nitrate and sulphate are added, the effects on CH4

23
production have been shown to be additive (Van Zijderveld et al., 2010).
Distiller’s grain contains high levels of sulphate, which has resulted in
intensive research on the effect of high-sulphate diets (also in combination with
high-sulphate drinking water). However, high-sulphate diets induce
polioencephalomalacia (Gould, 2000), which is caused by excessive
production of hydrogen sulphide (H2S) in the rumen.
Organic acids such as fumaric and malic acids have also been studied as
alternative hydrogen sinks in the rumen (Molano et al., 2008). The mitigating
potential of fumarate has been questioned (Ungerfeld et al., 2007), because it is
generally lower than that of nitrates and results have been inconsistent.
Plant compounds
Plant-bioactive compounds form a large and heterogeneous group and vary in
chemical structure. Tannins, saponins and essential oils have been reported as
the main compounds with anti-methanogenic activity in ruminants (Waghorn et
al., 2002; Hristov et al. 2013).
Tannins are polyphenolic substances widely distributed in plants which are
characterised by their ability to bind proteins in aqueous solutions. Tannin-
protein complexes involve both hydrogen-bonding and hydrophobic
interactions, causing a reduction in protein degradation in rumen conditions.
Tannins are anti-nutritional factors when dietary crude protein concentrations
are limiting production, because they reduce absorption of amino acids
(Waghorn, 2008). The anti-methanogenic effect of hydrolysed tannins is
caused by their inhibition of the rumen archaea, whereas condensed tannins act
indirectly by inhibition of fibre digestion (Goel and Makkar, 2012). In a review
by Hristov et al. (2013), tannins were reported to show good potential for
reducing CH4 emissions, by up to 20%.
Saponins are glycoside compounds present in many plants in which the
sugars units are linked to a triterpene or steroidal aglycone moiety. They
modify ruminal fermentation by their toxic effect on ruminal protozoa.
Therefore, saponins have the potential to enhance flow of microbial protein,
which is an alternative H2 sink in the rumen, and in this way increase the
efficiency of feed utilisation and reduce enteric CH4 production. However,
their effect is not always consistent, since it has been reported that saponins
can be inactivated by rumen bacterial populations and the saliva of adapted
animals (Newbold et al., 1997). The potential of essential oils as inhibitors of
CH4 production has been studied extensively in in vitro experiments
(Calsamiglia et al., 2008; Bodas et al., 2008; Benchaar et al., 2011). These
substances have an antimicrobial activity against rumen archaea by reducing

24
H2 availability. However, it is likely that the doses required for any substantial
mitigations in vivo are not economically feasible.
1.3.5 Manipulation of microbes
Defaunation refers to the removal of rumen protozoa. Rumen protozoa share a
symbiotic relationship with methanogens, participating in interspecies
hydrogen transfer. The literature reports contradictory results regarding
defaunation (McAllister and Newbold, 2008). However, some studies indicate
that defaunation may lower the amount of hydrogen in the system and thereby
reduce CH4 production (Vermorel and Jouany, 1989; Morgavi et al., 2010).
According to a review by Morgavi et al. (2010) defaunation decreases CH4
emissions by on average 10.5%. Their review also found that methane
production per mole of VFA, calculated according to Wolin (1960) using data
from Eugène et al. (2004) was 6.9% greater in faunated than in defaunated
animals (116-118 comparison) and digestibility of OM was 15 g/kg (2.2%)
higher in faunated animals. In addition, the efficiency of microbial N synthesis
was higher in faunated animals, which repartition fermented carbon from VFA
and gas production to microbial cells (Morgavi et al., 2010). In conclusion, it
seems that the lower CH4 emissions in defaunated animals can be entirely
attributed to changes in diet digestion and rumen fermentation pattern.
Significant efforts have been devoted to suppressing archaea and/or
promoting acetogenic bacteria in the rumen (Hristov et al., 2013). Vaccines
trigger the immune system of ruminants by a continuous supply of antibodies
against archaea to the rumen through saliva. Since the rumen methanogen
population present can differ based on diet and geographical location of the
host, applying a single-targeted approach it could be expected difficult its
implementation in in vivo conditions.
New approaches are under investigation, one of which involves identifying
genes encoding specific membrane-located protein as antigens to vaccinate
sheep (Buddle et al., 2011). Another involves generation of antisera against
subcellular fractions of this microorganism in in vitro conditions, reducing
microbial growth and CH4 production (Wedlock et al., 2013).
Reductive acetogenic bacteria reduce two moles of CO2 to acetate by
oxidation of H2 (Joblin, 1999). However, they are less efficient than archaea
populations, as H2 sinks in the rumen under normal ruminal conditions (Fievez
et al., 1999). Acetogenesis could take place more actively in the rumen when
methanogenesis is inhibited at increased hydrogen concentrations if dissolved
hydrogen concentrations increased as a result of suppressed CH4 production
(Le Van et al., 1998). In summary, in vitro approaches testing CH4 inhibitors
(see Figure 4) are useful for screening purposes (e.g. doses), but are still rather

25
far from explaining observed CH4 emissions from ruminants in in vivo
conditions. Long term in vivo studies are required to confirm results obtained
in in vitro conditions.
Figure 4. Cartoon showing the in vivo side-effects of dietary additives to inhibit CH4 production.
Reprinted from: An Introduction to Rumen Studies by J.W. Czerkawski, page 106. Copyright ©
(1986).
1.4 Techniques to measure methane emissions in vivo
1.4.1 Respiration chambers
Respiration chambers have been favoured for their accuracy and low
coefficient of variation (Blaxter and Clapperton, 1965; Grainger et al., 2007)
compared with other methods for measuring CH4 in in vivo conditions and
have been used widely to measure differences between diets (Figure 5). Indeed,
respiration chamber experiments have been performed for over 100 years and
provide the basis for our current understanding of energy metabolism in farm
animals (McLean and Tobin, 1987).
The principle of this technique is to collect all exhaled breath from the
animal and to measure gas concentrations (e.g. CH4). Gas concentrations are
corrected by continuous airflow to adjust the background concentrations. The
animals are kept in closed chambers for about 2-4 days with ventilation for
intake and exhaust air. The chamber gives an inflow and outflow of gas
concentrations (CO2, O2, and CH4) and therefore is possible to calculate the
energy balance of the animal. Methane flux (L/day) is calculated as CH4 flow =
Air flow × 106 × [CH4 Outflow (ppm) - CH4 Inflow (ppm)]. This technique has
been criticised for the fact that the animal inside the chamber is not
experiencing natural conditions and that the restriction could have
consequences for animal behaviour, especially feed intake, and could lower
heat production owing to the reduction in physical activity. Among practical

26
considerations, the technique is expensive, labour-intensive and not designed
for measuring a large number of animals simultaneously. Recently, in the large
scale EU project RuminOmics (www.ruminomics.eu), in vivo CH4 individual
data from 100 dairy cows was collected in respiration chambers at LUKE
(Natural Resources Institute Finland).
Figure 5. Respiration chambers. A) Schematic diagram of the open-circuit respiration chamber
showing air fluxes (adapted from Grainger et al. (2007). B) Research facilities at Poznan
University of Life Sciences, Poland (Source: http://globalresearchalliance.org/country/poland/
Accessed: 22 March 2017).
1.4.2 Spot sampling methods
GreenFeed system
The GreenFeed system (C-Lock Inc., Rapid City, South Dakota, USA) is a spot
gas sampling method which can be attached either to a concentrate feeder
station or an automatic milking system in farm conditions (Figure 6). It records
both CH4 and CO2 fluxes on an individual animal basis from the exhaled air
during breathing by the animal when eating small amounts of concentrate feed
released into a tray in a semi-enclosed hood (http://www.c-lockinc.com/).
GreenFeed is highly dependent on high frequency of animal visits per day,
which increases repeatability and certainty in estimating daily CH4 and CO2
emissions from individual animals. The duration of individual visits is
especially important for CH4 measurements, because most CH4 is eructated at
40- to 120-s intervals (Hammond et al. 2016). The system recognises the
individual animal by interfacing with an attached tag reader and the data are
stored online for further calculations.

27
Figure 6. The GreenFeed system (C-Lock Inc., Rapid City, South Dakota, USA). A) General
layout in the stand-alone concentrate feeder (Source: Hristov et al., 2015b). B) Online user
interface for visualizing methane (CH4) and carbon dioxide (CO2) fluxes. C) Carbon dioxide
recovery test for system calibration purposes.
The GreenFeed device uses a similar principle for measuring gas emissions
as in respiration chambers, where an active airflow is induced to capture
emitted air by integrating measurements of air flow, gas concentrations, and
detection of muzzle position (Zimmerman, 2011; Huhtanen et al., 2015b). The
ideal gas law is then used to convert the data in terms of mass fluxes and the
values obtained are adjusted for head position relative to the airflow.
Sniffer methods
Concentrations of both CH4 and CO2 in air released by eructation are recorded
by gas analysers throughout individual milking in robotic stations. Analysis of
changes in CH4 concentration provides information on frequency of eructation
and average CH4 release per eructation. The product of these variables
provides an estimate of CH4 emission rate for each milking.
Daily means are calculated, allowing for within-herd diurnal variation to be
taken into account, if necessary, for at least 7 days, as recommended by
Garnsworthy et al. (2012a), and combined into an overall mean for each cow.
Individual mean CH4 emission rates are then converted into daily CH4 output
using calibrations against chamber measurements (Garnsworthy et al., 2012a).
Thus, this technique does not measure CH4 emissions directly. Despite the
technique having been used to measure CH4 emissions on very large numbers
of animals in farm conditions, there are concerns over the accuracy,

28
repeatability and precision of the data obtained, which constrains the sensitivity
of the device to detect treatment differences in CH4 emissions (Hammond et
al., 2016). The sniffer methods are also highly dependent on the muzzle
position of the animal (Huhtanen et al., 2015b), an issue that significantly
increases the variation and compromises the reliability.
1.4.3 Tracer gas methods
Both sulphur hexafluoride (SF6) and CO2 techniques are based on the
concentration of a tracer gas for their measurements. One major requirement
for any tracer gas is that concentrations in the environment should be very low,
relative to the concentration of the tracer in collected samples, with
background gas concentrations accounted for (Berndt et al., 2014).
The SF6 technique
The SF6 technique is especially useful for measuring CH4 emissions in free-
ranging animals. General aspects of the SF6 technique are presented in Figure
7. Sulphur hexafluoride as a tracer gas is released from a bolus placed in the
rumen, gas samples are continuously collected from exhaled air in a canister
and the concentrations of SF6 and CH4 in the collected gas are analysed by
chromatographic methods. When SF6 release rate and gas concentration
(corrected for background) are known, CH4 flux can be calculated. Background
gas concentrations can be a problem in indoor conditions and therefore is
preferable to use this technique in grazing experiments (Hristov et al., 2016;
Dorich et al., 2015). Moreover, use of the SF6 technique to measure CH4
emissions in cannulated animals is not recommended because cannulation
introduces more variability into the SF6 technique with its head canister
(Beauchemin et al., 2012).
Although the SF6 procedure has been used for large-scale genetic and
nutritional evaluations, it remains labour-intensive, expensive and dependent
on technical specialists for operation and analysis (Hristov et al., 2013). The
use of SF6 has also been criticised since this chemical compound has a
greenhouse effect in the atmosphere (Berndt et al., 2014). Recent modifications
of the SF6 procedure have improved the accuracy of the technique (Deighton et
al., 2013).

29
Figure 7. The sulphur hexafluoride (SF6) tracer technique. A) Formula used in the calculations
according to Johnson et al. (1994). B) Sample collection apparatus (SLU, Uppsala, Sweden). C)
The SF6 technique under grazing conditions – back-mounting system for collection vessels
(Berndt et al., 2014).
The CO2 technique
Carbon dioxide is calculated as a function of heat production (heat calculated
from maintenance and production requirements) and CH4 emissions
determined by the product of the multiplication of CH4/CO2 ratio (ppm/ppm)
by CO2 production (L/day), as shown by Madsen et al. (2010). This technique
basically assumes that there is no variation in the efficiency of metabolisable
energy utilisation between animals for maintenance and production which not
makes biological sense. For instance high CH4/CO2 ratio can be as a result of
high emissions (more CH4) or high production efficiency (less CO2 per unit of
product). Therefore CH4/CO2 ratio cannot be a reliable indicator of CH4 fluxes
for individual animals. In the discussion section, the implications of CH4/CO2
ratio are discussed in detail.
1.4.4 The laser technique
A review by Chagunda (2013) summarises the potential of laser systems for
CH4 detection in dairy cows. Laser methane detection (LMD) equipment is
based on infrared absorption spectroscopy, using a semiconductor laser as a
collimated excitation source and using the second harmonic detection of
wavelength modulation spectroscopy to establish a CH4 concentration
measurement in parts per million-metre (ppm-m). The LMD equipment is able
to detect CH4 concentrations within a mix of gases in the environment.

30
Physical activity of the animal has a strong influence on CH4 emissions by this
technique and it has to be considered at individual animal level.
Despite of the fact that this novel technique is a non-invasive method, it is
still not widely used. However, a comparison study conducted in dairy cows
fed a diet of grass silage with 0.3 or 0.7 w/w of a concentrate supplement
demonstrated a high and positive correlation between measurements from the
LMD and the indirect open-circuit respiration calorimetric chamber (r = 0.8,
P<0.001) (Chagunda and Yan, 2011). However, the range in data was six-fold
higher, which increased the correlation coefficient. Within the practical range
for dairy cows (max 2-fold) the relationship was rather poor.
1.4.5 Proxies to measure in vivo methane
Novel non-invasive methods have been proposed to account for in vivo CH4
emissions in ruminants. Negussie et al. (2017) discussed the current potential
of available proxies as effective mitigation strategies. These methods can be
classified according to the chronological progression of nutrients through the
animal: (i) feed intake and feeding behaviour; (ii) rumen function, metabolites,
and microbiome; (iii) milk production and composition; (iv) hindgut and
faeces; and (v) measurements at the level of the whole animal (Negussie et al.,
2017). The authors concluded that most of proxies tend to be accurate only for
the production system and the environmental conditions under which they were
developed. As a result, the greatest shortcoming today is the lack of robustness
in their general applicability.
Both laser technique as the different proxies to measure in vivo CH4 in dairy
cows are not further discussed since they are beyond of the scope of the present
thesis.

31
2 Objectives
The overall aim of the studies presented in this thesis was to investigate the
contribution of different sources of variation to in vivo CH4 emissions from
dairy cows. Between-cow variation in CH4 emissions were further explored by
studying the effects of the measurement technique, animal-related factors and
diet effects. Specific objectives were to:
1. Compare two spot-sampling methods, i) the sniffer method and ii) the
flux method, for determining in vivo emissions from loose-housed dairy
cows.
2. Evaluate between-cow variability in different digestion and rumen
fermentation variables related to CH4 production and their contribution
to the observed individual animal variation.
3. Study the effects of graded replacement of late-harvested grass silage
and barley by highly digestible grass silage (early-harvested) on milk
production, CH4 and CO2 emissions and N efficiency.
4. Examine in depth the effects of graded replacement of late-harvested
grass silage and barley by highly digestible grass silage (early-
harvested) on the efficiency of ruminal and total tract digestion and
nutrient supply, in order to explain production responses in dairy cows.

32

33
3 Materials and methods
A general overview of the sources of variation contributing to between-cow
differences in CH4 emissions studied in this thesis is presented in Figure 8.
Figure 8. General layout showing sources of variation in CH4 emissions from dairy cows studied
in this thesis. Results from the flow study (Paper IV) are also included in the meta-analysis based
on an individual cow dataset (Paper II).
3.1 Paper I
Two spot-sampling methods for measuring CH4 emissions in cattle were
compared in dairy farm conditions. The gas emissions were measured using
portable units of the GreenFeed system (C-Lock Inc., Rapid City, SD) attached
either to a concentrate feeder or automatic milking system which was set up in
two different configurations (methods). In the first method (sniffer method),

34
both and CO2 concentrations were measured in close proximity to the muzzle
of the animal, and average concentrations or CH4/CO2 ratio were calculated. In
the second method (flux method), measurements of CH4 and CO2
concentrations were combined with an active airflow inside the feed troughs to
capture emitted gases coming from the animal. The flux method is the normal
set-up of the GreenFeed system and the purpose of the sniffer method was to
mimic the mechanism of commercially available gas analysers (Garnsworthy et
al., 2012a; Lassen et al., 2012). A muzzle sensor was used, allowing data to be
filtered according to the proximity to the cow’s head, allowing better estimates
of gas emissions for both methods. The proximity to the head adjustments for
each method was assessed by a study conducted in laboratory conditions using
a model cow’s head that emitted CO2 at a constant rate, by simulating different
cow head positions with respect to the manifold inlet.
The methods were compared in two on-farm studies conducted using either
32 (experiment 1) or 59 (experiment 2) cows in a switch-back design of 5 five
(experiment 1) or four (experiment 2) periods for replicate comparisons
between methods. In experiment 1, the experimental design was a cyclic
changeover with four blocks of eight mid-lactation Nordic Red cows, eight
diets and three experimental periods of 21 days. The eight treatments were
allocated to a 2 × 4 factorial arrangement consisting of two forages (mixtures
of grass and red clover silages), and increasing levels of CP in the diet by
gradually replacing ensiled barley grain with rapeseed expeller. Details of that
experiment are reported by Gidlund et al. (2017). Gas emissions for both
methods were recorded by two GreenFeed units attached to concentrate
feeders, which were switched to either sniffer or flux methods in the middle
and at the end of each experimental period. The cows were allowed to visit the
GreenFeed units every 7 h, and they were given eight 50 g servings of a
commercial concentrate at 40-s intervals during each visit.
Experiment 2, performed on Holstein-Friesian cows, lasted 40 days and
comprised four periods of 10 days. For method comparison purposes, the herd
was divided into two groups that were assigned to one of the two automatic
milking system units (AMS), which were retrofitted with both sniffer and flux-
method equipment and set up as follows: flux–sniffer–flux–sniffer (AMS 1)
and sniffer–flux–sniffer–flux (AMS 2). Cows were given unrestricted amounts
of total mixed ration (TMR; 60% forage, 40% concentrates on a DM basis) and
fed three times per day. In addition, the cows received a commercial
concentrate pellet during milking in the automatic system at a rate of 1 kg of
pellet per 7 kg of milk. Data obtained by each method were adjusted based on
filtering of muzzle position data as described previously.

