Effects of cashew nut shell extract and monensin on in vitro ruminal fermentation, methane production, and ruminal bacterial community

The objective of this study was to evaluate the effects of cashew nut shell extract (CNSE) and monensin on ruminal in vitro fermentation, CH 4 production, and ruminal bacterial community structure. Treatments were as follows: control (CON, basal diet without additives); 2.5 μ M monensin (MON); 0.1 mg CNSE granule/g DM (CNSE100); and 0.2 mg CNSE granule/g DM (CNSE200). Each treatment was incubated with 52 mL of buffered ruminal content and 500 mg of total mixed ration for 24 h using serum vials. The experiment was performed as a complete randomized block design with 3 runs. Run was used as a blocking factor. Each treatment had 5 replicates, in which 2 were used to determine nutrient degradability, and 3 were used to determine pH, NH 3 -N, volatile fatty acids, lactate, total gas, CH 4 production, and bacterial community composition. Treatment responses for all data, excluding bacterial abundance, were analyzed with the GLIM-MIX procedure of SAS v9.4. Treatment responses for bacterial community structure were analyzed with a PERMANOVA test run with the R package vegan . Orthogonal contrasts were used to test the effects of (1) additive inclusion (ADD: CON vs. MON, CNSE100, and CNSE200); (2) additive type (MCN: MON vs. CNSE100 and CNSE200); and (3) CNSE dose (DOS: CNSE100 vs. CNSE200). We observed that pH, acetate, and acetate: propionate ratio in the CNSE100 treatment were lower compared with CNSE200, and propionate in the CNSE100 treatment was greater compared with CNSE200. Compared with MON, CNSE treatments tended to decrease total lactate concentration. Total gas production of CON was greater by 2.63% compared with all treatments, and total CH 4 production was reduced by 10.64% in both CNSE treatments compared with MON. Also, compared with MON, in vitro dry


INTRODUCTION
Cattle produce an estimated 4.6 gigatons of CO 2equivalent units per year, and contribute approximately 65% to the total greenhouse gas emissions from the agricultural sector (Gerber, 2013).Among the greenhouse gases produced from enteric fermentation, 71% originate from ruminal fermentation (Opio et al., 2013).Therefore, it is crucial to study CH 4 mitigation approaches that target ruminal fermentation.Promising strategies to reduce CH 4 production involve feed additives, such as plant secondary metabolites, enzymes, and ionophores, (Hook et al., 2010;Watanabe et al., 2010;Hristov et al., 2013).Feed additives such as cashew (Anacardium occidentale) nut shell extract (CNSE) and monensin have exhibited selective antimicrobial activity in previous in vitro (Watanabe et al., 2010;Sarmikasoglou et al., 2023) and in vivo (Appuhamy et al., 2013) studies, respectively.
Cashew nut shell extract contains secondary metabolites, such as anacardic acid, cardanol, and cardol, which are phenolic compounds with observed antimicrobial (Kubo et al., 1993), antioxidative (Kubo et al., 2006), and antitumor activities (Itokawa et al., 1987).Previous reports have shown that these phenolic compounds, especially anacardic acid, exhibit bacteriostatic effects on gram-positive bacteria (Kubo et al., 1993).In addition, CNSE, which contains these phenolic compounds, has been also found to selectively inhibit gram-positive lactate-producing ruminal bacteria, such as Streptococcus bovis, which are believed to contribute to ruminal acidosis and bloat in feedlot cattle (Nagaraja and Titgemeyer, 2007;Watanabe et al., 2010;Compton, 2021).Finally, anacardic acid has been reported to decrease CH 4 emissions in vitro, and thus has been speculated to have negative effects on methanogenic archaea (Van Nevel et al., 1971).The extraction procedures, as well as mode of CNSE administration, are challenges for its supplementation in vivo.Previous studies that fed 20 to 37 g/cow per day of pelleted cashew nut shell reported an increase in propionate and decrease of CH 4 emissions (Shinkai et al., 2012;Mitsumori et al., 2014), but other studies that supplemented 30 g/cow per day of top-dressed heat-processed cashew nut shell found no effects (Branco et al., 2015).
Monensin is a carboxylic polyether ionophore antibiotic (Haney and Hoehn, 1967) commonly used as a coccidiostat for poultry, as well as a nonhormonal growth promoter in cattle (Russell and Strobel, 1989).Previous studies have shown that monensin inhibits the growth of H 2 -producing bacteria (Chen and Wolin, 1979), reducing H 2 production and thereby CH 4 (McGuffey et al., 2001).Furthermore, monensin is partially effective at decreasing CH 4 , with most of its methane inhibition related to improvements in feed efficiency (Hristov et al., 2013).More specifically, it has been observed to improve feed efficiency by 2.5% in dairy cows (Duffield et al., 2008) and by 6.4% in feedlot cattle (Duffield et al., 2012), as well as increasing the concentration of conjugated linoleic acid in milk fat in cattle (AlZahal et al., 2008).
Use of antibiotics as feed additives has led to concerns over the development of antimicrobial resistance, posing a potential threat to human health (Gadde et al., 2017).However, the efficacy of alternatives to antibiotics is not consistent in the literature and depend on their effects on the ruminal microbiome.Currently, very few studies have comprehensively evaluated CNSE as a potential modulator of ruminal microbial populations and fermentation patterns, and as a CH 4 mitigation strategy.Furthermore, pressing global demands to reduce antibiotic use merits the investigation of alternative solutions.Therefore, a comparison between the effects of monensin and CNSE on ruminal fermentation and bacterial communities warrants investigation.The hypothesis of the present study was that CNSE would modulate ruminal fermentation toward a reduction in CH 4 production and improve ruminal fermentation related to feed degradation similarly to or to a greater extent than monensin.Our objective was to determine the effects of CNSE and monensin on ruminal in vitro fermentation, CH 4 production, and ruminal bacterial communities.

Ethical Approval
The University of Florida Institutional Animal Use and Care Committee approved all the procedures for animal care and handling required for this experiment.

