Stability of Adult Emergence and Activity/Rest Rhythms in Fruit Flies Drosophila melanogaster under Semi-Natural Condition

Here we report the results of a study aimed at examining stability of adult emergence and activity/rest rhythms under semi-natural conditions (henceforth SN), in four large outbred fruit fly Drosophila melanogaster populations, selected for emergence in a narrow window of time under laboratory (henceforth LAB) light/dark (LD) cycles. When assessed under LAB, selected flies display enhanced stability in terms of higher amplitude, synchrony and accuracy in emergence and activity rhythms compared to controls. The present study was conducted to assess whether such differences in stability between selected and control populations, persist under SN where several gradually changing time-cues are present in their strongest form. The study revealed that under SN, emergence waveform of selected flies was modified, with even more enhanced peak and narrower gate-width compared to those observed in the LAB and compared to control populations in SN. Furthermore, flies from selected populations continued to exhibit enhanced synchrony and accuracy in their emergence and activity rhythms under SN compared to controls. Further analysis of zeitgeber effects revealed that enhanced stability in the rhythmicity of selected flies under SN was primarily due to increased sensitivity to light because emergence and activity rhythms of selected flies were as stable as controls under temperature cycles. These results thus suggest that stability of circadian rhythms in fruit flies D. melanogaster, which evolved as a consequence of selection for emergence in a narrow window of time under weak zeitgeber condition of LAB, persists robustly in the face of day-to-day variations in cycling environmental factors of nature.


Introduction
Circadian rhythms are manifestation of an organism's temporal adaptation to daily changes in cyclic environmental conditions. Therefore, the accuracy with which circadian timing systems maintain stable phase-relationship with cyclic environment is likely to be a key determinant of the organism's temporal adaptation in nature [1]. While it is likely that a large variety of time-cues (zeitgebers) have contributed to the evolution of circadian rhythms, most organisms seem to use light/dark (LD) cycles as the primary zeitgeber to time their cyclic behavioural and metabolic processes [2]. Besides LD cycles, circadian clocks of some organisms have been shown to synchronize to warm/cold (WC) cycles in otherwise constant laboratory condition [3][4][5]. While studies to examine circadian entrainment have invariably been carried out in the laboratory under simulated natural LD and WC cycles, such conditions markedly differ from natural environment in several aspects, including abrupt versus gradual changes in environmental factors, stochasticity and levels of these factors [6][7][8][9][10][11][12][13]. A few recent studies in fruit flies Drosophila melanogaster attempted to elucidate circadian entrainment of adult emergence and activity/rest rhythms under natural conditions [13,14]. During summers, flies show a third peak of activity, occurring during hottest and brightest part of the day, manifested in an additional ''afternoon peak'' [13]. Adult emergence under SN occurs mostly during dawn coinciding closely with humidity maxima and temperature minima of the day [14]. The gate-width of emergence, variance in peak-to-peak timing and fraction of flies emerging during night were shown to be modulated considerably in different seasons. The major differences in emergence rhythm between semi-natural (SN) and laboratory (LAB) conditions include much narrower emergence gate-width and reduced nighttime emergence in SN along with the finding that rhythmicity in emergence is rescued in genetically arrhythmic period null (per 0 ) flies [14]. Therefore, from these studies it is clear that several aspects of daily rhythms differ between natural and laboratory conditions. This raises concern about the validity of deductions made about rhythmic behaviours in nature based on laboratory studies, and whether they can be used to draw inferences regarding the evolutionary forces behind the origin of circadian clocks. For instance, a recent study asked whether early and late emergence chronotypes which evolved as a result of selection for morning and evening emergence under weak environmental cycles of the laboratory would persist in nature [15]. Surprisingly, not only early-late chronotypes persisted in these populations, under semi-natural conditions the phenotypes became even more striking, and showed no correlation with natural morning and evening transitions, unlike that observed under laboratory conditions. Such differential effect of LAB versus SN on the phase of emergence makes it relevant to ask how stability of circadian rhythms is influenced by the presence of multiple gradually changing zeitgebers of nature.
It is expected that precisely timed rhythmic phenomena would confer greater adaptive advantage to organisms living in cyclic environment compared to randomly occurring ones [16,17]. Furthermore, findings from some studies show association between circadian period close to 24 h and higher precision, suggesting that stability in circadian entrainment contributes to the evolution of circadian clocks in the face of cyclic fluctuations in nature [18][19][20]. To critically examine this, we carried out a laboratory selection study, where we imposed selection for adult emergence in a narrow window of time. As a result, selected populations evolved enhanced synchrony and accuracy in the phase of entrainment under LD cycles, and increased homogeneity and precision in clock period under constant darkness (DD) [21]. Furthermore, flies from selected populations displayed enhanced stability in circadian entrainment under a wide variety of environmental conditions created in the laboratory [22]. Although all our assessments of stability in circadian entrainment were performed under laboratory condition aided with relatively weak entraining cues -LD cycles with a light intensity of ,100 lux and constant temperature, humidity and ad libitum food, these studies clearly indicate that stability in circadian time-keeping evolves in response to periodic selection pressures under cyclic environment. We examined whether the mechanisms that evolve under the influence of weak selection pressures in the laboratory, contribute in any way to render the selected populations in having stable rhythmic behaviours under natural conditions which are intrinsically stronger, yet more labile. Moreover, the instance of reduction or enhancement of the differences in stability between selected and control populations under SN, could provide clues for the distinctions and relative contributions of environmental factors in determining stable circadian entrainment.
Stability of rhythmic behaviours in nature is thought to be determined by the interplay of stability of circadian pacemakers and of entraining signals [19]. Therefore, assessment of stability in circadian entrainment under SN is likely to provide a more realistic picture of how stability of rhythmic behaviour is maintained in nature. With this in mind, we examined emergence and activity rhythms of four populations each of selected and control flies under SN and LAB, and used various aspects of rhythms such as amplitude, synchrony and accuracy as measures of stability. To tease apart the role of light and temperature on stability in circadian entrainment, we assayed these rhythms in flies from selected and control populations under laboratory LD and WC cycles. Stability of emergence and activity rhythms of flies from populations selected for precise timing of adult emergence persisted under SN more robustly with higher stability, primarily mediated by light.