35
Data (CH4 and CO2 or CH4/CO2 ratio and corresponding fluxes) from the
on-farm studies were analysed with linear mixed models (PROC MIXED; SAS
Institute, 2008), taking into account the fixed effects of period, diet
(experiment 1), DMI (experiment 1) and the random effect of cow.
Repeatability was calculated as: Rep = δ2cow / (δ2cow + δ2residual). The
relationships between concentrations of CH4 and CO2 or CH4/CO2 ratio and
corresponding fluxes were estimated by linear regression using least squares
means for each cow.
3.2 Paper II
In Paper II, a meta-analysis based on an individual cow dataset was conducted
to investigate the effects of between-cow variation and animal variables related
to CH4 emissions from dairy cows. Data were collected from 40 change-over
studies comprising a total of 637 cow/period observations. Animal production
and rumen fermentation characteristics were measured for 154 diets in 40
studies; diet digestibility in the total tract was measured for 135 diets in 34
studies, digesta flow was measured for 103 diets in 26 studies, and ruminal
digestion kinetics was measured for 56 diets in 15 studies. The experimental
diets were based on silages (mainly grass with some legume and whole-crop
silage), with cereal grains or by-products as energy supplements, and soybean
or rapeseed meal as protein supplements. Average forage: concentrate ratio
across all diets on DM basis was 59:41. The diets were fed ad libitum either as
TMR or fixed amounts of concentrate with forage ad libitum. Finnish feed
tables values (LUKE, 2016) were used when starch and fat content in
concentrate ingredients was not reported.
Apparent diet digestibility was determined by total faeces collection (27
studies) or by faeces spot sampling (seven studies) using either acid-insoluble
ash (Van Keulen and Young, 1977) or indigestible neutral detergent fibre
(NDF) (Huhtanen et al., 1994) as internal markers. Digesta flow was assessed
using the omasal sampling technique (Ahvenjärvi et al., 2000) with the triple-
marker system (France and Siddons, 1986). Microbial N synthesis was
determined using 15N as a microbial marker except in two studies, where
purine-based derivatives were used. Rumen pool size was determined by
rumen evacuation and digestion and passage kinetics variables were calculated
using the compartmental flux method (Ellis et al., 1994). Rumen fluid samples
were mostly collected over 12 h after morning feeding to obtain fermentation
parameters. The main volatile fatty acids (VFAs) were used to determine ratios
between these (e.g. acetate: propionate) and production of both CH4 and CO2
per mol of volatile fatty acids (CH4VFA and CO2VFA, respectively) based on

36
stoichiometry principles according to Wolin (1960). Furthermore, CH4VFA in
addition to OM apparently digested in the rumen (OMADR) was used as the
basis to predict total CH4 production (g/day), which was further adjusted by
hydrogen sinks such as microbial cells (Czerkawski, 1986) and
biohydrogenation of fatty acids using the equation in the Karoline model
(Huhtanen et al., 2015c). The predicted total CH4 emissions were compared
with two empirical equations, those presented by Yan et al. (2000), and Ramin
and Huhtanen (2013), using data obtained from studies conducted in
respiration chambers.
Variance components analysis was used for the most relevant variables
associated with enteric CH4 production to calculate the random effects of:
experiment (Exp), Cow(Exp), Diet(Exp), Period(Exp) and residual variation. In
addition, repeatability values were determined as in Paper I. Single regression
models were developed based on their biological value in CH4 production. The
models included two random statements: a random intercept and slope of X1
with SUBJECT = Diet (Exp), and a random intercept with SUBJECT = Period
(Exp), using the TYPE = VC as covariance structure for both random
statements. The maximum likelihood method was used in the PROC MIXED
model syntax (version 9.3; SAS Institute Inc., Cary. NC). The purpose was to
remove both period and diet effects.
3.3 Paper III
A study was conducted at Röbäcksdalen experimental farm, Swedish
University of Agricultural Sciences, Umeå, Sweden (63º45’N; 20º17’E), to
investigate the effects of replacing late-harvested grass silage (LS) and barley
by early-harvested grass silage (ES) on performance and CH4 emissions in
dairy cows. Sixteen Nordic Red cows with mean BW of 635 ± 76.0 kg, at 79 ±
14.4 days in milk (DIM) and producing 34 ± 6.9 kg milk/day at the beginning
of the experiment were used in a replicate 4 x 4 Latin square design. Each 28-
day period comprised 14 days of diet adaptation followed by 14 days of data
collection. Cows were offered the diets ad libitum four times per day as TMR,
with free access to water, and were milked twice daily.
Two grass silages were prepared from the same primary growth of a third-
year ley dominated by timothy grass (Phleum pratense) with some red clover
(Trifolium pratense) harvested two weeks apart. A mixture of LS and rolled
barley was gradually replaced by ES (0, 33, 67, and 100% of the forage
component of the diet), in order to obtain four diets defined as: Late-cut (L),
late-early (LE), early-late (EL) and early-cut (E) silage. The proportion of
forage increased from 42 to 64 % on DM basis with increasing proportion of

37
ES in the TMR diet without considering extra concentrate supplementation
from the GreenFeed system attached to concentrate feeders. Heat-treated
solvent-extracted rapeseed was used as the protein supplement. The diets were
formulated to meet the metabolisable energy (ME) and metabolisable protein
(MP) requirements for 35 kg energy-corrected milk (ECM) per day.
Apparent diet digestibility was assessed by collecting grab samples of
faeces from the rectum of eight cows (two squares). Ash-free indigestible NDF
(iNDF) concentration was used as an internal marker to calculate diet
digestibility (Huhtanen et al. 1994). Milk yield was recorded daily, and
samples were taken for milk composition analysis at four consecutive milkings.
Gas emissions (CH4 and CO2) were measured using the GreenFeed system (C-
Lock Inc., Rapid City, SD) as described by Huhtanen et al. (2015b) and
Hammond et al. (2016). The GreenFeed system was programmed to allow each
animal to visit the two units at minimum 5-h intervals and they were given
eight 50 g servings of commercial concentrate at 40-s intervals during each
visit.
Chemical composition and feeding values of the diets were calculated from
the proportion of ingredients and their respective values. Energy-corrected
milk was calculated according to Sjaunja et al. (1990). Feed efficiency was
calculated as ECM yield (kg/d)/DMI (kg/d) and milk N efficiency (MNE) as
milk N [CP (g/day)/6.38]/N intake (kg/day). Methane and CO2 production
were calculated as mean daily production during the last 14 days of each
period. The experimental data were analysed by ANOVA for a replicate 4 × 4
Latin square design using the MIXED procedure of SAS (Version 9.3, SAS
Inst., Inc., Cary, NC) and orthogonal polynomial contrasts were used to
evaluate linear and quadratic effects of treatments.
3.4 Paper IV
The aim of Paper IV was to study the effects of the diets used in Paper III on
rumen fermentation, microbial N synthesis, diet digestion and digestion
kinetics using rumen-cannulated cows in a tie-stall system. This study was
conducted in parallel with the production study (Paper III). Four multiparous
rumen-cannulated Nordic Red cows averaging 676 ± 79 kg of BW, at 90 ±
19.1 days in milk and yielding 30.9 ± 6.27 kg of milk at the start of the
experiment were used in a balanced 4 x 4 Latin square design. The
experimental periods lasted for 21 days and were divided into 14 days of
adaptation and seven days of data collection. The cows were fed manually with
the experimental diets as TMR ad libitum and milked twice daily. To mimic
extra concentrate supply from the GreenFeed system, 1 kg DM/day of the same

38
feed was added to give similar diet composition in both studies. Orts were
recorded once daily and feeding rate was adjusted to provide 10% extra of the
previous calculated intake except during the 4 days of sampling from the
omasum, when feeding rate was restricted to 95% of previous ad libitum
intake.
Total tract digestibility was assessed as described in Paper III. Two rumen
evacuations were conducted, at 4 h after (d 12) and 1 h before (d 14) the
morning feeding, to give a representative estimate of rumen digesta pool size
and digestion kinetics. After the last rumen evacuation, in situ bags containing
2 g DM of early- or late-harvested grass silage were placed in the rumen for 24
h to evaluate the effects of diet composition on rumen fibrolytic activity. The
omasal sampling technique (Huhtanen et al., 1997), as modified by Ahvenjärvi
et al. (2000), was used for collection of digesta samples from the omasum on
day 17 to day 20, with 4 h intervals between the three sampling occasions each
day.
Omasal flow and ruminal digestibility of nutrients were calculated using the
reconstitution system based on a triple marker technique (Cr-EDTA, Yb-
acetate and iNDF; France and Siddons, 1986). Microbial protein synthesis was
determined using 15N as microbial marker (Broderick and Merchen, 1992).
Samples of rumen fluid (n = 8 time points) were collected at 1.5 h intervals on
day 21 to measure pH and VFAs concentrations. Production of CH4 per mol of
VFA (CH4VFA) was calculated based on VFA stoichiometry equations
(Wolin, 1960).
Calculation of omasal flow of nutrients was based on the triple-marker
method (France and Siddons, 1986) and daily marker doses recovered from
faeces according to Armentano and Russell (1985). True OM digestibility was
corrected for VFA flow according to Huhtanen et al., (2010). Flows of OM and
non-ammonia nitrogen (NAN) were corrected for microbial OM and microbial
NAN, respectively. Digestion kinetic variables were calculated by the
compartmental flux method (Ellis et al., 1994). The experimental data were
analysed by ANOVA for a 4 × 4 Latin square design using the MIXED
procedure of SAS (Version 9.3, SAS Inst., Inc., Cary, NC) and orthogonal
polynomial contrasts were considered to assess the effect of the diets.

39
4 Results
4.1 Paper I
The study conducted in laboratory conditions (robot) demonstrated that for the
sniffer method, the muzzle distance from the sampling point (0-30 cm) was a
key factor determining gas concentrations. Muzzle distance had no effect on
the recovery of CO2 with the flux method, regardless of the cow head position,
head movement or breath rate, and the variability was much smaller compared
with the sniffer method throughout all recovery tests. Wind (6 m/s) was the
only factor that clearly decreased the CO2 recovery of the flux method, to
about 85%.
In on-farm conditions (experiment 2), repeatability of muzzle position
across the cows was 0.74 and 0.82 when analysed using daily (n = 40) and
period (n = 4) data for each cow, respectively. When the flux-method data for
all cows were not filtered according to muzzle position, fractional time with
muzzle inside a manifold (i.e. from 0 to 1) and CH4 flux showed a positive
relationship (R2 = 0.31; P<0.001). Weak relationships were found between the
CH4 concentration (ppm) determined by the sniffer method and by the flux
method in both experiments. Between-cow coefficient of variation (CV) in
CH4 flux decreased from 21.2 to 17.6 % when using filtered data. The
CH4/CO2 ratio determined by the sniffer method was negatively (P<0.001)
related to muzzle position in experiment 2.
Total CH4 flux was similar in both experiments, but CO2 flux was
numerically greater in experiment 2. Both CH4 and CO2 concentrations
measured using the sniffer method were markedly lower in experiment 2
compared with experiment 1, indicating that the geometric structure of the
head-box (GreenFeed compared with automatic milking system) influences the
dilution of exhaled gases. The between-cow CV of CH4 emissions was greater
with the sniffer method compared with the flux method in both experiments.

40
However, between-cow CV values of the CH4/CO2 ratio for the flux and sniffer
methods were rather similar (6.4 and 6.6 %, respectively, in experiment 1 and
8.8 and 7.5%, respectively, in experiment 2). Repeatability of gas
measurements (CH4 and CO2) was generally high (>0.70), and the values were
similar between the experiments and methods.
The relationship between the sniffer method CH4 concentration and CH4
flux was significant (P=0.02) in experiment 2, but this was not replicated in
experiment 1 (P=0.11). The intercept (i.e. observed CH4 flux at zero CH4
concentration) was highly significant (P<0.001) for the sniffer method in both
studies, suggesting larger random error compared with the flux method. The
relationship between CH4/CO2 ratio and CH4 flux determined by the sniffer
and flux methods was statistically significant (P<0.01) in both studies, but R2
values were higher in experiment 1 than in experiment 2. In experiment 1, the
CH4/CO2 ratio was positively related to CH4 emissions per kilogram of DMI
when measured by the flux method, but not when measured by the sniffer
method.
4.2 Paper II
The dataset was representative of feeding conditions for dairy cows in
Northern Europe. The forage: concentrate proportion was 59:41 on DM basis
and dietary concentrations of CP and NDF were 160 ± 21.3 and 394 ± 54.8
g/kg DM respectively. Dry matter intake was on average 18.9±3.35 kg/day.
The variability and repeatability in molar proportions of VFA were generally
small. The variance component Diet(Exp) was on average two-fold larger than
the observed for Cow(Exp), except for butyrate. Low variability in the main
VFAs was reflected in calculated stoichiometric CH4VFA. Between-cow
variation and repeatability of CH4VFA were very low (0.010 and 0.10,
respectively), suggesting that rumen fermentation does not markedly contribute
to between animal variation in CH4 emissions. Total VFA concentration was
more repeatable (0.48) than molar proportions of individual VFA.
Organic matter digestibility (OMD) was within the expected range for good
quality grass silages (740 ± 39.9 g/kg). Between-cow variability in OMD and
NDF digestibility (NDFD) was highly significant (P<0.001), but rather small
(13 and 23 g/kg, respectively). For digestibility variables, the variance
component Diet(Exp) was the largest source of variation. Digestibility
variables had medium repeatability (Rep = 0.37 and 0.26 for OMD and NDFD,
respectively). The contribution of Cow(Exp) variance component to the
observed variation for both the OM and NDF pools per kg of BW was higher

41
than the observed for both fibre digestion and passage rate, and the same trend
was observed for repeatability values (rumen pools >0.70).
Differences in rumen ammonia N (RAN) concentration were mainly related
to differences in protein concentration and sources across the diets. Between-
cow variation in RAN was of greater magnitude (CV = 0.149) than the
variation in other N metabolism variables. However, repeatability for RAN
was similar to the efficiency of microbial N synthesis per kg of OM truly
digested in the rumen (OMTDR; Rep = 0.35 and 0.34, respectively). In
general, between-cow variability and repeatability of variables related to
ruminal N metabolism were greater than those related to rumen VFA pattern
and diet digestibility.
Although in vivo CH4 estimates based on total predicted CH4 and those
obtained by empirical equations were rather similar to each other (on average
392 g/day), the variation in total CH4 production was around two-fold (CV =
0.28) higher than observed for the empirical models evaluated. Random
variation in OM digested in the rumen across diets and studies could have
contributed to this. Methane yield was less variable than predictions based on
total CH4 emissions. The estimated CV values for total CH4 and CH4 yield
based on stoichiometric calculations were similar to those observed in
respiration chambers.
Stoichiometric CH4VFA increased (P<0.01) with increased OMD. For each
unit (g/kg) increase in OMD, CH4VFA decreased by 0.06 mmol/mol VFA. The
variation in OMD was closely related to the variation in NDFD and it was
reflected in the positive relationship (P<0.01) with digestion rate of potentially
digestible NDF (pdNDF). Organic matter digestibility increased as RAN
increased (P<0.01), but it was negatively (P<0.01) associated with microbial N
flow and the efficiency of microbial N synthesis. Rumen ammonia N decreased
(P<0.01) with increased passage rate of iNDF and molar proportion of
propionate. In addition, RAN concentrations were positively associated with
molar proportions of branched-chain VFA (BCVFA) in the rumen (P≤0.01)
(models not shown). Overall, the effects of digestion and fermentation
variables were additive.
4.3 Paper III
Differences between diets in which late-harvested grass silage (LS) and barley
were replaced by 0% early-harvested grass silage (ES) (L diet) and by 100%
early-harvested grass silage (E diet) in terms of dietary concentrations of
digestible organic matter (DOM) and metabolisable energy (ME) were as
expected, a consequence of different harvesting times. However, they were

42
slightly lower than expected for both silages, probably due to exceptionally
warm weather conditions during early summer. In both cases, silage
fermentation quality was good, as indicated by low pH and ammonia-N
concentrations.
The main differences in diet composition between treatments were related
to decreases in starch and increases in pdNDF supply due to graded addition of
ES in the diet. Dry matter intake decreased linearly (P<0.01) with increasing
proportion of early-harvested grass silage in the diet from 22.6 to 19.3 kg/day,
whereas digestibility of nutrients increased linearly (P<0.01). The greatest
numerical differences in digestibility between the L diet (0% ES) and the E diet
(100% ES) were observed for NDF and CP, which accounted for 122 and 100
g/kg, respectively. The higher digestibility observed E diet was consistent with
reduced faecal output of nutrients.
Decreased concentrate supplementation in the E diet did not have effect on
milk or ECM yield, except for milk protein which decreased linearly (P<0.01).
Milk fat concentration increased and protein concentration decreased with
increased proportion of early-harvested silage in the diet. The concentration of
milk urea N (MUN) increased linearly (P<0.01) from 9.7 to 11.9 mg/dL for
diet L and diet E respectively, but milk N efficiency was not influenced by the
diet. Total CH4 and CO2 production and gas emissions per kg of ECM were not
influenced (P>0.01) by the addition of early-harvested silage in the diet, but
CH4 yield increased linearly (P<0.01). This reflected differences in DMI and
OMD, and probably the composition of digested OM, among treatments.
Greater faecal output (g/kg DMI) of potentially digestible nutrients (NDS +
pdNDF) for diet L compared with diet E could counterbalance the reduced
enteric CH4 yield by increasing the potential for CH4 emissions from manure,
since more substrate is available for fermentation.
4.4 Paper IV
Overall, intake and milk production responses and total tract digestibility in
rumen-cannulated cows were consistent with observed trends described in
Paper III. Differences in rumen fermentation pattern between the diets were
only detected for the molar proportions of isovalerate and valerate, which
decreased linearly (P≤0.03) when the proportion of early-harvested silage
increased in the diet. Stoichiometric CH4VFA molar concentration was not
influenced by the dietary composition (P>0.10), which is consistent with
observations for the major volatile fatty acids, indicating that rumen
fermentation did not contribute to higher CH4 yield with increased proportion
of ES in Paper III. Both apparent and true ruminal OM digestibility increased

43
linearly (P≤0.04) with graded addition of ES silage in the diet and it was
consistent with linear increases (P<0.01) for both ruminal and total
digestibility of NDF and pdNDF. Organic matter and NDF flows into the
omasum were reduced (P<0.01) in response to decreased intake of these
nutrients in diets that included early-harvested silage.
There were no differences between treatments in terms of N intake or the
flow of different N fractions into the omasum (P>0.05), but the flow of feed N
into the omasum tended to decrease (P=0.08) as the inclusion rate of ES in the
diet increased. The ruminal N degradability and the total tract N digestibility
increased linearly (P≤0.04) with increased proportions of ES in the diet.
Decreased 15N enrichment (P=0.03) of rumen bacteria was observed in diets
containing early-harvested silage. Increased inclusion rate of ES in the diet
resulted in linear decreases (P<0.05) in both NDF and pdNDF pool sizes and
this was reflected in faster ruminal turnover time of NDF, which decreased
linearly (P<0.01) from 24.9 with the L diet (0% early-harvested silage) to 18.7
h with the E diet (100% early-harvested silage). No differences were observed
between treatments in passage rate of iNDF and pdNDF, but intake and
digestion rates of pdNDF linearly increased (P<0.01) as the proportion of ES
in the diet increased. Diet did not have any influence on ruminal in situ
degradation of silage DM or NDF. Improved OMD, numerically lower
efficiency of microbial and probably minor changes in the site of digestion
could explain the increased CH4 yield with increasing proportion of early-
harvested silage in the diet.

