Effect of low dietary concentrations of Acacia mearnsii tannin extract on chewing, ruminal fermentation, digestibility, nitrogen partition, and performance of dairy cows

The supplementation of dairy cows with tannins can reduce the ruminal degradation of dietary protein and urine N excretion, but high concentration in the diet can impair ruminal function, diet digestibility, feed intake, and milk yield. This study evaluated the effect of low concentrations (0, 0.14, 0.29, or 0.43% of diet in DM basis) of a tannin extract from the bark of Acacia mearnsii (TA) on milking performance, dry matter intake (DMI), digestibility, chewing behavior, ruminal fermentation, and N partition of dairy cows. Twenty Holstein cows (34.7 ± 4.8 kg/d, 590 ± 89 kg, and 78 ± 33 d in lactation) were individually fed a sequence of 4 treatments in 5, 4 × 4 Latin squares (with 21-d treatment periods, each with a 14-d adaptation period). The TA replaced citrus pulp in the total mixed ration and other feed ingredients were kept constant. Diets had 17.1% crude protein, mostly from soybean meal and alfalfa haylage. The TA had no detected effect on DMI (22.1 kg/d), milk yield (33.5 kg/d), and milk components. The proportions in milk fat of mixed origin fatty acids (16C and 17C) and the daily secretion of unsaturated fatty acids were linearly reduced and the proportion of de novo fatty acids was increased by TA. Cows fed TA had linear increase in the molar proportion of butyrate and linear reduction in propionate in ruminal fluid, whereas acetate did not differ. There was a tendency for the ratio of acetate to propionate to be linearly increased by TA. Cows fed TA had a linear reduction in the relative ruminal microbial yield, estimated by the concentrations of allantoin and creatinine in urine and body weight. The total-tract apparent digestibility of neutral detergent fiber, starch, and crude protein also did not differ. The TA induced a linear increase in meal size and duration of the first daily meal and reduced meal frequency. Rumination behavior did not differ with treatment. Cows fed 0.43% TA selected against feed particles >19 mm in the morning. There were tendencies for linear decreases in milk urea N (16.1–17.3 mg/dL), urine N (153–168 g/d and 25.5–28.7% of N intake), and plasma urea N at 6, 18, and 21 h postmorning feeding, and plasma urea N 12 h postfeeding was reduced by TA. The proportion of N intake in milk (27.1%) and feces (21.4%) did not differ with treatment. Reductions in urine N excretion and milk and plasma urea N suggest that TA reduced ruminal AA deamination, whereas lactation performance did not differ. Overall, TA up to 0.43% of DM did not affect DMI and lactation performance, while there was a tendency to reduce urine N excretion.


