Livestock Research for Rural Development 9 (4) 1997

Citation of this paper

In vitro gas production and washing losses of tropical crop residues for ruminants and pigs

J Ly*, Nguyen Van Lai, Lylian Rodríguez and T R Preston

University of Tropical Agriculture, Thu Duc, Thanh Pho Ho Chi Minh, Vietnam
* Present address: Swine Research Institute, carretera del Guatao km 1, Punta Brava, La Habana, Cuba


Crop residues consisting of the aerial part of cowpea (Vigna sinensis), groundnut (Arachis hypogea) or leaves of banana (Musa sapientum) and cassava (Manihot esculenta Crantz) were examined for washing losses (WL) of dry matter and in vitro gas production characteristics by comparing rumen fluid and fresh voided sow faeces as source of inoculum. Samples were used in the fresh state of after sun-drying. A washing time of 90 min appeared to be the more convenient time for determining WL values. On average, the sun-dried samples had 62.8% higher WL (P<0.001) than the fresh samples. A highly significant relationship (R2 = 0.70; P<0.001) was found between in vitro gas production over 96 hr from materials incubated with sow faeces and the same index obtained from cow rumen fluid as sources of inoculum. The washing effect had a minor influence (P<0.05) on the in vitro potential gas production (a + b) whereas the source of inoculum had no significant effect on the in vitro rate constant "c" of gas production. The type of crop residue significantly influenced (P<0.001) both the value of potential gas production and the rate at which it is produced.

It is suggested that inocula prepared either from cow rumen fluid or freshly voided sow faeces can be used for a rapid screening of the degradation characteristics of a relatively large amount of unknown samples.

Keywords: In vitro gas production, tropical crop residues, washing losses, rumen fluid, pig faeces


The use of crop residues in animal feeding is a very common practice in tropical countries. However, these feed resources have generally been directed to ruminant production, due to the high level of the cell wall fraction (Preston and Leng 1987). There is now an increasing interest in using certain crop residues in pig feeding, particularly for sows, as the aerial parts of several tropical crops are relatively rich in protein (Ly 1996). For example, the crude protein content in groundnut vines and leaves can reach 15.6 and 19.2% respectively (Buy Van Chinh et al 1997). In the case of cassava hay and leaves, the concentration of crude protein can be as high as 24.9% (Mararuf et al 1995: Wanapat et al 1997).

As with many other facets of livestock production, research in tropical countries on the nutritive value of crop residues lags behind such activities in temperate countries (Owen and Jayasuriya 1989). A low investment in research in tropical countries, most of which are undergoing structural adjustment programmes with severe curtailment of State spending, is one reason for the relatively low volume of research on tropical crop residues in their country of origin. There is a need to develop, for use in tropical countries, simple and cheap techniques which can be used to screen rapidly new feed resources. In this connection, recently developed techniques such as the dry matter washing loss (WL) and in vitro gas production could play an important role.

In previous papers (Ly and Preston 1997; Ly et al 1997) the washing loss technique and the characteristics of in vitro gas production have been studied in tropical leaves, as potential feeds for ruminants. In this communication, a study of the same type is reported concerning the potential of some tropical crop residues as feeds for ruminant and non-ruminant animals, especially pigs.


Materials and methods


Four crop residues were studied. Samples of the aerial part of cowpea (Vigna sinensis) and groundnut (Arachis hypogea) or leaves from cassava (Manihot esculenta Crantz) and banana (Musa sapientum) were collected at harvest. All the samples were obtained on farm with the exception of groundnut vines, which were from the local market at Thu Duc. Representative samples from every crop residue were cut in very small pieces with a knife and then used either in fresh state or sun-dried during three days. Groundnut vines were already wilted when they were obtained. In Table 1 is shown the DM content of the samples.

Experiment 1

The washing loss (WL) of the crop residues in fresh and sun-dried state were determined in duplicate for three representative samples collected during the same month at the farm (April 1997, during the dry season). The samples were washed in standard nylon bags in a semi-automatic washing machine during 30, 60, 90 and 120 min, as described previously by Ly and Preston (1997). The samples were not milled with the aim of minimizing losses of insoluble substances through the bag pores during washing. The dry residue contained in the bags was determined by microwave radiation until constant weight as recommended by Undersander et al (1993).

