7 (2006) 563–574

Physiology & Behavior 8

Individual variability and photic entrainment of circadian rhythmsin golden spiny mice

Rotem Cohen, Noga Kronfeld-Schor ⁎

Department of Zoology, Tel Aviv University, Tel Aviv, Israel 69978

Received 15 September 2005; received in revised form 4 December 2005; accepted 6 December 2005

Abstract

Golden spiny mice are diurnally active in most of their natural habitat. Their diurnal activity is ascribed to non-photic cues: competitiveexclusion from the nocturnal niche, or thermoregulatory considerations. Here we studied the entrainment of golden spiny mice to light.

In the laboratory, golden spiny mice were primarily nocturnal and displayed an unusual variety of rhythm patterns, with activity burstsoccurring during both activity and rest periods. Spontaneous shifts of activity rhythms between light phases were sometimes recorded. In all casesbut one, body temperature shifted in parallel with activity. Under DD conditions, the free running period (tau) of all individuals but one wasshorter than 24 h, and in all individuals but the same one it was shorter than tau under LL conditions.

In response to a 6 h phase delay, all individuals entrained to the new LD cycle in a relatively uniform way. During phase advance four out ofthe twelve individuals further delayed their activity and body temperature rhythms, and eight individuals advanced their activity rhythm, but there-entrainment took them over twice as long as to re-entrain to the phase delay.

We suggest that the golden spiny mouse is a nocturnal rodent whose circadian system developed the flexibility to be nocturnal or diurnalaccording to environmental conditions, or a nocturnal rodent in the process of turning diurnal, and that it has low sensitivity to the immediatemasking effect of light on activity.© 2005 Elsevier Inc. All rights reserved.

Keywords: Circadian rhythms; Acomys russatus; Diurnal activity; Masking

1. Introduction

Mammalian diurnality has attracted considerable attentionlately [e.g., 1–5]. It was suggested that since the common an-cestors to all mammals were nocturnal, the circadian system ofpresent-day nocturnal mammals is relatively uniform, while thatof diurnal mammals, which evolved independently at varioustimes [1,6,7] may be diverse [1,6]. To date, however, circadianrhythms have only been studied in a limited number of closelyrelated species [2,3,8], and this is especially true for diurnalspecies [1,9,10], so relatively little information exists on theircircadian system. Consequently, research into the circadian sys-tem of diverse species, and especially diurnal ones, is needed[1–3,6].

The golden spiny mouse (Acomys russatus) offers an inter-esting rodent model for the study of activity patterns and di-

⁎ Corresponding author. Tel.: +972 3 6405740; fax: +972 3 6409403.E-mail address: nogaks@tauex.tau.ac.il (N. Kronfeld-Schor).

0031-9384/$ - see front matter © 2005 Elsevier Inc. All rights reserved.doi:10.1016/j.physbeh.2005.12.010

urnality. This species has a limited distribution in rocky desertsin northeastern Egypt, southern Israel and Jordan [11]. In most ofits natural habitat, the golden spiny mouse is diurnal [12–17]; itis active in winter during the warmer hours of midday, and in thecooler mornings and afternoons during summer [12–17]. How-ever, in northeast Jordan, golden spiny mice were trapped duringthe night [18]. Furthermore, field experiments at Ein Gedi nearthe Dead Sea, have shown that when its nocturnal congener, thecommon spinymouse (A. cahirinus), is removed from the sharedhabitat, the golden spiny mouse shifts from a diurnal to a morenocturnal activity pattern [12,13,19,20]. This may explain whygolden spiny mice in northeast Jordan were trapped during thenight, since no common spiny mice were trapped during thatstudy [18]. These results suggest that interspecific competitionwith common spinymice results in exclusion of the golden spinymice from the nocturnal niche into the diurnal one. Nevertheless,in another habitat, the high mountains of the Sinai desert insouthern Egypt, golden spiny mice are diurnal even though thecommon spiny mice are absent from that area [14]. Ambient

30 40 50 60 70 80 90 100% Activity at night

0

1

2

3

4

5

6

7

8

Num

ber

of in

divi

dual

s

Fig. 1. The distribution of average percentage activity at night of naïve goldenspiny mice during 15 days under 12 :12 LD conditions. N=29.

