This post about the origin, evolution and adaptive function of biological clocks originated as a paper for a class, in 1999 I believe. I reprinted it here in December 2004, as a third part of a four-part post (the fourth part contains all the references). Later, I re-posted it here in 2005 and here in 2009. Thus, some of the information is out of date, the writing is still very 'academic', but main points still stand, I think/hope.

III. Whence Clocks?

Origin, Evolution, and Adaptive Function of Biological Clocks

The old saw about the early bird just goes to show that the worm should have stayed in bed. (Heinlein 1973)

Now darkness falls.

Quail chirps.

What use Hawk eyes?


Local/temporary and global/universal environments.

In the study of adaptive functions, usually the question is asked about a function A in the species X. This implies that the species X has evolved the function A as a response to recently and currently selective local environment a. One then studies the pertinent aspects of the environment a, the function A of X in the field and in the lab, and, perhaps, the similar function A2 of a similar species Y in the similar environment a2.

However, not all adaptive functions are adaptations to local environments. Some aspects of the environment are global. All organisms on Earth are adapted to its gravity. It can be safely assumed that even the first life form on the planet had this adaptation. Some environmental factors are global now but might not have been throughout the history of the planet, like the proportions of oxygen, nitrogen and other gases in the atmosphere. Some adaptations are very frequent although they do not seem to have a correlate in the external environment - for instance sexual reproduction. Such adaptations are result of the particularity of the mechanisms of evolutionary change on Earth, and might not have evolved if the life on Earth initially adopted different mechanisms of evolution. Other adaptations are correlated with an environmental factor which is global for a large number (but not all) of very diverse organisms, e.g., the properties of the marine environment.

Temporal structure of the environment.

Movements of the Earth, the Moon, and the Sun in relation to each other, and the rotation of Earth around its tilted axis, have tangible effects on all organisms inhabiting surface, or near-surface, habitats of the land and oceans of the planet. Only deep-soil Archaea, and the deep-oceanic organisms may conceivably escape from all the astronomically generated rhythms of the Earth's surface: daily cycle of light and darkness, the phases of the moon, the tides, and the seasons of the year.

The common properties of all these cycles are their precision and regularity - they are highly predictable. Organisms cannot evolve adaptations to rare and unpredictable events (e.g., meteor impacts), but can easily adapt to predictable and rapid cyclic events in the environment (Bornemisza 1955).

Why would organisms need to adapt to the natural cycles? In short: because they need to adapt to their environments, and the effect of the cycles is that each organism needs to adapt to more than one environment. For a rabbit, a meadow is a single spatial environment. However, it comprises of a number of temporal environments. The meadow by day and the same meadow at night are two very different worlds. If we multiply that with four seasons (spring, summer, fall, and winter environments) and at least two extreme values of intensities of moonlight at night, we arrive at the conclusion that the rabbit needs to adapt to twelve different environments!

All these cycles have been quite stable throughout the history of Earth. The alteration of night and day affected even the simplest unicellular organisms which were the only inhabitants of the planet for most of its history, as they rely on sunlight as the source of energy for metabolism and replication. As the organisms grew bigger, lived longer, and invaded other environments, longer cycles gained in importance, too. Seasonal changes in daylength are substantial towards the poles, and the annual changes in weather conditions are quite dramatic around equator.

Adaptations to environmental cycles.

What are the possible ways an organism may adapt to temporal changes? Certainly, an organism may use its sensory mechanisms to behaviorally react to the changes as they happen. When the sun comes up in the morning, the animal perceives the light and the warmth, wakes up and goes out to forage (or goes into the burrow to sleep). When nice weather and plentiful food arrive in spring, the animal mates and raises its young. In principle, this arrangement could work, as each species would evolve its own time-table of stimulus-responses to cues.

However, the extremely predictable changes are a form of information, and information is an important resource as it creates opportunities which can be exploited. It can be advantageous to an organism to be able to predict the changes before they occur. If a bird lets its body temperature drop while sleeping at night, and starts warming up at the crack of dawn, it will take some time, perhaps an hour or two, before its temperature is optimally high for foraging and predator avoidance. Another bird which can predict the time of dawn can start raising the temperature while it is still dark outside. At dawn, it is ready to go - the early bird gets the worm! The early bird leaves progeny - the other one goes extinct.

Another avian example: One bird saves the energy by keeping its reproductive system regressed during the winter. When spring brings nice weather, this bird grows its gonads, a process which takes a couple of weeks (longer in mammals). When the reproductive system is ready, provided the potential mates have also reacted the same way, the bird mates, lays eggs and raises hatchlings.

Another bird predicts the coming of spring by measuring the lengthening of the daylength (photoperiod). On the first day of nice weather, the bird is ready to mate. This bird is not wasting the first couple of weeks of bountiful food and warm air. A few weeks later, her chicks are out foraging and outcompeting the younger, smaller chicks of the first bird. The second bird can also expand its range towards the poles, as it can perform the whole reproductive effort within a very short window of opportunity, while the first bird would lay the first egg at the time when the bad weather has already returned.

It is obvious I love birds, so let me indulge in another avian example. As the winter gets closer, but the weather is still nice and the food abundant, a bird measures the shortening of the photoperiod and anticipates that the weather will change. She stops her reproductive effort early, she puts on large fat reserves, shrinks many internal organs, drops her old tattered feathers and grows new, healthy, and not so sexually attractive plumage. She gets together with other conspecifics of the area who were, up till then, fiercely territorial. Together they undergo a period of a flying exercise and fitness program. At a particular day of the year, the whole flock heads south to the distant warm winter feeding grounds. On their way there, the birds use the position of the sun to orient. The position of the sun changes during the day, but as these birds have the sense of time, they can compensate for the sun's movement across the sky, and will not get lost on the way. They will use the same methods to prepare for, accurately time, and precisely retrace the route on their flight back home in spring. Another bird which passively responds to the environment will stay at home until the first frost and will freeze together with her last clutch of newly-hatched chicks. If she attempts to migrate, she will have to do it alone, with no preparation, with insufficient energy stores, and with no idea how to navigate to the target environment and back.

