| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
1 Laboratory for Animal Nutrition and Animal Product Quality, Ghent University, 9090 Melle, Belgium
2 Nutrition and Food, Animal Sciences Group, Wageningen UR, 8200 AB Lelystad, The Netherlands
3 CECA/ICETA, Faculdade de Ciências, Universidade do Porto, Rua Padre Armando Quintas, 4485-661 Vairão VC, Portugal
4 Institute of Grassland and Environmental Research, Plas Gogerddan, Aberystwyth, SY23 3EB, UK
Corresponding author: V. Fievez; e-mail: veerle.fievez{at}UGent.be.
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Key Words: odd and branched-chain fatty acids • microbial protein • dairy cattle • milk
Abbreviation key: CCC = concordance correlation coefficient, CV = coefficient of variation, DAPA = diaminopimelic acid, MSPE = mean square prediction error, OBCFA = odd and branched-chain fatty acids, PB = purine bases, RMSE = root mean square error.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Several reports have been published on the use of urinary and milk allantoin excretion to estimate duodenal flow of microbial protein (reviewed by Dewhurst et al., 2000). Although milk allantoin seems most promising as an on-farm diagnostic tool, its use is obscured by its auto-correlation with milk yield (Dewhurst et al., 2000), underlining the need for research on other milk constituents as markers for the duodenal flow of microbial nitrogen.
The fatty acid composition of rumen bacteria is characterized by a large proportion of odd and branched-chain fatty acids (OBCFA) in their membrane lipids (pentadecanoic acid, C15:0; iso methyltetradecanoic acid, iso C15:0; anteiso methyltetradecanoic acid, ante-iso C15:0; heptadecanoic acid, C17:0; iso methylhexadecanoic acid, iso C17:0; anteiso methylhexadecanoic acid, anteiso C17:0, and heptadecenoic acid, C17:1) (Kaneda, 1991). Microbial odd chain fatty acids (C15:0 and C17:0) are formed through elongation of propionate or valerate, whereas precursors of branched-chain fatty acids (iso C15:0, iso C17:0, anteiso C15:0, anteiso C17:0) are branched-chain amino acids (valine, leucine, and isoleucine) and their corresponding branched short-chain carboxylic acids (isobutyric, isovaleric, and 2-methyl butyric acid) (Kaneda, 1991). In 1962, Keeney et al. suggested that the major source of these fatty acids in ruminant milk is from rumen bacterial origin rather than from animal synthesis. Although odd chain fatty acids or their anteiso isomers can be synthesized through incorporation of propionyl-CoA instead of acetyl-CoA, or methylmalonyl-CoA instead of malonyl-CoA, respectively, the contribution of the latter processes to OBCFA in milk from dairy cows is assumed negligible (Croom et al., 1981; Massart-Leën et al., 1983).
Hence, Keeney et al. (1962) suggested that bacterial OBCFA might be used as markers to quantify bacterial matter leaving the rumen. Since then, little attention has been paid to these fatty acids in ruminant research, but recently their presence in milk has been related to differences in feeding strategies (Dewhurst et al., 2000; Cabrita et al., 2003b) and rumen fermentation pattern (Vlaeminck et al., 2003). Moreover, based on large differences in microbial OBCFA pro-files, Keeney et al. (1962) and Dewhurst et al. (2000) suggested that OBCFA in duodenal digesta and milk could provide a qualitative description of the proportions of different classes of microbes leaving the rumen.
The objective of this work was to evaluate the potential of OBCFA to predict duodenal flow of microbial matter. For this purpose, data from 2 feeding experiments were used. Firstly, in experiment 1 (Bruinenberg et al., 2004), the relationship was investigated between the amounts of OBCFA and microbial matter in rumen contents, using purine bases (PB) and uracil as reference markers. Secondly, duodenal flow of microbial matter, using diaminopimelic acid (DAPA) and PB as reference markers, was related to milk secretion of OBCFA in experiment 2 (Hindle et al., accepted; van Vuuren et al., 2004). Thirdly, the relationships established in experiment 2 were validated using data of the experiments reported by Cabrita et al. (2003a, b). Finally, the predictive precision based on milk OBCFA was compared with that of previously published equations based on DMI and dietary characteristics (Oldick et al., 1999; Gosselink et al., 2003).
