Version of Record: [Link]

com/science/article/pii/S0377840116301055
Manuscript_913365029822cef5f517e6fa0c1829fe

1 Design, implementation and interpretation of in vitro batch culture experiments to

2 assess enteric methane mitigation in ruminants – a review

4 Yáñez-Ruiz D.R.a*, Bannink A.b, Dijkstra J.c, Kebreab E.d, Morgavi D.P.e, O´Kiely P.f,

5 Reynolds C.K.g, Schwarm A.h, Shingfield [Link], Yu Z.k, Hristov A.N.l

7 a
Estación Experimental del Zaidín (CSIC), Profesor Albareda, 1, 18008, Granada, Spain

8 b
Animal Nutrition, Wageningen UR Livestock Research, PO Box 65, Lelystad, The Netherlands

9 c
Animal Nutrition Group, Wageningen University, PO Box 338, 6700 AH Wageningen, The

10 Netherlands

11 d
Department of Animal Science, University of California, Davis 95616, USA

12 e
INRA, UMR1213 Herbivores, F-63122 Saint-Genès-Champanelle, and Clermont Université,

13 VetAgro Sup, UMR Herbivores, BP 10448, F-63000 Clermont-Ferrand, France

14 f
Teagasc, Animal & Grassland Research and Innovation Centre, Grange, Dunsany, Co. Meath,

15 Ireland

16 g
The University of Reading, Earley Gate, P.O. Box 237, Reading RG6 6AR, UK

17 h
ETH Zurich, Institute of Agricultural Sciences, Animal Nutrition, 8092 Zurich, Switzerland

18 i
Natural Resources Institute Finland (Luke), FI-31600 Jokioinen, Finland
19 j
Institute of Biological, Environmental and Rural Sciences, Aberystwyth University,

20 Aberystwyth, SY23 3EB, UK

21 k
Department of Animal Sciences, The Ohio State University, Columbus 43210, USA

22 l
Department of Animal Science, The Pennsylvania State University, University Park 16802,

23 USA

24

25 * Corresponding author: Dr. David R. Yáñez-Ruiz, Tel: +34 958 572757, Fax: +34 958
26 572753. Email: [Link]@[Link]

1
© 2017 published by Elsevier. This manuscript is made available under the Elsevier user license
[Link]
27 Abstract

28 In vitro fermentation techniques (IVFT) have been widely used to evaluate the nutritive

29 value of feeds for ruminants and in the last decade to assess the effect of different

30 nutritional strategies on methane production. However, many technical factors may

31 influence the results obtained. The present review has been prepared by the ‗Global

32 Network‘ FACCE-JPI international research consortium to provide a critical evaluation

33 of the main factors that need to be considered when designing, conducting and

34 interpreting IVFT experiments investigate nutritional strategies to mitigate methane

35 (CH4) emission from ruminants. Given the increasing and wide-scale use of IVFT

36 techniques, there is a need to critically review reports in the literature and establish what

37 criteria are essential to the establishment and implementation of in vitro techniques.

38 Key aspects considered include: i) donor animal species and number of animal used, ii)

39 diet fed to donor animals, iii) collection and processing of rumen fluid as inoculum, iv)

40 choice of substrate and incubation buffer, v) incubation procedures and CH4

41 measurements, vi) headspace gas composition and vii) comparability of in vitro and in

42 vivo measurements. Based on an evaluation of experimental evidence, a set of technical

43 recommendations are presented that allowing the harmonization of IVFT laboratory

44 methods employed with IVFT, of and procedures for feed evaluation, assessment of

45 rumen function and CH4 production.

46

47 Keywords: feed evaluation, in vitro gas production, methane, rumen, mitigation,

48 microbial inoculum

49 Abbreviations: DM, dry matter; DMI, dry matter intake; IVFT, in vitro fermentation

50 technique; NDF, neutral detergent fibre; OM, organic matter; VFA, volatile fatty acids.

51

2
52

53

54 1. Introduction

55 In vitro fermentation techniques (IVFT) that involve incubations of substrates with

56 rumen fluid have been used extensively to evaluate the nutritive value of ruminant

57 feeds. Measurements based on IVFT complement standard laboratory analysis of

58 chemical composition and therefore offer a rapid and less expensive alternative to the

59 determination of nutrient digestibility in vivo (Rymer et al., 2005). Also, application of

60 the IVFT to reduce the use of experimental animals represents an advantage when large

61 number of treatments needs to be tested. More recently, IVFT techniques have been

62 used to assess the potential of diet, dietary ingredients and modifiers of rumen

63 fermentation to decrease methane (CH4) emissions from ruminant livestock (Bodas et

64 al., 2008; Durmic et al. 2010). For many research groups with limited resources, the

65 use of in vitro tools is often the only option available for investigating potential agents

66 for CH4 mitigation. Depending on the research question, in vitro studies can be valuable

67 for screening and informing on the suitability for further evaluation in vivo. However, a

68 positive outcome in vitro does not guarantee that the same treatment will have a similar

69 effect in vivo. In some cases, IVFT results for feed evaluation and CH4 mitigation can

70 be misleading when the inherent characteristics of a batch culture system are not

71 carefully considered (reviewed by Dijkstra et al., 2005). Furthermore, the goals,

72 experimental design, results and conclusions of in vitro experiments require cautious

73 and considered interpretation. Often the findings from in vitro studies have little

74 relevance to commercial conditions, simply because in the amounts tested, an additive

75 would be too expensive for use on-farm, efficacy cannot be confirmed in vivo or have

76 detrimental effect on animal health and function.

3
 77 The FACCE-JPI ‗Global Network‘ project is an international initiative that

78 intends, among other goals, to develop reliable and robust guidelines for generating and

79 evaluating data from in vitro and in vivo experiments examining the potential to

80 mitigate greenhouse gas (GHG) emissions from ruminant livestock systems. The

81 present review is one output from the project that provides a critical evaluation of the

82 key points for consideration when planning and performing in vitro studies and

83 guidance for end users in the interpretation of experimental data from IVFT

84 experiments. Major emphasis is placed on factors that influence microbial activity

85 within in vitro systems typically used to assess CH4 production, and how these can be

86 balanced or accounted for, rather than simply providing a contrast and comparison of

87 reports in the scientific literature. It is not intended to provide an in-depth and

88 comprehensive description of different in vitro systems available, but rather provide an

89 appraisal of key aspects that are central to undertaking robust, representative and

90 reproducible in vitro experiments.

91

92 2. History and use of in vitro batch culture

93 Early in vitro studies focused on endpoint measurements such as the extent of substrate

94 degradation (Tilley and Terry, 1963). In the 1970‘s, researchers recognized that

95 measurement of fermentation gases in combination with dietary chemical composition

96 could be used to estimate both feed metabolisable energy content and ruminal organic

97 matter degradability. Czerkawski and Breckenridge (1975) developed a system that

98 involved recording the direct displacement of a piston by gases produced during the

99 fermentation of feeds by rumen fluid in a glass syringe. This was the basis of the

100 ‗Hohenheim Gas Test‘ developed by Menke et al. (1979, Table 1). The ‗syringe

101 technique‘ was originally developed to determine end-point fermentation of feeds after

4
102 24 h of incubation. Blümmel and Ørskov (1993) modified the technique by incubating

103 syringes in a water bath rather than a rotating incubator. By recording gas production at

104 more frequent intervals, the kinetics of fermentation could also be determined.

105 Wilkins (1974) described a different approach to measure fermentation kinetics

106 in vitro, whereby fermentation took place in a sealed vessel containing rumen fluid,

107 buffer and substrate, and a pressure transducer was used to measure gas accumulation in

108 the vessel headspace. In the simplest setup of this system, headspace pressure is

109 measured manually, as described by Theodorou et al. (1994, Table 1), while gas

110 samples are collected for the analysis of CO2, CH4 and/or H2 concentrations when gas

111 pressure is released (Tekippe et al., 2013). In the 1990‘s the first automated pressure

112 based systems were developed (Pell and Schofield, 1993, Table 1), providing real-time

113 measurements of gas accumulation allowing for a better understanding of the kinetics of

114 fermentation for a range of substrates (Groot et al., 1996). During the development of

115 these systems, it became increasingly clear that increased pressure within the

116 fermentation container could affect fermentation end-products (Jouany and Lassalas,

117 2002) and the rate and extent of fermentation (Tagliapietra et al., 2010). More advanced

118 systems periodically release and collect the gas via a solenoid valve (Cone et al., 1996;

119 Davies et al., 2000, Table 1) thereby avoiding the build-up of pressure. Even in

120 automated systems, analysis of gas composition (e.g. CH4) typically requires manual

121 injection of sample gases into a gas analyzer (Martínez et al., 2010; Pellikaan et al.,

122 2011).

123 Cornou et al. (2013) described the results of a ring-test evaluating the use of a

124 wireless system for automated gas release developed by Ankom (Ankom Technology,

125 Macedon, NY, USA). This system is being used in various laboratories, but still relies

126 on manual gas sampling and analysis.

5
127 More recently Muetzel et al. (2014) developed an automated gas measurement

128 system by which gas production is monitored in real-time via pressure sensors and the

129 proportion of CH4 and H2 in the vented fermentation gases is measured automatically by

130 gas chromatography. The main difference with previous automated systems (Cornou et

131 al., 2013) is that fermentation gases are collected and analysed by a computer-controlled

132 gas chromatograph, rather than being released into the air once a threshold pressure is

133 reached.

134 In vitro gas production systems have been used extensively for rapid screening

135 of chemical substances, plant species, plant extracts and dietary ingredients on CH4

136 emissions from rumen fermentation. Such experimental approaches have allowed the

137 mode of action of a range of chemicals (Busquet et al., 2005; Bodas et al., 2008; Garcia-

138 Gonzalez et al., 2008; Durmic et al., 2010) and dietary substrates (Patra and Yu, 2013;

139 Hatew et al., 2015) to be investigated. Use of IVFT offers the opportunity to evaluate a

140 broad spectrum of chemical agents alone or in a number of combinations over a wide

141 range of concentrations (e.g., Busquet et al., 2005; Garcia-Gonzalez et al., 2008).

142 However, this technique does not generate reliable information for agents that are only

143 effective for decreasing CH4 emissions over an extended period (Castro-Montoya et al.,

144 2015). Furthermore, results from screening studies (Bodas et al., 2008; Durmic et al.,

145 2010) are often inconclusive and may be conflicting due to variation in dosage,

146 chemical structure of the test substance or compound, diet, combination of treatments

147 applied, adaptation of rumen microbes or the form in which an agent is introduced into

148 the system (Cardozo et al., 2004, 2005). Substantial decreases of CH4 production in

149 vitro (Tan et al., 2011) have been reported, but in several cases these have been

150 accompanied by adverse effects on feed degradation, with the implication that a similar

151 effect may occur in vivo which would compromise diet digestibility and animal

6
152 performance.

153

154 Please insert Table 1 around here

155

156 3. Aspects to consider

157 3.1. Donor animal species and animal numbers

158 Different animal species (sheep, goats, cattle and buffaloes) may vary in their response

159 to the same CH4 mitigation strategy. The obvious recommendation is to use the same

160 species as donors of rumen fluid for in vitro incubations as the intended target species.

161 However, this is not always possible, and due to cost small ruminants are often used as

162 donors of rumen contents, even when cattle are the target species. One key question is,

163 therefore, whether sheep or goats can be used as suitable surrogates for cattle for the

164 study of CH4 production in vitro.

165 Bueno et al. (1999) compared the microbial biomass in bovine and ovine rumen

166 fluid and, although no intake data were reported, the inocula were adjusted to provide

167 the same microbial biomass. It was concluded that rumen fluid from sheep could

168 replace that from cattle or vice versa as rumen inoculum and that the two sources were

169 comparable under tropical feeding conditions. Cone et al. (2002) reported a comparison

170 of rumen samples collected in the same way from sheep and cattle maintained under

171 similar conditions. Incubations of 22 different feeds were performed. A close

172 association was observed (r = 0.98) for gas production at 24 (ranging from 100 to 325

173 ml/g OM incubated) and 48 (ranging from 150 to 350 ml/g OM) h for incubations with

174 rumen fluid from cows and sheep. However, the relationship based on the rate of gas

175 production (ml/g OM/h) was weaker (r = 0.79). Calabro et al (2005) compared rumen

176 fluid from buffalo and sheep as a source of inoculum and observed higher fermentation

177 rates and extent of degradation during incubations with rumen fluid from sheep.
7
178 Differences in fermentation kinetics were greater when fibre-rich substrates were tested,

179 such as straw and hay, but negligible for barley grain.

180 Muetzel et al. (2014) compared rumen fluid from cattle (Holstein × Jersey cows)

181 and sheep using a newly developed automated in vitro system and reported that total gas

182 production is unaffected by donor animal species. Prior to rumen sampling, cattle and

183 sheep had been adapted to a medium quality hay diet fed to meet maintenance energy

184 requirements for 14 days. However, proportions of CH4 in vented fermentation gases

185 were lower during incubations with rumen fluid from sheep than cattle (150 vs. 158

186 ml/L, P=0.003) associated with a lower proportion of acetate and a higher proportion of

187 propionate. There is no evidence of differences in methanogen communities in sheep

188 and cattle (Jeyanathan et al., 2011), which suggests that the differences are driven by H2

189 production from the bacterial or protozoal communities. No interaction between animal

190 species (sheep vs. cattle) and the type of substrate incubated (chicory, lucerne, ryegrass,

191 straw and white clover) was observed (Muetzel et al., 2014). Bueno et al. (2015)

192 compared in vitro CH4 production using rumen fluid from taurine dairy cattle (Bos

193 taurus taurus), zebu beef cattle (Bos taurus indicus), water buffaloes (Bubalus bubalis),

194 sheep (Ovis aries) and goats (Capra hircus) fed similar diets while testing the effect of

195 condensed tannins from an Acacia extract. Rumen fluid from cattle resulted in higher

196 CH4 per unit of degraded organic matter (OM) formation than rumen fluid from small

197 ruminants.