INTRODUCTION
Tannins are a diverse group of polyphenolic compounds capable of forming complexes with protein and other macromolecules (Bravo, 1998).Extensive reviews (Makkar, 2003;Herremans et al., 2020) have evaluated the effect of tannins on ruminal fermentation, diet digestibility, feed intake, and performance of ruminants.Tannins can limit protein and carbohydrate digestion in the rumen (Carulla et al., 2005) and intestines (Duodu et al., 2003), can affect the ruminal biohydrogenation of lipids (Khiaosa-Ard et al., 2009), microbial growth (Ahnert et al., 2015), and diversity (Vasta et al., 2019), and have the potential to affect protozoa population (Bhatta et al., 2009) and reduce methane emission (Jayanegara et al., 2012).Tannins are capable of binding protein and the possibility of reducing the ruminal degradation of dietary protein (to increase RUP supply) and urine N excretion has created interest on tannin supplementation for dairy cows (Aguerre et al., 2016(Aguerre et al., , 2020)).However, tannins can reduce ruminal microbial yield and the absorption of AA by the animal may not increase, even with increased proportion of dietary CP as RUP (Wang et al., 1996;MacAdam and Villalba, 2015).Further, tannins can decrease NDF digestibility (Reed, 1995) and can interfere with food acceptability due to astringency (Griffiths et al., 2013), potentially affecting eating and rumination behaviors.
The animal response to tannin supplementation is dependent on type of product and concentration in the diet (Aboagye et al., 2018).Tannins are classified as condensed (CT) or hydrolyzable (HT), based on their capacity to resist hydrolysis under mild acidic or alkaline conditions into sugars and phenolic carboxylic acids (Van Soest, 1994).The CT are considered to have higher binding capacity to protein than HT, and they are not degraded in the rumen (Makkar et al., 1995).Acacia mearnsii (TA; Black wattle), originally from Australia, has been cultivated as a source of wood, as well as tannin extract from its bark, for a variety of commercial applications (Tanac, Brazil).The tannin extract of TA is a source of CT (Ahmed et al., 2005).
In experiments evaluating TA in the diet of ruminants, a shift from urine N to fecal N excretion has been observed at concentrations in the diet greater than 0.9% of DM basis, associated with negative effects of TA on ruminal function and animal performance (Carulla et al., 2005;Ávila et al., 2015;Orlandi et al., 2015).Grainger et al. (2009) also observed that TA at 0.9 and 1.5% of estimated DMI partitioned N away from urine toward feces of grazing dairy cows, but also reduced DMI, digestibility of energy and CP, and milk yield.Additionally, Griffiths et al. (2013) highlighted that milk yield was reduced when TA was drenched daily to grazing dairy cows at approximately 0.6 to 2.9% of diet DM.
Conversely, Orlandi et al. (2020) detected that TA at 0.38% of DM did not affect ruminal microbial yield, NDF digestibility, and DMI in sheep.Interestingly, Alves et al. (2017) observed a reduction in methane excretion when TA was fed at 0.7% of diet DM to grazing dairy cows, with no effect on milk yield and DMI.The effect of graded concentrations of quebracho-chestnut tannin extract containing CT and HT on lactation performance and N partition of dairy cows was evaluated in 2 experiments (Aguerre et al., 2016(Aguerre et al., , 2020)).Aguerre et al. (2016) suggested that the optimum concentration of tannin extract was 0.45% of DM, based on the reduction in DMI and milk protein concentration and yield at higher concentrations in the diet, as well as a linear decrease in CP and NDF digestibility at those higher concentrations.The tannin extract induced a linear increase in fecal N excretion and a reduction in urine N, but did not affect the efficiency of N utiliza-tion for lactation.Aguerre et al. (2020) fed lactating dairy cows for 13 wk with the same CT and HT tannin extract (0.45 and 1.8% of DM) and detected the highest milk true protein concentration with 0.45% of DM.Although tannin extracts vary in their composition (Mueller-Harvey, 2006), it seems that the upper limit for tannin extract supplementation to dairy cows, without detrimental effects on ruminal function, digestibility, DMI, and lactation, is around 0.5% of diet DM.To our knowledge there are no experiments evaluating for lactating dairy cows the effects of low TA concentrations in the diet.
The objective of this study was, therefore, to evaluate the effect of 0.15, 0.30, and 0.45% of TA in DM basis on lactation performance, DMI, total-tract digestibility, ruminal fermentation, chewing behavior, and N partition of dairy cows.Our hypothesis was that TA supplementation up to 0.45% of DM would increase milk protein secretion and reduce urine N excretion by reducing the ruminal degradation of dietary protein.

MATERIALS AND METHODS
Experimental procedures were approved by University of Lavras Bioethics Committee in Utilization of Animals: Protocol 033/20.

Cows and Treatments
The experiment was conducted from July to October, 2020 in an open-walled, sand-bedded tiestall barn.Twenty Holstein cows (34.7 ± 4.8 kg/d, 590 ± 89 kg, and 78 ± 33 DIM; ±SD), 8 primiparous and 12 multiparous [mean = 2.40 ± 1.14 parity (2-7)], were assigned to the experiment.Cows were milked 3 times/d starting at 0500, 1300, and 1900 h in an adjacent herringbone parlor.Cows were allocated to five 4 × 4 Latin squares; allocation was performed first by parity (1 vs. >1) and then by milk yield and DIM.A sequence of 4 treatments, balanced for carry-over effects, was assigned to each cow for 21-d periods (each including a 14-d adaptation period).Treatments were 0, 0.15, 0.30, or 0.45% of TMR DM of TA (Tanac, Brazil).The concentration of CT in TA was 18.1% of DM basis as analyzed with a molecularly imprinted polymer solid-phase extraction and liquid chromatograph coupled to UV-vis detector (Martins et al., 2020).
The experimental diets were formulated 1 d before the start of experiment based on near-infrared reflectance spectroscopy analysis of forages and DMI prediction (NRC, 2001) for the mean cow during 7 d pre-experiment (34.7 kg/d, 3.7% milk fat, 590 kg BW, 78 DIM).The TA replaced citrus pulp in the TMR, whereas other feed ingredients were kept constant  1).The TMR for each treatment was mixed once per day in a 1.2 m 3 stationary vertical mixer (Unimix 1200, Casale) after weighting each feed with a precision scale (MOD B-520.Líder Balanças) and cows were fed at 0700 h.Feed was pushed up manually with a broom at least 10 times/d.The TA was mixed with the dietary soybean meal before addition to the mixer to improve homogenization to the TMR.The TMR offered and refusals per cow were measured daily and data during the last 7 d of each period were used for analysis of DMI.Feed was offered to each cow to allow for 10 to 15% of offered as daily refusal and was increased to be around 15% orts during the evaluation of particle size sorting behavior.Samples of whole plant corn silage, ensiled corn grain, and alfalfa haylage were collected weekly for DM determination with a microwave oven (by drying for 5 min and then in 3 min steps until a stable weight was obtained), and the TMR was adjusted accordingly.Individual feeds, each TMR, and orts per cow were sampled daily and formed weekly composites to represent the last 7 d of each experimental period.The composition of a treatment in ingredients was the total intake of an ingredient DM (orts assumed to have the same ingredient composition of the offered TMR on a DM basis) divided by total DMI.The composition of a treatment in nutrients was the total intake of a nutrient (TMR offered − orts per cow) divided by total DMI.

