If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
TI Food and Nutrition, PO Box 557, 6700 AN Wageningen, the NetherlandsAnimal Nutrition Group, Wageningen University, PO Box 338, 6700 AH Wageningen, the Netherlands
Various studies have indicated a relationship between enteric methane (CH4) production and milk fatty acid (FA) profiles of dairy cattle. However, the number of studies investigating such a relationship is limited and the direct relationships reported are mainly obtained by variation in CH4 production and milk FA concentration induced by dietary lipid supplements. The aim of this study was to perform a meta-analysis to quantify relationships between CH4 yield (per unit of feed and unit of milk) and milk FA profile in dairy cattle and to develop equations to predict CH4 yield based on milk FA profile of cows fed a wide variety of diets. Data from 8 experiments encompassing 30 different dietary treatments and 146 observations were included. Yield of CH4 measured in these experiments was 21.5 ± 2.46 g/kg of dry matter intake (DMI) and 13.9 ± 2.30 g/kg of fat- and protein-corrected milk (FPCM). Correlation coefficients were chosen as effect size of the relationship between CH4 yield and individual milk FA concentration (g/100 g of FA). Average true correlation coefficients were estimated by a random-effects model. Milk FA concentrations of C6:0, C8:0, C10:0, C16:0, and C16:0-iso were significantly or tended to be positively related to CH4 yield per unit of feed. Concentrations of trans-6+7+8+9 C18:1, trans-10+11 C18:1, cis-11 C18:1, cis-12 C18:1, cis-13 C18:1, trans-16+cis-14 C18:1, and cis-9,12 C18:2 in milk fat were significantly or tended to be negatively related to CH4 yield per unit of feed. Milk FA concentrations of C10:0, C12:0, C14:0-iso, C14:0, cis-9 C14:1, C15:0, and C16:0 were significantly or tended to be positively related to CH4 yield per unit of milk. Concentrations of C4:0, C18:0, trans-10+11 C18:1, cis-9 C18:1, cis-11 C18:1, and cis-9,12 C18:2 in milk fat were significantly or tended to be negatively related to CH4 yield per unit of milk. Mixed model multiple regression and a stepwise selection procedure of milk FA based on the Bayesian information criterion to predict CH4 yield with milk FA as input (g/100 g of FA) resulted in the following prediction equations: CH4 (g/kg of DMI) = 23.39 + 9.74 × C16:0-iso – 1.06 × trans-10+11 C18:1 – 1.75 × cis-9,12 C18:2 (R2 = 0.54), and CH4 (g/kg of FPCM) = 21.13 – 1.38 × C4:0 + 8.53 × C16:0-iso – 0.22 × cis-9 C18:1 – 0.59 × trans-10+11 C18:1 (R2 = 0.47). This indicated that milk FA profile has a moderate potential for predicting CH4 yield per unit of feed and a slightly lower potential for predicting CH4 yield per unit of milk.
Enteric fermentation is the main source of greenhouse gas (GHG) emissions from dairy cattle, with enteric CH4 amounting to 1.1 Gt per year, representing 46% of the total GHG emissions in dairy supply chains (
Tackling Climate Change Through Livestock—A Global Assessment of Emissions and Mitigation Opportunities. Food and Agriculture Organization of the United Nations (FAO),
Rome, Italy2013
). In view of these emissions and various mitigation options, there is a clear need for simple and inexpensive measurement techniques to estimate CH4 emissions from dairy cattle in commercial practice.
Various methods have been used to estimate CH4 production from ruminants. Production of CH4 can be measured in respiration chambers, which is an accurate but expensive technique, unsuitable for application on a large scale. Other methods, including the SF6 marker and CH4:CO2 ratio techniques, enable CH4 emissions to be determined in a larger number of animals but with higher between- and within-animal variation (
). For inventory and prediction purposes, many empirical equations have been developed to relate CH4 production per day or yield per unit of feed to feed intake and composition. Such equations have major limitations in predicting effects of mitigation strategies at the whole-farm level (
). Mechanistic models describing the mechanism of enteric feed degradation provide more accurate predictions of CH4 production than do empirical models (
). However, mechanistic models are more complex and require inputs that may not be commonly measured. In addition, forcing functions used and bias in parameter values adopted may limit the meticulousness of predicted model output. For these reasons, a simple and robust prediction equation of enteric CH4 yield from dairy cattle based on characteristics of feed or milk would be valuable for application on a large scale in GHG mitigation practices for the dairy industry.
Milk samples are frequently used in dairy farms to assess the nutritional and health status of dairy cattle and to obtain information on losses to the environment. For example, milk urea content is used to assess protein status of the animal and to estimate N excretion (
). Such relationships may be a result of lipid supplementation, which changes both CH4 production and milk FA profile, or may be a result of changes in diet composition, in view of CH4 production being associated with the VFA profile produced in the rumen (
supplemented cattle diets with different physical forms of linseed (crude, extruded, and oil). The most positive correlations between CH4 production (g/d) and milk FA concentrations were obtained for saturated FA (C6:0 to C16:0) and the most negative correlations for various trans-C18 FA. Milk OBCFA concentrations exhibited less strong correlations with CH4 production. These relationships may only apply to linseed supplemented diets.
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
, using only dietary oilseed supplementation (sunflower seed, linseed, and canola seed), best predicted CH4 production (g/d) by milk FA concentration of C16:0-iso (positive relationship) and cis-9 C17:1 (negative relationship).
