Eldin Leighton, PhD (pt 1): The Science of Animal Breeding

by Sep 17, 2020Genetics Research, Physical Health/Disease, Podcast, Working Dogs0 comments

Jessica Hekman: Welcome to the Functional Breeding Podcast. I’m Jessica Hekman, and I’m here interviewing folks about how to breed dogs for function and for health: behavioral and physical. This podcast is brought to you by the Functional Dog Collaborative, an organization founded to support the ethical breeding of healthy, behaviorally sound dogs. FDC’s goals include providing educational, social, and technical resources to breeders of both purebred and mixed breed dogs. You can find out more at www.functionalbreeding.org, or at the Functional Breeding Facebook Group, which is a friendly and inclusive community. I hope you have fun and learn something.


Jennifer Hekman: Hi, friends. This week I’m talking to Eldin Leighton. Eldin has a Ph.D. in Animal Breeding and has wound up a long career managing the breeding program at The Seeing Eye, which is the largest guide dog organization in the U.S. In this episode Eldin talks to us about how he got where he is now, and gets super nerdy about the ins and outs of selecting breeding animals for the traits you want. From cattle to seeing eye dogs. We love nerdy here. Eldin has had so much good information that this ended up being two episodes. So stay tuned for Part Two next week.


JH: Eldin, thank you so much for joining us today. I think you have a lot of expertise that’s going to be really useful for people who listen to this podcast. Normally I start out by asking people to talk about the dogs that they live with, but since you live with Jane’s (Russenberger) dogs then that is not going to be helpful for people to know. They’ve already heard about it. But I think it would be… Oh we lost one! There was one right there on the camera, but we lost one.


Eldin Leighton: She moved closer.


JH: So she could participate.




JH: But I think what is going to be really useful for people is to sort of know some details about your history. So how did you… You have a Ph.D. in Animal Breeding. How did you get into that area of work? How did you decide to go, to learn about that?


EL: In high school I was really fortunate to have a high school science teacher who came to my little high school in northeast New Mexico when I was in eighth grade. She taught eighth grade science. Her value was recognized and her expertise was recognized by the school system as exceptional. And so as my class advanced through the years, they kept adding more and more classes to her schedule. And for four of those five years that I was in junior high and high school, she was my science teacher. She was phenomenal. And she’s the one who got me hooked on genetics. So I knew when I went off to college that I wanted to study genetics. I was interested in beef cattle. I grew up on a ranch. Beef cattle were my life. And that’s what I went off to study, with an intent to apply genetics to the improvement of beef cattle. I had read about in the genetics texts in high school about mapping chromosomes in the fruit fly. And I had visions of maybe being able to help map the bovine genome, but that didn’t turn out to be part of my history. But she’s the one that got me hooked on science and genetics and I owe a lot of my direction in life to her early guidance. She was wonderful.




JH: That’s, that’s really lovely. I had such a bad experience in high school with science that I had to wait until after college to come around to it. So important when people get really good instruction on science early on in high school.


EL: She was the daughter of a medical doctor in Albuquerque who had trained at Tulane down in Louisiana, and during World War Two she was actually a personal secretary to Robert Oppenheimer.


JH: She sounds incredibly accomplished. So she inspired you, and then where did you go after high school?


EL: She did. Immediately left northeast New Mexico for south central New Mexico and New Mexico State University. I was there for four years. Graduated on schedule in the spring of 1971. I was accepted into graduate school at Iowa State for the fall of ‘71, moved to Iowa, learned what really cold winters are like that winter, and entered graduate school there to ultimately obtain a Ph.D. in Animal Breeding if I was smart enough to make it through the curriculum. And so two years later I had finished a master’s degree when I knew that I owed military service to the US government. And so I went on active duty through the ROTC program as a First Lieutenant in the Medical Service Corps, and spent three years on active duty working as the research geneticist at the US Army’s Division of Biosensor Research, which the popular press named The Superdog Project. That was a project started in 1968 by the US Army Veterinary Corps to genetically improve the military working dog. German Shepherds were widely used in Vietnam in many of the infantry platoons, so the dogs were playing a critical role. And the government was having a hard time then, just as they still are now, in finding an adequate number of high quality, healthy working dogs to meet those needs of the US military.




JH: I had no idea you are on that program. That is a program that a lot of people know really well because early neural stimulation came out of that program. And ENS is something that a lot of breeders use today to help socialize their puppies really early on in their puppies lives.


