general enquiries on this form should be made...

General enquiries on this form should be made to:Defra, Science Directorate, Management Support and Finance Team,Telephone No. 020 7238 1612E-mail: [email protected]

SID 5 Research Project Final Report

SID 5 (Rev. 3/06) of 29

NoteIn line with the Freedom of Information Act 2000, Defra aims to place the results of its completed research projects in the public domain wherever possible. The SID 5 (Research Project Final Report) is designed to capture the information on the results and outputs of Defra-funded research in a format that is easily publishable through the Defra website. A SID 5 must be completed for all projects.

This form is in Word format and the boxes may be expanded or reduced, as appropriate.

ACCESS TO INFORMATIONThe information collected on this form will be stored electronically and may be sent to any part of Defra, or to individual researchers or organisations outside Defra for the purposes of reviewing the project. Defra may also disclose the information to any outside organisation acting as an agent authorised by Defra to process final research reports on its behalf. Defra intends to publish this form on its website, unless there are strong reasons not to, which fully comply with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.Defra may be required to release information, including personal data and commercial information, on request under the Environmental Information Regulations or the Freedom of Information Act 2000. However, Defra will not permit any unwarranted breach of confidentiality or act in contravention of its obligations under the Data Protection Act 1998. Defra or its appointed agents may use the name, address or other details on your form to contact you in connection with occasional customer research aimed at improving the processes through which Defra works with its contractors.

Project identification

1. Defra Project code IF0149

2. Project title

Determining strategies for delivering environmental sustainable production in the UK ruminant industry through genetic improvement

3. Contractororganisation(s)

SAC Commercial LtdSir Peter Wilson BuildingKings BuildingsWest Mains RoadEdinburghEH9 3JG

54. Total Defra project costs £ 274,128(agreed fixed price)

5. Project: start date................ 01 March 2009

end date................. 31 March 2010

SID 5 (Rev. 3/06) of 29

6. It is Defra’s intention to publish this form. Please confirm your agreement to do so...................................................................................YES NO (a) When preparing SID 5s contractors should bear in mind that Defra intends that they be made public. They

should be written in a clear and concise manner and represent a full account of the research project which someone not closely associated with the project can follow.Defra recognises that in a small minority of cases there may be information, such as intellectual property or commercially confidential data, used in or generated by the research project, which should not be disclosed. In these cases, such information should be detailed in a separate annex (not to be published) so that the SID 5 can be placed in the public domain. Where it is impossible to complete the Final Report without including references to any sensitive or confidential data, the information should be included and section (b) completed. NB: only in exceptional circumstances will Defra expect contractors to give a "No" answer.In all cases, reasons for withholding information must be fully in line with exemptions under the Environmental Information Regulations or the Freedom of Information Act 2000.

(b) If you have answered NO, please explain why the Final report should not be released into public domain

Executive Summary7. The executive summary must not exceed 2 sides in total of A4 and should be understandable to the

intelligent non-scientist. It should cover the main objectives, methods and findings of the research, together with any other significant events and options for new work.

The identification of traits that affect economic and biological efficiency as well as green house gas (GHG) mitigation is crucial to achieve the maximum benefits from genetic improvement. There are several characteristics that are economically relevant for ruminant production which are associated not only with incomes (i.e., weaning weight, carcass weight and conformation) but also with enterprise costs through losses and production efficiency (i.e. offspring survival, resistance to diseases, feed intake and efficiency). Although many of these traits are already breeding goals for the UK breeding programmes, others have not been included yet as particular traits associated with mitigation of GHG emission, such as feed efficiency or even direct measurement of methane production. These traits are difficult and/or expensive to be recorded particularly under commercial farming conditions which has limited their implementation.

Evidence in the literature suggests that there is between-animal variation in methane emissions. Therefore, breeding directly for lower emissions is potentially feasible and may be an important tool for the mitigation of methane emissions from ruminant livestock. In addition, understanding the factors which influence the microbial population of the rumen is essential for improving productivity and mitigating GHG. From reviewing literature from various species there is evidence to suggest that the

Advances in molecular genetics have led to the development of dense panels of single nucleotide polymorphism (SNP) markers that cover the whole genome. This dense coverage means that markers are likely to be close to the quantitative trait loci (QTL) responsible for genetic variation in quantitative traits. This proximity increases the probability that marker alleles will be inherited in linkage with given QTL alleles throughout the population. This is called linkage disequilibrium (LD). We can therefore estimate the additive genetic effect (referred to as the genetic effect from here on) caused by the QTLs associated with markers and use this marker effect information to predict the genetic merit of other genotyped animals in the population. The principal focus for application of this technology in livestock breeding is in genomic selection (GS) where we are interested in the sum of all marker effects on the trait of interest to estimate the breeding value of the animal.

There is a strong likelihood that phenotypic information from crossbred animals will be utilised in training populations (the population of animals used to identify genetic effects) for GS and that the information estimated will be used to undertake selection in purebred populations. A simulation study was undertaken and showed that crossbred performances can be efficiently used in the genomic evaluation of purebred relatives, which is essential for crossbreeding breeding programmes in beef

SID 5 (Rev. 3/06) of 29

and sheep. This was more effective when the breeds involved in the cross were closely related. Considering that traits such as methane emissions and their correlated traits such as feed efficiency can only be measured on a subsample of the population, genomic selection is expected to be the most effective presently known methodology to achieve sufficient selection response in these traits.

The potential accuracy of GS is affected by the size of the training population, the heritability of the trait, and the genetic size of the population (effective size). Selection index theory was used to explore the impact of these factors on using genomic selection in selection for current and expanded beef and sheep breeding goals. Applying genomic selection to both males and females was associated with improvements of up to 14 and 21% in the beef terminal sire and maternal indexes and up to 33, 48 and 15% in the terminal sire, longwool and hill sheep indexes, respectively. Applying genomic selection to males only was associated with significant improvements, however, the magnitude of response was lower.

Considerable improvements in total economic response relative to the base index (under conventional selection) were achieved from applying genomic selection to the expanded index. Improvements of up to 56, 56, 36, 81 and 29% were observed in the beef terminal, beef maternal, sheep terminal, longwool and hill sheep indexes, respectively, when genomic selection was applied to both sexes. Further improvements in response were observed using only young sires for breeding.

The size of the training population was found to influence the economic response that can be achieved when including genomic information. Increasing the training population size was associated with a linear improvement in total economic response, where the highest response was achieved with the highest training population size of 5000. Smaller effective population size results in greater genomic selection accuracy and hence greater response to selection.

At the whole industry level the genetic response observed had favourable environmental impact in general, with the exception of the longwool sheep sector. However, these results should be interpreted with caution given that some traits associated with higher efficiency and lower losses, such as ewe longevity and animal survival, had a negative influence on GHG mitigation when they are considered on breeding animal basis. This may underestimate possible effects on GHG mitigation particularly in those systems in which the maternal traits are particularly relevant. Results may be different if GHG emissions are evaluated by unit of product. The evaluation of the genetic programmes per unit of product may imply re-defining the current expression of the breeding goals which have been defined until now on per breeding animal basis.

Large net incomes were obtained when genomic selection was applied simultaneously with the inclusion of survival and residual feed intake (RFI) in the breeding goals in both maternal (62% - 80%) and terminal (33% to 47%) beef breeding programmes. In sheep under the current breeding structure the ranges were 30% - 56%, 4% - 15% and 9% - 22% for the longwool, hill and terminal sheep, respectively. On the other hand, the net on-farm returns showed that their implementation was not economically viable in all cases until genotyping costs reduce, as they are expected to do. Further studies on optimising breeding structures may be very valuable for the sheep industry. In the case of the beef sector, the implementation of a training population following a multi-breed approach, may allow integrating of maternal and terminal breeds and therefore optimising the contributions from both sectors to the beef industry.

A number of delivery scenarios for GS are considered in a value proposition including, industry-driven GS, commercial company-driven GS, research provider-driven GS and a joint venture between industry and a research provider. These scenarios are compared to the “status quo” scenario to establish the additional benefit these scenarios provide. The benefits included farmer income and GHG benefits. Assuming that the cost of genotype tests will fall over the initial years and based on technology penetration of 35% for the beef industry the joint venture option promises the highest net present value of £24m for implementing a GS service (£88m for 100% penetration). For the sheep sectors a 34% penetration rate was assumed which is double that currently achieved. Across all sets of assumptions the joint venture option again came out best. Two assumptions were made about ram testing, either all candidate rams were genotyped or a two stage approach was adopted where only those rams with superior pedigree indexes were genotyped (30%). The net present value varied between £44m and £69m for these two approaches.

A stakeholder workshop was held in Edinburgh to assist in the scoping of a ruminant genetic improvement network. There was a strong consensus among stakeholders that a ruminant GIN could deliver valuable outcomes with respect to environmental sustainability if it focused on co-ordinated, strategically planned R&D and KE. It was recommended that Defra establish a ruminant GIN – ideally

SID 5 (Rev. 3/06) of 29

with other funding partners, to maximise the resources focussed on the problem of breeding ruminants for improved environmental sustainability.

Project Report to Defra8. As a guide this report should be no longer than 20 sides of A4. This report is to provide Defra with

details of the outputs of the research project for internal purposes; to meet the terms of the contract; and to allow Defra to publish details of the outputs to meet Environmental Information Regulation or Freedom of Information obligations. This short report to Defra does not preclude contractors from also seeking to publish a full, formal scientific report/paper in an appropriate scientific or other journal/publication. Indeed, Defra actively encourages such publications as part of the contract terms. The report to Defra should include: the scientific objectives as set out in the contract; the extent to which the objectives set out in the contract have been met; details of methods used and the results obtained, including statistical analysis (if appropriate); a discussion of the results and their reliability; the main implications of the findings; possible future work; and any action resulting from the research (e.g. IP, Knowledge Transfer).

1. BACKGROUND

Livestock account for up to 35-40% of world methane production, a large proportion (around 80%; de Haan et al, 1996) of which comes from enteric fermentation and a smaller proportion (around 20%; Safely et al., 1992) from anaerobic digestion in liquid manure. There are essentially three routes through which genetic improvement of livestock can help to reduce emissions per kg product, per head and at a UK national level: 1) as a result of improved productivity and efficiency at the individual animal or herd/flock level; 2) as a result of reducing ‘wastage’ (from infertility, disease and mortality) at the animal, or herd/flock level; and 3) as a direct response to selection on emissions, if or when these are measurable. A recent Defra report (Genesis Faraday Partnership, 2008) illustrated the substantial level of GHG emissions associated with production of meat from beef cattle. Genetic improvement of livestock is a particularly cost-effective technology, producing permanent and cumulative changes in performance. Mechanisms by which genetic tools could be used to reduce emissions per kg product include improving productivity and efficiency, reducing wastage at the herd or flock level, and selecting directly for reduced emissions. Recent research for the Committee on Climate Change has highlighted that a range of genetic improvement options in ruminants, combined with performance recording, provides a cost-effective mechanism of abating GHG from agriculture to help the UK meet Kyoto targets (Moran et al., 2008).

Conventional genetic evaluation methods may be very effective at reducing GHG emissions from livestock through routes one and two, but it has been hypothesised that the rate of improvement could be substantially enhanced by adopting emerging molecular genetic technologies. The potential GHG emission reductions offered by route three would be more readily harnessed using these emerging molecular genetic technologies. Rapid progress in, and falling costs of, genome sequencing and high throughput DNA-techniques have led to the availability of genome-wide applications to commercial livestock production. The new genotyping techniques, based on array technology, provide an efficient platform to genotype hundreds or even thousands of individual animals at relatively low costs. Currently, bovine single nucleotide polymorphism (SNP)-arrays with >54k SNPs are available commercially and their density is increasing rapidly. The ovine 50k SNP array has just become commercially available. Access to this extremely dense SNP information represents a unique genomic resource to investigate the genetic basis of traits, in particular those that are difficult to measure, such as methane production, feed efficiency, disease resistance, meat quality, etc. Novel approaches to genome-wide selection (GS) as proposed by Meuwissen et al. (2001) are already being implemented into genetic evaluation of cattle, primarily dairy cattle, to identify animals of high genetic merit (e.g. Guillaume et al., 2007; Goddard and Hayes, 2007). The method uses associations of large numbers of SNP markers across the whole genome with phenotypes, without prior QTL (quantitative trait loci) detection. It is expected to have a high potential, particularly for improving traits difficult to target by conventional selection techniques based on BLUP (Best Linear Unbiased Prediction) evaluation, such as disease resistance, meat quality, feed efficiency and reduced methane production. There are two steps involved in GS. In the first step, the effects of the markers (i.e., SNPs) are estimated in a ‘reference’ population where the individuals are genotyped and comprehensively measured for the trait of interest. In the second phase, the estimates of marker effects are used to obtain an estimate of the genetic merit of newborn animals that are only genotyped (but not measured). The second step could allow a much greater number of animals to obtain breeding values of comparable or better accuracy than achievable under conventional evaluation. Genetic improvement of sheep and beef cattle in traits directly or indirectly associated with GHG emissions is expected to have substantial effects on reduction of these emissions. However, the

SID 5 (Rev. 3/06) of 29

genetic basis of these traits and their genetic correlation with GHG emissions has to be known first for the development of environmentally sustainable breeding goals for the UK.

