Effects of cashew nutshell extract and monensin on microbial fermentation in a dual-flow continuous culture

The objective of this study was to compare cashew nutshell extract (CNSE) to monensin and evaluate changes in in vitro mixed ruminal microorganism fermentation, nutrient digestibility, and microbial nitrogen outflow. Treatments were randomly assigned to 8 fer-menters in a replicated 4 × 4 Latin square design with 4 experimental periods of 10 d (7 d for diet adaptation and 3 d for sample collection). Basal diets contained 43.5:56.5 forage: concentrate ratio and each fermenter was fed 106 g of DM/d divided equally between 2 feeding times. Treatments were 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). On d 8 to10, samples were collected for pH, lactate, NH 3 -N, volatile fatty acids (VFA), mixed protozoa counts, organic matter (OM), and neutral detergent fiber (NDF) digestibility. Data were analyzed with the GLIMMIX procedure of SAS. Orthogonal contrasts were used to test the effects of (1) ADD


INTRODUCTION
Feed additives such as antibiotics and secondary plant metabolites have been reported to modulate ruminal fermentation with observed effects on animal growth and health (McGuffey et al., 2001;Calsamiglia et al., 2007;Duffield et al., 2012).Ionophores, such as monensin, and secondary metabolites of plants, such as phenolic compounds, have been shown to improve energetic efficiency by decreasing ruminal amino acid deamination, and acetate to propionate ratio (Ruiz et al., 2001;Watanabe et al., 2010) and reducing environmental pollutants such as NH 3 and CH 4 (Shinkai et al., 2012;Hristov et al., 2013;Knapp et al., 2014).They have also been shown to suppress the development of metabolic disorders, such as acidosis, by selectively inhibiting the growth of lactate-producing ruminal bacteria (Kubo et al., 1993;Osborne et al., 2004;Watanabe et al., 2010).
Monensin is a carboxylic polyether ionophore antibiotic (Haney and Hoehn, 1968) commonly used as nonhormonal growth promoter in cattle (Russell and Strobel, 1989), and it is produced by the fermentation of Streptomyces cinnamonensis (Anadón and Martínez-Larrañaga, 2014).Feeding monensin has been demonstrated to inhibit H 2 -producing bacteria growth (Chen and Wolin, 1979), thus resulting in decreasing the H 2 production and subsequently the growth of methanogenic archaea (McGuffey et al., 2001).Regarding ruminal fermentation, studies showed that inclusion of monensin decreases butyrate production in vitro (Russell and Strobel, 1989), the rate of ruminal biohydrogenation of UFA in continuous culture fermenters (Fellner et al., 1997) and increases the concentration of conjugated linoleic acid in milk fat in dairy cattle (AlZahal et al., 2008).However, from January 1, 2006, the Economic and Social Committee of the European Union (Economic and Social Committee of the European Union, 1998) has prohibited the use of antibiotics, including monensin, as growth promoters, and therefore feed additives with similar effects would be desirable.
Cashew (Anacardium occidentale) nut production has been increased approximately by 123% worldwide in the last 2 decades (FAO, 1997), which resulted in a subsequent increase of its major byproduct-cashew nutshell.Cashew nutshell contains a mixture of anacardic acid, cardanol, and cardol in its pericarp fluid, which could potentially be used to modify ruminal fermentation and suppress CH 4 production (Watanabe et al., 2010;Shinkai et al., 2012;Compton et al., 2023).Previous studies have shown that cashew nutshell extract (CNSE), and in particular anacardic acid, exhibits selective antibacterial activity against gram-positive lactate-producing ruminal bacteria such as Streptococcus bovis, which is associated with the development of ruminal acidosis and bloat in feedlot cattle (Green et al., 2008;Watanabe et al., 2010), similar to the effects observed when monensin is used.Furthermore, a study with dry dairy cows fed 60% hay and 40% concentrate supplemented with 4 g of CNSE per 100 kg of BW exhibited an increase in ruminal propionate production (Shinkai et al., 2012), which improves energy efficiency in the rumen, and supports glucose and lactose synthesis in the body, resulting in greater milk yield (Danfaer, 1994;Lemosquet et al., 2009).
Therefore, we hypothesized that CNSE would exhibit similar effects to monensin on modulating mixed ruminal microorganism fermentation in vitro.Our objective was to evaluate the effects of monensin and 2 doses of CNSE on in vitro mixed ruminal microorganism fermentation, nutrient digestibility, and microbial nitrogen outflow in a high producing dairy cow ration using a dual-flow continuous culture system.

