Predicting methane emission of dairy cows using milk composition

Enteric methane (CH4) is produced as a result of microbial fermentation of feed components in the gastrointestinal tract of ruminant livestock. Methane has no nutritional value for the animal and is predominately released into the environment through eructation and breath. Therefore, CH4 not only represents a greenhouse gas contributing to global warming, but also an energy loss, making enteric CH4 production one of the main targets of greenhouse gas mitigation practices for the dairy industry. Obviously, reduction of CH4 emission could be achieved by simply reducing livestock numbers. However, the global demand for dairy products has been growing rapidly and is expected to further grow in the future. Therefore, it is critical to minimize environmental impact to produce high-quality dairy products. The overall aim of this PhD research was, therefore, to develop a proxy for CH4 emission that can be measured in milk of dairy cows. There are currently a number of potentially effective dietary CH4 mitigation practices available for the livestock sector. The results of Chapter 3 show that replacing fiber-rich grass silage with starch-rich corn silage in a common forage-based diet for dairy cattle offers an effective strategy to decrease enteric CH4 production without negatively affecting dairy cow performance, although a critical level of starch in the diet seems to be needed. Little is known whether host genetics may influence the CH4 emission response to changes in diet. Therefore, the interaction between host DGAT1 K232A polymorphism with dietary linseed oil supplementation was evaluated in Chapter 7. The results of Chapter 7 indicate that DGAT1 K232A polymorphism is associated with changes in milk composition, milk N efficiency, and diet metabolizability, but does not affect digestibility and enteric CH4 emission, whereas linseed oil reduces CH4 emission independent of the DGAT1 K232A polymorphism. Accurate and repeatable measurements of CH4 emission from individual dairy cows are required to assess the efficacy of possible mitigation strategies. There are several techniques to estimate or measure enteric CH4 production of dairy cows, including climate respiration chambers, but none of these techniques are suitable for large scale precise and accurate measurements. Therefore, the potential of various metabolites in milk, including milk fatty acids (MFA), as a proxy (i.e., indicators or animal traits that are correlated with enteric CH4 production) for CH4 emission of dairy cows gained interest. Until recently, gas chromatography was the principal method used to determine the MFA profile, but this technique is unsuitable for routine analysis. This has led to the application of Fourier-transform infrared spectroscopy (FTIR) for determination of the MFA profile. Chapter 2 provides an overview of the recent research that relates MFA with CH4 emission, and discusses the opportunities and limitations of using FTIR to estimate, indirectly via MFA or directly, CH4 emission of dairy cattle. The recent literature on the relationship between MFA and CH4 emission gives inconsistent results. Where some studies found a clear and strong relation, other studies consider MFA to be unreliable predictors for CH4 emitted by dairy cows. Even the studies that do find a clear relation between MFA and CH4 emissions do not describe similar prediction models using the same MFA. These discrepancies can be the result of many factors, including dietary composition and lactation stage. Additionally, literature showed that the major advantages of using FTIR to predict CH4 emission include its simplicity and potential practical application on a large scale. Disadvantages include the inability to predict important MFA for the prediction of CH4 emission, and the moderate power of FTIR to directly predict CH4 emission. The latter was also demonstrated in Chapter 9, in which the CH4 prediction potential of MFA was compared with that of FTIR using data from 9 experiments (n = 218 individual cow observations) covering a broad range of roughage-based diets. The results indicate that MFA have a greater potential than FTIR spectra to estimate CH4 emissions, and that both techniques have potential to predict CH4 emission of dairy cows, but also limited current applicability in practice. Much focus has been placed on the relationship between MFA and CH4 emission, but milk also contains other metabolites, such as volatile and non-volatile metabolites. Currently, milk volatile metabolites have been used for tracing animal feeding systems and milk non-volatile metabolites were shown to be related to the health status of cows. In Chapter 4, the relationship between CH4 emission and both volatile and non-volatile metabolites was investigated, using data and milk samples obtained in the study described in Chapter 3. In general, the non-volatile metabolites