Looking at sheep performance

Ways smarter tech can accelerate livestock genetic progress

I started writing about the case for smarter tech in cattle and sheep breeding programmes a while ago, and quickly realised there was more than would fit in a single blog post. In my previous post I described how technology can help us increase the pace of genetic progress, by:

  • Measuring what we can’t easily see, including a range of important traits such as efficiency, emissions and disease resistance; and
  • Reducing errors in recording and transcription.

There are two further areas where smart technology can and is helping:

Cut collection costs

How much does it cost to collect a phenotype for an animal?

A phenotype is the observable characteristics of an individual that result from both its genotype and the environment. When you collect sufficient phenotypic information about many individuals, you can adjust for environmental effects, and predict the genetic merit of that animals. Collecting phenotypes involves measuring details such as calving ease, weight gains, progeny survival and other characteristics of economic value to farmers and consumers.

Current technology for DNA analysis tell us more about the genetic similarity and differences between animals (their genetic relatedness) than their outright performance. There are relatively few simple gene interactions you can measure with a DNA test.

What DNA testing helps with is the challenge of recording which animals are related to each other, especially as pedigree recording gets harder with larger sets of animals. For the DNA test to be useful, it must be able to be statistically connected to animals that have had phenotypes measured.

As Dr Mike Coffey of SRUC, wrote in 2011, “In the age of the Genotype, Phenotype is king!

Phenotyping will continue to be required. Livestock breeders will need a pool of accurate and comprehensive phenotype observations to support both new and existing breeding objectives, connected to the growing pool of genotyped animals.

Measuring animals is time-consuming and expensive: the tedious task of observing and measuring characteristics of animals, linking these to the correct animal records, and transcribing the observations into computer systems.

With the need for ongoing, and potentially more detailed phenotype recording, technology is our friend. From automated weighing and electronic identification to cameras, sensors, and deep learning, properly configured and managed technologies reduce the cost of data collection – especially for larger operations and at commercial scale.

We’ve helped several organisations organise and automate their phenotype recording, integrating electronic identification and a range of other technologies, and streaming that data back to central databases.

Better buying behaviour

What really drives buying decisions?

Ultimately livestock breeding is driven by demand.  Livestock breeders focus their selection efforts on traits that drive economic performance such as:

  • Fertility;
  • Birth weight and survival;
  • Feed conversion efficiency and growth rates or milk production;
  • Carcass characteristics or milk characteristics;
  • Docility, resistance or resilience to disease;

and more.  In the livestock industry, this is presented as the predicted benefit to future generations (typically termed EPDs or EBVs depending on your species, breed, and country). The EBVs or EPDs are combined into economic indexes where each is given a $ or £ weighting based on its impact in a typical farm system and usually delivered as tabular reports and graphs to buyers.

And what do buyers select on?

  • In our experience, New Zealand dairy farmers make great use of the primary economic index – Breeding Worth or BW, combining that with decisions about breed and gestation length.
  • In the sheep industry, breed and breeder decisions are often the primary decider. Once a breeder is selected, the numbers are used to buy the best animals one can afford.
  • In the NZ and Australian beef industry, an expert tells me that for animals sold at auction, the closest purchase price correlation is with liveweight: in their opinion, animals are often valued on how big they are.

When sheep and beef farmers are presented with multiple EBVs and indexes, it can be overwhelming, and it is not surprising some farmers revert to assessing a ram visually, negating the value of genetics programmes.

I’m a great fan of Daniel Kahneman’s Thinking Fast, Thinking Slow, where he points out the challenge we have with quantifying difficult or unknown measures.

“If a satisfactory answer to a hard question is not found quickly, System 1 will find a related question that is easier and will answer it. I call the operation of answering one question in place of another substitution.”  – Daniel Kahneman, Thinking Fast, Thinking Slow.

If understanding EBV’s and indexes seems difficult, you’re most likely to substitute an easier question – how good the ram looks or how knowledgeable the breeder seems – without even realising you have done so.

How can technology help here? We can use technology to make ourselves smarter – or the questions simpler. For instance, there’s potential to create a data-driven system that analyses your current farm system and recommends which economic index or other measure you should use when choosing animals. Or perhaps a tool that with a few simple selections works out how much you should spend on rams or bulls, finds those at a sale that meet your needs, and returns a ranked short list that you can use in your final decisions.

Smarter technology for livestock buyers is a key area where the industry can make real progress on the rate of genetic gain, and a strong understanding of how people make decisions will be critical to its success.

The case for smarter tech in cattle and sheep breeding

My first job in livestock performance recording was with the Genetics Section, as it was called, at Ruakura Research Centre in New Zealand. I worked part time while studying at university, transferring research trial data off the government mainframe on reel-to-reel tape, and writing inbreeding coefficient calculation software.

