INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE NATURAL light-dark (LD) cycle is the most important environmental temporal cue for entraining circadian rhythms (16). A characteristic feature of the natural LD cycle is the gradual change in light intensity during dawn and dusk. Many animals begin or cease locomotor activity during these twilight periods. Dawn and dusk may be a part of the photic zeitgeber that contains important temporal information responsible for entrainment. However, relatively few studies examine the role of these transitional phases on photic entrainment (2, 9, 10, 13, 14, 19, 20).

Previous studies have demonstrated that simulated twilight periods alter the characteristics of entrained circadian rhythms compared with those produced by LD cycles without twilight transitions [square-wave LD cycles (LD_Sq)]. When deer mice in a 16:8-h LD cycle were transferred to an 8:8-h LD cycle, most animals failed to entrain to the new LD cycle (9). However, after simulated dawn and dusk transitions were added to the 8:8-h LD regimen, animals became entrained; adding artificial dawn and dusk transitions expanded the range of entrainment. Similarly, Boulos et al. (2) demonstrated that animals entrained to a 26-h LD cycle if it included twilight transitions; the same LD cycle without transitions failed to synchronize most animals. Boulos et al. (2) suggested that the inclusion of twilight transitions increased the strength of the LD zeitgeber. In addition, hamsters exposed to dawn, dusk, and square-wave light pulses exhibited different phase-response curves (3).

The phase angle between rhythms has also been examined in LD cycles that include twilight periods. Laakso et al. (14) examined the temporal patterns of melatonin secretion and locomotor activity in such LD cycles. In abrupt on-off LD cycles, melatonin levels reached a peak 4 h before lights on. In contrast, under LD cycles with gradual light on-off transitions, there was an elevated mean level of melatonin, and the peak level was phase delayed by 2 h. However, the phase angle between peaks of locomotor activity and melatonin level remained the same under both conditions.

This study was designed to further test the hypothesis that animals exhibit different circadian rhythm characteristics in a simulated natural light-dark cycle (LD_SN) than in an LD_Sq with the same photoperiod. Both activity (Act) and body temperature (T_b) were simultaneously recorded to allow us to compare responses of these two circadian rhythms.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Animals. Thirteen adult (160-180 g) male Syrian hamsters (Mesocricetus auratus; Simonsen Laboratories, Gilroy, CA) were provided water and food ad libitum. They were housed in individual cages (16 × 7.5 × 8 in.) within ventilated, light-tight enclosures (22.5 × 22.5 × 16 in.) at an ambient temperature of 25 ± 0.5°C. The animals were isolated from each other to prevent the transfer of social cues. Cages were changed weekly and animal health was checked daily. This study was conducted between mid-April and mid-July. All surgical and animal care procedures were performed according to the National Institutes of Health Guide for the Care and Use of Laboratory Animals.

Lighting system. The light source consisted of 40 light-emitting diodes (LED; Digi-Key, Thief River Falls, MN) in a 5 × 8 array format located in the center of each enclosure ceiling. The peak emission wavelength of each LED was 565 nm. Maximum intensity output of each 5 × 8 array was 0.05 µmol photons · (m²)¹ · s¹ (3.0 × 10¹² photons · (cm¹)¹ · s¹) measured at the bottom of the animal cage (10 in. below LED array). The linearity of intensity output of each LED matrix was measured using a spectrophotometer. A representative series of LED spectral profiles with increasing intensities (Fig. 1A) shows that the spectral profile was the same for each LED output intensity. Light intensity changed linearly across time (Fig. 1B).

View larger version (30K): [in this window] [in a new window]   Fig. 1.   Properties of the light-emitting diode matrix array. A: representative spectral profile shows stable spectrum at different output intensities. B: regression analysis shows linear intensity output across time. C: representative square-wave light-dark cycle (LDSq) profile. D: representative simulated natural light-dark cycle (LDSN) profile.