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
). The latter were obtained from one intensively managed grassland and 2 seminatural grasslands: one species-poor, composed mainly of meadow bird grass, and the other species-rich, consisting of 34% nonleguminous herbs, 11% legume, and 55% grasses on a DM basis. Cows were fed (twice daily) silage and a protein-rich concentrate (4.5 kg of DM/ d). The mean intake of silages was 14.0 kg of DM/d (on average, 13.9, 14.2, 13.9, and 13.7 kg of DM for diets 1, 2, 3, and 4, respectively). At 0600 and 1600 h, cows received 40 and 60% of the daily DM offered, respectively. After a 2-wk adaptation period, rumen contents were manually evacuated 4, 12, and 17 h after feeding. At each sampling time, ruminal contents were weighed and mixed thoroughly, and a subsample was frozen.
|
Analyses
Rumen and duodenal contents were freeze-dried before analysis. Purine bases and uracil were extracted from freeze-dried samples using perchloric acid, as described by Makkar and Becker (1999). In detail, 0.5 g of freeze-dried rumen or duodenal sample was weighed into a 25-mL screw-cap tube and hydrolyzed for 1 h at 95°C, using 3 mL of 12 M perchloric acid. After cooling, 25 mL of H2O was added and vortexed and the contents of the tube were filtered (S&S 589/2, Schleicher & Schuell MicroScience, Dassel, Germany), with recovery of the filtrate in a 50-mL volumetric flask. The final volume was brought to 50 mL using H2O. Part of the final solution was filtered (0.22-µm filter, Alltech, Lokeren, Belgium) and used for determination of PB and uracil by HPLC with a Partisil SCX column (10 µm, 250 x 4.6 mm, Alltech) guarded with a precolumn (7.5 x 4.6 mm, Alltech). The buffer solution (0.01 M KH2PO4) was run isocratically at 1 mL/min and the effluent monitored at 250 nm. Hydrolysis and analyses were performed in duplicate. Although the rumen microbes were not isolated from the rumen contents before uracil and PB analyses, contribution of nonmicrobial nucleic acids was assumed negligible 4 h after feeding (McAllan, 1982). Hence, both uracil and PB were used as references for changes in the amount of rumen microbial matter. Duodenal DAPA was analyzed according to Moore et al. (1985).
Lipids were extracted from 2.5 g of freeze-dried rumen contents using chloroform:methanol (2:1, vol/vol), adopted from Folch et al. (1957). Fatty acids in extracted lipids were methylated with NaOH in methanol (0.5 mol/L) followed by HCl in methanol (1/1; vol/ vol) (Raes et al., 2001). Nonadecanoic acid (Sigma, Bornem, Belgium) was used as the internal standard. The fatty acid methyl esters were extracted twice with 2 mL of hexane, and pooled extracts were evaporated to dryness under N2. The residue was dissolved in 1 mL of hexane.
Milk fat from thawed milk was extracted by the Röse-Gottlieb procedure (ISO-3889). Briefly, a homogenized sample (10 g) was weighed into a Mojonnier-type fat-extraction flask. Ammonium hydroxide solution (2 mL) was added and the mixture was shaken vigorously. Further, in a first extraction step, 10 mL of ethanol, 25 mL of diethyl ether, and 25 mL of petroleum ether were added and mixed gently. During the second extraction step, 5 mL of ethanol, 15 mL of di-ethyl ether, and 15 mL of petroleum ether was used, and only diethyl and petroleum ether (15 mL) were used during the final extraction step. After each extraction step, the supernatant was transferred into fat-collecting vessels after phase separation (30 min). The solvent was evaporated using a rotary evaporator at room temperature, and the extracted lipids were resolved in 20 mL of diethyl ether:petroleum ether (1:1, vol/vol). Tridecanoic and nonadecanoic acid (Sigma) were used as the internal standards and added before methylation. Fatty acids in extracted lipids were methylated as described previously. The fatty acid methyl esters were extracted twice with 2 mL of hexane. From the pooled extracts, 1 mL was analyzed immediately by gas chromatography for short- and medium-chain fatty acids (C4:0, C6:0, C8:0, and C10:0), and the reminder was evaporated to dryness under N2. The residue was dissolved in 1 mL of hexane.