A linear model was used (SAS 1993) to compare the WL of the crop residues. The model took into account the type of sample used (fresh or sun-dried) and the time spent for washing (30, 60, 90 and 120 min). An analysis of variance was conducted using the Minitab 10.2 software (Minitab Inc. State College, PA 16801, USA). In the appropiate cases the means were separated by the Duncan's multiple range test (Steel and Torrie 1980).

Experiment 2

Three samples of either unwashed or washed (90 min) crop residues from Experiment 1 were ground in a laboratory mill and thereafter used to study the characteristics of in vitro gas production following the method of Menke et al (1979) and Menke and Steingass (1988). Two sources of inoculum were used. Rumen fluid was obtained from three rumen fistulated Bos indicus heifers fed roughages as outlined by Ly et al (1997). Faeces just voided by three Mong Cai sows were also used as inoculum. The sows were fed on sugar cane juice and duckweed (Lemna sp) or ensiled cassava leaves as the only source of protein. No antibiotics of other antimicrobial substances were included in the diet of the sows. The faeces were suspended in a saline solution (1:3, w/w) in order to obtain a slurry and thereafter thoroughly mixed and filtered by the same method used for the rumen fluid.

The two sources of inoculum were mixed with a sodium and ammonia bicarbonate buffer (35 g NaHCO3 plus 4 g NH4HCO3 per litre), as outlined by Menke and Steingass (1988) in a ratio of 1:2 (v/v). About 200 mg of air-dry samples were introduced into glass syringes of 100 ml, and 30 ml of the buffered inoculum were sucked through a silicone tube attached to the needle top of each syringe. Then the gas bubbles were removed and the silicone tube was clamped, the position of the lubricated piston was recorded, and the syringes were placed in the water bath at 39oC as described elsewhere (Ly et al 1997). Gas production was recorded at 3, 6, 12, 24, 36, 48, 60, 72, 84 and 96 hr.

The dry matter (DM) of the filtered rumen fluid and the diluted sow faeces was determined by microwave radiation until constant weight (Undersander 1993). The pH values and short chain fatty acids (SCFA) concentrations of the media before and after incubation were determined as reported by Ly et al (1997).

The data for gas volume recording were fitted to the exponential equation p = a + b (1 - e-ct) according to McDonald (1981) where "p" is the gas production at time "t", "a + b" the potential gas production, and "c" the rate constant. A program prepared by Chen (1995) was used to compute the recorded information. A linear model (SAS 1993) was used to compare the in vitro gas production of the crop residues, taking into account the source of inoculum (rumen fluid or sow faeces) and the type of sample used (unwashed or washed).The analysis of variance was carried out as described in Experiment 1.



Experiment 1

There were no significant interactions among variables for the measurements studied. Washing time significantly affected (P<0.001) the WL values of samples (Table 2). The WL values were higher after 120 min compared to 30 and 60 min of washing time. Ninety min appeared not to be different in value from a lower or a higher washing time. The method of processing had a highly significant influence (P<0.001) on WL values in favour of sun-drying. On average, the sun-dried materials had WL values 62.8% higher than the crop residues in the fresh state.

Table 3 shows the individual WL values of the different crop residues studied, and the estimation of this index at zero time. The predicted WL values at zero time were 23.1 and 21.3% for linear and quadratic equations respectively, and this was independent of the method of processing (P>0.10). The comparison betwen both types of estimation did not show any significant effect either, and the R2 value for the quadratic estimation was only slightly higher (R2 = 0.55) than for the linear procedure of calculation (R2 = 0.51). Overall, the crop residues examined in the fresh state had a zero time WL value which was rather low (15.0%). The comparable value for sun-dried samples was practically double (29.3%).

A trend was found for the fresh materials to have higher correlation values (R2 = 0.59) than the corresponding figures for the same, sun-dried type of sample (R2 = 0.47) in the estimation of WL values at zero time. A practically constant, time-independent WL value for fresh banana leaves (19.4%) was noteworthy. However, the cowpea residues had a highly significant (P<0.001) time-dependent WL value, either in the fresh (R2 = 0.81 in both types of regressions), or in sun-dried state: (0.72<R2<0.74 in the case of the linear and quadratic effects).