The two lighting profiles used in this study were an LD_Sq (Fig. 1C) and an LD_SN (Fig. 1D). For the LD_Sq, LEDs were abruptly turned on to maximum intensity for a 14-h light period and then abruptly turned off for a 10-h dark period. The LD_SN also had a 14-h light period followed by a 10-h dark period in a 24-h cycle. However, the 14-h light period consisted of a 4-h dawn-mimicking period followed by a 6-h constant maximum intensity period and concluded with a 4-h period mimicking dusk. For the 4-h dawn and dusk periods of the LD_SN, the light intensity of the LED arrays was linearly changed between zero and maximum output in 256 steps, 59.25 s/step. Light intensity increased by 1.2 × 10¹⁰ photons · (cm²)¹ · s¹ at each step. Because the threshold irradiance to induce statistically significant phase response in hamsters is 7.4 × 10¹⁰ photons · (cm²)¹ · s¹ (17), the light intensity [3.0 × 10¹² photons · (cm²)¹ · s¹] of the light period in LD_Sq was above this threshold, thus ensuring that each animal received a 14-h photoperiod in the LD_Sq profile. On the other hand, during the simulated dawn in LD_SN, light intensity reached threshold intensity after 7 steps of intensity increase [7 × 1.2 × 10¹⁰ photons · (cm²)¹ · s¹ = 8.4 × 10¹⁰ photons · (cm²)¹ · s¹], which equals ~7 min. Light intensity dropped bellow threshold 7 min before the dark period. This gave the animals 13 h and 46 min of effective photoperiod in the LD_SN profile. Physiologically, we considered this 0.2-h difference in effective photoperiod between the two lighting conditions insignificant. Thus these physiologically identical photoperiods provided a constant marker for comparing the differences resulting from the two different lighting profiles. The total number of photons provided in LD_SN is 71.4% of that in LD_Sq. Twilight transitions in LD_SN provided 29.2% fewer photons.

The light intensity and light profile were controlled by a computer using software developed for this study. Because the LED output intensity was proportional to the applied current (Fig. 1A), this current was used to monitor the lighting regimens throughout the entire experiment. The current controlling output intensity was converted into a corresponding frequency by a voltage-current converter. This frequency was continuously recorded in parallel with the animal biotelemetry data. Period analysis of both lighting profiles demonstrated a 24-h period, and the variance ratio of both profiles was 1.0, indicating that a stable light cycle was provided to the animals throughout the study.

Biotelemetry. Animals were implanted intraperitoneally with a biotelemetry transmitter (VM-FH disc; Minimitter, Sunriver, OR) to monitor core T_b and ambulatory Act. For surgery, the animals were anesthetized with 3% isoflurane in pure medical-grade oxygen, administered using an adjustable isoflurane vaporizer (Viking Medical Products, Medford Lakes, NJ). To prevent hypothermia, all surgical procedures were carried out on a heating pad. Animals were transferred back to their housing enclosure and allowed 7-10 days to recover from surgery.

After recovery, animals were exposed first to LD_Sq for 6 wk and then to LD_SN for another 6 wk. T_b and Act were recorded at 5-min intervals throughout this study. The first 2 wk of each light exposure were considered an adaptive period, and these data were not analyzed.

Data analysis. Period analysis, based on the periodogram method developed by Enright (5), was first performed on all data to verify that all animals exhibited stable entrainment. Average 24-h waveforms (eductions) were generated from each animal to determine the daily onsets and offsets of Act and T_b. The predictive and reactive measures denoted in this study are equivalent to the concept of predictive and reactive homeostasis characterized by Moore-Ede (15). In this study, the intersections of the eduction curve with the daily mean level were denoted reactive onset and offset. In addition, the times of anticipatory Act and T_b (predictive onset and offset) were analyzed. Methods of determining these parameters are described below. Measured reactive and predictive onsets and offsets were then used to determine the relative phase change of each circadian rhythm in LD_Sq and LD_SN.

