The in vitro gas production technique (Menke et al., 1979; Theodorou et al., 1994) has been designed for the evaluation of ruminant feeds under optimal fermentation conditions, i.e., at an incubation pH (6.5-6.8) that favours maximum microbial activity. Those systems allow for maintaining incubation pH within a narrow range by adding an important proportion of bicarbonate ion to the incubation solution (Goering & Van Soest, 1970). However, the high rumen fermentability of diets rich in concentrates makes the pH range widely from 6.5 to even 5.5 (Hungate, 1966), clearly modifying rumen population and activity (Russell & Dombrowski, 1980; Hiltner & Dehority, 1983). Besides, concentrate feeds with a rapid rate of fermentation contribute themselves to enhance rumen acidification, but their potential in this sense cannot be evaluated in a strongly buffered medium. Feeds potential of acidification is generally assumed, though to a variable extent, for cereal grains (Chai et al., 2004; Lanzas et al., 2007) and even more for sources of soluble sugars (Huhtanen & Khalili, 1992; Hoover et al., 2006), but it should also be considered for highly fermentable fibrous feeds, such as citrus pulp or sugarbeet pulp, rich in pectins and fermentable hemicelluloses (Flachowski et al., 1993; DePeeters et al., 1997; Barrios-Urdaneta et al., 2003). In addition some cereal byproducts, such as wheat bran, may include important proportions of rapidly available starch that is fermented at a high rate by rumen microbes (Maes & Delcour, 2001).

Conventional incubation conditions with concentrate feeds give higher gas production results because of unrealistic high pH (Fondevila & Pérez-Espés, 2008; Bertipaglia et al. 2010), and therefore extrapolation to concentrate feeding is biased by incubation conditions. In previous work in our laboratory, we have modified incubation conditions of the in vitro gas production method by reducing the concentration of bicarbonate ion, thus reducing the buffering capacity (Kohn & Dunlap, 1998) and allowing for both the study of fermentation at low medium pH and the study of acidification properties of feeds.

The objective of this study was to evaluate in vitro different sources of highly fermentable carbohydrates under two different approaches: at an incubation pH (6.5) that ensures for an optimal rumen microbial fermentation, and at low buffered conditions that allow for the study of both the fermentation of feeds in conditions close to those occurring in high-concentrate feeding systems, and the acidification capacity of such feeds.

Five feeds differing in the nature of their carbohydrate fraction were chosen as substrates for in vitro incubation: maize grain (Dekalb 6667YG, MZ); sucrose (SU); wheat bran (WB); sugarbeet pulp (BP) and citrus pulp (CT). All substrates (except for SU) were milled through a sieve of 1 mm using a hammer mill (Retsch Gmbh/SK1/417449). Chemical composition of substrates is given in Table 1.

Four adult Rasa Aragonesa ewes (average weight 70 ± 2.7 kg) fitted with a rumen cannula, were used as donors of rumen inoculum. Animals were housed in the facilities of the Servicio de Experimentación Animal of the Universidad de Zaragoza. Maintenance and extraction procedures of rumen inoculum from donor animals were approved by the Ethics Committee for Animal Experimentation. Care and management of animals agreed with the Spanish Policy for Animal Protection RD 53/2013 (BOE, 2013), which complies with EU Directive 2010/63 (EC, 2010) on the protection of animals used for experimental and other scientific purposes.

In a first incubation trial, ewes were fed 600 g of alfalfa hay plus 300 g of barley straw for three weeks. Thereafter, in a second incubation trial diet was changed to 500 g of concentrate (0.60 ground barley grain, 0.20 ground maize grain and 0.20 soybean meal) plus 300 g of alfalfa hay for another three weeks, in order to get a rumen environment characteristic of concentrate diets. Along the third week of each period, rumen contents (~ 300 mL) were extracted before the morning feeding, pooled, filtered through cheesecloth and transferred to the laboratory in thermos bottles preheated to 39° C.

The studied substrates were chosen based on the varied chemical composition of their carbohydrate fraction (Table 1). Maize is a model of cereal grain, with a high proportion of starch, although it is available in the gut at a relative moderate rate compared to other grains (Offner et al., 2003; Lanzas et al., 2007). As a byproduct of cereal processing, wheat bran includes a variable but important proportion of starch (100 to 240 g/kg), highly available since it is from wheat, and its cell wall contains a major proportion of fermentable hemicelluloses (Maes & Delcour, 2001; FEDNA, 2010). Sugarbeet pulp has high proportions of both neutral detergent fibre (Fondevila et al., 1994; FEDNA, 2010) and soluble fibre, mostly pectins, that can be fermented at a similar rate than soluble starch (Fondevila et al., 2002), whereas CT is a source of both soluble fibre and soluble sugars (Hall, 2002; Barrios-Urdaneta et al., 2003). Finally, SU was chosen as a model for soluble sugars that are almost completely used by most rumen microbes in less than an hour (Hristov et al., 2005).

As a reference, in Experiment 1 substrates were compared under conventional incubation conditions, with inoculum from sheep given a forage diet, and maintaining incubation pH within a range that should allow for maximising potential fermentation. According to the volume of gas produced in vitro, CT in Experiment 1 was fermented at the highest extent during the 8 h incubation, but this was not manifested in higher total VFA concentration, partly because of the high magnitude of the error term (coefficient of variation 0.19), nor in the VFA pattern except for a higher butyrate molar proportion with the substrates rich in soluble fibre (CT and BP). Despite of differences in the nature of their carbohydrate fractions, gas production from BP and WB evolved similarly throughout the 8 h incubation period, whereas values at 6 and 8 h for the two non-fibrous substrates SU and MZ were lower than the other under these incubation conditions, probably because the inoculum induced by a forage diet, such as in Experiment 1, was less specialised in fermenting sugars or starch than that from a concentrate diet (Mould et al., 2005b).

