HOME HELP FEEDBACK SUBSCRIPTIONS ARCHIVE SEARCH TABLE OF CONTENTS		QUICK SEARCH:   [advanced] Author: Keyword(s): Year:  Vol:  Page:

This Article Abstract Full Text (PDF) Interpretive Summary Alert me when this article is cited Alert me if a correction is posted Services Similar articles in this journal Similar articles in PubMed Alert me to new issues of the journal Download to citation manager Citing Articles Citing Articles via HighWire Citing Articles via Google Scholar Google Scholar Articles by Zebeli, Q. Articles by Drochner, W. Search for Related Content PubMed PubMed Citation Articles by Zebeli, Q. Articles by Drochner, W.

Effects of Physically Effective Fiber on Digestive Processes and Milk Fat Content in Early Lactating Dairy Cows Fed Total Mixed Rations

University of Hohenheim, Institute of Animal Nutrition (450), Emil-Wolff-Str. 10, D-70599 Stuttgart, Germany

¹ Corresponding author: zebeli{at}uni-hohenheim.de

	ABSTRACT

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Data from recent research studies were analyzed quantitatively, and the random effect of experiment was assessed to define the physiological responses of dairy cows in early lactation to intake of physically effective neutral detergent fiber (peNDF). All studies were conducted with lactating Holstein cows (84.8 ± 3.54 days in milk) in Latin square designs, and feeds were offered ad libitum as total mixed rations (TMR). The peNDF was estimated by 2 measurement techniques, the NDF content of TMR multiplied by amount of dry matter (DM) retained on a 1.18-mm screen (peNDF_{> 1.18}) and NDF content of TMR multiplied by the proportion of DM retained by 19- and 8-mm Penn State Particle Separator screens (peNDF_{> 8}). Other factors, including concentrations of NDF, forage NDF, non-fiber carbohydrates, the amount of digestible organic matter of forages (FDOM), and the intake of ruminally degradable starch (RDSI) from grain in the diet were also investigated. The studied animal response variables included feed intake, ruminal fermentation, chewing activity, fiber digestibility, and milk production and composition. The ruminal pH (day mean) in this study ranged from 5.30 to 6.59. Using peNDF_{> 1.18} approach, the requirements for physically effective fiber in high-yielding dairy cows fed TMR in an ad libitum intake were estimated to be about 19% of ration DM or 4.1 kg/d or 0.6 kg/100 kg of body weight to maintain a ruminal pH of about 6.0. When peNDF was measured as peNDF_{> 8}, ruminal pH responded in a quadratic fashion but the confidence of estimation was lower (R² = 0.27) compared with the peNDF_{> 1.18} approach (R² = 0.67). Results of these data analyses showed that peNDF_{> 1.18} provided a satisfactory estimation of the mean ruminal pH (R² = 0.67) and NDF digestibility (R² = 0.56). Furthermore, peNDF_{> 1.18} was poorly, although positively, correlated to daily chewing (R² = 0.17), and rumination (R² = 0.24) activity. On the other hand, results from these analyses showed that milk parameters are less sensitive to the effects of dietary peNDF than other variables, such as ruminal pH, chewing activity, and fiber digestibility. Dietary FDOM correlated positively (moderately) to ruminal pH (R² = 0.24), daily chewing (R² = 0.23), and rumination (R² = 0.29) activity, whereas the daily RDSI from grain correlated negatively to ruminal pH (R² = 0.55) and positively to total volatile fatty acids (R² = 0.27). Inclusion of FDOM and RDSI from grain along with peNDF_{> 1.18} in the models that predict rumen pH further improved the accuracy of prediction. This approach appeared to further complement the concept of peNDF that does not account for differences in ruminal fermentability of feeds.

Key Words: physically effective fiber • dairy cow • rumen pH • chewing activity

	INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
High-concentrate diets for high-yielding dairy cows must contain sufficient physically effective fiber (i.e., fiber that stimulates rumination, saliva production, and rumen buffering) to prevent ruminal dysfermentation and subacute ruminal acidosis (SARA). The concept of physically effective fiber was created to amalgamate the chemical characteristics and particle size of forages, and to quantify its value to rumen function (Mertens, 2000). According to Lammers et al. (1996), physically effective NDF (peNDF) could be measured as a proportion of DM retained by the 19- and 8-mm Penn State Particle Separator (PSPS) screens multiplied by dietary NDF content (peNDF_{> 8}). Mertens (1997) determined peNDF as the proportion of DM retained by a 1.18-mm screen multiplied by dietary NDF (peNDF_{> 1.18}) using a dry-sieving technique. It is, how-ever, unclear which measure of peNDF provides the most accurate estimate of chewing, saliva production, and rumen buffering (Einarson et al., 2004). Although a number of recent studies have been conducted to investigate the effects of peNDF on rumen fermentation, feed intake, milk production, chewing activity, and nutrient digestibility in high-yielding, early lactation dairy cows, the results obtained from these studies are not conclusive. Differences in measurement and definition of dietary peNDF and interactions between levels of concentrate inclusion, forage and grain sources, and animal-response variables among studies make peNDF recommendations difficult. The NRC (2001) does not give requirements for peNDF due to lack of a standardized, validated method for measuring effective fiber in feeds and to establish requirements for effective fiber.

Part of the difficulty in assigning fiber requirements for high-yielding dairy cows related to the interpretation of response variables. Despite the fact that milk fat percentage is an easily measured parameter, low fiber in the diet can detrimentally affect animal health without significant milk fat depression (Mertens, 1997). Ruminal pH may be a better indication of ruminal health and optimal function, and a better basis for determining fiber requirements of dairy cows in early lactation than the maintenance of milk fat production (Allen, 1997; Mertens, 1997).

However, Beauchemin and Yang (2005) concluded that the models used to predict rumen pH should include both peNDF and fermentable OM intake. Because the peNDF concept relates only to the physical properties of fiber, inclusion of forage and grain fermentability characteristics in the models to predict animal response and to evaluate physical effectiveness of dairy cow diets would likely increase the estimation accuracy. Results from several studies showed that increasing the amount of digestible fiber of hay or corn silage in dairy cow diets increased digesta stratification, particle breakdown in the rumen as well as digesta turnover, forage intake, and fiber digestibility without compromising the physical effectiveness at stimulating chewing (Oba and Allen, 2000b; Tafaj et al., 2004b, 2005b). Krause et al. (2002b) found that replacing dry cracked corn with high-moisture corn in a TMR fed to dairy cows significantly reduced ruminal pH. Furthermore, Beauchemin and Rode (1997) reported that rapidly digested starch sources such as barley grain increase the need for effective fiber, suggesting an interaction between ruminal fermentability and physical characteristics of the ration.