Clocks as adaptations.

The competitive advantages of having an internal anticipatory system have played a role in the evolution of biological clocks (Pittendrigh 1965, 1967, 1993, DeCoursey 1990). Daily biological clocks, or circadian clocks, are endogenous, inherited timing devices which control rhythms of many physiological and behavioral functions. The clock can time changes in a variety of processes at any time of day that is advantageous to the organism, as well as translate the daily changes into rhythms with longer or shorter periods. The clocks, being organic structures, are not absolutely precise, so they have evolved the means by which external timing cues, like dawn and dusk, can entrain the phase of the clock to the local time.

The possession of the internal timing mechanism will also allow the organisms to invade novel ecological niches, e.g., polar regions, intertidal regions, migratory way of life, etc. As the predators and prey evolve biological clocks, the selective pressure on other organisms to do the same increases even more, as the biotic aspects of the environment get even more temporally complex in even more selectively meaningful ways. This may lead to coevolutionary arms-races around the circadian clock, as a predator tries to invade and the prey to escape a particular temporal niche. Unlike other arms-races, this kind will not lead to runaway selection of novel traits, as the clock is a closed loop (they may chase each other in circles forever). However, this may lead to an adaptive trend towards greater sophistication and flexibility of the timing mechanisms, including multi-oscillator systems, clocks modulated by the internal states, circannual clocks, etc.

The function of the clock in adaptation to the temporal changes in external environment is not its only function. Another hypothesis exists which states that one of the important roles of the timing mechanisms is the internal coordination of physiological and metabolic functions. Some cellular and organismal processes are incompatible with each other. If there is no possibility that two processes can be segregated spatially, then their temporal isolation is the only remaining solution. Theories differ in assigning relative importance to the two functions, and in stating which function arose in evolution first (Winfree 1990, pp.387-394, Edmunds 1988, pp.368-370, Pittendrigh 1961,1967, 1981, 1993, Enright 1970, 1975).

Another hypothesis is that the circadian clock is the organ of the sense of time (Daan 1995, Enright 1975). Unlike other sensory modalities, the clock does not respond to any form of energy from the environment. It measures the passage of time internally, and gives the non-directional physical time a direction ("arrow") by virtue of being in a living organism (Frazer 1996). The clock is, thus, an internal representation of time, a cognitive temporal map analogous to the cognitive spatial maps. This allows the organism to perform interval time measurement, distinction between before and after, measurement of duration of events, and learning of the local time of day (Shettleworth 1998, pp.333-378.). Presumably, only this last function requires that the clock, indirectly via other sensory modalities, be entrained to the local time.

Origin of timing mechanisms.

There are three main lines of thought concerning the origin, evolution and adaptive function of biological clocks (Winfree 1990). The first view assumes that in the beginning the organisms were arrhythmic. The cyclic nature of energetic availability and cycles of potentially degrading effects of the sun's ultraviolet rays on particular pigmented enzymes, provided the selective environment. A cell with a timer can predict the changes and adjust its metabolic activities to minimize energetic and material loss. This cell will outcompete the other cells in the Archeozoic sea (Pittendrigh 1967). The emergence of such a system so early in the evolution of life leads to prediction that the molecular mechanisms of circadian rhythmicity will be highly conserved among all organisms (Winfree 1990, p.389).

The second view assumes that the environment itself forces rhythmicity onto the early unicellular organisms (Goodwin 1966). To economize waste, the cell evolves modifier genes. Each of these gene products will have a role in facilitating a smooth transition from one to the next phase of the imposed cycle. As more and more such genes evolve, every state of the cycle comes under genetic control. Three billion years later "the cell might surprise itself to discover...when some scientist first puts it into constant conditions, that it shuffles its way spontaneously through almost the same cycle" (Winfree 1990, p.390). So, even though all the cogs and wheels were, the whole clock itself was never selected for.

The third view states that all biochemical processes are cyclic. Furthermore, this cyclicity is part of the definition of life. Some of the cycles are regular, and the periodicity of such oscillations can be modified by natural selection (Winfree 1990, pp.391-392). Flexibility in counteracting stabilizing homeostatic mechanisms can add another degree of freedom in which to search for optimization. The second and third views expect to see almost as many circadian mechanisms as there are species. If one speculates that the life originated in the shallow tidal pools, than the circatidal rhythms might have been the first to arise, either before or simultaneously with the circadian clocks.

There is nothing incompatible between the three views. They could have conceivably all contributed to the emergence of timing mechanisms. However, neither one of them explains why was it that a self-sustaining clock evolved. Why does a circadian rhythm persist in constant conditions for many days, months and years? The organisms were selected only for ability to cycle in a regularly cycling environment. Why is the clock not an hourglass mechanism which can be turned over by the environmental cues every day? Enright (1970, 1975) suggests that role of the clock in acquisition of temporal memories might have had an influence. Aschoff (cited in Enright 1970) suggested that an oscillator which is capable of a small number of cycles will be adaptive in times of weather catastrophes, during hiding and burrowing, and during hibernation. Such a timer, by virtue of capability for a few oscillations, inevitably can cycle indefinitely.