35

35.5

36

36.5

37

DayNight

Ave

rage

bod

y te

mpe

ratu

re (

C°)

Fig. 2. Average body temperature (±SE) during the light (white) and dark (gray)phase of naïve golden spiny mice during 15 days under 12 :12 LD conditions.N=29.

564 R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

temperatures in this habitat during winter may drop below 0 °C,and the authors ascribed the diurnal activity of the golden spinymouse in this area to thermoregulatory considerations [21]. Insum, golden spiny mice are diurnally active in most of theirnatural habitat (except from one record from Jordan), and theirdiurnal activity is ascribed to competitive exclusion from thenocturnal niche, or to thermoregulatory considerations.

Daily rhythms in mammalian physiology, hormone concen-trations, biochemistry, and behavior are all driven by an internalcircadian clock that is entrained by environmental cues or zeit-gebers [e.g., 22–24]. In mammals, the primary circadian clock islocated in the suprachiasmatic nucleus (SCN) [25]. In theabsence of external zeitgebers, the internal clock free runs, with aperiod (tau) of about 24 h. External cues can either entrain thecircadian clock or mask its effect on behavior. It is still unclearwhether the effect of temperature, as well as that of the commonspiny mouse on the golden spiny mouse activity and bodytemperature rhythms, results from a difference in the way that theclock is entrained to the light dark cycle, a difference in the waythat the clock is coupled to the behavioral rhythm (and otherrhythms that it controls), or masking. Previous studies suggestthat the golden spiny mouse has underlying nocturnal activityand temperature rhythms [17]: immediately upon removal fromits natural habitat at Ein Gedi to the laboratory, golden spinymice displayed nocturnal activity rhythms, or were active bothduring the light and dark periods, but did not show a diurnalactivity pattern [17].We concluded that this immediate inversionof phase preference, without evidence of a phase shift that wouldbe expected in the case of true entrainment [26–28], indicatesthat the diurnal activity of golden spinymice in the field, coupledwith the overt temperature rhythms, is probably not a result of anentrainment process [17]. Furthermore, laboratory experimentson captive golden spiny mice found that under controlled con-ditions, this species is nocturnally active [29]. However, in bothexperiments described above, wheel running was used to mea-sure activity pattern, which in itself can sometimes change andeven invert activity rhythms [5,27,30,31].

Based on the results described above, we hypothesized thatthe diurnal activity of the golden spinymice under controlled LD

conditions will show high variability, and that their circadiansystem will be less responsive than normal to photic inputs. Westudied for the first time the general activity and body tem-perature rhythms under controlled LD conditions, and the effectof photic input on the circadian system of animals from a goldenspiny mouse colony, originating from individuals trapped at EinGedi and bred in our laboratory. We examined entrainment tolight dark (LD) cycle, its free running rhythms under constantlight (LL) and dark (DD) conditions, and responsiveness tophase shifts of the LD cycle.

2. Methods

2.1. Animals and housing

29 golden spiny mice (12 males and 17 females) from ourbreeding colony at the I. Meir Segals Zoological Garden at TelAviv University were used in this study. The source of thebreeding colony (funded in 1995) is Ein Gedi, near the Dead Sea,where the two species (common and golden spiny mice) coexist.The mice were individually housed in 38×24×13 cm plasticcages, under controlled laboratory conditions of LD12:12 (unlessspecified differently) in an isolated–sound proof room, withambient temperature of 29 °C, and food (standard rodent chow)and water ad lib. Once every 2–3 weeks, at random times, weentered the controlled chambers in order to change food andwater.

2.2. Body temperature and activity

Body temperature and activity were continuously recordedusing one of two different systems: an integrated telemetryand data acquisition system (Dataquest III, Data Sciences Inc.,St. Paul, MN, USA, n=17) with a temperature-sensing trans-mitter (TA10TA-F20, Data Sciences, Inc., St. Paul, MN, USA);or a self-combined system including temperature sensitivesingle stage implanted transmitters (Sirtrack, New Zealand) andRX-900 scanner–receiver (Televilt, Sweden) for the measure-ment of core body temperature, and infra-red detectors

