A Plea for Scientific Thought in Breeding


There's a common feeling among dog breeders right now that bitches are "missing" more often than they used to -- that it's harder to get a bitch pregnant than it used to be.

Is this true?  Who knows?

If it is, why?  Again, who knows?

Many people today believe that more dogs are suffering from inherited health problems (such as hip dysplasia) than used to, and lots of people think they know why.  Is it true?  And if so, which of the proposed explanations is correct, if any?  How can you tell?

Lots of people are just certain that whatever extra-special diet they choose to feed their dogs is way, way better than whatever you're feeding your dog.  Why do they think so?  Why are they so deeply committed to whatever diet they've chosen for their dogs?  How can you tell whether you should change what you feed your dogs?

There are a lot of casual theories out there about all kinds of things, some of them intelligent and some irrational, very few tested.  Often, no one is even sure whether there's really a problem or not, or whether the question they have is a legitimate question.  Or rather, some people are sure, but they may well be wrong.

How is anybody supposed to know what's really going on?

Let's take that first case:  are bitches "missing" more often than they used to, and if so, why?  Finding out the truth would be more helpful than spreading questionable theories or changing how you manage your bitches depending on 'what people say', possibly for the worse.  This is a good example of a problem that anybody could investigate.


People who do not have a science background generally skip this step.  They feel like there is a problem, they segue from this to being convinced there is a problem, and they never look back.  Why do people commit so strongly to beliefs that have not (generally) been carefully thought through, much less tested?

The biggest problem with rational thought is that people aren't rational.  Scientists aren't either, but they learn to compensate:  that is what scientific training is for.  (You won't have been told that in your science classes, I bet, but it is true.)  The biggest logical flaw that gets in the way is called CONFIRMATION BIAS and it is an intrinsic part of being human.

Confirmation bias occurs when you first frame a hypothesis to yourself:  Hey, I think bitches are missing more often than they used to!  This sort of idea is likely to occur to you if you ever happen to have two bitches in a row miss.  As soon as you have this thought, you, being human, will start to notice evidence that supports your idea and fail to notice or forget evident that doesn'tEveryone does this.  It is impossible to not do this.  There are two ways to compensate:  first, be aware of the phenomenon (and really believe that it is a very strong bias and that you are not exempt).  Second, much more important, collect data.  What being aware of confirmation bias does for you is prod you to go to the trouble of collecting data so you can check to see if your feeling is right.

If you go back to your actual on-paper breeding records and count matings that occurred from, say, 1980 to 1990, and calculate the percentage of misses, and then do the same for the period from 2000 to 2010, you will have a good idea of whether the percentage of failed breedings has changed.

You do keep records, right?

Okay, if you didn't keep records or if you weren't breeding in the 80's or you didn't ever breed enough litters to get your sample size up enough to be meaningful, then you will need to find some long-time breeders and get a look at their records.  But it is in principle very easy to check and see whether there is actually a problem or whether you just feel like there is a problem.

What if there is a big difference?  Like you did the analysis and saw this:

  1970 -1979 1980 - 1989 1990 - 1999 2000 - 2010
% bitches bred that whelped a litter 91% 93% 92% 76%

I just made those numbers up!  But if you saw this, that would count as a really big difference!  Especially with (probably) more breeders using progesterone testing to pinpoint the correct days to breed, it looks like a huge problem exists.  But what if you got a lot smaller difference, like this:

  1970 -1979 1980 - 1989 1990 - 1999 2000 - 2010
% bitches bred that whelped a litter 91% 93% 92% 89%

How can you tell whether this represents a real decline or just a statistical fluke?  You would need either to do a simple statistical test -- your math department at your local college, or your kid's high school math teacher should be able to help with this -- or collect more data, or both.  Small difference are often no difference.  You want to err on the side of caution:  if you do not see a big difference, you more than likely do not see a difference at all!

But what if your small decrease is really pointing to a real difference?   Small sample sizes lead to big problems with knowing whether a difference you might see is real or not.  Only ten litters per year?  Not enough.  Get some friend's mating records and add her data to yours.  Get several friends.  You may find that as you add data, the pattern either evens out to show that there is no change over time, or that minor changes over time become big changes.

What if you find out that there actually has been no change in the proportion of missed breedings over time?

If there is not a problem, then there is not a problem and just feeling like there is doesn't make it so.  Just because two or three or four of your personal bitches missed in a row doesn't mean there is a broader pattern at work.  If there is nothing there, there is nothing there.  Do yourself a favor and turn your attention to something more useful, like figuring out why your own bitches are missing.

