The experiment was conducted in the Experimental Farm of the Centro Superior de Educação do Oeste (FECEO) dairy cattle sector in Guatambu-SC. Latitude (S) G: 27 M: 9 S: 9.52, Longitude (W) G: 52 M: 47 S: 14.22, humid Subtropical climate. The average temperature during the experimental period was 17.7˚C, and the relative humidity was 72%, as shown in Figure S1 , demonstrating that the temperature was within the required parameters, providing thermal comfort conditions. This work was submitted and approved by the Ethics Committee for the use of animals at UDESC under protocol number 8303290323, as well as with the rules issued by the National Council for Control of Animal Experimentation (CONCEA/Brazil).

Fourteen Jersey breed primiparous cows were used, with an average of 130 ± 3.2 DL and an average weight of 394.5 ± 14.7 kg. The cows were under a Compost Barn type of confinement regime with space per animal of 14.4 m² on a sawdust bed, availability of MX300 collective tilting drinking fountains (MX do Brasil®) with a capacity of 300 liters, in addition to an Eco fan system. Fan MX 160 (MX do Brazil®) controls the warehouse’s internal microclimate. Attached to the shed is the feeding lane, with a drinking fountain milking room used for a milking set of robotic systems (DeLaval VMS^TM V300). The milking system was a guided flow type, and the cows had free access to milking only two times during the day, with morning milking starting at 5:00 am and afternoon milking at 3:00 pm.

Our study used green powder denominate of BC, a commercial product from NutriQuest (Biocholine Powder®); the guaranteed levels expressed in the product description show levels above 32 g/kg of phosphatidylcholine. The additive is produced from the extract of selected plants with high choline content in the esterified form Azadirachta indica, Citrullus colocynthis, Trachyspermum ammi, and Achyranthes aspera [11] . This choline source is not hygroscopic, does not convert into trimethylamine, and does not accelerate the destruction of vitamins in the premix. It also helps to use the choline present in feeds used in the diet. A sample of this additive was used to quantify choline in BC using high-performance thin-layer chromatography, as described by Kupke and Zeugner [12] . The colorimetry method was used to determine tannins, phenolic, and flavonoids. The amount of tannins was measured by colorimetry, following method 52, Compêndio Brasileiro de Alimentação Animal, 2013, and of flavonoids was measured following the methodology described by Dazuk et al. [13] .

The experimental period was 28 days, with 14 initial days of adaptation and 14 days of data and sample collection. The animals with the same sample number were randomly divided into two groups based on body weight and average milk production over the last seven days; in order to have groups with similar characteristics for these variables. One group was defined as control (CONT—not-additive); and another group received the inclusion of BC in their diet at a dose of 10 g per animal per day, as suggested by Nunes et al. [1] , which was already included in the concentrate during its production.

The feed was provided individually at three different times during the day (6:00, 11:00, and 16:00, with each session lasting 2 hours), entirely delivered at the feeding alley, where the animals were restrained by a headlock and had access to water. The concentrate supplied by the robot during milking, limited to 2 kg per animal per day, was defined as a “milking diet”.

The total diet was formulated according to the nutritional needs required for each animal [14] , formulated based on 395 kg of body weight, for a production of 22 kg/animal/day of milk, DL of 130, using corn silage-based feeds, Tifton 85 hay and corn-based concentrate, cornmeal, soybean bran, wheat bran, soybean hulls, livestock urea, sodium bicarbonate, adsorbent (Milk Sacc®) and mineral core (Bovigold®). Ingredients (corn silage, hay, and concentrate) were mixed and made available to the animals, forming a partial mixed ration (PMR). The cows also received commercial pelleted concentrate (Cooper Alfa) provided during milking; being limited to 2 kg/cow/day (1 kg morning milking; 1 kg afternoon milking). Table 1 contains data on the chemical composition of feeds and PMR.

Note 1: The ingredients present in the concentrate were: ground corn (51.8%), soybean meal (32.14%), soybean hulls (5.24%), wheat bran (4.76%), urea (0.6%), sodium bicarbonate (1.67%), adsorbent (0.8%) and mineral core (3%). The guarantee levels of the mineral and vitamin core are Ca min. 231.40 max. 250 g; P min 40; S min 20 g; Mg min. 25 g; K min. 10 g; At min. 70 g; with min. 15 mg; Ass. min. 15 mg; Cr min. 20 mg; I. min. 40 mg; Mg. Min 2000 mg; min. 22 mg; Zn. Min 2850 mg. Vit A. 350,000 IU; Vit D3 min. 100,000 IU; Vit. E. 2000 IU.