Reactive onset and offset. The reactive measures were defined to evaluate the circadian changes after the light-dark transitions. The reactive measures of Act onset and offset and T_b onset and offset were determined using the daily mean crossing as a threshold for the wave eduction plot of each animal. The reactive onset time for Act (Act_R1) and T_b (T_bR1) was identified as the first point above the mean level that remained above the mean for more than three points (15 min) in a 1-h period (Fig. 2). The reactive circadian offset time for Act (Act_R2) and T_b (T_bR2) was identified as the last point before the eduction value dropped below the daily mean (Fig. 2). Sporadic short activity bursts that lasted <15 min (3 data points) were ignored.

View larger version (36K): [in this window] [in a new window]   Fig. 2.   Twenty-four-hour rhythms of activity (Act) and body temperature (Tb) records. P1 and P2, predictive onset and offset, respectively. R1 and R2, reactive onset and offset, respectively.

The active period () for reactive measures was calculated for each animal as the interval between the Act_R1 and Act_R2 (_R). The phase angle difference () between Act_R1 and T_bR1 was defined as the time interval between Act_R1 and T_bR1 (_R1). The between Act_R2 and T_bR2 was defined as the time interval between Act_R2 and T_bR2 (_R2). The circadian amplitude was calculated using a cosinor algorithm (7, 8).

Predictive onset and offset. The predictive measures were defined to evaluate circadian changes preceding and anticipatory to light-dark transitions. The predictive Act onset and offset and T_b onset and offset were a direct visual determination of Act and T_b changes from the wave eduction plot of each animal. The Act predictive onset time (Act_P1; Fig. 2) was determined by retrospectively tracing the plot curve to identify when an animal initially became active. The predictive offset time (Act_P2, Fig. 2) was the point before an Act eduction curve started to fall below the daily mean. The same technique was applied to determine the T_b onset (T_bP1) and offset (T_bP2).

The  for predictive measures (_P) was calculated as the interval between the Act_P1 and Act_P2. The between Act_P1 and T_bP1 was defined as the time interval between Act_P1 and T_bP1 (_P1). The between Act_P2 and T_bP2 was defined as the time interval between Act_P2 and T_bP2 (_P2). The period of anticipatory Act onset was defined as the time between Act_P1 and Act_R1. The period of anticipatory T_b onset was the difference between T_bP1 and T_bR1. The time intervals between Act_P2 and Act_R2 and between T_bP2 and T_bR2 were calculated as anticipatory offsets.

Statistical Analysis. Parameters describing Act and T_b rhythms were compared by the use of the paired t-test. Differences were considered significant at P < 0.05.

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Animals in this study entrained to both the LD_Sq and LD_SN cycles. However, the variance ratio was higher under LD_Sq than under LD_SN (Fig. 3A). The difference between the average variance ratio of Act in LD_Sq (0.30 ± 0.02) and in LD_SN (0.20 ± 0.02) was significant (t = 5.98, df = 12, P < 0.01). This was also true for T_b (0.55 ± 0.02 and 0.44 ± 0.02 for LD_Sq and LD_SN, respectively; t = 5.22, df = 12, P < 0.01). The circadian amplitudes (Fig. 3B) of Act and T_b were 38 ± 8 counts and 0.56 ± 0.02°C, respectively, in LD_Sq. In LD_SN, Act and T_b were 22 ± 4 counts and 0.42 ± 0.02°C, respectively. Animals exposed to LD_Sq exhibited significantly higher circadian amplitudes of daily Act (t = 3.77, df = 12, P < 0.01) and T_b (t = 5.85, df = 12, P < 0.01) than those exposed to LD_SN.

View larger version (18K): [in this window] [in a new window]   Fig. 3.   A: variance ratios of period analysis of Act and Tb both show significant decreases in LDSN. B: significant decrease in circadian amplitude of Act and Tb for animals exposed to LDSN. ** P < 0.01 vs. LDSq.