This quantitative study aimed to define the physiological responses of high-yielding dairy cows in early lactation to peNDF when estimated as peNDF_{> 1.18} and peNDF_{> 8}. Furthermore, based on the most sensitive animal response variable, an optimization of peNDF concentration in TMR fed ad libitum to this category of dairy cows was also intended. Possible interactions between dietary peNDF, NFC, the amount of digestible OM of forages (FDOM), and intake of ruminally degradable starch (RDSI) from grains composing TMR were also investigated.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Description of the Database
To conduct this quantitative study, an independent data set with data on animal characteristics, detailed ration components, and an evaluation of physical structure of the ration was generated. The data file containing 131 treatment means was generated from 33 experiments published from 1997 to 2005 (Table 1). Animals were lactating Holstein cows (84.8 ± 3.54 DIM) (mean ± SE) weighing between 570 and 886 kg and producing 23.1 to 49.3 kg of milk/d. Feeds were offered ad libitum as TMR. The level of DMI ranged from 16.9 to 28.3 kg/d, and the percentage of forage in TMR ranged from 27 to 75% of DM (49.2 ± 0.87%). Dietary NDF ranged from 18.2 to 48.1% of DM (30.9 ± 0.51%) and forage NDF (FNDF) ranged from 15.5 to 35.8% of DM (21.3 ± 0.36%). The peNDF_{> 1.18} content in TMR ranged from 4.2 to 37.4% of DM (21.1 ± 0.64%) and the content of peNDF_{> 8} ranged from 2.0 to 29.5% (15.2 ± 0.73%). The main characteristics of the database, including animal characteristics, investigated dietary factors, and animal response variables (hereafter referred to as response variables) are listed in Table 2. The response variables included data on feed intake, ruminal fermentation, chewing activity, fiber digestibility, and milk production and composition. All experiments included in this study were conducted in Latin square design.

Estimation of peNDF of Diets
As a prerequisite for inclusion in this literature study, articles were expected to give complete information on the components and chemical composition of rations, as well as on the physical evaluation of experimental diets (vertical dry-sieving technique). The peNDF_{> 8} content of TMR was determined by multiplying the proportion of DM retained by the 19- and 8-mm screens of PSPS by dietary NDF content (DM basis; Lammers et al., 1996).

Assuming that chewing activity is equal for all particles retained on a 1.18-mm sieve, Mertens (1997) proposed a system for estimating peNDF based on NDF concentration of feeds multiplied by the proportion of particles retained on a 1.18-mm sieve (peNDF_{> 1.18}) using the vertical oscillating dry-sieving technique. To estimate peNDF_{> 1.18} content of TMR in the present study, DM proportion retained on a 1.18-mm sieve, obtained either through the vertical oscillating sieving technique or the new version of PSPS (Kononoff et al., 2003b), was multiplied by NDF content (DM basis) of TMR.

Some studies reported only the particle size distribution of their forages but not of the TMR. Grain is also reported to possess effective fiber according to its physical form (ground, pelleted, or rolled; Mertens, 1997) and degradability characteristics (De Brabander et al., 2002). To account for peNDF_{> 1.18} content of concentrates in the present study, the data were used from Mertens (1997), who reported particle size distribution (proportion of particles retained on sieve 1.18 mm) for different concentrate feeds using the vertical oscillating sieving technique. In this case, total peNDF_{> 1.18} content of TMR was calculated as a sum of peNDF_{> 1.18} obtained from forages (NDF content of forages x DM retained on screens > 1.18 mm) with the estimate of concentrate peNDF_{> 1.18} (NDF content of grain x DM retained on screens > 1.18 mm, taken from tables of Mertens, 1997) according to their proportion in the TMR.

Estimation of FDOM
Nutritive value tables and chemical composition of feedstuffs from the United Kingdom (MAFF, 1990) were used to estimate the amount of FDOM for the present experimental diets. The content of FDOM was used to take into account in the analysis forage quality and the amount of fermentable OM of forages composing TMR. The United Kingdom tables express FDOM as in vitro digestible OM determined using rumen fluid-pepsin; for forages constituting diets of this study was set as follows (in g/kg of DM): corn silage (667), alfalfa hay (557), alfalfa silage (547), barley silage (684), mixed grass silage (587), oat silage (684), grass hay (sun-cured) (564). Furthermore, the total amount of dietary FDOM content was calculated based on the proportion of forages in the experimental TMR.

Estimation of Ruminally Degradable Starch of Grain
All published studies analyzed provided sufficient information about the components of concentrate mixture of their TMR. To account for differences of ruminal degradability characteristics among grains comprising experimental TMR in the present study, the in situ effective ruminal degradability of starch from grain was taken into consideration. Starch content of grain in TMR was either taken directly from articles or was taken from tables compiled by Sauvant et al. (2004). The amount of ruminally degradable starch (RDS; % of total starch) of grain mixture composing TMR was calculated according to the formula: RDS = pi x ERDi, where pi represents the proportion of dietary starch provided from grain i in the mixture, and ERD_i represents starch effective degradability for grain i, which was taken from in situ calculations made from Offner et al. (2003) for a fractional passage rate of 6%/h. To take into account the RDSI from grain in the analysis, RDSI was calculated by multiplying the daily DM intake with RDS of grain comprising TMR. A statistical description of the calculated in situ RDS and RDSI from grain of this study, including mean, range, and median, is given in Table 2.

Statistical Analyses
Data analysis was performed according to St-Pierre (2001) taking into account the random effect of the experiment, using PROC MIXED (version 8.2, SAS Institute, 2001). The variable experiment (hereafter referred as study) does not contain quantitative information and constitutes an additional variation source; it was therefore declared in the CLASS statement. Simple linear regression was performed to test the response of animal variables to dietary factors (refer to Table 2 for information regarding response variables and dietary factors). Therefore, dependent variables in the regression analysis included all response variables and the independent continuous variables included all dietary factors according to the following model:

where Yij = the expected outcome for the dependent variable Y (response variable) observed at level j of the continuous variable X (dietary factor) in the study i, a₀ = the overall intercept across all studies (fixed effect), s_i = the random effect of the study i (i = 1,..., 33), ß ₁ = the overall regressing coefficient of Y on X across all studies (fixed effect), X_ij = the value j of continuous variable X in study i, bi = the random effect of study i on the regression coefficient of Y on X in study i, and eij = the unexplained error.

