Saturday, 28 May 2011

Mate Select - good but no cigar (just yet)

Yesterday, the Kennel Club launched Mate Select - over-ambitiously hailing it as a breeding tool that will "transform" the health of pedigree dogs. (KC blurb here)

It does irritate the life out of me when the KC makes such bold claims and this one is on a par with one recent press release claiming that KC funding of the KC Genetics Centre at the Animal Health Trust would assist in the "eradication" of genetic disease.  Clearly nonsense. I was concerned, too, to read the AHT's Sarah Blott being quoted in the KC press release saying that Mate Select would "prevent" the decline of genetic diversity in pedigree dogs.  Mate Select will do nothing of the sort.

Indeed, in answer to one breeder's questions yesterday, Dr Blott replied that the hope for Mate Select is that it would "help to manage the rate of inbreeding - the percentage by which it climbs - to no more than 0.5% per generation."  Inherent in this reply is the acknowledgment that this is the best that one can hope for given the way pedigree dogs are being bred currently- which is never (or very rarely) to allow new blood in to halt the inexorable loss of genetic diversity.

I understand that the KC wants some positive headlines out of the launch of Mate Select - and it deserves some as it is a step forward. But the KC  must resist overstating the case in this way.

So does that make Mate Select a bad thing? No, it doesn't. It's actually a pretty good thing - and it has the potential to grow into something quite powerful. It's just that it's important to be aware of its flaws and its current limitations.

But first, let's look at the good bits:

Mate Select in its current form allows you to do three things:

• look up the co-efficient of inbreeding (COI) of a particular dog (ie how inbred it is) and compare it to the breed average.
• find the published health test results for an individual dog
• do a virtual mating. This is intended for use by breeders who already have a shortlist of potential mates for their dog to find out which would produce the least-inbred litter.

Now this is a facility that breeders in Finland have had available (in a rather different form) for some time thanks to a particularly progressive Kennel Club there. But, I have to say that Mate Select's "front-end" has been well-designed and it is extremely easy to use - and it is streets ahead of what is offered by some other Kennel Clubs (notably the American Kennel Club which should be making much more effort to tackle this issue).

Indeed, when you think back to where we were three years ago,  Mate Select is nothing short of a revelation. Then, there was very little recognition that inbreeding was a problem; barely anyone knew what a COI was, and there was very little understanding of why genetic diversity was so important - or why the lack of it was such a threat to pedigree dogs.  This despite the Kennel Club knowing the devastating findings of the Imperial College study that found, in the 10 dogs breeds they looked at, that most had lost 90 per cent of their unique genetic variants in just six generations. (Calboli et al, 2007)

We have, then, come a long way. I know too, that there has been a huge amount of effort put into the development of Mate Select and it is good to note the genuine excitement about it emanating not just from the team at the Animal Health Trust that has put so many hours into developing it,  but also from the Kennel Club itself. I hope they're enjoying the warm glow that comes from initiating (even it if was under duress originally) something that is genuinely a good idea.

Mate Select is free, too, and available to anyone with a computer.  And the goodness doesn't stop there - the team at the AHT (which includes Sarah Blott and the lovely Tom Lewis) clearly wants to improve it and is open to suggestions on how to make it better or more useful to breeders (see below for details of how to contact them).

Which brings me to the not-so-good bits.

Yesterday, there was a flurry of breeders, owners and researchers trying out Mate Select, both in the UK and abroad. And it had many scratching their heads, with the most common comment being: "The breed average COIs are much lower than I expected."

• Mate Select throws up an inbreeding average of  just 4.2% for Standard Poodles, for instance.  The much more comprehensive Standard Poodle Database, however, finds that it is around 17%. 

• Mate Select's inbreeding calculation for Irish Red and White Setters throws up a breed average of 16%.  But using a more complete breed database, the 10-generation COI for a typical IRWS is over 25%.

• Mate Select finds a breed average inbreeding of 2.4% for Nova Scotia Tolling Retrievers - but the worldwide breed database, which goes back to the handful of founders, shows that the true figure (for Tollers registered between 1999 and 2008) is 26%, making a Toller more than a full sibling to any other Toller.

In at least some breeds, then, Mate Select seriously under-reports the level of inbreeding.

So what's going on?

