Effects of protein source and lipid supplementation on conservation and feed value of total mixed ration silages for finishing beef cattle

Abstract

The objective of this study was to examine the conservation process and feed value of total mixed ration (TMR) silages. In exp. 1, we evaluated the fermentation pattern and aerobic stability of TMR silages containing different protein and lipid supplementations. In exp. 2, we compared the performance of finishing beef heifers fed those TMR silages. In both experiments, treatments were as follows: ensiled TMR with urea (U); ensiled TMR without a protein supplement at ensiling, but soybean meal supplemented at feeding to balance diet crude protein (CP) in exp. 2 (SMnf; where the acronym nf indicates nonfermented); ensiled TMR with soybean meal (SM); and ensiled TMR with rolled soybean grain (SG). Thirty-two Nellore heifers (313 ± 8.8 kg shrunk body weight [SBW]) were blocked by initial SBW, housed in individual pens, and enrolled in exp. 2 for 82 d. In exp. 1, treatment without a protein supplement (SMnf) had a lower content of CP, soluble CP, NH₃-N, pH, and Clostridium count compared with U (P ≤ 0.03). Lactic acid concentrations tended to be reduced for SMnf compared with U (P = 0.09). Ethanol concentration was reduced in SG compared with SM (P < 0.01). 1,2-Propanediol concentration was increased in SMnf compared with U (P < 0.01), reduced in SM compared with SMnf (P = 0.02), and increased in SG compared with SM (P = 0.02). Dry matter (DM) loss during fermentation was low and similar among treatments (~3.7%). All silages remained stable during 10 d of aerobic exposure after feed out. Considering fermentation traits, such as pH (≤4.72), NH₃-N (<10% of N, except for U treatment), butyric acid (<0.05 % DM), and DM losses (<3.70% DM), all silages can be considered well conserved. In exp. 2, diets were isonitrogenous because soybean meal was added to SMnf before feeding. Compared with SM, cattle fed SG made more meals per day (P = 0.04) and tended to have a decreased intermeal interval (P = 0.09). DM intake, average daily gain, final SBW, hot carcass weight, Biceps femoris fat thickness, and serum levels of triglycerides and cholesterol were increased for SG compared with SM (P ≤ 0.05). In brief, TMR silages exhibited an adequate fermentation pattern and high aerobic stability. The supplementation of true protein did not improve animal performance, whereas the addition of soybean grain as a lipid source improved the performance of finishing cattle fed TMR silages.

Introduction

Complete or total mixed rations (TMR) are produced by mixing forages, byproducts, concentrates, minerals, vitamins, and additives. From this mixture of ingredients, animals can consume the necessary nutrients to meet their maintenance and production requirements (Schingoethe, 2017). When consumed as a TMR without sorting of ingredients, more rumen fermentation and better use of nutrients should occur compared with feeding separate ingredients (NRC, 2001). Alternatively to daily preparation, TMR can be ensiled. Several studies show that ensiled TMR can be well preserved with minimal dry matter (DM) loss during storage and feed out (Weinberg et al., 2011; Chen et al., 2015; Hao et al., 2015; Kondo et al., 2016; Ning et al., 2017).

Several benefits have been associated with TMR silages, such as reductions in labor and the need for specialized machinery on-farm (if TMR silage is purchased), mixture uniformity and composition, reduced intensity of particle sorting in the feed bunk due to moisture redistribution (during storage), and the possibility of conserving wet byproducts with greater efficiency and high aerobic stability (Nishino et al., 2003; Weinberg et al., 2011; DeVries and Gill, 2012; Restelatto et al., 2019a).

Greater ruminal and total-tract starch digestibility have also been reported for ensiled compared with fresh TMR (Miyaji et al., 2017; Miyaji and Nonaka, 2018). Ensiling often increases starch availability, especially in cereals with higher prolamin content in their endosperm (e.g., flint corn and sorghum) due to proteolysis during silage fermentation (Benton et al., 2005; Hoffman et al., 2011; Der Bedrosian et al., 2012; Junges et al., 2017). In TMR silages, however, all dietary protein is exposed to proteolysis during storage. Thus, increases in diet soluble protein and rumen ammonia concentration were reported for TMR silages compared with fresh TMR (Kondo et al., 2016).

Nutrient changes caused by ensiling can lead to changes in TMR feed value. An increase in dietary soluble protein and ammonia concentrations can lead to a lower supply of metabolizable protein (MP) to animals fed TMR silages even with a possible increase in microbial protein synthesis provided by increased starch degradation in the rumen. A simulation with the current Beef Cattle Nutrient Requirements Model (NASEM, 2016) indicates that ensiling a typical finishing diet (based on dry-rolled corn and urea as a protein supplement) would change the balance of MP from positive to negative.

Thus, the objective of this study was to examine the conservation process and feed value of TMR silages with different protein sources and fat supplementation for finishing beef cattle. In the exp. 1, we investigated the TMR silage fermentation traits and aerobic stability. In the exp. 2, we examined whether cattle differ in performance when fed nonprotein nitrogen as urea or true protein from soybean meal, whether the true protein source can be ensiled with other TMR ingredients or should be supplied at feeding, and whether animals respond to a lipid source (rolled soybean grain) in diets with a high proportion of ensiled corn grain.

Material and Methods

Animal care and handling procedures were approved by the Ethics Committee for Animal Use of the Maringa State University (protocol number: 4647020718—CEUA/UEM).

Experiment 1

Preparing and ensiling the TMR

Four experimental TMR silages with different protein sources were prepared: U: TMR with urea; SM: TMR with soybean meal, SG: TMR with rolled soybean grain, and SMnf: TMR formulated with soybean meal, but the soybean meal was omitted from TMR at ensiling (the acronym nf indicates that SM was nonfermented, hence the SMnf TMR was ensiled without a protein supplement) (Table 1). Before mixing the rations, corn and soybean grain were processed with a roller mill (SEGU 30, Multiagro Implementos Agrícolas, Porto Alegre, Brazil).

Table 1.

Composition of experimental TMR (exp. 1 and 2)

	Treatment1
Item	U	SMnf	SM	SG
Ingredients, % DM
Sugarcane bagasse	13.0	13.0	13.0	13.0
Corn gluten feed	15.0	15.0	15.0	15.0
Dry-rolled corn	68.4	62.3	62.3	59.4
Urea	1.0
Soybean meal—non-ensiled		7.12
Soybean meal			7.1
Rolled soybean grain				10.0
Vitamin–mineral mix3	2.0	2.0	2.0	2.0
Limestone	0.6	0.6	0.6	0.6

¹U, ensiled with urea; SMnf, ensiled without a protein supplement at ensiling but soybean meal supplemented at feeding; SM, ensiled with soybean meal; SG, ensiled with rolled soybean grain.

²In exp. 1, soybean meal was omitted from SMnf TMR. In exp. 2, soybean meal was supplemented to SMnf TMR at feeding.

³Composition per kilogram: 160 g Ca, 64 mg Co, 800 mg Cu, 300 mg F, 48 mg I, 800 mg Mn, 24 g Mg, 110 g Na, 30 g P, 22 g S, 12 mg Se, 2,400 mg Zn, and 1,500 mg of monensin sodium.

Sixteen piles (four piles per treatment) were prepared individually, mixed manually, and packed in experimental silos (7-liters plastic buckets with lid) to achieve a density of 650 kg/m³ (as-fed basis). Fresh water was added onto the rations during mixing to adjust the DM content to 60%. Samples were collected from each pile to determine the fresh TMR composition. After 120 d of fermentation, the silos were weighed to determine fermentative loss and sampled to determine microbial counts, aerobic stability, and alterations in chemical composition caused by fermentation.

Aerobic stability test

Aerobic stability of TMR silages was evaluated in a room with controlled temperature (25 ± 2 °C) for 10 d. Fresh samples of TMR silages were placed in plastic buckets (3 kg), and the temperature was measured by a data logger placed in the center of the mass and programmed to collect temperature every 15 min (iMini, Impac Comercial e Tecnologia Ltda, São Paulo, Brazil). The ambient temperature was measured by two data loggers placed around the buckets. Aerobic stability was defined as the number of hours that the temperature of the silages remained stable before increasing more than 2 °C above ambient temperature (O′Kiely, 1993).