The methylated fatty acids were analyzed on a Hewlett-Packard 6890 gas chromatograph (Hewlett-Packard Co., Brussels, Belgium) with a CP-Sil88 column for fatty acid methyl esters (100 m x 0.25 mm x 0.2 µm; Chrompack Inc., Middelburg, the Netherlands). For ruminal fatty acids, the following temperature program was used: 150°C for 2 min, followed by an increase at 1°C/min to 158°C, held at 158°C for 28 min, then increased at 1°C/min to 196°C. For milk samples, the following temperature program was used: 70°C for 4 min, followed by an increase at 10°C/ min to 150°C, then increased at 1°C/min to 165°C, held at 165°C for 20 min, increased at 2°C/min to 170°C, held at 170°C for 10 min, increased at 4°C/min to 215°C, and held at 215°C for 20 min. Peaks were identified based on their retention times, corresponding with the standards (Sigma and NuChek Prep, Elysian, MN). Standard curves to determine the response factors for milk short-chain fatty acids were determined daily. Short-chain fatty acids in milk were quantified using tridecanoic acid as internal standard, whereas the other fatty acids were quantified with nonadecanoic acid.
Calculations
Independent data set to evaluate accuracy of prediction.
Predictive equations can only be evaluated properly by the means of an independent experiment. However, data on milk OBCFA are still very scarce. Hence, data from the experiments described by Cabrita et al. (2003a, b) were used to evaluate the accuracy of the predictive equations, although duodenal flow of microbial matter was not measured directly in these experiments but evaluated using the allantoin/creatinine ratio in spot urine samples. Hence, duodenal flow of PB in the independent data set (Cabrita et al., 2003a, b) was calculated from the allantoin/creatinine ratio in spot urine samples, assuming a constant daily urinary excretion of creatinine (931 µmol/ kg BW0.75; Gonzalez-Ronquillo et al., 2003), a constant endogenous purine derivative contribution (385 µmol/ kg BW0.75; Chen and Gomes, 1995), and the relation of urinary excretion of purine derivatives with duodenal flow of PB as described in Gonzalez-Ronquillo et al. (2003). Consequently, a number of assumptions have been made to calculate "actual" duodenal flow of microbial matter from measured urinary allantoin/creatinine ratios (e.g., daily creatinine excretion and endogenous allantoin excretion) in the independent data set (Cabrita et al., 2003a,b). Nevertheless, even with the assumptions made, relative differences in microbial flow, induced by dietary or animal factors, still might be evaluated and related to milk OBCFA secretion.
Prediction of duodenal flow of microbial matter using published equations based on DMI and dietary characteristics.
Furthermore, previously published equations based on DMI and dietary characteristics as described by Oldick et al. (1999) [equation 10 in Oldick et al. (1999); microbial PB (g/d) = [16.1 + 22.9 x DMI (kg/d) – 0.365 x DMI (kg/d) x DMI (kg/ d) –0.174 x NDF (g/kg of DM)] x 0.147] and Gosselink et al. (2003) [equation 3 in Gosselink et al. (2003); microbial PB (g/d) = (7.80 + 0.000119 x crude protein (g/kg of DM)) x DMI (kg/d) x 0.147] were used as an alternative to predict duodenal flow of microbial matter. As the equations described by Oldick et al. (1999) and Gosselink et al. (2003) were originally expressed as duodenal flow of microbial N, the values were converted to duodenal flow of PB using a PB-to-N ratio in mixed rumen bacteria of 0.147 (Clark et al., 1992; Perez et al., 1997; Volden et al., 1999; Martin-Orue et al., 2000).
Statistical Analyses
All statistical analyses were performed using SPSS 11.0 (SPSS software for Windows, release 11.0, SPSS, Inc., Chicago, IL).
Univariate analysis.
A split-plot analysis for repeated measures was used to evaluate the effect of dietary treatment, cow, experimental period, and sampling time on rumen pool size of OBCFA, uracil, and PB in experiment 1 according to:
 |
where Yijkl is the individual observation, µ the overall mean, Ti the effect of dietary treatment, Pj the effect of experimental period, Ck the effect of cow, Sl the effect of sampling time, TSil the interaction between treatment and sampling time, PSjl the interaction between period and sampling time, CSkl the interaction between cow and sampling time, and 
ijkl the residual error. Sampling time was the repeated measure. Dun-can’s multiple range test was performed to evaluate differences between dietary treatments.