Experiment 2

Some characteristics of four different samples from both inocula used in this experiment are shown in Table 4. There was no significant differences in DM concentration and pH between them, but the filtered sow faeces slurry had a significantly lower SCFA concentration (P<0.05) than the filtered cow rumen fluid. After 96 hr of in vitro incubation, equal gas production (7.5 and 7.8 ml) and pH values (6.94 and 7.06) were observed in 4 blanks of inocula from sow faeces and rumen fluid, respectively. SCFA production was negligible and independent of the source of inoculum (0.20 and 0.13 mmol, respectively).

Table 5 shows total gas and SCFA production after 96 hr of in vitro incubation. There was no statistically significant interaction (P>0.10) involving the source of inoculum on in vitro total gas production. However, a highly significant (P<0.001) material*washing effect interaction was observed. The type of inoculum had a highly significant influence (P<0.001) on total gas production over 96 hr, which was high when rumen fluid was used compared with sow faeces (34.1 and 25.3 ml/200 mg DM respectively).

Our data showed a reasonably close (R2 = 0.70) and highly significant relationship (P<0.001) between gas production over 96 hr from materials incubated in the pig faeces slurry (X= ml) and the same index obtained from cow rumen fluid (Y= ml) as shown in the equation:

Y = 8.03 + 1.03 X (Syx = 5.62).

Total gas production was also significantly affected by the type of crop residue examined, with the highest value recored for cassava leaves (32.4 ml/200 mg DM) and the lowest for banana leaves (25.3 ml/200 mg DM). Total gas production was only slightly influenced (P<0.10) by prior washing of the samples. In fact, washing caused an increase in gas production in the groundnut and cowpea crop residues, while it caused a decrease in this same index in cassava and banana leaves, respectively.

A weak inoculum*material interaction existed (P<0.05) for in vitro total SCFA production after 96 hr of incubation, but no other significant interaction was found. The effect of washing was more marked (P<0.001) in total SCFA production than the effect of inoculum source (P<0.01). The type of crop residue appeared to cause less differences (P<0.05) in this index. In fact there was a relatively small difference between the least and the greatest amount of total SCFA produced (1.00 and 1.25 mmol/200 mg of groundnut and cassava leaves respectively).

A highly significant interaction effect (P<0.001) was found for inoculum*material, inoculum*washing, and material*washing for the pH values of the in vitro incubation media after 96 hr. The rumen fluid showed significantly (P<0.001) higher pH values than sow faeces slurry (7.01 and 6.90 respectively). On the other hand, prior washing of the samples significantly increased (P<0.001) the pH from 6.93 to 6.97. The legume residues had significantly lower final pH values in the incubation milieu than the cassava and banana leaves.

Total gas production over 96 hr could not be explained by SCFA production (P<0.10) in the case of cassava leaves (R2 = 0.43) and banana leaves (R2 = 0.21). On the contrary, gas production over 96 hr was found to be directly associated to SCFA production in the case of groundnut vines. This relationship fitted the equation:

Y = 17.6 + 13.84X (Syx = 4.82),

where Y = ml of gas per 200 mg DM and X = mmol SCFA per 200 mg DM (R2 = 0.71; P<0.001). Similarly, gas production was found to be directly related to SCFA production from cowpea. In this case the relationship fitted the equation:

Y = 5.02 + 19.8X (Syx = 4.13).

In fact, cowpea had the strongest relationship between total gas and SCFA production over 96 hr (R2 = 0.91; P<0.001).

The results for the cumulative in vitro gas production and sample degradation charateristics are given in Table 6. The interaction concerning the type of crop residue (either inoculum*material or material*washing) had a highly significant influence (P<0.001) on the characteristics of in vitro gas production such as the potential gas production (a + b) and the rate of gas production "c". The inoculum*washing interaction was only significant (P<0.01) in the case of the potential gas production. The washing effect had a minor influence on the potential gas production (P<0.05), whereas the source of inoculum was without significant effect on the constant "c. The type of crop residue significantly influenced (P<0.001) both the value of potential gas production and the rate at which it was produced.