To take into consideration the unequal variance among studies, all dependent variables were weighted by the reciprocal of their squared standard error. In addition, an unstructured variance-covariance matrix (type = un) was performed at the random part of the model, as suggested by St-Pierre (2001) to avoid the positive correlation between the intercepts and slopes. When dietary factors were P < 0.05, their squared term was included in the model to test any likely quadratic relationship (second-order polynomial regression) and a variance components (type = vc) of variance-covariance structure was performed to avoid the positive correlation between the intercepts and slopes (St-Pierre, 2001). To fit the asymptotic relationship of ruminal pH or chewing index to dietary peNDF_{> 1.18}, a mathematical asymptotic function using PROC NLIN (DUD method; SAS Inst. Inc., version 8.2) according to the following model was used:

where Y represents the response variable; a, b, and c are the estimates, and X represents the dietary peNDF_{> 1.18}.

All significant dietary factors (P < 0.05) were further tested using the backward elimination multiple regression similarly to the algorithm reported by Oldick et al. (1999) and Firkins et al. (2001). To limit overparameterization of the model, a variance inflation factor less than 10 for every continuous independent variable tested was assumed, as suggested by Oldick et al. (1999). The best fit was chosen as the one with the lowest root mean square error (RMSE), higher determination coefficient (R²), and the highest Schwarz’s Bayesian criterion. For simplicity, only the best-fit equations of multiple regression (backward elimination) that further improved the relationship obtained from linear or polynomial regression are shown.

	RESULTS AND DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
Intakes of DM and peNDF
A summary of intakes for diets used in this quantitative study is given in Table 2. Intake of DM from 33 studies averaged 23.0 ± 0.17 kg/d or 3.5 ± 0.03 kg/100 kg of BW per d. Intakes of peNDF_{> 1.18} from 27 studies and peNDF_{> 8} from 15 studies averaged 4.9 ± 0.16 kg/d and 3.3 ± 0.19 kg/d, respectively. Table 3 shows the results of linear regression of DMI response to different dietary factors (for simplicity, only significant relationships are shown, P < 0.05). The analysis of backward elimination of multiple regression did not further improve the relationship obtained from linear regression, and therefore, the results are not shown. In general, relationships of dietary factors to DMI were low in this study. The dietary NDF negatively affected DMI (expressed as either kg/d or kg of DM/100 kg of BW per d), although the relationship was moderately higher when peNDF was measured as peNDF_{> 1.18} (R² = 0.21) and when BW of animals was included in the analysis (R² = 0.29). Dry matter intake also responded positively to NFC:NDF ratio of the diet. These responses are presumably connected with the ruminal physical fill effect of NDF and the high energetic necessity of dairy cows in early lactation. Cows in early lactation are often in negative energy balance because the ingested energy does not meet the energy requirements of lactation (NRC, 2001). The extent to which DMI of lactating dairy cows is regulated by distension in the reticulorumen depends upon the animal’s energy requirement and the filling effect of the diet offered (Allen, 2000). However, Allen (2000) stated that physical fill could limit feed intake at the low concentrate inclusion, but a metabolic rather than a physical constraint can be expected to be rate limiting at high concentrate inclusion. Beauchemin et al. (1994) reported an interaction (P < 0.01) between forage particle length (alfalfa silage chopped at 5- and 10-mm theoretical length of cut) and percentage of forage in the diet (35 or 65%). In that experiment, DMI was reduced by nearly 3 kg/d when forage percentage was increased from 35 to 65 with diets containing the long-chopped alfalfa silage, but was reduced by less than 0.5 kg/d with diets containing short-chopped silage. Tafaj et al. (2001) reported that reducing dietary hay particle size from 28.7 to 9.2 mm increased DMI by 13% only at a low-concentrate level (13% in DM), but when a high-concentrate diet (~40% in DM) was fed no differences were observed in sheep. Firkins et al. (2001) quantitatively analyzed results from literature in dairy cows with a comparable DMI to the present study (21.3 ± 2.6 kg/d) and reported a negative linear relationship among dietary NDF and FNDF and DMI. Recently, Beauchemin and Yang (2005) reported that reducing dietary peNDF_{> 8} from 11.5 to 8.9% ration DM was not a rate-limiting step for particulate passage and hence, for rumen physical fill and feed intake in Holstein cows fed corn silage-based TMR with a high concentrate level (~60% of ration DM).

View larger version (19K): [in this window] [in a new window]   Figure 1. Relationship of physically effective NDF (peNDF> 1.18) to observed ruminal pH value adjusted for the random study effect and unequal variance. The best-fit, quadratic regression pH = 5.06 ( ± 0.09) + 0.069 ( ± 0.008) peNDF> 1.18 –0.001 ( ± 0.0002) peNDF2> 1.18, root mean square error = 0.137, R2 = 0.67; Asymptotic function () pH = 6.34 – 1.39e(–0.08 peNDF> 1.18) using Proc NLIN (SAS Institute, 2001).

According to Figure 1, peNDF_{> 1.18} estimates ruminal pH with R² = 0.67 and an RMSE of 0.137 pH units. Increasing the dietary peNDF increased the ruminal pH quadratically. The same relationship was found when peNDF_{> 1.18} was expressed as intake (kg/d) or in-take per kg of BW (kg/100 kg of BW). It means that with increasing dietary peNDF_{> 1.18}, ruminal pH does not increase indefinitely, but rather attains an asymptotic plateau in response to dietary peNDF_{> 1.18}. Allen (1997) stated that roughages have a critical particle length and any increase thereafter does not further improve their physical effectiveness. However, in this study, the relationship between ruminal pH and dietary peNDF_{> 1.18} appeared to follow a quasi-linear course up to a pH value of approximately 6.0. Subsequently, the mathematical asymptotic function revealed an asymptotic relationship of ruminal pH with dietary peNDF_{> 1.18}, showing an approximated plateau at a ruminal pH of 6.2 and in response to about 30% dietary peNDF_{> 1.18} (Figure 1). Pitt et al. (1996) calculated ruminal pH from the equilibrium between ruminal acidity and buffering. In fact, they postulated that the crossover point representing acid-base equilibrium and the predicted steady state ruminal pH resulted at a ruminal pH of about 6.1. Mertens (1997) also reported a quadratic relationship between dietary peNDF_{> 1.18} and ruminal pH (R² = 0.71). Pitt et al. (1996) used data from sheep, beef cattle, and dairy cattle and observed a quadratic relationship between effective NDF and ruminal pH (R² = 0.52). In the study of Pitt et al. (1996), the sheep consumed a high-forage diet, which probably accounted for a higher plateau of ruminal pH in that study (~6.4).