Population Maintenance and Selection Protocol
Four replicate populations were initiated by selecting for flies that emerged during ExT7-8 (External Time 7-8, where time of the day is expressed as time difference in hour relative to the midpoint of dark phase [23]) under 12:12 h LD cycles. Anticipating evolution of reduced variance in clock properties we had named these populations as Precision Populations (PP; [21,22]). These populations were originated from four ancestral baseline populations of D. melanogaster which have been maintained under laboratory for several hundred generations under 12:12 h LD cycles at 25uC on banana-jaggery (BJ) food [24]. These baseline populations were derived from an ancestral population, namely the Ives population [25], which were wild caught from South Amherst, Massachusetts, USA, about 30 years ago. Those populations were maintained in the laboratory for about 110 generations under constant light (LL), constant temperature (2561uC) and constant humidity on a 14 day discrete generation cycle. Five replicate populations (B1.B5) were derived from the IV population [26,27], from which after about 360 generations, 5 sets of UU populations were derived [28]. The UU populations were maintained on a 21 day discrete generation cycle, and reared under LL regime in the laboratory for about 170 generations after which four populations (JB1.JB4) were derived from four UU populations (UU1, 2, 3, 5; [24]) from which after ,100 generations four LD (1.4) populations were derived. These populations served as the baseline populations for the selection experiments. We positioned the selection window 1 h (ExT7) after lights-on (ExT6) to select flies from emergence peak and to exclude those that emerge in 'startle response' to lights coming on. By doing so we ensured that flies that emerge as a result of 'masking' of emergence rhythm are not a part of the breeding population for the next generation. Four replicate control populations (CP 1-4 ) were also initiated along with four replicate selected populations (PP 1-4 ), in which flies emerging throughout the day were used as breeding population for the next generation. Therefore, CP populations experienced all other conditions similar to the selected populations except that they were not under any conscious selection for timing of emergence. For both selected and control populations, flies that emerged for four successive days (9 th -12 th day after egg collection) were taken to form breeding adults for the next generation. A total of 1200 adults per population, with approximately equal number of males and females, were maintained in plexiglass cages of 25620615 cm 3 dimension with BJ medium in petridish as food, which was changed every alternate day. Flies were fed with yeast-acetic-acid paste for 3 days before egg collection to induce greater egg production. Three days later, eggs were collected over 3 h window (ExT7-10) on BJ medium and approximately ,300 eggs were transferred to glass vials (18 cm height62.4 cm diameter) containing ,10 ml of BJ medium. Thus, both selected and control populations were maintained on a 21 day non-overlapping generation cycle. Exactly 48 and 16 such vials were set-up at every generation for each of the PP and CP stocks, respectively. The selection has being continuing for over 80 generations. Prior to the assays, all selected and control populations were subjected to one generation of common rearing condition to minimize non-genetic parental effects which may have been caused by the imposition of the selection protocol. The progeny of these flies will be henceforth referred to as ''standardized flies''.