1

2

3

4

5

6

7

8

9

10

11

a1

b1

c1

a2

b2

c2

10

12

14

16

18

20

22

24

10

12

14

16

18

20

22

24

0 8 16 0 8 16 0

201 Percentiles: 0.0 0.1 0.8 13.3 19.8 Percentiles: 4.9 5.8 5.8 6.0 6.3201

0 8 16 0 8 16 0

0 8 16 0 8 16 00 8 16 0 8 16 0

0 8 16 0 8 16 0 0 8 16 0 8 16 0

2005 Percentiles: 0.0 0.1 0.1 0.2 0.5 2005 Percentiles: 5.4 5.6 5.7 5.8

10

12

14

16

18

20

22

24

10

12

14

16

18

20

22

24

540 Percentiles: 0.2 2.7 12.0 15.3 18.2 540 Percentiles: 5.8 5.9 6.0 6.2 6.4

2

4

6

8

10

12

Fig. 3. Double daily plots of activity (a1–f1) and body temperature (a2–f2) of naïve golden spiny mice during 15 days under 12 :12 LD conditions. a— Representativeactogram of an individual which was continuously active during the dark phase. b — Representative actogram of an individual that showed bimodal activity pattern,with activity peak occurring either at the beginning and end of the dark phase. c— Representative actogram of an individual that showed bimodal activity pattern withone dominant peak towards the end of the light phase and one towards the dark phase. d— Representative actogram of an individual that showed one peak of activity,at or around the beginning of the dark phase. e— Representative actogram of an individual that was active both during the night and during the day and could not beassigned to any one of the above groups. f — Representative actogram of a diurnal individual that displayed most of its activity during the day, with another peaktowards the end of the night. Black bars at top of plot represent the light schedule.

565R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

d1

e1

f1

d2

e2

f2

3 Percentiles: 0.0 0.1 0.2 0.4 0.8 3 Percentiles: 5.6 5.9 6.0 6.1

0 8 16 0 8 16 0

0 8 16 0 8 16 0

0 8 16 0 8 16 0 0 8 16 0 8 16 0

0 8 16 0 8 16 0

5

10

15

20

8 16 0 8 16 0

5

10

15

20

Percentiles: 0.0 0.1 0.2 0.5 1.1500

10

12

14

16

18

20

22

24

10

12

14

16

18

20

22

24

500 Percentiles: 5.3 5.8 5.9 6.0 6.1

430 Percentiles: 5.6 5.9 5.9 6.0 6.2430 Percentiles: 0.2 0.5 1.8 4.5 8.5

2

4

6

8

10

12

14

1

2

3

4

5

6

7

8

9

10

11

Fig. 3 (continued).

566 R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

(Intrusion detector model MH10, Crow group, Israel) connectedto a PC. Data were collected using Strawberry tree software(n=12). The DSI system allowed continuous measurement ofactivity level and body temperature. However, it demands

operating the animals and implanting them with the transmitters,a procedure that by itself may affect activity rhythms.Furthermore, the number of animals we can test simultaneouslyin this system is limited to 12. The infra-red detectors system is

DD

LL

LD3

LD2

LD1

b

DD

LD3

LD2

LD1

LL

a

1

20

40

60

80

100

120

140

160

180

200

2005 Percentiles: 0.0 0.1 0.1 0.2 0.5

0 8 16 0 8 16 0

2005 Percentiles: 5.4 5.6 5.6 5.7 5.8

0 8 16 0 8 16 0 1

20

40

60

80

100

120

140

160

180

200

Fig. 4. A representative double daily plot of activity (a) and body temperature(b) rhythm of golden spiny mice during LD1, LL, LD2, DD and LD3 conditions.Black bars at top of plot represent the light schedule during the LD periods.

567R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

noninvasive, and allows monitoring of a larger sample size.Nevertheless, it does not provide us with information regardingbody temperature of the experimental animals. Therefore, thetwo systems were used in the study. Comparing the resultsobtained by the two systems, no differences were detected.Therefore, we analyzed the results collected using the twosystems together.

2.3. Animal surgery

Mice were anesthetized with isoflorane in medical grade oxy-gen using an anesthetic machine (Ohmeda, 1.5% vol, 1 L/min)and implanted with either DSI biotelemetry transmitters or thesingle stage implanted transmitters in the abdominal cavity. Boththe abdominal wall and the skin were sutured with absorbablesurgical suture, with cutting needle (5-0 Dexon) and the incisionwas treated with topical antibiotic (silver sulfadiazine 1%;Silverol Cream). Prophylactic antibiotics (Baytril 5% 24 mg/kg)and artificial tear ointment (to prevent desiccation) wereadministered preoperatively. At least two weeks of recoveryfrom surgery was allowed before initiation of data collection.

