A 4 x 4 Latin square design was used to investigate the effects of supplementary Chinese milk vetch (Astragalus sinicus L.) silage (MVS) on the dry matter intake and on the rumen environment in sheep offered ammonium bicarbonate-treated rice straw (ABRS). The ABRS was offered ad libitum and the MVS was included at levels of 0, 10, 20 or 30 % of dietary dry matter intake to create differing rumen environments. The intake of ABRS was reduced almost linearly with the increase in supplementary MVS, and the substitute rate was between 0.83 and 1.02. No significant treatment effects were observed on the pH of rumen fluid nor on the molar proportions or concentrations of rumen VFA. Rumen NH₃-N levels were significantly higher (P<0.05) in sheep fed on the higher levels of MVS than in those on the low MVS level, with the lowest in the non-supplemented group. All treatments produced pronounced diurnal variations in the rumen concentrations of NH₃-N, which tended to be similar in all treatments before feeding, peaked at 2 h post-feeding and then decreased. Rumen protozoal populations tended to decrease more quickly in the supplemented groups, suggesting that the presence of MVS may inhibit ruminal protozoa. The microbial protein concentration in rumen fluid increased in the presence of MVS, peaked at the 20% level of MVS-supplementation, and then decreased. The optimal level of MVS in ammoniated rice straw diet is thus likely to be about 20 %.

Ruminants offered low quality crop residues require additional nutrients to optimise productivity. In many crop residues the NH₃ generated in the rumen from degraded plant protein is often too low to support optimum microbial activity for ensuring efficient digestion. Pre-treatment with a source of ammonia such as urea or ammonium bicarbonate can greatly enhance both the intake and digestibility of straw, and will improve the productive performance of the animals (Liu 1995). When the requirement of the rumen microbes for nitrogen has been satisfied (met) then a small amount of undegraded protein available for post-ruminal digestion is desirable to maximize efficient use of the diet. However, for most developing countries the problem is in the limited availability of protein sources although great efforts have been and are being made to find alternative supplements (Preston 1986; Leng et al 1992).

There are many forages that may potentially manipulate digestive function in the rumen, and hence benefit the host animals. A wide range of forage supplements are available and are being evaluated including Leucaena, Gliricidia, cassava, etc. (Speedy and Pugliese 1992). It has been shown that supplementation of an ammoniated rice straw diet with Chinese milk vetch (Astragalus sinicus L.) silage (MVS) markedly improved the growth rate of ruminants, reduced the need for concentrates and decreased the feed cost for liveweight gain (Liu et al 1995).

The present research was conducted to investigate whether supplementation with MVS results in an improved rumen environment which is more favourable for the degradation of the ammonium bicarbonate-treated rice straw (ABRS).

Four Hu-Yang sheep weighing about 40±0.6 kg were fitted with permanent rumen cannulae of 35 mm internal diameter. The animals were maintained in individual metabolism crates and had free access to drinking water and mineralized salt blocks.

Both rice straw (Japonica, cv. ''Zhenongda 40") and Chinese milk vetch (Astragalus sinicus L.) at the flowering stage were obtained from the Experiment Farm, Zhejiang Agricultural University. The ABRS was prepared by the "stack method". One tonne of rice straw was treated with 100 kg AB and 300 kg water for 30 d at an ambient temperature of 15-20 °C (Liu et al 1992). The MVS was prepared in a bunker silo measuring 1 m³.

The trial was conducted within a 4 x 4 Latin square design. Within each experimental period of 24 d the first 14 d were for adaptation and the final 10 d for measurements. Intakes of ABRS and MVS were measured over 7 d (from day 15 to 21). The determination of rumen degradation of dry matter (DM) using the nylon bag technique was conducted on days 22 and 23, and further rumen parameters including pH and microflora were determined on day 24. Neutral detergent fibre (NDF) determinations were made on ABRS and MVS, and on the residual component for the nylon bag degradation.

Feeds were offered twice daily in two equal meals at 8:00 and 16:30 h. The ABRS was chopped to about 3-4 cm before feeding. All animals were offered the ABRS in excess of appetite in that the daily amounts were calculated to exceed that eaten on the previous day by about 10 % in order to avoid selective feed intake. The amount of straw to be fed every day was also used to calculate the ratio of MVS to diet, which was targeted at 0, 10, 20 or 30 % of diet DM. Daily subsamples of the diets offered were collected, and feed residues were weighed, sampled and bulked each morning for later analysis of DM and NDF.

