In vitro screening of the ruminal methane and ammonia mitigating potential of mixtures of either chestnut or quebracho tannins with blends of essential oils as feed additives

Abstract Tannins and essential oils (EOs) have been previously described for their properties to mitigate ruminal methane and ammonia. Their combination might be even more efficient, as they have different modes of action on rumen pathways. This study aimed to screen in vitro the mitigating properties of variously combining tannins with EO in a total of 48 treatments: 12 single additives, 10 mg of EO or 20 mg of tannins/g diet, to establish their basal efficiency; 36 combinations of 20 mg of tannins/g + 10–15 mg of EO. Quebracho (Q) and chestnut (C) tannins defined C, Q and C/Q groups of mixtures with EO blends, formulated with oregano, thyme and clove EO, citrus peel, carvacrol, thymol, eugenol, α-pinene, and bornyl acetate. Supplements were added to a control diet, which was also incubated alone as a basis for comparisons with supplemented treatments, in a total of six runs. Effects on rumen pH, protozoal count, and proportions of individual volatile fatty acids (VFA) were limited. The tannins extracts seemed to cause most of the mitigating effects by suppressing ammonia by up to 31% and methane yield by up to 15%, with the highest reductions obtained with tannins-based supplements. However, this occurred by contemporary affecting the feeding value of the substrate, as indicated by reductions in total VFA and in vitro organic matter digestibility. Overall, six mixtures of C and Q groups were the most efficient and they need further studies to understand the mechanisms of actions and the synergistic effects occurring among compounds. Highlights Combinations of tannins and essential oil compounds were screened in vitro Some combinations lowered methane yield and ammonia formation by up to 15 and 31% Essential oil compounds enhanced the mitigating properties of the tannins


Introduction
Ruminants physiologically produce methane mainly from the enteric fermentation of feed. Although enteric rumen fermentation by livestock is the largest single source of methane (CH 4 ) emissions in the EU-27 (European Environment Agency 2022), ruminants can convert biomass rich in fibre, which is not competitive with human consumption, into high-quality food protein (i.e., meat and milk) (Moss et al. 2000). However, effective strategies are required to cope with climate change. Previous studies, focussed on natural additives to diets of ruminants, reported several plant secondary metabolites as being efficient as rumen modulators. Among the groups of such metabolites, various tannins, natural essential oils (EO), and essential oil compounds (EOC) have been largely reviewed for their activity against ruminal formation of CH 4 and the mode of downregulating microbial pathways involved in methanogenesis (Cieslak et al. 2013;Honan et al. 2021;Zhang et al. 2021). Chemical structure seems to be important, explaining, for instance, differences in efficacy between condensed tannins (CT) and hydrolysable tannins (HT). The CT form a stronger bond with nutrients than HT, and they were suggested to decrease CH 4 by lowering fibre digestion and thus hydrogen formation (Jayanegara et al. 2015). Conversely, hydrolysable tannins can be degraded in the rumen, and their fractions might directly or indirectly inhibit rumen microorganisms involved in methanogenesis (Aboagye and Beauchemin 2019;Vasta et al. 2019;Cardoso-Gutierrez et al. 2021). Apart from CH 4 , tannins and EO may mitigate dietary protein degradation to ammonia (NH 3 ) in the rumen. Hereby, the tannins exhibit a protein binding property preventing access of the proteolytic microbes to dietary protein (Aboagye and Beauchemin 2019;Vasta et al. 2019), and the EO directly inhibits NH 3 -producing bacteria (Patra and Yu 2012). Mitigation properties are, in general, dosedependent and vary across compounds. First attempts to determine the optimal additive dosage for both tannins (Terranova et al. 2020;Menci et al. 2021), and EOC (Macheboeuf et al. 2008) were made previously.
These differences in mode of action and efficacy point towards the possibility of enhancing the extent of mitigation of ruminal CH 4 and NH 3 formation by combining compounds with different structural properties. This could be achieved within tannins, when combining CT and HT, within EO, when combining different EOC, or when combining tannins and EO. The combinations of various plant secondary metabolite classes were indeed considered successful previously both in vitro (mixtures of tannins: Sinz et al. 2019;Jayanegara et al. 2020; or mixtures of EO: Cobellis et al. 2016) and in vivo (tannins: Duval et al. 2016; EO: Belanche et al. 2020). Still, the combination of tannins and EOC has not yet been tested to the best of our knowledge.
For this reason, we tested the hypothesis that there are at least some combinations of HT and CT, of different EOC, and especially combinations of tannins and EOC, which enhance the mitigation potential of the single additives against ruminal CH 4 and NH 3 formation. The basis of such phenomena would be the different modes of action resulting in synergistic activities against the relevant rumen microorganisms. To test this hypothesis, an extensive in vitro screening approach was chosen. In this attempt, one HT and one CT source, as well as nine EOC sources, including four extracts from plants and five pure compounds, were mixed in various combinations, and their effect was compared to that obtained when incubated as single sources. The study also aimed at determining the combinations with minimal adverse effects on the fermentation characteristics in order to select additives with the best net mitigating potential.

