Genetic control of methane emission, feed efficiency and metagenomics in dairy cattle

The dairy industry faces the challenges of increasing production, remaining economically viable whilst simultaneously minimising impacts on the environment. The cost of feed is the highest variable cost of milk production, thus, improved feed efficiency is a strong wish. However, CH4 is a potent greenhouse gas with an energy value estimated as 2 -12% of the gross feed energy intake and thus represents a loss. There is, therefore, a need to identify the phenotypic and genetic relationships between efficiency of feed utilisation and CH4 production to ensure optimal breeding methods of increasing profitability and limiting environmental impact of dairy production. Feed is degraded and CH4 is produced by rumen microbes and not by the cow. The mechanisms which influence the composition of the rumen microbial community and how they, in turn, influence the feed efficiency and CH4 production of the host, are not well understood. Among the possible strategies, selective breeding has the benefit over others by being cumulative and persistent over generations. Genetic improvement through selection requires that phenotypes are recorded on large numbers of animals. Moreover, phenotypes must show variation, a portion of which must be genetic, and must have economic or societal value. Understanding the genetic co-variation behind and between these measures is crucial to simultaneous breeding for a more profitable and climate friendly dairy industry. However, the measurement of CH4 emissions, feed efficiency and the rumen microbiome under commercial conditions on a large scale is not a trivial task. The aim of this PhD project was to develop and integrate phenotyping measures for CH4 emission, feed efficiency and the rumen microbiome and to investigate their genetic potential for selective breeding. Firstly, in Chapter 2, improvements where made to the sniffer method of CH4 breath concentration recording in dairy cattle during automatic milking. An algorithm was developed to efficiently detect and correct for variable and random drift in time series between instruments and to detect when the cow’s head is out of the feed bin. Using linear mixed model methodology, repeated measures per cow were used to improve precision and control sources of inaccuracy such as sensor drift, background gas concentrations and diurnal variation, that were subsequently removed. Resultantly, highly repeatable phenotypes where obtained which demonstrated adequate agreement for the interchangeable use of two instruments. In Chapter 3, the ranking of cows under commercial conditions using the sniffer method was compared with the “gold standard” respiration chambers. Individual level correlations estimated as proxies for genetic correlations revealed a high correlation between sniffer-predicted CH4 production and CH4 production in the RC. These findings offer a proof of concept that sniffer CH4 phenotypes recorded over a week of lactation show substantial promise as large scale indicator traits for CH4 production using RC. In Chapter 4, genetic parameters were estimated between feed intake, milk production and CH4 breath concentration from sniffers over the course of the first lactation in Holstein cows in Denmark and The Netherlands. Through combining data between countries, genetic residual feed intake and breath gas concentrations were found to be significantly heritable, demonstrating that genetic improvement of feed efficiency and CH4 breath gas concentration is feasible in dairy cattle. The estimated genetic correlations from the largest dataset indicated that improved feed efficiency will also result in decreased gas emissions. Furthermore, including the breath gas concentrations in a multitrait genetic evaluation increased the accuracy of bull breeding values for gRFI, demonstrating an indirect economic value of CH4 and CO2 breath concentration phenotypes. In Chapter 5, we estimated the relative abundance of rumen bacteria and archaea and found a portion of these to be heritable in dairy cattle. The results demonstrate that host additive genetics has an influence on the abundance of some rumen bacteria and archaea. We detected significant associations between certain bacterial genera and differences in CH4 production of the host cow, further contributing to knowledge of the underlying biological mechanisms driving CH4 production of the host. We further extended quantitative genetic methods to estimate rumen microbial kinships between cows in place of additive genetic relationships. This enabled the quantification of variation in host CH4 production explained by the rumen microbial composition, expressed in the new term ‘microbiability’, as the relative proportion of host variation explained by associated microbes. Crucially the microbiability and the heritability of dairy cattle CH4 production were largely independent. Thus, selective breeding for reduced CH4 production can be extended by methods perturbing the rumen microbiota towards reduced CH4 production. In Chapter 6 (the general discussion), the value of method comparisons for phenotype development by comparatively quantifying sources of error between cheaper alternative methods and intensive gold standard methods was discussed. The primary constraint to breeding for improved feed efficiency and CH4 production remains the recording of feed intake on a large scale under commercial conditions and recording of “true” CH4 production. It was proposed that the accuracy of bull breeding values for both feed efficiency and CH4 production can be increased through the use of sniffer phenotypes in robot milking herds, using individual level correlations but a genetic correlation between sniffer phenotypes and RC CH4 production are still needed. The records required for estimating genetic correlations with meaningful standard errors can only be achieved through substantial financial investments, development of cheaper alternative methods of phenotype recording or international collaborations. Further to the general discussion, a portion of host phenotypic variation in CH4 production was found to be associated with the rumen bacterial and archaeal composition. However, research is needed to determine if microbial associations are causative and methods to direct desired changes in the rumen microbial composition are still needed to unlock the potential of this under-exploited resource. The methods developed for quantifying the microbial contribution to host phenotypic variation will be of value to inform research into complex microbial-associated phenotypes, such as diseases and digestion in dairy cattle, other livestock species and humans. This thesis therefore contributes to the understanding of the genetic variation in feed efficiency, methane emissions and rumen metagenome of dairy cows