The acute effects of rumen pulse-dosing of hydrogen acceptors during methane inhibition with nitrate or 3-nitrooxypropanol in dairy cows

Dietary methane (CH 4 ) mitigation is in some cases associated with an increased hydrogen (H 2 ) emission. The objective of the present study was to investigate the acute and short-term effects of acceptors for H 2 (fumaric acid, acrylic acid or phloroglucinol) supplemented via pulse-dosing to dairy cows fed CH 4 mitigating diets (using nitrate or 3-nitrooxypropanol), on gas exchange, rumen gas and VFA composition. For this purpose, 2 individual 4 × 4 Latin square experiments were conducted with 4 periods of 3 d (nitrate supplementation) and 7 d (3-nitrooxypropanol supplementation), respectively. In each study, 4 rumen cannulated Danish Holstein cows were used. Each additive for CH 4 mitigation was included in the ad libitum fed diet within the 2 experiments, to which the cows were adapted for at least 14 d. Acceptors for H 2 were administered twice daily in equal portions through the rumen fistula immediately after feeding of the individual cow. In Exp. 1 (nitrate), the treatments were CON-1 (no H 2 -acceptor), FUM-1 (fumaric acid), ACR-1 (acrylic acid) and FUM+ACR-1 (50% FUM-1 + 50% ACR-1). In Exp. 2 (3-nitrooxypropanol), the 3 treatments, CON-2 , FUM-2 , and ACR-2 , were similar to CON-1, FUM-1 and ACR-1 treatments, however the fourth treatment was PHL-2 (phloroglucinol). Gas exchanges were measured in respiration chambers, while samples of rumen liquid and headspace gas were taken in time series relative to feeding and dosing on specific days. Headspace gas was analyzed for gas composition and rumen liquid was analyzed for volatile fatty acid composition and dissolved gas concentrations. Headspace gas composition and dissolved gas concentration were only measured in Exp. 2. Dry matter intake was reduced


INTRODUCTION
Mitigation of enteric CH 4 from ruminants can include the use of 3-nitrooxypropanol (3-NOP) and nitrate.These additives have well-documented CH 4 mitigating effects (Feng et al., 2020;Kebreab et al., 2023).However, their use is often associated with an increased emission of enteric H 2 .This emerges as a consequence of their ability to reduce the methanogenic conversion of CO 2 and H 2 from rumen fermentation to CH 4 .The mode of action for 3-NOP was uncovered by Duin et al. (2016) to be a specific blocking of the methyl coenzyme M-reductase, which is a key enzyme in methanogenesis.For nitrate, the general perception is that nitrate acts The acute effects of rumen pulse-dosing of hydrogen acceptors during methane inhibition with nitrate or 3-nitrooxypropanol in dairy cows as an acceptor for H 2 in the rumen, during its reduction from nitrate (NO 3 -) to nitrite (NO 2 -) and further to ammonium (NH 4 + ), where both steps require reducing electrons in the form of H 2 (Ungerfeld and Kohn, 2006).Nevertheless, results from experiments with nitrate as a CH 4 mitigating additive show an increased H 2 emission, assigned to a direct inhibitory effect of nitrite on the methanogenic population (Zhou et al., 2012).Increased H 2 emission not only reflects an energetic loss for the animal, but further indicate a surplus of electrons in the rumen, a reduced redox state, which theoretically can hamper normal rumen fermentation (McAllister and Newbold, 2008) or potentially lead to other unwanted secondary effects, as reflected by decreased DMI (Maigaard et al., 2024a).
Competing H 2 consuming pathways exist naturally in the rumen (for overview see Ungerfeld, (2020)).One of the propionate synthesizing pathways is the reduction of fumaric acid to succinic acid that is further decarboxylated to propionate (succinate-propionate pathway).During this reduction, one mole of H 2 is being used for every mole of fumaric acid reduced to succinic acid (Ungerfeld et al., 2007).Fumaric acid has previously been shown to redirect H 2 away from methanogenesis, but with varying effects on CH 4 emission (Bayaru et al., 2001;van Zijderveld et al., 2011).
Another precursor for propionate is acrylic acid, which, in the acrylate pathway is reduced to propionate using one mole of H 2 for every mole of acrylic acid (Newbold et al., 2005).Like for fumaric acid, this precursor has also been investigated in vitro to redirect H 2 away from methanogenesis, and results from Newbold et al. (2005) revealed that both fumaric acid and acrylic acid added as free acids reduced methanogenesis.However, to our knowledge, in vivo application of acrylic acid has never been reported.Since the conversion of these organic acids to propionate occupies 2 naturally occurring pathways for propionate synthesis in the rumen, it is possible that a combined supplementation would redirect H 2 into both processes concurrently.
Rumen microbes can also reduce phenolic compounds such as phloroglucinol to acetate using H 2 or formate to deliver reducing elements (Patel et al., 1981;Krumholz et al., 1987;Armstrong and Patel, 1994).Thus, this reduction also yields a product, which potentially can be utilized by the animal.Martinez-Fernandez et al. (2017) found a decrease in H 2 and a concomitant increase in rumen acetate when phloroglucinol was supplemented to beef steers, whose CH 4 emission was mitigated using chloroform.Later, similar effects have been observed with in vitro studies incubating Asparagopsis taxiformis with both rumen fluid from dairy cows (Huang et al., 2023) and from goats (Romero et al., 2023).
Gaseous H 2 , either as expelled or in the rumen headspace is often used as an indicator of surplus of reducing elements in the form of H 2 in rumen liquid.Wang et al. (2016a) showed that measured dissolved gas concentrations of CH 4 and H 2 were greater than estimated concentrations based on headspace concentrations assuming equilibrium between the gas and liquid phase.These findings suggest that rumen fluid is supersaturated with H 2 , and that the degree of supersaturation may depend on when the last meal was ingested, wherefore the use of H 2 gas indicators (headspace or expelled) may not be ideal in understanding the kinetics in such electron acceptors during interventions for H 2 redirection.This suggests that initiatives to investigate such approaches should include measurements of dissolved gases using sensors or extraction approaches as also reported by Guyader et al. (2015) and Wang et al. (2014), respectively.
The objectives of this screening study were to investigate the acute and short-term ability of fumarate, acrylic acid or phloroglucinol to redirect H 2 when they were pulse-dosed in the rumen of dairy cows fed either nitrate or 3-NOP as CH 4 mitigation additives.This was evaluated based on cows' gas exchange, rumen gas and VFA composition.We hypothesized that excess H 2 derived from CH 4 mitigation could be redirected to the reduction of fumaric acid, acrylic acid or phloroglucinol reflected in increased rumen concentrations of propionate or acetate.