Comparative method in the study of clocks.

As temporal structure is a property of a global environment, the study of the mechanisms of time measurement will depend on our understanding of the origin, evolution and adaptive function of circadian clocks. Should we expect a mechanism as conserved as glycolysis, or use of ATP for energy, or mechanism of DNA transcription? Or should we expect that every narrow group of organisms has evolved a different mechanism? Or something in between? And how do we study the clocks in order to distinguish between phylogenetically conserved generalities and the ecologically determined novel particularities?

Comparative method is a "range of techniques that infer how one organism evolved by comparing what evolution produced in that case with what it produced in other cases...The comparative method... has three important applications... First, it enables us to directly test the historical assumptions tacit in adaptationist hypotheses. Second, it enables us to test the proposed link between environmental feature and adapted trait. Third, we can use it to make sense of the adaptationist claim about the explanatory priority of selection" (Sterelny and Griffiths 1999, pp.243-244).

An adaptationist hypothesis (e.g., about the adaptive function of the circadian clock) can be tested by checking the actual sequence of evolutionary changes to see if it is the one presumed by the hypothesis. A hypothesis may propose that a particular trait arose before another trait (e.g, hourglass before clock, circadian before circatidal rhythm, single before multi-oscillator systems, etc.). The hypothesis can be tested by comparing the traits in earlier and later representatives of a lineage (in the case of clocks as responses to a general feature of the environment - the entire tree of life).

Comparative method can also directly test the adaptationist hypothesis that adapted traits are responses to particular features of the environment by measuring if there is a tight correlation between a particular environment and a particular trait (e.g., are the deep-oceanic creatures, or cave and subterranean animals arrhythmic; is there a correlation between degrees of robustness of the circadian amplitude and the stationary vs. migratory habit, or burrowing vs. open field habitat).

Adaptationist extremist position is that convergent evolution illustrates the overwhelming power of natural selection over all other evolutionary mechanisms (genetic drift, sexual selection, etc.). But, convergence tells us nothing about the relative powers of selection and history unless we can somehow count all the possible convergences that have not happened - all the times history "won". Is a presence of a trait in a number of organisms the result of convergent evolution? This can be answered only by extensive use of comparative method. One cannot know if something is a convergence without a phylogenetic tree. If analysis of the tree reveals that the trait appeared repeatedly de novo, than it is a convergence and the force of natural selection is a primary explanation. If analysis of the tree reveals that all members of a lineage posses the trait, than it is not a convergence. The trait was probably adaptive for the common ancestor, but it might not be adaptive at all today. It could still be an adaptation as it may serve many different adaptive (exaptive) functions in different species and exist merely as phylogenetic inertia in others.

The adaptive hypothesis requires that the clock be phase-shiftable by an environmental cue and that it be temperature compensated. These are the two absolute requirements for its proper function. Every timing system will satisfy these minimal requirements no matter what the underlying mechanism may be. So, these two properties cannot be informative about the mechanisms, and if "we are to test the proposition of a common mechanism and use only formal properties, these must be of such a nature that selection can reasonably be dismissed as the agent responsible for their universal association with cellular clock systems; in short, the properties must lack adaptive value" (Pittendrigh and Bruce 1959) and the "study of features which do not seem to have any survival value...may be indicative of the structural properties of the system" (Pavlidis 1971).

So, is the circadian clock a universal adaptation, a case of convergent evolution, or a generatively entrenched relic of history? Here, one needs to make a sharp distinction between the circadian clock which is the timekeeping molecular mechanism in a single cell, and the circadian system which includes all the cellular clocks, all the sensory input pathways to the clocks, and all the mechanisms of translating clock states into overt rhythms of a whole organism (Underwood et al. 1997a). A clock is a molecular mechanism, while a system uses the molecular mechanisms to produce overt rhythms.

Circadian systems.

Data so far indicate that all the biological rhythms studied to date possess the basic properties of entrainability and temperature compensation. I have collected literature on observations and experiments in hundreds if not thousands of species of unicells, plants, fungi and animals, but the limitations of time and space prevent me from presenting them all in this abstract, so I will have to leave them for a later publication. All timing systems can be brought to match the local time, and will do so in a precise manner over a range of temperatures (if they could not they would be thermometers, not clocks). Temperature compensation varies with latitude in fruit flies (Sawyer et al. 1997). Critical photoperiod for induction of reproduction in mammals and birds, as well as the number of eggs in a clutch in birds, both traits regulated by the circadian system, correlates with latitude in many species (Baker 1937, 1938a,b., Daan and Aschoff 1975, Bronson and Heideman 1994). This indicates that the circadian clock (and all other circa-rhythms) is an adaptation, as the trait tightly correlates with the environment.

It is interesting to note that many organisms which live in caves and under ground still have circadian rhythms. For instance, Mediterranean blind mole-rats have circadian rhythms entrainable by light (David-Gray et al. 1998), as do the natural or induced disperser phenotypes of the African naked mole-rats (Goldman 1999, contrary to Davis-Walton and Sherman 1994). Cave-dwelling crickets have circadian rhythms which are nor entrainable by light, but by noise/silence cycles (Marques and Hoenen 1999).

Cave-dwelling catfish are apparently arrhythmic, but a rhythm emerges if the fish are exposed to a light-dark cycle (Trajano and Menna-Barreto 1999). Overt behavioral rhythmicity is dependent on intensities of light and levels of temperature in Japanese Newts (Nagai and Oishi 1999) and Leopard Geckos (Chris Steele, unpublished data). These data suggest that circadian systems in these organisms are "rudimentary organs" which were not entirely lost during the course of evolution from their surface-dwelling ancestors. Alternatively, one can argue that the clocks persist in these organisms due to their utility in orchestrating myriads of physiological events within the body, and that the persistence of a behavioral output is just a left-over from the past.

