Effects of direct-fed microbial supplementation on performance and immune parameters of lactating dairy cows

We evaluated the effects of supplementing bacterial direct-fed microbial (DFM) on performance, apparent total-tract digestibility, rumen fermentation, and immune parameters of lactating dairy cows. One hundred fourteen multiparous Holstein cows (41 ± 7 DIM) were used in a randomized complete block design with an experiment comprising 14 d of a covariate (pre-experi-mental sample and data collection) and 91 d of an experimental period. Cows were blocked based on energy-corrected milk (ECM) yield during the covariate period and the following treatments were randomly assigned within each block: (1) control (CON), corn silage-based total mixed ration without DFM; (2) PRO-A, basal diet top-dressed with a mixture of Lactobacillus animalis and Propionibacterium freudenreichii at 3 × 10 9 cfu/d; and 3) PRO-B, basal diet top-dressed with a mixture of L. animalis , P. freudenreichii , Bacillus subtilis , and Bacillus licheniformis at 11.8 × 10 9 cfu/d. Milk yield, dry matter intake (DMI), and body weight were measured daily, while milk samples for component analysis were taken on 2 consecutive days of each week of data collection. Feces, urine, rumen, and blood samples were taken during the covariate period, wk 4, 7, 10, and 13 for estimation of digestibility, N-partitioning, rumen fermentation, plasma nutrient status and immune parameters. Treatments had no effect on DMI and milk yield. Fat-corrected milk (3.5% FCM) and milk fat yield were improved with PRO-B, while milk fat percent and feed efficiency (ECM/DMI) tended to increase with PRO-B compared with PRO-A and CON. Crude fat digestibility was greater with PRO-B compared with CON. Feeding CON and PRO-A resulted in higher total volatile fatty acid concentration relative to PRO-B. Percentage of neutrophils tended to be


INTRODUCTION
Direct-fed microbials (DFM) have been explored as feed additives to improve feed efficiency and productivity in lactating dairy cows (Krehbiel et al., 2003).The effects of lactate-producing DFM, specifically Lactobacillus spp. on performance of lactating dairy cows are attributed to its antimicrobial effect (via production of hydrogen peroxide and bacteriocins), sustained lactic acid production, stabilization of ruminal pH, and enhancing the abundance of lactate-utilizers (Krehbiel et al., 2003;Zalán et al., 2005;Seo et al., 2010).Lactate-utilizers such as Propionibacterium spp.have also been explored as DFM, as they utilize lactic acid via pathways that favor propionate production, thereby improving feed efficiency (McDonald et al., 2010) and glucose availability for milk lactose synthesis (Oshio et al., 1987;Francisco et al., 2002).
The potential synergy between lactate-producers and lactate-utilizers can be harnessed by using multispecies DFM that include both groups.The goal is to improve rumen fermentation and enhance animal performance.Lactobacillus acidophilus (known as L. animalis) and Propionibacterium freudenreichii are 2 of the major DFM products used in lactating dairy cow diets with an aim of achieving sustained rumen propionate and intestinal lactate production, increased hepatic glucose synthesis, and overall productivity (Raeth-Knight et al., 2007;Boyd et al., 2011).Several studies have explored multispecies DFM and reported increased yields of milk, milk protein, milk fat, FCM, ECM, and ECM per unit of DMI (Boyd et al., 2011;West and Bernard, 2011;Dickey, 2016).However, the results of using multispecies DFM are equivocal as other studies reported no effect on milk yield and milk composition parameters (Raeth-Knight et al., 2007;Ferraretto and Shaver, 2015).Supplementing a mixture of L. animalis and P. freudenreichii warrants additional studies, including enhancing the synergy within the DFM-mix via inclusion of other bacteria species that could improve nutrient utilization and performance.
Bacillus subtilis and Bacillus licheniformis are sporeforming bacteria that have gained attention because of their potential to influence nutrient utilization, as well as their ability to increase productivity in lactating dairy cows (Qiao et al., 2010;Sun et al., 2013).Qiao et al. (2010) supplemented B. licheniformis to lactating dairy cows and reported improvements in FCM yield, milk protein concentration, and FCM/DMI.Similarly, Sun et al. (2013) supplemented B. subtilis to lactating dairy cows and reported improvements in SCC, milk yield, FCM yield, fat yield, and protein yield.Sun et al. (2010) likewise reported higher serum level of the antibody, IgG and the cytokine, IFN-γ in calves supplemented with B. subtilis.Therefore, both Bacillus spp.hold potential for improving performance if included in the DFM-mixture.
The objective of the current study was to evaluate the effect of 2 multispecies DFM (containing either L. animalis and P. freudenreichii, or L. animalis, P. freudenreichii, B. licheniformis, and B. subtilis) on DMI, milk yield, milk composition, ruminal fermentation, total-tract nutrient digestibility, plasma nutrient status, and immune parameters in lactating dairy cows.We hypothesized that both DFM supplements would increase nutrient utilization, shift ruminal fermentation toward greater propionate synthesis, increase milk yield and production efficiency and influence yield of milk components.In addition, we hypothesized that DFM supplements would improve the immune parameters of the lactating cows by decreasing SCC and increasing immune cell functions.

MATERIALS AND METHODS
All animal care and experimental procedures for this study were approved by the University of Florida Institutional Animal Care and Use Committee (protocol number 201810520).

Cows and Housing
The study was conducted at the University of Florida Dairy Unit (Alachua, FL).One hundred fourteen earlylactating multiparous Holstein cows that were on average (mean ± SD) 41 ± 7 DIM, 42 ± 8 kg milk/d milk yield, 675 ± 68 kg BW were used in the study.Study duration included a 7-d training period, during which all cows were fed the same basal diet fed throughout the study and were trained to use the Calan gate system; a 14-d covariate period for pre-experimental data and sample collection; and a 91-d experimental period, during which treatments were administered and samples and data were collected.The initial 7 d of the experimental period was considered an adaption period and data from this period were not used for statistical analysis because some of the cows refused to consume the entire treatment top-dressed on offered TMR. Cows were housed together in a freestall barn with sandbedded individual stalls.Cows were assigned to individual feeding gates (Calan Broadbent feeding system, American Calan Inc., Northwood, NH) according to the order of enrollment during the training period, when each cow was trained to eat from the assigned bunk.The experimental pens were equipped with 2 rows of fans (1 fan/6 linear meters).Low-pressure nozzles were mounted over the feeding bunk, facing away from the feed and onto the cows.Fans and nozzles were activated whenever ambient temperature reached 21°C.Water was available ad libitum within each pen.Beddings were cleaned twice a day and manure was flushed from pen twice daily via an automated flushing system.