44

45
5 Discussion
5.1 Measurement technique
5.1.1 Between-cow variability related to methods
The literature reports considerable between-animal variation in CH4 production
values across different measurement techniques. The influence of between-cow
variation compromises the repeatability of CH4 measurements and highlights
the need for revising the particularities of each technique, in order to minimise
confounding effects from undesirable sources of variation. Technical
limitations of the methods used in measuring CH4 production may explain why
it has been difficult to obtain consistent rankings in CH4 yields when animals
are measured on multiple occasions (Vlaming et al., 2008).
In the past, respiratory chambers have provided the most accurate and
reliable between-animal coefficient of variation (CV) for CH4 production by
farm animals compared with other techniques, due to the characteristics of the
equipment itself, and also due to the possibility of applying stronger controls
on the experimental animals and thus reducing variation in the CH4
measurements obtained. Blaxter and Clapperton (1965), using standard
respiration chambers, reported 7-8% between-animal variation in CH4
production, whereas studies conducted by Grainger et al. (2007) and Muñoz et
al. (2012) using the SF6 tracer technique reported greater between-cow CV
(16.4 and 19.3 %, respectively). Compared with respiration chambers and SF6
methods, studies using sniffer methods have reported even larger between-cow
variation, e.g. Garnsworthy et al. (2012a) observed considerable variation
between cows in their CH4 emission rate index (mg CH4/min) during milking
(CV = 33%). A study by Bell et al. (2014) using a similar sniffer system on 21
commercial farms (n = 1964 cows) showed that the extent of between-cow CV
for the CH4 production index varied from 22 to 67 % within farms. In addition,
there were six-fold differences between the farms reported in the study by Bell

46
et al. (2014), which might indicate difficulties in harmonisation of the sniffer
method between farms. Although dietary and animal factors could contribute to
variation in CH4 measurements, such differences between dairy farms are not
possible according to our current knowledge of factors influencing CH4
production Overall, the between-cow variability observed for the sniffer
methods within farms was much higher (17%) than reported in the dataset by
Yan et al. (2010) obtained from respiration chamber studies (n = 579 cows),
despite the large variability in individual animal and diet factors (BW: 379-733
kg; DMI: 7.5-25.0 kg/d; forage proportion: 18-100% of diet DM, NDF: 265-
604 g/kg DM). Therefore, it is likely that high CV values for a group of
animals in the same house, fed the same diet, reflect random error in
measurements rather than true between-animal variation.
In addition to the between-cow variation in CH4 production, it is also
important to consider the variation in the CH4/CO2 ratio, which is far from
being a constant value. In a dataset derived from studies conducted with dairy
cows in respiration chambers fed a wide range of diets (n = 157 observations;
30 diets), the CV of the CH4/CO2 ratio was 0.095 (Hellwing et al., 2013).
However, much greater variation in CH4/CO2 ratio (~0.15-0.20) has been
reported with the sniffer method (Lassen et al., 2012; Lassen and Løvendahl,
2013; Haque et al., 2014).
Variation in methane production with the GreenFeed system
The extent of between-cow variation and repeatability for CH4 production
(g/day) in studies conducted with dairy cows at Röbäcksdalen Research Centre
using the GreenFeed system is presented in Table 3.
The data in the Table 3 were taken from studies using either a Latin square,
cyclic change-over or switch-back design. The diets used in these experiments
were based on grass and can be considered representative of typical dairy cow
diets in the Nordic countries, with mean forage to concentrate ratio of 40-45%
on a DM basis. The concentrate supplements consisted principally of cereal
grains, some fibrous by-products from the food industry and protein
supplements, typically rapeseed meal. In all studies, the cows were fed a total
mixed ration ad libitum. Mean and residual standard deviation (SD) values for
CH4 production were obtained from least squares (LS) means for cows using
the general linear model procedure, while repeatability was assessed by the
covariance test using the mixed procedure of SAS (SAS Institute, 2008). The
use of a mixed model allowed the effects of diet and period to be removed, and
therefore only between-animal differences were considered. Total CH4
production was on average 435 g/day, in line with GreenFeed-measured data
for dairy cows reported in the literature (Dorich et al., 2015; Gidlund et al.,

47
2015; Hristov et al., 2015a). Assuming a constant value of 18.5 MJ gross
energy (GE) in DM in the forage, an estimated 6.5% of GE was lost in CH4.
This is in line with values reported by Yan et al. (2000; 2010) for cows in
respiration chambers fed similar grass silage-based diets. In this thesis, the
extent of between-cow variation (0.107) was higher than the residual variation,
which in turn reflected the high repeatability of the GreenFeed technique in
farm conditions (0.69) despite the contrasting dietary conditions across
experiments. Average repeatability across experiments (Table 1) was
consistent with repeatabilities presented in a previous report (>0.70; Huhtanen
et al., 2013).
Table 3. Total methane (CH4) production (g/day) in dairy cows and its variation across different
experiments conducted at Röbäcksdalen Research Centre (Umeå, Sweden) from 2012 to 2016
using the GreenFeed system
Study
Exp. Design Diet N
Per.
Mean, g/d
SD CV
Res SD
Res CV
1 Rep2
1 Latin square Forage
20
4 405
36.4
0.090
21.8 0.054
0.717
2 Latin square Forage
30
5 421
34.2
0.081
29.3 0.070
0.529
3 Latin square Straw
16
4 419
48.8
0.116
37.3 0.084
0.597
4 Switch-back Concentrate
16
3 451
53.0
0.118
22.1 0.049
0.844
5
Cyclic
-change over
For. x Conc.
16
4 443
48.8
0.110
37.3 0.084
0.597
6 Latin square Oats
16
4 454
54.7
0.121
21.7 0.048
0.860
7
Cyclic change
-over Protein
29
3 455
44.4
0.098
24.3 0.054
0.755
8
Cyclic change
-over Protein
24
3 395
36.5
0.092
31.1 0.079
0.512
9
Cyclic change
-over Protein
25
2 453
48.5
0.107
30.2 0.066
0.654
10 Switch-back Glycerol
22
3 452
60.1
0.133
27.6 0.061
0.814
1Residual coefficient of variation.
2Proportion of significant correlation coefficient (P<0.05).
The average between-cow CV for CH4 production across different
experiments using the GreenFeed system at Röbäcksdalen Research Centre
(Umeå, Sweden) was greater (0.107) than observed in carefully conducted
respiration chamber studies (0.072) (Blaxter and Clapperton, 1965), but even
lower to values observed with respiration chambers in studies conducted more
recently (0.178) (Grainger et al., 2007). In a recent review, Hammond et al.
(2016) concluded that published GreenFeed estimates of daily CH4 production
(mean values) are in agreement with those measured in respiration chambers.
In previous studies, those authors made direct comparisons between GreenFeed
and respiration chamber techniques with growing (Hammond et al., 2015) and
dairy cattle (Hammond et al., 2016b). Since the principle of the techniques is
different, comparison between the techniques using the same animals at the
same time is not possible. In the dairy cow study, two separate experiments

48
evaluating the same diets were conducted simultaneously, to compare the
measurement technique for CH4 production: experiment 1 used a randomised
block design for GreenFeed (40 cows) and experiment 2 used a Latin square
design for respiration chambers (four cows) (Hammond et al., 2016b). Both
techniques were able to detect similar dietary treatment effects, despite
differences in intake between studies, but the magnitude of the differences was
considerably different, e.g. 24% lower in experiment 1 using GreenFeed and
8% lower in experiment 2 using respiration chambers for maize silage
compared with grass silage as the only source of forage in the diet (Hammond
et al., 2016b). In the growing cattle study (Hammond et al., 2015), the
GreenFeed system provided an average estimate of CH4 production that was
not different from respiration chamber measurements using the same
experimental animals. However, GreenFeed and respiration chamber
techniques showed poor agreement. According to those authors, this is partly
attributable to the relatively small number of visits recorded by the GreenFeed
method. Arthur et al. (2017) recommend a minimum of 30 flux records, with
each record obtained from a minimum GreenFeed visit duration of 3 minutes.
As a spot-sampling technique, GreenFeed relies on the number of animal visits
during the day, whereas respiration chamber measurements are based on
integrated measurements at specific time intervals according to the system set-
up.
Thus, for the GreenFeed system, the greater the number of visits recorded,
the better the accuracy of the measurements. Results from Paper III (study 5
in Table 3), showed that the diurnal pattern in CH4 production recorded by the
GreenFeed system did not differ across diets and there was no clear diurnal
pattern (Figure 9). Conversely, in the study by Hammond et al. (2015) there
were some diurnal variations in number of visits.
An indirect comparison was performed by Huhtanen et al. (2016), by
comparing CH4 measured with the GreenFeed system in cattle with model-
predicted (six models) CH4 production. Mean CH4 production estimated by
GreenFeed was close to values predicted by models developed from respiration
chamber data (386 and 384 g/day, respectively). However, it was much higher
than that predicted by the model proposed by Ellis et al. (2007), which was
developed from data determined by different techniques (386 and 294 g/day,
respectively). Root mean square prediction error (RMSPE) ranged from 6.0 to
8.9 % of observed mean for models developed from the respiration chamber
data. Huhtanen et al. (2016) concluded that CH4 production estimated by the
GreenFeed system is consistent with values predicted by models derived from
large datasets from respiration chamber studies.

49
Figure 9. Mean diurnal pattern of methane (CH4) production observed in dairy cows (n=15) using
the GreenFeed system (Paper III). One cow was excluded owing to insufficient data.
Residual CV in general linear model analysis was on average 0.065 (range
0.048-0.084). Considering that 20 cows can be measured in one GreenFeed
unit in normal conditions, the probability of detecting biologically meaningful
differences in change-over studies is rather high. For example, in
quadruplicated 4 × 4 Latin square studies with a 2 × 2 factorial arrangement,
the probability of detecting differences (P<0.05) of 10, 7.5 and 5% between
main factors (n = 32) was >99, 93 and 61 %, respectively, for the highest
observed residual CV (0.084) obtained here. The corresponding probability
using the mean residual CV (0.065) was >99, 99 and 83 %, respectively.
GreenFeed compared with other methods to measure methane production
The GreenFeed technique has recently been compared with the SF6 tracer
technique. Dorich et al. (2015) performed a direct comparison between the two
techniques by measuring CH4 production in dairy cows with the same diet
(52:48 % on DM basis) fed either ad libitum or restricted feed to 90% of the
baseline DMI (cross-over design). The results showed that the SF6 tracer
method produces larger CV than obtained by the GreenFeed system, despite
the average values being virtually the same (468 and 467 g/day for GreenFeed
and SF6, respectively). When outliers were removed from the SF6 data, mean
value decreased to 405 g/day but, although variation was substantially reduced,
it still remained considerably higher than for GreenFeed data (CV = 0.41 and

50
0.22 for SF6 and GreenFeed, respectively). A moderately strong relationship
between CH4 production and DMI was observed for the GreenFeed system (R2
= 0.42) and a weak relationship for the SF6 technique (R2 = 0.17). Dorich et al.
(2015) attributed the higher variability in the SF6 measurements to the high
concentration of background gases, as a result of poor house ventilation in
indoor conditions. This is in agreement with findings by Hristov et al. (2016)
that correlation and concordance between the two methods are relatively low.
In addition, the difference between the methods was not consistent over time,
most likely influenced by house ventilation and background methane and SF6
concentrations. One major requirement for any tracer gas technique is that the
background concentrations in the environment should be low relative to the
concentration of the tracer in the samples collected (Berndt et al., 2014).
Paper I compared the GreenFeed system (flux method) with a modification
of the original system set-up, in order to mimic the mechanism of the sniffer
technique based on analysis of gas concentrations (Garnsworthy et al., 2012a).
Details of the experimental conditions applied in Paper I can be found in the
material and methods section of this thesis. Between-cow coefficient of
variation in CH4 was smaller for the GreenFeed system (range 0.11-0.18) than
for the sniffer technique (range 0.18-0.28). Although the repeatability of the
measurements by both methods was high for CH4 production (≥ 0.72) and
CH4/CO2 ratio (≥ 0.59), the relationship between the CH4 concentration (ppm)
determined by the sniffer method and CH4 production (g/d) determined by the
GreenFeed method was rather poor, as indicated by the low coefficient of
determination of the linear regression (R2 = 0.09). In contrast, Garnsworthy et
al. (2012a) reported a good relationship between methane emission index
measured by the sniffer method and respiration chambers.
In Paper I, CH4 values from the GreenFeed system were strongly related
either to DMI (experiment 1) or BW (experiment 2), whereas for the sniffer
method no significant relationships were observed for these variables.
Similarly, Garnsworthy et al. (2012a,b) and Bell et al. (2014) reported that
increased DMI was poorly associated with increases in CH4 emission rate.
Since DMI is the main driver determining CH4 production in ruminants, as
determined in large datasets derived from respiration chamber studies (Yan et
al., 2000; Ramin and Huhtanen, 2013), sniffer values lack biological value in
terms of animal physiology mechanisms related to CH4 production. In addition,
in both the study by Garnsworthy et al. (2012a) and Paper I, the intercepts of
regressions predicting fluxes from CH4 concentrations were highly positive.
Theoretically this is not possible (positive flux at zero concentration) and it
most likely reflects random variation in the gas concentration measurements.
High repeatability in both CH4 concentrations and CH4/CO2 ratio for the sniffer

51
method were at least partly associated with greater variability of the data and
not necessarily the accuracy of the technique. Results from both laboratory and
on-farm studies indicated that, for the sniffer method, the muzzle distance from
the sampling point is a critical factor in determining the concentrations. Indeed,
the repeatability of muzzle position was as high as 0.82 for experiment 2 in
Paper I. This may seem surprising, but different characteristics of animal
behaviour can be highly repeatable (e.g. Napolitano et al., 2005). In
experiment 1 in Paper I, repeatability of number of visits was also highly
variable (0.50-0.68). In addition to the muzzle position, differences in manifold
geometry and number of cow visits between concentrate feeders and automatic
milk stations seemed to contribute to larger variation in CH4 and CO2
concentrations in the sniffer method in experiment 2, which were similar to
values reported by de Haas et al. (2013). From an image of a cow breathing
(Figure 10), it is clear that exhaled air goes in two directions at a near 90-
degree angle and that small changes in head position can influence measured
CH4 concentrations by the sniffer method. Therefore, the combined effects of
the smaller head-box for concentrate feeders attached to the GreenFeed unit
compared with the automatic milking system and the muzzle position may add
non-accounted variation to the predictions.
Figure 10. Image of a cow breathing, showing the direction of breath from each nostril.
http://1.bp.blogspot.com/_6Zl8x3ZGFMY/TQTIr9xDjXI/AAAAAAAAG48/05AQZdFZFHI/s16
00/cow_breath.jpg (Accessed 8 May, 2014).
Overall, Paper I showed that CH4 measured by the sniffer method is a poor
indicator for ranking animals for selection purposes, whereas GreenFeed
showed more realistic results in terms of CV and goods agreement with
respiration chamber data, both in direct and indirect comparisons.

52
5.1.2 CH4/CO2 ratio
Carbon dioxide comes from fermentation of the feed in the gastrointestinal
tract and also from tissue mobilisation, whereas CH4 can only be produced
from enteric fermentation in the rumen and to a limited extent in the hindgut.
However, CO2 and CH4 production are positively associated because both are
highly correlated with DMI (Pinares-Patiño et al., 2007). Therefore it can be
expected that the CH4/CO2 ratio is not constant. In a dataset derived from
respiration chamber studies conducted with dairy cows (157 observations, 30
diets), the CH4/CO2 ratio varied between 0.053 and 0.105 (Hellwing et al.,
2013). However, in Paper III, which compared the effects of graded
replacement of late-harvested and early-harvested grass silage and barley, the
CH4/CO2 ratio (g/kg) was rather constant (33.2-34.2), despite the large
differences in dietary carbohydrate composition. Because both ECM yield and
BW were similar between the diets, calculated CO2 production was also
similar. In Paper III, moderate relationships were observed between the
CH4/CO2 ratio (R2 = 0.56) or CH4 production (R2 = 0.57) predicted according
to Madsen et al. (2010) and observed CH4 production, but predicted values
were on average 11% lower. However, it should be noted that the CH4/CO2
ratio was measured by the GreenFeed mode that resulted in much better
relationship between gas ratio and CH4 flux than the gas ratio measured in the
sniffer mode. In Paper I, the CH4/CO2 ratio (ppm/ppm) was 0.107 and 0.088
for the GreenFeed method and 0.094 and 0.100 for the sniffer method in
experiment 1 and 2, respectively. In that study, the better relationship between
CH4/CO2 ratio and CH4 flux than the corresponding relationship between CH4
concentration and flux suggests that CH4/CO2 ratio could be more useful in
ranking cows as emitters than CH4 concentration, as is the case for the sniffer
method.
The problem with CH4/CO2 is that it can be influenced by the CH4 and the
CO2 concentrations, both of which can vary because of different biological
mechanisms. High CH4/CO2 ratio can result from increased CH4 production as
a consequence of increased DMI and/or high CH4 yield and from improved
feed efficiency due to reduced CO2 production per unit intake as the result of
allocation to both milk and body tissues. Conversely, in addition to low CH4
emissions, low CH4/CO2 ratio can also result from mobilisation of body
tissues, which produces CO2 but not CH4.
Monte Carlo simulation
One problem with using CH4/CO2 and estimated CO2 production for predicting
total CH4 production is that it is not known whether the gas ratio changes due
to increased CH4, decreased CO2 or both. To evaluate this, a Monte Carlo

53
simulation study was conducted to evaluate the effects of efficiency of ME
utilisation on predictions of total CH4 production. The following default values
were used: DMI 20 kg/d (CV = 2.4 kg), CH4 production was adjusted for
differences in DMI (-0.35 g per kg/DMI deviation from 20 kg/day; Ramin and
Huhtanen, 2013); a value of 0.10 was used for CV in CH4 production and 11.5
MJ/kg DM for dietary ME concentration; a constant value of 70 MJ/d was
assumed for maintenance heat production; yield of ECM was calculated as
0.62 × (ME intake – ME for maintenance)/3.14 (MJ/kg ECM); and total
estimated heat production (MJ/d) was calculated as 70 + ECM yield (kg/d) ×
1.92 MJ/kg (3.14 / 0.62 – 3.14). A herd of 100 cows was simulated 1000 times
assuming CV values of 0.06, 0.08 and 0.10 for kl (efficiency of ME utilisation
for lactation above maintenance; default = 0.62). The correlation between DMI
and dietary ME concentration and between CH4 production and the efficiency
of ME utilisation was assumed to be 0. Irrespective of CV for kl, the efficiency
of ME utilisation for lactation was consistently negatively related to residual
CH4 production (observed – predicted), i.e. CH4 production estimated from
CH4/CO2, and predicted CO2 production was overestimated with improvements
in feed efficiency (FE; Table 4).
The greater the variation in kl, the stronger the negative relationship
between FE and residual CH4 production was. With no or small (e.g. CV =
0.04) variability in kl, the CH4/CO2 ratio was not associated with FE (results
not shown), but with a CV value of 0.06 for the correlation between CH4/CO2
and FE, the probability of a positive correction coefficient was 0.907 and the
probability of a significantly positive (r > 0.165) correlation was 0.369 (Table
4). With higher CV values in FE (0.08 and 0.10), the probability of a
significant relationship between CH4/CO2 and FE was high. Examples of the
simulation results representing average correlations (CV = 0.08) are shown in
Figure 11. The CH4/CO2 ratio was positively related to predicted CH4
prediction (Figure 11). The relationships were broadly similar to in vivo data
(Paper I) when the measurements were based on the flux method (R2 = 0.55
and 0.23 in experiment 1 and 2, respectively), indicating that default values and
variations used in the simulation were relevant. However, the relationships
between gas ratio and CH4 flux were weaker (R2 = 0.27 and 0.09) when the
CH4/CO2 ratio was measured with the sniffer method (Paper I). The results of
this simulation indicate that care should be exercised when applying
predictions of CH4 production based on determined CH4/CO2 ratio and
predicted CO2 production in breeding programmes, since it may favour cows
with low FE.

