Effect of a garlic and citrus extract supplement on performance, rumen fermentation, methane production, and rumen microbiome of dairy cows

The aim of this trial was to determine the effect of a garlic and citrus extract supplement (GCE) on the performance, rumen fermentation, methane emissions, and rumen microbiome of dairy cows. Fourteen multiparous Nordic Red cows in mid-lactation from the research herd of Luke (Jokioinen, Finland) were allocated to 7 blocks in a complete randomized block design based on body weight, days in milk, dry matter intake (DMI), and milk yield. Animals within each block were randomly allocated to a diet with or without GCE. The experimental period for each block of cows (one for each of the control and GCE groups) consisted of 14 d of adaptation followed by 4 d of methane measurements inside the open circuit respiration chambers, with the first day being considered as acclimatization. Data were analyzed using the GLM procedure of SAS (SAS Institute Inc.). Methane production (g/d) and methane intensity (g/kg of energy-corrected milk) were lower by 10.3 and 11.7%, respectively, and methane yield (g/kg of DMI) tended to be lower by 9.7% in cows fed GCE compared with the control. Dry matter intake, milk production, and milk composition were similar between treatments. Rumen pH and total volatile fatty acid concentrations in rumen fluid were similar, whereas GCE tended to increase molar propionate concentration and decrease the molar ratio of acetate to propionate. Supplementation with GCE resulted in greater abundance of Succinivibrionaceae , which was associated with reduced methane. The relative abundance of the strict anaerobic Methanobrevibacter genus was reduced by GCE. The change in microbial community and rumen propionate proportion may explain the decrease in enteric methane emissions. In conclusion, feeding GCE to dairy cows for 18 d modified rumen fermentation and microbiota, leading to reduced methane production and intensity without compromising DMI or milk production in dairy cows. This could be an effective strategy for enteric methane mitigation of dairy cows


INTRODUCTION
Methane (CH 4 ) is one of the main contributors to global greenhouse gas emissions and therefore to climate change. Compared with carbon dioxide (CO 2 ), CH 4 has a shorter lifetime in the atmosphere (12 vs. 1,000 yr), but 28 times higher 100-year global warming potential (EPA, 2020). The environmental impact of ruminant livestock has attracted increasing attention because of the direct emission of CH 4 from the fermentation of feed in the rumen. In the anaerobic environment of the rumen, complex feed compounds are fermented by microorganisms to VFA. During this process, CO 2 and hydrogen (H 2 ) are produced, which are then used by methanogens to produce CH 4 Ungerfeld, 2020). Enteric CH 4 not only affects the environment but also leads to a loss of, on average, 6% of the total feed energy consumed by high-producing dairy cows (Niu et al., 2018), which makes it an economic issue for farmers.
Research suggests that enteric CH 4 mitigation is possible either by inhibiting methanogenic archaea directly by reducing hydrogen (H + ) production or by alternative pathways to utilize H + (McAllister and Newbold, 2008;Martin et al., 2010). Different strategies to achieve CH 4 mitigation were tested in the recent past to use one of the above-mentioned mechanisms, with dietary manipulation and supplementation with feed additives being extensively researched (Haque, 2018;Sun et al., 2021). Dietary manipulations such as utilizing highquality forages (e.g., corn silage) in the diet (Hassanat et al., 2013) or replacing fiber with starch by including concentrates high in fermentable carbohydrates (Jiao et al., 2014) may result in significant CH 4 reduction. This is a common practice in the diets of high-yielding dairy cows to reduce their CH 4 production potential.