Blood samples were collected on days 1, 14, 21, and 28 for blood count and metabolic biochemistry using needles and vacuolated tubes; tubes without anticoagulants were used for biochemical analyses and with anticoagulant (EDTA) for hematologic analyses. The tubes were kept refrigerated at 10˚C until arrival at the laboratory (box with styrofoam and ice). Subsequently, the blood was centrifuged at 7000 rpm for 10 minutes to separate the serum. Then, a serum fraction was stored in 1.8-ml microtubes (Eppendorf) and frozen (−15˚C) for subsequent analysis. During random moments of the experiment, hay, silage, concentrate, and total PMR were collected from the control and treatment groups, and the material was used to form a pool that represents the experimental period for subsequent chemical analysis of the feed.

Individual milk analyses were collected from each animal on days 1, 14, 21, 25, and 28, using the automatic collection system (DeLaval VMS^TM V300), stored at 10˚C in styrofoam boxes with ice until arrival at the laboratory, and subsequently, analyses of the proximate composition of the milk were carried out; the remainder was frozen in Eppendorf for analysis of the milk’s fatty acid profile.

In the robotic milking system (DeLaval VMS^TM V300), individual production data for each animal were obtained during the experimental period, and data on Somatic cell counts (SCC), body condition score (BCS), and individual pelleted concentrate intake. Milk production corrected for 4% fat (4% FCM) was estimated by the equation proposed by the NRC [14] : FCM = 0.4 × (kg of milk produced) + 0.15 × (% fat) × (kg of milk produced).

On the last day of the experiment, rumen fluid was collected using a silicone esophageal probe attached to a vacuum system. The pH of the collected material was measured within 1 minute after collection, and the material was then filtered using double gases. The rumen fluid was stored in microtubes and frozen at –80˚C until analysis.

To conduct chemical composition analyses of the feeds used, homogeneous samples of the total diet of both experimental groups were collected. These samples were pre-dried in a forced air oven at 55˚C for 72 hours, removed from the oven, and weighed to determine the partial DM content, followed by grinding in a Wiley-type mill (Marconi, model: MA340), using a sieve of 1-mm mesh. The pre-dried and ground samples were heated at 105˚C to obtain the DM and the mineral material in a muffle furnace at 600˚C [15] . Crude protein (CP) was determined using the micro-Kjeldahl method (Method 984.13, AOAC, 1997). To determine the neutral detergent fiber (NDF) content, the samples were placed in polyester bags (Komarek, 1993) and treated with a neutral detergent solution in an ANKOM 200® fiber analyzer at 100˚C for 1 hour. Acid detergent fiber (ADF) concentrations were determined according to AOAC (1997, method 973.18). The ether extract portion of the samples was quantified using an automatic SER158 Fat Extractor (VELP® Scientifica).

Ten fecal samples were collected during the experimental period of three days (d 22, 23, 24, 25, and 26), with the material collected directly from the cow’s rectal ampulla. The material was frozen, and at the end of the experiment, a pool of all collections in that period was formed with 500 g. The material was dried in a forced circulation oven at 56˚C for 72 h. It was then ground in a knife mill, removed from the oven, weighed to determine the partial DM content, and ground in a Wiley-type mill (Marconi, model: MA340) using a 2 mm mesh sieve. The pre-dried and ground samples were heated at 105˚C to obtain the MS and the mineral material in a muffle furnace at 600˚C [15] .

Samples were analyzed according to AOAC [16] : DM, method 930.15; CP, method 976.05; ether extract (EE), method 920.39 and ash, method 942.05. The concentrations of NDF and ADF were determined according to Van Soest et al. [17] methodology without adding sodium sulfite or alpha-amylase.

We used indigestible neutral detergent fiber (iNDF) to determine apparent digestibility, as described by researchers [18] . The feed and feces samples were incubated in bovine rumen for 288 hours, washed, and dried in a forced ventilation oven. NDF and ADF concentrations were determined to calculate digestibility [19] .

The hematologic variables were obtained using tubes of blood with EDTA collected, using a VET3000 hematologic analyzer (equip®), which determines the following variables: leukocytes, lymphocytes, granulocytes, monocytes, total erythrocytes, and platelet count in cells ×10³/µl, total red blood cell count (×10⁶/µl), hemoglobin concentration (g/dL) and hematocrit (%).