, using a larger variety of diets (3 experiments, 10 dietary treatments) in which fat supplementation was a major source of dietary variation, evaluated relationships between CH4 yield per unit of feed and milk FA profile in dairy cattle. Their prediction equation included milk FA concentration of C17:0-anteiso and cis-13 C18:1 (positive relationship) and trans-10+11 C18:1 and cis-11 C18:1 (negative relationship). The various models to predict CH4 emission in these 3 studies have only a few milk FA in common. This might be a result of the small number of experiments and the limited variation in dietary treatments as well as the analytical methods used to elucidate milk FA profile. Data from a greater number of experiments containing a wider variety of diets are required to firmly assess the potential of milk FA profile as an indicator of CH4 yield. A greater number of experiments also allows quantification of between-study variability or heterogeneity of the correlation between milk FA concentrations and CH4 yield.
The aims of this study were to perform a meta-analysis to quantify relationships between CH4 yield and individual milk FA concentrations in lactating dairy cattle while quantifying the heterogeneity of these relationships, and to develop equations to predict CH4 yield (per unit of feed and per unit of milk) based on milk FA profile of cows fed a wide variety of diets. Such equations may ultimately be used to estimate CH4 yield from dairy cattle under field conditions to fulfill the need for simple, inexpensive measurement techniques.
Materials and Methods
Data Collection
For inclusion in the present meta-analysis, studies were required to have CH4 production measured using respiration chambers and milk FA profile elucidated using gas chromatography. Four studies designed as 4 × 4 Latin squares from the University of Reading (Reading, UK) and 4 studies designed as randomized block experiments from Wageningen University (Wageningen, the Netherlands) met these requirements and were included (Table 1). The procedures to determine CH4 production and milk FA profile are described by
for the Wageningen studies. The 8 studies represented 30 different dietary treatments and 146 individual observations encompassing a variety of diets. Studies 1, 3, 4, 7, and 8 contained lipid treatments, whereas studies 2, 5, and 6 did not contain any lipid treatment. Studies 4 and 5 comprised diet treatments with different forage types and contents, and studies 2, 6, 7, and 8 contained various nonlipid additives. Animals were described by treatment diet composition, DMI, milk yield, milk composition, milk FA profile, and CH4 production (Tables 2 and 3). Methane yield was expressed per unit of feed (g/kg of DMI) and per unit of fat- and protein-corrected milk (g/kg of FPCM), where FPCM (kg/d) = [0.337 + 0.116 × milk fat (%) + 0.06 × milk protein (%)] × milk production (kg/d) (
Basal diet comprising 37.5% corn silage, 12.5% grass silage; treatments comprising 3.5% DM: control diet with calcium salts of palm oil fed once daily vs. control diet with calcium salts of palm oil fed twice daily vs. milled rapeseed diet fed twice daily vs. coconut oil diet fed twice daily.
Reynolds, C. K., D. J. Humphries, P. Kirton, L. A. Crompton, J. A. N. Mills, and D. I. Givens. 2010. Methane production by lactating dairy cows fed diets containing allicin, glycerol or naked oats. Page 120 in Proc. 4th Intl. Greenhouse Gases Anim. Agric. Conf., Banff, Canada.
van Gastelen, S., E. C. Antunes Fernandes, K. A. Hettinga, G. Klop, S. J. J. Alferink, and J. Dijkstra. 2014. Replacing grass silage with maize silage affects rumen fermentation characteristics and enteric methane production in dairy cattle. In Proc. 39th Anim. Nutr. Res. Forum, Utrecht, the Netherlands. Universiteit Utrecht, Utrecht, the Netherlands.
Dietary inclusion of diallyl disulfide, yucca powder, calcium fumarate, an extruded linseed product, or medium-chain fatty acids does not affect methane production in lactating dairy cows.
29% grass silage, 22% corn silage, 2% wheat straw, 47% concentrates; supplemented with palm oil vs. supplemented with calcium fumarate plus mix of lauric acid, myristic acid and linseed oil.
Dietary inclusion of diallyl disulfide, yucca powder, calcium fumarate, an extruded linseed product, or medium-chain fatty acids does not affect methane production in lactating dairy cows.
41% grass silage, 35% corn silage and 24% concentrates; control vs. extruded linseed supplement vs. diallyldisulfide supplement vs. mixture of caprylic acid and capric acid supplement.
Some of the milk FA profile analyses did not allow identification of certain individual milk FA but did identify certain FA together as one fraction. When these FA were individually identified in other studies, they were grouped together. Milk FA fractions were expressed in g/100 g of total milk FA. Fatty acid fractions with an average study concentration <0.1 g/100 g of milk FA were excluded from the data set.
Statistical Analysis
Random-Effects Model Analysis
Relationships between CH4 yield per unit of feed and per unit of milk, and individual milk FA concentrations for dairy cattle were meta-analyzed using the metafor package (version 1.6–0;
) in R (version 2.15.2; R Foundation for Statistical Computing, Vienna, Austria). The effect size of these relationships for each of the 8 studies was estimated by correlation coefficients (values in Supplemental Tables S1 and S2; http://dx.doi.org/10.3168/jds.2014-8268). The correlation coefficients were obtained by linear regression using individual animal data. In contrast to treatment mean data, use of individual animal data prevents one from ignoring variation of CH4 production and milk FA concentrations at the animal level due to, for example, parity and DMI level. The correlation coefficients were transformed via Fisher’s r-to-Z transformation, Z = tanh−1(r), where tanh−1 = the inverse hyperbolic tangent, which is defined as follows: tanh−1(r) = 0.5 × ln[(1 + r)/(1 − r)] (
). This transformation ensures more stable variance and normality. To obtain the average true effect, the meta-analytic model applied is given by
[1]
where yi is the observed effect in the ith study, μ is the average true effect, ui is the variability among the true effect induced by study, and ei is the sampling error with ei ~N(0, vi). The sampling variance, vi, is known based on the number of observations per study. The study effect, ui, was taken into account as a random factor. This model is referred to as the random-effects model.