Effect of a garlic and citrus extract supplement on performance, rumen fermentation, methane production, and rumen microbiome of dairy cows
Supplementation of these diets with feed additives that either directly inhibit methanogens or modify rumen fermentation leads to less CH 4 production, which may result in a further reduction in CH 4 emission. According to a meta-analysis of enteric CH 4 mitigation strategies conducted by Arndt et al. (2022), 3 product-based strategies-increasing feeding level, decreasing grass maturity, and decreasing dietary forage-to-concentrate ratio-decreased CH 4 per unit of meat or milk by, on average, 12% and increased animal productivity by a median of 17%. Five absolute strategies-CH 4 inhibitors, tanniferous forages, electron sinks, oils and fats, and oilseeds-decreased daily CH 4 by, on average, 21%. Novel feed supplements such as the chemical 3-nitrooxypropanol (3-NOP; Melgar et al., 2020) as well as natural compounds like essential oil blends (Belanche et al., 2020) or seaweed (Roque et al., 2019a) have shown promising results in reducing CH 4 emissions from ruminants.
Other plant secondary compounds from garlic and citrus fruits have been researched to reduce CH 4 production by modifying the rumen microbiome. Garlic oil and its active compounds showed, in vitro, a strong reduction in CH 4 production and the relative abundance of methanogens (McAllister and Newbold, 2008;Patra and Yu, 2015) and a reduced acetate-topropionate ratio with similar VFA production (Busquet et al., 2005;Kamra et al., 2012) and OM digestibility (Soliva et al., 2011). These suggest a direct inhibition of rumen methanogens without affecting microbial fermentation. In the rumen of sheep, it was observed that feeding garlic oil increased the diversity of rumen methanogens (Ohene-Adjei et al., 2008) and feeding allicin extracted from garlic reduced CH 4 production and methanogens and tended to reduce protozoa (Ma et al., 2016). Citrus fruit extract and its active compounds showed lower CH 4 production, acetate-to-propionate ratio, and relative abundance of methanogens in vitro. In heifers (Balcells et al., 2012) and steers (Seradj et al., 2018) fed citrus extract, the rumen microbiome, especially the relative abundance of lactate-consuming bacteria, was modified, which was associated with an increase in rumen propionate and a decrease in lactate concentrations. In an in vitro experiment with a batch culture (Ahmed et al., 2021a) and rumen simulation technique (RUSITEC; Eger et al., 2018), supplementation with garlic granules (Allium sativum) and citrus extract (Citrus aurantium) (GCE) led to a decrease in the abundance of methanogenic archaea, a 54% or 95% reduction in CH 4 production, respectively, and an increase in VFA production. Dairy cows fed GCE showed significantly reduced CH 4 production and higher milk yield (Vrancken et al., 2019). The supplementation of GCE has also demonstrated a reduction in CH 4 produc-tion of crossbred beef steers (Roque et al., 2019b) and Holstein veal calves (Brand et al., 2021). Therefore, supplementation with GCE could be one solution to reduce CH 4 emissions from ruminants.
On the basis of these findings, we hypothesize that GCE might be able to reduce enteric CH 4 emissions of lactating dairy cows by altering the ruminal microbial community without affecting animal performance. The objective of this study was to investigate the effect of GCE on CH 4 emissions, performance, rumen fermentation, and rumen microbiome of lactating dairy cows.