The objectives of this project can broadly be divided into a consideration of the beef and sheep sectors in the UK in relation to the ability of current breeding tools and GS methods to improve the environmental impact of these sectors, to look at the structure of the populations with respect to training population design and to quantify the economic aspects of delivering GS in sheep and beef with a view to how technology uptake may be enhanced. Additionally the project scoped the formation of a ruminant genetic improvement network. The various individual tasks required to address these objectives are delivered through a series of key pieces of work described in the sections of this report. The background to the environmental impact of current and new traits has been addressed through an extensive review of the literature (section 2) and also in collaboration with Wall et al., 2010 (Defra project IF0182). A big question for beef and sheep with regard to GS is the ability to use crossbred animals to provide phenotypic records. Section 3 outlines a simulation study undertaken to investigate this issue. The core of this study addressed the impact of GS on beef and sheep breeding goals (section 4) both current and expanded to include traits identified in section 2. This work developed the inclusion of GS in a selection index framework. The impact of GS on farmer returns and environment for the whole of UK beef and sheep production assuming various levels of technology uptake is addressed in sections 5 and 6. Section 6 goes on to explore how GS might be delivered in the UK through the use of a value proposition. Finally, a summary of the genetic improvement network scoping exercise can be found in section 7.

2. LITERATURE REVIEW

Livestock production has traditionally been an important economic activity as well as source of food and provider of public goods such as biodiversity and landscape values. However the role and relevance of livestock production is challenged by the increasing concern regarding its contribution to greenhouse gas (GHG) and other pollutant emissions, particularly in regards to methane and nitrogen.

Beef production systems can be successfully modified through the manipulation of genetic components underlying the expression of the economically relevant traits (Amer et al. 2007), which can also translate into the reduction of GHG emissions (Moran et al. 2008; Gill et al. 2010). The identification of the traits that affect economic and biological efficiency as well as GHG mitigation is crucial to achieve the maximum benefits from genetic improvement. There are several characteristics that are economically relevant for beef production which are associated not only with incomes (i.e., weaning weight, carcass weight and conformation) but also with enterprise costs through losses and production efficiency (i.e. calf survival, resistance to diseases, feed intake and efficiency). Although many of these traits are already breeding goals for the UK beef breeding programmes, some have not been included yet, in particular traits associated with mitigation of GHG emission, such as feed efficiency or even direct measurement of methane production. These are, in general, traits that are difficult and/or expensive to be recorded particularly under commercial farming conditions.

Genomic selection (GS) is a new genetic tool that may provide the opportunity to enhance the genetic progress that is currently being achieved and facilitates the inclusion of these relevant but difficult-to-measure traits. Genomic selection is based on the estimation of the genetic merit of breeding animals using a large number of genetic markers (single nucleotide polymorphisms, SNP) that cover the whole genome (Goddard and Hayes 2007). Panels of high density of SNPs recently became commercially available making possible the implementation of GS in current breeding programmes.

Improvements of production efficiency will lead to more significant contributions to GHG mitigation. Some of the potential new traits may play a significant role in the reduction of GHG emissions, such as residual feed intake, mature weight and survival. Some other new traits such as meat quality may be less directly relevant for the protection of the environment but play an important role in the long term sustainability of supply chains.

Residual feed intake (RFI) is calculated as the difference between actual and predicted feed intake based on “standard requirements” for production and live weight maintenance. Residual feed is the feed that cannot be accounted for by any of these processes. Therefore, differences in RFI are related to the ability of the animal to be more efficient and consume less feed for the same output (van der Werf, 2004). Furthermore, the international literature indicates that RFI is a heritable trait that is not only an indicator of animals that are more efficient but also that produce a lower volume of methane. Therefore, selection to improve efficiency by reducing RFI offers an opportunity to reduce feed intake, improve GHG mitigation without compromising growth performance, and without the possible correlated response of increasing cow size (van der Westhuizen et al. 2004; Arthur and Herd 2008). Nevertheless, more research on the consequences on maternal traits, reproduction, meat quality and animal welfare is needed for a complete evaluation of the impact of selection for RFI on the beef production systems (Herd and Arthur 2008).

SID 5 (Rev. 3/06) of 29

The high cost involved in recording feed intake, and the limited numbers of animals that can be tested has created interest in identifying genomic markers associated with feed efficiency and integration of this marker information into breeding programmes (Barwick et al. 2009). IGF-1 has received substantial attention as a potential indirect marker for feed efficiency. A review of the literature suggests that IGF-1 is heritable and associated with various economically important traits including feed efficiency as well as production traits. In particular IGF-1 has been shown to be positively correlated with RFI (Johnson et al. 2002; Moore et al. 2003, Moore et al. 2005). IGF-1 is simple and inexpensive to measure and it can be measured in early life suggesting its potential for use as an indirect selection criterion for genetic improvement of both feed efficiency and carcass traits (Kahi et al. 2003; Moore et al. 2003). More recently, this marker lost support in breeding as concerns were raised by the inconsistencies of its important associations across different populations at different measurement times. Further research is warranted to fully understand possible genetic antagonisms before the use of IGF-1 as an indicator trait for RFI can be recommended in breeding programmes.

The following potential future breeding goal traits, which may play a relevant role on enhancing profit, efficiency and GHG mitigation, were identified as part of this review: mature weight, shear force, docility, RFI and calf survival. Mature weight is currently in the process of being incorporated into genetic evaluations in UK and therefore will be considered in the next steps of this project. Genetic parameters for the potential future breeding goals and their association with the current ones are relevant for the implementation of breeding programmes orientated to more sustainable beef production, as well as for the inclusion of genomic selection. The estimates of heritability available in the literature for the potential future breeding goals indicated that there is scope for genetic improvement of most of the traits. Lower values were reported only for calf survival. In terms of genetic correlations, the availability of information was poor or non-existent in some cases. For example genetic correlations of RFI with other traits are concentrated on growth traits but no estimates were available for the association with reproduction or meat quality traits. The difficulties of measuring these traits have not only delayed their utilization in breeding programmes but also the estimation of genetic parameters. Regarding the effect of these traits on GHG emission, the information reported by Wall et al. (2010) is discussed in this review.

Mitigating GHG emissions from livestock relies on obtaining accurate and repeatable measurements of GHG emissions from animals which is a challenge, particularly under commercial conditions. At present there are several methods available for measuring methane emissions in individual animals. These methods generally fall into one of two categories: (1) direct and (2) indirect methods. Direct methods comprise of partial or total enclosure of animals and indirect methods involve the use of isotopic and non-isotopic tracers and estimate methane emissions based on characteristics of rumen fermentation (Pinares-Patino and Clark 2008). From reviewing these techniques it is clear that although chambers incur higher cost and are more labour intensive, they provide more accurate measures of methane emissions. Therefore, novel and robust approaches have to be developed using measurement techniques in total enclosure facilities as a reference method. The newly developed proxy methods should enable the industry to cost-effectively measure methane on a large number of animals under practical conditions.

It is important to investigate whether evidence exists for genetic variation of methane in ruminants. If this does exist then this provides a useful opportunity to mitigate GHG by exploiting between-animal variation in emissions, by directly selecting animals for lower emissions of GHG. This would provide a relatively cost-effective method to produce a long-term reduction in GHG emissions (Vlaming et al. 2008). Research in the literature indicates that large between-animal variation in methane emissions exists in ruminants, however the evidence is relatively limited, most likely due to the difficulty in collecting suitable data on methane production of animals in a powerful genetic design. Due to the limited evidence available in beef cattle, this review included information from other ruminants as well as humans. Within- and between-animal variation in methane emissions have been reported from studies using respiration chambers as well as the SF6 tracer technique (Blaxter and Clapperton 1965; Lassey et al.1997; Boadi and Wittenberg 2002; Robertson and Waghorn 2002; Pinares-Patino et al. 2003; Vlaming et al. 2008). However, there are contradictory results in the literature which report a lack of persistence of animal differences in methane emissions over time (Pinares-Patino et al. 2000; Pinares-Patino et al. 2003; Goopy and Hegarty 2004; Munger and Kreuzer 2008). Achieving accurate and repeatable measures of methane emissions is a difficult task and therefore the lack of persistence in methane emissions described in these studies may be influenced by the measurement techniques. This strongly indicates a requirement for obtaining large sample sizes with accurate and repeatable measures of methane emissions. Furthermore, diet variability within and between experiments will affect methane emissions, therefore it is essential to feed a highly standardized diet in these genetic studies to limit variation in methane emissions. On the whole, evidence in the literature suggests that there is between-animal variation in methane emissions, therefore, breeding directly for lower emissions is potentially feasible and may be an important tool for the mitigation of methane emissions from ruminant livestock.

The microbial population of the gastrointestinal tract is the key to determining the quantity and range of nutrients available for absorption by the animal (Hegarty 2004). Understanding the factors which influence the microbial population of the rumen is essential for improving productivity and mitigating GHG. Understanding the host’s influence on the population of microbes in the gastrointestinal tract is important for mitigating GHG. From

SID 5 (Rev. 3/06) of 29

reviewing literature from various species there is evidence to suggest that the host does influence the microbial population of the digestive tract (Hackstein et al. 1996; Zoetendal et al. 1998; Tannock et al. 2000; Zoetendal et al. 2001; Seksik et al. 2003; Vanhoutte et al. 2004; Ley et al. 2005; Stewart et al. 2005; Guan et al. 2008; Shi et al. 2008). The variation observed between animals regarding methane emissions may also be due to factors associated with digestion (e.g. rumen retention time, ruminal capacity and pH), which have an impact on the microbial population of the gastrointestinal tract (Hegarty 2004; Kumar et al. 2009). Furthermore, understanding the role of rumen microbes in the digestive process is crucial to optimising productivity of animals and altering GHG emissions (Attwood et al. 2008). Methanogens are a component of the microbial community of the rumen also known as methane-producing archaea. In order to inhibit methanogenesis it is essential to investigate the diversity of methanogens present in ruminants and genetic variation in methane production of these microbes (Attwood et al. 2008; Ouwerkerk et al. 2008). To date microbial populations of the rumen and gastrointestinal tract have not been well characterised. However enormous genetic diversity within the microbial population of the gastrointestinal tract has been reported (Zoetendal et al. 2006). Recent advances in genomics has allowed for the sequencing of whole genomes of microbes. This allows for a better understanding of the functions of microbes and how they interact with the surrounding environment (Attwood et al. 2008; Attwood and McSweeney 2008). This tool may be an important contributor in mitigating GHG emissions from ruminants (Attwood et al. 2008; Attwood and McSweeney 2008).

3. SIMULATION OF THE USE OF GENOMIC SELECTION IN CROSSBRED POPULATIONS.

Based on the simulation study it can be concluded that crossbred performance can be efficiently used for genomic selection of purebred populations. The accuracies of genomic breeding values from crossbred performance are higher when closely related breeds are crossed than when distantly related or unrelated breeds are crossed. Therefore, population analysis of the genomic relationship of breeds to be crossed is expected to give information on its usefulness for selection of purebred performances. In the present simulation no purebred performance information was used for genomic evaluation of the selection candidates. Wei (1992) and Dekkers (2007) showed that selection of purebred animals based on combined purebred and crossbred performances will be the most efficient procedure for improvement of crossbred performances of commercial animals. However, the present study showed that the impact of ignoring dominance effects can substantially reduce the accuracy of additive genomic breeding values (GBVs), depending on the relative importance of dominance variation in relation to additive genetic effects. Therefore, inclusion of dominance effects in the genomic breeding evaluation, as shown by Toro and Varona (2009), is recommended, particularly for the traits with low heritability for which dominance effects are expected to be most important. This is expected to result in more accurate estimates of additive GBV and may also give the opportunity to exploit dominance effects for example in a mate selection programme.

Using crossbred information from completely unrelated populations substantially reduced the accuracy of the estimated GBV and therefore specification of the relationship between the crossbred breeds is necessary in order to use crossbred information for genomic selection efficiently. Inclusion of dominance effects in genomic evaluation models is important when the crossed breeds are unrelated or distantly related because in these cases ignoring dominance will substantially decrease the accuracies of GBV depending on the dominance heritability. In contrast, when the crossed breeds are closely related, ignoring dominance showed a low effect on the accuracy of estimated GBV especially when the additive heritability is high.

Generally the simulation showed that genomic selection results in sufficiently high accuracies of estimated GBV even for lowly heritable traits and thus can be successfully used to select purebred animals using purebred as well as crossbred performances in the training population. Considering that traits such as methane emission and traits correlated to it such as feed efficiency can only be measured using proxy methods (e.g. methane detector gun) or electronic feeders on a subsample of the population, genomic selection is expected to be the most effective methodology to achieve sufficient selection response in these traits. This simulation provides the basis for an efficient improvement of those traits. Further simulation using a deterministic approach (developed or validated based on simulation studies) is used in this report to reflect the efficiency of genomic selection based on the given UK population structure in sheep and beef.

4. ECONOMIC RESPONSE THAT CAN BE ACHIEVED FROM INCLUDING NEW TRAITS IN THE CURRENT BEEF AND SHEEP INDEXES UNDER CONVENTIONAL AND GENOMIC SELECTION

Selection index theory was applied to investigate the genetic response when using the beef and sheep indexes under conventional selection or genomic selection (GS) given the structure of the UK beef and sheep industries. The addition of new traits, was also investigated to mitigate GHG emissions without compromising meat quality and animal welfare traits. Bespoke selection index software developed by AbacusBio Ltd., New Zealand was used.