Actual in vitro gas production from unwashed crop residues at 3, 6, 12, 24, 48 and 96 hr could be predicted from the WL of the materials (Table 7). This relationship was calculated from pooled data of samples incubated with either cow rumen fluid of sow faeces slurry, taking into account that it had previously been observed an interdependence (R2 = 0.70; P<0.001) between both source of inocula in total in vitro gas production after 96 hr of incubation (see Table 5), and that there was no inoculum influence (P>0.10) on the calculated rate constant "c" values (Table 6).

In this connection, the highest correlation value found was at 24 hr (R2 = 0.72; P<0.01). This type of association could not be established from washed crop residues (R2<0.10; P>0.10).


In this study an attempt was made to determine if some properties, such as the washing loss of DM in different types of crop residue, could be associated with the degradation characteristics when using different sources of inoculum. Some practical implications can be derived from these relationships, considering the suggestion of Chermiti et al (1996) that washing loss of DM in fact represents the detergent soluble substances, or cell contents of fibrous feeds. The proposal of Menke et al (1979) and Menke and Steingass (1988) that in vitro gas production can describe the pattern of feed degradation in the rumen is well known. However, the use of this technique for samples previously washed by the WL method, could provide an estimation of the pattern of microbial degradation of different fractions of the cell wall in the large intestine of non-ruminant animals. This follows from the understanding that in this type of animal, including the pig, the cell contents disappear from the gastrointestinal tract before the ileo-caecal valve (Ly 1996).

From the point of view of the method of estimation of washing loss, evidence is provided that the water solubility of some fractions of feeds is practically complete after 90 min of washing time, thus confirming previous observations made with trees and shrubs leaves (Ly and Preston 1997). On the other hand, it was found that the in vitro gas production during the first 24 hours of incubation may be largely explained by the amount of the soluble fraction of fibre-rich crop residues such at those used in the present investigation.

It could be assumed that a wide range of values in the characteristics of in vitro gas production was to be expected in such different types of crop residues such as leaves of banana and cassava, and the aerial part of groundnut and cowpea. In this connection, there is no explanation for every phenomenum observed in the pattern of in vitro gas production. Interpretations of the results are also made more difficult by the fact that there is little information available considering the in vitro degradability of non-cereal feed materials from the wet tropics.

Without doubt, more studies are needed to develop the newer techniques such as the washing loss and in vitro gas production for application in non-ruminants. In pigs, there is some information on the application of other more conventional in vitro and in situ techniques (see for example, Metz and Van der Meer 1985; Graham et al 1989).

It is hypothesised that the increase in in vitro gas production brought about by prior 90 min washing, observed in crop residues of legumes, could be a consequence of the elimination of antinutritional factors which are soluble in water, such as anti-tryptic factors (D'Mello 1992). The reversal could be true in the case of cassava and banana leaves. It might be considered that, after sun-drying, cassava leaves are practically devoid of cyanogenic compounds (Kumar 1992). Therefore the washable fraction of cassava leaves should be an adequate substrate for sustaining in vitro gas production. In this connection Maaruf et al (1995) have found that cassava leaves are highly degradable due to the high ratio of arabinose to xylose in the hemicellulose fraction. Arabinose sugars are very water soluble. Tannins present in banana leaves would not be water-soluble compounds in the fresh state of this material.

The close relationship found between cow rumen fluid and sow faeces slurry, as inocula for the in vitro gas production test of feed materials, suggests that either ecosystem could be used as a tool for a rapid screening of the degradation characteristics of feed samples. In fact, a noteworthy parallelism was observed in data arising from incubation with both digestive ecosystems of the crop residues used in this investigation. Nevertheless, since the preparation of the sow faeces inoculum was rather empirical, more information is needed to describe the pattern of substrate degradation in a digestive ecosystem such as the sow large intestine.


Thanks are due to Miss Tran Thi Quynh My, from the staff of the University of Ttropical Agriculture at Ho Chi Minh City, for excellent technical assistance. This work was supported by the Swedish Agency for Research Cooperation with Developing Countries (SAREC).