MATERIALS AND METHODS
Two animal experiments (Exp. 1 and Exp. 2) were conducted to assess short-term effects of supplementation of acceptors for H 2 during CH 4 inhibition.The experiments were conducted at Aarhus University, AU Viborg -Research Centre Foulum, Denmark, in March 2022, and February 2023 for Exp. 1 and 2, respectively.Both experiments fulfilled the guidelines of the Danish Ministry of Food, Agriculture and Fisheries, The Danish Veterinary and Food Administration under act.No 474 (May 15, 2014) and executive order no.2028 (December 14, 2020) under consideration of the Arrive Guidelines (Percie du Sert et al., 2020).

Animals, Experimental Design and Housing
Four lactating rumen, duodenum and ileum cannulated Danish Holstein cows were enrolled in each experiment in a 4 × 4 Latin square (LS) design.In Exp 1., all cows were multiparous (4th parity).At the initiation of the experiment, cows were 182 ± 106 d in milk (mean ± SD), milk yield was 34.3 ± 9.96 kg, and BW was 693 ± 54.3 kg.Each experimental period lasted 3 d, where cows were housed in respiration chambers for the entire period (3 d).Cows were milked twice daily before feeding at 0515 h and 1615 h.
In Exp. 2, 2 cows were primiparous and 2 were multiparous (5th parity), and at the initiation of the experiment, cows were 167 ± 14.0 d in milk (mean ± SD), milk yield was 24.8 ± 6.16 kg, and BW was 627 ± 93.6 kg.Each experimental period lasted 7 d, and cows were housed outside respiration chambers for the first 2 d of the period, whereafter they returned to the chambers for 5 consecutive days.Cows were milked twice daily before feeding at 0615 h and 1630 h.
In both experiments, when cows were outside chambers, cows were loose housed in individual pens with slatted floor and a cubicle bedded with mattress and sawdust.

Diets and Treatments
In both experiments, cows were fed ad libitum with a TMR twice daily at 0525 h and 1625 h in Exp. 1 and at 0625 and 1640 h in Exp. 2. Around 65% of the daily ration was fed in the morning and 35% was fed in the afternoon.The TMR's consisted of grass clover silage (perennial ryegrass, hybrid ryegrass, white clover, and red clover), corn silage, spring barley, rapeseed cake, sugar beet pulp and mineral supplements (see chemical composition of feed ingredients in Supplemental Table S1, Maigaard et al., (2024b)).The TMR for each experiment (Table 1) was formulated according to NorFor (Volden, 2011) with a forage to concentrate ratio of 56:44 (on a DM-basis) in both experiments.Different additives to mitigate CH 4 were included in the TMR in each experiment, and before initiation of the first period in each experiment, cows were adapted to the CH 4mitigating diet for at least 14 d.Within each experiment, there were 4 treatments, where one represented a control treatment and 3 represented supplementations of different H 2 acceptors that were administered twice daily in equal portions through the rumen fistula immediately after feeding of the individual cow.
In both experiments, daily collection and weighing of feed refusals were used to determine feed intake.Samples of TMR and refusals were collected every day in each experiment for daily DM determination.Individual feed ingredients were sampled once per week and pooled to represent period 1 and 2 and period 3 and 4, respectively, before subsequent analyses.Subsamples of TMR were pooled across days to represent each period.All samples were stored at −20°C until pooling and later chemical analyses.
Milk yield was recorded by continuous flowmeters at every milking throughout the 2 experiments.Milk was sampled from a continuous flow sample collected at every milking during the experimental periods, and samples were subsequently analyzed for fat, protein, lactose monohydrate, and urea concentration (Eurofins Steins Laboratorium A/S, Vejen, Denmark).

Rumen and Blood Sampling
Rumen samples were taken from each cow in a time series relative to feeding/dosing in each experiment.Exp. 1 was originally designed for obtaining continuous measurements of dissolved H 2 in rumen fluid using a sensor mounted in the rumen of the cow, but these measurements of dissolved H 2 failed due to technical challenges.Therefore, rumen fluid samples were collected from 2 of the cows on d 2, and from the 2 other cows on d 3. Samples were taken immediately before dosing (0.0 h), and at 3 time points after dosing (1.0, 3.0, and 5.0 h) of the treatments.Rumen fluid (30 mL) was sampled from the ventral ruminal sac through the rumen fistula using a syringe attached to a 90-cm stainless steel rumen sampler devise (Bar Diamond Inc., Parma, ID).The sample was transferred to a Greiner tube for immediate pH measurement using a glass electrode connected to a meter (Meterlab PHM 220,Radiometer,Brønshøj,Denmark) and subsequent transferred to tubes that were stored (−20°C) for later analysis of VFA, alcohols, L-lactate, and ammonia concentration.
In Exp. 2 rumen samples (fluid and headspace gas) were obtained from all cows on d 2 and 7 of each period.

Maigaard et al.: Hydrogen acceptors during methane mitigation
On d 2, samples were taken at 4 time points relative to morning dosing (−1.0, 0.5, 1.5, and 5.0 h), whereas, on d 7, samples were taken at 8 time points relative to morning dosing (−1.0, 0.0, 0.5, 1.5, 2.0, 3.0, 5.0, and 7.0 h).Rumen fluid samples were obtained and handled as described for Exp. 1, however, for Exp. 2, 50 mL of rumen fluid was initially collected to be able to retain 35 mL of rumen fluid in the syringe that was closed with a T-valve for immediate dissolved gas extraction.Headspace gas was sampled after the cow had been standing up for at least 2 min and as the first procedure at every time point.A silicone tube terminating in a suction head was threated through a hole in the rumen cannula (tightened with silicone) and attached to the top of the internal cannula collar.A 3-way valve was mounted on the outside end of the tube, wherefrom samples were taken using a 20 mL syringe.To prevent contamination with residual air in the device, 10-15 mL was discarded, whereafter the next syringe volume was aspirated back into the rumen.Then a 15 mL sample was taken, and 5 mL was discarded through a needle, and the remaining 10 mL was transferred to an evacuated glass vial (Labco Limited) for later gas analysis.
Blood was sampled by venipuncture of the tail vein into Na-heparin vacutainers on d 2 and 7 at 2 time points relative to dosing (1.5 and 5 h after dosing).Vacutainers were placed on ice until centrifugation at 3000 × g av at 4°C for 20 min and plasma was stored at −20°C until analysis.respiration chambers as described by Thorsteinsson et al. (2023).Chamber and cow were confounded within experiment.Before, during, and after the experiments, a sequence of recovery tests for CH 4 and CO 2 were performed and used to correct measured values at chamber and gas level.For Exp. 1, the recovery across the used tests averaged 99.9 ± 1.38 and 99.6 ± 1.17% for CH 4 and CO 2 , respectively.For Exp. 2, the recovery percentage for CH 4 and CO 2 averaged 100.7 ± 1.66 and 100.1 ± 1.34%, respectively.For other gases besides CH 4 and CO 2 , an average recovery percentage of these 2 gases was used for the correction.Entering/leaving the chamber in relation to samplings was through an air-lock system.Since Exp. 1 originally was designed to obtain measurements of dissolved H 2 that corresponded with gas exchange measurements from chambers, the resolution of chamber measurements was increased on d 2 and 3 of the experimental period by increasing the actual time of measuring in the specific chamber.Therefore, all cows were measured during d 1 in each period, whereafter and with greater resolution, only 2 cows were measured on d 2 and the 2 remaining cows were measured on d 3.This resulted in a total of 2 d of measuring per cow per period in this experiment.In Exp. 2, measurements were obtained over 5 consecutive days.

