With Ardipithecus radius (about 4.5 million years ago) we have the strongest evidence so far that hominins have adopted bipedalism. Earlier fossils, including the earlier Ardipithecus kadabba, are too fragmentary to be very sure. Even “Ardi” was not bipedal quite the way we are. She had a somewhat diverging big toe, and arms and hands well-adapted for suspension, suggesting she was bipedal on the ground, but still spent a lot of time in trees.
We’ve seen bipedalism before on Logarithmic History. Bipedalism allowed ancestral dinosaurs to overcome the tight coupling of locomotion and respiration that prevents sprawling lizards from breathing while they run. But human bipedalism, with no counterbalancing tail, is different. As far as we know it evolved only once in the history of life (or maybe twice if Oreopithecus was bipedal).
In part human bipedalism is related to the general primate phenomenon of having grasping hands. Both humans and macaques, for example, devote separate areas of the brain (within the somato-sensory cortex, specifically) to each finger on each hand. Brain areas for the toes, by contrast, are more smooshed together.
Human bipedalism is more specifically related to tradeoffs in locomotion in great apes. Other great apes pay a big price for being the largest animals well-adapted for moving around under and among branches: great ape locomotion on the ground is particularly inefficient. Chimpanzees spend several times as much energy knuckle-walking on all fours as you would expect based on comparisons to similar sized quadrupedal mammals. Remarkably, chimpanzees don’t take any more energy walking on two legs than they do walking on all fours, even though they aren’t at all well-adapted to bipedalism. Humans by contrast take a little less energy to walk around than a same-size four-legged mammal, and way less than a chimp. (No one has yet done the research to figure out how (in)efficient humans are when getting around all fours. More research needed!) A recent study shows that the same was true of Ardipithecus: she was an efficient bipedal walker on the ground. In other words, being good at climbing trees (although not as good as a chimp) didn’t hurt her when it came to getting around bipedally on the ground.
That said, efficiency isn’t everything. Human beings are lousy at sprinting – try outsprinting your dog, or a squirrel for that matter. Our top speed is less than half that of a chimpanzee.
So there’s a tradeoff between the efficiency advantages of bipedalism (at least compared to knuckle walking), and the loss of speed. It may be that bipedalism evolved initially in an environment where predation pressure wasn’t very intense, and the need for speed was not as great. This argument has been made for Oreopithecus, living on an island in the Mediterranean. Perhaps Graecopithecus initially enjoyed a similar isolation, and freedom from predation, associated in some way with the drying and flooding of the Mediterranean.