Ruminal plant cell wall digestibility estimated from digestion and passage kinetics utilizing mathematical models

EISEVIER

Animal Feed Science and Technology 52 ( 1995 ) 159-l 73

Ruminal plant cell wall digestibility estimated from digestion and passage kinetics utilizing mathematical models P. Huhtanen*q’, S. Jaakkola’, U. Kukkonen Department ofAnimalScience, University ofHelsinki, P.O. Box 28, FIN-00014 Helsinki, Finland Received 23 December 1993; accepted 6 July 1994

Abstract

Methods to describe the digestion and passage kinetics for estimating ruminal neutral detergent fibre digestibility (RNDFD) with simple mathematical models were evaluated in cattle fed a diet of hay, barley and urea (600, 396 and 4 g dry matter (DM) kg-’ total DM) at two levels of intake. The parameters of digestion kinetics were determined by in situ incubation and rumen evacuation technique. Digesta passage kinetics was estimated from duodenal and faecal Cr and Yb concentrations either from exponential decline in marker concentration or by using two-compartmental models with gamma age dependency in the first compartment. Increasing the feeding level from 40 to 80 g DM kg- ’live weight0.75 decreased (P< 0.05) thedigestibilityofDM (0.775 vs. 0.749), NDF (0.758 vs. 0.707) and cell solubles (0.838 vs. 0.824). The amount of total digesta, digesta DM content and the amount of DM in the rumen increased as feed intake increased. The relative increase in rumen pool size of digestible NDF was greater than in that of indigestible NDF (0.72 vs. 0.34). Using in situ digestion kinetics data and the exponential decline in marker concentration in a simple rumen digestion model yielded RNDFD estimates which were much lower than those based on rumen evacuation data, and only about 0.70 of the total in vivo NDF digestibility. This was partly because of the slower (PC 0.01) rate of digestion observed with the in situ method than estimated from rumen evacuation (0.0458 vs. 0.0607), and partly because of an underestimation of rumen residence time by the passage model. Incorporation of the selective retention of feed particles in the digestion model yielded RNDFD estimates which were 0.13 units higher than those based on exponential marker concentration decline curves. The kinetics parameters for this model were calculated from duodenal marker concentrations using two-compartmental passage models. The RNDFD * Corresponding author. ’Present address: Institute of Animal Production, 3 1600 Jokioinen, Finland.

Agricultural Research Centre of Finland, FIN-

0377-8401/95/$09.50 0 1995 Elsevier Science B.V. All rights reserved SSDZO377-8401(94)00700-4

160

P. Huhtanen et al. /Animal Feed Science and Technology 52 (1995) 159-173

estimates were 0.94-0.96 of the total in vivo NDF digestibility when the passagekinetics parameters, estimated with the two-compartmental model, were used in combination with the rumen evacuation derived rate of digestion. The values for RNDFD were higher when Cr and faecal sampling were used rather than Yb and duodenal sampling to estimate passage kinetics. Keywords: Cattle, beef; Digesta passage kinetics; Modelling

1. Introduction

Potential extent and rate of digestion and rate of passage are the determinants of ruminal fibre digestibility. Generally, ruminal fibre digestibility is determined as the ratio (intake-duodenal flow) /intake. It can also be calculated from the kinetics parameters using mathematical models (Allen and Mertens, 1988). Waldo et al. ( 1972) developed the first model including digestible and indigestible fibre fractions, and rates of digestion and passage. This model is limited by the assumption that the probability of particles escaping from the rumen is independent of their age and size. The flow of feed particles from the rumen is, however, a selective process. Both the size (Dixon and Milligan, 1985 ) and specific gravity of particles (desBordes and Welch, 1984) affect the rate of passage. Slower passage rate of digestible than indigestible fibre (Tamminga et al., 1989; Huhtanen and Khalili, 199 1; Jaakkola et al., 199 1) is also an indication of selective retention of feed particles in the rumen. Digesta kinetics parameters derived from duodenal sampling using models with time dependency have shown two pre-duodenal compartments (Pond et al., 1988). The first compartment probably represents the release from the non-escapable to the escapable pool and the second represents passage from the escapable pool. Selective retention of potentially digestible fibre is necessary to maximize ruminal fibre digestion (Allen and Mertens, 1988 ), and not more than IO20% of the potentially digestible fibre passes out from the rumen (Kennedy and Murphy, 1988). Various methods have been used to determine the constraints of fibre digestion. The rate of digestion (&) is usually determined by incubation in vitro (Smith et al., 197 1) and in situ but it can also be calculated from rumen evacuation data (Aitchison et al., 1986; Robinson et al., 1987). Rumen evacuation derived estimates for kd of NDF have been higher than those based on in situ incubation (Aitchison et al., 1986; Tamminga et al., 1989). Rate of passage (b) is estimated from exponential decline in ruminal or faecal marker concentrations after a pulse dose of markers in the rumen. This approach does not take into account the two ruminal compartments and selective retention of fibre in the rumen. Using these kp values with simple digestion models may therefore underestimate ruminal fibre digestibility. The objective of this study was to compare the estimates of ruminal NDF digestibility obtained from different models when various methods were used to