54
Table 4. The effects of the variability in the efficiency of ME utilisation to milk production on the
relationships between feed efficiency (FE) and residual of CH4 production (default observed –
predicted from gas ratio and CO2 production) and between gas ratio /CH4/CO2) and feed
efficiency (kg ECM/kg DMI)
CV of k
l
1
Relationship
Mean r
Min r
Max r
r2 > 0
P-value3 <0.05
0.06
FE vs. CH
4
Res
-0.751
-0.866
-0.549
1.000
1.000
CH
4/CO2 vs. FE 0.129 -0.265 0.498 0.907 0.369
0.08
FE vs. CH
4
Res
-0.828
-0.905
-0.699
1.000
1.000
CH
4/CO2 vs. FE 0.285 -0.042 0.600 0.996 0.898
0.10
FE vs. CH
4
Res
-0.875
-0.939
-0.753
1.000
1.000
CH
4/CO2 vs. FE 0.410 0.077 0.697 1.000 0.997
1kl = Efficiency of ME used for lactation.
2Proportion of correlation coefficient < 0 (FE vs. CH4 Residual) or >0 (Ratio vs. FE).
3Proportion of significant correlation coefficient (P<0.05) in 1000 simulations.

55
Figure 11. An example of simulation results for a 100 cow herd when CV of efficiency of ME utilisation was assumed to be 0.08. A) Methane (CH4) emissions
as a function of CH4/CO2 ratio, B) Feed efficiency as a function of CH4/CO2 ratio and C) Residual of CH4 emissions as a function of feed efficiency.

56
5.2 Animal-related factors
Evidence from several studies indicates that CH4 production in cattle is partly
under genetic control, and therefore it could be possible to decrease CH4
production through genetic selection for low-emitting animals (de Haas et al.,
2011; Pickering et al., 2015; Negussie et al., 2017). Indeed, studies conducted
in sheep (Pinares-Patiño et al., 2013) and Dutch dairy cows (de Haas et al.
2011) have shown medium heritability values for total CH4 production (0.29
and 0.35, respectively). In the long run, the success of strategies to mitigate
CH4 production in cattle will rely on how the mitigation target is defined and
the implications of the chosen variable in practice. Animal breeders have
defined four different CH4 phenotypes for genetic selection purposes that have
been widely used during the past decade (de Haas et al., 2017; Negussie et al.,
2017).
From a general perspective, total CH4 production (CH4, L/day, g/day or
MJ/day) is the clean trait that animal breeders want to improve in terms of CH4
mitigation strategies in ruminants (de Haas et al., 2017). However, as
mentioned before, total CH4 production is strongly positively correlated with
DMI or gross energy intake (Yan et al., 2000; Ramin and Huhtanen, 2013;
Hristov et al., 2013). One of the limitations of considering CH4 production as
an isolated mitigation target is that it is a poor indicator of the efficiency of
utilising dietary energy, and thus lacks economic value. In addition, reliable
measurement of CH4 production based on DMI by individual animals still
represents a major challenge in large-scale practical farming.
Efforts aimed at reducing CH4 production from ruminants should also
consider reducing CH4 production per unit of edible product. For meat-type
animals, CH4 intensity (CH4/unit of edible product, g/kg) is usually measured
in terms of kg of BW or carcass gain. In dairy cattle, by default, it is quantified
in terms of g CH4 per kg of milk, or preferentially per kg energy-corrected milk
(ECM). Methane intensity is mostly influenced by milk production level in
dairy cows and BW gain (gr/day) in growing animals. In addition, BW
influences CH4 intensity via CH4 produced from maintenance feed. At
individual animal level, this means that the total energy requirement per kg of
milk produced is reduced by dilution of the energy requirement for
maintenance and hence the cows are more efficient in feed conversion. In a
global perspective, CH4 intensity should be considered a target to mitigate
production considering that the demand for ruminant products (beef, milk) is
likely to increase in the future. This is the case especially in tropical countries,
where there is great potential for substantial reductions in CH4 production in

57
the near future by improving management and nutrition of the animals (FAO,
2016).
Methane yield (CH4/DMI, g/kg or proportion of GEI) describes the
arithmetic ratio between daily CH4 production (output) related to the DMI or
GEI (input) per animal. Therefore, this criterion is important in understanding
the mechanism of digestive physiology and rumen microbiology involved in
enteric CH4 production. While CH4 yield may better explain the biological
mechanisms involved in CH4 production among CH4 phenotypes, the use of
ratio traits has been criticised by animal breeders, as the genetic parameters
may not truly represent the trait under consideration, because there is always
extra variability of the denominator trait (Pickering et al., 2015). As an
alternative to overcome this issue, calculation of residual CH4 (observed minus
predicted CH4 production) has been suggested, based on its advantages in
terms of statistical properties (de Haas et al., 2017). In a large dataset (n =
1000) of dairy cows across Europe fed the same diet within-farm
(RuminOmics EU project), both high and low CH4 emitters were ranked
according to residual between observed and predicted CH4 production taking
into account the effects of DMI, BW and period (takes into account possible
variation in diet composition within herd). However, the use of residual CH4
on-farm conditions still remains unpractical because DMI cannot be
determined for individual animals.
Methane production expressed in terms of CH4 yield is probably the most
appropriate CH4 trait in order to understand biological mechanisms involved in
between-animal variation in CH4 production in experimental conditions.
However, difficulties in measuring DMI limit its use on-farm conditions. As
indicated before, total CH4 production is mainly driven by DMI and CH4
intensity by production level. Therefore, for the purposes of the present
discussion, animal factors influencing between-cow variation in CH4
production are addressed in terms of CH4 yield. This trait allows integration in
a more comprehensive manner of physiological mechanisms such as: rumen
fermentation, passage rate, digestibility and ruminal N metabolism.
5.2.1 Effects of rumen fermentation pattern
Methane can only be produced from available substrate for enteric
fermentation; in other words, its rate of production relies on the type of feed
ingested by the animal. The amounts of specific VFAs produced in the rumen
(i.e. acetate, propionate and butyrate) change depending on the diet. These
VFAs are the major determinant of the amount of H2 produced, and
consequently CH4 produced, in the rumen (Wolin, 1960; Czerkawski, 1986;
Van Soest, 1994). Hydrogen production is a thermodynamically unfavourable

58
reaction, but methanogens scavenge H2 and relieve this inhibition (Russell and
Rychlik, 2001). Equations based on stoichiometric principles (Czerkawski,
1986; Van Soest, 1994) demonstrate that acetate and butyrate production
promotes CH4 production, whereas high propionate production acts as an H2
sink and consequently reduces CH4 production (Johnson and Johnson, 1995;
Moss et al., 2000). In such a scenario, it is expected that increased proportion
of concentrates in the diet reduces CH4 production in cattle by promoting
propionate fermentation in the rumen. In feed-lot type diets fed to growing beef
cattle (>90% of concentrate in the diet on DM basis) substantial reductions in
CH4 production have been reported (Johnson and Johnson, 1995). However,
within the common ranges used in dairy cow diets the effect of concentrate
proportion on CH4 energy losses is marginal until 59% and only tends to
decrease at 70% of concentrate supplementation (Ferris et al., 1999). Methane
yield is 3% of GE intake of a grain ration and 6% of a roughage ration
(Johnson et al., 1995). Jonker et al. (2016) performed a study in sheep in order
to evaluate the effects of graded substitution of lucerne silage with maize silage
or maize grain in a diet fed at a fixed DMI level (2% of BW) on rumen
fermentation characteristics in both in vivo and in vitro conditions. A quadratic
effect was observed for both supplements on CH4 yield, to a maximum at 50%,
and thereafter it decreased more rapidly for the maize grain supplement.
Ruminal fermentation pattern was significantly related to CH4 yield with the
ratio of (acetate + butyrate) / (propionate + valerate) and the propionate
concentration alone being the best single predictor of CH4 yield (Jonker et al.,
2016).
In the meta-analysis approach by Ramin and Huhtanen (2013), mixed
regression equations were developed for predicting CH4 yield (CH4-E/GE;
kJ/MJ) in ruminants. Dry matter intake as a proportion of BW, OMD at
maintenance level and dietary fat concentration were the major factors
predicting CH4-E/GE. When rumen fermentation pattern was determined, the
effect of CH4VFA on CH4 yield was significant and it was the best predictor
among VFA measurements. Regression coefficient of CH4VFA on CH4 yield
was close to that expected from stoichiometric relationships.
In order to demonstrate the relationship between CH4VFA and CH4 yield,
primary data were taken from a study conducted by Kittelmann et al. (2014) on
118 low- and high-CH4 emitting sheep selected from a group of 340 animals
(Figure 12). The sheep were fed the same diet (2.2 times maintenance). Daily
CH4 production from individual animals was measured in respiration chambers
(2-3 days) and rumen fluid samples were taken by stomach tube before
feeding. To calculate stoichiometric CH4VFA from acetate to propionate ratio
(AP) from Kittelmann et al. (2014) dataset, the molar proportion of butyrate

59
was estimated using the relationship between AP and butyrate from the dataset
in Paper II. In that study, rumen fermentation pattern and CH4 yield in
respiration chambers (2-3 days) from individual animals were measured on two
occasions. There was a positive (P<0.001) relationship between CH4VFA and
CH4 yield (Figure 12). Two approaches were taken into account to calculate
stoichiometric CH4VFA (x-axis) as follows: ‘observed’ refers to relationship
based on reported VFA values and ‘predicted’ is based on expected CH4
production assuming only changes in the molar proportions of major VFAs,
but constant amount of fermentable substrate, without considering other animal
or dietary factors. For each unit increase in CH4VFA, CH4 yield increased by
0.053 and 0.037 g/kg DMI for ‘observed’ and ‘predicted’, respectively.
Figure 12. Relationship between stoichiometric CH4VFA (Wolin, 1960) and observed CH4 yield
in sheep (n = 118 animals) fed a standard lucerne pellet diet (19% CP, 43% NDF and 10 MJ
ME/kg DM) at 2.2 times the maintenance ME requirement (CSIRO, 2007). Calculated from
Kittelmann et al., (2014; supplementary data). Reproduced with the author’s permission.
Differences between observed and predicted CH4VFA responses in
observed CH4 yield can be related to improved diet digestibility, as the positive
relationship between these variables suggests (Paper II). Based on the
relatively low coefficient of determination (R2 = 0.16), although significant,
this analysis demonstrated that rumen fermentation pattern explains a relatively
small proportion of the variation in CH4 yield in sheep. Use of the Wolin

60
equation (Wolin, 1960) to calculate CH4VFA can be criticised mainly because
it assumes that all fermented substrates are expressed in terms of hexose
equivalents (C6H12O6), and due to the fact that it does not take into account
microbial cells as a H2 sink (Czerkawski, 1986). However, this approach, based
on stoichiometric principles, is still valid, since a major part of the substrate
available for fermentation comes from dietary carbohydrates, which in turn are
converted mostly to glucose units, and to the fact that acetate production is a
major factor for CH4 production in rumen fermentation conditions.
5.2.2 Variability and repeatability of volatile fatty acids in the rumen
In Paper II, low variation in rumen fermentation variables was observed when
accounting for differences between diets across changeover studies in dairy
cows. The CV for molar proportions of individual main VFAs ranged from
0.022 to 0.061. Among these, propionate and butyrate made similar
contributions to the total variation in the Diet(Exp) variance component (CV =
0.055 and 0.061, respectively). The between-cow variation observed in
CH4VFA derived from fermentation pattern (Paper II) was very low (CV =
0.010). The CV in CH4VFA calculated from Kittelmann et al., (2014) data was
higher (0.033), but still rather low to account for the variation in observed CH4
yield. However, differences in rumen fluid sample collection method could
have influenced the variation. In most cases in Paper II, the samples were
collected during a 12-h sampling period at 1.5 h intervals through a rumen
cannula in dairy cows, whereas in Kittelmann et al. (2014) rumen fluid was
taken once using a stomach tube in sheep. In that sheep study, the 118 animals
used for rumen fluid collection were intentionally selected as high and low
CH4 emitters from a larger group of animals (340), which could have
contributed to the higher CV values obtained compared with the dataset in
Paper II. Results from the variance components analysis performed in Paper
II basically confirmed findings obtained from analysis of data reported by
Kittelmann et al. (2014) and models proposed by Ramin and Huhtanen et al.
(2013). In Paper II, variance component analysis for major VFAs in terms of
concentrations (mmol/L) was not performed, but they are included in Table 5.
Overall, comparisons between individual VFAs (Table 5) demonstrated that
between-cow CV is a more important source of variation for VFA
concentrations than for molar proportions of VFA. As a consequence, VFA
concentrations display higher repeatability values. Among individual VFAs,
butyrate concentration was more repeatable (Rep = 0.55) than acetate or
propionate (Rep = 0.46 and 0.33, respectively). This implies that animal
physiology factors such as passage rate or VFA absorption through the rumen
wall may have a major impact on VFA concentrations, whereas molar

61
proportions of individual VFA reflect changes in fermentation pattern mainly
as a consequence of the type of diet ingested by the cow. The data in Paper II
suggested that that repeatability of VFA fermentation pattern decreased as a
function of time, which may indicate changes in microbial population over
time.
Table 5. Variance components and repeatability estimates for major volatile fatty acids (VFAs) in
rumen fermentation of dairy cows fed typical grass-silage based diets in the Nordic countries.
For further details, see Paper II
Variance component1
Diet
Cow
VFA
units
SD
CV
SD
CV
Rep
2
Acetate
mmol/L
3 2.4 0.032 3.4 0.046 0.46
mmol/mol
4 14.9 0.022 7.4 0.011 0.28
Propionate
mmol/L
1.2
0.058
1.3
0.062
0.33
mmol/mol
10.4 0.055 4.6 0.025 0.06
Butyrate
mmol/L
1.0
0.071
1.4
0.098
0.55
mmol/mol
7.9 0.061 6.6 0.051 0.23
1Diet(Exp) = diet within experiment; Cow(Exp) = cow within experiment respectively.
2Rep = Repeatability calculated as Rep = δ2 Cow / (δ2 Cow + δ2 Residual).
3Volatile fatty acids concentrations.
4Volatile fatty acids molar proportions.
Variation in VFA concentrations can be related to differences in
bicarbonate secretion, saliva production, short chain fatty acid absorption and
fluid passage rate out of the rumen. These factors are more likely related to
animal physiology than rumen microbiome and may be partly genetically
controlled. On the other hand, variations in VFA pattern are mainly related to
diet composition and to a smaller extent variations in rumen microbiome.
Variation in the physical structure and size of the rumen, as well as the
intensity of contractions and rate of passage of digesta are all expected to have
an influence on the rumen microbial community (Roehe et al., 2016).
Since microbial populations in the rumen act directly on the available
substrate by modulating the rates of VFA production, it could be expected that
the possible effects of the rumen microbiome on enteric CH4 production would
be also reflected in variation in the VFA fermentation pattern. However, the
relatively small CV calculated in both studies does not support a major
contribution of the rumen microbiome to between-animal CV in CH4 yield.
Different metabolic pathways in rumen fermentation pattern, e.g. acetogenesis,
would weaken the relationship between CH4VFA and CH4 yield. Although

62
induction of rumen acetogenesis was proposed by Van Nevel and Demeyer
(1995) as an interesting alternative to reduce CH4 production in ruminants, all
attempts to establish it have failed so far (Fievez et al., 1999). Because
quantitative relationships between CH4VFA and CH4 yield were close to the
theoretical potential both when analysed from treatment mean data (Ramin and
Huhtanen, 2013) and from individual animal data from Kittelmann et al.
(2014), any other major fermentation pathways are unlikely. Emissions of free
H2 can be an alternative fate of ruminal H2 production, but with normal diets its
contribution is likely to be minimal. Conversely, it could be high with
halogenated hydrocarbons, the increase in hydrogen production is generally of
a similar order of magnitude to the decrease in CH4 production as discussed by
Hristov et al. (2015a).
It seems that between-animal variability in CH4 production and rumen
fermentation pattern is greater for high concentrate diets than the estimated for
forage diets or mixed diets. In a study by Roehe et al. (2016) with Aberdeen
Angus and Limousine steers, average CV in total CH4 was 0.166 and 0.283 and
in CH4 yield 0.180 and 0.263 for low and high concentrate diets, respectively.
Similar differences between forage and concentrate diets were observed in a
study by Herd et al. (2016) (0.125 and 0.217 in total CH4 and 0.100 and 0.207
in CH4 yield, respectively). Consistently, analysis of primary data from the
study of Jaakkola and Huhtanen (1993) indicated increased between-animal
variability in CH4VFA with increased proportion of concentrate in the diet
(CV: 0.016, 0.020 and 0.096 with 25, 50 and 75% concentrates on DM basis,
respectively). In Paper II, it was not possible to estimate variance components
separately for the low and high concentrate diets (average 41% on DM basis),
but animal + residual variance was greater for high concentrate diets,
indicating greater between-cow variation with high than low concentrate diets
in rumen fermentation pattern.
Individual animal data from a study conducted with primiparous cows by
Zhu et al. (2014) showed that the estimated repeatability was rather low for the
main VFAs: acetate (0.03), propionate (0.18), and butyrate (0.03). The acetate:
propionate ratio had slightly higher repeatability compared to individual VFAs.
Moreover, a study by Robinson et al., (2010) showed repeatabilities in the
magnitude of 0.20 for VFA profiles in 708 sheep. In the same study,
phenotypic correlations between individual VFAs and short-term CH4
production (1 h) were relatively low, ranging from 0.15-0.20. An experiment
by Pinares-Patiño and Clark (2010) in lactating cows under grazing conditions
showed very small animal variation in rumen fluid osmolarity (CV < 0.05),
which may indicate little variation in rumen fermentation pattern.