4.1 Beef indexes (terminal sire index and maternal index)

SID 5 (Rev. 3/06) of 29

4.1.1 Model description

Selection criteria and breeding goals: The base indexes were constructed to mimic the current terminal sire and maternal index using information available from the UK national genetic evaluations data. A second index was created by adding five new traits to the selection criteria and breeding goals including: residual feed intake (RFI), shear force (SF), birth survival, weaning survival and docility score (DS). Detailed information about the current and new recorded traits are presented in Table 4.1 and the breeding goal traits in Table 4.2.

Table 4.1 Current and new recorded traits in the beef terminal sire (TS) and maternal (M) index.

Recorded trait name1 HeritabilityPhenotypic

Variance RepeatabilityIndex

BWT-direct, kg 0.23 9.035 0.299 TS, MBWT-maternal, kg 0.06 9.035 0.299 MWT200-direct, kg 0.33 807 0.393 TS, MWT200-maternal, kg 0.07 807 0.393 MWT400, kg 0.40 1590 0 TS, MMSC, score (1-15, higher = higher muscularity) 0.27 1.315 0 TS, MFD, mm 0.29 6395 0 TS, MMD, mm 0.26 2518 0 TS, MGL-direct, days 0.29 23.8 0.391 TS, MGL-maternal, days 0.05 23.8 0.391 MCD-direct, score (1-5, higher = more assistance) 0.12 1.015 0.217 TS, MCD-maternal, score (1-5, higher = more assistance) 0.05 1.015 0.217 MCI, days 0.09 2368 0 MAF, year 0.20 0.221 0 MLS, parity 0.11 6.059 0 MMW, kg 0.51 3779 0 MRFI-test, kg DMI/animal/ growing period (24 months)2 0.34 95852 0

TS, M

SF, kg2 0.25 3.26 0 TS, MBSV-direct, score (0=not survive, 1=survive)2 0.10 0.026 0 TS, MBSV-maternal, score (0=not survive, 1=survive)2 0.03 0.026 0 TS, MWSV-direct, score (0=not survive, 1=survive)2 0.11 0.013 0 TS, MWSV-maternal, score (0=not survive, 1=survive)2 0.05 0.013 0 TS, MDS, score (1-6, higher = less docile)2 0.33 0.430 0 TS, M

1Definition of trait name abbreviations: BWT, birth weight; WT200, weight at 200 days; WT400, weight at 400 days; MSC, muscle score; FD, fat depth; MD, muscle depth; GL, gestation length; CD, calving difficulty; CI, calving interval; AF, age first calving; LS, lifespan; RFI, residual feed intake; SF, shear force; BSV, birth survival (survival to 3 days); WSV, weaning survival (from day 4 to weaning); DS, docility score.2New recorded traits included in the index

Sources of information: Two breeding program structures were considered in the present research: (i) current breeding structure (CBP) and (ii) a young sire programme (YSP). Sources of information for each breeding structure were calculated from UK national genetic evaluations data. This included numbers of records for each trait on the candidate, paternal half sibs, mother, mother’s sisters and progeny (paternal and maternal progeny). For the new traits, no sources of phenotype information were included for RFI, SF and DS as these will not be measured in a commercial setting. However sources of information were included for survival traits as this information is already recorded.

Table 4.2 Heritability, phenotypic variance and economic weights for current and new profit traits beef terminal sire (TS) and maternal (M) index.

Profit trait name1 Heritability Phenotypic variance Economic weight IndexWT200-maternal, kg 0.07 807 0.73 MCW, kg 0.44 860 0.7 TS, MCFS, score (1-15) 0.13 4.099 0 TS, MCCS, score (1-15) 0.11 9.833 6.7 TS, MGL-direct, days 0.29 23.8 -1.17 TS, MCD-direct, days 0.12 1.015 -2.88 TS, MCD-maternal, days 0.05 1.015 -2.19 MCI, days 0.09 2368 -0.83 MAFC, year 0.20 0.221 -48.11 MLS, parity 0.11 6.059 6.63 MMW, kg 0.51 3779 -0.23 MRFI-growing, DMI/animal/ period (24 months)2 0.30 95852 -0.069 TSRFI-breeding DMI/animal/ period (12 months)2 0.30 46916 -0.066 TS, MSF, kg2 0.25 3.260 0 TS, MBSV-direct, score (0=not survive, 1=survive)2 0.10 0.026 266.8 TS, M

SID 5 (Rev. 3/06) of 29

BSV-maternal, score (0=not survive, 1=survive)2 0.03 0.026 303.6 TS, MWSV-direct, score (0=not survive, 1=survive)2 0.11 0.013 266.8 TS, MWSV-maternal, score (0=not survive, 1=survive)2 0.05 0.013 303.6 TS, MDS, score (1-6, higher = less docile)2 0.33 0.430 0 TS, M

1Definition of trait name abbreviations: WT200, weight at 200 days; CW, carcass weight; CFS, carcass fat score; CCS, carcass conformation score; GL, gestation length; CD, calving difficulty; CI, calving interval; AFC, age first calving; LS, lifespan; MW, mature weight; RFI-growing, residual feed intake for slaughter calves; RFI-breeding, residual feed intake for breeding cows; SF, shear force; BSV, birth survival (survival to 3 days); WSV, weaning survival (from day 4 to weaning); DS, docility score.2New profit traits included in the index

Breeding program structure: For both breeding program structures the proportions of male and female candidates which were selected and generation intervals for male and females were estimated from UK national genetic evaluations data (Table 4.3).

Table 4.3 Proportion of candidates selected and generation intervals of the two breeding programmes.

Breeding programmeYSP CBP

Proportion of male candidates selected 0.0607 0.0375Proportion of female candidates selected 0.533 0.533Male generation interval (years) 3 5.583Female generation interval (years) 6.245 6.245

Scenarios simulated: Selection index theory was applied to investigate the response of the beef terminal sire index different scenarios given the structure of the UK beef industry:

Scenario 1: Base index: investigating the current selection index (no new traits). Both breeding structures were considered (CBP and YSP). The base index was investigated using conventional selection (BLUP) alone and conventional plus genomic selection. For the genomic selection effective population sizes of Ne 100, 300 and 500 and training population size (phenotypes and genotypes recorded) of 1000, 2000, 3000, 4000 and 5000 were considered. In this scenario, genomic selection was considered for both males and females as well as for males only as it may be too expensive to genotype both sexes.

Scenario 2: Base index + new traits included as profit and recorded traits (provided they can be recorded). In this scenario, of the five new traits considered as recorded traits information sources were only available for birth survival and weaning survival. The five new traits were also included as profit traits. In this scenario conventional selection of both breeding program structures were considered (CBP and YSP). Genomic selection was applied to these programmes, where effective population sizes of Ne 100 and Ne 300 and number of recorded phenotypes of 2000 and 5000 were considered. Genomic selection was considered for both males and females as well as for males only. Two steps were considered for the CBP. In this first step no selection pressure was applied to the new profit traits (i.e. no economic weights were included). In the second step selection pressure was applied to the following profit (breeding goal) traits in addition to the base breeding goals: residual feed intake, birth survival and weaning survival (i.e. economic weights were included for these traits in the index equations of the new traits). For the YSP only the second step was considered where selection pressure was applied to residual feed intake and survival traits in addition to the base breeding goal traits.

Modelling the addition of GBV: Genomic information was included in the selection index model based on the theory of Dekkers (2007). Accuracies of GBVs are predicted based on trait heritability, number of phenotyped animals in the training population, the number of QTL underlying the trait and the effective population size (Ne) (Daetwyler et al. 2008; Goddard 2009).

4.1.2 Results

4.1.2.1 Scenario 1: Base index

The total economic responses of the two breeding programmes based on conventional selection and the economic responses of different genomic selection strategies relative to the current scenario (CBP) under conventional selection is presented in Table 4.4 for the terminal sire index and Table 4.5 for the maternal index.

Terminal sire index: Under conventional selection, the economic response was similar between the two breeding structures (CBP and YSP), however a slightly higher response (3%) was achieved when selecting younger sires. This shows that for a terminal sire breeding goal shorter generation interval in the YSP compensates for the loss of information in younger sires as young sires have a lower number of information sources from relatives and lower selection intensity. Including genomic information increased the economic response to selection in all genomic selection scenarios in both breeding structures, however the magnitude of the response in comparison to conventional selection was much higher when selecting younger sires (up to 35% increase when genomic selection was applied to males and females and up to 27% increase when genomic

SID 5 (Rev. 3/06) of 29

selection was only applied to males compared to the base index) in comparison to the current breeding structure (up to 14% increase when genomic selection applied to males and females and up to 9% increase when genomic selection was only applied to males compared to the base index).

Table 4.4 Total economic response in £/animal (%) of different breeding programmes based on conventional selection and genomic selection of the current beef terminal sire index.

CBP YSPConventional Selection

3.39 (100) 3.49 (103)

Ne100b Ne300b Ne500b Ne100b Ne300b Ne500b

Genomic selection applied to males and femalesGS1a 3.55 (105) 3.46 (102) 3.43 (101) 3.92 (116) 3.68 (109) 3.62 (107)GS2 a 3.66 (108) 3.52 (104) 3.47 (102) 4.17 (123) 3.83 (113) 3.73 (110)GS3 a 3.75 (111) 3.57 (105) 3.51 (104) 4.34 (128) 3.96 (117) 3.82 (113)GS4 a 3.82 (113) 3.61 (106) 3.54 (104) 4.46 (132) 4.06 (120) 3.90 (115)GS5 a 3.88 (114) 3.65 (108) 3.57 (105) 4.56 (135) 4.14 (122) 3.97 (117)Genomic selection applied to males onlyGS1a 3.48 (103) 3.43 (101) 3.41 (101) 3.83 (113) 3.64 (107) 3.59 (106)GS2 a 3.54 (104) 3.46 (102) 3.43 (101) 4.02 (119) 3.76 (111) 3.68 (109)GS3 a 3.60 (106) 3.49 (103) 3.45 (102) 4.15 (122) 3.85 (114) 3.75 (111)GS4 a 3.64 (107) 3.51 (104) 3.47 (102) 4.24 (125) 3.93 (116) 3.81 (112)GS5 a 3.69 (109) 3.54 (104) 3.49 (103) 4.31 (127) 4.00 (118) 3.87(114)

aGS 1 to 5 refers to training population sizes of 1000, 2000, 3000, 4000 and 5000.bNe, Effective population size.

Maternal index: Under conventional selection, the economic response was much lower (8% less) with the YSP compared with the CBP. This was because of the higher accuracy of breeding values of older sires due to the availability of phenotypic records on more relatives. Incorporating genomic information on both males and females increased the economic response to selection in both breeding structures considered in this study; however the magnitude of response was much higher in the YSP, up to 32% greater than the base index of the current structure compared with up to 21% improvement that was achieved when applying genomic selection to the CBP. Incorporating genomic information of only males also increased the economic response to selection however to a smaller extent than incorporating genomic information of both sexes. The CBP achieved an improvement of up to 13% whereas the YSP was associated with an improvement of up to 23% compared with the base index.

Table 4.5 Total economic response in £/animal (%) of different breeding programmes based on conventional selection and genomic selection of the current beef maternal index.

CBP YSPConventional Selection 4.19 (100) 3.86 (92) Ne100b Ne300b Ne100b Ne300b

When GS is applied to males and femalesGS1a 4.72 (113) 4.47 (107) 4.97 (119) 4.46 (106)GS2 a 5.07 (121) 4.72 (113) 5.55 (132) 4.93 (118)When GS only applied to malesGS1a 4.51 (108) 4.36 (104) 4.71 (112) 4.32 (103)GS2 a 4.75 (113) 4.52 (107) 5.14 (123) 4.68 (112)

aGS 1 and 2 refer to training population sizes of 2000 and 5000, respectively.bNe, Effective population size.

The size of the training population influences the economic response that can be achieved when including genomic information. Increasing the training population size was associated with a linear improvement in total economic response, where the highest response was achieved with the highest training population size of 5000. However, this was constrained by the breeding program structure and Ne. Larger training population sizes had more impact when younger sires were selected. Furthermore, the rate of economic response was higher for Ne of 100 than 300 or 500.

4.1.2.2 Scenario 2. Base index + new traits included as profit and recorded traits

Terminal: When no selection pressure was applied to the new profit traits, a slightly higher response was achieved with the YSP (+1%) compared with the CBP under conventional selection. In contrast, when direct selection pressure was applied to all traits except SF and DS, the YSP performed less well (-6%) than the CBP. In this scenario, the lower generation interval of younger sires did not compensate for the increased information sources of relatives of older sires in the CBP. When none of the new traits were subject to direct selection no improvement is observed from applying genomic selection. This is mainly because genomic selection is having an unfavourable impact on RFI which cancels out any improvement achieved in the other profit traits. Considerable

SID 5 (Rev. 3/06) of 29

improvements in total economic response relative to the base index were achieved from applying genomic selection to the new index when RFI and survival are also subject to selection pressure (non zero economic weights). When genomic selection was applied to both males and females improvements of up to 80% (Figure 4.1) were observed and up to 66% when genomic selection was only applied to males (Figure 4.2). When direct selection pressure was applied to RFI and survival traits YSP performed better than the CBP across all genomic selection scenarios. A considerable improvement in economic response of up to 80% was achieved in the YSP compared with the CBP (56%) when genomic selection was applied to both sexes (Figure 4.1).