All procedures were conducted in accordance with and ap-proved by the Institutional Animal Ethics Committee (L-02-45).

2.4. Experimental protocol

Body temperature and activity were monitored continuouslyduring all the experiments. First, all 29miceweremonitored for 15(using the self-build system) or 24 (using the DSI system) daysunder a 12:12 LD regime. Twelve individuals (6 males and6 females) of the 17 that were monitored for 24 days using the DSIsystem, were then placed under constant light (LL) conditions for20 days. At the end of this period, a 12:12LD cyclewas reinstated.After 40 days, the animals were placed in constant darkness (DD)for another 60 days. A 12 : 12 LD cycle was then reinstated foranother 30 days. This was followed by a 6 h phase advance and,24 days later, a 6 h phase delay of the LD cycle.

2.5. Data analysis

Data were collected at 6-min intervals in all experiments,and were analyzed using the Clocklab program (ActimetricsWilmette, IL). We attempted to define activity onset and offseton both the Clocklab program using various settings accordingto published papers (e.g., onset— the first active bout that waspreceded by at least 20 min of inactivity, offset— the last bin ofactivity that was followed by at least 20 min of inactivity [32],onset — 30 min of activity, subsequent to 3 h of inactivity[33]), and by visual inspection, but could not determine eitheronset or offset in most individuals. Therefore, response tophase shifts (phase delay and phase advance) was analyzed,using clear peak of activity in each individual (e.g., Fig. 4). Theday of re-entrainment after a phase advance was defined as theday after the onset of the activity peak no longer advanced, andthe day of re-entrainment after a phase delay was defined as theday after which the onset of the activity peak no longer delayed.Chi-square periodogram was used to calculate the periods offree running rhythms. The Clocklab program was also used togenerate all actograms. Microsoft Excel was used to generatethe figures. Data are presented as means±SEM; statistical testsused include paired t-test, repeated measures, one-way andtwo-way ANOVA, and Scheffe post hoc test (Statistica).

3. Results

3.1. 12 :12 LD cycle

During the first LD period, all individuals (N=29) displayeddaily activity and temperature rhythms that were entrained to the24 h LD cycle. FFT (Fast Fourier Transformation, Clocklab)analysis showed that in addition to a significant 24 h rhythm,some individuals displayed other significant rhythms withperiods such as 12, 6, 8 and 4 h. Activity rhythms were verydiverse, and we therefore classified individuals for which over50% of their activity occurred during the dark phase asnocturnal, and individuals for which 50% of their activityoccurred during the light phase as diurnal. Based on thisclassification, 25 out of the 29 individuals tested were nocturnal,and 4 were diurnal (Fig. 1). Average body temperature of allindividuals was significantly higher (t(26)=−5.16, Pb0.01)during the dark phase (Fig. 2), and increased activity and body

568 R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

temperature were coincident (Fig. 3). No significant differencesin body temperatures between sexes were found, and nointeraction between sex and light phase (2-way ANOVA).Therefore, the results for males and females were combined.

Golden spiny mice showed extremely diverse activity pat-terns. Eight out of the 29 individuals were continuously activeduring the dark phase (e.g., Fig. 3a). Eight showed a bimodalactivity pattern, with activity peak occurring either at the be-ginning and end of the dark phase (e.g., Fig. 3b), or one peaktowards the end of the light phase and one towards the darkphase (e.g., Fig. 3c). Seven individuals had one peak of activity,at or around the beginning of the dark phase (e.g., Fig. 3d), andanother two were active both during the night and during the dayand could not be assigned to any one of the above groups (e.g.,Fig. 3e). The four diurnal individuals displayed most of theiractivity during the day, with another peak towards the end of thenight (e.g., Fig. 5f).

In all individuals, activity bursts were also common duringthe resting period, and in most of them it was impossible to

DD

LD3

DD

LD3

a1 a

b1 b

3004 Percentiles: 0.0 0.1 0.3 0.9 9.9

0 8 16 0 8 16 0

0 8 16 0 8 16 0

120

130

140

150

160

170

180

190

200

120

130

140

150

160

170

180

190

200

315 Percentiles: 0.0 0.1 0.1 0.3 0.8

Fig. 5. Double daily plots of the effect of returning to LD from constant light conditirepresentative golden spiny mouse individual that abruptly shifted its activity and bothat shifted its activity and body temperature through phase transients to the entrainedperiods.

pinpoint activity onset or offset, neither by visual inspection northe Clocklab program (see Methods).