Experimental Design, Treatments, and Feeding
The basal TMR diet (Table 1) was formulated to meet or surpass the energy and protein requirements of lactating dairy cows expected to produce a minimum of 42 kg of milk/d, with 3.50% milk fat, and 3.20% milk protein.Diets were formulated using NDS Professional (RUM&N, Reggio, Emilia, Italy) based on the Cornell Net Carbohydrate and Protein System equations (CNCPS v 6.5).Cows were fed twice daily at 0600 and 1200 h, with 60 and 40% of the total daily allotment, respectively.Feed offered was adjusted daily based on the intake of the preceding 3 d and fed ad libitum with a minimum daily refusal ranging from 5 to 10%.
The experiment utilized a randomized complete block design with 38 blocks, each containing 3 experimental units (cows).Cows were blocked by ECM yield during the covariate period, and 1 of 3 treatments were randomly assigned to each experimental unit within each block.Treatments included: (1) control (CON, with no DFM supplement), (2) a mixture of Lactobacillus animalis LA-51 and Propionibacterium freudenreichii PF-24 (PRO-A; supplemented at 3 × 10 9 cfu/head per day), and (3) a mixture of L. animalis, P. freudenreichii, Bacillus licheniformis, and Bacillus subtilis (PRO-B; supplemented at 11.8 × 10 9 cfu/head per day).Both PRO-A and PRO-B treatments were suspended in lactose powder.The PRO-A and PRO-B treatments were mixed with 100 g of dried molasses and top-dressed on the TMR, whereas the control treatment group received only 100 g of molasses as top-dress.
The amount of feed offered and left over were measured and recorded for individual cows daily.Approximately 200 g of the TMR and refusals from each cow were taken 3 times weekly and dried in a forced-air oven at 55°C for 48 h.Dried samples were composited on weekly basis and stored for later analysis.Corn silage samples were taken 3 times weekly, dried, and composited; while premixed concentrate and individual concentrate ingredients were sampled once every 3 wk, dried, and stored for later analysis.Composited TMR, refusals, corn silage, and concentrate ingredients were ground in Wiley mills (A.H. Thomas Scientific, Philadelphia, PA) to pass through a 1.0-mm sieve and stored for later analysis.

Milk Yield and Milk Components
Cows were milked twice daily at 1000 and 2200 h.Milk yield values were recorded electronically (AfiLab mini laboratory, Afimilk Ltd., Kibbutz Afikim, Israel).Milk samples were collected from morning and evening milkings on 2 d (4 milkings) during every week of the study, starting from the covariate period, and analyzed for milk fat, true protein, lactose, and MUN using a Fourier Transform Spectrometer model FTS 500 (Bentley Instruments Inc., Chaska, MN) at Southeast Milk Dairy Laboratory (Bellevue, FL).Somatic cell count was analyzed using the Bentley flow cytometer model FCM 500 (Bentley Instruments Inc., Chaska, MN).Milk components were adjusted for the milk yield at each milking (morning and evening) and used to calculate the daily component yield and concentrations.The ECM yield was calculated as [(0.3246 × milk yield) + (12.86 × milk fat yield) + (7.04 × milk protein yield)] (NRC, 2001).The 3.5% FCM yield was calculated as [(0.4324 × milk yield) + (16.218 × milk fat yield)] (NRC, 2001).Weekly mean values were calculated from the daily values and used for statistical analysis.

Body Weight, Body Condition, and Energy Balance
Cows were automatically weighed twice daily after each milking on a walk-in electronic scale (Afi-Weigh, S.A.E.Afikim) located in the exit lane of the milking parlor.Daily values of each cow were averaged per week for statistical analysis.Body condition scores were determined once weekly by the same trained evaluator, using a 1 to 5 scale, with 0.25 increments (Ferguson et al., 1994).Energy balance was calculated using daily caloric intake, which was estimated using daily DMI and energy density of the diet after subtracting daily calories required for maintenance, milk production, and growth (NRC, 2001).
where NE c = net energy consumed, NE M = net energy required for maintenance, NE L = net energy output in milk, and NE G = net energy required for growth.

Nutrient Composition and Digestibility
Fecal samples were collected during the covariate period, and each of the 4 sampling periods, on wk 4, 7, 10, and 13 of the experiment.During each of these sampling periods, fecal grabs were collected over 3 consecutive days from the rectum every 3 h, starting at 0500, 0600, and 0700 h on sampling d 1, 2, and 3, respectively.Fecal grab samples were oven-dried at 55°C for 5 d and composited per cow per sampling period.Left-over TMR samples were collected for each cow on each day of fecal sampling and dried at 55°C for 48 h.Composited TMR, refusals, and fecal samples were ground in Wiley mills (A.H.Thomas Scientific, Philadelphia, PA) to pass through a 1.0-mm sieve and stored for later analysis.Ground samples from each cow and sampling periods were incubated in situ for 12-d, in replicates of 6, allotted equally to 2 cannulated cows.Analysis of NDF was performed post-incubation for estimation of indigestible NDF, which was the marker used to estimate the digestibility coefficient (Krizsan et al., 2012).
Offered TMR, refusals, and fecal samples were analyzed for DM (105°C for 12 h), CP, NDF, ether extract, starch, and ash concentrations, which were used to calculate nutrient digestibility.Nitrogen content was measured by rapid combustion using a Macro Elemental N analyzer (AOAC International, 2000;Vario MAX CN, model no. 25.00-5003;Elementar, Hanau, Germany).The N values were multiplied by 6.25 to calculate the CP composition for each sample.Nonsequential analysis of NDF was done with an Ankom 200 Fiber Analyzer (Ankom Technologies, Macedon, NY) using heat-stable α-amylase and sodium sulfite (aNDF;method 2002.04, AOAC International, 2012).Crude fat content was determined by solvent extraction, using an Ankom XT15 extractor (Ankom Technologies, Macedon, NY).Starch was analyzed with near-infrared reflectance spectroscopy (NIRS; DS2500 Feed Analyzer, Foss A/S, Hillerod, Denmark) at the University of Florida Forage Evaluation Support Laboratory (Gainesville, FL).Dietary ash was estimated by overnight combustion samples in a muffle furnace at 550°C.

