Samples of Kochia scoparia and Atriplex dimorphostegia (stems and leaves) were harvested at mid bloom stage from the Salinity Research Station of Birjand University with clay-sandy soil (pH = 7.4 and EC = 10.3 dS/m) and salty irrigation water (pH = 6.8, EC = 5.5 dS/m, and Na⁺ = 24.7 meq/L). Samples of each plant were composited and sub-samples were taken from each composite and then dried in a forced air oven (65°C) for 48 h. Samples were ground using a Wiley mill (two-mm screen) and analyzed for total N (Kjeldahl method, Kjeltec 2300 Autoanalyzer, Foss Tecator AB, Hogans, Sweden), neutral detergent and acid detergent fiber (aNDF and ADF, respectively, Van Soest et al., 1991) and ash (AOAC, 2000, ID 942.05). Heat-stable α-amylase was used in the aNDF assay (Udén et al., 2005). Non-protein N was chemically determined using trichloroacetic acid 78 (1 g/100 ml) solution (Licitra et al., 1996).

Eight dual-flow continuous culture fermenters (Hannah et al., 1986) were inoculated with ruminal fluid (1,030 ml) from a Holstein cow. The cow was surgically fitted with a ruminal cannula made of soft plastic, housed in a well-ventilated barn, had free access to water, and was fed a total mixed diet that was formulated to meet its nutrient requirements (NRC, 2001). The diet consisted of 1.8 kg alfalfa hay, 9.5 kg corn silage, 0.7 kg wheat straw, and 4.4 kg concentrate mix on a DM basis. The concentrate mix was composed of 31% ground corn, 20% dried molasses, 16.5% soybean meal, 20% soybean hull, 1.5% yeast, 2% blood meal, and 9% vitamin-mineral mix. Ruminal fluid was collected using a vacuum pump approximately 2 h after the morning feeding (09:30), strained through four layers of cheese cloth, and anaerobically transported to the laboratory in a prewarmed insulated thermos.

The experimental period was 10 d in length, with seven d for adaptation followed by 3 d for sampling. Each fermenter was continuously infused with artificial saliva containing 0.5 g of urea/L (pH = 8.25). Solid and liquid dilution rates were adjusted daily to approximate 0.05 and 0.10 h⁻¹ respectively, by regulation of saliva input and filtrate removal rates. Temperature of fermenter contents was maintained at 38.6°C±0.1 by an electrical heater and the pH at 6.5±0.5 by the addition of five N HCl or five N NaOH through automated peristaltic pumps that were controlled by an electronic data acquisition system (Daisy Lab^®). Fermenters were constantly purged with N₂ gas (40 ml/min) to preserve anaerobic condition. Fermenters were supplied daily with 75 g DM of pelleted Kochia scoparia or Atriplex dimorphostegia by an automated feeding mechanism adjusted to deliver the diet in eight equal portions over a 24-h period to establish steady-state conditions. Solid and liquid effluent weights were recorded daily at 09:00 and discarded until the last three days of the experiment. On the last 3 days, fermenter effluents were maintained at 2 to 4°C by a cold water bath to retard microbial metabolism. Total effluent for each 24 h of the 3-d sample period was recorded, combined and homogenized (Power Gen Model 700) and a 1-L sample was removed by vacuum aspiration and stored at −20°C for analysis. Fresh composite samples were used for total nitrogen (N), ammonia-nitrogen (NH₃-N) and volatile fatty acid (VFA) analyses. Freeze-dried samples were ground through a 1-mm screen and used for all other analyses (DM, OM, ash, aNDF, ADF, and purine). Bacterial samples (fluid-associated bacteria) were collected from fermenter contents on the last sampling day by straining the total content of each fermenter through two layers of cheese cloth. This strained fluid was centrifuged at 1,000 g for 10 min to remove feed particles and any protozoa that may have been present. The supernatant was centrifuged at 20,000 g for 20 min to isolate bacterial cells. The resulting supernatant fluid was discarded and the precipitate (bacterial cells) was re-suspended in de-ionized water for freeze drying. Bacterial cells were used to determine the concentration of purines that were used as a bacterial marker.