Feed Analysis
Composite samples per period of feed ingredients, orts per cow, and TMR per treatment were dried in a forced-air oven at 55°C for 72 h and ground to pass a 1-mm diameter mesh screen (Wiley mill, Thomas Scientific).The DM concentration was determined at 105°C for 24 h and ash was at 550°C for 8 h.Samples were sent to a commercial laboratory (3rLab/Rock River Laboratories, Lavras, Brazil) for determination of CP with a Kjeldahl steamer distillator (AOAC International, 2002), NDF by filter bag technique with heat-stable α-amylase (ANKOM Technology; Schlau et al., 2021), ether extract (AOAC, 1990), and starch with α-amylase and amyloglucosidase and colorimetry for

Performance and Milk Composition
The mean milk yield and DMI of d 16 to 21 of each treatment period were used to compare treatments.An aliquot of the yield of each milking was collected and composited in duplicate per cow per day in proportion to the yield of each milking.Composite milk samples were stored under refrigeration (at 5°C) in flasks containing 2-bromo-2-nitropropane-1-3-diol preservative and manually agitated daily until shipping to a commercial laboratory.Milk components (CP, casein, lactose, fat, and TS), MUN, SCC, and milk fatty acids (FA) were measured by mid-infrared analysis (Nexgen FTS/FCM.Bentley Instruments Inc.) at the Laboratory of the Paraná State Holstein Breeders Association (APCBRH, Curitiba, Brazil).Milk energy secretion (Mcal/d) was calculated (NASEM, 2021): [(0.0929 × % fat) + (0.0547 × % CP) + (0.0395 × % lactose)] × kg of milk.ECM (kg/d) was calculated as: Milk energy secretion/0.70(assumes 0.70 Mcal/kg of milk with 3.7% fat, 3.2% CP, and 4.6% lactose).The feed efficiencies were calculated as milk yield/DMI and ECM/DMI.The BW was measured on d 16 to 20 of each treatment period immediately after the morning and afternoon milking with an individual walk-over weigh scale (Eziweigh2, Tru-Test) and a mean value was generated per cow per period.The BCS was assessed on a 1 to 5 scale (Wildman et al., 1982) on d 21 of each treatment period and was the mean of 3 independent evaluators.

Ruminal Fermentation
On d 21 of each treatment period, samples of ruminal fluid were obtained with a flexible oro-gastric tube, with the goal of having minimal saliva contamination by discarding the first flow of ruminal fluid.Samples were obtained at 10.6 ± 0.12 h after the morning feeding (0700 h), at random within square.Samples of strained ruminal fluid were frozen in liquid nitrogen and stored at −20°C, and then thawed and centrifuged at 4°C at 8,855 × g for 15 min.The supernatant was analyzed for VFA by gas-liquid chromatography (CP 3800 Gas Chromatography Varian, Varian Chromatography Systems), with a capillary column [CP-Wax 58 (FFAP) CB, Varian Analytical Instruments].Another strained ruminal fluid sample (10 mL) was mixed to a 36% formaldehyde solution (10 mL) before refrigeration at 5°C for total protozoa counting.Samples were stained according to Dehority (1984) and total protozoa was enumerated with an optical microscope in a Neubauer chamber (Warner, 1962).

