The source of the animals used in experiments were acquired by the Committee of Animal Experimentation and were performed under the institutional guidelines for animal experiments of the College of Animal Science, Inner Mongolian Agricultural University, China. The experiment was performed according to recommendations proposed by the European Commission (1997) to minimize the suffering of animals.After the experimental procedure, the experimental sheep will not be killed, and they will continue to provide help for the future research work and provide suitable living conditions for them in the future life.
Diet and experimental design
The CS and CA were both harvested from an Agricultural Field of Xiwuzhumuqin town, Xilinhot, China. We had permission to use this field for our study, and all experiments were conducted in accordance with local legislation.The CS at maturity stage was obtained at 5 d of field exposure after corn harvest, and the stovers were chopped into approximately 2 cm by a chopper (TS420; Qufu Machinery Equipment Co., Ltd., China). The CA at growth stage was obtained after 7 d of harvest, and the small branches and leaves were crushed to 4 mm with a crusher (FW100, Guangming Machine Co., Ltd, Beijing, China). The mixing ratio and chemical composition of CS and CA are shown in Table 1.
Based on the feeding methods used on the Mongolian Plateau, the diets in this study were prepared from CS, CA, corn meal, sunflower cake, soybean cake, flaxseed cake, calcium bicarbonate, salt, and mineral-vitamin premix (Inner Mongolia Zheng Da, Hohhot, China). The treatment diets contained similar concentrate mixtures and fine feed, but different pro-portions of CS and CA. The treatments were (1) corn stover (CS); (2) caragana (CA); (3) CS 80% + CA 20% (MT 1); (4) CS 50% + CA 50% (MT 2); (5) CS 20% + CA 80% (MT 3). The ration ingredient composition is shown in Table 2. The Table 2 contained the corn stover treatment (CST), caragana treatment(CAT), MT1, MT2 and MT3. The roughage in the diet consisted of 20.66–82.64% CS on a DM basis, while the concentrate consisted of 9.97% corn grain, 0.99% sunflower cake, 1.66% Soybean cake, and 3.30% Flaxseed cake. The diet contained calcium bicarbonate, vitamin premix and mineral premix. The Mineral-vitamin premix composition were 2.4×104 mg/kg Fe; 7×103 mg/kg Cu, 3×104 mg/kg Zn; 1.25×104 mg/kg Mn; 30 mg/kg I, 850 mg/kg Co, 5×108 IU/kg Vitamin A and Vitamin D3, 1×108 IU/kg. The diet was formulated to meet the nutritional requirements of Mongolian sheep . Five diets were formulated to meet the animals’ requirements for energy, protein, minerals, and vitamins based on Cornell-Penn-Miner 
Sixty healthy Aerbasi cashmere goats (BW = 30 ± 2.35 kg; mean ± SD) were used in a randomly experiment. Groups were balanced for parity and production during each period. The diet were assigned randomly to one of 5 groups (12 sheep/group) in a completely randomized design. They were penned individually in a well-ventilated shed that was cleaned weekly. Before starting the trial, all goats were vaccinated against clostridial diseases and treated for internal parasites. The diet was provided with an allowable refusal level of 10% DM of the diet. The animals were fed the diets three times a day at 8:00, 13:00 and 19:00, and clean drinking water was available at all times. After a 10 days adaptation period, the 45 days growth trial was started. Feed offered and feed residues were weighed daily along the growth trial to calculate the net intake of feed. The refused feed was collected daily, weighed, and individually bulked for analysis. Samples of the daily refusals from each animal and feed offered were dried at 105°C for 24 hours to determine the DM intake.
When the growth trial experiment was completed, the apparent nutrient digestibility of the experimental diets was taken in animals in metabolic cages equipped with separate feeders, an automatic watering trough, and a system for the separate collection of feces and urine. Twelve goats (30 ± 2.35 kg) in each treatment were selected for the digestibility trial. The experiment consisted of 14 days of adaptation to the diet in individual stalls with a concrete floor, 2 days of adaptation to the fecal collection bags, and 5 days of feces collection [36–37]. The bags (25 cm2) were made of nonwoven textile (100 g/m2) material and a sample weight to bag surface area ratio of 20 mg DM/cm2 was used . Offered and refused food were recorded daily and composite samples, pooled on an individual animal basis at the end of each collection period, were dried at 105°C for 24 hours to determine the DM intake. Another sample of the diet was ground in a hammer mill with a 1-mm screen and stored until analysis. Feces were collected daily and weighed and 20% of the total daily excretion was dried at 70°C for 48 hours and pooled on an individual animal basis at the end of the collection period. The daily spot urinary samples were collected for 2 h with bowls attached with a harness to each animal 4 h after the morning feeding. Ten ml urine sample was diluted with 40 ml of a 0.036 N H2SO4 solution and stored at –20°C for later analysis.
The feedstuff was recorded every day and it was weighed for 3 consecutive days to determine the dry matter intake (DMI), organic matter intake (OMI), crude protein intake (CPI), neutral detergent fiber intake (NDFI), acid detergent fiber intake (ADFI), acid detergent lignin intake (ADLI), half cellulose intake (HCI) and cellulose intake (CELI). The particle size of feedstuff was sampled by utilizing the Penn State Particle Separator as described . Apparent nutrient digestibility was calculated utilizing indigestible nutrition (12-d ruminal incubation in 25-µm-pore-size bags) as an internal marker, based on the concentration of indigestible nutrition in the diet. The pH was measured with a glass electrode pH meter (STARTER 100/B, OHAUS, Shanghai, China), the ammonia-N content was analyzed by using steam distillation of the filtrates 23, the concentration of organic acid were measured by high performance liquid chromatography methods as described .
Sample of feedstuff were dried for 48 h at 65℃ in a forced-air oven and ground to pass through a 1-mm screen in a Wiley mill before analysis. The contents of organic matter (OM), crude protein (CP), neutral detergent fiber (NDF), acid detergent fiber (ADF), acid detergent lignin (ADL), half cellulose (HC) and cellulose (CEL) were tested in accordance with the reference of the AOAC International . The content of NDF, ADF, ADL, HC and CEL were analyzed according to the Ankom A200 fiber analyzer (Ankom Technology, Macedon, NY) by the method of Van Soest .
Statistical analyses of chemical composition, intake, apparent digestibility, particle size distribution and fermentation were performed by one-way analysis of variance of SAS (SAS 9.0, SAS Inc. Chicago, USA) and the statistical model is: Yij = μ+ Ti + Eij, where Yij is observation for dependent variables, μis overall mean, Ti is effect of treatment, and Eij is residual error. Statistical difference among means were tested by least significant difference multiple comparison. Polynomial contrasts were used to test linear, quadratic and cubic responses to CS and CA mixture ratios. The significant difference was declared at P < 0.05.