In vivo methods involving internal and external markers

In vivo methods are the most logical to evaluate degradation of feeds in the rumen. The protocols require animals fitted with cannulae in the reticulo-rumen, the abomasum, or proximal duodenum⁽Reference Stern, Bach and Calsamiglia^4, Reference Stern, Varga, Clark, Firkin, Huber and Palmquist³⁹⁾. Also they require suitable methods for determining digesta flow rates and for differentiating microbial protein from dietary protein in the digesta that flows to the small intestine. These procedures are labour intensive and require considerable investment. In addition, increasing concern for animal welfare limits the applicability of such methods. For these reasons, only a few animals can be used in in vivo experiments, which could be unreliable due to the large variation observed among animals⁽Reference Stern, Bach and Calsamiglia⁴⁾. The in vivo procedure also relies on the accurate estimation of the flow of microbial protein to differentiate it from the feed protein reaching the duodenum. For this purpose several microbial markers have been used. These may be classified as internal markers that are inherently present in micro-organisms and include diaminopimelic acid, aminoethylphosphonic acid and nucleic acids (DNA or RNA) or external markers (that are added to the rumen to label the micro-organisms) including³⁵S, ¹⁵N, ¹⁴C, ³H and ³²P, etc. Despite the use of several markers, there is no single ideal marker to estimate the ruminal microbial protein yield⁽Reference Broderick and Merchen⁴⁰⁾.

There is a need for a more practical, repeatable and cheaper method for measuring ruminal degradation of various feeds, particularly for commercial laboratories which do not have access to fistulated animals. In spite of the limitations with the in vivo procedure, a routine alternative procedure would only be acceptable following validation against in vivo measurements. Different investigators have used several alternative procedures but without reaching a consensus on the suitability of a unified approach.

In sacco method to estimate feed degradation

The in sacco technique was first suggested by Quin et al. ⁽Reference Quin, Van der Wath and Myburgh⁴¹⁾ and it has since been used by others to estimate utilisation of either forages⁽Reference Van Keuren and Heinemann⁴²⁾ or concentrates and high-protein feeds⁽Reference Mehrez and Ørskov⁵⁾. Interest in the technique has intensified since Mehrez & Ørskov⁽Reference Mehrez and Ørskov⁵⁾ critically assessed the factors causing variability in DM and N degradability. They concluded that as long as the bags were large enough to allow free movement of substrate within, the technique could be extremely useful as a rapid guide to determine nutrient disappearance, particularly the rate and extent of nutrient disappearance from the rumen. All modern systems of feeding ruminants^{(1, 2, 43)} require an estimation of the amount of feed protein escaping ruminal degradation. This estimation is obtained by the in sacco technique, which is probably the best-known simple and reliable method to assess the degradability of DM and protein in the rumen^(1–Reference Thomas^3, Reference Mehrez and Ørskov^{5, 43)}.

The in sacco method requires the use of fistulated animals, which limits its routine use by the commercial laboratories. However, it is widely applied by researchers since it requires fewer measurements, is relatively less labour intensive and so is cheaper as compared with the in vivo method. The in sacco method involves the sealing of feed samples within nylon, polyester or Dacron bags, which are then suspended in the rumen of sheep or cattle for varying periods of time, followed by determination of the DM and protein in the washed residues. The technique allows the test feed to be incubated in the ruminal environment (i.e. pH, temperature and CO₂), but unlike the normal situation the feed is not subjected to mastication and rumination. Despite its widespread use, the technique has inherent errors that must be taken into account, particularly if comparisons of degradation among different laboratories are to be made. Table 1⁽Reference Cottrill and Evans^44–Reference Verbic, Ørskov, Zgajnar, Chen and Znidarsic-Pongrac⁴⁷⁾ shows possible sources of variations in the use of the in sacco method among different laboratories in terms of bag size, sample size, particle size and time (h) of incubation used by different authors. The assumption that the N leaving the bag during washing in water, at 0 h of incubation, is completely degraded may not be true⁽Reference Chaudhry and Webster^10, Reference Mahadevan, Erfle and Sauer⁴⁸⁾. Extensive loss of feed material at 0 time will lead to an overestimation of degradability. Although Table 1 and Table 2⁽Reference Chaudhry^15, Reference De Smet, De Boever, De Brabander, Vanacher and Boucque^49–Reference Noziere and Michalet-Doreau⁵²⁾ present information in relation to degradable crude protein values only, it is assumed that similar variations in practice will also cause variations in DM and organic matter degradability values.

