Introduction

Despite widespread acknowledgment of the important contribution of the Australian livestock-working dog to livestock industries and the rural economy, many aspects that influence livestock-working dog performance and breeding and selection success have not been previously been quantified. To optimise dog performance and success rates, and thus minimise so-called wastage, research is required to provide evidence-based information of direct relevance to those who breed, train, handle, work and trial livestock-working dogs.

In our ambitious project we are interested in two questions.

 The first is:
How much influence does genetics have on the various working behaviours and working dog types?

 A second question is:
Can we produce a platform through which working-dog breeders can improve selection of working dogs?

Each question required a different approach. First, to assess the influence of genetics on working dog success, we examined the problem from two angles: both from the bottom up (from the DNA to the behaviour) and from the top down (from the behaviour to the underlying breeding value).

1. How much influence does genetics have on the various working behaviours and working dog types?

Bottom-Up Approach

To try and discover the individual genes that have the most impact on working behaviour, geneticists typically use panels of genetic markers to locate the influential genes. For this project we employed a combination of genotyping arrays (which comprise vast samplings of single letter differences in the dog DNA at over 170,000 locations in the genome) and also next-generation sequencing that enables us to read nearly every letter of the dog's 2.5-billion letter DNA genome.

The value of using genotyping arrays is that they have relatively low cost and are relatively easy to analyse. This allows us to examine trends in many dogs. However, this ease comes at the cost of being able to get only a quite low resolution picture of what is happening in the DNA. To overcome this, we try to use as many dogs as possible and to sensibly group dogs for comparison, so that in the broader DNA landscape the dogs have very little difference between them, but if they then have quite different working behaviour, this will enable the signals that come from the relevant genes to 'stand out' from the genetic background. We can also use families to assess the differences, and this tends to give a lower intensity but broader signal in our data.

One method that we employed was to use the arrays to compare the DNA profiles of Australian Working Kelpies (Working Kelpie Council registered) with Australian Kelpies (Australian National Kennel Council registered) (Arnott, Peek, et al. 2015). By examining a modest group of dogs from each cohort, we were able to detect regions of DNA that were relatively stable (fixed) in each group and where the groups were quite different from one another in the segments of DNA beneath the signals. The results revealed that the major selected factor in working success for WKC centred on genes that enable dogs to feel pain and to form fear-based memories.

The full DNA sequences enable us to look between the markers and determine what is happening at the individual DNA letter level. The DNA differences that underlie subtle differences between dog behaviours are unlikely to be as stark as those that completely disable the production of proteins. Instead, we expect to find subtle differences in the switches and dials that promote the fine control of gene expression. Proving causation for these observed differences is often difficult. Nonetheless, it is well worth doing because it allows us to better understand the biology of behaviour. We expect work in this portion of the project to continue for some time.

Top-Down Approach

Another way to understand the genetics of behavioural differences is to apply a top-down approach. For this portion of the project, we have applied statistical genetics techniques based on the actual observed differences between behaviours within families of dogs to calculate heritabilities for the various behavioural traits. These analyses tell us that many of the traits that working-dog breeders and handlers care about have strong inherited components. This analysis justifies our use of the bottom-up approach to investigate the genetic differences further.

2. Can we produce a platform through which working-dog breeders can improve selection of working dogs?

A significant outcome from the top-down approach is that it provides the framework for a practical solution that will help working-dog breeders to better select breeding dogs and to identify other kennels that have similar breeding goals to their own. Those in the livestock industries may be familiar with the concept of estimated breeding values. Estimated breeding values look at the traits that are similar for dogs within families and different for dogs between families. The breeding value of a dog is its predicted capacity to pass on good genes (which need not be individually identified) to its progeny. This is the method that is scientifically preferred to enable breeders to achieve genetic improvement in traits that are considered to be complex. Estimated breeding values account for environmental differences between dogs. Environmental differences can be things such as access to high-quality training and exercise.

The major part of this project was to devise a method by which dogs could be assessed to identify these differences. We needed to compromise between evaluation methods that might be highly accurate (such as individual assessments by trained animal-behaviour consultants) with those that are accessible to a broader range of participants. In the end, we elected to base our ultimate breeding value calculations on owner-handler-based assessments of dog abilities. While we appreciate this may have lower accuracy than other methods, we still expect that by sampling a broad range of related dogs, we will ultimately determine the flow of good genes among the dogs that will enable us to give some overview of their relative talents. By sampling more dogs, we can use statistical averaging to arrive at higher accuracy evaluations for breeding dogs higher in the pedigree.