SN and LAB Conditions
It is likely that levels of light, temperature and humidity experienced by the flies in glass vials/tubes are different from what was measured by the Drosophila Environmental Monitor (DEnM), and therefore we prefer to refer our outdoor experimental conditions as semi-natural (SN). Condition used in the LAB was either 12:12 h LD cycles with ,100 lux light intensity at ,25uC, or 12:12 h WC cycles (29/25uC) in constant darkness (DD). The reason behind the choice of light intensity of ,100 lux for the light phase of LD was because these flies were reared under this light intensity for about 80 generations. Assays under SN were conducted in an outdoor enclosure kept under canopy [14] inside the campus of Jawaharlal Nehru Centre for Advanced Scientific Research, Bangalore (12u599N 77u359E). During the assays under SN, light, temperature and humidity were continuously monitored using DEnM System from Trikinetics, USA.

Adult Emergence Assay
For the adult emergence rhythm assay, eggs were collected from standardized populations and transferred at a density of ,300 eggs per vial containing ,10 ml of food. In each environmental regime, 10 such vials were used for each of the four replicate populations, and four such replicates were used for selected and control populations each. Thus, for the adult emergence assay, under each environmental regime, a total of 40 vials were used for PP and 40 for CP populations. Emergence rhythm of PP and CP was assayed under LAB and SN conditions during spring (18 th to 23 rd February 2011). Vials kept under LAB (LD and WC) and SN conditions were closely monitored for the emergence of the first fly and thereafter assayed regularly at every 2 h interval for at least 5 successive days and the number of emerging adults in every 2 h period recorded. To assess the percentage of flies that emerged close to the selection window, we assayed during ExT5-10 the number of adults emerging every 1 h. To estimate the percentage of flies that emerge in 1 or 2 h interval, we calculated the total number of flies that emerged from each vial per cycle for all the five days. To obtain percentage emergence, the number of flies that emerged during 1 or 2 h assay interval was divided by the total number of flies that emerged in that particular cycle. The data from each replicate population were averaged over five consecutive cycles. The mean values obtained from four such replicate populations were plotted as a function of external time [23]. We also estimated the gatewidth of emergence by calculating time-interval between the start and end of emergence on each day, with start and end defined as emergence greater than or less than an arbitrary cutoff of 5% of daily emergence.

Activity/Rest Assay
The activity/rest behaviour of flies was assayed for a minimum of 8-10 days under LAB (LD and WC) and SN conditions during spring (12 th to 19 th March 2011). For this, eggs were collected from the standardized populations and transferred into vials containing ,10 ml of food at a density of ,300 eggs per vial. Two-day old virgin males were introduced singly into activity tubes to record their activity/rest behaviour using Drosophila Activity Monitors from Trikinetics, USA. The activity/rest behaviour of flies was recorded continuously for 8-10 days. Under each environmental regime, about 32 flies were used for each replicate population, and four replicate populations each were used for PP and CP. The daily waveform of activity rhythm of each fly was obtained by dividing hourly collected activity data by the total amount of activity of that fly in that cycle. The waveform plotted is averaged across four replicate populations each having ,32 flies. The phase of morning activity peak was taken as the time point in the morning showing maximum activity. The comparisons of afternoon and evening peaks are based on the differences seen in the waveforms.