Table 1 Some of the variations in the protocols among selected laboratories to obtain in sacco data from fistulated sheep

DCP, degradable crude protein; NA, not available.

Table 2 Some of the variations in the protocols being used by different laboratories to obtain in sacco data using fistulated cattle

DCP, degradable crude protein; C, concentrate; S, silage.

Beside microbial contamination within the bag, there are numerous other sources of errors that affect the in sacco DM and N disappearance from feeds. The importance of sample weight in a given bag size has been emphasised by Bullis et al. ⁽Reference Bullis, Haj-Manouchehri and Know⁵³⁾ who observed reduced DM digestibility with increased weight in the bag. This finding agreed with the finding of Van Keuren & Heineman⁽Reference Van Keuren and Heinemann⁴²⁾ who showed that sample weight influenced DM digestibility, at least when short incubation times were used; the difference tended to disappear with longer periods of incubation. Also, oven drying of silage samples at high temperatures was found to reduce N degradability and solubility⁽Reference Lopez, Hovell, Manyuchi and Smart⁵⁴⁾ of these samples. Additionally, Noziere & Michalet-Doreau⁽Reference Noziere, Michalet-Doreau and D'Mello⁵⁵⁾ reported that grinding and pre-wetting underestimates degradation rates due to the increased microbial colonisation. Machine washing of residues overestimates solubles and particulate losses but it is less subjective than hand washing⁽Reference Cockburn, Dhanoa, France and Lopez⁵⁶⁾. Huntington & Givens⁽Reference Huntington and Givens⁵⁷⁾ reported that bag pore size less than 15 μm can reduce degradation by restricting microbial colonisation and diversity and trapping fermentation gases. However, bag pore size of more than 40 μm can cause losses of solubles and undegradable particles. Furthermore, the animal effects and bag incubation sequence also contribute to the variation in results among laboratories⁽Reference Nocek⁵⁸⁾.

The disappearance of DM is also affected by the diet fed to the host animal⁽Reference Caton, Freeman and Galyean⁵⁹⁾. While these effects make it difficult to compare feeds for degradation across studies, DM or protein degradation of a feed is not entirely a function of the feed, but also affected by the ruminal conditions, so variation across studies is expected. More troublesome aspects of the in sacco method do exist. The pH inside the nylon bag has been shown to be lower than that outside the bags, especially when small pore-sized bags were used⁽Reference Marinucci, Dehority and Loerch⁶⁰⁾. The microbial population inside the bags also differed, both in composition and concentration, from that of the outside of the bag. For example, both protozoa and bacterial populations were found to be lower inside the bags⁽Reference Marinucci, Dehority and Loerch^60–Reference Meyer and Mackie⁶¹⁾. This could be due to the limited micro-environment that existed within the bag involving a single ingredient of smaller size with limited exposure to rumen microbes perhaps due to the bag size and its pores.

Analysis of digesta from nylon bags incubated in vitro showed that some nutrients escaped the nylon bags before being digested⁽Reference Marinucci, Dehority and Loerch⁶²⁾. The microbial attachment to feeds incubated in sacco is frequently not measured, though several studies have shown high levels of contamination of incubated feed with rumen microbes⁽Reference Huntington and Givens⁵⁷⁾. All these sources of errors increase variability of predictions of degradability among laboratories. In spite of being widely used and standardised, the application of this methodology needs to address two points: first, the fraction assumed to be completely degraded, and the DM and N disappearance during this step could simply be due to DM or N washed out of the bag; second, the microbial contamination of feeds within the bags. The first point would overestimate degradation and the second point would underestimate it⁽Reference Madsen and Hvelplund⁶³⁾. The significance of these two factors would be important depending on the type of feed being analysed. However, the in sacco method is still the reference method in most countries; the reason is probably that the degradability is measured in the rumen and, therefore, from a biological point of view it is more reliable than those of the in vitro methods⁽Reference Hvelplund, Weisbjerg, Deaville, Owen, Rymer, Adesogan, Huntington and Lawrence⁶⁴⁾. But as in vivo and in sacco methods require fistulated animals they cannot be accepted as methods for routine screening of feedstuffs. Therefore, there is a need for a viable and accurate in vitro method to estimate the degradability of feeds in the rumen.