In the random-effects model, the variability among studies (or heterogeneity), which is regarded to be induced by experimental circumstances, is assumed to be normally distributed with variance τ2 such that ui ~N(0, τ2). The model was fit with REML. Heterogeneity (τ2) was expressed as percentage of the total variability in the effect size (τ2 plus sampling error), yielding the I2 statistic (
). In case of negative values of I2, a value of 0% was adopted; I2 values greater than 50% indicate substantial heterogeneity. The transformed correlations were tested for homogeneity with the Q-statistic (
), which follows a χ2 distribution. Average true correlation coefficients and their boundaries of the 95% CI were back-transformed to raw correlation values for convenience of interpretation. Estimates of average true correlations were declared significant at P ≤ 0.05 and tendencies at 0.05 < P ≤ 0.10. The strength of the estimated average correlations is interpreted as small if 0.10 ≤ |r| < 0.30, medium if 0.30 ≤ |r| < 0.50, and large if |r| ≥ 0.50 (
) were applied using PROC MIXED (SAS Institute Inc., Cary, NC). This enabled analysis of fixed effects of independent variables as well as the effect of study, which was taken into account as a random factor. The general model for single and multiple regression is represented as
[2]
where Yij is the dependent variable (ith study 1, ..., 8, jth observation 1, ..., 146) and Xijk is the value of the kth explanatory variable (k = 1, 2, …, p). The overall intercept B0 and the overall regression coefficients of Y on X across all studies comprise the fixed-effects part of the model for k different parameters. The random effect of the ith study on the overall intercept B0, si, together with the unexplained residual error, eij, comprise the random-effects part of the model, with both assumed to be normal. Random effects were modeled with (co)variance matrices that were fitted with an unstructured approach, providing that matrices converged. In cases of nonconvergence, (co)variance matrices were fitted with a compound symmetry. When matrices still did not converge, they were fitted with variance components. No random effect on slope was included in the multiple regression analysis to prevent overparameterization. A selection procedure for multiple regression was performed using a stepwise procedure (PROC GLMSELECT in SAS), retaining the experiment effect in every step, with CH4 yield being the dependent variable and stepwise selection of FA based on the Schwarz Bayesian information criterion, where lower values indicate better model adequacy. All available single FA or FA combinations were included in the selection. For predicting CH4 yield per unit of feed, milk fat and milk protein contents (g/100 g of milk) were included in the selection as well. Parameter estimates for fixed effects were declared significant at P ≤ 0.05. Adjusted dependent variable values were calculated based on regression parameters of the final model to determine r or R2 values corrected for experiment effect (
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
. In the present data set, the forage proportion varied between 50 and 80% of total diet, with forage consisting of grass silage and corn silage in ratios ranging from 0:100 to 100:0 (all on a DM basis). It remains questionable, though, to what extent the present data set represents the variety of diets supplied on commercial dairy farms. In particular, the large variation in forage proportion (fraction of total diet) and composition (type of forage and quality of forage) in practice is not completely represented in the 8 studies included. As in the previous analyses cited above, the data used in the present study include measurements for diets that include supplemental lipids and other ingredients with potential to reduce CH4 yield.
Study effect can be taken into account as either a fixed or random factor in meta-analysis and leads to a fixed-effects model or a random-effects model, respectively. In contrast to fixed-effects models that make a conditional inference only about the number of studies included in the analysis, random-effects models estimate the unconditional inference about a larger set of studies of which the studies included in the analysis are assumed to be a normally distributed random sample (
). Therefore, including the study effect as a random factor, not as fixed factor, is in better agreement with the aim of this meta-analysis to search for a generally applicable prediction equation for CH4 yield with milk FA concentration as input.
Random-Effects Model Analysis Correlation per Unit of Feed
The concentration of C16:0 in milk fat was moderately positively related to CH4 yield (g/kg of DMI), and concentrations of C6:0, C8:0, and C10:0 in milk fat tended to be weakly positively related to CH4 yield (Table 4), which is largely in agreement with previous findings (
), which subsequently yields H2 and ultimately CH4 by methanogenic archaea. Milk C4:0 concentration was not significantly related to CH4 yield. With the exception of C4:0, de novo FA synthesis in the mammary gland of C16 and shorter FA is inhibited in the presence of unsaturated long-chain FA (
). The fact that dietary unsaturated FA generally reduce CH4 yield may explain why concentrations of C4:0 were not related to CH4 yield, in contrast to other even-chain de novo synthesized FA. Moreover, C4:0 in milk fat does not require acetate for its production as it can be produced directly from β-hydroxybutyrate derived from the blood and thus only partly originates from de novo FA synthesis using acetate.
Table 4Estimated true correlation coefficients with standard errors and P-values, P-values for heterogeneity test statistic (Q-value P), and heterogeneity as a fraction of total variability (I2) for the correlation between CH4 yield per unit of feed and per unit of milk (fat- and protein-corrected milk, FPCM), and milk FA concentration
Milk FA
CH4 yield (g/kg of DMI) vs. milk FA concentration
CH4 yield (g/kg of FPCM) vs. milk FA concentration
As for milk FA concentration of C4:0, no significant relationship was found between CH4 yield (g/kg of DMI) and concentrations of C12:0 or C14:0. Based on genetic and herd clustering of milk FA,
found that C4:0 and C12:0 differed from the general pattern of the other FA in the group of de novo synthesized even-chain FA. Milk FA C12:0 and C14:0 are not only synthesized de novo in the mammary gland, but can also originate from dietary C12:0 and C14:0 (e.g.,
). Ingredients such as palm kernel expeller and extracted coconut, with relatively large proportions of C12:0 and C14:0 in fat, are commonly included in dairy cattle diets and were present in various diets in the current data set. Dietary C12:0 and C14:0 inhibit CH4 production (
The effect of dietary fats on methane emissions, and its other effects on digestibility, rumen fermentation and lactation performance in cattle: A meta-analysis.