100

127144

156

135143

121

158

180

140156

0

20

40

60

80

100

120

140

160

180

200

BLUP BASE

BLUP NEW

Ne100 P2000

Ne100 P5000

Ne300 P2000

Ne300 P5000

Econ

omic

resp

onse

per

ann

um (%

)

CBP

YSP

Genomic Selection

Figure 4.1 Total economic response (%) of the terminal sire index when new traits were incorporated as recorded and profit traits with direct selection pressure on RFI and survival traits where genomic selection was applied to both males and females relative to the BLUP base scenario (current index adjusted for the correlated response in the new traits).

100

127137

145132 137

121

150

166

136149

0

20

40

60

80

100

120

140

160

180

BLUP BASE

BLUP NEW

Ne100 P2000

Ne100 P5000

Ne300 P2000

Ne300 P5000

Econ

omic

resp

onse

per

ann

um (%

)

CBP

YSP

Genomic Selection

Figure 4.2 Total economic response (%) of the terminal sire index when new traits were incorporated as recorded and profit traits with direct selection pressure on RFI and survival traits where genomic selection was applied only to males relative to the BLUP base scenario (current index adjusted for the correlated response in the new traits).

Maternal: When no selection pressure was applied to the new profit traits, a higher economic response was achieved with the CBP (+19%) compared with the YSP under conventional selection. When direct selection pressure was applied to all traits except SF and DS, the CBP outperformed the YSP to an even greater extent (+21%) . The lower generation interval of younger sires did not compensate for the increased information sources of relatives of older sires in the CBP. A considerable improvement in total economic response was achieved from applying genomic selection to both males and females: up to 18% improvement when none of the new traits were subject to direct selection and up to 56% improvement when direct selection pressure was applied to RFI and survival traits (Figure 4.3) in the CBP. When genomic selection was applied to only males an improvement of up to 11% was achieved when none of the new traits were subject to direct selection and up to 43% improvement when direct selection pressure was applied to RFI and survival traits (Figure 4.4) in the CBP. When direct selection pressure was applied to RFI and survival traits, the YSP performed better than the CBP across all genomic scenarios (when applied to males and females) except for Ne 300 and P2000 where the CBP performed better (+9%) (Figure 4.4). When genomic selection was only applied to males the YSP outperformed the CBP for Ne100 and P5000 (+9%) and the YSP performed less well than CBP for Ne300 and P2000 (-9%) (Figure 4.4).

100

121

143156

133143

100

146

168

125

144

0

20

40

60

80

100

120

140

160

180

BLUP BASE

BLUP NEW

Ne100 P2000

Ne100 P5000

Ne300 P2000

Ne300 P5000

Econ

omic

resp

onse

per

ann

um (%

)

CBP

YSP

Genomic Selection

Figure 4.3 Total economic response (%) of the maternal index when new traits were incorporated as recorded and profit traits with direct selection pressure on RFI and survival traits where genomic selection was applied to both males and females relative to the BLUP base scenario (current index adjusted for the correlated response in the new traits).

100

121

135143

128134

100

135

152

119

134

0

20

40

60

80

100

120

140

160

BLUP BASE

BLUP NEW

Ne100 P2000

Ne100 P5000

Ne300 P2000

Ne300 P5000

Econ

omic

resp

onse

per

ann

um (%

)

CBP

YSP

Genomic Selection

Figure 4.4 Total economic response (%) of the maternal index when new traits were incorporated as recorded and profit traits with direct selection pressure on RFI and survival traits where genomic selection was applied only to males relative to the BLUP base scenario (current index adjusted for the correlated response in the new traits).

SID 5 (Rev. 3/06) of 29

Both breeding programs performed the same for the remaining genomic selection scenarios (when only applied to males). A considerable improvement of 68% was achieved with genomic selection (Ne100, P5000) in the YSP compared with the 56% that was achieved with the CBP under the same scenario (Figure 4.3). This is slightly less when genomic selection was only applied to males (Figure 4.4).

In the same way as was observed with the base index, the highest response was achieved with a lower Ne (100) and a higher number of phenotypic records in the training population (5000). In the figures above two of the genomic selection strategies performed similarly (i) Ne100, P2000 and (ii) Ne300, P5000. The negative impact of having a larger effective population size in (ii) was compensated for by having a larger number of recorded phenotypes in the training population.

4.2 Sheep indexes (terminal sire, longwool, and hill sheep indexes)

4.2.1 Model descriptionSelection criteria and breeding goals: The base indexes were constructed to mimic the current terminal sire, longwool, and hill index using information available from the UK national genetic evaluations data. A second index was created by adding the new traits to the selection criteria and breeding goals including. Detailed information about the current and new breeding goal traits in Table 4.6 to 4.8.

Table 4.6 Heritability, phenotypic variance and economic weights for current and new profit traits in the terminal sire sheep selection index.

Trait1 Heritability Phenotypic variance Economic weightLWt (kg) 0.280 3.31 2.660FWt (kg) 0.290 4.84 -1.760RFI-lamb 0.300 83.0 -0.056Lamb survival (yes/no) 0.030 0.090 15.561: LWt: lean weight (kg), FWt: Fat weight (kg), RFI-lamb: residual feed intake for growing lambs over 4 months growing period (kg DMI/head), and lamb survival as binary trait (yes/no).

Sources of information: The number of records for each trait on the candidate and its relatives (mother, progeny, paternal half sibs, and mother’s sisters) were estimated based on pedigree and performance data from EGENES. Data for one breed from each sector (Texel for terminal, Bluefaced for longwool and Blackface for hill) were used. The expected response to selection was estimated for males and females separately. Two breeding program structures were considered: (i) the current breeding programme (CBP) in which rams are used for service for multiple seasons, and (ii) a young ram programme (YRP) in which rams are used for breeding for one year only.

Table 4.7 Heritability, phenotypic variance and economic weights for current and new profit traits in the longwool sheep selection index.

Trait1 Heritability Phenotypic variance Economic weightSLage (kg) 0.230 2784 -0.068Conf15 (score) 0.120 1.120 0.550LNWt (kg) 0.360 1.050 5.870MS (kg) 0.450 47.80 -0.500LSB- D (count) 0.050 0.300 4.700RFI-Lambs (kg) 0.300 75 -0.067RFI-Ewes (kg) 0.300 1228 -0.050Longevity (years) 0.080 1.51 11.20Footrot (incidence) 0.045 0.149 -0.196FECS (log count) 0.150 0.800 0FECN (long count) 0.200 0.300 0Lamb Survival (yes/no) 0.030 0.090 38.821: SLage: slaughter age in days, Conf15: Conformation score, MS: mature size (kg), LSB-D: direct litter size at birth (count/year), RFI-Ewe: residual feed intake of breeding ewes for one year (kg DMI/head/year), Longevity: Ewe longevity in years, Footrot: footrot incidence (yes/no), FECS: Strongyle faecal egg count (log of count), FECN: Nematode faecal egg count (log of count), the other traits are defined before in Table 4.6

SID 5 (Rev. 3/06) of 29

Table 4.8 Heritability, phenotypic variance and economic weights for current and new profit traits in the hill sheep selection index.

Trait1 Heritability Phenotypic variance Economic weightCurrent index traitsCFat (kg) 0.25 0.477 -0.1393CMus (kg) 0.35 0.6607 0.5507MS (kg) 0.43 23.522 -0.0522MAT (kg) 0.1 12.806 0.1900LSR (number/year) 0.05 0.566 0.09518WK (kg) 0.12 6.384 0.3840New traits in the indexRFI-Lambs (kg) 0.3 85 -0.0349RFI-Ewes (kg) 0.3 809 -0.0218Ewe longevity (years) 0.03 0.78 0.664Footrot (yes/no) 0.045 0.149 -0.0001FECS (log of count) 0.15 0.8 0.0FECN (log of count) 0.20 0.3 0.0Lamb survival (yes/no) 0.03 0.09 0.45861: MAT: litter wt weaned, LSR, litter size reared (number/year), 8WK: eight week weight, the other traits are defined before in Tables 4.6 and 4.7.

Breeding program structure: The proportions of male and female candidate selected were estimated from the pedigree data as the actual proportion of animals born who appeared later on in the pedigree file as a parent for at least one time. For terminal sire sheep, 8.08% of the males born were found to become sires and for females, 45.37% of the females born became dams. For longwool, 7.0% of the males born were found to become sires and for females, 40.15% of the females born became dams. For hill sheep, 3.84% of the males born were found to become sires and for females, 45.3% of the females born became dams. Similarly, generation interval for males and females was estimated from the pedigree as the average age of the parents when their progeny were born. The estimate of generation interval was 3.02 years for sires and 3.56 years for females in terminal sire, 2.52 years for sires and 3.44 years for females for long wool, and 2.3 years for sires and 3.63 years for females of hill sheep. For the young ram programme, the generation interval for males was assumed to be 1.5 years.

Scenarios simulated: Selection index theory was applied to predict genetic and economic response to selection for different traits and under different sources of information scenarios. The scenarios evaluated in the simulation were:

Scenario 1: Base index (+ correlated response in the new traits): investigating the current selection index (no direct selection pressure on the new traits). Both breeding structures were considered (CRP and YRP). The base index was investigated using conventional selection (BLUP) alone and conventional plus genomic selection. For the genomic selection effective population sizes of Ne 200 and 500 and training population size (phenotypes and genotypes recorded) of 2000 and 5000 were considered. In this scenario, genomic selection was considered for both males and females as well as for males only as it may be too expensive to genotype both sexes.

Scenario 2: Base index + new traits included as profit and recorded traits provided they can be recorded (direct selection pressure was applied to the new traits). In this scenario conventional selection of both breeding program structures were considered (CRP and YRP). Genomic selection was applied to these programmes, where effective population sizes of Ne 200 and Ne 500 and number of recorded phenotypes of 2000 and 5000 were considered. Genomic selection was considered for both males and females as well as for males only.

Modelling the addition of GBV: Genomic information was included in the selection index model based on the theory of Dekkers et al. (2007). Accuracies of GBVs are predicted based on trait heritability, number of phenotyped animals in the training population, the number of QTL underlying the trait and the effective population size (Ne) (Daetwyler et al. 2008; Goddard 2009).

4.2.2 Results

4.2.2.1 Scenario 1: Base index (+correlated responses in the new traits)

The total economic responses of the two breeding programmes based on conventional selection and the economic responses of different genomic selection strategies relative to the current ram scenario (CRP) under conventional selection is presented in Table 4.9 for the terminal sire index, Table 4.10 for the longwool index and

SID 5 (Rev. 3/06) of 29

Table 4.11 for the hill sheep index. For genomic selection, the results are presented when both males and females are assumed to be genotyped and for the scenarios when assuming that only males are genotyped.

Terminal sire index: Genomic information can have a remarkable favourable effect on the overall annual expected response to selection. The expected gain when including GBV for both males and females in the CRP was in the range of 10 to 33% when the new traits were not directly selected for (correlated response). The potential improvement due to GBV was even higher in the YRP, between 19 to 65%. The overall response to selection when genotyping only the males was still higher than the conventional breeding system but of course less effective than genotyping both males and females. The response to selection when only males were genotyped in the CRP was 5 to 17% more effective. Applying genomic selection through males only in YRP can still result in 12 to 45% better response.

Table 4.9 Total economic response in £/animal (%) of different breeding programmes based on conventional selection and genomic selection of the current sheep terminal sire index (+ correlated response in the new traits).

CRP YRPConventional Selection 0.627 (100) 0.587 (94) Ne200b Ne500b Ne200b Ne500b



Longwool index: Comparing the CBP with the YRP, both indexes performed similarly under conventional breeding system in the base longwool index (+ correlated response in new traits). This shows that the effect of loss of information in the YRP (as young rams have less number of recorded relatives) was about the same as the gain due to shorted generation interval in YRP. The use of genomic selection was shown to have the potential to accelerate the overall response to selection. When assuming that both males and females were genotyped in a CBP, the overall response to selection was expected to be 15 to 48% higher. The expected improvement due to GBV was even higher with the YRP, up to 69%. Although less effective than genotyping males and females, the overall response to selection when genotyping only the males was still higher than the conventional breeding system with no direct selection on the new traits (base index).

Table 4.10 Total economic response in £/animal (%) of different breeding programmes based on conventional selection and genomic selection of the current sheep longwool index (+ correlated response in the new traits).

CRP YRPConventional Selection 0.602 (100) 0.596 (99) Ne200b Ne500b Ne200b Ne500b



Hill sheep index: Applying genomic information had a favourable impact on the hill sheep index. The expected gain when including GBV for both males and females in the CRP was in the range of 3 to 15%. For the YRP, the potential of improvement was even higher and was up to 20%. Genotyping the males only in CBP was still more effective than conventional breeding alone. However, the improvement was smaller compared with genotyping both males and females. The YRP superseded the CRP when using GBV when the training population was large (5000) and the effective population size was small (2000).

Table 4.11 Total economic response in £/animal (%) of different breeding programmes based on conventional selection and genomic selection of the current hill sheep index (+ correlated response in the new traits).

