Six wethers (19.30±1.17 kg body weight (BW), mean±standard deviation), fitted with ruminal fistula, were assigned to two groups for cross-over design with a 14-d interval. Each experimental period lasted 30 d, consisting of 27 d of adaptation and 3 d for sample collection. Goats were fed a basal diet without (control) or with GO infusion. Based on the dose of GO in lamb research (0.01 g/kg BW) (Chaves et al., 2008) and our previous study in Saanen dairy goats (0.01, 0.02 and 0.03 g/kg BW) (Zhu, 2011), the dose 0.04 g/kg BW was used in the present research. Accordingly, the dose of GO for each goat was 0.8 g/d. The basal diet was formulated (Table 1) according to Feeding Standard of Meat-producing Sheep and Goats (NY/Y 816-2004; Ministry of Agriculture, China, 2004) in order to fulfill requirements for the growth of goats. Diets for restricted intake consisted of grass hay (Leymus chinensis, 0.38 kg/d dry matter (DM)) and concentrate (0.22 kg/d DM). Once offering concentrate to the goat, GO product was infused directly and immediately by a 1-ml syringe via ruminal fistula. For each goat, the infusing lasted a few seconds. The ingestion of concentrate was completed simultaneously for each goat. Afterwards, hay was offered. Diet and GO infusion were offered in equal portions at 08:00 and 18:00 h. All goats had free access to fresh water throughout the experiment. The commercial GO (stored at 4°C; Henan Yuanhua Biotechnology Co., Ltd., China) used in the experiment was produced by steam distillation (150°C) which was used to collect the products that evaporate with steam. The main components of GO were sulfur compounds, such as the predominant diallyl trisulfide (29.3%) and diallyl disulfide (31.3%). No FA was detected in GO used in this study. Goats were fed individually in pens and managed in accordance with guidelines regarding animal care provided by the College of Animal Science and Technology in Nanjing Agricultural University.

Feed intake for individual goats was recorded on the last 3 d of each experimental period. Feed samples were dried at 65°C for 24 h in a forced-air oven, ground through a 1 mm screen, and analyzed for DM, crude ash content, ether extract and crude protein using standard procedures (AOAC, 1990). The concentrations of acid detergent fiber (ADF) and neutral detergent fiber (NDF) were determined by sequential procedures (pretreatment with α-amylase for NDF) (Van Soest et al., 1991), and the results were expressed inclusive of residual ash.

Ruminal contents from multiple sites within the rumen of each animal were collected by suction via ruminal fistula before (0 h) and at 2, 4, 6, 8, and 10 h after morning feeding on the last 3 d of each period. The pH value was measured immediately using a portable pH meter (HI 8424; Hanna, Singapore). Then ruminal contents were collected into tubes. The 0-h samples were collected in triplicate. The tubes were capped and placed in an ice bucket and immediately transferred to the laboratory. Duplicates of 0-h samples were immediately stored at −25°C for latter DNA extraction and FA analysis. The others were vortexed for 1 min, and then centrifuged at 1,000×g for 10 min. A 1-ml sample of the supernatant was acidified with 1 ml of 0.2 mol/L HCl and frozen at −25°C for latter ammonia nitrogen (NH₃-N) analysis by spectrophotometer (UV-2450; Shimadzu, Japan) (Broderick and Kang, 1980). One ml of the supernatant was mixed with 0.2 ml 25% (w/v) metaphosphoric acid including crotonic acid (64.64%, w/v) as an internal standard and frozen at −25°C for latter VFA analysis by gas chromatograph (GC-14B; Shimadzu, Japan) equipped with capillary column (Supelco 41491-03B, 30 m×0.32 mm×0.25 μm film thickness; Bellefonte, USA) (Qin, 1982). One ml of the supernatant was stored at −25°C for microbial crude protein (MCP) analysis by spectrophotometer (UV-2450; Shimadzu, Japan) (Makkar et al., 1982).

Ruminal contents collected at 0 h on d 28, 29, and 30 were used for DNA extraction. Total DNA was extracted from 1.5 ml homogenized ruminal contents by bead-beating for 3 min using a mini-bead beater (Biospec Products, USA), followed by phenol-chloroform extraction (Zoetendal et al., 1998). DNA was then precipitated with ethanol and pellet was resuspended in 70 μl of Tris-HCl/EDTA (TE) buffer.