Differences in duodenal flow of DAPA and PB and milk secretion of OBCFA from experiment 2 were evaluated according to:
 |
where Yijk is the individual observation, µ the overall mean, Ti the effect of dietary treatment, Pj the effect of experimental period, Ck the effect of cow, and ijk the residual error. Effect of period and cow were treated as random effects. Duncan’s multiple range test was performed to evaluate differences between dietary treatments. For milk OBCFA, means were calculated per cow and diet from 8 analyses.
Linear regression analyses.
In experiment 1, individual data were used for linear regression analyses. For milk OBCFA in experiment 2, means were calculated per cow and diet from 8 analyses. The backward elimination procedure of multiple regression was performed using the regression procedure of SPSS to develop equations to predict microbial matter in rumen (experiment 1) and duodenal flow of microbial matter (experiment 2).
Regression equations were evaluated based on the root mean square error (RMSE) and coefficients of variation (CV). Moreover, a first evaluation of the prediction accuracy of the regression equations was performed using the statistic PRESS (Rawlings, 1988). This statistic is computed as follows:
 |
where yi is the ith observation for the dependent variable, 
i–i is the prediction of observation i using a model estimated without the ith observation, and n is the number of observations in the data set used for parameter estimation. The statistic PRESS divided by n can be considered an estimate of the mean square prediction error (MSPE). Although this statistic cannot detect lack of accuracy due to inadequacies of the sample used to represent the population studied, the MSPE estimate is useful when independent datasets are not available (Rawlings, 1988).
Independent datasets to evaluate accuracy of prediction.
The regression equations developed in this study were validated with independent datasets using the MSPE (Bibby and Toutenberg, 1977):
 |
where n is the number of observations, and yi and i are the observed and predicted values, respectively. The square root of MSPE is expressed in the same units as the observed values and a comparison of the root MSPE as a percentage of the observed mean provides an indication of the overall error of prediction.
Accuracy and precision of the equations were further evaluated based on the concordance correlation coefficient (CCC) (Lin, 1989; St-Pierre, 2003). The CCC is calculated as:
 |
with 
c the concordance correlation coefficient, the Pearson correlation coefficient, and Cb the bias correction factor which is calculated as:
 |
with 
o and p the standard deviation of the observed and predicted values, respectively, and µo and µp the observed and predicted mean. The Pearson correlation coefficient reflects precision, i.e., the degree to which individual predictions cluster about the regression line. The bias correction factor reflects accuracy, i.e., the degree to which the regression line adheres to the 45° line through the origin (the concordance line). The CCC reflects the degree to which individual predictions adhere to the concordance line. The bias correction factor consists of a scale shift (= o/p) and a location shift relative to the scale (= [µo – µp]/[ o x p]1/2) (Lin, 1989; St-Pierre, 2003).
Comparison with previously published prediction equations.
Using the same independent data-sets, the accuracy of prediction equations based on milk secretion of OBCFA, developed in this study, was compared with predictions by 2 previous published equations (Oldick et al., 1999; Gosselink et al., 2003).
|
RESULTS AND DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
. Dietary treatments with 40% grass silage from intensively managed grassland showed smaller pool sizes of both PB and uracil, which decreased with time after feeding. As could be expected, rumen pool sizes of PB and uracil were closely related [PB (g) = 3.54(SE = 0.051) x uracil (g), n = 48, r2 = 0.991, RMSE = 2.90].
|
). Changes in rumen pool size of OBCFA were closely associated with changes in uracil and PB, confirming the potential of OBCFA to predict rumen pool size of microbial matter (Table 3; Figure 1). A plot of predicted vs. residual values (i.e., observed values – predicted values) did not reveal any pattern (data not shown). Patterns in residuals can indicate that the assumptions of regression procedures are not met or that required independent variables are missing (Draper and Smith, 1981).