Laboratory analyses
Dry matter concentration in feed ingredients and TMR samples was determined by drying at 60°C for 48 h (Åkerlind et al., 2011).Sample preparation and chemical analyses (ash, CP, crude fat, starch, NDF, and indigestible NDF (iNDF)) in feed samples was performed as described in Maigaard et al. (2024a).Digestibility of OM in silages and concentrates was determined as also described in Maigaard et al. (2024a).
For the analysis of dissolved gases in rumen fluid in Exp. 2, the procedure followed Wang et al. (2014).Briefly, a 20-mL syringe with 10 mL of pure gas N 2 (>99.99%) was connected with a valve to the 50-mL syringe containing the rumen fluid (35 mL).The 10 mL gas N 2 was injected into the syringe with rumen fluid and the syringe was shaken for 5 min with 300 rpm using an orbital shaker (IKA-KS 501) to extract the dissolved gases into the gas phase attaining an equilibrium between the liquid and gas phases.After shaking, the extracted gas was transferred to the 20 mL-syringe and the volume of gas was recorded before transfer of 10 mL to an evacuated glass vial for gas analysis.The temperature of the rumen fluid was measured using an infrared thermometer (Max, Schou Company A/S, Denmark) and the salinity was measured using a handheld salinity tester (Pocket Pro, Hach, Germany).
Concentrations of CO 2 in rumen headspace samples and gas samples from rumen fluid gas extraction (Exp. 2) was measured using gas chromatography (GC; Trace 1310, Thermo Scientific) with a split/splitless injector at 150°C and a thermal conductivity detector (TCD) at 200°C using a 30 m × 0.25 mm × 8 µm Rt-Q-BOND column (Restek) with helium as carrier gas at 1.7 mL/ min flow.The GC was programmed with a ramped temperature starting at 50°C for 1.75 min and then increased from 50 to 150°C at 20°C/min.For determination of CH 4 , H 2 , O 2 and N 2 concentrations, a split/ splitless injector at 150°C and a TCD at 200°C was used.A 30 m × 0.53 mm × 50 µm TG-BOND Msieve 5A column (Thermo Scientific) was used with argon as carrier gas at 4 mL/min flow.The GC was programmed with a ramped temperature starting at 80°C for 7 min and then increased from 80 to 150°C at 20°C/min.
In rumen fluid, VFA was analyzed using GC (Trace 1310, Thermo Scientific) as described in Kjeldsen et al. (2023).Glucose and L-lactate in rumen fluid was analyzed with membrane-immobilized substratespecific oxidases (YSI 2900D Biochemistry Analyzer, YSI Inc.).A Randox Ammonia Kit-AM1015 was used to determine NH 4 in phosphate buffered samples.In Exp. 2, formate, succinate, benzoate, and heptanoate was analyzed in samples from d 7 taken at 3 out of the 8 time points relative to dosing (−1.0, 1.5 and 5.0 h).These were analyzed as described in Jensen et al. (1995)  Alcohols in rumen fluid and plasma (Exp.2) was analyzed using headspace GC-MS according to Kristensen et al. (2007).

Calculations and Statistical Analyses
In both experiments, feed intake and milk yield were averaged over the days within the experimental periods (i.e., 3 and 7 d for Exp. 1 and 2, respectively).Yieldweighted averages of milk fat, protein and lactose concentration of milk was used for the calculation of energy-corrected milk yield (ECM; 3.14 MJ/kg) according to Sjaunja et al. (1991).Gas emissions from respiration chambers were averaged over the days of gas measurements (i.e., 2 and 5 d for Exp. 1 and 2, respectively).Calculation of gas yield (g/kg of DMI) was based on DMI from the corresponding period.Thus, data on feed intake, milk yield and gas emission represented each cow within period in each experiment and summed up to 16 observations per experiment.Statistical analyses were performed using R 4.2.2 (R Core Team, 2022) and no observations were discarded for the analysis.For the analysis of feed intake, milk yield and gas emissions, a Maigaard et al.: Hydrogen acceptors during methane mitigation linear mixed model fitted with REML was applied for each experiment using the lmer function from the lme4 package (Bates et al., 2015): where Y ijk is the dependent response variable, µ is the overall mean, TREAT is the fixed effect of treatment (Exp.1: i = CON-1, FUM-1, ACR-1, FUM+ACR-1.Exp.2: i = CON-2, FUM-2, ACR-2, PHL-2), P is the fixed effect of period (k = 1 to 4), δ is the random effect of cow (c = 1 to 4), and ε ikc is the random residual error assumed to be independent and normal distributed.
Calculations of concentrations of dissolved gases (dCH 4 and dH 2 ) in rumen fluid were according to Wang et al. (2016b): where C dgas is the dissolved gas in rumen fluid (dCH 4 , µM, or dH 2 , µM), C dgas is the gas concentration measured in the gas phase of the 20-mL syringe at equilibrium after extraction (gCH 4 , µL/L, or gH 2 , µL/L), V g is the gas volume (mL) in the 20-mL syringe after extraction, assuming 1 atm in the gas phase, V l is the volume (mL) of the rumen fluid in the 50-mL syringe, α gas is the Bunsen absorption coefficient (L/L) for CH 4 and H 2 , calculated according to Wiesenburg and Guinasso (1979): where T is the temperature in K (273 + temperature in °C).Concentrations of dissolved CO 2 in rumen fluid was estimated combining equations from Wang et al. (2016b) and Hille et al. (2016) as described in Wang et al. (2019) with minor modifications: 6 8346 1 2817 10 3 7668 10 where C TdCO 2 is the total dCO 2 concentration in the original rumen fluid (mM), C eTdCO 2 is the total dCO 2 concentration in the rumen fluid at equilibrium after extraction (mM), C gCO 2 is the gCO 2 concentration measured in the gas phase of the 20-mL syringe at equilibrium after extraction (L/L), V g is the gas volume (mL) in the 20-mL syringe after extraction, V l is the volume (mL) of the rumen fluid in the 50-mL syringe, pK CO 2 is the dissociation constant of bicarbonate set to be 6.11, α CO 2 is the Bunsen absorption coefficient for CO 2 (mol/L•atm), and T is the temperature in K (273 + temperature in °C).
To account for variation in contamination with atmospheric air in rumen headspace samples from Exp. 2, these were corrected by assuming zero concentrations of O 2 and N 2 in rumen headspace.Rumen headspace O 2 and N 2 concentration of uncorrected samples averaged 2.58 ± 3.71% and 11.6 ± 14.5%, respectively.
In both experiments, hourly gas emissions (CH 4 , H 2 and H 2 to CO 2 ratio) were averaged within hour per cow per period to obtain 24 observations per cow per period.Measurements were adjusted to align time of feeding/dosing between cows in chambers since chamber procedures (milking, feeding etc.) were carried out chamber by chamber with a time span of approximately 1 h from cow 1 to 4. Hourly gas emissions, rumen variables (4 observations per cow per period in Exp. 1 and 8 observations (d 7) per cow per period in Exp. 2), and plasma alcohol concentrations (4 observations per cow per period in Exp.2; d 2 and 7) were analyzed using a linear mixed model fitted with REML and the lme function: where Y, µ, TREAT, P, δ, and ε is as described above, HOUR ij is the fixed effect of hour of the day (l = 1 to 24 for gas data) and hour relative to feeding/dosing (j = 0.0, 1.0, 3.0 and 5.0 h and j = −1.0,0.0, 0.5, 1.5, 2.0, 3.0, 5.0, 7.0 h for rumen data in Exp. 1 and 2, respectively), (TREAT × HOUR) ij , is the 2-way interaction effect between TREAT and HOUR.A compound symmetry structure was chosen to account for correlation among repeated measurements within period for each cow, as this resulted in a lower Akaike's information criterion value than a continuous first-order autoregressive structure.
To investigate possible adaptation of rumen variables throughout each period in Exp. 2, measurements obtained at corresponding time points on d 2 and 7, respectively, (−1.0, 0.5, 1.5 and 5.0 h) were analyzed using a similar model as above but also included the fixed effect of sampling day and the 2 2-way interaction effects between treatment and sampling day and hour relative to feeding and sampling day.The correlation structure (compound symmetry) accounted for repeated measures within cow, day, and period.
Estimated marginal means (EMM) and standard error of the means were computed using the emmeans package (Lenth, 2023) and a type 2 ANOVA were used to report P-values.Statistical differences were declared if P ≤ 0.05 and tendencies were declared when 0.05 < P ≤ 0.10.Pairwise comparisons were based on Tukey's multiple comparisons.Normality of residuals and homogeneity of variance were evaluated using qq-plots and residual plots, respectively.Dissolved H 2 , hourly H 2 emission and hourly H 2 to CO 2 emission had heterogeneous variance and data was log-transformed (natural log; ln) to obtain homogeneity of variance.The EMMs from the model without log-transformation were presented in the tables and figures along with the P-values or pairwise comparisons associated to the logtransformed data.
For investigation of correlations between rumen dissolved gases and headspace proportions of gases (CH 4 , H 2 , and CO 2 ), linear functions were fitted using data obtained from rumen samples from d 2 and 7 of the experimental periods of Exp. 2. For correlations between headspace gas proportion and average gas yield obtained from respiration chambers in Exp. 2, the average headspace gas samples obtained on d 7, along with the corresponding average gas yield was used for fitting the linear functions.

