When A Selective Breeder is not a Geneticist

Robert Bakewell by John Boultbee (Wikipedia image)

It is a little known fact that most guppy breeders breed their finned animals according to precepts laid down by Robert Bakewell (1725–1795), an eighteenth century farmer. He is the father of selective breeding. Selective or artificial breeding is intentional breeding for a trait or set of traits, like large bodies and long fins in guppies. In a sense artificial selection has going on since plants and animals were first domesticated thousands of years ago. The guppy breeder practices selective breeding when he or she culls guppies from a breeding colony or selects a pair of guppies to produce the next generation of a strain.

What makes Bakewell special is that he laid down a canon for selectively breeding plants, cattle, sheep, and horses. You owe that hamburger or steak you are munching on to Bakewell who transformed cattle from plowing machines to meat factories. Father of the modern hamburger? Dean of steaks?

I love the picture of Bakewell shown above. He is the quintessential English gentleman, at home on a horse he designed with the other products of his designs grazing in the distance.

Was Robert Bakewell a geneticist? Some people might argue that he is. But they are wrong. Genetics was first employed in agricultural improvement by the person who enjoys much more name recognition than Bakewell. His name of course was Gregor Mendel, the acknowledged father of genetics.

Both Robert Bakewell and Gregor Mendel were involved in agronomy, which can be defined as the science and technology of producing and using plants for food, fuel, feed, fiber, and reclamation. Mendel did his studies of pea plants to discover the principles that would allow the farmers in his local area to increase the health and yield of their crops. But while Bakewell may be regarded as a superb synthesizer of all the previous knowledge about selectively breeding plants and animals, Mendel created entirely new knowledge. He uncovered the underlying principles of breeding.

Bakewell was not a geneticist, while Mendel was...because Mendel discovered genes. You cannot be a geneticist without studying genes. And if the study of genes is not the focus of your breeding efforts, than you are following in the footsteps of Bakewell, not Mendel.

Which makes Alan Bias a very peculiar animal indeed. I think I would characterize him as a gentleman breeder of guppies in the tradition of Bakewell. I have never met Alan, but when I do (if I do!), I would almost expect to smell the faint odor of cow dung on him, as if he had just walked in from the pasture with his fashionable wellies on.

In our correspondence in the past year he has often referred to himself as a farmer, despite the fact he works in high tech and has no cow to his name, just a room full of guppies. There are cattle and sheep in his background and probably in his blood. That is what makes his perspective on guppies so interesting to me. He thinks about guppies as Bakewell might have thought about his herd of cattle or the horse under him. That is fully evident in the two papers he has just written and published on Guppy Designer, and in other places. They are called "Breeding Strategies and Genetic Manipulation in Guppies" and "Genetic Strategies and Genetic Consequences in Guppies."

But is Bias a geneticist? I know that he is, because I know he studies genetics. He conducts crossing experiments to discover the genes involved in the colors and swords of his beloved swordtails. But everything is seen through the eyes of a farmer examining his herd. He sees the herd not the animal. Or he sees the animal as a member of the herd. And it is the herd that he sees himself...I am looking for a word here..."curating?" I know I am badly mixing my metaphors here, having gone from the barnyard to the museum in mid-sentence. But I blame Bias for that. He defies classification.

What makes Bias special as a friend and fellow guppy breeder is that he is breeder of show quality guppies who does not subscribe to any specific show standards. How can you be a breeder of show quality guppies without membership and bragging rights in a show guppy organization? Read his paper about selecting breeders and you will see excellent Bakewellian advice about avoiding sway-backed females and enhancing animal vigor. And like the farmer intent upon preserving the unique characteristics of a breed of cattle, he indeed thinks of guppies in terms of breeds. Which makes his study of guppy genes very peculiar indeed, at least from my perspective. I do not have strains in my fish room. I have phenotypes. And they are temporary, because I am always about two generations away from crossing them with another phenotype that would be classified as a different strain by show guppy breeders. Alan stubbornly sticks to the idea of "guppy breeds" and appears to think he owes it to future generations of guppy breeders to preserve the swordtail strains. He thinks about the whole animal, not just the colors on the skin or the flow of the fins. He thinks like a farmer.

Alan uses genetics in his fish room like a farmer a pitchfork in the barn, It is a tool for him. Not the whole story. That is hard for me to understand because guppies are for me a mystery novel where the book itself disappears (the paper and the ink) as I wander around in the imaginary world created between the covers. For Alan, I suspect he still sees and loves the individual guppies. Not as a wet pet, but as a farmer sees a cow or sheep.

