DNA USA
Page 12
Clan Donald genealogists connect Somerled to Colla Uais through a direct patrilineal line with twelve, or thirteen in some sources, named ancestors. There is an obvious discrepancy here, as the average generation time would have been almost sixty years each to stretch from AD 330 to AD 1100, the approximate year of Somerled’s birth. Although some of them doubtless took advantage of their elevated status to continue having children into old age, the first son would surely have been born well before his father reached sixty. While this is a fine Gaelic pedigree, it must be mistaken, because Somerled’s Y chromosome has all the hallmarks of a Norse Viking. The clan of Sigurd is decidedly Norse. Chromosomes from his clan are very rare in Britain in those regions not settled by Vikings, such as inland Wales and Ireland. The only places in Britain where this Y-chromosome clan is found in abundance are Orkney and Shetland with their well-known history of Norse settlement. Indeed, these two archipelagoes belonged to the king of Norway until 1468, when they were annexed by Scotland to make up for a dowry that was never paid. Y chromosomes in the clan of Sigurd are very frequent in Scandinavia, especially Norway, and also abundant in Iceland, which was settled from Norway from the tenth century. Furthermore one particular marker is able to distinguish Sigurd chromosomes from Norway from those with an origin in Denmark, whose Viking fleets raided the east coasts of England and Scotland. The Somerled chromosome belongs firmly to the Norwegian type. Consequently there is very little chance of a Sigurd chromosome, like Somerled’s, being found among Irish Gaels in the fourth century, the time of Colla Uais.
How can these two different conclusions, the historical and the genetic, possibly be reconciled? The Clan Donald Society explanation is that, at some point in the line of Irish kings that make up the well-researched Gaelic pedigree from Colla Uais to Somerled, there was what boils down to a nonpaternity event. At least one man was not the biological father of the son recorded in the traditional genealogy. Mark Macdonald has suggested a colorful explanation that points a finger to the link between Fergus and his son Godfrey, born around AD 830. This was a turbulent time in northern Europe, and all countries bordered by the sea lived in fear of a Viking raid. The Vikings took full advantage of their reputation and developed a very successful invasion strategy with all the features of a protection racket.
The best known of these adventures was planned and executed by Hrolf, better known as Rollo, from Ålesund on the west coast of Norway. Camping out at the mouth of the river Seine in northern France, he blockaded Paris, which lies a hundred miles upstream, and sent regular raiding parties to wreak havoc in the capital. After a while the French king, forever afterward known as Charles the Simple, sued for peace and Rollo struck a deal: We will leave you alone, indeed we will promise to protect you from other Viking raids and all your other enemies into the bargain. In exchange, you will give us Normandy. Charles agreed, and, in 911, Rollo, true to his word, cleared Normandy of the king’s enemies and became the first duke. One hundred fifty years later his descendant William, Duke of Normandy, led the successful invasion of Britain in 1066.
According to Mark Macdonald’s ingenious theory to accommodate Somerled’s Norse Y chromosome with his traditional Gaelic genealogy, Rollo wasn’t the only one in the protection business. This time the Norse racketeer was Gutfrith, who tried his luck with Fergus, chief of the MacUais clan from Oriel, around the shores of Loch Foyle in the north of Ireland, who were having a hard time resisting the expansion of their neighbors to the west, the Ui Neill. This time the price for protection was marriage to Fergus’s daughter and a change of name to Godfrey mac Fergus to make his accession to the chiefship of the clan appear legitimate. Thereafter the chiefs carried Godfrey’s Viking Y chromosome all the way down to Somerled himself.