Synchrony and Accuracy of Emergence and Activity Rhythms
Synchrony and accuracy was used as measures of interpopulation/individual, and intra-population/individual variance in daily rhythm of selected and control populations [21,29]. To estimate synchrony in emergence rhythm of each replicate population, mean phase-relationship (y) of the emergence peak relative to lights-on was obtained for each of the 10 vials, by averaging it across 5 cycles. The inverse of standard deviation (SD) across 10 such average y values was used as a measure of ''synchrony'' in emergence rhythm of each replicate population. The mean values obtained from four such replicate populations were used to calculate the synchrony of the population. Similarly, mean y of morning activity peak for each individual fly was obtained by averaging it across 10 cycles. The inverse of SD across 32 such mean y values was used as a measure of synchrony of activity rhythm within each population. The mean values obtained from four such replicate populations were used to calculate the synchrony of the population. ''Accuracy'' was estimated as the reciprocal of SD of day-to-day y of emergence or activity rhythms. To estimate accuracy of emergence rhythm of each replicate population, we first calculated the inverse of SD of daily y of emergence peak for each vial across 5-days and then averaged it across all 10 vials for each replicate population. The mean values obtained from four such replicate populations were used as ''accuracy''. Similarly for activity rhythm, we estimated reciprocal of SD of daily y of morning activity peak across 10 days for each individual and then averaged it across all flies. Synchrony and accuracy have a unit of h 21 .

Statistical Analyses
The emergence and activity during the selection window and their waveforms, gate-width of emergence, synchrony and accuracy of emergence and activity rhythms were analyzed separately in mixed model analysis of variance (ANOVA) treating replicate populations (Block-B) as random factor, environmental condition (E), population (P) and time point (T) as independent fixed factors crossed with replicates. Mean values of emergence at every 1 or 2 h and activity at every 1 h were the dependent variables used as inputs in the mixed model ANOVA. For specific measures, the mean values of each replicate population of the dependent variables such as emergence and activity during the selection window, gate-width of emergence, synchrony and accuracy of emergence and activity rhythms were the inputs for the model. ANOVA was followed up by post-hoc multiple comparisons using Tukey's test. 95% Comparison Interval (95%CI) calculated using minimum significant difference in Tukey's test [30] was used in the figures as error bars to facilitate visual hypothesis testing [31]. Therefore, overlapping error bars would imply that the values do not differ statistically. The waveforms of emergence rhythm of PP and CP flies were additionally analyzed using Kolmogorov-Smirnov two-sample test to assess whether or not they were different. All analyses were implemented on STATISTICA TM for Windows Release 5.0 B [32].

Persistence of Robust Emergence Rhythm Under SN
As we have shown previously [21], under LAB, PP showed enhanced peak of emergence compared to CP, and this difference persisted even more robustly under SN (Figures 1, 2B). In addition, daily emergence waveform of PP under SN was more robust than CP, marked by enhanced emergence peak and reduced gate-width ( Figures 1B, 2B, 3B). This was further confirmed in Kolmogorov-Smirnov's test, which revealed that while emergence waveform of PP and CP was significantly different from each other under both LAB (p,0.05) as well as SN conditions (p,0.01), emergence waveform of each both PP (p,0.05) and of CP (p,0.05) were significantly different when compared among LAB and SN.
While more PP flies emerged during the selection window (ExT7-8) compared to CP under both conditions (Figures 2A, B,  3A), PP flies emerged in greater numbers during the selection window under SN than in LAB. Concurrent to increase in percentage of PP flies emerging during selection window, a decrease prior-to (ExT6-7) and an increase after (ExT8-9), associated with reduced lights-on anticipatory emergence was observed under both conditions (ExT5-6; Figure 2A  greater numbers compared to LAB ( Figure 3A). Enhanced amplitude of emergence rhythm in PP, suggests that under SN emergence rhythm of PP was more robust than CP.