The rumen degradation of ABRS and of MVS was determined in sacco (Orskov 1985) as described elsewhere (Liu et al 1995). Duplicate samples in dacron bags were suspended in the rumen of sheep offered one of the four dietary treatments respectively, and were removed from the rumen after 48 hours of incubation. DM and NDF determinations were made on the residues and 48 h rumen degradability (D48) was then calculated for each component.

Rumen liquor samples were withdrawn by tube through the rumen cannulae before, and 3 and 6 h after morning feeding, and 50 ml subsamples were immediately frozen at -20 °C for later determination of microbial protein (MP). The remainder of the samples was strained through four layers of surgical gauze and 1 ml of 30% H₂O₂ was added to suppress microbial activity. Subsamples of the strained rumen liquor, 20 ml, were decanted into test tubes, mixed with an equal volume of 20% formaldehyde solution and stored at -4 °C for protozoa counts. Further samples were then stored at 4 °C until analysed for NH₃-N and volatile fatty acids (VFA).

The pH of rumen liquor was determined immediately after removal using a pH meter (PHS-3c Model).

Duplicate samples of rumen fluid, 5 ml, were placed in test tubes, and mixed with 5 ml of KOH solution (2 mol). The rumen NH₃-N concentration was then determined by steam distillation into boric acid and titration with dilute hydrochloric acid (10 mmol).

Duplicate 2 ml samples of the rumen fluid were placed in graduated tubes, and mixed uniformly with 8 ml of distilled water (approximately 4 °C). One ml of 25 % ortho-phosphoric acid was added and after thorough mixing the samples were allowed to stand in a refrigerator (4 °C) for 30 minutes. The test tubes with rumen fluid were then centrifuged at 3000 rpm for 10 min. The supernatant was decanted into another test tube, capped and stored in a refrigerator at 4 °C until analysed using gas liquid chromatography (Model 663-30, Hitachi). Samples were injected into a 2 m x 3 mm glass column packed with porapak Q (80 mesh). The temperature of the injector, oven and detector were 220 °C, 260 °C and 260 °C respectively. Nitrogen gas was used as a carrier at a flow rate of 30 ml/min.

Ciliate protozoa were counted by the method of Ogimoto and Imai (1981). Two ml of formaldehyde-preserved samples were mixed with 3 ml methylgreen-formaldehyde-saline solution and shaken gently. The mixture was held overnight and then pipetted into a counting chamber (haemocytometer) of 0.4 mm. The protozoa were counted in 4 microscopic fields. The counts were repeated four times and 16 microscopic fields were thus counted for each sample.

Rumen liquor was thawed and the samples taken at the three sampling times were mixed thoroughly and strained through four layers of surgical gauze. Microbial determinations were determined by the method of Zinn and Owens (1986) based on purines. Microbial protein was estimated from the ratio of purines to N of isolated bacteria. Yeast RNA was used as a standard.

ABRS and MVS were sampled daily and dried at 60 °C. These samples were bulked and subsampled for further analysis: the DM was determined at 105 °C, and crude protein (CP) was determined by the macro-Kjeldahl method (AOAC 1990). The NDF content was analysed as outlined by Van Soest et al (1992). The pH value of MVS was determined using a precise pH meter (PHS-3c Model).

Data were analysed by one way analysis of variance. The differences among means for the four treatments were tested using Duncan's new multiple range test (Steel and Torrie 1980).

The CP of the ABRS and MVS were 9.6 and 18.9% respectively (Table 1). The MVS had a bright-yellow colour, an acid smell, and a pH value of 4.1, all indicative of a high-quality silage.

Table 1: Chemical composition of ammoniated rice straw and Chinese milk vetch silage
	Ammoniated rice straw	Milk vetch silage
p H	-	4.1
D M (g/kg)	89.2	13.2
Composition of DM (g/kgDM)	83.8	87.5
N x 6.25	9.6	18.9
NDF	67.1	54.9
ADF	41.4	38.0
Lignin	4.9	3.3

The actual MVS supplementation levels were 0, 9.4, 19.1 and 26.9 % of diet DM (Table 2), very close to the target levels. Straw intake declined linearly with the increase in supplementary MVS levels (P<0.05) and the substitution rate was high (0.83-1.02). Total DM intake did not differ between dietary treatments.

