Load-bearing adaptation of women’s spines


Blogging on Peer-Reviewed Research

Those of you who have been pregnant, or have been a partner to someone who has been pregnant, are familiar with one among many common consequences: lower back pain. It’s not surprising—pregnant women are carrying this low-slung 7kg (15lb) weight, and the closest we males can come to the experience would be pressing a bowling ball to our bellybutton and hauling it around with us everywhere we go. This is the kind of load that can put someone seriously out of balance, and one way we compensate for a forward-projecting load is to increase the curvature of our spines (especially the lumbar spine, or lower back), and throw our shoulders back to move our center of mass (COM) back.

Here’s the interesting part: women have changed the shape of individual vertebrae to better enable maintenance of this increased curvature, called lordosis, and fossil australopithecines show a similar variation.

The first step is to document the phenomenon. The posture and gait of pregnant women were studied kinematically through their pregnancy, and yes, they do extend their lower spines to shift the COM. The diagrams at the bottom, below, show the situation: on the left (c) is a non-pregnant woman with the center of mass marked. It’s located at the base of the spine, roughly centered over the hip joint. In (d), the woman is braced to hold her spine in the non-pregnant posture, and the weight of the fetus pushes the COM forward, to a place that would put her out of balance. In (e), you can see the preferred posture in pregnancy: the s-curve of the back is increased, moving the COM back over the hip joint.

COM and lumbar lordosis during pregnancy.
a, Quadrupedal chimpanzee, non-pregnant. b, Quadrupedal chimpanzee, pregnant with no change in sagittal position of the COM with respect to the postural support base. c, Bipedal human female with typical lumbar lordosis and COM in approximate sagittal alignment with the hip. At a given 0.005-m COM distance from the hip, a 409-N upper body generates 2 N m torque at the hip. d, Pregnant human female with anteriorly translated COM, lacking positional adjustment of lumbar lordosis. The force of gravity, when more distant from the hip, generates a larger hip moment and an unstable upper body. With pregnancy, a 511-N upper body and a COM at 0.032 m from the hip increases the torque to 16 N m. e, Typical pregnant human female with naturally extended back and recovered COM by means of increased lumbar lordosis, a stable positional alignment with reduced hip torque (1.5 N m) but with exacerbated spinal shearing load. Open circle with cross hairs, COM in sagittal plane; filled circle, hip position in sagittal plane; arrow, direction of gravitational force.

Now let’s take a look at the anatomy. These are not measurements from the same women in the kinematic study, but from collections of skeletal remains where age and sex of the individuals were well-established from post-mortem records. The qualitative observation is that the three caudalmost lumbar vertebrae, and especially L3, show a characteristic sex difference. L4 and L5 in both sexes are typically wedge-shaped, to bend the spine forward; L3 in men tends to be more columnar, while L3 in women is also wedge-shaped, promoting a greater curvature.

(click for larger image)

Sex differences in the lumbar vertebral column of human males and females.
Female values are shown by filled bars, male values by open bars. a, Wedging angle of vertebral bodies, angles greater than 0u are kyphotic (thoracic-type wedging), whereas angles less than 0u are lordotic (lumbar-type wedging). Females present a longer series of dorsally wedged vertebrae; L3, L4 and L5, whereas males are lordotic at only two levels, L4 and L5. b, Prezygapophyseal (prezyg.) area, adjusted by geometric mean for overall vertebral size. The female area is significantly larger than the male area at L2, L3, L4 and L5, indicating that females bear a greater proportion of spinal load along the dorsal pillar, which is consistent with fetal loading patterns identified during pregnancy. c, Prezygapophyseal angle. The female facets are significantly more oblique at L2, L3 and L5, conferring greater resistance to forward displacement of lumbar vertebrae. In a-c, n 5 59 males, 54 females. Data are means and s.d. d, Diagram of lumbar region in males and females, showing contrasting mean wedging patterns and anatomical structures within the dorsal pillar (including zygapophyses) and ventral pillar (vertebral bodies). e, Difference in vertebral body shape in males and females at L3. There are equivalent angles of excursion yet there is greater upper body extension in the female spine. The inherent dorsal wedging shape of the female L3 relative to the non-wedged male L3 generates less shearing force when the upper body is repositioned by means of lower back extension, as occurs during fetal loading.

The quantitative measurements back the observation up. The charts on the left above show the shape of the vertebra, and the area and angle of the zygapophyseal process. Just look at (a), though. That chart illustrates the degree of wedging of individual vertebrae; a bar that goes downward means the vertebra is trapezoidal, with the narrow end to the back, bending the spine foward; bars that extend upwards means the vertebra is wedge-shaped in the other direction; and bars that are close to the baseline mean the vertebra is columnar. Women’s vertebrae are black, men’s are white. Look in particular at L3; women’s L3 has a bar that is downward, while the men’s L3 is more columnar.