Reactive onsets and offsets and associated measures. For reactive measures (Fig. 4A), average Act_R1 occurred at zeitgeber time (ZT) 13.1 ± 0.2 and 12.0 ± 0.3 in LD_Sq and LD_SN, respectively. This 1.1-h circadian Act_R1 advance from LD_Sq to LD_SN was significant (t = 7.29, df = 12, P < 0.01). Act_R2 times, which occurred at ZT 23.1 ± 0.2 and 23.6 ± 0.3 in LD_Sq and LD_SN, respectively, showed no significant difference between lighting conditions.

View larger version (24K): [in this window] [in a new window]   Fig. 4.   Onset (open symbols) and offset times (filled symbols) for Act and Tb. A: reactive onset of Act (ActR1; ) and Tb (TbR1; ) and reactive offset of Act (ActR2; ) and Tb (TbR2; ) in LDSq and LDSN. In LDSN, ActR1 advanced significantly (1.1 h), as did TbR1 (0.7 h). ActR2 remained unchanged between 2 lighting conditions, but TbR2 was significantly delayed by 0.6 h in LDSN. B: predictive onsets for Act (ActP1; ) and Tb (TbP1; ) and predictive offsets for Act (ActP2; ) and Tb (TbP2; ). ActP1 advanced significantly (0.3 h), whereas TbP1 showed 0.6-h delay. ActP2 remained unchanged between 2 lighting conditions. TbP2, however, was significantly delayed by 0.6 h in LDSN. * P < 0.05, ** P < 0.01 vs. LDSq.

Mean circadian T_bR1 was at ZT 12.1 ± 0.2 and 11.4 ± 0.2 in LD_Sq and LD_SN, respectively (Fig. 4A). The 0.7-h advance from LD_Sq to LD_SN was significant (t = 3.58, df = 12, P < 0.01). Mean T_bR2 was at ZT 23.4 ± 0.2 and 24.0 ± 0.3 in LD_Sq and LD_SN, respectively. T_bR2 was significantly delayed by 0.6 h in LD_SN (t = 2.27, df = 12, P < 0.05). The average _R1 in LD_Sq (1.0 ± 0.2 h) was not significantly different from that in LD_SN (0.6 ± 0.2 h; Fig. 5A). The _R2 in LD_Sq (0.3 ± 0.1 h) also did not differ significantly from that in LD_SN (0.4 ± 0.1 h). The average _R (Fig. 5B) was 9.9 ± 0.3 and 11.6 ± 0.5 h in LD_Sq and LD_SN, respectively. The _R was significantly lengthened from LD_Sq to LD_SN by 1.7 h (t = 2.63, df = 12, P < 0.05).

View larger version (15K): [in this window] [in a new window]   Fig. 5.   Phase relationship between Act and Tb. Filled bars, LDSq; open bars, LDSN. A: phase angle difference () between ActR1 and TbR1 remained unchanged between 2 lighting conditions, whereas between ActP1 and TbP1 was significantly shortened by 0.9 h. in offset times remained unchanged between lighting conditions. B: in LDSN, active periods for reactive (R) and predictive measures (P) lengthened significantly by 1.0 and 1.7 h, respectively. * P < 0.05, ** P < 0.01 vs. LDSq.

Predictive onsets and associated measures. Using predictive measures, average Act_P1 occurred at ZT 11.7 ± 0.2 and 11.4 ± 0.2 in LD_Sq and LD_SN, respectively (Fig. 4B). Act_P1 showed a significant advance of 0.3 h from LD_Sq to LD_SN (t = 2.42, df = 12, P < 0.05). Act_P2 occurred at ZT 22.6 ± 0.3 and 23.2 ± 0.3 in LD_Sq and LD_SN, respectively. Act_P2 showed no significant difference between lighting conditions.