Selected Flies have a Narrower Gate of Emergence Under SN
Under LAB, selected populations had a significantly reduced gate-width compared to controls, which persisted even more robustly under SN (Figures 2A, 3B), with gate-width difference between PP and CP was greater compared to that seen in LAB (Figures 2A, B, 3B). ANOVA on gate-width data revealed a statistically significant effect of E (F 1,3 = 179.6; p,0.0008), P (F 1,3 = 153.9; p,0.001) and E6P interaction (F 1,3 = 5.64; p,0.04). Post-hoc multiple comparisons using Tukey's test revealed that gate-width of emergence in both populations was reduced significantly under SN compared to LAB, which suggests that emergence rhythm was more robust under SN, however, under both conditions gate-width of PP flies was significantly narrower compared to CP ( Figure 3B).

Selected Flies Display Enhanced Synchrony and Accuracy in Emergence Rhythm Under SN
The synchrony and accuracy of emergence rhythm of PP was significantly higher compared to CP (Figure 3C, D). ANOVA revealed a statistically significant effect of E (synchrony-F 1,3 = 15.81; p,0.02; accuracy-F 1,3 = 17.54; p,0.02) and P (synchrony-F 1,3 = 19.2; p,0.02; accuracy-F 1,3 = 38.2; p,0.008), while the effect of E6P interaction was statistically not significant (synchrony-F 1,3 = 1.1; p = 0.3; accuracy-F 1,3 = 0.73; p = 0.4). Posthoc multiple comparisons using Tukey's test revealed that, under both conditions synchrony and accuracy of emergence rhythm were significantly higher in PP compared to CP ( Figure 3C, D), but of a given population did not differ between SN and LAB. Persistence of enhanced synchrony and accuracy indicates that emergence rhythm of selected flies continues to be more stable under SN compared to controls.

Selected Flies Display Enhanced Morning Peak of Activity Under SN
Selection also altered the waveform of activity rhythm, consequently in LAB, morning activity peak and daytime activity of PP was enhanced and nighttime activity reduced compared to CP ( Figure 4A, B). Under SN while morning and evening activity peaks of PP was significantly greater than CP (Figure 4A), daytime and nighttime activity of PP and CP did not differ ( Figure 4B). Flies displayed a third peak of activity (afternoon peak; Figure 4A, B). The amplitude of afternoon peak in PP was significantly reduced compared to CP (Figure 4a). ANOVA revealed a statistically significant effect of T (F 23,69 = 90.04; p,0.0001), E6T (F 23,69 = 76.9; p,0.0001), P6T (F 23,69 = 3.64; p,0.0001), and E6P6T interaction (F 23,69 = 3.40; p,0.0001), whereas the effect of E (F 1,3 = 0.05; p = 0.8), P (F 1,3 = 0.09; p = 0.7) and E6P interaction was statistically not significant (F 1,3 = 0.09; p = 0.7). Post-hoc multiple comparisons using Tukey's test revealed that, under LAB, activity during morning peak (ExT6-8) including the selection window (ExT7-8) and during evening peak (ExT16-18) was significantly enhanced and during nighttime (ExT0-4) significantly reduced in PP flies compared to CP. Similarly, under SN, activity of PP during morning peak (ExT7-8) and evening peak (ExT18-20) was significantly greater than CP. In addition, PP flies were significantly less active during afternoon (ExT13-16) compared to CP ( Figure 4A). Persistence of increased activity during morning and evening peaks compared to controls under SN, suggests increased stability of activity rhythm in selected flies.