CRP YRPConventional Selection 0.202 (100) 0.178 (88)

SID 5 (Rev. 3/06) of 29

Ne200b Ne500b Ne200b Ne500b



4.2.2.2 Scenario 2. Base index + new traits included as profit and recorded traits (selection pressure on new traits)

Terminal: The overall annual economic responses of different scenarios and as percentage relative to the base index (+correlated responses in the new traits) are shown in Figures 4.4 and 4.5. For genomic selection, the results are presented when both males and females are assumed to be genotyped (Figure 4.4) and for the scenarios when assuming that only males are genotyped (Figure 4.5). There will be a slight gain in overall economic response due to indirect and direct selection on the two new profit traits in the index and using BLUP evaluation (EBV). The correlated response due to adding the new traits with no direct selection pressure was about 1% for both the CRP and the YRP using EBV. When applying direct selection pressure on the two new profit traits in the index, the expected overall economic response was 2% over the base index for both the CRP and the YRP. The expected overall annual economic response using EBV was about 6 to 7% better for the CRP compared with the corresponding indexes of YRP. This shows that the effect of loss of information in the YRP (as young rams have less number of recorded relatives) was much higher than the gain due to faster breeding turnaround due to lower generation interval in the YRP. Genomic information can have a remarkable favourable effect on the overall annual expected response to selection. The expected gain when including GBV for both males and females in the CRP was in the range of 13 to 36% when the new traits were selected for directly (using non-zero economic weights in the index) (Figure 4.4). The effect of GBV alone when the new profit traits were subjected to direct selection was in the range of 10 to 34% better selection response compared with the conventional selection when the new profit traits were directly selected for. The potential improvement due to GBV was even higher in the YRP, between 21 to 68% when the new traits were directly selected for (Figure 4.4). The response to selection when only males were genotyped in the CBP was 8 to 20% better when the new traits were subjected to direct selection and correspondingly in the YRP 14 to 47% improvement was achieved when the new traits were directly selected for (Figure 4.5). The effectiveness of genomic selection was different among different training and effective population sizes. The effect of the size of the training population was substantial on the overall expected economic response when using GBV and its effect was more apparent when the effective population size was smaller.

100 102

122136

113123

95

141

168

121

143

0

20

40

60

80

100

120

140

160

180

BLUP BASE

BLUP NEW

Ne200 P2000

Ne200 P5000

Ne500 P2000

Ne500 P5000

Econ

omic

resp

onse

per

ann

um (%

)

CRP

YRP

Genomic Selection

Figure 4.4 Total economic response (%) of the terminal sire index when new traits were incorporated as recorded and profit traits with direct selection pressure on RFI and lamb survival where genomic selection was applied to both males and females relative to the BLUP base scenario (current index adjusted for the correlated response in the new traits).

100 102112

120108 113

95

128

147

114

130

0

20

40

60

80

100

120

140

160

BLUP BASE

BLUP NEW

Ne200 P2000

Ne200 P5000

Ne500 P2000

Ne500 P5000

Econ

omic

resp

onse

per

ann

um (%

)

CRP

YRP

Genomic Selection

Figure 4.5 Total economic response (%) of the terminal sire index when new traits were incorporated as recorded and profit traits with direct selection pressure on RFI and lamb survival where genomic selection was applied only to males relative to the BLUP base scenario (current index adjusted for the correlated response in the new traits).

Longwool: The overall annual economic responses of different scenarios and as percentage relative to the base index (+correlated responses in the new traits) are shown in Figures 4.6 and 4.7. The expected overall economic response was exceeded in the base index when applying direct selection pressure on the new profit traits in the index. Genomic information combined with direct selection can result in 46 to 81% faster selection response compared with the base index in the CRP. The expected improvement due to GBV was even higher with the YRP more than 2 folds higher when the new traits were directly selected for (compared with the base index) (Figure 4.6). Although not as effective as genotyping both males and females, when the traits were directly selected for

SID 5 (Rev. 3/06) of 29

the expected gain in YRP with genotyping only the males was 2.8 to 3.3 times more effective than the base index and it was also more effective than conventional breeding in YRP when the new traits were directly selected for (Figure 4.7). The results showed that genomic selection has high potential to accelerate response to selection under both the CBP and YRP (more effective) even if only the male candidates were genotyped.

100121

155

181

146

168

123

173

209

150

190

0

50

100

150

200

250

BLUP BASE

BLUP NEW

Ne200 P2000

Ne200 P5000

Ne500 P2000

Ne500 P5000

Econ

omic

resp

onse

per

ann

um (%

)

CRP

YRP

Genomic Selection

Figure 4.6 Total economic response (%) of the longwool index when new traits were incorporated as recorded and profit traits with direct selection pressure on RFI, longevity, footrot and lamb survival where genomic selection was applied to both males and females relative to the BLUP base scenario (current index adjusted for the correlated response in the new traits).

100

121

138

155

128

147

123

157

182

141

169

0

20

40

60

80

100

120

140

160

180

200

BLUP BASE

BLUP NEW

Ne200 P2000

Ne200 P5000

Ne500 P2000

Ne500 P5000

Econ

omic

resp

onse

per

ann

um (%

)

CRP

YRP

Genomic Selection

Figure 4.7 Total economic response (%) of the longwool index when new traits were incorporated as recorded and profit traits with direct selection pressure on RFI, longevity, footrot and lamb survival where genomic selection was applied only to males relative to the BLUP base scenario (current index adjusted for the correlated response in the new traits)

Hill Sheep: The overall annual economic responses of different scenarios and as percentage relative to the base index (+correlated responses in the new traits) are shown in Figures 4.8 and 4.9. Applying direct selection pressure on the new traits using conventional breeding did not improve the expected overall annual response. Comparing the CBP with the YRP using conventional EBV selection showed that the expected overall annual response was 12% better with the CRP compared with using YRP. The inclusion of genomic information resulted in an increase in expected overall annual economic response. The expected gain when including GBV for both males and females in the CBP was between 9 to 29% when the new traits were selected for directly (using non-zero economic weights in the index) (Figure 4.8). For the YRP, the potential of improvement was even higher, up to 38% when the new traits were directly selected for (Figure 4.8). Genotyping the males only in CBP was still more effective than conventional breeding alone (Figure 4.9). However, the improvement was smaller compared with genotyping both males and females. The YRP superseded the CBP when using GBV when the training population was large (5000) and the effective population size was small (2000). The opposite picture can be seen when the training population was small (2000) and the effective population size was large (500). In this case the response to the CBP would be larger than that from the YRP when either only males were genotyped or both sexes were genotyped.

100 100

116129

109117

88

117

138

104

119

0

20

40

60

80

100

120

140

160

BLUP BASE

BLUP NEW

Ne200 P2000

Ne200 P5000

Ne500 P2000

Ne500 P5000

Econ

omic

resp

onse

per

ann

um (%

)

CRP

YRP

Genomic Selection

Figure 4.8 Total economic response (%) of the hill sheep index when new traits were incorporated as recorded and profit traits with direct selection pressure on RFI, longevity, footrot and lamb survival where genomic selection was applied to both males and females relative to the BLUP base scenario (current index adjusted for the correlated response in the new traits).

100 100110

118

105111

88

109

124

100110

0

20

40

60

80

100

120

140

BLUP BASE

BLUP NEW

Ne200 P2000

Ne200 P5000

Ne500 P2000

Ne500 P5000

Econ

omic

resp

onse

per

ann

um (%

)

CRP

YRP

Genomic Selection

Figure 4.9 Total economic response (%) of the hill sheep index when new traits were incorporated as recorded and profit traits with direct selection pressure on RFI, longevity, footrot and lamb survival where genomic selection was applied only to males relative to the BLUP base scenario (current index adjusted for the correlated response in the new traits).

5. ECONOMIC AND ENVIRONMENTAL IMPACT OF GENOMIC SELECTION AND NEW BREEDING GOAL TRAITS AT THE INDUSTRY LEVEL FOR THE BEEF CATTLE AND SHEEP SECTORS

5.1 Introduction

SID 5 (Rev. 3/06) of 29

The evaluation of the responses from the economic selection indexes indicated increases in the economic benefits from including residual feed intake (RFI) and survival traits in the breeding goal and the utilization of genomic selection (GS) in the beef cattle sector. These traits, in addition to ewe longevity and genetic resistance to footrot, showed to have also favourable effects in the sheep breeding programmes. However, the evaluation of different possible GS strategies should consider improved responses at national scale and accounting for training population costs. The results of section 4 showed the relevance of the effective population (Ne) and training population (TP) sizes on the magnitude of the responses, given their impact on the accuracy with which genomic breeding values are estimated. This clearly indicates that training population size should be considered when investigating optimal GS strategies. The aims of this study were to (i) investigate the economic impact at the whole beef and sheep industry level of improving efficiency thorough the inclusion of new traits and the implementation of GS and (ii) assess the effect on GHG mitigation.

5.2 Methodology

5.2.1. Breeding structures, selection strategies and breeding goalsFive sectors were considered in sheep and beef production, each has a specific breeding programme: maternal and terminal beef sector and hill, longwool and terminal sheep sectors. Two breeding goals were evaluated in each of these five sectors. These breeding goals were the current one and an extended breeding goal with additional traits. These new traits are associated with production efficiency and/or GHG emissions as described in section 2.

Profits at the whole industry level were calculated for the two selection strategies (EBV selection and genomic selection) but considering the current breeding programme (CBP) and an alternative young sire breeding programme (YSP), in which only young bulls or rams were used as breeding sires for one year. Ages of selection candidates, proportion selected and recorded trait information modelled in the selection index equations were based on average industry practice as determined from a summary analysis of the national database used in the UK for the genetic evaluations of beef cattle and sheep. The selection strategies were the current one based on EBV and genomic selection in which EBV and GBV were combined. For the “traditional” selection (only EBV) indices used on the current criteria were applied for males and females. However, in the case of genomic selection the genomic information was only used for male selection. Due to the lower selection intensity in females their contribution to the total genetic gain is of small magnitude and does not justify the use of high density SNP panels given the high cost of this technology at present. Female cattle were selected based on EBV for current and survival traits, whilst ewes were selected for current traits, as well as lamb survival and longevity. Selection in males took into account RFI but only as GBV in both cattle and sheep. Selection of rams also considered GBV for footrot. Other traits such as beef shear force, cattle docility and genetic resistance to internal parasites in sheep, were not considered in this study. Because no economic or GHG values were assigned to them, they do not contribute to the calculation of profit or GHG emissions. In the case of GS, the same combinations of Ne and TP sizes considered in the previous chapter were assumed.

Based on the combinations of different breeding structures, selection strategies, breeding goals, Ne and TP sizes a maximum of 20 scenarios were investigated. However, the results of a reduced number of scenarios will be presented here (Figure5.1). These are the result of considering the expanded breeding goals (current breeding goal traits plus new traits) in the current breeding structure. Responses were lower in the case of the current breeding goal, whilst the modification of the breeding structure to the young sire one resulted in small increases of responses in some cases. The reference for comparisons is the current system in place.

5.2.2 Genetic responsesSelection index theory was applied to predict the genetic responses that were used for the calculation of the economic profit and GHG emissions for the whole industry. Detailed explanations of the methodology and parameters used were given in section 4.

5.2.3 Economic evaluation at industry levelCumulative marginal net discounted return from 10 years of selection with benefits considered over a 20-year horizon was the primary criterion for comparison of the different scenarios. Discounted incomes were calculated for each of the goal traits based on the annual genetic gain in the trait units and their economic values discounted by the specific genetic expression coefficients considering time and number of expressions of the genetic progress, and the number of sires from the breeding programmes required to mate the industry females, which varies according to the level of technology assumed in each sector (100% in maternal and terminal beef cattle; 23% in hill sheep; 9% in longwool sheep; 15% in terminal sheep, Amer et al. 2007). A discount rate of 7% was used when discounting genetic expressions of goal traits over time. The numbers of bulls and rams used were based on the information presented by Amer et al. (2007).

SID 5 (Rev. 3/06) of 29

Figure 5.1 Diagram of the scenarios considered for the beef and sheep breeding programmes. Ne values are those used in beef cattle. In the case of sheep the Ne sizes were 300 and 500 animals

The costs included were: (1) the cost of genotyping the selection candidates (£200, £100 or £50/sample) during the 10 years of selection discounted at 7%, and (2) the costs associated with the TP, which included genotyping and RFI measurements (£1700/animal). No additional costs of recording survival were accounted for as the information is already available in the current databases. Other costs associated with the implementation were considered similar to the current breeding programme and therefore not added to the total cost.

5.2.4 GHG emissionsThe volumes of GHG emissions were calculated taking into account the genetic gains, discounted genetic expression coefficients and the estimates of emissions of CO2 equivalent per unit of change per breeding animal (Wall et al. 2010). These estimates, which considered methane and nitrous emissions, were obtained from the modelling of UK beef production system to calculate the GHG emissions when production traits were modified. Not all traits were considered due to modelling limitations and lack of information on the associations among traits (Wall et al. 2010).

5.3 Results and discussion

5.3.1 Beef cattle5.3.1.1 Discounted incomes and GHG mitigation in beef Table 5.1 shows the discounted income and GHG emissions for the current sheep and beef breeding programmes based on the current breeding structure, breeding goals and selection strategy, which are the reference point of comparison.

Table 5.1 Net income and GHG emission estimated for the current sheep and beef breeding programmes CURRENT BREEDING GOAL AND STRUCTURE – SELECTION BASED ON EBV

Cattle SheepMaternal Terminal Hill Longwool Terminal

Net income (000 £) 62,580 23,846 5,957 2,406 9,453GHG emissions (000 ton CO2e) -1,543 -508 -137 -103 - 7

Table 5.2 shows the discounted income values of the different scenarios for current breeding programme, which are presented in absolute terms (£) and as proportionate increases relative to the net returns for the current breeding programme. The inclusion of the survival traits in the current breeding program with selection based on EBVs increased significantly the net returns in both maternal and terminal breeding programmes. This improvement was larger in the maternal than in the terminal breeding programme given a relative higher expression of this trait in the maternal system in which survival at birth and weaning were added as new breeding goals (54% vs. 28%).