|
|
) suggest the ratios PB:OBCFA and uracil:OBCFA in mixed rumen bacteria are 1.16 and 0.34, respectively. Bas et al. (2003) reported a mean OBCFA content of 8.97 g/kg of DM in mixed rumen bacteria. Assuming PB and uracil contents of 10.5 and 2.59 g/kg of DM, respectively (Perez et al., 1997; Volden et al., 1999; Martin-Orue et al., 2000), we calculated the ratios PB:OBCFA and uracil:OBCFA in mixed rumen bacteria to be 1.17 and 0.29, respectively. Obviously, the ratios observed in this experiment are in line with these calculations. Nevertheless, the ratios are clearly susceptible to variation as indicated by the estimated root MSPE of 16.4 and 14.6% of the observed mean for prediction of PB and uracil, respectively (equations 1 and 3 in Table 3). Because OBCFA and PB in rumen contents are the weighted sum of different microbial populations, it was anticipated that the ratio might be affected by variation in the composition of the rumen microbial population. Indeed, significant differences are reported in both OBCFA (Ifkovitz and Ragheb, 1968; Viviani, 1970), and uracil and PB contents (Arambel et al., 1982; Obispo and Dehority, 1999) between different strains of rumen bacteria. Particle-associated rumen bacteria contain generally more OBCFA (9.98 vs. 7.52 g/kg of DM; Bauchart et al., 1990) compared with liquid-associated rumen bacteria. The opposite is true for PB and uracil [9.59 vs. 13.0 g/kg of DM (Perez et al., 1997; Volden et al., 1999; Martin-Orue et al., 2000) and 2.57 vs. 2.68 g/kg of DM (Volden et al., 1999) for PB and uracil, respectively]. Hence, variation in rumen proportions of particle- or fluid-associated bacteria clearly affects the ratios PB:OBCFA and uracil:OBCFA in mixed rumen bacteria. Although proportions of particle- and fluid-associated bacteria were not determined, the negative relation of time after feeding with the residuals from the equation relating OBCFA to PB (rpearson = – 0.542, P = 0.001) or uracil (rpearson = –0.528, P = 0.001) suggests that variation in the proportions of particle- and fluid-associated bacteria did occur in this experiment. The negative relation of the residuals with time after feeding indicates a relatively higher contribution to the rumen bacterial pool of fluid-associated bacteria 4 h after feeding compared with 12 and 17 h after feeding. This finding is not surprising because fiber degradation and the contribution of particle-associated bacteria is expected to change with time after feeding.
Relation of rumen pool size of C17:0 with PB and uracil.
Although a strong relationship was observed between rumen pool size of OBCFA and microbial matter, it was affected by variation in the composition of the microbial population in the rumen. Little information is available on the ratio of individual OBCFA to PB and uracil in rumen bacteria, although data from Hussein et al. (1995) suggest relatively low variation in the ratio of C17:0 to PB in mixed rumen bacteria. Because rumen OBCFA, PB, and uracil are of microbial origin, the ratios OBCFA:PB and OBCFA:uracil in whole rumen contents reflect the ratio in mixed rumen bacteria. In the current experiment, the ratio of C17:0 to PB and of C17:0 to uracil in rumen content was the least variable (data not shown). Hence, rumen pool size of C17:0 (Table 2), instead of total OBCFA, was used to estimate rumen pool sizes of PB and uracil (equations 2 and 4 in Table 2; Figure 1).
Relationship of Duodenal Flow of DAPA and PB with Milk Secretion of OBCFA (Experiment 2)
The results of experiment 1 suggested that OBCFA in rumen contents are strongly related to microbial matter. However, for practical reasons, predictions of rumen microbial protein yield based on milk OBCFA would be of most value. Indeed, milk can be collected noninvasively and contains considerable amounts of microbially derived OBCFA (Fievez et al., 2003).
Duodenal flow of PB and DAPA.
Dietary treatment influenced duodenal flow of microbial matter (Table 4). Flow of microbial matter to the duodenum was greater for the potato starch and wheat starch diet, probably due to differences between diets in true ruminally degraded OM and/or microbial growth efficiency (Oba and Allen, 2003). Duodenal flow of DAPA and PB were closely related [PB (g/d) = 2.87(SE = 0.043) x DAPA; n = 16; r2 = 0.997; RMSE = 2.38 g/d].