198 Although microbiota of ruminant species housed in close contact and fed a similar diet

199 may be of a similar composition, the microbial ecology of rumen samples between

200 sheep and goats, for example, may vary due to differences in dentition, eating and

201 ruminating behaviour, digestive tract physiology and ruminal retention time (Ammar et

202 al., 2004). For this reason, collection of inoculum from animals of the same target

8
203 species fed a diet containing the same feedstuffs would be recommended. Differences

204 due to feeding behaviour and diet composition can to some extent be overcome by the

205 collection of rumen samples before morning feeding, when the effect of diet

206 composition on rumen metabolites or microbiota are likely to be minimized (Martinez

207 et al., 2010) (see section 3.3).

208 There remains some uncertainty on the number of animals that need to be

209 sampled to provide a representative sample of rumen inocula. Several studies using

210 different ruminant species (e.g., Pinares-Patiño et al., 2003; Waghorn et al., 2006; Yan

211 et al., 2006) have reported that CH4 emissions per unit dry matter intake vary between

212 individual animals. Such variation has been associated, among other factors, with

213 differences in the rumen microbiome associated with between-animal variation in

214 passage rates, rumen volume and morphology, eating behaviour, etc. (Kittelman et al.,

215 2014). Martínez et al. (2010) observed consistent differences in CH4 production and H2

216 recovery in vitro during incubations with rumen liquor collected from 6 different sheep

217 fed the same diet. Such differences can only be explained by differences in microbial

218 populations or activities in the starting inocula or variation in the survival or activity of

219 microbes over the incubation period. To reduce the effects of unusual rumen inocula, it

220 has been suggested that, normally, at least 3 animal sources should be used to provide a

221 representative source of rumen inoculum (Editorial, Animal Feed Science and

222 Technology, 2012). However, this might be difficult when large ruminants are used.

223 Another issue to consider is the number of independent incubation runs to be conducted

224 in different days, which should be at least 3. Unfortunately, there is no work available

225 assessing in statistical terms the minimum number of donor animals needed to provide a

226 representative source of rumen inoculum.

227

9
228 RECOMMENDATION: Where possible the target animal species should be used as the

229 donor of rumen fluid. Sampling before feeding is advantageous for minimizing diet by

230 animal interactions. We recommend using 3 or more (optimum) or 2 (minimum)

231 animals as donors of rumen inoculum and at least 3 independent incubation runs.

232

233 3.2. Donor animal diet

234 Diet composition and nutrient intake are major factors affecting both microbial

235 populations in the rumen and microbial activity of rumen inoculum (Mould et al.,

236 2005). Compared with ruminants fed high-concentrate diets, a greater proportion of

237 fibrolytic bacteria and methanogenic archaea can be expected in rumen fluid collected

238 from animals on high-forage diets (Demeyer and Fievez, 2000). However, the extent to

239 which concentrate feeds affect rumen digestion and microbial populations may depend

240 on the source and proportion of concentrate ingredients in the diet, as well as forage

241 quality (Dijkstra, 1994).

242 Martínez et al. (2010) assessed the effect of feeding sheep diets differing in

243 forage:concentrate ratios (F:C; 70:30 vs 30:70) and forage source (alfalfa hay vs grass

244 hay) on rumen fermentation and CH4 production. In vitro CH4 production (per g of

245 incubated DM) was increased by decreases in the F:C ratio of diets fed to donor animals

246 or when Lucerne hay was replaced by grass hay. Differences in the F:C ratio altered

247 pH and the activity of certain glycoside hydrolases (carboxymethylcellulase, xylanase

248 and amylase) in rumen fluid. Forage type also influenced NH3 content and

249 carboxymethylcellulase activity in rumen fluid. These results suggest that feeding donor

250 animals a diet similar to the substrate to be incubated in vitro may be advantageous.

251 Of particular importance is the observation that some mitigation strategies have

252 been evaluated using rumen fluid from animals fed poor quality feeds. This raises the

10
253 question of whether outcomes on the efficacy of mitigation agents under these

254 circumstances can be considered reliable when donor animals have not been fed a diet

255 of similar characteristics as that offered to target animals. The effects of differences in

256 diet composition fed to donor animals may be minimized by obtaining rumen fluid

257 immediately before feeding. Huntington and Givens (1998) obtained fluid from cows

258 fed either a silage:barley diet (80:20) or a barley straw diet. Although microbial activity

259 of rumen fluid collected on the straw diet was lower than the silage:barley diet, this did

260 not alter the gas production profile, which was attributed to sampling prior to morning

261 feeding.

262 The diet fed to donor animals also needs to be considered when testing additives

263 as CH4 mitigation agents. Mateos et al. (2013) reported that the effect of garlic oil and

264 cinnamaldehyde on in vitro fermentation and CH4 production varied depending on

265 whether the donor animals were fed a typical dairy diet (alfalfa hay:concentrate 50:50)

266 or a fattening diet (barley straw: concentrate 15:85). Observations from several in vitro

267 studies suggest that the effects of essential oils on rumen function are pH-dependent,

268 and this also appears to be true for garlic oil and some of its components (Cardozo et al.

269 2005; Kamel et al. 2008). Cardozo et al. (2005) found that garlic oil had a more

270 pronounced impact on rumen volatile fatty acids (VFA) profile at low compared with

271 high rumen pH (5.5 versus 7.0), an effect explained by differences in the status of the

272 active molecules (i.e. dissociated or un-dissociated) possibly mediated by changes in

273 rumen pH. However, batch cultures are usually highly buffered systems allowing

274 ruminal microorganisms to grow for a prolonged period despite the accumulation of

275 fermentation end-products. In such cases, factors other than pH (e.g. microbial

276 composition) may explain differences in the efficacy of additives. Hatew et al. (2015)

277 provided further evidence on the importance of diet fed to the donor animal.

11
278 Experiments involved the incubation of the same substrate (grass silage or beet pulp)

279 with rumen inoculum obtained from donor cows fed on diets that differed in starch

280 source (native vs. gelatinized maize grain) and starch level (270 vs 530 g/kg concentrate

281 DM). A higher level of starch and gelatinized rather than native maize were found to

282 lower gas and CH4 production after 24-h incubations.

283 Level of feed intake is also an important consideration, given that a higher DM

284 intake (DMI) lowers retention time in the rumen, decreasing the amount of time

285 available for feed degradation of feeds and hence ruminal digestibility (Clauss et al.,

286 2007). Rumen pH, proteolytic and cellulolytic activities are thought to be influenced by

287 the level of DMI which can in turn influence growth rates and the metabolic activity of

288 inoculum used in in vitro systems. Increasing feeding frequency will generally lower

289 diurnal variation in rumen fermentation parameters. For example, in lactating cows

290 increasing feeding frequency from two to six times-daily was found to decrease post-

291 feeding variation in rumen pH, osmolality, VFA and NH3 concentrations (Le Liboux

292 and Peyraud, 1999).

293 The period of adaptation to a given diet by the donor animal probably needs to

294 be revisited. It is common to collect rumen fluid from animals fed a diet for 2 weeks.

295 However, there are indications that the methanogenic archaeal population requires an

296 adaptation period of around 30 days after a change in diet (Williams et al., 2009). Also

297 Monteils et al., (2012) reported that protozoa counts in the rumen needed 25 days to

298 stabilize after different dietary changes. Given the lower generation time that protozoa

299 and archaea have as compared to bacteria (Dehority, 2003), further research is required

300 to assess the effect of dietary treatments on the adaptation time needed for

301 methanogenesis tests.

12
302 RECOMMENDATION: The diet fed to donor animals should be similar in composition

303 to the substrate incubated in vitro. Care should be taken to ensure sufficient buffering

304 capacity when investigating diets or dietary ingredients or additives that promote

305 differences in rumen pH (see section 3.4), in particular when samples are not taken

306 immediately before feeding. It is recommended that donor animals are fed a diet at a

307 restricted level of feeding as frequent meals to ensure constancy of diet composition and

308 digestion, minimize variation in feed intake and avoid diurnal variation in rumen

309 fermentation.

310

311 3.3. Rumen fluid sampling: time, location and processing

312 In all in vitro fermentation systems it is essential to create an environment, which, for

313 any set of parameters, mimics the fermentation in a specific section of the gastro-

314 intestinal tract in vivo (e.g. reticulo-rumen or caecum). Therefore, the inoculum should

315 be representative of that environment with respect to both the composition and

316 abundance of the microbial population. For in vitro systems to be robust (i.e.,

317 reproducible over time and representative of conditions in vivo), the inoculum must

318 meet certain criteria. Making a valid assessment of whether a given study has met these

319 criteria may be problematic, as often-essential information is not reported. Given the

320 precision of gas release kinetic techniques relative to degradability at a set end-point

321 over extended periods, variations in inoculum characteristics due to host animal effects,

322 nutrition and sampling time, as well as sample preparation and inoculation, can have

323 substantial cumulative effects on in vitro fermentation. It seems pertinent, not only to

324 permit comparison between studies, but also to limit potential errors, to have a set of

325 accepted guidelines and standard procedures for preparing inoculum for measuring CH4

326 in vitro, as proposed for animal studies in vivo. Such guidelines should include host

13
327 animal management, sampling techniques (time, location, alternatives) and inoculum

328 preparation.

329

330 Sampling time:

331 Diurnal changes of the rumen microbiome, both in terms of abundance and metabolic

332 activity have been documented. Concentrations of viable microbial populations in the

333 rumen typically decrease 4 h post-feeding, due to dilution with feed, water and saliva,

334 and peaks at 6-12 h post feeding (depending on diet and level of feed intake) (Leedle et

335 al., 1982; Dehority, 2003). Furthermore, microbial abundances were found to decline

336 during the degradation of available nutrients. Cone et al. (1996) observed that the rate of

337 fermentation was highest when rumen fluid was collected after morning feeding,

338 although rumen sampling time had no effect on the total gas production. Menke and

339 Steingass (1988) stated that sampling rumen contents just before feeding lowered

340 variation in composition and activity of the inoculum and minimized the influence of

341 diet fed to donor animals. However, Payne et al. (2002) observed that the total gas

342 production from both starch and ground straw were less variable between replicate

343 bottles and between weeks of collection when rumen fluid inoculum was collected

344 either 4 or 8 h post feeding, compared with samples collected just before, or 2 h after

345 feeding. Presumably, the activity of the inoculum is determined just as much by time of

346 rumen sampling relative to feeding as by feeding pattern and eating time. Furthermore,

347 microbial diversity and activity is at its lowest before feeding and while this reduces the

348 variability, it may not reflect the ‗true‘ effect that the feed/additive may have on much

349 diverse population and their activities present after feeding. This deserves further

350 research in the future.

351

14
352 RECOMMENDATION: There is no general recommendation on the ideal rumen

353 sampling time as it depends on the objectives of a specific experiment. Given the

354 difficulty of minimizing diurnal variation, rumen fluid samples are best collected

355 immediately before feeding, based on a consistent protocol for dietary access by donor

356 animals across experiments. When a series of studies are conducted over time, feeding

357 and sampling procedures should be kept as identical as possible.

358

359 Sampling procedure: Rumen digesta is comprised of different fractions (large and small

360 particulate matter and liquid). Ample evidence exists on the different abundance and

361 diversity of bacteria associated with the liquid and solid ruminal contents (Pei et al.,

362 2010). The pH and VFA concentration varies between different sections of the reticulo-

363 rumen in cattle (Bryant 1964). De Visser et al. (1993) indicated that rumen VFA

364 concentrations were about 20% higher for the entire evacuated rumen content compared

365 with calculations based on samples of rumen fluid collected in a standardized manner.

366 Storm and Kristensen (2010) indicated differences of 0.4 to 0.6 pH units and of 40 to 50

367 mM VFA between the central and ventral regions of the rumen, with the lowest pH and

368 highest VFA concentrations in the medial rumen. Rumen fermentation parameters have

369 also been found to differ between samples collected at different locations in the rumen

370 (Shen et al., 2012), implying possible differences in microbial abundance and activity

371 within the rumen.