MATERIALS AND METHODS
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 and Diets
Eight fermenters of a dual-flow continuous culture system were used in a replicated 4 × 4 Latin square design with a complete randomized arrangement of treatments.Treatments were (1) control, without feed additives (CON); (2) monensin sodium salt (MON); (3) single dose of cashew nutshell extract (CNSE100); and (4) double dose of cashew nutshell extract (CNSE200).
Treatments were fed at the following doses: MON at 2.5 μM, final concentration in the fermenter, 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 concentration of monensin was chosen based on previous dose response studies to avoid elimination of protozoa after the initial dosing, so that its effect would be limited to bacteria (Karnati et al., 2009a;Sylvester et al., 2009).The concentrations of the 2 doses of CSNE were selected according to manufacturer guidelines and are comparable to the those used in previous studies (Watanabe et al., 2010).
The same basal TMR (Table 1) was used for all treatments and was formulated according to the NRC (2001) recommendations for a high producing-lactating Holstein cow with 680 kg of BW, and milk production of 45 kg/d with 3.5% fat.Nutrient composition of feeds was determined in samples ground through a 1-mm screen in a Wiley mill (model N°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 analyzed sequentially using the Ankom system (Appendix A ADF method in Ankom Technology, 2014;Schlau et al., 2021) 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).Before the experiment, the corn silage was dried for 72 h at 60°C in a forced-air oven (Heratherm, Thermo Scientific, Waltham, MA), until it reached approximately 90% DM.All feed ingredients were ground through a 2-mm screen in a Wiley mill (model N°2; Arthur H. Thomas Co., Philadelphia, PA).
Each fermenter received 106 g/d DM distributed equally between 2 feedings at 0800 and 1800 h.The CNSE100 and CNSE200 were added as dry products to their respective diets at 2 daily equal doses.According to Sigma protocol and by previous studies (Sylvester et al., 2009;Capelari and Powers, 2017;Shen et al., 2017), monensin is insoluble to water, thus MON was diluted using absolute ethanol.A stock solution (100×) was made before the experiment, stored in −20°C, and pipetted into the fermenters immediately before both morning and evening feedings.The final ethanol concentration in the fermenters was less than 1.0% vol/ vol.Equal volume of absolute ethanol was pipetted to the other fermenters to account for any effects from absolute ethanol.

Dual-Flow Continuous Culture System Operation
A dual-flow continuous culture system as described by Hoover et al. (1976) and adapted by Paula et al. (2017) was used for this experiment.The fermentation vessels used had an average total volumetric capacity of 3 L and were filled at 1.82 L when inoculated until the solid effluent outflow port.Ruminal fermentation is simulated in this system through continuous agitation (100 rpm), infusion of N 2 gas to displace oxygen, constant temperature (39°C), and infusion of artificial saliva (Weller and Pilgrim, 1974) at 3.05 mL per minute to individually regulate passage rates of liquid (11% h −1 ) and solid (5.5% h −1 ) effluents of digesta (Brandao and Faciola, 2019;Brandao et al., 2020).
This experiment consisted of 4 fermentation periods of 10 d each (40 d of fermentation total).Ruminal fluid was collected approximately 3 h after morning feeding from 2 ruminally cannulated Holstein cows in mid lactation (108 ± 9 DIM on average) 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) 3 wk before start and until completion of the experiment.Ruminal digesta was collected from the ventral, central, and dorsal areas of the rumen, strained through 2 layers of cheesecloth, and transferred into prewarmed thermoses.Ruminal fluid from both cows were homogenized (50:50), and 1.82 L was inoculated to each fermenter.Fermenters were prewarmed and under continuous flush of N 2 gas during inoculation.