Urine Sampling and Analysis
Urine samples were collected twice daily, on 2 consecutive days during each of the sampling periods, by mild manual stimulation of the rear udder escutcheon area.Urine samples were collected at 0800 and 1500 h on d 1 and at 0900 and 1600 h on d 2 of sampling.Samples were filtered through 2 layers of cheesecloth, and 20 mL of samples of urine were mixed with 80 mL of 0.036 N sulfuric acid and stored at −20°C for later analysis of urea-N, allantoin, uric acid, creatinine, and urine N. Urine samples were sent to the Forage Evaluation Support Laboratory for ammonium sulfate analysis and urinary N estimation.Concentration of allantoin was determined colorimetrically using the procedure of Chen and Gomes (1992).Uric acid concentration was determined using the Infinity Uric Acid Liquid Stable Reagent from Thermo Fisher Scientific Inc. (Middletown, VA).Urea-N and creatinine were determined using a colorimetric detection kit from Arbor Assays (Ann Arbor, MI).The average daily urine output was estimated using creatinine concentration as a marker by dividing the expected daily creatinine excretion by observed urine concentration of creatinine (Cobianchi et al., 2012).The daily creatinine excretion was estimated from the assumption of 24.05 mg/kg BW of creatinine (Chizzotti et al., 2008).The total purine derivative (PD) concentration was calculated as the sum of the urinary concentration of allantoin and uric acid, while absorbed purines (AP, mmol/d) was calculated as a function of excreted PD, using the equation: PD = 0.85 × AP + 0.512 × BW 0.75 , as described by Cobianchi et al. (2012).

Nitrogen Partitioning
Daily intake of N was estimated from DMI and TMR CP percentage, fecal N was calculated from fecal weight estimate and fecal CP percentage, while milk N was obtained from addition of milk true protein N and MUN.Retained N was calculated as N intake minus the sum of fecal N, urinary N, and milk N. The ruminal microbial N (g/d) was calculated as a function of AP (mmol/d), using the equation: microbial N = (70 × AP)/(0.83× 0.116 × 1,000), where the coefficient 70 amounts to the N content in purines (mg N/mmol), 0.83 assumes 83% digestibility for microbial purines (mmol/d), and 0.116 implies 11.6% purine-N relative to total N in bacteria (Chen and Gomes, 1992).

Rumen Fluid Sampling and Analysis
Rumen fluid samples were collected during the covariate period and each of the 4 sampling periods.Samples were collected 2 h after morning feeding.Approximately 600 mL of rumen fluid were collected using an oro-ruminal probe and a suction pump (Ruminator; profs-products.com, Wittibreut, Bayern, Germany).About 100 mL of initial rumen fluid samples for each cow were discarded to prevent contamination of the samples with saliva.Subsequent samples were observed visually for color and texture, and if there was any suspicion of saliva contamination, the sample was discarded and a fresh rumen fluid sample was collected.Rumen samples were strained through 4 layers of cheesecloth, acidified to a pH of 2 using 50% sulfuric acid solution, and stored at −20°C for later analysis.Acidified samples were centrifuged at 5,400 × g for 20 min at 4°C.The supernatants were used for analysis of VFA and NH 3 -N concentrations.No internal standard was added to the rumen samples for consideration of potential loss of VFA by volatilization; hence, caution should be exercised in interpreting total VFA values.

Blood Sampling and Analysis
Blood samples were taken from the coccygeal vein or artery at 0800 h once during each of the experimental sampling periods in evacuated tubes containing anticoagulant K 2 -EDTA (Vacutainer, Becton Dickinson, Franklin Lakes, NJ).Within 2 h of collection, blood samples were analyzed for total and differential cell counts including white blood cells (WBC) and neutrophil, lymphocyte, monocyte, eosinophil, and basophil (count/μL) using a ProCyte Dx hematology analyzer (IDEXX Laboratories Inc., Westbrook, ME).Another 20 mL of blood was collected into sodium heparin evacuated tubes (Vacutainer).Half of these blood samples (10 mL) were kept on ice after collection and were centrifuged at 2,500 × g for 20 min at 4°C within 30 min of sampling to harvest the plasma fraction.Plasma samples were transferred into snap cap microcentrifuge tubes (Eppendorf AG, Hamburg, Germany) and stored at −20°C for later analysis of glucose, nonesterified fatty acids (NEFA), BUN, and BHB.Plasma glucose concentration was determined using an enzymatic assay kit (Teco Diagnostics, Anaheim CA).Plasma NEFA and BHB concentrations were determined using commercial kits, NEFA-HR (2) and Autokit 3-HB Assay, respectively (Fujifilm Wako Diagnostics U.S.A. Corp., Mountain View, CA).The BUN concentration was estimated using a colorimetric detection kit from Arbor Assays (Ann Arbor, MI).A ~10-mL blood sample in a second heparin tube was kept at ambient temperature and used for analysis of neutrophil phagocytosis and neutrophil oxidative burst activities, and another portion used for measurement of CD62L and CD44 expression in granulocytes, CD4+ αβ T cells, CD8+ αβ T cells, uncommitted αβ T cells, γδ+ T cells, CD21+ B cells, CD14+ monocytes, and NK cells.

