Lumbar vertebrae in hominids: six or five?

So have we decided that early hominids had five lumbar vertebrae?

I ask because I was reading a paper by Owen Lovejoy (2005) the other day, and came across this passage:

While most observers have concluded that both STS-14 and STW-431 exhibit six lumbar vertebrae, Haeusler et al. recently presented detailed arguments that these two specimens exhibit only five lumbar vertebrae (as well as KNM-WT 15000, a specimen of H. erectus also
described as having six lumbar vertebrae; see below), but with the last thoracic essentially having lumbar-like function. Their arguments are based on traditional anatomical analysis (presence/absence and location of rib facets, etc.), and need not be considered further here. Even they conclude that all "three fossils differ from the majority of modern humans in that it is the seventh presacral vertebra where the orientation of the articular facets changes" (Haeusler et al. 2002, p. 634), thus making lordosis possible over a longer range of the lower back (Lovejoy 2005:102).

</p>

There are two questions here: first, how many lumbar vertebrae (5 or 6) did early hominids have, and second, how many vertebrae (whether lumbar or thoracic) did they have functioning as part of the lumbar curve?

Haeusler addresses the first question with anatomical examination of the three most complete vertebral columns from early hominids: Sts 14, STW 431, and KNM-ER 15000. From the abstract:

Thus, in Sts 14 the sixth last presacral vertebra has on one side a movable rib. In Stw 431, the corresponding vertebra shows indications for a rib facet. In KNM-WT, 15000 the same element is very fragmentary, but the neighbouring vertebrae do not support the view that it is L1. Although in all three fossils the transitional vertebra at which the articular facets change orientation seems to be at Th11, this is equal to a large percentage of modern humans (Haeusler et al. 2002:621).

The problem in all three cases is that the incompleteness of the column allows doubt about the number of thoracic vertebrae. Their argument is based on the morphology of the putative L 1 / T 12 in Sts 14 and STW 431. The most difficult case seems to be KNM-WT 15000 (the Nariokotome skeleton). Haeusler et al. (2002) discuss the critical differences as follows:

The principal difference between Brown et al.'s (1985) interpretation of five and Walker and Leakey's (1993) six lumbar vertebrae presumably rests on the twelfth thoracic vertebra that the latter authors reported to be missing. This assumption might be based on the impression of an abrupt increase in the sagittal corpus height from vertebra Y to AR/BA (cf. Figure 7). Metrical dimensions, however, do not show abrupt changes between these two segments and hence do not oppose a position of vertebra Y immediately cranial to AR/BA (Figure 8). Therefore, Walker and Leakey's (1993) inference that a vertebra is missing at the thoraco-lumbar border is questionable.
On the other hand, due to its poor preservation, the morphology does not allowidentification of vertebra AR/BA either as the first lumbar element (cf. Walker and Leakey, 1993) or as Th 12, as preferred here. The dorsal part of the body together with the pedicles is missing and thus it is unknown whether or not this vertebra had costal facets. Moreover, almost nothing is preserved of the processus spinosus. The sequence in the morphologies of the processus spinosi from the caudally adjacent vertebra AA/AV ( = L 2 according to Walker & Leakey, 1993) to that of vertebra Y ( = Th 11), however, makes it rather unlikely that a further -- not preserved -- vertebra came in between them in addition to AR/BA. This suggests that AR/BA, in fact, represents Th 12 instead of L 1, and AA/AV would become L 1 instead of L 2. The position of the anticlinal vertebra seems to support this hypothesis. In the modern human spine it is typically at the level of Th 11, with a range from Th 10 to Th 12 (Double, 1912; Danforth, 1930; Ankel, 1967). In KNM-WT 15000, the processus spinosus of Th 11 still projects caudally, which means that Th 12 must have been the anticlinal vertebra. On the other hand, the morphologies of the processus spinosi of vertebra Y and AA/AV -- though fragmentary -- are not far from that of an anticlinal vertebra. This again suggests that AA/AV represents L 1, and thus AR/BA becomes Th 12 (Haeusler et al. 2002:632-633).

Ultimately, Haeusler et al. (2002) and Lovejoy (2005) come down to the same point: if you want to figure out how many of which type of vertebrae there are, then you need to know how many vertebrae there are, and none of the early hominid vertebral columns are complete. It would be easier to say that the last thoracic was actually the first lumbar if you knew that it would be T 13 instead of T 12.

From that perspective, Lovejoy's argument really gains traction concerning the function of the putative first lumbar in the lordosis. This is important to his interpretation of the lower back morphology of apes:

The implications are quite profound. Since apes exhibit the opposite change of lumbar column reduction (Fig. 2), the demonstrably more lordotic column in Australopithecus than occurs even in most H. sapiens signals skeletal reorganization assuring virtually complete abandonment of arboreal activity in favor of permanently terrestrial life. That is, if climbing induces the entire elimination of any mobility in the lower backs in (essentially quadramanual) arboreal apes, it is, by extension, antithetical to arboreality in a species in which so many other critically important adaptations for climbing had also been abandoned (see below) (Lovejoy 2005:102).

Lovejoy traces the functional necessity for lumbar reduction in apes to the reduction in the m. erector spinae:

Apes can only have afforded such reduction of their erector spinae if it was simultaneously accompanied by "passive" spinal immobilization so as to prevent habitual Euler buckling of the column (that is, by virtue of the combined deformation of its individual disks), though shortening of the column has largely eliminated this prospect in apes. Such changes were very likely also associated with other modifications of the thorax (and abdomen), including an "imagination" of the spine ventrally into the thorax (Fig. 6), a character which increases column rigidity and which was presumably associated with their more posterolaterally repositioned shoulder girdle than occurs in monkeys (ibid.).

Well, I don't know if the five lumbar explanation is quite baked in yet, but it might take another vertebral column to turn it around.

References:

Lovejoy CO. 2005. The natural history of human gait and posture Part 1. Spine and pelvis. Gait Posture 21:95-112. DOI link

Haeusler M, Martelli SA, Boeni T. 2002. Vertebrae numbers of the early hominid lumbar spine. J Hum Evol 43:621-643. DOI link