People are always wasting a tremendous of time and energy and money trying to solve problems that don't exist, like worrying that computer screens or power lines cause cancer, or that alar on apples did, or that silicone breast implants caused whatever they were supposed to cause, or that there is a "sudden acceleration syndrome" that afflicts some cars when you hit the breaks, or what have you.  It is not okay to worry about problems that don't exist.  The time and energy and money invested in solving imaginary problems could go toward solving real problems, rather than just being burned for no reason.

A background in science will make you learn to let go of theories that are wrong.  At least sometimes -- scientists can cling like grim death to wrong theories too, while evidence piles up against them like floodwaters behind a dam, until the dam finally bursts and they are knocked loose.  This is like what happened when all of geology was finally forced to admit that continents do in fact drift, for example.  But science does have a way of forcing a focus on real evidence and thus converging eventually on something like truth.  Whereas there's no limit to how long nonsense can propagate through society, if it 'feels' right and nobody's checking.


2)  BUT IF IT TURNS OUT THAT BITCHES REALLY ARE MISSING MORE OFTEN THAN THEY USED TO, WHY?  In other words, what do you do when you find your first impression was right and there really is a problem?

a)  Maybe modern overprocessed pet foods, laden with chemicals and dyes, are to blame?

b)  Maybe there are some nasty new strains of canine herpes that has become prevalent, and many bitches are catching it when they go to the stud dog and then aborting their litters?

c)  Maybe dogs are catching West Nile Virus?  Like humans, they barely show symptoms when they catch it, but maybe if a bitch catches it for the first time while she is pregnant, she will abort her litter?

d)  Maybe purebreds are becoming too inbred and, like mothers who are Rh negative, bitches are now tending to have incompatible blood factors which kill their puppies?

e)  Maybe a whole lot more breeders are using AI's (artificial insemination) and the AI techniques are not always sound?

f)  Maybe ordinary household cleaning solutions are poisoning bitches and causing them to miss?

g)  Something else?

How do you know which of these, if any, are contributing to the problem?

There is a very strong tendency for people who do not have a science background to think of one of these possible explanations, or of some other explanation, and then to cling to that theory come hell or high water.  Nothing will knock them loose.  They are not interested in evidence -- or at least, they only notice or pay attention to evidence that supports their theory.  This goes double -- triple -- quadruple -- if the theory is connected to some broader ideology, as with a belief in the evils of modern pet foods.

It is a lot more useful to ask yourself, What should I see if hypothesis 'a' is correct? And what evidence would disprove hypothesis 'a'?  And then march right down the line of possible hypotheses, testing each one in turn.

For example, you would go back and collect actual data about what you were feeding your bitches at various time intervals, and you would go collect data from other breeders.  Breeders who made one dramatic switch in feeding protocols would be especially useful.  So would breeders who did not switch foods at all.  Anybody who was switching every year from one fad to the next would be very annoying to fit into this data set, maybe impossible.  Leave them out.  Compare the percentage of bitches that miss given different feeding protocols.  Then you would be able to check and see whether there is a correlation between pet food fed and likelihood of missing a breeding.

If canine herpes, an easily communicable virus that will cause abortions and death of young puppies, is to blame, one would expect young bitches to miss their first breedings and then be fine after that.  After they'd caught it, they should be immune.  If several strains were being traded around, you'd still expect to see younger bitches missing more often, but they might be vulnerable to several different strains.  You'd also expect to see postnatal puppy mortality from herpes, which would be a lot uglier than just a miss.  There's other things you could look for:  maybe bitches that have been shown a lot and just been around other dogs a lot would not tend to miss nearly as often as bitches that had been kept at home?  After all, the girls who are out at shows should catch lots of things and be immune to them by the time they come to be bred.

If West Nile was to blame -- something that occurred to me because a beagle guy I know reported a LOT of misses the same year WNV was such a news item -- then you'd expect a huge uptick in misses in any area of the country right after WNV got established, then a return to normal after the bitches caught the virus and became immune.

Rh or other similar incompatibilities ought to be associated with smaller litter sizes, not just misses -- I think. Not an expert.   I'm not sure dogs have problems like this, but maybe.  An increase in inbreeding per se does not make sense, by the way, because different breeds vary a lot in how inbred they are, whereas we're assuming an increase in misses across all breeds at a particular time.  Logic is very helpful in weeding out weak hypotheses.  If you cared to test this, you would compare misses in a more highly inbred breed vs a more outbred breed, like Canaan dogs, which were feral until very recently.