Flow Cytometry and Blood Immune Parameter Analysis
Neutrophil phagocytosis and oxidative burst activities were measured from whole blood samples after an Escherichia coli challenge as described by Martinez et al. (2012).From each blood sample, 100 μL of whole blood was pipetted into each of 4 polystyrene roundbottom tubes (12 × 75 mm).Exactly 10 μL of 5 μM dihydrorhodamine 123 (DHR, Sigma-Aldrich Co., St. Louis, MO) solution (prepared by adding 50 μL DHR from a 500 μM stock with 450 μL of PBS) was added to all tubes, to load neutrophils with DHR.Tubes were gently vortexed and incubated at 37°C for 10 min while kept on a rotating nutator (BD, San Jose, CA).After incubation, 1 of 4 tubes was used as a negative control for the analysis of oxidative burst and phagocytosis.Another tube served as positive control for oxidative burst by adding 10 μL of phorbol 12-miristate 13-acetate (PMA, Sigma-Aldrich) from a 20 μg/mL PMA solution (prepared by mixing 10 μL of 1 mg/mL PMA stock with 490 μL of PBS).To estimate phagocytosis, samples in the remaining 2 tubes were cocultured with propidium iodide-labeled E. coli (from a bacterial suspension of 5 × 10 5 cells/μL) at a bacteria: neutrophil ratio of 40:1, based on a premeasured neutrophil concentration of each sample.Phagocytosis and oxidative burst activities were measured using a Becton Dickinson Accuri C6 digital analyzer flow cytometer with a 488nm excitation wavelength (Becton Dickinson, Franklin Lakes, NJ).The flow cytometry output data files were analyzed using FlowJo software (version 10.7.1, Treestar, Palo Alto, CA).Granulocytes, monocytes, and lymphocytes were identified according to their size and granularity after sequentially gating live single cells, as earlier described by Marrero et al. (2021).Using the forward and side scatter plots, subsets were gated to isolate singlets, and then to distinguish granulocytes subset from lymphocytes and monocytes.The mean fluorescence intensity (MFI) of propidium iodide staining was used to estimate the level of propidium iodidestained bacteria phagocytized by granulocytes, while the MFI of oxidized DHR was used to estimate the average oxidative burst intensity among granulocytes.
A 100-μL aliquot of whole blood from sodium heparin treated tubes was transferred to round bottomed FACS tubes, and 2 mL of 1× PBS solution was added to each sample, followed by vortex and centrifugation at 700 × g at 4°C for 3 min.Supernatants were aspirated and cell pellets were loosened by scraping the tubes on the tube rack (Lo Celso et al., 2011).Resulting cell pellets were resuspended in 2.5 mL of phosphate bicarbonatebuffered cell suspension and washing solution (Gey's solution), followed by centrifugation, aspiration of supernatant, and unsticking of cells.The suspension and washing procedures were done twice, followed by the addition of 2 mL 1× PBS, vortexing, centrifugation, aspiration, discard of supernatant, and drying of tube on paper towel.Cells were thereafter incubated with 10 μL rat IgG (#I4131, Sigma-Aldrich, Saint Louis, MO) for 10 min at 4°C, with the aim of blocking fragment crystallizable receptor binding and the consequential false positives.Fluorochrome-conjugated antibodies were added to stain the IgG-incubated cells.One panel included antibodies against: CD62L (BAQ92A, Alexa Fluor 647, #WS0515B-100, Kingfisher), CD44 (IL-A118, FITC, #MCA2433F, Bio-Rad), CD4 (CC30, PerCP, #MCA834GA, Bio-Rad), and CD8 (CC63, RPE, #MCA837PE, Bio-Rad).A second panel contained antibodies against CD21 (CC51, FITC, #MCA5953F, Bio-Rad), TCRδ (GB21A, Alexa Fluor 647, #WSC0578B-100, Kingfisher), and CD14 (TÜK4, PE, #MCA1568PE, Bio-Rad).Propidium iodide (10 μL) was added to samples for exclusion of dead cells during analysis.Samples were vortexed after addition of antibodies and were incubated at 4°C for 30 min.The stained cells were resuspended, vortexed, centrifuged, and washed in 2 mL of PBS 1× for 3 min, at 4°C, after which supernatant were aspirated and discarded.Cells were then resuspended in 500 μL of PBS 1×, vortexed, with measurements done using Becton Dickinson Accuri C6 digital analyzer flow cytometer (Becton Dickinson, Franklin Lakes, NJ).The resulting flow cytometry output data files were analyzed using FlowJo software (version 10.7.1, Treestar, Palo Alto, CA) Granulocytes and lymphocytes were distinguished accordingly to size and granularity, after gating on viable cells and single cells.The percentage of cells positive for each antibody were based on gated cells, as described earlier (Marrero et al., 2021).

Statistical Analysis
The experimental design was a randomized complete block design, with a total of 38 blocks and individual cows serving as experimental units.Data from the first 7 d of the experiment were excluded from statistical analysis for reasons earlier described.One hundred fourteen cows were enrolled in the experiment; however, one cow from the CON group had to be removed prematurely before the first sampling period, due to development of toxic mastitis for unknown reasons.Data from this cow were excluded from statistical analysis.Daily measures for parameters such as milk yield, milk components, BW, and DMI were averaged into weekly data for ANOVA.For parameters such as rumen VFA, rumen NH 3 -N, digestibility, urine parameters, blood, and immune parameters, single measurements taken per experimental unit per sampling period, were used for the ANOVA.Somatic cell scores were estimated from SCC, using the equation SCS = log 10 (SCC/12.5)/log 10 2, before analysis.The normal distribution of data residuals was evaluated using the UNIVARIATE procedure of SAS.The Shapiro-Wilk statistic was used to determine if the residuals were normally distributed (value ≥0.10), otherwise data were log-transformed.Data were analyzed using the GLIMMIX procedure of SAS version 9.4 (SAS Institute Inc., Cary, NC).
The average pre-experimental measures for each experimental unit was used as covariate within the statistical model.Data were analyzed as repeated measures over time using the auto-regressive type 1 [AR(1)] as the time-series variance-covariance structure as this was the model fit that resulted in the smallest corrected Akaike's information criterion, and cow was used as the subject.Below is the linear additive model for this analysis: where µ = overall mean; β i = random effect of block, i = 1 to 38; T j = fixed effect of treatment, j = 1 to 3; W k = fixed effect of time, k = 1 to 12 for DMI, milk, and BW parameters, and k = 1 to 4 for all other parameters; (T × W) jk = treatment × time interaction; CV l is the fixed effect of pre-experimental covariate data; C(T) jm = random effect of cow m nested in treatment j; and ε ijklm = residual random error of experiment.For all data, denominator degree of freedom was estimated using the first order Kenward-Roger option in the MODEL statement.Adherence of residuals to normal distribution and homogeneity of variance were checked after fitting the final model, using the histogram and residual plots in StudentPanel of SAS (version 9.4, SAS Institute Inc.).Post hoc tests were adjusted for pairwise multiple mean comparisons using Bonferroni adjustment.All data are presented as least squares means ± average standard error of the mean.Statistical significance and tendencies were declared at P ≤ 0.05 and 0.05 < P ≤ 0.10, respectively.

Intake and Production Performance
Direct-fed microbial supplementation did not influence DMI of cows (P = 0.84), which averaged 24.0 kg/d (Table 2).Net energy balance tended to increase (P = 0.06) with PRO-A supplement compared with PRO-B and CON (5.7 vs. 4.3 and 4.6 Mcal/d ± 0.6).Fat-corrected milk yield increased (P = 0.05) by 2.1 and 1.9 kg/d for the PRO-B when compared with the CON and PRO-A, respectively (42.3 vs. 40.2and 40.4 kg ± 0.67).Similarly, PRO-B increased (P = 0.05) milk fat yield by 0.11 kg/d compared with CON (1.53 vs. 1.42 kg/d ± 0.03) with no difference with PRO-A (1.44 kg/d).Moreover, PRO-B tended to increase ECM/DMI efficiency (P = 0.06; 1.73 vs. 1.66 and 1.67 ± 0.02) and milk fat percent (P = 0.07; 3.82 vs. 3.60 and 3.65% ± 0.07) compared with CON and PRO-A.Both PRO-A and PRO-B tended to increase MUN compared with CON group (P = 0.08; 13.2 and 13.0 vs. 12.4 mg/dL ± 0.29).Treatments had no effect on milk yield and other measured milk composition pa-rameters, including milk protein and lactose (P ≥ 0.26).Similarly, no differences were observed between treatments for BW gain and BCS (P ≥ 0.17; Table 2).