Laboratory analysis

Aqueous extracts of fresh and ensiled TMR were prepared by mixing 25 g of fresh (wet) sample and 225 g of distilled water for 2 min in a blender and filtering through cheesecloths. After measuring the pH (pH meter model Tec5, Tecnal, Piracicaba, Brazil), an aliquot was frozen at −20 °C for fermentation products analysis, and a second aliquot was diluted in ringer solution (10^–1 to 10^–6) and pour plated in selective media for microbial counts.

Malt extract agar (M137, Himedia, Mumbai, India) acidified to pH 3.5 with lactic acid was used for yeast and mold counts. Lactic acid bacteria (LAB) were counted in de Man, Rogosa, and Sharpe agar (7543A, Acumedia, Michigan, EUA). The plates were incubated aerobically at 30 °C for 2, 3, and 4 d for counting of LAB, yeasts, and molds, respectively. For Clostridium and aerobic spore counts, the diluted extracts (10^–1 to 10^–3) were pasteurized at 80 °C for 10 min. The medium used for Clostridium counts was reinforced Clostridial agar with the addition of neutral red and D-cycloserine (Jonsson, 1990). Aerobic spores were enumerated in the plate count agar. The Clostridia plates were placed in anaerobic jars and maintained in a biochemical oxygen demand incubator at 37 °C for 5 d. The aerobic spore plates were incubated in an incubator at 34 °C, and counting was performed after 2 d. The number of microorganisms was counted as colony-forming unit (cfu) and expressed as log₁₀.

After thawing, the aqueous extract was centrifuged at 10,000 × g for 15 min at 4 °C to obtain the supernatant, which was used to measure the concentrations of fermentation end products. Ammonia (Chaney and Marbach, 1962) and lactic acid (Pryce, 1969) concentrations were determined by colorimetric methods using a spectrophotometer (model Janway 6305, Marconi, Piracicaba, Brazil) with wavelengths of λ = 630 and 565 nm, respectively. Silage volatile fatty acids and alcohols were determined by gas chromatography-mass spectrometry (GCMS QP 2010 plus, Shimadzu, Kyoto, Japan) using a capillary column (Stabilwax, Restek, Bellefonte, PA; 60 m, 0.25 mm ø, 0.25 μm Crossbond carbowax polyethylene glycol). The injector was set to 220 °C and split mode injection (split ratio 20.0). The carrier gas was helium with flow and pressure adjusted to retain linear velocity (30.0 cm/s). The column oven temperature was held initially at 40.0 °C for 3 min, increased to 45.0 °C at 5.0 °C/min, then isothermal at 45 °C for 2 min, further increased to 60.0 °C at 5.0 °C/min, then isothermal at 60 °C for 3 min, further increased to 220 °C at 10.0 °C/min, and then held at 220 °C for 5 min. Ion source temperature was set to 200 °C and interface temperature to 240 °C. The mass spectrometer was set to record scans in the range of 31.0 to 250.0 m/z, every 0.30 s. Compounds were identified based on their retention time and mass spectra and quantified with external standards.

Fresh and ensiled TMR subsamples were also collected to evaluate the long-chain fatty acid profiles. The samples were dried in a freeze-dryer (Alpha 1–4 LD plus, Martin Christ Gefriertrocknungsanlagen GmbH, Osterode, Germany) and ground using a coffee grinder with a 0.3-mm sieve (DCG-20BKN, Cuisinart, Stamford, CT, USA). The fat was extracted according to Bligh and Dyer (1959). Fatty acid methyl esters were obtained in a solution of n-heptane and KOH/methanol (method 5509; ISO, 1978). Fatty acid methyl esters were separated using a gas chromatography (Trace Ultra 3300, Thermo Fisher Scientific, MA, USA) equipped with a flame ionization detector and an SP-2560 cyanopropyl capillary column (100 m × 0.25 mm i.d., 0.25-µm film thickness). Fatty acid methyl esters were identified by comparing the retention times with those of standard methyl esters and were quantified using tricosanoic acid methyl ester as an internal standard (Sigma-Aldrich, São Paulo, Brazil) as described by Joseph and Ackman (1992). Theoretical detector correction factors were used to calculate the fatty acid concentrations as described by Visentainer (2012).

Freeze-dried ground samples were analyzed for monensin by liquid chromatography-mass spectrometry at the Upscience Brasil-Labs Solutions, Hortolândia, SP, Brazil. Sample preparation and liquid chromatography-mass spectrometry methods were as described in detail by Blanchflower and Kennedy (1996).

Fresh TMR (day 0) and TMR silage (day 120) subsamples were dried in an air-forced oven at 55 °C for 72 h to determine the DM content and then ground using a Wiley mill with a 1-mm screen (Marconi MA340, Piracicaba, Brazil). After grinding, absolute DM was obtained by oven drying at 105 °C (method 934.01; AOAC, 1990). In dried samples of fresh TMR, the following parameters were determined: soluble carbohydrates using the phenol-sulfuric acid method (Hall, 2014), buffering capacity (Weissbach, 1967), fermentability coefficient (Weissbach et al., 1974), crude protein (CP) by Kjeldahl calculated as total N × 6.25 (method 984.13), and soluble CP (Licitra et al., 1996). The soluble CP concentration was also determined in dried samples of the TMR silages.

Experiment 2

Preparing and ensiling the TMR

The rations described in Table 1 were prepared at the Iguatemi Farm (23°21′13″S, 52°04′27″W; 550 m altitude) in May 2018. Before mixing the rations, corn and soybean grains were processed in a roller mill (SEGU 30, Multiagro Implementos Agrícolas, Porto Alegre, Brazil). The mean particle sizes were 2.66 ± 0.04 mm for rolled corn and 4.10 ± 0.23 mm for rolled soybean. All ingredients were mixed for 5 min in a mixer wagon (VMN 6.0 PA, Nogueira S/A Máquinas Agrícolas, São João da Boa Vista, Brazil) and packed in Ag-bag silos (1.8 m diameter, Pacifil, Sapiranga, Brazil) using a bagger machine (SEGU 30, Multiagro Implementos Agrícolas, Porto Alegre, Brazil). The treatments were identical to exp. 1, but soybean meal was supplied onto SMnf-TMR at feeding to balance diet CP. Similarly to exp. 1, fresh water was added during TMR mixing to adjust the DM content to 60%. All diets were formulated to meet or exceed the nutrient requirements of finishing Nellore heifers (NASEM, 2016), and CP content was adjusted to 13.0% of DM. The ensiled TMR were stored for 120 d before feeding.

Animals, facilities, and collections

Thirty-two Nellore heifers (313 ± 8.8 kg shrunk body weight [SBW], 24 mo old) were housed in individual pens with a concrete floor, feed bunk, and water bowl. The feeding period lasted for 82 d (first 21 d for adaptation to facilities and high proportion of concentrates in TMR silages, and 61 d for diet comparison). During the adaptation period, all heifers received a diet increasing in concentrate proportion over three periods of 7 d. From 0 to 7 d of adaptation, the diet contained (% DM) 40% corn silage, 22% corn gluten feed, 9.5% U-TMR silage, 9.5% SMnf-TMR silage, 9.5% SM-TMR silage, and 9.5% SG-TMR silage; from 8 to 14 d, 28% corn silage, 18% corn gluten feed, 13.5% U-TMR silage, 13.5% SMnf-TMR silage, 13.5% SM-TMR silage, and 13.5% SG-TMR silage; and from 15 to 21 d, 15% corn silage, 13% corn gluten feed, 18% U-TMR silage, 18% SMnf-TMR silage, 18% SM-TMR silage, and 18% SG-TMR silage. At the end of the adaptation period, animals were weighed after 16 h of fasting (during the night), blocked according to the SBW, and randomly assigned to the four dietary treatments (U, SM, SMnf, and SG). Every morning, silages were unloaded manually using a fork and offered once daily (0900 hours) in amounts of approximately 5% in excess of daily intake (as-fed basis). Feed refusals were collected and weighed daily to determine DM intake (DMI). Daily DMI variation was calculated as the difference between the DMI on the current day and the DMI on the previous day (Bevans et al., 2005). Offered feed and orts were composed by week to determine DM content and chemical composition.