Working Traits of Importance to Kelpie Breeders and Handlers

The training of the Australian Working Kelpie has been a topic of several books designed to assist trainers and handlers to get the most from their dogs. We saw these publications as a solid resource to better understand the traits perceived as important in working dogs under Australian field conditions. The publications were analysed to determine the frequency of authors' use of common terms that describe working manoeuvres, skills, and general behavioral attributes. For the purposes of this study, we defined working manoeuvres as a sequence of movements used in herding; working skills as an ability used in herding, learned through training, or experience; and general behavioural attributes as personality traits ascribed to an individual dog. The initial purpose was to obtain an indication of behaviors relevant to herding dogs. Additionally, we sought to test the level of concordance among authors in the use of the most common terms (Early, Arnott, et al. 2014).

Figure 1. Top 10 terms for general behavioural attributes

Figure 2. Top 10 terms for specialist working attributes of farm dogs

Next, we conducted a general survey of farm dog owners and handlers. By this survey (accessible via: http://sydney.edu.au/vetscience/research/animal_behaviour/farmdog/), we were able to accomplish a number of goals of import to our wider analysis. The first was to assess the economic value of the livestock-herding dog (Arnott, Early et al. 2014), while another was to establish environmental factors that might impact working-dog success (Arnott, Early, et al. 2014).

Can Dog Handlers Assess Their Own Dogs?

An important goal in establishing a viable working dog selection program is to have a tractable means of assessing dog phenotypes. We assessed owner-handler abilities to assess dogs, by having the owners score their own dogs (and those of other competitors) by questionnaire based on viewing a common anonymised video of dogs working during a "yard trial." The questionnaire results of the owner handlers were then compared with the assessments of the same dogs made by Kelpie handlers and trial judges that were generally regarded as "expert" according to the Working Kelpie Council.

The level of agreement between the groups of video reviewers was very good providing confidence that non-expert reviewers of dog skills can still yield usable information for genetic analysis in a practical context.

Table 1. Agreement between owners' scores of dogs versus experts' scores

NB: Traits 'cast' and 'force' were measured on the quality with which they were performed whereas 'bite' and 'bark' by the frequency with which the dog performed them when working.

Heritabilities of Livestock Working Traits in Australian Working Kelpies

While to date we are in the early phase of data collection (the majority of the work of the project has been related to the design of the evaluation resources), we have nonetheless been able to determine some preliminary heritabilities for important farm dog phenotypes. Heritabilities are one of the key genetic parameters required to produce estimated breeding values for working behaviour traits. The estimated breeding values will provide a tool to assist Kelpie breeders to identify other breeders producing dogs comparable to their own, or with traits that complement those they are seeking to develop in their breeding programs.

Table 2 shows the preliminary heritability estimates (between 0 and 1) and associated standard errors for behaviour traits in a cohort of Australian Working Kelpies

Summary

The provision of a resource that quantifies the performances across a range of traits will provide an unprecedented opportunity for breeders to identify other lines of dogs with qualities similar to those that they value in their breeding program. Breeders will be readily able to identify other kennels from outside of their local area that produce dogs like their own. This will enable them to make better use of outcrossing in their breeding programs. There will be reduced risk of purchasers obtaining dogs that are of a kind unsuited to their needs.

Our study has not only already contributed much new information about kelpies, their special qualities and how to get the best of out of them but has also put in place a process for assessing dogs on their performance and breeding potential. This has laid a strong foundation for ongoing research and development that will continue to deliver ever more detailed information to interested parties.

This project was funded by Meat and Livestock Australia (MLA) and the Rural Industries Research and Development Corporation (RIRDC) with invaluable in-kind support from the Working Kelpie Council of Australia (WKC).

Other participants in the Farm Dog Project are: Professor Paul McGreevy (lead researcher), Dr. Elizabeth Arnott (PhD Scholar), and Dr. Jonathan Early (PhD Scholar).

References

1.  Arnott ER, Early JB, et al. Environmental factors associated with success rates of Australian stock herding dogs. PLoS One. 2014;9(8):e104457.

2.  Arnott ER, Early JB, et al. Estimating the economic value of Australian stock herding dogs. Animal Welfare. 2014;23(2):189–197.

3.  Arnott ER, Peek L, et al. Strong selection for behavioural resilience in Australian stock working dogs identified by selective sweep analysis. Canine Genetics and Epidemiology. 2015;2(6).

4.  Early JB, Arnott ER, et al. Manual muster: a critical analysis of the use of common terms in Australian working dog manuals. Journal of Veterinary Behavior - Clinical Applications and Research. 2014;9(6):370–374.