Better Breeding with Automated Milking Systems

Collecting and analysing vast amounts of data from robot milkers could speed up genetic improvement.

Image: robot milking system.

With almost a quarter of its dairy herds using automated milking systems (AMS), Denmark is already looking into ways to harness this potential for genetic improvement.

AMS, when coupled with an increasing array of in-line milk sensors, can generate masses of data that could measure important traits.

However, says Uffe Lauritsen of the Danish milk recording organization RYK, this great potential comes with challenges. How data from these farms are shared - or possibly not shared - and how these issues are dealt with in the next few years will determine the future shape of dairy genetic improvement, he told the recent North American Conference on Precision Dairy Management.

Denmark has 845 AMS herds averaging 168 cows each on milk recording. At 22 per cent of all dairy herds and 27 per cent of dairy cows, it has the highest AMS concentration of any country in the world.

Introduced in 1992, the technology has spread worldwide, especially in the past few years with just over 8,000 AMS farms reported in 2009 more than 90 per cent of them in northwestern Europe. North America has relatively few AMS farms, but the number is growing as the technology becomes more aligned with the style of dairying on this side of the Atlantic.

The number of milkings per cow on AMS farms averages between 2.5 and more than 3.0 times per day, although 10 per cent of operations average less than two milkings per day. Time between milkings can vary widely.

Data Collection Opportunity

These milkings create a great opportunity to increase data collection frequency and accuracy. In-line sensors to monitor milk quality and milk flow can generate this data to measure important traits. Studies have been done on the best methods to record every milking with a view to finding an economical way of handling this data.

One option is to transfer these huge volumes of data to the milk recording agency, but it could raise concerns about capacity and costs for storage and analyses. Nevertheless, the Danish milk recording agency is taking the bold step of capturing all AMS herd data, and managing it within its database.

The most exciting development as this technology progresses are trait measurements from in-line sensors: milk quality, electro-conductivity, milk flow and milking time, as well as a host of milk component data that will soon be available.

Measuring traits more consistently by machine improves accuracy. A study reported by Lauritsen used electronic milk meters to measure milk flow. It showed improved accuracy of measuring milking speed. That resulted in heritability estimates increasing to 0.3 from 0.2, and greater accuracy of estimated breeding values (EBVs) for milking speed-much more accurate than human estimates.

Milk Recording Challenge

However, increasing use of AMS and parlour technology poses a major challenge. Once dairy farmers have invested in the technology, they may be reluctant to participate in and pay for milk recording services. This could detract from the usefulness of dairy herd improvement (DHI) to provide management, benchmarking and genetic improvement information.

While the technology could open the door to more information about a large number of new traits, loss of DHI participation could prevent these traits from being useful from a genetic improvement viewpoint. Other considerations include time and effort to collect samples for lab analysis, and AMS software not working with the DHI system.

Computer Compatibility

The ability to share masses of data captured on AMS farms is crucial to turning it into useful information for the dairy industry for better management information, benchmarks and genetic improvement. One dairy farmer, quite familiar with AMS data and DHI data streams, compared the lack of standardization of data protocols among milking equipment manufacturers and the DHI system with the standardization of couplers for hydraulic hoses on field equipment.

AMS and automated milking parlour software for trait definition and data transfer differ among manufacturers and even models of the same make. Every parameter is defined differently, depending on the brand of robotic milking system, Lauritsen's group concluded in their study. Each must be uniquely identified in their data.

DHI organizations, since they are supported by their member producers, may balk at the cost of such customization. The answer to this dilemma may lie in DHI organizations worldwide sharing their approach in capturing and managing this sort of complex data.

Standardization is vital for integration and efficiency. Milking equipment manufacturers, DHI organizations and software providers need to pursue standardized data transfer protocols for the good of the industry, and the farmers who support them by buying products and services to better manage their herds.

Breeding For AMS Traits

The Danish study also defines future projects. They include developing more accurate EBVs for milking speed, an economically important trait in large parlours as well as in AMS. Others are breeding for cows more suitable for AMS and improvement of functional traits.

A different set of attributes may make a cow well-suited for milking in an AMS versus a parlour or tiestall. At present we lack that information. The Danes plan to look at possibilities to develop breeding values for that ability. Traits that could affect a cow's suitability for AMS might include udder conformation, teat placement and shape, body size and shape, and temperament.

In Canada, a small study recently looked at the milking ability or desire of cows to use an AMS. Although it showed some relation to milk production-cows giving more milk tended to want to use the AMS more-the study lacked enough numbers to identify any genetic traits that make cows more or less adapted to AMS milking. This points out the need for accurate, routine capture of quality cow data from automated systems to develop worthwhile genetic parameters.

Even More Information

Improvement of breeding values for functional traits shows great promise. Data recording from AMS may improve accuracy of EBVs for existing functional traits in health, fertility, conformation and temperament. Data already captured by automated milking include live weight, milk yield per quarter, temperature, conductivity, blood in the milk, activity measurements and chewing activity.

In-line analysis systems developed over the past few years are close to becoming widely available commercially. They provide much more cowside information about important milk components.

The Afi-Lab developed by Afimilk can analyse for butterfat and protein, and estimate somatic cell count. DeLaval's Herd Navigator system, prototyped in European herds, can estimate progesterone, useful in reproduction management, Lactate dehydrogenase (LDH), an udder health indicator, and urea and Beta Hydroxy Butyrate (BHB), indicators of rumen health and ketosis levels.

Expanding trait measurements through milking systems promises great potential for industry benefit if the information can be shared and used for common good. A key step in encouraging producers to continue using DHI services will be improving compatibility and data sharing among equipment manufacturers and DHI organizations.

These steps would let the industry take full advantage of the wealth of information automated systems provide for genetic improvement and overall profitable farm management. Without progress in compatibility and data sharing we may soon be drowning in data on the farm, but with dramatically reduced value of information for DHI and genetic improvement.

This article appeared in the May 2010 edition of Milk Producer Magazine.


Lauritsen, U. and A. Fogh, Huge Potential in Data from AMS from A Breeder's Perspective. Proceedings of the First North American Conference on Precision Dairy Management. Toronto March 2-5, 2010. pp. 28-29.

deKoning, CJAM. Automatic Milking - Common Practice on Dairy Farms. Proceedings of the First North American Conference on Precision Dairy Management. Toronto March 2-5, 2010. pp. 52-67.

Author: Blair Murray - Dairy Genetic Improvement Specialist
Last Reviewed: August 2010


For more information:
Toll Free: 1-877-424-1300
Author: Blair Murray - Dairy Genetic Improvement Specialist,OMAFRA
Creation Date: 17 August 2010
Last Reviewed: 17 August 2010