Digestibility
The total-tract apparent digestibility of OM, NDF, starch, and CP was estimated by fecal sampling on d 16 to 20 of each treatment period.Spot samples obtained around 0800, 1300, and 1800 h each day were frozen and formed period composites on an equal fresh basis.Fecal DM and nutrient concentration were determined as already described for feeds and orts.Acid insoluble ash (Van Keulen and Young, 1977) in the consumed TMR and in orts and feces per cow were used to estimate the daily fecal DM excretion based on intake of internal marker divided by fecal marker concentration.Intake (kg/d) of digestible OM (DOMI), NDF, starch, and CP was calculated.

Microbial Yield
Relative ruminal microbial yield was estimated with urine spot samples obtained simultaneously to fecal sampling by subvulvar stimulation (d 16-20 of each treatment period at 0800, 1300, and 1800 h each day).For the evaluation of allantoin (Alla) concentration in urine, spot samples (120 mL/d) were diluted with 20% sulfuric acid (5 mL/d) throughout the sampling procedure and were immediately refrigerated at 4°C.At the end of each day, a 4% sulfuric acid solution was added (5 mL of urine and 20 mL of acid) and samples were frozen at −20°C.The Alla was analyzed as in Young and Conway (1942).Creatinine (Crea) was analyzed in urine samples refrigerated during sampling and frozen at the end of each day.The Crea was analyzed with a colorimetric assay based on Jaffe's reaction with alkaline picrate (Creatinina Doles, Doles Reagentes e Equipamentos para Laboratórios, Brazil).Relative ruminal microbial yield was estimated, using the Alla to Crea ratio multiplied by BW measured 2 times/d during d 16 to 20 (Chen et al., 1995).A ratio between the estimated relative ruminal microbial yield and DOMI was calculated to estimate the efficiency of microbial synthesis.

Chewing and Sorting
During d 17 to 19 of each treatment period, rumination and eating behaviors were monitored by visual observation at 5-min intervals continuously for 24 h/d as in Pereira et al. (1999).Buccal activities were rumination, eating, drinking, and idleness.Eating, rumination, and chewing (eating + rumination) were calculated (min/d and min/kg DMI).

Oliveira et al.: TANNIN FOR DAIRY COWS
Individual feeding observations were combined and separated into meals using a meal criterion (i.e., the minimum duration of time between meals) calculated for each cow.Meal criteria were calculated for each cow using methods described by DeVries et al. (2003); in summary, a software package (MIX 3.1.3;MacDonald and Green, 1988) was used to fit normal distributions to the frequency of log 10 -transformed intervals of time between recorded eating observations.If the interval of time between 2 recorded observations of eating exceeded the determined meal criterion, this was classified as a different meal.The number of different meals in a day was termed meal frequency (meals/d).Total meal time (min/d) was the total eating time (eating observations/d × 5 min), plus all the nonfeeding intervals shorter than the length of the meal criterion of each cow.Meal duration (min/meal) was calculated as the total daily meal time divided by the meal frequency.Finally, mean meal size was calculated per cow per day as the ratio between daily DMI (kg/d) and meals per day (Coon et al., 2018).The duration and time postfeeding of the longest daily meal was calculated.The duration of the first (initial) daily meal was measured with a stopwatch.Five evaluators observed the behavior of all cows, individually, after offering feed at 0700 h until the last cow finished its first meal.
Particle size sorting behavior in periods of the day was evaluated on d 17 to 19 of each treatment period with the Penn State Particle Separator with the 19-and 8-mm diameter screens and pan (Lammers et al., 1996).The particle distribution and weigh of the offered TMR and available orts of each cow was measured at 0700 (at feeding), 1200, 1900, and 0500 h (during morning milking).The predicted intake (as-fed basis) of particles on each screen was % TMR retained on screen × kg of TMR consumed.The observed intake of particles was % TMR retained on screen × kg of TMR offered − % orts retained on screen × kg of orts.The selection index (Leonardi and Armentano, 2003) was 100 × (observed intake/predicted intake).Sorting values below 100% represent selective refusal, above 100% represent preferential intake, and equal to 100% represent no selection.A mean sorting value was generated per screen, per cow, and per period.The proportion of daily DMI in the morning (0700 to 1200 h), afternoon (1200 to 1900 h), and night (1900 to 0700 h) were also determined.