Table 2: Dry matter intake (g kgLW0.75) of ammoniated rice straw and Chinese milk vetch silage (MVS) by sheep
	MVS level (%diet DM)#
Feeds	0	10	20	30	SEM
Amm. straw	53.5a	48.2b	43.6bc	40.8c	±0.79
MVS	0	5.0	10.3	15.0
Total	53.2	53.2	53.9	55.8	±0.73
Substit. rate	-	1.02	0.93	0.83

#Levels actually determined were 0, 9.4, 19.1 and 26.9%, respectively.
a,b,c Means with different superscripts within rows are significantly different (P<0.05).

There were little differences in the rumen degradabilities of DM and NDF of both feedstuffs when incubated in the rumen of sheep fed the different diets (Table 3). However, the 48 hour degradability of DM and NDF in MVS was higher than for ABRS.

Table 3: Degradation (%) after 48 hours incubation of dry matter and neutral detergent fibre of ammoniated rice straw (ABRS) and Chinese milk vetch silage (MVS) in the rumen of sheep offered the four different diets
	MVS level (% diet DM)#				SEM
	0	10	20	30
Dry matter loss in 48 hr
MVS	65.5	66.3	67.0	67.2	±0.49
ABRS	49.4	46.7	50.1	50.2	±0.98
NDF loss in 48 hr
MVS	53.8	54.8	55.5	56.0	±0.64
ABRS	42.1	39.0	42.1	43.4	±1.22

There was little diurnal variation within the time interval of pH measurements (Table 4), although the average pH values tended to increase with an increase in the supplementary MVS level. Rumen NH₃-N levels increased linearly as the MVS substitution level increased.

Table 4: Mean pH and NH3-N concentrations in the rumen of sheep offered the four different diets
	MVS level (%, diet DM)
	0	10	20	30	SEM
Post-prandial pH
0 hr	6.57	6.60	6.59	6.64	±0.02
3 hr	6.72	6.76	6.80	6.81	±0.03
6 hr	6.61	6.62	6.62	6.66	±0.02
Overall	6.63	6.66	6.67	6.70	±0.02
Post-prandial NH3-N (mg/100ml)
0 hr	8.9	10.6ab	10.2ab	11.3a	± 0.4
3 hr	15.1d	19.5c	22.1bc	26.0a	± 0.9
6 hr	12.6ab	13.9ab	15.1a		±0.8
Overall	11.7	14.2	15.4	17.0	±1.0

a,b,c,d Means with different superscripts within rows are significantly different (P<0.05)

The VFA concentrations were not affected by supplementary MVS levels or by sampling time (Table 5), suggesting that the diets produced similar fermentation patterns in the rumen. The ratios of acetic to propionic acids were 3.66, 3.71, 4.10 and 3.86 in the 0, 10, 20 and 30 % MVS-supplemented diets respectively. Molar proportions were typically around 75, 25 and 5 mol/100 mol total VFA for acetic, propionic and butyric acids respectively.

Table 5: Mean VFA concentrations in the rumen of sheep offered the four different diets to appetite(mM/100ml)
	MVS level (%, diet DM)				SEM
	0	10	20	30
Overall means
Acetate	6.06	5.83	6.44	6.19	±0.25
Propionate	1.68	1.60	1.58	1.61	±0.09
Butyrate	0.36	0.41	0.50	0.58	±0.02
Post-prandial total VFA
0 hr	9.00	8.45	8.72	8.05	±0.34
3 hr	7.74	7.41	8.20	8.91	±0.39
6 hr	7.56	7.69	8.64	8.18	±0.49
Overall	8.10	7.84	8.52	8.38	±0.45
Ac:Pr	3.61	3.64	4.08	3.84