Selected Populations Exhibit Enhanced Synchrony and Accuracy in Activity Rhythm Under SN
Under both conditions, PP flies showed enhanced synchrony and accuracy compared to CP (Figure 4C, D). ANOVA revealed a statistically significant effect of E (synchrony-F 1,3 = 17.1; p,0.02; accuracy-F 1,3 = 6.8; p,0.04), P (synchrony-F 1,3 = 10.1; p,0.04; accuracy-F 1,3 = 13.3; p,0.03), while the effect of E6P interaction was statistically not significant (synchrony-F 1,3 = 3.1; p = 0.1; accuracy-F 1,3 = 0.1; p = 0.9). Post-hoc multiple comparisons using Tukey's test revealed that while synchrony and accuracy of activity rhythm in PP was significantly higher compared to CP under both LAB and SN ( Figure 4C, D), it did not differ between SN and LAB. Persistence of enhanced synchrony and accuracy under SN indicates enhanced stability of activity rhythm in the selected flies.

Stability of Emergence Rhythm Under Warm/Cold (WC) Cycle
In order to study whether light or temperature cycles contributed more to the enhanced stability of emergence and activity rhythms under SN, we assayed these rhythms in the LAB under cycles of 12 h warm phase (29uC) and 12 h cool phase (25uC) in DD and compared it with those assayed under LAB LD at 25uC. While emergence peak occurred significantly earlier under WC compared to LD, PP flies displayed enhanced emergence peak compared to CP ( Figure 5A). ANOVA on emergence data revealed a statistically significant effect of T (F 11,33 = 178.4; p,0.0001), E 6 T (F 11,33 = 807.5; p,0.0001), P 6 T (F 11,33 = 6.7; p,0.0001), and E6P6T interaction (F 11,33 = 9.1; p,0.0001), while the effect of E (F 1,3 = 1; p = 0.3) and P was statistically not significant (F 1,3 = 1; p = 0.3). Post-hoc multiple comparisons using Tukey's test revealed that under WC, emergence in PP during ExT2-4 was significantly higher compared to CP ( Figure 5A).
The tightness of emergence gate, evolved as a correlated response to selection, was not seen under WC ( Figure 5B), which is associated with reduced synchrony in emergence rhythm in PP. Although accuracy in both populations was reduced in WC, PP flies had higher accuracy in emergence compared to CP ( Figure 5C, D). ANOVA revealed a statistically significant effect of E (gate-width-F 1,3 = 16.7; p,0.02; synchrony-F 1,3 = 78.4; p,0.003; accuracy-F 1,3 = 22.4; p,0.02), P (gate-width-F 1,3 = 176.6; p,0.0009; synchrony-F 1,3 = 16.2; p,0.02; accuracy-F 1,3 = 21.6; p,0.01), and E6P interaction (accuracy-F 1,3 = 0.02; p,0.04), while the effect of E6P interaction on gate-width and synchrony was statistically not significant (gate-width-F 1,3 = 0.9; p = 0.4; synchrony-F 1,3 = 6.2; p = 0.08). Post-hoc multiple comparisons using Tukey's test revealed that gate-width of PP did not differ from CP under WC, while in LD it was significantly narrower than CP ( Figure 5B). Synchrony of emergence rhythm was reduced considerably in WC compared to LD, and PP flies had enhanced synchrony compared to CP in LD but not in WC ( Figure 5C). Although, accuracy of emergence rhythm was significantly reduced under WC compared to LD, under both conditions it was significantly higher in PP compared to CP ( Figure 5D). These results suggest that while within population features of emergence rhythm (amplitude and accuracy) of selected flies were higher under WC compared to controls, among population features (synchrony and gate-width) do not differ.
The amplitude of emergence rhythm was reduced under WC, and flies showed increased preference to emerge during the cold phase ( Figure 5E). ANOVA on emergence data under SN, LD and WC showed a statistically significant effect of T (F 11,33 = 227.0; p,0.0001), E6T (F 22,66 = 252.0; p,0.0001), P6T (F 11,33 = 116.1; p,0.0001), and E6P6T interaction (F 22,66 = 106.3; p,0.0001), while the effect of E (F 2,6 = 0.8; p = 0.4) and P was statistically not significant (F 1,3 = 1.3; p = 0.3). Post-hoc multiple comparisons using Tukey's test revealed that emergence peak of PP was significantly higher under SN compared to LD, while it was reduced under WC compared to LAB. The emergence rhythm seems to be differently modulated by external stimuli but PP flies continue to display enhanced robustness in emergence rhythm compared to CP under SN compared to laboratory LD and WC.