372 Rumen cannulation is considered the reference method allowing the collection of

373 representative samples of rumen digesta from donor animals (Komarek, 1981;

374 Kristensen et al., 2010). Access to surgically-modified animals is not universal, and

375 therefore less invasive techniques, such as oral stomach probing, have been used as an

376 alternative. In the relatively few studies that have compared sampling through the

15
377 rumen cannula or by stomach probing, differences in fermentation profile and

378 microbiota have been reported in some (e.g., Geishauser and Gitzel, 1996; Duffieldet

379 al., 2004), but not all cases (e.g., Lodge-Ivey et al., 2009; Shen et al., 2012; Terré et al.,

380 2013). Part of the discrepancy between studies may reflect differences in the procedures

381 used to avoid salivary dilution and contamination, the type of samples collected and

382 rumen sampling site. Stomach probing results in the collection of samples containing a

383 high proportion of liquid, whereas sampling via a rumen cannula allows both solid and

384 liquid digesta fractions to be obtained. Differences in the methods used to collect rumen

385 samples are of greater relevance when treatments are not expected to have the same

386 effect on microbial populations attached to solids or inhabiting the liquid phase

387 (Martínez et al., 2010). The study of Shen et al. (2012), attributed the differences

388 between samples collected via cannula or stomach tube to rumen sampling site, as a

389 consequence of the probe not being inserted to a depth sufficient to reach the ventral

390 sac. Accurate probe insertion to a desired location within the rumen is extremely

391 challenging in small ruminants. In a recent study, Ramos-Morales et al. (2014) found

392 that stomach-tubing in sheep and goats detected the same differences in rumen

393 fermentation due to species, diet or sampling time as sampling via the cannula.

394 However, certain differences were more readily detected in rumen cannula samples,

395 while substantial differences in the bacterial community structure were detected

396 between the sampling methods.

397 Faeces has also been used as an alternative source of inoculum to rumen fluid

398 (El Shaer et al., 1987). Cultures of ruminal or faecal microorganisms appear to result in

399 similar fermentation processes (El-Meadaway et al., 1998). However, fewer

400 microorganisms in faecal inocula may result in lower degradation capacity and

401 decreased gas production (Cone et al., 2002; Vàradyovà, et al., 2005), a longer lag

16
402 phase and a slower rate of degradation at the outset (Mauricio et al., 2001).

403 Furthermore, for poor-quality forages, there is only a weak relationship between gas

404 production during incubations with inocula sourced from faeces and rumen fluid (El-

405 Meadaway et al., 1998; Varadyova et al., 2005). Dhanoa et al. (2004) proposed a

406 method allowing mathematical adjustments to convert or translate the degradation

407 profiles produced by faecal inoculum to correspond with ruminal fluid, but this requires

408 application of different prediction equations for each group of feeds incubated.

409 Rumen contents collected post-morten at abattoir can also be used as an

410 alternative to rumen fluid (Mould et al., 2005). Several IVFT have been performed

411 using rumen fluid collected from slaughtered cattle, sheep, buffalo and dromedary

412 (Haddi et al., 2003; Salem, 2005). To date, there are no reports directly comparing the

413 use of rumen fluid from slaughtered animals with oral or rumen sampling in the same

414 animal. Such an approach requires sampling of rumen contents soon after slaughter, as

415 well as the same criteria for other sources of inocula being met. While the intake and

416 diet composition of slaughtered animals are not known, access to entire rumen contents

417 allows the collection of representative samples that can also be used to inform on

418 nutrient supply. If sampling of rumen cannulated animals is not possible, collection of

419 rumen content from slaughtered animals may prove a viable alternative.

420

421 RECOMMENDATION: To be as representative of the rumen environment as possible,

422 samples of ruminal contents for the preparation of in vitro inoculum need to be

423 collected from several locations. This is more feasible in large ruminants and requires

424 the collection of rumen contents from animals fitted with rumen cannula following a

425 clearly defined and standardized sampling protocol. Stomach tubing or faecal inoculae

426 may serve as an alternative for ranking purposes, but quantitative data using these

17
427 alternatives may differ from sampling of rumen contents in cannulated animals.

428 Stomach tubing should be performed by well-trained persons to minimize salivary

429 dilution.

430

431 Preservation of inoculum: Anaerobiosis is essential to culture rumen microorganisms,

432 methanogenic archaea, in particular (Joblin, 2005). Certain experiments may require

433 rumen inoculum to be stored until culturing in vitro. Short-term storage (<1 h), for

434 example during transport from donor animals to the IVFT laboratory, should exclude

435 exposure to air. It is equally important that any increase in headspace pressure does not

436 cause CO2 to go into solution, thus lowering pH. Excessive fermentation due to

437 extended storage at 39°C should be restricted to prevent any microbial group from

438 becoming dominant and modifying the composition of the original inoculum.

439 The influence of storage time and temperature on neutral detergent fibre (NDF)

440 degradation by rumen microorganisms has been investigated (Robinson et al., 1999).

441 Studies involved the use of the ANKOM end-point system to examine the effect of

442 delaying inoculum storage at 39ºC by up to 6.5 h, or up to 48 h after storage over a

443 range of temperatures from – 22 and 39 °C. No apparent effect on 48 h end-point

444 degradation of medium-term (6.5 h) storage compared with long-term storage (48 h)

445 was identified. Authors concluded that no storage method, irrespective of temperature,

446 would maintain rumen inoculum activity for up to 48 h that support normal

447 fermentation in vitro. Subsequent works (Cone et al., 2000; Hervás et al., 2005; Prates

448 et al., 2010) presented data on the effect of using rumen fluid directly or stored

449 anaerobically at 39°C for increments of up to 24 h or at –24°C for 1, 3, 10, 40 or 76

450 days before use. In general, gas production in terms of kinetics and cumulative yields,

451 decreased as storage period increased. Final gas volumes were similar when rumen fluid

18
452 was stored up to 4-6 h compared with no prior storage. Gas production rates were lower

453 for inocula stored for 8 or 24 h, while gas production was considerably decreased by

454 extended storage at -20ºC of more than 10 days. Microbial activity was lowered by

455 freezing, with the decrease being substrate-dependent with the degradation of pure

456 starch and cellulose being less affected compared with lucerne hay and barley straw.

457 Freezing in liquid nitrogen is preferred over storage at -20ºC (Prates et al., 2010) and

458 thawing of small volumes (approximately 20 ml) at 39ºC for 2 min. The time taken for

459 freezing and thawing appears to be as equally important as storage temperature. Overall,

460 studies suggest that preservation of rumen fluid at 0ºC for up to 6 h offered a practical

461 alternative, where necessary, to freshly collected inocula. Protozoa are lost after

462 freezing which could have an impact on the fermentation. There are, however, no

463 reports documenting the effect of different preservation methods on methanogenic

464 activity.

465

466 RECOMMENDATION: Fresh rumen fluid maintained under anaerobic conditions at

467 39ºC should be inoculated into in vitro vessels as soon as possible, ideally within 1 h

468 post collection. When this is not possible, rumen fluid can be preserved at 0-4 ºC for up

469 to 6 h or frozen in liquid nitrogen following addition of a cryo-protectant (i.e. 15%

470 glycerol) for longer periods for use as inoculum. In either case, implementation of

471 standardized procedures to avoid undesirable variation in microbial activity is highly

472 recommended.

473

474 Preparation of inoculum prior to incubation: Method of preparation also influences the

475 microbial activity of rumen inoculum. The rumen microbiome consists of three sub-

476 populations of microbes: those in the fluid phase, adherent to the particulate phase

19
477 (further divided into loosely- and firmly-associated with the feed particles) (Cheng et

478 al., 1993) or attached to the epithelium (Sadet et al., 2007). The latter tends to be

479 primarily involved in the release of ammonia from urea absorbed across the rumen

480 epithelium and, as such, has only a minor role in feed degradation, and for this reason

481 does not need to be sampled (Mueller et al., 1984). Fluid and particulate associated

482 bacterial populations (Kim et al., 2004) and methanogenic archaea (Shin et al., 2004)

483 differ in growth characteristics and in the activities of most enzymes (Moharrery and

484 Das, 2001). Microorganisms in rumen liquid (20 to 30% of total microbes) including

485 free-living bacteria and bacteria detached from solid substrate, have little direct

486 involvement in structural carbohydrate digestion (Miron et al., 2001). Microbes attached

487 to feed particles predominate (70 to 80% of microbial matter and microbial ATP

488 production) and play a key role in feed particle digestion in the rumen (Miron et al.,

489 2001; Trabalza-Marinucci et al., 2006). Consequently, once samples of whole rumen

490 contents have been collected, the problem arises of how to effectively detach the

491 microorganisms associated with feed particles. Failure to do so will result in a high

492 proportion remaining to be attached after filtration, while the use of multiple layers of

493 cheesecloth, muslin or surgical gauze leads to different microbial fractions being

494 retained and inoculated in vitro. For example, use of a cloth with 50 µm pores would

495 substantially lower the number of large protozoa (which range from to 15–250 µm in

496 size, Dehority, 2003) in inocula. Williams and Coleman (1991) reported that most

497 protozoa are retained using a 100 µm pore size filter. Therefore, if total protozoa are to

498 be included in the microbial culture, a 250 µm pore size cloth should be used.

499 Furthermore, some of the physical methods used to detach particle-associated microbes

500 (e.g., a stomacher or maceration of rumen contents in a food processor) may also cause

501 cell damage. Rymer et al. (1999) examined four methods of inoculum preparation:

20
502 strained, blended fluid, ―stomacher‖ or strained plus blended residues. No consistent

503 treatment effect was observed, apart from blending which tended to decrease substrate

504 degradation. Authors concluded that there was little advantage from blending inoculum,

505 particularly in light of the risk of exposing microorganisms to oxygen. Even though

506 homogenizing rumen fluid may increase numbers of particle-associated bacteria in the

507 inoculum, Pell and Schofield (1993) excluded this step on the grounds that it i)

508 introduced an extra procedure into the laboratory protocol, ii) increased the risk of

509 exposing rumen microorganisms to oxygen, iii) increased the quantity of gas released

510 from the blanks, and iv) had no obvious effect on the results of IVFT.

511 A number of techniques that have been proposed for detaching microorganisms

512 from rumen feed particles that involve various combinations of chemical and physical

513 treatments and h yield removal rates of between 20 and 80% (Hristov et al., 1999;

514 Ranilla et al., 2001). While the use of these techniques is essential when collecting

515 microbial biomass to determine their composition to accurately assess passage of

516 nutrients of microbial origin to the intestine, it is not clear whether a standard protocol

517 for microbial detachment should be applied for in vitro methods. In recent work, Soto et

518 al. (2013) studied the development of the microbiota in different in vitro rumen

519 simulation systems inoculated with intact or filtered rumen fluid from goats. Incubation

520 of filtered rumen fluid fraction in batch culture resulted in lower microbial diversity

521 compared with non-filtered rumen fluid inoculum. Substantial growth of fibrolytic

522 bacteria and methanogenic archaea over the first 24 h partially compensated for low

523 numbers in the inoculum due to filtering.

524

525 RECOMMENDATION: Filtration of rumen fluid using the same pore size across

526 incubation runs is a straightforward method for preparing inoculum suitable for in vitro

21
527 experiments. A larger pore size results in greater numbers of small particulate

528 associated bacteria and protozoa. A pore size of 250 m is recommended. Use of

529 multiple layers of cheesecloth is not recommended due to inconsistencies in pore size.

530

531 3.4. Substrate and incubation buffer

532 Substrate:

533 Provided donor animals are fed the same or a similar diet as that tested in vitro, there

534 remains an uncertainty of the choice of substrate to be used in the evaluation of

535 additives on CH4 production in vitro. The composition of diets fermented in vitro

536 determines the production of dissolved H2 that serves as a substrate for methanogens.

537 Most in vitro studies have tested additives in incubations with a single substrate, but

538 there are reports on the effects of additives using different fermentation substrates in a

539 single experiment. Machm ller et al. (2001 investigated effects of medium chain fatty

540 acids in incubations containing high or low amounts of fiber. The efficacy of monensin

541 has been examined using corn meal or timothy hay (Russell and Strobel, 1988) and corn

542 meal or soyabean hulls (Pellikaan et al., 2011) as substrates. In certain instances a

543 substrate by additive interaction has been observed. Pellikaan et al. (2011) reported a

544 complete inhibition of CH4 production during the first 30 h of incubation, irrespective

545 of substrate composition (soyhulls or maize), with bromoethanesulphonate (BES) and

546 cinnamaldehyde. However, CH4 production from soybean hulls with BES or

547 cinnamaldehyde was 65% lower compared with soyhulls without additive after 72 h,

548 suggesting an influence on adaptation, whereas no CH4 was produced after 72 h

549 production when maize was incubated with these additives. Subsequent studies covering

550 a range of plant extracts and fatty acids (Castro-Montoya et al., 2012; Klevenhusen et

551 al., 2012; O‘Brien et al., 2013) have provided further evidence that responses of CH4 to

22
552 a given additive differs depending on the feed substrate incubated. For example,

553 addition of fatty acids were found more effective in lowering CH4 during incubations

554 containing higher proportions of concentrate ingredients, an effect attributed to greater

555 protonation of fatty acids at a lower pH (e.g. Zhou et al. 2015). While specific

556 experiments have provided examples of substrate by additive interactions, drawing firm

557 conclusions on the magnitude of these effects remains challenging. Nevertheless, there

558 is a need for end users to recognize that specific characteristics of incubated substrates

559 impact on the outcomes of IVFT and be aware of a possible mismatch between the diet

560 of the donor animal and incubated substrate.

561 Implementing a standardized protocol for preparing substrates to be incubated is

562 also critical in allowing for between IVFT comparisons (Rymer et al., 2005). The most

563 critical issue appears the methods used to dry fresh material, such as grass. Comparisons

564 of freeze-drying with oven drying at 60 or 105ºC are often contradictory. There are

565 reports on the effect of feed processing on CH4 production. Nevertheless, a priori

566 freeze-drying is the method of choice for minimizing cell damage that potentially alters

567 the dynamics of microbial attachment, substrate degradation and altering bioactive

568 compounds (Rymer et al., 2005).