P. Huhtanen et al. /Animal Feed Science and Technology 52 (199.5) 159-I 73

161

determine the constraints of fibre digestion. The rate of digestion was measured by in situ incubation and the rumen evacuation technique. The passage kinetics was estimated using two markers (Cr-mordanted versus Yb-labelled hay), two sampling sites (duodenum versus rectum) and different passage models, or it was calculated as a ratio of estimated duodenal flow and rumen pool size. Some preliminary results of this work have been reported earlier (Huhtanen and Kukkonen, 199 1). The results for digesta passage kinetics are reported in Huhtanen and Kukkonen ( 1994).

2. Materials and methods Animals, diets and experimental procedures are described in detail in Huhtanen and Kukkonen ( 1994). Six ruminally and duodenally cannulated Friesian bulls (initial live weight (LW) 500 kg) were given a diet of (g kg- ’on DM basis ) chopped hay (600)) rolled barley (396) and urea (4). The diet was given in a switch-back design at levels of 40 (L) and 80 (H) g DM kg-’ LW”.75.Digestibility was determined by the total collection method and neutral detergent tibre (NDF) digestion kinetics by in situ incubation and rumen evacuation. Digesta passage kinetics were determined from duodenal or faecal Cr and Yb excretion curves and from rumen evacuation. 2.1, Digestion models The disappearance of NDF from the nylon bags was fitted to the equation of Orskov and McDonald (1979) originally proposed for ruminal protein degradation: P=a+b(

1-eekdr)

(1)

where P is NDF disappearance at time t, a is rapidly degraded NDF, b is degradable NDF and /cdis the rate of NDF digestion. In the following calculations (a + 6) is considered as one entity, i.e. potentially digestible NDF (D). Ruminal NDF digestibility (RNDFD) was calculated using the model proposed by Waldo et al. (1972): RNDFD, =p k,D kd+kp

(2)

where kd and D are as defined above and kp is the rate of passage from the rumen. Rate of passage was obtained from the exponential decline in Cr and Yb concentration in either duodenal or faecal samples. This model assumes that the probability of particles escaping from the rumen is independent of their age or size. Allen and Mertens ( 1988 ) proposed a model in which selective retention of feed particles is incorporated:

162

P. Huhtanen et al. /Animal FeedScience

and Technology 5.2 (199s) 159-I 73

RNDFD,, In this model kd and D are as defined earlier, k, is the rate of release from the rumen non-escapable to the escapable pool, and k, is the rate of passage from the escapable pool. The passage kinetics parameters were obtained from duodenal Cr and Yb excretion curves fitted to two-compartmental models with 1 to 4 orders of gamma age dependency in the first compartment. The first age-dependent compartment was assumed to be the non-escapable pool, and k, was calculated as the reciprocal of residence time in this compartment. Calculation of RNDFD from rumen evacuation data was according to the following model: RNDFD, =p kd kd+kp

(4)

In this model kd and kp are the rates of digestion and passage for total NDF. However, the rates of disappearance of digestible and indigestible fibre from the rumen are different, and the rates determined will not be first-order (Allen and Mertens, 1988). The model is valid only for the digestible fraction of NDF. RNDFD derived from a model including the two fibre fractions based on the potential digestibility is actually similar to that used for the in situ data: k,D RNDFD, =p kd+k,

(5)

The rates of passage and digestion for potentially digestible NDF were calculated from rumen evacuation data as described in the previous paper (Huhtanen and Kukkonen, 1994). Potential NDF digestibility of feed, ruminal and faecal samples was determined by a 12 day in situ incubation. The total NDF digestibility from digestion and passage kinetics was calculated by assuming a similar kd for NDF digestion in the rumen and hind-gut, and that the distal colon represented 0.50 of the post-duodenal transit time (TT). Ruminal digestibility was calculated using passage kinetics from duodenal sampling. Post-ruminal NDF digestibility was calculated as: NDFDf-NDFDd+

(D-NDFDf)

x ( 1-e-~xo.5*)

where NDFDf and NDFDd are estimates for NDF digestibility calculated using the passage kinetics parameters obtained from faecal and duodenal sampling. The passage parameters derived from the best-fit models were used. 2.2. Statistical analyses Digestibility data were analysed using the model: yok=Ai+Pj+Tk+eok