Furthermore, our analysis also shows that dietary NDF, FNDF, and FDOM positively influenced ruminal pH, whereas both the NFC:NDF ratio and RDSI from grain indicated a linear negative effect (Table 4). This result is partly in accordance with Allen (1997), who found a positive correlation only between FNDF and ruminal pH (R² = 0.63) and not between total dietary NDF and pH. Firkins et al. (2001) also reported that ruminal pH in dairy cows correlated positively to FNDF and negatively to nonstructural carbohydrates in the ration. The estimated RDSI from grain in this study represents both the amount of ruminally degradable starch from grain and DMI. Firkins et al. (2001) found that an increase in DMI would still result in an increase in ruminally degradable starch, despite a reduction in the percentage of ruminally degraded starch caused by the increase of passage rate. In this context, Stone (2004) stated that high-producing cows might have an increased risk of SARA simply due to higher DMI. As expected in the present study, RDSI from grain correlated negatively to daily ruminal pH, explaining 55% of its variation (Table 4). This study shows that the intake of rapidly fermentable carbohydrates is more important than total NFC percentage of the diet to be taken into account in terms of avoiding SARA in high-yielding dairy cows.

Using the inverse regression, Mertens (1997) reported that a peNDF intake of 4.4 kg/d or a concentration of 22.3% of ration DM is needed to maintain a mean pH of 6.0. According to the results of our analysis, an intake of peNDF_{> 1.18} of either 4.1 kg/d, or of 0.58 kg/100 kg of BW, or a concentration of ~19% of ration DM is needed to maintain a pH of 6.0 (Table 5).

The results of analysis of multiple regression using the backward elimination procedure, where all significant dietary factors were included, are shown in Table 6. To avoid overparameterization of the model, FDOM and RDSI were separately analyzed. This analysis showed that dietary peNDF_{> 1.18} (quadratically) and FDOM (linearly) significantly affected (P < 0.05) ruminal pH. The same effect on ruminal pH was observed when RDSI along with peNDF_{> 1.18} was tested in the multiple backward elimination regression, only that RDSI negatively correlated to ruminal pH. In fact, the concept of peNDF does not account for differences in fermentability of feeds. The FDOM (expressed as g/kg of DM) is an indicator particularly of the quality (ruminal degradability) of forage in TMR, whereas RDSI (kg/d) accounts for daily intake of ruminally degradable starch from grain. The positive relationship of FDOM to pH (R² = 0.24) showed that FDOM affects the rumen conditions and should be considered when formulating TMR for dairy cows. Similarly, Allen (1997) reported a linear positive relationship (R² = 0.18) between ruminal pH and the amount of OM truly degraded in the rumen. Inclusion of FDOM along with peNDF_{> 1.18} in the model slightly increased the accuracy of estimation of ruminal pH from R² = 0.67 to R² = 0.72. The positive effect of FDOM on ruminal pH was probably a result of the positive effect of good-quality forage in promoting digesta stratification and turnover in the rumen and consequently stimulating forage intake and chewing activity (Tafaj et al., 2004b, 2005b). On the other hand, inclusion of RDSI in the model with peNDF_{> 1.18} increased the accuracy of estimation of ruminal pH up to 75%. Krause et al. (2002b) found that diets containing finer haylage particles and high-moisture corn reduced mean ruminal pH more than diets containing coarser haylage and dry ground corn. Furthermore, Beauchemin and Rode (1997) reported that rapidly digested starch sources, such as barley grain, increase the need for effective fiber. The results of the present study indicate that ruminal pH is influenced both by dietary components affecting chewing and salivary secretion, and by those affecting ruminal carbohydrate fermentation. Nevertheless, even though FDOM and RDSI serve as an indication of forage quality and of the intake in ruminally degradable starch, respectively, we are aware that in vitro- and in situ-estimated FDOM and RDSI cannot completely represent the in vivo ruminal degradation of OM of the experimental TMR used in this study. More research with known in vivo FDOM and RDSI is needed to validate this assumption.

However, results of this study indicate that accounting for dietary physically effective fiber is a more efficient procedure to assess effective fiber adequacy of dairy cow ration than simply taking into account dietary NDF or FNDF. In this context, the PSPS constitutes a useful on-farm choice for frequent on-site examination of ration particle size and ration physical effectiveness, thus increasing the efficacy for controlling SARA on dairy farms. But, as this study demonstrates, not only physically effective fiber, but also the amount of digestible OM of forages and grain fermentability affects ruminal pH, which are not accounted for when using PSPS to calculate peNDF.

Effects of dietary factors on ruminal pH were not completely consistent with the effects on ruminal VFA content or molar percentage of acetate, propionate, or butyrate, which were not affected to the same extent or were not affected at all compared with ruminal pH (Tables 4 and 6). The effects of dietary particle size on the production and absorption of VFA are often contradictory. Reducing dietary particle size might increase the ruminal degradation rate of fiber, but this does not inevitably lead to an increase of rumen OM degradability or the production of VFA, as particulate passage rate can also be increased (Soita et al., 2003). However, Allen (2000) stated that decreasing forage particle size might increase DMI, which could also increase the availability of ruminally digested OM and hence the production of VFA.

