The pelvis is often called the keystone of upright locomotion. More than any other part of our lower body, it has been radically altered over millions of years, allowing our ancestors to become the bipeds who trekked and settled across the planet.

But just how evolution accomplished this extreme makeover has remained a mystery. Now a new study in the journal Nature led by Harvard scientists reveals two key genetic changes that remodeled the pelvis and enabled our bizarre habit of walking on two legs.

“What we’ve done here is demonstrate that in human evolution there was a complete mechanistic shift,” said Terence Capellini, professor and chair of the Department of Human Evolutionary Biology and senior author of the paper. “There’s no parallel to that in other primates. The evolution of novelty — the transition from fins to limbs or the development of bat wings from fingers — often involves massive shifts in how developmental growth occurs. Here we see humans are doing the same thing, but for their pelves.”

Anatomists have long known that the human pelvis is unique among primates. The upper hipbones, or ilia, of chimpanzees, bonobos, and gorillas — our closest relatives — are tall, narrow, and oriented flat front to back. From the side they look like thin blades. The geometry of the ape pelvis anchors large muscles for climbing.

In humans, the hipbones have rotated to the sides to form a bowl shape (in fact, the word “pelvis” derives from the Latin word for basin). Our flaring hipbones provide attachments for the muscles that allow us to maintain balance as we shift our weight from one leg to the other while walking and running.

In their new paper, the international team of researchers identified some of the key genetic and developmental shifts that radically resculpted the quadrupedal ape pelvis into a bipedal one.

“What we have tried to do is integrate different approaches to get a complete story about how the pelvis developed over time,” said Gayani Senevirathne, a postdoctoral fellow in Capellini’s lab and study lead author.

Senevirathne analyzed 128 samples of embryonic tissues from humans and nearly two dozen other primate species from museums in the U.S. and Europe. These collections included century-old specimens mounted on glass slides or preserved in jars.

The researchers also studied human embryonic tissues collected by the Birth Defects Research Laboratory at the University of Washington. They took CT scans and analyzed histology (the microscopic structure of tissues) to reveal the anatomy of the pelvis during early stages of development.

“The work that Gayani did was a tour de force,” said Capellini. “This was like five projects in one.”

The researchers discovered that evolution reshaped the human pelvis in two major steps. First, it shifted a growth plate by 90 degrees to make the human ilium wide instead of tall. Later, another shift altered the timeline of embryonic bone formation.

Most bones of the lower body take shape through a process that begins when cartilage cells form on growth plates aligned along the long axis of the growing bone. This cartilage later hardens into bone in a process called ossification.

In the early stages of development, the human iliac growth plate formed with growth aligned head-to-tail just as it did in other primates. But by day 53, the growth plates in humans evolved to radically shift perpendicularly from the original axis — thus shortening and broadening the hipbone.

“Looking at the pelvis, that wasn’t on my radar,” said Capellini. “I was expecting a stepwise progression for shortening it and then widening it. But the histology really revealed that it actually flipped 90 degrees — making it short and wide all at the same time.”

Another major change involved the timeline of bone formation.

Most bones form along a primary ossification center in the middle of the bone shaft.

In humans, however, the ilia do something quite different. Ossification begins in the rear of the sacrum and spreads radially. This mineralization remains restricted to the peripheral layer and ossification of the interior is delayed by 16 weeks compared to other primates — allowing the bone to maintain its shape as it grows and fundamentally changing the geometry.

“Embryonically, at 10 weeks you have a pelvis,” said Capellini as he sketched on a whiteboard. “It looks like this — basin-shaped.”

To identify the molecular forces that drove this shift, Senevirathne employed techniques such as single-cell multiomics and spatial transcriptomics. The team identified more than 300 genes at work, including three with outsized roles — SOX9 and PTH1R (controlling the growth plate shift), and RUNX2 (controlling the change in ossification).

The importance of these genes was underscored in diseases caused by their malfunction. For example, a mutation in SOX9 causes campomelic dysplasia, a disorder that results in hipbones that are abnormally narrow and lack lateral flaring.

Similarly, mutations in PTH1R cause abnormally narrow hipbones and other skeletal diseases.

The authors suggest that these changes began with reorientation of growth plates around the time that our ancestors branched from the African apes, estimated to be between 5 million and 8 million years ago.

They believe that the pelvis remained a hotspot of evolutionary change for millions of years.

As brains grew bigger, the pelvis came under another selective pressure known as the “obstetrical dilemma” — the tradeoff between a narrow pelvis (advantageous for efficient locomotion) and a wide one (facilitating the birth of big-brained babies).

They suggest that the delayed ossification probably occurred in the last 2 million years.

The oldest pelvis in the fossil record is the 4.4-million-year-old Ardipithecus from Ethiopia (a hybrid of an upright walker and tree climber with a grasping toe), and it shows hints of humanlike features in the pelvis.

The famous 3.2-million-year-old Lucy skeleton, also from Ethiopia, includes a pelvis that shows further development of bipedal traits such as flaring hip blades for bipedal muscles.

Capellini believes the new study should prompt scientists to rethink some basic assumptions about human evolution.

“All fossil hominids from that point on were growing the pelvis differently from any other primate that came before,” said Capellini. “Brain size increases that happen later should not be interpreted in a model of growth like chimpanzee and other primates. The model should be what happens in humans and hominins. The later growth of fetal head size occurred against the backdrop of a new way of new way of making the pelvis.”

This research was funded in part by the National Institutes of Health.