Ethical approval
The rumen fluid donors and rumen fluid collection were carried out according to the ARRIVE guidelines (Kilkenny et al. 2010) and were approved by the responsible Swiss authority (cantonal veterinary office of Zurich; licence no. ZH113/18).

Feed additives
As a source of CT, the extract from the quebracho bark (Q; Quebracho Crown Atg powder, Figli Di Guido Lapi S.P.A., Italy) was selected, whereas an extract from chestnut wood (C; Silvafeed ENC powder, Silvateam, Italy) was the HT source. The tested EOC sources included four plant extracts and five pure compounds. Extracts were from oregano (Origanum vulgaris; main active EOC, according to the manufacturer's technical data sheet: carvacrol and b-caryophyllene), from thyme (Thymus vulgaris; main active EOC: thymol, p-cimene, c-terpinene, linalool, carvacrol), from clove (Eugenia caryophyllata Thunb.; main active EOC: eugenol, b-caryophyllene), and from citrus peel (Citrus aurantium dulcis; main active EOC: limonene). The five pure EOC were those prevalent in oregano, thyme and clove EO, namely carvacrol, thymol and eugenol, and, in addition, a-pinene and bornyl acetate. The oregano and thyme extracts were purchased from Italfeed s.r.l, Milan, Italy, while the other EO sources were obtained from Moellhausen S.P.A., Milan, Italy.
Twelve EOC blends were formulated, each by mixing two or three EOC as listed in Table 1. All supplements contained EOs from oregano or thyme or their main active EOC, carvacrol and thymol, in pure form, because mitigating potential of these EO/EOC had been previously reported (Benchaar et al. 2007;Cobellis et al. 2016;Joch et al. 2015;Rofiq et al. 2021). Moreover, Macheboeuf et al. (2008) identified carvacrol and thymol based-EO as the additives effective at the lowest dosage (1 mM or 150 mg/L) to modulate rumen fermentation; this in contrast to other natural EO or the respective pure constituents. The other EO/ EOC sources were included only in two of the mixtures each.

Experimental design and diets
A total of 49 treatments (1 control, 48 ones containing additives) were investigated with the Hohenheim Gas Test apparatus (HGT), in addition to rumen blank and two feed standards, as described by Menke and Steingass (1988). Each treatment containing additives was incubated once each in a total of six runs providing six independent replicates. The control diet, the blank (only incubation medium: rumen fluid and buffer solution with a proportion of 1:2), the HGT standard hay, and the HGT standard concentrate were all repeated twice in each of the six runs, for a total of twelve replicates each. In total, 56 syringes were incubated per run in a changing order of places in the rotor in every run. This experimental design considered the recommendations made by Y añez- Ruiz et al. (2016) concerning the number of runs and replicates. However, modifications were necessary to allow the simultaneous screening of all treatments in each individual run. Therefore, a higher number of independent runs, in contrast the minimum of three runs suggested, were conducted to allow statistical significance to be observed and to identify possible outliers.
The hay and concentrate standards had been supplied by the Institute of Animal Nutrition, University of Hohenheim, Stuttgart, Germany, for later data adjustments as described by Menke and Steingass (1988). The control diet was composed, (g/kg of dry matter (DM)), of ryegrass hay, 700, maize, 120, barley, 90 and soybean meal, 90. The supplements were added to the control diet as follows (see Table 1): 12 treatments were obtained by adding tannin or EO/EOC sources alone, namely 20 g/kg of C, Q or C/Q (10 g/kg each) (3 treatments) or 10 g/kg of each EO/EOC source (9 treatments); 36 treatments were obtained by adding a mixture containing 20 g/kg of one or 10 g/kg each of both tannin sources and 10-15 g/kg of each of the 12 EOC blends as detailed in Table 1. Each treatment was indicated in the text with an alphanumerical code consisting of the letters C, Q or C/Q (tannin source) and a number from 1 to 12 identifying the EOC blend contained (see Tables 2 and 3).

Animals and preparation of rumen fluid
Rumen fluid was obtained before the morning feeding from a total of three cannulated lactating Original Brown Swiss cows. Each run was conducted with rumen fluid from one individual donor cow only, as reported previously by Terranova et al. (2018Terranova et al. ( , 2020, with each donor used for two runs in an alternating way. This procedure was preferred over generating analytical replicates with rumen fluid from the same cow or a mix of rumen fluid in order to be able to include all treatments in each run and to generate six true replicates by keeping the natural variation among individual donor animals in terms of rumen fluid properties. The donor cows were fed with a diet composed (g/kg DM) of maize silage, 550; grass silage, 380; ryegrass hay, 20; dairy concentrate (UFA-243, UFA AG, Switzerland), 50. Animals had free access to water. Once collected, the rumen fluid was stored anaerobically in sealed pre-warmed bottles and then immediately strained through four layers of cheesecloth as the barn and laboratory are located a few metres from each other. For the incubations, rumen fluid was added to the buffer medium in a proportion of 1:2 (v/ v). The buffer medium was prepared accordingly to Menke and Steingass (1988) and it was continuously kept under carbon dioxide (CO 2 ) flow to establish anaerobic condition.