Perhaps Alan Bias is something new entirely and that is why I have trouble classifying him. While the show guppy hobby is in a long term decline because of the cost of putting on shows and traveling to them, perhaps Alan is the first of a new breed: somebody deeply committed to individual breeds of guppies, farming them, like a gentleman farmer.

I highly recommend Alan's papers to guppy breeders who have the same independence and restless mind as his. Heck, I recommend them to all guppy breeders.

There is an interesting application of his approach to guppy breeding in a critique he has just written. It is on his swordtail blog as "Science as a Tool and Benefit to Guppy Breeders." Bias analyzes an article written by Jim Alderson, who is a member of the IFGA (International Fancy Guppy Association), one of two show guppy organizations in North America (the other is IGEES - International Guppy Exhibition and Education Society).

In a paper called "Follow the Best Genetics," Alderson encourages out-breeding as a tool for producing hybrid vigor in IFGA strains and for producing hybrid show entries. But Bias counters with the observation that out-crossing, as a tool for curing inbreeding depression risks outbreeding depression. You could potentially lose the breed, in Alan's judgement. Alan suggests you use science to ensure that you do not subject your guppy strain to inbreeding depression in the first place.

In Alderson's seminars I have heard him say that the IFGA guppy is highly homozygous across its loci, suggesting that the various classes of guppies are highly inbred. But how does he know this? Is this just another example of what is called "anecdotal evidence" precisely because it is not documented and has no scientific corroboration? As it turns out there is good evidence for IFGA strains lack of genetic diversity. A scientific study was made of guppies sampled from IFGA stock.

The complete title of the paper is "Are Designer Guppies Inbred? Microsatellite Variation in Five Strains of Ornamental Guppies, Poecilia reticulata, Used for Behavioral Research" (Zebrafish, Volume 5, Number 1, pp. 39-48). The authors are Bleakley, Eklund, and Brodie. Find it here: http://faculty.virginia.edu/brodie/edb3pdfs/Zebrafish2007.pdf.

Trust a scientific paper to ask a question whose answer is obvious to everyone.

Or is it?

Have modern guppies, despite their visual diversity, lost the original genetic diversity that once made them a hardy, fertile and long-lived fish? Is that diversity permanently lost?

The reason the scientists who wrote this paper want to know the answer to this question is for the potential use of the designer guppy as a model organism for behavioral research:

Inbred lines are commonly used to map quantitative trait loci (QTL) and candidate genes underlying behavior,investigate the heritability of all types of traits,and provide tester strains against which other strains, inbred or outcrossed, may be compared during behavioral research. (p. 39)

In other words, highly inbred designer guppies are valuable to the genetic researcher because they reduce the amount of genetic variability that has to be taken into account when investigating an isolated trait. That is the argument for using an inbred laboratory strain rather than a wild caught strain. Indeed laboratory strains that begin as fresh imports from the wild, lose their original genetic diversity after ten generations of inbreeding.
You would think that using wild guppies would be more useful to the researcher attempting to isolate and study a single trait. But wild guppies are actually more genetically diverse than domestic strains of guppies.
That’s where hobbyists can be handmaidens to scientists. Most hobbyists do not maintain large populations of their favorite strains. They tend to practice selective breeding where the most desirable male and female are selected to be parents of the next generation. This “bottleneck” reduces genetic diversity by reducing the number of gene variants (alleles) passed on to the next generation.

The authors choose as their laboratory specimens guppies they obtain from the website “Angels  Plus,”in Olean NY.  The owner, S. Rybicki, is well connected to the IFGA and his guppies are largely derived from show winning IFGA breeders. In fact four of  the strains employed by the authors have IFGA strain names: Half-Black Green, Half-Black Yellow, Blue and Snakeskin. The fifth strain, “Red Cobra” is not a recognized IFGA strain. The term “cobra” is used by Asians to describe the snakeskin phenotype, so it is possible it is an Asian import.

The IFGA guppy standards for its color classes is analogous to the American Kennel Club standards for dog breeds. It sets strict standard for body shape, fin shape and length and color depth and coverage. Like most conformance judging standards the IFGA standard has led to highly homozygous strains, including those selected by the scientists. The IFGA standard restricts its guppies to defined color patterns.

Indeed Japanese guppy breeders have long used IFGA guppies in their breeding experiments because they are so highly homozygous, at least in terms of color and pattern. While the judging standards in Europe and Asia have tended to encourage outcrossing, with more flexible and less defined color classes, the senior judges and elder mentors in the IFGA have actively campaigned against crossing the IFGA color class strains, and especially outcrossing to foreign strains. As a matter of fact, in the same issue of the IFGA Bulletin as the Alderson article is published, Alderson warns against crossing IFGA strains to import guppies.