), which might explain why no overall correlation of concentrations C12:0 and C14:0 in milk fat and CH4 yield was found in this study. However, it should be noted that the level of either C12:0 or C14:0 fed was not available for most of the studies included in this meta-analysis. A part of C14:0 is desaturated to cis-9 C14:1 by Δ9-desaturation in the mammary gland. This desaturation activity is regulated by genetics (
). These contrasting mechanisms together may result in milk cis-9 C14:1 not being related to CH4 yield in the present study. In contrast with our study,
found milk FA concentrations of all even-chain de novo synthesized FA between C4:0 and C16:0 to be positively related to CH4 production. However, in the study of
, dietary treatments differed only in physical form of linseed.
Concentrations of OBCFA in milk FA are related to molar proportions of VFA in the rumen and have potential to be used as rumen microbial markers or indicators of the type of VFA formed and as predictors of CH4 yield (
). Of the various OBCFA in milk fat in the present study, only C16:0-iso tended to be weakly positively related (r = 0.22) with CH4 yield (g/kg of DMI). This weak positive relationship is in line with
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
. An increased level of fiber in the diet generally results in increased CH4 yield and is associated with increased concentrations of C14:0-iso in milk fat (
reported negative and positive correlations of milk C15:0-anteiso concentration with NDF and starch, respectively. These 2 feed components have counteracting effects on CH4 yield (
), which might explain why the concentration C15:0-anteiso in milk fat was not significantly related to CH4 yield in the present study. This result is in line with
, who suggested that the concentration of C15:0-anteiso was relevant only in prediction of butyrate proportion in the rumen that is associated with dietary sugars (
) and may indicate that CH4 yield and milk C15:0-anteiso concentration are not related.
No significant relationships with CH4 yield (g/kg of DMI) were found for concentrations of C15:0 and C17:0 in milk fat (Table 4). Two different hypotheses on the relationship between diet composition and these odd-chain milk FA concentrations have been investigated.
reported increased milk C15:0 concentration when feeding increased proportions of grass silage and dietary NDF levels, which was attributed to more abundant membrane lipids from rumen microbes. A high NDF content in the feed is associated with increased CH4 yield (
reported milk concentrations of C15:0 and the sum of C17:0 and cis-9 C17:1 to be positively related to propionate concentration in the rumen as these are synthesized from propionate de novo (
). Propionate production is negatively related to CH4 production, suggesting a negative relationship between milk odd-chain FA concentration and CH4 yield. In the present meta-analysis, odd-chain FA concentrations in milk fat were not significantly related to CH4 yield. It should be noted that the concentration of cis-9 C17:1, a desaturation product of C17:0 in the mammary gland, was not available in all experiments included in the present study. Grouping these 2 fractions together based on a broad database might shed new light on the relationship between CH4 yield and odd-chain FA concentration in milk, as also reported by
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
, relationships between concentrations of OBCFA in milk and CH4 production were generally rather minor or absent, and also less than expected based on theoretical relationships between rumen fermentation products and CH4 yield (e.g.,
Several long-chain unsaturated FA in milk originate from dietary oils and their biohydrogenation products formed in the rumen. Higher concentrations of these FA in cattle diets, which are known to reduce DM and NDF digestibility, are negatively associated with CH4 yield (e.g.,
The effect of dietary fats on methane emissions, and its other effects on digestibility, rumen fermentation and lactation performance in cattle: A meta-analysis.
). Concentration of cis-9,12 C18:2 in milk fat tended to be negatively related to CH4 yield (g/kg of DMI), r = −0.25, in line with expectations. In general, replacing grass silage with corn silage reduces CH4 yield, and corn silage is rich in cis-9,12 C18:2 and increases the proportion of this FA in milk fat (
stated that the potential to increase milk FA concentration cis-9,12,15 C18:3 was limited, with some positive effects upon feeding protected oilseed supplements in particular. Oilseeds in protected form may not decrease methanogenesis (
) and may not result in a significant negative correlation between CH4 yield and milk cis-9,12,15 C18:3 concentration. The absence of a significant relationship between milk cis-9,12,15 C18:3 concentration and CH4 yield is in line with
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
included a linseed treatment, which is high in cis-9,12,15 C18:3, and indicated a moderate negative correlation between concentration of cis-9,12,15 C18:3 in milk fat and CH4 production (g/d). Additional evaluation of the rumen microbial metabolism of cis-9,12,15 C18:3 thus seems to be necessary to better understand these contrasting findings.
Various trans- and cis-C18:1 milk FA concentrations were negatively related to CH4 yield (g/kg of DMI), as shown in Table 4. In general, trans-11 C18:1 is the major monounsaturated biohydrogenation intermediate of both cis-9,12,15 C18:3 and cis-9,12 C18:2 (e.g.,
). These PUFA inhibit CH4 production, and milk trans-11 C18:1 concentration is therefore expected to be negatively related to CH4 yield. With reduced rumen pH values, the predominant biohydrogenation pathway of cis-9,12 C18:2 may shift to trans-10 C18:1 (
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
). These observations explain the strong negative correlation (r = −0.56) obtained between milk trans-10+11 C18:1 concentration and CH4 yield. Like trans-10+11 C18:1, cis-11 C18:1 and cis-12 C18:1 and the fraction trans-6+7+8+9 C18:1 result from biohydrogenation of both cis-9,12 C18:2 and cis-9,12,15 C18:3 (e.g.,
), and this explains their significantly negative strong (r = −0.53; cis-11 C18:1), moderate (r = −0.40; cis-12 C18:1), and weak (r = −0.27; trans-6+7+8+9 C18:1) relationship of concentrations of these FA with CH4 yield.