The results clearly show that large net incomes were obtained when genomic selection was applied simultaneously with the inclusion of survival and RFI in the breeding goals in both maternal and terminal breeding programmes. The increase in the response varied from 62% to 80% and 33% to 47% in maternal and terminal breeding programmes, respectively. The higher responses in the maternal breeding programmes may be associated with a relative larger response in the traits of lower heritability (Daetwyler et al. 2008; Goddard 2009). The maternal breeding goal traits tend to have lower heritabilities than those in the terminal breeding goal. The

SID 5 (Rev. 3/06) of 29

CURRENT BREEDING PROGRAMME STRUCTURE

EBV

Current breeding goal + new traits

EBV

Ne=100 TR=2000

Ne=100 TR=5000

Ne=300 TR=2000

Ne=300 TR=2000

EBV+GEBV

Current breeding goal


EBV


EBV

Ne=100 TR=2000

Ne=100 TR=5000

Ne=300 TR=2000

Ne=300 TR=2000

EBV+GEBV


EBV


EBV

Ne=100 TR=2000

Ne=100 TR=5000

Ne=300 TR=2000

Ne=300 TR=2000

EBV+GEBV

Current breeding goal

implementation of GS will also increase of information contributed by GBV from relatives/progeny faster than with BLUP evaluation. This will have a relative more favourable effect on accuracies in life traits such as lifespan which are in the maternal breeding goal. As previously reported (i.e. Daetwyler et al. 2008; Goddard 2009), smaller Ne and larger TP lead to higher responses to genomic selection. Although the difference is small the increase tended to be larger in the maternal than in the terminal breeding scheme.

Table 5.2 Net incomes for the expanded breeding goals under current selection (EBV) and genomic selection in maternal and terminal current breeding programme structures

EBVGenomic selection

Ne =100 Ne = 300TP = 2000 TP = 5000 TP = 2000 TP = 5000

Maternal breeding programmeNet income (£ 000) 96,125 106,352 112,621 101,328 105,918Relative value2 1.54 1.70 1.80 1.62 1.69GHG (000 ton CO2e) - 1,731 - 1,845 - 1,863 - 1,806 - 1,843Relative value2 1.12 1.20 1.21 1.17 1.19Terminal breeding programmeNet income (£ 000) 30,610 33,099 34,996 31,795 32,980Relative value2 1.28 1.39 1.47 1.33 1.38GHG (000 ton CO2e) - 451 - 467 - 480 - 459 - 466Relative value2 0.89 0.92 0.95 0.90 0.92

1 Level of adoption of genetic evaluations assumed in maternal and terminal beef cattle to be 100% (Amer et al., 2007); 2 Relative to the net income or GHG volume obtained in the current breeding programme with the current breeding goals (absolute values in Table 5.1)

In agreement with the results reported by Moran et al. (2008) genetic improvement contributes to GHG mitigations (Table 5.2). To facilitate the comparison, the mitigation levels were also expressed relative to the mitigation of the current maternal and terminal breeding programmes. Smaller impacts were observed in terms of mitigation compared to net income in the case of the maternal breeding programmes. In the terminal breeding programmes the beneficial effect was reduced. This may be due to the fact that survival traits increase GHG emissions when GHG values were estimated per breeding female (Wall et al. 2010).

5.3.1.2 Costs in beefThe cost of implementing GS in the UK beef sector following the previous assumption of 100% penetration of improved genetics can be seen in Table 5.3. The cost is broken down into the training population component, either with or without creating RFI phenotypes in the training population, and the ongoing genotyping costs associated with implementing GS. The cost of implementation is considered for a genotyping cost of either £200, £100 or £50 to provide a range of possible future scenarios.

Table 5.3 Cost of implementing genomic selection within the UK terminal sire and maternal beef sectors.Genotyping cost per animal (£)200 100 50

Scheme Cost (£000’s)Training population 2000 animals (no RFI) 1000 500 250Training population 5000 animals (no RFI) 400 200 100Training population 2000 animals, RFI recorded 9500 9000 8750Training population 5000 animals, RFI included 3800 3600 3500Ongoing genotyping for 10 years of terminal sire selection 5480 2740 1370Ongoing genotyping for 10 years of maternal sire selection 17800 8900 4450

For the same scenarios net returns were much more significant in the maternal than in the terminal beef breeding programmes. The largest net returns were obtained by a training population of 5000 animals when the Ne was 100. However, in the case of Ne of 300 animals the net returns were larger with the smaller TP. Reducing the cost of genotyping increased net returns but the relative ranking of the scenarios remains the same.

5.3.2 Sheep5.3.2.1 Discounted incomes and GHG mitigation in sheepThe discounted income values and volume of GHG emissions of the different scenarios in sheep are presented in Table 5.4. Under the current breeding programme structure and selection strategy, the inclusion of survival and ewe longevity in the breeding goals increased slightly the net incomes in the hill and terminal sheep (4%-5%) but had a more significant effect in the longwool breeding programme (18%). The inclusion of genomic selection and residual feed intake in the breeding programmes had a favourable effect in the three sheep sectors. The largest

SID 5 (Rev. 3/06) of 29

and lowest relative responses were observed in the longwool and hill breeding programmes, whilst the values were intermediate in the terminal sheep. In the current breeding structure the ranges were 30% - 56%, 4% - 15% and 9% - 22% for the longwool, hill and terminal sheep, respectively (Table 5.4).

Table 5.4 Net incomes for expanded breeding goals under current selection (EBV) and genomic selection in the hill, longwool and terminal sire current breeding programme structures

Hill Sheep1 EBVGenomic selection

Ne = 200 Ne = 500TP = 2000 TP = 5000 TP = 2000 TP = 5000

Net income (£ 000) 6,031 6,440 6,846 6,190 6,473Relative value2 1.05 1.08 1.15 1.04 1.09GHG (000 ton CO2e) - 145 - 187 - 239 - 132 - 193Relative value2 1.05 1.36 1.74 0.96 1.40Longwool sheep1

Net income (£ 000) 2,852 3,358 3,751 3,116 3,535Relative value2 1.18 1.40 1.56 1.30 1.47GHG (000 ton CO2e) - 54 - 69 - 78 - 62 - 70Relative value2 0.52 0.67 0.76 0.60 0.68Terminal sire sheep1

Net income (£ 000) 9,861 10,764 11,491 10,326 10,822Relative value2 1.04 1.14 1.22 1.09 1.14GHG (000 ton CO2e) - 16 - 33 - 40 - 25 - 33Relative value2 2.19 4.47 5.45 3.36 4.58

1 Levels of adoption of genetic evaluations assumed were 23%, 9% and 15% for hill, longwool and terminal sire sheep, respectively (Amer et al., 2007); 2 Relative to the net income or GHG volume obtained in the current breeding programme with the current breeding goals (absolute values in Table 5.1)

The responses in the terms of GHG mitigation to the inclusion of the new breeding goal traits were very different among sheep sectors in the current breeding structure when new traits were considered. In the case of the hill sheep a small favourable effect was observed, whilst the effect in relative terms was very significant in the terminal sheep. On the other hand, the contribution to GHG mitigation was reduced by 50% in the longwool sheep. The increase of GHG emission (reduction of mitigation) in the longwool sheep may be explained by the relative high importance of ewe longevity in the breeding goal which is also associated with an increase in the GHG emission according to Wall et al. (2010) when the modelling was on a female breeding base.

Genomic selection and RFI had a favourable effect on improving mitigation. In the case of the longwool sheep this effect was reducing the negative effect, probably due to a favourable effect of GS on RFI. The reduction decreased from 50% to values between 40% and 24% depending of the scenarios (combination of Ne and TP sizes). In the terminal and hill sheep breeding programmes the levels of improvement compared to the current scenarios were of very high and moderate magnitude, respectively. The large effect observed in the terminal sire sheep may be related to an important effect of the growth selection criteria that were considered for the calculation of the GHG emissions in addition to RFI. Relative lower influence was observed in the hill sheep. In this sector some of the relevant maternal traits are associated with larger gas emissions (Wall et al. 2010). However, the impact of these traits changes when emission are considered per unit of product.

5.3.2.2 Costs in sheep

The cost of implementing GS in the UK sheep sector follow the previous assumption that the levels of adoption of improved genetics were 23%, 9% and 15% for hill, longwool and terminal sheep, respectively (Amer et al. 2007), Table 5.5. The cost is broken down into the training population component and the ongoing genotyping costs associated with implementing GS. The cost of implementation is considered for a genotyping cost of either £200, £100 or £50 to provide a range of possible future scenarios.

Table 5.5 Cost of implementing genomic selection within the UK terminal sire and maternal beef sectors.Genotyping cost per animal (£)200 100 50

Scheme Cost (£000’s)Training population 2000 animals (no RFI) 1000 500 250Training population 5000 animals (no RFI) 400 200 100Ongoing genotyping for 10 years of terminal sire selection 14400 7200 3600

SID 5 (Rev. 3/06) of 29

Ongoing genotyping for 10 years of longwool sire selection 2700 1350 675Ongoing genotyping for 10 years of hill sheep selection 12000 6000 3000

The discounted net returns for the three sectors considering these costs were predominantly negative because the levels of net incomes were lower than the implementation and running costs.

5.4 Final commentsThe results presented here confirmed that genetic improvement resulted in increased economic results and GHG mitigation in both sheep and beef cattle sector.

The inclusion of new breeding goal traits such RFI and survival, which are related to production efficiency and GHG mitigation, as well as the implementation of GS showed in general a favourable environmental effects, which the exception of the longwool sheep sector. However, these results should be interpreted with caution given that some traits associated with higher efficiency and lower losses, such as ewe longevity and animal survival, had a negative influence on GHG mitigation when they are considered on breeding animal basis (Wall et al. 2010). This may underestimate possible effects on GHG mitigation particularly in those systems in which the maternal traits are particularly relevant. Results may be different if GHG emissions are evaluated by unit of product. The evaluation of the genetic programmes per unit of product may imply re-defining the current expression of the breeding goals which have defined until now on per breeding animal basis.

The incorporation of the innovations considered in this study improved the discounted incomes of all sheep and beef cattle sectors. Besides these favourable effects of GS and the new traits, the net returns showed that their implementation was not economically viable in all cases. In the sheep sectors the results indicated the economic return did not pay the genotyping costs given the breeding structures and the genotyping prices considered in this study. Given the favourable economic and environmental results that can be achieved, further studies on optimising breeding structures may be very valuable for the sheep industry. The inclusion of RFI and survival traits and the use of GS had favourable impact on net economic returns in both maternal and terminal beef breeding programmes. However, the magnitude of the discounted returns was not large enough in the terminal beef sector to compensate for the costs of the training population. In the case of the beef sector, the implementation of a training population following a multi-breed approach may allow integrating maternal and terminal breeds and therefore optimising the contributions from both sectors to the beef industry.

Economic and environmental contributions were affected by TP and Ne sizes. Although it is not possible to modify Ne, optimizing GS strategy in practice should consider the real undetermined beef cattle effective population sizes. Assuming responses possible with smaller Ne may lead to erroneous conclusions on biological and economic terms. Larger TP are related to higher biological responses and higher costs. However, in certain circumstances the higher accuracy obtained by larger TP sizes overcompensate the higher costs.

6. ECONOMIC EVALUATION OF A NATIONAL GENOTYPING SERVICE: THE UK BEEF INDUSTRY

An economic analysis of the impact of a genome wide selection scheme (GS) on the beef and sheep industries was conducted. This analysis was based on different implementation options (established in an earlier report), the results of an expert consultation and additional economic models developed for this report. The different implementation options are:Status Quo – No genome wide selection (GS) service. The status quo situation, addresses the scenario where nothing will happen from now on to encourage and support a GS service. Option 1 – Industry-driven GS. The first option is that an industry body leads a newly founded umbrella organisation to provide the services for a national genotyping service. All necessary tasks would be outsourced to specialised service providers, e.g. laboratory, IT, marketing and data storage. The main functions of the new organisation would be to liaise with relevant stakeholders in the service and to provide a single point of contact for all issues relating to the GS service. Option 2 – Commercial company-driven GS. The second option could either be seen as a single service provider-led option or a joint venture of commercial service providers, leading the national beef genotyping service. The beef industry would be included with a seat on the board and a primary characteristic of this option is that it would mainly be profit driven. Option 3 – Research provider-driven GS. The third option is a national genotyping service, led by a research provider. A research organisation would be at the forefront of the technology and could use this knowledge to bring the technology to the market.Option 4 – Joint Venture between industry and research provider. The fourth option is a joint venture between an industry body and a research provider.

SID 5 (Rev. 3/06) of 29

For the beef industry the results of the expert consultation provided feedback on a set of key assumptions and factors needed for the economic analysis. The following scenario was considered the most likely to structure the genetic breeding program - Ne = 300 animals, TP = 2000 animals, inclusion of RFI testing and cost of genotyping starts at £200 and reduces by £50 per year to a base of £50.For the sheep industry the increase in response due to utilising GS was averaged across sectors and assumed to be ~15%, with penetration of the technology expected to double from the current 17% to 34%.

These scenarios are compared to the “status quo” scenario to establish the additional benefit these scenarios provide. The status quo model reflects the current UK maternal breeding programs in which selection indices are based on EBV computed using Best Linear Unbiased Prediction statistical methodology.