Total VFA was negatively affected by peNDF, whether expressed as peNDF_{> 8} (R² = 0.25) or peNDF_{> 1.18}, (R² = 0.12). This decrease of VFA in the rumen fluid with increasing dietary peNDF might be attributed to the positive effect of peNDF on rumen motility. Taylor and Allen (2005) stated that as ruminal motility increases, VFA molecules are expected to be replenished at the rumen wall more rapidly, increasing the concentration gradient across the ruminal epithelium and rate of VFA absorption. Increasing the dietary NFC:NDF ratio, NFC content, or RDSI from grain increased the ruminal VFA linearly, whereas acetate to propionate ratio and butyrate percentage decreased (Tables 4 and 6). This is in accordance with the results found by Firkins et al. (2001), who reported that acetate to propionate ratio correlated positively to FNDF and negatively to nonstructural carbohydrates in the diet of dairy cows.

Chewing Activity
Total chewing time ranged from 425 to 969 min/d (691 ± 11.2 min/d; mean ± SE), and rumination time ranged between 151 and 632 min/d (434 ± 8.29 min/d; mean ± SE). The mean value of rumination found here is consistent with that found by Beauchemin et al. (1994), who reported that high-producing dairy cows consuming large quantities of DM tended to ruminate more than 360 min/d unless digestive upset occurs. This was equivalent to a ruminative minimum of 16 min/kg of DM for 22 kg/d of DMI. About 14% of 99 treatment means used for rumination in this study did not achieve a rumination time of 360 min/d; and 25% of the reports gave values lower than 16 min/kg of DM (data not shown). Although the mean DMI in this study was about 1.03 kg/d higher than in that of Beauchemin et al. (1994), if we consider these criteria, it could be stated that these results do not represent a sufficient rumination activity. Table 2 gives an overview of the chewing indices. The range of time spent chewing and ruminating was high in this study; cows spent 17.9 to 47.1 min of chewing per kg of DM (30.1 ± 0.59 min/kg of DM) and 54.3 to 160 min/kg of NDF (103 ± 2.48 min/kg of NDF), whereas chewing time spent per kilogram of peNDF_{> 1.18} was 167 ± 9.67 min. Sudweeks et al. (1981) proposed that chewing corrected for DMI as a criterion for physical effectiveness of forages. They further proposed values equal to or greater than 30 min/kg of DM as suitable for limiting the risk of digestive disorders. Based on the recommendation of Sudweeks et al. (1981), 51% of 99 treatment means analyzed here were below this criterion, ranging from 17.9 to 29.9 min/kg of DM (data not shown). De Brabander et al. (2002) suggested that dairy cows should achieve between 59 and 72.8 min of chewing time per kg of DM from forages to prevent ruminal disorders and milk fat depression. Only 2 treatment means out of 99 were below this criterion in this study (result not shown). Tafaj et al. (2005c) estimated that for dairy cows to achieve a chewing time of 74 min/kg of DM from a long-chopped hay, diets should contain at least 10.7% long-chopped hay-crude fiber, which in turn corresponded to 28% NDF or 19% peNDF and 60% slowly degradable concentrate in the diet. Furthermore, Tafaj et al. (2005c) reported that rations should contain at least 10.0% long-chopped hay-crude fiber to maintain 3.4% milk fat, which is close to 10.7% long-chopped hay-crude fiber needed to achieve a chewing time of 74 min/kg DM from long-chopped hay.

In Table 7, the significant linear relationships between dietary factors and chewing parameters are given, and in Table 8, the results of the analysis of multiple regression using backward elimination procedure are summarized. Total chewing and rumination time are positively affected by peNDF, FNDF, and FDOM contents of the ration, albeit to a lower extent than on ruminal pH. Total chewing activity was positively affected by dietary NFC, presumably because of the positive effect of dietary NFC on DMI. These results agree with those of Firkins et al. (2001), who reported that increasing dietary FNDF and nonstructural carbohydrate concentrations linearly increased chewing activity (min/kg of NDF). Similarly, Beauchemin and Yang (2005) reported a linearly increased chewing time from 702 to 783 min/d, and rumination time from 441 to 494 min/d with increasing peNDF_{> 8} in the diet from 8.9 to 11.5%. Based on this finding, Beauchemin and Yang (2005) postulated that a level of peNDF_{> 8} above 10% in the dairy cow diet is required to avoid reduction of chewing activity.

View larger version (16K): [in this window] [in a new window]   Figure 2. Relationship of physically effective NDF (peNDF> 1.18) to observed chewing index adjusted for the random experiment effect and unequal variance. The best-fit, quadratic regression chewing time = 628.8 ( ± 13.8) –38.8 ( ± 1.35) peNDF> 1.18 + 0.73 ( ± 0.03) peNDF2> 1.18; root mean square error = 24.82; R2 = 0.91; Asymptotic function () chewing time = 80.4 + 664e(–0.94 peNDF> 1.18) using Proc NLIN (SAS Institute, 2001). To achieve 150 min/kg of peNDF> 1.18 the ration should contain about 20% peNDF> 1.18.

Milk Production and Composition
In this study, cows produced between 23.1 and 49.3 kg of milk/d and were between 9 and 170 d in milk (84.8 ± 3.54 DIM; Table 2). All articles included in this study gave information about milk production and its composition. This is probably because milk parameters are easy to record, but as observed in this study, they did not satisfactorily respond to dietary factors considered.

Table 11 shows the results of linear regression of response of milk parameters to different dietary factors (only significant relationships are shown). The analysis of backward elimination of multiple regression did not further improve the relationship obtained from linear regression, and therefore no results are shown here. Milk yield was negatively affected by dietary NDF and FNDF and positively by NFC:NDF ratio, presumably because of their effect on DMI and energy concentration of the diet. Firkins et al. (2001) reported a negative correlation between milk yield and FNDF. Milk protein was not significantly affected by dietary factors studied here. However, the milk fat:protein ratio increased linearly with increasing peNDF_{> 1.18}, FNDF, and FDOM in the diet. Milk fat percentage increased linearly with increasing dietary FNDF and FDOM. Firkins et al. (2001) reported that milk fat content responded in a quadratic fashion to dietary FNDF.

Failure to observe effects of dietary peNDF on milk yield and fat percentage contrasts with the effect of increasing dietary peNDF_{> 1.18} on increasing ruminal pH and fiber digestibility. It is apparent that milk parameters are less sensitive to the effects of dietary peNDF than are other variables, such as ruminal pH, chewing activity, and fiber digestibility. It is well recognized, however, that cows are in negative energy balance in early lactation; thus, animals must mobilize fat (NRC, 2001) and consequently, milk fat content artificially increases. On the other hand, milk fat is a variable that is closely related to dietary fat and the genetic merit of the animal. Mertens (1997) concluded that ruminal pH may be a better indication of ruminal health and optimal function than the maintenance of milk fat production in this stage of lactation for dairy cows.