Plasma Urea N
Blood samples were collected from the coccygeal vessels of each cow immediately before the start of the first daily meal after the morning milking (0700 h) and at 1.5, 3, 6, 12, 18, and 21 h after first meal on d 18 of each treatment period.Samples were collected in tubes with EDTA and were centrifuged at 2,000 × g for 10 min at room temperature.Plasma was obtained and frozen at −20°C for urea N determination (PUN) with an enzymatic assay based on the conversion of urea to glutamate, oxidation of NADH, and measurement of urea at 340 nm (Ureia UV, Doles Reagentes e Equipamentos para Laboratórios).

Nitrogen Partition
For the evaluation of total N content of urine, spot samples (120 mL/d) were diluted with 20% sulfuric acid (5 mL) throughout the sampling procedure and were immediately refrigerated at 4°C.At the end of each day, the pH of the sample was adjusted to <3 (2.40 ± 0.45) with drops of 95% sulfuric acid (Êxodo Científica) and samples were frozen at −20°C.The endogenous Crea excretion was 0.2380 mmol/kg BW (26.93 mg/ kg BW), the mean value of Valadares et al. (1999), Lee et al. (2019), Chizzotti et al. (2008), andPereira et al. (2021).Urine volume (L/d) was endogenous Crea excretion/Crea concentration in urine.Fecal N and urine N concentrations were determined with a Kjeldahl steamer distillator (AOAC International, 2002).The N concentration in milk was milk CP/6.38 (Ipharraguerre and Clark, 2005).The proportion of N intake (% of daily intake) and excretion (g/d) of urine N, milk N, and fecal N were also calculated.
Significance was declared at P ≤ 0.05 and tendencies at 0.05 < P ≤ 0.10.

Diets
The experimental diets had similar concentrations in DM of ingredients and nutrients (Table 1).The CP was 17.1% of DM, mostly from soybean meal and alfalfa haylage.Starch was 29.8% of DM (ranging from 28.5 to 31.1% of DM), mostly from silages of whole plant corn and rehydrated corn grain.The TA replaced citrus pulp in the diet, citrus pulp ranged from 6.0 to 6.4% of diet DM.The actual concentration of TA in the consumed diets were: 0.14, 0.29, and 0.43% of DM.The actual TA concentrations in the diet were slightly different from the planned (0.15, 0.30, 0.45% DM TA), and were used to report results in tables.

Intake, Performance, and Milk FA
Intake and milking performance are in Table 2.There was no detected effect of treatment (P ≥ 0.14) on variables related to milk components content or daily secretion, SCC, feed efficiency, BW, and BCS.The DMI was not affected by TA concentration in the diet (P ≥ 0.41).
Differences were detected on the secretion (g/d) of milk FA (Table 3).More TA in the diet induced linear increases in the secretion of UFA (P = 0.03) and oleic FA (P = 0.02) and there was a tendency for linear increase in MUFA secretion (P = 0.10).Milk FA profile (g/100 g FA) was also affected by TA concentration.Cows fed more TA had a linear reduction in the proportion of mixed FA (P < 0.01) and had a linear increase in proportion of short-chain de novo FA (P < 0.01) in milk fat.The proportion of palmitic FA was linearly reduced (P = 0.05) and there were linear tendencies (P = 0.10) for UFA to be increased and saturated to be reduced by TA.There were tendencies (P = 0.08) of quadratic changes in the proportions of oleic and stearic FA in milk fat in response to TA.

Ruminal Fermentation, Microbial Yield, and Digestibility
The TA affected ruminal fermentation profile and microbial yield (Table 4).The increase in diet TA concentration induced a linear reduction in ruminal propionate proportion (P = 0.02) and a linear increase in butyrate proportion (P = 0.03).Acetate proportion did not differ (P ≥ 0.23).There was a tendency for TA to induce a linear increase in the acetate to propionate ratio in ruminal fluid (P = 0.09).The concentrations of total VFA (P ≥ 0.12) and total protozoa (P ≥ 0.26) did not differ.The estimated indicators of relative ruminal microbial yield and efficiency, based on Alla and Crea concentrations in urine and BW, were linearly reduced by the increase of TA in diet (P ≤ 0.04), but the totaltract digestibility of nutrients did not differ (P ≥ 0.25).The TA had no detected effect on DOMI and on the intakes of digestible NDF and CP (P ≥ 0.22), but there was a linear increase in digestible starch intake with more TA in the diet (P < 0.01).The daily excretion of urine (24.8 L/d) was not affected by TA concentration in the diet (P ≥ 0.17).