This is history massaged for political purposes, as history so often was. It suited Godfrey to side with the Gaels, and he used his credentials to gain lands in Scotland by helping Kenneth mac Alpin finally to defeat the Picts and become Scotland’s first king in AD 843. Godfrey died ten years later, in 853, by which time he had added the title Righ Innse Gall, “king of the foreign lands,” to taoiseach of Oriel. The Gaelic ancestry concocted by Godfrey was used to good effect by Somerled and his descendants, who make up the Clan Donald. Somerled himself gained his reputation for clearing the Vikings out of the west coast of Scotland and the Hebrides, so a solid Gaelic ancestry was a distinct advantage, even if he knew it to be false. Though there were hints of a Norse association in the clan, and Somerled married a daughter of the king of Norway, for the intervening 850 years his Gaelic pedigree was never really in doubt. Only in the early years of the twenty-first century did genetics finally uncover the truth. Although as yet unsupported by historical evidence, Mark Macdonald’s theory is well worth serious attention and certainly provides future Clan Donald historians with a target for their researches. If he is right, and Somerled’s Norse Y chromosome entered the clan genealogy with Godfrey mac Fergus in the ninth century, then how appropriate it is that the current chief of the clan, Lord Macdonald, one of the volunteers in my original DNA study and who does carry the Norse Y chromosome, is also Godfrey. All this from a withered fragment of DNA that travels blindly from father to son, oblivious to the bloodshed and betrayals that smooth its path from one generation to the next.
Because it is the largest family study so far, the Clan Donald DNA project has been able to examine the internal intricacies of the clan structure. As such it is a superb example of the potential for genetics to place individuals on particular branches defined by DNA, and to examine in greater detail how this compares with what is already known from the careful examination of the genealogical records that has been done, first in parallel and then in combination.
I have seen close up the way in which genetics, and geneticists, sometimes proclaim that their science makes all that went before both unnecessary and irrelevant. I had the good fortune to experience at firsthand the first forays of modern genetics into medicine and, through ancient DNA, into archaeology and both followed rather similar patterns. It must be disconcerting to have spent a lifetime in an area, to have researched a particular subject in great depth, only to be told in no uncertain terms by a young geneticist (which we all were once) with no experience in your specialty that you were completely mistaken, that your expertise was eclipsed by genetics and had no future. I can remember the brashest “young Turks” in the field saying things like “biochemistry is dead”—a relief to many, I am sure—but meaning that there was no longer any need to research things like enzymes and proteins. All you needed to know was the DNA sequence and everything would flow from that. This was pure hubris, and there will be many a medical specialist, now retired, who grins with quiet satisfaction when he reads that, ten years and billions of dollars later, the Human Genome Project has achieved very little as far as alleviating or even untangling the suffering caused by disease. The claims of ten years ago that we were about to witness the greatest medical advance since antibiotics have proved, thus far, to be thoroughly hollow.
However, in genealogy the two approaches have not clashed. This is partly because few scientists strayed into the field, so it never became a great source of career advancement or research funds, and also because the genealogy world was so well organized and mature. Genetics could never on its own replace the painstaking effort in the records, but it can and has added a brilliant new tool to the genealogists’ kit. The two do work hand in hand very well, and the Clan Donald Project, orchestrated from the United States, is a prime example.
Although it might be thought that the official clan historians might resent the genetic conclusion that Somerled and the chiefly line of Clan Donald has a Norse rather than Gaelic origin, that has not happened. Genetics is not a dirty word within the clan, as the sheer size of the DNA project testifies. So, in addition to the discovery of the chiefs’ Norse ancestry, what else has been revealed about the clan? I am well aware that in genealogy there is nothing as interesting as your own family history
, and nothing as dull as someone else’s, so I hope you will forgive my delving deeper into this clan’s genetic history by way of illustration. Every family is different, of course, but most of what happens in other families also happens in Clan Donald.
