The last Neanderthal.

The mechanism of the Neanderthal extinction and their replacement by modern humans of African origin is one of the most discussed issues in paleoanthropology. Central to this discussion are the questions of the chronological overlap between Neanderthal populations and modern humans in Western Eurasia and the precise geographical circumstances of this overlap. For a long time, the Vindija (Croatia) site was considered to provide solid evidence for a long survival of Neanderthals in Central/Southern Europe. Not only did directly dated Neanderthal remains from layer G1 of the site provide radiocarbon ages postdating the most widely accepted transition time of 40–35,000 radiocarbon years ago (1), but the same layer also yielded a type of split-based bone points commonly assigned to the Aurignacian (2), a stone artefact industry of the early Upper Paleolithic that, to date, only yielded human remains of a modern nature (3). For some, this situation implied the possibility of a long and complex interaction between the two groups of hominins in this region and also falsified the notion of a systematic association between defined archaeological assemblages and specific biological populations at the time of the replacement. In PNAS, Devièse et al. (4) provide new radiocarbon dates for the same Vindija Neanderthal samples, dating them to before 40,000 C B.P., significantly older than previous efforts dating this material to 29–28,000 and 33–32,000 radiocarbon years (1). The bone points of layer G1 could not be dated, but the range of ages obtained from faunal and human samples in this layer suggests taphonomic mixing as a likely mechanism to explain their stratigraphic association in this part of the Vindija stratigraphic sequence. The situation in Vindija is therefore not at all exceptional, and previous results can be explained by the effect of sample contamination and layer admixture. The first message delivered by the new study is one of prudence. Radiocarbon is the most precise method of directly dating human fossil remains. However, at the limit of its application range around 45–40,000 calendar years ago, which unfortunately corresponds to the period of replacement of the last Neanderthals, it is highly sensitive to contamination. For a radiocarbon date of 25,000 y, 1% contamination by modern carbon will produce an age that is 1,628 y too young, but for a radiocarbon date of 42,500 y, the apparent age shifts 8,455 y toward the present (Fig. 1). This length of time is in the higher range of estimates for possible overlap between Neanderthals andmodern humans at the scale of Western Eurasia (3, 5). It is therefore critical when dating organic material extracted from fossil bones to authenticate it as composed of degraded proteins, essentially collagen, of the bone itself. From this perspective, the development of extraction techniques implementing “ultrafiltration” in order to eliminate small contaminants represented a major step forward (6). Ever since, sample pretreatments have witnessed several important improvements. In practical terms, these successive advances mean that, for the transition period, the large number of radiocarbon dates on bone samples produced before 2004 should be used with great caution, if not simply forgotten. To overcome contamination issues in an even more thorough way, Devièse et al. (4) used a method based on the extraction of hydroxyproline, an amino acid specific to collagen. Hydroxyproline dating relies on significant bone sample sizes, as the dated carbon atoms are restricted to those deriving from a single amino acid making up roughly 10% of all amino acid positions of mature collagen type I. Hence, bone sample sizes for hydroxyproline dating are larger than Fig. 1. Contamination effect on radiocarbon dates. The black dots represent three examples of radiocarbon dates without contamination, and the red dots represent the effect of only 1% contamination by modern carbon on the same samples.