The question has been asked: Why do modal profiles for populations sometimes obtain their most likely matches with other populations?
To make a simple analogy: imagine we are comparing three boxes, each of which contains ten pieces of fruit.
Box A contains 2 apple, 0 bananas, 8 oranges, and 0 pears.
Box B contains 5 apples, 3 bananas, 2 oranges, and 0 pears.
Box C contains 2 apples, 4 bananas, 3 oranges, and 1 pear.
Imagine that each box is like a population, and each type of fruit is like an allele value. Suppose that instead of looking for genetic profiles, we are looking for types of fruit.
First, we will obtain a modal fruit for each box. In Box A, oranges are most frequent, and so the modal fruit is an orange. In Box B, the modal fruit is an apple. In Box C, the modal fruit is a banana.
Now, suppose we have an orange, and we want to predict where it is most likely. The orange is from Box C, but we don't know that. The orange is most likely in Box A, then Box C, then Box B. This is analogous to a modal genetic profile for one population that obtains its highest match in another population. The reason it doesn't match its actual population of origin is that it is not distinctively typical enough of Box A. However, in a statistical sense, an orange is more typical of A than C, and more "A-like" than "C-like." If fruits were alleles (genes), then it is possible that oranges originated in Box A or another unknown box closer to Box A than to Box C. In this case, the likelihood match with A would yield information previously unknown from the recent origins of the fruit in Box C.
Now, suppose we have a pear. That pear is most likely in Box C; in fact, since pears are not found in any other boxes, it must be from Box C or else a box for which we don't have fruit data. This is analogous to a rare allele. Note that a pear is not the modal fruit for Box C, but that it nevertheless is distinctively typical of Box C.
Genetic data is considerably more complex, because: (1) there are many more kinds of fruit (allele values at 13 loci); (2) each genetic profile is not one fruit, but a set of 26; (3) there are many more boxes (currently over 300 populations); (4) fruit moves between boxes (admixture between populations); (5) fruit randomly disappear or multiply in each box over time (genetic drift); (6) fruit spontaneously acquire new characteristics (mutation); (7) two fruits can randomly generate an intermediate kind of fruit (recombination); and many other factors.
Note that genetic data is fundamentally different than genealogical information. Genetic data includes material inherited far back into prehistory; genealogical data extends back only a few generations. Genetic material is transmitted as a partial set of each parent's genes at each generation; genealogical data simply knows who the parents are, not what was inherited from them. Genetic material is transmitted somewhat unequally at each generation; genealogical data assumes a person has an equal share of inheritance from each ancestor. Genetic data concerns what genes actually are carried inside a person; genealogical data considers only where a person is from, their name, their religion, and other external information.
To continue the boxes and fruits analogy: imagine a person is from Box A (Box A is their ethnicity), and their genetic profile is an apple. Their profile (apple) is most likely in Box B, and equally likely to occur in Box A or Box C. However, just because their most likely match is Box B does not require they are 3/10 pear or 2/10 orange. Further, it does not imply that they are not from Box A. It is a likelihood calculation.
However, if that person from Box A is a pear, then their most likely match will be Box A. In fact, their matches for other boxes will be of virtually zero likelihood. This is like a rare allele value highly distinctive of Box A relative to the other boxes.
DNA Tribes uses highly distinct genetic profiles that contain complex and detailed information about each person. For any individual profile, the likelihood of obtaining an exact match anywhere in the world is infinitesimal. Matches are obtained by finding where each profile is most likely. Because genes (allele values) correlate strongly with geography, individual profiles tend to exhibit precise geographical patterns of matches. However, top matches will not always correspond to known ethnicity. However, when they do not, the most likely match gives an important clue as to a person's strongest genetic affinities that paper genealogy might not reveal.
Genes contain detailed information that human memory omits or forgets. Genetics and genealogy ask two fundamentally different kinds of questions. However, the information derived from each source can enrich understanding of the other. Genetic data can put paper genealogy in a broader context, showing how a person's family history relates to world history. Genealogical information can put genetic data in context, showing how genetic material was recently transmitted to an individual.
