When did the modern human pattern of childbirth arise? New insights from an old Neandertal pelvis.

H uman birthing is difficult owing to a tradeoff between large neonatal brain size and maternal pelvic dimensions, which are constrained by aspects of bipedal biomechanics (1, 2). The net effect is that human neonatal head size closely matches maternal pelvic dimensions, unlike in our closest living relatives, the great apes, whose pelvic dimensions are larger than neonatal head sizes. This size relationship, along with a twisted birth canal shape, makes human parturition mechanically difficult and results in a unique pattern of ‘‘rotational’’ birth (Fig. 1). As in humans, and unlike great apes, monkey neonatal head size closely matches the mother’s pelvic dimensions. Although birth rotation occurs in some monkeys, the rotation pattern is different from that in modern humans, with monkey neonates exiting the birth canal facing forward (1). Therefore, in both apes and monkeys, it is relatively easy for a mother to guide her infant out of the birth canal, keep the umbilical cord from wrapping around the neck, and extract mucous from the nose and mouth to facilitate breathing. All of these critical activities are much more difficult for a human mother, whose infant emerges facing backwards, and pulling on the infant from this position also risks serious neck injury. For these reasons, humans uniquely engage in assisted birth (obligate midwifery). Although unassisted birth, the norm for nonhuman primates, does occasionally occur in humans, assisted birth and its myriad social implications are the human norm crossculturally (1). Obstetrical difficulties and some form of birthing assistance may have arisen at the outset of hominin evolution in concert with bipedalism (3), yet it seems likely based on 2 preserved Pliocene female pelves, A.L. 288-1 (Australopithecus afarensis) and Sts 14 (Australopithecus africanus), that birth in australopiths was nonrotational, with the long axis of the neonate’s head oriented transversely as it moved through the 3 birth canal planes (2). The extant human pattern of rotational birth, along with intensified and obligate midwifery, most likely emerged later when a critical threshold of brain size attainment in Homo was added to the shared hominoid pattern of broad, rigid shoulders (4). But when did this happen? Until now, the consensus has been that rotational birth was not present in Early Pleistocene Homo (5), recently confirmed by the discovery of a largely complete 1.4–0.9 million-year-old female Homo erectus pelvis from Gona, Ethiopia (6), with the primitive nonrotational pattern persisting until the Middle Pleistocene (5). Resolving the appearance of rotational birth more precisely has been hindered by the paucity of fossil pelvic remains, especially from females, because of the fragility of pelvic bones. In a recent issue of PNAS, Weaver and Hublin (7) report on a virtual reconstruction of the pelvis of the Tabun C1 adult female Neandertal which indicates, surprisingly, that Neandertals retained the primitive birth pattern into the Late Pleistocene, and that the shift to a rotational birth pattern occurred later in human evolution than previously thought. Over the past 25 years, rotational birth has been inferred for archaic Homo sapiens based largely on 3 specimens (3): Sima 1, a 600,000-year-old male pelvis from the Sima de los Huesos, Spain (8, 9); a 260,000-year-old left os coxa belonging to a partial female skeleton from Jinniushan, northeastern China (10, 11); and a 60,000-year-old partial pelvis from a male Levantine Neandertal, Kébara 2 (12). Nonetheless, it has always been recognized that extrapolating female birth canal shape from male pelvic remains is uncertain. Moreover, whereas the inlet anatomy of the female Jinniushan specimen can be estimated (11), the midplane and outlet dimensions are unknown. For these reasons, Weaver and Hublin have returned to a fossil discovered nearly 80 years ago. The Tabun C1 pelvis belongs to a 60,000to 100,000-yearold female Neandertal skeleton and was partially reconstructed and described in 1939 (13). Her pelvic remains consist of portions of the left pubis and ilium, as well as the right pubis, ischium, and ilium. The advent of computerized virtual reconstruction allows highly accurate restorations of fragmented and variably distorted fossil elements, including mirror imaging of damaged or missing elements. The Weaver and Hublin recon-