The SBW was recorded at the beginning of the comparison period and every 28 d. The average daily gain (ADG) was determined as the slope of the SBW linear regression on days of diet comparison. Feed efficiency was computed as ADG/DMI. From the individual DMI and ADG data, diet net energy was estimated using the calculations described by Zinn and Shen (1998). Energy requirement for gain was calculated as: $\mathrm{E}\mathrm{g}\left(\mathrm{M}\mathrm{c}\mathrm{a}\mathrm{l}/\mathrm{d}\right)=(0.0608\times \mathrm{B}{\mathrm{W}}^{0.75})\times \mathrm{A}\mathrm{D}{\mathrm{G}}^{1.119}$, where BW is the mean SBW. Energy requirement for maintenance was calculated as: $\mathrm{E}\mathrm{m}(\mathrm{M}\mathrm{c}\mathrm{a}\mathrm{l}/\mathrm{d})=0.077\times \mathrm{B}{\mathrm{W}}^{0.75}$. Diet net energy for maintenance was estimated by the following equation: $\mathbf{N}\mathbf{E}\mathbf{m}(\mathrm{M}\mathrm{c}\mathrm{a}\mathrm{l}/\mathrm{k}\mathrm{g}\mathrm{D}\mathrm{M})=((-\mathrm{b}-{({\mathrm{b}}^2-4\mathrm{a}\mathrm{c})}^{0.5})/2\mathrm{a})$, where $\mathrm{a}=-0.877\times \mathrm{D}\mathrm{M}\mathrm{I},\mathrm{b}=((0.877\times \mathrm{E}\mathrm{m})+(0.41\times \mathrm{D}\mathrm{M}\mathrm{I})+\mathrm{E}\mathrm{g})$, and $\mathrm{c}=-0.41\times \mathrm{E}\mathrm{m}$. Diet net energy for gain was calculated as: $\mathbf{N}\mathbf{E}\mathbf{g}(\mathrm{M}\mathrm{c}\mathrm{a}\mathrm{l}/\mathrm{k}\mathrm{g}\mathrm{D}\mathrm{M})=(0.877\times \mathrm{N}\mathrm{E}\mathrm{m})-0.41$.

The particle size distribution of TMR silages and orts was determined using a Penn State particle separator (Kononoff et al., 2003) at 30-d interval, and the results were pooled. The sorting index was calculated by the observed intake of each fraction retained in each sieve expressed as a percentage of the predicted intake (as-fed basis; Leonardi and Armentano, 2003). Feeding behavior was evaluated on day 38 of the comparison. Animals were visually monitored for 24 h by eight trained evaluators. The criteria for assessing the behaviors were reviewed by all evaluators, and then they simultaneously recorded behaviors of a subset of heifers to ensure agreement beforehand. Eating and ruminating activities were recorded at 10-min interval, and the daily pattern was estimated assuming a constant behavior between the observations (Maekawa et al., 2002). Chewing activity was obtained as the sum of eating and rumination activities in minutes. The number of meals per day, meal size (obtained dividing DMI by number of meals), meal length (obtained by dividing eating time by number of meals), intermeal interval (obtained by dividing [1,440 − eating time] by number of meals), and intake rate (DMI in g divided by eating time) was also calculated.

Fecal score and pH were evaluated during four consecutive days from 45 to 48 d of the comparison. Four visual scores were considered for feces: 1 = runny: liquid consistency, splatters on impact, spreads readily; 2 = loose: may pile slightly and spreads and splatters moderately on impact and setting; 3 = soft: firm but not hard, piles but spreads slightly on impact and settling; and 4 = dry: hard, dry appearance, original form not distorted on impact and settling (Ireland-Perry and Stallings, 1993). For pH evaluation, fresh feces were collected every morning and evening during four consecutive days. A solution was made by mixing 15 g of feces and 100 mL of distilled water (Ireland-Perry and Stallings, 1993), and the pH was recorded after 2 min (pH meter model Tec5, Tecnal, Piracicaba, Brazil). The remaining samples of each sampling period were stored in separate plastic bags and then frozen at −20 °C. At the end of the collection period, samples within the same pen were composited on an equal wet weight basis, resulting in approximately 250 g of feces for DM content determination as described for feed samples in exp. 1.

On day 57 of comparison, blood samples were collected from the external jugular vein of each animal at 6 ± 1 h after feeding, into separate tubes without or with anticoagulant (K₃EDTA). After blood collection, the tubes were centrifuged at 4,000 × g for 15 min at 4 °C to obtain serum and plasma, respectively. Serum and plasma were stored in 2-mL microtubes at −80 °C until the analysis using commercial kits (Gold Analisa Diagnóstica Ltda, Belo Horizonte, Brazil). Serum was analyzed for total protein (Proteínas-Totais-PP Cat. 418), triglycerides (Triglicérides-PP Cat. 459), and cholesterol (Colesterol-PP Cat. 460), and plasma samples were analyzed for glucose (Glicose-PP Cat. 434) and urea (Uréia-PP Cat. 427).

Carcass traits

At the final weighing, carcass traits were evaluated by ultrasound (Aloka SSD500). Images were collected using a 17-cm, 3.5-MHz probe. Ribeye area and rib fat thickness were measured between the 12th and 13th rib transversally to the Longissimus muscle. Marbling score (1 to 10) was recorded from the 11th to 13th rib longitudinally to the Longissimus muscle. Biceps femoris fat thickness was also recorded. A single trained technician scanned all animals. Images were analyzed using Bia Pro Plus software (Designer Genes Technology). Afterward, animals were transported 135 km to a commercial abattoir and slaughtered according to animal welfare and slaughter practices established by the local sanitary inspection. The hot carcass weight (HCW) was recorded, and the dressing was calculated as HCW/SBW. Liver abscess score was evaluated according to incidence and severity, as described by Brink et al. (1990): 0 (without abscess), 1 (one or two abscesses with <2.5 cm or abscess scars), 2 (three to four abscesses <2.5 cm), and 3 (one abscess >2.5 cm or several small abscesses).

Protein degradability

Ruminal protein degradability was estimated by two methods: in situ assay and nitrogen (N) fractionation (Cornell Net Carbohydrate and Protein System [CNCPS] method). For the in situ assay, TMR silage samples were dried at 55 °C for 72 h and ground in a Wiley mill (Marconi MA340, Piracicaba, Brazil) with a 5-mm screen and weighed in woven bags (10 × 20 cm; 50 μm porosity; Ankon Technology, Macedon, NY, USA). Approximately, 5 g was placed in each bag, which were incubated in the rumen of two cannulated crossbred dry cows for 0, 12, 24, and 36 h. Bags were inserted in reverse order and recovered altogether. Immediately after removal, bags were submerged in cold water (0 °C) for 5 min and washed in a washing machine (three cycles followed by a final spin). Washed bags were dried in an air-forced oven at 55 °C for 72 h and weighed, and their contents were ground through a 1-mm screen using a Wiley Mill for measuring total N concentration (Kjeldahl method described above). Ruminal degradability was calculated using the first-order approach $\left[\mathrm{k}\mathrm{d}/(\mathrm{k}\mathrm{d}+\mathrm{k}\mathrm{p})\right]$. Fractional passage rates (liquid, concentrate, and forage) were estimated using DMI, dietary forage level, and SBW (Tylutki et al., 2008; NASEM, 2016) values. Forage and concentrate fractional passage rates were used to predict the solid fractional passage rate as follows: kp solids = kp forage × % forage in diet + kp concentrates × % concentrates in diet. Rumen-degraded protein (RDP) of each TMR was determined as RDP (%) = A (%) + B (%) × [kd/(kd + kp)], and rumen-undegraded protein (RUP) was determined as RUP (%) = 100 – RDP (%).