In any case, Experiment 1 was carried out as a standard reference, but the major interest of the present work was to compare substrates in incubation conditions close to those found with concentrate-rich diets, such in Experiment 2, where the inclusion of an inoculum from a concentrate diet and the low buffering potential of the incubation solution make the procedure closer to natural incubation conditions in intensive feeding. In this case, acidification caused by fermentation of substrates incubated in a medium with a low buffering capacity did not differ between MZ, BP and WB, pH being reduced in 1.2 units from that of inoculum after 10 h of incubation, with most of this response occurring in the first 2 h (Fig. 2). The magnitude of this drop increased to 1.4 pH units when CT was incubated, reaching a final pH of 5.08 after 10 h of incubation. For SU, the buffering capacity of the medium was exhausted from 4 h onwards, and pH fell from 5.45 to 4.00 at 10 h because of its high rate and extent of fermentation. This should probably be linked to the high concentration of lactic acid recorded at 8 h incubation, in agreement with that reported by Strobel & Russell (1986). Therefore, the inclusion of high levels of sucrose sucrose in concentrate diets may contribute to enhance the risk of acidosis, because of its fast fermentation rate, thus reducing the potential microbial activity over this and the whole diet. In fact, fermentation was almost stopped from 6 h onwards, probably because of the worsening of incubation conditions. These results make recommend a low level of sugar inclusion in ruminant diets (Vallimont et al., 2004).

If gas production volumes observed after 8 h incubation in both experiments are contrasted, fermentation of CT in Experiment 2 ( Fig. 3) was proportionally reduced in 0.29 with respect to that recorded in Experiment 1 (Fig. 1). In Experiment 2, incubation pH with this substrate dropped to 4.8 after 6 h incubation, and continued falling afterwards (Fig. 2). This should explain why rate of gas production from CT was reduced from 4 h onwards in Experiment 2, instead of continuing increasing like in Experiment 1. The high content in both soluble sugars and soluble fibre of CT (Table 1) should promote a high rate of fermentation that was also manifested on a high VFA concentration (Table 3), thus leading to a low incubation pH. These altered conditions should prevent from a higher DMd for this substrate. It can be considered that, since incubation conditions became unfavourable for microbial activity after 6 h, fermentation of citrus pulp was uncompleted, mostly affecting the fibrous fraction since fibrolytic activity is minimal before that time (Fondevila et al., 2002; Mould et al., 2005a). In this regard, straw diets for sheep supplemented with 0.42 citrus pulp did not apparently behaved differently in vivo than those with the same proportion of barley grain, with a minimum pH of 5.8 after 4 h (Barrios-Urdaneta et al., 2003). These results, however, cannot be directly extrapolated to practical feeding conditions, since citrus pulp composition may widely vary among batches. In this sense, from 79 batches of this by-product, Hall (2002) reported that proportions of soluble sugars and soluble fibre varied from 125 to 402 and from 252 to 437 g/kg DM, respectively.

When comparing results from both experiments, it should be considered that gas production in Experiment 2 could be reduced to a certain extent by a lower contribution of indirect gas coming from the buffering of produced VFA (Beuvink & Spoelstra, 1992) because of the reduced concentration of bicarbonate ion in the poorly-buffered medium. Besides, drop of pH as incubation proceeds was higher with the inoculum from a concentrate diet (Experiment 2), negatively affecting microbial community, in a higher extent to fibrolytic bacteria (Russell & Dombrowski, 1980). This should partly explain why gas production from BP at 8 h of incubation was proportionally 0.27 lower in Experiment 2 than in Experiment 1, mainly considering that rumen microbes from concentrate-rich conditions are less capable to ferment fibre than those from a forage diet, as suggested by Calsamiglia et al. (2008). In contrast to gas production results, total VFA concentration was higher in Experiment 2, but this may be explained by the self-fermentation of the inoculum itself, which harbours a higher concentration of soluble nutrients, whereas gas production results were corrected in both experiments by the volume of gas produced from the blanks of inoculum (incubated without substrate), thus minimising such bias.

In contrast, fermentation of MZ increased in a proportion of 0.28 under conditions of Experiment 2, indicating that rumen microbial population promoted by forage diets is not adapted to ferment the moderately available vitreous starch of maize (Nagadi et al., 2000). Finally, values observed with WB and SU were affected to a minor extent by incubation conditions (0.09 proportionally lower and 0.04 higher gas volume in Experiment 2 than 1, respectively). The mixed characteristics as a source of fermentable fibre but with a noticeable proportion of highly fermentable starch (Lanzas et al., 2007) in case of WB, and the ease for a rapid utilization of mono- and disaccharides for most rumen microorganisms in SU (Hristov et al., 2005) should explain those minor differences.

In conclusion, the extent and rate of fermentation of sugars and soluble fibre fractions from citrus pulp are of a higher magnitude than those from other carbohydrates, such as the highly available starch and hemicellulose from wheat bran. In contrast, starch from other sources like maize grain is fermented at a slower rate, similar to that of the cell wall and soluble fibre of sugarbeet pulp. These last three feeds contribute at a minor extent to acidification of the medium, and therefore their impact on microbial fermentation should be lower. In contrast, sucrose as a source of soluble sugars has a high acidification potential, which may reduce its potential fermentation from 4 h onwards by challenging the microbial activity when include at a high level. The practical implications of these responses, in order to choose the level of each carbohydrate source in high-concentrate diets will depend on the objectives and constraints of the feeding system itself.

Compared to conventional in vitro incubation conditions, a poorly-buffered medium allows for the study of fermentation of substrates differing in the nature of their carbohydrate fraction, when used in high-concentrate diets.