EL: Yes. Dr. Battaglia and I have had a few interesting conversations about the program. He’s a big proponent of ENS. And I think he often gives the US Army credit for having introduced it to the dog world or something like that. But yes, I was there for three years. I was the Property Book Officer in addition to being the research geneticist. And so my last job was to dispose of the equipment and the dogs when the Army closed it all down in 1976. And by the end of June, 1976, it was all dispersed. So the U.S. government has a long history of starting breeding programs, keeping them going just long enough that they begin to reap the benefit of getting themselves organized and putting in a generation or two of selection. And then budget crunches come and they cut the programs and they all disappear.


JH: One of my friends in graduate school had been working on the latest iteration of that, and that program was closed down exactly as you describe after just exactly that many generations. And he changed schools and came and worked elsewhere.


EL: The TSA program at Lackland was the same way. There was a breeding program at Front Royal, Virginia, which has been up and down, up and down, up and down as budgets come and budgets go. So yeah. I’ve seen it several times.


JH: So then you went back to school?


EL: I did. I went back to graduate school after I left active duty. Then there I finished my Ph.D. I was very fortunate to have access to 650,000 records on registered Hereford calves. And we were able to take a first look at the effect of region of the country where the calves were raised as a fixed effect, producing some of the variation observed in weaning weights. And in 1977-78 working with 650,000 records on a mainframe computer was quite a challenge. We learned a lot about magnetic tapes and we pined away for (laughter) access to rotating disks and disk packs that were maybe 20 megabytes in size as the largest one. But we learned a lot about how to process records sequentially, reading them off tape and doing things in as efficient a manner as we could.


So, yes. We published that paper a few years later. That one was really interesting. It didn’t get much attention until quite recently. And I’ve noted that the beef industry has taken notice, and they’ve now decided that including a term in their genetic prediction models for region of the country where calves were raised accounts for a bit of the variation that otherwise would be placed into the error term of their models. And so they can make a little bit better correction of the observed phenotypic variation by including that term in the model. But it took 30 years to pass before that was recognized as being something that maybe should be considered.




JH: Yeah, you’d think it would be obvious, but…


EL: Well, it’s a matter of how do you collect the data? And how do you get the animals coded in a way that it properly reflects the part of the country where they really were raised. And, you know, technology has finally enabled us to do many things that we dreamed about doing 30 years ago, but we really didn’t have the technology to do it. But the cloud, the ability of a cell phone to connect to a website and allow you to enter data when you might be miles away from a telephone line or any kind of an internet connection that your phone can find a cell phone signal. And you’ve got a connection to the world. And it’s really revolutionized what we can do.


JH: It’s an amazing time. And I think we’re just now starting to really come to terms with what the implications of that are for how we breed animals and the depth of what we can do now that we couldn’t do before.


EL: Yeah, I agree.


JH: All right. So you got your Ph.D. and then you had some other jobs doing other things before you ended up working with guide dogs.


EL: Well, that’s an interesting aspect of the time. (laughter) I was still a graduate student at Iowa State in 1977, when I had a phone call from The Seeing Eye, asking if I would be willing to do some consulting for them. And we chatted for a while and I agreed to go back to Morristown and get to know them a little bit and then give them an opportunity to get to know me. But that all came about because an advisor to our breeding program with the Army project was Dr. Don Patterson, the head of Medical Genetics at the University of Pennsylvania. Wayne Riser was also one of our advisors, one of the founders of OFA, just as an aside. So I got to know both of those, but Dr. Patterson especially was a member of the Board of Trustees of The Seeing Eye. And after he and I got to know each other through working on the Army project together, he suggested to The Seeing Eye that they consider inviting me to do some consulting for them. And so that relationship began in 1977 while I was still in graduate school. And that’s how I got started in the work with The Seeing Eye and with dog breeding outside of the Super Dog Project.


But when I finished the Ph.D. in ‘79, I needed a real job. So I accepted a position back at New Mexico State on the faculty and was in the Animal Science department as an assistant professor. My primary role was to teach a class to graduate students on research design. And I did a lot of statistical consulting in the department with graduate students who had data they needed to analyze and they needed some expertise and statistics and computer processing to get that accomplished. So I did that for four years.