There are a number of reasons for the discrepancies. First,  Mate Select uses the Kennel Club's electronic database as its raw material. The database was not computerised until 1980 and and the very earliest dogs recorded in it (picked up from the pedigree data for those early dogs) were born in the 1960s.  Mate Select treats the first dogs in the database as unrelated founders - ie. of having a COI of 0% - when that will almost never be the case.

This is, in fact, the flaw with any COI program unless the data includes the true breed founders.  The Imperial study mentioned above was similarly handicapped. As you can see from the table below, it found that there were over 15,000 founders for the GSD, for example - crazy when the reality is that German Shepherds descend from just a handful of original dogs.

Click to enlarge

Mate Select also treats the furthest-back dogs in the pedigrees of imported dogs (between three and five generations are recorded) as having a COI of 0% which, except perhaps in the case of a new breed, is never going to be the case. In fact, those dogs that Mate Select marks down as having a COI of 0% could (and in many cases do) go back to original UK stock and in reality could be extremely inbred.  That Mate Select does not include this missing data means that the inbreeding data it generates is flawed - quite severely in some cases, and particularly in breeds where there have been a lot of imports.

Another reason some of Mate Select's breed-average COIs are low is because it is based on the COIs of registered dogs from the last year for which complete records are held (currently 2010). This, in fact, makes some sense as the current inbreeding levels are probably the most relevant. However, again, it does not reflect the true level of inbreeding and the figure could be very skewed in breeds with few litters registered per annum. A rolling mean of five years would, perhaps, be a better idea.

Although some breeders on the canine genetic lists have welcomed Mate Select, others have expressed concern that people will be falsely reassured as to the extent of inbreeding/level of genetic diversity in their breed. A low breed average COI suggests there is no need to panic when the reality is that some breeds are in dire straits. The very low genetic diversity in Tollers, for instance, recently prompted Finnish researcher, Dr Katariina Maki to call for an outcross to another breed (pdf of Dr Maki's statement here).

The other concern was that it will be impossible in some cases for breeders to breed litters with a COI below the breed average because of the way it is worked out.  Mate Select suggests that its breed average for Malinois is 2.2% (almost certainly because of very few litters, many sired by imports) - but a breeder on one show list yesterday said that it would impossible to do a mating using their dogs that would get anywhere near this low a figure. This makes it unnecessarily awkward for breeders - and confusing for puppy-buyers.

So do we write off Mate Select then? Well, no. It has its uses. Although it should not be seen as a measure of a breed's extant genetic diversity, COIs of individual dogs will be fairly accurate, at least in terms of recent inbreeding, unless there are imports in the pedigree (and in many UK breeds there may be none). More health/genetics data will be added in time. There is, I understand, a plan for the system to flag up when individual dogs have been over-used at stud, which would be very welcome (and long overdue) given the enormous damage popular sires can do. I hope there is a plan to include effective population sizes too - a more useful measure of genetic diversity than COI alone.

For experienced breeders, nothing will replace in-depth pedigree analysis (something that is impossible to do with Mate Select as it gives no pedigree information at all) and they will undoubtedly continue to use alternative breed databases, which are very often mantained by the most dedicated of breed enthusiasts.

But, for casual breeders and pet owners, Mate Select offers a really simple, intuitive way to find out a bit of info on individual dogs, encouraging them to think about genetic diversity and, particularly, to breed for lower COIs without all the hassle and hard work of trawling through endless pedigree information.

For that reason, it gets the PDE Blog's seal of approval. Albeit one with caveats. Another one being, incidentally, that the person who wrote Mate Select's guide to genetic diversity should be shot. It starts: "At any single gene, diversity can be measured as the degree of homozygosity (the inheritance of the same version of a gene from both dam and sire) which is observed.  The more homozygous a population the less diverse it is.  One of the main factors leading to increased homozygosity or loss of diversity is inbreeding. "

As Pat Burns ("Terrierman") observed in a private email to me yesterday:

One of the ways you can tell an expert is thay they do not feel the need to toss around four-syllable words to a lay audience.  Picasso, in the end, could draw a raging bull with one line.

An old sailor, who knows how to teach sailing, does not talk about port and starboard, bow and aft, pintles and gudgeons, rodes and stays.  He talks about left and right, front and back, hinges, ropes, and cables.  Paint it simple, say it simple.   And the more important the message the simpler it needs to be.