569 With regards to particle size, using cereals as substrates resulted in particle size

570 having little effect on in vitro fermentation, provided that the grain kernel had been cut

571 (Lowman et al., 2002), although some evidence to contradict this argument exists with

572 maize (Rymer et al., 2005). With fibrous and more slowly degraded feeds, fermentation

573 rate increases as particle size decreases (Lowman et al., 2002) and it seems likely that

574 this is a consequence of increased surface area as a result of grinding, thereby allowing

575 better microbial access. The most common procedure is to mill the substrate through a 1

576 mm screen and store under dry, cool and dark conditions in sealed containers prior to

23
577 use. Adoption of a standardized approach to sample preparation may enable comparison

578 between independently produced in vitro fermentation data of different feeds.

579

580 RECOMMENDATION: It is recommended that a range of substrates that reflect the

581 types of feeds used in commercial production systems are used in the initial screening of

582 new additives, unless the objective of an in vitro experiment requires a predefined

583 substrate. Freeze-drying is preferred to oven drying for the drying of high moisture

584 substrates.

585

586 Incubation medium and rumen fluid:medium ratio (RF:M)

587 There is considerable variation in the composition of the medium used for in vitro

588 studies reported in the literature. It is important to make a distinction between the term

589 ‗medium‘ (i.e., a solution containing a number of components including buffering

590 agents, trace elements, true protein and reducing agents and ‗buffer‘ (Williams, 1998 .

591 The types of buffers used in IVFT and the implications on fermentation has been

592 comprehensively reviewed (Rymer et al., 2005). A medium with a high buffering

593 capacity, when used in IVFT, may be disadvantageous because it creates conditions that

594 are not representative of the rumen in vivo. This is particularly important when

595 assessing the effectiveness of CH4 mitigations strategies that may rely on a decrease in

596 rumen pH that include an inhibition of fibrolytic bacteria and/or methanogens (Argyle

597 and Baldwin, 1988; Van Kessel and Russell, 1996; Navarro-Villa et al., 2011).

598 Furthermore, through the prevention of a sharp decline in pH, a highly buffered medium

599 may increase acetic:propionic acid ratios more than would otherwise occur in vivo,

600 which impacts on the availability of H2 for CH4 production (Lana et al., 1998).

24
601 The effect of the incubation medium and rumen fluid to medium ratio (RF:M)

602 on IVFT has been investigated (Rymer et al., 1999; Pell and Schofield, 1993; Cone et

603 al., 2000). Such studies have demonstrated that increases in the proportion of rumen

604 fluid is associated with a decrease in lag phase and a higher rate of gas production,

605 while the effect on total gas production varies. Navarro-Villa et al. (2011) investigated

606 the effect of variable RF:M ratios (1:2, 1:4, and 1:6) in IVFT involving the incubation

607 of different amounts (0.3, 0.5, and 0.7 g) of three contrasting feeds (barley grain, grass

608 silage and barley straw) . The results indicated that CH4 per unit of DM degraded was a

609 more appropriate unit for expressing in vitro CH4 output than CH4 per unit of DM

610 incubated. Incubation of 0.3 g dried milled feed in 50 ml of in vitro culture containing

611 1:2 RF:M was an acceptable combination for 24 h incubations allowing for a decline in

612 pH declined and maximizing the difference in CH4 output between substrates.

613 For a given substrate, pH in the incubation vessel should ideally mimic that in

614 the rumen, i.e. between 6.0-7.0 for forage based diets and 5.5- 6.0 for concentrate based

615 diets. Patra and Yu (2013) evaluated IVFT containing different bicarbonate

616 concentrations (80, 100, and 120 mM) in buffer. Results indicated that bicarbonate

617 concentrations above 80 mM should be avoided to minimize non-microbial CO2

618 production associated with changes in pH. A recent comparison of two buffers

619 commonly used in IVFT (McDoughall‘s (McDoughal, 1948) and Mould‘s (Mould et

620 al., 2005) buffer indicated that buffer composition had no effect on total gas production

621 (Muetzel et al., 2014). These findings are in direct contrast to earlier reports that a

622 higher phosphate concentration decreased gas production over a 9 h fermentation period

623 (Mould et al., 2005). The differences the latter authors reported, however, were about

624 4% and decreasing over incubation time. A trend towards higher gas production for

625 McDoughall buffer compared with Mould‘s buffer may be related to a higher carbonate

25
626 concentration (Muetzel et al., 2014). The buffer composition had no effect on CH4

627 production or on the percentage of CH4 released or on VFA production. However, the

628 proportions of major VFA were altered; molar proportion of acetate was higher and that

629 of propionate was lower for incubations with Mould´s buffer compared with

630 McDougall´s buffer. It is possible that differences in the phosphate to carbonate ratio

631 and associated changes in pH may be responsible (Broudiscou et al., 1999). Based on

632 the work of Kohn and Dunlap (2008) a recent report demonstrated that by adjusting the

633 concentration of buffer bicarbonate, pH can be reasonably well controlled at specific

634 target pH (6.50, 6.25, 6.00, 5.75 and 5.50) during 12 h incubations (Amanzougarene et

635 al., 2015). What is less clear is whether different antimethanogenic additives perform

636 similarly in IVFT over a range of pH.

637

638 RECOMMENDATION: Available data do not allow for recommendations on an ideal

639 RF:M. A ratio of 1:2 appears to generate the most reliable results for 24 h incubations.

640 However, this ratio and the amount of substrate incubated needs to be considered on

641 the basis of the frequency of gas sampling and the duration of the incubation depending

642 on the research objectives. Bicarbonate concentration in buffer may influence

643 methanogenesis. To minimize non-metabolic CO2 production use of buffers containing

644 bicarbonate concentrations above 80 mM should be avoided. Furthermore, adjusting

645 the concentration of bicarbonate in the buffer or the RF:M ratio offers the possibility of

646 setting a target pH according to the substrate incubated.

647

648 3.5. Incubation procedure and CH4 measurements

649 In vitro gas production systems are typically conducted over intervals of between 16 to

650 72 h. As such, IVFT do not mimic important physiological processes in vivo such as

26
651 ruminal digesta turnover. Removal of soluble particles in the liquid medium may have

652 adverse effects on microbial fermentation, by decreasing the amount of soluble

653 substrate available for microbial growth or may conversely stimulate activity (Roger et

654 al., 1990). Soto et al. (2013) reported that the numbers of all quantified microorganisms

655 (total bacteria, protozoa, methanogens, fungi, Fibrobacter succinogenes and

656 Ruminococcus flavefaciens) declined sharply during 24 h to 72 h of incubation. This is

657 likely due to the exhaustion of fermentable substrate and the accumulation of

658 fermentation end products. Different substrates (soluble carbohydrates, starch, pectins,

659 cellulose, hemicellulose and protein) that are fermented can be degraded at different

660 rates also have a variable contribution to CH4 production (Bannink et al., 2006). In

661 cows, mean retention time of NDF components in the reticulorumen is around 28 h

662 (using external markers, Schwarm et al., 2015) or 28 to 60 h (using intrinsic stable

663 isotope labelling techniques, Warner et al., 2014), being longer than the standard in

664 vitro incubation time of 24 h. Retention time in the reticulorumen is related to particle

665 size (Schwarm et al. 2008). It can be argued that NDF fermentation during in vitro

666 incubations probably approaches a plateau after 24 h given that relatively small feed

667 particles (ca. 1 mm) are usually incubated.

668 Some IVFT have tested the effects of different substrates on CH4 production based on

669 the collection of a single gas sample after 24 h (García-González et al., 2008). Such an

670 approach may result in gas pressure in the headspace exceeding a given threshold (48

671 kPa) and consequently an impairment of microbial activity (Rymer et al., 2005;

672 Taglapietra et al., 2010). Several protocols, such as those described by Theodorou et al.,

673 1994, Cone et al., 1996 and Davies et al., 2000, stipulate that headspace gases being

674 released at pre-determined intervals or when a pre-set threshold of pressure is reached

675 (Muetzel et al., 2014). Venting the gas produced requires that CH4 concentration is

27
676 measured simultaneously, given that different CH4 concentrations can be expected

677 depending on substrate. Although the most common method to measure CH4

678 concentration on gas samples collected from incubation vessels is by using gas

679 chromatography, as for in vivo measurements, other techniques are available and have

680 been used such as infrared methane analysis (Goel et al., 2008, Cobellis et al., 2015)

681 and absorption of CO2 (Fievez et al., 2005).

682 For successful IVFT, it is essential that experimental treatments be randomly

683 allocated to bottle positions. The order in which bottles are inoculated also needs to be

684 randomized across bottle positions and treatments. Randomization serves to minimize

685 possible confounding effects of bottle position (or water baths used), treatments and

686 timing of inoculation. Such an approach is analogous to the random allocation of

687 treatments to animals for in vivo experiments. It is also important to consider

688 establishing incubations with a single, identical source of medium, and a single,

689 identical source of substrate. Finally, independently of the number of independent

690 incubation runs performed, different replicates (bottles) of each treatment need to be

691 included. Three bottles minimum is sensible as it allows possible outliers to be

692 identified.

693

694 RECOMMENDATION: The duration of the incubation should be adjusted based on the

695 composition and physical properties of substrate incubated that determines the

696 frequency of gas sampling required for measuring CH4 production. Gas composition

697 should be determined at the same time gas pressure is vented. It is highly desirable that

698 the timing of inoculation and the allocation of treatments with respect to bottle position

699 are randomized as much as possible. It is recommended that 3 bottles be used per

700 unique treatment in each incubation run.

28
701

702 3.6. Headspace gas composition

703 It is well established that H2 concentration can affect the thermodynamics of

704 fermentation and the growth rate of hydrogenotrophic methanogens in the rumen

705 (Janssen, 2010). Hydrogen produced in the rumen is present in two forms, as dissolved

706 H2 and as H2 gas, but methanogens only utilize dissolved H2 (Wang et al., 2014).

707 Reports on the influence of headspace gas composition on in vitro gas production and

708 rumen fermentation are scarce. In one study, Patra and Yu (2013) noted that initial CO2

709 headspace, but not N2 headspace, was positively correlated with CH4 production after

710 fermentation. This prompted the hypothesize that headspace gas composition, CO2 in

711 particular, which is in exchange with H2CO3/HCO3− in the medium, depending on

712 concentration, acid-base balance and gas pressure, may affect fermentation

713 characteristics and gas production in ruminal in vitro cultures. A range of initial

714 headspace composition of in vitro cultures has been reported, including 100% CO2

715 (Anderson et al., 2003; Weimer et al., 2005), 100% N2 (Hoover et al., 1976) and a

716 mixture of gases typical of an anaerobic chamber (85% N2, 10% H2, and 5% CO2; Zhou

717 et al., 2011; Patra et al., 2012). Patra and Yu (2013) investigated the effects of three

718 different headspace gases (N2 + CO2 + H2 in the ratio of 90:5:5, 100 % CO2, and 100 %

719 N2) and the interaction with type of substrate (alfalfa hay or alfalfa hay and concentrate)

720 and media bicarbonate concentration on gas and CH4 production. Methane production

721 was much higher when CO2 was present in the headspace. It is conceivable that

722 equilibrium is established between CO2 dissolved in the inoculum and CO2 in

723 headspace gas (Alford, 1976), such that a higher concentration of CO2 in the headspace

724 would result in a greater concentration of dissolved CO2 in the media. Higher CH4

725 production corresponding to CO2 in the headspace may be explained by an immediate

29
726 and greater availability of CO2 in the inoculum that serves as the electron acceptor for

727 the primary hydrogenotrophic methanogenesis pathway. An increase in dissolved CO2

728 may also promote growth and activity of methanogens. Even though CH4 production in

729 the study of Patra and Yu (2013) was greater when the headspace contained CO2 rather

730 than N2, total or net gas production was lower for the former compared with the latter.

731 Further investigations are required to understand the impact of a mixture of CO2 and N2

732 that best mimics rumen gas composition in vivo.

733

734 RECOMMENDATION: Owing to the effect of headspace gas composition on gas

735 production, including CH4 production, it is recommended that all future studies should

736 both consider this as an influencing factor and report headspace gas composition.

737 Rumen fluid should be flushed continuously, and once added into the medium,

738 continued to be flushed for at least 10 min before incubation is initiated. Following the

739 addition of inoculum into the bottle, the headspace should be flushed continuously until

740 the bottle is sealed.

741

742

743

744 Please insert Table 2 around here

745

746 3.7. In vitro versus in vivo

747 Numerous studies have examined the influence of antimethanogenic compounds on

748 CH4 production in vitro, but few have undertaken a simultaneous evaluation in vivo and

749 in vitro. Direct comparison of effects in vitro and in vivo would allow a better

750 interpretation of IVFT data and inform on the treatments suitable for further evaluation

30
751 in vivo. When addressing inconsistences between results from in vitro and in vivo

752 studies it is worthwhile considering:

753 i) The accuracy of in vitro systems to predict the CH4 production of a given

754 diet per unit feed intake or digested matter

755 ii) The ability to simulate the direction of changes (not absolute values) in CH4

756 production when anti-methanogenic agents are tested relative to an

757 appropriate control.