The protein degradability of TMR silages was also estimated based on N fractionation (CNCPS method). Ammonia concentration was determined as described in exp. 1 (Chaney and Marbach, 1962). Dry samples were ground to pass a 1-mm screen using a Wiley mill (Marconi MA340, Piracicaba, Brazil) to determine soluble protein, acid-detergent insoluble nitrogen, and neutral-detergent insoluble nitrogen contents using the methods described by Licitra et al. (1996). After these measures, N fractionation was determined using CNCPS v.6.5 (A1, A2, B1, B2, and C fractions; Van Amburgh et al., 2015). Ruminal degradability was calculated using the first-order approach $\left[\mathrm{k}\mathrm{d}/(\mathrm{k}\mathrm{d}+\mathrm{k}\mathrm{p})\right]$ (Van Amburgh et al., 2015). Fractional passage rates (liquid, concentrate, and forage) were estimated using DMI, dietary forage level, and SBW (Tylutki et al., 2008; NASEM, 2016) values. Forage and concentrate fractional passage rates were used to predict the solid fractional passage as follows: kp solids = kp forage × % forage in diet + kp concentrates × % concentrate in diet. Liquid fractional passage rate was considered for soluble fractions (A1 and A2), whereas the solid fractional passage rate was used for insoluble fractions (B1 and B2). The RDP of each TMR was determined as RDP (%) = A1 (%) × (kd/[kd + kp liquids]) + A2 (%) × (kd/[kd + kp liquids]) + B1 (%) × (kd/[kd + kp solids]) + B2 (%) × (kd/[kd + kp solids]). The RUP was calculated as $\mathrm{R}\mathrm{U}\mathrm{P}\left(\mathrm{\%}\right)=100-\mathrm{R}\mathrm{D}\mathrm{P}\left(\mathrm{\%}\right)$. The microbial protein and RUP fluxes were estimated based on the protein degradability data and energy calculations, and the protein requirements were calculated according to NASEM (2016).

Laboratory analysis

The procedures used to determine DM, CP, and soluble carbohydrates were the same as described in exp. 1. Ash concentration was determined by complete combustion in a muffle furnace at 600 °C for 5 h (method 942.05). Ether extract (EE) was quantified by the Soxhlet method (method 963.15) according to AOAC (1990). Neutral detergent fiber (NDF) was assayed with sodium sulfite and a heat-stable amylase (Mertens, 2002), and acid detergent fiber (ADF, Van Soest, 1967) was determined sequentially in a Fiber Analyzer (TE-149, Tecnal, Piracicaba, Brazil). The content of non-fiber carbohydrates was calculated as $\mathrm{N}\mathrm{F}\mathrm{C}=100-(\mathrm{C}\mathrm{P}+\mathrm{N}\mathrm{D}\mathrm{F}+\mathrm{E}\mathrm{E}+\mathrm{a}\mathrm{s}\mathrm{h})$.

Statistical analysis

Statistical analysis was performed using the Mixed procedure of SAS (v 9.4). Silage characteristics (exp. 1) were analyzed as a completely randomized design with the following model: ${{\mathrm{Y}}_{\mathrm{i}\mathrm{j}}}_{}=\mathrm{\mu }+{\mathrm{T}}_{\mathrm{i}}+{\epsilon }_{\mathrm{i}\mathrm{j}}$, where µ: overall mean, T_i: fixed effect of treatment (i = U, SMnf, SM, or SG), and ε _ij: residual error (j = 1 to 4). Animal performance data (exp. 2) were analyzed as a randomized complete block design with the following model: ${{\mathrm{Y}}_{\mathrm{i}\mathrm{j}}}_{}=\mathrm{\mu }+{\mathrm{B}}_{\mathrm{i}}+{\mathrm{T}}_{\mathrm{j}}+{\epsilon }_{\mathrm{i}\mathrm{j}}$, where µ: overall mean, B_i: random effect of block (i = 1 to 8), T_j: fixed effect of treatment (j = U, SMnf, SM, or SG), and ε _ij: residual error. In exp. 1 and exp. 2, means were compared by orthogonal contrasts. The contrasts included U vs. SMnf, SMnf vs. SM, and SM vs. SG. In exp. 2, the first contrast was defined to test whether cattle differ in performance when fed nonprotein nitrogen as urea or true protein from SM. The second contrast was defined to test whether the true protein can be ensiled. The last contrast was defined to test whether the animals respond to fat supplementation in a diet with a high proportion of ensiled corn grain proportion. Differences between treatments were declared if P ≤ 0.05 and trends considered if 0.05 < P ≤ 0.10.

Results

Experiment 1

The characteristics of the fresh and ensiled TMR are presented in Tables 2 and 3.

Table 2.

Characteristics of the fresh and ensiled TMR (exp. 1)

	Treatment1					P-value contrast
Item	U	SMnf	SM	SG	SEM	U vs. SMnf	SMnf vs. SM	SM vs. SG
Fresh TMR
DM2, % as fed	59.7	60.0	60.1	60.3	0.27	0.48	0.85	0.63
pH	5.24	5.13	5.25	6.24	0.033	0.03	0.02	<0.01
CP, % DM	13.1	10.1	13.3	13.2	0.12	<0.01	<0.01	0.55
Soluble CP, % CP	51.1	36.5	34.3	35.6	0.66	<0.01	0.04	0.24
Soluble carbohydrates, % DM	1.42	1.31	1.53	2.07	0.089	0.41	0.09	<0.01
Buffering capacity, g lactic acid/kg DM	23.1	18.0	22.8	21.0	0.40	<0.01	<0.01	<0.01
Fermentability coefficient3	60.5	60.7	60.7	61.3	0.25	0.59	0.93	0.14
LAB, log cfu/g as fed	4.07	3.96	4.12	4.16	0.091	0.39	0.25	0.76
Yeasts, log cfu/g as fed	3.81	3.76	3.77	3.66	0.050	0.53	0.93	0.16
Molds, log cfu/g as fed	3.95	3.86	4.00	3.98	0.116	0.57	0.40	0.87
Ensiled TMR
DM, % as fed	60.6	61.9	60.7	61.6	0.53	0.11	0.14	0.23
CP, % DM	13.2	10.6	13.5	13.3	0.16	<0.01	<0.01	0.38
Soluble CP, % CP	71.3	61.8	61.0	61.4	0.56	<0.01	0.33	0.61
NH3-N, % total N	26.7	8.57	8.38	9.37	0.81	<0.01	0.87	0.38
pH	4.72	4.24	4.23	4.22	0.060	<0.01	0.99	0.83
Lactic acid, % DM	3.70	2.45	2.36	2.67	0.413	0.09	0.90	0.69
Acetic acid, % DM	1.29	1.03	1.07	1.17	0.159	0.27	0.87	0.67
Butyric acid, % DM	<0.05	<0.05	<0.05	<0.05	—	—	—	—
Ethanol, % DM	0.55	0.71	0.74	0.41	0.076	0.14	0.77	<0.01
1,2-Propanediol, % DM	0.05	0.18	0.11	0.18	0.020	<0.01	0.02	0.02
2,3-Butanediol, % DM	0.11	0.06	0.07	0.09	0.023	0.16	0.84	0.54
LAB, log cfu/g as fed	5.59	4.47	4.55	4.65	0.574	0.21	0.92	0.91
Yeasts, log cfu/g as fed	3.37	3.24	3.50	3.43	0.200	0.65	0.38	0.80
Molds, log cfu/g as fed	<2.00	<2.00	<2.00	<2.00	—	—	—	—
Clostridia, log cfu/g as fed	3.77	2.45	3.07	3.19	0.363	0.03	0.26	0.82
Aerobic spores, log cfu/g as fed	3.97	3.62	3.66	3.61	0.262	0.36	0.91	0.87
DM loss, %	3.96	3.63	3.56	3.58	0.138	0.12	0.75	0.95
Aerobic stability, h	>240	>240	>240	>240	—	—	—	—

¹U, ensiled with urea; SMnf, ensiled without a protein supplement at ensiling but soybean meal supplemented at feeding; SM, ensiled with soybean meal; SG, ensiled with rolled soybean grain.