I left New Mexico State and went to the University of Maryland to accept a position in their Animal Science department where I was in charge of the research program for the Wye Angus Herd. The Wye Angus Herd is a world famous herd of Angus cows that was built by Arthur Houghton on the eastern shore of Maryland beginning in 1935. Well he bought the farm in 1935, and in 1938 he brought Angus cattle to the farm. And the herd was a well known herd through the ‘60s, ‘70s and ‘80s. They never succumbed to the temptation the Angus breed as a breed succumbed to, that is selecting for just increased size and growth. They focused instead on maintaining maternal ability of the mother cows. And even now I think they’re best well known for the mothering ability and the maternal aspects of the females that are produced by the herd. But I was there for six years, and had a wonderful time helping organize the records into a database that allowed them to really begin making some genetic improvement on the records.


By then I had already begun building a database for The Seeing Eye. I started that in 1980 when I was still a consultant. The computer technology was rapidly changing in those years. The IBM PC, I think, came along in 1982, or ‘83. And the Radio Shack Model 2 was a predecessor of the IBM PC. We started The Seeing Eye’s record keeping system using that Radio Shack Model 2. And by the time we retired that system, we had several thousand records already collected on the dogs from the 1970s and the 1980s.


So anyway, I’m kind of mixing up both progression and where I worked, but six years at the University of Maryland. Then I worked as a statistical consultant in the DC area for a few years and maintained my consulting relationship with The Seeing Eye. And eventually that led to my accepting a full-time position with The Seeing Eye in the fall of ‘95. And I remained with The Seeing Eye in a full time position then until I retired from them in 2015. So a few months after trying retirement for a little while, I decided that wasn’t much fun. And I started Canine Genetic Services, which is my private consulting company. And we’re now extremely busy trying to build IWDR, or The International Working Dog Registry. So.




JH: And that’s very much what I have in here to talk about. But before we get into that, can you tell us a bit about what your job was at The Seeing Eye? What you did? What some of the things that you saw changed genetically during the time you were there?


EL: Sure. In 1980, while I was just getting started in my job at New Mexico State, The Seeing Eye asked me if I would write them a new breeding plan. And they said at the time they needed to improve hip quality because they were losing at least a third of the young dogs that came back to begin training before they could ever start training because of poor hip quality. So, these were about half German shepherds and half Labradors. The hip quality losses were greater in the shepherds than in the Labradors. But they were still losing Labradors as well, even at young ages, and the dogs were around 14 months of age when they came back to be radiographed.


So a young dog at that age, which already shows evidence of hip dysplasia, osteoarthritic changes in the hip joint, are dogs that are not good breeding candidates. If it’s not too severe, they probably can work for a number of years and do really well. But those are not dogs you want to keep for breeding. So they were really keen on improving hip quality. And at the same time, they wanted to be sure they did not lose the ability of dogs to be trained. So that was the second major goal they wished to have the breeding program focus on. So I wrote a plan in 1980, which basically is still the plan they’re following. I mean, it’s been tweaked along the way. As technology has advanced its components have been dramatically improved over the years.


But the plan included using “estimated breeding values” for hip quality as one of the components for identifying the best young dogs to keep for breeding. But of course, in 1980, there was no ability to record anything in a database, or to calculate estimated breeding values. So I had to create one. And they hired me to create the Radio Shack Model 2 driven database. And even though it was crude and rudimentary, I did the programming that calculated estimated breeding values. We adjusted records for contemporary group effects by calculating deviations from contemporary group means. Then we did the calculations on those deviated records to produce the estimated breeding values that helped us identify the best young dogs coming from families that were more likely to have improved hip quality. So we started all of that, oh, it was in place by probably sometime in 1981.




JH: So probably this is a good point for me to ask. I’m going to have more questions about exactly what you saw at The Seeing Eye but I think a lot of people probably listened to that last set of things that you said and thought, “What are estimated breeding values?” What is, what did you say? The average group mean?


EL: Oh, the contemporary group mean.


JH: Yeah. So maybe you could explain what an estimated breeding value or EBV is.


EL: So everyone understands, I think, that the genetic component that an animal inherits from its parents is fixed at conception, essentially. There are slight modifications that could happen after conception, but those are rare. So for practical purposes, consider that it’s fixed at conception. Everything that happens to that developing embryo after conception is in one way or another an environmental influence that modifies the expression of that genetic complement to become the phenotype we can measure as the embryo grows, as it’s born, as it becomes a growing individual and a young adult, and we measure phenotypes of all kinds.