Experimental Procedure
On d 5 of each period, subsamples of homogenized liquid and solid effluents from each fermenter, and artificial saliva were collected to determine the abundance of background 15 N (Calsamiglia et al., 1996).Then 0.1733 g ( 15 NH 4 ) 2 SO 4 10.2% atom excess (Sigma-Aldrich Co., St. Louis, MO) was dosed to each fermenter to instantaneously label the NH 3 -N pool.Artificial saliva was reformulated, and ( 15 NH 4 ) 2 SO 4 was continuously added as a marker at a rate of 0.077 g/L until the end of each fermentation period.
On d 8, 9, and 10, the digesta containers of liquid and solid were partially submerged into an ice-cold water bath (<2°C) to prevent further microbial fermentation.On the last day (d 10) of each experimental period, microbial samples were collected according to the modified method of Krizsan et al. (2010) and described in detail in Brandao et al. (2018).The obtained microbial pellets were freeze-dried, grounded by mortar pestle, and later analyzed for 15 N enrichment, total N, and DM.

Collection of Data and Samples
On the last 3 d of each experimental period, data and samples were collected for analyses of VFA, NH 3 -N, and 15 N concentration (N metabolism), protozoal counts, and nutrient digestibility.

Daily Average pH, VFA, NH 3 -N, Lactate Concentrations, and Protozoal Counts
Before morning feeding on d 8, 9, and 10 of each period, fermenter pH was measured with a portable pH meter (Thermo Scientific Orion Star A121, Thermo Scientific, Waltham, MA), and pooled samples representing the whole day of fermentation were also collected from the homogenized sample (solid and liquid effluents) for determination of VFA, NH 3 -N, and lactate concentration.For VFA and NH 3 -N, 10 mL of sample from the mixed digesta effluent was strained through 4 layers of cheesecloth and immediately acidified with 100 μL of 50% H 2 SO 4 and stored at −20°C.For lactate, 1 mL of sample from the mixed digesta effluents was strained through 4 layers of cheesecloth and stored at −20°C.
For protozoal counts, an aliquot of 2 mL from the fermenter contents and the composite of the effluent containers were collected at 3 and 24 h after morning feeding, and were immediately fixed in formalin solution, and protozoal counts were determined by microscopy (Dehority, 1993;Oldick and Firkins, 2000).The protozoal counts from the fermenter and effluents were used to calculate the generation times of protozoa based on Sylvester et al. (2004), by using the following equation: Ruminal digestibility of nutrients from the basal diet was determined from samples collected on the last 3 d of each period.Digesta samples were collected before the morning feeding to represent 24 h of fermentation.
Briefly, once per day, before morning feeding, the liquid and solid effluent containers of the same fermenter were combined, mixed for 30 s, and then a sample of 500 g of digesta was collected and stored in − 20°C for later determination of DM, NDF, and 15 N abundance.
On the last day of each experimental period (d 10) the microbial content was isolated as described by Krizsan et al. (2010).Briefly, the content of each fermenter was blended for 30 s, filtered through 4 layers of cheesecloth, and washed with 0.9% saline solution.The filtrate content was centrifuged (Allegra X-15R Centrifuge, Beckman Coulter, Brea, CA) at 1,000 × g for 10 min to remove any protozoa and feed particles, and the supernatant was collected and ultracentrifuged (Sorvall RC-5B Refrigerated Superspeed Centrifuge, DuPont Instruments, Wilmington, DE) at 11,250 × g for 20 min.The pellet was resuspended in 200 mL of McDougall's solution, and spun (Sorvall Lynx 4000 Centrifuge, Thermo Scientific, Waltham, MA) at 16,250 × g for 20 min.Finally, the pellet was resuspended in dH 2 O and transferred into a plastic container and stored in −20°C for further evaluation of DM, total N, and 15 N abundance.
All background, digesta, and bacteria samples were freeze-dried and ground with a mortar and pestle at least 24 h after the completion of freeze-drying process.