63
It is important to note that in the studies by both Zhu et al. (2014) and
Robinson et al. (2010), the rumen fluid samples were collected by stomach
tubing, which could alter VFA concentrations due to saliva contamination
(more diluted samples) compared with the rumen fluid samples used in the
study by Pinares-Patiño and Clark (2010) and in Paper II, in which grab
samples were collected directly through the rumen cannula. Another
consideration is the effect of time of rumen fluid collection, which in addition
to diet composition also influences the VFA production rate (Bergman, 1990).
In Paper II, rumen fluid samples were taken on several occasions.
Nevertheless, despite the differences in animals and experimental conditions,
all the data found in different studies display a similar general trend, which
indicates small variation in rumen fermentation pattern.
Overall, observed CV and repeatability in rumen fermentation pattern are
much smaller than those in CH4 production, suggesting that variations in the
rumen microbiome are not likely be the major factor influencing between-
animal variation in CH4 production. Much smaller repeatability of CH4VFA
than CH4 yield (0.20 compared with 0.59) calculated from the primary data
published by Kittelmann et al. (2014) is in line with this suggestion.
Repeatability in CH4VFA is more ‘time-dependent’ than estimated
repeatability for VFA concentrations.
The effects of microbiome could derive indirectly from differences in the
physical structure and size of the rumen, while the intensity of contractions and
rate of passage of digesta can also be expected to have an influence on the
rumen microbial community, as suggested by Roehe et al. (2016). Therefore,
there is a limited room to select low CH4 emitters by taking into account only
rumen fermentation pattern due to the rather small repeatability values, which
in turn may compromise the accuracy of animal selection.
5.2.3 Passage rate and associated factors
In a study by Pinares-Patiño et al. (2003) on sheep, fractional passage rate
(reciprocal of mean retention time; MRT) of particulate matter was strongly
and negatively (R2 = 0.57) related to CH4 yield. More recently, Goopy et al.
(2014) selected 10 low and 10 high emitters from 170 ewes and found that,
over the measurement period, the difference between low and high groups in
CH4 yield was 2.7 g/kg DM intake. High emitters had 5.5 h longer particulate
mean retention time (MRT). Particulate and fluid MRT explained 56% and
69% of the variation in CH4 yield, respectively (Goopy et al. 2014). When
expressed per hour difference in MRT, CH4 yield (g/kg DM intake) declined
by 0.48 and 0.49 g/h in the study by Pinares-Patiño et al. (2011) and Goopy et

64
al. (2014), respectively. A simulation study by Huhtanen et al. (2016) also
showed a positive relationship between MRT and CH4 production.
Increased retention time of rumen digesta is related to higher reticulorumen
volume and consequently higher CH4 production per unit intake can be
expected. Higher OM rumen pool size was observed for high CH4 yield
emitters (7.4 L) compared with low CH4 yield emitters (5.9 L) in a study by
Goopy et al. (2014) conducted on ewes fed 1.2-fold the maintenance
requirement. In an earlier study by Pinares-Patiño et al. (2003) in sheep at a
fixed level of intake, a high and positive correlation between CH4 production
and rumen fill was observed (0.84). Robinson et al. (2010) suggested two main
mechanisms which may explain reduced VFA concentrations at larger rumen
volumes: i) with increased volume of rumen water VFA concentrations are
diluted, and ii) changes in the absorptive surface area could reduce VFA
absorption through the rumen wall.
It seems that the effects of MRT on CH4 yield are similar when variation in
MRT is related to increased feeding level of group of animals or between
individual animals fed the same level of intake. According to Yan et al. (2000),
proportion of CH4 energy decreased 0.78 %-units per multiple of maintenance,
whereas Ramin and Huhtanen (2013) reported 0.7 kJ/MJ per 1 g/kg BW
increase in DMI. Both estimates are close to a 10% reduction in the
maintenance requirement. It is also possible that physiological mechanisms are
involved in the relationship between MRT and CH4 yield when MRT is
affected by individual animal variation or by feeding level. According to the
Karoline model (Huhtanen et al., 2015c), reduced CH4 with increased intake is
associated with reduced digestibility, repartitioning of fermentation products
between gases and VFA compared with microbial cells and uptake of H2 by
microbes.
Link between passage rate and digestibility
The differences in CH4 yield between low- and high-emitting sheep have been
associated with digesta retention time (Pinares-Patiño et al., 2003, 2011;
Goopy et al., 2014), with diet digestibility being significantly (Pinares-Patiño
et al., 2011) or numerically (Goopy et al., 2014) lower in low-emitting sheep.
In the study by Goopy et al. (2014), for each kg increase in rumen particulate-
phase MRT, CH4 yield increased by 11.5 g/kg DMI in combined individual
animal data for low and high CH4 emitters (R2= 0.56). The modelling approach
by Huhtanen et al. (2016) predicted similar relationships between MRT and
OMD. Positive correlations between CH4 production and cellulose digestibility
have been reported in sheep (Pinares-Patiño et al., 2003), and positive

65
correlations between NDF digestibility and CH4 production in dairy cows
(Pinares-Patiño and Clark, 2010).
The simple equation proposed by Waldo (1970) for calculating digestibility
from digestion and passage rates [Digestibility = digestion rate / (digestion rate
+ passage rate)] indicates that with increased passage (reduced MRT),
digestibility decreases when digestion rate is constant. The effect of passage
rate on digestibility calculated by a biologically more correct two compartment
model considering selective retention of particles in the rumen also predicts
reduced digestibility with shorter retention time in the rumen (Allen and
Mertens, 1988).
Schiemann et al. (1971) presented individual data for eight cows fed either
at maintenance or production level. There was a strong positive relationship
between diet digestibility and CH4 yield in both cases (Figure 13). In that
respiration chamber study with dairy cows (Schiemann et al., 1971) and in the
modelling study by Huhtanen et al. (2016), the slope between digestibility and
CH4 yield was about three-fold the average CH4 yield, suggesting that
incremental digestion produced more CH4 per unit of digested OM than the
diet on average. It is possible that between-cow differences in OMD result
mainly from digestion of the slowly digestible NDF fraction, which can
produce more acetate and CH4. The positive relationships found for OMD and
NDF in relation to molar proportion of acetate, and the negative relationship
found for propionate (Paper II), support this suggestion.
Figure 13. Relationship between gross energy (GE) digestibility and CH4 yield at individual
animal level based on data from Schiemann et al. (1971).

66
In Paper II, improved OMD was positively associated with the molar
proportion of acetate and CH4VFA and negatively with the molar proportion of
propionate. Thus, the effects of OMD and CH4VFA are additive. Because most
of the variation in OMD was due to NDFD (cow variance for NDFD was
three-fold that for digestibility of ND solubles) and CH4VFA ratio was
positively related to dietary NDF concentration, incremental digestibility
increased CH4VFA.
Between-cow variation in OMD was small (SD = 10 g/kg; CV = 0.013).
Mehtiö et al. (2016) reported a value of 12.3 g/kg for between-cow variation in
OMD determined using acid-insoluble ash as an internal marker. Similarly,
small between-animal variation (CV = 0.012) was observed in a meta-analysis
of individual cow data from 21 studies using acid insoluble ash as a marker
(Huhtanen et al., 2015a). These values are consistent with the 0.016 in OMD
predicted by the Karoline model (Huhtanen et al., 2016) using the same
between-animal CV (0.085) as the dataset in Paper II. Based on Ørskov et al.
(1988) and the results described above, between-animal variation in
digestibility is strongly influenced by MRT.
In general, between-cow variation in digestibility can explain only a small
part of the observed variation in CH4 production. Similarly, analysis of the data
from respiration chamber studies (Yan et al., 2000, 2010; Ramin and
Huhtanen, 2013) indicates that CH4 yield decreases by about 10% per multiple
of maintenance increase in feeding level. This effect is much greater than
observed decreases of approximately 2-3% in OMD per multiple of
maintenance (Yan et al., 2002; Huhtanen et al., 2009).
Link between passage rate and efficiency of microbial protein synthesis
As discussed before, between-animal variation in VFA pattern and digestibility
cannot explain observed effects of between-animal variation and feeding level
on CH4 yield. One possible mechanism can be improved efficiency of
microbial synthesis in the rumen associated with increased passage rate.
Because bacteria pass with digesta, their growth rate increases with increasing
digesta passage rate in the rumen. Increasing passage rate by nutritional
manipulation could be one strategy to decrease the relative impact of
maintenance energy and improve growth efficiency (Hackman and Firkins,
2015). The relationship between passage rate and microbial efficiency can be
demonstrated from the positive relationship between feed intake and microbial
efficiency. It is well-known that increased feed intake would increase ruminal
passage rate (e.g. NRC, 2001) and reduce microbial retention time, and thus
increase microbial cell yield per unit of energy fermentation by diluting
maintenance expenditure (Russell et al., 1992). Several studies have shown

67
that the efficiency of microbial N synthesis is positively related to feed intake
(Chen et al., 1992; Volden, 1999; Broderick et al., 2010). The relationship
between passage rate and the efficiency of microbial N synthesis could be
expected to be similar when the differences in passage rate derive from
differences in feeding level or from between-animal differences.
With improved efficiency of microbial synthesis, more fermented carbon is
partitioned to microbial cells instead of VFA and fermentation gases. In
addition, microbial cells are more reduced than fermented carbohydrates
(Czerkawski, 1986; Van Soest, 1994) and act as a H2 sink. According to
Czerkawski (1986), at microbial hydrogen uptake of 8.1 g/kg cells:
Production of hydrogen, mol 2A + P + 4B +3V + L
Utilization of hydrogen, mol 2P + 2B + 4V + L + 4CH4 + 8.1 (kg cell
DM),
where A. P, B, V, L and CH4 are the amounts of acetic, propionic, butyric,
valeric and lactic acid and CH4 (mol) produced, respectively.
Methane production would clearly have been overestimated in Paper II if
microbial uptake of H2 had not been included in stoichiometric predictions.
Applying the equations of Czerkawski (1986), CH4 yield per mol of VFA was
0.25-0.26, which is considerably lower than the 0.312 calculated by Wolin
(1960) equations for VFA ratio in the example. In in vitro studies, the recovery
rate of metabolic H2 varies between 78 and 96 % (Demeyer, 1991).
Considering a mean H2 recovery of 90%, CH4 production should be 10% lower
than the stoichiometric fermentation equation suggests (Moss et al., 2000). In
Paper II, microbial N efficiency was negatively relatively related to OMD,
indicating that the effects of these variables on CH4 yield were additive. The
positive relationship between rumen ammonia N concentration and OMD is
consistent with this. Variation in rumen ammonia N concentration in animals
fed the same diet reflects differences in the balance between microbial
synthesis and protein degradation.
Between-cow CV in parameters related to passage kinetics (0.077-0.090)
and rumen digesta pools (0.130) was much greater than in parameters related to
rumen fermentation pattern or digestibility. Other studies have also indicated
similar or higher between-animal variation in passage rate or retention. In the
study by Pinares-Patiño and Clark (2010), the CV of mean retention time
determined using particle marker was 0.209, whereas CV of rumen evacuation
derived from lignin passage rate was 0.14. Between-cow CV of the passage
rate of chromium-mordanted straw was 0.10-0.11 in a study by Ørskov et al.
(1988). Variables related to ruminal N metabolism also showed high between-

68
animal variation (ammonia N concentration 0.149, microbial N flow 0.078,
efficiency of microbial N synthesis 0.078). In studies by Pinares-Patiño et al.
(2003) and Pinares-Patiño and Clark (2010), between-animal CV in microbial
N flow was 0.209 and 0.179, respectively, and between-animal CV in
microbial N efficiency was 0.166 and 0.180, respectively. In both studies,
microbial N was estimated from urinary purine derivative excretion.
In Paper II, passage rate of iNDF was moderately repeatable (0.38).
Earlier studies with sheep (Faichney, 1993) and cattle (Ørskov et al., 1988)
found that the ranking of animals on the basis of rumen fractional passage rate
was consistent among diets and feeding levels. Smuts et al. (1995) reported
that rumen digesta retention time is a repeatable physiological trait (Rep =
0.45). Other physiological traits that can reflect variation in passage rate, such
as OMD (rep = 0.365), microbial N flow (rep = 0.509), microbial N efficiency
(rep = 0.354) had moderate repeatability. It can be concluded that both
variability and repeatability are greater for physiological animal-related factors
such as passage rate and rumen pool size than for rumen fermentation pattern,
which is more related to microbial ecology in the rumen.
In the modelling approach by Huhtanen et al. (2016), the coefficient of
variation of predicted CH4 yield was 0.052 for cows and 0.045 for sheep (DMI
= 20 and 1 kg/day respectively)when the simulations were made using the
variation in MRT of iNDF. Variation in model predictions was smaller than
observed CV in animal studies. This may be because i) fixed intake was used
in the modelling approach and ii) possible variation in rumen fermentation
pattern was not taken into account. In addition, iii) random measurement errors
increase observed between-animal CV of CH4 production, especially when
measurements are based on single observations from one animal. Differences
in predicted microbial efficiency were the main contributor to variation in CH4
production. A proportion of between-animal variation can be related to
digestive processes. As a consequence, selecting animals for low CH4
production could lead to selection of low digester animals. Finally, a summary
of the effects of animal-related factors on selecting low and high CH4 emitters
is presented in Figure 14. The contribution of isolated factors and mechanisms
was discussed above.
5.3 Dietary factors
Mitigation of greenhouse gas emissions in cattle production systems should
focus on reducing CH4 production per unit of edible product. Methane
intensity, measured as the ratio CH4/ECM, is by default the most important
trait in dairy production, which requires further research. Forage quality and

69
the level of concentrate supplementation are the most important practical tools
available on dairy farms to optimise production. Because these factors
represent a high proportion of total diet, they can also have a critical impact on
CH4 production.
Figure 14. Summary of animal-related factors influencing CH4 yield as a criteria to identify both
low and high CH4 emitters. (Modified from slide of Dr Sidney Leahy presented at METHAGENE
Training School on Rumen Microbial Ecosystem. University of Porto, Porto, Portugal. September
11 – 14, 2016, (https://twitter.com/METHAGENE?ref_src=twsrc%5Etfw). Reticulorumen figures
were taken from www.scanvetpress.com (accessed 1 October, 2016; Copyright © (2016).
Dietary carbohydrate composition affects digestion site, fermentation
pattern and digestibility of different nutrients. In practice, it is determined by
the forage to concentrate ratio, forage type, forage maturity and concentrate
source. The results in this thesis (Papers III and IV) demonstrated that it was
possible to reduce the amount of concentrate supplementation by early
harvesting to improve forage quality, without compromising the performance
or increasing CH4 production or N excretion per kg ECM, and even improving
feed efficiency.
Although the reported effects of forage maturity at harvest (digestibility) are
variable (Thomas, 1987; Kuoppala et al., 2008), improved digestibility
increases DMI (Huhtanen et al., 2007), and consequently ME intake at a fixed

70
level of concentrate. Therefore it is likely to increase nutrient intake, with
improved forage digestibility and increased production and decreases in CH4
production per kg of ECM. Paper IV showed reduced rumen NDF pool size
with increased inclusion of early-cut silage in the diet. This is an indication that
rumen fill was not a limiting factor in intake with higher inclusion of better
quality forage in the diet. Rumen NDF pool size decreases with increasing
digestibility (Bosch et al., 1992; Rinne et al., 2002; Paper IV), suggesting that
this strategy can work even at higher production levels.
Increased concentrate supplementation is often considered as an effective
strategy to reduce CH4 production. With high grain diets in a feedlot situation,
CH4 losses may drop to approximately 3% of gross energy (Johnson et al.,
1993), which is much lower than the 6-7% of gross energy reported for typical
grass silage-based diets for dairy cows (e.g. Yan et al., 2000). However, within
typical ranges of concentrate supplementation for dairy cows, the effects are
relatively small (Sauvant and Giger-Reverdin, 2009; Ramin and Huhtanen,
2013). For example, in the study by Ferris et al. (1999), CH4-E/GE only tended
to decrease when the proportion of concentrate gradually increased from 37 to
70%. Although the effects of concentrate level on CH4 yield are not very large
for dairy cow diets, increased feed intake and production will most likely
decrease CH4 production per unit of product.
When using high concentrate diets for dairy cows, it is important to be
aware that some concentrate ingredients such as cereal grains and soybean can
be used directly as human food or more efficiently in monogastric animals with
minimal CH4 production. With high concentrate diets, the special advantage of
ruminants in human food production – microbial digestion of fibre in the
rumen – is also partly or completely neglected. The results in Papers III and
IV indicated that with increased concentrate proportion in the diet and reduced
forage quality, more potentially digestible nutrients were excreted in faeces. It
is possible that the greater faecal output of fermentable substrate with increased
concentrate at least partly compensates for the lower CH4 yield from rumen
fermentation. In addition, the carbon footprint of feed production should be
taken into account when comparing different nutritional mitigations strategies.
According to Mogensen et al. (2014), carbon footprint is 1065 and 671 g CO2-
eq/kg DM for barley grain and grass silage, respectively, in Danish conditions.
This difference corresponds to 14 g CH4 (1 g CH4 = 28 g CO2) when replacing
1 kg DM of grass silage with barley. It is also important to consider the
contribution from soil carbon storage or loss potential from different land uses
and manure systems when identifying appropriate strategies for reducing
greenhouse gas emissions from dairy production.