In vitro rumen fermentation experiment
The incubation was performed following the method of Menke and Steingass (1988) as modified by Soliva and Hess (2007) and was carried out at AgroVet-Strickhof research centre (Lindau, Switzerland). For each run, the same amount of basal diet (200 mg DM) was prepared, and the feed additives (amounts reported in section 'Experimental design and diets') were added on top as a supplementation to the basal diet (control). This procedure was chosen, as the feeding value of the supplements was assumed to be equal to zero. Afterwards, the HGT glass syringes were EO: essential oils; EOC: essential oil compounds.
closed and stored at room temperature. A fixed volume of 30 mL of the warm buffered rumen fluid was added through the inlet of each syringe. Straight afterwards, the syringes were placed in the pre-heated incubator (39 C) for 24 h. For each run, the treatments were placed differently in the HGT incubator to avoid effects of the syringe position in relation to the rotational axis.

Laboratory analyses
For compositional analysis, the ryegrass hay and the concentrate ingredients (corn, barley, soybean meal) were first pre-dried at 60 C for 48 h and then ground with an ultra-centrifugal mill (Model ZM 200, Retsch GmbH, Hann, Germany) to pass through a 1 mm sieve. The DM content was determined by oven drying at 105 C until a stable weight, and total ash content, and indirectly the organic matter (OM) content by ashing at 550 C for 3 h. Nitrogen content was determined with the Kjeldahl method, and crude protein (CP) was calculated as N Â 6,25. An Ankom XT10 Extraktor (Astori tecnica, Brescia, Italy) was used to determine ether extract with AOCS Official Method Am 5-04 method (AOCS 2004). The concentrations of ash-free neutral detergent fibre (aNDFom), ash-free acid detergent fibre (ADFom), and acid detergent lignin (lignin(sa)) were analysed with Filter Bag Technique accordingly to the Van Soest et al. (1991) protocol and using the Ankom Fibre Analyser A200 (Astori tecnica Brescia, Italy). All samples were analysed in duplicate. Table 3. Effect of treatment on gas production (GP) and yield (GY, expressed per gram of diet DM), CH 4 and CO 2 production and yield (expressed per gram of diet DM), in vitro organic matter digestibility (IVOMD, expressed per gram of diet DM) and net dissolved hydrogen after 24 h of in vitro incubation. After 24 h of incubation in the HGT syringes at 39 C, the volume of gas produced (mL/day) was read from the calibrated glass syringes. The incubation liquid was removed from the syringes, whereas the syringes still containing gas phase were stored in ice until gas analysis. Within the same day the incubation liquid was analysed for pH (pH Metre 913, Metrohm Suisse SA, Zofingen, Switzerland), NH 3 concentration (pH Metre 632 equipped with NH 3 -selective gas membrane electrode 6.0506.010, Metrohm), and protozoa count (after 1:1 dilution in formaldehyde solution and with a B€ urker counting chamber with a depth of 0.1 mm, BlauBrand, Wertheim, Germany). A 4-mL sample of incubated fluid was centrifuged at 2600 Â g for 5 min at 4 C, and 2 mL of the supernatant were rapidly frozen (-20 C) for later volatile fatty acid (VFA) analysis (HPLC La Chrom, L-7000 series, Hitachi Ltd, Tokyo, Japan, equipped with a UV detector). The gas phase was sampled from the syringes with a gas-tight Hamilton syringe and analysed for concentrations of CH 4 and CO 2 with a gas chromatograph (GC-TCD 6890 N, Agilent Technologies, Wilmington, NC, USA).

Calculations and statistical analysis
Net gas production (GP) was calculated as the gas volume read from the test syringes subtracted by the average of gas volume of the blank syringes incubated in the same run and adjusted with the quotients obtained fermenting HGT standard hay and standard concentrate following Menke and Steingass (1988). Each run was considered successful when the quotients 'expected GP from the HGT hay standard (49.6 mL/24 h)/its recorded GP', and 'expected GP from the HGT concentrate standard (61.1 mL/24 h)/its recorded GP' ranged between 0.9 and 1.1 (Menke and Steingass 1988). Methane and CO 2 production were calculated as the product of GP and their corresponding concentration in the gas phase. Data of the control treatment that were not related to the amount of diet incubated (i.e., GP, mL/day; CH 4 and CO 2 production, mL/day; NH 3 , mmol/ L; total VFA, mmol/L) were adjusted to consider the lower DM amount incubated in the control treatment (200 mg) compared to the final DM weight of supplemented treatments (210 mg ± 2.0 mg). The unadjusted values were reported in brackets in Tables 2 and 3.
In vitro OM digestibility (IVOMD) was estimated according to the following equation (Menke and Steingass 1988): where GP is the net gas production, CP is the crude protein content of the incubated substrate, and TA is the total ash content of the incubated substrate. The concentration of the net hydrogen dissolved in the incubation liquid (net H 2 ) was calculated as reported by Wang et al. (2014), and values were corrected with the blanks' concentrations, as previously described for GP. The following equation was used: All treatments including supplementation were statistically compared to the control by the following linear model (SAS Institute Inc., Cary, NC, USA, 010): where Y ijk ¼ variables, m¼ overall mean, T i ¼ fixed effect of the i th treatment (49 treatments); r j ¼ random effect of j th run (6 runs) and e ijk residual error. Differences were declared significant at P < 0.05. When the treatment effect was significant, orthogonal contrasts were applied to compare each least-squares means of the supplemented treatments with that of the control. In case of statistical significance (P < 0.05) these differences were marked in the tables by printing the mean value in bold. Additionally, orthogonal contrasts were applied to determine the differences within the treatments containing tannins.