Intake and Milk Yield
In Exp.1, cows on FUM-1 and FUM+ACR-1 had greater DMI than cows on ACR-1, however not significantly greater than cows on CON-1 Similarly, cows on ACR-1 did not show significant lower DMI than cows on CON-1, although the numerical decrease was 11% (P = 0.12; Table 2).Milk and ECM-yield did not differ among treatments (all P > 0.10) and averaged 34.4 ± 5.34 kg milk and 33.2 ± 3.22 kg ECM, across treatments.Milk fat, protein, and lactose concentrations did not differ among treatments (all P > 0.10), however milk urea concentration was greater in ACR-1 cows compared with CON-1 cows (4.51 vs. 4.08 mmol/L; P = 0.04).

Gas Exchange
Results on gas exchange from both experiments have been reported in Table 2. Figure 1 (A-F) illustrates hourly CH 4 and H 2 emission (g/h) from respiration chambers and hourly ratio between H 2 and CO 2 , to correct for possible changes in DMI within each hour.
Across treatments, CH 4 production varied from 319 to 345 g/d in Exp. 1, and there was no treatment effect.Production of CO 2 (g/d) was lower in cows on ACR-1 compared with CON-1, whereas yield (g/kg of DMI) of CO 2 was only numerically greater in cows on ACR-1 compared with CON-1.There was no treatment effect on H 2 production (g/d) nor H 2 yield (g/kg of DMI), and the level of H 2 production varied from 2.48 to 4.71 g/d, across treatments.For hourly CH 4 emission (g/h), a TREAT × HOUR interaction (P = 0.02; Figure 1A) reflected the diurnal variation in CH 4 emission related to feeding/dosing.For hourly H 2 emission (g/h), there was a tendency for a TREAT × HOUR interaction (P = 0.09; Figure 1B).This was a result of a peak in H 2 emission for CON-1 compared with the other treatments for up to 1 h after morning feeding and again 1-2 and 5 h after afternoon feeding.In addition, hourly H 2 emission seemed lowered for FUM-1, ACR-1 and FUM+ACR-1 especially after afternoon feeding and dosing, compared with the level around and before feeding (16.45 h).A similar pattern was observed for the hourly ratio of H 2 to CO 2 emission (Figure 1C).
Across treatments in Exp. 2, CH 4 production varied from 102 to 203 g/d and daily CH 4 production was 49.8 and 35.0%lower in cows on ACR-2 and PHL-2, respectively, compared with CON-2, but when corrected for DMI (g CH 4 / kg of DMI), these treatment differences were only tendencies (P = 0.07).Production of CO 2 (g/d) was lower in cows on ACR-2 compared with CON-2, whereas yield (g/kg of DMI) of CO 2 was greater in cows on ACR-2 compared with CON-2.The level of H 2 production varied from 19.2 to 25.5 g/d, across treatments, and there were no treatment effects  for daily production (g/d) nor yield of H 2 (g/kg of DMI).
There was a TREAT × HOUR interaction for hourly CH 4 emission (P < 0.01; Figure 1D), which revealed differences in CH 4 emission rate between treatments, within hour.For hourly H 2 emission, a significant interaction (P < 0.01; Figure 1E) showed that the emission peaked 1 h after both morning and afternoon feedings for CON-2 and PHL-2.However, only H 2 emission for ACR-2 in the morning but both FUM-2 and ACR-2 in the afternoon, were lower than CON-2 at 1 h after feeding.For hourly H 2 to CO 2 ratio (Figure 1F), these differences were only present for FUM-2 compared with CON-2 in the afternoon.