Tables 1 and 2 indicate some of the variations of the most commonly tested methods using fistulated sheep or cattle, i.e. the same feeds give different values in different laboratories. For example, Cottrill & Evans⁽Reference Cottrill and Evans⁴⁴⁾ reported effective crude protein (CP) degradability (), where c is degradation rate of b and k is rumen outflow rate) in fistulated sheep of 51 and 82 % for fishmeal and soyabean respectively. In an unrelated study, Djouvinov et al. ⁽Reference Djouvinov, Nakashima, Todorov and Pavlov⁵⁰⁾ used fistulated cattle and reported ECPD of 28 and 77 % for fishmeal and soyabean respectively. In contrast, Kristensen et al. ⁽Reference Kristensen, Moller and Hvelplund⁵¹⁾ used fistulated cattle and reported ECPD of 50 and 73 % respectively for different samples of the same feeds. Tables 1 and 2 show that different amount of feeds were incubated using various dimensions of bags in each of the above three studies. Additionally, Cottrill & Evans⁽Reference Cottrill and Evans⁴⁴⁾ reported ECPD of 33 % for maize using fistulated sheep compared with 52 % reported by De Smet et al. ⁽Reference De Smet, De Boever, De Brabander, Vanacher and Boucque⁴⁹⁾ for another sample of maize using fistulated cattle. On the other hand, variation was also observed in ECPD for cattle where variable ECPD of 28 v. 50 % for fishmeal were reported by different authors⁽Reference Djouvinov, Nakashima, Todorov and Pavlov^50–Reference Kristensen, Moller and Hvelplund⁵¹⁾. Similar variations for the ECPD of 34 v. 77 % for soyabean meal were also reported⁽Reference De Smet, De Boever, De Brabander, Vanacher and Boucque^49–Reference Djouvinov, Nakashima, Todorov and Pavlov⁵⁰⁾. This clearly shows the inconsistency between laboratories which may be because different diets offered to the same host animal have different effects on degradability of the same feed. Additionally, different time (h) of incubation, different sample size and different dimension of in sacco bags gave different degradability values for the same feeds. Nevertheless, Table 3⁽Reference Madsen and Hvelplund⁶⁵⁾ presents variation in the in sacco ECPD results for the same feed in the same laboratory. Madsen & Hvelplund⁽Reference Madsen and Hvelplund⁶⁵⁾ reported different values for feeds using fistulated cattle. In their study they tested a minimum of three samples for the same feed obtained from different sources and the ECPD was calculated as the average of all values obtained. Table 3 shows the variations that could occur in the in sacco method for the same feed, although feeds were selected on the basis of their almost similar CP contents in DM. These variations could partly be attributed to the differences in variety, agronomic conditions and processing methods to obtain these feeds. Such variations in the in sacco estimates raise the question: when formulating a ration for ruminants, which of the reported values should be followed? This question necessitates the need to either standardise the in sacco method by following a standard protocol with minimum sources of variations or more preferably develop alternative methods which are more consistent and which do not require the use of surgically modified animals to estimate degradability of ruminant feeds.

Table 3 Variation in the in sacco data for the same feed within the same laboratory⁽Reference Madsen and Hvelplund⁶⁵⁾

CP, crude protein; ECPD, effective crude protein degradability.

In vitro methods to estimate nutrient degradation

Numerous in vitro methods have been used in the past as alternatives to the in sacco method. These methods involve buffers, chemical solvents, rumen fluid and enzymes that are either commercially available or extracted from rumen contents. Another approach is to use gas production as an indirect measure of in vitro digestion. In vitro techniques are considered less expensive than the in vivo and in sacco methods, and they offer the possibility of analysing both the residue and the metabolites of microbial degradation. In contrast, more physiological techniques are complicated by the fluxes out of the rumen or out of the incubation bags. In vitro methods may ultimately allow for the control of various factors that alter the feed degradation (microbial, animal, environment) and, therefore, allow for the uniform characterisation of feeds for DM and protein degradation.

