Feasibility of using abattoir generated data and BCMS records for carcass trait evaluations

Project number:                    72401

Lead contractor:                   SAC / Egenes

Partners:                                   DairyCo, HCC

Start & end date:                   01 November 2011 – 30 April 2012

 

The Problem

A number of national datasets contain information which could be used to produce EBVs for carcass traits based on abattoir data. These include abattoirs, BASCO, breed societies, milk recording organisations and British Cattle Movement Service (BCMS). Work is needed to draw the relevant data together and to create a consolidated pedigree and performance file.

 

Project Aims

The overall aim of this feasibility project was to assimilate, cleanse, salvage, validate and characterise abattoir and BCMS data then report descriptive statistics for the resulting dataset. The subsequent use of these data (in another follow on project) is expected to be for genetic evaluations.

The specific objectives are;-

1. The construction of computer systems to allow automatic upload and preparation of abattoir (ABP), BCMS and DARDNI data. Construction of processes to find and update the sire primary ID for BCMS and abattoir records in as many places as possible (BASCO, milk recording databases, archive files from breed societies).

2. Extraction, validation and QA of data. Reporting of numbers matched per breed, per sire etc. Phenotypic analysis and reporting of carcass weight, carcass fat and carcass conformation. Reporting on numbers of sires and offspring per sire broken into informative categories such as sire breed, sire breed category (dairy/beef), year of birth of offspring. Reporting on efficiency measures such as age at given slaughter weight and predicted methane emissions.

 

Approach

Objective 1: Consolidate available data

Carcase data for all breeds from abattoirs were collated into a single database. Animal identity was reformatted and matched to BCMS to produce a national cross reference file. This was then matched to other sources such as BASCO, milk recording organisations (MROs) and breed society files where available. The programming has enabled a data and pedigree file to be produced as a combination of data from all available data sources, for both beef and dairy bulls where data exists. The consolidated dataset also contains the “best data from different sources for the same animal (e.g. sire from one record and dam from the other).

Objective 2: Characterise merged and validated data

Data produced in objective 1 was validated, collated by cross referencing between all available and appropriate data sources and then extracted for analysis. The extraction processes has been automated and built into the national genetic evaluation systems with appropriate quality assurance tests, validation, logging and data salvaging. Reporting includes:

1. Number of sires

2. Number of offspring per sire

3. Breakdown by breed

4. Record counts for data salvaged/lost

5. Means for age, weight and carcass conformation at slaughter broken down into sire, breed, year of slaughter etc.

6. Estimates of efficiency differences between sires where possible

 

Results

Existing industry (abattoir) and government data were used to produce a consolidated dataset of carcass traits for beef and dairy cattle. The overall aim of this feasibility study was to assimilate, cleanse, salvage, validate and characterise abattoir and British Cattle Movement Service (BCMS) data followed by descriptive statistics for the resultant dataset. The subsequent use of these data is expected to be for genetic evaluations and so analyses undertaken in this study were designed to reveal information on the suitability of these data for genetic evaluations.

Initially, 3 million individual carcass abattoir records (from three abattoirs) and approximately 48 million BCMS animal records were made available for this project. Using intelligent string matching, 82% of the individual carcass records could be matched to a BCMS individual animal record, resulting in a dataset of 2,435,875 for further investigation. The three traits available from abattoir records were net carcass weight, conformation and fat class. Matching to BCMS data provided information on animal movements, breed, dates of birth and death, in addition to dam and sire identities. Sire, which is not compulsory to record, was recorded for approximately 23% of animal records and the level of recording was generally higher in more recent years (11% in 2001 and 23% in 2011). Dates of animal birth and death in BCMS enabled determination of age at slaughter, and the average daily gain for net carcass weight. Across all breeds, the averages for the slaughter population aged from 3 to 36 months for net carcass weight, days to slaughter, average net carcass weight daily gain, conformation and fat class were 323.7 kg, 743 days, 0.45 kg, -R, and +3 respectively.