3.2. LL/DD

All individuals (n=12) displayed free running rhythms underLL and DD conditions. Under LL conditions, the free runningperiod (FRP) of all individuals was longer than 24 h (24.47±0.05, range: 24.14–24.73 h, e.g., Fig. 4). Under DD conditions,the free running rhythm period of all individuals but one (female201) was shorter than 24 h (23.72±0.1, range: 23.41–24.66 h,e.g., Fig. 4), and in all individuals but the same one it wasshorter than the FRP of the same individual under LL conditions(t(11)=−5.08, Pb0.001).

Upon returning to LD from constant light conditions (LL orDD) two different patterns were observed: 1) an abrupt shift inactivity and body temperature rhythms (e.g., Fig. 5a1,2), or 2)a shift through phase transients to the entrained rhythm (e.g.,Fig. 5b1,2).

2

2

0 8 16 0 8 16 0

3004

0 8 16 0 8 16 0120

130

140

150

160

170

180

190

200

120

130

140

150

160

170

180

190

200

315 Percentiles: 5.6 5.8 5.9 6.0 6.3

Percentiles: 5.3 5.8 5.9 6.0 6.3

ons (LL or DD) on activity (a1, b1) and body temperature (a2, b2) rhythms of ady temperature rhythms (a), and a representative golden spiny mouse individualrhythm (b). Black bars at top of plot represent the light schedule during the LD

569R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

Percentage of nocturnal activity under LD conditions sig-nificantly decreased during the course of experiments. At thefirst LD phase, 77.4±4.2% of activity occurred during the darkphase, while at the LD phase at the end of the LL (LD2) and DD(LD3) experiments, only 57.25±5.7% and 59.2±5.0% of ac-tivity (respectively) occurred during the dark phase (Fig. 6a,repeated measure ANOVA, F(2,22)=5.54, Pb0.01, ScheffePb0.05). Average body temperature during the night under LDconditions showed the same trend as activity, i.e., decreasedsignificantly from 36.6±0.3 °C during the first LD period to36.1±0.3 and 36.0±0.2 °C during LD2 and LD3 periods, re-spectively (Fig. 6b, repeated measure ANOVA, F(2,22)=6.45,Pb0.05, Scheffe Pb0.05). An opposite trend, although notsignificant, was observed in average body temperature duringthe day (Fig. 6b, repeated measure ANOVA, F(2,22)=1.22,NS). Nevertheless, average body temperature during the nightduring LD1 and LD3 was significantly higher than during theday (LD1: t(11) = 6.35, Pb0.05, LD2: t(11) =0.85, NS, LD3:t(11)=2.21, Pb0.05). During LD2 there was no significant dif-ference between average body temperature during the day andduring the night, although the latter was a bit higher (Fig. 6b).

50

55

60

65

70

75

80

85

LD1 LD2 LD3

% a

ctiv

ity a

t ni

ght

35

35.5

36

36.5

37

LD1 LD2 LD3

Cor

e bo

dy t

empe

ratu

re (

c0 )

Night

Day

*b

a *

Fig. 6. a — Percentage of nocturnal activity under LD1, 2, 3 (12:12 LDconditions). Percentage nocturnal activity significantly decreased during the courseof experiments (repeated measure ANOVA, Scheffe Pb0.05). b— Average bodytemperature during the night under and day under LD1, 2, 3 (12:12 LDconditions). Average body temperature during the night decreased significantlyfrom during the first LD period (repeated measure ANOVA, Scheffe Pb0.05).⁎Significant difference from LD2 and LD3.

During LD2 (after LL), one individual (440) who was noc-turnally active during LD1, turned diurnal when LD was rein-serted after LL, and its body temperature was higher duringthe day. After two weeks, it spontaneously shifted its activityrhythms ca 10 h to a more nocturnal activity (Fig. 7a1). Bodytemperature, that was high during the night during LD1, washigher during the day in LD2, remained high during the daywhen the shift in activity occurred, together with another peak inbody temperature which paralleled the shifting activity (Fig. 7a2).