Depending on their FA composition, different dietary lipids result in variable biohydrogenation products. Milk trans-16+cis-14 C18:1 concentration tended to be moderately negative related to CH4 yield (r = −0.35). Elevated levels of trans-16 C18:1 have been found in duodenal digesta (
) upon supplementing diets with feed ingredients rich in cis-9,12,15 C18:3. trans-16+cis-14 C18:1 also appeared as an in vitro rumen biohydrogenation product of cis-9,12,15 C18:3 (
), which might suggest that milk cis-14 C18:1 is derived from cis-9,12,15 C18:3. cis-13 C18:1 was increased in milk when increased contents of corn silage were fed (
). In this meta-analysis, the relationship tended to be weakly negative (r = −0.26) and may suggest that cis-13 C18:1 is derived from both cis-9,12 C18:2 and cis-9,12,15 C18:3. The positive response of milk cis-13 C18:1 concentration to dietary linseed oil supplementation (
Biohydrogenation, duodenal flow, and intestinal digestibility of trans fatty acids and conjugated linoleic acids in response to dietary forage: Concentrate ratio and linseed oil in dairy cows.
Milk cis-9 C18:1 concentration was not significantly related to CH4 yield in the present meta-analysis. The absence of a significant relationship is in line with the analysis of
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
The effect of dietary fats on methane emissions, and its other effects on digestibility, rumen fermentation and lactation performance in cattle: A meta-analysis.
Biohydrogenation, duodenal flow, and intestinal digestibility of trans fatty acids and conjugated linoleic acids in response to dietary forage: Concentrate ratio and linseed oil in dairy cows.
). Nonetheless, cis-9 C18:1 is converted into trans-C18:1 isomers in the rumen, possibly contributing to the negative correlations between concentrations of several milk trans-C18:1 isomers and CH4 yield as obtained in this study and discussed earlier (e.g.,
). Δ9-Desaturation of C18:0 in the mammary gland is another mechanism by which cis-9 C18:1 appears in milk. Furthermore, dietary C18:0, being a possible substrate for Δ9-desaturase, was not observed to inhibit CH4 production (
The effect of dietary fats on methane emissions, and its other effects on digestibility, rumen fermentation and lactation performance in cattle: A meta-analysis.
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
, who found a negative relationship, possibly coming from dietary linoleic and linolenic acid, which were biohydrogenated to C18:0 that was absorbed. Milk C20:0 is an elongation product by action of elongase enzymes on C18:0 from the diet or from body fat. The fact that milk C20:0 concentration and CH4 yield were not related in the present meta-analysis is in line with C18:0 concentration not being related overall to CH4 yield.
Evaluation of Heterogeneity
Milk C6:0, C8:0, C10:0, and C16:0 concentrations were not substantially heterogeneously correlated to CH4 yield per unit of feed (Table 4). These relatively low heterogeneities may indicate good precision of the estimated relationships and the simplicity of the mechanism determining the relationship. Next to de novo FA synthesis and feed, body fat is another source of milk C16:0 (
). Nonetheless, the de novo FA synthesis that regulates milk C16:0 concentration is positively associated with CH4 yield. Milk C16:0 from body fat and feed may not induce heterogeneity among the relationship. Unlike that of other saturated even-chain FA, milk C12:0 and C14:0 concentrations showed substantial heterogeneity among their true correlation with CH4 yield. Variation in diet composition, in particular when ingredients relatively rich in C12:0 and C14:0 are supplied to cattle (discussed in a previous section), may explain such heterogeneity. The sum of all de novo synthesized even-chain FA concentrations (C4:0 to C16:0, including C12:0 and C14:0) did not show any heterogeneity among its true correlation (Table 5).
Table 5Estimated true correlation coefficients with standard errors and P-values, P-values for heterogeneity test statistic (Q-value P), and heterogeneity as a fraction of total variability (I2) for the correlation between CH4 yield per unit of feed and milk FA concentration
The correlation of milk C14:0-iso, C15:0-anteiso, C15:0, C17:0 (Table 4), and of combined OBCFA concentrations; namely, C14:0-iso+C16:0-iso and C15:0+C17:0 (Table 5), were all substantially heterogeneous (I2 ≥ 61.1%). The heterogeneity observed for these milk OBCFA concentrations might reflect the variation in microbial species and activities in the rumen. All concentrations of individual C18:1 fractions appeared to be substantially heterogeneous (I2 > 50%), except for the trans-6+7+8+9 C18:1 fraction, which was totally homogeneous (Table 4). The combination of these 4 different FA may balance out all heterogeneity but does not indicate homogeneity for the correlation of the 4 individual FA concentrations. The correlation of the concentration of a combined fraction of milk cis-11+12+13 C18:1 with CH4 yield was also homogeneous, even though substantial heterogeneity was observed using the individual milk FA concentrations (Table 5). Substantial heterogeneity was apparent for correlations between CH4 yield and concentrations of all combined fractions of C18:1 isomers that contained trans-10+11 C18:1 and a tendency for heterogeneity when trans-16+cis-14 C18:1 was added to the fraction trans-6+7+8+9 C18:1 + cis-11+12+13 C18:1 (Table 5). Various concentrations of fractions with milk trans-C18:1 isomers and cis-9,12 C18:2 (I2 = 59.0%) did not, therefore, appear to be precise indicators of CH4 yield, whereas milk cis-11+12+13 C18:1 and trans-6+7+8+9 C18:1 concentration did. The C18:1 isomers in milk other than isomers containing a cis-9 double bond mainly originate from the rumen, but microorganisms and enzymes responsible for their production are not well characterized, and candidate bacterial species have yet to be cultivated (
). Isolation of bacterial species may help to better interpret heterogeneity and homogeneity observed.