71
6 Conclusions
 The GreenFeed system (flux method) was shown to be a useful tool
for measuring CH4 emissions from large numbers of animals in on-
farm conditions. Repeatability was high, while between-animal
variation and measured emissions were within expected biological
ranges.
 Methane emissions measured by the sniffer method were poorly
correlated to CH4 flux measured by the GreenFeed system. Head
position had a strong influence on measured CH4 values. A sniffer
method based on CH4/CO2 ratio was better correlated to CH4 flux than
CH4 concentration.
 Repeatability and between-cow variation in stoichiometric CH4
production per mole of volatile fatty acids (VFA) were small and can
only make a minor contribution to observed between-cow variation in
CH4 emissions. Variation and repeatability were greater for ruminal
VFA concentrations than molar proportions.
 Between-cow variability in digestibility was small, but repeatability
was moderate.
 Greater between-cow variability and repeatability was observed in
digesta passage rate and rumen pool size variables.
 Between-cow variation in digesta passage rate-associated variables
can explain more of the between-cow variation in CH4 emissions than
rumen fermentation patterns associated with differences in rumen
microbial population.
 Decreased CH4 emissions with increased digesta passage rate are
related to reduced diet digestibility, improved efficiency of microbial
protein synthesis, which repartitions fermented carbon from VFAs and
gases to microbial cells, and uptake of hydrogen by microbial cells.
Selection for low CH4 emissions can decrease the efficiency of cell
wall digestion.
 By improving forage digestibility, the amount of concentrate
supplementation could be reduced and milk production level could be

72
maintained, without increasing CH4 emissions or nitrogen excretion
per unit of product.
 The depression in digestibility from maintenance level to production
level was greater for diets based on medium-quality silage and a high
level of concentrate than for diets based on high-quality silage and a
moderate level of concentrate supplementation. The difference was
mainly due to lower digestibility of potentially digestible neutral
detergent fibre (pdNDF).
 Higher CH4 yield with increased proportion of early-harvested silage
was not related to rumen fermentation pattern. The differences were
mainly related to higher total digestibility of organic matter and
especially to higher apparent organic matter digestibility in the rumen.
 Feed efficiency in terms of ECM yield per kg dry matter intake
improved with increased inclusion of early-harvested silage in the diet.
No difference in the efficiency of nitrogen utilisation was observed.
 Ruminal and total tract NDF digestibility improved with increased
inclusion of early-harvested silage in the diet, reflecting differences in
intrinsic characteristics of fibre and negative effects of higher starch
content in diets with increased proportion of late-harvested silage.

73
7 Future perspectives
Based on the results obtained in this thesis, future studies focusing on the study
of between-animal variation in CH4 emissions should consider:
 Investigate the potential for ranking cows as high and low CH4
emitters and for establishing links with their productive performance
(i.e. dry matter intake, milk yield), their ability to digest fibre, rumen
microbial ecology and fermentation characteristics etc.
 Quantify the real effect of animal variation on methane emissions,
measure the repeatability of specific animal characteristics as a tool
for animal breeders, identify biomarkers from low CH4 emitters (i.e.
fatty acids in rumen bacteria) and then suggest protocols for future
research and develop useful CH4 mitigation strategies for dairy cows.
 Examine why rumen fermentation pattern is not enough to explain
individual differences in CH4 emissions.
 Compare digestibility and microbial community, e.g. when selecting
for low emitters then also select for low digesters.
 In selection of low methane emitters, determine the relationships
between digestibility and CH4 emissions.
 Study whether increasing feed conversion is a more effective way to
reduce CH4 emissions than selecting for low CH4 emitters.

74

75
8 Popular scientific abstract
Climate change is kind of every day’s issue that contemporary society has
to deal with. The growing human population represents a constant threat to
the ecosystems since it demands increased amount of food and natural
resources to supply its demands. Greenhouse gas (GHG) emissions to the
atmosphere is just one example as a consequence of livestock production.
Due to the particularities of digestive tract of ruminants, they are able to
convert non-edible foods (i.e. forages) into highly valuable products for
human consumption such as milk. However, ruminants also produce
methane (CH4) which contributes significantly to global warming. Several
attempts to account for CH4 emissions around the world and across
different production systems have been made but many of them fail in get
realistic numbers, especially at large farm scale.
Accurate and reliable methods for measuring CH4 emissions in dairy cows
at individual-animal basis are needed in order to develop successful CH4
mitigation targets in cattle production. The first study of the present thesis
demonstrated GreenFeed is a reliable tool for ranking animals as low CH4
emitters in farm conditions. Gas concentrations from sniffer method were
poorly correlated to dry matter intake (DMI) and it lacks of biological value. In
addition to be repeatable, GreenFeed proved to be an accurate method to
measure CH4 emissions despite of being based on spot sampling.
Interest on selecting low emitters as a long term strategy to mitigate
methane CH4 emissions from ruminants has increased significantly mainly due
to promising heritability values. Traditional selection for total CH4 emissions
may have an impact on selecting efficient animals since the selection of low
CH4 emitters could lead to select for low fibre digesters which is not very
convenient for farmers’ income. A meta-analysis of experiments conducted in
the Nordic countries was performed to investigate the effects of animal-related
factors on the variation in in vivo CH4 yield emissions. Results from study 2,

76
showed that among of the studied animal-related factors, passage rate is the
key variable in modulating CH4 yield emissions since their contribution to the
observed between-cow variation was much higher than digestibility, microbial
N synthesis or rumen fermentation patterns. Since passage rate is positive and
strongly correlated to DMI, it may be a strong evidence to support selecting
individual animals for feed efficiency rather than selection for low CH4
emissions.
In the Nordic countries, diets for dairy cows are based on grass silage
forages. By harvesting in an early stage the forage, it is expected that forage
quality improved compared to late cut harvest. Combined results from two
experiments (studies 3 and 4), conducted at the same time in either 16 intact
cows (production trial) or 4 rumen-cannulated cows (flow study), demonstrated
that by the graded addition of early-cut silage in the diet is possible to reduce
concentrate supplementation in practical diets without compromising milk
production or total CH4 emissions. Differences in forage digestibility partly
explained the differences between treatments. Although, medium quality
forage (late-cut silage) and increased amount of concentrate was able to reduce
CH4 yield, higher faecal output of potential digestible nutrients was observed
in these treatments compared with early-cut silage diets. This compromises
give a final recommendation on a life-cycle assessment approach since
potential nutrients are wasted to the environment and its CH4 production was
not measured in those studies. Future GHG mitigation strategies have to
consider all nutrient outputs of the system to better understanding the real
impact of ruminants on the environment.

77
9 Populärvetenskaplig sammanfattning
Exakta och tillförlitliga metoder för att mäta produktionen av CH4 från
mjölkkor på individnivå behövs för att utveckla realistiska mål för
begränsningar av utsläppen av växthusgaser från nötkreaturssektorn. Den första
studien i denna avhandling visar att GreenFeed är ett pålitligt verktyg för att
rangordna djuren som hög- eller lågemitterande kor och utan att behöva flytta
djuret från sin naturliga miljö under mätningarna. GreenFeed har utvecklats för
att i realtid mäta flöden av CO2 och CH4 från en större grupp djur i en
besättning genom upprepade regelmässiga individbaserade mätningar under
flera dagar. Att bara mäta gaskoncentrationer med den så kallade Sniffer
metoden visade sig vara dåligt korrelerat till kornas konsumtion.
Konsumtionen är den viktigaste faktorn som bestämmer den totala
produktionen av CH4 hos en mjölkko. Sniffer metoden bedöms därför vara
missvisande och sakna biologisk relevans för att mäta kornas metanproduktion.
På grund av den individuella variationen borde det kunna vara möjligt att
välja ut kor som ger lägre metanproduktion som en långsiktig strategi för att
minska klimatpåverkan från idisslare. Att ensidigt selektera för en målsättning
kan dock påverka andra funktionella egenskaper, och just metanproduktionen
har visat sig ha ett nära samband med kornas förmåga att smälta växtfibrer.
Den här doktorgradsavhandlingen undersökte också hur andra funktionella
egenskaper relaterade till mjölkkornas matsmältning inverkade på
metanproduktionen. I en metaanalys där flera försök med mjölkkor, som
genomförts i de nordiska länderna, användes för att undersöka vilka
djurrelaterade faktorer som har störst inverka på metanproduktionen visade sig
passagen av fodret ut ur våmmen vara mycket viktigare än fodrets smältbarhet,
den mikrobiella proteinsyntesen eller jäsningsmönstret av flyktiga fettsyror i
våmmen. Eftersom fodrets passagehastighet är positivt korrelerat till
foderkonsumtionen innebär det att genom att hellre avla för bättre
foderutnyttjande än låg metanproduktion per se har man större förutsättningar

78
för att minska metanproduktion per kg mjölk än om man gör ett direkt urval av
kor med låg produktion av CH4. Detta eftersom det senare kan leda till nedsatt
smältbarhet och fodereffektivitet.
I de nordiska länderna baserar sig utfodringen av mjölkkor i hög grad på
gräsensilage. Genom att skörda fodret tidigt i säsongen uppnår man ett
högkvalitativt vallfoder. Detta innebär att man kan spara kraftfoder i
produktionen jämfört med att utfodra ett senare skördat vallfoder i foderstaten.
I två olika försök i den här doktorgradsavhandlingen undersöktes effekten av
att gradvis ersätta ett sent skördat gräsensilage och kraftfoder med ett energirikt
och tidigt skördat gräsensilage på mjölk- och metanproduktionen samt mer
detaljerat med avseende på fodrets omsättning i korna. Produktionsförsöket
visade att det är möjligt att ersätta kraftfoder med ett tidigt skördat gräsensilage
i foderstaten till mjölkkor och upprätthålla produktionen utan att öka
metanproduktionen hos korna. Även om det senare skördade ensilaget och
mera kraftfoder minskade metanproduktionen relaterat till konsumtionen
utsöndrades mer näringsämnen i kornas träck än när tidigt skördat gräsensilage
utfodrades. Slutgiltigt poängterar detta vikten av en helhetlig betraktning av
nötkreaturssektorns klimatpåverkan för att minska metanproduktionen från
idisslarna i framtiden.

79
References
Ahvenjarvi, S., A. Vanhatalo, P. Huhtanen, and T. Varvikko. (2000). Determination of reticulo-
rumen and whole-stomach digestion in lactating cows by omasal canal or duodenal sampling.
British Journal of Nutrition 83, 67-77.
Allen, M.S. and Mertens D.R. (1988). Evaluating constraints on fiber digestion by rumen
microbes. Journal of Nutrition 118, 261-270.
Appuhamy, J.A.D., Strahe, A.B., Jayasundara, S., Wagner-Riddle, C., Dijkstra, J., France, J. and
Kebreab, E. (2013). Anti-methanogenic effects of monensin in dairy and beef cattle: A meta-
analysis. Journal of Dairy Science 96, 5161-5173.
Armentano, L.E. and Russell, R.W. (1985). Method for calculating digesta flow and apparent
absorption of nutrients from non-representative samples of digesta. Journal of Dairy Science
68, 3067-3070.
Aronson, E.L., Allison, S.D. and Helliker, B.R. (2013). Environmental impacts on the diversity of
methane cycling microbes and their resultant function. Frontiers in Microbiology 225, 48-56.
Arthur, P.F., Barchia, I.M., Weber, C., Bird-Gardiner, T., Donogue, K.A., Herd, R.M. and R.S.
Hegarty. (2017). Optimizing test procedures for estimating daily methane and carbon dioxide
emissions in cattle using short-term breath measures. Journal of Animal Science 95, 645-656
Beauchemin, K.A., Kreuzer, M., O'Mara, F. and McAllister, T.A. (2008). Nutritional
management for enteric methane abatement: a review. Australian Journal of Experimental
Agriculture, 48, 21-27.
Beauchemin, K.A., Coates, T., Farr, B. and McGinn, S.M. (2012). Technical Note: Can the sulfur
hexafluoride tracer gas technique be used to accurately measure enteric methane production
from ruminally cannulated cattle? Journal of Animal Science 90, 2727-2732.
Bell, M.J., Potterton, S.L., Craigon, J., Saunders, N., Wilcox, R.H., Hunter, M., Goodman, J.R.
and P.C. Garnsworthy. (2014). Variation in enteric methane emissions among cows on
commercial dairy farms. Animal, 8:9, 1540-1546.
Benchaar, C. and Greathead, H. (2011). Essential oils and opportunities to mitigate enteric
methane emissions from ruminants. Animal Feed Science and Technology 166-167, 338-355.
Bergman, E.N. (1990). Energy contributions of volatile fatty acids from the gastrointestinal tract
in various species. Physiological Reviews 70, 567-590.
Berndt, A., Boland, T.M., Deighton, M.H., Gere, J.I., Grainger, C., Hegarty, R.S. Iwaasa, A.D.,
Koolaard, J.P., Lassey, K.R., Luo, D., Martin, R.J. Martin, C., Moate, P.J., Molano, G.,

80
Pinares- Patiño, C., Ribaux, B.E. Swainson, N.M., Waghorn, G.C. and Williams, S.R.O.
(2014). Guidelines for use the sulphur hexafluoride (SF6) tracer technique to measure enteric
methane emissions from ruminants. Pages 166. M.G. Lambert, ed. New Zealand Agricultural
Greenhouse Research Centre, New Zealand.
Blaxter, K.L. and Clapperton, J.L. (1965). Prediction of the amount of methane produced by
ruminants. British Journal of Nutrition 19, 511-522.
Bodas, R., López, S., Fernández, M., Garcia-González, R., Rodríguez, A.B., Wallace, R.J. and
González, J.S. (2008). In vitro screening of the potential of numerous plant species as
antimethanogenic feed additives for ruminants. Animal Feed Science and Technology 145,
245-258.
Bosch, M.W., Lammers-Wienhoven, S.C.W., Bangma, G.A., Boer, H., van Adrichem, P.W.M.
(1992). Influence of stage of maturity of grass silages on digestion processes in dairy cows. 2.
Rumen contents, passage rates, distribution of rumen and faecal particles and mastication
activity. Livestock Production Science 32, 265-281.
Broderick, G.A., Huhtanen, P., Ahvenjärvi, S., Reynal, S.M. and Shingfield, K.J. (2010).
Quantifying ruminal nitrogen metabolism using the omasal sampling technique in cattle - A
meta-analysis. Journal of Dairy Science 93, 3216-3230.
Broderick, G.A. and Merchen, N.R. (1992). Markers for quantifying microbial protein synthesis
in the rumen. Journal of Dairy Science 75, 2618-2632.
Buddle, B.M., Denis, M., Attwood, G.T., Altermann, E., Janssen, P.H., Ronimus, R.S., Pinares-
Patino, C.S., Muetzel, S. and Wedlock, N. (2011). Strategies to reduce methane emissions
from farmed ruminants grazing on pasture. The Veterinary Journal 188, 11-17.
Calsamiglia, S., Busquet, M., Cardozo, P.W., Castillejos, L. and Ferret, A. (2008). Essential oils
as modifiers of rumen microbial fermentation. Journal of Dairy Science 90, 2580-2595.
Capper, J.L., Caddy, R.A. and Bauman, D.E. (2009). The environmental impact of dairy
production: 1944 compared with 2007. Journal of Animal Science 87, 2160-2167.
Chagunda. M.G.G. (2013). Opportunities and challenges in the use of the Laser Methane detector
to monitor enteric methane emissions from ruminants. Animal 7, 394-400.
Chagunda, M.G.G. and Yan, T. (2011). Do methane measurements from a laser detector and an
indirect open-circuit respiration calorimetric chamber agree sufficiently closely? Animal Feed
Science and Technology 165, 8–14
Chen, X. B., Chen, Y.K., Franklin, M.F. and Ørskov, E.R. (1992). The effect of feed intake and
body weight on purine derivative excretion and microbial protein supply in sheep. Journal of
Animal Science 70, 1534-1542.
CSIRO (2007). Nutrient Requeriments of Domesticated Ruminants. CSIRO Publishing,
Collingwood VIC, Australia.
Czerkawski, J.W. 1986. An Introduction to Rumen Studies. Robert Maxwell, M. C., Oxford, UK.
de Haas, Y., Pszczola, M., Soyeur, H., Wall, E. and Lassen, J. (2017). Invited review: Phenotypes
to genetically reduce greenhouse gas emissions in dairying. Journal of Dairy Science 100,
855-870.
de Haas, Y., J. W. van Riel, R. F. Veerkamp, W. Liansun, and N. Ogink. (2013). On-farm
methane measurements in exhaled air of individual Dutch cows obtained during milking using
Fourier transformed infrared methods. Advances in Animal Biosciences 4, 391.