Some emergent properties of circadian systems are lost only to reappear again and again when the need for them arises, presumably because it is relatively easy to re-invent them since all the clock parts are already present. For instance tidal rhythms have been found in birds and lizards (Daan and Koene 1981, Sawara et al. 1990, Wikelski and Hau 1995). Lunar cycles appear in some terrestrial species of plants, invertebrates, mammals, birds and reptiles (Abrami 1972, Skutelsky 1996, Brigham et al. 1999, Chris Steele, unpublished data).

Circadian rhythmicity does not disappear in artificially imposed relaxed selective environments. After 600 generations in constant conditions, fruit flies retain precise circadian rhythmicity in behavior and pupal eclosion (Sheeba et al. 1999) suggesting that rhythmicity is important in regulating cyclic biochemical processes within an individual fly, as much as in proper phasing to the environmental cycles. Similar conclusion can be drawn from the fact that it is easy, by experimental manipulation or by genetic selection, to eliminate behavioral but not physiological circadian rhythms in quail (Guyomarc'h' et al. 1998, Zivkovic et al. 1999).

Interestingly, it took a long time to discover circadian rhythms in the laboratory Nematode C.elegans. In this animal, the so-called "clock-genes" are present, the ortholog of PER cycles with an ultradian rhythm with a period of about 6 hours and is thought to regulate timing of developmental events (Jeon et al. 1999). The circadian rhythmicity in this animal was not completely eliminated through many generations of laboratory rearing practices, yet the involvement of circadian clock genes in timing of development is apparently compatible with their use for circadian behavioral timing.

Archaea and most Eubacteria are arrhythmic. For Archaea, the explanation might lie in their ecology - they usually inhabit the most inhospitable regions of the planet including deep-oceanic hydrothermal vents, deep soil, rocks, salt deposits, polar regions, underground hot-water springs, where the rhythms of the planetary surface might not have any direct effect on their survival. Some Cyanobacteria have circadian rhythms as they need to temporaly separate incompatible reactions of photosynthesis and nitrogen fixation (Johnson and Golden 1999). Real Bacteria, being Prokaryotes, might just be capable of a rapid response to the direct environmental fluxes, and a timer might impose too rigid a control in, on their scale, essentially unpredictable environments.

Cellular clocks.

Molecular clocks are invisible to selection. Circadian systems, being in direct interaction with the environment, are more likely to show correlations with the environment and a faster rate of change over evolutionary time scales. It is perfectly safe to state that all mammalian systems are built basically the same way, which differs from the avian, reptilian, insect or plant systems. Some data also exists suggestive of correlations between some formal circadian properties and specific ecological niches.

All the circadian systems contain, as their parts, cellular circadian clocks. Nothing a priori dictates that the cellular clocks need to have, or not have, exactly the same molecular mechanisms of rhythmicity. It is conceivable that a cellular clock can be just a rudimentary timer, even an hourglass, and that the multicellular organization of the system provides temperature compensation, and entrainment properties. Also, even if the dynamics of the mechanism are the same, it is not necessary that the actual molecular players remain the same.

What do the actual molecular data show? Each circadian clock cell in all organisms is a self-sustaining circadian oscillator, resettable directly or indirectly by light, and it is temperature compensated (reviewed by Dunlap 1999). The dynamics of the mechanism are very similar in cyanobacteria, plants, fungi, insects and mammals. The molecular clock is a negative feedback transcription-translation-based oscillator in all of the studied organisms. Fungi, insects and mammals use PAS-domain-containing proteins as positive elements. Insects and mammals have pairs or families of negative elements. Several genes can be found both in the fruit fly and in the mouse., e.g., per, tim, clk and cyc in flies have homologs in per1, per2, per3, tim, clock and bmal in mice. Drosophila dbt is a sequence homolog to casein kinase 1.

These are the similarities. How about differences? All negative elements cycle (in expression) in fungi and insects, but only some in mouse. No positive element cycles in fungi, one in the fruit fly and a different one in the mouse. The phases at which elements act are shifted - fungi and mammals have a day-phase clock, while the fruit fly has a night-phase clock - a complete change in internal logic of resetting by light.

So it seems that nature is using some well-conserved themes but is mixing and matching the details. Is there a consensus today about the evolution of circadian function on Earth? Past five years have seen an avalanche of new molecular data. Every few weeks a paper is published, it seems, which leads to a change in one's opinion on the evolutionary conservatism of the clock. A surprising finding of a similar gene or function in two very distinct organisms points to overall great similarity of mechanisms. Next paper discovers a seemingly unexplainable difference in mechanisms between relatively closely related organisms, and suggests that each species has freedom to tinker with the basic clock mechanism. The field as a whole is oscillating on this issue. If there is anything like a consensus, it is the idea that clock is always running and is full of redundant elements. As long as a wheel or a cog can be removed or replaced without stopping the clock, the overall mechanism will stay the same, although the individual players may, one by one, all be replaced over long periods of evolutionary time.