Total N of effluent samples was determined by macro-Kjeldal procedure (AOAC, 2000) and the bacterial N was determined using a true spec nitrogen analyzer (True Spec Series Leco, USA). Concentration of NH₃-N in effluent samples was determined by direct stem distillation using a Kjeltec 2300 Autoanalyzer (Foss Tecator AB, Hogans, Sweden). Purine concentration in effluent and bacterial cells was determined using the method of Zinn and Owens (1986). Purine content of effluent and bacteria was used to partition flow of effluent N into bacterial and dietary nitrogen. Effluent VFA concentration was measured by gas chromatography (Hewlett-Packard, model 5880A, Palo Alto, CA) with a Carbopack DA/0.3% Carbowax 20M column (Supelco, Bellefonte, PA).

Some chemical composition of Kochia scoparia and Atriplex dimorphostegia is shown in Table 1. In agreement with Riasi et al. (2008) the DM, ash, aNDF and ADF content of Kochia was lower than Atriplex. However, there are some contradictions in chemical composition of halophytic forages between assays. In the present study CP content of Kochia was lower than some previous reports (Cohen et al., 1989; Madrid et al., 1996; Riasi et al., 2008). Differences among studies could be due to forage production condition, stage of harvesting, leaf:stem ratio or genetic variation (Gihad and El Shaer, 1992; Benjamin et al., 1995; El-Shatnawi and Abdullah, 2003). The NPN/CP ratio of Kochia and Atriplex (0.032 and 0.053, respectively) was relatively lower than the common forages such as alfalfa hay. This finding is important, because NPN content can influence CP quality of forages for ruminants (Cohen et al., 1989; Ben Salem et al., 2004).

Nutrient digestibility of Kochia scoparia and Atriplex dimorphostegia is shown in Table 2. There was no difference (p>0.05) between true DM digestibility of Kochia compared with Atriplex. In general, DM digestibility determined using in vitro methods is lower than digestibility determined using in vivo and in situ methods (Stern et al., 1997). Consistent with this observation, true DM digestibility of Kochia (397 g/kg) and Atriplex (426 g/kg) (Table 2) was lower than DM digestibility of Kochia and Atriplex determined using an in situ procedure (444 and 472 g/kg, respectively) (Riasi et al., 2008).

Although the OM content of Kochia was higher than Atriplex (898 vs. 854 g/kg) (Table 1), Atriplex had higher true OM digestibility compared with Kochia. Ghadaki et al. (1975) determined nutritive values of introduced and native range plants grown under natural conditions in Iran and showed that the shrubs had higher percentages of lignin, and there was a relationship between their lignin content and in vitro true OM digestibility. However, it should be noted that moisture, pressure, and heat applied to continuous culture diets during pelleting may reduce the OM digestibility (Stern et al., 1997).

Kochia and Atriplex had about 50% true CP digestibility and there was no difference (p>0.05) between the two Halophytes. The CP digestibility of Kochia and Atriplex (Table 2) was higher than values reported by Danesh Mesgaran and Stern (2005). However, concordant with Danesh Mesgaran et al. (2004) and Riasi et al. (2008), there were no differences (p>0.05) between CP digestibilities of the two halophytic plants.

The aNDF digestibility of Atriplex was higher (p<0.05) than that of Kochia while ADF digestibility showed the opposite effect (p<0.05). These results may be related to chemical composition of plants regarding variations in attachments of hemicellulose to phenolic rings (Moore and Cherney, 1986) and/or a limitation in availability of certain nutrients such as valerate, branched-chain VFA and amino acids that are required by cellulolytic bacteria (Hoover, 1986; Hussein et al., 1991). In agreement with Mansfield et al. (1994) who observed a negative relationship between NFC concentration of diets and NDF digestibility values, our study showed that NFC concentrations of Kochia and Atriplex (376 and 221 g/kg, respectively) (Table 1) may have affected their NDF digestibility. A lower concentration of valerate with Atriplex (Table 3) might have affected its ADF digestibility (Table 2) (Hoover, 1986).