Urinary Metabolites and Nitrogen Partitioning Study
Supplementation of DFM had no effect on the urinary concentrations of PD and its constituents, allantoin, Mean values in the same row with different superscripts differ according to P-value for treatment effect. 1 Cows were fed a corn silage-based diet, with either no supplement (CON), supplemented at 3 × 109 cfu/cow per day with DFM combination of Lactobacillus animalis and Propionibacterium freudenreichii (PRO-A), or supplemented at 11.8 × 10 9 cfu/cow per day with DFM combination of L. animalis, P. freudenreichii, Bacillus subtilis, and Bacillus licheniformis (PRO-B).

2
T × W = effect of interaction between treatment and week of study.and uric acid (P = 0.24 and P = 0.35, respectively; Table 4).However, cows fed PRO-B had lower (P < 0.01; 301.3 vs. 388.1 ± 22.2 mmol/d) PD excretion and purine absorption compared with CON, while PRO-A cows tended to have lower (P = 0.10; 327 mmol/d) PD excretion and purine absorption compared with CON.
Nitrogen intake and milk N yield were lower for PRO-A when compared with CON (P < 0.01; 541.6 vs. 586.8± 15.2 g/d N intake; and 171.9 vs. 190.5 ± 5.8 g/d milk N yield; Table 5), with no difference between PRO-B (557.2 g/d N intake and 178.0 milk N yield) and other treatments.No treatment effects were observed on fecal N output, urine N output, retained N, and milk N efficiency (P > 0.10).Nonetheless, PRO-B supplemented group had lower (P < 0.01) daily micro-bial N yield relative to CON (198.4 vs. 272.1 ± 18.9 g/d) while being similar to PRO-A, which tended to be lower compared with the CON (P = 0.09; 220.2 g/d).

Ruminal Fermentation
Direct fed microbial supplementation had no effect (P > on rumen ammonia N and rumen pH (Table 6); however, PRO-B supplementation decreased (P = 0.05; 82.5 vs. 96.4 and 98.0 ± 1.05 mM) total VFA concentration compared with CON and PRO-A.Similarly, PRO-A supplementation increased (P = 0.05) molar proportion of valeric acid in the rumen compared with CON (8.44 vs. 6.95 mol/100 mol), with no difference to PRO-B (8.02 mol/100 mol).A treatment × week interaction effect was observed (P = 0.02) for acetate-to-propionate (A:P) ratio.On wk 3 of the study, PRO-A tended to have a lower A:P T × W = effect of interaction between treatment and week of study.Mean values in the same row with different superscripts differ according to P-value for treatment effect. 1 Cows were fed a corn silage-based diet, with either no supplement (CON), supplemented at 3 × 10 9 cfu/cow per day with DFM combination of Lactobacillus animalis and Propionibacterium freudenreichii (PRO-A), or supplemented at 11.8 × 10 9 cfu/cow per day with DFM combination of L. animalis, P. freudenreichii, Bacillus subtilis, and Bacillus licheniformis (PRO-B).

2
T × W = effect of interaction between treatment and week of study.
ratio compared with CON (P = 0.06; 1.98 vs. 2.18), while on wk 6 of the study, PRO-A tended to have a lower A:P ratio compared with PRO-B (P = 0.06; 1.92 vs. 2.13).Treatments showed no effects on other parameters including molar proportion of acetate, propionate, and butyrate along with A:P ratio (P > 0.10).

Plasma Metabolites and Blood Cell Count
Treatments had no effect (P > 0.10) on plasma concentrations of NEFA, BHB, glucose, and BUN (Table 7).Treatments had no effect on counts of erythrocytes (RBC), the proportion of RBC out of total blood cell (hematocrit), and the average volume of RBC (mean corpuscular volume; P > 0.10; Table 8).Similarly, no treatment effects (P > 0.10) were observed on hemoglobin parameters including hemoglobin concentration, mean corpuscular hemoglobin count and concentration.Treatments had no effect (P > 0.10) on leukocyte concentration (WBC), as well as proportion of the different WBC, including lymphocytes, monocytes, eosinophils, and basophils; however, PRO-A supplementation decreased the proportion of neutrophils relative to CON (P = 0.03; 21.6 vs. 34.2± 3.30%), with no difference between both treatments in comparison to PRO-B (28.2%).No differences were observed between treatments on platelet counts and platelets distribution width (PDW; P > 0.10).T × W = effect of interaction between treatment and week of study. 3 Microbial N = (70 × absorbed purines) ÷ (0.83 × 0.116 × 1,000).Mean values in the same row with different superscripts differ according to P-value for treatment effect. 1 Cows were fed a corn silage-based diet, with either no supplement (CON), supplemented at 3 × 10 9 cfu/cow per day with DFM combination of Lactobacillus animalis and Propionibacterium freudenreichii (PRO-A), or supplemented at 11.8 × 10 9 cfu/cow per day with DFM combination of L. animalis, P. freudenreichii, Bacillus subtilis, and Bacillus licheniformis (PRO-B).
2 Rumen fluid was sampled on wk 4, 7, 10, and 13 of the study. 3T × W = effect of interaction between treatment and week of study.

Immune Parameters
The PRO-B treatment tended (P = 0.08) to increase the expression of the proteoglycan, CD44 in granulocytes when compared with CON (2,554.6 vs. 2,349.3 ± 107.7 MFI), as evident from MFI (Table 9).No difference was however observed between PRO-A (2,464.9MFI) and other treatments.No differences among treatments were observed (P = 0.39) in expression of the adhesion molecule, CD62L in granulocytes.No differences were observed in the percentage of T-helper lymphocytes (CD4+ cells; P = 0.62), as well as their expression of CD44 (P = 0.24), CD62L (P = 0.59) and T-cytotoxic/suppressor lymphocytes (CD8+ cells; P = 0.12).However, PRO-B supplementation increased (P = 0.05) expression of CD62L on CD8+ cells rela- T × W = effect of interaction between treatment and week of study.Mean values in the same row with different superscripts differ according to P-value for treatment effect. 1 Cows were fed a corn silage-based diet, with either no supplement (CON), supplemented at 3 × 10 9 cfu/cow per day with DFM combination of Lactobacillus animalis and Propionibacterium freudenreichii (PRO-A), or supplemented at 11.8 × 10 9 cfu/cow per day with DFM combination of L. animalis, P. freudenreichii, Bacillus subtilis, and Bacillus licheniformis (PRO-B).
2 T × W = effect of interaction between treatment and week of study.
3 RBC = red blood cells (million red blood cells per microliter or M/μL). 4MCV = mean corpuscular volume (fL or 10 −15 L), which is the average volume of a red blood corpuscle (cell). 5MCH = mean corpuscular hemoglobin (pg or 10 −12 grams), which is the average weight of hemoglobin (Hb) in the RBC. 6MCHC = mean corpuscular hemoglobin concentration (g/dL), which is the average concentration of Hb in the RBC volume.