Effect of Laboratory Warm/Cold (WC) Cycles on Activity/ Rest Rhythm
PP flies displayed enhanced morning activity peak compared to CP under LD but not in WC ( Figure 5F). Flies from both populations showed reduced amplitude of morning activity peak under WC compared to LD ( Figure 5G). ANOVA on activity data revealed a statistically significant effect of T (F 23,69 = 199.8; p,0.0001), E6T (F 23,69 = 79.2; p,0.0001), P6T (F 23,69 = 7.2; p,0.0001), E6P6T interaction (F 23,69 = 9.7; p,0.0001), while the effect of E (F 1,3 = 0.05; p = 0.8) and P was not statistically significant (F 1,3 = 0.09; p = 0.7). Post-hoc multiple comparisons using Tukey's test showed that while under LD, PP flies exhibited enhanced activity during morning (ExT6-8) and evening peaks (ExT16-18), activity of PP and CP flies did not differ under WC. Enhanced morning peak of activity in PP flies observed under LD did not persist under WC. This lack of any difference in the activity profiles of PP and CP under WC ( Figure 5F) indicates that activity rhythm is more stable under LD cycles, whereas temperature cycles reduce the stability of activity rhythm in PP flies.

Discussion
Accurate knowledge of time of the day is of paramount importance for organisms living under fluctuating environments. In the present study, we elucidated effect of semi-natural conditions on the stability (amplitude, synchrony and accuracy) of daily rhythms in D. melanogaster populations selected for emergence in a narrow window of time. In a previous study we had shown that such selection eliminated flies from both ends of the emergence waveform and favoured those that emerged during the selection window, and this eventually resulted in the evolution of increased stability in emergence and activity rhythms [21]. Such differences in stability between selected and control populations persisted under a wide variety of environmental conditions [22]. Interestingly, the enhanced peak and tighter waveform of emergence in selected flies as observed under LAB persists even more robustly in SN (Figures 1, 2), which suggests that natural conditions are more conducive for the expression of circadian phenotypes. In nature, emergence was mostly restricted to early hours of morning, which coincided with high humidity and low temperature ( Figure 2B). Although emergence in the LAB and SN conditions was essentially at the same time, enhanced amplitude of the rhythm under SN suggests a role for other factors such as humidity and temperature in the regulation of amplitude and profile of emergence rhythm. Flies tend to emerge in association with gradually increasing light and/or temperature during dawn, suggesting that fly emergence in nature is regulated by multiple environmental cycles. Such emergence waveforms are consistent with those reported in a previous study [14], where emergence was found to be tightly gated around dawn, gradually tapering down towards the afternoon. However, this trend of consolidation of emergence under SN appears to not only additively contribute to the tightness and increase in the amplitude of daily rhythms, it seems to preferentially aid the selected populations more than the controls. Apparently, random fluctuations present in the SN environment are not nearly as strong a deterrent as perhaps high amplitude of light cycles or its waveform, to strengthen the selected phenotype even more than in native selection conditions.
The activity rhythm of flies entrained to natural zeitgebers with its morning peak synchronized to dawn and evening peak to dusk. Besides, flies displayed increased activity in the middle of the day, manifested in an additional afternoon peak [13]. The afternoon peak was significantly reduced in selected flies, however, they exhibited enhanced morning and evening activity peaks, and reduced variation in the phase of activity rhythm compared to controls ( Figure 4). Therefore, flies selected for precise timing of emergence maintained greater stability in emergence and activity rhythms compared to controls even under SN, where it is challenged by stochasticity associated with diurnal changes. The afternoon peak is likely to be a stress response to bright light (,2500 lux) on a warm day. This could be tested in future studies by giving flies access to natural light keeping temperature and humidity constant, or to conditions where the natural light alone is excluded, while flies experience natural variation in temperature, humidity and other factors.