Chewing and Sorting
The TA had effect on chewing behavior (Table 5).There was a tendency for a quadratic effect of TA on eating time (316 min/d as highest on 0.43% and 302 min/d as lowest on 0.29%, P = 0.09); no other effect of TA on variables describing eating and rumination behaviors was detected (P ≥ 0.14).The TA induced a linear reduction in meal frequency (P = 0.01) and linearly increased mean meal size (P = 0.02).There was a quadratic increase in meal duration induced by TA (P = 0.04).Cows fed TA had a linear increase in first meal duration (P = 0.04).Cubic effects of TA concentration in the diet were detected for particle size sorting behavior in the morning and night (P ≤ 0.10).The TA at 0.43% of DM apparently induced rejection of long (>19 mm) feed particles in the morning, while cows in the other treatments selected in favor of long particles (P < 0.01).The TA at 0.43% also tended to induce selection in favor of particles retained on the 8-mm screen, whereas cows fed the other treatments rejected the 8-mm screen (P = 0.06).
The selection indices of particles over the 24-h period (0700 to 0700 h) were all close to 100% (indicating no preferential selection).

Urea N in Milk and Plasma and N Partition
Variables describing urea N in milk and plasma and N partition are in Table 6.There was a tendency for a linear reduction in MUN with more TA in the diet (P = 0.10).Cows fed increased levels of TA had linear reduction in PUN 12 h after feeding (P = 0.01), and there were tendencies of reduced PUN also at 6, 18, and 21 h after feeding (P ≤ 0.08).There were tendencies (P = 0.10) for the excretion of N in urine to be linearly reduced by TA, as well as urine N as a proportion of N intake.The excretion of N in feces and in milk were not affected by TA (P ≥ 0.35).The efficiency of N excretion in milk was 27.1% of daily intake and fecal N was 21.4% of daily intake.