Random-Effects Model Analysis Correlation per Unit of Milk
Positive relationships were obtained between CH4 yield (g/kg of FPCM) and the milk FA concentrations of C10:0, C12:0, C14:0, cis-9 C14:1, C15:0, and C16:0. The concentration of C14:0-iso tended to be positively related to CH4 yield. Negative relationships or a tendency for a negative relationship were obtained between CH4 yield and the milk FA concentration of C18:0, cis-9 C18:1, cis-11 C18:1, trans-10+11 C18:1, and cis-9,12 C18:2. These relationships and the absence of significant relationships between CH4 yield and milk FA concentrations of C17:0, cis-9,12,15 C18:3, and C20:0 are largely in agreement with the correlations obtained when CH4 yield was expressed per unit of feed. In contrast to CH4 yield per unit of feed, no positive relationships were obtained between CH4 yield per unit of milk and milk FA concentration of C6:0 and C8:0, and a negative relationship was observed for the concentration of C4:0. Moreover, no significant relationship was obtained between the milk FA concentrations of trans-6+7+8+9 C18:1, cis-12 C18:1, cis-13 C18:1, and trans-16+cis-14 C18:1 and CH4 yield per unit of milk. This might be due to the fact that various biohydrogenation intermediates associated with a reduction in CH4 yield per unit feed (Table 4) are associated with milk fat depression (e.g.,
), which negatively affects the amount of FPCM, thus yielding more CH4 yield per unit of milk. A reduced correlation strength for trans-10+11 C18:1 per unit of feed and per unit of milk (r = −0.56 vs. r = −0.16, respectively) is in line with this. The negative relationship between CH4 yield per unit of milk and the concentration of C18:0 in milk fat may be explained by the concentration of C18:0 being decreased during lactation (
), when milk yield decreases and next CH4 yield per unit of milk increases. Milk FA concentration of the fraction odd-chain C5:0 to C15:0 and of cis-9 C18:1 are decreased and increased, respectively, during negative energy balance in early lactation (
) when cows are producing at a high level. High milk production is associated with lower CH4 yield per unit of milk, which may explain why milk FA concentrations of C15:0 and cis-9 C18:1 were positively and negatively related to CH4 yield per unit of milk, as also shown by
. Concentrations of C15:0-anteiso and C16:0-iso in milk fat were not significantly related to CH4 yield per unit of milk, possibly because the proportion of branched-chain FA does not vary during lactation (
Most of the milk FA that were significantly correlated, or tended to be correlated, with CH4 yield per unit of milk showed relatively low heterogeneity, with the exception of C14:0-iso, C15:0, and cis-11 C18:1 (Table 4). In general, milk FA concentrations may therefore be regarded as precise indicators of CH4 yield per unit of milk.
Correlations between CH4 yield (g/kg of FPCM) and combined milk FA concentrations were significantly positive for C10:0+C12:0+C14:0, C10:0+C12:0+C14:0+C16:0, even-chain C4:0 to C16:0, and C16:0+C18:0, and were significantly negative or tended to be negative for the various combined fractions consisting of C18:1 isomers and for the sum of C18:2 and C18:3 (Table 6). The strongest positive and negative correlations were not stronger than those based on single FA fractions: 0.36 versus 0.36 for C10:0+C12:0+C14:0+C16:0 and C12:0, and −0.43 versus −0.45 for trans-10+11 C18:1+cis-9+11 C18:1 and cis-11 C18:1, respectively. The fractions of even-chain saturated FA and C18:1 isomers were not or not substantially heterogeneously correlated. Concentrations of combined fractions of milk OBCFA; namely, C15:0+C17:0 and C14:0-iso+C16:0-iso, were not related to CH4 yield (g/kg of FPCM), which is in line with the absence of such relationships of CH4 yield per unit feed. In general, most combined and single concentrations of milk FA showed less heterogeneity in the correlation with CH4 yield per unit of milk than with CH4 yield per unit of feed.
Table 6Estimated true correlation coefficients with standard errors and P-values, P-values for heterogeneity test statistic (Q-value P), and heterogeneity as a fraction of total variability (I2) for the correlation between CH4 yield per unit of milk and milk FA concentration
Mixed Model Regression Analysis for CH4 Yield per Unit of Feed
Mixed model fits to evaluate the potential to predict CH4 yield (g/kg of DMI) using selected milk FA concentrations (g/100 g of FA) resulted in R2 values of 0.15 and 0.17 for the concentration of milk C16:0 and the fraction C6:0+C8:0+10:0+C16:0, respectively (Figure 1). Thus, the concentration of the fraction C6:0+C8:0+10:0+C16:0 did not show a substantially stronger positive relationship with CH4 yield than the concentration of C16:0 alone. The R2 values were 0.20, 0.41, 0.31, and 0.41 for the prediction with milk FA concentrations of trans-6+7+8+9 C18:1, cis-11 C18:1, cis-11+12+13 C18:1, and C18:1 (trans-6+7+8+9+10+11+16 C18:1+cis-11+12+13+14 C18:1), respectively. Concentrations of C18:1 fractions that were grouped together did not result in a substantially stronger negative relationship with CH4 yield than concentrations of single C18:1 fractions.
