What if we could breed sheep that emit less methane and are more feed efficient?
That’s the core idea behind SustainSheep. Researchers from Ireland, the United Kingdom, New Zealand, France, Uruguay and Norway are collaborating to explore how animal breeding can reduce the carbon footprint of sheep production.
A key technology is Portable Accumulation Chambers (PAC)—mobile chambers used to quantify enteric methane emissions from individual sheep under field conditions. This is methane mainly produced in the rumen during digestion. PAC measurements are used to investigate the genetics of enteric methane emissions and feed efficiency (how efficiently feed is converted into products such as meat), with the aim of developing new breeding goals that contribute to achieving climate targets.
In Norway, NMBU and the Norwegian Sheep and Goat Breeders’ Association (NSG) are partners in the project. The Norwegian contribution focuses on identifying proxies of ewe feed intake, estimating the impact of breeding for lower methane emissions and/or improved feed efficiency, and assessing how genetic change in these traits affects greenhouse gas (GHG) emissions per kilogram of product (emission intensity)—specifically sheep meat and wool.
Feed intake is costly and labour intensive to obtain in forage-based systems. The potential to predict dry matter intake (DMI) in ewes using PAC measurements was investigated using data from the GrassToGas project (2019–2023). Ewes from two breeds—Norwegian White Sheep (NWS; large modern breed) and Old Norwegian Spæl (ONS; small extensive breed)—were monitored across two feeding trials using silage and fresh-cut grass. In both trials, DMI and PAC measurements were available.
Methane (CH4), carbon dioxide (CO2), and oxygen (O2) emissions, along with body weight (BW) and eating time (ET), were used as predictors in machine learning models (data-driven method where patterns are found from the data to make predictions without being explicitly programmed). The best model, including PAC traits, BW, and ET, achieved an R² of 0.77 and a mean absolute percentage error (MAPE) of 13.8% on test data. CH4 and CO2 emissions showed strong individual-level correlations with DMI (r? = 0.76–0.81), suggesting their utility as proxies for feed intake. Interestingly, simpler models using only PAC traits were nearly as effective for ranking animals, indicating their potential for scalable intake estimation in breeding programs and pasture-based systems.
To assess the long-term impact of breeding for reduced enteric methane emissions and improved feed efficiency, a simulation model—a digital twin—has been developed. This tool mimics the Norwegian breeding program for NWS, allowing testing of various breeding strategies in a virtual environment. The model is now in its final stages of development and will be used to evaluate the effect of different breeding strategies, like the effect of different weightings of enteric methane emissions in the breeding goal, as well as the impact of varying numbers of ewes phenotyped and genotyped in the ram circles annually.
The digital twin simulates genetic and phenotypic changes over multiple generations. It begins with a base population of 5,000 ewes randomly mated over 5,000 generations to establish genetic diversity. From this base population, a founder population is sampled to initiate breeding cycles. All individuals are assigned a genetic profile, including quantitative trait loci (QTLs), which are genes or chromosomal regions that affects a specific trait.
All traits are defined with parameters such as heritability, sex-specific expression, and age of measurement. Traits modelled include direct and maternal effects for growth (e.g., birth, spring, autumn, and slaughter weights), carcass traits (slaughter class and fat group), wool traits (weight and class), reproductive traits (lamb number and teat size), and enteric methane emissions. Since some traits are genetically correlated, their relationships are accounted in the simulations to ensure that the results are realistic and useful in practice.
Each breeding cycle involves selecting rams in different categories (test ram, progeny tested rams and elite rams) based on their total merit index and mating with a defined number of ewes. The offsprings get updated genotypes and phenotypes and breeding values are estimated for all traits. The simulation also includes culling and database updates, ensuring that only selected offspring are retained for future breeding.
Based on simulated changes in phenotypic values from the digital twin—such as live weight, number of weaned lambs, and enteric methane emissions—the whole-farm model HolosNorSheep will be used to calculate emission intensities. This will estimate the impact of breeding decisions on GHG emissions per kilogram of meat and wool produced.
This is how SustainSheep will help promote sustainable and climate-friendly sheep production, in Norway and internationally.
Breeding can be a cost-effective way of reducing enteric methane (CH4) emissions from sheep, by direct selection for CH4 emissions or indirectly through improved productivity. Currently there is under-adoption of these measures as benefits are not captured by the market. Sustain Sheep builds on the successful Joint Call ERA-Net project Grass To Gas and will create infrastructure for incorporation of low environmental impact into national breeding schemes that dovetails into the IPCC inventory. Sustain Sheep is unique as all partner countries (Ireland, United Kingdom, New Zealand, France, Norway and Uruguay) have invested in the same CH4 measurement technology (Portable Accumulation Chambers).
Sustain Sheep will 1) review the current scientific status of the potential to breed sheep for reduced CH4 emissions (WP1), 2) investigate genetics of CH4 emission and feed efficiency and model the mitigation potential from breeding and give recommendations for new breeding goals (WP2), 3) forecast uptake rates, cost and abatement of breeding for reduced CH4 emissions (WP3) and 4) determine the best mechanism for dissemination and implementation of project result to maximize stakeholder involvement (WP4).
In Norway, genetic and genomic analyses of data from the Norwegian Sheep Recording System and the genomic reference population recorded for CH4 emission will be done by NSG (WP2.1). Machine learning to investigate relationships between GHG emission, ewe live weight and feed intake will be applied and validated by NMBU (WP2.2). The expected response to selection for breeding goals including enteric CH4 and/or feed efficiency will be predicted by NMBU using a digital twin of the Norwegian sheep breeding scheme and data from WP2.1, and the effect on emission intensities (kg CO2-equivalents per kg product) will be investigated using a whole-farm model. The impact of alternative breeding goals on GHG emissions at national level will be quantified using IPCC methodology (WP2.3).