Chemical Analyses
VFA and NH 3 -N.Ruminal fluid samples from the fermenters were thawed at room temperature, centrifuged at 10,000 × g for 15 min at 4°C, and the supernatant was collected for subsequent analyses.The determination of NH 3 -N concentration was done as described previously by Broderick and Kang (1980).Briefly, the samples were analyzed by the phenol-hypochlorite method in a 96-well flat-bottom plate.Absorbance was measured at 620 nm using a spectrophotometer (Spec-traMax Plus 384 Microplate Reader).For VFA analyses, subsamples of the supernatants collected and processed according to the method described by (Ruiz-Moreno et al., 2015).A crotonic-metaphosphoric acid solution was added to the subsamples at a 1:5 ratio and stored at − 20°C overnight, and then centrifuged again at 10,000 × g for 15 min at 4°C.The supernatant was mixed with ethyl acetate in 2:1 ratio, vortexed, allowed to settle, and finally the top layer transferred to a chromatography vial for determination of acetate, propionate, butyrate, isobutyrate, valerate, and isovalerate by gas chromatography (Agilent 7820A GC, Agilent Technologies, Palo Alto, CA) with a flame ionization detector and a capillary column (CP-WA × 58 FFAP 25 m 0.53 mm, Varian CP7767, Varian Analytical Instruments, Walnut Creek, CA) maintained at 110°C, with injector temperature at 200°C and detector at 220°C.Lactate was determined using a kit (R-Biopharm, Darmstadt, Germany) based on the procedure of Niederholtmeyer et al. (2010).
Percent Atom 15 N Analysis.Approximately 0.02 g of background, digesta and bacteria samples combined with 2.0 mm zirconia beads were ground (Precellys 24, Bead Mill Homogenizer, Bertin, France) at 5,500 × g for 10 s at RT for 15 N analysis.Based on expected 15 N concentrations 4, 2, and 1 mg of ball-milled sample of background, effluent, and bacteria, respectively, were weighed into 8 × 5 mm tin capsules (Elemental microanalysis, Devon UK) using a microbalance (Excellence Plus XP Micro Balance, Mettler-Toledo GmbH, Laboratory and Weighing Technologies, CH-8606 Greifensee, Switzerland).Then 35 μL of K 2 CO 3 (10 g/L) was added to each capsule and placed in a forced-air oven at 40°C to dry overnight (Reynal et al., 2005).Background, digesta, and bacteria samples were analyzed for DM, ash, and concentration of total N as described previously.
Calculations for N Metabolism and Digestibility of Nutrients.The total N in digesta effluent was divided into 3 fractions: NH 3 -N flow, dietary-N flow and microbial N flow.The dietary-and microbial-N flows combined constitute the non-ammonia N (NAN).Flows of NH 3 -N; NAN; and microbial N were calculated based on Calsamiglia et al. (1996) and Bach and Stern (1999), by using the following equations: Where Y ijkl is the response variable, μ is overall mean, L i is the effect of Latin square, P i is the effect of period, F(L) ki is the effect of fermenter (F) within square, T i is the effect of treatment, and E ijkl is the residual error.Square (L), period (P) and fermenter (F) within square were considered as random effects.Orthogonal contrasts were used to test the effects of (1) ADD, the control compared with all treatments with additives (CON vs. MON, CNSE100 and CNSE200); (2) MCN, treatment with monensin compared with those with CNSE (MON vs. CNSE100 and CNSE200); and (3) DOSE, the single dose compared with the double dose of CNSE (CNSE100 vs. CNSE200).Significance was declared at P ≤ 0.05 and a tendency was considered when 0.05 < P ≤ 0.10.

RESULTS AND DISCUSSION
While time course data were collected, the variation pattern was similar among the treatments over time and there was no time by treatment effects observed for any of the parameters, thus all data were presented as either the overall mean (pH) or the 24-h pool values (VFA, lactate and NH 3 -N).

Fermenter pH
Addition of MON, CNSE100, and CNSE200 in the diet had no effect on fermenter pH (Table 2).Previous studies that evaluated the effects of monensin on ruminal pH in dairy cows reported an increase of ruminal pH in transition cows upon subclinical acidosis challenge (Green et al., 1999), while other studies reported the lack of effect in monensin-treated dairy cows fed diets intended to induce acidosis (Mutsvangwa et al., 2002) or no-acidosis challenge (Haïmoud et al., 1995;Ruiz et al., 2001).Starch abundance promotes the proliferation of lactate-producing bacteria; however, in presence of monensin their growth is inhibited (Russell and Hino, 1985;Callaway and Martin, 1997).In Mutsvangwa et al. (2002), ruminal lactate was lower than our study (0.02 mM vs. 0.51 mM), which may explain differences observed in pH between studies.It appears that 5 mM is a threshold that monensin would affect lactate and ruminal pH (Green et al., 1999).
Regarding CNSE, a previous in vitro study, in which CNSE was supplemented to a diet with 41.2% NDF, ruminal pH quadratically decreased (Watanabe et al., 2010), while in another study, in which CNSE was supplemented to a diet with 29.7% NDF, and 29.9% starch, there were no effects on pH (Compton et al., 2023).In our study, we fed a diet with 30.3% NDF, and 26.8% starch (DM basis) and the ruminal pH observed in our CNSE treatments were lower than Watanabe et al. (2010), and similar to Compton et al. (2023).Differences in response in ruminal pH could be attributed to dietary differences among studies, notably in regards to starch and NDF levels.The effects of CNSE on pH warrant further evaluation and a wider range of starch and NDF concentrations.