758 i) Flachowsky and Lebzien (2012), using data from Moss and Givens (1997)

759 reported a poor relationship (r2=0.264) in CH4 production obtained in vivo and in vitro

760 methods. Bhatta et al. (2007) compared IVFT measurements of CH4 production with the

761 SF6-technique across a range of diets. Methane production (ml/g DM) estimated from

762 48 h in vitro gas production was higher than measurements in vivo for all diets. Of

763 particular note is that the average of CH4 production at 24 h and 48 h was closely

764 correlated with values based on SF6 (R2=0.78, 5 diets and 4 animals used) (Table 2).

765 Blümmel et al. (2005) conducted a study to compare feed intake, digestibility and CH4

766 production by open-circuit respiration measurements in sheep fed 15 untreated, sodium

767 hydroxide (NaOH) treated and anhydrous ammonia (NH3) treated wheat, barley and oat

768 straws also evaluated using IVFT. Total daily CH4 production, calculated from in vitro

769 fermentation characteristics (i.e., true degradability, SCFA ratio and efficiency of

770 microbial production) and OM intake were found to be closely correlated with CH4

771 emissions (L/d) measured in respiration chambers (y = 2.5 + 0.86x, R2 = 0.89, P <

772 0.001, 15 diets and 4 animals used; y=CH4 production in vivo; x=CH4 predicted in

773 vitro). Intake of OM measured in vivo OM intake was also used to calculate CH4

774 production in vitro. As such OM intake was common to both the independent and the

775 dependent variable, which could explain the close association between in vitro and in

31
776 vivo measurements. It is important to point that Blümmel et al. (2005) and Bhatta et al.

777 (2007) did not use the same animals from the in vivo trials as donor of rumen fluid for

778 the in vitro incubations.

779 More recently, Hatew et al. (2015) reported a study comparing measurements of CH4

780 production in vitro and in vivo, and those were conducted simultaneously (animals

781 adapted to the same substrate as incubated in bottles were used as a source of rumen

782 inocula for 72 h in vitro incubations). Measurements of CH4 production for 24 h in vitro

783 (expressed per unit of OM incubated) were found to be moderately correlated (R2 =

784 0.54; P = 0.04, 4 diets and 16 animals used) with in vivo CH4 production (when

785 expressed per unit of estimated rumen-fermentable OM) across a range of diets

786 differing in source and amount of starch in dietary concentrates. However, no

787 association was found when in vivo CH4 production was expressed per unit of ingested

788 OM (R2 = 0.04; P = 0.88). More research is needed using a wider range of diets

789 representing different production systems.

790 ii) Few direct comparisons of antimethanogenic compounds in vitro and in vivo are

791 available (Table 2). Martínez-Fernández et al. (2013) compared the effectiveness of

792 bromchloromethane and propyl propane thiosulfinate to inhibit CH4 production in vitro

793 and in vivo. Even though both compounds were found to decrease CH4 production in

794 vitro by as much as 90% per unit of DM intake responses in goats were much lower (-

795 33%), although measurements in vivo and in vitro were not made simultaneously. Two

796 newly developed molecules (ethyl-3-nitrooxy propionate, ENP, and 3-nitrooxypropanol,

797 3NOP) have also been evaluated in vitro and in vivo (Martinez-Fernandez et al., 2014).

798 Both compounds were given to sheep at two different doses of 50 and 500 mg/animal

799 per day, corresponding to around 10 and 100 mg/L of rumen content or concentrations

800 of 68 and 681 μM, respectively. Administration of 500 mg/d of h 3NOP at decreased

32
801 CH4 production by -29% on d 14, which was much lower than a value of -95%,

802 determined in vitro (Martínez-Fernández et al., 2014). Similar differences in responses

803 to Japanese horseradish oil have been reported between in vitro and in vivo studies

804 (Mohammed et al., 2004). Much larger decreases in CH4 production were observed in

805 vitro (-89%) than in vivo (-18.7%, Table 2), findings that are in agreement a recent

806 meta-analysis (Hristov et al. 2012). Such discrepancies in the effectiveness of test

807 compounds when given in similar doses may be explained by a combination of several

808 factors: (1) test compounds used are typically administered in 1-2 pulses via the ruminal

809 cannula that often coincide with feeding times, and as a consequence not be rapidly and

810 well mixed with rumen contents; (2) differences in the degradation rate of the active

811 compounds in vitro and in vivo; (3) decrease in microbial density and changes in

812 bacterial community structure of rumen contents during processing as inoculum for in

813 vitro studies associated with the exposure of microorganisms to oxygen and the removal

814 of solids during filtration (Soto et al., 2012); (4) potential washout of these compounds

815 from the rumen or absorption through the rumen wall and (5) adaptation of the rumen

816 microbial ecosystem to the tested compound in vivo that is not emulated by inoculated

817 microbiota in vitro.

818 A different scenario as described above has been recently reported by Castro-Montoya

819 et al. (2015): a blend of essential oil was effective reducing daily emissions of methane

820 in dairy cattle and emissions relative to body weight in beef cattle, interestingly, these

821 effects were not observed in vitro regardless of the technique used to replicate in vivo

822 results (IVFT or consecutive batch culture). This might be due to differences in the

823 mode of action of the essential oils in vitro and in vivo, which merits attention for future

824 research.

825

33
826 RECOMMENDATIONS:

827 In vitro CH4 production is more closely correlated with in vivo CH4 production across a

828 range of feeding regimes when in vivo CH4 emissions are expressed per unit of

829 degraded material rather than per unit of material ingested, although the ratio of

830 substrate (feed) to volume of rumen fluid needs to be considered. The ability of IVFT to

831 reliably predict effects in vivo is affected by adaptation of the rumen inoculum to the

832 substrate tested in vitro. IVFT offers a valuable tool for the study and screening of anti-

833 methanogenic additives before testing in vivo. It is recommended that in vitro data are

834 confirmed in vivo before conclusions on the effectiveness of feed ingredients or

835 additives for lowering CH4 production are drawn given that inhibition potential is often

836 overestimated in vitro.

837

838 Please insert Table 3 around here

839

840 3.8. Units to express CH4 production

841 In vitro CH4 production is usually expressed as ml (or g or mmol) of CH4/g of

842 substrate incubated (Bodas et al., 2008), ml (or g or mmol) of CH4/g of substrate

843 degraded (García-González et al., 2008), or ml of CH4/ml of gas produced (Goel et al.,

844 2008). As rumen fermentation modifiers may affect substrate degradation in a dose

845 dependent manner (Russell and Strobel, 1988), effects on CH4 production are better

846 expressed per unit of substrate degraded, rather than per unit of substrate incubated. For

847 example, Navarro-Villa et al. (2011) highlighted the importance of measuring substrate

848 degradability over the incubation period based on the observation that the ranking of

849 CH4 production potential of incubated feeds (barley grain, grass silage, barley straw)

850 differed depending on whether CH4 was expressed per unit DM incubated or DM

34
851 degraded. Methane output/g DM incubated (CH4/DMi) was in the order barley grain >

852 grass silage > barley straw, whereas when CH4 output was expressed per g DM

853 disappeared (CH4/DMd) the ranking was barley straw > grass silage > barley grain. This

854 last observation is consistent among all quantities of feed, and ratios of rumen fluid to

855 buffer combinations, and concurs with findings from in vivo studies, which compared

856 low and high concentrate diets (Yan et al., 2000).

857 To calculate the volume of CH4 produced per unit of substrate degraded, net

858 values of gas and CH4 production, as well as net amount of substrate degraded, are

859 needed. These are obtained by blanks (i.e., flasks without substrate which contain only

860 inoculum and medium), which are necessary to correct for gas, CH4 and fermentation of

861 residual OM in the inoculum (Rymer et al., 2005; Navarro-Villa et al., 2011). The

862 CH4/DMd values assume that all DM which disappeared is digested, and that no

863 undigested soluble or particulate DM passed through the sintered glass crucible when

864 the post-incubation DM residue was being isolated. Total gas or VFA produced is not as

865 precise as substrate degraded but a good, simple proxy when degradation of substrate

866 can not be determined.

867

868 RECOMMENDATIONS:

869 Regardless of the duration of the incubation in vitro, the amount of CH4

870 produced should ideally be expressed relative to the amount of substrate degraded,

871 rather than the amount of substrate incubated. If not possible, it is recommended to be

872 expressed relative to total gas or VFA produced.

873

874 4. Conclusion

35
875 There is no standard protocol for assessing enteric methane mitigation in ruminants

876 using in vitro gas production technique, as conditions need to be adjusted according to

877 the research question. However, numerous technical issues relating to donor animals,

878 microbial inoculum and general procedures need to be considered (Table 3) to ensure

879 the objectives of experiments can be properly fulfilled. This would allow harmonization

880 of laboratory methods, better interpretation of results and facilitate inter-studies

881 comparisons. Results from in vitro gas production technique studies need to be carefully

882 interpreted before assessing mitigation strategies in vivo.

883

884 Acknowledgements

885 This review is part of the FACCE-JPI ‗Global Network‘ project and the ‗Feeding and

886 Nutrition Network‘ of Livestock Research Group, in the Global Research Alliance for

887 Agricultural Greenhouse Gases ([Link]). Authors

888 acknowledge national funding from INIA (Spain, project MIT01-GLOBALNET-EEZ),

889 the Ministry of Economic Affairs (The Netherlands; project Global Research Alliance

890 on Agricultural Greenhouses Gases, BO-20-007-006), USDA-NIFA (USA), French

891 National Research Agency through the program FACCE-JPI program, Agricultural

892 GHG Research Initiative for Ireland (AGRI-I), Department for Environment Food &

893 Rural Affairs (UK), BLW (Switzerland) and Academy of Finland, Helsinki, Finland

894 (Project 281337).

895

896 References

897

898 Alford, J. S., Jr., 1976. Measurement of dissolved carbon dioxide. Can. J. Microbiol. 22,

899 52–56.

36
900 Amanzougarene, Z., Schauf, S., Fondevila, M., 2015. Control of pH in vitro for the

901 simulation of fermentation in high concentrate feeding conditions. AIDA, XVI

902 Jornadas sobre producción Animal, I, 215-217 (ISBN 978-84-606-7969-1).

903 Ammar, H., S. López, S. Andrés, M.J. Ranilla, R. Bodas, J.S. González., 2008. In vitro

904 digestibility and fermentation kinetics of some browse plants using sheep or goat

905 ruminal fluid as the source of inoculum. Anim. Feed Sci. Technol. 147, 90-104.

906 Anderson, R. C., Callaway, T.R., Van Kessel, J.A.S., Jung, Y.S., , Edrington, T. S.,

907 Nisbet, D. J., 2003. Effect of select nitrocompounds on ruminal fermentation; an

908 initial look at their potential to reduce economic and environmental costs

909 associated with ruminal methanogenesis. Bioresour. Technol. 90, 59–63.

910 Argyle, J.L., Baldwin, R.L., 1988., Modeling of rumen water kinetics and effects of

911 rumen pH changes. J. Dairy Sci. 71, 1178-1188.

912 Bannink, A., Kogut, J., Dijkstra, J., Kebreab, E., France, J., Tamminga, A., Van

913 Vuuren, A.M., 2006. Estimation of the stoichiometry of volatile fatty acid

914 production in the rumen of lactating cows. J. Theor. Biol. 238, 36–51.

915 Blümmel, M., Givens, D.I., Mossb, A.R., 2005. Comparison of methane produced by

916 straw fed sheep in open-circuit respiration with methane predicted by

917 fermentation characteristics measured by an in vitro gas procedure. Anim. Feed

918 Sci. Technol. 123–124, 379–390.

919 Blümmel, M., Ørskov, E.R., 1993. Comparison of in vitro gas production and nylon bag

920 degradability of roughages in predicting feed intake in cattle. Anim. Feed Sci.

921 Technol. 40, 109–119.

922 Bodas, R., Lopez, S., Fernandez, M., García-González, R., Rodrıguez, A.B., Wallace,

923 R.J., Gonzalez, J.S., 2008. In vitro screening of the potential of numerous plant

37
924 species as antimethanogenic feed additives for ruminants. Anim. Feed Sci.

925 Technol. 145, 245–258.

926 Broudiscou, L.P., Papon, Y., Broudiscou, A.F., 1999. Optimal mineral composition of

927 artificial saliva for fermentation and methanogenesis in continuousculture of

928 rumen microorganisms. Anim. Feed Sci. Technol. 79, 43–55.

929 Bryant, A. M., 1964. N. Z. J. Agric. Res. 7, 695

930 Bueno, I.C.H., Brandi, R.B., Franzolin, Benetel, G., Adibe G., Abdalla, L., Muir, H.L.

931 2015. In vitro methane production and tolerance to condensed tannins in five

932 ruminant species. Anim. Feed Sci. Technol. 205, 1-19.

933 Busquet, M., Calsamiglia, S., Ferret, A., Cardozo, P., Kamel, C., 2005. Effects of

934 cinnamaldehyde and garlic oil on rumen microbial fermentation in a dual flow

935 continuous culture. J. Dairy Sci. 88, 2508–2516.

936 Calabrò, S., Lopez, S., Piccolo, V., Dijkstra, J., Dhanoa, M.S., France, J., 2005.

937 Comparative analysis of gas production profiles obtained with buffalo and sheep

938 ruminal fluid as the source of inoculum. Anim. Feed Sci. Technol., 123, 51-65.