Experimental Design
The experiment was a complete randomized block design with 3 runs.Run was used as blocking factor.Each treatment had 5 replicates, with 2 used to determine the in vitro dry matter degradability (IVDMD) and in vitro neutral detergent fiber degradability (IVNDFD), and 3 used to determine the pH, total lactate, VFA, NH 3 -N, gas production, and bacterial composition.

Monensin and Cashew Nut Extract Treatments
Four treatments were used in this study: (1) control without feed additives (CON); (2) monensin sodium salt (MON); (3) low dose of cashew nut shell extract (CNSE100); and (4) double dose of cashew nut shell extract (CNSE200).Treatments were provided at the following doses: MON at 2.5 μM, final concentration, monensin sodium salt (M5273, Sigma-Aldrich Chemicals, Burlington, MA), CNSE100 and CNSE200, which corresponds to 0.1 and 0.2 mg CNSE granule/g DM, respectively (SDS Biotech K.K., Tokyo, Japan).The CON treatment was supplied with neither additive.The concentration of monensin was chosen based on previous dose-response studies with protozoal cultures to ensure that monensin will not eliminate protozoa after the initial dosing so that its effect would be limited to bacteria (Karnati et al., 2009;Sylvester et al., 2009).The concentrations of the 2 doses of CNSE were selected according to manufacturer guidelines and are comparable to the ones used in previous studies (Watanabe et al., 2010).
According to manufacturer protocol and as established by previous studies (Capelari and Powers, 2017;Shen et al., 2017), monensin is insoluble in water; thus, MON was diluted in absolute ethanol.A stock solution (100×) was made before the experiment, stored at −20°C, and pipetted into serum vials immediately before inoculation with rumen fluid.An equal volume of absolute ethanol was pipetted to the remaining serum vials, to account for any effects from absolute ethanol.The respective CNSE doses were weighed individually in a precision scale (Mettler Toledo, Greifensee, Switzerland) and supplemented to the respective bottles.

Feeding and Management
Two cannulated Holstein cows in mid-lactation (108 ± 9.00 DIM on average) were used as ruminal inoculum donors.Cows were fed twice a day a TMR with 38% corn silage, 19% ground corn, 13% soybean meal, 11% cotton seed, 9% citrus pulp, 8.5% mineral premix, and 1.5% palmitic acid supplement (on a DM basis) from 3 wk before start until completion of the experiment.Neither monensin nor CNSE were included in the TMR fed to the ruminal inoculum donors.

In Vitro Fermentation
Approximately 3 h after morning feeding, ruminal contents were manually collected in equal proportions from each cow and strained through 4 layers of cheesecloth, transferred into prewarmed thermoses (Coleman, Chicago, IL), transported to the laboratory, and immediately mixed together.The ruminal fluid used for incubations was then added to a buffered prewarmed (39°C) medium (Goering and Van Soest, 1970) in a 1:2 ratio (rumen fluid: artificial saliva).The medium was continuously infused with CO 2 to maintain the anaerobic environment for the rumen fluid inoculum.
The in vitro experiment was conducted by using serum vials.Twenty serum vials of 160-mL volume were used, containing 0.50 g of substrate and the corresponding inclusion rate for each treatment.As shown in Table 1, the experimental substrates were formulated to provide the same concentration of nutrients regardless of treatment and according to the NRC (2001) recommendations for lactating Holstein cows with 680 kg BW and milk production of 45 kg/d with 3.5% fat, 3.0% protein, and 4.8% lactose.Nutrient compositions of feed ingredients were determined in samples ground through a 1-mm screen in a Wiley mill (model no.2; Arthur H. Thomas Co., Philadelphia, PA) and sent for analyses to SDK Laboratories (Hutchinson, KS).Samples were analyzed for DM (Shreve et al., 2006), ash (AOAC, 1990, method 942.05),NDF and ADF sequentially using the Ankom system (Schlau et al., 2021;Appendix A in Ankom Technology, 2014) with ash and heat-stable α-amylase and sodium sulfite for NDF, starch (Hall, 2009), crude fat (Modified AOCS Am5-04), and CP (AOAC International, 2000;method 990.03).The nutrient composition from each feed ingredient was used as a reference for formulation of the experimental substrates.All feed ingredients were ground through a 2-mm screen in a Wiley mill (model no.2; Arthur H. Thomas Co., Philadelphia, PA).Before grinding, corn silage was dried for 72 h at 60°C in a forced-air oven (Heratherm, Thermo Scientific, Waltham, MA), until DM was approximately 90%, allowing for proper grinding and storage.
Each treatment had 5 serum vials, and each experimental run had 5 blank serum vials (buffered rumen fluid only) with no substrate, to correct for background gas production.Buffered rumen fluid (52 mL) was added to the 160-mL serum vials containing Ankom F-57 filter bags or substrate, and a continuous stream of CO 2 was flushed into the vials during the whole inoculation process.Serum vials were closed with rubber stoppers, crimped with aluminum seals, and then placed at 39°C for 24 h in a mild shaking air-forced incubator.

Gas Pressure and CH 4 Concentrations
Gas pressure was measured 24 h after incubation using a pressure transducer (JYB-KO-M Pressure Transducer, Kunlun Tech Co Ltd., Beijing, China) for determination of total gas concentration.Based on the conditions in the laboratory, the following equation was used to convert pressure to volume: gas volume (mL) = [gas pressure (psi) × 4.8843] + 3.1296.
Headspace gas (10 mL) was collected from each serum bottle at 24 h, using a sealed gas injection needle, for determination of CH 4 concentrations.Concentration of CH 4 was analyzed by gas chromatography (Agilent 7820A GC; Agilent Technologies, Palo Alto, CA).A flame ionization detector was used with a capillary column (Plot Fused Silica 25 m by 0.32 mm, Coating Column pressure was 20 psi with a flow of 1.8552 mL/ min.Detector makeup flow was 21.1 mL/min.The carrier gas was N 2 , and the run time was 3 min.The total CH 4 (mL) was calculated by multiplying the calculated gas volume (mL) with the measured relative concentration of CH 4 (mg/mL), and then converting it to volume (mL) based on ρ = 0.619 kg/m 3 , at 39°C, and 1 atm.