Tilley & Terry⁽Reference Tilley and Terry¹⁷⁾ developed an in vitro method to estimate the apparent DM digestibility of feeds for ruminants in the laboratory. The method has two stages. In the first stage, a feed sample is incubated at 38°C in rumen fluid, which is diluted with a buffer solution resembling saliva and saturated with carbon dioxide. After 48 h, the incubation is stopped and the incubation mixture filtered. The feed residues are subsequently incubated for another 48 h with pepsin-HCl. The main disadvantage of the method is that rumen fluid is required, which is obtained from fistulated animals, and may not be available in all laboratories. Since the use of fistulated animals is not desirable, the need for alternative in vitro methods has arisen. There are many difficulties that are associated with the in vitro fermentation studies. These include the requirements to standardise the fermentation process, measurement of fermentation profiles and the access to fistulated ruminants to obtain rumen inocula. Therefore, several methods have been developed to measure nutrient degradation by using various enzyme preparations involving cellulases, proteases, lipases and amylases individually or as mixtures. Ruminal protease was enriched from Bacteroides amylophilus and used for studies of degradation of several protein sources⁽⁴³⁾. Alternatively, other methods have used commercially available proteases. Some of these in vitro methods are discussed in the following sections.

In vitro methods involving solubility in solvents and buffers

Several researchers have attempted to characterise feed nutrients according to their solubility in aqueous solutions such as saline and buffers, autoclaved rumen fluid (ARF) and water (cold, hot or distilled)⁽Reference Little, Burroughs and Woods^8–Reference Chaudhry and Webster^10, Reference Mahadevan, Erfle and Sauer^48, Reference Pichard and Van Soest^66–Reference Stern and Satter⁷¹⁾. However, most of the available reported literature on feed solubility focuses on protein solubility. N solubility varies greatly for different feedstuffs. For example, while N of brewer's grains was only 3 % soluble in borate–phosphate buffer, oat N was 55 % soluble⁽Reference Krishnamoorthy, Sniffen, Stern and Van Soest⁶⁹⁾. Buffer-soluble N is comprised mostly of NPN, such as ammonia, urea, nitrates, amino acids and small peptides⁽Reference Pichard and Van Soest^66–Reference Krishnamoorthy, Sniffen, Stern and Van Soest⁶⁹⁾. Nucleic acid N is the major NPN fraction that is not soluble in neutral buffer, but generally it is low in quantity and it is underestimated by N analysis using the Kjeldahl procedure⁽Reference Lyttleton, Butter and Baily⁷²⁾. In addition to NPN, some true protein is soluble to varying degrees among feeds.

Protein solubility is influenced by various factors associated with the solvent or extraction procedure. Changes in chemical composition of solvent can have pronounced effects on N solubility. For example, substitution of ammonium chloride with sodium chloride in Burroughs' mineral mixture (BMM) resulted in increased N solubility for several concentrates⁽Reference Nocek, Herbein and Polan⁷⁰⁾. It is difficult to draw meaningful conclusions as to the effect of solvent composition on N solubilisation, because of the interaction that may exist among feedstuffs, methodologies and solvents. For example, in one study, N solubility was found to be higher in BMM or 0·15 m-sodium chloride than in ARF⁽Reference Crawford, Hoover, Sniffen and Crooker⁶⁷⁾, whereas some researchers have reported the opposite response⁽Reference Waldo and Goering⁷³⁾ and others found insignificant difference among three solvents⁽Reference Lamport and Preiss⁷⁴⁾. It is noteworthy that similar concentrate feeds were analysed in each of the above studies.

Comparisons of ARF, NaCl, BMM and hot water revealed that the differences among feeds were most pronounced for ARF, suggesting that this method may best separate different feeds according to protein solubility⁽Reference Waldo and Goering⁷³⁾. However, while homogeneous variance within samples was detected in the above experiment, the use of ARF may not be repeatable from laboratory to laboratory, or from time to time, as ARF composition can very depending upon the diet and other biological differences between donor animals.

The pH of a solvent has been shown to influence N solubility of concentrate feed protein⁽Reference Wohlt, Sniffen and Hoover^9, Reference Waldo and Goering⁷³⁾. About 85 % of soyabean meal protein was soluble in aqueous HCl (pH 2·0) and water (pH 7·2), while less than 10 % was soluble at pH 4⁽Reference Waldo and Goering⁷³⁾. Significant differences in solubility of isolated soyabean protein and casein were also observed when pH of ARF and BMM varied from 5·5 to pH 6·5⁽Reference Wohlt, Sniffen and Hoover⁹⁾. However, the effect of pH on protein solubility often is measured in neutral pH solutions, even though rumen fluid is slightly acidic when donor animals consume high-grain-based diets. As sodium chloride has almost no buffering capacity and bicarbonate–phosphate buffer has an unstable pH, a borate–phosphate buffer has been suggested as an appropriate substitute to measure N solubility without variation due to fluctuation in pH⁽Reference Krishnamoorthy, Muscato, Sniffen and Van Soest⁷⁵⁾.