Results
The control diet contained, as analysed, 907 g DM/kg and, per kg of DM, 925 g OM, 150 g CP, 26 g ether extract, 465 g aNDFom, 251 g ADFom and 30 g lignin(sa).

Properties of the incubation liquid
The pH of the incubation liquid (data not shown in the table) ranged from between 6.77 and 6.85 with no significant deviations from control. The protozoa count was mostly not influenced by the treatments, and the values ranged between 1.05 and 2.10 Â 10 4 cells per mL of incubation liquid (Table 2). Still, the protozoa count increased with treatments C-6, C-8, C/Q-9, C/Q-11 by up to 2.12 Â 10 4 cells per mL compared to the control. All mixture additives significantly reduced NH 3 formation by 17% to 31% of total. The mixtures containing EOC and C had the highest effect on NH 3 (declines by 23% to 31%), while the reduction potential of the Q or C/Q groups containing EOC ranged between declines of 17% and 23% of total. Tannins as single additives significantly lowered the NH 3 production (C by 10%, Q by 13% and C/Q by 16% of total), illustrating a lower efficacy against NH 3 without EOC in comparison to mixture additives. However, these differences were statistically significant only for six additives of the C-group (containing blends C-1, C-2, C-3, C-4, C-11 and C-12) that produced lower NH 3 (P < 0.001) than the pure C additive (data not shown). The total VFA production was reduced by most of the supplementation treatments. Considering the individual VFA, proportions of acetate (average 676 mmol/ mol VFA) and butyrate (average 103 mmol/mol) were not affected by any treatment. Propionate proportion decreased with thymol, clove EO, and citrus peel supplementation (by 4.7% of total on average), and it increased with C-2, C-5, Q-4, Q-5, C/Q-11 (by 4.6% on average) in comparison to control. Consequently, the acetate-to-propionate ratio increased when thymol, clove EO or citrus peel were added, and it decreased with the above-mentioned mixtures. Iso-butyrate proportion increased by up 56% when carvacrol, clove EO, citrus peel, a-pinene or most of the C/Q group additions were supplemented. Proportions of valerate and iso-valerate were not affected by the treatments, except for oregano EO and thymol supplements, which increased their formation. Among the tannins supplemented as pure or as a mixture, no significant differences of individual VFA proportions and total VFA produced were observed (data not shown).

Fermentation gas production and composition
The GP was generally decreased when any supplement was added (Table 3), and this was significant with most treatments. In detail, GP was decreased with single additives by 5.5 to 8.1% or with mixtures of additives: group C treatments by 6.7 to 9.8%, group Q treatments by 6.8 to 14.9%, and with group C/Q treatments by 7.6 to 12.2%, respectively, without significant differences between the mixtures (data not shown). GY decreased with the addition of C, C/Q and carvacrol addition as well as with all mixtures, except for C-11. Methane production and CH 4 yield followed a similar pattern as GP in terms of mitigating potential. When tannins were supplemented, either as single additives or in combination with an EOC blend, a decrease in CH 4 production by 6.4 to 13.9%, and in CH 4 yield by 7.9 to 15.0% was observed, with only few exceptions where there was no significant effect. Treatments Q-10, Q-11, Q-12 did not have a significant effect on either CH 4 production or yield, and with C-6 only CH 4 production was significantly affected. When supplemented alone, any EO/EOC source significantly influenced CH 4 formation. The above-described pattern changed when CH 4 was expressed per unit of digestible OM (dOM) or per moles of total VFA produced. Tannins when supplemented as single additives decreased CH 4 per unit of dOM, but not per moles of total VFA produced. Considering the thirtysix multiple mixtures, 15 mixture additives were effective in decreasing CH 4 per unit of dOM (nine from group C, four from group Q and two from group C/Q), whereas nine decreased CH 4 expressed per moles of VFA (two from group C, five from group Q and two from group C/Q). Only six treatments (C-4, C-10, Q-2, Q-7, Q-8, C/Q-8) lowered CH 4 regardless the way it was expressed. The absolute amount of CO 2 produced within 24 h was significantly reduced with 38 of the 48 treatments by on average reduction of 9.8% in comparison to control, whereas ten supplements (five single EO sources, five mixtures), remained ineffective. Expressing the CO 2 per unit of feed DM, a similar pattern was obtained; however, none of the EO/EOC sources supplemented alone resulted in a significant reduction. The CH 4 -to-CO 2 ratio was not influenced by any of the treatments, even though in the a-pinene and Q-4 treatments the ratio tended to be higher when compared to the control (P < 0.10).