Systems for Estimation of Physically Effective Fiber and Response Variables
For physical evaluation of forages or TMR, different wet and dry sieving techniques have been proposed. Murphy and Zhu (1997) compared 9 methods (3 dry and 6 wet) for evaluating particle size distribution of feedstuffs. Based on median particle size estimates (central tendency), they concluded that 6 of those (including all dry sieving techniques tested) had relatively consistent estimates of particle size distribution. Lammers et al. (1996) developed the simplified method for evaluating particle size distribution of forages and TMR using PSPS. The PSPS is based on the properties of Standard S424 of the American Society of Agricultural Engineers (ASAE, 1998) and has been proven to generate similar results to the vertical oscillating sieving method (Lammers et al., 1996; Buckmaster, 2000; Teimouri Yansari et al., 2004). The PSPS device contains 2 sieves and a bottom pan. Using the PSPS, particle distribution is determined by separating particles according to size; > 19 mm, between 19 and 8 mm, and < 8 mm (Lammers et al., 1996).

The new version of PSPS (Kononoff et al., 2003b) is constructed from 3 sieves with pores measuring 19, 8, and 1.18 mm and a solid bottom pan, permitting the estimation of peNDF_{> 1.18} according to Mertens (1997). Teimouri Yansari et al. (2004) estimated peNDF_{> 1.18} of forages (alfalfa and corn silage) and TMR using both a vertical oscillating sieve and the new PSPS and reported very similar results. The PSPS is a quick, cost-effective method, and produces consistent results in measuring particle size distribution of forages and TMR. These properties made the PSPS method a valuable on-farm choice to estimate forage and TMR particle size distribution around the world.

As shown in Table 2, the amount of physically effective NDF in TMR was higher when estimated as peNDF_{> 1.18} than when estimated as peNDF_{> 8}. This is because peNDF_{> 1.18} contains a large pool of particles retained on the lower screens (i.e., particles greater than 1.18 mm and smaller than 8 mm). Research results (Kononoff and Heinrichs, 2003a; Kononoff et al., 2003a; Plaizier, 2004) showed that the proportion of this pool could range from 30 to 50% in the TMR. Beauchemin et al. (2003) reported a 50% higher peNDF of TMR when it was estimated as peNDF_{> 1.18} compared with peNDF_{> 8}, because peNDF_{> 8} does not consider particles < 8 mm or steam-rolled concentrate, whereas peNDF_{> 1.18}, does.

Results of this study showed that differences in the quantity of peNDF_{> 8} and peNDF_{> 1.18} were reflected by their effects on the response variables studied. Thus, regarding the chewing activity, the peNDF_{> 8} affected rumination time to a slightly higher extent than did peNDF_{> 1.18} (R² = 0.27 vs. R² = 0.24), whereas total chewing time was more accurately estimated when peNDF was expressed as peNDF_{> 1.18} (R² = 0.17 vs. R² = 0.13). On the other hand, the accuracy of estimation for ruminal pH and NDF digestibility was higher when peNDF was expressed as peNDF_{> 1.18}. Although the content of peNDF_{> 8} is lower than peNDF_{> 1.18} in the TMR, the effectiveness of peNDF_{> 8} to stimulate rumination is higher. Buckmaster (2000) proposed an effective fiber index that weights NDF content by particle size. To calculate the effective fiber index of a ration, NDF content of each fraction of particles retained on each sieve is multiplied by a relative effectiveness coefficient, which is different for every particle fraction retained on the screens of PSPS. Based on these relative effectiveness coefficients, Buckmaster (2000) stated that particles > 19 mm are twice as effective for stimulating rumination and contributing to a rumen mat as those between 8 and 19 mm, whereas particles < 8 mm have one-fifth of the effectiveness of particles between 8 and 19 mm. This contrasts with the peNDF_{> 1.18} method of Mertens (1997), which equally weights particle mass > 1.18 mm in stimulating chewing activity and neglects the rest. Krause et al. (2002b) corrected the particle size distribution of TMR by that of orts and found a good relationship only between chewing and rumination activity with DMI from particles retained on the 19-mm screen of PSPS (r = 0.61 each), but not with DMI from particles retained on the 8-mm screen or the pan. Furthermore, Krause et al. (2002b) found no relationship between ruminal pH and DMI from particles retained on the 19-and 8-mm screens of PSPS.

Results of ruminal pH and NDF digestibility in the present study did not support dietary peNDF measured as the proportion of particles retained on 8 mm sieve as the best indicator to express the physical effectiveness of TMR, even though peNDF_{> 8} slightly better estimated rumination time. Indeed, the relationship between chewing and rumination time and ruminal pH was low in this study (r = 0.25 and r = 0.33, respectively; results not shown). Cassida and Stokes (1986) estimated saliva flow rates of 150, 177, and 300 mL/min during resting, eating, and ruminating, respectively. Using these estimated flow rates, the contribution in saliva production in the present study would be 112, 46, and 130 L/d for resting, eating, and ruminating time, respectively (data calculated based on chewing times given in Table 2). It could be stated that decreased saliva flow associated with a decreased rumination time could be partly balanced by saliva secretion during resting time. Krause et al. (2002b) reported that salivary buffering is only one of many factors determining ruminal pH.

Discrepancies between peNDF_{> 8} and peNDF_{> 1.18} to estimate rumination time, ruminal pH, and fiber digestibility observed in this study indicate that these concepts cannot be used interchangeably. Although the peNDF_{> 8} concept represents particles retained on the 8-mm sieve, which are expected to provoke an intensive saliva output, peNDF_{> 8} was not the best parameter to estimate ruminal pH and fiber digestibility.