were more closely related to CH4 emissions than the volatile metabolites. More specifically, the results indicate that CH4 intensity (g/kg fat- and protein-corrected milk; FPCM) may be related to lactose synthesis and energy metabolism in the mammary gland, as reflected by the milk non-volatile metabolites uridine diphosphate-hexose B and citrate. Methane yield (g/kg dry matter intake) on the other hand, may be related to glucogenic nutrient supply, as reflected by the milk non-volatile acetone. Based on the metabolic interpretations of these relationships, it was hypothesized that the addition of both volatile and non-volatile metabolites in a prediction model with only MFA would enhance its predictive power and, thus, leads to a better proxy in milk for enteric CH4 production of dairy cows. This was investigated in Chapter 5, again using data and milk samples described in Chapter 3. The results indicate that MFA alone have moderate to good potential to estimate CH4 emission. Furthermore, including volatile metabolites (CH4 intensity only) and non-volatile metabolites increases the CH4 emission prediction potential. The work presented in Chapters 3, 4 and 5, was based upon a small range of diets (i.e., four roughage-based diets in which grass silage was replaced partly or fully by corn silage) of one experiment. Therefore, in Chapter 6, the relationship between CH4 emission and the milk metabolome in dairy cattle was further quantified. Data (n = 123 individual cow observations) were used encompassing a large of roughage-based diets, with different qualities and proportions of grass, grass silage and corn silage. The results show that changes in individual milk metabolite concentrations can be related to the ruminal CH4 production pathways. These relationships are most likely the result from changes in dietary composition that affect not only enteric CH4 production, but also the profile of volatile and non-volatile metabolites in milk. Overall, the results indicate that both volatile and non-volatile metabolites in milk might provide useful information and increase our understanding of CH4 emission of dairy cows. However, the development of CH4 prediction models revealed that both volatile and non-volatile metabolites in milk hold little potential to predict CH4 emissions despite the significant relationships found between individual non-volatile metabolites and CH4 emissions. Additionally, combining MFA with milk volatile metabolites and non-volatile metabolites does not improve the CH4 prediction potential relative to MFA alone. Hence, it is concluded that it is not worthwhile to determine the volatile and non-volatile metabolites in milk in order to estimate CH4 emission of dairy cows. Overall, in comparison with FTIR, volatile and non-volatile metabolites, the MFA are the most accurate and precise proxy in milk for CH4 emission of dairy cows. However, most of MFA-based models to predict CH4 emission tend to be accurate only for the production system and the environmental conditions under which they were developed. In Chapter 8 it was demonstrated that previously developed MFA-based prediction equations did not predict CH4 emission satisfactory of dairy cows with different DGAT1 genotypes or fed diets with or without linseed oil. Therefore, the greatest shortcoming today of MFA-based CH4 prediction models is their lack of robustness. Additionally, MFA have restricted practical application, meaning that most MFA retained in the current CH4 prediction models cannot be determined routinely because of the use of gas chromatography. The MFA that can be determined with the use of infrared spectroscopy are however no promising predictors for CH4 emission. Furthermore, MFA have only a moderate CH4 prediction potential. This together suggests that it might not be the best option to focus in the future on MFA alone as a proxy for CH4 emission of dairy cows. The FTIR technique has a low to moderate CH4 prediction potential. However, FTIR has a great potential for practical high throughput application, facilitating repeated measurements of the same cow potentially reducing random noise. Results of this thesis also demonstrated that FTIR spectra do not have the potential to detect differences in CH4 emission between diets which are, in terms of forage level and quality, commonly fed in practice. Moreover, the robustness of FTIR spectra is currently unknown. Hence, it remains to be investigated whether FTIR spectra can predict CH4 emissions from dairy cows housed under different conditions from those under which the FTIR-based prediction equations were developed. It is therefore concluded that the accuracy and precision to predict CH4 emission using FTIR needs to increase, and the capacity of FTIR to evaluate the differences in CH4 emission between dairy cows and different types of diets needs to improve, in order to actually be a valuable proxy for CH4 emission of dairy cows.