P. Huhtanen et al. /Animal Feed Science and Technology 52 (1995) 159-I 73

163

whereA, Pand Tare animal (i= 1...6), period (j= 1,2) and treatment (k= 1,2) effects. Rumen evacuation data were analysed using a model for split-plot design:

in which A, P and Tareas defined in the previous model, H is the effect of evacuation time (I= 1...3) with XSpCCtiVC iIItCraCtiOnS, and and are main-plot and sub-plot errors. The estimates for RNDFD using kpvalues obtained from different markers (Cr versus Yb, sampling sites (duodenum versus faeces) and passage models; 1 to 4 orders of gamma age dependency in the first compartment) were analysed according to a split-plot design as described by Huhtanen and Kukkonen ( 1994). The RNDFD estimates from different models were compared to in vivo NDF digestibility by calculating residual mean squared error (RSME ) and using regression analyses. eijk

eij&

3. Results The average N and NDF contents of the diet were 18.6 and 480 g kg-’ DM. Increasing feed intake decreased (PC 0.05) the digestibility of all dietary constituents (Table 1). The depression was greater for the cell wall fraction, but the difference was also significant (P-c 0.05 )for the cell solubles (OM -NDF) . The rate of NDF digestion in situ tended to increase (P= 0.056) and potential extent of digestion to decrease (P=O.O64) as DM intake increased (Table 2). Feeding level had no effect on kdderived from rumen evacuation but the values were on average 32.4% higher (P-c 0.01) than those estimated from in situ incubation. Potentially digestible NDF was lower when estimated from the in situ degradation curve than when determined by a 12 day incubation (769 vs. 804 mgg-‘). to increase with The amount of total digesta in the rumen tended (P=O.O78) DM intake (Table 3). Digesta DM content (PcO.01) and consequently rumen Table 1 The effect of feeding level on the digestibility of dietary components Diet

DM OM Nitrogen NDF ADF Cellulose Hemicellulose Cell solubles

L

H

0.775 0.794 0.772 0.758 0.743 0.779 0.763 0.838

0.749 0.761 0.736 0.707 0.694 0.723 0.708 0.824

SEM

Significance

0.0058 0.0059 0.0062 0.0087 0.0100 0.0086 0.0082 0.0033

* 1 * * * ** ** *

Significance: *P
164

P. Huhtanen et al. /Animal FeedScience and Technology 52 (1995) 159-l 73

Table 2 The effect of feeding level on the NDF digestion kinetics Diet

In situ a (me g-‘) b (me g-‘) D (mpg-‘) k., (h-l) Rumen evacuation kd (h-l)

L

H

135 645 780

60 698 758

SEM

Significance

20.9 24.1 6.3

0 0 0

0.0409

0.0507

0.0026

o

0.0604

0.0609

0.003 1

NS

Significance: “P
Digesta (kg) Digesta DM (g kg-‘) Pool size (kg) DM Crude protein NDF DNDF INDF ADF ADL Cellulose Hemicellulose RNA-N (g)

L

H

46.7 84.6

52.3 112.0

3.92 0.59 2.27 1.17 1.10 1.18 0.20 0.98 1.09 5.27

5.86 0.92 3.48 2.01 1.47 1.81 0.25 1.56 1.67 9.22

SEM

Significance

1.68 2.37

0 I

0.048 0.008 0.032 0.039 0.039 0.021 0.005 0.020 0.013 0.085

*** Q* *** *** ** *** ** ** n* ***

Significance: “P
DM pool (PC 0.00 1) were higher in cattle fed the H-diet than in those fed the Ldiet. Rumen NDF pool was proportionally 0.5 3 greater (PC 0.00 1) at high than at low DM intake. Because the potential NDF digestibility of ruminal digesta increased from 5 16 to 575 mg g- ’ (P-c 0.05) as DM intake increased, the difference in the pool size of digestible NDF was greater (0.72) than in that of indigestible NDF (0.34) or acid detergent lignin (0.25). Rumen pool size of microbial N estimated from purine bases of nucleic acids (Zinn and Owens, 1986) increased relatively more with DM intake than did the DM pool (0.75 vs. 0;49). Increasing DM intake decreased RNDFD estimated from in situ NDF kinetics using the kp value derived from the exponential decline in duodenal or faecal Cr and Yb concentrations (Table 4)) but the difference tended to be significant

P. Huhtanen et al. /Animal FeedScience

and Technology 52 (1995) 159-I 73

165

Table 4 The effect of feeding level on ruminal NDF digestibility estimated using parameters derived from in situ NDF kinetics, exponential decline in faecal or duodenal marker concentration and rumen evacuation Method -

Site

Marker

Diet

SEM

Significance

L

H

Cr Yb Cr Yb

0.537 0.497 0.557 0.516

0.497 0.470 0.516 0.493

0.0200 0.0146 0.0126 0.0143

o NS o NS

RE RE RE

INDF NDF DNDF”

0.587 0.621 0.656

0.546 0.587 0.621

0.0128 0.0105 0.0103

o o 0

RE RE RE

NDF DNDF” DNDFb

0.754 0.754 0.713

0.701 0.702 0.677

0.0089 0.0089 0.0033

* * ***

kd

kP

IS IS IS IS

Ma Ma Ma Ma

IS

IS IS RE RE RE

D D F F

IS, in situ; Ma, marker; RE, rumen evacuation; D, duodenal sampling; F, faecal sampling; L, low; H, high. Significance: ‘PC 0.10, *P< 0.05, -P<:O.OOl; NS, not significant. SEM, standard error of the mean. “Determined from faecal samples. bDetermined from feed samples.