Sampling for Lactate, VFA, NH 3 -N, IVDMD, IVNDFD
At the end of the 24-h incubation, serum vials were opened, and pH was immediately measured using a portable pH meter probe (Orion Star A121, Thermo Fisher Scientific, Waltham, MA).Subsamples were collected from vials with substrate for later determination of lactate, VFA, NH 3 -N, and bacterial community contents.Samples were stored at −20°C for lactate, VFA, and NH 3 -N and at −80°C for bacterial community abundance.
The IVDMD and IVNDFD were estimated from the vials with the substrate preweight into the Ankom bags.

Lactate Concentration
The ruminal fluid used for lactate analysis was first boiled at 100°C for 10 min and then centrifuged at 10,000 × g for 10 min at 4°C.The supernatants were collected and used for lactate analysis.The UV method was used for the determination of d-and l-lactic acid concentrations by using the d-Lactic Acid/l-Lactic Acid Kit from R-Biopharm (Darmstadt, Germany).The analysis procedure was adapted from the manufacturer's protocol.Specifically, 20 μL of sample, 80 μL of deionized water, a 100-μL mix of glycylglycine buffer and l-glutamic acid, 20 μL of NAD, and 2 μL of glutamate-pyruvate transaminase suspension were loaded in a 96-well UV microplate, incubated at room temperature (RT) for 5 min, and read at 340 nm on a microplate spectrophotometer (Spectra Max 340 PC, Molecular Devices Corporation, Sunnyvale, CA).Then, 2 μL of d-lactate dehydrogenase solution was added to each well and incubated at RT for 180 min, and absorbance was read to determine the concentration of d-lactic acid.Then 2 μL of l-lactate dehydrogenase solution was added to each well and incubated at RT for another 180 min, and absorbance was read to determine the concentration of l-lactic acid.A 99% of recovery of spiked concentration for both d-and l-lactic acid was considered as non-inhibitory dilution.

VFA Analysis
Samples were processed according to Ruiz-Moreno et al. (2015).Briefly, the ruminal fluid used for VFA analysis was thawed at RT and centrifuged at 10,000 × g for 15 min at RT; then the supernatant was mixed with crotonic acid and metaphosphoric acid solution and frozen overnight.The sample was then thawed at RT and centrifuged at 10,000 × g for 15 min at 4°C, and the supernatant was collected.The final supernatant was mixed with ethyl acetate (2:1 ethyl acetate: supernatant), vortexed, and allowed to settle, with the top layer being transferred to a chromatography injection vial for analysis.Concentrations of acetate, propionate, butyrate, valerate, isobutyrate, and isovalerate in samples were analyzed via gas chromatography (Agilent 7820A GC, Agilent Technologies, Palo Alto, CA) with a flame ionization detector and a capillary column (CP-WAX 58 FFAP, 25 m, 0.53 mm, Varian CP7767, Varian Analytical Instruments, Walnut Creek, CA) where temperature was maintained at 110°C, with injector temperature at 200°C and detector at 220°C.

NH 3 -N
Concentration of NH 3 -N in samples was analyzed according to Chaney and Marbach (1962).Samples were thawed at RT and centrifuged at 10,000 × g for 15 min at 4°C, and the supernatant was analyzed using the phenol-hypochlorite method in a 96-well flat-bottom plate.Absorbance was measured with a spectrophotometer (SpectraMax Plus 384 Microplate Reader, Molecular Devices, San Jose, CA) at 620 nm.

Chemical Analyses
The IVDMD and IVNDFD were calculated after 24 h of incubation.The procedure followed Cappellozza et al. (2023).Briefly the preweight Ankom bags were taken out of serum vials, washed with tap water until effluent was clear, and dried in a forced-air oven set at 60°C for 48 h.Dried residue samples were analyzed for NDF using an Ankom200 Fiber Analyzer (Ankom Technology, Macedon, NY) with heat-stable α-amylase and sodium sulfite.Residue weights and their NDF concentrations were used to calculate IVDMD and IVNDFD.

DNA Extraction, PCR Amplification, and rRNA Sequencing
Total genomic DNA from ruminal samples were extracted using the Quick-DNA Fecal/Soil Microbe Miniprep Kit (D6010, Zymo Research Corporation, Irvine, CA), following the manufacturer's instructions.Before storage in −80°C, the extracted DNA concentration was measured using a Qubit Fluorometer (Invitrogen, Waltham, MA).
DNA was sequenced according to Kozich et al. (2013).Amplification with PCR was performed in a C1000 Touch Thermal Cycler (Bio-Rad Laboratories, Hercules, CA).The V4 region of the 16S rRNA gene was amplified by dual-index universal bacterial primers (515 forward: 5′-GTGCCAGCMGCCGCG-GTAA-3′; 806 reverse: 5′ GGACTACHVGGGTWTC-TAAT-3′; Caporaso et al., 2011) through an initial denaturation of 5 min at 95°C, followed by 30 cycles of 30 s at 95°C, 30 s at 55°C, 1 min at 72°C, and 5 min for final elongation at 72°C.Forward and reverse primers, as well as small DNA fragment contaminants, were removed using a 1% low-melting agarose gel extraction kit (National Diagnostics).Amplicons were then purified and normalized using a SequalPrep plate kit (Invitrogen, Waltham, MA), and the DNA concentration was measured with a Qubit Fluorometer (Invitrogen, Waltham, MA).Adapters were added to the amplicons, and the DNA library was constructed by equally pooling all the amplicons together and using quantitative real-time PCR for quality check.Sequencing was performed using a MiSeq Reagent Kit V2 (2 × 250 cycles run; Illumina) in an Illumina MiSeq platform at the Interdisciplinary Center for Biotechnology Research at the University of Florida (Gainesville, FL).Sequences were deposited at the Sequence Read Archive of the National Center for Biotechnology Information (https: / / www .ncbi.nlm.nih.gov/sra) under access no.PRJNA772193.