|
). The prediction equations showed promising low CV of 10.6 and 10.1% for duodenal flow of PB and DAPA, respectively (equations 5 and 7 in Table 5). Indeed, the CV values obtained in the current experiment were in line with or considerably lower than CV values reported by others who developed equations to predict duodenal flow of microbial matter based on DMI and dietary characteristics (e.g., CV = 8.2 to 14.2%, Oldick et al., 1999; CV = 29 to 30%, Gosselink et al., 2003). Nevertheless, direct comparison of CV values should be interpreted with caution, as residual variation in the predictive equations of Oldick et al. (1999) and Gosselink et al. (2003) might be partly due to the inherent variation in flow measurements and microbial markers used in the series of experiments of their databases. Obviously, this might not have been the origin of the residual variation of our predictive equations, as they were based on data of one experiment only. Despite the promising low CV, prediction equations based on milk secretion of OBCFA were biased, as indicated by the linear pattern in the residuals (data not shown).
|
(equations 6 and 8). These prediction equations were no longer biased (data not shown). However, we did not observe the expected decrease in the CV value when predictions were based on C17 fatty acids compared with predictions based on total OBCFA (Table 5). Because earlier investigations of the predictive value of milk allantoin were partially discounted because of a mutual correlation with milk yield, it is worth noting that there was only a weak relationship between daily milk production and duodenal flow of microbial matter (rpearson = 0.160 and 0.361 for DAPA and PB flow, respectively, P > 0.10) in the present study. In addition, milk production explained only 17.8% of the variation in milk secretion of OBCFA (P >0.10).
Evaluation of the prediction accuracy of equations based on milk OBCFA with independent data set.
Although the relatively low estimated root MSPE, reasonably close to RMSE (Table 5), supports the validity of the fitted regression model (Neter et al., 1996), comparison of MSPE values obtained with an independent data set is the best way to evaluate the accuracy of different equations. The results of the present validation using an independent data set (Cabrita et al., 2003a,b) is presented in Figure 2. Predictions based on milk secretion of C17 fatty acids (equation 6) were more accurate compared with predictions based on milk secretion of OBCFA (equation 5), resulting in a root MSPE of 20.0 and 13.5% of the observed mean for equations 5 and 6, respectively. Predictions based on milk secretion of OBCFA were less concordant compared with predictions based on C17 fatty acids (Figure 2). The lower CCC was mainly due to a lower accuracy (0.633 and 0.957 for equations 5 and 6, respectively), whereas precision was higher for predictions based on milk secretion of OBCFA (0.851 and 0.753 for equations 5 and 6, respectively). The lower accuracy for predictions based on milk secretion of OBCFA was attributed to a higher mean bias as illustrated by the higher location shift (1.06 and –0.090 for equations 5 and 6, respectively) and the significant difference of the intercept from 0 when regressing predicted values on observed values (Figure 2). As depicted in Figure 2, the intercept and the slope of the regression of the predicted values on the observed values were not different from 0 and 1, respectively, when regression was based on milk secretion of C17 fatty acids (P < 0.05).
|
Comparison of the prediction equations based on milk OBCFA with published prediction equations based on DMI and dietary characteristics.
The root MSPE values of the equation reported by Oldick et al. (1999) (22.6%) and the equation based on the secretion of OBCFA (equation 5) were similar. The equation of Gosselink et al. (2003) was the least accurate, with a root MSPE of the observed mean of 34.5%. Both equations based on DMI and dietary characteristics (Oldick et al., 1999, Gosselink et al., 2003) showed low CCC values, which was mainly due to a relatively low accuracy (0.347 and 0.283 for equations of Oldick et al., 1999 and Gosselink et al., 2003, respectively). Indeed, predictions based on DMI and dietary characteristics (Oldick et al., 1999, Gosselink et al., 2003) were not able to accurately reflect changes in microbial flow to the duodenum, as additionally indicated by the low standard deviation of the predicted values (Figure 2). It should be noted that the original data sets used by Oldick et al. (1999) and Gosselink et al. (2003) for estimation of the regression equations were treatment means, resulting in predictions of duodenal flow of microbial matter for an "average cow." In contrast, regression equations developed in this study were based on individual cow observations as animals notoriously vary in duodenal flow of microbial matter on the same diet. These differences are accounted for when using milk OBCFA, but are largely neglected in equations based on dietary characteristics only.
In general, prediction of duodenal flow of microbial matter seems most accurate when based on milk secretion of OBCFA, and in particular, C17 fatty acids. Nevertheless, as more data become available, more trials will be included in this kind of evaluation to ensure the evaluation is not biased by trial effects.