Mean circadian T_bP1 was at ZT 9.7 ± 0.3 and 10.3 ± 0.2 in LD_Sq and LD_SN, respectively (Fig. 4B). T_bP1 showed no significant difference between LD_Sq and LD_SN. T_bP2 was at ZT 22.9 ± 0.3 and 23.5 ± 0.3 in LD_Sq and LD_SN, respectively. T_bP2 was significantly delayed by 0.6 h in LD_SN (t = 2.30, df = 12, P < 0.05). The average _P1 was 2.0 ± 0.4 h in LD_Sq and significantly decreased to 1.1 ± 0.1 h in LD_SN (Fig. 5A; t = 2.13, df = 12, P < 0.05). The _P2 in LD_Sq (0.2 ± 0.03 h) did not differ from that in LD_SN (0.3 ± 0.01 h). The average _P (Fig. 5B) was 10.9 ± 0.3 and 11.9 ± 0.4 h in LD_Sq and LD_SN, respectively. The _P was significantly lengthened from LD_Sq to LD_SN by 1.0 h (t = 4.32, df = 12, P < 0.01).

Anticipatory behavior. Animals exposed to LD_Sq typically exhibited an anticipatory Act onset (Act_P1  Act_R1). Such intervals are shown as shaded areas in Fig. 6A. The length of this anticipatory activity shortened under LD_SN (Fig. 6B), completely disappearing in some animals. The duration of the mean anticipatory activity significantly decreased from 1.4 ± 0.2 to 0.6 ± 0.2 h in LD_Sq and LD_SN, respectively (Fig. 7; t = 6.69, df = 12, P < 0.01).

View larger version (37K): [in this window] [in a new window]   Fig. 6.   Changes in anticipatory intervals during onset and offset in LDSq and LDSN. Shaded areas denote anticipatory intervals for Act and Tb in LDSq (A) and LDSN (B). Intervals associated with onsets were shorter, whereas those associated with offsets remained unchanged.

View larger version (14K): [in this window] [in a new window]   Fig. 7.   Mean duration of anticipatory onset and offset interval for Act (ActR1P1 and ActR2P2, respectively) and Tb (TbR1P1 and TbR2P2, respectively). In LDSN, anticipatory onset intervals decreased significantly by 0.8 and 1.3 h for Act and Tb, respectively. However, offset intervals were not affected in either lighting condition. ** P < 0.01 vs. LDSq.

An anticipatory rise in T_b also occurred before and during the corresponding anticipatory Act, as shown in the shaded area between T_bP1 and T_bR1 (Fig. 6A). The mean duration of anticipatory T_b in LD_Sq decreased significantly in LD_SN, from 2.4 ± 0.3 to 1.1 ± 0.1 h, respectively (Fig. 7; t = 3.29, df = 12, P < 0.01). The T_b maximum generally appeared ~5 min after the corresponding peak of Act (Fig. 6), although the rise of T_b started earlier than the increase of Act. Act and T_b also showed different onset slopes. Generally, the slope of anticipatory T_b was shallower than that of anticipatory Act. However, after Act reached its first peak, changes in T_b tended to parallel Act, as shown in Fig. 6. The average anticipatory offsets of both Act and T_b were similar and showed no differences in either LD cycle (Fig. 7). The anticipatory offset of Act was 0.4 ± 0.2 and 0.4 ± 0.1 h in LD_Sq and LD_SN, respectively. The anticipatory offset of T_b was 0.5 ± 0.2 and 0.5 ± 0.1 h in LD_Sq and LD_SN, respectively.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

This study demonstrates that animals exhibit different circadian characteristics when exposed to 24-h LD cycles that vary in the rate of their LD transitions. In LD_SN, both Act_R1 and Act_R2 were phase advanced compared with those in LD_Sq. T_bR1 was also advanced; however, T_bP1 did not change between lighting conditions. The Act offsets remained unchanged between lighting conditions, whereas the T_b offsets were slightly delayed in LD_SN. The _R1 was not affected by the different LD cycles; however, the _P1 was shortened in LD_SN. In addition, the between reactive and predictive offsets, including the reversed phase relationship to the onset measures, remained unchanged in either lighting environment. A longer active period of the circadian activity rhythm was exhibited in LD_SN. The anticipatory Act and T_b onsets were shortened in LD_SN. Most of the changes in the circadian parameters between LD_Sq and LD_SN (i.e., onset , increased active period, and decreased anticipatory interval) occurred during the circadian onset.