Animals and Experimental Design
The experiment was conducted at the Natural Resources Institute Finland (Luke) under Regional State Administrative Agency permission ESAVI/7012/2019 in accordance with the guidelines established by the European Community Council Directive 2010/63/EU for animal experiments complying with the ARRIVE guidelines (Kilkenny et al., 2010). Fourteen healthy multiparous Nordic Red cows in mid-lactation from the research herd of Luke (Jokioinen, Finland) were used in this experiment. The sample size of 7 replicates per treatment was calculated before the study by assuming the power of the study (1 − β) as 0.85, the level of significance (α) in a one-sided test to be 0.05, and an assumption to detect at least a 10% reduction in daily CH 4 production. To minimize the differences between the groups, the animals were allocated to 7 blocks in a complete randomized block design on the basis of BW (650 ± 49 kg), DIM (110 ± 37 d), DMI (28.2 ± 1.6 kg/d), and milk yield (39.8 ± 3.9 kg/d). The 2 animals within each block were randomly allocated to a diet without (control; CTRL) or with GCE applied at 44 g/d (Table 1). The GCE consisted of garlic and citrus extract (50%) and limestone and vegetable oil (50%); it contained 960 g/kg DM as fed, and 377 g/kg ash, 79 g/ kg CP, 53 g/kg ether extract, and 151 g/kg crude fiber on a DM basis. Half of the daily GCE (22 g/d) was mixed with 3 kg of TMR offered in the morning (0800 h) and the other half during the evening (1630 h) feeding, and animals were allowed to consume the mixtures for 2 h. The proportion of consumed TMR was recorded, and the leftovers were top-dressed on the main TMR for the rest of the feeding period. The same procedure was followed for the CTRL treatment without adding GCE. The main portions of TMR were delivered to cows by an automatic feeding wagon (TR Feeding Robot, Pellon Group Ltd.) at 1000, 1300, 1830, and 2000 h. The TMR had a forage-to-concentrate ratio of 50:50 on a DM basis and comprised grass silage and concen-4610 trate pellets. The formulation of TMR and the nutrient composition of grass silage and concentrate pellets are presented in Table 1. The grass was ensiled from swards of mixed timothy (Phleum pratense) and meadow fescue (Festuca pratensis) with a formic acid-based ensiling additive (AIV 2 Plus, Eastman Ltd.) applied at a rate of 5.0 L/t of fresh grass. The grass was lightly wilted before ensiling to achieve a DM content of 25%. All cows received the same TMR. The cows were housed in a freestall barn (140 freestalls) with controlled access to their feed using individual feed bins controlled by neck collars and automated opening gates (RIC, Insentec), with free access to water and salt blocks during the adaptation period. The cows were milked twice daily at 0700 and 1645 h in a 2 × 6 tandem milking parlor. The animals were monitored daily for health problems, and any abnormalities and infections were recorded and treated according to the general barn guidelines. All cows completed the whole experimental period (18 d for each cow and 60 d in total) without major symptoms that could affect the results. The experimental period for each block of 2 cows (one for each of the CTRL and GCE groups) consisted of 14 d of adaptation followed by 4 d of CH 4 measurements inside the open circuit respiration chambers, with the first day being used for acclimation. During the chamber measurements, the cows were milked in situ. Each block of 2 cows entered the experiment 7 d after the previous block to ensure the availability of respiration chambers for CH 4 measurements. The cows were restrained using a neck yoke in a dedicated platform (180 × 126 cm) covered with a rubber mat and had continuous access to experimental feeds, fresh water, and salt blocks. The cows from each block were randomly allocated to different chambers.

Measurements and Chemical Analyses
Feed refusals were measured on a daily basis, and daily feed intake was determined as the amount of TMR offered minus refusals for every cow. Cows were weighed on a daily basis except for 4 d during CH 4 measurements in respiration chambers by a walkthrough static scale (Pellon Group) every time they left the milking parlor.
Grass silage and concentrates were collected over sampling days, kept at 4°C, composited for each block, and stored at −20°C until chemical analysis using routine procedures as described by Ahvenjärvi et al. (2018). Silage DM content and consequently DMI were corrected for the loss of volatiles according to Huida et al. (1986). Milk yields were recorded gravimetrically (Pellon SAC) throughout the trial. Milk samples, taken from 4 consecutive milkings starting in the afternoon of d 16 for each block while the cows were in respira-tion chambers, were preserved with Bronopol tablets (Valio Ltd.) and stored at 4°C until analysis of fat, protein, lactose, total solids, urea, and SCC (MilkoScan FT6000, Foss Electric). The cows entered the respiration chambers in the morning (0900 h) of d 15 for 4 d to provide 3 full days of measurements. Over 3 d, gas exchanges were recorded by measuring concentrations of CH 4 , CO 2 , and H 2 in the inlet and outlet air using dedicated analyzers (Columbus Instruments) located in the Minkiö dairy barn (Jokioinen, Finland). Respiration chambers and their calibration are described in detail by Bayat et al. (2022). Briefly, gas analysis was configured to allow automatic measurements at 3.5-min intervals from each chamber and the reference air. The zero and span calibrations of analyzers were conducted at the beginning of every 3-d measurement for each block of cows using standard gases (AGA Ltd.). The concentration of measured gases and airflow (corrected based on standard temperature and pressure) were recorded and monitored (Oxymax v. 4.86,Columbus Instruments), and the data captured were used for further calculations. Environmental control of temperature across a range from 15 to 22°C and relative humidity of 50 to 70% was maintained through an adjustable air conditioning system.