939 Cardozo, P., Calsamiglia, S., Ferret, A., Kamel, C., 2004. Effects of natural plant

940 extracts on ruminal protein degradation and fermentation profiles in continuous

941 culture. J. Anim. Sci. 82, 3230–3236.

942 Cardozo, P., Calsamiglia, S., Ferret, A., Kamel, C., 2005. Screening for the effects of

943 natural plant extracts at different pH on in vitro rumen microbial fermentation of a

944 high-concentrate diet for beef cattle. J. Anim. Sci. 83, 2572–2579.

945 Castro-Montoya, J., De Campeneere, S., Van Ransta, G., Fievez, V., 2012. Interactions

946 between methane mitigation additives and basal substrates on in vitro methane

947 and VFA production. Anim. Feed Sci. Technol. 176, 47– 60.

38
948 Castro-Montoya, J., Peiren, N., Cone, J.W., Zweifel, B., Fievez, V., De Campeneere, S.,

949 2015. In vivo and in vitro effects of a blend of essential oils on rumen methane

950 mitigation. Livest. Sci. 180, 134–142.

951 Cheng, K.J., McAllister, T.A., Mathiesen, S.D., Blix, A.S., Orpin, C.G., Costerton,

952 J.W., 1993. Seasonal changes in the adherent microflora of the rumen in high-

953 arctic Svalbard reindeer. Can. J. Microbiol. 39, 101-8.

954 Clauss, M., Schwarm, A., Ortmannb, S., Streichb, W.J., Hummel, J., 2007. A case of

955 non-scaling in mammalian physiology? Body size, digestive capacity, food intake,

956 and ingesta passage in mammalian herbivores. Comp. Bioch. Physiol. 148, 249–

957 265.

958 Cobellis, G., Petrozzi, A., Forte, C., Acuti, G., Orrù, M., Marcotullio, M.C., Aquino, A.,

959 Nicolini, A., Mazza, V., Trabalza-Marinucci, M., 2015. Evaluation of the Effects

960 of Mitigation on Methane and Ammonia Production by Using Origanum vulgare

961 L. and Rosmarinus officinalis L. Essential Oils on in Vitro Rumen Fermentation

962 Systems. Sustainability. 7, 12856-12869.

963 Cone, J.W., Van Gelder, A.H., Bachmann, H., 2000. Influence of inoculum source,

964 dilution and storage of rumen fluid on gas production profiles. In: Gas Production:

965 Fermentation Kinetics for Feed Evaluation and to Assess Microbial Activity.

966 Proceedings of the EAAP Satellite Symposium on Gas Production, Wageningen,

967 The Netherlands. Proc. Br. Soc. Anim. Sci., 15–16.

968 Cone, J.W., Van Gelder, A.H., Bachmann, H., 2002. Influence of inoculum source on

969 gas production profiles. Anim. Feed Sci. Technol. 99, 221–231.

970 Cone, J.W., Van Gelder, A.H., Visscher, G.J.W., Oudshoorn, L., 1996. Influence of

971 rumen fluid and substrate concentration on fermentation kinetics measured with a

39
972 fully automated time related gas production apparatus. Anim. Feed Sci. Technol.

973 61, 113–128.

974 Cornou, C., Storm, I.D., Hindrichsen, I., Worgan, H., Bakewell, E., Yáñez-Ruiz, D.,

975 Abecia, L., 2013. A ring test of a wireless in vitro gas production system. Anim.

976 Prod. Sci. 53, 585–592.

977 Czerkawski, J.W., Breckenridge, G., 1975. New inhibitors of methane production by

978 rumen micro-organisms.

979 Davies, Z.S., Mason, D., Brooks, A.E., Griffith, G.W., Merry, R.J., Theodorou, M.K.,

980 2000. An automated system for measuring gas production from forages inoculated

981 with rumen fluid and its use in determining the effect of enzymes on grass silage.

982 Anim. Feed Sci. Technol. 83, 205–221.

983 De Visser, H., Huisert, H., Klop, A., Ketelaar, R.S. 1993. Autumn-cut silage as

984 roughage component in dairy cow rations. 2. Rumen degradation, fermentation

985 and kinetics. Neth. J. Agric. Sci. 41, 221–234

986 Dehority, B.A., 2003. Rumen Microbiology. Nottingham: Nottingham University Press.

987 Demeyer, D., Fievez, V. 2000. Ruminants and environment: methanogenesis. Ann.

988 Zootech. 49, 95–112.

989 Dhanoa, M.S., France, J., Crompton, L.A., Mauricio, R.M., Kebreab, E., Mills, J.A.N.,

990 Sanderson, R., Dijkstra, J., López, S., 2005. Technical note: A proposed method

991 to determine the extent of degradation of a feed in the rumen from the degradation

992 profile obtained with the in vitro gas production technique using feces as the

993 inoculum. J. Anim. Sci. 82, 733-746.

994 Dijkstra, J., 1994. Simulation of the dynamics of protozoa in the rumen. Brit. J. Nut. 72,

995 679-699.

40
 996 Dijkstra, J., Kebreab, E., Bannink, A., France, J., Lopez, S., 2005. Application of the

997 gas production technique to feed evaluation systems for ruminants. Anim. Feed

998 Sci. Technol. 123, 561–578.

999 Duffield, T., Plaizier, J.C., Fairfield, A., Bagg, R., Vessie, G., Dick, P., Wilson, J.,

1000 Aramini, J., McBride, B., 2004. Comparison of techniques for measurement

1001 ofrumen pH in lactating dairy cows. J. Dairy Sci. 87, 59–66.

1002 Durmic, Z., Hutton, P., Revell, D.K., Emms, J., Hughes, S., Vercoe, P.E., 2010. In vitro

1003 fermentative traits of Australian woody perennial plant species that may be

1004 considered as potential sources of feed for grazing ruminants. Anim. Feed Sci.

1005 Technol. 160, 98–109.

1006 Editorial, 2012. Use of replicates in statistical analyses in papers submitted for

1007 publication in Animal Feed Science and Technology. Anim. Feed Sci. Technol.

1008 171, 1-5.

1009 El Shaera, H.M., Omeda, H.M., Chamberlain, A.G., Axforda, R.F.E., 1987. Use of

1010 faecal organisms from sheep for the in vitro determination of digestibility. J.

1011 Agric. Sci. 109, 257-259

1012 El-Meadaway, A., Z. Mir, P. S. Mir, M. S. Zaman, L. J. Yanke., 1998. Relative efficacy

1013 of inocula from rumen liquor or fecal solution for determining in vitro

1014 digestibility and gas production. Can. J. Anim. Sci. 78, 673–679.

1015 Fievez, V., O.J. Babayemi and D. Demeyer, 2005. Estimation of direct and indirect gas

1016 production in syringes: A tool to estimate short chain fatty acid production

1017 requiring minimal laboratory facilities. Anim. Feed Sci. Technol. 123-124, 197-

1018 210.

41
1019 Flachowsky, G., Lebzien, P., 2012. Effects of phytogenic substances on rumen

1020 fermentation and methane emissions: A proposal for a research process. Anim.

1021 Feed Sci. Technol. 176, 70-77.

1022 García-González, R., Lopez, S., Fernandez, M., Bodas, R., Gonzalez, J.S., 2008.

1023 Screening the activity of plants and spices for decreasing ruminal methane

1024 production in vitro. Anim. Feed Sci. Technol. 147, 36–52.

1025 Geishauser, T., Gitzel, A., 1996. A comparison of rumen fluid sampled by oro-ruminal

1026 probe versus rumen fistula. Small Rumin. Res. 21, 63–69.

1027 Goel, G., Makkar, H.P.S., Becker, K., 2008. Effect of Sesbania sesban and Carduus

1028 pycnocephalus leaves and fenugreek seeds (Trigonella foenum-graecum L.) and

1029 their extracts on partitioning of nutrient from roughage and concentrate based

1030 feeds to methane. Anim Feed Sci Technol. 147, 72–89.

1031 Groot, J.C.J., Cone, J.W., Williams, B.A., Debersaques, F.M.A., Lantinga, E.A., 1996.

1032 Multiphasic analysis of gas production kinetics for in vitro fermentation of

1033 ruminant feeds. Anim. Feed Sci. Technol. 64, 77–89.

1034 Haddi, M.L., Filacorda, S., Meniai, K., Rollin, F., Susmel, P., 2003. In vitro

1035 fermentation kinetics of some halophyte shrubs sampled at three stages of

1036 maturity. Anim. Feed Sci. Technol. 104, 215-225.

1037 Hatew, B., Conea, J.W., Pellikaana, W.F., Podesta, S.C., Bannink, A., Hendrik, W.H.,

1038 Dijkstra, J., 2015. Relationship between in vitro and in vivo methane production

1039 measured simultaneously with different dietary starch sources and starch levels in

1040 dairy cattle. Anim. Feed Sci. Technol. 202, 20–31

1041 Hervás, G., Frutos, P., Giráldez, F.J., Mora, M.J., Fernández, B., Mantecón, A.R., 2005.

1042 Effect of preservation on fermentative activity of rumen fluid inoculum for in

1043 vitro gas production techniques. Anim. Feed Sci. Technol., 123–124, 107–118.

42
1044 Hoover, W. H., B. A. Crooker, and C. J. Sniffen., 1976. Effects of differential solid-

1045 liquid removal rates on protozoa numbers in continuous cultures of rumen

1046 contents. J. Anim. Sci. 43, 528–534.

1047 Hristov A.N., Lee, C., Hristov, R., Firkins, J.L. 2012. A meta-analysis of variability in

1048 continuous-culture ruminal fermentation and digestibility data. Journal of Dairy

1049 Science 95, 5299-307 ·

1050 Hristov, A. N., T. A. McAllister, F. H. Van Herk, K. -J. Cheng, C. J. Newbold and P. R.

1051 Cheeke. 1999. Effect of Yucca schidigera on ruminal fermentation and nutrient

1052 digestion in heifers. J. Anim. Sci. 77, 2554-2563.

1053 Huntington, J.A., Givens, D.I. 1998. The effects of host diet on the cumulative gas

1054 production profile of grass hay and high temperature dried grass. In vitro

1055 Techniques for Measuring Nutrient Supply to Ruminants. In: E.R., Deaville, E.,

1056 Owen, A.T., Adesogan, C., Rymer, J.A., Huntington, T.L.J., Lawrence (Eds.),

1057 Occ. Publication no. 22. Br. Soc. Anim. Sci. 224–226.

1058 Janssen, P.H. 2010. Influence of hydrogen on rumen methane formation and

1059 fermentation balances through microbial growth kinetics and fermentation

1060 thermodynamics. Anim. Feed Sci. Technol. 160, 1–22

1061 Jeyanathan, J., Kirs, M., Ronimus, R.S., Hoskin, S.O., Janssen, P.H., 2011. Methanogen

1062 community structure in the rumens of farmed sheep, cattle and reddeer fed

1063 different diets. FEMS Microbiol. Ecol. 76, 311–326.

1064 Joblin, K.N., 2005. Classical methods for isolation, enumeration, cultivation nand

1065 functional assays of rumen microbes. Methanogenic archaea. In Methods in Gut

1066 Microbial Ecology for Ruminants. Springer, Edited by H.P.S. Makkar and C.S.

1067 McSweeney. ISBN-10 1-4020-3790-2 (HB).

43
1068 Jouany, J.P., Lassalas, B., 2002. Gas pressure inside a rumen in vitro systems stimulates

1069 the use of hydrogen. Presented at 3. Joint RRI-INRA Gastrointestinal Tract

1070 Microbiology Symposium. Beyond Antimicrobials. The Future of Gut

1071 Microbiology, Aberdeen, GBR (2002-06-12 - 2002-06-15).

1072 Kamel, C., Greathead, H.M.R., Ranilla, M.J., Tejido, M.L., Carro, M.D., 2008. Effects

1073 of allicin and diallyl disulfide on in vitro fermentation of a mixed diet. Anim.

1074 Feed Sci. Technol. 145, 351–363.

1075 Kim, C. H., J. N. Kim, J. K. Ha, S. G. Yun, S. S. Lee., 2004. Effects of dietary addition

1076 of surfactant Tween 80 on ruminal fermentation and nutrient digestibility of

1077 Hanwoo steers. Asian-Aust. J. Anim. Sci. 17:337-342.

1078 Kittelmann, S., Pinares-Patiño, C.S., Seedorf, H., Kirk, M.R., Ganesh, S.V., McEwan,

1079 J.C., Janssen, P.H., 2014. Two Different Bacterial Community Types Are Linked

1080 with the Low-Methane Emission Trait in Sheep. PLoS ONE 9(7): e103171.

1081 Klevenhusen, F., A. Muro-Reyes, R. Khiaosa-ard, B. U. Metzler-Zebeli, and Q.

1082 Zebeli. 2012. A meta-analysis of effects of chemical composition of incubated

1083 diet and bioactive compounds on in vitro ruminal fermentation. Anim. Feed Sci.

1084 Technol. 176:61–69.

1085 Kohn, RA., Dunlap, TF., 2008. Calculation of the buffering capacity of bicarbonate in

1086 the rumen and in vitro. J. Anim. Sci. 76, 1702-1709.

1087 Komarek, R.J., 1981. Rumen and abomasal cannulation of sheep with specially

1088 designed cannulas and a cannula insertion instrument. J. Anim. Sci. 53,790–795.

