Where Did Rollers Come From
By Dick Cryberg
Published on Spinner Magazine, Vol 1 May/June 2015
Many people develop an interest in discovering who their ancestors
were and where they lived. This can become quite a passion. A passion
that has never caught my fancy I guess. Although, it would be interesting
to know if I had some cousin x times removed that was a horse thief.
We often ask these same types of questions about our pigeons. How
were all the different breeds developed? Specifically, the big
question is what were the ancestors of Birmingham Rollers. This
ancestor question is of interest, but I do not think anyone believes
it will tell us how to breed a winner. Just like my horse thief
ancestor does not make me a horse thief.
Until quite recently about all we had to go on in determining where Rollers came from is spoken and written history that really only goes back accurately maybe 150 years. Before that it was pretty much pure guess work. But, with the modern developments in learning what DNA sequences can tell us we can now put some real science behind such a question. Mike Shapiro and his team of scientists have done exactly that1,2. I am going to try and explain in English part of the kind of data they needed to perform their work. But first, Figure 1 shows the results they discovered. This chart is a sort of development tree going back in time. At the far left is the wild C livia. As you move to the right you see the lines split and eventually end in a breed name for some variety of pigeon on the right. If you look at the way the breeds are arranged many breeds are very close to other breeds that you would logically expect to be closely related. There are also some surprises. For instance Iranian Tumblers are quite far from the other acrobatic breeds.
Figure 1. Family Tree for Domestic Pigeons
In order to construct a family tree like this from DNA data what you need to do is find some elements in the DNA that will have changed somewhat, but not a lot, over the time frame of interest. The time frame in the case of pigeons is more or less 5000 years. You have probably read about work in humans using mitochondrial DNA. Mitochondrial DNA is genetic material carried exclusively in the cell organelles called mitochondria. This DNA is separate from the DNA that makes up chromosomes. It is quite stable stuff as most changes tend to be lethals and eliminated. So, change is very slow. In constructing a family tree for humans we know that humans left Africa at least 50,000 or more years ago. So, you need to look at DNA that is pretty stable, but not perfectly stable over that time period. Using such data in humans allows us to trace back the pathways that lead to modern Euopeans and Asians and Pacific Islanders for instance. We can get a pretty good idea of the geographical routes those humans took to get to their present locations by examination of the mitochondrial data. But, mitochondria are too stable to look at relationships among pigeon breeds, so Shaprio and crew needed some other type of DNA probe. They picked a structure that is called a microsatellite for their first attempts.
Microsatellites are simply DNA repeat structures. As you probably know DNA is composed of structures called bases that carry the DNA information. There are four bases A, G, C and T. It is not important to know exactly what A, G, C and T stand for. They are simply the first letter of the chemical names of the four bases. DNA consists of long streches of these four bases arranged into the chromosomes. You could describe a chromosome perfectly simply by listing these bases in the order they occur on each chromosome. For instance a sample hunk of a chromosome might read ATCCGTTAGCTA. A typical chromosome is many millions of bases long. Only a small amout of any chromosome is actually a gene. Most of the chromosome, more or less 99%, is made of what are called non coding bases. Non coding simply means this part of the chromosome does not get transcribed into any useful protein. Microsatellites are simply short lengths of such non coding bases where there is a regular repeated sequence of bases. For instance CACACACACACACACACACA. Or AGTAGTAGTAGTAGTAGTAGTAGTAGTAGT. In each of these cases the repeat length is ten but in the first case only two bases are repeated while in the second three bases are repeated. Definitions are somewhat arbitrary but in general the number of repeats in a microsatellite ranges from 10 to 100. The number of bases per repeat unit can range from 1 to 6.
Microsatellites are ideal to study how breeds evolved from each other over a 5000 year time period. What makes them useful is the number of repeats can change occasionally, but not often. So, lets just take a simple example of one microsatellite and see how it can be useful. Take a bunch of breeds of pigeons and determine the microsatellite length in each breed. Maybe our data shows we have three groups of pigeons. One group has this microsatellite 15 repeats long, the second group is 16 repeats long and the third group is 17 repeats long. It would be reasonable to think that for each of these three groups an individual breed within the group is likely closely related to the other breeds within the group. And, at the same time that breed would be much more distantly related to breeds within the other two groups. Well, that is sure a start. But, when microsatellites mutate they can either grow longer or shorter with pretty much equal probability. So, one microsatellite is not enough to draw solid conclusions.
