Scientific papers usually don't faithfully convey exactly how the researchers came up with the idea, or the chronological order in which the investigation proceeded. And there is a good reason for that - papers need to be standardized so other scientists can easily read them, understand them, replicate them and use them to perform further research.
But sometimes, a paper is honest about the process. It is wonderful - and shows that scientists are human, with a great sense of humor - when #OverlyHonestMethods sneak into the text of a scientific paper, surprising and rewarding the careful reader with an 'easter egg'.
One such paper - on the effects of moon phase of sleep quality - just came out in Current Biology.
The first thing I noticed was that the data were collected in 2000-2003. Why did it take a decade to publish? Was it just sitting on a back burner of a PI for years after the student left the lab? Did it have to go through many rounds of peer review in several journals until it finally managed to get published? None of those reasons, actually! See for yourself:
We just thought of it after a drink in a local bar one evening at full moon, years after the study was completed.
But jokes aside, this is also a great example of a paper that usefully re-visits and re-analyzes old data sets. Of course, the authors emphasize the positives of this post hoc approach - nobody at the time of the study could possibly know that the data would be analyzed in this way, so there were no possible subconscious psychological effects - it was a truly triple-blind study:
Thus, the aim of exploring the influence of different lunar phases on sleep regulation was never a priori hypothesized, nor was it mentioned to the participants, technicians, and other people involved in the study.
On the other hand, a study specifically designed to test for moon-phase effects on sleep quality would have been designed differently to ensure it has just the right controls and that maximum information can be derived from the data.
Research in chronobiology is frustratingly slow. In circadian research, each day is just one data point, so each study has to keep subjects in isolation for many days. In the study of lunar rhythms, each month is a data point and the subjects need to be kept in isolation for many months.
To determine if a rhythm is generated by an internal timer (daily or monthly) as opposed to being a direct behavioral response to environmental cycles requires a whole battery of tests, which are hard and time-consuming enough in circadian research, and twenty eight times more so in circalunar rhythm research
Back in the 1960s, it was possible to keep (well compensated) human subjects in isolation rooms for long periods of time (see pioneering research by Wever and Aschoff in the underground bunker in Andechs, Germany). Likewise, animal subjects can be kept and monitored in isolation chambers for long periods of time.
As lunar rhythms are more "messy" than daily rhythms, more data over more time are necessary for the robust statistical analysis. And, due to ethics creep, it is not certain that either animal or human studies of such scope can be approved and performed any more. So, one has to be creative and get quality information out of imperfect experimental protocols (just like we cannot wait to observe multiple cycles of 17-year cicadas, but have to invent creative, short-term approaches instead).
But this time, the researchers were just lucky! Their data-set came from an old experiment which was designed well enough for this new purpose. The key is they had LOTS of data. Their subjects came in to the sleep lab many times and a number of different parameters were measured. Ideally, each subject would stay in the lab for a few months instead of just four days at a time. But having such a huge data set allowed them to weave together a patchwork of fragmented data into a large, trustworthy whole. Each first night of the test was eliminated from the data due to potential influence of the previous day (and the so-called "weekend effect", as people tend to change sleep times on their days off). Each phase of the moon was covered by multiple subjects multiple times. So they could employ powerful statistics to tease out the effects of the moon phase on various parameters of sleep quality.
And they found some interesting stuff! My colleague Dina Fine Maron has covered the paper in greater detail here. In short, human subjects with no access to information about moon phase, or any ability to perceive the moon itself or its light intensity, nonetheless slept about 20 minutes shorter on the nights of full moon, mostly due to taking roughly 5 minutes longer to fall asleep in the evening than on a night of the new moon. Levels of melatonin, hormone released by the pineal gland during the night, were lower during full moon nights as well. Some of the age and sex differences cannot be explained at this time due to imperfect experimental design - and that is OK. I'd rather see new interesting information coming out of an old data set, than never seeing it at all just because it cannot be "just perfect".
There are many claims around about lunar periodicities in all sorts of human behavior. For some of those, there is no evidence the claims are true. For others, there is strong evidence the claims are not true. But a few subtle effects have been documented. This paper adds another set with persuasive statistics.
Is this a demonstration that there is a working circalunar clock in humans, operating endogenously, and independently from the actual moon? It's not possible to tell yet. Those kinds of demonstrations (just like for circadian clocks) require a battery of tests, starting with documenting multiple cycles (I'd say at least three complete monthly cycles) in complete isolation, ability of artificial moonlight to phase-shift the phase of the rhythm in a predictable manner (consistent with a Phase-Response Curve), and hopefully identification of body structures or cellular components which are devoted to generation of the rhythms, with at least some hint of the mechanism how they do it.
We are far from it yet even in animals we can manipulate in lab and field studies. Much work has been done over the decades in the study of lunar and circalunar rhythms in various animals, mostly aquatic and intertidal ones. There are documented lunar cycles (but not necessarily internal lunar clocks) in a variety of organisms, including sponges, cnidaria, polychaetes, aquatic insects, and many different crustaceans including crayfish.
In the terrestrial realm, antlions possess internal lunar clocks, but many other species show modifications of behavior during different phases of the moon, including honeybees, rattlesnakes, ratsnakes, some rodents, some lizards, and lions.
The gravitational force of the moon is so weak that it can affect only very large bodies of water on the Earth's surface. It cannot even affect smaller lakes and rivers. There is no theoretical mechanism by which any molecule or cellular structure in a human body can be so sensitive as to detect the gravity of the moon. So that hypothesis is out.
In field studies, animals can see and synchronize to the changing night-time intensity as the moon goes through its phases. But in the lab, as in the case of this study, there are no visual clues to the moon phase for the subjects, and, since they had no idea the data would be analyzed for moon phases, they probably did not pay attention to that before they entered the light-isolation lab.
With both gravity and light eliminated as potential clues, the internal clock remains the strongest hypothesis. But it's still a hypothesis that needs to be tested before one can state with any certainty that it is the case.
As for evolutionary explanations for the existence of a putative lunar rhythm of humans? I would be very careful about this. Demonstrating that any trait is actually an adaptation (and not an exaptation or side-effect of development, or something else) is an incredibly difficult task. Just because something seems "obviously useful" does not make it an adaptation. It is an error of hyperadaptationism to pronounce a trait an adaptation just because it exists, and then to tack on a semi-plausible scenario as to how it may have been selected for. Evolutionary biology is much more rigorous than that kind of lazy armchair speculation.
Sure, if our ancestors actually had lunar clocks as adaptations, it is possible that the mechanism for it may still remain, even if in a weak state, in at least some of today's humans. But maybe not. And like a rudimentary organ, it does not seem to have any obviously useful function for humans living in the modern society. Twenty minutes of less sleep, that's all. But it's good to know. So we can find good use to those extra twenty minutes, perhaps come up with new scientific hypotheses over a pint with colleagues at a local pub.
Reference: Cajochen et al., Evidence that the Lunar Cycle Influences Human Sleep, Current Biology 23, 1–4, August 5, 2013, http://dx.doi.org/10.1016/j.cub.2013.06.029
Images: top: by NASA, bottom: from the paper.