Embedded Files
No, our ancestors didn't spend 3,000 years traveling in a circle.
FTDNA's new Globetrekker Beta is the latest in a line of algorithm-based YDNA migration predictors such as PhyloGeographer and SNP Tracker. Essentially these programs utilize ever growing YDNA SNP trees, Ancient samples, and modern testing data (SNPS, Country of Origin, SNP frequency) to recreate your line's theoretical journey across time and place.
Of course, for many of us, knowing how we got here from the beginning, generation after generation, is truly the pot of gold at the end of the YDNA rainbow. But like anything else in life, algorithms like those used by Globetrekker and others can only become remotely accurate when supplied with overwhelming data. Essentially, it can only give out, what you put in.
When FTDNA states that Globetrekker is "...based on the largest database of high-coverage Y-DNA sequences, ancient DNA results from archaeological remains, and user-reported ancestral locations," they aren't lying. Despite the criticisms of some in the community, the company is largely responsible for keeping the YDNA testing industry relevant to the general public. Its Discover platform has many great offerings, most notably the Ancient Connections and Time Tree tabs which have one-upped some of the most beloved advantages of sites like Y-Full. It remains true that FTDNA does not allow for the uploading of YDNA tests from other platforms which still makes Y-Full an important player in the industry. But they are putting much more data into Globetrekker than any other algorithm based migration estimator; that much we can be certain.
However, it's simply not enough. Despite FTDNA's sizable database of BIG Y testers and the ever increasing scientific database of ancient DNA, there is just nowhere near enough data to make reliable migration estimates for most SNPS. This is especially true for testers in rare or under-tested Haplogroups.
It is true, that as YDNA testing exponentially expands, the Globetrekker tool will become more accurate. Over time it may possibly be pumping out compelling migrations for select clades like Haplogroup R, which makes up nearly half of all FTDNA Big Y testers. FTDNA does informs users that the tool will need to be constantly fine-tuned, "These are best estimates and will change over time as more people test their Y-DNA and provide information about their paternal line ancestry." What they don't do, at least not loud enough for my liking, is warn users how drastically bad those best estimates might be at this very moment and seemingly for a long time to come.
This is a hypothetical tool. Sure, there is the fine print Caveats section provided by FTDNA which acknowledges the major shortcomings of the estimator. But in some cases these shortcomings do not result in a few inaccurate missteps on a map, but in pathways that do not remotely recreate any semblance of an accurate migration path.
A) Globetrekker assumes present day populations live near their ancestral locations which we know is not always true.
B) FTDNA's customer base is tremendously skewed towards the West (the United States and Europe) and many cultures and ethnicities are drastically underrepresented. Going further than that, many population-based genetic studies into haplogroup frequency such as 2012 Rootsi Study of Haplogroup G come from old data and many were based on STRs and not SNPs.
C) The algorithm assumes slow and steady movement over time, as if each father's son journeyed just a few miles over the horizon on their march to whatever terminal location is determined. This of course, doesn't work for many migrations caused by war, famine, colonization, captivity, resettlement, where long distances were traveled in just one life time. In addition, there is something to be said about your ancestors staying put and not going anywhere at all! In some cases for thousands of years!
D) The tool ignores negative findings. Such as where a certain SNP has not been found in ancient DNA. If a number of ancient samples have been sequenced in a particular region, and a certain SNP has not been identified, that is also data. In certain situations this could be useful in determining where migrations did not occur.
Some of these problems are NOT fixable without ancient DNA informing nearly every step of a haplotype's journey or without many tens of thousands of diverse testers guiding the algorithm to more accurate assumptions. At best these issues make the tool a feel good mosaic of guesswork, at worst it is just outright irresponsible.
In regard to Haplogroup G specifically, Globetrekker is not just bad, it is astoundingly wrong. It currently places G-L30 as forming in the Balkans around 15,000 BC. It has the G-M406 signature, which formed around 12,500 BC, arising in Macedonia. Several other subclades of M406 like G-FGC5089, G-M3302, G-M3240, it purports migrated in a big circle from Thrace to Bulgaria, North Macedonia, and back towards the Aegean. Sadly, this led one Globetrekker user to invent an explanation for his ancestor's cyclical journey, "It looks like my M406 ancestors wandered southeastern Europe for thousands of years, maybe they were following game or were pushed back where they came from by another group."
Nevermind, the fact that the oldest G-L30 were found in some of the first farming villages of Anatolia like Barcin and Boncuklu 7,000 - 8,000 BC. Later this clade was found along the Danube, coinciding with the spread of farming into Europe around 6,000 BC. Additionally, we also have established time and again, via overwhelming evidence from ancient DNA, that G-M406 was predominantly in Central Anatolia until at least the Bronze Age or later.
Clearly (hopefully?) the Beta version of Globetrekker isn't yet accounting for ancient G Haplogroup samples. It also is clearly being skewed by early studies putting Greece as a hotspot of G-M406 frequency as well as the tug of so many G-M406 European testers. No ancient G-M406 individuals have been found in Europe until the Roman era, of the many, many sequenced Neolithic Europeans, none have been G-M406, that is data too my friends.
Of course the algorithm is programmed to deliver a slow and steady transition of G-M406 sub-clades into Europe. While this pace would fit the first farmers who slowly expanded up the rivers of Europe, it is likely not the case with G-M406 which occupied the same Central Anatolian orbit for thousands of years and then seemingly spread over great distances rather quickly. Our best guess on G-M406 migration out of Anatolia is that is spread rapidly over short periods of time via Mediterranean colonization, the Roman slave trade (see Logical Steps). Some may have been seeded rather suddenly in the city of Rome via the slave trade and spread more gradually through Europe after its decline. G-M406 also likely spread via small groups moving over the Caucuses and up the rivers of Eastern Europe and as well into Arabia which on the wings of Islam would have rapidly spread into the Far East. Again, these are scenarios that will always be very difficult for an algorithm to model accurately.
To come full-circle, we are left with the unfortunate scenario that FTDNA's greatest strength, getting casual genealogists to purchase next-generation YDNA sequencing, may also mean its users are more susceptible to taking Globetrekker as fact. Without reading and understanding the fine print, these customers may draw misinformed conclusions about the journeys of their male ancestors. FTDNA is expected by its users to publish tools based on science and research. Right now FTDNA's Globetrekker is getting a lot awfully wrong, and that of course is not doing the field or its customers any favors in the long-run.
FTDNA user: "My ancestors seem to have felt for southern Denmark: After first arrival there they fled the Ice Age to Spain, and returned. Then they visited Lithuania, and returned. Next to Sweden, and returned,"
Did they? Really?
As for the rest of us who read the fine print and/or spend way too much time obsessing over the footprints of history...we may very well be cursed with repeatedly dispelling misguided assumptions and crushing the dreams of casual users with copy/paste explanations that basically say, "HEY, Globetrekker is fun to look at it, but it doesn't really mean much....yet"
Page updated
Google Sites
Report abuse