Bioinformatics and Analyses
Sequenced amplicons were processed using the DADA2 pipeline (version 1.16) in R (Callahan et al., 2016), and taxonomy assignment was performed using the Bayesian Ribosomal Database Project classifier trained with the RDP_train_set_18 database (Cole et al., 2014;Edgar, 2018).Briefly, paired-end raw reads were demultiplexed, and the quality profiles of the forward and reverse readings were separately inspected, filtered, and trimmed based on the relationship between error rates and quality scores.Forward and reverse readings were merged, chimeras were removed, and an amplicon sequence variants table was created.The resulting tables were converted into a phyloseq object for downstream analyses.(McMurdie and Holmes, 2013).Before further data analysis, we calculated the coverage of the data set according to Good (1953) to evaluate whether the numbers of sequences obtained for each sample were adequate to provide representativeness of the bacterial community (Supplemental Table S1; https: / / doi .org/ 10 .6084/m9 .figshare.24038319;Sarmikasoglou, 2023).
After rarefaction, all samples had coverage >80% and thus were considered representative.
To protect against amplicon sequence variants (ASV) with small means and large coefficients of variation, ASV not seen more than 3 times in at least 20% of the samples were removed.Sequencing depth was normalized by the minimum library size (10,100 sequences per sample) to perform all microbiome analyses.The batch effect introduced by our experiment block design was tested and corrected with the MultiBaC package in R (RStudio 3.0.1,https: / / www .r-project .org;Ugidos et al., 2022).
Treatment effects were considered as fixed effects, and run was considered random effect.Orthogonal contrasts were used to test the effects of (1) additive inclusion (ADD: the control compared with all treatments with additives [CON vs. MON, CNSE100, and CNSE200]); (2) additive type (MCN: treatment without CNSE compared with those with CNSE [MON vs. CNSE100 and CNSE200]); and (3) CNSE dose (DOS: the 0.1 mg CNSE granule/g DM compared with the 0.2 mg CNSE granule/g DM dose [CNSE100 vs. CNSE200]).
Bacterial α-diversity indices (Chao1 and Shannon) were calculated with the R package phyloseq (McMurdie and Holmes, 2013).Comparison of the overall bacterial communities, implemented by using the Bray-Curtis dissimilarity, was visualized by principal coordinates analysis, and the statistical differences among samples were measured by PERMANOVA using the R package vegan (Oksanen, 2007).Orthogonal contrasts (ADD, MCN, and DOS) were used to test the effects of the treatments on genus differential abundance using the R package limma (Ritchie et al., 2015).Significance was declared at P ≤ 0.05, and 0.05 < P ≤ 0.10 was considered a tendency.