Subsequently, the serum was stored in microtubes at −20˚C until analysis. Total protein (PT), albumin (AL), glucose, urea, triglycerides, and total cholesterol levels were evaluated in serum. The ANALISA® kits were used in semi-automatic equipment (Bio Plus 2000®). The concentration of globulins will be calculated using the PT-AL equation.

Lipoperoxidation is a highly rapid reaction formed by the breakdown of polyunsaturated fatty acids, which are usually measured by their products, mainly TBARS, among which malondialdehyde (MDA) is the primary one [20] . To evaluate this product, the reaction of thiobarbituric acid (TBA) with serum samples was used, which, in the presence of MDA, results in a pink product that can be read at 532 nm. Briefly, 20 µL of samples were mixed with 55 µL of distilled water, 100 µL of orthophosphoric acid (0.2 M), and 25 µL of TBA (0.1 M). A spectrophotometric reading was taken after 45 minutes of incubation at 37˚C. Results were expressed in nM MDA/mL. Serum glutathione S-transferase (GST) activity was measured based on the method described by Habig et al. [21] and expressed as U GST/mg protein. The measurement of SOD activity was based on the inhibition of the radical superoxide reaction with adrenaline, as described by McCord and Fridovich [22] . In this method, SOD present in the sample competes with the detection system for radical superoxide. One SOD unit is defined as the amount of enzyme able to slow down the speed of adrenaline oxidation by 50%. The oxidation of adrenaline leads to the formation of colored product adrenochrome that might be detected by spectrophotometry. SOD activity was determined by measuring the speed of adrenochrome formation observed at 480 nm in a reaction medium containing glycine-NaOH (50 mM, pH 10.0) and 1 mM adrenaline. Specific activity was expressed as SOD units per mg of protein.

For the measurements of ketosis in animals, blood assessment of beta-hydroxybutyrate was carried out using a device from the Free Style Optium Neo human line, which determines the concentration of ketone bodies through a drop of blood taken from the coccidial vein and placed on the tape of the device. The reference values for BHB, according to Duffield et al. [23] , BHB ≤ 1.1 mmol/L—reflects the absence of ketosis; BHB ≥ 1.2 mmol/L subclinical ketosis and BHB; 3.5 mmol/L clinical ketosis.

Ruminal fluid samples were thawed at 5˚C and manually shaken for homogenization. Aliquots of 1 mL of the supernatant of ruminal fluid samples were collected in polypropylene microtubes (2 mL) and then centrifuged for 5 minutes (12,300 × g). Then, 250 μL of the supernatant was transferred to a new microtube containing 250 μL of formic acid. The mixture was manually shaken and centrifuged for 3 min. After centrifugation, 250 μL of the supernatant of the mixture was collected in another polypropylene tube previously containing 500 μL of isoamyl alcohol (3-methyl-1-butanol) solution (638.28 μg∙mL⁻¹ in methanol), used as an internal standard, and were homogenized and centrifuged again. 650 μL of sample was inserted into a 2-mL injection vial; 1 μL of extract was injected into a gas chromatograph equipped with a flame ionization detector (GC-FID; Varian Star 3400, Palo Alto, USA) and an autosampler (Varian 8200CX, Palo Alto, USA) in split mode (1:10) at 250˚C. The carrier gas used was hydrogen at a constant pressure of 20 psi. The analytes (acetic, propionic, butyric, valeric, and isovaleric acids) were separated by a CP-Wax 52CB capillary column (60 m × 0.25 mm; 0.25 μm stationary phase thickness). The initial column temperature was set at 80˚C for 1 min and increased to 120˚C at a rate of 8˚C min⁻¹, then to 230˚C for 20˚C min⁻¹, where it remained for 1 min. The detector temperature was set to 250˚C. Method validation comprised the following parameters: selectivity, linearity, linear range, repeatability, precision, limit of detection (LOD), and limit of quantification (LOQ) for acetic, propionic, butyric, and isovaleric acids. Linearity was assessed by calculating a regression equation using the least squares method. LOD and LOQ values were obtained by sequential dilutions to signal-to-noise ratios of 3:1 and 6:1, respectively. Precision was assessed by analysing the repeatability of six replicated samples. Accuracy was determined by recovering known amounts of the standard substances added to a diluted sample ( Table S1 ). Valeric acid was expressed as the equivalent of isovaleric acid. The results were expressed in mol 100 mol⁻¹ of each short-chain fatty acid (SCFA) in the ruminal fluid.