(P
166

P. Huhtanen et al. /Animal Feed Science and Technology 52 (1995) 159-173

Table 5 The effect of feeding level on ruminal NDF digestibility estimated using in situ NDF digestion kinetics and digesta passage parameters estimated from duodenal sampling with non-linear models (GnG 1, n = 1 to 4 order of gamma age dependency in the first compartment) Marker

Diet

Passage model GlGl

G2Gl

G3Gl

G4Gl

Best-W

Cr

Low High SEM Significance

0.671 0.603 0.0166 *

0.656 0.613 0.0184 NS

0.663 0.622 0.0167 NS

0.663 0.636 0.0122 NS

0.658 0.617 0.0178 NS

Yb

LOW

0.594 0.584 0.0200 NS

0.616 0.582 0.0146 NS

0.630 0.596 0.0138 NS

0.643 0.607 0.0137 NS

0.630 0.605 0.0123 NS

High SEM Significance

’ The model with the smallest residual mean squared error. Significance: *P-C0.05; NS, not significant. Table 6 Comparison of methods to estimate rate of NDF digestion in estimating ruminal NDF digestibility in cattle fed at two levels of intake Marker

Cr

Diet

LOW

High SEM Significance Yb

LOW

High SEM Significance

Best-fit” linear

Non-linear IS

RE

IS

RE

0.537 0.470 0.0200 0

0.618 0.531 0.0147 *

0.658 0.617 0.0178 0

0.730 0.681 0.0047 **

0.497 0.470 0.0146 NS

0.582 0.532 0.0086 *

0.630 0.605 0.0123 NS

0.710 0.671 0.0048 9z8

’ The model with the smallest residual mean squared error. Significance: “P
The comparison of the effects of different methods to estimate NDF digestion and passage kinetics on RNDFD are shown in Table 6. When passage kinetics were based on the best-fit two-compartmental model rather than exponential decline in duodenal Yb concentration, the values for RNDFD were 0.13 units higher. A further increase of 0.07 units was obtained by replacing /cddetermined by in situ incubation with that obtained from rumen evacuation. A combination of the two-compartmental passage model and rumen evacuation derived /cdproduced RNDFD estimates very similar to those derived completely from rumen evacu-

P. Huhtanen et al. /Animal Feed Science and Technology 52 (1995) 159-l 73

167

ation and dietary rather than faecal INDF content (low DM intake, 0.7 10 vs. 0.7 13; high DM intake, 0.671 vs. 0.677). Assuming a similar rate for NDF digestion in the rumen and hind-gut, and that the distal colon represented 0.50 of the post-duodenal transit time, the proportion of ruminal digestion of total NDF digestion averaged 0.95 (Table 7). The proportion of NDF digestion occurring in the hind-gut increased slightly as feed intake increased. The estimates for total NDF digestibility were close to in vivo NDF digestibility when kd values derived from rumen evacuation were used. The comparison of RNDFD from different models and in vivo NDF digestibility is shown in Table 8. The values predicted from in situ data and linear passage models were lower (PC 0.00 1) than in vivo digestibility. The agreement was not good in terms of intercept and slope of the regression equation, which were significantly different from 0 and 1, and the correlations were relatively low. Incorporation of selective retention in the digestion model did not improve the correlation but decreased the bias. Using the kd values obtained from rumen evacuation rather than in situ digestion decreased the bias, improved the correlation and also provided regressions with intercepts and slopes closer to 0 and 1. A combination of in situ digestion kinetics and rumen evacuation derived k, improved the prediction of RNDFD as compared to values derived from in situ and linear passage models. 4. Discussion 4.1. Digestibility and rumen pool size Increasing DM intake caused a well-known depression in the digestibility of the diet. This was mainly associated with lower cell wall digestibility resulting from increased rate of passage. Faster passage rate with increasing DM intake was observed with both marker and rumen evacuation techniques (Huhtanen and Kukkonen, 1994). Staples et al. ( 1984), Robinson et al. ( 1987) and Bare ( 1989) also reported reduced rate of digestion with increasing intake. Slightly, although significantly, lower digestibility of cell solubles in cattle fed at high rather than at low DM intake may be related to improved microbial protein synthesis (Robinson et al., 1985) and excretion of undigested microbial residues in the faeces. Assuming that the proportion of rumen microbes in the liquid phase and those attached to the particles was not affected by intake, it can be calculated from rumen microbial pool size and liquid ( CoEDTA) and solid (NDF) passage rate (Oldham, 1984) that the production of microbial protein per kg DM intake increased 3 1% as DM intake increased. The finding of higher DM content in rumen ingesta as feed intake increased agrees with the observations of Hartnell and Satter ( 1979) and Robinson et al. ( 1987). As concluded by Robinson et al. ( 1987), physiological adaptation of rumen capacity with decreasing feed intake appears to result in a greater decrease in rumen DM components than in total rumen volume. It is also noteworthy that

