Early birds, late owls, and the ecological role of intra-population chronotype variation

While the molecular mechanisms underlying variation in chronotypes within populations have been studied extensively, the ultimate selective forces governing it are poorly understood. The proximate cause is variation in clock genes and protein expression, which produces variation in tau (period length of the circadian clock), with early individuals having shorter tau. We studied within-population variation in foraging activity times of two Acomys species in the field. This variation manifested in a regular and consistent sequence of individual foraging activity that is positively and strongly correlated with variation in tau. Thus, variation in circadian clock period length (tau) appears to be the mechanism underlying the regular pattern of intraspecific temporal partitioning. Late chronotypes also spent more time torpid than earlier ones, suggesting an energetic cost to this strategy and possible tradeoffs. We suggest that variation in tau is an adaptive mechanism to reduce competition between individuals within a population.

The diel activity patterns of animals have attracted significant scientific interest focusing on their 2 evolution, ecological significance, regulating mechanisms, and impacts of anthropogenic activities 3 (1-5). The evolution of diel rhythms is driven by the 24 h light-dark cycle and its derivatives, 4 including both environmental conditions and ecological interactions. In the past two decades a 5 growing volume of research has addressed interspecific niche partitioning along the diel axis, with a 6 multidisciplinary approach that involves community ecology, evolutionary biology, and biological 7 rhythms research (2, 4). This research ranged from macroevolutionary patterns to studies of 8 behavioral plasticity in response to human activity and artificial illumination and their bearing on 9 species coexistence (e.g., (5,6)). 10 Concurrently biological rhythms research has developed dramatically and while a significant part of 11 it focuses on humans and model organisms, the past decade has seen a growing literature on animal 12 biological rhythms. The 24 h light and dark cycle is a strong and predictable environmental pattern; 13 daily rhythms in behavior and physiology are governed by a biological clock that is synchronized to 14 this universal cue. When biological clocks synchronize or entrain to the environment (usually to the 15 light/dark cycles), they not only show the same period as the zeitgeber (cue) cycle (24 h), but also 16 establish a stable phase relationship with the zeitgeber, called the phase of entrainment (7, 8). 17 In spite of millions of years of natural selection to a strong universal cue, a significant variation in 18 the period length of the internal biological clock occurs, resulting from variation in clock genes and 19 protein expression (9, 10); consequently different individuals synchronize differently to the 20 light/dark cycle, resulting in variability in their phase of entrainment relative to the environmental 21 cue or zeitgeber (e.g., sunrise), called chronotypes . Chronotype describes the manifestation of 22 daily rhythms, which are the outcome of an interaction between the endogenous circadian clock and 23 environmental conditions (7, 11). They correlate with the free-running period of the circadian clock 24 (the endogenous circadian clock period under constant environmental conditions, called tau), which 25 has a significant genetic component, with early chronotypes displaying a shorter tau than late 26 chronotypes (9, [12][13][14]. Here we investigate the role that this genetic and phenotypic variation may 27 3 have in the coexistence of individuals within a population and hence the selective forces that may 1 be at play maintaining this variation.

2
Individual variation in daily rhythms under free-living conditions has been the focus of intensive 3 research in humans during the last decades, with earlier individuals described as larks and late 4 individuals as owls (7, 9, 13). Variability in chronotypes was also described in different animal 5 species, including rodents, birds, fishes, insects and others, but these experiments were conducted 6 almost exclusively under controlled laboratory conditions; very few studies attempted to address 7 variation of chronotypes in natural populations of species other than humans. Even fewer studies 8 (14-16) focused on the selective forces driving this variation which are yet to be understood (8).

9
Several studies in recent years have demonstrated intraspecific patterns of temporal partitioning that 10 implicated intraspecific competition (17,18). We studied intraspecific variation in chronotypes in 11 two coexisting wild rodent species, its adaptive significance and proximate underlying mechanisms, 12 and the selective forces driving it.

13
In an earlier study we found regular intraspecific patterns of foraging activity among golden spiny 14 mice (Acomys russatus), with specific individuals tending to arrive and forage early (early 15 chronotype) and others to arrive and forage later (late chronotype) during the day (19). Persistence 16 of intraspecific variation in chronotypes in natural populations suggests that a specific chronotype 17 entails benefits or costs which depend on environmental conditions, and have fitness consequences.

18
Under our field experimental setup, where food was supplemented and replenished every morning, 19 we found that early individuals spent less time torpid (controlled reduction of body temperature (20,20 21)) than late individuals (19). As use of torpor was related to a decrease in reproductive success in 21 this species (22), it appears that early chronotypes had a selective advantage in this study system.