1089 Kristensen, N.B., Engbæk, M., Vestergaard, M., Harmon, D.L., 2010. Technical note:

1090 ruminal cannulation technique in young Holstein calves: effects ofcannulation on

1091 feed intake, body weight gain, and ruminal development at six weeks of age. J.

1092 Dairy Sci. 93, 737–742.

44
1093 Lana, R.P., Russell, J.B., Van Amburgh, M.E., 1998. The role of pH in regulating

1094 ruminal methane and ammonia production. J. Anim. Sci. 76, 2190-2196.

1095 Le Liboux, S., Peyraud, J.L., 1999. Effect of forage particle size and feeding frequency

1096 on fermentation patterns and sites and extent of digestion in dairy cows fed mixed

1097 diets. Anim. Feed Sci. Technol. 76, 297-319.

1098 Leedle, M.P. Bryant, R.B. Hespell., 1982. Diurnal variations in bacterial numbers and

1099 fluid parameters in ruminal contents of animals fed low- or high-forage diets.

1100 Appl. Environ. Microbiol. 44, 402–412

1101 Lodge-Ivey, S.L., Browne-Silva, J., Horvath, M.B., 2009. Technical note: bacterial

1102 diversity and fermentation end products in rumen fluid samples collectedvia oral

1103 lavage or rumen cannula. J. Anim. Sci. 87, 2333–2337.

1104 Lowman, R.S., Theodorou, M.K., Cuddeford, D., 2002. The effect of sample processing

1105 on gas production profiles obtained using the pressure transducer technique.

1106 Anim. Feed Sci. Technol. 97, 221–237.

1107 Machm ller, A., Dohme, ., oliva, C.R., Wanner, M., Kreuzer, M., 2001. Diet

1108 composition affects the level of ruminal methane suppression by medium-chain

1109 fatty acids. Aust. J. Agric. Res. 52, 713–722.

1110 Martinez-Fernandez, G., Abecia, L., Martin-Garcia, A.I., Ramos-Morales, E., Hervas,

1111 G., Molina-Alcaide, E., Yanez-Ruiz, D.R., 2013. In vitro–in vivo study on the

1112 effects of plant compounds on rumen fermentation, microbial abundances and

1113 methane emissions in goats. Animal 7, 1925–1934.

1114 Martínez-Fernández, G., L. Abecia , A. Arco , G. Cantalapiedra-Hijar , A. I. Martín-

1115 García, E. Molina-Alcaide ,M. Kindermann , S. Duval , Yáñez-Ruiz, D.R., 2014.

1116 Effects of ethyl-3-nitrooxy propionate and 3-nitrooxypropanol on ruminal

45
1117 fermentation, microbial abundance, and methane emissions in sheep. J. Dairy Sci.

1118 97, 3790–3799.

1119 Martínez, M.E., Ranilla, M.J., Tejido, M.L., Ramos, S., Carro, M.D., 2010. The effect

1120 of the diet fed to donor sheep on in vitro methane production and ruminal

1121 fermentation of diets of variable composition. Anim. Feed Sci. Technol. 158,

1122 126–135.

1123 Mateos, I., Ranilla, M.J., Tejido, M.L., Saro, C., Kamel, C. and Carro, M.D., 2012. The

1124 influence of diet type (dairy versus intensive fattening) on the effectiveness of

1125 garlic oil and cinnamaldehyde to manipulate in vitro ruminal fermentation and

1126 methane production. Anim. Prod. Sci. 53, 299–307

1127 Mauricio, R.M., Owen, E., Mould, F.L., Givens, I., Theodorou, M.K., France, J.,

1128 Davies, D.R. and Dhanoa, M.S., 2001. Comparison of bovine rumen liquor and

1129 bovine faeces as inoculum for an in vitro gas production technique for evaluating

1130 forages. Anim. Feed Sci. Technol. 89, 33-48.

1131 McDougall, E.J., 1948. Studies on ruminant saliva, 1. The composition and output of

1132 sheep‘s saliva. Biochem. J. 43, 99–109.

1133 Menke, K.H., Raab, L., Salewski, A., Steingass, H., Fritz, D., Schneider, W., 1979. The

1134 estimation of the digestibility and metabolisable energy content of ruminant

1135 feeding stuffs from the gas production when they are incubated with rumen liquor.

1136 J. Agric. Sci., 93, 217–222

1137 Menke, K.H., Steingass, H., 1988. Estimation of the energetic feed value obtained from

1138 chemical analysis and in vitro gas production using rumen fluid. Anim. Res. Dev.

1139 28, 7–55.

1140 Miron, J., Ben Ghedalia, D., Morrison, M.. 2001. Invited Review: Adhesion

1141 mechanisms of rumen cellulolytic bacteria. J. Dairy Sci. 84, 1294-1309.

46
1142 Mohammed, N., N. Ajisaka, Z. Lila, K. Hara, K. Mikuni, K. Hara, S. Kanda, and H.

1143 Itabashi., 2004. Effect of Japanese horseradish oil on methane production and

1144 ruminal fermentation in vitro and in steers. J. Anim. Sci. 82, 1839–1846.

1145 Moharrery, A., Das, M., 2001. Correlation between microbial enzyme activities in the

1146 rumen fluid of sheep under different treatments. Reprod. Nutr. Dev. 41, 513-29.

1147 Monteils, V., Rey, M., Silberberg, M., Cauquil, L., Combes, S., 2012. Modification of

1148 activities of the ruminal ecosystem and its bacterial and protozoan composition

1149 during repeated dietary changes in cows. J. Anim. Sci. 90, 4431–4440.

1150 Mould, F.L, Kliem, K.E., Morgan, R., Mauricio, R.M. 2005. In vitro microbial

1151 inoculum: A review of its function and properties. Anim. Feed Sci. Technol. 123–

1152 124, 31–50.

1153 Moss, A.R., Givens, D.J., 1997. Methane production from a range of feedstuffs as

1154 determined in vitro using the cumulative gas production technique and compared

1155 with that measured in vivo. Proc. Brit. Soc. Anim. Sci. 194.

1156 Mould, F.L., Morgan, R., Kliem, K.E., Krystallidou, E., 2005. A review and

1157 simplification of the in vitro incubation medium. Anim. Feed Sci. Technol.

1158 123,155–172.

1159 Mueller, R.E., Asplund, J.M., Lannotti, E.L., [Link] changes in the epimural

1160 bacterial community of young lambs as revealed by scanning electron

1161 microscopy. Appl. Environ. Microbiol. 47, 715–723.

1162 Muetzel, S., Hunt, C., Tavendale, M.H., 2014. A fully automated incubation system for

1163 the measurement ofgas production and gas composition. Anim. Feed Scie

1164 Technol. 196, 1-11.

47
1165 Navarro-Villa, A., O‘Briena, M., López, ., Bolandb, T.M., O‘Kiely, P., 2011.

1166 Modifications of a gas production technique for assessing in vitro rumen methane

1167 production from feedstuffs. Anim. Feed Sci. Technol. 166, 163– 174.

1168 O'Brien, M., Navarro-Villa, A., Purcell, P. J., Boland, T. M., O'Kiely P., 2013.

1169 Reducing in vitro rumen methanogenesis for two contrasting diets using a series

1170 of inclusion rates of different additives. Anim. Prod. Sci. 54, 141–157.

1171 Patra, A. K., J. Stiverson, Z. Yu., 2012. Effects of Quillaja and yucca saponins on

1172 communities and select populations of rumen bacteria and archaea, and

1173 fermentation in vitro. J. Appl. Microbiol. 113, 1329–1340.

1174 Patra, A.K., Yu, Z., 2013. Effects of gas composition in headspace and bicarbonate

1175 concentrations in media on gas and methane production, degradability,and rumen

1176 fermentation using in vitro gas production techniques. J. Dairy Sci. 96, 4592–

1177 4600.

1178 Payne, J.S., Hamersley, A.R., Milligan, J.C., Huntington, J.A., 2002. The affect of

1179 rumen fluid collection time on its fermentative capacity and the stability of rumen

1180 conditions in sheep fed a constant diet. Proc. Br. Soc. Anim. Sci. 165.

1181 Pei, C.X., Mao, S.Y., Chenga, Y.F., Zhua, W.Y., 2010. Diversity, abundance and novel

1182 16S rRNA gene sequences of methanogens in rumen liquid, solid and epithelium

1183 fractions of Jinnan cattle. Animal. 4, 20-29.

1184 Pell, A.N., Schofield, P., 1993. Computerized Monitoring of Gas Production to Measure

1185 Forage Digestion In Vitro. J. Dairy Sci. 76, 1063-1073.

1186 Pellikaan, W., Hendriks, W., Uwimana, G., Bongers, L., Becker, P., Cone, J., 2011. A

1187 novel method to determine simultaneously methane production during in vitro gas

1188 production using fully automated equipment. Anim. Feed Sci. Technol. 168, 196–

1189 205.

48
1190 Pinares-Patiño, C.S., Ulyatt, M.J., Lassey, K.R., Barry, T.N., Holmes, C.W. 2003.

1191 Rumen function and digestion parameters associated with differences between

1192 sheep in methane emissions when fed chaffed lucerne hay. J. Agric. Sci. 140,

1193 205–214.

1194 Prates, A., de Oliveira, J.A., Abecia, L., Fondevila, M., 2010. Effects of preservation

1195 procedures of rumen inoculum on in vitro microbial diversity and fermentation.

1196 Anim. Feed Sci. Technol. 155, 186–193.

1197 Ramos-Morales, E., Arco-Pérez, A., Martín-García, A.I., Yáñez-Ruiz, D.R., Frutos, P.,

1198 Hervás, G. 2014. Use of stomach tubing as an alternative to rumen cannulation to

1199 study ruminal fermentation and microbiota in sheep and goats. Anim. Feed Sci.

1200 Technol. 198, 57–66

1201 Ranilla M.J., Tejido, M.L., Carro M.D., 2001. Comparación de diversos métodos de

1202 desligamiento de bacterias ruminales asociadas a la fase sólida de la digesta en un

1203 sistema in vitro (Rusitec). ITEA (Vol. extra), 22, pp. 382-384.

1204 Robinson, P., Campbell, M., Fadel, J., 1999. Influence of storage time and temperature

1205 on in vitro digestion on neutral detergent fibre at 48h, and comparison to 48h in

1206 sacco neutral detergent fiber digestion. Anim. Feed Sci. Technol. 80, 257-266.

1207 Roger, V.R., G. Fonty, S. Komisarczuk-Bondy, Gouet, P., 1990. Effects of

1208 physicochemical factors on the adhesion to cellulose avicel of the rumen bacteria

1209 Ruminococcus flavefaciens and Fibrobactor succinogenes subsp. succinogenes.

1210 Appl. Environ. Microbiol. 56:3081-3087.

1211 Russell, J.B., Strobel, H.J., 1988. Effects of additives on in vitro ruminal fermentation: a

1212 comparison of monensin and bacitracin. Another Gram-positive antibiotic. J.

1213 Anim. Sci. 66, 552–558.

49
1214 Rymer, C., Huntington, J.A., Givens, D.I., 1999. Effects of inoculum preparation

1215 method and concentration, method of inoculation and pre-soaking the substrate on

1216 the gas production profile of high temperature dried grass. Anim. Feed Sci.

1217 Technol. 78, 199–213.

1218 Rymer, C., Huntington, J.A., Williams B.A., Givens, D.I., 2005. In vitro cumulative gas

1219 production techniques: History, methodological considerations and challenges.

1220 Anim. Feed Sci. Technol. 123–124, 9–30

1221 Sadet, S,, Martin, C., Meunier, B., Morgavi, D.P., 2007. PCR-DGGE analysis reveals a

1222 distinct diversity in the bacterial population attached to the rumen epithelium.

1223 Animal. 1, 939-944.

1224 Salem, A.F.Z.M., 2005. Impact of season of harvest on in vitro gas production and dry

1225 matter degradability of Acacia saligna leaves with inoculum from three ruminant

1226 species. Anim. Feed Sci. Technol. 123, 67-79.

1227 Schwarm, A., Ortmann, S., Wolf, C., Streich, W.J., Clauss, M., 2008. Excretion patterns

1228 of fluid and different sized particle passage markers in banteng (Bos javanicus)

1229 and pygmy hippopotamus (Hexaprotodon liberiensis): two functionally different

1230 foregut fermenters. Comp. Biochem. Physiol. A, 150: 32-39

1231 Schwarm, A., Schweigel-Röntgen, M., Kreuzer, M., Ortmann, S., Gill, F., Kuhla, B.,

1232 Meyer, U., Lohölter, M., Derno, M., 2015. Methane emission, digestive

1233 characteristics and faecal archaeol in heifers fed diets based on silage from brown

1234 midrib maize as compared to conventional maize. Arch. Anim. Nutr. 69, 159-176.

1235 Shen, J.S., Chai, Z., Song, L.J., Liu, J.X., Wu, Y.M., 2012. Insertion depth of oral

1236 stomach tubes may affect the fermentation parameters of ruminal fluid collected

1237 in dairy cows. J. Dairy Sci. 95, 5978–5984.

50
1238 Shin, E.C., Choi, B.R., Lim, W.J., Hong, S.Y., An, C.L., Cho, K.M., Kim, Y.K., An,

1239 J.M., Kang, J.M., Lee, S.S., Kim, H., Yun, H.D., 2004. Phylogenetic analysis of

1240 archaea in three fractions of cow rumen based on the 16S rDNA sequence.