²DM content adjusted by addition of water.

³FC = DM (% as fed) + soluble carbohydrates (g/kg)/buffering capacity (g/kg).

Table 3.

Fatty acid profile and monensin concentration in fresh and ensiled TMR (exp. 1)

	Treatment1					P-value contrast
Item	U	SMnf	SM	SG	SEM	U vs. SMnf	SMnf vs. SM	SM vs. SG
Fresh TMR
Monensin, mg/kg DM	33.6	35.4	32.9	33.3	1.34	0.37	0.23	0.86
Fatty acids, % total fatty acids
C16:0	18.9	18.4	18.9	18.0	0.36	0.41	0.43	0.13
C18:0	5.21	5.40	4.71	5.01	0.242	0.59	0.08	0.40
C18:1 trans-9	2.29	2.30	1.76	1.74	0.283	0.98	0.21	0.97
C18:1 cis-9	25.7	25.8	25.3	23.7	0.17	0.66	0.09	<0.01
C18:2 cis-9,12	45.7	45.8	46.6	46.8	0.67	0.93	0.41	0.83
C18:3 cis-9,12,15	1.07	1.10	1.42	3.40	0.050	0.71	<0.01	<0.01
Other fatty acids	1.16	1.19	1.32	1.36	0.046	0.65	0.08	0.55
Ensiled TMR
Monensin, mg/kg DM	31.5	33.1	30.8	31.1	1.25	0.73	0.23	0.29
Fatty acids, % total fatty acids
C16:0	18.4	18.5	18.6	18.1	0.19	0.40	0.65	0.15
C18:0	4.33	4.42	4.35	5.02	0.231	0.97	0.83	0.11
C18:1 trans-9	1.84	1.48	1.42	1.69	0.235	0.24	0.86	0.54
C18:1 cis-9	26.7	25.3	24.4	23.8	0.24	<0.01	0.02	<0.01
C18:2 cis-9,12	46.3	48.1	48.5	46.9	0.53	0.02	0.63	0.15
C18:3 cis-9,12,15	1.24	1.29	1.59	3.61	0.127	0.09	0.13	<0.01
Other fatty acids	1.17	0.91	1.19	0.86	0.120	0.89	0.14	0.76

¹U, ensiled with urea; SMnf, ensiled without a protein supplement at ensiling but soybean meal supplemented at feeding; SM, ensiled with soybean meal; SG, ensiled with rolled soybean grain.

Comparison between nonprotein N and true protein as a protein source to TMR silage

In fresh TMR, U resulted in higher pH (P = 0.03), CP (P < 0.01), soluble CP (P < 0.01), and buffering capacity (P < 0.01) compared with SMnf. Concentrations of DM, soluble carbohydrates, fermentability coefficient, and counts of LAB, yeast, and molds were unaffected (P > 0.10) by protein source. Monensin concentration and proportions of fatty acids did not differ between U and SMnf treatments (P > 0.10).

In TMR silage, U resulted in higher pH (P < 0.01), CP (P < 0.01), soluble CP (P < 0.01), NH₃-N (P < 0.01), and lower 1,2 propanediol content (P < 0.01). Clostridium counts were higher (P = 0.03) and lactic acid tended to be higher (P = 0.09) in TMR formulated with U compared with SMnf. Contents of DM, acetic acid, butyric acid, ethanol, and 2,3-butanediol; counts of LAB, yeasts, molds, and aerobic spores; as well as DM loss and aerobic stability were similar between U and SMnf (P > 0.10). Similar to fresh TMR, monensin concentration was similar between U and SMnf silages (P > 0.10). Nevertheless, the proportion of C18:1 cis-9 was higher (P < 0.01) in U compared with SMnf. The proportion of C18:2 cis-9,12 was lower (P = 0.02), and the proportion of C18:3 cis-9,12,15 tended to be lower in U compared with SMnf (P = 0.09). The proportions of C16:0, C18:0, C18:1 trans-9, and other fatty acids remained unchanged due to protein source (P > 0.10).

Effect of ensiled and non-ensiled true protein source

Fresh TMR with SM had less soluble CP (P = 0.04), higher pH (P = 0.02), CP (P < 0.01), and buffering capacity (P < 0.01) and tended present higher content of soluble carbohydrates (P = 0.09) compared with SMnf treatment. The SM fresh TMR presented higher proportion of C18:3 cis-9,12,15 (P < 0.01) in comparison with SMnf. The proportion of C18:0 (P = 0.08) and C18:1 cis-9 (P = 0.09) had a trend to be lower, whereas other fatty acids (P = 0.08) had a trend to be higher in SM compared with SMnf. The DM content, fermentability coefficient, and counts of LAB, yeasts, and molds were not affected by protein source (P > 0.10). Monensin concentration and proportions of fatty acids C16:0, C18:1 trans-9, and C18:2 cis-9,12 were not changed between SMnf and SM fresh TMR (P > 0.10).

In TMR silages, CP was higher (P < 0.01) and 1,2-propanediol was lower (P = 0.02) in SM than in SMnf. A lower proportion of C18:1 cis-9 was found in SM compared with SMnf (P = 0.02). Values of pH, soluble CP, NH₃-N, lactic acid, acetic acid, butyric acid, ethanol and 2,3-butanediol, microbial counts (yeast, molds, LAB, Clostridia, and aerobic spores), aerobic stability, DM loss, monensin, and proportions of C16:0, C18:0, C18:1 trans-9, C18:2 cis-9,12, C18:3 cis-9,12,15, and other fatty acids were similar between SMnf and SM (P > 0.10).

Effect of lipid source

Fresh TMR with SG had higher pH (P < 0.01) and soluble carbohydrates (P < 0.01) and lower buffering capacity (P < 0.01) than SM. The SG treatment had a lower proportion of C18:1 cis-9 (P < 0.01) and greater proportion of C18:3 cis-9,12,15 (P < 0.01) than SM fresh TMR. Values of DM, CP, soluble CP, fermentability coefficient, and microbial counts (LAB, yeasts, and molds) were similar between SM and SG (P > 0.10). Monensin content and proportions of C16:0, C18:0, C18:1 trans-9, C18:2 cis-9,12, and other fatty acids did not change between SM and SG.

In TMR silages, SG as protein source resulted in lower concentration of ethanol (P < 0.01) and higher concentration of 1,2-propanediol (P = 0.02) than that in SG. Similarly to fresh ration, TMR silage with SG had lower proportion of C18:1 cis-9 (P < 0.01) but higher C18:3 cis-9,12,15 (P < 0.01) compared with SM treatment. Values of DM, CP, soluble CP, pH, NH₃-N, lactic acid, acetic acid, butyric acid, 2,3-butanediol, microbial counts (LAB, yeasts, molds, Clostridia, and aerobic spores), DM loss, and aerobic stability remained similar between SM and SG. Monensin content and proportions of C16:0, C18:0, C18:1 trans-9, C18:2 cis-9,12, and other fatty acids did not differ between SG and SM TMR silages.

Experiment 2

The chemical composition of offered TMR silages is described in Table 4, whereas animal performance data are presented in Tables 5–7.

Table 4.