In vitro organic matter digestibility and net dissolved hydrogen
The supplementation of tannins as single additives or as mixtures and carvacrol significantly lowered the calculated IVOMD (Table 3). The level of reduction in IVOMD was more comparable between C addition and the C group (by 5.7% vs. 5.5% on average), than between Q addition and the Q group (by 3.5% vs. 7.1%) and C/Q addition and the C/Q group (by 4.5% vs. 7.2%); however, no significant difference was observed among individual additives containing tannins (data not shown). The net H 2 concentration dissolved in the incubation liquid declined with most of the mixtures of the C and Q groups by up to 12% in comparison to control diet (Table 3).

Discussion
Various tannins and EOC were previously shown to have a potential to mitigate ruminal CH 4 and NH 3 ; however, their combinations have not been tested yet, to the best of our knowledge. As synergistic or antagonistic effects may occur among these compounds, an in vitro screening approach was chosen in this study, aiming to select the best combinations of tannins and EOC with the best mitigating properties concerning CH 4 and NH 3 among those tested, while having no or only minimal adverse effects.
In the present experiment, the CH 4 and NH 3 reductions seemed to be mainly associated to the tannin content in the supplements, as the pure EO/EOC did not determine significant effects on CH 4 and NH 3 at the concentration studied, in comparison to the control. By comparing the treatments containing tannins, either pure or as a mixture, the effects described were not generally statistically different, except for the NH 3 decline, which was particularly evident for six combinations of supplements containing the chestnut extract. The general decline in feeding value observed in the case of mixtures supplementation can be explained by the presence of tannins, for both total VFA and IVOMD reductions, and EO/EOC, for total VFA reduction.
In our study, additives causing the lowest reduction of feeding values were particularly of interest, as additives only able to reduce absolute CH 4 formation cannot reduce the environmental impact of ruminant. For this purpose, in the following sections special emphasis was given to CH 4 expressed per unit of digestible OM (dOM) and per mole of VFA produced, where it was possible.