So the estimated breeding value is an attempt to take all the data that we can accumulate on a particular phenotype and predict what the relative genetic value is of that animal as a replacement breeder. So it’s attempting to predict the worth of that genetic complement which we can’t see, but we know that it’s fixed at conception. And that the only view we have of it is what we can measure phenotypically on the animals as they grow and develop. So that’s what an EBV is, an estimated breeding value. It’s based on statistical methods that rely on having a large volume of well-measured data, properly coded data. And if you have that, and you’ve got a phenotype that is influenced by the genetic material that an individual inherits from its parents, then you can make genetic improvement if you get things organized correctly, and you apply the principles.


I sometimes will liken it to compound interest. It does build over generations, just as the value of an investment builds over years of compounding interest over time, and generations, and the ability to make genetic improvement builds in a similar way. But the turnover has to be measured in generation interval, not in years. I mean, you can put it in terms of years, but it’s generally better to look at it in terms of a generation interval.




JH: So it’s basically applying selection pressure, using data, statistical data. And part of what you’re saying about the improvement over years… I mean, obviously, it takes multiple generations to continue to apply pressure and to see continued improvement. But I think also what you’re talking about is that with more generations, you have more data. And so your predictions are better. Am I right about that?


EL: Those points are true. Yes, with more generations you certainly have more data, assuming you’ve put in place the procedures for collecting data in a uniform way across time. And that requires that you have, you know, proper methods available for measuring a phenotype. And that you apply those. In the beef cattle world, it’s relatively easy to think about measuring the weight of a calf, the day you decide you’re weaning all the calves from their mothers in the fall, usually in spring calving operations. And so a weaning weight is something you can visualize.


Measuring hip quality in dogs when they’re 14 months of age is common in the working dog world. That’s something you can visualize if you think about taking the radiographs when the dogs are that age, but age is an important variable in the extended view hip score.


Just as a bit of an aside because the OFA, for example, requires dogs to be 24 months of age before they will issue a definitive statement on a dog’s hip quality. Most working dog organizations can’t afford to keep large numbers of dogs around until they’re two years of age just to get an assessment of hip quality. So they do it when they can at the oldest age possible. But it’s in the neighborhood of 14 or 15 months usually. And we know that that will miss some dogs that begin to show additional signs of arthritic change, or joint conformation change, as they age from 15, say, or 14 months to 24 months, but there’s a trade off and cost to the organization that has to be reckoned with. And so that’s why they’re usually radiographed or evaluated for hip quality at such a young age. And, well, where were we? We were talking about contemporary groups and evaluating hip quality back in the early days.


So a contemporary group is simply a group of animals that have been managed in a similar way. My introduction to the concept came from the dairy industry when I was in graduate school and we learned that dairy cows managed on a dairy farm in Illinois, in the month of January are likely to produce milk in a different volume, on average, than dairy cows in Georgia, in January. Measured in the same month, but there’s a contemporary group difference that’s somewhat attributed to the difference in climate between Illinois and Georgia. And, of course, a difference attributable to the way in which the cows are managed on their separate farms.


So by defining a contemporary group to be all cows that have been managed on a farm in a given month, as an example, then you’ve got a smaller group of animals that define that contemporary group. And now we have much more sophisticated ways of adjusting for the contemporary group effect. But way back then, the technique the statisticians had developed was to calculate the contemporary group mean, then deviate or subtract that mean from the observation of every single cow that made up that contemporary group. Now, you’ve got a set of deviations within that contemporary group that range from some minus number to some plus number. And if you sum those together, you should have zero because they’ve all been adjusted for the mean. If you do the same thing for the cows raised in Georgia, you’ve now got the the mean taken out of the measure of milk production. And now the deviation of the cows in January in Georgia also sum to zero. And those deviations now represent an effect that’s been adjusted for the contemporary group mean.


We did the same thing with the hip quality measurements at The Seeing Eye beginning with those early days in the 1980s where we were adjusting for contemporary group means of the dogs evaluated in a given month within breeds. So the German shepherds evaluated in January of 1981 would have had an average hip score, and as the average began changing, you wanted to adjust for that change before you did your modeling. So you could continue to identify the dogs that were genetically superior. And so that’s the way we adjusted for the contemporary group effect.


And that, I hope, explains at least in principle, how it’s done. The linear models that are used now do it in a much more sophisticated way that simultaneously estimates the effect of that contemporary group term, in addition to all the other terms that are in the model.