Chemical composition of offered TMR silages (mean ± SD; exp. 2)

	Treatment1
Item	U	SMnf	SM	SG
DM, % as fed	60.6 ± 1.26	63.5 ± 1.75	60.7 ± 0.97	61.6 ± 1.36
CP	13.2 ± 0.42	13.5 ± 0.49	13.8 ± 0.25	13.5 ± 0.19
N fractionation2, % N
A1	26.7 ± 2.70	6.02 ± 0.57	8.37 ± 1.86	9.02 ± 1.41
A2	44.6 ± 3.02	41.3 ± 0.65	52.6 ± 1.78	52.4 ± 1.71
B1	22.2 ± 0.96	40.7 ± 0.96	31.7 ± 0.59	31.3 ± 0.65
B2	2.25 ± 0.27	6.35 ± 0.58	1.59 ± 0.58	1.81 ± 0.50
C	4.28 ± 0.35	5.60 ± 0.64	5.76 ± 0.73	5.49 ± 0.46
NDF	29.1 ± 2.77	29.7 ± 2.57	29.7 ± 2.31	30.8 ± 4.26
ADF	14.1 ± 2.00	14.4 ± 1.59	13.8 ± 2.04	15.6 ± 2.87
EE	3.72 ± 0.29	3.67 ± 0.24	3.54 ± 0.12	5.04 ± 0.15
Non-fiber carbohydrates	50.6 ± 2.46	48.2 ± 2.32	49.1 ± 2.21	46.0 ± 4.29
Soluble carbohydrates	1.19 ± 0.21	1.72 ± 0.08	1.34 ± 0.21	1.80 ± 0.10
Ash	4.54 ± 0.22	4.99 ± 0.30	4.81 ± 0.23	4.72 ± 0.13

¹U, ensiled with urea; SMnf, ensiled without a protein supplement at ensiling but soybean meal supplemented at feeding; SM, ensiled with soybean meal; SG, ensiled with rolled soybean grain.

²Nitrogen fractionation according to CNCPS (Higgs et al., 2015).

Table 5.

Performance and carcass traits of feedlot Nellore heifers fed TMR silages and calculations of net energy and TDN of TMR silages (exp. 2)

	Treatment1					P-value contrast
Item	U	SMnf	SM	SG	SEM	U vs. SMnf	SMnf vs. SM	SM vs. SG
Initial SBW, kg	318	313	311	311	8.8	0.70	0.90	0.99
Final SBW, kg	389	389	390	407	4.8	0.90	0.91	0.02
DMI, kg/d	7.91	8.18	8.02	9.43	0.449	0.67	0.79	0.03
Daily DMI variation, %	6.64	5.30	5.96	5.96	0.496	0.08	0.36	0.99
ADG, kg/d	1.20	1.23	1.23	1.49	0.075	0.81	0.97	0.02
Feed efficiency	0.153	0.150	0.154	0.160	0.0074	0.80	0.68	0.56
Dressing, %	53.9	54.8	54.7	54.1	0.45	0.17	0.89	0.29
HCW, kg	209	214	213	220	2.3	0.12	0.69	0.05
Backfat thickness, mm	4.56	5.95	5.03	6.20	0.565	0.11	0.26	0.14
Biceps femoris fat thickness, mm	6.56	7.82	7.65	9.34	0.469	0.09	0.80	0.01
Marbling score at 12th rib, 0 to 10	3.35	3.53	3.56	3.63	0.132	0.35	0.91	0.67
Longissimus muscle area at 12th rib, cm2	59.0	59.6	62.2	65.5	1.99	0.82	0.37	0.23
Liver abscess score, 0 to 3	0	0	0	0	—	—	—	—
Diet energy2
NEm, Mcal/kg DM	1.97	1.92	1.96	1.96	0.076	0.64	0.70	0.99
NEg, Mcal/kg DM	1.32	1.27	1.31	1.30	0.067	0.62	0.68	0.98
TDN3, % DM	81.1	79.4	80.8	80.7	2.39	0.64	0.69	0.98

¹U, ensiled with urea; SMnf, ensiled without a protein supplement at ensiling but soybean meal supplemented at feeding; SM, ensiled with soybean meal; SG, ensiled with rolled soybean grain.

²Calculated from animal performance data.

³Total digestible nutrients calculated from animal performance data.

Table 7.

Calculations of protein supply, using treatment means, in Nellore heifers fed TMR silages (exp. 2)

	Treatment1
	U	SMnf	SM	SG	U	SMnf	SM	SG
Item	CNCPS				In situ
Requirement of MP for maintenance1, g/d	308	309	309	314	308	309	309	314
Requirement of MP for gain1, g/d	280	297	292	331	280	297	292	331
Total MP requirement1, g/d	588	605	601	645	588	605	601	645
Microbial CP2, g/d	606	612	614	681	606	612	614	681
Microbial MP2, g/d	388	392	393	436	388	392	393	436
RDP supply, g/d	808	791	808	930	921	806	936	1,059
RDP balance, g/d	+202	+179	+194	+249	+315	+194	+322	+378
RDP balance, %	+33.4	+29.2	+31.6	+36.5	+52.0	+31.6	+52.5	+55.5
RUP supply, g/d	236	314	299	343	123	299	170	214
MP from RUP, g/d	183	243	231	266	95	231	132	165
Total MP supply, g/d	570	634	624	702	483	623	525	601
Total MP balance, g/d	−17	+29	+23	+56	−105	+18	−76	−44
Total MP balance, %	−3.0	+4.8	+3.9	+8.7	−17.8	+2.9	−12.6	−6.8

¹U, ensiled with urea; SMnf, ensiled without a protein supplement at ensiling but soybean meal supplemented at feeding; SM, ensiled with soybean meal; SG, ensiled with rolled soybean grain.

²Requirements of MP for maintenance and gain, and microbial CP and MP were calculated according to NASEM (2016).

Table 6.

Feeding behavior, fecal characteristics, and blood parameters of feedlot Nellore heifers fed TMR silages (exp. 2)

	Treatment1					P-value contrast
Item	U	SMnf	SM	SG	SEM	U vs. SMnf	SMnf vs. SM	SM vs. SG
Eating, min/d	222	189	186	209	33.6	0.52	0.94	0.62
Ruminating, min/d	275	249	253	303	26.6	0.52	0.93	0.17
Chewing, min/d	497	439	439	512	39.0	0.32	0.99	0.18
Meals, /d	9.80	9.00	9.00	10.6	0.58	0.41	0.97	0.04
Meal size, kg DM/meal	0.808	0.942	0.906	0.875	0.0618	0.16	0.69	0.71
Meal length, min/meal	22.2	20.4	20.2	20.2	3.02	0.68	0.97	0.99
Intermeal interval, min	127	143	145	119	10.6	0.32	0.90	0.09
Intake rate, g DM/min	46.4	57.8	47.8	45.0	7.75	0.33	0.38	0.79
Particle sorting index, % as fed
8 to 19 mm	106	105	105	105	0.5	0.14	0.55	0.49
1.18 to 8 mm	101	100	100	100	0.2	0.38	0.61	0.90
Pan	96.6	97.3	96.6	96.6	0.50	0.34	0.32	0.99
Fecal score (1 to 4)	2.94	2.91	2.86	2.96	0.059	0.78	0.55	0.26
Fecal DM	23.5	23.9	24.2	23.7	0.50	0.58	0.69	0.43
Fecal pH	6.34	6.27	6.27	6.07	0.077	0.53	0.95	0.08
Glucose, mg/dL	62.6	63.9	65.7	62.3	3.26	0.78	0.71	0.46
Urea, mg/dL	37.4	38.6	38.7	36.9	1.61	0.61	0.99	0.43
Total protein, g/dL	7.05	7.60	7.18	7.31	0.127	<0.01	0.03	0.48
Triglycerides, mg/dL	17.6	20.2	20.0	25.1	1.87	0.35	0.93	0.05
Cholesterol, mg/dL	172	183	193	256	13.0	0.59	0.58	<0.01

¹U, ensiled with urea; SMnf, ensiled without a protein supplement at ensiling but soybean meal supplemented at feeding; SM, ensiled with soybean meal; SG, ensiled with rolled soybean grain.