Let us start with the chiefly line, which is Somerled’s own. We can be sure of that because the same Y chromosome is found among McDougalls who descend from Dugall, the elder brother of Ranald, whose patrilineal descendants are the clan chiefs of Clan Donald. The number of Y-chromosome markers has been extended from the original seven that I used to study the clan in the late 1990s to thirty-seven and, in some cases, even more. Using the chiefly DNA as reference point, 120 of the 935 participants who had enrolled when I examined the results carry either an identical or very close Y chromosome as defined by the markers. However, they are not all called MacDonald or McDonald. There are a lot of McConnells and McDonnells and McDaniels, which confirms these as alternative spellings to MacDonald. There are also McKeans, a couple of Alexanders and Wilsons, a McCain, a Gordon, a Douglas, and even a Campbell, the sworn enemies of Clan Donald since the Massacre of Glencoe in February 1692. Thirty-eight members of Clan Donald were murdered by the Campbells, who had been on the opposite side during the Jacobite uprising of 1689. What drenched the incident in infamy was that the Campbells had accepted the hospitality of the Macdonalds and had been their guests when the killing started. The massacre is still remembered in the Highlands. The memory caused an uproar only a few years ago when a Campbell was employed as the manager of the Glencoe Visitor Center, so much so that he was forced to quit. Some of these names are known spelling variants of a clan surname, such as Alexander, which is a common form of MacAlastair. Others with no obvious surname relationship, like Wilson and Campbell, may well be the result of “aristocratic diffusion,” as they are likely to have come about from illicit liaisons. Somewhere in the past they got the Clan Donald chromosome without the surname. In my own study I found a couple of MacArthurs with the Somerled chromosome. The MacArthurs were hereditary pipers to the Clan Donald chiefs, which might have brought them within breeding range. They also hail from Islay, which was a Clan Donald stronghold during the fourteenth to sixteenth centuries. Either way some MacArthurs certainly carry the Somerled chromosome.
The much larger Clan Donald DNA Project has been able to look more closely at the internal genealogy of Somerled’s descendants within the clan. This is a great achievement, and I think looking in a little more detail will be rewarding, not so much for what it says about an important Scottish clan, but as a superb example of what can be achieved with the combination of genetics and genealogy. Using an array of Y-chromosome markers, the Clan Donald DNA Project has compared the genetic signatures of nine men who have very solid paper genealogies back to the key ancestor, John, Lord of the Isles, who died in 1386. He is labeled “E” on Figure 2. He was the great-grandson of Donald MacRanald (C), from whom the clan takes its name. Donald was Somerled’s grandson and one of his sons, Alastair (D), is the ancestor of the MacAlastairs, who, being descended from Donald, are embraced by the clan. Unlike the MacAlastairs, the MacDougalls, not being descended from Donald but linked to Somerled through his eldest son, Dugall (B), are not part of Clan Donald even though they share the same Y chromosome.
Figure 2. Genetic genealogy of Clan Donald. Y-chromosome mutations are shown as explosions () adjacent to the links on which they occur. Lengths of vertical links are approximately proportional to the number of generations. Details of the mutations are given in the appendix. Labeled individuals are identified in the appendix.
The men in the bottom rank (M-V) have well-documented genealogies back to John, Lord of the Isles. Three are chiefs—Ranald Alexander MacDonald, twenty-fourth of Clanranald (M); Ranald MacDonell of Glengarry (P); and Sir Iain MacDonald of Sleat (R). There are also two chieftains; Allan MacDonald of Vallay (Q) and David MacDonald of Castle Camus (S), while the other five men are neither chiefs nor chieftains. That is the genealogy, now what about the genetics? This is where it gets complicated. What I want to do now is to look at how closely a genealogy based solely on the genetics resembles the Clan Donald genealogy as constructed from the records. It is a question I am often asked by families who are considering embarking on a genetics project. Rather than interrupt the narrative flow, I have put the details in the appendix.
Of the thirty-eight markers tested in this genealogy, thirteen showed genetic variation within the clan. Mostly there was only one mutation from the original Somerled Y chromosome at each of the thirteen markers, but one of them has mutated three times and another five times. These recurrent mutations are a real problem in trying to reconstruct an accurate family tree based on the genetics in the absence of a well-documented genealogy. In this case the very same mutation has happened three separate times in different parts of the family tree. We call these “parallel” mutations. Without the genealogy, it would look at first sight as if there had only been one mutation, and the inclination would be to group all the men whose Y chromosome carried it as if they were descended from the man in whom the mutation had first occurred. This is a familiar issue for geneticists who are trying to reconstruct past events using genetic evidence from people alive today. My colleagues and I had to struggle with it when dealing with hundreds of mitochondrial DNA sequences, where parallel mutations are also an occupational hazard.
