Models, predictions, and the fossil record of modern human origins

Abstract

It is clear from the recent contents of this journal and others that the debate about modern human origins continues unabated. My own foray into this subject has dealt with some of the genetic evidence. It has become increasingly clear to me that much of the genetic evidence is indeterminate and that both African replacement and multiregional models can explain observed patterns of genetic variation.1,2 Consider, for example, the finding that many traits show higher genetic diversity within sub-Saharan African populations. While this finding can be interpreted as indicating a greater age for African populations, thus supporting a recent African origin, it can also be explained by a larger long-term African population size, which is compatible with both a recent African origin and multiregional evolution. Population geneticists have long known of the problems of unraveling population history from genetic data. Relationships between populations can reflect either common ancestry or migration.3–7 When considering the predictions of different models, it is critical to make sure that the predictions are unique to each model. I suggest that the problem of indeterminate results is also characteristic of some analyses of the fossil record. In particular, the results from several studies proclaimed as proof of a recent African origin are also compatible with a multiregional model. One approach to analyzing the fossil record has been to compare fossil samples from different geographic regions across time by using some form of biological distance measure.8–11 A useful comparison would be between fairly recent modern human fossil samples (,30 kya) and earlier samples (35 to 1001 kya) across the major geographic regions of the Old World.10,11 The most relevant distances are those across time periods. Are the distances within regions less than the distances between regions? Several analyses have shown that more recent modern samples are morphologically more similar to earlier samples from Africa and the Middle East than to earlier samples within their geographic region. For example, it has been suggested that recent modern samples from Europe (e.g., Cro-Magnon) are more similar to older samples from Africa and the Skhul-Qafzeh samples in the Middle East than to earlier Europeans (Neandertals).8–11 These findings are often taken as support for a recent African origin because this is the type of pattern we would expect to see if all recent modern humans came from Africa within the last 100,000 years. I will not discuss here debates over sample composition, measurements used, or specifics of chronology. My purpose is to examine the underlying assumptions of such studies, and to that end I will take the reported distances as given. Further study can always help us refine our measurements and analyses, but this is of little utility if we do not examine underlying assumptions and make sure that our interpretations are based on valid predictions of the models. Assuming that the distances between fossil samples across time and space are an accurate reflection of past history, it is clear that such results are compatible with a recent African origin. This finding would reject a multiregional model only if the results do not agree with the predictions of a multiregional model. What are these predictions? It is common to see statements to the effect that multiregional evolution predicts that the greatest similarity across time will be within geographic regions. According to this prediction, recent Europeans should be more similar to Neandertals than are fossil samples from other regions at roughly the same time period. This assumption is apparent in Waddle’s design matrix for the multiregional model, in which she predicts that the smallest biological distances will occur within geographic regions.10 The assumption was made most recently with reference to the extraction of Neandertal mitochondrial DNA:12 Krings and colleagues stated that ‘‘whereas the Neandertals inhabited the same geographic region as contemporary Europeans, the observed difference between the Neandertal sequence as modern Europeans do not indicate that it is more closely related to modern Europeans than to any other population of contemporary humans.’’ At first glance, this assumption seems to make sense. After all, given that multiregional evolution incorporates isolation by distance, we would expect, in any given generation, a pattern of population endogamy. At an aggregate level, this would translate into regional endogamy, so that the vast majority of a generation’s genes would come from ancestors one generation earlier in the same geographic region. It is then assumed that the accumulated ancestry over many generations would reflect this pattern, so John H. Relethford, Department of Anthropology, State University of New York College at Oneonta, Oneonta, NY 13820.