A refreshed BCMS database was obtained which included movement records. The herd identity was encrypted and so no information was available on the holdings themselves although animals could be grouped by holding by time and contemporary groups formed. This was a significant improvement on the data expected to be available to the project in comparison to a previous extract which only included holding of birth and death and the number of movements. The average number of locations for an animal to be reared (for a period of at least two months) was 1.7.

The major breeds present in the carcass population (with over 100,000 animals) were described as Limousin, Aberdeen Angus, Holstein Friesian, Charolais, Hereford, Simmental, and Belgian Blue, and these accounted for 92% of the animals present in the matched abattoir/BCMS dataset. A major proportion of animals described as beef breeds were cross-breds and generally take the name of the sire breed. Dam breed records emphasise that dairy cows are a major component of beef production with Holstein Friesian being the most common dam breed accounting for 46% of the slaughter population.

A pedigree file was created for BCMS records by matching to other national data sources (NMR, CIS, HUK, BASCO) and to itself. This resulted in a (super) pedigree file of over 50 million animals going back a maximum of 13 generations, and sire was available for 25% of the slaughter population (an increase of 2.3%). The super pedigree tended to lead to greatest improvement in sire records for dairy breeds, which were generally low in BCMS. The across beef and dairy super pedigree can be used to explore additive and non-additive (heterosis, recombination losses) genetic effects for traits relevant across both breeds (where data exists), such as carcass and product quality and safety traits but also cattle health/disease traits.

The proportions of animals with sire records are given by breed in the table below:

Percentage of animals with sire records with and without the super pedigree information

Breed

Count

No. of sires in super pedigree

% sire records abattoir/BCMS

% sire records

super pedigree

HIGHLAND0S

2918

1950

57.23

66.83

STABILIS0R

7482

4951

59.6

66.17

LUING

3971

2278

57.34

57.37

SOUTDEVO

15457

8479

53.92

54.86

LINCRED_U

1638

831

50.61

50.73

GALLOWAY

3222

1558

48.29

48.36

WELSBLAC

12202

5503

44.39

45.10

SUSSEX

3517

1557

41.97

44.27

DEVON

3419

1371

39.66

40.10

SALER

13335

5115

38.1

38.36

ABERANGU

499458

185935

36.75

37.23

SHORTHORN

17115

5802

32.32

33.90

HEREFORD

198362

60196

30.02

30.35

BLONDAQ

59581

15892

26.33

26.67

SIMMENTAL

187147

48263

25.41

25.79

CHAROLAIS

295021

65889

21.93

22.33

LIMOUSIN

500047

104330

20.01

20.86

JERSEY

2822

586

4.29

20.77

AYRSHIRE

7570

1298

11.03

17.15

MERHIS_U

5377

881

14.28

16.38

HOLSFRIE

347051

48526

5.59

13.98

SWRE&WH_U

2421

337

1.03

13.92

BROWSWIS

3268

381

4.31

11.66

MONTBELIAR

7724

839

7.52

10.86

BELGBLUE_U

159271

13277

7.41

8.34

 

Genetic analyses were performed on a subset of the data for animals with a Charolais sire, which consisted of 17,125 records after editing. Heritability estimates for net carcass weight, conformation and fat class were 0.31, 0.24, and 0.14. Similar results were seen in a within Limousin breed parameter estimation analysis. These results provide strong indication of the existence of genetic variation in the studied traits. This, in turn, suggests that improving carcass quality traits through genetic selection is entirely possible, thereby warranting more detailed investigation of their genetic background, particularly their relationship with other traits of importance and within, between and across breeds.

The results of this feasibility study indicate that genetic analysis for carcass traits is realistic, particularly for breeds which make up a major part of the carcass population and have sufficient information on the sire. Encouraging the recording of sire identity by farmers in BCMS would further improve the usefulness of future data.

 

Planned activity

The results of this study may lead to further research to undertake genetic analyses.