Figure 1Relationships between CH4 yield (g/kg of DMI) and selected milk FA concentrations. Milk total C18:1 comprises trans-6+7+8+9 C18:1, trans-10+11 C18:1, cis-11 C18:1, cis-12 C18:1, cis-13 C18:1, and trans-16+cis-14 C18:1 for experiments 1 to 4 and 6 to 8. The different symbols identify the 8 individual experiments described inTable 1 (see Figure 2 for explanation).
), and P ≤ 0.003 for all parameters. The model acceptably reproduced the CH4 yield (Figure 2), as the slope and intercept of the regression between observed and predicted values did not differ from 1 and 0, respectively (results not shown). No clear patterns appear in the observed versus predicted values of CH4 yield and residual versus predicted values of CH4 yield (Figure 3), which does not indicate substantial bias or heteroscedasticity. The obtained R2 was larger than for concentrations of single milk FA fractions but lower than the value (0.73) that
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
are also included in the present analysis. The larger number of studies included in our meta-analysis, including a wider variety of diet compositions, may have resulted in the lower R2 value of Equation [3]. A similar issue may hold when comparing the multiple regression analysis of
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
expressed CH4 production in grams per day. Furthermore, the substantial heterogeneity of the correlation of CH4 yield and milk FA concentrations of several C18:1 isomers found in the present analysis may limit a precise prediction of CH4 yield (g/kg of DMI) using milk FA profile.
Figure 2Observed versus predicted CH4 yield per unit of feed from Equation [3], including experiment as a discrete class variable with experiment effect not shown. The different symbols identify the 8 individual experiments. The line of unit slope represents the line of equivalence.
Figure 3Residuals (predicted minus observed) and predicted CH4 yield per unit of feed from Equation [3], including experiment as a discrete class variable with experiment effect not shown. The different symbols identify the 8 individual experiments.
Equation [3] of the present meta-analysis has various milk FA concentrations in common with previously reported equations. Milk C16:0-iso concentration did appear in the best equation of
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
included all trans-C18:1 FA concentrations in their second-best equation based on milk FA concentrations and DMI. Milk cis-9,12 C18:2 concentrations was also present in the best equation of
. However, milk cis-9,12 C18:2 concentration was positively related to CH4 in that equation, whereas it was negatively related in Equation [3] of the current meta-analysis.
Different FA concentrations available for selection in different studies may hamper the development of a universally valid CH4 prediction equation based on milk FA concentrations. A stepwise selection was performed for concentrations of all milk FA that were available in 7 experiments to extend the number of available FA with C14:0-iso, trans-6+7+8+9 C18:1, cis-11 C18:1, and trans-16+cis-14 C18:1. Methane yield was best predicted by the concentration of C16:0-iso (positively related) and the concentrations of cis-11 C18:1 and trans-16+cis-14 C18:1 (both negatively related), with P ≤ 0.02 for all regressors and R2 = 0.55. The concentration of cis-11 C18:1 was also present in the equation of
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
. However, the equation based on 7 experiments with R2 = 0.55 is not preferred over Equation [3] because it is based on one less experiment and the coefficient of determination is just marginally higher. Total milk fat and milk protein content were never selected and might illustrate the value of concentrations of single milk FA to predict CH4 yield.
In addition to gas chromatography, Fourier transform infrared spectroscopy (FTIR) is often applied to quantify FA concentrations in milk. To achieve a reasonable accuracy of milk FA concentrations with FTIR, FA should have an average concentration of ≥2.45 g/100 g of FA (
). Restricting the selection of milk FA with this threshold concentration would have resulted in a best equation based on positive relationships of concentrations of C14:0, C16:0, and C18:0 and R2 = 0.29. More recently,
found concentrations of milk C4:0, C6:0, C8:0, C10:0, C12:0, C14:0, C16:0, C18:0, all trans-C18:1, cis-9 C18:1, all cis-C18:1, and some groups of FA in milk to be sufficiently accurately determined by FTIR to be used in milk payment systems. Including these FA concentrations mentioned by
in a selection procedure resulted in a best equation with a negative relationship of all trans-C18:1 concentration, where it is noted that the trans-C18:1 fraction consisted of all trans-C18:1 available in studies 1 to 4 and 6 to 8. The R2 value was 0.43 for this equation, lower than that obtained for Equation [3]. Several milk FA with lower concentrations that appear in various equations published previously are not available when milk FA profile is determined using FTIR. Furthermore, concentrations of combined FA fractions, several of which are accurately determined with FTIR, did not substantially increase the potential for predicting CH4 yield (Figure 1). Compared with gas chromatography, the current performance of FTIR therefore limits the potential for predicting CH4 yield based on milk FA profile.
Mixed Model Regression Analysis for CH4 Yield per Unit of Milk
The best multiple regression to predict CH4 yield using concentrations of milk FA was as follows:
[4]
where R2 = 0.47 after correction for experiment effect (
) and P ≤ 0.006 for all parameters. The model acceptably reproduced the CH4 yield (Figure 4), as the slope and intercept of the regression between observed and predicted values did not differ from 1 and 0, respectively (results not shown). No clear patterns appear in the observed versus predicted values of CH4 yield and residual versus predicted values of CH4 yield (Figure 5), which does not indicate substantial bias or heteroscedasticity. The obtained R2 value was somewhat smaller than that for the prediction of CH4 yield per unit of feed. Equation [4] also contains milk C16:0-iso and trans-10+11 C18:1 concentration (like Equation [3]) to predict CH4 yield per unit of feed. Therefore, this may be regarded as further evidence for concentrations of milk C16:0-iso and trans-10+11 C18:1 being appropriate predictors of CH4 yield.