P. Huhtanen et al. /Animal Feed Science and Technology 52 (1995) 159-l 73

168

Table I Partitioning NDF digestion in cattle fed at two levels of intake using NDF digestion kinetics parameters derived from in situ incubation or rumen evacuation and digesta passage kinetics parameters from duodenal and faecal sampling Diet

NDF digestibility

LOW

IS

High SEM Significance LOW

RE

High SEM Significance

Rumen

Hind-gut

Total

Rumen/total

0.630 0.605 0.0124 NS

0.035 0.035 0.0035 NS

0.665 0.640 0.0120 NS

0.948 0.946 0.0055 NS

0.710 0.671 0.0048 1

0.029 0.034 0.0030 NS

0.739 0.706 0.0070 *

0.960 0.951 0.0040 NS

Significance: *P
Site

Marker

Model

Intercept=

Slopeb

RSME”

Biasd

Correlation

kd

k,

IS

IS IS IS IS IS IS

Ma Ma Ma Ma RE RE RE

D D F F

Cr Yb Cr Yb INDF NDF DNDF

L L L L

0.55-* 0.6 l** 0.47** 0.56” 0.38’ 0.32 0.36

0.36 0.26” 0.49O 0.35 0.62 0.68 0.59

0.232 0.252 0.199 0.236 0.168 0.132 0.099

0.229** 0.249** 0.1970.233*” 0.166’*, 0.128”* 0.095””

0.555 0.294 0.520 0.298 0.618 0.567 0.491

IS IS

Ma Ma

D D

Cr Yb

NL NL

0.53” 0.45*

0.32. 0.45

0.103 0.105

0.096 0.105

0.357 0.382

RE RE RE RE

Ma Ma Ma Ma

D D F F

Cr Yb Cr Yb

L L L L

0.46”* 0.42* 0.30” 0.34”

0.47** 0.56* 0.72” 0.69

0.161 0.178 0.129 0.162

0.1580.176*-’ 0.128#* 0.1 60”*

0.815 0.723 0.857 0.758

RE RE

Ma Ma

D D

Cr Yb

NL NL

0.13 0.05

0.85 0.97

0.032 0.038

0.027# 0.034”

0.824 0.842

Significance:“Pi0.10;*Pt0.05;LPi0.01;”*<0.001. IS, in situ; Ma, marker; RE, rumen evacuation; D, duodenal sampling; F, faecal sampling; L, linear model (exponential decline in marker concentration); NL, non-linear model (best-fit model of twocompartmental models with 1 to 4 order of gamma age dependency in the first compartment ) . ’Intercept significantly different from 0. b Slope significantly different from 1. ’Residual mean squared error. d The difference between methods significant.

P. Huhtanen et al. /Animal Feed Science and Technology S2 (I 995) 159- 173

169

potential digestibility of NDF in rumen digesta was greater at high than low DM intake. This adaptation may decrease the depression in NDF digestibility as feed intake increases. Rumen evacuation method resulted in a higher rate of NDF digestion than the in situ degradation method. This agrees with the earlier observations of Aitchison et al. ( 1986)) Tamminga et al. ( 1989) and Huhtanen and Khalili ( 199 1). Lower numbers of cellulolytic bacteria (Meyer and Mackie, 1986) and lower particleassociated carboxymethylcellulase and xylanase activities within the bags than in the surrounding digesta (Huhtanen and Khalili, 1992) are consistent with these observations. The relatively small open surface area of the bags available (in the present study 3 1% of the total area) may have hindered free inflow and outflow of microbes and probably also some essential nutrients. Underestimation of the rate of digestion by the in situ technique could also be due to the lack of rumination (Aitchison et al., 1986). Kinetics of NDF digestion estimated from in situ data tended to underestimate the potential extent of digestion compared to the value obtained with a 12 day incubation. This suggests that the maximal extent of NDF digestion is not attained within 96 h, the longest incubation period used in the present study. Other studies (Robinson et al., 1986; Lippke et al., 1986) also suggest that maximal extent of NDF digestion is not always attained within 72 or 96 h time periods, often used as end-points of in vitro or in situ incubations. 4.2. Digestion models Ruminal NDF digestibility estimated from in situ digestion kinetics and exponential decline in duodenal marker concentration was 0.705 (Cr) and 0.660 (Yb) as a proportion of the total tract NDF digestibility. Low values for the proportion of ruminal NDF digestion were reported by Poore et al. ( 1990) and Messman et al. ( I99 1) using the same approach. For bromegrass hay the values ranged from 0.44 to 0.77, depending on the model used to estimate digestion kinetics and the maturity of forage (Messman et al., 199 1). These values were considerably lower than the values of 0.90-0.95 reported by Jones et al. ( 1988) using intestinally cannulated cattle fed similar diets. Grant and Weidner ( 1992 ), using a typical k, value of 0.05 for dairy cow diets and in vitro digestion kinetic data, calculated that only 0.30-0.44 of the potentially digestible bromegrass NDF was digested in the rumen. There are at least three points which should be considered as reasons for low values of RNDFD based on in situ digestion kinetics and exponential decline in marker concentration curves. First, in situ incubation tended to underestimate the rate of digestion. The possible reasons for this were discussed earlier. Using the kd from rumen evacuation rather than the in situ increased RNDFD by approximately 0.07 units, and also resulted in a considerably higher correlation between RNDFD and NDF digestibility in the total tract. A small correlation, relatively large intercept and small slope suggest that the in situ method was not very sensitive in detecting differences in NDF digestion kinetics which affected NDF digestibility. The rates of