Rumen Variables and Plasma Alcohols
Total VFA concentration in Exp. 1 was lower in cows on ACR-1 compared with cows on CON-1, whereas total VFA concentration was similar for cows on FUM-1, FUM+ACR-1, and CON-1 (Table 3).Acetate proportion (% of total VFA) was greatest in cows on ACR-1 and lowest in FUM-1, though both groups were not different from CON-1 and FUM+ACR-1.For propionate, butyrate, and caproate proportions, there were a TREAT × HOUR interaction.Figure 2A, 2B, and 2C illustrates proportions of acetate, propionate, and butyrate over the hours relative to feeding/dosing in Exp. 1. Propionate proportions increased 1 h after feeding for ACR-1 and FUM+ACR-1 cows, compared with CON-1 and FUM-1 cows, but there was no difference among treatments after 3 and 5 h after feeding.Propionate proportions seemed to increase later relative to feeding (3 h) for FUM-1 compared with CON-1 cows, although pairwise comparisons revealed no statistical difference.For butyrate, proportions were lowest for ACR-1 and FUM+ACR-1 cows 1 h after feeding compared with CON-1, and only ACR-1 remained lowest throughout the sampling period.For valerate proportions, values were generally lowest for ACR-1 and FUM+ACR-1 cows, compared with CON-1 and FUM-1.L-lactate concentration was greater in cows on FUM-1 compared with CON-1, and numerically increased in cows on FUM+ACR-1, compared with CON-1 cows.
In Exp. 2 there were no TREAT × DAY interaction and no effect of sampling day for all rumen VFA proportions (except propionate), glucose, L-lactate and ammonium concentrations, pH and salinity (Table S2, Maigaard et al. (2024b)).For propionate proportion, there was no difference among treatments on d 2, but on d 7, cows on ACR-2 had greater rumen propionate proportions than cows on CON-2.A similar interaction revealed that cows on ACR-2 had greater dissolved CH 4 concentration and tended to have greater headspace CH 4 compared with cows on CON-2 on d 2, whereas no treatment differences were observed on d 7.
Table 4 summarizes rumen fermentation characteristics from cows in Exp. 2 sampled on d 7. Total VFA concentration was lower in cows on ACR-2 compared with cows on CON-2, whereas cows on FUM-2 and PHL-2 did not differ in total VFA from cows on CON-2.There was a TREAT × HOUR interaction for acetate, propionate, and butyrate proportions (Figure 2D, 2E, and 2F).Acetate proportions were similar across treatments until 3 h after feeding, where FUM-2 cows had lower acetate proportions than CON-2 and PHL-2 cows, but similar to ACR-2 cows.Cows on PHL-2 did not show greater acetate proportions than CON-2 cows at any time point but showed numerically greatest acetate proportion from 1.5 to 5 h after feeding.Propionate proportion increased for ACR-2 cows after feeding and peaked after 1.5 h, where the proportion of propionate was greater than for cows on CON-2, FUM-2, and PHL-2.Propionate proportions for ACR-2 cows remained greater than for CON-2 cows throughout the remaining sampling period.At 1.5 h after feeding, propionate proportion also peaked for FUM-2 cows and was greater than for cows on CON-2 and remained greater until 5 h after feeding.At these time points, butyrate proportion was lowest for ACR-2 cows compared with cows on CON-2 and remained lower from 0.5 h to 2 h after feeding.A similar pattern appeared for cows on FUM-2, however, these values were only numerically lower than for cows on CON-2.For succinate, a TREAT × HOUR interaction was a result of a peak in succinate concentrations for FUM-2 cows of 3.52 mM at 1.5 h after feeding, where succinate proportions averaged 0.09 mM for the other treatments.At 5 h after feeding, succinate proportion for FUM-2 was not different from cows fed CON-2.L-lactate concentration was increased in FUM-2 cows compared with CON-2 cows.For formate, ACR-2 cows had generally greater concentrations than CON-2 cows, especially at 5.0 h after feeding, which was reflected in a TREAT × HOUR interaction.Concentrations of dissolved CH 4 and proportions of headspace CH 4 were generally greatest in cows on CON-2, although multiple comparisons between treatments showed no differences.Dissolved concentration of H 2 was not different between treatments, whereas headspace H 2 proportion was greater in cows on ACR-2 compared with CON-2, whereas cows on FUM-2 and PHL-2 had similar headspace H 2 proportion as CON-2.Rumen and plasma alcohols have been presented in Table 4. Rumen ethanol, methanol, and 1-propanol concentrations were generally lowest before feeding (−1 and 0 h), increased shortly after feeding (0.5 h), and remained increased throughout the sampling period.Rumen methanol concentration was different and rumen ethanol concentration tended to be different between treatments.Rumen methanol concentrations were 2.2-fold greater in cows on PHL-2 compared with cows on CON-2 and plasma methanol concentration was also greater in PHL-2 cows compared with CON-2 cows.

Relationship between Measures of Gas
Figure 3A and 3B illustrate the correlations between rumen dissolved concentrations and headspace proportions of CH 4 and H 2 , respectively, in the rumen of cows sampled on d 2 and d 7 (n = 192) in Exp. 2. For CH 4 , a positive and strong correlation was observed (Figure 3A; r = 0.84), whereas only a moderate correlation was observed for H 2 (Figure 3B; r = 0.41).For CH 4 and H 2 , there were strong and positive correlations between headspace gas proportion and average chamber gas yield (r = 0.68 (Figure 3C) and r = 0.78 (Figure 3D) for CH 4 and H 2 , respectively).For CH 4 , a positive and moderate relationship between dissolved concentration and the corresponding average gas yield from respiration chambers was observed (r = 0.54, Figure 3E).For H 2 , there was no relationship between these measures, and the correlation coefficient was weak (r = 0.28, Figure 3F).For CO 2 , there was no correlation between dissolved concentrations and headspace proportions (r = 0.08).In addition, the relationship between rumen pH and dissolved CO 2 showed that dissolved CO 2 concentrations increased (min = 15.8 mM, max = 71.2mM) as rumen pH increased (min = 5.38, max = 7.09) following a curvilinear convex relationship (data not shown).

DISCUSSION
In the present experiments, the duration of the individual experimental periods was short (3 and 7 d for Exp. 1 and 2, respectively).This hinders evaluation of long-term effects, thus only acute and short-term responses to acceptors for H 2 in a CH 4 mitigated rumen environment were evaluated and discussed in the following.

Fumaric Acid Supplementation
The decrease in H 2 emission rate (g/h) upon fumaric acid supplementation shortly after feeding in Exp. 2 was accompanied by an increased propionate concentration in rumen fluid roughly at the same time point, although the increase in propionate lasted for up to 5 h after feeding, whereas H 2 emission rate only was different at 1 h after feeding.This was also the case for H 2 emission rate (g/h) in Exp. 1, while there was no immediate effect on propionate.Assuming that fumaric acid quickly associates to the liquid fraction in the rumen, where the passage rate out of the rumen is fast (15%/h; Krämer et al. (2013), this suggests that the added fumaric acid was either washed out or converted to succinate (which peaked 1.5 h after feeding) within hours from dosing.If H 2 availability is a kinetic limitation for the H 2 capturing effect of fumaric acid (Schulmand Cows were supplemented with water (CON-1), fumaric acid (FUM-1), acrylic acid (ACR-1) or the combination of FUM and ACR (FUM+ACR-1) in Exp. 1. and Valentino (1976), and the substrate availability (fumaric acid) is also low due to potential wash-out, this could explain why there was no additional continuation in the decrease in H 2 emission rate in the hours following the instant significant treatment response at 1 h after feeding/dosing.Hence, the overall effect on H 2 yield (g/kg of DMI) only decreased numerically by 38 and 6% upon fumaric acid supplementation in Exp. 1 and 2, respectively.The numerical decrease in Exp. 1 is in line with our previous results, when nitrate and fumaric acid were fed in combination for a longer duration (Maigaard et al., 2024c).Yet, the diurnal profiles of H 2 emission in both experiments suggested that the decrease in H 2 emission occurred relatively shortly after feeding in both experiments (within 1 h).However, if hourly DMI (eating rate) was affected by the sudden Cows were supplemented with water (CON-2), fumaric acid (FUM-2), acrylic acid (ACR-2) or phloroglucinol (PHL-2) in Exp. 2.