Ruminal contents collected at 0 h on d 28, 29, and 30 were used for FA determination. Lipid extraction was performed as described by Folch et al. (1957) with some modifications. Briefly, the homogenized ruminal contents (10 g) were mixed with 20 ml chloroform/methanol (2/1, v/v). After dispersion, the whole mixture was agitated for 20 min in an orbital shaker at room temperature. The homogenate was centrifuged (1,200×g for 10 min) to recover the liquid phase. Then the liquid phase was washed with 5 ml 0.9% NaCl solution. After vortexing 2 min, the mixture was centrifuged at 1,200×g for 10 min to separate the two phases. The upper phase was removed by siphoning, the lower chloroform phase containing lipid was evaporated at 50°C under nitrogen. Methylation of FA was performed with 2 ml 0.5 mol/L methanolic KOH, followed by rocking (about 3 min) in a water bath at 50°C under nitrogen until the oil beads disappeared. The flasks were cooled and then 2 ml 14% boron trifluoride/methanol was added. Afterwards, the flasks were incubated in a water bath at 50°C under nitrogen for 3 min. After cooling, one ml n-heptane was added followed by vortexing. The upper organic phase was rinsed with saturated NaCl solution, and then transferred to 2-ml centrifuge tube containing anhydrous sodium sulfate.

Fatty acid methyl esters were analyzed according to the reported methods (Lu et al., 2005; Lou et al., 2010) with some modifications using a gas chromatography (6890N series; Agilent Technology, USA) equipped with mass spectrometry (5973 series; Agilent Technology, USA). Fatty acid separation was performed on a fused silica capillary column (HP-5MS, 100 m×0.25 mm×0.20 μm; Supelco, Bellefonte, USA). Ultrapure helium was used as carrier gas, the injector pressure was held constant at 43.9 kPa. Analysis of FA required injection of 1 μl with split ratio of 10:1 at a constant flow rate of 1 ml/min by carrier gas. Injector temperature was maintained at 280°C. Initial column temperature was 80°C, increased to 220°C at a rate of 6°C/min (held for 13 min), and then raised at 10°C/min to a final temperature of 280°C (held for 15 min). The electron-impact ionization mass spectrometer was operated as follows: ionization voltage, 70 eV; ion source temperature, 230°C; mass scan range, 29 to 450. Mass spectral identification of FA was carried out by comparing spectra with those in the NIST98 (National Institute of Standards and Technology, USA). Results for each FA were expressed as a percentage of the sum of all identified FA.

Real-time PCR was performed on an ABI 7300 system (Applied Biosystems, USA) associated with 7300 Sequence Detection Systems Software (version 2.2.1; Applied Biosystems, USA). The primers described below were synthesized and the PCR reagents were provided by Sangon (Sangon Biotech Co., Ltd., China).

According to the method from Boeckaert et al. (2008), 16S rRNA gene of the Butyrivibrio group of bacteria was quantified by Butyrivibrio-specific forward and reverse primers B395f (5′-GYG AAG AAG TAT TTC GGT AT-3′) and B812r (5′-CCA ACA CCT AGT ATT CAT C-3′), respectively. Quantitative PCR (qPCR) reaction mixture (20 μl) consisted of 10 μl of 2×FastStart Universal SYBR Green Master (Roche Applied Science, Germany), 0.6 μl forward primer (10 μmol/L), 0.6 μl reverse primer (10 μmol/L), 2 μl DNA and 6.8 μl PCR-grade H₂O. Quantitative PCR amplification was performed using the following conditions: 1 cycle of 2 min at 50°C, 10 min at 95°C, 40 cycles of 15 s at 95°C, 30 s at 55°C, and 45 s at 72°C. Quantification of 16S rRNA gene copies in each sample was performed in triplicate, and the mean value was calculated. The Butyrivibrio group bacterial 418-bp PCR product amplified by specific primers (Boeckaert et al., 2 0 08) from the strain B. fibrisolvens DSM 3071 was used as a template for standard curves. Standard curves were constructed using at least five different 10-fold serial dilutions of PCR product ranging from 9.82 copies to 9.82×10⁷ copies of product per μl. The efficiency of amplification was calculated as follows: efficiency = (10^(−1/slope)-1). The bacterial 16S rRNA gene copies were plotted against the C_T value used to estimate the number of copies in each sample. PCR reaction mixture (50 μl) consisted of 1 μl DNA extract, 1 μl of each primer (10 μmol/L), 1 μl deoxynucleotide triphosphate mixture (10 mmol/L each), 5 μl 10×PCR amplification buffer (50 mmol/L KCl, 10 mmol/L Tris-HCl pH 8.5, 0.1% Triton X-100), 6 μl MgCl₂ (25 mmol/L), 0.25 μl Taq polymerase (5 U/μl) and lastly PCR-grade H₂O. PCR amplification was carried out in Eppendorf Mastercycler^® Gradient (Eppendorf, Germany), with an initial DNA denaturation step at 94°C for 5 min, followed by 34 cycles of denaturation at 94°C for 30 s, annealing at 55°C for 30 s, and extension at 72°C for 40 s, and final extension at 72°C for 7 min. The amplicons were measured in triplicate using a NanoDrop-1000 spectrophotometer (Labtech International, UK) and analyzed by electrophoresis on a 1.5% agarose gel (w/v) containing ethidium bromide.