The variation is very large, though, which I’ll come back to later.

Now here are some measurements from a pair of australopithecine spines. It doesn’t quite have the same distribution as the above graph, but keep in mind that this one is from two individuals, while the modern human skeletal data is from 59 males and 54 females. Basically, though, the data show that one individual has more columnal vertebrae, while the other is more wedge-shaped. There is no independent determination of the sex of these two individuals, though, so all we can really say is that we see a morphological difference in the lumbar vertebrae that parallels a sex difference we see in modern populations.

Australopithecine lumbar lordosis and prezygapophyseal angle.
a, Angle of lumbar vertebral wedging for Australopithecus africanus specimens Sts 14 (red) and Stw 431 (blue). Sts 14 shows a wedging pattern similar to that in modern human females, comprising the three caudalmost lumbar vertebrae, L4, L5 and L6. Although the preserved lumbar column of Stw 431 is less complete than that of Sts 14, the caudalmost levels are preserved well enough to identify a different wedging pattern. The dorsal wedging sequence of Stw 431 includes only one lumbar vertebra, at the last lumbar level. In this manner, Stw 431 is unlike Sts 14 and modern human females and is more similar to modern human males in having a shorter region of lordotic lumbar vertebrae. b, The prezygapophyseal angle of the preserved lumbar region for Sts 14 and Stw 431. The larger angles of Sts 14 relative to those of Stw 431 mirror the modern human female-male pattern in that Sts 14 presents more oblique angles and therefore greater coronal orientation of the prezygapophyseal facets than Stw 431.

So far, so good. I can believe that the authors have identified a statistical difference in the anatomy of lumbar vertebrae between males and females. However, I have some significant disagreements with the evolutionary interpretations of the paper. They claim to have identified evidence of an evolutionary novelty, but they haven’t tested the alternative hypothesis, that this is not an evolutionary adaptation, but a physiological one, and they haven’t adequately distinguished cause and effect.

My first thought on reading the results was that this is an example of developmental plasticity. Bones are flexible; they respond to stress with changes in shape and size that accommodate them to the pattern of activity they experience. This is an indirect evolutionary adaptation, of course — that bones have this response is a product of their genetic and developmental potential. However, the shape of an individual vertebra may not be so precisely specified, but may emerge as a product of the strains put upon it.

I’d make an alternative hypothesis. The female L3 vertebra is not wedge-shaped because women need to bear a fetal load, but instead, because women bear a fetal load, the L3 vertebra is wedge-shaped. In particular because their own data shows a significant amount of variability in vertebral shape, I’d be hesitant to assign a direct genetic cause on the pattern.

Unfortunately, the data in this paper do not touch on this possibility. All of it is from either pregnant women, or from skeletal remains of adults of child-bearing age. What I’d like to see is some developmental information, especially measurements of lumbar vertebrae in pre-pubertal children. If the difference precedes the child-bearing experience, then I’d agree that they’ve found a sexual dimorphism that could have an evolutionary cause.

Other data I’d like to see: is there a difference in vertebral morphology between women who have had children and those who have not? Another sex difference that could generate variation in vertebral morphology besides pregnancy is breast size; like carrying a fetus, women have another forward projecting weight that can shift the center of mass. Do large-breasted women have a consistent change in vertebral morphology that isn’t found in small-breasted women or men? How does obesity affect vertebral shape?

The authors have identified an interesting sexual dimorphism, but I think the paper was far too quick in assigning an evolutionary selective cause for the difference, and that it did not adequately examine the more likely (to my mind, at least) explanation of physiological adaptation.

Whitcome KK, Shapiro LJ, Lieberman DE (2007) Fetal load and the evolution of lumbar lordosis in bipedal hominins. Nature 450(7172):1075-8.


  1. #1 David Marjanovi?, OM
    December 15, 2007

    Perhaps the curve to spine was crucial to four footed and knuckle walking apes for some reason, but has not been lost in the move to upright posture?

    No. Look at the chimp again (at the top of the first figure): it doesn’t have the curvature.

  2. #2 David Marjanovi?, OM
    December 16, 2007

    Fifteen pounds? I think PZ’s partner must be a squirrel. What fraction of women gain fifteen pounds in pregnancy? Fifteen to twenty kilos is more like it, but most women gain more.

    My mother says 10 kg all-inclusive, and I’m the first of four children, so I dare say she speaks from experience… My youngest sister had 4.39 kg at birth. She was rather globular.