Buffer-soluble CP appears to be more readily degraded in the rumen than insoluble N. There was a close association between buffer-soluble CP and N disappearance from synthetic fibre bags suspended in the rumen, especially when feeds were incubated in sacco for short periods of 1 h⁽Reference Huntington and Givens^57, Reference Stern and Satter⁷¹⁾. In addition, the amount of buffer-soluble N in a feed was correlated strongly to increases in rumen ammonia concentration after feeding⁽Reference Waghorn⁷⁶⁾. However, other researchers found no consistent relationship between N solubility in various aqueous solvents and ammonia concentration⁽Reference Crooker, Sniffen, Hoover and Johnson⁶⁸⁾. N that is available to microbes is converted to ammonia which is used for microbial protein synthesis which is considered as true protein due to its high amino acid profile. Soluble NPN is utilised by rumen bacteria and solubility in mineral buffer of pure proteins was correlated strongly to degradation rate in rumen fluid⁽Reference Hamilton, Ashes and Camichael⁷⁷⁾.

Some researchers have reported a poor correlation between N solubility in buffer and in vivo protein degradation across several feeds⁽Reference Stern and Satter⁷¹⁾. This result was not surprising given that the amount of soluble feed protein appears to have no relationship with degradation rate of the remaining insoluble protein fraction of a given feed⁽Reference Mahadevan, Erfle and Sauer^48, Reference Nocek, Herbein and Polan⁷⁰⁾. Therefore, one can conclude that estimation of N solubility in neutral buffers may predict the amount of N immediately available to bacteria, but it may not be accurate for all feeds as discussed above and N solubility in aqueous solutions is not sufficient to determine protein degradation rate or protein degradability. However, as suggested by Chaudhry & Webster⁽Reference Chaudhry and Webster¹⁰⁾, positive correlations between some soluble protein fractions and degradation of feeds do require further investigations to test this approach for its much wider application.

In vitro methods using enzymes

Rumen digestion is mainly linked to the cellulolytic activity of the microbial flora, which represents its specificity and advantage to utilise cellulose-rich feeds. Researchers have used commercially available cellulolytic enzymes, often extracted from fungi, to reproduce this activity; hence many enzymic methods have been proposed for their use to predict feed digestibility. These methods differ on the basis of the nature and level of the enzymes and the target feeds and whether a pre-treatment (chemical or enzymic) is necessary or not⁽Reference Aufrere, Graviou, Demarquilly, Verite, Michhalet-Doreau and Chapoutot⁷⁸⁾. These methods are widely used for forages, and have been applied to by-products, concentrates and mixed feeds produced by the agro-food industry. For various types of forages, prediction involving enzymes is higher, perhaps due to the enzyme specificity and greater activity than the chemical methods and comparable with that obtained in vitro ⁽Reference Aufrere and Guerin⁷⁹⁾. In addition, cellulase methods can be used for mixed rations and permanent pastures. With forages containing tannins, organic matter digestibility prediction is generally poor when cellulolytic enzymes are used. This can be due to the fact that enzymes are used at pH values that are different from those prevailing in the rumen, enabling possible release of tannins and their subsequent linkage to protein. Another possible explanation is that some tannins might have inhibited the enzyme activity, especially on cellulose. Malestein et al. ⁽Reference Malestein, Klooster and van't Cone⁸⁰⁾ showed clear differences in starch degradation upon incubation with α-amylase or rumen fluid. However, Cone & Vlot⁽Reference Cone and Vlot⁸¹⁾ concluded that it was not possible to accurately predict the rate of starch degradation by rumen fluid as with enzymic degradation. Addition of non-amylolytic enzymes, such as cellulase, protease, lipase, xylanase and pectinase, did not enhance starch degradation. Starch granules contain components other than amylose and amylopectin and possibly need non-amylolytic enzymes for full degradation. These enzymes may be found in the rumen micro-organisms.