In the laboratory immediately after collection, the milk samples were analyzed using infrared equipment (LactoStar Funke Gerber®), quantifying the proximate composition of fat, protein, lactose, and total milk solids.

Lipid extraction was performed using the method of Bligh and Dyer [24] with adaptations to determine the fatty acid profile in milk. 1.5 g of samples, 0.8 mL of water, 5 mL of methanol, and 2.5 mL of chloroform were added to a 15 mL polypropylene tube, and mechanical stirring was carried out for 30 min. Next, 2.5 mL of chloroform solution and 1.5% NaSO₄ were added to promote a two-phase system. This mixture was stirred for 2 minutes and then centrifuged for 15 minutes at 2000 rpm. The lipids obtained from the chloroform phase were subjected to fatty acid analysis.

FA methylation was performed using a transesterification method proposed by Hartman and Lago [25] . 1 mL of 0.4 M KOH methanolic solution was added to the extracted lipids in a test tube and vortexed for 1 min. The samples were kept in a water bath for 10 minutes at boiling temperature. Subsequently, they were cooled to room temperature, and 3 mL of 1 M methanolic H₂SO₄ solution were added, vortexed, and kept in a water bath for 10 min. After cooling, 2 mL of hexane was added and centrifuged at 2000 rpm for 10 min. Finally, hexane with fatty acid methyl esters (FAME) was subjected to chromatographic analysis.

A gas chromatograph model TRACE 1310 with a flame ionization detector (Thermo Scientific) was used to determine FAME. One microliter of samples was injected into a split/splitless injector, operated in split mode with a 1:20 ratio at 250˚C. Hydrogen was used as the carrier gas at a constant flow rate of 1.5 mL/min. Separation of FAMEs was performed using an RT 2560 chromatography column (100 m × 0.25 mm × 0.20 μm film thickness, Restek, USA). The initial oven temperature was programmed at 100˚C for 5 min and increased to 180˚C at an 8˚C/min rate. Then, it increases to 210˚C at a rate of 4˚C/min, and finally up to 250˚C, increases to 20˚C/min, and maintains for 7 min in isothermal. The detector temperature was kept constant at 250˚C. FAME compounds were identified by comparing the experimental retention time with those of the authentic standard (FAME Mix-37, Sigma Aldrich, St. Louis, MO). The results were presented as a percentage of each FA identified in the lipid fraction, considering the factor equivalent to the size of the FAME chain for FID and the ester conversion factor for the respective acid, according to Visentainer and Franco [26] .

All variables were subjected to the Shapiro-Wilk W-test, which revealed a normal data distribution. Skewness, kurtosis, and homogeneity were evaluated using the Levene test, and linearity was used using linear regression. All data were analyzed using the SAS MIXED procedure (SAS Inst. Inc., Cary, NC, USA; version 9.4), with the Satterthwaite approximation to determine the denominator degrees of freedom for the fixed effects (day, treatment, and day × day interaction) test in a completely randomized design. Fat-corrected milk 4%, dairy efficiency, feed conversion, sum of milk production, and average production gain were tested for fixed treatment effects using animal (within each group) as random variable. All other variables (milk production, composition, quality, serum biochemistry, blood count) were analyzed as repeated measurements and tested for fixed effects of treatment, day, and treatment × day interaction (group) and animal as random variables. All results obtained in d1 for each variable were also included as covariates; however, the command for covariates was removed from the model when p > 0.05. The mean comparison analysis was performed based on PDIFF, the Student’s test was chosen due to the experimental condition of only two groups. All results were reported as LSMEANS followed by standard error. Significance was defined when p ≤ 0.05.

The product has the following chemical composition: 978 g/kg of DM; 885 g/kg of organic material; 40 g/kg protein; 456 g/kg NDF; 367 g/kg ADF; and 52 g/kg EE. Analysis of the BC allowed identifying 88.5% of the composition of the commercial product, of which 56% was total choline (10.6% phosphatidylethanolamine, 16.8% phosphatidylinositol, 55.2% phosphatidylcholine, 9.59% lyso phosphatidylcholine, and 7.81% phosphatidylcholine “natural choline conjugates”), 8.72 g/kg total tannins, 2.53 g/kg total flavonoids.