JH: Yeah, so basically what you’re talking about is, you’re talking about building a statistical model, which is a term not everybody may be familiar with. But the idea is that you’re going to be able to say, what are all the things that affect the trait that you’re looking at? So if you’re interested in hip scores in dogs, what affects that? Some amount of that is going to be genetics, and that’s what you’re really interested in selecting for. But you have to also take into account the fact that, you know, well, some dogs were, you know, raised by this group of people. Some dogs were evaluated by this group of people. You know, some dogs… Just things were different this year than they were the other year. And you’re going to have to be able to put all that into your calculations.


And, actually, those of you who aren’t clear on what the term model means, in this case, you could just think of it as the calculations, right? And so you want to know all the parameters that you put into your calculations and how you decide what affects what. So you were back in the ‘80s, trying to make computers from Radio Shack (laughter) do all of this to help you improve hip scores, while maintaining trainability at The Seeing Eye?




EL: We were. Yes, we, at the same time we developed the scoring scheme for the hip quality, we also implemented a trainability score that the trainers of The Seeing Eye’s dogs put on the dogs to simply reflect, “What do you think is the overall trainability and general value of this dog as a working, seeing eye dog compared to others you’ve seen recently?” And I think it’s really important to understand that last phrase I added to that. Compared to those you’ve seen recently. So the trainability score was intended to always reflect, at a point in time, the range of values observed in dogs over the scale. We used a nine point scale. They still use a nine point scale. And the best dogs they’ve seen recently are intended to be scored nine. The worst dogs they’ve seen recently are intended to be scored one. And the others are arrayed somewhere along that scale. But what defines “best” changes over time. So we’re always using that score to reflect the best dogs at a point in time. And the genetic models take into account then the temporal effect, the time effect, by including the contemporary group in the model. 


And over time that has worked reasonably well for The Seeing Eye. Part of what’s made it work well, though, was a small, dedicated group of people were given the responsibility of getting that score recorded on pretty much every dog that came back to begin training. Even the young dogs that were determined to have a health problem were looked at by these people to get a trainability score. And granted, they didn’t have the benefit of putting those dogs in training for a month, which is when the dogs that do enter training actually received their score, at the end of the first month, roughly. But they still took a look at that dog and attempted to make an assessment of its worth as a working seeing eye dog if it didn’t have the health problem that’s just been identified and has been determined as the reason why that dog would not be put into training.


So because we had the data on pretty much every dog that came back to The Seeing Eye, we had a really nice set of data for identifying families that tended to produce the dogs that succeeded in training most often. And that worked and it worked. And it still works. And they’re still using it. So what are we now, 40 years on from when we began?




JH: That’s amazing. Yeah, and you were talking about some other issues you had with trying to see the improvement on the hips. And I think that also speaks to the importance of having good phenotypes and how hard it can be to measure a good phenotype. Right. So you were talking to me before we started recording about how you measured hip laxity and issues with hips early on, and how your measurements improved later, and what that reflected on the results of your selection pressure.


EL: Yeah. In the early days, in the early 1980s, the only technique that had been widely adopted in the canine world for assessing hip quality was the extended view score. And the OFA was the primary organization providing that score and we’ve already spoken a bit about their requirement to see radiographs on dogs that are at least two years of age before they’ll make a definitive statement.


So The Seeing Eye was fortunate in that they had a long standing relationship with the University of Pennsylvania School of Veterinary Medicine. And Dr. Darrel Biery in radiology, a board-certified radiologist on faculty, was very interested in helping The Seeing Eye improve hip quality. And so he began reading the extended view hip scores, probably in 1978 or so. And we used a nine point scale for grading back in, well, we adopted it even then. Because the Army project had a nine point scale. There was no other reason except that we already had that bit of history. And Dr. Biery had read radiographs for the Army project in the years when Dr. Riser was beginning to move into retirement, and so he and I knew each other a little bit from having crossed paths back with the Army project, but the Army scale was one to nine for everything. So that’s why The Seeing Eye scale continues to be one to nine. Once inertia sets in, it’s hard to change, I guess. But yeah, the nine point scale meant that in hip quality, the best scores were to receive a nine. Dr. Biery didn’t like to use a nine. So we had lots and lots of dogs that had sevens and eights. The scale really had too many points on it. So seven and eight meant that the dogs had nice quality hips. Six was not used very often. Five was analogous to the borderline category of the OFA, and then four, three, two and one represented worse levels of hip quality.