I have been curious for years if the IFGA standard has led to highly inbred guppies. What has caused me to hesitate about coming to that conclusion is the fact that IFGA strains are maintained by hundreds of breeders who freely exchange guppies at shows and privately. I have viewed the organization as supporting a de facto large population of guppy strains, certainly under artificial selection pressure to conform to certain visible phenotypes, but otherwise genetically diverse.

The scientific study answers this question of the role of the IFGA standard in American guppy heterozygosity. The scientists make the following observations about the IFGA strains they obtained.

Most guppy body colors other than the wild type are hemizygous or homozygous recessive traits. F1 and F2 progenies were assessed for coloration upon adulthood and found to breed true in all cases, including the Red-Cobra, which was assessed after the other four strains. Strains that breed true for such body coloration are likely to exhibit reduced genetic diversity for at least body color and fin morphology, as those traits have been under strong artificial directional selection. (p. 40)

At this point in their investigation, the authors encountered a phenomena well-recognized by hobbyists who first attempt to raise so-called designer guppies. The designer guppies are fragile, they die off. The scientists started with eight males and females of each strain, but all of the IFGA strains with the exception of the snakeskin strain died back to a single founding male and female. They then bred the strains for 6 to 10 generations. As a comparison, they imported a wild strain of guppies from Trinidad. After between 6 to 10 generations they sampled the DNA from each of the strains. Let’s see what their DNA tests found.

Although no strain of designer guppies was completely homozygous for all the loci sampled, they were more inbred than a wild strain of fish. We found broad concurrence between several measures indicative of inbreeding: allelic diversity, levels of observed versus expected heterozygosity, and inbreeding coefficients. Four of the designer strains experienced severe bottlenecks within the lab and overall reduced effective population sizes; however, the strains retained some genetic diversity across all five of the sampled loci. Although single females became founders in three of the designer strains, they were almost certainly multiply mated prior to the bottleneck and are likely to have stored sperm, allowing more variation to pass into the next generation than in traditional isofemale lines. Individuals that died of natural causes were included in the sampling, potentially increasing measures of observed heterozygosity by including purged deleterious alleles that may have contributed to their deaths. However, the alleles sampled are expected to be neutral and thus should not be impacted by selection operating through inbreeding depression. (pp. 43-44)

The authors take into account the use of linebreeding by IFGA breeders. In linebreeding the breeder maintains at least three different lines of the strain, crossing them occasionally to try to combat inbreeding depression and loss of diversity. It might be argued that the lab’s initial problems in keeping the IFGA strains alive led to more inbred guppies then found in a leading IFGA breeder’s tanks. This is how the scientists respond to this argument:

The small number of founders combined with the bottlenecks experienced within the lab may have reduced genetic diversity compared to the original strains maintained by the breeder; however, the Snakeskin and Red Cobra strains did not experience bottlenecks while in the lab but exhibited similar levels of inbreeding to the other strains. (p. 44)

Actually I think the scientists inability to keep their original breeding colonies alive unwittingly simulated what actually happens in the hobby. Most hobbyist start a strain with a trio (male and two females). It is called the “founder effect” but it acts just like a bottleneck. Furthermore, the practice of selecting a breeder male and breeder female, or a small number of breeders, amounts to the same thing as forcing guppy diversity through a bottleneck.

An interesting finding is that the IFGA guppies do not appear to have come from a single point of origin in the wild habitats of the guppy (Trinidad, Venezuela and Guyana).

The designer guppies carried alleles not found in the wild fish, with all identified alleles falling within reported size variation for alleles for all markers, suggesting that the designer strains do reflect ancestors from multiple geographical locations, violating the assumption of shared geographical origin. However, this violation potentially results in underestimating inbreeding in the designer strains compared to the wild-type reference population. The measured magnitude of f in the designer strains compared to the wild-type reference population combined with measures of observed heterozygosity (H0) and expected heterozygosity (He) and significantly greater allelic diversity in the wild-type fish compared to designer strains therefore provide strong evidence that the designer strains are inbred. (p. 45)

Another interesting point raised by the authors is how different the IFGA strains are from each other.

In general, designer strains were more different from each other than any designer strain was from the wild population. Allelic diversity within a locus for many designer strains was near zero, while the wild fish were more variable at all loci. (p. 45)

This scientific study finally confirmed something I had puzzled over for a long time. Are IFGA guppies highly inbred? The answer is an emphatic yes.