Remove all AI litters from your analysis and you should be able to find out whether misses are increasing independent of the use of AIs.

Some chemicals used to clean labs have been shown to have a very deleterious effect on reproduction in mice.  You can't generalize from this to, say, Lysol.  But could Lysol have an effect?  Who knows?  You could compare the breeding success of people who use only bleach, like me, with people who use Lysol.  There's a theory that antimicrobial soaps (which really may tend to let resistant strains of bacteria build up in the house) may be causing a problem. I don't know.  Dogs are so thoroughly exposed to horrible bacteria in the yard and park and street and so forth that it's hard to imagine antimicrobial soaps having any effect on them, no matter what effect they have on bacterial growth in the house.

And if none of the prevalent hypotheses explain a phenomenon that is nevertheless real?  Back to the drawing board, and do some thinking.  Look at different breeders' records and see if any patterns emerge.  One doesn't form hypotheses in isolation, but in response to what one sees in the world.


Why do I get to write about scientific thought?

I have a master's degree from the U of Illinois from what was then the Department of Ecology, Ethology, and Evolution.  I went in as a PhD candidate but found I really did not care for the research end of things -- collecting data as a profession can be grindingly tedious.  I don't mind analyzing data, though I'm no stats whiz; there's a hard-to-put-into-words but very real excitement to watching information emerge from a heap of data.  I like reading the results of other peoples' research, depending on exactly what their work involves, despite the desperately boring style in which professional articles are written in academia.  I'm still interested in ecology, animal behavior and evolutionary theory.  I'm also interested in genetics (can you tell?), medicine, psychology, sociology, anthropology, history, politics, and economics.  My brother is a history guru and I think I absorbed some of the social science interests from him.  I teach.  I don't "do" science, not professionally. 

So, no, I'm not a scientist.  But I have a background in science, and years and years of thinking about science.  And I got tired of watching people leap to unsupported conclusions and then cling to them like their hope of heaven.  Which I'm sure not everybody does.  But too many.

Stuff I've personally heard from people who actually breed dogs and should know better:

"But hip dysplasia can't be genetic -- it skips generations."

"But she got out and was bred by a mutt, so all her puppies now will be mixed breeds."

"Well, yes, his shoulder is upright, and hers is too, but you really can't tell what you'll get in the puppies -- it's all chance anyway."

"First they say that raised food bowls help prevent bloat, then they say that they cause bloat -- you just can't tell what's true, it's all guesses."

"Just take her out of the ring and breed her -- having a litter will mature her mentally and physically."

These statements range from so deeply wrong you can't even tell where to start to rebut them (top two);

to wrong, stupid, and likely to lead to structurally unsound puppies (third);

 to idiotic and lazy -- for the Lord's own sake, track down WHO said what and when you find out it was an excellent study from Purdue that discovered that raised food bowls cause bloat, stop worrying about How can we know truth and get rid of your raised bowls;

to untested but silly on the face of it.  I mean, take her out of the ring and breed her and it will be a year before she's back in the ring.  After a year she will be a year older, which is frequently associated with maturation.


    *  The desire to test a belief is part of scientific thinking.

    *  The ability to come up with multiple possible explanations for why something might work is part of         scientific thinking.

    *  The ability to prevent yourself from "falling in love" with a hypothesis and looking only for data which supports it (confirmation bias) is part of scientific thinking.

    *  The ability to look for proper data that already exists, or to design an experiment that will allow you to test a belief or a possible explanation, is part of scientific thinking -- as is knowing that your experiment should be designed to allow your hypothesis to be disproved.

   *   The understanding that it will take methodical measurement of different characteristics throughout the experiment to reach a conclusion -- not just a general subjective impression --  is part of scientific thinking.

    *  The ability to draw a conclusion based on the data and stick with it, even if it doesn't agree with what you expected to begin with, is part of scientific thinking.

   *   The ability to see how your conclusion should be restricted (how generalizable it is, or isn't) is part of scientific thinking.


Is this of practical use?

If you change the nutrition regimen for a group of dogs and record their response to the change, that is an experiment.  That is what dog-food companies do.  Or if you want to give folic acid to some bitches and not to others and see how this particular nutrient affects the proportion of puppies born with birth defects, that is an experiment.  Any time you are manipulating part of the environment to see what effect that has on some aspect of dog growth or development, you are doing an experiment -- or you are if you are being methodical, rigorous, and keeping good objective records of what you do and what you find.