Effects on Ruminal Fermentation Characteristics
Buffered ruminal fluid pH from serum vials was measured after 24 h of incubation and was lower (DOS, P = 0.02) in CNSE100 compared with CNSE200 (Table 2).In a previous semicontinuous culture study, 3 levels of CNSE (50, 100, and 200 μg/mL) were supplemented in a high-grain substrate (70:30, concentrate: forage), and ruminal pH had a dose-dependent reduction at all levels of supplemented CNSE (Watanabe et al., 2010).In addition, similarly to our findings, the 100-μg/mL dose had a lower ruminal pH compared with 200 μg/ mL, which indicates that concentration of CNSE at 200 μg/mL would potentially affect the growth of bacteria involved in lactate metabolism in the rumen.One hypothesis on the potential underlying mechanism would be that 200 μg/mL of CNSE would exhibit greater bactericidal activity against lactate-producing bacteria.However, previous studies in humans reported that Lactobacillus spp., gram-positive lactate producers, were able to metabolize phenolic compounds, thus making them resistant to the bactericidal effects of these compounds (Lee et al., 2006).In general, higher doses of toxins (e.g., phenolic compounds) can impose greater selection pressure on mutations that confer resistance to them (Raymond, 2019); thus, the concentration of 200 μg/mL of CNSE would result in the development of tolerance to lactate-producing bacteria.Consequently, higher levels of lactate would result in more substrate for lactate utilizers, such as Megasphaera spp., which, in our study, exhibited greater abundance in CNSE200 compared with CNSE100.
To our knowledge none of the aforementioned hypotheses have been previously published; however, the limited diet adaptation time that serum vial culture provides does not allow for conclusive statements.Therefore, future studies should focus on evaluating the bactericidal potential of CNSE, as well as tolerance development by lactate-producing bacteria.
In addition, data from a previous growth inhibition study showed that Ruminobacter amylophilus, Succinivibrio dextrinosolvens, and Selenomonas ruminantium in pure culture were resistant to CNSE at doses ≥50 μg/mL (Wakai et al., 2021).Interestingly all of the aforementioned genera and species are present in the ruminal microbiome and produce lactate; thus we would point out these microbes as potential targets for further research regarding their lactic acid metabolism in the presence of 0.2 mg CNSE granule/g DM.Moreover, the lack of effects on ADD and MCN on ruminal pH (Table 2) indicates that the tested feed additives have similar effects to ruminal pH; thus the presence and type of the tested feed additives seem to not influence the ruminal pH.Previous studies conducted in dairy cows reported no effects in monensin-treated cows under acidosis challenge (Mutsvangwa et al., 2002) or no acidosis (Haïmoud et al., 1995;Ruiz et al., 2001), whereas others that evaluated monensin supplementation in transition dairy cows under subclinical acidosis found an increase (Green et al., 1999) (Dennis et al., 1981).In our study, the CNSE doses tended to increase the relative abundance of Streptococcus spp.by 26%, tended to decrease lactate production by 20% and the relative abundance of Butyrivibrio spp.by 6.3% compared with MON, and also had numerically greater ruminal pH compared with MON, which would suggest that some lactate-producing bacteria have been suppressed by CNSE doses in a greater extent.However, the absence of MCN effect in Eubacterium spp.and Ruminococcus spp.indicates the potential selectivity of CNSE toward these genera.Therefore, further studies are needed to evaluate the potential of CNSE to suppress the accumulation of lactate and subsequently mitigate the development of metabolic diseases such as ruminal acidosis.
Data on total VFA concentration and individual molar proportions are presented in Table 2. Total concentration of VFA and molar proportions of butyrate, valerate, isobutyrate, and isovalerate were similar across treatments, but there were no significant contrasts for these variables either.Acetate molar proportion was lower in CNSE100 compared with CNSE200 (DOS; P = 0.04), propionate molar proportion was greater in CNSE100 compared with CNSE200 (DOS; P < 0.01), and the acetate-to-propionate ratio was lower in CNSE100 compared with CNSE200 (DOS; P < 0.01).Previous studies that evaluated the effects of CNSE on VFA reported that 500 μg of CNSE per milliliter of buffered ruminal fluid (approximately 7.5 g CNSE/100 kg BW, assuming a rumen content equivalent to 15% of BW; Van Soest, 1994) increased the total short-chain fatty acid (SCFA) concentrations when supplemented with a 50:50 forage: concentrate substrate in serum culture (Oh et al., 2017), whereas others reported a reduction or lack of effects on SCFA at doses of 4 g of CNSE/100 kg BW per day in dairy cattle (Shinkai et al., 2012;Konda et al., 2019).Similar to reports among dairy cattle, supplementation of 2 to 4 g of CNSE/100 kg BW to sheep had no effects on SCFA concentration (Kang et al., 2018).In our study, the total SCFA concentrations were averaged among treatments to 87.38 mM, and, in accordance with previous reports, no effect was observed (data not shown).Regarding individual VFA, previous studies reported a decrease in acetate, followed by an increase in propionate in response to CNSE, whereas for butyrate some reports have found reduced or no effect on its concentration (Shinkai et al., 2012;Oh et al., 2017;Konda et al., 2019).In our study, we evaluated more VFA in addition to the SCFA and found no effects on ADD; however, the DOS effect indicates that a greater inclusion rate of CNSE shifts the fermentation to greater acetate and lower propionate.Overall, the effects of CNSE on VFA vary because of variations in basal diet composition, DMI, animal category, dose tested, and specific treatment.Finally, the lack of effects of ADD and CNSE in total or individual VFA (Table 2) suggests that both MON and the 2 doses of CNSE tested may be affecting VFA production in similar ways, despite their different modes of action against bacteria.
Regarding the NH 3 -N concentration, no effects by any of the orthogonal contrasts were observed (Table 3).Similar to our study, previous trials have reported no effects on NH 3 -N when monensin was supplemented in lactating cows' diets in a continuous flow (Karnati et al., 2009;Ye et al., 2018) as well as when technical cashew nutshell liquid (TCNSL), which contains mainly cardanol, was administered to multiparous lactating dairy cows (Branco et al., 2015).
Regarding the IVDMD, we detected an increase (MCN, P = 0.05) in MON compared with CNSE treatments and a tendency to decrease (ADD, P = 0.09) when additives were compared with control (Table 3).Greater IVDMD has been reported in in vitro supplementation of raw CNSE (50-200 μg/ mL) regardless of the dose (Watanabe et al., 2010), whereas in vivo supplementation of TCNSL at 30 g/cow per day to multiparous lactating Holstein cows had no effect (Branco et al., 2015).Monensin effects on diet digestibility vary in the literature, where some studies have reported no effects (Osborne et al., 2004;Tebbe et al., 2018;Boardman et al., 2020;e Silva et al., 2021), but others have found greater DMD in grain-fed cattle (Dinius et al., 1976).
Finally, no ADD, MCN, or DOS effects were observed for IVNDFD.Consistent with our findings, previous continuous flow studies supplemented 0.34 μM (Ye et al., 2018) and 2.5 μM (Karnati et al., 2009) monensin and reported no effects on NDF, whereas others that supplemented haylage with a premix of monensin so-dium salt at 22 mg/kg of feed to multiparous lactating cows found that monensin tended to increase NDFD compared with the control group (Osborne et al., 2004).Regarding CNSE, Branco et al. (2015) reported that supplementation of TCNSL by 30 g/cow per day to multiparous lactating Holstein cows tended to increase NDFD compared with the control group.Overall, all the additives tested tended to hinder IVDMD and monensin had a greater IVDMD compared with CNSE, indicating that using CNSE to improve nutrient utilization requires further investigation.
Degradability characteristics were evaluated following the procedure described by Cappellozza et al. (2023); substrates were put into the Ankom bags before incubation and incubated together for 24 h.According to Schlau et al. (2021), the aforementioned procedure is dependent on the feed used as substrate and in general exhibits lower fermentability compared with other studies in which the substrate is transferred to fiber bags after incubation (Jiang et al., 2020).Moreover, the agitation of vials with the substrate in the bags could disturb microbial activity, especially considering the volume of feed relative to the surface area of fiber bags.Therefore, the degradability characteristics of this study could be affected by a potential lower fermentability of the serum vials, which is intrinsic to the methodology.

Effects on Total Gas and CH 4 Production
Total 24-h gas production showed a decrease of 2.70% (ADD, P = 0.02) when the feed additives were compared with CON (Table 2), whereas both the 24-h total CH 4 production and total CH 4 per gram of DM incubated had 10.6% and 10.4% decrease (MCN, P = 0.04), respectively, when cashew treatments were compared with MON (Table 2).
Previous studies have highlighted the anti-methanogenic potential of ionophores, such as monensin (Russell and Strobel, 1989;McGuffey et al., 2001;Hristov et al., 2013), as well as phenolic compounds, such as CNSE and TCNSL (Hook et al., 2010;Watanabe et al., 2010;Branco et al., 2015).Therefore, as expected, in the present study the total gas production was decreased in response to all tested feed additives, which indicates that both monensin and CNSE can decrease total gas production from ruminal fermentation in a similar pattern.Moreover, all additives tested tended to exhibit a lower IVDMD, which further explains the decrease in total gas and indicates an inhibitory effect on fermentation.Regarding the MCN effect in total CH 4 and CH 4 per gram of DM incubated, that indicates that the anti-methanogenic potential of CNSE is greater compared with monensin.Of note is the lack of ADD effect, which indicates that CH 4 production was not affected by the additive tested.In the present study, no differences were found for total CH 4 production between the additives.In previous studies, the effects of monensin and CNSE on CH 4 production have been inconsistent.Regarding monensin, previous studies have reported a decrease in CH 4 production (Sj et al., 2005;Odongo et al., 2007), whereas others have reported no effects (Karnati et al., 2009;Grainger et al., 2010).Concerning CNSE, some previous studies have reported a decrease in CH 4 production (Watanabe et al., 2010;Shinkai et al., 2012), but others have reported no effects (Branco et al., 2015).In addition, CH 4 production keeps the H 2 concentration lower by shifting the fermentation by H 2 -producing saccharolytic bacteria and protozoa toward greater ATP-yielding pathways (Wolin et al., 1997); thus, to acquire more conclusive data regarding the potential of monensin and CNSE on H 2 mitigation by H 2 -producing bacteria, further research should focus on assessing the H 2 production and also detecting the abundance of H 2 -producing bacteria in the microbiome.Overall, the total gas and CH 4 data indicate that monensin and CNSE are shifting the ruminal fermentation toward lower total gas and CH 4 production.