Intake and Production Performance
Treatments had no effect on DMI in agreement with previous studies supplementing a mixture of Lactobacil-lus acidophilus and Propionibacterium freudenreichii to near-mid-lactation Holstein cows fed a TMR of 60.8% forage (47.9% corn silage, 12.9% alfalfa hay; using L. acidophilus LA747 and P. freudenreichii PF24; Raeth-Knight et al., 2007) and to mid-lactation Holstein cows fed a TMR of 39.1% forage (31.6% corn silage, 7.5% alfalfa hay; using L. acidophilus NP51 or NP45 and P. freudenreichii NP24; West and Bernard, 2011).However, other studies observed a tendency for decreased intake with Lactobacillus spp.and P. freudenreichii supplementation to mid-lactation Holstein cows fed a TMR of 54.2% forage (27.1% corn silage and 27.1% alfalfa silage; using L. acidophilus NP51 and P. freudenreichii NP24; Ferraretto and Shaver, 2015) and to mid-lactation Holstein cows fed a TMR of 69.0% forage (51.5% corn silage, 8.90% alfalfa silage, and 8.58% alfalfa hay; Lawrence et al., 2021).While the stage of lactation and proportion of forage in diet could explain the disparity in DMI between studies, neither of these factors were distinct for the studies under comparison and therefore cannot explain the discrepancy.Lawrence et al. (2021)  Mean values in the same row with different superscripts differ according to P-value for treatment effect. 1 Cows were fed a corn silage-based diet, with either no supplement (CON), supplemented at 3 × 10 9 cfu/cow per day with probiotic combination of Lactobacillus animalis and Propionibacterium freudenreichii (PRO-A), or supplemented at 11.8 × 10 9 cfu/cow per day with probiotic combination of Lactobacillus animalis, Propionibacterium freudenreichii, Bacillus subtilis, and B. licheniformis (PRO-B).

2
T × W = effect of interaction between treatment and week of study.DFM, which may have lowered intake by increasing ruminal propionate.Similarly, Ferraretto and Shaver (2015) speculated greater ruminal propionate in response to supplemented DFM for lower intake; however, rumen fermentation was not measured this study.Lack of treatment effects on starch digestion and ruminal propionate may have contributed to unchanged DMI in the present and previous studies (Raeth-Knight et al., 2007;West and Bernard, 2011).
A positive correlation was reported between milk yield and parameters such as DMI, molar proportion of butyrate and propionate, while little to no relationship was reported between milk yield and either total VFA or acetate concentration (Seymour et al., 2005).No treatment effect was observed on milk yield, which could be attributed to absence of differences in DMI and the proportions of propionate and butyrate in the current study.Lack of milk yield responses in the current study agreed with previous studies supplementing a combination of Lactobacillus spp.and P. freudenreichii (Raeth-Knight et al., 2007;Ferraretto and Shaver, 2015;Lawrence et al., 2021).However, other studies have reported improved milk yield with DFM supplementation.West and Bernard (2011) observed a tendency for greater milk yield in DFM-supplemented groups, while Boyd et al. (2011) reported increased milk yield with a mixture of L. acidophilus and P. freudenreichii supplemented to mid-lactation cows fed a TMR of 48% forage (20.2% corn silage, 15.5% alfalfa hay, and 12.3% ryegrass silage).Boyd et al. (2011) and West and Bernard (2011) both supplemented DFM to lactating cows and reported an increase in milk yield in cows fed TMR diets with relatively higher proportions of concentrate (52 and 60.9%, respectively), as opposed to other studies which fed less than 50% concentrate.This suggests an interaction between DFM supplementation and dietary readily fermentable energy supply.Further studies that assess the effect of readily fermentable energy on the effectiveness of multispecies DFM in dairy cow diets are needed.
Supplementing a mixture of L. animalis, P. freudenreichii, B. subtilis, and B. licheniformis increased milk fat yield and concentration, which may be attributed to increased crude fat digestibility, since no differences were observed in fiber digestibility and body fat mobilization.The effects of PRO-B on improved crude fat digestibility may have increased absorption and availability of fatty acids for milk fat synthesis or secretion, while the lack of this effect on milk fat with PRO-A might be attributable to absence of changes in digestibility.Increased 3.5% FCM yield due to PRO-B supplementation is attributed to the greater milk fat yield observed with this treatment.A similar response on 3.5% FCM was observed earlier by supplement-ing a mixture of L. acidophilus and P. freudenreichii; however, no Bacillus spp.were added in the DFM mix (West and Bernard, 2011).In contrast, no effects were observed with DFM supplementation on milk fat yield in other studies (Raeth-Knight et al., 2007;Ferraretto and Shaver, 2015).Sun et al. (2013) supplemented a strain of Bacillus subtilis to early-lactation Holstein cows and reported an increase in milk fat yield and 4% FCM yield, with no difference in milk fat percentage.Qiao et al. (2010) compared the effects of B. subtilis and B. licheniformis on dairy cows and reported that only B. licheniformis improved 4% FCM, with no effect on milk fat percentage.Another plausible explanation for increased milk fat yield is increase rumen concentration of acetate, which has a positive correlation with milk fat yield (Seymour et al., 2005).Qiao et al. (2010) reported a greater acetate concentration for the group supplemented with B. licheniformis, which might explain the higher milk fat yield reported by these authors.In the current study however, PRO-B had no effect on molar proportion of acetate, and had lower total VFA concentration relative to CON and PRO-A, which suggests that increased crude fat digestibility is a more likely explanation for the increased milk fat yield observed for PRO-B.Based on the increased crude fat digestibility observed for PRO-B and not for PRO-A, and given the fact that Bacillus spp.produces exogenous lipases (Gupta et al., 2004;Bracco et al., 2020), we speculate that Bacillus spp.may have contributed to greater crude fat digestibility and concomitant increase in 3.5% FCM.
Increased feed efficiency is one of the major performance improvements attributed to DFM supplementation, primarily with lactate-utilizing or lactate-producing bacteria (Krehbiel et al., 2003).Feed efficiency increases either when performance increases given the same level of feed intake, or when feed intake decreases while maintaining the same level of performance.Either approach to improve feed efficiency could be due to increased ruminal or postruminal nutrient utilization, or decreased energy expenditure due to improved immune status.Feed efficiency (ECM/DMI) tended to be greater along with greater crude fat digestibility with PRO-B supplemented cows.However, increased crude fat digestibility cannot solely explain the increased feed efficiency because crude fat only constitutes 7% of the experimental diet.We speculate that higher DM and NDF digestibility, and postabsorptive metabolism may have contributed to greater feed efficiency.While postabsorptive metabolism was not measured, only numerical responses were observed on DM and NDF digestibility.Feed efficiency was measured daily while other response variables such as nutrient digestibility, rumen fermentation, and blood profile were measured every 3 wk.As such, it is possible that because of sampling schedule we may have missed treatment differences in these parameters.In addition, changes in ruminal microbial population due to combined effects of Lactobacillus spp.and Propionibacterium spp.(West and Bernard, 2011) and increased nutrient utilization toward more efficient metabolic pathways may contribute toward greater energy efficiency (Krehbiel et al., 2003).
The MUN values are within the optimal target range of 10 to 16 mg/dL expected from cows fed according to NRC recommendations (Jonker et al., 1998).The tendency for increased MUN for both DFM supplemented groups may suggest less utilization of N in the rumen for microbial protein synthesis, which Nousiainen et al. (2004) suggested might imply lower dietary N utilization efficiency.Although there was no indication of SARA in any of the treatment groups based on the ruminal pH, PRO-A tended to reduce microbial N, while PRO-B decreased microbial N in comparison to CON.Both effects imply less microbial proliferation and a decrease in NH 3 capture by microbes, which may explain the increased MUN for both groups.

