How Machine Learning Models can help in the Selection of High Feed Efficiency and Low Methane Emitting Dairy Cattle.

Stephen Ross, Huiru Zheng (Editor), Haiying Wang (Editor), Tianhai Yan (Editor), Masoud Shirali

Research output: Contribution to conferencePoster

2 Downloads (Pure)

Abstract

Methane (CH4) makes up 44% of all GHGs emitted by the livestock sector, 39% of which is emitted through the process of Enteric Fermentation (EF) in ruminants. Cattle are the primary source of CH4 emissions within the livestock sector, contributing 77% of all CH4 emissions through EF. Whilst dietary strategies have proven to be effective in lowering the CH4 emissions of cattle, they face limitations in that they are only effective for as long as they are being implemented, with the risk of the cattle microbiome adjusting to the additives over time. Therefore, genetic selection through breeding has been suggested as a promising alternative, due to the proven heritability of CH4 emissions and Feed Efficiency traits in cattle, as well as the permanence of the effects. To identify cattle low in CH4 emissions and high in Feed Efficiency for genetic selection through breeding, Machine Learning modelling has been used to great effect.

Many studies have been conducted comparing the effectiveness of traditional linear models against advanced non-linear models in the prediction of Feeding Efficiency and CH4 emissions. Advanced models such as Random Forests and Neural Networks have shown to outperform their traditional counterparts, as they are able to better handle the complex relationships between cattle traits and target phenotypes. In partnership with AFBI, it is hoped that this PhD Project will result in the development of a statistical analysis, as well as novel multi-trait machine learning models, which can uncover new insights into the Feeding Efficiency complex of dairy cattle and accurately predict their Green House Gas contributions in Northern Ireland. The project will be developed on top of the state of the art N.I.FAB database provided by AFBI. It is also hoped that proxy traits for Feeding Efficiency and CH4 emissions can be identified so that the analysis and models developed can be applied on a commercial scale for the largest possible impact on livestock Green House Gas emissions based on genetic selection through breeding.
Original languageEnglish
Publication statusUnpublished - 14 Nov 2022
EventQUB-AFBI PhD Student Conference 2022 - Riddel Hall, Belfast, Northern Ireland
Duration: 14 Nov 2022 → …

Conference

ConferenceQUB-AFBI PhD Student Conference 2022
Country/TerritoryNorthern Ireland
CityBelfast
Period14/11/22 → …

Keywords

  • Methane
  • Feeding Efficiency
  • Machine Learning
  • Selective Breeding
  • Green House Gases
  • Proxies
  • Diary Cattle

Fingerprint

Dive into the research topics of 'How Machine Learning Models can help in the Selection of High Feed Efficiency and Low Methane Emitting Dairy Cattle.'. Together they form a unique fingerprint.

Cite this