22
Intra-population morphological variation has been studied empirically with a theoretical basis that 23 related it to intra-specific competition, with variation related to resource breadth and availability 24 (23). In the current study we hypothesized that in a similar manner, foraging order is determined by 25 chronotype, which in turn is determined by tau, and that intra-population variation in tau, as a 26 proximate mechanism driving variation in chronotype (and hence foraging sequence), will be 27 4 likewise selected for to reduce resource use overlap and thus reduce intraspecific competition at the 1 diel niche axis. To test this hypothesis, we measured the order of arrival to a foraging patch 2 (foraging sequence) and use of torpor in golden and common spiny mice (A. cahirinus) under semi-3 natural conditions, and the free-running period lengths (tau) of the foraging individuals under 4 controlled laboratory conditions. 5

6
The experiment yielded a total of 550,000 foraging logs and 300,000 body temperature records. For 7 one common spiny mouse (from enclosure 3) there were no body temperature data, and three 8 golden spiny mice (two from enclosure 3 and one from enclosure 2) were not recaptured and 9 consequently we could not measure their torpor use or period length; these individuals were 10 excluded from the data analysis.

11
Common spiny mice 12 Individuals from the same enclosure entered the foraging patches in a constant order during the 13 experimental period (Fig 1A). The free running period length (tau) of common spiny mice ranged 14 between 23.7 h to 24.3 h ( Fig. 2A).

15
Our statistical models suggested that individuals that had shorter period length arrived earlier to  Golden spiny mice 1 Individuals from the same enclosure entered the foraging patches in a constant order during the 2 experimental period ( Fig 1B). Furthermore, the frequency distribution of the period length of the 3 golden spiny mice individuals is dispersed between 23.2 h to 25.4 h (Fig. 2B). In enclosure # 2 4 where four golden spiny mice were placed, two pairs of individuals overlapped almost entirely in 5 tau, and the four individuals were fairly close in tau as well ( Fig 2B). Hence no correlations could 6 be discovered and this enclosure was excluded from the following analyses.

7
Individuals that had shorter period length (τ) arrived earlier to the foraging patches (Estimate=  suggests that tau has a significant genetic component (9, [12][13][14], so it appears that the order of 9 arrival to forage in a food patch, returns in energy acquisition, and the consequent energy 10 conserving strategy required, are strongly influenced by the genetic makeup of the individual.

11
We note that the entire pool of mice studied constitutes random sampling by trapping the local free- enclosures are limited (4-6), they cannot be taken to fully represent tau in the native population. 16 Even with this limitation, a clear pattern emerges in seven of eight populations. We thus suggest 17 that in the field, intra-population variation in tau, as a proximate mechanism for driving variation in 18 chronotype, is selected for to reduce resource use overlap or intraspecific aggression (see below) 19 and thus to reduce intraspecific competition at the diel niche axis.  However, it is also possible that being active at different times reduces direct, aggressive 1 interactions between individuals of the same species: in a previous study, we found that only rarely 2 did two individuals forage concurrently in a foraging patch and in the infrequent cases when this 3 occurred, 90% of the individuals left within 2 seconds (19).

4
The question whether the deviation of the internal clock period from 24 hours has any specific 5 function or causal meaning has attracted attention of scientists for years. Traditionally, scientists 6 considered the deviation of tau from 24 hours a simple biological imperfection which is corrected 7 daily by a Zeitgeber (light exposure), and therefore does not lead to serious functional deficits.

8
Accordingly, most studies of this variation focused on its proximate casual mechanism, trying to 9 identify the molecular mechanisms generating the circadian phenotypic diversity, and not on its that different chronotypes will have a selective advantage during different times. 17 The source of variation in chronotypes of spiny mice is currently unknown. In humans, genetic with older individuals having earlier chronotypes was found in in rodents (e.g.,(36)). Spiny mice 26 may live for 2-3 years in natural conditions, so it is entirely possible that their tau changes with age. 27 We know not the ages of the individual spiny mice studied, nor the age structure of the population. While it would be interesting to speculate that the significant intrapopulation variation in tau 2 reflects both selective pressures at the individual level and age-driven individual variation, further 3 research is required to address this question.

4
In humans, extreme chronotypes are usually interpreted as syndromes (e.g., delayed and advanced reduce the hazards of sleep to the whole group. When group size drops, active sentinel behavior is 10 adopted (37). Thus intraspecific variation in tau may have evolutionary significance for anti-11 predator vigilance and for coexistence between competing individuals. Possibly, it has other roles 12 that should be further explored.

13
In sum, our study suggests that the significant variation in tau that is frequently found among 14 individuals is not a mere 'biological imperfection' but has evolutionary significance and is selected 15 for; our study system demonstrates its role in reducing intraspecific competition between 16 individuals within a population. (considered as repeated measures for each individual, as explained in the Statistical analysis, p.13). 16 Monitoring foraging time 17 We used auto-monitored foraging patches, comprising a plastic tray (25 cm diameter), in which 2L 18 of local soil were mixed with 3 g of broken sunflower seeds (3g seeds per individual). In each 19 enclosure, we placed two round antennas (20 cm diameter) beneath two foraging trays and each Use of torpor: torpor temperature threshold was defined according to Willis (40). For each day, we 10 calculated individual total time torpid. Torpor bout duration was determined as time between the 11 beginning of the torpor bout until body temperature returned above the torpor threshold. For each 12 individual, we calculated the mean time spent torpid across days.