170

P. Huhtanen et al. /Animal Feed Science and Technology 52 (1995) 159-l 73

NDF digestion determined by the two methods were poorly correlated (r= 0.20, n= 12). Second, the kp values calculated from the exponential concentration decline curves in duodenal digesta and faeces underestimate the residence time of cell walls in the rumen, and consequently RNDFD. Using the model proposed by Allen and Mertens ( 1988) to consider the selective retention of particles in the rumen, increased RNDFD estimates by 0.13 units. Combining rumen evacuation derived kd with the two-compartmental passage model resulted in RNDFD estimates of 0.96 (Cr) and 0.94 (Yb) of the total NDF digestibility. This suggests that external markers do not necessarily underestimate rumen residence time of cell walls as was concluded earlier by Robinson et al. ( 1987) and Tamminga et al. ( 1989). This discrepancy may partly be explained by the differences in the physical form of the marked feed and, more importantly, by the model used to estimate passage kinetics. We gave marked feed as chopped hay while Robinson et al. ( 1986) used Cr-mordanted particles between 0.2 and 1.0 mm. In earlier studies, simple digestion models were used to compare rumen evacuation versus in situ incubation and marker excretion (Aitchison et al., 1986; Tamminga et al., 1989). In this approach, the selective retention of feed particles is included in the passage rate determined by rumen evacuation but not in that calculated from exponential decline in marker concentration. In the present study, the reciprocal of k, determined as the exponential decline in faecal marker concentration represented only 0.75 of the compartmental retention time obtained from duodenal sampling using best-fit two-compartmental model. In addition, the simple firstorder kp of the digestion model considering the selective retention is a function not only of the release from the non-escapable to escapable pool and release from the escapable pool but also of digestion rate (Allen and Mertens, 1988 ) . For example, if the simple first-order k, for INDF is 0.025, then the &, for DNDF (kd=0.06) is 0.0156, i.e. proportionally 0.375 slower. In our study, the difference in compartmental retention time between duodenal and faecal sampling was relatively small, in agreement with the short compartmental post-duodenal retention time determined by introducing the markers into the duodenum (Huhtanen and Kukkonen, 1994). Consequently, the differences in RNDFD were small when passage kinetics were determined either from duodenal or faecal samples (0.032 for Cr and 0.016 for Yb). This implies that large errors are not made in estimating RNDFD by applying the passage kinetics data from faecal sampling using two-compartmental models. When the digestion of cell walls shifts from the rumen to the hind-gut, e.g. with ground forages, it may be more approariate to use passage kinetics parameters estimated from duodenal sampling. The present results. demonstrate the importance of the selective retention of feed particles in the rumen to maximize fibre digestion. On average, estimates of RNDFD were 25% higher when the models considering the selective retention were used. Selective retention is based on particle size (Dixon and Milligan, 1985 ) and specific gravity of the particles (desBordes and Welch, 1984). Gas produced during the fermentation delays the increase in specific gravity of particles thereby