2015)
. In general, the sealing of vials in the serum vial culture would prevent the CO 2 from venting, the sodium bicarbonate from buffering, and subsequently the pH from dropping (Kohn and Dunlap, 1998).In cases when the pH is below 6, this would result in methanogenesis inhibition (Van Kessel and Russell, 1996).In our study, the absence of venting would probably explain why the mean pH was close to 6.0, which is near the threshold of a potential suppression of methanogenesis.Finally, other studies have indicated that CO 2 accumulation in the headspace increases the availability of CO 2 in buffered ruminal fluid as the electron acceptor, and promotes CO 2 -reducing methanogens (Patra and Yu, 2013).Therefore, further studies should determine the number and activity of archaea using molecular techniques to elaborate on the influence of headspace CO 2 on methanogenesis.

Effects on Rumen Bacterial Community
A total of 1,840,767 reads were generated from 16S rRNA gene sequencing, from which 1,288,574 highquality sequences were retained for analysis after filtering, denoising, merging, and removing chimeras with the DADA2 pipeline.Based on those sequences, a total of 2,482 taxa (from kingdom, phylum, class, order, family, genus) were identified after taxonomy assignment.
The profile of bacterial communities is reported in Figure 1.The principal coordinates analysis indicates that the treatments contributed to 33% of the variation in community distances (P < 0.01).The analyses of treatment effects on richness and diversity at the ASV level of the bacterial community are presented in Table 4. Chao 1 richness index was greater (DOS, P = 0.04) in CNSE100 compared with CNSE200, whereas no effects were observed for the rest of the contrasts (Table 4).Regarding the Shannon diversity index, the feed additives had lower diversity (ADD; P < 0.01), MON had lower diversity (MCN; P < 0.01), and CNSE100 had greater diversity compared with CNSE200 (DOS; P < 0.01).Regarding monensin, previous studies found no effect on richness and a decrease in diversity (Schären et al., 2017;Shen et al., 2017;Melchior et al., 2018), which are in accordance with our findings.Moreover, previous studies reported that CNSE (4-6 g/100 kg BW) decreased both richness and diversity (Maeda et al., 2021).Overall, it seems that monensin and CNSE function in the same way on the structure of ruminal bacterial community; however, further research needs to be done in that direction.
To better understand the possible effects of monensin and CNSE on ruminal microbiome, we analyzed their effects on relative abundance at genus level.Based on the obtained taxa, a total of 20 out of 56 observed Principal coordinates analysis plots of Bray-Curtis dissimilarity matrix at amplicon sequence variant level, comparing the treatment effects on community structure of ruminal bacteria.Experimental treatments were CON, control (experimental diet); MON, monensin (experimental diet plus 2.5 μM monensin sodium salt); CNSE100, experimental diet plus 0.1 mg CNSE granule/g DM; and CNSE200, experimental diet plus 0.2 mg CNSE granule/g DM.Contrasts were ADD = CON vs. MON, CNSE100, CNSE200; MCN = MON vs. CNSE100, CNSE200; DOS = CNSE100 vs. CNSE200.The PERMANOVA analysis indicated that the treatments contributed to 33% of the variation in community distances (P = 0.001).genera had relative abundance greater than 1% (Table 5).The most abundant bacterial genus detected was Prevotella, which had a decrease of 4.30% (ADD, P < 0.01) when the feed additives were compared with CON (Table 5).In the rumen, Prevotella spp.are among the most predominant genera and produce various extracellular digestive enzymes, such as amylases, xylanases, peptidases, and deaminases (Griswold et al., 1999;Shen et al., 2018); thus they are important in the digestion and fermentation of nutrients, especially dietary protein and non-cellulosic polysaccharides (Wallace, 1996;Stevenson and Weimer, 2007;Bekele et al., 2010).Previous studies have highlighted that in general Prevotella spp.are sensitive to monensin; however, some species, such as Prevotella bryantii, seem to exhibit a resistance (Callaway andRussell, 1999, 2000;Ferme et al., 2008).Regarding CNSE, a previous report in a Rusitec system found that both Prevotella ruminicola and P. bryantii decreased their relative abundances when 50 to 200 μg/ mL of CNSE oil was included (Watanabe et al., 2010).In our study, the decrease in Prevotella relative abundance in response to the tested feed additives further validates the aforementioned reports and additionally indicates that the CNSE mode of action is strain independent within the Prevotella genus.
The relative abundance of Succiniclasticum decreased by 11.46% (MCN, P < 0.01) when monensin was compared with cashew treatments, but increased by 3.17% (DOS P = 0.02) between the cashew treatments (Table 5).The Succiniclasticum spp.are nonmotile, rod-shaped, anaerobic gram-negative bacteria associated with carbohydrate fermentation and, more specifically, on converting the succinate to propionate (van Gylswyk, 1995;Henderson et al., 2015).Due to their increased levels on high-grain diets, it has been suspected that they contribute to the synthesis of total free-ruminal LPS (Petri et al., 2020); however, this remains to be elucidated.Previous studies, found a decrease in Succiniclasticum spp.when haylage was supplemented with monensin at 33 mg/kg feed DM (Kim et al., 2014), whereas others observed an increase when monensin was supplemented at 5 μM (0.35 mg/ mL; Shen et al., 2017).In vitro, a liquid CNSE extract product supplemented at 166 μg/mL reduced CH 4 emission by 18.0% and was able to increase Succiniclasticum spp.population in a 48-h time span (Danielsson et al., 2014).In our study, we found a decrease in propionate production from CNSE200 compared with CNSE100 (Table 2), which indicates a lower conversion rate of succinate to propionate, thus lower relative abundance from the Succiniclasticum spp. in CNSE200 (Table 3).Overall, our findings are consistent with the aforementioned in vitro data; however, more research in pure culture and at species level are needed to elucidate the effect of monensin and CNSE on Succiniclasticum spp.
Treponema relative abundance was decreased by 8.28% (ADD, P = 0.02) when the feed additives were compared with CON (Table 5).Generally, Treponema spp. is found in the rumen and is involved in the degradation of plant polysaccharides (Ziołecki, 1979), as well as promoting Fibrobacter succinogenes activity to degrade cellulose (Stanton and Canale-Parola, 1980).Similar to our findings, previous studies observed a decrease in the relative abundance of Treponema spp. in response to monensin (Shen et al., 2017) and CNSE (Watanabe et al., 2010) supplemented to lactating dairy cow diets.Thus, monensin and CNSE could potentially exhibit bactericidal activity against Treponema spp.; however, further pure culture studies would provide further knowledge.