Chemical composition of offered TMR silages

Offered TMR silages had similar values of DM, CP, NDF, ADF, NFC, soluble carbohydrates, and ash. The TMR with SG had greater content of EE. Nitrogen fractionation also differed among offered TMR silages. The main changes were observed in the TMR with U, which had more A1 and less B1 fraction than other TMR. The SMnf, which was supplied with soybean meal at feeding, had a greater proportion of B2 fraction.

Comparison between nonprotein N and true protein as a protein source to TMR silage

Protein source (U or SMnf) did not affect (P > 0.10) initial and final SBW, DMI, ADG, feed efficiency, HCW, dressing, backfat thickness, marbling score, Longissimus muscle area, liver abscess score, and diet energy concentration (NEm, NEg, or total digestible nutrient). Feeding behavior variables (eating, ruminating, chewing, meals per day, meal size, meal length, intermeal interval, and intake rate), particle sorting, fecal score, DM and pH, and blood concentrations of glucose, urea, triglycerides, and cholesterol were unaffected by protein source (P > 0.10).

Daily intake variation tended (P = 0.08) to be greater and Biceps femoris fat thickness tended to be lower (P = 0.09) in animals fed U compared with SMnf treatment. The concentration of total protein was greater (P < 0.01) in the blood of heifers fed SMnf than U treatment. Protein calculations indicated a lower supply of RUP and total MP in U than in SMnf treatment. The balance of total MP was negative in U but positive in SMnf treatment.

Effect of ensiled and non-ensiled true protein source

Animal performance, carcass traits, liver abscess score, diet energy, feeding behavior, sorting index, fecal pH, DM and score, and blood parameters were similar among heifers fed SMnf or SM (P > 0.10). Animals fed SMnf TMR had a higher concentration of total protein in blood compared with animals fed SM TMR (P = 0.03). Calculations of protein supply based on in situ method indicated a greater supply of RUP and total MP in SMnf than in SM, but no difference was observed when the calculations were based on the CNCPS model.

Effect of lipid source

Initial SBW, daily intake variation, feed efficiency, carcass dressing, backfat thickness, marbling score, Longissimus muscle area, liver abscess score, diet energy concentration, eating and ruminating times, meal size, meal length, particle sorting index, fecal score, fecal DM, and blood glucose, urea, and total protein were similar (P > 0.10) between SM and SG treatments. However, compared with SM, the SG treatment increased DMI (P = 0.03), final SBW (P = 0.02), ADG (P = 0.02), HCW (P = 0.05), and Biceps femoris fat thickness (P = 0.01). Cattle fed SG made more meals per day (P = 0.04) and tended to have a decreased intermeal interval (P = 0.09). Fecal pH tended to decrease (P = 0.08) in SG compared with SM. Heifers fed SG TMR silage had greater blood concentrations of triglycerides (P = 0.05) and cholesterol (P < 0.01). Protein calculations revealed a greater supply of microbial protein and total MP in SG than in SM treatment. The balance of total MP was negative in both SM and SG based on the in situ method but positive in both treatments based on the CNCPS model calculations.

Discussion

Ensiling complete rations modifies nutrient composition and availability (Bueno et al., 2020). In this study, we examined the conservation and feed value of TMR silages containing different protein supplements for finishing heifers.

In exp. 1, all TMR silages were well conserved and stable upon air exposure (>240 h of aerobic stability). Despite the low levels of soluble carbohydrates (<2.1% DM), the relatively high DM content (60%) associated with a low buffering capacity and a moderate initial LAB counts in fresh rations led to TMR silages with low DM loss during fermentation (<4%), confirming that TMR silages are generally easy to ensile (Weinberg et al., 2011; Miyaji et al., 2017; Restelatto et al. 2019a, 2019b).

Lactic acid was the most abundant fermentation end product, indicating satisfactory LAB development, which certainly constrained DM losses during anaerobic storage (McDonald et al., 1991). Acetic acid was the second acid found in the TMR silages, and its strong antifungal activity potentially helped to protect the TMR silages from fungal spoilage during feed out (Nishino et al., 2003; Nishino and Hattori, 2007; Wang and Nishino, 2008; Restelatto et al., 2019a). All alcohols evaluated in our study (ethanol, 1,2-propanediol, and 2,3-butanediol) were found in low concentrations, which is similar to data reported by previous publications and consistent with the low DM loss in all TMR silages (Nishino et al., 2003, 2004; Wang and Nishino, 2008). In our study, the yeast counts were relatively low, and mold counts were below the detection limit, which potentially contributed to the high aerobic stability of all TMR silages. The prolonged storage period might have contributed to the high aerobic stability in our TMR silages (Wang et al., 2016; Restelatto et al., 2019a).

As expected, the TMR silage formulated with urea presented the greatest ammonia concentration (+204%), higher pH value (+13%), and more Clostridia (+21%) compared with other treatments. Urea is prone to hydrolysis by urease enzyme during silage fermentation (Lessard et al., 1978). Taken together, this information suggests that conversion of urea to ammonia might have increased the buffering capacity at the onset of fermentation resulting in an increased pH value of this treatment compared with the other treatments. Nevertheless, in addition to the greater number of Clostridium spores, their development was likely curtailed due to the relatively high DM content as indicated by the very low levels of butyric acid. Moreover, the content of lactic acid tended to be increased in U-TMR silage probably due to the increase in buffering capacity, which often stimulates lactic acid formation in silages (Owens et al., 1969; Huber and Santana, 1972; Custódio et al., 2016).

During ensiling, changes in long-chain fatty acid profile may occur and are mainly driven by lipase and lipoxygenase activities when pH is still high (>5) during the onset of fermentation (Liu et al., 2018). In our trial, however, changes in fatty acids were minimal. Overall, fatty acid profiles were similar among treatments and between fresh and ensiled TMR. The immediate sealing of the experimental silos certainly provided rapid oxygen depletion and acidification, inhibiting the oxidation and saturation of fatty acids.

The fate of feed additives during ensiling is seldom reported in the literature. In our study, the concentrations of monensin were slightly reduced in TMR silages (by −6%) compared with fresh TMR, without a marked difference among treatments. This finding suggests that supplemented monensin was partially metabolized during silage fermentation; however, the final concentrations were within recommended levels for finishing beef cattle (NASEM, 2016). Despite these results, further research is warranted to verify whether the ingredients and silo management affect the recovery of feed additives in TMR silages.

In exp. 2, the treatments provided different profiles of N fractionation and MP supply. However, contrary to our hypothesis, higher MP supply did not improve animal performance. Indeed, few numerical positive responses were noted when the finishing heifers were supplemented with a true protein source compared with urea, as discussed below.

Several authors reported lower DMI in cattle fed silages containing higher NH₃-N content (Rook and Gill, 1990; Cushnahan et al., 1995; Huhtanen et al., 2002). In our trial, the TMR silage containing urea tended to increase DMI variation (+20%) compared with the TMR silage supplemented with SMnf; however, DMI was similar between U and SMnf. Galyean et al. (1992) reported lower ADG and feed efficiency in steers due to a daily intake variation of 10%, whereas Soto-Navarro et al. (2000) and Zinn (1994) did not find differences in steers performance due to a daily intake variation of 10% and 20%, respectively. Soto-Navarro et al. (2000) also concluded that the animals may become adapted to a 10% intake variation, which is a greater threshold than the variation observed in our trial (from 5.30% to 6.64%). In our study, the change in intake variation was not sufficient to affect ADG, HCW, and backfat thickness.

Blood total protein was greater in animals fed SMnf compared with TMR ensiled with SM or U, suggesting a greater MP supply by the SMnf-TMR than that provided by TMR silages containing ensiled soybean meal or urea. This finding was consistent with the predicted flux of MP. Nevertheless, the ensiled soybean meal (SM-TRM silage) did not change animal performance and carcass traits compared with SMnf-TMR. Hence, soybean meal may be ensiled with other ingredients and at least partially replaced with urea when preparing TMR silage for finishing cattle.

In our study, animal performance was markedly improved when the soybean meal was replaced with soybean grain, a combined source of protein and lipids. Feeding supplemental fat to finishing cattle typically improves ADG and feed efficiency (Zinn, 1988, 1989; Brandt and Anderson, 1990) or only feed efficiency (Krehbiel et al., 1995; Zinn and Shen, 1996; Ramirez and Zinn, 2000) because the fat has a NEg value that is approximately 3-fold increased compared with corn (NASEM, 2016). Nonetheless, in our trial, the SG-TRM silage did not alter the feed efficiency and the estimated concentration of diet NEg, indicating that the higher ADG (+21%) was mainly a result of increased DMI in heifers fed SG-TMR silage.

Several studies report similar or lower DMI for cattle fed high-grain diets containing supplemental fat (Zinn, 1989, 1992; Krehbiel et al., 1995; Ramirez and Zinn, 2000), but other studies reported higher DMI for cattle fed diets supplemented with fat (Silva et al., 2007, 2018; Jacovaci, 2019). Often, fat supplementation stimulates cholecystokinin and glucagon-like peptide 1 secretion and decreases ghrelin and ruminal motility (Allen, 2000; Bradford et al., 2008). Those physiological changes are generally related to fat decreasing DMI. However, such effects seem dependent on diet concentration of fat. Haaland et al. (1981) reported a curve linear response to fat supplementation, with higher DMI, ADG, and feed efficiency for the intermediate dose of supplemental fat (~5% of DM). In our study, the inclusion of 10% of SG increased diet EE from ~3.5% to ~5%, and DMI by 18% because the animals made more meals (10.6 vs. 9.0/d) without altering meal size.

The motive for the higher DMI induced by SG remains unknown, but two mechanisms can be hypothesized. First, SG may have increased TMR palatability compared with SM. On the other hand, the fat present in SG might have altered liver signaling to satiety (Allen, 2000). In high-grain diets with a high proportion of ensiled corn as in the current trial, propionate is likely the major anaplerotic metabolite of the tricarboxylic acid cycle within meals. Propionate stimulates oxidation of the existing pool of acetyl-CoA increasing hepatic energy charge and generating a satiety signal according to the hepatic oxidation theory (Allen et al., 2009). However, other metabolites such as lactate (originated from endogenous metabolism and rumen fermentation) become a greater contributor to anaplerosis of the tricarboxylic acid cycle between meals and delay hunger, as propionate production from the previous meal subsides (Allen and Bradford, 2012). Meanwhile, supplemental fat could prevent the hypophagic effects of other fuels such as lactate (Allen, 2000; Allen and Bradford, 2012). Lactate must be converted to pyruvate to be metabolized and fat supplementation may decrease the activity of pyruvate dehydrogenase (Begum et al., 1982), which should inhibit pyruvate oxidation in favor of carboxylation to oxaloacetate. If supplemental fat decreases the conversion of pyruvate to acetyl-CoA, then hunger might be stimulated sooner. In our study, SG tended to decrease intermeal interval compared with SM; hence, the hyperphagic effect of SG could be from a similar mechanism.

The higher DMI for SG was also associated with slightly lower fecal pH compared with SM treatment (−3.2%), and this effect is possibly due to an increase in fermentable organic matter passage to the hindgut (Wheeler and Noller, 1977). However, the absence of liver abscess and the similarity of fecal scores among all treatments (values close to 3 or normal consistency) suggest the absence of digestive disturbances.

The higher ADG promoted by SG led to an improvement in HCW given that the dressing percentage was similar among treatments. Changes in other carcass traits were less consistent. Zinn (1992) observed increased HCW, dressing percentage, Longissimus muscle area, KPH, and backfat thickness numerically higher for steers fed steam-flaked corn or steam-flaked wheat-based diets containing 6% of supplemental fat, such as yellow grease or cottonseed oil. In our study, besides the higher ADG promoted by SG compared with SM, backfat thickness and marbling score were similar among treatments. Brandt and Anderson (1990) fed steers flaked milo-based diets containing no added fat or 3.5% supplemental fat, such as soybean oil, tallow, or yellow grease. They reported increased HCW only in steers supplemented with yellow grease and increased dressing percentage due to supplemental fat regardless of the source. Andrae et al. (2001) reported that steers fed high oil corn-based diets exhibited increased marbling scores compared with those fed a conventional corn-based diet despite the lack of difference in backfat thickness. A similar result was reported by Gillis et al. (2004) in heifers fed dry-rolled corn-based diets supplemented with 4% of corn oil compared with the heifers fed the control diet (no added fat). However, the studies cited above were conducted with Bos taurus breeds, while the animals in our study were Nellore heifers (Bos indicus), which have leaner meat with lower marbling scores (Marshall, 1994). On the other hand, the Biceps femoris fat thickness was greater (+18%) for cattle fed SG-TMR silage compared with SM-TMR silage. Perhaps, an extended finishing period would result in more benefits in carcass traits of heifers fed SG.

However, SG-TMR silage led to an increase in blood triglycerides (+20.3%) and cholesterol (+24.6%) concentrations compared with SM treatment, which is expected when fat sources are included in diets (Bauchart, 1993). Our results were consistent with other studies evaluating fat supplementation for beef cattle (Bindel et al., 2000; Belal et al., 2017).

The calculated fluxes of N fractions revealed that all TMR silages provided a positive balance of RDP; however, the RDP values assessed by the in situ method were increased compared with those estimated by CNCPS. The limitations related to the in situ technique are well known (Michalet-Doreau and Ould-Bah, 1992), but the major problem is associated with fraction A (soluble fraction), which is assumed to be completely degraded in the rumen (Ørskov and McDonald, 1979). Fraction A includes not only the rapidly degraded nonprotein nitrogen (e.g., ammonia and free amino acids) but also soluble proteins, which are not instantaneously degraded and small undegraded particles that escape through the bag pores. Studies have shown that the breakdown rates of soluble proteins are highly variable (Broderick et al., 1988; Hedqvist and Udén, 2006). Using an in vitro technique, Peltekova and Broderick (1996) estimated that 20% of the soluble protein in alfalfa silage escaped rumen digestion, whereas Ahvenjärvi et al. (2018) reported that 12.5% of soluble N in timothy silage escaped rumen degradation in the liquid phase. These findings suggest an overestimation of fraction A by the in situ method; thus, RDP values estimated using in situ method in our study are consequently increased. Although replacing soybean meal with urea did not affect animal performance, based on protein supply calculations, treatments containing true protein supplements met or exceeded the requirements, whereas U-TMR silage led to a slightly negative MP balance. Hence, lower levels of urea inclusion in TMR silage than that used in the current study (<1% DM) would be recommended. Past studies that examined the inclusion of urea in corn silages led to the conclusion that levels of approximately 0.5% to 0.6% of DM would be optimum (Huber et al., 1968; Polan et al., 1968). Regarding TMR silages, more studies are warranted to define and predict the pattern of proteolysis for a given ingredient or ingredient combinations during both ensiling and rumen fermentation. These findings will ultimately allow the MP requirements of the animals to be adequately met.

Conclusions

The TMR silages showed adequate fermentation pattern, low fermentative loss, and high aerobic stability. Supplementation with soybean meal (true protein source) at feeding did not improve the performance of finishing Nellore heifers fed TMR silage. If used in finishing diets, the true protein source can be ensiled with other ingredients as a TMR silage. The inclusion of soybean grain as a protein and lipid source improved the performance of finishing beef heifers compared with the inclusion of soybean meal.

Glossary

Abbreviations

ADF

acid detergent fiber

ADG

average daily gain

CP

crude protein

DM

dry matter

DMI

dry matter intake

EE

ether extract

HCW

hot carcass weight

LAB

lactic acid bacteria

MP

metabolizable protein

NDF

neutral detergent fiber

NEg

net energy for gain

NEm

net energy for maintenance

NFC

non-fiber carbohydrates

RDP

rumen-degraded protein

RUP

rumen-undegraded protein

SBW

shrunk body weight

TDN

total digestible nutrients

TMR

total mixed ration

Conflict of interest statement

The authors declare no real or perceived conflicts of interest.