And as people who have studied extended view scores and hip quality measurements probably know, the extended view is a view from which it’s very difficult to measure joint laxity. And the veterinarians, the orthopedists especially, began to realize that the presence of joint laxity could be a really important factor in determining the propensity of a dog to develop early signs of osteoarthritis. Dogs with more laxity have a hip joint that doesn’t fit together so well. And the more severe the conformation changes in the hip score, or I’m sorry, in the hip phenotype, could be attributed to the presence of joint laxity.


So sometime in the, probably, 1986 or thereabouts, Gail Smith at the University of Pennsylvania, an orthopedic surgeon, became interested in trying to develop a technique for measuring joint laxity. And a lot of research was done by him in the lab, and on dogs that came to Penn routinely until he had developed a technique for using a distractor bar to directly measure the degree of joint laxity present in a set of hips that was in a unique position where the distractor bar permitted applying a bit of pressure and seeing how much the degree to which the head of the femur moves away from the acetabulum, the center of the acetabulum.


And so eventually that became known as the PennHIP Score. And The Seeing Eye agreed early on to help him evaluate the efficacy of that phenotype because young dogs measured at 14 months of age in the extended view scale, we know, are often not exhibiting their final phenotype. And so the early research with PennHIP was to evaluate, “Is a score obtained on a six month old dog different from one obtained on a 12 month or an 18 month old dog or a 24 month old dog?” And so they did all that research and those papers are in the literature and they were able to show rather definitively using some dogs at The Seeing Eye, but dogs of many breeds from lots of places, that the PennHIP score is rather highly repeatable from at least six months of age onward. And there’s some evidence that maybe it even can be evaluated in dogs as young as four months of age. But I think you’re probably generally safe to take a six month old score as pretty much a good reflection of joint laxity. And it’s as good as one obtained at 12 or 18 months of age or somewhere in between.


So The Seeing Eye began using PennHIP on all the dogs probably in 1991 or thereabouts. So the scores were being obtained, but to prove the value of the scoring scheme and to avoid introducing any change in the PennHIP score attributable to genetic selection, the scores were all sequestered by Penn. So The Seeing Eye just knew that the scores had been obtained, but the radiographs went to Penn. They read them. The radiographs and the scores were kept at Penn, and we had no access to them. The papers were published. And finally in 1995 the scores became available, it might have been 96. But it was along that time period. And it was then that we could begin calculating estimated breeding values on the PennHIP score. That turned out to be a highly heritable phenotype. Every analysis I’ve done of it, it’s about 60% heritable…




JH: Which is really good, by the way. Yeah you were about to compare it…


EL: Compared to the extended view scores that generally in both the dhepherds and the Labradors in our early analyses were more like 25 or 30% heritable. So that’s high enough that you can make genetic improvement, but you need to use a lot more technology in terms of genetic modeling to sort out which are the best young dogs from families that are going to produce high quality hips. In other words, the dog’s own phenotype is not necessarily a very good predictor of its genotype when the heritability is as low as 25, or 30%. You need to know, and this is why the OFA’s recommendation is that you look at a dog’s own phenotype. You look at the phenotypes of its siblings, the phenotypes of its parents, and the phenotypes of its grandparents. And if all of those are good or excellent, then there’s a good chance that that young dog is from a line that is going to produce and transmit the genes for high quality hips.


But there are not that many people who have… If you even go look in the OFA’s publicly accessible database, it’s hard to find families of dogs with that much data. Even now. In any breed. Yeah, there are some examples where a few dedicated breeders have been recording the hip quality scores for generations, but that’s not the norm. And unfortunately, the hip quality hasn’t changed much in the general canine population. In The Seeing Eye’s population, hip quality improved rather dramatically starting in about the third generation, and I think it took us that long to get things organized, and to get enough data in place that the models began working. And after the third generation, the hip quality began improving using the same techniques for evaluating phenotypes that we had been using all along.


And by the fifth or sixth generation, it was hard to find a dog which had a score below four. And as time passed, even those began disappearing. So then we were in the position of selecting among shades of excellent or shades of acceptable hips. Not excellent, but acceptable hips. And if all you’ve got are shades of normal, it’s really hard now to make any change because everything is phenotypically so similar, there’s not much difference to work with. And that’s when it really became important for The Seeing Eye to change the phenotype from an extended view score completely to incorporating the joint laxity measure. And the estimated breeding values just helped us a little bit more because it took into account not only the dog’s phenotype, but also the phenotype of its relatives. And by then we were using much improved statistical software for calculating the breeding values. And we were fitting models that included sex of the dog, contemporary group of the dog, and then the genetic component, and all the relationships among the the ancestors and so forth. And so it really worked.