Digestibility and Rumen Fermentation
Direct-fed microbial supplementation is expected to optimize ruminal fermentation and subsequently improve total-tract nutrient digestibility (Qiao et al., 2010).While the modes of action of DFM are quite variable depending upon bacterial species used in the mixture, the effects on nutrient digestibility are mediated via increased fiber-degrading bacterial proliferation.Some studies have however reported increased NDF digestibility with DFM composed of either B. licheniformis or a combination of L. acidophilus and P. freudenreichii (Qiao et al., 2010;Boyd et al., 2011).Philippeau et al. (2017) reported increased cellulase activities in cows supplemented with multispecies DFM composed of either Propionibacterium spp.and L. plantarum or Propionibacterium spp.and L. rhamnosus, although the authors observed no effect on NDF digestibility.However, we are still lacking adequate understanding of how dietary inclusion of bacterial DFM influences nutrient digestibility.In the current study, only crude fat digestibility was increased by PRO-B supplementation.This absence of effect on most digestibility parameters might be due to the composition of the experimental diet used in the current study or affinity of the strains of DFM to improve a nutrient component but not others.There are no other studies examining the effects of a DFM mixture of Lactobacillus spp., Propionibacterium spp., and Bacillus spp. on crude fat digestibility.Bacillus is a major extracellular lipase-producing bacterial genus (Gupta et al., 2004;Bracco et al., 2020).Several strains of P. freudenreichii were implicated in production of lipases and esterases, although their enzymes were inactive on fatty acids with more than 4 C atoms in the chain (Dupuis et al., 1993).The presence of lipase activity of B. subtilis and B. licheniformis may have increased crude fat digestibility in the current study.The diet used in this study had 7.27% crude fat, with 77.4% of this crude fat originating from concentrate fraction.Unsaturated fatty acid-rich triglycerides are the most abundant form of lipid in the concentrate fraction of diets, and are highly susceptible to microbial lipase induced hydrolysis in the rumen (Dawson et al., 1977;Palmquist and Jenkins, 1980;Lourenço et al., 2010).Such increase in lipolysis could increase the concentration of free unsaturated fatty acids in the rumen, which could negatively affect microbial proliferation and rumen fermentation.We speculate this might be responsible for the lower total VFA observed for PRO-B supplemented cows in the current study.The lower total VFA, lower concentrations of AP and lower microbial N observed in the current study, may be indicative of reduced microbial proliferation and ruminal fermentation.Further studies should investigate the effect of multispecies DFM containing Bacillus spp. on fat utilization in ruminants as they may allow higher fats to be fed, such as in diets for early-lactation cows.Unlike the reduction in total VFA by PRO-B in the current study, Qiao et al. (2010) observed no effect of B. subtilis on total VFA, whereas greater VFA concentration was reported when B. licheniformis was supplemented, which may have been due to the higher NDF digestibility in the rumen of the supplemented group.The exclusive use of the same strains of B. licheniformis and B. subtilis as used in the current study resulted in increased digestibility of NDF, DM, and starch when different forage types and grains were used in an in vitro rumen fermentation study (Pan et al., 2022).Lack of PRO-A effects on total VFA concentration is consistent with reports of Raeth-Knight et al. (2007) and Lawrence et al. (2021) who supplemented a mixture of Lactobacillus spp.and Propionibacterium spp.
The proportion of acetate observed in the current study was relatively lower than previous studies probably due to the low NDF digestibility observed for all treatments.Previous studies supplementing Lactobacillus spp.and P. freudenreichii observed no effect on acetate proportion (Raeth-Knight et al., 2007;Lawrence et al., 2021) while propionate proportion tended to increase with the DFM mixture (Raeth-Knight et al., 2007).Sun et al. (2013)  propionate ratio, suggesting a beneficial effect of these strains on rumen fermentation traits.

Urine and Nitrogen Metabolism
The lower N intake observed for PRO-A cows compared with CON was surprising given that no differences were observed in DMI, and same experimental diet were fed to all cows.The lower N intake observed for PRO-A might be due to feed sorting or sampling error while collecting TMR and refusal samples.The lower milk N also observed for PRO-A might be a consequence of the lower N intake.Lower microbial N observed with PRO-B may be attributed to lower microbial proliferation, which subsequently led to lower total VFA as observed for this group.
Urine output was estimated using creatinine concentration from spot urine samples collected twice each day for 2 d. Lee et al. (2019) validated creatinine as marker and optimized procedure for spot sampling for accurately estimating urine volume.No significant difference was observed on average creatinine concentration, and urine volume between total urine collection and spot urine samples collected at 12, 6, and 4 time points over a 3-d period (Lee et al., 2019), suggesting that spot sampling schedule used in the present study can successfully identify the differences in urine outputs influenced by dietary treatments.The lower PD excretion and AP concentration observed for PRO-B in this study, suggests that DFM may have influenced ruminal bacteria N turnover, and perhaps reduced microbial proliferation.Nevertheless, milk protein percentage or yield was not affected.