If you want to determine the mode of inheritance of a trait, that will be observational.  You may set up the conditions under which the observation will be done -- you choose which bitch to mate to which dog -- but fundamentally you will be counting puppies with trait A versus trait B.  I did an exactly similar observational study out of casual curiosity when I found that petunia seeds I had collected were producing some white seedlings (true whites -- no chlorophyll) -- I simply counted green and white seedlings as I thinned them and compared the ratio I got, from about 150 thinned-out seedlings, to the expected Mendelian ratio for a single-gene autosomal recessive trait.  I drew on background knowledge to make the hypothesis that this was a single-gene autosomal recessive trait.  Most good hypotheses have reasons behind them; they aren't plucked randomly from the aether.  In my case, I happened to know that corn seeds with a similar white mutation are sold by biological supply companies for use in teaching genetics, and in corn the mutation involves a simple autosomal recessive.  (The data fit this hypothesis quite well for my petunias).

Test breedings of all kinds fall into the observational category of study.

When Isabell (Introduction to Genetics for Dog Breeders) noticed that some male puppies showed an early descent of testicles, she did a very nice, if limited, study of the correlation of early testes descent with cryptorchidism -- a fine example of precisely how a single breeder can, and ought to, contribute to the base of knowledge that all breeders can use to improve their breeding programs.  Any single breeder can only do a limited study -- small sample size is the most important constraint on a single breeder.  But even a very small-scale study can yield strong suggestions about how to avoid cryptorchidism, or whatever.  Isn't it nice to know that the daughters of males with early testes descent tended to pass on less cryptorchidism than average to their sons?


That's all very well, but could I, as a small breeder, actually do any kind of useful experiment?

Yes, you could.  After all, Isabell did!

If you want to find out whether docking dewclaws inhibits puppy growth or interferes with normal personality development, you could do an experiment -- you could dock the dewclaws of half the puppies in each litter, choosing the puppies at random, and record subsequent growth, vigor, and personality traits.  You'd have to decide how to define "vigor" and how to describe personality traits.  It would take a lot more than one litter to reach conclusions -- tiny sample sizes are one of the most common and most serious sources of error in reaching sound conclusions.  But over time, even with small litters, you could accumulate enough puppies to make your own personal study meaningful.  (Meaningful for your breed.  There is no reason to expect your findings to apply to breeds with substantially different personality type -- Cavaliers are not the same as Basenjis!  It's important to restrict your conclusions.)

You could do something similar if you decided to feed half, but only half (and randomly selected!) of your dogs a bones-and-raw-foods diet.  You'd have to decide what to record and how to record it in each case.  You could decide to socialize every other litter with children at three weeks of age and see how the puppies react to children when they grow up.  How would you define behaviors when watching puppy-child interactions?  You could let half your puppies watch their trained mothers enthusiastically retrieve objects when they are six weeks old and see if this treatment tends to produce puppies that retrieve well when they start their own training at six months of age.

These sorts of experiments are hard to do, not because the question you're trying to answer is necessarily particularly intractable, but because the tendency is to make an a priori (before testing) decision that treatment X (BARF diet, say) is probably the best diet.  Well, if it's the best, don't you want all your dogs on that diet?  So rather than try to find out whether the diet actually makes a difference by doing a randomized experiment, you just put all your dogs on the diet and then try to spot improvements that might be attributable to the diet.  This is a very strong impulse, but it leads to very weak conclusions.


Even when you don't, or can't, conduct your own tests, you ought to be evaluating the claims that other people make from a scientific perspective, not from a credulous perspective.  Especially sensational claims of the I alone have discovered how to defeat The Establishment and restore Our Dogs to the Natural Health that is their Birthright type.  Yes, it's nice to be open-minded -- but not so much, as they say, that your brains fall out.

You can't know everything about everything -- suppose you encounter a claim that eating dark-green vegetables is good for you because the chlorophyll in the vegetables provides you with oxygen (a claim actually made by a particular "nutritionist" of the current day); I suppose that if you know nothing about botany and less about physiology, you might not instantly realize how utterly nonsensical such a claim actually is.  It's another example of so wrong you can't even start to explain why its wrong.  However, if you come across a claim that on the face of it seems odd and off-the-mainstream, then, especially if it's got that tell-tale Only I have discovered tone, why not read up on the subject a bit so that you can make a reasoned judgment?


What is the alternative to scientific thinking?


If a breeder can't, or won't, form rigorous questions, ask for good evidence, collect hard objective data -- then she will almost certainly fall into one or more of the very common fallacies that afflict us all.