Indeed, the procedures involving the use of commercial proteases offer potential advantages over other techniques, particularly in terms of the labour and speed of operation. Proteases from different origins have been tested by several researchers to estimate ruminal protein degradation; however, the most commonly used is the one obtained from S. griseus as reported by Chaudhry⁽Reference Chaudhry^14, Reference Chaudhry¹⁵⁾ and Krishnamoorthy et al. ⁽Reference Krishnamoorthy, Sniffen, Stern and Van Soest⁶⁹⁾. Table 4⁽Reference Wohlt, Sniffen and Hoover^9, Reference Poos-Floyd, Klopfenstein and Britton^13–Reference Chaudhry^15, Reference Krishnamoorthy, Sniffen, Stern and Van Soest^69, Reference Roe, Chase and Sniffen^82–Reference Cone, Van Gelder, Steg and Van Vuuren⁸⁷⁾ gives the list of enzymes studied in some recent publications on the topic.

Table 4 Summary of the selected commercial enzymes used by different authors to estimate in vitro degradation of feeds

C, concentrate; F, forage.

CP degradation using five different commercially available proteases was compared with that of the in sacco method for several concentrate feeds⁽Reference Poos-Floyd, Klopfenstein and Britton¹³⁾. The solubility of CP was determined by filtering after separate incubation with S. griseus protease, papain, bromelain, ficin and Aspergillus oryzae protease. Though absolute degradation with protease was different from that observed in sacco, all enzymic degradations were significantly correlated to the in sacco data, supporting the possible use of enzymes to detect relative differences among feeds for degradation. S. griseus protease has become popular for prediction of protein degradation of feeds but the optimal pH for the enzyme is 8, which is greater than the rumen pH. However, it has been used at both the rumen pH with either reduced activity⁽Reference Chamberlain and Thomas^88, Reference Terramoccia, Puppo, Rizzi and Martillotti⁸⁹⁾ or reasonable response⁽Reference Chaudhry^14, Reference Chaudhry¹⁵⁾ or higher or optimum activity at the higher pH⁽Reference Krishnamoorthy, Sniffen, Stern and Van Soest⁶⁹⁾.

Protease extracted from S. griseus was used because its activity was comparable with the protease of B. amylophilus. Blackburn⁽Reference Blackburn⁹⁰⁾ selected a pure culture of B. amylphilus strain H18 to study the nature of ruminal microbial proteases. This micro-organism was selected due to its ability to hydrolyse casein rapidly, its simple nutrient requirements and its utilisation of ammonia in preference to preformed amino acids and peptides. All these characteristics simplify the measurement of the endproducts of protein degradation⁽Reference Krishnamoorthy, Sniffen, Stern and Van Soest⁶⁹⁾.

In the in vitro method as proposed by Krishnamoorthy et al. ⁽Reference Krishnamoorthy, Sniffen, Stern and Van Soest⁶⁹⁾, 6·6 enzyme units (IU) of S. griseus protease were used per g of sample DM to break down the peptide bonds of the feedstuffs. The sample (0·5 g) was incubated in 40 ml of a borate–phosphate buffer (pH 8·0) for 1 h, then 10 ml protease solution (0·33 IU/ml) were added. Roe et al. ⁽Reference Roe, Chase and Sniffen⁸²⁾ observed a fixed ratio of enzyme:CP and pH of phosphate buffer solution of 6·7 more effective for studying the enzyme degradation of feeds. These researchers assumed that all the protein that remained insoluble after an incubation of 18 h for concentrate feed and 48 h for forages was potentially rumen undegradable. The procedure then consisted of incubation of the test feed for the appropriate time and then filtration of the whole contents through a filter paper. After washing the residues with distilled water, the N in the residue was estimated by the Kjeldahl method. Degradability of the feedstuffs was then calculated as percentage of the total CP as follows:

A later modification of this method was proposed by Licitra et al. ⁽Reference Licitra, Van Soest, Schadt, Carpino and Sniffen⁸⁵⁾ who used a fixed ratio of enzyme:true protein as determined by tungstic acid precipitation. The authors found a significant effect on the estimate of degradable N when compared with the original procedure. They also reported a significantly higher degradation when the buffer solution had a pH of 8 instead of 6·7. This finding differed from that of Krishnamoorthy et al. ⁽Reference Krishnamoorthy, Sniffen, Stern and Van Soest⁶⁹⁾ where optimum pH for enzyme activity was 8. In fact, Licitra et al. ⁽Reference Licitra, Lauria, Schadt, Caprpino, Sniffen and Van Soest⁸⁴⁾ suggested that the pH of the buffer solution should be similar to rumen conditions despite the pH of 8 required for the optimum enzyme activity.