What Shapiro did was look at 32 microsatellites in each of the breeds of pigeons he examined. He then took this data and subjected it to a particular kind of statistical analysis. This statistical analysis generated the first pass relationships between 70 different breeds of pigeons and is the data reported in paper 1. He was able to establish fairly solid relationships between 40 breeds and grouped them more or less into five families which shared common ancestors after divergence from the wild rock dove. The branch of the tree that leads to the Birmham Roller traces back to a split with a common ancestor where the twig that split off formed the Budapest Short Faced Tumbler. At a later date the line split off the Tippler and then the West of England Tumbler. Some time later the Birmingham Roller split off and after that split the Portuguese Tumbler and finally the Parlor Roller split off into distinct breeds. This family tree says that the Birmingham Rollers closest relatives are the West of England Tumbler and Portuguese Tumbler.
In a later paper2 these relationships were further refined. In this second paper other DNA probes were used to refine the microsatellite data using 1.48 million DNA loci followed by a similar statistical analysis of the data and led to modestly different conclusions. In this refined approach the Birmingham Roller split from a common ancestor that it shared with the Oriental Roller. Later the English Long Faced Tumbler and Parlor Roller split from a common ancestor they shared with the Birmingham Roller. Only 40 breeds were involved in this second study so elimination of the West of England Tumbler and Portuguese Tumbler from the tree is not meaningful as they were not included.
One of the very interesting findings to me was that Iranian Tumblers are not at all closely related to all the other air performing breeds. The Iranian Tumbler is most closely related to the Shakhsharli and Lahore. Both the Iranian Tumbler and Shakhsharli are only occasional single flippers. It would appear based on this data that a tumble mutant may have happened at least twice independantly to lead to this division. Or, an alternative explanation is that breed crosses were done to move the performance trait into these Iranian breeds followed by enough back crosses to pure Iranian birds to destroy genetic evidence of the cross. The only real way to tell which happened would be to learn the exact DNA nature of the performance mutants and see if Rollers and Iranian birds share exactly the same mutation in one of their performance genes. If they did this would show breed crossing happened.
What does all this tell us about the genetics behind rolling? The implication is there was a long time ago some pigeon that was an air performer. This original bird likely did nothing much besides single flips. With breeding time other mutants happened that impacted air performance. These mutants resulted in a variety of different types of performance ranging from single frequent flips to coop tumblers to birds that rolled horizontally, etc. The whole variety of performances we see in today’s breeds. Very often mutants that impact some particular aspect of a pigeon act in far more than simply an additive fashion. To take a fairly well understood example consider muffed feet. Muffed feet are simply a combination of the mutant that causes grouse legs plus the mutant that causes slipper. Neither of those mutants by themselves produce spectacular leg feathering. A grouse legged bird might have toe feathers an inch long and slipper about the same. Yet, the combination produces feathers as long as six inches on the toes. Shapiro has also shown that all crested breeds contain as one essential element the peak crest mutant. Yet combinations of peak crest plus other mutants produce the head feathering found on a Jacobin. Likely the same is true of air performance. That is combinations of two mutants that each by themselves may not produce a great deal of performance can result in birds much better than simply additive effects.
We know from many peoples experience that if you cross a Birmingham Roller to a non performing breed it takes many back crosses to recover a half decent roll. A modern Birmingham Roller likely consists of the original tumble mutant plus several mutants that have happened since then that all contribute to todays outstanding performance. A reasonable guess, based on the difficulty of recovering roll after an outcross, would be someplace around three or six mutants in addition to that original tumble mutant. Add one or two more and you have a Parlor Roller.
References:
1. Sydney A. Stringham, Elisabeth E. Mulroy, Jinchuan Xing, David Record, Michael W. Guernsey, Jaclyn T. Aldenhoven, Edward J. Osborne, and Michael D. Shapiro; Current Biology 22, 302–308, February 21, 2012
2. Michael D. Shpiro, Zev Kronenberg, Cai Li, Eric T. Domyan, Hailin Pan, Michael Campbell, Hao Tan, Chad D. Huff, Haofu Hu, Anna I. Vickrey, Samdra C. A. Nielsen, Sydney A. Stingham, Hao Hu, Eske Willerslev, M, Thomas P. Gilbert, Mark Yandell, Guojie Zhang, Jun Wang; Science 1 March 2013: Vol. 339 no. 6123 pp. 1063-1067.