Assessment of effects of laboratory light and temperature cycles on circadian entrainment, and on stability parameters of emergence and activity rhythms revealed that flies from selected populations has enhanced and accurate peak of emergence rhythm under LD and WC. However, under WC, synchrony and gate-width of selected flies did not differ from controls, which is probably because these flies were never exposed to temperature changes in the laboratory. This is reflected also in the lack of difference in robustness between selected and control flies when assayed under different ambient temperatures [22]. It is therefore likely that amplitude and stability in the phase of emergence peak are affected by light and temperature in a manner such that the differences between both populations persist, whereas light and temperature have differential effects on the gate-width and withingroup synchrony. Whereas, for activity rhythm, light and temperature differentially affect selected populations such that robustness of morning activity peak of selected flies observed under LD, is diminished considerably under WC, rendering them as stable as controls. Since we find that differences in several features of rhythms between selected and control populations did not persist in WC, it is possible that enhanced stability of rhythms seen under SN is mostly owing to the effect of light. However, an important caveat to note here is that temperature in Bangalore during the course of the study varied between 18 to 30uC, but our choice of temperature for WC cycles was a square wave of 29:25uC. We chose this temperature range considering the fact that the maximum and minimum temperatures under SN were prevalent only for short durations of time. We suspected that 12:12 h square wave of high (30uC) and low (18uC) temperature may be stressful for flies. Therefore, although it is likely that stability of circadian rhythms in selected flies is primarily lightmediated under semi-natural conditions, a complex interaction of a multitude of environmental variables, perhaps acting in a hierarchical manner, regulate the stability of their circadian rhythms in nature. Besides light and temperature fluctuations, precision of circadian rhythms is also known to be affected by variations in photoperiod, which was reported to be higher around equinoxes, and lower during midsummer (in shrews) and during midwinter (in hamsters) [33]. Rigorous and systematic studies under natural conditions along with simulated laboratory conditions with varying light and temperature profiles mimicking natural conditions are required to elucidate how stability of circadian clocks is maintained in the face of daily and seasonally changing environments, and how interaction of these two environmental cycles influences stability of circadian entrainment in nature.
In summary, our study demonstrates the effect of semi-natural condition on stability of circadian rhythms in fruit fly D. melanogaster populations selected for emergence in a narrow window of time. Stability of circadian rhythms in flies from selected populations persisted even more robustly under seminatural conditions displaying enhanced amplitude, synchrony and accuracy compared to controls, which suggests robustness of their time-keeping systems. Stability in emergence and activity rhythms of flies from the selected populations seems to stem from stability in their circadian entrainment primarily due to light, although role of temperature and other factors cannot be completely ruled out. Although under WC, phase of emergence peak of both PP and CP was altered, PP flies continued to show enhanced emergence peak compared to controls. (B) Gate-width of emergence rhythm in PP and CP populations under LD and WC. Under LD, PP had narrower gate of emergence compared to controls whereas in WC the gate-width of CP and PP did not differ. (C, D) Synchrony, accuracy of emergence rhythm under LD and WC cycles. Although synchrony and accuracy is reduced under WC, the difference in accuracy between PP and CP is still maintained. (E) Comparison of emergence profile of PP and CP under SN, LD and WC. PP flies have greater amplitude of emergence peak under SN compared to LD and WC. (F, G) Activity rhythm of PP, CP flies under LD and WC. Under LD, PP flies exhibit enhanced morning activity peak compared to controls which did not persist in WC. In figure 5A, E, F and G the square-wave above the Figure represents 12:12 h LD and WC cycles in LAB and vertical dotted lines represents the transition between warm and cool phase of WC, and in Figure 5E vertical dotted lines represent the transition between warm and cool phase of WC and dark and dark transition of LD. All other details are same as in Figure 2. doi:10.1371/journal.pone.0050379.g005 To the best of our knowledge, this is the first study of its kind demonstrating the effect of day-to-day variations of several zeitgebers present in real natural environment on stability of circadian time-keeping.