I quote the Bible when I talk about inbreeding. This is an ancient wisdom and even the illiterate pagans know it is wrong to lie with their sisters.

Fucking your sister?  Not a good game plan for humans OR dogs.  Said plain, and easy to remember, eh?

Now there's a strapline.

To help make Mate Select better, write to


  1. "... people will be falsely reassured as to the extent of inbreeding/level of genetic diversity in their breed - with a low breed average COI suggesting there is no need to panic when the reality is that some breeds are in dire straits."

    Given the talent for dog fanciers to cherrypick the "data" that supports what they wish to believe and ignore all else, I am afraid this detail transmogrifies my reaction from "The perfect is the enemy of the good" to "Oh dear. Worse than useless."

  2. For me, its clear to see that 'Mate Select' is in need of much fine tuning ! Knowing the COI is useful but the health status of each dog is much more important. Would it not be wiser to mate two HEALTHY dogs with a higher COI - rather than go for a low result based merely on In-breeding but end up with puppies, born to parents that are genetically compromised and passing on severe issues ?

  3. "But, for casual breeders and pet owners, Mate Select offers a really simple, intuitive way to find out a bit of info on individual dogs, encouraging them to think about genetic diversity and, particularly, to breed for lower COIs without all the hassle and hard work of trawling through endless pedigree information."

    Sorry, I don't agree with this at all. All the pet buyer will get is a false sense of reassurance from those numbers. I was literally shocked by the discrepancies in the dogs I looked at. And the fact that you must have the dog's exact name makes it worse than useless.

    The problem with import pedigrees is especially troublesome, because there are a lot of imported dogs in the UK show lines. is a much, much better example and one that should have been followed by the KC. You can enter a kennel name, part of a name, and get list that matches. Very simple and easy to navigate, does test breedings, COI including a complete COI for the entire known pedigree, and ancestor percentages.

    I AM a casual breeder, I AM also a pet owner, and I found Mate Select useless.

  4. Margaret Sierakowski28 May 2011 at 22:00

    I wouold rather know the health status AND the COI, and breed healthy dogs with a low COI from healthy parents with a low COI
    Mate Select will improve over the next ten years as the KC gets another two or three generations into the data base
    And even if the COIs they give at the moment are articially low, at least it has got a lot of people thinking and talking, and has increased awareness of the risks of high COIs
    The programme isnt perfect yet but its a good move in the right direction

  5. "Would it not be wiser to mate two HEALTHY dogs with a higher COI - rather than go for a low result based merely on In-breeding but end up with puppies, born to parents that are genetically compromised and passing on severe issues ?"

    Are those our only choices? Is it that difficult to find two healthy dogs that are not closely related?

    1. Even two inbreed dogs can be bred together to produce a low COI, when those individuals are highly inbreed but not related to each other, as are some imports! There are imports that arnt just bred from British stock and have their own genetic differences, producing pups with low inbreeding when bred with our indiginous dogs, Some imports having genetic diversity of their own country!!! and even if the COI number is higher than stated on the KC Mate Select site it still shows dogs that are less inbred when a hypothetical mating is used and a COI number assigned!

  6. A really good start. We need accurate as possible data to convince people this will help, otherwise ammunition is provided to counter such a good initiative.

    A further concern is the lack of comparability. We know COI has to be pinned to a generation for it to be meaningful Mate Select does not allow this, nor viewing the pedigrees behind the front end calculation.

    Thus in Finnish Lapphunds there are dogs with 8/9 gen on the system being compared to those with 4 generations of data. We need to correct this - by adding more generations now rather than waiting for reproduction to add these for us!

    A rolling 3-5 year mean breed COI would again be more representative and a visual trend analysis, help to illustrate the point. Data verification/entry error is also an does KC verify data of imported dogs in there cross-checking with other KC databases?

    Mary Starling

  7. "Would it not be wiser to mate two HEALTHY dogs with a higher COI - rather than go for a low result based merely on In-breeding but end up with puppies, born to parents that are genetically compromised and passing on severe issues ? "

    If you had a choice? Increasing the average COI of the two parents results in gene loss. Period. Since so few dogs are bred out of any one litter (only 20% in the UK), resulting in gene loss right there, yes, keeping the COI less than that of the parents is important if you want to keep that registry closed and not lose genes right and left. If a breeding that increases the COI is done, best to outcross and lower it again with the resulting pups.