It is believed that diets with less than 7% long particles (particles retained on the top screen of the separator) put cows at increased risk of SARA, particularly if these diets are also borderline or low in chemical fiber content (Grant et al., 1990a). However, increasing chemical fiber content may compensate for short particle length (Beauchemin et al., 1994). On the other hand, diets having excessive long forage particles can paradoxically increase the risk of SARA, especially when long particles are unpalatable and sortable (K. M. Krause and G. R. Oetzel, unpublished data). Several researchers (Calberry et al., 2003; Leonardi and Armentano, 2003; Beauchemin and Yang, 2005) reported different particle size distributions for TMR and orts for cows fed TMR ad libitum, indicating selective consumption. By feeding Holstein cows a corn silage-based TMR, we found (Junck et al., 2004) that across treatments (4 different theoretical particle lengths, 5.5, 8.1, 11, and 14 mm), difference between the offered TMR and orts were 26% for particles retained on the 8-mm screen (i.e., sum of particles retained on 19- and 8-mm sieves of PSPS). This difference was only 9% for particles retained on the 1.18-mm screen (i.e., sum of particles retained on 19-, 8-, and 1.18-mm sieves of PSPS) (in both cases were more long particles in orts than in the offered TMR). This indicates that if we evaluate our TMR only based on particles retained on the 8-mm screen (i.e., peNDF_{> 8}), our estimation to evaluate effective fiber content of the truly consumed TMR was biased 26% compared with only 9% bias if we estimated the peNDF of TMR based on particles retained on the 1.18- mm screen (i.e., peNDF_{> 1.18}). This was because cows sorted against coarse particles of TMR and preferred to consume short particles and crushed corn kernels, which are more palatable and digestible. However, this sorting against the coarse particles was more evident for TMR containing the longest levels of corn silage (11 and 14 mm) compared with TMR containing the shortest levels (8.1 and 5.5 mm), which showed a lower sorting rate. It seems, therefore, that an advantage of estimating peNDF of the ration based on particles retained on 1.18-mm screen consists in this, that this system, better reflects the TMR really consumed by dairy cows and can reduce the estimation bias related to sorting consumption. Under practical conditions on dairy farms, the evaluation of particle size distribution is mainly carried out in TMR and not in orts. For this reason, the evaluation of the ration based only on the pool of long particles of the original TMR does not appear to be an adequate approach for estimation of physical effectiveness in dairy cows fed TMR ad libitum.

	CONCLUSIONS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 
This study showed that the concentration and intake of peNDF_{> 1.18} should be considered in formulation of TMR for high-yielding dairy cows. Results of this data analysis showed that peNDF_{> 1.18} provided a satisfactory estimation of mean ruminal pH and fiber digestibility. Indeed, the peNDF_{> 1.18} approach explained 67% of the variation in ruminal pH. On the other hand, peNDF_{> 1.18}, was poorly, although positively, correlated to daily chewing (R² = 0.17) and rumination (R² = 0.24) activity. Results from this analysis showed that milk parameters are less sensitive to the effects of dietary peNDF than are other variables, such as ruminal pH, fiber digestibility, and chewing activity.

Using the peNDF_{> 1.18} approach, the requirements for physically effective fiber in high-yielding dairy cows fed TMR in an ad libitum intake were estimated to be about 19% of ration DM (4.1 kg/d or 0.6 kg/100 kg of BW) to maintain a ruminal pH of about 6.0. Inclusion of FDOM and RDSI from grain in the model together with peNDF_{> 1.18} appeared to be advantageous in improving the accuracy of estimation. This approach appeared to complement the concept of peNDF that does not account for differences in ruminal fermentability of feeds. Moreover, this study showed that the intake of rapidly fermentable carbohydrates is more important than total NFC percentage of the diet to be taken into account in terms of avoiding SARA in high-yielding dairy cows.

When peNDF was measured as the proportion of particles retained on an 8-mm PSPS screen (peNDF_{> 8}), ruminal pH responded in a quadratic fashion but the confidence of estimation was lower (R² = 0.27) compared with peNDF_{> 1.18}. Furthermore, measuring physically effective NDF as peNDF_{> 1.18} might be more realistic in terms of expression of effective fiber in the TMR truly consumed by dairy cows.

Results of this study indicated that accounting for dietary physically effective fiber is a more efficient procedure to assess effective fiber adequacy of dairy cow rations than simply taking into account dietary NDF or FNDF. In this context, the PSPS constitutes a useful on-farm choice for frequent on-site examination of ration particle size and ration physical effectiveness.

Received for publication June 22, 2005. Accepted for publication September 29, 2005.

	REFERENCES

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS AND DISCUSSION CONCLUSIONS REFERENCES

 


Allen, M. S. 1995. Fiber requirements: Finding an optimum can be confusing. Feedstuffs 67:13–16.

Allen, M. S. 1997. Relationship between fermentation acid production in the rumen and the requirement for physically effective fiber. J. Dairy Sci. 80:1447–1462.[Abstract]

Allen, M. S. 2000. Effect of diet on short-term regulation of feed intake by lactating dairy cattle. J. Dairy Sci. 83:1598–1624.[Abstract]

Allen, D. M., and R. J. Grant. 2000. Interactions between forage and wet corn gluten feed as sources of fiber in diets for lactating dairy cows. J. Dairy Sci. 83:322–331.[Abstract]

ASAE. 1998. S424: Method of determining and expressing particle size of chopped forage materials by sieving. ASAE Standards. Am. Soc. Agric. Eng., St. Joseph, MI.

Bal, M. A., R. D. Shaver, A. G. Jirovec, K. J. Shinners, and J. G. Coors. 2000. Crop processing and chop length of corn silage: Effects on intake, digestion, and milk production by dairy cows. J. Dairy Sci. 83:1264–1273.[Abstract]

Beauchemin, K. A. 2000. Managing rumen fermentation in barley based diets: Balance between high production and acidosis. Adv. Dairy Technol. 12:109–125.

Beauchemin, K. A., B. I. Farr, L. M. Rode, and G. B. Schaalje. 1994. Effects of alfalfa chop length and supplementary long hay on chewing and milk production of dairy cows. J. Dairy Sci. 77:1326–1339.[Abstract]

Beauchemin, K. A., and L. M. Rode. 1997. Minimum versus optimum concentrations of fiber in dairy cow diets based on barley silage and concentrates of barley or corn. J. Dairy Sci. 80:1629–1639.[Abstract]

Beauchemin, K. A., L. M. Rode, and M. V. Eliason. 1997. Chewing activities and milk production of dairy cows fed alfalfa as hay, silage, or dried cubes of hay or silage. J. Dairy Sci. 80:324–333.[Abstract]

Beauchemin, K. A., and W. Z. Yang. 2005. Effects of physically effective fiber on intake, chewing activity, and ruminal acidosis for dairy cows fed diets based on corn silage. J. Dairy Sci. 88:2117–2129.[Abstract/Free Full Text]

Beauchemin, K. A., W. Z. Yang, and L. M. Rode. 2003. Effects of particle size of alfalfa based-dairy cow diets on chewing activity, ruminal fermentation, and milk production. J. Dairy Sci. 86:630–643.[Abstract/Free Full Text]

Boddugari, K., R. J. Grant, R. Stock, and M. Lewis. 2001. Maximal replacement of forage and concentrate with a new wet corn milling product for lactating dairy cows. J. Dairy Sci. 84:873–884.[Abstract]

Buckmaster, D. M. 2000. Particle size in dairy cows. Pages 109–128 in Recent advances in animal nutrition. P. C. Garnsworthy and J. Wiseman, ed. Nottingham University Press, Nottingham, UK.