The data on present value for the industry benefits, number of animals selected and genotyped, as well as the cost of genotyping were used as inputs in this economic analysis. The present values for beef included the farm profits (Tables 5.1 and 5.2) and the economic contribution of the GHG mitigation. The economic models also incorporated statistics on breed proportions of animals slaughtered obtained from BCMS data in order to take into account a multi-breed approach including terminal and maternal breeding sectors. Table 6.1 summarises the economic results expressed as net present values for the implementation options and scenarios described.

Table 6.1 Net present values from implementing GS service in the UK beef industry assuming all selection candidates genotypedImplementation Option Present Value (‘000s)

35% uptake 100% uptakeStatus Quo -£71 £255Industry body £5,540 £18,038Commercial company £6,759 £17,531Research Institute -£151 £6,699Industry and Research Institute Joint Venture £23,911 £59,437

The results in Table 6.1 illustrate that scenario 3 (especially with delivery option 4) is the most promising one. This positive result for option 4 in all scenarios is strengthened by the feedback from the expert consultation. The majority of the respondents value delivery option 4 as the “best option” and also rate option 4 as the option which scored the highest points in a matrix of options and attributes (simplicity of operation, one-stop-shop, cost effectiveness, future proofing and effective communication). For the sheep industry an assumption has been made that two stage selection will operate and only 30% of the selection candidates will be genotyped (Table 6.2). This approach could be applied to the beef industry which would further improve the expected net present value.

Table 6.2 Net present value from implementing GS service in UK sheep industry assuming 30% of selection candidates genotyped

Delivery Option Results (000s)Status Quo £3,662Industry body £62,127Option 2a: Commercial company – favourable £65,804Option 2b: Commercial company – unfavourable £22,423Research Institute £65,979Industry and Research Institute Joint Venture £69,359

7. GIN WORKSHOP

To fulfil the ‘GIN scoping’ parts of the present project, we have: Held meetings with the scientific co-ordinators of the oilseed rape and wheat GINs (OREGIN and WGIN)

at Rothamsted Research, to become more familiar with their aims, operation, strengths and weaknesses. Reported on current structures for genetic improvement of ruminants in Britain. Created and populated a Sharepoint site to collate information relevant to a potential ruminant GIN. Hosted a ‘GIN scoping’ workshop, with representatives from across the ruminant livestock-derived food

supply chain, livestock levy bodies, researcher providers and government departments.

SID 5 (Rev. 3/06) of 29

The GIN scoping Workshop was held in Edinburgh on 4 and 5 June 2009. Sixty seven people attended all or part of it. The Workshop provided valuable information on research needs, and the need for, and remit and functions of, a ruminant GIN.Based on this, our key recommendations are:

There is a strong consensus among stakeholders, with some provisos addressed below, that a ruminant GIN could deliver valuable outcomes with respect to environmental sustainability. We therefore recommend that Defra establish a ruminant GIN – ideally with other funding partners, to maximise the resources focussed on the problem of breeding ruminants for improved environmental sustainability.

To be effective, and retain stakeholder support, a ruminant GIN must have sufficient core funding to make a real difference in several of the research themes identified.

The ruminant GIN should:o Focus strongly on a co-ordinated, strategically planned, programme of R&D, and associated

resources, with directly related KE;o Provide a forum for linking private and public interests in ruminant breedingo Achieve strong interaction and partnership with end users;o Achieve strong interaction and partnership with other groups working in this area internationally;o Achieve strong interaction, and no duplication of effort, with other KE networks, particularly those

listed later;o Focus on reducing GHG emissions (rather than all environmental sustainability traits), but in a

systems context, including trade-offs;o Broker satellite R&D.

If there are insufficient resources to begin work in all themes now, priority for Defra funding should be given to:

o Breeding for environmental sustainability in a wider systems context;o Better tools for measurement of environmental sustainability traits (especially GHG emissions);o Understanding farmer (and other stakeholder) needs and behaviour; o Co-ordinated data (phenotype and genotype) collection, storage and use.

Although research on ‘Economic impact of alternative options at national/farm level and goals/targets/monitoring’ attracted least votes, there is an argument that funding this work would be a useful way of focussing effort and maximising benefit from the research spend on ruminant breeding for environmental sustainability (NB some of the answers will emerge from existing projects).

Research involving genetic evaluation customers and providers is likely to be critical to successful uptake of new breeding tools. (There is a history of successful projects, many LINK projects, in this area.) However, approaches that stimulate ‘pull’ for evaluations and new tools from the commercial sector, as well as producing tools for the purebred sector, should be encouraged.

8. CONCLUSIONS AND IMPLICATIONS

The majority of breeding goals aimed at genetic response per mating made result in a decrease in GHG emissions from the system.

Improving uptake of improved genetic stock for breeding will have a positive effect on farmer returns whist having a positive environmental impact. The inclusion of new traits in the breeding goal such as feed efficiency and survival will have further positive impacts.

The use of genomic selection will further enhance genetic response in the breeding goals and is currently cost effective for the beef sector, though an initial training population investment will be required.

In sheep breeding the use of genomic selection with current breeding structures in the short term is not cost effective. However, the technology shows promise and with novel approaches to breeding structure, genotyping strategy and assuming that the genotyping cost will reduce over time the application of genomic selection in sheep breeding will make a positive impact.

In an assessment of delivery options the most favourable was some form of joint venture between the sheep and beef industry and a research provider.

The national ruminant breeding goals have an indirect influence on GHG emissions through raising the efficiency of production. However, recent work (Wall et al., 2010 Defra IF0182) demonstrated there can be conflict between a selection goal to develop a profitable and biologically sustainable animal and the goal to exclusively reduce GHG emissions.

The main finding of this study is that the majority of breeding goals currently utilised in beef and sheep result in a decrease in GHG emissions from the system and therefore more effective ways of increasing the response to genetic selection and the penetration of improved genetic stock (e.g. greater use of EBVs) will enable a more rapid reduction in GHG emissions per unit of product. This is also relevant when considering traits that influence the life expectancy of animals in the breeding and growing population. The environmental cost of maintaining the suckler herd or sheep flock has to be allocated to the terminal product.

SID 5 (Rev. 3/06) of 29

Some potential new traits such as residual feed intake (RFI), mature weight and survival, may play a significant role in the reduction of GHG emissions. The way in which these traits are treated in the breeding goal does have a significant effect. For example in the terminal beef breeding goal the inclusion of RFI and calf survival weighted by profit value alone resulted in a large increase in returns but had a small negative impact on GHG mitigation. This demonstrates that traits such as RFI will need to be used within a GHG specific breeding goal in order to capture their benefits. Work will be required on finding ways to reduce the cost of developing recorded populations for the RFI trait. This recording step is essential for both traditional and genomic selection. For genomic selection, a multi-breed training population approach may allow the integration of the terminal and maternal breeds and therefore optimising the contributions from both sectors to the beef industry. This approach is likely to be of less use when considering between sheep sector improvement due to the diverse nature of sheep breeding goals. Some other new traits such as meat eating quality may be less directly relevant for the protection of the environment but play an important role in the long term sustainability of supply chains.

The inclusion of new traits in the breeding goals of hill, longwool and terminal sire sheep, as well as the implementation of genomic selection, had favourable effects on net income at the industry level. Changes in farm profits and GHG emission did not follow the same direction or have the same magnitude in the three sheep sectors. The extreme cases were the responses in the longwool sheep breeding sector, in which the highest improvements in net incomes were accompanied by substantial increases in GHG emissions. However, this measure of negative GHG impact only occurs if we consider the female breeding animal in the breeding goal – at the level of kg of product the correlation is positive (Wall et al.,2010 Defra IF0182) .

Currently, the most (cost) effective way in which genetic gain can be increased in the beef and sheep sectors is by improved uptake of existing breeding values when making breeding animal purchases. The need to record the phenotype of animals in order to enter the genetic evaluation system has always been seen as a stumbling block since farmers often view recording as a nuisance. GS may provide an opportunity to overcome this barrier by creating a large section of the breeding/multiplying community which does not record phenotypes but utilises genomic technology to assign breeding values to breeding stock. There will still be a need for a large degree of phenotypic recording but this effort can be concentrated in specialist herds and flocks and include more fruitful areas such as phenotypes for previously unrecorded traits in novel populations.

For example it will be possible to use data on carcass and meat characteristics gathered in abattoirs to assign breeding values to other animals in the population. There is a strong likelihood that such populations providing phenotypes will be composed of crossbred individuals. Based on the simulation undertaken in this study, crossbred performance can be efficiently used for genomic selection of purebred populations. The accuracies of genomic breeding values (GBV) from crossbred performance data are higher when closely related breeds are crossed than when distantly related or unrelated breeds are crossed. The present study showed that the impact of ignoring dominance effects can substantially reduce the accuracy of additive GBV, depending on the relative importance of dominance variation in relation to additive genetic effects. However, using crossbred information from completely unrelated populations substantially reduced the accuracy of the estimated GBV and therefore specification of the relationship between the breeds used for crossing is necessary in order to use crossbred information for genomic selection efficiently.

Economic and environmental contributions were affected by training population and effective population sizes. Although it is not possible to modify effective population size, optimizing GS strategy in practice should consider the real undetermined beef cattle effective population sizes. Assuming responses possible with smaller effective population size or extrapolating results achieved in other breeds or sectors may lead to erroneous conclusions in biological and economic terms. Larger training populations are related to higher biological responses and higher costs. However, in certain circumstances the higher accuracy obtained by larger training population sizes outweighs the higher costs.

Generally, the crossbreeding simulation showed that GS results in sufficiently high accuracies of estimated GBV even for lowly heritable traits and thus can be successfully used to select purebred animals using purebred as well as crossbred performance in the training population. Considering that traits such as methane emission and traits correlated to it such as feed efficiency can only be measured using proxy methods (e.g. methane gun) or electronic feeders on a subsample of the population, genomic selection is expected to be the most effective methodology to achieve sufficient selection response in these traits. This simulation provides the basis for an efficient improvement of those traits. Further simulation using a deterministic approach (developed or validated based on simulation studies) is used in this report to reflect the efficiency of genomic selection based on the given UK population structure in sheep and beef.

The relative magnitudes of the responses, compared to the current scenarios, varied for the different breeding sectors depending on the traits considered in each of these sectors. For the three sheep sectors, although the increases ranged between 4% and 56%, the additional net income would not cover the cost of genotyping in the unlikely event that it were to remain at the current initial development cost. There is good evidence that development of genotyping technology results in considerable cost reduction over time and for this reason the

SID 5 (Rev. 3/06) of 29

cost benefit study considered a stepped reduction in genotyping cost over the initial years of the breeding scheme to a minimum level of £50/genotype. This minimum value was considered as the least price as there will still be associated sampling and administration costs regardless of how inexpensive the actual genotyping is.

The issue of low or negative profitability of GS due to the current high cost of the high-density SNP panels has been raised in other studies (Albers, 2010; Goddard and Hayes, 2009). An alternative GS strategy based on the use of low-density SNP panels, in which SNPs are evenly spaced across the genome and the high density missing information is imputed from key relatives with higher density genotypes, is being investigated with encouraging results (Goddard and Hayes 2009; Habier et al. 2009). This approach is still at an early experimental phase though the cost-effectiveness of GS would improve under such an approach given the lower prices of the low-density panels. This is an alternative option to be analysed in future studies taking into account not only the reduced costs but also the likely loss in selection accuracy compared to the high-density SNP panel that has to be investigated for the different breeding programmes. An alternative approach to reducing genotyping costs is through developing genotyping schemes that again only use high density SNP panels on key animals and use the information established to predict breeding values in offspring.

In separate consultations with stakeholders in the sheep and beef industry of the UK and in terms of the delivery of GS, the option that was identified as most effective in terms of adoption of the technology and industry wide coordination of approach was a proposed joint venture between industry and research provider/s.

RecommendationsBased on the new results from the economic evaluation and the experience in New Zealand, the following recommendations should be considered for the future for a UK GS service:

1. Collection of high quality DNA (i.e. blood) from all sires in performance recording flocks is needed so that current and recent industry performance records do not get lost to future training and discovery efforts.

2. New and creative ways of incentivising sheep breeders to record more traits more accurately including disease traits, maternal performance traits, and any other traits linked to GHG production need to be found to enable subsequent exploitation of GS and the harvesting of its benefits.

3. All stakeholders in GS need to keep up to date with research methods that will increase accuracy and decrease implementation cost of GS. This includes work that is developing low density SNP arrays onto which high density information can be imputed from genotyped relatives.

4. More breeding strategy studies that look at how to save costs (e.g. sire tests, and/or two stage selection), and the importance of trait recording in industry animals.

5. We advise against information nucleus approaches because of concerns that predictions will not hold up in different subpopulations. This is because of highly complex mechanisms (epigenetics, imprinting, gene expression control) associated with QTLs leading to gene x gene and gene x environment interactions at the QTL level, notwithstanding the breakdown in LD between markers and QTLs over time and to very distant relatives.

Future WorkAt the present time SAC is leading a consortium of researchers in a bid to create a ruminant genetic improvement network (Tender: IFO119 - Ruminant GIN: Genetic Improvement Network for Ruminant Livestock).

Associated ResearchThe modelling of the sheep sectors in this project has been done in conjunction with work undertaken for the Scottish Government Project: Development of a National Genotyping Service to Deliver Key Benefits for the Scottish Sheep Sector.

