Trial 1 aimed to determine a suitable level of PBB to replace the concentrate using an in vitro gas test. The concentrate mixture was replaced by PBB at levels of 0, 10%, 20%, 30%, or 40%. The ratio of forage to concentrate was 50:50 (w/w) in the mixed substrate with Chinese wild ryegrass hay (CRH) as the only forage, with a mixture of corn meal and soybean meal (SBM) (60:40, w/w) as the concentrate. The PBB used in the trial was a pelletized product obtained from Zhejiang Province in China. A total of 200 mg substrate dry matter (DM) was used in all treatments. Six syringes (duplicates) were incubated for a treatment simultaneously, with 3 samples used for the gas test and the others for determining pH and the content of volatile fatty acid (VFA) and ammonia nitrogen after 24 h incubation.

Based on the chemical composition of PBB and the results from Trial 1, a lactation trial (Trial 2) was designed using 24 multiparous Holstein cows (day in milk [DIM] = 170.4±35; milk yield = 30±3 kg/d; body weight [BW] = 580±13 kg). The animals were divided into 12 blocks based on DIM and milk yield and randomly allocated to two dietary treatments: basic diet with or without PBB replacing 20% of the concentrate mixture. The cows were housed in a tie-stall barn, and fed and milked at 06:30, 14:30, and 21:00 h. All diets contained 50% forage, which was 23% corn silage, 11% alfalfa pellet, 13% CRH, and 3% beet pulp. All animals had free access to drinking water. The ingredients and composition of the concentrate in the experimental diets are presented in Table 1.

The GP was determined using the technique of Menke and Steingass (1988) using 100-mL calibrated glass syringes (Model Fortuna, Haberle Labortechnik, Oberer Seesteig 7, DE-89173 Lonsee-Ettlenschieβ, Germany). About 200 mg of each substrate were introduced into the syringes fitted with plungers. The syringes were filled with 30 mL of a medium comprising 10 mL of rumen fluid and 20 mL of buffer solution as described by Menke and Steingass (1988). The rumen fluid was collected from three donor sheep fed twice daily on a mixture of CRH and concentrate (70:30, w/w) before morning feeding, then mixed, squeezed through four layers of gauze and kept in a water bath at 39°C with CO₂ saturation. All laboratory handling of the rumen fluid was carried out under continuous flushing with CO₂. In vitro fermentation studies were conducted in triplicate. During each incubation run, three blanks were included simultaneously to correct the GP values for gas release from endogenous substrates. The glass syringes were immediately placed in an electrically heated isothermal oven equipped with a rotor at 39°C, which was rolling continuously at 1 rpm. The GP value was recorded at 2, 4, 6, 9, 12, 24, 36, and 48 h of incubation. To describe the GP dynamics over time, the following equation was used to fit the data to the model described by Ørskov et al. (1980): GP = a+b (1−e^−ct), where, GP = the cumulative GP (mL) at time t (h), a = the GP from the immediately soluble fraction (mL), b = the GP from the insoluble fraction (mL), (a+b) = the potential GP (mL), c = the GP rate constant (%/h) and a, b, and c are constants.

The ammonia-N, pH and VFA content of the fluids were determined after incubation for 24 h by using methods described by Hu et al. (2005).

The experiment lasted for 8 weeks, following a week for the cows to adapt to the diets. Milk sampling devices (Waikato Milking Systems NZ Ltd., Hamilton, New Zealand) were attached to the milking machines to measure milk weight and collect samples. Milk production was recorded for all three milking times. A 50-mL aliquot of milk was collected weekly at each milking, proportional to yield (4:3:3, composite). The composited milk sample, with added Bronopol tablets (milk preservative, D & F Control Systems, San Ramon, CA, USA), was stored at 4°C for later analysis of fat, protein and lactose by infrared analysis (Laporte and Paquin, 1999) using a four-channel spectrophotometer (Milk-o-Scan, Foss Electric A/S, Hillerød, Denmark).

The PBB was ground to pass through a 1-mm sieve for subsequent analysis. The DM contents were determined according to method No. 942.05 (AOAC, 1997). The samples were analyzed for crude protein (CP) (method 988.05), acid detergent fiber (ADF) (method 973.18, AOAC, 1997) and neutral detergent fiber (NDF) by the method of Van Soest et al. (1991). Amylase, but not sulfite was used in the determination of NDF. Both NDF and ADF were expressed exclusive of residual ash. Samples of SBM, CRH and corn meal were milled to pass through a 1-mm screen and then stored for later determination of chemical composition as described above.