However, data exist which challenge the notion that transcription/translation loops of clock genes are mechanisms of all aspects of circadian rhythmicity (Hall 1996, Lakin-Thomas and Johnson 1999, Lakin-Thomas 2000). In an alga, Acetabularia, circadian rhythmicity persists in cells from which the nucleus was removed and the transcription of non-nuclear DNA was pharmacologicaly blocked. However, this rhythmicity damps out after several cycles, and re-introduction of the nucleus restarts the rhythm with the phase dictated by the nucleus (Edmunds 1988). These data suggest that the circadian clock in this organism consists of two parts - a nuclear (transcription/translation loop) and a cytoplasmic clock. These two clocks interact with each other. A computer model was developed (Roenneberg and Merrow 1998, 1999) which shows how such a feedback system of two oscillators can account for stability of period, persistence of rhythm in prolonged constant conditions, temperature compensation, resistance to environmental noise, and entrainment properties of the cellular clock.

There are also data suggestive that some of the circadian rhythms are and others are not controlled by the "clock genes". Null mutations, deletions and continuous overexpression of per gene in fruitflies does not entirely abolish circadian rhythmicity. The behavioral activity rhythms of fruitflies are characterized by two peaks. Evening peak is assumed to be the time when the insects are foraging. The morning peak is the time when the insects are involved in reproductive activities: search, courtship and mating.

Elimination of cycling of PER protein eliminates the evening peak. However, the morning peak persists for several more cycles, then the period gets shorter (ultradian), and finally dissolves into arrhythmicity (Renn et al. 1999, Helfrich-Forster 2000, Gvakharia et al. 2000, Kaneko et al. 2000).

In the giant silkmoth, PER protein never enters the nucleus in the pacemaker cells responsible for overt behavioral rhythmicity. However, it does enter the nucleus in some peripheral tissues (Sauman and Reppert 1996), suggesting that a transcription/translation loop is part of some but not all cellular circadian clocks.

Two separate circadian systems can be observed in female birds of some species. One system, driving rhythms such as locomotor activity, feeding activity, body temperature, oxygen consumption and melatonin release is sensitive to light and melatonin and can be readily entrained to light-dark cycles. The other circadian system drives all the mentioned overt rhythms, but also the circadian rhythm of ovulation and oviposition. This rhythm is not directly entrained by light cycles. It tends to free-run through a portion of the day. As the two systems interact, and as they show two different periods (24h for the entrained system and some other value for the other system), the two rhythms continuously change mutual phase relationships. Some phase relationships are "permissive" and others are "forbidden". As the ovulatory rhythm intrudes into the "forbidden zone", its interaction with the other system makes it phase-jump back to the starting (default) phase from which it starts free-running again. Interaction between two clocks may be neural (e.g., between two hypothalamic pairs of nuclei) or hormonal (perhaps sex steroid hormones). The main system consists of the pineal gland, the retinae, and a pair of hypothalamic nuclei. The brain clocks are not self-sustained oscillators - they need a daily input from either the eyes or the pineal, depending on the species. It is not known if the retinal/pineal clocks are self-sustained oscillators in the long term. Their presence is necessary for circadian rhythmicity, but a feedback signal from the hypothalamus to these structures might be necessary, too. Data so far cannot distinguish between the hypothesis that cellular clocks in the retina or pineal are autonomous clocks which entrain other parts of the system, or that a feedback loop between damped oscillators in the center and the periphery are necessary for sustained circadian output. Likewise, the ovulatory circadian rhythm may involve an interaction between cellular circadian clocks in the brain and in the ovary, or a hormonal feedback loop which does not even utilize any clock-genes and in which no single cell acts as a clock.

The idea that circadian rhythm, or similar mechanisms are involved in other aspects of biological timing is strong and some links have been made, particularly, with the timing of developmental events. Mutations of the per gene affect the developmental time in fruitflies (Kyriacou et al. 1990). The so-called heterochronic genes in C. elegans turned out to be orthologs of per and tim (Jeon et al.1999), in an organism with no circadian rhythms. However, a 90 minute cycle of chairy, which times the formation of somites in the chick embryo does not involve protein synthesis (Palmeirim and Pourquie, cited in Pennisi 1997).

To summarize, the current view of circadian rhythm generating mechanism is that it is a transcription/translation feedback loop between clock-genes and their protein products. However, the challenges to this view are coming from within the field. Some circadian rhythms are probably generated by cellular clocks consisting of two interacting loops, one being nuclear, the other cytoplasmic. As these clock-cells are organized in tissues, the communication between individual cells in that tissue confers novel properties to the circadian system of which individual cells are not capable. As most organisms, even unicells, posses more than one circadian clock, interactions between different clock-tissues allows for evolution of more complex temporal phenomena, including photoperiodic time measurement, lunar, tidal and circannual rhythms, and continuously consulted clocks. Some circadian systems are perhaps not even utilizing the individual clock-cells, but may be properties of neuroendocrine feedback loops between two or more clock or non-clock tissues (Cassone and Menaker 1984).

It is difficult to glean from the published data if this is so, but one can hypothesize that the individual variation is far greater in the properties of the whole circadian systems than in the properties of molecular circadian clocks. Period of circadian rhythmicity of the individual cells might be very similar between individuals of the same population, but the periods of the overt rhythms vary greatly, as do the emergent properties, e.g., photoperiodic sensitivity, or the critical photoperiod (e.g., Heideman and Bronson 1991, Heideman et al. 2000) due to differences in development of the whole circadian systems and hormonal influences on their properties. If this hypothesis is true, than Natural Selection will have to operate on the mechanisms of integration of the whole systems to a greater degree than on the sequences of clock-genes themselves. If cellular clocks are viewed as Lego blocks, and the circadian systems as the structures built out of these blocks, then it becomes apparent that most evolutionary changes will be in the way the blocks are put together, rather than in the shapes, sizes and colors of the blocks themselves.

Inferring Rhythmic Behavior and Physiology from Vertebrate Fossils .