General effects of tannins
The general effects associated with tannin supplementation seemed to be mainly related to the included dosage and their chemical structure (Cardoso-Gutierrez et al. 2021). Concerning the first, in this study the minimal effective dosage of tannins (20 g/kg equivalent to 133 mg/L incubation liquid) was supplemented, as proposed in the meta-analysis by Jayanegara et al. (2012) to obtain significant mitigation of CH 4 production. With such a dosage the reduction obtained in CH 4 and NH 3 formation was significant, even though other authors had reported no effect on CH 4 until the dosage applied was higher than 50 g/kg DM (Hassanat and Benchaar 2013). Nevertheless, 50 g/kg was also reported as the maximal dosage to avoid adverse effects on feeding value, particularly with CT, as reviewed by Mueller-Harvey (2006). Generally, the tannin depression of both CH 4 and NH 3 was reported to be dose-dependent, with greater dosage also associated with a worsening of the feeding value. However, both the in vitro conditions and the fermented diet, influence the effects measured. Indeed, in a recent in vitro study similar effects on CH 4 and NH 3 production by adding the double dosage of chestnut wood extract to the basal diet were obtained (Cappucci et al. 2021) as half of this dosage was supplemented in this study. In the study of Menci et al. (2021), absolute CH 4 production was reduced by 6% by supplementing 30 g tannins/kg diet, whereas CH 4 expressed per gram of degraded DM did not vary across the treatments, due to the concomitant reduction of in vitro fibre degradability. The same authors suggested that the effective dosage for tannins might depend on CP content in the diet: the higher the CP content and the higher the dosage of tannins has to be, to obtain mitigating effects. On the other hand, the reduction of the feeding value with higher concentrations of tannins must be considered.
Particularly interesting effects were obtained by Terranova et al. (2018), who tested tannins from chestnut leaves at a comparable concentration to the current trial and obtained greater reductions of both CH 4 yield (-28%) and NH 3 formation (-32.2%). The highly effective mitigating capacity of chestnut leaves obtained in their study might be partially explained by the content of other active plant secondary metabolites in the fermented leaves (Mannelli et al. 2019;Terranova et al. 2020). Indeed, despite the slightly higher IVOMD depression reported in their study (9. 8% Terranova et al. 2018 vs. 5.7% present study), the authors obtained a significant reduction of CH 4 both expressed per unit of dOM and VFA produced, conversely to our findings where CH 4 per mole of VFA was not reduced. This can be partially explained by the similar total VFA reduction obtained in comparison to this study. Results of individual VFA changed instead, with a higher and a lower concentration of propionate and acetate, respectively, found in the study by Terranova et al. 2018, while no differences were measured in this study, as previously reported by Lavren ci c and Pirman (2021). Concerning the mitigation of NH 3 production, results in literature seem to be generally more consistent, as also reported by Hassanat and Benchaar (2013), who obtained a significant reduction of NH 3 formation (-20 and À24%) with 20 g/kg of chestnut or quebracho commercial tannin extracts (same dosage adopted in this study), but no effect on absolute CH 4 . More recently, chestnut extract at 15 or 30 g/kg diet reduced NH 3 formation by up to 17%, but CH 4 yield was not influenced (Sarnataro et al. 2020). Accordingly, the tannin extract from chestnut wood supplemented at 2.5 g/kg of DM intake lowered ruminal NH 3 formation in crossbred steers without influencing total VFA, protozoa number and failing to reduce daily CH 4 production (Aboagye et al. 2018). Similarly, Krueger et al., (2010) did not state significant differences in animal performance and rumen fermentation when supplementing chestnut extract at 14.9 g/kg in a high-grain diet. Nevertheless, promising results in vivo, concerning ruminal methane and/or ammonia, were obtained both with quebracho (Piñeiro-V azquez et al. 2018;Norris et al. 2020) and chestnut (Liu et al. 2011) extracts with a dosage above 30 g/kg, levels which, however, might be not economically sustainable for farmers. However, in this study both tannin extracts similarly lowered absolute CH 4 , CH 4 per unit of DM, dOM (P > 0.05, data not shown). The results differed from what was reported previously by Jayanegara et al. (2015) for whom chestnut tannin, in contrast to quebracho extract, had a greater ability to decrease CH 4 while lowering less IVOMD and VFA total production. In this study, similar effects with C and Q were obtained in contrast to the control diet, and both lowered CH 4 yield with a dosage (20 g/kg) that had been often unsuccessful under previous in vitro study conditions. Such differences might be partially explained by the optimal dietary tannin-todiet CP (150 g/kg) ratio applied in this study, as with increasing dietary CP contents increasing amounts of tannins supplementation are needed to generate the mitigating properties in vitro (Menci et al. 2021).

Effects of combinations of condensed and hydrolysable tannins
As the purpose of the study was to evaluate whether combining various compound types could improve the mitigating capacity of the single compounds, a mixture of C and Q in a 1:1 ratio was fermented both alone and in combination with EOC blends. Few studies tested the effect of chestnut and quebracho extracts as a mixture on rumen fermentation characteristics. Menci et al. (2021), similarly to this study, did not observe a performance improvement between the single addition of quebracho and a mixture of chestnut: quebracho in a ratio of 3:2, except for a greater in vitro mitigation of NH 3 obtained with the mixture. Concerning in vivo effects, Aboagye et al (2018) compared the supplementation of chestnut extract either alone or in combination with quebracho extract (1:1) at 2.5 and 15 g/kg of inclusion levels in beef cattle fed high forage diet. In their experiment, ruminal NH 3 was reduced, irrespective of type and inclusion level; however, CH 4 was not influenced regardless the way it was expressed, and animal performance and total VFA production did not decrease as well (Aboagye et al. 2018). Such findings are consistent with reports described for single supplementation of chestnut or quebracho (Hassanat and Benchaar 2013;Aboagye et al. 2018). Indeed, a significant reduction of emitted CH 4 was reported less frequently. Nevertheless, a previous study with dairy cows (Duval et al. 2016) reported a difference of 2.6 g/kg DMI of CH 4 produced with 18 g/kg of a chestnut: quebracho (1:2) supplementation. Interestingly, the difference was not significant before 90 days of supplementation had passed, suggesting a prolongated exposure is needed to obtain CH 4 reduction, differently from what was found for ruminal NH 3 that was already lowered at the first measurement time point of 45 days (Duval et al. 2016).