One solution is to consider all the possible family trees that could be drawn and then work out which one requires the least number of parallel mutations. That is a reasonable procedure for the construction of a really large-scale tree, like the maternal genealogies of Europe for example, when the aim is to get close enough to the true tree to be able to draw some general conclusions. But it isn’t appropriate in a family genealogy project, where it is very important for individuals to be accurately placed. The path of least resistance, or “maximum parsimony” as it is known in the trade, is not designed to be absolutely accurate, just roughly right. In fact I doubt whether the most parsimonious tree is ever absolutely accurate, and it can be misleading. Take the case in Clan Donald, where the same mutation has happened three times. Maximum parsimony without the accompanying genealogy would come up with a tree with the least parallel mutations, and it would be wrong.
Another important message from the Clan Donald DNA Project concerns the estimates made when two men try to find out how long ago their common ancestor lived by comparing their Y-chromosome results. It stands to reason that the more generations have passed since that ancestor was alive, the more likely that their Y chromosomes will differ from one another. This is because mutations accumulate over time, and the longer the time, the more mutations will have occurred. While this is a perfectly accurate general statement, it can be wildly misleading in individual cases. It works well when large groups are involved, so in Clan Donald, the Y chromosomes of the ten descendants of John, Lord of the Isles (E on the genealogy in Figure 2), have experienced a total of twenty-five mutations. How many generations will have passed to give that figure? The Y-chromosome-marker mutation rate is generally reckoned to be about 0.31 percent (i.e. 0.0031) per marker per generation, so the average number of generations needed to generate twenty-five mutations in thirty-eight markers at this rate is 25/(38 × 0.0031), which comes to 212. Since we know the genealogy, we know the true figure to be 139 generations, so while the 212 estimated by the genetic data alone is too high, it is not wildly off. This is when we are dealing with a fairly large group, and since the Clan Donald DNA project is the largest of its kind, other family studies will have fewer participants, and the estimates will tend to be more inaccurate. However, when you look at individual branches, the real problem becomes apparent. And it is precisely these smaller-scale issues that most genealogists I hear from look to genetics to solve.
Very frequently two men, who might share the same surname and whose Y chromosomes are similar but not identical, ask how long ago their common ancestor lived. The straightforward reply is to give an average
figure, based on the same sort of calculations we have just used. But a close look at the Clan Donald genealogy shows how very misleading this can be for individual branches, even when, as we have seen, the results for the whole clan are reasonably close to expectations. You don’t need sums to see this, just look at the diagram. Suppose the first two men in the bottom row asked just that question. We know from the genealogy that both men are descended from Ian Moidartach MacDonald (K), the eighth chief of Clanranald, who died in 1584, and that the two men, the current twenty-fourth chief (M) and a Mr. P. M. Macdonald (N), are separated from their common ancestor by eleven generations on one side and twelve on the other, making a total of twenty-three. Their Y chromosomes have accumulated a total of nine mutations, five on one line and four on the other since they last shared a common ancestor. This gives an estimate that they are separated from each other not by the twenty-three generations that we know to be the case from the genealogy, but by 9/(38 × 0.0031), which is 76 generations and more than three times as many.
At the other extreme, J. J. Macdonald (O) and Allan Douglas MacDonald of Vallay (Q) last shared a common ancestor in John, Lord of the Isles (E), who died in 1386. The genealogy shows the two men to be separated by thirty-nine generations, yet their Y chromosomes are only three mutations different. The estimate from the genetics puts the number of generations separating the two men at 3/(38 × 0.0031), which is 25 and far fewer than in reality.
Though I have chosen extreme examples I hope I have shown enough to convince you of the great inaccuracy of these estimates for individual cases. Not only do I believe they are next to useless, they are also misleading. Many people naturally think that increasing accuracy will come by increasing the number of markers tested. It will not. The random nature of mutation and the complications of parallel mutation will see to it that no estimate of the time to a common ancestor for any two men will be much more than a shot in the dark, even with the abundance of markers that are now available. And it is a fallacy to think that using more markers is the answer. Even with a perfectly respectable thirty-eight markers, any treatment of the genetic results on their own would appear to prove that J. J. MacDonald (O) and Allan Douglas MacDonald (Q) were far more closely related than Ranald Alexander MacDonald (M) and P. M. Macdonald (N), whereas the reality is the complete opposite. Caveat emptor.