I recalled what he had said of the pineal gland, and wondered what he saw with this preternatural eye. (H.P.Lovecraft, From Beyond)

From the perspective of a paleontologist, a question arises if one can infer the type of rhythmicity displayed by an animal from the fossilized remains of its bones. One asks if the fossils can tell us if particular animals were diurnal or nocturnal, and if their reproduction and migration were seasonal.

Before one starts to hypothesize about dinosaurian sleeping habits, a study of living vertebrates is in order. In an oversimplified (and Eurocentric) scheme, most of the living mammals are nocturnal, birds are diurnal, an reptiles are bimodal (dawn and dusk). But a closer look at the exceptions might be more revealing. Reptiles in the tropics are more likely to develop nocturnal habit, while at higher latitudes and/altitudes diurnality might be more common. Many diurnal birds temporarily switch to nocturnality during migration, when the production of energy favors the cool of the night. In mammals, it is mostly the small animals that are nocturnal.

Large mammals of the polar regions, like seals and walruses are diurnal, while large mammals of the Serengeti are usually bimodal.

If diurnality is taken as the default condition, the question to ask is: what adaptation to nocturnal life can be seen in the skeleton of an animal. Adaptations to nighttime activity involve adaptations to cold and adaptations to dark.

If we assume that dinosaurs were ectothermic, virtually all of them should have been nocturnal because of their large size. A large body mass dissipates heat more slowly, and is more likely to overheat. Endotherms produce their own heat but are also thought to possess more sophisticated methods of getting rid of surplus thermal energy, so if some of the dinosaurs were endotherms, they might as well have been diurnal. Also, the largest tropical mammals, like elephants, keep their core body temperature much lower than the mammalian average (20-30 C, as opposed to 37). No dinosaur was small enough to exhibit daily torpor.

Adaptations to dark involve sensory systems. There are three possible ways to adapt to life in the dark:

- evolving very sensitive eyes,

- evolving very sensitive hearing and smell, perhaps even echolocation, at the expense of largely degenerated vision, and

- evolving both very sensitive vision and other sensory modalities.

Of the three cases, the first and the third are repeatedly observed in nocturnal vertebrates. Think of owls, lemurs and cats. The second solution is employed by animals that live in caves, hollow trees or in underground tunnels. Dinosaurs were too large for this kind of fossorial habitat.

In all of these senses, increase of acuity is likely to be associated with the relative increase of size of that organ. In a fossil skeleton, enlarged ears are unlikely to be identifiable. Also, importance of the auditory acuity might be small in a world where the movements of predators and prey alike produce minor earthquakes. Detection of the enemy at a large distance might be more important, so emitting and hearing infrasound (like elephants and giraffes) could have evolved in Dinosaurs. A strong sense of smell will result in enlarged olfactory bulbs which will leave their imprints on the interior of the skull. Large eyes require large sockets. Binocular vision would also translate into a more sophisticated central processing mechanisms of visual information. Large orbits with a big overlap of visual fields, as well as large olfactory bulbs were found in the skulls of T.rex. Most bird-like dinosaurs (eg.Troodon) also had disproportionately large eyes. Were they nocturnal? Who did they hunt at night? Was their prey sleeping at the time? Where would a Brachiosaur hide for an overnight sleep? Was one guardian always awake? A careful analysis of the ecology of a certain area at a certain time might shine some light on these questions, since both temporal niches are likely to have been filled.

As for annual rhythms, they might be preserved as growth rings in the bones, reflecting the seasonal changes in food or mineral availability, UV-radiation, disease or activity (Hermann and Danielmeyer 1994). Timing of seasonal reproduction and migration is likely to have been determined by photoperiodic response. Unfortunately, living vertebrates have variable anatomical organization of the photoperiodic system. While the pineal gland and its hormone melatonin seem to be necessary in mammals, in other vertebrates pineal, parapineal organ, eyes, hypothalamus and gonads may or may not be involved. In vertebrate evolution, however, there is a trend for the pineal and parapineal organs to gradually lose a photoreceptive function and gain a secretory role.

In fish, amphibians and reptiles, pineal is a fully developed photoreceptive organ. In birds it is still responsive to light, but the histology suggests a neuroendocrine organ instead of a photoreceptor. Mammalian pineal has completely lost ability to sense light.

Parapineal organ is a highly developed "third eye" with a lens in most fish, amphibians and some reptiles (lizards, Sphenodon). In other reptiles, as well as all birds and mammals, parapineal is gone. Even pineal itself is missing in some organisms like crocodile, dugong, armadillo, anteater, sloth and pangolin, or is very small as in elephants, rhinos, manatees, whales, monotremes, marsupials and bats. The existence and size of the pineal gland and/or parapineal organ seems to correlate well with the climate and latitude.

Global species, like horses, cows, sheep and rabbits have large pineals. Polar animals, like seals, sea lions and walruses tend to have the largest pineals. This fact, together with the hypothesis that the pineal is somehow involved in thermoregulation, prompted Roth and Roth (1980) to argue that dinosaurs were cold-blooded and restricted to tropical climates, since most Dinosaur skulls do not show evidence of the presence of the pineal-parapineal complex. In all vertebrates except primates, pineal gland leaves an imprint on the inside of the skull on the roof of the brain. Parapineal lies in a hole in the skull called the pineal foramen. Only in Diplodocids there was evidence for the existence of the parapineal organ, and only in some bird-like Dinosaurs for the pineal itself. In any case, existence of parapineal foramina is easily perceived in fossil skulls, and it is the study of living organisms' roles of these organs that needs to be correlated with the daily and seasonal rhythms, as well with thermoregulation.