Essential oils
General effects of essential oils In literature, studies on various EO sources provided were described for their effects on methanogens and hyper-NH 3 -producing bacteria (Benchaar et al. 2007;Patra and Yu 2012;Joch et al. 2015;Rofiq et al. 2021). In many studies, the EO amount supplemented was reported as a concentration in the buffered rumen fluid, instead of presenting the level of inclusion on a diet DM basis. Accordingly, the levels investigated in this study were 67 mg/L (10 g/kg diet) for the individual EO/EOC sources and 67-100 mg/L (10-15 g/kg) in the incubation liquid for the EOC blends tested (Table  1). Like with the tannins, the concentration supplemented is of great importance. This was highlighted by Patra and Yu (2012), as an increasing dosage (from 250 to 1000 mg/L) depressed CH 4 and NH 3 formation more and, at the same time, drastically impaired feed degradability.
In this study, adding 67 mg/L of each EO/EOC source (10 g/kg) similarly lowered the total VFA production and slightly negatively affected the IVOMD, with no mitigating effects either on CH 4 or NH 3 formation. Each of the EOCs listed in Table 1 was selected since was reported previously to inhibit rumen fermentation. Therefore, the lack of relevant mitigation properties in the present seemed to be related to the low concentrations tested. Variations in response type to various EO supplementation were previously attributed to chemical structures, in addition to the EO dosage (Macheboeuf et al. 2008;Cobellis et al. 2016). Indeed, Macheboeuf et al. 2008 reported minimal efficient dosages for affecting CH 4 , NH 3 , and total VFA production that varied depending on EO type or EO pure constituents, namely: 1-2 mM (correspondent to 150-300 mg/L) for carvacrol and thyme natural EO; 2-3 mM (300-450 mg/L) for thymol and oregano natural EO. Accordingly, higher concentrations of oregano EO (1000 mg/L; Patra and Yu 2012), with carvacrol as its main constituent ('1000 mg/L; Joch et al. 2015), or thymol (500 mg/L; Pirondini et al. 2015) extensively reduced CH 4 and NH 3 (reported only for oregano EO) by up to 87% and 62%, respectively. Generally, such reductions were associated with extensive adverse effects on total VFA produced, with reductions of these by up to 60% (Joch et al. 2015;Pirondini et al. 2015). In the same in vitro screening, Joch et al. (2015) reported a lower effect (up to 28%) by adding chemically different EOC, such as pure limonene, eugenol, a-pinene and bornyl acetate at about 1000 mg/L of concentration. Despite differences revealed in the previous studies, the lower concentrations of EO and EOC tested in this study did not allow to distinguish among them based on the effect on rumen fermentation, except for minor differences in single VFA proportions, GP and CO 2 , as reported in Tables 2 and 3. Such minor differences failed to confirm clearly higher mitigation properties expected from oregano and thyme EO and their related main constituents (carvacrol and thymol) as previously reported (Macheboeuf et al. 2008;Joch et al. 2015). Still, the dosage is suggested to be a central issue since low supplementation in vivo (50 mg/kg of DM intake) of either thymol or carvacrol did not alter either rumen fermentation characteristics or nitrogen utilisation (Benchaar 2020(Benchaar , 2021.

Effects of blends of essential oils
It is worth mentioning that some commercial additives (Agolin RuminantV R and Crina V R Ruminant) based on blends of EOC claim a CH 4 reduction (Tomkins et al. 2015;Belanche et al. 2020). Castro-Montoya et al.
(2015) tested a particularly low concentration of Agolin RuminantV R , containing coriander oil, geranyl acetate and eugenol both in cattle (0.2 g/day) and in vitro (30 mg/L incubation fluid). In their in vitro study, the dosage was not enough to determine significant effects, even if a significant reduction of CH 4 production was obtained during the in vivo trial (-15% with 0.2 g/day, dosage recommended by the manufacturer), as it was recently reported in a meta-analysis by Belanche et al. (2020). Likewise, 16.7 mg/L either of Crina Ruminants and Agolin Ruminant (according to the authors being equivalent to 1 g/head and day) did not influence CH 4 (Pirondini et al. 2015), although the former had reduced by over 22% the total VFA. In the in vitro study, the lack of a mitigating effect on CH 4 for Agolin Ruminant was related to the low dosages tested, as suggested by the authors of the study (Castro-Montoya et al. 2015). Conversely to Agolin Ruminant, Crina Ruminants supplemented to beef cattle (1-2 g/day) and sheep (0.1 g/day), did not influence rumen fermentation, CH 4 and NH 3 formation (Newbold et al. 2004;Tomkins et al. 2015). Other attempts of supplementing three-way EO blends (containing among others oregano, rosemary, cinnamon, and eucalyptus EOs) in vitro have been made previously. Some of these, tested at a concentration of about 800 mg/L, reduced both CH 4 and NH 3 with only marginal interference with feed degradability (Cobellis et al. 2016). In this study, no EOC blend was tested without a tannin source; however, the EO/EOC blend concentrations (67-100 mg/L) contained in the EO-tannins mixtures were far below the concentrations that had been successful in mitigating CH 4 and NH 3 formation in the studies described above. Nevertheless, the results obtained with the mixtures in this study were novel as they either also contained a tannin source or because the EO and EOC contained were combined differently from what had been tested previously.