During LD3, two individuals were diurnally active, but shift-ed their activity to a more nocturnal activity after ca 40 daysunder LD (e.g., Fig. 7b1). In this case, body temperature wascoupled with activity and shifted in parallel (Fig. 7b2). Oneindividual was nocturnal and entrained to the photoperiod dur-ing the first 40 days of LD3, but then started free running again(tau=24.98). After ca 25 days, when activity onset coincidedlights off, it entrained its rhythm once again with the daily pho-toperiod (Fig. 7c1,2). Another individual was unstable duringLD3. It was nocturnally active for ca 20 days, and then shifted tobeing active around lights offset for about 14 days, and thenturned active mainly during the light period with another com-ponent towards light onset.

3.3. Phase shift

Since we were unable to determine activity onset and offset(see Methods), we chose a clear peak of activity, and studied itsadjustment to the phase shift (e.g., Fig. 8a,b). Four individuals(e.g., Fig. 8c,d) delayed their activity as a result of both phasedelay and phase advance, and were analyzed separately. In orderto compare the rest of the individuals (n=8), average time ofpeak activity occurrence during the last 10 days of the LD cyclebefore the shift was set as 0, and the changes in the activityrhythms were described in relation to that time point (Fig. 9). Inthese individuals, when lights were delayed 6 h, activity peakre-entrained after 6.12±0.69 days (range=2–8). When lightswere advanced 6 h, it took activity peak much longer to re-entrain (t(7)=−10.9, Pb0.001): 15.12±0.74 days (range=13–19). Re-entrainment after phase delay was not only faster, butalso smoother and more uniform as compared to that after phaseadvance (Fig. 9).

4. Discussion

We found that under laboratory 12 :12 LD conditions, goldenspiny mice are primarily nocturnal: 25 out of 29 individualsmeasured in this study concentratedmost of their general activityduring the dark period, and their body temperature was sig-nificantly higher during this time. This result is in accord withprevious laboratory experiments that used running wheels tostudy activity rhythms of golden spiny mice, and found thatunder controlled laboratory conditions, they showed activity andoxygen consumption rhythms that are typical for nocturnalmammals [17,29,34,35], even though this species is diurnallyactive in most of its natural habitat [13–17, but see Ref. 18]. Thevariety of rhythm patterns displayed by the golden spiny micewas vast. There were no two individuals with the same activity

b2

a2

c2

LD1

LL

LD2

DD

LD3

DD

LD3

b1

a1

c1

0 8 16 0 8 16 0

0 8 16 0 8 16 0 0 8 16 0 8 16 0

0 8 16 0 8 16 0

Percentiles: 0.0 0.1 0.4 8.2 16.8440440 Percentiles: 5.5 5.8 5.8 5.9 6.2

1

10

20

30

40

50

60

70

80

1

10

20

30

40

50

60

70

80

440 Percentiles: 0.0 0.1 0.4 8.2 16.8

120

130

140

150

160

170

180

190

200

120

130

140

150

160

170

180

190

200

120

130

140

150

160

170

180

190

200

120

130

140

150

160

170

180

190

200

201201 Percentiles: 0.0 0.1 0.8 13.3 19.8 Percentiles: 4.9 5.8 5.8 6.0 6.3

440 Percentiles: 5.5 5.8 5.9 6.2

0 8 16 0 8 16 0

0 8 16 0 8 16 0

Fig. 7. Double daily plots of activity (a1–c1) and body temperature (a2–c2) of individuals that spontaneously shifted their activity and/or body temperature rhythms.a— During LD2 (after LL), individual 440 who was nocturnally active during LD1, turned diurnal when LD was reinserted (a1), and its body temperature was higherduring the day. After two weeks, it spontaneously shifted its activity rhythms ca 10 h to a more nocturnal activity (a1). Body temperature (a2), that was high during thenight during LD1, was higher during the day in LD2, remained high during the day when the shift in activity occurred, together with another peak in body temperaturewhich paralleled the shifting activity. b—During LD3, one individual was diurnally active, but shifted its activity to a more nocturnal activity after ca 40 days under LD(b1). In this case, body temperature was coupled with activity and shifted in parallel (b2). c—One individual was nocturnal and entrained to the photoperiod during thefirst 40 days of LD3, but then started free running again (tau=24.98). After ca 25 days, when activity onset coincided lights off, it entrained its rhythmonce againwith thedaily photoperiod (c1). In this case too, body temperature was coupled with activity and shifted in parallel (c2). Black bars at top of plot represent the light scheduleduring the LD periods.