Rumen Fermentation and Microbial Analysis
On the last day of the sampling period for each block, a 500-mL rumen fluid sample was collected from each cow at about 1000 h using stomach tubing (Ruminator, Khurana et (20); organic selenium (10); α-tocopheryl acetate (2,000); and d-biotin (30).
Geishauser). Rumen fluid collected from the ventral sac of the rumen was immediately measured for pH using a portable pH meter (VWR International). A subsample of 5.0 mL of rumen fluid was immediately preserved with 0.5 mL of saturated mercury (II) chloride and 2.0 mL of 1 M sodium hydroxide and stored at −20°C for VFA determination. Another subsample of 15.0 mL of rumen fluid for the determination of ammonia was immediately preserved with 0.3 mL of sulfuric acid (50%, vol/vol) and stored at −20°C. Rumen VFA and ammonia N concentrations were determined according to Huhtanen et al. (1998) andMcCullough (1967), respectively.
For microbial analyses, samples were aliquoted into 2-mL tubes, snap-frozen in dry ice, and stored at −80°C until DNA extraction. Total DNA was extracted from 0.5 mL of rumen liquid, as described by Rius et al. (2012). For microbial amplicon sequencing, universal primers 515F and 806R (Caporaso et al., 2011) targeting the bacterial 16S rRNA gene V4 region, and 316F and 539R primers (Sylvester et al., 2004) targeting the ciliate protozoa 18S rRNA gene were used. Libraries were prepared and sequenced on the Illumina MiSeq platform using 2 × 250 bp chemistry in the Finnish Functional Genomics Centre (Turku, Finland). Demultiplexing of sequences, adapter removal and sorting sequences by barcode was performed by the sequencing data provider. The sequence reads are available in the National Center for Biotechnology Information Sequence Read Archive under BioProject PRJNA872093.

Calculations
Energy-corrected milk was calculated using the equation proposed by Sjaunja et al. (1990) based on milk fat, protein, and lactose yields. Methane, CO 2 , and H 2 production were calculated by multiplying air flow and differences in gas concentrations in the inlet and outlet chambers. Methane yield was calculated as grams of CH 4 emitted per kilogram of DMI by individual animals per day, and CH 4 intensity was calculated as grams of CH 4 emitted per kilogram of ECM from individual animals per day. Data on SCC were log 10 -transformed before statistical analysis to ensure normal distribution.

Statistical Analysis
Data were analyzed using the GLM procedure of SAS (version 9.4, SAS Institute Inc.), with block and treatment included as fixed effects in the model. The natural variation between experimental animals was corrected by blocking the animals on average DMI and milk yield, collected 2 wk before the start of the trial when all the cows received the CTRL diet. Least squares means and standard error of means are reported, and differences between treatments were considered significant when P ≤ 0.05 and a tendency was recorded when 0.05 < P ≤ 0.10.
Rumen microbial community α-diversity was calculated using the Shannon index, Pielou evenness estimate, and observed number of ASVs. For α-diversity, all samples were subsampled to the same depth, equivalent to the lowest number of reads per sample. To evaluate whether treatment was significantly associated with rumen microbial community composition, between-sample diversity was calculated as Bray-Curtis dissimilarities following Hellinger transformation and visualized using principal coordinate analysis (PCoA) as implemented in the MicrobiotaProcess R package (Xu and Yu, 2021). The significance of groups was evaluated by distancebased permutational multivariate ANOVA (adonis) and defined at P < 0.05 level after 999 permutations.
The treatment effect on individual microbial taxa was evaluated by ANOVA as described above. Before the test, low-abundance taxa with <0.01% relative abundance across all samples were filtered out. The number of reads was log base transformed [log 2 (x + 1)] and standardized by data centering. For easier interpretation of results, the number of reads of microbial taxa identified as significantly affected by treatment in ANOVA analysis was converted to compositional data Khurana et al.: METHANE PRODUCTION FROM COWS FED ON GARLIC-CITRUS SUPPLEMENT and presented as relative percent abundances within each microbial category.