P. Huhtanen et al. /Animal FeedScience and Technology 52 (1995) 159-I 73

171

contributing to the selective retention of fibre in the rumen ( Wattiaux et al., 1992 ) . As a result of selective retention, the passage rate of indigestible fibre from the rumen is higher than that of indigestible fibre (Egan and Doyle, 1985; Tamminga et al., 1989; present study). The relative difference in the passage rate of indigestible and digestible fibre was not related to the particle size, indicating a preferential retention of particles with high potential digestibility irrespective of their size (Huhtanen et al., 1993 ). The third factor explaining the low values for RNDFD, especially those based on in situ data and exponential decline in marker concentration, may be that a large proportion of post-ruminal fibre digestion indeed does occur. However, the values for RNDFD as a proportion of total NDF digestion obtained from our experiments using duodenally cannulated cattle and a double-marker system have consistently been close to 100% (Khalili and Huhtanen, 1991; Jaakkola et al., 199 1; Huhtanen and Jaakkola, I993 ). Also, the short post-duodenal retention time in the fermentation compartments (e.g. Poore et al., 1991; Huhtanen and Kukkonen, 1994) does not allow extensive cell wall digestion in the hind-gut, especially if the hind-gut fermentation is characterized by a lag phase similar to the rumen fermentation. In the present study, ruminal digestion of NDF as a proportion of total NDF digestion when calculated from digestion and passage kinetics was 0.95, similar to values obtained using duodenally cannulated cattle. Also, the results from experiments using the INDF ratio in duodenal and faecal particles (Huhtanen et al., 1993) and the mobile nylon bag technique (Vanhatalo et al., 1992) consistently indicate that only a small proportion of NDF digestion occurs in the caecum and colon. In the present study, the values for RNDFD and the proportion of RNDFD of total NDF digestibility calculated using the INDF ratio of rumen digesta and feed were 0.530 and 0.723. This approach assumes that the digesta present in the rumen and leaving the rumen have the same composition. This is not the case, since large particles have a slower passage rate than small particles and the passage rate of INDF is higher than that of DNDF. Therefore these values represent the minimum contribution of the rumen to NDF digestion. Because the values were higher than those based on in situ digestion kinetics and the exponential decline in marker concentration, it can be concluded that these methods used to estimate the constraints of NDF digestion will underestimate RNDFD. 5. Conclusions Using in situ digestion kinetics data and passage data calculated from exponential decline in marker excretion in a simple digestion model severely underestimated ruminal NDF digestibility. This was partly because of an underestimation of the rate of digestion by the in situ incubation and partly because the passage model ignores the retention in the rumen of non-escapable particles. Incorporation of a selective retention in the digestion model and using digesta passage kinetics estimated from duodenal marker concentrations and two-compartmental

172

P. Huhtanen et al. /Animal Feed Science and Technology 52 (1995) 159-173

models resulted in much higher RNDFD estimates. A combination of two-compartmental passage models with rumen evacuation derived rate of digestion values yielded RNDFD estimates which were 0.94-0.96 of the total NDF digestibility. This suggests that markers probably describe the passage of cell walls fairly accurately provided they are given in the same physical form as unlabelled feed. Using faecal rather than duodenal sampling will slightly increase the values for RNDFD but the error is much smaller than that caused by the underestimation of the rate of digestion by in situ incubation.

References Aitchison, E., Gill, M., France, J. and Dhanoa, MS., 1986. Comparison of methods to describe the kinetics of digestion and passage of tibre in sheep. J. Sci. Food Agric., 37: 1065-1072. Allen, M.S. and Mertens, D.R., 1988. Evaluating constraints of fiber digestion by rumen microbes. J. Nutr., 118: 261-270. Bee, U.-B., 1989. Utilization of grass silage by sheep: effects of supplementation with different amounts of ground barley. Norw. J. Agric. Sci., 3: 25 l-263. desBordes, C.K. and Welch, J.G., 1984. Influence of specific gravity on rumination and passage of indigestible particles. J. Anim. Sci., 59: 470-475. Dixon, R.M. and Mill&m, L.P., 1985. Removal of digesta components from the rumen of steers determined by sieving techniques and fluid, particulate and microbial markers. Br. J. Nutr., 53: 347-362. Egan, J.K. and Doyle, P.T., 1985. Effect of intraruminal infusion of urea on the response in voluntary intake by sheep. Aust. J. Agric. Res., 36: 483-495. Grant, R.J. and Weidner, S.J., 1992. Digestion kinetics of fiber: influence of in vitro buffer pH varied within physiological range. J. Dairy Sci., 75: 1060-1068. Hartnell, G.F. and Satter, L.D., 1979. Determination of rumen fill, retention time and ruminal turnover rates at different stages of lactation. J. Dairy Sci., 48: 38 l-392. Huhtanen, P. and Jaakkola, S., 1993. The effect of the forage preservation method and the proportion of concentrate on digestion of cell wall carbohydrates and rumen digesta pool size in cattle. Grass Forage Sci., 48: 155-165. Huhtanen, P. and Khalili, H., 199 1. Sucrose supplements in cattle given grass silage based diet. 3. Rumen pool size and digestion kinetics. Anim. Feed Sci. Technol., 33: 275-287. Huhtanen, P. and Khalili, H., 1992. The effect of sucrose supplements on particle-associated carboxymethylcellulase (EC 3.2.1.4) and xylanase (EC 3.2.1.8) activities in cattle given grass-silagebased diet. Br. J. Nutr., 67: 245-255. Huhtanen, P. and Kukkonen, U., 199 1. Evaluation of models to describe neutral detergent tibre digestion in cattle fed at two levels of dry matter intake. Anim. Prod., 52: 593, abstract. Huhtanen, P. and Kukkonen, U., 1994. Comparison of methods, markers, sampling sites and models in estimating digesta passage kinetics in cattle fed at two levels of intake. Anim. Feed Sci. Technol., in press. Huhtanen, P., Kukkonen, U., Jaakkola, S. and Kallio, M., 1993. The effect of particle size on passage rate of indigestible and digestible fiber. J. Dairy Sci., 76 (Suppl. 1): 407, abstract. Jaakkola, S., Huhtanen, P. and Hissa, K., 199 1. The effect of cell wall degrading enzymes or formic acid on fermentation quality and digestion of grass silage by cattle. Grass Forage Sci., 46: 75-87. Jones, A.L., Goetsch, A.L., Stokes, S.R. and Colberg, M., 1988. Intake and digestion in cattle fed warm and cool season grass hay with or without supplemental grain. J. Anim. Sci., 66: 194-203. Kennedy, P.M. and Murphy, M.R., 1988. The nutritional implications of differential passage of particles through the ruminant alimentary tract. Nutr. Res. Rev., 1: 189-208.