Effects of mixtures of tannins and essential oils
Under the present experimental conditions, the addition of EOC blends to tannins improved the mitigation potential of such compounds both on CH 4 (particularly expressed per total VFA moles) and NH 3 formation; and this with only marginally adverse effects on feed OM degradability. Most of the tannin-EOC mixtures lowered absolute CH 4 production and yield, except for three mixtures in which Q extract was combined with blends numbered 10, 11 and 12, each containing both oregano and thyme EO mixed with citrus peel, a-pinene and bornyl acetate, respectively. Such difference is difficult to explain as other parameters considered were coherent to other mixtures with mitigating properties. CH 4 was also expressed per unit of digestible OM (CH 4 /dOM) and mole of total VFA produced (CH 4 /VFA) to consider the slight reductions of feeding value determined by the tannins-EOC mixtures. When CH 4 was expressed per unit of digestible OM and mole of total VFA produced, only six mixtures (C-4, C-10, Q-2, Q-7, Q-8, C/Q-8) caused a concomitant decrease of CH 4 /dOM and CH 4 /VFA (Table 3) when related to the control diet. It is noteworthy that, some EOC blends with similar composition combined with tannins (both C and Q) had similar mitigating effects. For instance, EOC blends 4 and 10 combined with C tannins or EOC blends 2 and 8 combined with Q tannins. Particularly EOC blends 4 and 10 shared the citrus peel content, whereas EOC blends 2 and 8 shared the eugenol content. This might suggest an optimal synergistic effect of limonene (the main active compound of citrus peel) with C extract and eugenol with Q extract, respectively, and this was independent from the presence of thymol-carvacrol EOC (blends 2 and 4) or thyme-oregano EO (blends 8 and 10). It is also interesting to note that Q-8 was the only additive treatment that did not negatively affect VFA total production while lowering CH 4 and NH 3 (Table 2). Therefore, with certain combinations of EOC-tannins, the presence of EO and EOC enhanced the efficacy of tannins in reducing rumen CH 4 , even with low concentrations of the latter, which were previously insufficient to reveal differences in vitro (Macheboeuf et al. 2008;Castro-Montoya et al. 2015).
Another parameter considered was the net dissolved metabolic hydrogen as it is one of the main substrates for CH 4 production (Wang et al. 2014), and of which the rumen microbiota controls the balance between methanogenesis and other competitive pathways (Ungerfeld 2020). Thus, considering the lower net H 2 at lower amounts of CH 4 produced, the mixtures with reduced values were likely to either directly inhibit methanogenesis or to trigger alternative microbial fermentation that produced less H 2 . Interestingly, these two mechanisms were previously attributed to HT (direct inhibition of methanogenesis) and CT tannins (reduction of fibre degradability), respectively (Goel and Makkar 2012;Cardoso-Gutierrez et al. 2021). Indeed, most of the tannins-EOC mixtures belonging to the C and Q groups significantly reduced net H 2 , whereas the two modes of action seemed to counteract each other when HT and CT were combined in the C/Q group. However, effects on net H 2 cannot be fully attributed to tannin content as both C and Q tannins without EO/EOC blends were not effective. Nevertheless, the rumen microbiota was not evaluated, and further studies are needed to confirm this hypothesis, as almost all feed additives did not decrease the protozoa count (H 2 -producing microorganisms), conversely to what was reported by Salami et al. (2018) supplementing either HT or CT. Since protozoa have a significant role in digesting fibre, the lack of effect on them likely prevented a considerable reduction in feeding value.
Different from CH 4 , the NH 3 formation was mitigated by all types of additives containing tannins, with some treatments of the C group being particularly efficient. The inhibiting effect of the mixtures was likely due to the tannin extract component, as tannins were effective on NH 3 also as pure additives, and this was not observed with pure EO/EOC. Still, the tannin-EOC mixtures seemed to have a synergistic effect against NH 3 production, particularly for the mixtures of the C-group. Indeed, the level of reduction of some mixtures was greater (up to 31%) if compared to the pure chestnut extract addition. This effect can probably be explained by the combination of the proteinbinding properties of the tannins and the direct inhibition of NH 3 -producing bacteria caused by the EOC (Patra and Yu 2012;Sinz et al. 2019;Honan et al. 2021). Besides, there was also an unexpected result. The proportion of the iso-butyric acid (expressed as mmol per mol of total VFA), which is formed by deamination of valine (Allison 1978), significantly increased when most of the EO-tannin mixtures of the C/Q group were included (except for C/Q-1 and C/Q-12), suggesting a greater deamination of valine in contrast to control. This was not the case with the C or Q groups, as previously reported by Jayanegara et al. (2015) supplementing either quebracho or chestnut extract.

Conclusions
Tanniferous extracts of both, chestnut and quebracho turned out to cause most of the mitigating properties in CH 4 and NH 3 formation found with the mixture treatments. The role of the EO/EOC, instead, was to facilitate the extent of the mitigations, in particular when CH 4 production was related to the total VFA and digestible OM, which was relevant to account for the slight adverse effects determined by all the additives on the feeding value. Considering all results, only six mixture treatments turned out to be really promising. Therefore, further studies are needed to understand the mechanisms of actions and the synergistic effects of these mixtures. Finally, in vivo studies will confirm whether or not the effect of the studied mixtures are effective in live animals and will evaluate the persistence of the effect on CH 4 and NH 3 mitigation in a long-term period.