VFA and Lactate
Data for VFA and lactate concentrations are presented in Table 2. Total concentration of VFA and molar proportions of propionate, valerate, isobutyrate, isovalerate, and acetate: propionate ratio were similar across treatments.
Butyrate molar proportion was greater in CON compared with the rest of the treatments (ADD; P = 0.04) and was lower in MON compared with both CNSE treatments (MCN; P = 0.02).While it is less common to observe a decrease in ruminal butyrate, decreasing butyrate has been reported with the inclusion of monensin (Russell and Strobel, 1989;Shen et al., 2017) and CNSE (Watanabe et al., 2010) in vitro.Similar to monensin, CNSE exhibits selective antimicrobial activity against ruminal gram-positive bacteria that produce H 2 , formate, and butyrate, such as R. flavefaciens, R. albus, Eubacterium ruminantium, and Butyrivibrio fibrisolvens (Watanabe et al., 2010).Therefore, our observations in butyrate were expected; however, not many studies reported effects on butyrate yet probably because of its minor proportion on the total VFA concentration.Of note, there was no effect of DOSE which suggests that butyrate molar proportion is independent of the inclusion level of CNSE.
Acetate molar proportion tended to be lower in the MON treatment compared with the CNSE treatments (MCN; P = 0.07).Similar to our results, Schären et al. (2017) fed a 60% concentrate diet to multiparous Holstein cows with the inclusion of monensin, provided through a controlled released capsule at 335 mg/d and a blend of essential oils administered through pelleted concentrate at 1 g per cow/d starting at d 21 after parturition.The molar proportion of acetate was not affected compared with the control; however, decreased in comparison to the essential oil blend (Schären et al., 2017).The essential oil blend composed of eugenol among other ingredients, which is a phenol-based substance similar to CNSE.Thus, the tendency to decrease acetate in MON compared with CNSE in our study is consistent with previous studies.2 Contrasts: ADD = Control vs. MON, CNSE100, CNSE200; MCN = MON vs. CNSE100, CNSE200; DOSE = CNSE100 vs. CNSE200.
In previous continuous culture fermentation studies where monensin was dosed at 25 ppm (Jenkins et al., 2003), 2.89 μM (Fellner et al., 1997) and 11 g/909 kg DM (Ye et al., 2018), it decreased the acetate: propionate (A:P) ratio by 28.3%, 76.5%, and 17.4% respectively.Contrarily, Mathew et al. (2011) provided a 50:50 forage: concentrate diet and dosed monensin at 12 g/909 kg DM to lactating Holstein cows and reported no effect in A:P ratio.Similar to the in vivo study by Mathew et al. (2011), we observed no effect of the respective contrasts for propionate and A:P.The absence of effect and the inconsistency among the responses can be justified by the diverse mode of action of monensin against gram-positive bacteria (Schären et al., 2017), as well as, by the way that monensin is administered (liquid vs. solid).More specifically, previous studies highlighted that monensin binds to feed and microorganisms when supplemented as solid, while that binding is reduced when it is administered diluted in liquid (Chow et al., 1994).
No effects were observed on lactate concentrations among the orthogonal contrasts tested (Table 2).Monensin has been reported to mitigate the effects of rapid carbohydrate fermentation by selectively inhibiting lactate-producing bacteria (Dennis et al., 1981), and decreasing deamination through the reduction of the obligate amino acid fermenters and other active deaminators (Firkins et al., 2006).Lactate is approximate 10 times stronger than VFAs (Nagaraja and Titgemeyer, 2007), thus increases of its concentration in the rumen is often associated with a reduction in ruminal pH, which in our study was not observed; however, this indicates that in the current study VFAs played a major role in the regulation of fermenter pH.In accordance with our observations, Ye et al. (2018) dosed 11 g/ 909 kg DM, which equates to approximately 0.34 μM MON, and found no effects on lactate concentration.In regards to CNSE, previous in vitro studies observed no changes in ruminal lactate concentrations (Watanabe et al., 2010).Interestingly both Watanabe et al. (2010) and Sarmikasoglou et al. (2024) found that CNSE doses had numerically lower lactate concentration (~36%), when compared with control.