2
Based on 3 samples per cow per period (−1.0, 1.5 and 5.0 h relative to feeding/dosing). 3 Means and SEM are from model without log-transformation, whereas P-values are from log-transformed model.
pulse-dosing of the treatment, this could have biased the results and thus be the reason for a somewhat lower peak in H 2 emission immediately after feeding.Using hourly CO 2 emission as proxy for DMI, hence rumen fermentation rate, the hourly ratio between H 2 and CO 2 would (partly) account for such possible changes in eating rate, and these results (H 2 to CO 2 ratio) do confirm the pattern observed for hourly H 2 emission.
Although direct statistical comparisons between the 2 experiments is impossible, the greater response in propionate proportions in Exp 2. may also be related to the greater H 2 availability from CH 4 mitigation using 3-NOP compared with nitrate.
Fumaric acid can potentially be metabolized to acetate, yielding H 2 rather than consuming H 2 (Ungerfeld et al., 2007).Such effects would have counteracted the effects of fumaric acid on H 2 emission, but since rumen acetate proportions decreased upon fumaric acid supplementation (at 3 and 5 h after feeding) in Exp. 2, it is unlikely that fumaric acid was metabolized to acetate in significant amounts in the present experiments.
In both experiments, rumen L-lactate concentrations were increased 2.8 and 3.6-fold by fumaric acid supplementation, compared with CON-1 and CON-2, respectively.Lactate is another intercellular electron carrier, that captures electrons in the reduction of pyruvate to lactate.Therefore, it appears unexpected that lactate concentrations increased, if fumaric acid redirected excess electrons to its own reduction, which would have left less electrons for the reduction of pyruvate to lactate, thus a lower lactate concentration.The rumen concentration of lactate is normally low, but accumulates if the production rate of pyruvate surpasses the utilization rate of lactate to VFA (Ungerfeld, 2020).Fumaric acid may instead serve as a catalyzer for the rumen fermentation being easily fermentable energy for specific microbes, which thereby increased the overall fermentation rate, although this was not reflected in increased total VFA concentrations, yet VFA absorption rates are unknown.
Increases in rumen alcohol concentrations have previously been reported as a consequence of CH 4 mitigation (Guyader et al., 2017;Kjeldsen et al., 2023), reflecting increased ratio of co-factors in the reduced (NADH) compared with oxidative (NAD + ) state (van Lingen et al., 2016).However, if fumaric acid supplementation contributed to decrease the NADH to NAD + ratio by capturing H 2 , this was not reflected in rumen alcohol concentrations in the present study, since alcohol concentrations did not differ between FUM-1 and CON-1 cows.

Acrylic Acid Supplementation
Total VFA concentration was reduced upon acrylic acid supplementation as a consequence of reduced DMI, and changes in VFA profile further supported limitations in fermentable substrate or available energy with lower proportions of butyrate, valerate, and caproate, and increased proportions of branched-chain VFAs (isobutyrate and isovalerate; Exp. 1) as a consequence of microbial CP degradation (Allison, 1978;Tamminga, 1979).Nevertheless, propionate proportion increased 1 h after dosing in Exp. 1, whereas the increase remained significant and remarkable from 0.5 h after dosing in Exp. 2. Along with similar changes in hourly H 2 emission rate as was observed for fumaric acid, this suggests that some H 2 was redirected to the reduction of acrylic acid to propionate.In Exp. 1, the combined supplementation of fumaric acid and acrylic acid (at 50% of the molar quantity of each treatment) was hypothesized to result in effects equal to the sum of the individual effects of fumaric acid or acrylic acid.Evaluated across variables, this seemed to be confirmed.
As for fumaric acid, there were no significant changes in H 2 emission (production and yield) due to acrylic acid supplementation.In Exp. 1, a numerical reduction in H 2 emission was observed, whereas in Exp. 2, numerical increases in H 2 yield and H 2 headspace proportions were observed.Thus, increased rumen H 2 explains the elevated formate concentrations with formate normally being a sink for H 2 .In both experiments, we observed huge animal variation in response to the treatments (data not shown), which possibly explains the lack of significant effects on H 2 .In Exp. 2, one animal on ACR-2 had on average 80% greater H 2 yield than the average of the 3 other animals on ACR-2, and the difference between individual days was even greater, indicating a great day to day variation.Similarly in Exp. 1, one cow on CON-1 had 100% greater H 2 yield than the average of the 3 other animals on the same treatment, averaged across the chamber period.We did not observe any effects that could be ascribed to such variation in H 2 emission levels, and it is perceived to be a true biological response that needs further investigation.
To our knowledge, there is no available literature documenting the effects of in vivo application of acrylic acid.In the present experiments, acrylic acid decreased cows' DMI rather substantial, yet the decrease only appeared significant in Exp. 2. When investigating the changes in DMI over the days of the periods, the decrease in DMI appeared acute and reversible, yet the periods were short (data not shown).Both acrylic acid and fumaric acid are relatively weak acids with acid dissociation constants (pK a ) of 4.25 and 3.03, respectively and increased ruminal pH shortly after dosing rules out that the observed DMI reductions were related to acute rumen acidosis due to acid dosing.However, acrylic acid is highly reactive and can cause skin irritations (Thermo Fisher Scientific, 2009), wherefore the suspension was dosed in the central rumen content to prevent direct epithelial contact.Newbold et al. (2005) reported no inhibiting effect on microbial numbers and DM degradation when adding acrylic acid as sodium salts.However, such measures were not obtained in the present experiments to elucidate if acrylic acid as an acid negatively would have affected the rumen microbial environment.Yet, it is unknown why acrylic acid decreased DMI substantially.These DMI reductions also explained the reductions in CH 4 production (g/d) since the decreasing effects diminished when CH 4 production was corrected for DMI.In addition, decreased CO 2 production and increased CO 2 yield along with increased oxygen consumption (g/kg of DMI) support that DMI was reduced with mobilization of adipose tissue as a compensation.

