Authors: Carolina Garcia-Baccino (INRAE), Andres Legarra (INRAE)

Under real productive rearing conditions, challenge events in sheep and goats are sometimes unrecorded and from unknown source. However, frequent recording allows to observe anomalous patterns in the flock showing up as additional variability on recorded traits.

To detect an unrecorded challenge, you need a series of daily recordings (milk yield, feed intake, growth) spanning at least one season. Then, challenges manifest as extra variation among individuals. To avoid the scale effect of increasing trait (growth, milk yield), analyse the natural logarithm of the daily coefficient of variation, log(CV). The daily log(CV) is analyzed using a mixture model with two components, one is “normal” variation and the other one is “extra” variation. Analysis may be done in R, for instance with package normalmixEM. On output there is an indicator variable from 0 to 1 tells the probability of a day being a “challenge”. These values can be used as indicators of “challenge” days or can be directly used in norm reaction models.

Author: Andres Legarra (INRAE), Fernando Macedo (INRAE)

Accuracies of genomic prediction are important to predict future genetic progress and to choose the adoption of different selection schemes, e.g. pedigree-based versus genomic ones. A method frequently used requires careful analysis of data involving cross-validation (e.g. hiding records and trying to predict them). Such methods for sheep and goats are very cumbersome and require access to raw data from genetic evaluation. We are essentially interested in accuracy at birth, when selection decisions are more important. The “LR” method (from Linear Regression) estimates these accuracies from sets of consecutive proofs. Pick a set of “focal” animals with same age and information - for instance, young rams or young females. The set has to consist on at least 50 animals for the results to be reliable. We compare genetic proofs (genetic evaluations) of these “focal” animals, at birth (“old” proofs), versus “more recent” proofs of the same animals, one (or more) years later, when they have more information (progeny records or own phenotype). The correlation between “old” and “most recent” proofs across the set of animals (r(w,p)) is the ratio of initial and final accuracy such that the increase in accuracy from initial to final is (1/r(w,p)) - 1. For instance if r(w,p) is 0.8 it means that the accuracy increases by relative 25% from birth to the most recent evaluation. Thus high values of r(w,p) imply high accuracies at birth, and very low values imply that proofs at birth are little accurate.

Authors: Donagh Berry (TEAGASC) and Jean-Michel Astruc (IDELE) In sheep and goats, compared to cattle, smaller within-country populations in selection and higher relative cost of genotyping and performance recording are among the main hindrances to the development of genomic selection. An international cooperation leading to across-country genetic evaluations and conception of an optimised and affordable genomic platform might generate great benefits for stakeholders in terms of genetic progress on resilience and efficiency traits.The essential tools required for establishing an international evaluation and identifying the best panel of genomic markers across breeds have been put in place and are expected to be used widely beyond the project.A template of agreement for sharing and pooling data was proposed and signed by 10 organisations. As performance recording and models of genetic evaluation for similar traits are somehow different across country, a survey on the situation in each partner country was undertaken. Harmonised formats of files for exchanging pedigree, phenotypes and genotypes were adopted by the 10 involved organisations. A codification of breeds was set up. Allele frequency information acquired from 14 meat and dairy sheep breeds were used to detect the more informative SNP markers across populations. Connectedness between countries was assessed on a first batch of populations, underlining the relevance of implementing across country evaluation.As a main outcome of these first achievements, case studies of international evaluations will be developed for determining the best technical options and highlighting the benefits for breeders.

Authors: F. Tortereau, C. Marie-Etancelin, INRAE, France Feed efficiency is a trait whose selection meets both economic and agri-environmental challenges. The objective of farmers is to breed animals that consume less feed while maintaining their production level. Genetic selection of efficient animals is possible but requires the recording of individual feed intakes by means of automatic feeders, which remain very expensive devices. Most of the breed cannot afford to invest in these automates. The identification of proxies for feed intake is therefore a challenge we have to overcome in order to enable breeding companies to include feed efficiency in their breeding objectives. The proxies we proposed to consider in the SMARTER project are to be easily and non-invasively sampled and analysed at a reasonable cost. Among these proxies, we will first try to benefit from already recorded traits such as body weights and average daily gains. We will also focus on biological markers that can be measured in fluids that breeders are used to sample: blood and faeces. From blood, we will get genotypes and metabolites (either through specific determination or through NMR spectra). From faeces, we will get NIRS spectra that will be analysed in comparison with NIRS spectra obtained from the food itself. All these proxies can be considered separately to predict feed intake but also combined with data integration methods to benefit from all these proxies. For research purpose, additional proxies such as ruminal data (microbiota, volatil fatty acids…) will be considered because of their direct link with feed efficiency, but it is currently difficult to include them in the proposed proxies to sample routinely

Authors: Gabriel Ciappesoni et alia, INIA-Uruguay

Abstract focused on practical and relevant aspects of measuring greenhouse gases emissions from sheep, based on short-term measurements using portable static accumulation chambers with an internal volume of 870 L (Goopy & al.2011; Hegarty.2013). This method allows measuring (and ranking) large groups of animals in a fast and inexpensive way. After weaning, animals are placed 2 to 3 times (1 week between rounds) in sealed chambers for 40-50 mn, after 3 weeks of constant feeding in terms of quantity and type of feed. On the measurement day with the animal placed into the chamber, CH4, CO2, and O2 are recorded using a portable multi-gas detector (in parallel with a background estimation) every 10 or 20 minutes. Air temperature and pressure are also registered for the calculation of methane emission at standardized conditions. Multi-gas detector calibration, bump tests and chambers leak tests are performed routinely. Sealing of the chamber is mandatory to guarantee isolation, which is highly recommended. Transparent chambers are used to reduce stress, accounting for animal welfare. Records of bodyweight are necessary to estimate actual gas volume in the chamber and to estimate methane intensity. Also, dry matter intake on the measurement day and previous days are required to assess methane yield. J.P.Goopy, R.Woodgate, A.Donaldson, D.L.Robinson, R.S.Hegarty (2011), Validation of a short-term methane measurement using portable static chambers to estimate daily methane production in sheep. Animal Feed Science and Technology, 166, 219-226.

R.S.Hegarty (2013), Applicability of short-term emission measurements for on-farm quantification of enteric methane. Animal, 7(s2), 401-408.

Authors: F.Tortereau et alia, INRAE, France Feed efficiency is a trait of major interest because breeding efficient animals is not only cost-saving but also leads to a decrease of environmental impacts. The genetic improvement of this trait requires feed intakes to be recorded individually. Sheep and goats are gregarious animals that are reared in group so either several individual devices are available (one for each individual) or one device can handle the feeding of several animals. Automated concentrate feeders (ACF) exist for small dairy ruminants. They deliver concentrate in the milking parlour, but the hypothesis is that all the delivered concentrate is eaten by the animal. Other ACF have been developed, mainly for research purposes. Very few devices enable the recording of forage intakes (AFF) when animals are reared in groups whereas small ruminants are mainly fed with forage. At INRAE, we developed ACF that can be used in ad libitum or restricted version, AFF and automated water dispensers. These three feeders return data for each visit (quantity and duration). This high-throughput recording is of high value. In combination with other biological data, feed intakes can be analysed to study feed efficiency but also responses to nutritional or infectious challenges. The succession of visits between animals gives hints for social interactions’ studies. The duration of visits is also included in behavioural analyses. These large datasets can also help in identifying major events (technical, meteorological, health …) and then to test the resilience of animals under these stressful events. Current research programs aim at identifying proxies for feed intakes automated feeders, remain expensive devices.