Nitrogen Flows and Metabolism
No effects of treatment were observed on NH 3 -N concentration, or any of the parameters related to N flow or metabolism (Table 3).Other studies have reported no effects on NH 3 -N or N utilization when monensin was added to lactating cows' diets in vitro (Karnati et al., 2009a;Ye et al., 2018) and when technical cashew nutshell liquid (TCNSL) that contains mainly cardanol was fed to multiparous lactating dairy cows (Branco et al., 2015).Additionally, the effects of monensin on NH 3 -N and N utilization appears to be independent of the way that monensin is administered.More specifically, previous studies have either infused monensin diluted in ethanol (Karnati et al., 2009a), or supplemented as feed pellet (Ye et al., 2018), and observed similar responses on NH 3 -N and N utilization.
Overall, our results indicate that CNSE inclusion tends to improve microbial N utilization compared with MON.More specifically, we observed a tendency for microbial N flow (MCN; P = 0.07) and available N efficiency (MCN; P = 0.07) to increase with CNSE inclusion.The amount of NH 3 that is incorporated in microbial N fraction is typically defined by energy availability (Salter et al., 1979;Hoover and Stokes, 1991).Under excess NH 3 concentrations, the degraded N is absorbed through the rumen epithelium as NH 3 and could eventually end up excreted in the urine as urea, which would be a nutritional loss, metabolic burden, and an environmental pollutant (Broderick and Albrecht, 1997).In our study, the greater microbial N flow (MCN; P = 0.07) and available N efficiency (MCN; P = 0.07) indicate better microbial N utilization.A possible explanation would be based on the selective bactericidal activity of CNSE and MON.More specifically, phenolic compounds such as anacardic acid, cardanol, and cardol kill bacteria mostly by biophysical disruption of their membrane (Kubo et al., 2003), whereas MON acts mostly biochemically by forming ion-channels in cell membranes and disrupting the homeostatic mechanism responsible for maintaining intra-and extra-cellular ion concentrations across the cell membrane of microbial cells (Callaway et al., 2003).Additionally, comparing it with other sources of cashew nutshell liquid, that contain less anacardic acid and more cardanol, such as TCNSL, CNSE, which contains more anacardic acid would potentially be more efficient in regards to N utilization.Future studies should focus on comparing the effects of different sources of cashew nutshell liquid in regards to N metabolism.