DISCUSSION
The supplementation of low concentrations of TA in the diet of lactating dairy cows reduced urine N excretion and did not affect the total-tract apparent digestibility of nutrients, DMI, and lactation performance.There was a tendency for the excretion of urine N to be reduced in approximately 9% of the highest value (168 g/d) with 0.43% TA in diet (153 g/d).The reduction in estimated urinary N excretion when a mixture of CT and HT was fed at 0 or 1.8% of diet in DM basis was 11% (Aguerre et al., 2020) and the meta-analysis of Herremans et al. (2020) predicted an average reduction of 11% in urine N in dairy cows supplemented with various sources of tannins.Tannins, in general, are known to reduce ruminal deamination of AA and urinary N loss (Makkar, 2003;Frutos et al., 2004).Ávila et al. (2015) observed that TA in the diet of steers (at 1.5% of DM) increased the duodenal flow of nonammonia, nonmicrobial N, and AA.This is a plausible explanation for the lower MUN and PUN observed with more TA in the diet.However, the ruminal degradation of carbohydrates could also affect MUN, PUN, and urine N, independently of ruminal CP degradation.The increase in digestible starch intake may have been involved in the reduction in N loss in urine and milk (Reed, 1995).Although the experimental diets had similar ingredient composition (difference in 0.4% of DM, citrus pulp vs. TA), DMI, and the digestibility of starch and NDF, the effect of TA on eating and sorting behavior apparently affected digestible starch intake, even at similar intakes of digestible NDF and energy (DOMI).The total-tract digestibility of CP was not affected by TA, suggesting that the reduction in ruminal protein degradation did not induce the formation of indigestible protein within the lower digestive tract (Mueller-Harvey, 2006).
The negative environmental impact of urine N was reduced with TA supplementation in this experiment, urine N is considered to be more detrimental to the environment than fecal N (Dijkstra et al., 2013;Lessa et al., 2014).The capacity of TA to reduce urine N excretion in ruminants has been shown, although at concentrations in the diet that reduced intake, digestibility, and animal performance (Carulla et al., 2005;Grainger et al., 2009).The TA at about 0.9% of diet in DM basis, fed with a fixed amount per day of partial mixed ration to grazing dairy cows, reduced the daily urine N excretion and had no effect on milk yield and DMI (Pozo et al., 2022).Tannins excreted in feces can have beneficial effect on manure N emissions (Sliwinski et al., 2004;Powell et al., 2011;Sepperer et al., 2020), leading to additional environmental advantage.However, potential economic return to TA supplementation driven by milking performance and feed efficiency was not detected in this experiment, similarly to that observed in the meta-analysis of Herremans et al. (2020).
The TA concentrations in diet resulted in mean intakes of 30.7, 64.4, and 95.9 g/d of TA on treatments 0.14, 0.29, and 0.43, respectively.The TA up to 96 g/d had no effect on total-tract diet digestibility, DMI, and lactation performance in this experiment.The reduction in the estimated relative ruminal microbial yield is a plausible explanation for the lack of positive response in cow performance, even with greater RUP supply with TA (Wang et al., 1996;MacAdam and Villalba, 2015).2022) and is known to inhibit microbial yield at high concentrations in the diet (Carulla et al., 2005;Orlandi et al., 2015).Curiously, although the estimated relative ruminal microbial yield was negatively affected by low concentrations of TA in this experiment, diet digestibility was not impaired by the reduction in microbial yield.There was no observed difference in total protozoa concentration, suggesting that the reduction in microbial yield was driven by action on ruminal bacteria.Changes induced by TA in ruminal bacteria diversity (Silva de Sant'ana et al., 2022) or the compensatory effect of intestinal digestion on ruminal degradation (Oba and Allen, 2003) may explain the similarity in total-tract digestibility at lower ruminal microbial yield with TA.
2 Chewing = eating + rumination. 3 Minutes from the daily feeding after the morning milking at 0700 h.
cal significance, they support the concept that TA can affect the ruminal biohydrogenation of UFA, even at low concentrations in the diet.The supplementation of tannins to improve the nutraceutical properties of dairy products through manipulation of milk fat composition has been extensively investigated (Purba et al., 2020).
Our data suggest that TA increased the secretion (g/d) of UFA in milk, suggestive of a reduction in ruminal biohydrogenation of FA.The CT is able to depress the biohydrogenation of UFA by changes in ruminal microbiome composition (Vasta et al., 2019).Cows fed more TA in the diet also had increased proportion of short-chain de novo FA and decreased proportion of mixed FA, at the same milk fat concentration and yield.Pasture rich in CT reduced the proportion of preformed and increased the proportion of de novo FA in milk fat of grazing dairy cows (Turner et al., 2005).
Grazing dairy cows fed a partial mixed ration with TA tended to have lower proportion of 16:0 + 16:1 FA and had lower proportion of C16:1 cis-9 and 14:1/14:0 and 16:1/16:0 Δ 9 -desaturase ratio indices than cows fed a control diet (Pozo et al., 2022).On the contrary, a blend of CT and HT increased the C14: 1/ 14: 0 ratio index in milk fat of sheep (Buccioni et al., 2015).The changes in milk FA profile are not easily predictable and the effect on milk quality may vary, but they suggest that low concentrations of TA could affect ruminal function and milk FA secretion and profile.
Ruminal fermentation profile was also affected by TA supplementation at low concentrations.Cows fed increasing concentrations of TA in the diet had increased proportion of butyrate, lowered proportion of propio-nate, and similar proportion of acetate in total VFA.A tendency was observed for a linear increase in the acetate to propionate ratio in ruminal fluid with more TA in the diet.This change in ruminal fermentation profile may have been mediated by changes in rumen microbiome (Silva de Sant'ana et al., 2022), probably related to a direct effect of TA on ruminal microbes and does not support a reduction in methane emission in response to TA (Williams et al., 2019).At the low concentrations evaluated in our study, TA affected ruminal fermentation profile, in addition to the reduction in the estimate of relative ruminal microbial yield.
Effects of TA were detected on eating behavior.The effect of tannin on eating behavior may be driven by the capacity of tannin to bind protein in saliva, leading do astringency (Mueller-Harvey, 2006).Cows fed more TA in the diet had a longer first meal and less frequent and, on average, larger meals during the day.Although there were effects of TA on eating behavior and particle size selection, daily DMI was not affected by TA, but digestible starch intake may have been affected by feed sorting behavior.There was no evidence that diet acceptability was a major issue within the TA concentrations supplemented.Rumination behavior did not differ, coherent with the absence of an effect of TA on NDF digestibility.The selection indices of particles over the 24 h period (0700 to 0700 h) were all close to 100%, suggesting that sorting was not a significant issue in this experiment.
urine excretion is usually attributed to tannin as a result of reduced water intake.Urine excretion was not affected by TA in our experiment, suggesting that TA concentrations in the diet were adequate for lactating dairy cows to maintain hydration.
The ratio of urine N to fecal N was 1.25 on 0% TA, 1.31 on 0.14% TA, 1.30 on 0.29% TA, and 1.14 on 0.43% TA.Excessive N intake increases urine N excretion exponentially, whereas fecal N increases linearly in dairy cows (Castillo et al., 2000).Based on this survey of the literature, the urine N to fecal N ratio can range from less than 1 at low N intake (<400 g/d) to more than 2 at high N intake (>700 g/d).However, fecal N may have been under-predicted in our experiment, although there was no treatment effect on fecal N excretion, because it was lower (130 vs. 184 g/d) than the mean value for dairy cows (32.1 kg of milk/d, 21.4 kg of DMI/d, 16.4% CP in DM basis) reported in a recent meta-analysis (Bougouin et al., 2022), at similar N intake (597 vs. 569 g/d) and urine N excretion (162 vs. 175 g/d) to our experiment.Residual N was apparently high in our experiment (143 g/d), but did not differ by treatment (P ≥ 0.51).Assuming that acid insoluble ash may underestimate fecal N excretion by 15 to 20% relative to total fecal collection (Lee and Hristov, 2013), fecal N excretion may have been underestimated by 20 to 30 g/d in our experiment.Under this scenario, TA had no detectable effect on the excretion of N in milk.The effect of low concentrations of TA on cows fed limited dietary protein supply needs further evaluation.
The slope and intercept of linear regressions of treatment least squares means with linear effect of TA concentration in the diet (0, 0.14, 0.29, 0.43% DM) were generated to estimate the response per unit of TA in DM (Table 7).The slopes were standardized for the intercept of each variable (slope/intercept × 100%) and the positive integer was generated.Among these sensitive variables, the proportional variation in the estimated relative ruminal microbial yield was the most affected by the increase of TA in the diet and milk FA profile was the least affected.The PUN was proportionally more affected by TA in the diet than MUN.Based on the coefficient of determination of the linear regressions, first meal duration was a better predictor of the effect of TA on eating behavior than meal frequency.Urine N as a proportion of daily N intake had the lowest linear correlation with diet TA concentration.Nutritional strategies capable of avoiding or compensating for the negative impact of TA on ruminal microbial yield may be avenues to induce positive lactation response to TA at low concentrations in the diet.Based on regression, the reduction of N excretion in urine in response to 0.5% TA in the diet should be around 14.6 g/d (−29.2/2).