The relative abundance of Lentimicrobium increased by 25.2% (MCN, P = 0.01) and 6.70% (DOS, P < 0.01), when monensin was compared with cashew treatments and between the cashew treatments, respectively (Table 5).Lentimicrobium spp.belong to the Bacteroidetes phylum, and are a strictly anaerobic, nonmotile, gram-negative genus that possess hydrolytic properties (Wirth et al., 2019).Previous studies have reported the presence of Lentimicrobium spp. in the rumen; however, their functions in the rumen remain to be elucidated (Solden, 2018;Islam et al., 2021).To our knowledge, no previous studies have evaluated the Butyrivibrio relative abundance was increased by 11% (ADD, P = 0.01), when CON was compared with the feed additives (Table 5).Butyrivibrio spp.are capable of using complex plant structural polysaccharides, such as xylan and pectin (Bryant and Small, 1956;Hungate, 2013), to produce butyrate.In vitro, it has been observed that 5 μM monensin decreased the relative abundance of Butyrivibrio spp.(Shen et al., 2017), and that 3.13 μg/mL of CNSE inhibits the growth of Butyrivibrio fibrisolvens D1 (Watanabe et al., 2010).These findings are not in agreement with our data; however, we dosed 2.5 μM of MON and the inhibitory concentration of Butyrivibrio spp. was determined in pure culture.Thus, more research needs to be done toward the effect of CNSE on Butyrivibrio spp. in mixed culture.
Succinivibrio relative abundance had an increase of 14.4% (ADD, P = 0.03), in response to the feed additives compared with CON (Table 5).Members of Succinivibrio genus can use glucose and produce acetate and succinate (Bryant and Small, 1956).Additionally, some Succinivibrio strains express enzymes related to nitrogen-containing compound degradation and assimilation (Patterson and Hespell, 1985;Hailemariam et al., 2020).In vitro supplementation of 5 μM monensin (Shen et al., 2017) or CNSE (50-200 μg/mL; Watanabe et al., 2010) has been shown to increase the relative abundance of Succinivibrio, which is consistent with our findings.Our study supplemented lower doses of monensin and CNSE, which indicates that the promoting effect on the relative abundance of Succinivibrio can be achieved in lower concentrations than those previously tested.
The relative abundance of Megasphaera decreased by 24.53% (MCN, P < 0.01) and increased by 9.15% (DOS, P < 0.01) when monesin was dosed compared with cashew treatments and between the cashew treatments, respectively (Table 5).Megasphaera use lactate and prevent it from excessive accumulation in the rumen, thus preventing the development of ruminal acidosis (Chen et al., 2019).Especially, some species such as Megasphaera elsdenii have been found to ferment lactate 6 times faster than glucose (Hino and Kuroda, 1993); thus they are major direct-fed bacterial candidates to be used against acute ruminal acidosis (Monteiro et al., 2022).No effects on Megasphaera have been reported in regard to growth or metabolism in presence of monensin (Marounek et al., 1993).Regarding CNSE, the relative abundance of Megasphaera increased in vitro (Watanabe et al., 2010), but it has not been observed in vivo (Shinkai et al., 2012).Our findings indicate that the cashew treatments did not promote the growth of Megasphaera, which may be attributed to the lower doses used in our experiment.
On Schwartzia relative abundance, a decrease of 6.33% (ADD, P = 0.05) was observed, when CON was compared with the feed additives (Table 5).The genus Schwartzia is asaccharolytic, and its members are able to ferment succinate and produce propionate (van Gylswyk et al., 1997).Previous reports found negative association of Schwartzia and CH 4 emissions (Cunha et al., 2017), whereas others found a positive association of Schwartzia and CH 4 emissions after monensin supplementation (Ran et al., 2021).Our findings are consistent with the latter report, but to our knowledge no published reports exist on the effects of CNSE on Schwartzia relative abundance.Overall, more research is needed to elucidate the effects of monensin and CNSE in regard to Schwartzia physiology.
The relative abundance of Mogibacterium increased by 12.42% (MCN, P = 0.02), when CON was compared with the feed additives (Table 5).Mogibacterium is a commonly found genus in the rumen of cattle (Freetly et al., 2020) and sheep (Mi et al., 2018), and also has been associated with greater CH 4 emission from steers (Wallace et al., 2015).Monensin supplemented with vinasse substrate in vitro with ruminal fluid collected from the Maiwa yak decreased CH 4 production and did not affect the relative abundance of Mogibacterium (Xue et al., 2021).In contrast, CNSE inclusion resulted in an increase in the relative abundance of Mogibacterium in vitro (Danielsson et al., 2014).Our findings, are consistent with the previous reports, thus further validating that monensin decreases and CNSE increases the relative abundance of Mogibacterium.Finally, more research should be focused on the association of Mogibacterium with CH 4 emission in the presence of CNSE.
The relative abundance of Dialister increased by 14.3% (ADD, P = 0.05) in response to additives and by 18% (MCN, P = 0.04) in response to cashew nut shell treatments compared with MON (Table 5).In general, Dialister has been positively correlated with high-efficiency lambs (Ellison et al., 2017) and cattle (Mu et al., 2019).In addition, Dialister has been reported to promote starch and cellulose degradation, thus enhancing amylase and carboxymethyl-cellulase activity (Wang et al., 2016).Monensin has been reported to decrease the relative abundance of Dialister when supplemented at 33 mg/kg diet DM with a ~50% gain diet in vitro (Ran et al., 2021).However, in our study, Dialister increased its relative abundance, and cashew nut shell treatments increased the relative abundance of Dialister compared with monensin.
The relative abundance of Acidaminococcus decreased by 22.7% (MCN, P < 0.01) in response to cashew nut shell treatments compared with the MON, and had an increase by 4.3% (DOS, P < 0.01) when CNSE200 was compared with CNSE100 (Table 5).Acidaminococcus has similar characteristics to Dialister, in regard to starch and cellulose degradation in the rumen (Wang et al., 2016).Our results indicate that monensin enhances but CNSE suppresses the growth of Acidaminococcus.
The relative abundance of Pseudobutyrivibrio increased by 22.7% (MCN, P = 0.04) in response to cashew nut shell treatments compared with MON (Table 5).Pseudobutyrivibrio constitutes a commonly found fibrolytic bacterial genus in the rumen (Thoetkiattikul et al., 2013).Similar to our findings, previous reports have found a reduction in the relative abundance of Pseudobutyrivibrio in the presence of monensin in vitro (Shen et al., 2017) and no effects in the presence of TCNSL in vivo (Branco et al., 2015).
Overall, in vitro procedures such as serum vial culture experiments are commonly used for assessing the effects of substrates and additives on enteric CH 4 production.Compared with in vivo, the in vitro procedures reduce the effect of the animal, reduce the costs, and are easier to control.However, their limitations should be considered before conclusions are extrapolated to in vivo conditions.One important aspect is that the rumen microbial communities in vitro are not fully adapted to the tested diets, which could further affect CH 4 production.Therefore, in vitro systems importance is evident; however, the outcomes from them should be qualified as preliminary and further validated by research with greater microbial adaptation potential.