The genetics section was based in an old converted house, where we sat around at large, wooden, public service desks, surrounded by high stacks of computer printouts, all painstakingly bound and labelled for future use. We were the leading edge of genetic improvement and livestock performance recording.

That was nearly thirty years ago of course, and the face and capability of modern technology has radically changed. Interestingly however, many of the practices in livestock recording industries still reflect that past golden age, and it is only recently that the software tools and databases of – let’s be generous and say – 15 years ago have started to be refreshed.

In this, the first of two articles about technology in livestock breeding, I propose that we could make much more effective use of smart technologies to increase the rate of genetic progress and address commercially important, but hard to measure, animal characteristics. In my next post, I’ll examine how technology could reduce the cost of phenotype collection (I might even explain what a phenotype is), and encourage better use of improved genetics by commercial producers.

Measure what you can’t see

In our traditional performance breeding tools, we focused on things that farmers could readily measure: kilograms and counts. Numbers of live progeny, and kilograms of liveweight, milk, and wool. Good news, most of those production traits are heritable and we’ve made good progress over the last 30+ years.

So how do you measure characteristics that are important in modern farming systems?

  • Meat eating quality, so that consumers can repeatably have a great eating experience;
  • Feed conversion efficiency, converting inputs into product more efficiently, reducing greenhouse gas emissions per unit of product, and making the farming system more profitable;
  • For that matter, greenhouse gas emissions (where this is driven by livestock genetics rather than inoculation by a specific set of gut microorganisms);
  • Urine nitrate concentration, and hence one key environmental impact of extensive livestock farming;
  • Disease resistance and the response of animals to a variety of disease and parasite challenges;
  • Behaviour of animals around people and other livestock, including how they handle stressful environments such as being moved; and
  • Longevity, the ability of female animals to raise progeny season after season, reducing the substantial cost of replacement animals.

There are proxies for many of these measures of course. Breeding for growth rates or milk production have arguably improved greenhouse gas efficiency for example, but in some breeding systems a change in mature weight of animals has increased emissions. Progeny tests and laboratory measures have been used in key programmes, but they may not help us with routinely identifying the genetic outliers that will lead the next leap in genetic progress.

New measurement and sensing technologies offer real potential to help with these “hard to measure” areas of animal performance in the coming years. Accelerometer and microphone technologies can identify individual animal eating habits, heats and parturition (birth) dates. 3D and multispectral cameras tell us about carcass and meat product characteristics, and additional characteristics of milk. Increasingly, this data will be collected in-line or in near-real-time, providing a rich stream of data that could be analysed for many purposes.

The next generation of animal recording and genetic analysis systems must be built to handle this variety of real-time, stream data: or at least the results of analysing it.

Fewer errors, more progress

A primary driver of any livestock recording and animal evaluation system is to enable breeders and commercial producers to make better decisions about the animals they use in breeding. Computers don’t select animals: people do. Where a producer chooses an animal because they like the look of its eyes, or its stance, or its colour, and ignores the potential impact of the animal on their herd, the results will be at best random, and often detrimental.

Formal breeding schemes with EBVs and indexes seek to inform better decisions about the breeding merit of animals, but EBVs can be limited by the information available:

  • Accuracy of recording parentage and animal relationships;
  • Incorrect allocation of records to the wrong animals;
  • Transposition and recording errors when capturing data; and
  • Failing to account for the impact of environmental effects such as the feeding and management regimes of groups of animals, the age of the mother, or whether an animal was reared as a single or twin.

Technology is playing a substantial role in improving the accuracy of EBVs, notably through genomic DNA analyses resolving the fraught process of parentage recording and contributing substantially more information, earlier in each animals’ life-cycle. Better facilitation and handling of genomic data collection is well overdue in animal recording systems, and I’m pleased to see this being addressed.

In addition to genomics, electronic identification (EID) and automated recording systems can remove many identification and data capture areas, and the ability to feed this data seamlessly into modern evaluation systems without having to manually manipulate data will provide another leap forward.

Recording management groups properly has been a real limiting factor in many breeding programmes, and is one of the key hesitations in extending these to commercial producers. I believe that sensors that identify eating and movement behaviours, and location or proximity to other animals, will help us to automatically and transparently solve the problem of recording management groups and regimes. This will provide another substantial step forward in removing the noise of environmental effects.

Of course, more accurate EBVs is still only a piece of the puzzle. Helping producers to make use of this information effectively is another, and something I’ll address in my next post.

 

Rezare Systems is a bespoke software design and development company specialising in the agriculture sector. We have special expertise in building livestock recording and management systems, and tools for data collection and integration. Learn how Rezare Systems can assist your business.