Plasma and Immune Parameters
Concentration of BHB and NEFA were within normal range throughout the study.With positive energy balance and increasing DIM, NEFA concentration decreased for all treatment groups.The positive energy balance was observed for all groups and suggests that body fat mobilization did not contribute toward greater milk fat yield in the present study.The absence of treatment effect on BHB, NEFA, and glucose might be attributed to the use of mid-lactation dairy cows for this study and the use of high energy density of the experimental diet resulting in limited potential metabolic stresses and subsequently low body fat mobilization.The lack of effect of DFM supplementation on BUN agrees with findings of Boyd et al. (2011), who observed no DFM effect on serum urea N; however, West and Bernard (2011) reported lower serum urea N with a mixture of Lactobacillus spp.and Propionibacterium spp.Variable treatment responses on BUN may be attributed to differences in CP levels of the experimental diets.Although Boyd et al. (2011) and the present study formulated diets for ~17.5% CP, West and Bernard (2011) fed 20.3% CP.We can speculate that probiotics lower serum urea N and improve N efficiency in diets with high CP levels where N efficiency may not be optimal.However, in the present study, margins for improvement in N efficiency may be minimal because of lower CP concentrations in the experimental diets; however, we are still lacking studies validating this speculation.
Neutrophils are the first recruited immune cell to sites of infection, playing an important role in initiating an innate inflammatory immune response (Ezzat Alnakip et al., 2014;Alhussien et al., 2015).An increase in percentage of circulating neutrophils could imply activation of immune response, especially if accompanied by an increase in neutrophil function and expression of relevant markers.Although treatment differences were observed in the percentage of circulating neutrophils, no differences were observed in the viability or effector functions of these neutrophils as measured by oxidative burst and phagocytosis activities, as well as the mean fluorescence intensity of these activities.Furthermore, the expression of CD44 on granulocytes tended to be greater in the PRO-B group compared with CON, which might imply lesser granulocytes activation in CON cows compared with PRO-B cow, which is opposed to what could be implied from considering neutrophil percentage alone.Though the CON and PRO-B were similar in percentage neutrophil, the higher activation of neutrophils in PRO-B cows as evident from CD44 expression makes the effect of PRO-B more desirable.Regulation of CD62L has been said to control the trafficking of lymphocytes to and from peripheral lymph nodes, with the shedding of CD62L from cell membrane of T cells signifying their activation (Yang et al., 2011).CD62L levels on CD8+ T cells were lower in the PRO-A CD8+ T cells compared with those of PRO-B, indicating higher shedding of this adhesion molecule from the T cells.This may indicate higher activation of the CD8+ T cells from the PRO-A group relative to PRO-B cows, which may indicate greater immune responses in the PRO-A-supplemented group.In ruminants, the γδ+ T cells have been described to have potential for cytotoxic activity, respond to mycobacterial protein, and having roles of being regulatory T cell that produce the antiinflammatory cytokines, IL-10 and TGF-β (Baldwin et al., 2021).Result from the current study showed a significant interaction between treatment and week for the proportion of γδ+ T cells.The decrease or tendency to decrease the proportion of γδ+ T cells at one or more time points in the supplemented groups suggests that DFM may have affected the immune system of supplemented cows in some way.The effect of PRO-A on γδ+ T cells was more prolonged with a consistent decrease or tendency for decrease of these cells as observed for wk 6 and 12, and wk 3 and 9, respectively.On the other hand, PRO-B showed tendency to lower proportion of γδ+ T cells compared with CON only at wk 6.Further studies are however needed to fully understand the implications of these findings on cow health and performance.

CONCLUSIONS
Supplementation of the Bacillus containing DFM, PRO-B, improved dietary crude fat digestibility in lactating Holstein dairy cows, with a consequential increase in milk fat and FCM yields and a tendency for greater ECM/DMI efficiency.The results observed in this study validates our hypothesis that a multispecies DFM that includes Bacilli has potential to alter nutrient utilization and increase performance of lactating dairy cows, in the absence of any discernible effects on rumen fermentation, blood profile, and N efficiency.In addition, the effects of DFM on immune parameters is inconclusive despite desirable effects with PRO-B on expression of CD44 in granulocytes and for PRO-A in expression of CD62L in CD8+ T cells.
Oyebade et al.: DIRECT-FED MICROBIAL SUPPLEMENT IN DAIRY COW RATION
Oyebade et al.: DIRECT-FED MICROBIAL SUPPLEMENT IN DAIRY COW RATION Oyebade et al.: DIRECT-FED MICROBIAL SUPPLEMENT IN DAIRY COW RATION Oyebade et al.: DIRECT-FED MICROBIAL SUPPLEMENT IN DAIRY COW RATION
Oyebade et al.: DIRECT-FED MICROBIAL SUPPLEMENT IN DAIRY COW RATION

Table 2 .
Oyebade et al.: DIRECT-FED MICROBIAL SUPPLEMENT IN DAIRY COW RATION Effect of direct-fed microbial (DFM) supplementation on intake and performance of lactating dairy cows

Table 3 .
Oyebade et al.: DIRECT-FED MICROBIAL SUPPLEMENT IN DAIRY COW RATION Effect of direct-fed microbial (DFM) supplementation on apparent total-tract digestibility in lactating dairy cows 2

Table 4 .
Effect of direct-fed microbial (DFM) supplementation on urinary metabolites in lactating dairy cows

Table 5 .
Oyebade et al.: DIRECT-FED MICROBIAL SUPPLEMENT IN DAIRY COW RATION Effect of direct-fed microbial (DFM) supplementation on nitrogen metabolism of lactating dairy cows 2

Table 6 .
Effect of direct-fed microbial (DFM) supplementation on rumen fermentation in lactating dairy cows

Table 7 .
Oyebade et al.: DIRECT-FED MICROBIAL SUPPLEMENT IN DAIRY COW RATION Effect of direct-fed microbial (DFM) supplements on plasma metabolites in Holstein cows

Table 8 .
Effects of direct-fed microbial (DFM) supplementation on hematological parameters in lactating dairy cows

Table 9 .
observed greater starch digestion with Oyebade et al.: DIRECT-FED MICROBIAL SUPPLEMENT IN DAIRY COW RATION Effects of direct-fed microbial supplementation on blood cells phenotypes and immune parameters in lactating dairy cows reported lower acetate and increased propionate proportion with B. subtilis, while Qiao et al. (2010) reported no effect on acetate and propionate with B. subtilis but a greater acetate proportion with B. licheniformis.Recently, Dias et al. (2022) reported that supplementation of B. licheniformis and B. subtilis to feedlot cattle reduced the acetate: