Animals, experimental design and treatments
All animal care and handling procedures were approved by the Animal Care and Use Committee of the Universidade Federal de Viçosa, Brazil (protocol CEUAP-UFV 31/17). Animals used in this study were provided by the Beef Cattle Farm of Animal Science Department at the University Federal de Viçosa, Viçosa-MG, Brazil, where the study as carried out, from July to December 2017. After the study, all cows were maintained on Beef Cattle Farm herd.
Thirty-eight pregnant multiparous Nellore cows, with an average body weight (BW) of 515±11 kg, body conditions scores (BCS) of 5.5±0.25 and 230±10 gestation days were used. Animals were randomly divided into eight paddocks with seven hectares each, evenly covered with Urochloa decumbens grass, with free access to water and feeders.
The experimental design was completely randomized, with two treatments as following: NS-control; SS-cows supplemented for the 60 pre-partum days (gestation period from 230 to 290-d). The NS cows, received only a mineral mixture (MM) as loose mesh, ad libitum, during gestation. SS cows were group-fed with 90 kg of supplement during the pre-partum period (1.5 kg/d), accompanied by MM offered ad libitum supplied separately in additional feeders. The compositions of supplement, MM and pasture are shown in Tables 1 and 2. Treatments were randomly assigned to paddocks: six paddocks with five cows each and two with four, totalizing 19 cows per treatment. Feeders were equipped with creep-feeding and sheltered, with space of 0.3 m per cow.
The supplement was a loose mesh formulated to contain 30% crude protein (CP) as fed to meet around 40% of CP maintenance requirements, according to BR-Corte . Supplement was always provided at 11:00h to minimize any interference of animal grazing behavior . After calving, cows remained in the same paddocks, but received only MM ad libitum until 45 lactation days.
Experimental procedures and sampling
Cows were weighed on two consecutive days at the beginning of the experiment (60-d pre-partum), and 7-d before the expected calving day to quantify the average daily gain pre-calving (ADGpre). Cows were weighed after calving and at the end of the experiment period also on two consecutive days (45-d) to quantify the average daily gain post-calving (ADGpost). Calves remained with dams during the experiment and were weighed immediately after birth, and also at 45 and 90-d. Body condition scores (BCS) were also recorded on a scale from 1 to 9, as recommended by NRC , by three experienced persons at the beginning of the experiment, upon calving and 45-d post-partum.
During the breeding season, starting on December 12, cows were synchronized, and fixed time artificial insemination (FTAI) was performed on December 23. Pregnancy diagnosis was made via transrectal ultrasonography 30-d after FTAI. The number of days from parturition to re-conception was calculated for each cow and pregnancy rate.
Every 30-d, grass samples were collected by hand-plucked sampling to evaluate the forage selected by animals. Samples were collected by cutting at the ground level from five delimited areas (0.5 x 0.5 m), selected randomly in each paddock to quantify DM and DMpd. In these circumstances, all the samples were weighed, oven-dried (55°C) and then ground to pass through 1- and 2-mm screens in a Wiley mill (model 3, Arthur H. Thomas, Philadelphia, USA).
Intake and digestibility assay
To evaluate intake and digestibility, a trial was run for 9-d on day 45 before the estimated parturition date (around 245-d of gestation). Titanium dioxide (TiO2) was used to estimate the fecal excretion of animals, which was wrapped in paper cartridges (20 g per animal/day) and inserted with a metal probe via the esophagus at 12:00h . The first 5 trial days were used to adapt animals to TiO2. Fecal samples were collected immediately after defecation or directly from the rectum on the last 4 days (one sample/day) at 18:00h, 14:00h, 10:00h and 06:00h. Feces samples were over-dried (55°C) and ground to pass through 1- and 2-mm screens in a Wiley mill (model 3, Arthur H.Thomas, Philadelphia, USA). Then 25 g from all 4 days were pooled.
Indigestible neutral detergent fiber (iNDF) was used to estimate pasture dry matter intake (DMI) . It was assumed that supplement consumption equaled the amount offered per animal/day.
