C. Elegans do it. The circadian rhythm is something that may be a global property of terrestrial life. Regardless of sunlight then living things tend to operate on a daily routine, from rest to activity, and from high to low metabolic activity.
The exact biochemical origins of this internal clock have been somewhat elusive. In last week's Nature two new works by O'Neill et al. shed some more light on the subject. A possibility has been that a transcription/translation feedback loop governing expression of certain 'clock' genes played a role in setting the 24 hour timer in organisms. O'Neill and colleagues seem to have found good evidence that there are additional, possibly superior, 'time-keeping' processes at play. In essence these are chemical 'oscillators' that behave like a well-tuned pendulum. Intriguingly this type of mechanism was already known to operate in the ancient cyano-bacteria. In tandem then perhaps both the purely chemical and gene mechanisms act like a self-correcting clock, keeping life to a consistent 24 hour timetable. The genetic coding for the chemical clock seems likely to be shared amongst organisms like ourselves and ancient bacteria.
This is all very interesting. However, it also raises a number of questions that I've not seen discussed in detail in these or related experiments. 24 hours is the rotation period of the modern Earth. The Earth-Moon system has been in constant dynamical evolution since the formation of the Moon about 4.53 billion years ago following a massive proto-planet collision. At present the gravitational tides due to the Moon are dissipating energy at a rate of a few Terawatts and slowing the Earth's rotation by about a couple of milliseconds a century. Other variations, like changing ice-caps, solar tides, even tectonic shifts tend to obscure this slowdown on short timescales but over millions of years there is little doubt that the Earth's spin has been slowing. At the same time angular momentum conservation means that the Moon is receding from us at a few centimeters a year - a fact confirmed by laser ranging.
The upshot is that it's quite possible that 4 billion years ago the Earth's daylength was only 12 hours. Geological evidence is scarce to non-existent that far back, but studies of material deposited on what were once tidal shorelines indicate that around 600 million years ago the day length was certainly more like 22 hours, and the slowdown should have been more extreme in the further past. So the intriguing question to ask is how the biochemical clocks, be they the genetic or chemical variety, adjust over the millenia to that shift? Or, to be provocative, is there some way we could use our understanding of the evolution of these mechanisms to independently test the physical changes to Earth rotation over hundreds of millions to billions of years?
Celestial mechanics probed by paleogenetics? That sure sounds like fun.