  8. Jess said: "...Since so few dogs are bred out of any one litter (only 20% in the UK), resulting in gene loss right there, ..."

    Frankly... aren't most breeders attempting to DISCARD certain genes and increasing the homozygosity of other genes? For example, I breed a dome headed breed... I don't want flat skulls, so why wouldn't I discard those pups that do not have the skull type I need for my breed? Yes- that means gene loss... but "we" as breeders of this breed DON'T WANT those genes.
    In the same breed... we want a specific type of eye, a specific size, a specific type of coat...etc.etc.etc.... so why wouldn't we discard those dogs that do not have the traits we wantneed/require? and discard those dogs (and genes) ?

    This can be said for ALL purebred breeds.

    Inbreeding/linebreeding has its uses for producing/maintaining/increasing liklihood of specific type/structural points required within any given breed.

    One of the problems *I* believe which has led to todays current problem(s) in purebred dogs- is that breeders in previous years promoted the use of "carriers"... and how to "do it safely". There is NO way to use carriers "safely" within a breeding program. If you use carriers- you'll produce more carriers (and thus, more "affecteds")... Sure, testing is necessary - but MANY genetic ailments don't even show up until dogs are 5, 6 7 years old or more... by that time, many have already produced puppies, which have likely in turn also produced puppies... so those "carriers" have contributed MORE carriers along the way.

    INBREEDING and linebreeding in and of themselves are not the culprits...............

    Its earlier breeders use of (and PROMOTION of using) carriers --- which never should have been used in the first place--- but they justified it by using "structurally" and "type" strong dogs... I understand WHY..... but anyone with a brain should have realised what would happen to those breeds in future years.... producing more and more carriers.

    The future of purebred dogs NOW... is bleak.
    I honestly don't believe any new system will fix things because every breed is riddled with genetic problems... so even if kennel clubs encourage (or even force) breeders to outcross to other breeds... we're simply going to be introducing new (or more) carriers of more genetic problems than we may already have within our breed....

    It will never end.

  9. Sometimes I will see a comment or a post by a breeder that literally shocks me with it's ignorance. Since the advent of PDE, that has happened more and more often. Yours doesn't quite take the Moron Prize, Anon, but it's close.

    Any breeding within a closed registry loses genes unless it is very carefully managed, meaning that as many dogs as possible are bred, and contributions by any one dog are limited. For your information, this is how endangered species are bred. Line-breeding and inbreeding and choosing dogs based on very narrow characteristics by their very nature result in gene loss. Once those genes are gone you cannot get them back without an outcross, either to another line or another breed. Look at the increasing number of diseases being associated with homozygosity in the major histocompatibility complex. Selecting for homozygosity in regards to appearance results in a homozygous MHC, the immune system genes. Frankly, I'd rather have a dog that is healthy and at low risk for disease than one that has a perfect head. But that's just me. Perhaps you have *other* priorities.

    Your ignorance about using carriers in a breeding program is absolutely stunning. Luckily, not all breeders are so, for lack of a better word, stupid. Here's an excellent link about using carriers, for those who are interested:

  10. Jess,

    Please be more polite with your postings. There is no need to call people 'stupid' when no doubt they are well intentioned.

    I have looked at my Breed's average and it's far lower than expected.

    Not surprising though, when many, many of the early generations are not even included.

    Then we need to know what 'average' actually means ? Is it the average of all COIs based on say 9 generations, or the average of EVERY dog listed within a Breed ?

    Two of mine are less than 1% and the side message talks about 19 generations but also tells me that ONLY 4 generations have been used ?

    So, if my two extremely low COI results make up the AVERAGE - no wonder the Breed Average is what it is.

    What is the point in making reference to 19 Generations if only 4 are accepted. Does this not evidence the inaccuracy of KC records ?

    First Anonymous !

  11. What dome-headed-breed-Anonymous seems unable to grok is that you cannot discard individual "genes."

    You can only discard whole dogs. The whole organism is the unit of currency in the breeding of higher animals.

    So with that hideous insufficiently domed head, out go all the other genes residing in that horribly flawed individual.

    You cannot be homozygous for all the terribly important fancy traits and retain heterozygosity for trivialities, such as a functional immune system.