RESULTS
Dry matter intake was similar (P ≥ 0.66) between treatments ( Table 2). The intake of GCE and CTRL mash concentrates was not different (P = 0.18). The BW of cows tended (P = 0.08) to be lower for GCE than for CTRL. Milk yield (P = 0.80) and ECM yield (P = 0.57) were not different between treatments (Table 3). Similarly, protein, fat, and lactose concentrations and yields and SCC were unaffected (P ≥ 0.31) by treatment. Milk production efficiency calculated as milk yield or ECM divided by DMI was also not affected (P ≥ 0.34) by GCE.
Methane production (g/d) was 10.3% lower (P = 0.018) for GCE than for CTRL (Table 4). The GCE also indicated a tendency (P = 0.06) toward lower CH 4 yield (g/kg of DMI) of 9.7%. Methane intensity (g/ kg of ECM) was 11.7% lower (P = 0.039) for GCE compared with CTRL. Daily H 2 and CO 2 emissions did not differ (P ≥ 0.11) between the treatments.
Rumen pH and total VFA concentration were not affected (P ≥ 0.32) by GCE (Table 5). The molar concentration of propionate tended (P = 0.089) to be higher and that of caproate lower (P = 0.029) for GCE. Tendencies toward a lower molar ratio of acetate-topropionate and acetate plus butyrate-to-propionate (P ≤ 0.09) were observed for animals fed GCE.
Bacterial, archaeal, and ciliate protozoan community composition was explored at the phylum, family, and species levels, respectively. The α-diversity of rumen bacteria, archaea, and ciliate protozoa, estimated as  Intake of GCE-TMR mix with or without garlic and citrus extract supplement (0 or 44 g/d).  Sjaunja et al. (1990).
3 Data transformed using log 10 for statistical analysis and back-transformed for interpretation. Mean values in the same row with different superscripts differ (P < 0.05). 1 CTRL = control; GCE = garlic and citrus extract supplement. Mean values in the same row with different superscripts differ (P < 0.05). 1 Shannon index, Pielou evenness, and observed number of ASVs, was not affected (P ≥ 0.11) by treatment ( Table 6). Rumen bacterial and archaeal community structure, assessed by PCoA, did not indicate differences between CTRL and GCE treatments (adonis test P > 0.1; Figure 1A and 1B). However, the ciliate protozoa community structure was influenced by the dietary additive (P = 0.017; Figure 1C).
Family and genus WCHB1-41 affiliated with phylum Verrucomicrobiota were lower (P = 0.02), whereas Suc-cinivibrionaceae were more abundant (P = 0.04) in the GCE compared with the CTRL diet (Table 7). At the species level, both WCHB1-41 sp. (P = 0.02) and Prevotella sp. (P = 0.05) were less abundant in the GCE compared with the CTRL group.