P. Huhtanen et al. /Animal Feed Science and Technology 52 (1995) 159-I 73

173

Khalili, H. and Huhtanen, P., 1991. Sucrose supplements in cattle given grass silage-based diet. 2. Digestion of cell wall carbohydrates. Anim. Feed Sci. Technol., 33: 263-273. Lippke, H., Ellis, W.C. and Jacobs, B.F., 1986. Recovery of indigestible fiber from feces of sheep and cattle on forage diets. J. Dairy Sci., 69: 403-4 12. Messman, M.A., Weiss, W.P. and Ericson, D.O., 1991. Effects of nitrogen fertilization and maturity of bromegrass on in situ ruminal digestion kinetics of fiber. J. Anim. Sci., 69: 115 l-l 16 1. Meyer, J. and Mackie, RI., 1986. Microbial evaluation of the intraruminal in sacculus technique. Appl. Environm. Microbial., 5 1: 622-629. Oldham, J., 1984. Protein-energy relationship in dairy cows. J. Dairy Sci., 67: 1090-l 114. (arskov, E.R. and McDonald, I., 1979. The estimation of protein degradability in the rumen from incubation measurements weighted according to the rate of passage. J. Agric. Sci., 92: 499-503. Pond, K.R., Ellis, W.C., Matis, J.H., Ferreiro, H.M. and Sutton, J.D., 1988. Compartmental models for estimating attributes of digesta flow in cattle. Br. J. Nutr., 60: 571-595. Poore, M.H., Moore, J.A. and Swingle, R.S., 1990. Differential passage rates and digestion of neutral detergent fiber from grain and forage in 30, 60 and 90% concentrate diets fed to steers. J. Anim. Sci., 68: 2965-2973. Poore, M.H., Moore, J.A., Eck, T.P. and Swingle, R.S., 1991. Influence of passage model, sampling site and dosing time on passage of rear earth-labelled grain through Holstein cows. J. Anim. Sci., 69: 2646-2654. Robinson, P.H., Sniffen, C.J. and van Soest, P.J., 1985. Influence of level of feed intake on digestion and bacterial yield in the forestomach of dairy cattle. Can. J. Anim. Sci., 65: 437-444. Robinson, P.H., Fadel, J.G. and Tamminga, S., 1986. Evaluation of mathematical models to describe neutral detergent residue in terms of its susceptibility to degradation in the rumen. Anim. Feed Sci. Technol., 15: 249-27 1. Robinson, P.H., Tamminga, S. and van Vuuren, A.M., 1987. Influence of declining level of feed intake and varying proportions of starch in the concentrate on rumen ingesta quantity, composition and kinetics ingesta turnover in dairy cows. Livest. Prod. Sci., 17: 37-62. Smith, L.W., Goering, H.K., Waldo, D.R. and Gordon, C.H., 1971. In vitro digestion rate of forage cell wall components. J. Dairy Sci., 54: 71-76. Staples,, CR., Fernando, R.L., Fahey, Jr., G.C., Berger, L.L. and Jaster, E.H., 1984. Effect of intake of a mixed diet by dairy steers on digestion events. J. Dairy Sci., 67: 995-1006. Tamminga, S., Robinson, P.H., Vogt, M. and Boer, H., 1989. Rumen ingesta kinetics of cell components in dairy cows. Anim. Feed Sci. Technol., 25: 89-98. Vanhatalo, A., Dakowski, P. and Huhtanen, P., 1992. Ruminal and intestinal degradation ofgrass cut at different stages of growth. Proceedings 14th General Meeting of the European Grassland Federation. Lahti, Finland. European Grassland Federation, pp. 4 12-4 15. Waldo, D.L., Smith, L.W. and Cox, E.L., 1972. Model of cellulose disappearance from the rumen. J. Dairy Sci., 55: 125-129. Wattiaux, M.A., Satter, L.D. and Mertens, D.R., 1992. Effect of microbial fermentation on functional specific gravity of small forage particles. J. Anim. Sci., 70: 1262-1270. Zinn, R.A. and Owens, F.N., 1986. A rapid procedure for purine measurement and its use for estimating net ruminal protein synthesis. Can. J. Anim. Sci., 66: 67-166.