On trial day 5, forage was collected by hand-plucked sampling at each paddock separately. These samples were used to estimate voluntary dry matter intake and forage digestibility.
On the trial last day, spot urine samples (5 mL) were collected 4 h before and after administering the supplement. Then a 10-mL compound was prepared with 5 mL of the urine collected in the morning and afternoon. Urine samples were diluted in 40 mL of H2SO4 (0.036 N) and then frozen (-20°C).
On 30- and 45-d post-calving, milking was performed to estimate milk production. In order to empty udders, calves were separated from their mothers from 15:00h to 17:45h, when they were reunited to dams and allowed to suckle. At 18:00h, calves were once again separated from dams until the next morning. At 06:00h on the next day, cows were milked immediately after an injection of 20 UI of oxytocin (10 UI/mL; Ocitovet®, Brazil) in the mammary vein and the produced milk was weighed. The exact time when each cow was milked was recorded. Calves were kept away from their mothers until the next milking at 06:00 to obtain a 24-hour milk production. Next 30 mL of milk were separated from each cow to evaluate milk composition. Total production was corrected to 4% fat, according to NRC .
By taking calving day as day 0, blood samples were collected before feeding on days -30, 0, 15, 30, 45. Blood samples were collected by jugular vein punctur, using vacuum tubes with a clot activator and gel for serum separation (BD Vacutainer® SST® II Advance®, São Paulo, Brazil) to quantity: blood nitrogen urea, total protein, albumin, triglycerides, total cholesterol, high density lipoprotein (HDL), nonesterified fatty acid (NEFA), beta-hydroxybutyrate (βHB), insulin, insulin-like growth factor (IGF-1), total triiodothyronine (T3), total thyroxine (T4), progesterone (P4) contents (only on 30 and 45-d). A tube with EDTA and sodium fluoride (BD Vacutainer® Fluorinated/EDTA, São Paulo, Brazil) was used to quantity the plasma glucose concentration. After collection, samples were centrifuged at 3600 × g for 20 min. Serum and plasma were immediately frozen at -20°C until analyzed.
The forage, feces and supplement samples were analyzed following the procedures described by Brazilian National Institute of Science and Technology in Animal Science (INCT-CA)  for: dry matter (DM; index INCT-CA method G-003/1), ash (index INCT-CA method M-001/1), crude protein (CP; index INCT-CA method N-001/1), neutral detergent fiber corrected for ash and protein (apNDF; index INCT-CA method F-002/1). Indigestible neutral detergent fiber (iNDF)  was processed at 2 mm and quantified by in situ incubation procedures with nonwoven textile bags (100 g/m²) for 288 h. The fecal samples were evaluated for titanium contents using acid digestion process with concentrated sulfuric acid followed by the addition of 30% hydrogen peroxide solution and further quantification by spectrophotometry (INCT-CA method M-007/1).
With the blood samples, Bioclin® kits (Belo Horizonte, Brazil) was employed to quantity urea (K056), total protein (K031), albumin (K031), triglycerides (K117), total cholesterol (K083), HDL (K071) and glucose (K082). NEFA and βHB were analyzed using Randox® kits (FA115 and RB1007, Antrim, UK). Uric acid, creatinine and urea in urine were analyzed with kits Bioclin® (K0139, K067 and K056, Belo Horizonte, Brazil). All the above-mentioned analyses were determined by an automated biochemical analyzer (Mindray, BS200E, Shenzhen, China). Allantoin in urine was analyzed by the colorimetric method .
Insulin, Total T3, Total T4 and progesterone contents were analyzed by kits Beckman (33410, 33830, 33800 and, 33550 Beckman Coulter®, Brea, USA). IGF-1 contents were quantified with kits DiaSorin® (California, USA) in an automated chemiluminescence analyzer (Liaison®, Saluggia, Italy). Milk was analyzed for protein, fat, lactose and total solids content using infrared spectroscopy (Foss MilkoScan FT120, São Paulo, Brazil).
(See Calculations and Statistical analyses in the Supplementary Files)