A drawback in relying on the proteolytic activity of just one specific bacterium, as opposed to a group of micro-organisms, is that the whole range of activities towards the different nitrogenous substrates found in the rumen may not be present in a single enzyme. Russell⁽Reference Russell, Gilchrist and Mackie⁹¹⁾ showed that not all species of bacteria can degrade and utilise nitrogenous substrate to the same extent. This suggests that the use of a single enzyme, even if extracted from a rumen micro-organism, may not be appropriate for an accurate estimation of the total proteolytic activity of the whole rumen fluid. Luchini et al. ⁽Reference Luchini, Broderick and Combs⁹²⁾ compared the activity of a mixture of trypsin, carboxypeptidase B, chymotrypsin and carboxypeptidase A with that of the strained rumen fluid in incubations with fifteen feeds. Degradation rates using strained rumen fluid ranged from 0.008 to 0.250/h. However, results using the enzyme mixture as the inoculum source, detected no differences in degradation rates among feeds. This indicates that the commercial enzymes employed did not mimic the activity of the strained ruminal fluid.

The use of commercial enzymes has not provided consistent results. Theoretically, there may be a problem that only one of the many microbial sources of proteases that exist in the rumen is used to prepare a purified enzyme. More serious, however, is the possibility that the protease activity present in the purified enzyme does not exist in the rumen. A comparison of protein degradation by crude extract from the rumen and by S. griseus protease showed variations in degradation for several feeds⁽Reference Mahadevan, Sauer and Erfle³⁷⁾. Furthermore, some researchers have questioned the use of any commercial protease for the estimation of ruminal protein degradation due to differences in specificity and mode of action of proteases⁽Reference Kopecny, Vencl, Kyselova and Brezina⁸⁶⁾. However, Chaudhry⁽Reference Chaudhry^14, Reference Chaudhry¹⁵⁾ has supported the use of S. griceus due to its ability to predict protein degradation of both purified substrates or commonly used feeds. In fact, the enzyme-based estimates for these feeds were reasonably compared with their in sacco counterparts. Therefore, this enzyme deserves further attention in standardising and validating its use to estimate rumen degradation of feeds.

In vitro methods involving gas production

A number of reports have described adaptations of the first stage of the in vitro digestibility method of Tilley & Terry⁽Reference Tilley and Terry¹⁷⁾ to permit the measurement of the volume of gas produced by fermenting feedstuffs⁽Reference Menke, Raab, Salewski, Steingass, Fritz and Schneider^11, Reference Krishnamoorthy, Rymer and Robinson^12, Reference Raab, Cafantaris, Jilg and Menke^93–Reference Cone, VanGelder and Driehuis¹⁰⁰⁾. The procedure used for gas collection and measurement ranges from the use of calibrated syringes⁽Reference Menke, Raab, Salewski, Steingass, Fritz and Schneider¹¹⁾ and pressure transducers⁽Reference Theodorou, William, Dhanoa, McAllan and France⁹⁴⁾ to computerised monitoring⁽Reference Pell and Schofield⁹⁶⁾. Rumen fluid obtained from fistulated cattle or sheep was used for fermentation of substrates. The rumen fluid had been diluted in either bicarbonate or phosphate or bicarbonate–phosphate buffers or in a modified medium⁽Reference Raab, Cafantaris, Jilg and Menke⁹³⁾ which was a complex buffer containing micro-minerals, cysteine, resazurin as well as both phosphate and bicarbonate mineral mixtures. According to Pell & Schofield⁽Reference Pell and Schofield⁹⁶⁾, gas is produced from metabolic energy sources, and they measured the potential of different sources (monosaccharides, polysaccharides, pectin, starch, cellulose and hemicellulose) for conversion to CO₂ and CH₄. Gas produced has been reported to be primarily from the fermentation of digestible carbohydrates by the activity of rumen microbes. However, differences were found between researchers who suggested that gas production was also affected by other factors such as the nature of the buffer, and source and/or handling of the fermenting micro-organisms⁽Reference Theodorou, William, Dhanoa, McAllan and France^94, Reference Cone, VanGelder and Driehuis^100–Reference Harris, Barlet and Chamberlain¹⁰⁷⁾. This aspect requires further attention when such in vitro studies are involved to estimate degradation of ruminant feeds.