Quantification of B. proteoclasticus was conducted according to Paillard et al. (2007) using forward (SA-FW, 5′-TCC GGT GGT ATG AGA TGG GC-3′) and reverse (SA-RV, 5′-GTC GCT GCA TCA GAG TTT CCT-3′) primers. Reaction mixture (20 μl) consisted of 10 μl of 2×FastStart Universal Probe Master (Roche Applied Science, Germany), 0.8 μl forward primer SA-FW (10 μmol/L), 1.6 μl reverse primer SA-RV (10 μmol/L), 0.5 μl molecular beacon probe (5′-FAM-CCG CTT GGC CGT CCG ACC TCT CAG TCC GAG CGG-DABCYL-3′) (Invitrogen, China) (10 μmol/L), 2 μl DNA and 5.1 μl PCR-grade H₂O. The qPCR amplification program was the same as for the Butyrivibrio group bacteria. Dilutions of PCR product (1,506 bp) amplified by forward 8f (5′-AGA GTT TGA TCC TGG CTC AG-3′) and reverse 1510r (5′-GGC TAC CTT GTT ACG A-3′) primers (Ding and Yokota, 2004) from 16S rRNA gene clone A23 (accession No. HQ326602) were used for constructing specific standard curves. The 10-fold serial dilutions of PCR product ranged from 6.06 copies to 6.06×10⁷ copies of product per μl. The PCR amplification mixture of clone A23 was the same as for B. fibrisolvens DSM 3071. Amplification conditions were as follows: 94°C for 3 min, 30 cycles of 94°C for 30 s, 57°C for 30 s and 68°C for 1.5 min, and a final extension of 68°C for 7 min. The amplicons were measured and analyzed as above.

Data were analyzed using PROC MIXED of SAS (version 8.01; SAS Institute, Inc., USA) (SAS, 2000) according to the model Y_ijk = μ+G_i+P_j+T_k+e_ijk, where Y_ijk was the response variable, μ was the overall mean, G_i was the random effect of goat i, P_j was the effect of period j, T_k was the effect of treatment k, and e_ijk was the random residual error. For the statistical analysis of ruminal fermentation characteristics (pH, NH₃-N, MCP, and VFA), sampling time (H_r) and interaction of treatment with sampling time (T×H)_kr were added to the model, and analyzed using repeated measures of PROC MIXED with the covariance type auto-regressive order 1 (AR (1)). The effect of sampling time was repeated measures. Means were separated by using the PDIFF option in the LSMEAS statement. Pearson correlation was analyzed using PROC CORR procedure to investigate the relationship between microbial abundance and FA proportion in the rumen. Significances were determined at p<0.05; differences of p<0.1 were discussed as trend.

No refusals of concentrate in the two groups were observed. The DM intake of concentrate for each goat was 0.22 kg/d. Garlic oil had no influence on the intake of hay (0.33 kg/d for control and 0.32 kg/d for GO group).

The effects of GO on ruminal fermentation are shown in Table 2. A slight reduction (p<0.05) in ruminal pH was found in GO group (6.49) as compared with control (6.55). There were increases (p<0.05) in the concentrations of NH₃-N and MCP upon GO infusion. Garlic oil had no influence on total VFA concentration, molar proportions of individual VFA, or ratio of acetate to propionate. There were interactions between treatment and sampling time for total VFA concentration, propionate molar proportion and ratio of acetate to propionate (p<0.05). The temporal changes showed that total VFA concentration and propionate molar proportion reached maxima at 2 h after morning feeding in control, whereas the maxima occurred at 6 h in GO group (Figure 1). The ratio of acetate to propionate in control reduced sharply after morning feeding, whereas it remained relatively stable in GO group (Figure 1).

Details regarding FA proportion are summarized in Table 3. Generally, GO had no effects on C4 to C14 FA proportions, while reducing C15 and C16 FA proportions (p<0.05) and increasing the proportions of C18 FA, TVA, c9,t11-CLA and t10,c12-CLA (p<0.05) in the rumen. Garlic oil reduced saturated FA (SFA) proportion (p<0.05), whereas GO increased polyunsaturated FA (PUFA) proportion (p<0.05) and tended to increase monounsaturated FA (MUFA) proportion (p = 0.076). The values of TVA/ (c9,t11-CLA+TVA) and C18:0/(TVA+C18:0) were reduced by GO (p<0.05).

The abundance of Butyrivibrio group bacteria and B. proteoclasticus in ruminal contents is listed in Table 4. Quantitative PCR efficiency for all assays was between 90% and 110% (data not shown). Correlations of DNA standards for all real time PCRs were >0.97 (data not shown). Results showed that GO had no influence on the Butyrivibrio group bacterial abundance, whereas GO tended to reduce B. proteoclasticus abundance (p = 0.058). Correlation analysis showed that there was a low correlation between B. proteoclasticus abundance and C18:0/(TVA+C18:0) (r = 0.02, p = 0.910).