Body weight and BCS results are shown in Figure S2 . BW and BCS of the evaluated animals had no significant difference between the groups.

Productive performance results are presented in Table 2 . Feed intake showed no statistical difference when considering the adaptation and experimental periods ( Table 2 ; p > 0.05). Milk production during the experiment period was higher in BC cows compared to the control group, with an interaction between treatments × day (p ≤ 0.05), with this production being 11.55% higher compared to the control group of animals (p ≤ 0.05). The average milk production kg/day tended to be higher in relation to the control group of animals, and the sum of milk produced per cow during d15-28, throughout the experimental period, was more significant than the control group of animals. Considering the period from 1 to 28 days, we found that there was a treatment x day interaction for milk production ( Figure 1 ; p ≤ 0.05); being greater in animals that consumed BC on some days. Milk production corrected to 4% fat was 19.1% higher in cows that consumed BC compared to CONT (p ≤ 0.05) in the experimental period (d15 to 28). No effect of the treatment was found on productive efficiency and feed conversion (p > 0.05).

Note 1: Means followed by letters written differently (a, b) on the same line illustrate a treatment effect when considering p ≤ 0.05 and a tendency to be different when p > 0.05 and ≤ 0.10; Note 2: Details of the treatment × day interaction were presented in Figure 1 and Supplementary material 4; ³Milk production corrected for 4% fat (4% FCM) was estimated by the equation proposed by the NRC (2001): FCM = 0.4 × (kg of milk produced) + 0.15 × (% fat) × (kg of milk produced).

The proximate milk composition results are presented in Table 2 and Table S2 . The results show a treatment effect for the percentage of protein, lactose, and total solids, which is more significant in cows that consumed BC than in the control group (p ≤ 0.05). This difference is related to the interaction of treatment × days on d14, d21, and d28 (p ≤ 0.05). The variables protein, lactose, and total solids were higher in the milk cows in the treatment group (p ≤ 0.05). The percentage of fat tended to be different, with the fat in the milk produced by animals in the treatment group being higher compared to the percentage of fat in the control animals, with a statistical difference in the interaction of treatment × days on d14 and d28, higher in the group of animal treatment (p ≤ 0.05).

There was a day effect for fat, protein, and total solids in milk, data presented. Over time, fat levels increased in the animals in the treatment group (p ≤ 0.05); this did not happen in the control group. Protein in milk was lower on day 14 compared to other collection days in the cows in the control group (p ≤ 0.05). Total solids in milk differed in both groups, but the highlight was in the treatment group, which increased over time.

The results of the milk fatty acid profile are described in Table 3 . The percentage of C15:0 and C21:0 fatty acids was lower in the milk of cows in the treatment group, with an interaction between treatment versus day on d14 and d28 (p ≤ 0.05). The effect of treatment for C17:0 was observed, with milk from cows in the treatment group being lower. The percentage of C18:0 tended to the interaction between treatment × day, being higher in the milk produced by the treatment group of animals than in the control group (d-28). The treatment for C18:1n9c had an effect that was more significant in cows fed with BC (p ≤ 0.05). Treatment × day interaction for C18:2n6c was observed, being greater in cows that consumed additive on d14 and d28. The percentage of C18:3n6 was higher in the milk of treatment cows compared to the control group of animals (p ≤ 0.05). The sum of unsaturated fatty acids was higher in the milk of cows in the treatment group compared to the control group (p ≤ 0.05). There was significant interaction between treatment × day on d14 and d28 (p ≤ 0.05), highlighting this more significant amount of unsaturated fatty acids. Profile of other fatty acids showed no statistical difference (p > 0.05).

Note 1: Means followed by letters written differently (a, b) on the same line illustrate a treatment effect when considering p ≤ 0.05 and a tendency to be different when p > 0.05 and ≤ 0.10; Note 2: Details of the treatment × day interaction and day effect were presented in Supplementary Material 4.

A day effect for C15:0 (pentadecanoic), which in the milk of control cows increased on days 14 and 28 compared to d1 (p ≤ 0.05), while C18:2n6c (linoleic) also increased in cows that consumed BC on days 14 and 28 compared to d1 (p ≤ 0.05). Unsaturated fatty acids had a day effect in both groups, but with opposite effects; i.e., in the milk of control cows, UFA levels decreased, while in the treatment cows, they increased (p ≤ 0.05; Table S2 ).