Counts and Generation Time of Mixed Protozoa
Total protozoal counts in the fermenters were greater in CON compared with the rest of treatments (ADD; P = 0.01), but the daily effluent flow of cells was not affected by treatment (Table 4).Contrary to previous literature findings, in our study we found that 2.5 mM of monensin decreased protozoa population; however, the effects of different fermentation systems and studies within the same systems do not allow for conclusive suggestions.Fermenter protozoal counts were approximately 150-fold lower than those observed in the rumen in vivo (Benchaar et al., 2006) and 10-fold lower than those observed other in vitro studies (Hoover et al., 1976;Karnati et al., 2009a).Previously, Karnati et al. (2009a,b) evaluated the effects of unsaturated fat, monensin, or bromoethanesulfonate on protozoa counts in continuous culture system with a multistage protozoa-retention filter.In our study, we operated our fermenters without a protozoa-retention filter and this limitation resulted in quite low protozoal counts; however, the pattern of effluent flow counts in our study is similar to previous continuous flow studies (Karnati et al., 2009a), indicating a consistent effluent flow and fermentation.
There was a tendency of protozoal generation time to decrease in response to additives (ADD; P = 0.08).Prolonged monensin feeding has been previously shown to structurally change protozoal membrane, reducing protozoal populations, and increasing protozoal generation time (Dennis et al., 1986;Karnati et al., 2009a;Sylvester et al., 2009;Ye et al., 2018).It is important to note the way monensin was administered.Previous studies administered monensin similarly to our study (infused as liquid) and observed no effect on generation time (Karnati et al., 2009a), while others supplemented monensin mixed and processed with other feed com- 2 Contrasts: ADD = CON vs. MON, CNSE100, CNSE200; MCN = MON vs. CNSE100, CNSE200; DOSE = CNSE100 vs. CNSE200. 3 Total N = NH 3 -N + NAN (Bach and Stern, 1999).8 ENU (efficiency of N use) = (g of microbial N/g of available N) × 100 (Bach and Stern, 1999).9 Microbial efficiency = g of microbial N/kg of OM truly digested (Calsamiglia et al., 1996). 10Rumen degraded protein N supply = g/d.
11 Rumen undegraded protein N supply = g/d.For context, protozoa numbers reported herein are, on average, 2-3 orders lower than in the rumen, which is commonly reported in continuous culture experiments (Karnati et al., 2009b).
2 ADD = Control vs. MON, CNSE100, CNSE200; MCN = MON vs. CNSE100, CNSE200; DOSE = CNSE100 vs. CNSE200.ponents into pellets and found an increase in generation time (Ye et al., 2018).Therefore, further research should be centered on assessing the different modes of monensin administration to the rumen protozoal populations.Lastly, CNSE exhibited similar effects to MON on generation time of protozoa; however, the underlying mechanism of CNSE on generation time of protozoa remains to be elucidated.In our study, protozoa types from counts were classified based on visual identification, following the procedure described by Dehority, (1993).We observed greater numbers of Entodinium compared with isotrichids in all treatments (data not shown).Similar to previous studies, the Entodinium remained abundant, while the isotrichids remained lower, in all treatments (Karnati et al., 2009b).No effects were observed on the contrasts tested for either protozoa type.The high growth rate of small Entodinium and their respective abundance in our study suggests that they were able to maintain their population even though they can pass through our liquid phase filter.Thus, our fermenters seemed able to maintain Entodinium populations; however, in lower numbers compared with in vivo (Benchaar et al., 2006), or continuous cultures that used protozoa-retention filter (Karnati et al., 2009a).Lastly, we acknowledge that our identification observations have not been validated with 18S sequencing techniques, thus inferences on the protozoa data reported herein should be interpreted carefully.
Overall, the CNSE showed similar responses to MON on protozoal population, and its mode of action has been speculated to affect metabolic activity of archaea associated with ruminal protozoa (Vogels et al., 1980;Watanabe et al., 2010); however, more research needs to be done to elucidate the underlying mechanism for CNSE.Lastly, considering the overall low levels of protozoal counts, the increase of microbial N in response to CNSE treatments, and that the generation time tended to be lower compared with control; we could assume that the contribution of protozoa to the microbial N flow would be quite low.Consequently, lower contribution from protozoa to microbial N flow, would improve ruminal N metabolism through the enhancement of bacterial protein synthesis (Koenig et al., 2000).