Enteric Methane Emissions
In this study, GCE decreased daily CH 4 emissions significantly compared with CTRL. A similar CH 4reducing effect of GCE was observed in various in vitro studies (Eger et (Roque et al., 2019b;Vrancken et al., 2019;Brand et al., 2021;Bitsie et al., 2022).
The CH 4 reduction in vitro was almost 95% by GCE added to a dairy cow diet with rumen fluid in a RUSITEC system (Eger et al., 2018). This was achieved by alteration of the archaeal community without negatively affecting rumen fermentation. In the present trial, the lower abundance of Methanobrevibacter at the genus level in the GCE group establishes the inhibitory effect of GCE on methanogens, leading to reduced enteric CH 4 emissions. The organosulfur compounds in garlic lead  Mean values in the same row with different superscripts differ (P < 0.05).
1 to the direct inhibition of methanogenic archaea in the rumen by inhibiting the enzyme 3-hydroxy-3-methylglutaryl coenzyme A reductase. This enzyme catalyzes the synthesis of the isoprenoid units in methanogenic archaea as they have unique membrane lipids that contain glycerol joined by ether linkages to long-chain isoprenoid alcohols (Busquet et al., 2005). In another long-term (38-d) RUSITEC experiment (Brede et al., 2021), CH 4 production was significantly reduced until d 18, which was also due to a selective effect on archaeal community composition. Ahmed et al. (2021a), in an in vitro batch culture experiment conducted on rumen fluid from sheep, observed that GCE dosed at 20% of substrate (50% grass: 50 % concentrate) led to an up to 54% reduction in CH 4 production. Although the forage-to-concentrate ratio in the present trial was similar to 50:50, the dosage of GCE was lower and might explain the lesser reduction in enteric CH 4 emission in vivo. It may not be possible to observe a similar CH 4 reduction effect in in vitro and in vivo trials as physiological processes come into play in live animals compared with batch cultures.
In beef cattle on a feedlot diet supplemented with 15 g/d GCE included in an alfalfa pellet, a reduction in CH 4 yield of 23.2% (Roque et al., 2019b) was observed, and with 16 g/d GCE in a pellet, a reduction in CH 4 yield of 24.6% (Bitsie et al., 2022) was observed. The tendency for reduced CH 4 yield in the current trial was lower than in the previous trials. The daily dosage of GCE in the beef trials was similar but higher per unit of DMI, which may explain the higher CH 4 reduction. Moreover, the feedlot diets were low in forage (fiber) and high in corn (starch), although a forage effect was observed by Bitsie et al. (2022) as the steers fed 41.5% corn silage emitted more CH 4 on a grams per kilogram of DMI basis compared with steers fed the 15% corn silage diet. The feedlot diets are high in concentrates, and greater CH 4 emissions are expected from cattle fed 50% forage from grass silage in the diets, as observed in the present trial. A 22.8% decrease in CH 4 production by preweaning Holstein-Friesian bull calves fed milk replacer and a 52.3% concentrate starter feed and supplemented with GCE at 4 g/d for 70 d could also be attributed to lower dietary fiber levels (Brand et al., 2021).
The use of grass silage-based rations in milk production is typical practice in Norway, with feed rations for dairy cows comprising 45% and 42% silage and concentrates, respectively (Åby et al., 2014). Lower CH 4 emissions from feeding corn silage compared with grass silage have been reported in several studies (Evans, 2018). Vrancken et al. (2019) observed that feeding 15 g/d GCE to Holstein Friesian dairy cows fed a TMR with 55% grass silage and corn silage decreased CH 4 production by 20.7% compared with when cows were not fed GCE. In this trial, a comparatively lesser reduction in enteric CH 4 emissions may be attributed to the proportion of grass silage, indicating that GCE is more effective in CH 4 reduction with lower forage proportion and more starch fermentation in the rumen.
Indeed, antimethanogenic feed additives must exhibit a persistent mitigation effect to ensure successful practical application. It is noteworthy that the CH 4mitigating effects of feed additives have been mostly investigated in short-term studies, including the experiment reported herein. However, the ability of rumen microbes to adapt to some feed additives could impair their long-term persistent effects on enteric CH 4 reduction (Kumar et al., 2014). Transient effects of CH 4 mitigation have been reported in cattle studies involving supplementation of feed additives such as ionophores (Sauer et al., 1998;Guan et al., 2006), essential oils (Klop et al., 2017) and lipids (Woodward et al., 2006;Muñoz et al., 2021). Although there is limited information about the persistency of the antimethanogenic effects of GCE, Vrancken et al. (2019) observed that CH 4 emission was significantly reduced when GCE was supplemented in the diets of Jersey and Holstein-Frie-  sian cows over a 12-wk period. Thus, further long-term in vivo studies are required to confirm the persistency of GCE supplementation on enteric CH 4 reduction and the rumen microbiome.

Feed Intake and Milk Yield
Intakes of the total diet or the GCE mixed in TMR fed for 2 h were similar to the CTRL, indicating no effect of GCE on palatability. This observation is similar to some of the feed trials conducted with Holstein dairy heifers, cows, or calves fed garlic cloves or powder at a rate of 7 or 10 g/kg of DMI, respectively, which caused no depression in feed intake, even when applied at high dosages (Gholipour et al., 2016;Rossi et al., 2018). Ahmed et al. (2021b), in their study with sheep, also reported no negative effect of GCE supplementation on total feed intake, even at the highest dose of 10 g/ kg of DMI.
In this study, ECM and milk yield were similar in CTRL and GCE-supplemented groups. This is contradictory to the observation of Vrancken et al. (2019), who found that feeding GCE led to increased milk yield by Jersey and Holstein Friesian cows. In theory, a decrease in CH 4 production could provide more energy that is partitioned into milk. However, in this study, CH 4 production was reduced by 10.3% and CH 4 intensity by 11.7%, and these effects could be too small to detect an effect on milk yield. However, in recent metaanalyses of CH 4 inhibitors with a larger CH 4 reduction than in this study, a tendency for lower milk yield for 3-nitrooxypropanol (Kim et al., 2020;Yu et al., 2021) or no relationship between reduced CH 4 production and milk yield (Ungerfeld, 2018) was found.