Figure 4Observed versus predicted CH4 yield per unit of milk (fat- and protein-corrected milk yield, FPCM) from Equation [4], including experiment as a discrete class variable with experiment effect not shown. The different symbols identify the 8 individual experiments. The line of unit slope represents the line of equivalence.
Figure 5Residuals (predicted minus observed) and predicted CH4 yield per unit of milk (fat- and protein-corrected milk yield, FPCM) from Equation [4], including experiment as a discrete class variable with experiment effect not shown. The different symbols identify the 8 individual experiments.
Restricting the selection of milk FA with a threshold concentration of 2.45 g/100 g of FA for reasonable accuracy using FTIR would have resulted in a best prediction equation based on negative relationships of concentrations of C4:0 and cis-9 C18:1, and R2 = 0.28. Including the FA concentrations mentioned by
in the selection procedure resulted in a best equation based on negative relationships of concentrations of C4:0, cis-9 C18:1, and total trans-C18:1, and R2 = 0.36. It is noted that all FA available in studies 1 to 4 and 6 to 8 were included in the selection procedure because trans-10+11 C18:1 is the only trans-C18:1 fraction available in all 8 studies. Similar to prediction of CH4 yield per unit feed, these results indicate that current performance of FTIR limits the potential for predicting CH4 yield per unit of milk based on milk FA profile, compared with gas chromatography.
The present meta-analysis showed that milk FA profile has moderate potential to predict CH4 yield. Further improvement in ability of milk FA-based models to predict CH4 yield may be achieved by distinguishing diets with or without lipid supplements and by including diet composition characteristics. The present data set was too small to allow separate analyses of lipid-supplemented and non-lipid-supplemented diets, and more data may be required to analyze possible effects of lipid supplements on the relationship. On diet composition,
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
already indicated that the combination of milk FA profile and diet characteristics may improve prediction performance. For practical application, this requires knowledge of diet composition (including forage to concentrate ratio and chemical composition of feed consumed), which may not always be available.
Conclusions
Various FA concentrations in milk fat appeared to be weakly, moderately, or strongly related to CH4 yield per unit of feed. Milk C6:0, C8:0, C10:0, C16:0, and C16:0-iso concentrations showed positive relationships, whereas milk cis-9,12 C18:2, trans-10+11 C18:1, cis-11 C18:1, cis-12 C18:1, cis-13 C18:1, and trans-16+cis-14 C18:1 concentrations showed weak to strong negative relationships with CH4 yield. Milk OBCFA concentrations C14:0-iso, C15:0-anteiso, C15:0, and C17:0, as well as milk C4:0, C12:0, C14:0, cis-9 C14:1, C18:0, cis-9 C18:1, cis-9,12,15 C18:3, and C20:0 concentrations were not significantly related to CH4 yield per unit of feed. When expressing CH4 yield per unit of milk, relationships of milk C10:0, C15:0-anteiso, C16:0, C17:0, trans-10+11 C18:1, cis-11 C18:1, cis-9,12 C18:2, cis-9,12,15 C18:3, and C20:0 concentrations were in line with relationships expressing CH4 yield per unit of feed. Concentrations of C18:1 isomers in milk fat showed more heterogeneity among the true correlation with CH4 yield per unit of feed than concentrations of the saturated even-chain FA C6:0, C8:0, C10:0, and C16:0. Mixed model multiple regression resulted in various milk FA included in optimal equations to predict CH4 yield per unit of feed and per unit of milk, with C16:0-iso and trans-10+11 C18:1 concentrations appearing in both equations. These regressions indicated moderate potential for using milk FA profile to predict CH4 yield per unit of feed and per unit of milk.
Acknowledgments
The financial support of UK Defra projects LS3656 and AC0209 and Marks and Spencer (London, UK) is gratefully acknowledged for studies at the University of Reading. Authors acknowledge financial support of SenterNovem, an agency of the Dutch Ministry of Economic Affairs, to obtain part of the data on DMI, milk production, and CH4 production.
Tackling Climate Change Through Livestock—A Global Assessment of Emissions and Mitigation Opportunities. Food and Agriculture Organization of the United Nations (FAO),
Rome, Italy2013
Biohydrogenation, duodenal flow, and intestinal digestibility of trans fatty acids and conjugated linoleic acids in response to dietary forage: Concentrate ratio and linseed oil in dairy cows.
Prediction of enteric methane output from milk fatty acid concentrations and rumen fermentation parameters in dairy cows fed sunflower, flax, or canola seeds.
The effect of dietary fats on methane emissions, and its other effects on digestibility, rumen fermentation and lactation performance in cattle: A meta-analysis.
Reynolds, C. K., D. J. Humphries, P. Kirton, L. A. Crompton, J. A. N. Mills, and D. I. Givens. 2010. Methane production by lactating dairy cows fed diets containing allicin, glycerol or naked oats. Page 120 in Proc. 4th Intl. Greenhouse Gases Anim. Agric. Conf., Banff, Canada.
van Gastelen, S., E. C. Antunes Fernandes, K. A. Hettinga, G. Klop, S. J. J. Alferink, and J. Dijkstra. 2014. Replacing grass silage with maize silage affects rumen fermentation characteristics and enteric methane production in dairy cattle. In Proc. 39th Anim. Nutr. Res. Forum, Utrecht, the Netherlands. Universiteit Utrecht, Utrecht, the Netherlands.
Dietary inclusion of diallyl disulfide, yucca powder, calcium fumarate, an extruded linseed product, or medium-chain fatty acids does not affect methane production in lactating dairy cows.
00617-1&id=gr1.jpg){kind=link}
00617-1&id=gr2.jpg){kind=link}
00617-1&id=gr3.jpg){kind=link}
00617-1&id=gr4.jpg){kind=link}
00617-1&id=gr5.jpg){kind=link}