But it turned out that there was not a high genetic correlation between the extended view score, which is similar to the OFA score, and the PennHIP score. We summarized all those results in a paper published in PLOS ONE, February of 2019, where the results of 30 years of selection are summarized to try and put it all together in one place. Where we talk about the low genetic correlation between an extended view score and the PennHIP score. And now if I were starting over with a new population of dogs on which improving hip quality was an important criterion for selection, I would devote it entirely to the PennHIP score as the method for evaluation. Maybe that helps explain a bit about it. I sort of rambled a bit.



JH: Yeah, no. I think it helps enormously. There’s been a lot of conversation… I think there’s a lot of curiosity among dog breeders about how best to improve the hip scores, or hip health, let’s say. Or I think you’re using hip quality as sort of a more generic term, right? So how to best improve hip quality in the dogs that they’re producing. And a lot of people do OFA and a lot of people do PennHIP, and there’s not a lot of people who really know why they might choose one over the other. And it’s important to know that there’s some evidence out there that PennHIP really is a better indicator, and as you say, gives more data to select on.


EL: And it’s important to understand that one of the biggest risk factors for developing osteoarthritis with age is joint laxity. And PennHIP is a score which directly measures joint laxity. So a hip that is tighter at a young age, the research has shown, is less likely to develop osteoarthritic changes as the dog ages. 


The beautiful study done by Purina with the Labrador retriever brothers. They had 20 or 25 pairs of Labrador retrievers, full sibs, males, that were followed for a lifetime in the Purina Canine Research Center out in St. Louis. And that study showed definitively that joint laxity was an important component. But it also was an important component of, as a risk factor for developing osteoarthritis. But that study also showed the huge impact that can be placed on the expression of joint laxity by weight of the dog. So the study used one dog as the control for the other in the sense that one Labrador male was chosen to be the dog that received as much food as it wished to eat in 15 minutes. And the next day it’s full brother received 75% that amount as its meal for the day. So the full brother that received the smaller ration was kept leaner.


And the study was extremely clear that the dogs which were kept leaner, developed osteoarthritic changes at a slower pace than the dogs that were allowed to eat as much as they wanted in 15 minutes. And those dogs tended to very quickly become overweight, and they stayed that way. And their lifetime was shortened. The lean dogs lived longer. They were later in life before they developed osteoarthritic changes. It’s a beautiful set of work, papers published by the group from Purina. I think that work was mostly published in the early 2000s, maybe the late 1990s.




JH: We’ll look it up and I will provide links to that paper and to the other one that you mentioned for people who want to track that down, the paper that you published in PLOS.


Yeah, so I think that’s an important takeaway is that genetics provide risk. And how we manage the dog has a lot to do with what phenotype ends up appearing. So you can have an increased risk for actually having osteoarthritis later in life, but how the dog is managed has a big effect on whether the dog actually has a problem later in life. And really, what we’re measuring with these hip quality scores earlier on is risk. And I think it’s important to recognize that it’s not 100% that the dog is going to have problems later. But it is a very important indicator.


But I also thought it was fascinating that you pointed out that you can get this information from PennHIP at about six months reliably. And you don’t have to wait until two years of age. That was something I didn’t know until you said it just now. And I think it could be huge for some breeders to recognize that they don’t have to wait that long before making decisions, at least based on hips. They may have to wait longer, you know, to get other data to make decisions going forward on their dogs but the hips don’t need to hold them up until two years. They can get some useful information out of PennHIP much earlier than that




EL: Oh, yeah, yeah. That’s a really important advantage of the PennHIP score.


JH: Yeah, that’s huge. Um, so were there other things at The Seeing Eye that you, so you were selecting on trainability? And was that also amenable to selection?


EL: Yes. That’s, that’s really hard to assess in a quantitative way, except to show that over time, the genetic components steadily increased in value compared to the base, which was set back in the early 1980s when we began recording that score. And so the trainability score continued to increase. But the the one comment that was the most telling to me was when the trainers who were there when I began consulting in the late ‘70s, and up through the 1980s, would tell me how much better the dogs were in the late 1990s than the dogs they had to train when they were young in their careers in the 1980s. And then, you know, they would retire and others would come along. And over time, they all could see that the dogs that were coming along more recently, were just better dogs in lots of ways.