This first evaluation suggests that milk OBCFA could be used as a marker for microbial material, particularly when accurate measurements of DMI are lacking. Moreover, predictive equations in this first evaluation were developed using a relatively small data set with limited variation in dietary treatments. Hence, it is suspected that enlarging the data set will result in more accurate prediction equations. Enlarging the data set with additional experiments is currently in progress. Additionally, the effect of dietary fat levels on this relationship should be elucidated, as dairy cows were not fed supplemental fat in the current experiments. Indeed, increased dietary fat supply (mainly 18-carbon fatty acids and C16:0) reduces the proportion of OBCFA in duodenal contents and milk (Fievez et al., 2003). This decrease in OBCFA with higher dietary fat levels may be due to (partial) inhibition of de novo synthesis of OBCFA by rumen bacteria at higher dietary fat levels (Demeyer et al., 1978) or dilution with greater concentrations of even-chain fatty acids in milk. Furthermore, additional studies are needed to identify the effects of other factors, such as lactation stage and contribution of OBCFA from body fat to milk OBCFA, particularly during periods of negative energy balance and parity.
|
CONCLUSIONS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
Received for publication July 2, 2004. Accepted for publication November 18, 2004.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTS AND DISCUSSIONCONCLUSIONSACKNOWLEDGEMENTSREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  M. Craninx, A. Steen, H. Van Laar, T. Van Nespen, J. Martin-Tereso, B. De Baets, and V. Fievez Effect of Lactation Stage on the Odd- and Branched-Chain Milk Fatty Acids of Dairy Cattle Under Grazing and Indoor Conditions J Dairy Sci, July 1, 2008; 91(7): 2662 - 2677. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
O. A. Rego, S. M. M. Regalo, H. J. D. Rosa, S. P. Alves, A. E. S. Borba, R. J. B. Bessa, A. R. J. Cabrita, and A. J. M. Fonseca Effects of Grass Silage and Soybean Meal Supplementation on Milk Production and Milk Fatty Acid Profiles of Grazing Dairy Cows J Dairy Sci, July 1, 2008; 91(7): 2736 - 2743. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
P. A. Abrahamse, J. Dijkstra, B. Vlaeminck, and S. Tamminga Frequent Allocation of Rotationally Grazed Dairy Cows Changes Grazing Behavior and Improves Productivity J Dairy Sci, May 1, 2008; 91(5): 2033 - 2045. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Vlaeminck, G. Mengistu, V. Fievez, L. de Jonge, and J. Dijkstra Effect of In Vitro Docosahexaenoic Acid Supplementation to Marine Algae-Adapted and Unadapted Rumen Inoculum on the Biohydrogenation of Unsaturated Fatty Acids in Freeze-Dried Grass J Dairy Sci, March 1, 2008; 91(3): 1122 - 1132. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. J. Dewhurst, J. M. Moorby, B. Vlaeminck, and V. Fievez Apparent Recovery of Duodenal Odd- and Branched-Chain Fatty Acids in Milk of Dairy Cows J Dairy Sci, April 1, 2007; 90(4): 1775 - 1780. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
H. Rulquin, B. Graulet, L. Delaby, and J. C. Robert Effect of different forms of methionine on lactational performance of dairy cows. J Dairy Sci, November 1, 2006; 89(11): 4387 - 4394. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
B. Vlaeminck, V. Fievez, S. Tamminga, R. J. Dewhurst, A. van Vuuren, D. De Brabander, and D. Demeyer Milk odd- and branched-chain Fatty acids in relation to the rumen fermentation pattern. J Dairy Sci, October 1, 2006; 89(10): 3954 - 3964. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
J. L. Firkins, A. N. Hristov, M. B. Hall, G. A. Varga, and N. R. St-Pierre Integration of Ruminal Metabolism in Dairy Cattle J Dairy Sci, March 1, 2006; 89(e_suppl_1): E31 - E51. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. P. Alves, C. Marcelino, P. V. Portugal, and R. J. B. Bessa Short Communication: The Nature of Heptadecenoic Acid in Ruminant Fats J Dairy Sci, January 1, 2006; 89(1): 170 - 173. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||