Bui Van Chinh, Le Viet Ly and Nguyen Huu Tao 1997 Storage and use of groundnut vines as animal feed. In: Better use of locally available feed resources in sustainable livestock-based agricultural systems in South-East Asia. Proceedings of a FAO Regional Seminar-Workshop. Phnom Penh p 44-47

Chen X B 1995 XBC Laboratory. An utility for processing data of feed degradability and in vitro gas production. Version 4.0. Rowett Research Institute. Aberdeen

Cherrmiti A, Nefzaoui A, Teller E, Vanbelle M, Ferchichi H and Rokbani N 1996 Prediction of the voluntary intake of low quality roughages by sheep from chemical composition and ruminal degradation characteristics. Animal Science 62:57-62

D'Mello J P F 1992 Nutritional potentialities of fodder trees and fodder shrubs as protein sources in monogastric nutrition. In: Legume trees and other fodder trees as protein sources for livestock (A Speedy and P L Pugliese, editors) FAO Animal Production and Health Paper No 102. Rome p 145-160

Graham H, Lowgreen W and Aman P 1989 An in vitro method for studying digestion in the pig. 2. Comparison with in vivo ileal and faecal digestibility. British Journal of Nutrition 61:689-698

Kumar R 1992 Anti-nutritional factors, the potential risks of toxicity and methods to alleviate them. In: Legume trees and other fodder trees as protein sources for livestock (A Speedy and P L Pugliese, editors) FAO Animal Production and Health Paper No 102. Rome p 145-160

Ly J 1996 Una reseña corta sobre avances en estudios de procesos digestivos en cerdos alimentados con dietas tropicales no convencionales. Revista Computadorizada de Producción Porcina 3(1):8-18

Ly J and Preston T R 1997 An approach to the estimation of washing losses in leaves of tropical trees. Livestock Research for Rural Development 9(3):5-9

Ly J, Nguyen Van Lai and Preston T R 1997 A study of washing losses and in vitro gas production characteristics of nine leaves from tropical trees and shrubs for ruminants. Livestock Research for Rural Development 9(3):26-33

Maaruf K, Christensen D A and Campbell G L 1995 Comparison of protein and cell wall degradation of selected tropical and temperate roughages. In: Increasing animal production with local resources (Guo Tingshuang, editor) Proceedings of the Second International Conference. China Forestry Publishing House. Beijing. pp 163-171

McDonald E 1981 A revised model for the estimation of protein degradability in the rumen. Journal of Agricultural Science, Cambridge 96:251-252

Menke K H, Raab L, Salewski A, Steingass H, Fritz D and Schneider W 1979 The estimation of the digestibility and metabolizable energy content of ruminant feedingstuffs from the gas production when they are incubated with rumen liquor. Journal of Agricultural Science, Cambridge 93:217-222

Menke K H and Steingass H 1988 Estimation of the energetic feed value obtained from chemical analysis and in vitro gas production using rumen fluid. Animal Research and Development 28:7-55

Metz S H M and Van Der Meer J M 1985 Nylon bag and in vitro technique to predict in vivo digestibility of organic matter in feedstuffs for pigs. In: Digestive physiology in the pig (A Just, H Jorgensen and J A Fernandez, editors) National Institute of Animal Science Institute Report No 590. Kobenhavn p 373-376

Owen E and Jayasuriya M C N 1989 use of crop residues as animal feeds in developing countries. In: Research and Development in Agriculture. Longman 6(3):129-138

Preston T R and Leng R A 1987 Matching ruminant production systems with available resources in the tropics and subtropics. Technical Center for Agricultural and Rural Cooperation. Pernambul Books, Armidale pp 245

SAS 1982 SAS user's guide. Statistical Analysis Systems Institute. Cary NC

Steel R G D and Torrie J H 1980 Principles and procedures of statistics. Mc-Graw-Hill Book Co. New York

Undersander D, Mertens D R and Thiex N 1993 Forage analysis procedures. National Forage Testing Association. Omaha pp 154

Wanapat M, Pimpa O, Petlum A and Boontao U 1997 Cassava hay: a new strategic feed for ruminants during the dry season. In: Better use of locally availabe feed resources in sustainable livestock-based agricultural systems in South-East Asia. Proceedings of a FAO Regional Seminar-Workshop. Phnom Penh p 26-29

Received 1 July 1997

Return to top