The circadian clock, of which pineal is a part, is also involved in orientation by sun compass, night-sky orientation and magnetoreception. These mechanisms, although useful for all animals, are particularly necessary for long distance migrants, as some Dinosaurs were shown to be. A wide variety of animals has deposits of magnetic ferri-oxide thought to be involved in magnetoreception (Wiltschko and Wiltschko 1995). In vertebrates these deposits are usually concentrated within the ethmoid bone where they are innervated by the ophthalmic branch of the trigeminal nerve. I t is likely that Dinosaurs possessed these deposits, and it would be very interesting to compare the species with respect to quantity, location, particle-size and orientation of biomagnetite. Unfortunately, study of this phenomenon in living animals is still in its pioneering stages, so any theorizing concerning Dinosaurs would still be pure speculation.

In summary, analysis of vertebrate fossil skulls in light of relative sizes of sensory organs including eyes, olfactory bulbs, pineal and parapineal organs and magnetite particles, can give us insights into spatio-temporal physiology, ecology and behavior of Dinosaurs and other extinct vertebrates.

Tests of Adaptive Function.

As there are two main hypotheses regarding adaptive function of biological rhythms, there are also two lines of research involved in testing these hypotheses. The hypothesis that the primary role of a circadian clock is to orchestrate physiological and biochemical functions and events within a body has usually been tested by looking at longevity of either normal or clock-altered organisms placed in unnatural light cycles. For instance, fruitflies kept in different periods of light-dark cycles, as well as various clock-mutants have been reported to have different lengths of lifespans (Pittendrigh and Minis 1972). Similar findings were obtained form a study of hamsters (Hurd and Ralph 1998). The fact that breeding 600 generations in constant darkness did not lead to the loss of any circadian function in fruitflies adds another argument in support of this hypothesis (Sheeba et al. 1999).

The hypothesis that behavioral adaptations to timing of environmental events, including the temporal organization of natural enemies, has also been recently addressed. In cyanobacteria, colonies of various period mutants were subjected to competition assays in various lengths of light-dark cycles (Ouyang et al. 1998). Invariably, the strain whose endogenous period more closely matched that of the environmental cycle won the competition.

In rodents, lesions of the suprachiasmatic nucleus renders these animal arrhythmic even in light-dark cycles. In field conditions, lesioned individuals were more susceptible to predation than their intact conspecifics in the same area. This effect was greater when the predator pressure was greater (DeCoursey et al. 1997, 2000). These data suggest that correct timing of activity is crucial for survival of prey species. However, these studies do not address the question if, in absence of the clock, organisms could have evolved alternative mechanisms of temporal control.

In my opinion, the best way to study adaptive function of circadian clocks is to study coevolution of temporal parameters in two or more species of natural enemies. For instance, one can study circadian rhythmicity in a venomous snake, e.g., rattlesnake and its venom-resistant prey, the ground squirrel. In areas of sympatry, at what times of day are the two species active, and does that differ from the populations in allopatry? Is there a circadian rhythm in snake's venom production, venom composition and toxicity, amount of venom injected in a bite, inclination to bite in response to a stimulus? Likewise, is there a circadian rhythm of resistance to venom and in evasive behaviors in the squirrels? Are those rhythms different in allopatry? Who is "winning": the predator or the prey?

The same kind of questions can be asked in other systems, e.g., resistant predator and poisonous prey (garter snake and tiger salamander); resistant predator and venomous prey (desert mouse and scorpion); venomous defender and non-resistant intruder (guard honeybee and Death's Head Sphinx Moth); non-resistant host and venomous parasitoid (American cockroach and it's wasp prey), etc. Collectively, results of these experiments would give insight into the dynamics of evolution of circadian clock properties in their adaptive contexts.

A more complex system, one involving more than two species, might be more difficult to study, but would be even more informative. For instance, there is a case (Sara Oppenheim, personal communication) where a single specialist wasp parasitoid (Cardiochiles nigriceps) lays its eggs in larvae of two closely related moth species Heliothis virescens (HV) and Heliothis subflexa (HS). The two moths inhabit two very different plant hosts: HV feeds on tobacco, tomato and cotton, HS only on a Physallis plant. The host plants synthesize their deterrent chemicals (e.g., nicotine) evenly throughout the day. However, they show a circadian rhythm in synthesis of volatile compounds induced by the insect herbivory. These volatiles attract the wasps to the plant which carries the moth larvae. Plants emit different quantities and compositions of volatiles dependent on the species of moth which is attacking them. Wasps, in turn, are active bi-modally during the day - morning and evening only. HV lives on the surface of the leaf and is very vulnerable to the attacks of the parasitoid. On the other hand, HS quickly bores its way into the seed lantern of the plant where the wasp cannot reach it. A tobacco plant has enough leaf surface to accommodate a very large number of HV, but the Physallis plant has only a limited number of lanterns available at any time. It is not known, but it would be very interesting to discover, if the two moth species have evolved different temporal aspects of their biology. For instance, do HV larvae all emerge simultaneously in large numbers and find safety in numbers, while HS emerge one at a time as each individual needs time to search for and find a lantern? At what time of day do the larvae emerge, climb, feed, drop on the ground and pupate? Is there a difference in the sensitivity to light in two species as the HS hidden in a lantern is more protected from environmental light than the HV which sits out in the open? If there are differences in clock properties do they also translate into differences in developmental timing, frequency of courtship song, photoperiodic time-measurement and the ability of one of the species to invade higher latitudes? How are the plants, the wasp, and the two moths tracking each other in time - who is "winning" the race around the circadian clock? Can the changes in circadian properties be considered a part of the mechanism by which the two moth species speciated from each other in relatively recent past? Answers to these and similar questions would provide great insights into evolution and adaptive function of circadian systems.