Phloroglucinol Supplementation
It was expected that supplementation of phloroglucinol during inhibition of methanogenesis using 3-NOP would result in decreased emission of H 2 along with increased proportions of acetate in rumen fluid.Such effects were observed in vitro using halogenated compounds as methanogenesis inhibitor (Huang et al., 2023;Romero et al., 2023), and similar promising effects have been documented in vivo in steers (Martinez-Fernandez et al., 2017) and dairy goats (Romero et al., 2022), where H 2 yield (g/kg of DMI) was reduced with 50 and 74%, respectively when phloroglucinol, at similar inclusion rates as in the present study, was combined with CH 4 inhibitors.Though, there were no indications of H 2 capture in the present experiment since there was no effects on H 2 expelled nor rumen H 2 (dissolved or in headspace), not even at time points shortly after feeding where phloroglucinol and H 2 availability was expected to be excessive.Alternative reducing elements may originate from formate (Krumholz et al. (1987), which most likely is why Martinez-Fernandez et al. (2017) found remarkably lower formate concentrations in the rumen as a consequence of phloroglucinol supplementation.Contrary, rumen formate concentration was unaffected, but numerically increased in the present study, yet it should be noted that formate analysis was only based on 3 of the 8 samples on d 7.In contrast, concentration of heptanoate was 40% lower in cows on phloroglucinol, and since heptanoate synthesis from glucose requires 2 mol H 2 per mol of heptanoate (Guyader et al., 2017), this suggests limitations in H 2 availability which could be caused by competition from phloroglucinol, although heptanoate synthesis is of minor quantitative importance in the overall H 2 balance.In addition, stoichiometrically, 1 mol of phloroglucinol and 1 mol of H 2 produce 2 moles of acetate and 2 moles of CO 2 (Tsai et al., 1976).Cows received on average 22.8 ± 2.71 g phloroglucinol/kg of DMI, which corresponded to 0.18 mol/kg of DMI.A complete metabolization of phloroglucinol in the rumen would theoretically capture 0.364 g H 2 /kg of DMI.As such a decrease in H 2 yield was not observed, it could be speculated if the added phloroglucinol instead competed with methanogenesis for H 2 , and thereby explained the tendencies for a decrease in CH 4 yield of 3.09 g/kg of DMI (0.192 mol/kg of DMI) in PHL-2 treated cows.If 1 mol of CH 4 requires 4 moles of H 2 , this reduction would correspond to a capture of 0.770 mol H 2 /kg of DMI in methanogenesis, which is more than twice as much as the added phloroglucinol would theoretically capture.Nevertheless, as the stochiometric relationship showed production of CO 2 , and there were no indications of increased CO 2 production (neither as rumen dissolved or headspace gas or as expelled) and only small and insignificant increases in acetate proportions in the present study was observed, the added phloroglucinol most likely was not metabolized, although the applied dose (2.3% of DM) was similar to the previous studies (~3% of DM in Martinez-Fernandez et al. (2017) and 2% of DM in Romero et al. (2022)).Tsai and Jones (1975) provided evidence that the rumen microbial strains Streptococcus bovis and Coprococcus were capable of metabolizing phloroglucinol under rumen conditions.However, it appears likely that microbes with such capacity are in low abundancy in a rumen environment not adapted to phloroglucinol or other phenolic compounds (Petri et al., 2014).Huang et al. (2023) found that a sequential batch incubation for a longer duration showed decreasing effects of phloroglucinol on H 2 compared with a shorter 24 h incubation, which indicated that adaptation was important.In our case, this presumed need for adaptation was the main reason for the change toward longer period durations in Exp. 2 (7 d) compared with Exp. 1 (3 d).However, there was no interaction between the PHL-2 treatment and day for H 2 measures, that could have indicated a decreasing H 2 trend over time, and hence supported a need for adaptation, knowing that the 7 d period may already have been too short, compared with the 10 d of adaptation in the study by Romero et al. (2022) or 16 d in the study by Martinez-Fernandez et al. (2017).
It should be noted that rumen and plasma methanol concentrations was increased 2.2 and 3.9-fold respectively, upon phloroglucinol supplementation.Rumen and plasma alcohol concentrations have not previously been reported in studies with phloroglucinol.To our knowledge, methanol is not a direct product of phloroglucinol degradation (Tsai et al., 1976;Patel et al., 1981;Armstrong and Patel, 1994), yet apparently somehow related to phloroglucinol.

Relationship between Measures of Gas
Initiatives to investigate changes in ruminal dissolved gas concentrations during CH 4 mitigation and possible H 2 redirection was carried out in the 2 experiments.In Exp. 1, a H 2 sensor (H 2 -500 X; Unisense, Aarhus, Denmark) was mounted in the ventral rumen for continuous 6 h measurements, following a similar procedure as described in Guyader et al. (2015).For unknown reasons, the sensor response declined over time, thus data was omitted and not reported.From Exp. 2, spot measurements of dissolved gas concentrations obtained from extraction procedures were reported instead.Considering an assumed 30% CH 4 reduction by 3-NOP (Kebreab et al., 2023) in the present study, dissolved CH 4 concentrations were within the ranges of what has been reported elsewhere in dairy cows (Wang et al., 2016b;Wang et al., 2019), and similarly for dissolved H 2 when methanogenesis was inhibited with 3-NOP (Melgar et al., 2020;Alemu et al., 2023), considering the applied dose and the cows' production level.Rumen headspace gas proportions was corrected for possible contamination with ambient (atmospheric) air, assuming that any O 2 and N 2 in the sample originated from ambient contamination.A linear regression between N 2 and O 2 concentration in the samples from Exp. 2 showed a slope of 3.81 (r = 0.997; data not shown).The ratio between N 2 and O 2 in atmospheric air is 3.71 (78/21), which supported that the presence of N 2 and O 2 in the samples originated from ambient air.Thus, the correction eliminated possible bias between samples.However, such contamination could originate from either the sampling itself or be naturally present in the rumen due to engulfed air during eating/rumination or leaking cannulas.The latter has been suggested in studies observing decreased headspace concentrations of CH 4 an H 2 with increased concentrations of N 2 and O 2 in cannulated animals compared with intact animals (Moate et al., 2013;Wang et al., 2019), although the ratio between N 2 and O 2 differed substantially from atmospheric air in the study by Moate et al. (2013).The relationships between the different measures of gas seemed to differ for CH 4 and H 2 .Wang et al. (2016a) suggested that the rumen fluid is supersaturated with CH 4 and H 2 , meaning that headspace gas concentration of gases (consequently also expelled) may not fully reflect changes in dissolved gas concentrations.Although H 2 is produced by microbes present in the liquid phase of the rumen, H 2 gas poorly dissolves in water or rumen fluid because it is a non-polar gas, and since it does not ionize, the solubility is independent of pH (Hegarty and Gerdes, 1999).This may explain why several rumen samples showed zero or very low dissolved H 2 concentrations with corresponding headspace H 2 concentration of values greater than zero (Figure 3B), indicating that H 2 very quickly shifts from the liquid to the gas phase (headspace) in the rumen.This explains why dissolved H 2 correlated weakly (although positively) or did not correlate at all with the corresponding headspace or chamber measurements, respectively (Figure 3B and 3D), while there was a stronger positive correlation between headspace proportion and chamber measurements of H 2 (Figure 3F).Wang et al. (2014) similarly found no correlation between dissolved H 2 and headspace H 2 .For CH 4 , the overall correlation between dissolved gas and headspace was somewhat stronger (Figure 3A) than for H 2 and appeared also to be stronger for the values close to zero, compared with the corresponding H 2 correlation (Figure 3B).This was consequently reflected in the relationship between headspace CH 4 and chamber measurements as well.The Bunsen absorption coefficient for CH 4 in water (or rumen fluid; assuming zero salinity and 1 atm pressure) at 39°C is 0.02563 L CH 4 /L of water, whereas it for H 2 is nearly halved (0.01662 L H 2 /L of water) (Wiesenburg and Guinasso, 1979).This supports that the capacity for keeping the gas in the liquid phase is greater for CH 4 than for H 2 , assuming equal partial pressure, and may explain the differences in their dissolved vs. headspace relationships.The relationships between headspace concentrations of gases (CH 4 and H 2 ) and chamber measurements appears as expected, since the headspace gas is being expelled from the animal thus measured in the chamber, however, to our knowledge, these relationships have not previously been illustrated.
In contrast to CH 4 and H 2 , CO 2 does ionize in water (or rumen fluid) to produce carbonic acid (H 2 CO 3 ), which then dissociates into bicarbonate (HCO 3 -) and protons (H + ), which further can dissociate into carbonate (CO 3 2-) and 2 protons (H + ).These reactions (involving different phases of CO 2 ) play an important role in the rumens' buffering system, and pH is determining the phase of CO 2 within these reactions (Hille et al., 2016).This explains the relationship between dissolved CO 2 and rumen pH in the present study; at greater rumen pH, relatively more CO 2 is in the form of bicarbonate (HCO 3 -).This agrees with the lack of relationship between dissolved CO 2 and headspace CO 2 , since rumen pH affects whether CO 2 is in a form that can shift to the headspace (as CO 2 ) or not (as bicarbonate).