81
de Haas, Y., Windig, J.J., Calus, M.P.L., Dijkstra, J., de Haan, M., Bannink, A. and Veerkamp,
R.F. (2011). Genetic parameters for predicted methane production and potential for reducing
enteric emissions through genomic selection. Journal of Dairy Science 94, 6122-6134.
Deighton, M.H., O’Loughlin, B.M., Williams, S.R.O., Moate, P.J., Kennedy, E., Boland, T.M.,
Eckard, R.J. (2013). Declining sulphur hexafluoride permeability of polytetrafluoroethylene
membranes causes overestimation of calculated ruminant methane emissions using the tracer
technique. Animal Feed Science and Technology 183, 86-95
Demeyer, D.I. (1991). Quantitative aspects of microbial metabolism in the rumen and hindgut.
Pages 217-237. In J. –P. Jouany (ed.). Rumen microbial metabolism and ruminant digestion.
INRA Editions, Paris.
Dorich, C.D., Varner, R.K., Pereira, A.B.D., Martineau, R., Soder, K.J. and Brito, A.F. (2015).
Short communication: Use of a portable, automated, open-circuit gas quantification system
and the sulfur hexafluoride tracer technique for measuring enteric methane emissions in
Holstein cows fed ad libitum or restricted. Journal of Dairy Science 98, 2676-2681.
Ellis, J.L., Kebreab, E., Odongo, N.E., McBride, B.W., Okine, E.K., and France, J. (2007).
Prediction of methane production from dairy and beef cattle. Journal of Dairy Science 90,
3456-3466.
Ellis, W.C., Matis, J.H., Hill, T.M. and Murphy, M.R. (1994). Methodology for estimating
digestion and passage kinetics of forages. Pages 682–756 in Forage Quality, Evaluation, and
Utilization. G. C. Fahey, Jr., M. Collins, D. R. Mertens, and L. E. Moser, ed. Am. Soc.
Agron., Madison, WI.
Eugène, M., Archimede, H. and Sauvant, D. (2004). Quantitative meta-analysis on the effects of
defaunation of the rumen on growth, intake and digestion in ruminants. Livestock Production
Science 85, 81-97.
FAO. (2016). The future of food and agriculture – Trends and challenges. Rome. Available:
http://www.fao.org/3/a-i6583e.pdf
FAO. (2011). World Livestock 2011 – Livestock in food security, FAO, Rome.
Ferris, C.P., Gordon, F.J., Patterson, D.C., Porter, M.G. and Yan, T. (1999). The effect of genetic
merit and concentrate proportion in the diet on nutrient utilization by lactating dairy cows.
Journal of Agricultural Sciences 132, 483-490.
Fievez, V., Piattoni, F., Mbanzamihigo, L. and Demeyer, D. (1999). Evidence for reductive
acetogenesis and its nutritional significance in ostrich hindgut as estimated from in
vitro incubations. Journal of Applied Animal Research 16, 1-22.
France, J. and Siddons, R.C. (1986). Determination of digesta flow by continuous marker
infusion. Journal of Theoretical Biology 121, 105-120.
Garnsworthy, P.C., Craigon, J., Hernandez-Medrano, J.H. and N. Saunders. (2012a). On-farm
methane measurements during milking correlate with total methane production by individual
dairy cows. Journal of Dairy Science 95, 3166-3180.
Garnsworthy, P.C., Craigon, J., Hernandez-Medrano, J.H. and N. Saunders. (2012b). Variation
among individual dairy cows in methane measurements made on farm during milking.
Journal of Dairy Science 95, 3181-3189.
Gerber, P.J., Steinfeld, H., Henderson, B., Mottet, A., Opio, C., Dijkman, J., Falcucci, A. and
Tempio, G. (2013). Tackling climate change through livestock: a global assessment of

82
emissions and mitigation opportunities. Food and Agriculture Organization of the United
Nations (FAO), Rome.
Gill, M., Smith, P. and Wilkinson, J.M. (2010). Mitigating climate change: the role of domestic
livestock. Animal 4(3), 323-333.
Godfray, H.C.J., Beddington, J.R., Crute, I.R., Haddad, L., Lawrence, D., Muir, J.F., Pretty, J.,
Robinson, S., Thomas, S.M. and Toulmin, C. (2010). Food security: The challenge of feeding
9 billion people. Science, 327, 812-818.
Goel, G. and Makkar, H.P.S. (2012). Methane mitigation from ruminants using tannins and
saponins. Tropical Animal Health and Production, 44,729-739.
Gould, D.H. (1998). Poliencephalomalacia. Journal of Animal Science 76, 309-314.
Goopy J. P., Donaldson, A., Hegarty, R., Vercoe, P.E., Haynes, F., Barnett, M. and Oddy, V.H.
(2014). Low-methane yield sheep have smaller rumens and shorter rumen retention time.
British Journal of Nutrition 111, 578-585.
Grainger, C., Clarke, T., McGinn S.M., Auldist, M.J., Beauchemin, K.A., Hannah, M.C.,
Waghorn, G.C., Clark, H. and Eckard, R.J. (2007). Methane emissions from dairy cows
measured using the sulfur hexafluoride (SF6) tracer and chamber techniques. Journal of Dairy
Science 90, 2755-2766
Gidlund, H., Hetta, M. and Huhtanen, P. (2017). Milk production and methane emissions from
dairy cows fed a low or high proportion of red clover silage and an incremental level of
rapeseed expeller. Livestock Science 197, 73-81.
Gidlund, H., Hetta, M., Krizsan, S.J., Lemosquet, S. and Huhtanen, P. (2015). Effects of soybean
meal or canola meal on milk production and methane emissions in lactating dairy cows fed
grass silage-based diets. Journal of Dairy Science 98, 8093-8106.
Guan L., Nkrumah, D.J., Basarab, J.A. and Moore, S.S. (2008). Linkage of microbial ecology to
phenotype: correlation of rumen microbial ecology to cattle's feed efficiency. FEMS
Microbiology Letters 288, 85-91.
Hackmann, T.J. and Firkins, J.L. (2015). Maximizing efficiency of rumen microbial protein
production. Frontiers in Microbiology 6, 465. https://doi.org/10.3389/fmicb.2015.00465
Hammond, K.J., Waghorn, G.C. and Hegarty, R.S. (2016a) The GreenFeed system for
measurement of enteric methane emission from cattle. Animal Production Science 56, 181-
189.
Hammond, K.J., Jones, A.K., Humphries, D.J., Crompton, L.A. and Reynolds, C.K. (2016b).
Effects of diet forage source and neutral detergent fiber content on milk production of dairy
cattle and methane emissions determined using GreenFeed and respiration chamber
techniques. Journal of Dairy Science 99, 7904-7917.
Hammond, K.J., Humphries, D.J., Crompton, L.A., Green, C. and 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. Animal Feed Science and Technology 203, 41-52.
Haque, M.N., Cornou, C. and Madsen, J. (2014). Estimation of methane emission using the CO2
method from dairy cows fed concentrate with different carbohydrate compositions in
automatic milking system. Livestock Science 164, 56-67.

83
Hegarty, R.S. (2004). Genotype differences and their impact on digestive tract function of
ruminants: A review. Australian Journal of Experimental Agriculture 44, 459-467.
Hellwing, A.L.F., Lund, P., Madsen, J. and Weisberg, M.R. (2013). Comparison of enteric
methane production predicted from the CH4/CO2 ratio and measured in respiration chambers.
Advances Animal Biosciences 4, 557.
Herd, R.M., Velazco, J.I., Arthur, P.F. and Hegarty, R.F. (2016). Associations among methane
emission traits measured in the feedlot and in respiration chambers in Angus cattle bred to
vary in feed efficiency. Journal of Animal Science 94, 4882-4891.
Herrero, M., Havlík, P., Valin, H., Notenbaert, A., Rufino, M.C., Thornton, P.K., Blümmel, M.,
Weiss, F., Grace, D. and Obersteiner, M. (2013). Biomass use, production, feed efficiencies,
and greenhouse gas emissions from global livestock systems. Proceedings of National
Academy of Sciences 110, 20888-20893.
Herrero, M., Gerber, P., Vellinga, T., Garnett, T., Leip, A., Opio, C., Westhoek, H.J., Thornton,
P.K., Olesen, J., Hutchings, N., Montgomery, H., Soussana, J.-F., Steinfeld, H. and
McAllister, T.A. (2011). Livestock and greenhouse gas emissions: The importance of getting
the numbers right. Animal Feed Science and Technology, 166-167, 779-782.
Hristov, A.N., Oh, J., Giallongo, F., Frederick, T., Harper, M.T., Weeks, H., Branco, A.F., Price,
W.J., Moate, P.J., Deighton, M.H., Williams, S.R.O., Kindermann, M. and Duvall, S. (2016).
Short communication: Comparison of the GreenFeed system with the sulfur hexafluoride
tracer technique for measuring enteric methane emissions from dairy cows. Journal of Dairy
Science 99, 5461-5465.
Hristov, A.N., Oh, J., Giallongo, F., Frederick, T.W., Harper, M.T., Weeks, H.L., Branco, A.F.,
Moate, P.J., Deighton, M.H., Williams, S.R.O., Kindermann, M. and Duval. S. (2015a). An
inhibitor persistently decreased enteric methane emission from dairy cows with no negative
effect on milk production. Proceedings of the National Academy of Sciences 112, 10663-
10668.
Hristov, A.N., Oh, J., Giallongo, F., Frederick, T., Weeks, H., Zimmerman, P. R. (2015b) The
Use of an automated system (GreenFeed) to monitor enteric methane and carbon dioxide
emissions from ruminant animals. Journal of Visualized Experiments 103, e52904, 1-8.
Hristov, A.N., Oh, J., Firkins, J.L. Dijkstra, J., Kebreab, E., Waghorn, G., Makkar, H.P.S.,
Adesogan, A.T., Yang, W., Lee, C. and Gerber, P.J. (2013). Special topics – Mitigation of
methane and nitrous oxide emissions from animal operations: I. A review of enteric methane
mitigation options. Journal of Animal Science 91, 5045-5069.
Huhtanen, P., Ramin, M. and Hristov, A.N. (2016). Validation of the GreenFeed system against
model predicted methane emissions. Journal of Animal Science 94, E-Suppl. 5, 726-727
(abstract 1521).
Huhtanen, P., Cabezas-Garcia, E.H., Krizsan, S.J. and Shingfield, K.J. (2015a). Evaluation of
between-cow variation in milk urea and rumen ammonia nitrogen concentrations and the
association with nitrogen utilization and diet digestibility in lactating cows. Journal of Dairy
Science 98, 3182-3196.
Huhtanen, P., Cabezas-Garcia, E.H., Utsumi, S. and Zimmerman, S. (2015b). Comparison of
methods to determine methane emissions from dairy cows in farm conditions. Journal of
Dairy Science 98, 3394–3409.

84
Huhtanen, P., Ramin, M. and Udén, P. (2015c). Nordic dairy cow model Karoline in predicting
methane emissions: 1. Model description and sensitivity analysis. Livestock Science 178, 71-
80.
Huhtanen, P., Krizsan, S., Cabezas Garcia, E. H., Hetta, M. and Gidlund, H. (2013). Repeatability
and between cow variability of enteric CH4 and total CO2 emissions. Advances Animal
Biosciences 4, 588.
Huhtanen, P., Ahvenjärvi, S., Broderick, G.A., Reynal, S.M. and Shingfield, K.J. (2010).
Quantifying ruminal digestion of organic matter and neutral detergent fiber using the omasal
sampling technique in cattle – A meta-analysis. Journal of Dairy Science 93, 3203-3215.
Huhtanen, P., Rinne, M. and Nousiainen, J. (2009). A meta-analysis of feed digestion in dairy
cows. 2. The effects of feeding level and diet composition on digestibility. Journal of Dairy
Science 92, 5031-5042.
Huhtanen, P., Rinne, M. and Nousiainen, J. (2007). Evaluation of the factors affecting silage
intake of dairy cows: a revision of the relative silage dry-matter intake index. Animal 1, 758-
770.
Huhtanen, P., Brotz, P.G. and Satter, L.D. (1997). Omasal sampling technique for assessing
fermentative digestion in the forestomach of dairy cows. Journal of Animal Science 75, 1380-
1392.
Huhtanen, P., Kaustell, K. and Jaakkola, S. (1994). The use of internal markers to predict total
digestibility and duodenal flow of nutrients in cattle given six different diets. Animal Feed
Science and Technology 48, 211-227.
Hungate, R.E. (1966). The rumen and its microbes. Academic Press. New York.
IPCC (2007). Climate Change 2007: Mitigation. Contribution of Working Group III to the Fourth
Assessment Report of the Intergovernmental Panel on Climate Change. Cambridge, United
Kingdom and New York, NY, USA. (Available at:
http://www.ipcc.ch/publications_and_data(ar4/wg3/en/contents.html) [2016-06-25].
IPCC (2013). 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 United Kingdom and New York, NY, USA. (Available at:
http://www.climatechange 2013.org/report/) [2016-06-25].
Jaakkola, S. and Huhtanen, P. (1993). The effects of the forage preservation method and the
proportion of concentrate on nitrogen digestion and rumen fermentation in cattle. Grass and
Forage Science 48, 155-165.
Jenkins, T.C. (1993). Lipid metabolism in the rumen. Journal of Dairy Science 76, 3851-3863.
Jentsch, W., Schweigel, M., Weissbach, F., Scholze, H., Pitroff, W. and Derno, M. (2007).
Methane production in cattle calculated by the nutrient composition of the diet. Archives of
Animal Nutrition 67, 10-19.
Joblin, K.N. (1999). Ruminal acetogens and their potential to lower ruminant methane emissions.
Australian Journal of Agricultural Research 50, 1307-1313.
Johnson, K.A. and Johnson, D.E. (1995). Methane emissions from cattle. Journal of Animal
Science 73, 2483-2492.

85
Johnson, K.A., Huyler, M., Westberg, H., Lamb, B. and Zimmerman, P. (1994). Measurement of
CH4 emissions from ruminant livestock using a sulphur hexafluoride tracer technique.
Environmental Science and Technology 28, 359-362.
Johnson, D.E., Hill, T.M., Ward, G.M., Johnson, K.A., Branine, M.E., Carmean, B.R. and
Lodman, D.W. (1993). Ruminants and other animals. Pages 199-229 in M. A. K. Khalil, ed.
Atmospheric methane: sources, sinks, and role in global change. Springer-Verlag, Berlin
Heidelberg, Germany
Jonker, A., Lowe, K., Kittelmann, S., Janssen, P.H., Ledgard, S. and Pacheco, D. (2016). Methane
emissions changed nonlinearly with graded substitution of alfalfa silage with corn silage and
corn grain in the diet of sheep and relation with rumen fermentation characteristics in vivo and
in vitro. Journal of Animal Science 94, 3464-3475.
Kittelmann, S., Pinares-Patiño, C.S., Seedorf, H., Kirk, M.R., Ganesh, S., McEwan, J.C. and
Janssen, P.H. (2014). Two different bacterial community types are linked with the low-
methane emission trait in sheep. PLoS ONE 9, 19. //dx.doi.org/10.1371/journal.pone.0103171
Kearney, J. (2010). Review: Food consumption trends and drivers. Philosophical Transactions of
the Royal Society B 365, 2793-2807.
Knapp, J.R., Laur, G.L. Vadas, P.A., Weiss, W.P. and Tricarico, J.M. (2014). Invited review:
Enteric methane in dairy cattle production: Quantiying the opportunities and impact of
reducing emissions. Journal of Dairy Science, 97, 3231-3261.
Krause, D.O., Nagaraja, T.G., Wright, A.D.G. and Callaway, T.R. (2013). Board-invited review:
Rumen microbiology: Leading the way in microbial ecology. Journal of Animal Science 91,
331-341.
Kuoppala, K., Rinne, M., Nousiainen, J. and Huhtanen, P. (2008). The effect of cutting time of
grass silage in primary growth and regrowth and the interactions between silage quality and
concentrate level on milk production of dairy cows. Livestock Science 116, 171-182.
Lassen, J. and Løvendahl, P. (2013). Heritability for enteric methane emission from Danish
Holstein cows using a non-invasive FTIR method. Advances Animal Biosciences 4, 280.
Lassen, J., Løvendahl, P. And Madsen, J. (2012). Accuracy of non-invasive breath methane
measurements using Fourier transform infrared methods on individual cows. Journal of Dairy
Science 95, 890-898.
Leng, R.A. (2008). The potential of feeding nitrate to reduce enteric methane production in
ruminants. A report to the department of climate change. Commonwealth Government of
Australia, Canberra. www.penambulbooks.com.
Le Van, T.D., Robinson, J.A., Ralph, J., Greening, R.C., Smolenski, W.J., Leedle, J.A.Z. and
Schaefer, D.M. (1998). Assessment of reductive acetogenesis with indigenous ruminal
bacterium populations and Acetitomaculum ruminis. Applied and Environmental
Microbiology 64, 3429-3436.
LUKE. (2016). Finnish Feed tables.
Accessed Jul. 1, 2016. https://portal.mtt.fi/portal/page/portal/Rehutaulukot/feed_tables_english
Madsen, J., Bjerg, B.S., Hvelplund, T., Weisbjerg, M.R. and Lund, P. (2010). Methane and
carbon dioxide ratio in excreted air for quantification of the methane production from
ruminants. Livestock Science 129, 223-227.

86
Martinez-Fernandez, G., Denman, S.E., Yang, C., Cheung, J., Mitsumori, M. and McSweeney,
C.S. (2016). Methane inhibition alters the microbial community, hydrogen flow, and
fermentation response in the rumen of cattle. Frontiers in Microbiology 7, 1122, 1-14.
McAllister, T.A. (2011). Greenhouse gases in animal agriculture - Finding a balance between
food production and emissions. Animal Feed Science and Technology 166-167, 1-6.
McAllister, T.A. and Newbold, C.J. (2008). Redirecting rumen fermentation to reduce
Methanogenesis. Australian Journal of Experimental Agriculture 48, 7-13.
McAllister, T.A., Okine, E.K., Mathison, G.W. and Cheng, K.J. (1996). Dietary, environmental
and microbiological aspects of methane production in ruminants. Canadian Journal of Animal
Science 76, 23-243.
McDonald, P., Edwards, R.A., Greenhalgh, J.F.D., Morgan, C.A., Sinclair, L.A. and Wilkinson,
R.G. (2011). Animal Nutrition, 7th ed. Harlow, UK: Prentice Hall.
McLean, J.A. and Tobin, G. 1987. Animal and Human Calorimetry. Cambridge Univ. Press, New
York, NY.
Mehtiö, T., Rinne, M., Nyholm, L., Mäntysaari, P., Sairanen, A., Mäntysaari, E.A., Pitkänen, T.
and Lidauer, M.H. (2016). Cow-specific diet digestibility predictions based on near-infrared
reflectance spectroscopy scans of faecal samples. Journal of Animal Breeding and Genetics,
133, 115-125.
Mogensen, L., Kristensen, T., Nguyen, T.L.T., Knudsen, M.T. (2014). Method for calculating
carbon footprint of cattle feeds e including contribution from soil carbon changes and use of
cattle manure. Journal of Cleaner Production 73, 40-51.
Molano, G., Knight, T.W. and Clark, H. (2008). Fumaric acid supplements have no effect on
methane emissions per unit of feed intake in wether lambs. Australian Journal of
Experimental Agriculture 48, 165-168.
Morgavi, D.P., Forano, E., Martin, C. Newbold, C.J. (2010). Microbial ecosystem and
methanogenesis in ruminants. Animal (4:7), 1024-1036.
Moss, A.R., Jouany, J.P. and Newbold, J. (2000). Methane production by ruminants: its
contribution to global warming. Annals of Zootechnology 49, 231-253.
Muñoz, C., Yan, T., Wills, D.A., Murray, S. and Gordon, A.W. (2012). Comparison of the sulfur
hexafluoride tracer and respiration chamber techniques for estimating methane emissions and
correction for rectum methane output from dairy cows. Journal of Dairy Science 95, 3139-
3148.
Napolitano, F., Grasso, F., Bordi, A., Tripaldi, C., Pacelli, C., Saltalamacchia, F. and De Rosa, G.,
(2005). On-farm welfare assessment in dairy cattle and buffaloes: evaluation of some animal-
based parameters. Italian Journal of Animal Science 4, 223-231
Naturvårdsverket (2008) (Available at: http://naturvardsverket.se(Samar-miljon/Klimat-och-
luft/Stastistik -om-luft/Utslappsstatistik/) [2016-06-25].
Naturvårdsverket (2013). National Inventory Report, Swedish Environmental Protection Agency,
Stockholm, Sweden.
Negussie, E., de Haas, Y., Dehareng, F., Dewhurst, R.J., Dijkstra, J., Glender, N., Morgavi, D.P.,
Soyeurt, H., van Gastelen, S., Yan, T. and Biscarini, F. (2017). Invited review: Large-scale
indirect measurements for enteric methane emissions in dairy cattle: A review of proxies and