CONCLUSIONS
The supplementation of low concentrations in the diet of TA to lactating dairy cows tended to induce a linear reduction in MUN, PUN, and urine N excretion.The TA induced minor changes in milk FA secretion and profile and ruminal butyrate proportion was linearly increased and propionate was decreased by TA.The TA affected eating and sorting behavior, but daily DMI was not affected.The low concentrations of TA did not affect the total-tract apparent digestibility of nutrients, but were capable of reducing the relative ruminal microbial yield.Lactation performance did not differ by treatment.Overall, TA up to 0.43% of DM did not affect DMI and lactation performance, whereas it tended to reduce urine N excretion.
Oliveira et al.: TANNIN FOR DAIRY COWS (Table

Table 1 .
Composition and particle size distribution of treatments containing 0, 0.14, 0.29, and 0.43% of DM of Acacia mearnsii tannin extract

Table 2 .
Oliveira et al.:TANNIN FOR DAIRY COWS Dry matter intake, lactation performance, SCC, feed efficiency, BCS, and BW of cows fed with treatments containing 0, 0.14, 0.29, and 0.43% of DM of Acacia mearnsii tannin extract

Table 3 .
Oliveira et al.:TANNIN FOR DAIRY COWS Milk fatty acid secretion and profile of cows fed treatment diets containing 0, 0.14, 0.29, and 0.43% of DM of Acacia mearnsii tannin extract

Table 4 .
Oliveira et al.:TANNIN FOR DAIRY COWS Ruminal molar VFA proportions, total protozoa concentration, total-tract apparent digestibility, intake of digestible nutrients, estimated relative ruminal microbial yield (urinary allantoin/creatinine × BW), and urine excretion of cows fed treatment diets containing 0, 0.14, 0.29, and 0.43% of DM of Acacia mearnsii tannin extract acted on a diet of excessive N supply.The TA can affect ruminal microbial diversity (Silva deSant'ana et al.,

Table 5 .
Oliveira et al.:TANNIN FOR DAIRY COWS Chewing and meal behavior, proportion of daily intake, and particle size sorting behavior of cows fed treatment diets containing 0, 0.14, 0.29, and 0.43% of DM of Acacia mearnsii tannin extract