Rumen Fermentation
Rumen pH was not affected by GCE supplementation. This is in accordance with in vitro studies in RUSITEC with garlic oil (Soliva et al., 2011) and in batch culture with GCE (Ahmed et al., 2021a). Moreover, this is consistent with observations reported in previous in vivo studies feeding garlic oil to ewe lambs , dairy goats (Kholif et al. (2012), and dairy cows (Yang et al., 2007). The GCE did not affect the daily output of VFA; however, a shift in fermentation pattern was observed.
A tendency for higher propionate proportion and lower acetate-to-propionate ratio in GCE-fed cows was similar to that observed when supplementing Bioflavex (HealthTech BioActives S.L.U.; with main components naringin, neohesperidine, and poncirin) to a feedlot diet fed to Friesian steers, suggesting the effect of citrus extracts in GCE in modulating the activity of rumen microbiota (Seradj et al., 2014). Soliman et al. (2020) also observed higher propionate concentration and lower acetate-to-propionate ratio in Holstein Friesian dairy cows fed a flavonoid-rich fruit and vegetable juice. Enteric CH 4 emissions have been negatively associated with an increase in propionate production and a decrease in the acetate-to-propionate ratio (Wang et al., 2018). After CH 4 , propionate is the principal alternative H + sink, although H 2 concentration was observed to be similar between both treatments.
The proportion of a minor VFA, caproate, was negatively correlated with both CH 4 production (g/d) and CH 4 yield (g/kg of DMI) in sheep fed alfalfa pellets (Jonker et al., 2019). The lower caproate concentrations we observed in GCE, despite lower CH 4 production and CH 4 yield, are in contrast to the results from the above study. However, Zhu et al. (2015) observed that rapid lactate degradation led to a gradual accumulation of ncaproate in high concentrations in the fermentation pit used for the production of Chinese strong-flavor liquor. It is possible that reduced lactate concentration in the GCE-fed group led to the lower caproate concentration. Because lactate concentration was not measured in this trial, it may not be possible to attribute the lower caproate production in the GCE group to lesser availability of lactate substrate in the rumen.