The advantage of the gas production systems is that they can be automated, thus reducing the labour input. However, automated gas production methods are expensive and may not handle large numbers of samples. While manual methods are considered cheap, they are labour intensive and restricted in capacity. The results obtained from automated or manual systems are dependent on several procedural details. Table 5⁽Reference Theodorou, William, Dhanoa, McAllan and France^94, Reference Pell and Schofield^96–Reference Williams, Givens, Owen, Axford and Omed¹⁰²⁾ shows the effect of several factors on gas production. In addition to these factors, the results varied with the type of system and the source, activity and consistency of the rumen fluid used⁽Reference Krishnamoorthy, Sniffen, Stern and Van Soest⁶⁹⁾.

Table 5 Factors affecting the accuracy of in vitro gas production technique involving rumen fluid (RF)

In vitro method involving faeces

It is well established that certain amounts of cellulose and hemicellulose are fermented in the large intestine, because many bacterial species in the rumen are also represented in the hind gut from where bacterial residues are subsequently passed in the faeces⁽Reference Van Soest¹⁰³⁾. Therefore, the suspension of sheep faeces in buffer might be capable of acting as an inoculum for the initial fermentation of feed samples, in place of rumen liquor. In vitro digestibility determined by using sheep faeces as an inoculum correlated well with the in vivo digestibility⁽Reference El Shaer, Omed and Chamberlain¹⁰⁴⁾. These researchers demonstrated the potential of using micro-organisms from faeces instead of rumen fluid in the two-stage procedure of Tilley & Terry⁽Reference Tilley and Terry¹⁷⁾. Akhter et al. ⁽Reference Akhter, Owen, Tembo, Theodorou and Deaville¹⁰⁵⁾ showed potential for cattle faeces to be used as an alternative to rumen liquor that is collected from rumen-fistulated sheep for use in the in vitro digestibility assay of forages. When a 48 h acid pepsin digestion, the second stage of the Tilley & Terry technique⁽Reference Tilley and Terry¹⁷⁾, was included, the organic matter digestibility values and the ease of filtration of undigested residues were increased. They also investigated the accuracy of estimating the organic matter digestibility of eight forages determined by using rumen liquor from three sheep (y) and faeces from two cows (x). All regressions between sheep rumen fluid and cow faeces-based organic matter digestibility were significant (P < 0·001) with residual standard deviations of between ± 0·019 and ± 0·022. In another study Akhter & Hossain⁽Reference Akhter and Hossain¹⁰⁶⁾ concluded that fresh or frozen freeze-dried cow faeces were satisfactory and repeatable substitutes for rumen liquor as a source of micro-organisms for in vitro digestibility assay for forages. If proven to work under most conditions and satisfactorily standardised among laboratories, this method provides an opportunity to overcome the main disadvantage of the methods that require surgically prepared animals to obtain fresh rumen liquor.

Harris et al. ⁽Reference Harris, Barlet and Chamberlain¹⁰⁷⁾ successfully demonstrated an experiment using dairy cow faeces rather than using the traditional rumen fluid in the gas pressure transducer technique. It was observed that cumulative gas production from the test feeds for faecal inoculum showed a correlation of R ² 0·95 with data obtained from the use of rumen fluid. Nsahlai & Umunna⁽Reference Nsahlai and Umunna¹⁰⁸⁾ compared rumen fluid inocula with reconstituted sheep faeces to predict in vivo digestibility and intake and concluded that gas production using faecal inoculum was positively related to gas production using rumen fluid inoculum particularly at 48 h (R ² 0·85) of incubation. It was further confirmed by these authors that in vitro DM digestibility estimated using reconstituted sheep faecal inoculum had a positive correlation (R ² 0·88) with in vitro DM digestibility measured using rumen fluid. Since sheep faeces are much more easily obtained than rumen fluid, the faecal inoculum method would seem to have a distinct advantage in use. This advantage may be of special value as the use of fistulated animals for the purpose of nutritional studies has been criticised due to the cost and animal welfare implications of keeping fistulated animals.