And I… trainability is a really complex phenotype. We never tried to break it down into all its many parts at The Seeing Eye. And I think the reason that it worked was because we had that small group of really experienced trainers who were always giving the score. But that’s not particularly a technique that is transferable to like a breed of dogs that are scattered across a large geographic area and cannot all be evaluated by a small group of highly skilled evaluators.


So you need other techniques, and I know Dr. Serpell at the University of Pennsylvania has done a lot of work to develop the C-BARQ as a tool for trying to characterize a dog’s behavior. There’s a lot of good science behind his C-BARQ. In developing the C-BARQ, he also developed something called the Behavior Checklist (BCL), which The Seeing Eye started using to help him evaluate C-BARQ. I don’t know, don’t remember now, what timeframe that all was happening in the late 1990s? I think? Well, no, it was later than that. His early papers on C-BARQ were published in 2002, I think. So it was somewhere in maybe the early 2000s.


But anyway, the Behavior Checklist is another tool that we’re now attempting to validate and to get described in the literature to show that it’s a method for providing a uniform scoring scheme for describing aspects of a dog’s behavior in ways that can be taught to a large number of people. I think that’s a really key difference between what we hope to accomplish with the Behavior Checklist and what The Seeing Eye has been able to implement with the trainability score. So the Behavior Checklist is a tool for scoring aspects of behavior by responding to, or by providing scores to, each of 52 components which describe some aspect of behavior. And then on a scale from Absent to Severe, on a five point scale, the observer provides a response. And we know that among those 52 questions, they basically boil down to somewhere between eight and 11, or 12, different components of behavior. And those components are, to some degree or another, genetically related. Some much more so than others.


We can begin to sort all that out when we have a large number of dogs, evaluated by skilled observers who have learned what these questions on the Behavior Checklist actually mean, and who have learned what the scores Absent to Severe reflect. And so we’ve actually built training videos and we’ve done training classes, and when COVID is over we’ll probably be doing more of those. In the meantime, we may be offering online training for people to learn how to use the Behavior Checklist as an alternative to the C-BARQ for evaluating aspects of behavior.


And let me quickly say that there’s nothing wrong with C-BARQ. But then The Seeing Eye had literally thousands of dogs evaluated by puppy raisers using the C-BARQ. Guiding Eyes also had many thousands of dogs evaluated. A master’s thesis was done several years ago now looking at variation in the C-BARQ scores within these two highly selected populations of dogs being bred for work as guides. And over time, the variation the C-BARQ could pick up was steadily diminishing in these two populations, I think because they were responding to genetic selection. So with response to genetic selection, one expects the genetic merit to improve but you also expect genetic variation to disappear. And eventually, unless you are working with a huge population, the animals will become genetically identical, maybe 100 generations of selection. But, you know, theoretically that’s the end point.


Well, the Behavior Checklist provides us with a way of evaluating aspects of behavior that restores some of the variation that’s not captured by the C-BARQ. Though I don’t want to imply at all that there’s anything wrong with the C-BARQ. But it was developed to focus on dogs owned by the general public. It was not developed as a tool for these highly selected populations of working dogs. And I think that’s a statement Dr. Serpell would agree with. The Behavior Checklist, on the other hand, I think has utility in these working dog populations that is different from the utility of the C-BARQ in evaluating aspects of behavior. Okay?




JH: Yeah. So that’s a massive amount of work that you did over your 40 years working with The Seeing Eye. And I’m now realizing that we have talked for more than an hour, and that it might be a good time to pause, and perhaps split this episode into two episodes. So we will be back next week with more from Dr. Eldin Leighton on the IWDR. And what he’s working on now, which is hopefully something that’s also going to be super relevant to a lot of the people listening to this podcast. But for this round, thank you so much, Eldin. This was really insightful. I learned a lot, and I hope other people did too. So thank you so much for taking so much time to explain this stuff.


EL: You’re welcome. And thank you for the invitation. It’s been fun.


JH: Absolutely. So more soon.

Thanks so much for listening. The Functional Breeding Podcast is a product of the Functional Dog Collaborative and was produced by Sarah Espinosa Socal. Come join us at the Functional Breeding Facebook group to talk about this episode or about responsible breeding practices in general. To learn more about the Functional Dog Collaborative, check out www.functionalbreeding.org. Enjoy your dogs.

Copyright © 2021 Functional Dog Collaborative. All rights reserved.