Table 6: Mean rumen protozoa counts and microbial protein (MP) concentrations in the sheep offered four different diets
	MVS level (%, diet DM)
	0	10	20	30	SEM
Post-prandial protozoa counts (x 105 /ml)
0 hr	1.78	1.79	2.00	1.92	±0.13
3 hr	1.61	1.44	1.70	1.58	±0.10
6 hr	1.55	1.25	1.31	1.33	±0.09
Overall	1.65	1.48	1.67	1.61	±0.11
MP concentration in rumen fluids (mg/ml)
	2.58	2.93	3.23	2.82	±0.12

The numbers of protozoa declined with time after feeding and tended to decrease more quickly with a supplement than without although the differences between diets were not statistically significant (P>0.05). The microbial protein (MP) concentrations in the rumen fluid increased with increasing level of MVS, reached a peak value with 20 % MVS, and then levelled off.

The present study was designed to determine whether the potentially different rumen environments produced by increasing levels of MVS would induce differences in the degradation of ammoniated rice straw. The results were inconsistent in respect of the degradation of ABRS as well as of MVS in the four rumen environments. Lack of difference in degradability of straw and of vetch silage may be because the conditions for cellulolysis of straw were adequate in all diets, as indicated by the rumen parameters measured.

The pH values (Table 4) showed typical rumen buffering capacity when straw diets were ingested, and were optimal for the growth of cellulolytic bacteria as suggested by Mould et al (1983). Similar values were reported by Djajanegara and Doyle (1989) and Mgheni et al (1993) in sheep fed untreated, urea-supplemented or urea-treated rice straw diets. The rate of degradation of straw is optimal when rumen pH is maintained at 6.7±0.15; cellulolysis is inhibited below 6.0-6.1 (Mould et al 1983). The mean overall pH of 6.67 obtained for all diets suggested that optimal pH conditions existed in the rumen for cellulolysis.

The level of NH₃-N that supports maximum digestion of straw in the rumen will vary between diets. The critical level of NH₃-N has been variously reported as between 50 and 280 mg of NH₃-N per litre of rumen liquor (Durand 1987). The level of NH₃-N obtained in this study (Table 4) suggests that all diets were adequate in NH₃-N to support optimal fermentation. It would therefore be expected that microbial growth and digestion of the straw was maximal in all diets. This then would account for our failure to detect differences in the degradation of the DM and NDF of ABRS and of MVS when incubated in the rumen of sheep fed these diets. However, the higher levels of NH₃-N in the rumen coinciding with higher levels of MVS may have resulted in an accumulation of NH₃-N and an increase in N loss due to incomplete utilization of the available NH₃-N by the microbes for protein synthesis.

Supplementation with MVS did not markedly influence the concentrations of VFA (Table 5). With all diets the sheep maintained a typical roughage type of fermentation with a high proportion of acetate. However the ratio of acetate:propionate altered with the level of the MVS. It increased with increasing level of MVS, peaked at 20 % MVS and then levelled off.

The MP concentrations in the rumen fluids were consistent with the growth rate achieved by heifers offered the same diets (Liu et al 1995). When MVS accounted for 20 % of the diet, the rumen MP concentrations were significantly higher than in the non-MVS group (P<0.05). Supplementation at a 30 % level of MVS resulted in MP concentrations, and liveweight gains, lower than with 20 % MVS, while protozoal biomass tended to decrease more quickly with MVS than without. These changes may be associated with some as yet unqualified anti-nutritional factors in the MVS.

Navas-Camacho (1994) observed that protozoal populations in the rumen dramatically decreased when sheep were supplemented with leguminous tree leaves having high or medium saponin contents. Further studies are indicated to confirm the findings of the present work.

The rumen conditions for fibre digestion were generally good in the present study where ammonium bicarbonate-treated rice straw was used as the basal diet. These favourable conditions may account for our failure to detect differences in the rumen degradation of the DM and NDF of ABRS and of MVS. The largest amount of microbial biomass and highest ratio of acetic to propionic acids were achieved in the rumen when 20% of Chinese milk vetch silage was included in the diet. These patterns and byproducts of rumen fermentation were confirmed by the levels of liveweight gain achieved by heifers offered similar diets (Liu et al 1995).

rskov E R 1985 Evaluation of crop residues and agro-industrial by-products using the nylon bag method. Research and Development in Agriculture. 4(2):65-69

Received 15 November 1996