In most circadian studies, when light is used as the zeitgeber to entrain animals, the transition between light and dark is abrupt. However, studies by Kavanau and colleagues (9, 10, 13) have shown that a gradual transition between the light and dark periods not only facilitates reentrainment to a new LD cycle but also increases the range of entrainment. As a result of these studies, it has been suggested that twilight periods contain important photic temporal information that allows animals to maintain their synchrony with the rhythmically changing environment (2, 9, 10, 13, 14, 19, 20). In addition, results using simulated burrow system studies revealed that Syrian hamsters and flying squirrels maintained in an LD_Sq were exposed to relatively small amounts of light per day (4, 18). DeCoursey (4) further suggested that phase shifts occurred at circadian Act onset during the light-to-dark transition. Because brief light exposures during the day were not considered to have a phase-shifting effect, and animals ceased Act before the beginning of the light period, it was hypothesized that the transition from light to dark may have occurred when the circadian clock was reset daily.

Anticipatory Act and T_b in LD_Sq and LD_SN. This study demonstrates that the animal's Act in LD_Sq rose above the daily mean ~1 h before the onset of dark. This manifests as a significant amount of anticipatory behavior. Peak Act did not occur until after the onset of dark. These results are consistent with light-sampling behavior reported in other studies on nocturnal rodents (4) and with activity rhythms of animals in a simulated burrow system (18). Observations from flying squirrels housed in a simulated den showed a vigorous preening activity lasting ~1 h before their departure from the simulated den (4). The anticipatory behavior, as seen in this study, is internally driven by the circadian pacemaker.

Alternatively, when animals were exposed to LD_SN, the anticipatory Act was significantly attenuated. In some animals, anticipatory Act could not be detected. Kavanau (10) has also observed that white-footed mice housed in LD_Sq cycles warm up to full Act gradually, over a 1- to 3-h period. However, if a dusk period is provided, animals display a burst of high Act in 5-15 min. Shortening of the anticipatory interval could arise as a result of one or more mechanisms. First, the presence of additional temporal cues during the simulated dusk period may make the anticipatory Act unnecessary. Second, LD_Sq may produce a stronger masking effect than LD_SN. In this case, anticipatory Act could be viewed as the result of a light-induced negative masking effect (inhibition) on Act in nocturnal animals. Dawn and dusk transitions may remove such light masking.

T_b also exhibited a significant shortening of the anticipatory rise between LD_Sq and LD_SN. Decrease of anticipatory T_b was mainly a result of the advance of T_bR1. T_bP1 was unaffected by the two different lighting conditions, which is in contrast to the onset of Act_P1.

Examination of the anticipatory behavior during circadian offsets shows an almost identical phase relationship between predictive and reactive offsets for Act and T_b in both lighting conditions. This suggests that neither Act nor T_b anticipates the light transition during dawn. In addition, the fact that neither Act_P2 nor Act_R2 showed difference between the two lighting environments suggests that the circadian timing system could be relatively insensitive to light changes during the dark-to-light transition. It is reasonable to hypothesize that the circadian timing system of hamsters is relatively insensitive to light changes during dawn. We can further suggest that the light changes during dusk and the consequential changes in the circadian onsets are more important for daily resetting of the circadian pacemaker.

Circadian activity in LD_Sq and LD_SN. In addition to attenuation of the anticipatory measures, our results demonstrate that animals exhibited an earlier Act onset and T_b rise under LD_SN. It is possible that during the gradual change in light intensity, a threshold is reached at which circadian changes are initiated, resulting in the phase shift in Act onset and T_b rise. Usui et al. (19, 20) reported that rats exposed to non-square-wave LD cycles initiated circadian activity at a specific light intensity. Moreover, their results suggested that animals have the ability to sense the direction of the light intensity change. For example, in an ascending saw-tooth waveform, the gradual intensity increase was followed by a brief dark pulse. The animal would remain active until the light intensity reached a particular level, and then the activity would abruptly cease. Later, animals initiated circadian Act at the same light intensity level when exposed to a descending saw-tooth waveform. The studies of Kavanau and colleague (11-13) used a twilight zeitgeber to study the illuminance preferences of nocturnal animals, including rodents and primates, and showed that the animals were most active at a specific illuminance level and could detect light changes during twilight periods. Kavanau and Peters (13) raised the possibility of a shift between photopic and scotopic vision during twilight periods that triggers physiological changes. These physiological changes may then result in circadian behavioral changes.