Calberry, J. M., J. C. Plaizier, M. S. Einarson, and B. W. McBridge. 2003. Effects of replacing chopped alfalfa hay with alfalfa silage in a total mixed ration on production and rumen conditions of lactating dairy cows. J. Dairy Sci. 86:3611–3619.[Abstract/Free Full Text]

Cassida, K. A., and M. R. Stokes. 1986. Eating and resting salivation in early lactating dairy cows. J. Dairy Sci. 69:1282–1292.[Abstract/Free Full Text]

Clark, P. W., and L. E. Armentano. 1997. Influence of particle size on the effectiveness of beet pulp fiber. J. Dairy Sci. 80:898–904.[Abstract]

Clark, P. W., and L. E. Armentano. 1999. Influence of particle size on the effectiveness of the fiber in corn silage. J. Dairy Sci. 82:581–588.[Abstract]

Clark, P. W., and L. E. Armentano. 2002. Influence of particle size on the effectiveness of the fiber in alfalfa silage. J. Dairy Sci. 85:3000–3007.[Abstract/Free Full Text]

De Brabander, D. L., J. L. De Boever, J. M. Vanacker, and N. E. Geerts. 2002. Evaluation and effects of physical structure in dairy cattle nutrition. Pages 182–197 in Recent developments and perspectives in bovine medicine, Proc. XXII World Buiatrics Congr., Hanover, Germany. M. Kaske, H. Scholz, and M. Höltershinken, ed. Tierärtzliche Hochschule, Hanover, Germany.

Einarson, M. S., J. C. Plaizier, and K. M. Wittenberg. 2004. Effects of barley silage chop length on productivity and rumen conditions of lactating dairy cows fed total mixed ration. J. Dairy Sci. 87:2987–2996.[Abstract/Free Full Text]

Eun, J.-S., K. A. Beauchemin, S.-H. Hong, and W. Z. Yang. 2004. Effects of mechanical processing on the nutritive value of barley silage for lactating dairy cows. J. Dairy Sci. 87:4170–4177.[Abstract/Free Full Text]

Fernandez, I., C. Martin, M. Champion, and B. Michalet-Doreau. 2004. Effect of corn hybrid and chop length of whole-plant corn silage on digestion and intake by dairy cows. J. Dairy Sci. 87:1298–1309.[Abstract/Free Full Text]

Firkins, J. L., M. L. Eastridge, N. R. St-Pierre, and S. M. Noftsger. 2001. Effects of grain variability and processing on starch utilization by lactating dairy cattle. J. Dairy Sci. 79(E. Suppl.):E218–E238.

Garrett, E. F., M. N. Pereira, K. V. Nordlund, L. E. Armentano, W. J. Goodger, and G. R. Oetzel. 1999. Diagnostic methods for the detection of subacute ruminal acidosis in dairy cows. J. Dairy Sci. 82:1170–1178.[Abstract]

Grant, R. J., V. F. Colenbrander, and D. R. Mertens. 1990a. Milk fat depression in dairy cows: Role of particle size of alfalfa hay. J. Dairy Sci. 73:1823–1833.[Abstract]

Grant, R. J., V. F. Colenbrander, and D. R. Mertens. 1990b. Milk fat depression in dairy cows: Role of silage particle size. J. Dairy Sci. 73:1834–1842.[Abstract]

Harvatine, D. I., J. E. Winkler, M. Devant-Guille, J. L. Firkins, N. R. St-Pierre, B. S. Oldick, and M. L. Eastridge. 2002. Whole linted cottonseed as a forage substitute: Fiber effectiveness and digestion kinetics. J. Dairy Sci. 85:1988–1999.[Abstract/Free Full Text]

Junck, B., M. Tafaj, Q. Zebeli, R. Funk, H. Steingass, and W. Drochner. 2004. Einfluss der Häcksellänge auf Strukturwert und Qualität der Maissilage (Influence of length of the cut on the physical evaluation and quality of corn silage). Page 96 in Proc. 116th VDLUFA Congress, Rostock, Germany. (Abstr.) VDLUFA-Verlag, Darmstadt, Germany.

Kohn, R. A. 2000. Three conditions of the ruminal milieu that determine pH. Proc. 25th Conf. Rumen Function, Chicago, IL. Department of Animal Science, Michigan State University, East Lansing. Online. Available http://www.msu.edu/user/rumen/index.htm

Kononoff, P. J., and A. J. Heinrichs. 2003a. The effect of corn silage particle size and cottonseed hulls on cows in early lactation. J. Dairy Sci. 86:2438–2451.[Abstract/Free Full Text]

Kononoff, P. J., and A. J. Heinrichs. 2003b. The effect of reducing alfalfa haylage particle size on cows in early lactation. J. Dairy Sci. 86:1445–1457.[Abstract/Free Full Text]

Kononoff, P. J., A. J. Heinrichs, and D. A. Buckmaster. 2003b. Modification of the Penn State forage and total mixed ration particle separator and the effects of moisture content on its measurements. J. Dairy Sci. 86:1858–1863.[Abstract/Free Full Text]

Kononoff, P. J., A. J. Heinrichs, and H. A. Lehman. 2003a. The effect of corn silage particle size on eating behavior, chewing activities, and rumen fermentation in lactating dairy cows. J. Dairy Sci. 86:3343–3353.[Abstract/Free Full Text]

Krause, K. M., and D. K. Combs. 2003. Effects of forage particle size, forage source, and grain fermentability on performance and ruminal pH in midlactation cows. J. Dairy Sci. 86:1382–1397.[Abstract/Free Full