87
their potential for use in management and breeding decisions. Journal of Dairy Science 100,
2433-2453.
Newbold, C.J., El Hassan, S.M., Wang, J., Ortega, M.E. and Wallace, R.J. (1997). Influence of
foliage from African multipurpose trees on activity of rumen protozoa and bacteria. British
Journal of Nutrition 78, 237-249.
NRC. (2001). Nutrient Requirements of Dairy Cattle. 7th rev. ed. Natl. Acad. Press, Washington,
DC.
Ørskov, E.R. and Ryle, M. (1990). Energy Nutrition in Ruminants. Elsevier Science Publishers
LTD.
Ørskov, E.R., Ojwang, I. and Reid, G.W. (1988). A study on consistency of differences between
cows in rumen outflow rate of fibrous particles and other substrates and consequences for
digestibility and intake of roughages. Animal Production 47, 45-51.
Patra, A.K. and Zhongtang, Y. (2014). Essential oils affect populations of some rumen bacteria in
vitro as revealed by microarray (RumenBactArray) analysis. Frontiers in Microbiology 6,
297. https://doi.org/10.3389/fmicb.2015.00297
Pickering, N.K., Oddy, V.H., Basarab, J., Cammack, K., Hayes, B., Hegarty, R.S., Lassen, J.,
McEwan, J.C., Miller, S., Pinares-Patiño, C.S. and de Haas, Y. (2015). Invited review: genetic
possibilities to reduce enteric methane emissions from ruminants. Animal 9, 1431-1440.
Pinares-Patiño, C.S., Hickey, S.M., Young, E.A., Dodds, K.G., MacLean, S., Molano, G.,
Sandoval, E., Kjestrup, H., Harland, R., Hunt, C., Pickering, N.K. and McEwan, J.C. (2013).
Heritability estimates of methane emissions from sheep. Animal 7:s2, 316-321.
Pinares-Patiño, C.S., Lassey, K.R., Martin, R.J., Molano, G., Fernandez, M., MacLean, S.,
Sandoval, E., Luo, D. and Clark, H. (2011). Assessment of the sulphur hexafluoride (SF6)
tracer technique using respiration chambers for estimation of methane emissions from sheep.
Animal Feed Science and Technology 166, 201-209.
Pinares-Patiño, C.S. and Clark, H. (2010). Rumen function and digestive parameters associated
with methane emissions in dairy cows. Proceedings of the 4th Australasian Dairy Science
Symposium 2010. Pages 86-93.
Pinares-Patiño, C.S., D’Hour, P., Jouany, J.P. and Martin, C. (2007). Effects of stocking rate on
methane and carbon dioxide emission from grazing cattle. Agriculture, Ecosystems and
Environment 121, 30-46.
Pinares-Patiño, C.S., Ulyatt, M.J., Lassey, K.R., Barry, T.N. and Holmes, C.W. (2003). Rumen
function and digestion parameters associated with differences between sheep in methane
emissions when fed chaffed lucerne hay. Journal of Agricultural Science, 140, 205–214.
Ramin, M. and Huhtanen, P. (2013). Development of equations for predicting methane emissions
from ruminants. Journal of Dairy Science 96, 2476-2493.
Rinne, M., Huhtanen, P. and Jaakkola, S. (2002). Digestive processes of dairy cows fed silages
harvested at four stages of grass maturity. Journal of Animal Science 80, 1986-1998.
Robinson, D. L., Goopy, J.P., Hegarty, R.S. and Vercoe, P.E. 2010. Repeatability, animal and sire
variation in 1-hr methane emissions and relationship with rumen volatile fatty acid
concentrations. Abstract no. 712 in Proceedings 9th World Congress in Genetics Applied to
Livestock. Book of Abstracts. German Society of Animal Science, Leipzig, Germany.

88
Roehe, R., Dewhurst, R.J., Duthie, C.A., Rooke, J.A., McKain, N., Ross, D.W., Hyslop, J.J.,
Waterhouse, A., Freeman, T.C., Watson, M. and Wallace, R.J. (2016). Bovine host genetic
variation influences rumen microbial methane production with best selection criterion for low
methane emitting and efficiently feed converting hosts based on metagenomics gene
abundance. PLoS Genet. 12:e1005846. http://dx.doi.org/10.1371/journal.pgen.1005846
Russell, J.B. 2002. Rumen Microbiology and its Role in Ruminant Nutrition, 1st ed. Ithaca, NY.
Russell, J.B. and Rychlik, J.L. (2001). Factors that alter rumen microbial ecology. Science 292,
1119-1122.
Russell, J.B., O’Connor, J.D., Fox, D.G., Van Soest, P.J. and Sniffen, C.J. (1992). A net
carbohydrate and protein system for evaluating cattle diets: I. Ruminal fermentation. Journal
of Animal Science 70, 3551-3561.
Sauvant, D. and Giger-Reverdin, S. (2009). Modelling of digestive interactions and methane
production in ruminants. INRA Production Animales 22, 375-384.
SCB (statistiska centralbyrån) (2016) (Available at:
http://www.statistikdatabasen.scb.se/pxweb/sv/ssd/START_JO_JO0103/HusdjurL/?rxid=da84
6e89-fed5-46f4-b789-63a11b6d7580) [2016-06-25].
Schiemann, R., Jentsch W. and Wittenburg, H. (1971). Zur Abhängigkeit der Verdaulichkeit der
Energie und der Nährstoffe von der Höhe der Futteraufnahme und der
Rationszusammensetzung bei Milchkühen. Arch. Tierernähr 21, 223-240.
Sjaunja, L.O., Baevre, L., Junkkarinen, L., Pedersen, J. and Setälä, J. (1991). A Nordic proposal
for an energy corrected milk (ECM) formula. Pages 156–157 in Recording of Animals—State
of the Art 1990. EAAP Publ. 50. Perform. Centre Agric. Publ. Doc. (PUDOC), Wageningen,
the Netherlands.
Smith, J., Sones, K., Grace, D., MacMillan, S., Tarawali, S. and Herrero, M. (2013). Beyond
milk, meat, and eggs: Role of livestock in food and nutrition security. Animal Frontiers 3, 6-
13.
Smuts, M., Meissner, H.H. and Cronjé, P.B. (1995). Retention time of digesta in the rumen: Its
repeatability and relationship with wool production of Merino rams. Journal of Animal
Science 73, 206-210.
Steinfeld, H., Gerber, P., Wassenaar, T., Castel, V., Rosales, M. and de Haan, C. (2006).
Livestock’s Long Shadow: Environmental Issues and Options. Rome: Food and Agriculture
Organization of the United Nations.
Storm, I.M.L.D., Hellwing, A.L.F., Nielsen, N.I. and Madsen, J. (2012). Methods for measuring
and estimating methane emission from ruminants. Animals 2, 160-183.
Thomas, C. 1987. Factors affecting the substitution rates in dairy cows on silage based rations.
Pages 205-218 in: Recent Advances in Animal Nutrition. W. Haresign and D.J.A Cole. (ed.)
Butterworths, London, UK.
Van Keulen, J., and Young, B.A. (1977). Evaluation of acid-insoluble ash as a natural marker in
ruminant digestibility studies. Journal of Animal Science 44, 282–287.
Van Nevel, C.J. and Demeyer, D.I. (1995). Feed additives and other interventions for decreasing
methane emissions. In: Biotechnology in animal feeds and animal feeding. R. Wallace and A.
Chesson (eds.). VCM, Weinheim. Pages 329-349.

89
Van Soest, P.J. (1994). Nutritional Ecology of the Ruminant. 2nd ed. Ithaca, NY: Cornell
University Press.
Van Vugt, S.J., Waghorn, G.C., Clark, D.A. and Woodward, S.L. (2005). Impact of monensin on
methane production and performance of cows fed forage diets. Proceedings of New Zealand
Society of Animal Production 65, 362-366.
Van Zijderveld., Gerrits, W.J.J., Apajalahti, J.A., Newbold, J.R., Dijkstra, J., Leng, R.A. and
Perdok, H.B. (2010). Nitrate and sulfate: Effective alternative hydrogen sinks for mitigation of
ruminal methane production in sheep. Journal of Dairy Science 93, 5856-5866.
Vermorel M. and Jouany J.P. (1989). Effects of rumen protozoa on energy utilization by wethers
of two diets based on ammonia-treated straw supplemented or not with maize. Asian-
Australasian Journal of Animal Science 2, 7–17.
Vlaming, J.B., Lopez-Villalobos, N., Brookes, I.M., Hoskin, S.O. and Clark, H. (2008). Within-
and between-animal variance in methane emissions in non-lactating dairy cows. Australian
Journal of Experimental Agriculture 48, 124-127.
Volden, H. (1999). Effects of level of feeding and ruminally undegraded protein on ruminal
bacterial protein synthesis, escape of dietary protein, intestinal amino acid profile, and
performance of dairy cows. Journal of Animal Science 77, 1905-1918.
Waldo, D.R. (1970). Factors influencing the voluntary intake of forages. In: R.F. Barnes et al.
(eds.) Proceedings of the National Conference on Forage Quality, Evaluation and Utilisation,
Pages E1-22.
Waghorn, G.C. (2008). Beneficial and detrimental effects of dietary condensed tannins for
sustainable sheep and goat production—Progress and challenges. Animal Feed Science and
Technology 147, 116-139.
Waghorn, G.C., Tavendale, M.H. and D. R. Woodfield. (2002). Methanogenesis from forages fed
to sheep. Proceedings of New Zealand Grassland Association 64, 167-171.
Wedlock, D.N., Janssen, P.H., Leahy, S.C., Shu, D. and Buddle, B.M. (2013). Progress in the
development of vaccines against rumen methanogens. Animal, 7(s2), 244-252.
Weld, K.A. and Armentano, L.E. (2017). The effects of adding fat to diets of lactating dairy cows
on total-tract neutral detergent fiber digestibility: A meta-analysis. Journal of Dairy Science
100:1766-1779.
Wolin, M.J. (1960). A theoretical rumen fermentation balance. Journal of Dairy Science 40,
1452-1459.
Wolin, M.J., Miller, T.L. and Stewart, C.S. (1997). Microbe-microbe interactions In: The Rumen
Microbial Ecosystem. 2nd ed. (Ed. P. J. Hobson and C. S. Stewart), Blackie Acad. Profess.
London. Pages: 467-491.
Yan, T., Mayne, C.S., Gordon, F.G., Porter, M.G., Agnew, R.E., Patterson, D.C., Ferris, C.P. and
Kilpatrick, D.J. (2010). Mitigation of enteric methane emissions through improving efficiency
of energy utilization and productivity in lactating dairy cows. Journal of Dairy Science 93,
2630-2638.
Yan, T., Agnew, R.E. and Gordon, F.J. (2002). The combined effects of animal species (sheep
versus cattle) and level of feeding on digestible and metabolizable energy concentrations in
grass silage based diets of cattle. Animal Science 75, 141-151.

90
Yan, T., Agnew, R.E., Gordon, F.J. and Porter, M.G. (2000). The prediction of methane energy
output in dairy and beef cattle offered grass silage-based diets. Livestock Production Science
64, 253-263.
Zhu, Z., Kristensen, L., Højberg, O., Poulsen, M., Lassen, J., Noel, S.J. and Løvendahl, P. (2014).
Variation among dairy cows in rumen liquid fermentation characteristics. Proceedings, 10th
World Congress of Genetics Applied to Livestock Production. Vancouver, Canada. August 17-
22.
Zimmerman, P. (2011). U.S. Patent 7966971, “Method and system for monitoring and reducing
ruminant methane production”, U.S. Patent and Trademark Office, June 28, 2011.

91
Acknowledgements
There are so many people who have supported me in several ways in order to
complete this big challenge, some more than others but all of their
contributions are highly appreciated. During my time as a PhD student, I have
improved significantly my scientific, practical and communication skills. Such
professional goals only come true when one has the opportunity of sharing
with colleagues which encourage you to be more curious, methodical,
supportive and passionate for science without the ridiculous jealousness of
someone in the academia who wants to keep the “secrets” for him/herself. At
the end, science is also about of taking part and be responsible of the
consequences of our actions. As a good friend of mine said: “We are blessed
creatures because God creates nature – to be discovered and also understood
– for us as Animal Scientists. It has to be our main skill my friend, don’t forget
about it”.
After this long introduction, first I would like to express my appreciation to
all present and former colleagues at the Department of Agricultural Research
for Northern Sweden, my family and friends in Colombia and over the world.
Financial support to conduct my research is also acknowledged. Since there is
not room to acknowledge everyone as I would like, I have dedicated the
following lines to:
Professor Pekka Huhtanen, my main and fantastic supervisor! You are
enthusiastic, patient, creative and always supportive. For me it has been an
honor and a pleasure working and sharing with you Pekka. One plus one
(coming ideas) give as a result much more than 2!!! That level is only possible
to reach when someone is so passionate and dedicated as you are about
science. Thanks a lot Pekka for showing me the way to continue the scientific
journey and always believing in your students. Finally, as our friend Kari
Tapio used to said: “Myrskyn Jälkeen”. Kiitos!!!

92
Dr. Sophie Krizsan, my co-supervisor. For practicalities and going directly
to the point, you’re the best one! I enjoyed your guidance and good advices
Sophie. Please send my regards to Tim and colleagues – nice guys!
Professor Kevin Shingfield (in memoriam), my co-supervisor. Role model
to be a better scientist and pursue my dreams in my scientific career. It was my
pleasure to have a chance of sharing with you, even just for a short time.
Dr. Mohammad Ramin, my new co-supervisor. Thanks Mohammad for
your hospitality and friendship during my time in Umeå. We passed very good
times together. Regards from Dr. Prada, believe me!
Professor Kjell Martinsson and Dr. Mårten Hetta, former and current
heads of NJV Department for giving me the chance to be part of the Animal
Science section as a PhD student. Support with administrative issues from
Britt-Inger Nyberg, Gun Bernes, and Johanna Wallsten is also highly
appreciated.
Professor, Kerstin Huss-Danell, thanks for your kindness and
encouragement to learn the Swedish language. God morgon!
Animal Nutrition laboratory, thanks to Ann-Sofi Hahlin and Lars
Wallgren for your big help with samples processing and analytical work.
Mycket bra!
Röbäcksdalen barn crew, for their support with experimental work with
the dairy cows. Karim, Annika, Evelina, Abbas, Malin, Viktor, James, Joakim,
Emma, Stig. For all of you guys: Tack!
Dr. Merko Vaga, you have been much more than my brother during our
time in Umeå. You also were my first English language teacher when I came
(something that I will really appreciate for the rest of my career), and to me,
the best example of self-confidence and determination to achieve goals in life.
We came together to the North and now we are finishing this chapter in our
lives. New challenges are coming for us! Thanks man!!!
Helena Gidlund, thank you very much for your kindness and big support
during my PhD studies. I have learn good tips from you about how to be more
organized with data from barn experiments. For sure, that knowledge is for the
entire life. Tack så mycket!

93
Professor Edenio Detmann, my friend, you showed me that biodiversity
even in the Northern hemisphere, is much more than looking at “the white bird
and the blue one” and just merely describing them. You gave me good tips to
overcome and survive the long periods of cold and darkness. Now, the sun is
shining for the Three Little Birds! Nós também compartilhamos o orgulho e o
privilegio de ser Zootecnistas! Eh nóis!!!
Dr. Rafael Leite, The wise guy! Advices and role model to be a better
person in life. “Tudo bem Rafa, mas também me deixa reclamar... kkkkkk”.
Brincadeira! Meu mais sincero Muito Obrigado, meu amigo!
Degong Pang, fruitful discussions about research and life in the North and
in the UK. You’re so determined and committed! No doubts about your good
performance during the rest of your PhD studies. Trust in yourself, and don’t
forget: be water my friend!
Xiaoxia Dai, for the professional and big help conducting my flow study.
Ohhh thank youuuu Daisy!!! Pancakes are not enough to express my gratitude.
Former and current interns, PhD students and Postdocs, sorry If I
forgot someone: Abdulai, Miriam, Annika Höjer, João, Natalie, Marcia, Niko,
Maria, Salvatore, Raimonda, Petra, Cynthia, Dora, Farhang, Zhenjiang. You
guys made my life easier and enjoyable in the North, thank you very much!
To my beloved family, always and forever in my thoughts even though far
away from home: Hermógenes, Maria Ligia, Gabrielina, Diana, Luz Stella,
Ligia Janeth, Luisa, Felipe and Jerónimo. This PhD degree is also yours!
Ustedes más que nadie merecen este logro. Mil gracias por todo!
Professor Luiz Gustavo Nussio and his Forage Quality and Conservation
Research Team (QCF; University of São Paulo, Brazil). Words are not enough
to express my gratitude. Muito obrigado mesmo pela oportunidade de fazer
meus estudos de mestrado na ESALQ com o apoio de um fantástico time!
Aos amigos da Viola Quebrada. Valeu Hagas! Rodrigo, Delci, Salim,
Adenilson, Marcão, Marconi, Cesinha, Roberto; vcs sabem que sempre fico a
disposição quando precisarem!

94
Juan Carlos Araujo Cabarcas, Parcero! Muchas gracias por los buenos
consejos y también por liderar las tertulias académicas. Comparto su
preocupación y búsqueda constante de soluciones para ver algún día a nuestra
patria en mejores manos. Seguimos en la jugada!
A todos los amigos salseros y bachateros de Umeå, Daniela, Mayte,
Daniel, Alessia, Anastasia, todos!. Como dice Celia: “my English is not very
good looking”. Abrazo a todos y como dice El Joe: “Y para ti también”
Daniel Eduardo Rodríguez Aguilar, Yo veré mano, lo quiero ver
haciendo el doctorado, con toda! Saludos y muchos éxitos amigo!
Profesor Alvaro Wills Franco, cátedras magistrales en Nutrición de
Rumiantes en la Universidad Nacional de Colombia.
Dr. Marko Kass, My friend, Estonian diplomacy suits you very well! I told
you… I did!
Finally but no less important, I would like to express my appreciation to the
institutions and people whom made possible the funding for my PhD studies at
SLU:
RuminOmics EU Project (no. 289319 of the European Community 7th
Framework Programme: Food Agriculture, Fisheries and Biotechnology).
Swedish Research Council – FORMAS (project no. 220-2011-1247).
Con especial afecto para todos los Ingenieros Zootecnistas
que un día soñaron con serlo a pesar de las dificultades
Keywords:
Tack så mycket! Kiitos! Obrigado! Thank you! Gracias!
Umeå, Sweden, 3th April 2017.