Interesting that the Mycenaeans plot so differently here (and on other amateur calcs) compared to the Lazaridis PCA. Also of note is how much more southerly they are compared to Sicilians and south Italians here, whereas in the Lazaridis plot they're very close. Any reason for such contrast?
PCAs aren't calibrated at cardinal points. They are not made to see who is more north or who is more south, which is irrelevant. A PCA, a statistical procedure, is based on components. These components influence the position of the various samples in a PCA, and the samples in turn affect each other, and every calculator is different from the other. Lazaridis uses more sophisticated tools, while ancient samples on gedmatch are often of low quality or of variable quality. Always check how many SNPs are used in the calculator's evaluation. Ancient samples less and less than any modern sample. Anyway Mycenaeans are not much more southerly compared to Sicilians and south Italians in this PCA, Mycenaeans are simply more Neolithic-shifted (Greece-Neolithic or Sardinian-shifted), and the position of the Sardinians, which are a genetically isolated population, still varies greatly from admixture calculator to admixture calculator.
I read somewhere that only one of the Mycenaean samples is of really good quality. Do you happen to know which one it was?
Likely I9041? Based on gedmatch SNPs all the Mycenaeans seem to be of lower quality than Minoans. But I could be wrong.
Pratt, i was also surprised at the variation. The only two Mycenaeans that group tightly are great friends with the South Italians. As for the one far to the left near the neolithic groups...why is it plotting so far out? I'm beginning to question whether he/she's an actual mycenaean and not just a foreign visitor. I know they are mostly Anatolian, but that one plots too far from the rest.
Were the Mycenaens this diverse?
You mean Mycenaean I9010? Because it has greater percentage of Mediterranean/EEF.
I would be dishonest if I tell you I know the reason. In forums there are already too many users who have the answers, often wrong, and never ask the right questions.
Trying to find a common sense answer it may be that the ancient samples, already from the start lower in quality than the modern ones, in the process of converting to gedmatch uploadable formats, may lose even more quality. On gedmatch all these samples tend to have very few components, and this could distort their position in a PCA as well. Especially with the oldest samples, many SNPs had not yet formed, and on gedmatch all these samples have less SNPs evaluated than modern samples.
Another common sense explanation it may be that the ancient samples are different because they have always been so: southern Europe has already been inhabited and the process of assimilation of new migrants (Yamnaya-like or Caucasian-like or whatever) has never been homogeneous, because that is how it happens also in real life. Someone blends with newcomers first and more than others. Others, on the other hand, mingle less, but then, culturally, are assimilated by the dominant culture imposed by new migrants. For a whole picture we need more ancient samples.