  12. Please do not rise to the bait of anon dome headed dog lovers comment. They are quite obviously trying to provoke.
    Pity them and their hydrocephalus dogs.
    Now, if you are genuine dome headed dog lover, stop being a coward and tell us your breed and name. Quite simple, and very easy.

  13. I fully support any effort get the "idea" that genetic diversity matters and that line- breeding IS inbreeding and we cannot carry on in a similar way for the next 40 years across to breeders.
    So well done KC....for all its flaws its a start and it available and has been said a few years ago the idea was heresy in the dog world.
    Have to add still is for the majority !

    The part that made me sit up and read again was the explanation of COi levels
    It clearly says the KC considers a COi of over 25% an unacceptable risk and has therefore banned father/daughter, mother/son and sibling matings , surely then any COi above 25% whether its the result of close inbreeding or the result of prolonged line-breeding over many gens poses a similar unacceptable risk and should likewise be banned.

  14. Yes, Jan, it should - and percentage at ban threshold should be much, much lower than 25!!!

    Anonymous of the dome-headed dogs, I for one think you courageous. Why is everybody angry? He/She is telling the truth!

    IF what you breed for is a thing called "type" and that thing is becoming progressively more narrowly defined - e g, if THE important thing is the dome-shaped head, and such and such a length and twist of tail, and such and such a set of ears and mass of wrinkles. Why, she´s right! Inbreeding is the only way to achieve it and the more of it, the better!
    That is the way in which the modern versions of a number of old "breeds" (dog varieties) were created. They didn´t occur naturally as "type" in the sense "conformation" - they were created. They were crafted. Right? The old breeders sacrificed very much for that "type" and the more recent ones have gone on sacrificing - intelligence, capability, health, movement, immune system function - but they got what they opted for - "type", ribbons and praise. In effect, they traded genetic diversity for a domed head and are pleased with the swap. Some of them even realize that you can´t get both, unless you have a very large number of founders and a very large number of years to start with. Right?

    OK, some of us here don´t want to breed or own domed heads. We like to breed or own dogs.
    May they vary in looks, as long as there is health, we think. May they look as they wish, as long as the "type inside" - the Collie behaviour, or the Terrier behaviour - is there!

    So ideas about what dogs are differ, but why think Anonymous is provocative? I think she tells plainly how things look from her horizon.

  15. Dear First Anonymous-

    This past week, I sat with my heart in my mouth, waiting to hear whether the latest anti-breeding bill in my state (I live in the US) had passed into law. I have done this very same thing, every year, for three years now. The past couple of years, the breeders group in my state has been able to talk sense about how these laws will affect ethical breeders adversely, and they did not pass. This year's law did pass. Luckily, it was amended in such a way that my breeding program will not be affected.


    After the Purebred Paradox conference, HSUS, who is behind or endorses pretty much every anti-breeding bill here in the US, has a whole new bag of tricks. Now they will write bills, these people who have never bred so much as a hamster, using the principles behind population genetics, etc., principles that should be applied by the individual breeder, NOT BY LAW, and these proposed laws will probably pass.

    Why, you ask, am I so convinced that the nitty gritty of my breeding program down the COI of each litter will be placed under the auspices of the state, instead of under MY auspices, where I can make the best decisions for MY dogs? Because of breeders like our carrier-hating anonymous, who have NO FREAKING CLUE about genetics, or how to apply them. These people will be left gasping like fish out of water when called on to answer to the assertions of the HSUS Center for Science and POLICY. They are going to have to understand the science. Period.

    These people are going to drag my breeding program down to hell with them, and after arguing with them for almost two years now, I reserve the right to be rude. Don't like it, don't read it.

    Jan: absolutely not. The ONLY way to set a realistic COI threshold is to know the average COI of the breed. Set it at the average, then reevaluate the pedigrees in ten years time, and lower it to the new average. It will take time, but this is not something you can be arbitrary about. In some breeds it may be relatively easy to get a low COI, in others, near impossible.

    Or you could set it up so that the litter COI must be lower than the average of the parents. This will result in a lowering of the COI across the board in the breed, without the problems associated with an arbitrary number.

  16. Jess you are clearly a troubled person !

    One has to read something before making a decision as to it's content.

    If you are fundamentally rude - please find the Gene causing this and do something about it.

    I do so hope that you Breed dogs that are far more friendly and socialised than you - seem to be.

    It's never to late to self reflect ?

    First anonymous.