CONCLUSION
Pulse-dosing of fumaric acid and acrylic acid to cows fed 3-NOP or nitrate increased rumen propionate proportions for periods up to 5 h after dosing or more, and the effect seemed greater and more stable when CH 4 was inhibited using 3-NOP rather than nitrate, thus the effect may depend on the availability of H 2 .However, these effects were not associated with any noteworthy changes in gas measures (emission, dissolved, headspace), except for a reduced hourly H 2 emission rate at 1 h after dosing, or any changes in competing sinks for rumen H 2 .Phloroglucinol did not result in increased rumen acetate proportions and did not affect any H 2 measures on the short-term, and the results indicated that phloroglucinol was not degraded.Changes in H 2 headspace or emission were not good proxies for changes in dissolved H 2 in rumen fluid.

Figure 1 .
Figure 1.Diurnal methane (CH 4 ), hydrogen (H 2 ), and H 2 to carbon dioxide (CO 2 ) ratio emissions (g/h) for dairy cows in Exp. 1 (A, B and C) respectively, and in Exp. 2 (D, E and F), respectively.Cows in Exp. 1 were fed a diet with nitrate (15 g/kg of DM) and supplemented with water (CON-1), fumaric acid (FUM-1), acrylic acid (ACR-1) or the combination of FUM and ACR (FUM+ACR-1) twice daily through the rumen fistula and cows in Exp. 2 were fed a diet with 3-NOP (60 mg/kg of DM) and supplemented with water (CON-2), fumaric acid (FUM-2), acrylic acid (ACR-2) or phloroglucinol (PHL-2) twice daily through the rumen fistula.Each point represents Estimated Marginal Means of the variables within the following hour and error bars represent standard error of the means.Letters represent significant differences (P < 0.05) if the letters within hour differs.Letters in B, E and F are based on log-transformed estimates.

Figure 2 .
Figure 2. Rumen proportions of acetate, propionate and butyrate (% of total volatile fatty acids (VFA)) at different time points relative to feeding/dosing of dairy cows in Exp. 1 (A, B and C) respectively, and in Exp. 2 (D, E and F), respectively.Cows in Exp. 1 were fed a diet with nitrate (15 g/kg of DM) and supplemented with water (CON-1), fumaric acid (FUM-1), acrylic acid (ACR-1) or the combination of FUM and ACR (FUM+ACR-1) twice daily through the rumen fistula and cows in Exp. 2 were fed a diet with 3-NOP (60 mg/kg of DM) and supplemented with water (CON-2), fumaric acid (FUM-2), acrylic acid (ACR-2) or phloroglucinol (PHL-2) twice daily through the rumen fistula.Each point represents Estimated Marginal Means of the variables within the following hour and error bars represent standard error of the means.Letters represent significant differences (P < 0.05) if the letters within hour differs.

Figure 3 .
Figure3.Relationship between different measures of methane (CH 4 ) and hydrogen (H 2 ).A and B shows the relationships between dissolved (d) gas concentration (µM) and headspace (h) gas proportion (% of volume) of CH 4 and H 2 , respectively, in rumen if dairy cows in Exp. 2 that were fed a diet with 3-NOP (60 mg/kg of DM) and supplemented with water, fumaric acid, acrylic acid, or phloroglucinol, twice daily through the rumen fistula (n = 192).C and D shows the relationships between the average headspace gas proportion and average chamber gas yield (g/ kg of DMI) of CH 4 and H 2 , respectively, on d 7 in Exp. 2 (n = 16).E and F shows the relationship between the average dissolved concentrations and average chamber yield of CH 4 and H 2 , respectively, on d 7 in Exp. 2 (n = 16).Lines illustrate the linear regression (y = x) and P-values for the corresponding slope coefficient is presented along with the corresponding correlation coefficient (r) for the regression.
Maigaard et al.: Hydrogen acceptors during methane mitigation Maigaard et al.: Hydrogen acceptors during methane mitigation

Table 2 .
Maigaard et al.:Hydrogen acceptors during methane mitigation Gas exchange measurements of dairy cows in Exp. 1 fed a diet with nitrate (15 g/kg of DM) and in Exp. 2 fed a diet with 3-NOP (60 mg/kg of DM).Cows were supplemented with different treatments 1 twice daily through the rumen fistula

Table 3 .
Maigaard et al.:Hydrogen acceptors during methane mitigation Rumen fermentation characteristics of dairy cows in Exp. 1 fed a diet with nitrate (15 g/kg of DM) and supplemented with different treatments 1 twice daily through the rumen fistula

Table 4 .
Maigaard et al.:Hydrogen acceptors during methane mitigation Rumen fermentation characteristics, rumen dissolved and headspace gas, rumen alcohols and plasma alcohols of dairy cows in Exp. 2 on d 7 fed a diet with 3-NOP (60 mg/kg of DM).Cows were supplemented with different treatments 1 twice daily through the rumen fistula