The in situ disappearance of DM and CP for SBM, CRH and corn meal was determined using three ruminally cannulated sheep (BW = 35±5 kg) housed in individual stalls. The basal diet (% of DM) consisted of 70% CRH and 30% concentrate mixture fed twice daily at a level to meet the requirement for 1.3 times maintenance. The samples were ground to pass through a 4-mm screen in a mill (Thomas-Wiley Laboratory Mill; Arthur H. Thomas, Philadelphia, PA, USA). The nylon bags (8×12 cm; 40-mesh pore size) containing 5 g of each sample were tied to the end of a 14-cm nylon line and then placed in the ventral sac of the rumen through the ruminal cannula to incubate for 24 h. After removal from the rumen, the bags were rinsed thoroughly in cool running tap water until the wash water ran clear. The samples were dried in an oven at 65°C for 48 h and weighed to determine the residue weight. The residues and original diet samples were ground to pass through a 1-mm screen in a Cyclotec mill (Tecator 1093; Tecator AB, Höganäs, Sweden) before analysis of DM and CP. The in situ degradation of DM and CP was calculated based on the differences in weight between the residues and original diet samples.

The effects of PBB on GP, fermentation parameters, milk yield and milk composition were analyzed using the one-way analysis of variance, with themeans compared using Duncan’s new multiple range test at a level of significance of 0.05 (SAS, 2000)..

Table 2 presents the results for chemical composition and in sacco degradability of PBB, SBM, corn meal and CRH. The CP content in PBB was 21.6% (DM basis), higher than that of corn meal (9.2%), but lower than SBM (47.9%). The contents of NDF (47.1%) and ADF (38.3%) of PBB were higher than those of corn (14.3% and 6.7%) and SBM (14.0% and 10.7%), but lower than CRH (70.2% and 62.6%). The mean disappearance of DM in the rumen was 88.2% in PBB, higher than that of corn meal (72.6%), but lower than that of SBM (94.6%). This shows that PBB is a highly digestible feed resource with a CP level comparable with that of alfalfa hay (MoA, 2004). The calculated net energy for lactation (NE_L) content was lower than that of corn, much higher than that of CRH, but comparable with that of SBM. The basic diet with PBB replacing 20% of the concentrate mixture may meet the requirement of NE_L for milk production at about 20 to 30 kg/d (MoA, 2004).

As shown in Table 3, with an increasing proportion of PBB in the concentrate, the GP and potential GP values of the mixed substrates decreased, but decrease of GP and potential GP were only significant from the replacement level of 40% and 30% (p<0.05) respectively. Gas production gives a measure of the degradation of substrate, particularly the carbohydrate fraction (Menke et al., 1979). These results have revealed that replacing concentrate with PBB for ruminants might not affect digestion if the replacement level was at less than 30%.

Maintaining a stable rumen environment is critical. In this experiment, there was no difference (p>0.05) in pH among the five replacement levels (Table 3). The concentration of ammonia-N was significantly lower at the 10% replacement level of PBB in the mixed substrate than at other levels. However, it was still within a suitable range of ammonia-N concentrations (>5 mg/dL) to ensure maximum microbial growth in vitro (Satter and Slyter, 1974) and that required for optimal fiber digestion (Kennedy et al., 1992). The concentration of total VFA was enhanced significantly by replacing PBB at levels of 30% to 40%, indicating that the availability of energy from PBB was adequate (Srinivas and Gupta, 1997). Replacing PBB at a level of 40% enhanced the molar proportion of acetate (p<0.05) and depressed the proportion of propionate (p<0.05). This change may have a negative effect on energy use because propionate increases energy use efficiency by decreasing energy losses on inter-species hydrogen transfer and methane production in the rumen (Van Houtert, 1993). Based on the above results (both chemical composition and in vitro degradation), a 20% replacement level of concentrate by PBB was chosen for the dairy cows’ diet.

As shown in Table 4, milk yield was not affected when PBB was included in the dairy cows’ diet at a level of 20% (24.2 vs 23.5 kg). However, including PBB significantly increased the percentage of milk fat (p<0.05), but had no significant effects on the percentages of milk protein, lactose, total solids and solids-not-fat (p>0.05).

The higher fat content may be attributed to the higher content of NDF and ADF (Table 2) and the slightly higher proportion of acetate (Table 3) in PBB compared with the other concentrate ingredients such as corn and SBM. This is because the contents of acetate and butyrate have a positive correlation with milk fat concentration while propionate has a negative correlation (Sutton et al., 1993, 1998). Eastridge et al. (2009) found that an increase in ADF results in a higher milk fat content and thus a higher fat-corrected milk yield for cows fed straw. In the present study, the intake of fibrous materials should be higher in the diet containing PBB than in the control diet containing no PBB.