SCC results are presented in Figure 1 . There was no difference between groups for SCC during the experimental period (p > 0.05). However, there was a tendency (p = 0.07) in the interaction between treatments × day, being lower in animals in the group that consumed BC compared to the control group on some days of evaluation. Numerically, the cell count (SCC) was lower in animals that consumed BC throughout the experiment (p > 0.05) ( Figure 1 ).

Blood count results are presented in Table 4 and Table S3 . There was a treatment effect on erythrocyte count and hemoglobin concentration, being more significant in the treatment group compared to the control group (p ≤ 0.05), with an interaction between treatment × day in the experimental period (d14, d21, and d28). Hematocrit showed a statistical difference between the treatment and the control, higher in cows that consumed BC due to the interaction between treatment × day (d14 and d28). The number of platelets in animals in the treatment group was lower compared to the control group, with an interaction between treatment × day on d21 and d28 (p ≤ 0.05). There was no statistical difference for the other variables analyzed in the blood count: leukocytes, lymphocytes, granulocytes, and monocytes (p > 0.05).

A day effect on hemoglobin concentration increased in animals in the treatment group (p ≤ 0.05). Platelet count had a day effect in both groups, with an increase in control cows and a decrease in treatment cows (p ≤ 0.05); however, in both situations, they remained within the reference values for the breed. No effect of day for erythrocyte, leukocyte, granulocyte, neutrophil and monocyte counts; as well as hematocrit percentage (p > 0.05).

Table 4 and Table S3 describe the serum biochemistry and metabolism analyses during the experimental period. It is noted that there was an effect of treatment on the enzymatic activity of ALT and AST; that is, it was lower in the serum of animals in the treated group (p ≤ 0.05), and there was an interaction between treatment × day (d21 and d28). There was an interaction between treatment × days for the uric acid variable on d14 and d21, which was higher in the treatment group compared to the control group. Glucose levels differed in the control group, being smaller and interacting between treatment × day (p ≤ 0.05; d21). The variables cholesterol, urea, total protein, albumin, globulin, GGT enzyme, and BHB did not show statistical differences during the experimental period (p > 0.05).

The effect of day for AST and ALT in cows in the control group was that the activity of these enzymes increased, which did not occur in the treatment group (p ≤ 0.05). Uric acid levels had an effect of day, being higher on days 14 and 21 when compared to d 1 and 28 in cows that consumed BC (p ≤ 0.05). There was also an effect of day on glucose; in both groups, it increased throughout the experiment (p ≤ 0.05).

TBARS levels were lower in cows that consumed BC than in the control (p ≤ 0.05); however, there was no effect of day or treatment × day interaction. GST activity impacted day, treatment, and interaction, emphasizing more significant enzyme activity in the serum of animals that consumed BC (p ≤ 0.05). SOD activity had an effect of day and treatment × day interaction, being higher on day 28 in the blood of cows in the treatment group (p ≤ 0.05).

Note 1: Means followed by letters written differently (a, b) on the same line illustrate a treatment effect when considering p ≤ 0.05 and a tendency to be different when p > 0.05 and ≤ 0.10; Note 2: Details of the treatment × day interaction and day effect were presented in Supplementary Material 5.

The results of the SCFA profile of the rumen fluid are described in Table 5 and Table S4 . There was no effect of treatment and no treatment × day interaction for SCFA (p > 0.05). However, treatment and interaction affected the proportion of acetic, propionic, butyric, isovaleric, and valeric fatty acids (p ≤ 0.05). The percentage of acetic and butyric acid was higher in animals in the treatment group on D14 and D28 (p ≤ 0.05) compared to the control. Propionic and isovaleric acid were in lower proportions in cows in the treatment group compared to the control on days 14 and 28 (p ≤ 0.05). There was a treatment effect for the proportion of valeric acid, which was more significant in the animals fed with BC (p ≤ 0.05). There was no effect of day on variables of the SCFA profile in the rumen (p > 0.05).

Note 1: Means followed by letters written differently (a, b) on the same line illustrate a treatment effect when considering p ≤ 0.05 and a tendency to be different when p > 0.05 and ≤ 0.10; Note 2: Details of the treatment × day interaction and day effect were presented in Supplementary Material 6.

The diet digestibility results are presented in Table 5 and Table S4 . The treatment affected the CP’s apparent digestibility coefficient, which was higher in cows that consumed BC (p ≤ 0.05). The treatment did not influence the digestibility of other nutrients (organic matter, fiber, EE—p > 0.05).