1241 Anaerobe. 10, 313-319.

1242 Soto, E.C., Molina-Alcaide, E., Khelil, H., Yáñez-Ruiz, D.R., 2013 Ruminal microbiota

1243 developing in different in vitro simulation systems inoculated with goats‘ rumen

1244 liquor. Anim. Feed Sci. Technol. 185, 9-18.

1245 Soto, E.C., Yáñez-Ruiz, D.R., Cantalapiedra-Hijar, G., Vivas, A., Molina-Alcaide, E.,

1246 2012. Changes in ruminal microbiota due to rumen content processing and

1247 incubation in single-flow continuousculture fermenters. Anim. Prod. Sci. 52, 813–

1248 822.

1249 Storm, A.C., Kristensen, N.B., 2010. Effects of particle size and dry matter content of a

1250 total mixed ration on intraruminal equilibration and net portal flux of volatile fatty

1251 acids in lactating dairy cows. J. Dairy Sci. 93, 4223–4238

1252 Tagliapietra, F., Cattani, M., Bailoni, L., Schiavon, S. 2010. In vitro rumen

1253 fermentation: Effect of headspace pressure on the gas production kinetics of corn

1254 meal and meadow hay. Anim. Feed Sci. Technol. 158, 197–201.

1255 Tan, H.Y., Sieo, C.C., Abdullah, N., Liang, J.B., Huang, X.D., Ho, Y.W., 2011. Effects

1256 of condensed tannins from Leucaena on methane production, rumen fermentation

1257 and populations of methanogens and protozoa in vitro. Anim. Feed Sci. Technol.

1258 169, 185–193.

1259 Tekippe, J.A., Tacoma, R., Hristov, A.N., Lee, C., Oh, J., Heyler, K.S., Cassidy, T.W.,

1260 Varga, G.A., Bravo, D., 2013. Effect of essential oils on ruminal fermentation and

1261 lactation performance of dairy cows. J. Dairy Sci. 96, 7892-7903.

51
1262 Terré, M., Castells, L., Fàbregas, F., Bach, A., 2013. Comparison of pH, volatile fatty

1263 acids, and microbiome of rumen samples from preweaned calved obtainedvia

1264 cannula or stomach tube. J. Dairy Sci. 96, 5290–5294.

1265 Theodorou, M.K., Williams, B.A., Dhanoa, M.S., McAllan, A.B., France, J., 1994. A

1266 simple gas production method using a pressure transducer to determine the

1267 fermentation kinetics of ruminant feeds. Anim. Feed Sci. Technol. 48, 185–197.

1268 Tilley, J.M.A., Terry, R.A., 1963. A two-stage technique for the in vitro digestion of

1269 forage crops. J. Br. Grass Soc. 18, 104–111.

1270 Trabalza-Marinucci, M., Poncet, C., Delval, E., Fonty, G., 2006. Evaluation of

1271 techniques to detach particle-associated microorganisms from rumen contents.

1272 Anim. Feed Sci. Technol. 125, 1–16.

1273 Van Kessel J.A.S., Russell, J.B., 1996. The effect of pH on ruminal methanogenesis.

1274 FEMS Microbiol. Ecol. 20 , 205–210.

1275 Vàradyovà, Z., M. Baran and I. Zeleňák., 2005. Comparison of two in vitro

1276 fermentation gas production methods using both rumen fluid and faecal inoculum

1277 from sheep. Anim. Feed Sci. Technol. 123-124:81-94.

1278 Waghorn, G.C., Woodward, S.L., Tavendale, M., Clark, D.A., 2006. Inconsistencies in

1279 rumen methane production—effects of forage composition and animal genotype.

1280 International Congress Series 1293, 115– 118.

1281 Wang, X.Z., Sunb, P.H. Janssenb, S.X. Tanga, Z.L. Tana., 2014. Responses of methane

1282 production and fermentationpathways to the increased dissolved hydrogen

1283 concentrationgenerated by eight substrates in in vitro ruminal cultures. 2014.

1284 Anim. Feed Sci. Technol. 194, 1–11.

52
1285 Warner, D., Dijkstra, J., Hendriks, W.H. & Pellikaan, W.F., 2014. Stable isotope-

1286 labelled feed nutrients to assess nutrient-specific feed passage kinetics in

1287 ruminants. J. Sci. Food Agric. 94, 819-824.

1288 Weimer, P. J., B. S. Dien, T. L. Springer, and K. P. Vogel., 2005. In vitro gas

1289 production as a surrogate measure of the fermentability of cellulosic biomass to

1290 ethanol. Appl. Microbiol. Biotechnol. 67, 52–58.

1291 Wilkins, J., 1974. Pressure transducer method for measuring gas production by

1292 microorganisms. Appl. Microbiol. 27, 135–140.

1293 Williams, A.G., Coleman, G.S., 1992. The Rumen Protozoa. New York, NY: Springer-

1294 Verlag.

1295 Williams, B.A., 1998. Poster discussion session report. In: Deaville, E.R., Owen, E.,

1296 Adesogen, A.T., Rymer, C.,

1297 Williams, Y.J., Popovski, S., Rea, S.M., Skillman, L.C., Toovey, A.F., Northwood,

1298 K.S., Wright, A.D.G., 2009. A vaccine against rumen methanogens can alter the

1299 composition of archaeal populations. Appl. Environm. Microbiol. 75, 1860–1866.

1300 Yan, T., Agnew, R.E., Gordon, F.J., Porter, M.G., 2000. Prediction of methane energy

1301 output in dairy and beef cattle offered grass silage-based diets. Livest. Prod. Sci.

1302 64, 253–263.

1303 Zhou X, Zeitz JO, Meile L, Kreuzer M, Schwarm A (2015) Influence of pH and the

1304 degree of protonation on the inhibitory effect of fatty acids in the ruminal

1305 methanogen Methanobrevibacter ruminantium strain M1, J Appl Microbiol, in

1306 press, doi: 10.1111/jam.12955.

1307 Zhou, Z., Q. Meng, Z. Yu., 2011. Effects of methanogenic inhibitors on methane

1308 production and abundance of methanogen and cellulolytic bacteria in in vitro

1309 ruminal cultures. Appl. Environ. Microbiol. 77, 2634–2639.

53
1 Table 1.

2 Typical in vitro fermentation techniques (IVFT) used to quantify CH4 (and CO2 and H2) production (adapted from Rymer et al., 2005).

Duration
Inoculum Incubatio Inoculum: Dietary Gas
Device and Buffer of Pressure Gas sampling and
Inoculum collection n volume, medium substrate, release/colle
In vitro volume reference incubation, control analysis (GC)
time ml ratio mg cted
system h
Manual/End Yes,
Menke et Syringes kept Liquid Before Menke et al.
30 1:3 24 200-300 -point moveable Manual
al. (1979) in rotor phase feeding (1979)
sampling glass piston
Theodorou Liquid Manual/End No,
Bottle in Before Theodorou
et al. and solid 60 1:5 24-72 500-1000 -point pressure Manual
incubator feeding (1993)
(1994) phases sampling increases
Manual/ No,
Mauricio et Bottle in Liquid Before Theodorou
100 1:10 n.s. 1000 End-point pressure Manual
al. (1999) incubator phase feeding (1993)
sampling increases
Pell and Bottle and Goering and Manual/End No, Manual, CH4
Liquid 2 h after
Schofield stirrer kept in 10 1:5 Van Soest n.s. 100 -point pressure estimated by
phase feeding
(1993) incubator (1970) sampling increases difference to CO2
Automated/
Bottle in Fixed
Cone et al. Liquid 2 h after Steingass
shaking water 100 1:3 48 400-500 pressure Yes Manual
(1996) phase feeding (1983)
bath
Automated/
Liquid Fixed
Davies et Bottle in Before Theodorou
and solid 100 1:10 n.s. 1000 pressure Yes n.s.
al. (1998) incubator feeding (1993)
phases
Menke and Automated/
Cornou et Bottle in Liquid
n.s. 60 1:3 Steigass 72 500 Fixed Yes Manual
al. (2012) incubator phase
(1988) pressure
Muetzel et Bottle in Liquid Mould et al. Automated/
n.s. 60 1:5 48 600 Yes Automated GC
al. (2014) incubator phase (2005) Automated
3

4 GC, gas chromatograph; n.s., not specified

9 Table 2.

10 Comparison of results based on in vitro gas production and studies in vivo evaluating the effect of a) diets or b) feed additives on methane production

Diets Animal species Adaptation IVFT used CH4 in vivo R2, P value Reference
/Intake level period
Straw ± NH3/NaOH Sheep / Maintenance 10 days Menke et al. Open circuit 0.89, P<0.0001 Blümmel et al.
(1979) 3 respiration chamber, (2005)
g/kg OM
AH, CS+SBM, Cattle / Maintenance 12 days Menke and SF6 ml/kg DMI 0.98, P<0.0001 Bhatta et al. (2007)
RG+SBM, RS+SBM, Steingass (1988)
SG+SBM1
GS:CO 60:402 Cattle / 95 % voluntary DMI 12 days Cone et al. Open circuit 0.04, P= 0.878 (a) Hatew et al. (2015)
(1996) respiration chamber 0.54, P= 0.040 (b)
g/kg OMI (a) or
OMFR (b)
Additive4 Dose in vitro // Animal Adaptation IVFT used/ CH4 in vivo Change in CH4 (%, L/kg DMI) Reference
Inclusion rate in species period incubation time in vitro // In vivo
vivo
Horseradish oil 0.17, 0.85 and 1.7 Cattle 14 days Russel and Headhood collection -18, -89 // Mohamed et al.
g/L // 20 g, kg DM Martin 1984/6 h chamber -18.6 (2004)
PTS 100, 320 µl/L // 50, Goats 7 days Theodorou et al. Open circuit -28, -96 // Martínez-Fernández
100, 200 mg/L (1994)/24 h respiration chamber -13, -18 et al. (2013)
rumen content
BCM 100, 320 µl/L // 50, Goats 7 days Theodorou et al. Open circuit -94, -96 // Martínez-Fernández
100, 160 mg/L (1994)/24 h respiration chamber -34, - 45 et al. (2013)
rumen content
E3NP 25, 50 µM // 100 Sheep 30 days Theodorou et al. Open circuit -90, - 96 // Martínez-Fernández
mg/d (1994)/24 h respiration chamber -21 et al. (2014)
 3NOP 33, 66 µm // 100 Sheep 30 days Theodorou et al. Open circuit -99, -99 // -24 Martínez-Fernández
mg/d (1994), 24 h respiration chamber et al. (2014)
Blend essential oils 0.25, 0.5, 1, 2, 5, 15, Cattle 42 days Theodorou et al. Open circuit 0 // - 15 (dairy cattle), 0 (beef Castro-Montoya et
30 mg/L // 0.2 g/day (1994), 24 h, 72 respiration chamber cattle) al. (2015)
h
11 1 AH=alfalfa hay, CS= corn silage, SBM=soyabean meal, RS=rice straw, SG=sudan grass; 2GS= grass silage + 4 types of concentrate containing low or high amounts of slowly or rapidly

12 degraded sources of starch; 3CH4 estimated from fermentation end products. 4PTS: propyl thyosulfanite, BCM: brochloromethane, E3NP: ethyl-3-nitrooxy propionate, 3NPOP: and 3-
13 nitrooxypropanol, 3NOP
14 Table 3.

15 Summary of technical recommendations on the use of in vitro gas production methods for measuring methane production

Aspect Recommendation
Donor animals
Species Donor animals should be the same as target species
Numbers 3 or more animals where possible, minimum 2 animals.
Replications Minimum of 3 independent incubation runs
Diet Donor animals should be fed the same diet as incubated or of similar nutrient composition
Adaptation Collect inoculum from animals that have been adapted to treatment for at least 2 weeks

Microbial inoculum
Collection Collection before feeding unless the experimental design requires a different sampling regimen.

Ideally maintain inocula under anaerobic conditions at 39ºC and use within 1 h. Preservation at 0ºC up to 6 h or after freezing in liquid N are
Preservation viable alternatives.
Processing Filter through 250 µm pore size cloth.

Procedure
Use a range of feeds for additives screening. Include a blank substrates to control for intra and inter variability and as a reference for comparisons
Substrate incubated between laboratories.
No single buffer is recommended. For all incubations adjust bicarbonate concentration to maintain a target culture pH culture. Concentrations
Buffer composition should not exceed 80 mM.
Inoculum:medium ratio Adjust depending on the duration of incubation and frequency of gas sampling.

Duration of incubation Adjusted according to the composition and particle size of substrate based on retention time in the rumen in vivo.
 Gas measurements Determine CH4 and CO2 when gas pressure is vented.
Preferably express data relative to substrate (DM, OM, NDF) degradation. If this is not possible, data should be expressed relative to total gas,
Units CO2 or VFA production.

Mandatory technical Numbers of donor animals used to obtain inocula. Composition of diet fed to donor animal and the time on diet. Inoculum collection time relative
details to be reported in to feeding. Headspace gas composition. Pore size of filter used to strain rumen fluid. Buffer and media composition. Rumen inoculum storage
peer-reviewed conditions, both temperature and the time from collection to inoculation. Ratio of medium to rumen inoculum. Substrate processing procedures,
publications frequency of gas sampling and CH4 measurements
16

17
18