But then there is the second and more important question. How useful are these guppies for the geneticist, including a hobbyist geneticist like me interested in dissecting the color genetics of the guppy? The answer is: VERY! IFGA guppies are a wonderful tool for exposing minute pattern and color differences that would be lost in a cross with a more genetically diverse strain from the wild.

Inbred lines increase the power of such studies. For example, inbred lines can be used to finely partition the effects of minute genetic changes, such as single nucleotide polymorphisms on complex phenotypes in Drosophila.  While the designer guppy strains are not clonal, as the Drosophila strains above are, they nevertheless offer the opportunity to better control genetic influences on phenotype to enhance our understanding of the impact of genetics and all types of environmental influences on the expression and evolution of traits that are ecologically important. (p. 45)

Well then. Bring on the clones.

So I find myself agreeing with Alderson that out-crossing is a useful method for ensuring genetic diversity within a strain, and agreeing with Alan that it carries with it the danger of out-breeding depression. The best approach is to design a judging standard that does not back you into the corner in the first place. And that requires a firm grasp of modern scientific breeding principles.

It is in the light of this fact that I find (as Alan does) Alderson's denunciation of guppy genetics in the final paragraph of his paper peculiar. Alderson writes:

Guppies are a tapestry that we can learn to paint with a few genetic principles, trial and error,and records. Those who claim to study guppy genetics do so by making the crosses and dissecting the results, not by predicting the outcome in advance, claiming to have discovered the inheritance of a new trait, until the next cross is made that does not follow the rules. Breeding guppies may someday be a science, but I will be pushing up daisies by then.

The comment reveals that Mr. Alderson has little apparent knowledge of the methodology of science or he chooses to ignore it. The classical methodology for geneticists is something called "forward genetics" where phenotypes are identified and then crossed, then analyzed in terms of frequency of resulting phenotypes to arrive at the theory of the gene or genes producing a particular phenotype. Mendel in his garden did not come up with his laws before conducting his research. He conducted his research first and then analyzed them statistically to come up with his novel theories. He then tested the theory by evaluating its ability to predict the outcome of future crosses. This is exactly how the Singapore scientist Dr. Violet Phang studies guppies in such papers as "Genetic Basis of the Variegated Tail Pattern in the Guppy, Poecilia reticulata" (Zoological Science 16(3):431-437, 1999). You study the trait by doing crosses and analyzing the results. Then you formulate the theory that predicts how the trait is inherited.

I have to conclude that Mr. Alderson has not chosen to study the science to support his view that little is known about guppy genetics. The developmental biology of color cell theories I study actually come from studies in other fish, particularly the zebrafish. (My Guppy Extracts document contains hundreds of citations of studies.) For example, I discovered in zebrafish research that the absence of yellow color cells (xanthophores) means stripe formation fails. I wondered if this was true in guppies as well. In the case of snakeskins, when you incorporate mutations that affect the xanthophores, the snakeskin pattern fails to form. So the prediction becomes that a homozygous European Blau snakeskin will fail to develop the snakeskin pattern. Turns out this is one cross I have not tried yet, but it is in progress. Most of my theories are not novel at all. They have been discovered by scientists in other model organisms.

In fact the paper on the Mc1r gene (discussed in a previous blog) shows that the guppy shares much of its color genetics with other animals, something called "homology" in the scientific literature or "evolutionarily conserved." So the study of guppy genetics includes researching the color genetics of other fish, indeed of other vertebrates. There has been a huge amount that we have learned about the developmental biology of color cells in the past twenty years, particularly in the last ten years. Perhaps Mr. Alderson has no interest in this subject, which is not surprising since he claims he can reach his results with a few simple rules of thumbs which he calls "genetic principles."But Alderson cannot have his cake and eat it too. Speculating that a trait may be due to an autosomal or sex-linked gene is like speculating that birds can fly because they have wings. The scientist goes beyond that and identifies the specific structures and aerodynamics that enable flight. A genetic principle is arrived at by recording observations about phenotypes and interpreting the action of genes, otherwise speculation about trait inheritance belongs to a selective breeding canon. His "Follow the Best Genetics" paper owes much of its language to Mendel and classical geneticists but on close examination he is really talking about selective breeding. Why he should use the language of genetics to describe his selective breeding methods and then abjure genetics makes no sense at all.

You might conclude that I am against selective breeding. But I am a selective breeder. Modern geneticists practice selective breeding. It is not about being for or against selective breeding or genetics. Both have a role to play in breeding guppies. And that is why I think Alan Bias is the future of the hobby. He shows how to breed pedigreed guppies using both selective breeding and genetics and he has the good sense and taste of a gentleman guppy farmer. Read his papers.