Microbial Population
The abundance of bacteria was not affected by GCE, which is in agreement with previous studies where the flavonoid naringin (Oskoueian et al., 2013) or garlic oil (Patra and Yu, 2015) were used in in vitro experiments using rumen fluid from male or lactating dairy cattle. Succinivibrionaceae is the dominant family among rumen Proteobacteria (Tapio et al., 2017) and is known to play an important role in succinate production through hydrogen utilization. This not only enables competition with hydrogenotrophic methanogens for substrate, but succinate is also a precursor for propionate production (McCabe et al., 2015;Bailoni et al., 2021), with an estimated 70 to 100% of the propionate produced via the succinate pathway (Van Lingen, 2017). Ramayo-Caldas et al. (2020) observed that Succinivibrionaceae are associated with improved feed efficiency, reduced CH 4 emissions, and higher propionate concentration in lactating Holstein cows. This supports the significant abundance of the Succinivibrionaceae family and a tendency to higher propionate concentration and lower CH 4 production in GCE-fed cows in this trial. Conversion of succinate to propionate was further aided by the significantly higher abundance of the Succiniclasticum genus in the GCE. According to van Gylswyk (1995), Succiniclasticum is characterized as a gram-negative and nonmotile species, unable to ferment carbohydrates, amino acids, or mono-, di-, and tricarboxylic acids other than succinate to propionate.
The relative abundance of strict anaerobic Methanobrevibacter genus was lower with GCE. This resulted in a comparable decrease in enteric CH 4 emissions, considering that the dominant methanogens in the rumen fluid were related to the genus Methanobrevibacter (Danielsson et al., 2012). This is in line with Eger et al. (2018), who showed reduced abundance of several OTU of the Methanobrevibacter genus in an in vitro experiment with GCE.
The abundance of rumen ciliate protozoa is a function of energy and nitrogen availability and the quality of diet in the rumen (Vogels et al., 1980). For example, the Ostracodinium genus possesses high cellulolytic activity, and their increased relative abundance is known to be associated with high-forage diets (Dehority and Odenyo, 2003;Bailoni et al., 2021). According to Vogels et al. (1980), the somatic association between ciliate protozoa and methanogenic bacteria can be observed in all species of the genera Epidinium, Entodinium, and Ostracodinium in addition to a few others; however, the differences observed in their frequency of association are not concretely understood. Further, the association of methanogens with subspecies of one species of ciliate protozoan may differ substantially, and therefore the difference in their numbers in control and treatment groups cannot be categorically elucidated. For example, the range of association of methanogens to subspecies longinucleatum of Entodinium species is 5 to 10%, whereas this association with caudatum subspecies is 20 to 50% (Vogels et al., 1980). The occurrence of Isotrich ciliates in high numbers in GCE compared with CTRL may not imply any differences because, according to Vogels et al. (1980), no externally associated methanogens have been observed to be associated with Isotrich ciliates. Alternatively, Belanche et al. (2015) in their experiment with 8 Texel-crossbred fauna-free sheep observed that post-inoculation holotrich-faunated sheep exhibited a 22% population of Isotricha spp. and produced 11.8 times more CH 4 per cell than the total protozoa population observed in fully faunated sheep. In a recently published meta-analysis reporting data from 79 in vivo experiments, Dai et al. (2022) observed a positive association between isotrichids and CH 4 emissions. The meta-analysis also illustrated that a positive association exists between CH 4 emissions from entodiniomorphids and dairy cows but not with small ruminants and beef cattle. Ranilla et al. (2007), in an in vitro trial, compared 4 protozoa species, Entodinium caudatum, Isotricha intestinalis, Metadinium medium, and Eudiplodinium maggii, from monofaunated and conventional wethers. The trial demonstrated that compared with Isotricha spp., E. caudatum increased CH 4 production and protein degradation, displaying a negative effect on rumen fermentation for the host.
The overall differences in ciliate protozoa populations among the CTRL and GCE-supplemented groups indicate the rumen-modifying potential of the GCE supplement. This is in line with the observations of Wanapat et al. (2008), who observed a difference in ciliate protozoa numbers in incubated rumen fluid of Holstein Friesian crossbred steers fed diets supplemented with various levels of garlic powder.

CONCLUSIONS
Feeding GCE to dairy cows for 18 d reduced CH 4 production and intensity without compromising feed intake or milk production of dairy cows. The GCE supplement modified rumen fermentation and microbiota, as shown by a tendency for higher propionate concentration and lower acetate-to-propionate ratio, a higher relative abundance of the Succinivibrionaceae family of bacteria acting as a hydrogen sink, and a lower relative abundance of the CH 4 -producing Methanobrevibacter genus. Hence, supplementation of dairy cattle with GCE could be an effective strategy for enteric CH 4 mitigation in dairy cows.

ACKNOWLEDGMENTS
This work was funded by Mootral SA (Rolle, Switzerland). For the rumen microbiome data analysis, Ilma Tapio was supported by funding from the European Union's Horizon 2020 research and innovation programme under grant agreement No. 101000213. The scientific and technical staff of Luke (Jokionen, Finland) are appreciated for designing and conducting the animal experiment, for technical support, care of experimental animals, assistance in sample collection, and chemical analysis of samples. The authors greatly acknowledge the support of Bart Tas (Mootral Ltd., Abertillery, United Kingdom) for critically reviewing the manuscript. The financial support from Luke for contributing to the writing of this paper is acknowledged. We thank the Finnish Functional Genomics Centre supported by University of Turku, Åbo Akademi University, and Biocenter Finland for sequencing. Ruchita Khurana and Tassilo Brand are employees of Mootral GmbH. The authors have not stated any other conflicts of interest.