Nine feedstuffs were evaluated in this study. Two cereal grains (corn and oat), two wheat by-products (wheat bran and distilled wheat), four protein meals (coconut meal, sunflower meal, soybean meal and rapeseed meal) and muiumba seeds were analysed. Feeds were chosen in order to cover a wide range in chemical composition and in vitro digestibility. Mature muiumba seeds were collected in the semiarid region of the Cunene basin (Angola) during the dry season. Samples were dried in a forced-air oven (Venticell, Munchen, Germany) at 60°C for 24 h for dry matter (DM) determination. Dried samples were ground to pass 1 mm screen (Retsch, Cutting mill, model SM1, Haan, Germany) and stored in airtight flasks at room temperature for subsequent laboratorial analysis. For starch and non-starch polysaccharides (NSP) analysis, samples were ground to pass a 0.5 mm screen (Cyclotec, Tecator, Hoganas Sweden).
Samples were analysed in quadruplicate using the procedures described by AOAC, Association of Official Analytical Chemists (1990) for ash (#942.05), crude protein (#954.01) and ether extract (#920.39). Neutral detergent fibre (NDF) was determined by the detergent procedures of Van Soest et al. (1991). Sodium sulphite was not added and heat stable amylase (Sigma A3306, St. Louis, Ill, USA) was only added to cereal grains and wheat by-products samples. Klason lignin (KL) was determined using a two stage sulphuric acid hydrolysis according to the method proposed by Theander and Westerlund (1986).
Low molecular weight sugars (LMW) were analysed according to the methods of Knudsen (1997) with minor modifications. Samples of approximately 500 mg were weighed into 50 mL centrifuge tubes. Fifteen mL of an ethanol solution (500 mL/L), including 2 mL of an internal standard (arabinose, 1 mg/mL), were added and the samples sonicated and extracted in a water bath for 60 min at 65°C. During extraction the tubes were mixed (vortex mixer) three times and finally centrifuged (2200 g, 20 min). An aliquot of 1 mL was removed from the supernatant and diluted with 5 mL of water and analysed by anion-exchange chromatography with pulsed amperometric detection (HPAEC-PAD). The HPAEC system used consisted of the model ICS-5000 (Thermo Fisher Scientific Inc., Dionex, USA) equipped with an ED40 electrochemical detector (Thermo Fisher Scientific Inc., Dionex, USA) with a working gold electrode. Low molecular weight sugars were analysed by isocratic elution with 0.05 M NaOH containing 0.002 M of Ba(OH)2 at 0.5 mL/min with a CarboPac PA-20 column (Thermo Fisher Scientific Inc., Dionex, USA). Sugar composition from the total and insoluble non-starch polysaccharides (NSP) were analysed by isocratic elution with 0.018 M NaOH containing 0.002 M of Ba(OH)2 at 1 mL/min with a CarboPac PA-1 column (Thermo Fisher Scientific Inc., Dionex, USA). The injection volume was 10 μL, and the column and detector temperature was set at 25°C. All individual sugar values were converted to their equivalent polysaccharide value using appropriate conversion factors; 0.88 for pentoses and deoxyhexoses and 0.9 for other hexoses.
The starch content was determined according to the enzymatic method proposed by Salomonsson et al. (1984), with amyloglucosidase from Aspergillus niger (Sigma, A3042; 6000 U/ml) and heat stable amylase (Sigma A3306, Ill) using a spectrometer (U2000, Hitachi Ltd., Japan).
Total, soluble, and insoluble NSP were determined using the Knudsen (1997) procedures with minor modifications. Briefly, samples were weighed into 50 mL centrifuge tubes. Acetate-buffer with CaCl2 (0.1 M; 0.02 M; pH 5.0; 9.8 ml), termostable α-amylase (Termamyl, Novo Nordisk A/S.; 100 μL) was added and the samples were incubated for 1 h at 100°C. Complete degradation of starch was achieved after 2 h with β-glucanase free amyloglucosidase from Aspergillus niger (Boehringer Mannheim GmbH, Cat No. 1202 367:135 U/ml; 100 μL) at 60°C. Soluble NSP were precipitated with 4 volumes of an ethanol solution (990 mL/L) for 1 h on ice bath, the tubes centrifuged (4600 rpm; 10 min) and the supernatant discarded. Residues were washed twice with an ethanol solution (850 mL/L) and once with acetone and dried in a hood overnight. Polysaccharides in starch free residues were treated with 12 M H2SO4 (35°C 60 min) and hydrolysed to monosaccharides with 1 M (100°C. 120 min). An internal standard (2-deoxyglucose 1 mg/mL; 500 μL) was added to the hydrolysate aliquot. Neutral sugar analysis was performed by HPAEC-PAD (section 2.2.1.). The uronic acids were measured spectrophotometrically with D-galacturonic acid solutions as standards according to the method described by Scott (1979).
The soluble NSP in the starch-free residue were extracted by a phosphate buffer (Englyst et al. 1982) at neutral pH (0.2 M; 100°C; 60 min; pH 7.0) and the neutral and acidic sugars in the insoluble NSP were analysed as described for total NSP.
Two cows, weighing 500 ± 36 kg live weight and fitted with permanent ruminal cannulae (Bar Diamond Inc., Parma ID, USA), were used to collect rumen fluid. Cows were housed in individual stalls in a ventilated barn. The diet consisted of meadow hay and commercial concentrate making 75:25 ratio on a DM basis and was offered at 1.2 times the maintenance requirements (Thomas, 2004). Animals were fed in equal portions at 0800 and 1600 h and had free access to water and mineral-vitamin blocks. The concentrate was offered to the animals prior to the hay. Rumen fluid samples were drawn 2 h after the morning feed into pre-warmed insulated flasks previously filled with CO2. Rumen fluid was strained through 4 layers of cheesecloth and kept at 39°C under CO2.
In vitro organic matter digestibility (IVOMD) was determined according to Tilley and Terry (1963) modified by Marten and Barnes (1980). Each sample was incubated in quadruplicate in one series. Fermentations were in 50 mL centrifuge tubes containing 250 mg of sample and 25 ml of buffered rumen fluid solution. Immediately after addition of buffer solution, flasks were closed with rubber stoppers fitted to a Bunsen valve. After 48 h of incubation a pepsin/HCl solution was added and the incubation was prolonged by 24 h.
Rumen fluid used to measure in vitro gas production was mixed (1:2 v/v) with an anaerobic buffer/mineral solution (Cone et al. 1996). All laboratory handling was under continuous flushing with CO2. Samples (500 mg) were accurately weighed into 250 mL serum bottles (Schott, Mainz, Germany) and incubated in 60 ml buffered rumen fluid. Each sample was incubated in two series, completed on different days. Gas production was recorded every 20 min for 72 h using a fully automated system (Cone et al. 1996).
Calculations and statistical analysis
Gas curves were fitted by iteration to a mono-phasic model as described by Groot et al. (1996) as:
where: A = estimated asymptotic gas production; B = time of incubation at which half of the asymptotic gas production has been formed; C = sharpness of the switching characteristic for the profile.
Maximum rate of gas production (R
G) and the time at which maximum rate of gas production is reached (TR
G) were calculated according to Yang et al. (2005) as:
Fractional rate of substrate fermentation (R) was calculated using the equation of Groot et al. (1996) as:
All data were expressed in terms of ml of gas/g OM fermented.
In vitro organic matter digestibility and in vitro gas production parameters were analysed according to a completely randomized design using GLM of SAS (SAS, 2004). Differences between treatment means were determined using Tukey’s test with a predetermined significance level of P < 0.05.