ReferencesAlbers, G.A.A. (2010) Genomic selection in poultry and pig breeding: a breakthrough technology? Proceedings

of the British Society of Animal Science.(Belfast, UK).Amer, P. R., Nieuwhof, G. J., Pollott, G. E., Roughsedge, T., Conington, J., and Simm, G. (2007). Industry

benefits from recent genetic progress in sheep and beef populations. Animal 1 (10), 1414-1426.Arthur, J. P. F. and Herd, R. M. (2008). Residual feed intake in beef cattle. Revista brasileira de Zootecnia 73,

269-279.Attwood, G. and McSweeney, C. (2008). Methanogen genomics to discover targets for methane mitigation

technologies and options for alternative H-2 utilisation in the rumen. Australian Journal of Experimental Agriculture 48 (1-2), 28-37.

Attwood, G. T., Kelly, W. J., Altermann, E. H., Moon, C. D., Leahy, S., and Cookson, A. L. (2008). Application of rumen microbial genome information to livestock systems in the postgenomic era. Australian Journal of Experimental Agriculture 48 (6-7), 695-700.

Barwick, S. A., Wolcott, M. L., Johnston, D. J., Burrow, H. M., and Sullivan, M. T. (2009). Genetics of steer daily and residual feed intake in two tropical beef genotypes, and relationships among intake, body composition, growth and other post-weaning measures. Animal Production Science 49 (5-6), 351-366.

SID 5 (Rev. 3/06) of 29

Blaxter, K. L. and Clapperton, J. L. (1965). Prediction of Amount of Methane Produced by Ruminants. British Journal of Nutrition 19 (4), 511-522.

Boadi, D. A. and Wittenberg, K. M. (2002). Methane production from dairy and beef heifers fed forages differing in nutrient density using the sulphur hexaflouride (SF6) tracer gas technique. Canadian Journal of Animal Science 82, 201-206.

Daetwyler, H. D., Villanueva, B., and Woolliams, J. A. (2008). Accuracy of predicting the genetic risk of disease using a genome-wide approach. Plos One 3 (10), e3395.

de Haan, D., Steinfeld, H., and Blackburn, H. (1996). Livestock and the environment: Finding a balance. Report to the World Bank, FAO and USAID. http://www.fao.org/docrep/x5303e/x5303e00.HTM

Dekkers, J. C. M. (2007). Marker-assisted selection for commercial crossbred performance. Journal of Animal Science 85, 2104-2114.

Genesis Faraday Partnership. (2008). A study of the scope for the application of research in animal genomics and breeding to reduce nitrogen and methane emissions from livestock based food chains. Final Report to Defra: Project AC0204.

Gill, M., Smith, P., and Wilkinson, J. M. (2010). Mitigating climate change: the role of domestic livestock. Animal 4 (3), 323-333.

Goddard, M. E. (2009). Genomic selection : prediction of accuracy and maximization of long term response. Genetica 136, 245-257.

Goddard, M. E., and Hayes, B. J. (2007). Genomic selection. Journal of Animal Breeding and Genetics 124, 323-330.

Goddard, M. E. and Hayes, B.J. (2009). Genomic selection based on dense genotypes inferred from sparse genotypes. Proceedings of the Association for the Advancement of Animal Breeding and Genetics 18, 26-29. Barossa Valley, Australia.

Goopy, J. P. and Hegarty, R. S. (2004). Repeatability of methane production in cattle fed concentrate and forage diets. Journal of Animal and Feed Sciences 13, 75-78.

Guan, L. L., Nkrumah, J. D., Basarab, J. A., and Moore, S. S. (2008). Linkage of microbial ecology to phenotype: correlation of rumen microbial ecology to cattle's feed efficiency. Fems Microbiology Letters 288 (1), 85-91.

Guillaume, F., Fritz, S., Boichard, D., and Duret, T. (2007). Use of SNP for marker assisted selection in French dairy cattle. Book of Abstract of the 58th Annual Meeting of the European Association for Animal Production, No. 13, Dublin, Ireland.

Habier, D., Fernando, R.L. and Dekkers, J.C.M. (2009) Genomic selection using low-density marker panels. Genetics 182, 343–353.

Hackstein, J. H. P., Langer, P., and Rosenberg, J. (1996). Genetic and evolutionary constraints for the symbiosis between animals and methanogenic bacteria. Environmental Monitoring and Assessment 42 (1-2), 39-56.

Hegarty, R. S. (2004). Genotype differences and their impact on digestive tract function of ruminants: a review. Australian Journal of Experimental Agriculture 44 (4-5), 458-467.

Herd, R. M. and Arthur, P. F. (2009). Physiological basis for residual feed intake. Journal of Animal Science 87 (14), E64-E71.

Johnston, D. J., Herd, R. M., Kadel, M. J., Graser, H. U., Arthur, P. F., and Archer, J. A. (2002). Evidence of IGF-I as a genetic predictor of feed efficiency traits in beef cattle. Proceedings of the 7th World Congress on Genetics Applied to Livestock Production. Montpellier, France.

Kahi, A. K., Barwick, S. A., and Graser, H. U. (2003). Economic evaluation of Hereford cattle breeding schemes incorporating direct and indirect measures of feed intake. Australian Journal of Agricultural Research 54 (10), 1039-1055.

Kumar, S., Puniya, A. K., Puniya, M., Dagar, S. S., Sirohi, S. K., Singh, K., and Griffith, G. W. (2009). Factors affecting rumen methanogens and methane mitigation strategies. World Journal of Microbiology & Biotechnology 25 (9), 1557-1566.

Lassey, K. R., Ulyatt, M. J., Martin, R. J., Walker, C. F., and Shelton, I. D. (1997). Methane emissions measured directly from grazing livestock in New Zealand. Atmospheric Environment 31 (18), 2905-2914.

Ley, R. E., Backhed, F., Turnbaugh, P., Lozupone, C. A., Knight, R. D., and Gordon, J. I. (2005). Obesity alters gut microbial ecology. Proceedings of the National Academy of Sciences of the United States of America 102 (31), 11070-11075.

Meuwissen, T. H. E., Hayes, B. J., and Goddard M. E. (2001) Prediction of total genetic value using genome-wide dense marker maps. Genetics 157, 1819-1829.

Moore, K. L., Johnston, D. J., Graser, H. U., and Herd, R. (2005). Genetic and phenotypic relationships between insulin-like growth factor-I (IGF-I) and net feed intake, fat, and growth traits in Angus beef cattle. Australian Journal of Agricultural Research 56 (3), 211-218.

Moore, K. L., Johnston, D. J., Herd, R. M., and Graser, H. U. (2003). Genetic and non-genetic effects on plasma insulin-like growth factor I (IGF-I) concentration and production traits in angus cattle. Proceedings of the 15th Association for the Advancement of Animal Breeding and Genetics. Melbourne, Australia.

Moran, D., MacLeod, M., Wall, E., Eory, V., Guillaume, P., Matthews, R., McVittie, A., Barnes, A., Rees, B., Moxey, A., Williams, A., and Smith, P. (2008). UK marginal abatement cost curves for the agriculture and land

SID 5 (Rev. 3/06) of 29

http://apps.isiknowledge.com/full_record.do?product=UA&search_mode=GeneralSearch&qid=2&SID=X2j62GJp7eL563CggaN&page=1&doc=1&colname=WOS

use, land use change and forestry sectors out to 2022, with qualitative analysis of options to 2050. Final report to the Committee on Climate Change (RMP4950). http://www.theccc.org.uk/reports/supporting-research/

Munger, A. and Kreuzer, M. (2008). Absence of persistent methane emission differences in three breeds of dairy cows. Australian Journal of Experimental Agriculture 48 (1-2), 77-82.

Ouwerkerk, D., Turner, A. F., and Klieve, A. V. (2008). Diversity of methanogens in ruminants in Queensland. Australian Journal of Experimental Agriculture 48 (6-7), 722-725.

Pinares-Patino, C., Ulyatt, M. J., Holmes, C. W., Barry, T. N., and Lassey, K. R. (2000). Inter-sheep variation in methane emission, a tool for mitigation. Proceedings of 2nd international methane mitigation conference. Novosibirsk, Russia.

Pinares-Patino, C. S. and Clark, H. (2008). Reliability of the sulfur hexafluoride tracer technique for methane emission measurement from individual animals: an overview. Australian Journal of Experimental Agriculture 48 (1-2), 223-229.

Pinares-Patino, C. S., Ulyatt, M. J., Lassey, K. R., Barry, T. N., and Holmes, C. W. (2003). Persistence of differences between sheep in methane emission under generous grazing conditions. Journal of Agricultural Science 140, 227-233.

Robertson, L. J. and Waghorn, G. C. (2002). Dairy industry perspectives on methane emissions and production from cattle fed pasture or total mixed rations in New Zealand. Proceedings of the New Zealand Society of Animal Production 62, 213-218.

Safely, L. M., Casada, M. E., Woodbury, J. W., and Roos, K. F. (1992). Global methane emissions from livestock and poultry manure. US Environmental Protection Agency. EPA/4001/1-91/048.

Seksik, P., Rigottier-Gois, L., Gramet, G., Sutren, M., Pochart, P., Marteau, P., Jian, R., and Dore, J. (2003). Alterations of the dominant faecal bacterial groups in patients with Crohn's disease of the colon. Gut 52 (2), 237-242.

Shi, P. J., Meng, K., Zhou, Z. G., Wang, Y. R., Diao, Q. Y., and Yao, B. (2008). The host species affects the microbial community in the goat rumen. Letters in Applied Microbiology 46 (1), 132-135.

Stewart, J. A., Chadwick, V. S., and Murray, A. (2005). Investigations into the influence of host genetics on the predominant eubacteria in the faecal microflora of children. Journal of Medical Microbiology 54 (12), 1239-1242.

Tannock, G. W., Munro, K., Harmsen, H. J. M., Welling, G. W., Smart, J., and Gopal, P. K. (2000). Analysis of the fecal microflora of human subjects consuming a probiotic product containing Lactobacillus rhamnosus DR20. Applied and Environmental Microbiology 66 (6), 2578-2588.

Toro, M. A., and Varona, L. (2009). Including dominance effects in genomic selection. Conference on Next Generation Sequencing: Challenges and Opportunities. 1st – 3rd October, Barcelona, Spain.

van der Werf, J. H. J. (2004). Is it useful to define residual feed intake as a trait in animal breeding programs? Australian Journal of Experimental Agriculture 44 (4-5), 405-409.

van der Westhuizen, R. R., van der Westhuizen, J., and Schoeman, S. J. (2004). Genetic variance components for residual feed intake and feed conversion ratio and their correlations with other production traits in beef bulls. South African Journal of Animal Science 34 (4), 257-264.

Vanhoutte, T., Huys, G., De Brandt, E., and Swings, J. (2004). Temporal stability analysis of the microbiota in human feces by denaturing gradient gel electrophoresis using universal and group-specific 16S rRNA gene primers. Fems Microbiology Ecology 48 (3), 437-446.

Vlaming, J. B., Lopez-Villalobos, N., Brookes, I. M., Hoskin, S. O., and Clark, H. (2008). Within-and between-animal variance in methane emissions in non-lactating dairy cows. Australian Journal of Experimental Agriculture 48 (1-2), 124-127.

Wall, E., Ludemann, C., Jones, H., Audsley, E., Moran, D., Roughsedge, T., Amer, P. (2010).The potential for reducing greenhouse gas emissions for sheep and cattle in the UK using genetic selection. Final Report for Defra.

Wei, M. (1992). Combined crossbred and purebred selection in animal breeding. PhD thesis Wageningen Agricultural University, The Netherlands.

Zoetendal, E. G., Akkermans, A. D. L., and de Vos, W. M. (1998). Temperature gradient gel electrophoresis analysis of 16S rRNA from human fecal samples reveals stable and host-specific communities of active bacteria. Applied and Environmental Microbiology 64 (10), 3854-3859.

Zoetendal, E. G., Vaughan, E. E., and de Vos, W. M. (2006). A microbial world within us. Molecular Microbiology 59 (6), 1639-1650.

Zoetendal, Erwin G., Akkermans, Antoon D., kkermans-van Vliet, Wilma M., de Visser, J., and de Vos, Willem M. (2001). The host genotype affects the bacterial community in the human gastrointestinal tract. Microbial Ecology in Health and Disease 13 (3), 129-134.

SID 5 (Rev. 3/06) of 29

http://www.theccc.org.uk/reports/supporting-research/

References to published material9. This section should be used to record links (hypertext links where possible) or references to other

published material generated by, or relating to this project.

Refereed Conference Proceedings

1. Navajas EA, Sawalha R, Duthie C-A, Roehe R and Roughsedge T. 2010. Response to genomic selection in the Scottish Blackface breeding programme. Proceedings of the British Society of Animal Science, Belfast.

2. Navajas EA, Duthie CA, Amer PR, Sawalha RM, Roehe R and Roughsedge T. 2010. Impact of Genomic Selection for Residual Feed Intake and Calf Survival in Beef Cattle on Profit and Greenhouse Gas Mitigation. Proceedings of the 9th World Congress of Genetics Applied to Livestock Production.

3. Duthie C, Sawalha R, Navajas E, Roehe R, Roughsedge T. 2010. Economic response that can be achieved from including genomic information to the terminal sire index of beef cattle. Proceedings of the British Society of Animal Science, Belfast.

4. Sawalha, RM, Navajas EA, Duthie CA, Roehe R and Roughsedge T. 2010. Selection Response in UK Sheep Breeding Program with the Application of Genomic Selection and Inclusion of Production Efficiency Traits Proceedings of the 9th World Congress of Genetics Applied to Livestock Production.

SID 5 (Rev. 3/06) of 29

general enquiries on this form should be made...

Documents