Nutrient Digestibility
No effects of treatment were observed on the nutrient digestibility parameters measured, including OM, CP and NDF (Table 5).Consistent with our findings, previous in vitro reports that tested monensin at 0.34 μM in pelleted form (Ye et al., 2018) and 2.5 μM diluted in ethanol (Karnati et al., 2009a) levels, found no effects on NDF and OM digestibility.However, in our study there was a tendency for NDF digestibility (ADD; P = 0.06) to increase in response to monensin, and CNSE inclusion, compared with control.In addition, Osborne et al. (2004) observed that monensin supplementation at 22 mg/kg of feed (DM basis) to multiparous lactating Holstein cows tended to increase NDF digestibility compared with the control group.Regarding CNSE, Watanabe et al. (2010) found that at all levels of raw CNSE (50-200 μg/ mL) increased apparent DM digestibility compared with control.Additionally, tendencies were observed when TCNSL was evaluated by Branco et al. (2015) where the authors found that supplementation of TCNSL at 30 g/cow per day to multiparous lactating Holstein cows tended to increase NDF digestibility compared with a control group.
The reduced protozoal counts in our study would result in a subsequent lower rate of protozoal predation, which would potentially allow greater proliferation rate by cellulolytic bacteria and greater NDF digestibility.Defaunation of protozoa in continuous culture have increased NDF digestibility (Karnati et al., 2009a).In our study, we did not defaunate the fermenters, and operated the fermenters without a protozoa-retention filter, resulting in quite lower protozoal counts com- 2 ADD = CON vs. MON, CNSE100, CNSE200; MCN = MON vs. CNSE100, CNSE200; DOSE = CNSE100 vs. CNSE200.
pared with in vivo (Benchaar et al., 2006) and those observed in other in vitro studies (Karnati et al., 2009a).Therefore, we could suggest that the reduced protozoal populations in our fermenters, would allow cellulolytic bacteria to proliferate in a greater extent, thus resulting in the tendency observed for NDF degradability.Overall, nutrient digestibility data demonstrate that monensin and CNSE tend to affect nutrient digestibility in a similar way, thus CNSE could be an alternative to ionophores in lactating cows' diets.

CONCLUSIONS
Butyrate concentration was reduced in all treatments compared with CON and also, MON treatment reduced butyrate concentration compared with CNSE treatments.Additionally, MON treatment tended to have lower microbial N-flow and efficiency of N use and no differences in regards to nutrient digestibility compared with CNSE treatments.Our findings demonstrate that from a ruminal fermentation standpoint, it is possible to replace monensin supplemented as liquid with CNSE; however, the lack of effects in nutrient digestibility, and inconsistencies with previous in vivo studies warrant further experiments to be conducted.Further evaluations should be focused on comparing the effects of CNSE and monensin in solid form on ruminal fermentation, as well as, on determining molecular mode of actions of CNSE on microbial cells and the effects of CNSE on animal performance.
Sarmikasoglou et al.: FEED ADDITIVES AND RUMINAL FERMENTATION excess of 15 N in NAN effluent = % atom15 N in NAN effluent sample − % atom 15 N in background.Dietary-N flow and microbial efficiency were determined based onBach and Stern (1999) as follows: Dietary-N flow (g/d) = effluent g of microbial N Rumen undegradable protein and RDP were calculated based on (NRC, 2001) model, by using the following equations: RUP-N flow (g/d) = NAN flow -effluent g of microbial N, and RDP -N supply (g/d) = total N intake -dietary-N flow.Ruminal digestibility of OM, CP, and NDF was determined based on(Soder et al., 2013) as follows: average pH, pool concentrations of VFA, NH 3 -N, and lactate, nutrient digestibility (DM, OM, CP, and NDF), total nitrogen flow, NAN flow, microbial N flow, dietary-N flow, ENU, and microbial efficiency were analyzed with GLIMMIX procedure of SAS as a replicated 4 × 4 Latin square design, with the model:

Table 1 .
Sarmikasoglou et al.: FEED ADDITIVES AND RUMINAL FERMENTATION Ingredient and chemical composition of basal diet 1Expressed as a percent of DM, unless otherwise stated.2Provided (per kg of DM): 955 g of NaCl, 3,500 mg of Zn, 2,000 mg of Fe, 1,800 mg of Mn, 280 mg of Cu, 100 mg of I, and 60 mg of Co.

Table 2 .
Sarmikasoglou et al.: FEED ADDITIVES AND RUMINAL FERMENTATION Effects of cashew nutshell extract (CNSE) and monensin (MON) on in vitro mixed ruminal microorganism fermentation parameters in a dual-flow continuous culture system

Table 3 .
Sarmikasoglou et al.:FEED ADDITIVES AND RUMINAL FERMENTATION Effects of cashew nutshell extract (CNSE) and monensin (MON) on N metabolism in a dual-flow continuous culture system

Table 4 .
Effects cashew nutshell extract (CNSE) and monensin (MON) on counts and generation time of mixed protozoa in a dual-flow continuous culture system

Table 5 .
Sarmikasoglou et al.: FEED ADDITIVES AND RUMINAL FERMENTATION Effects of cashew nutshell extract (CNSE) and monensin (MON) on true nutrient digestibility in a dual-flow continuous culture system