CONCLUSIONS
Total gas was decreased in all treatments compared with CON, and total CH 4 production was lower in cashew nut shell treatments compared with MON.Additionally, pH and acetate concentration were lower, whereas propionate concentration was greater in CNSE100 compared with CNSE200.Finally, the relative abundances of Prevotella, Treponema, and Schwartzia were lower but the relative abundances of Butyrivibrio and Succinivibrio were greater in all treatments compared with CON.Our results contribute to the current body of knowledge toward a better understanding of the effects of CNSE in ruminal fermentation, and help direct future research regarding the identification of alternatives to ionophores, such as monensin.Overall, future continuous culture and in vivo studies, as well as more in-depth analyses of the ruminal microbiome, such as whole-genome shotgun sequencing, are needed to further validate the efficacy of CNSE in modifying rumen fermentation.
Sarmikasoglou et al.: FEED ADDITIVES AND IN VITRO RUMINAL FERMENTATION

Table 1 .
Sarmikasoglou et al.: FEED ADDITIVES AND IN VITRO RUMINAL FERMENTATION Ingredients and chemical composition of experimental substrate Sarmikasoglou et al.: FEED ADDITIVES AND IN VITRO RUMINAL FERMENTATION

Table 2 .
in ruminal fluidSarmikasoglou et al.:FEED ADDITIVES AND IN VITRO RUMINAL FERMENTATION Effects of cashew nut shell extract (CNSE) and monensin sodium (MON) on pH, NH 3 -N, lactate, and total gas and CH 4 in serum vial culture .Overall, the comparison of monensin and CNSE provides new insights on the potential of CNSE as an alternative to monensin.To our knowledge, this is the first study comparing monensin and CNSE that reports ruminal pH; thus, further studies should focus on assessing the effects of monensin and CNSE on ruminal pH and evaluate any potential effects on ruminal bacterial species and their lactic acid metabolism.Samples were also collected after 24-h incubation to evaluate lactate, VFA, and NH 3 -N concentrations.Lactate concentration tended to increase (MCN, P = 0.06) in MON compared with CNSE treatments, but no effects of ADD or DOS were observed (Table 2).Both monensin and CNSE have been previously reported to selectively suppress the growth of lactate-producing bacteria, such as Butyrivibrio fibrisolvens, Eubacterium ruminantium, Lactobacillus ruminis, Ruminococcus albus, Ruminococcus flavefaciens, and Streptococcus bovis among others 1 Experimental treatments: CON = control (experimental diet); MON = monensin (experimental diet plus 2.5 μM monensin sodium salt); CNSE100 = experimental diet plus 0.1 mg CNSE granule/g DM; CNSE200 = experimental diet plus 0.2 mg CNSE granule/g DM. 2 Contrasts: ADD = CON vs. MON, CNSE100, CNSE200; MCN = MON vs. CNSE100, CNSE200; DOS = CNSE100 vs. CNSE200. 3P = acetate: propionate ratio.4otal gas production, mL/g of DM incubated.5Total C 4 (mL), based on ρ = 0.619 kg/m 3 , at 39°C, and 1 atm.6 CH 4 , mg/g of DM incubated. p

Table 3 .
al., Sarmikasoglou et al.: FEED ADDITIVES AND IN VITRO RUMINAL FERMENTATION Effects of cashew nut shell extract (CNSE) and monensin sodium on in vitro DM degradability (IVDMD) and in vitro NDF degradability (IVNDFD) in serum vial culture 2

Table 4 .
Sarmikasoglou et al.: FEED ADDITIVES AND IN VITRO RUMINAL FERMENTATION Effects of cashew nut shell extract (CNSE) and monensin sodium in richness and diversity of the ruminal bacterial community in serum vial culture

Table 5 .
Effects of cashew nut shell extract (CNSE) and monensin sodium on relative abundance of ruminal bacterial community composition at genus level Lentimicrobium spp. in response to monensin or CNSE; however, our findings indicate that monensin suppresses the growth whereas CNSE treatments enhance the growth of Lentimicrobium spp.
effects on