Another possibility is that an integration of photic properties, such as duration and intensity, provides the critical photic signal for entrainment. In a study by Nelson and Takahashi (17), hamsters were most sensitive to photic stimulation of longer duration, with 5 min being the minimum. It is reasonable to hypothesize that the circadian pacemaker, after reaching its photic threshold, integrates photic information during its light-sensitive period and uses this information to initiate circadian onset but is rather insensitive to any brief light intensity fluctuations. Such underlying mechanisms could result in stable and relatively precise circadian onsets in an LD_SN environment.

In this study, hamsters expressed a longer active period and reduced circadian amplitude when exposed to LD_SN. At the LD transition, there was a higher maximum Act level in LD_Sq, which may have been a result of the sudden release from the masking effect of light. The core T_b had already risen to the subjective night level during the T_b anticipatory interval, fully supporting the animal's Act burst. Alternatively, this circadian pattern may result from the absence of temporal cues before the rapid light-to-dark transition, resulting in the animal's heightened anticipation of the onset of dark. Once the LD_Sq dark period began, the animals displayed vigorous locomotor activity. The gradual intensity changes during the dusk-mimicking period in LD_SN could provide additional temporal information, allowing gradual initiation of circadian activity without such anticipation and expenditure of anticipatory metabolic energy.

Interactive relationship between Act and T_b in LD_Sq and LD_SN. In both lighting conditions used in this study, T_b rise always preceded Act onset. Fuller et al. (6) reported that the squirrel monkey produced an anticipatory rise in core T_b by increasing metabolic heat production and reducing peripheral heat loss before dawn. In humans, as reported by Aschoff et al. (1), T_b started to rise ~3 h before waking. In both studies core T_b rise occurred before light onset in LD_Sq. These observations are consistent with our data and suggest that the daily rise in T_b is regulated by the circadian pacemaker and not a result of either the daily onset of locomotor activity or the LD transition.

Difference in photic information between LD_Sq and LD_SN. The analysis of the variance ratio in our study demonstrates differences in the stability of the circadian rhythm waveforms between LD_Sq and LD_SN, further suggesting that these two lighting conditions may have provided different photic information. Although light is accepted as the most potent zeitgeber for entrainment, the masking effect of light itself cannot be excluded when evaluating LD effects on the circadian rhythms. However, the dim light intensity used in this study should have minimized the role of any responses attributable to masking. If masking does play a role in affecting photic entrainment, the difference in variance ratio could imply that the animals expressed stronger masking effects in LD_Sq because of the abrupt changes during LD transitions. In contrast, when LD_SN was provided, the masking effect was not as dominant because of the gradual LD transitions. We postulate that the gradual LD transition may have provided other temporal cues, resulting in changes in circadian rhythms. The lack of or lessened masking effect could explain why animals exhibited an earlier and abrupt circadian onset of activity.

In conclusion, this study shows that the LD_SN can effectively entrain animals and that animals express different entrainment patterns under LD_SN than under LD_Sq. Because the lighting conditions in LD_SN are more similar to that of the natural habitat, these different photic entrainment characteristics may more precisely resemble the profiles of photic entrainment in the natural environment. The circadian activity rhythm is responsive to light intensity changes as well as lighting profile (LD_Sq vs. LD_SN). However, T_b is less sensitive to light intensity and profile than to the duration of light (or dark) period. In the nocturnal hamster, the circadian pacemaker seems to be more sensitive to light changes during dusk and relatively insensitive to light during dawn, suggesting that the daily resetting occurs at dusk during circadian onset.

Perspectives