Effects of clocks on evolution.

We have so far seen that there is individual variation in timing mechanisms (Aschoff 1998); that this variation is heritable; that timing is an important adaptive function, hence that clocks are products of evolution by natural selection. Let us now see if the possession of a clock can, in turn, direct or restrict the possible paths of further evolutionary change.

As noted above, it seems that biological clock mechanisms, perhaps even the same as circadian ones, are involved in timing of developmental events (Saunders 1972, Kyriacou et al. 1990, Moore et al. 1998). From this one can predict that changes in circadian clock properties will also have an effect on embryonic development, and vice versa. Thus adaptive changes in behavioral timing may be constrained by the needs of normal development, and heterochronic mechanisms of evolution might effect a change of fitness of the adults due to the change in physiological or behavioral synchronization.

In insects, developmental events are controlled by hormonal communication involving prothoracic glands and gonads (Nijhout 1994, 1999). Prothoracic glands are sites of circadian clocks in some insects (Vafopolou and Steel 1991, 1998). In some systems, a wasp parasitoid injects a virus into its moth victim (egg or larva) resulting in gonadectomy. The result is a change of timing of developmental events and disruption of pupation in the end. Is it possible that the lack of gonadal hormones broke one part of the feedback loop of the timer responsible for development?

As we have noted several times so far, circadian clocks are also involved in time measurement at other temporal scales from hours to years. For instance, mutations in fruitfly per gene change not just the period of circadian rhythms of behavior but also ultradian frequencies of the courtship songs (Dowse et al. 1987, Kyriacou and Hall 1980, Kyriacou et al. 1993). Likewise, sensitivity to olfactory cues is under circadian control (Krishnan et al. 1999). Although these mutations might not alter the timing of courtship activity during the day (this morning peak of sexual activity does not seem to be controlled by the PER-based mechanism), the changes in song frequency, receptivity to the song frequency, and timing of sensitivity to pheromones might greatly reduce the chance of a wild type and a mutant actually recognizing each other as mates and mating in a mixed population setting, leading to sympatric speciation via behavioral and temporal isolation due to a single nucleotide change.

On the other end of the spectrum, it is possible to construct a model in which a two-oscillator system measuring photoperiod (Warman and Lewis 1997) and another three-oscillator system controlling a free-running circannual rhythm interact to control precise timing of emergence of 17- and 13-year periodic cicadas (Williams and Simon 1995).

Likewise, many changes which various animals underwent under domestication, e.g., albinism, temperament, loss of seasonality, change in developmental time and rate of maturation, etc., point to changes in the clock systems under artificial selection (e.g., Trut 1999).

Can we go further? Circadian clocks are ubiquitous in all multicellular and many unicellular organisms. They provide a whole range of adaptive mechanisms involving development, physiology and behavior. Is it possible to have life without clocks? In my opinion, and data are needed to disprove it, without multi-oscillator circadian systems there would be no possibility for Life to invade oceanic surface, intertidal and terrestrial environments. Some components may be secondarily lost, but the process of invasion necessitated the presence of functioning circadian, lunar, tidal and photoperiodic clocks. Back in depths of the ocean, some kind of clock, not necessarily circadian, would be necessary for internal synchronization of complex organisms, and without such a timer, there would be no complex Metazoa on Earth. This leaves the bottom-dwelling microorganisms as the only organisms which do not need a clock, but will still exhibit some kinds of biochemical cycles, as life itself is defined by cycles (Bonner 1993, Kauffman 1993, 1995, Goodwin 1963, Maynard-Smith and Szathmary 1999).

If we go to another planet and find something that might be alive, how can we decide if it really is. Does it take energy from the environment, stores it, cycles it, uses it and dissipates it? Does it grow and reproduce? Does the offspring resemble parents? Is there a diversity of organisms on the planet? One more question: Is its environment periodic and does it have innate cycles corresponding in period to the environmental cycle? If yes, it is alive, as it possesses an universal adaptation which non-life cannot acquire or evolve. Endogenous rhythmicity is a diagnostic property of Life, more so than the underlying chemistry, as even living forms which do not use DNA as hereditary material, or are even not carbon-based, will still cycle in sync with the star and the moon of their native planets.


Circadian clock without DNA–History and the power of metaphor

Basics: Biological Clock

The Clock Metaphor

The New Meanings of How and Why in Biology?

Some hypotheses about a possible connection between malaria and jet-lag

Evolutionary Medicine: Does reindeer have a circadian stop-watch instead of a clock?

The Mighty Ant-Lion

Are Zombies nocturnal?

City Of Light: Insomniac Urban Animals

Me and the copperheads–or why we still don’t know if snakes secrete melatonin at night

Diversity of insect circadian clocks – the story of the Monarch butterfly

Biological Clocks in Protista

Do sponges have circadian clocks?

Daily Rhythms in Cnidaria

Carolus Linnaeus’s Floral Clocks

Clock Classics: It All Started with the Plants

Chestnut Tree Circadian Clock Stops In Winter

Flirting under Moonlight on a Hot Summer Night, or, The Secret Night-Life of Fruitflies

Too Hard for Science? Centuries to Solve the Secrets of Cicadas

Circadian Clocks in Microorganisms

Clocks in Bacteria I: Synechococcus elongatus

Clocks in Bacteria II: Adaptive Function of Clocks in Cyanobacteria

Clocks in Bacteria III: Evolution of Clocks in Cyanobacteria

Clocks in Bacteria IV: Clocks in other bacteria

Clocks in Bacteria V: How about E.coli?