570 R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

a2

Phase delay

Phase advance

b2b1

a1

Phase delay

Phase advance

190

200

210

220

230

240

250

260

270

3004 Percentiles: 0.0 0.1 0.3 0.9 9.9

0 8 16 0 8 16 0

0 8 16 0 8 16 0

0 8 16 0 8 16 0

0 8 16 0 8 16 0

223 Percentiles: 0.0 0.1 0.1 0.3 0.8

190

200

210

220

230

240

250

260

270

223 Percentiles: 5.7 5.8 5.9 6.0 6.1

3004 Percentiles: 5.3 5.8 5.9 6.0 6.3

180

190

200

210

220

230

240

250

260

270

190

200

210

220

230

240

250

260

270

Fig. 8. Double daily plots of activity (a1, b1) and body temperature (a2, b2) of golden spiny mice during a 6 h phase delay and a 6 h phase advance of the 12 :12 LDcycle. a — A representative individual that delayed its activity and body temperature in response to phase delay and advanced its activity and body temperaturerhythms in response to the phase advance. b— A representative individual that delayed its activity and body temperature in response to phase delay and responded tothe phase advance by further delaying its activity and body temperature rhythms until entrained to the new LD cycle. Black bars at top of the plot represent the lightschedule during the normal LD periods.

571R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

pattern, and all rhythm types were displayed (nocturnal, diurnal,crepuscular, morning types, evening types, etc.). Furthermore, inmost individuals, it was impossible to determine the time ofactivity onset or offset, either by visual inspection or by theClocklab program, and activity bursts were also commonlydisplayed during the rest period. To the best of our knowledge,such extreme variation in activity rhythms in a single species israre. Usually, even in species that exhibit both diurnal andnocturnal patterns, distinct activity patterns are described, andsome features are common for the species [but see Refs. 10,36].For example, in the Nile grass rat, Arvicanthis niloticus, wheel-running activity of both diurnal and nocturnal individuals hadbimodal activity rhythms, with peaks of activity occurring at thebeginning and end of the active phase under LD, LL and DDconditions [32]. Furthermore, without a running wheel, generalactivity in this species is always diurnal [30]. Even in the degu(Octodon degus), which displays distinct inter-individual varia-tion in the timing and expression of circadian activity and bodytemperature rhythms, three chronotypes (morning, intermediate[crepuscular], and evening) were described [37]. Nocturnal ac-tivity was also observed in this species when exposed to a

running wheel [27]. Furthermore, on all occasions, prominentcrepuscular episodes of behavioral activity and body temperaturewhile entrained to a LD cycle were exhibited, with clear onsetand offset of activity [27,37]. Similarly high inter-individualvariability in circadian activity patterns was previously observedin subterranean rodents that are subjected to relatively moreconstant habitat characteristics than rodents that live above ground[e.g., 38–43]. It was suggested that in these species, the circadianclock is not very strongly coupled to the photic input [42], and thatnon-photic cues may play an important role. Based on the highindividual variation of the circadian rhythms of golden spiny micefound in the current study, and on the well documented effect ofnon-photic cues on their activity pattern in the field [12,13,19–21], we suggest that similar to subterranean species, non-photiccues may play an important role in the determination of activitypatterns in this species.

This suggestion was further supported by other results ob-tained in the current study. As stated before, activity onset andoffset were difficult to determine. Under LD conditions, allindividuals showed activity bursts during the resting period,and were not strictly diurnal or nocturnal. During the different

-5 0 5 10 15

1

5

9

13

17

21

25

29

33

37

41

45

49

53

57

61

Phase delay

Phase advance

Fig. 9. A schematic representation of peak activity re-entrainment to the 6 h phasedelay and a 6 h phase advance of the 12 :12 LD cycle of the 8 individuals thatresponded as expected to the shifts in the light schedule (see Fig. 8).

572 R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

LD phases of the experiments, spontaneous shifts of activityrhythms from a more nocturnal to a more diurnal activity or viseversa were recorded in several individuals. It is impossible to sayfrom the current results whether these shifts result from a shift inthe clock, splitting, or masking, and this point deserves furtherstudy. Following constant LL, individuals significantly reducedtheir activity level during the dark period by 20% (from anaverage of 77% to 57%). Exposure to DD conditions did notchange that pattern, and the activity level during the dark periodupon their return to LD conditions remained 57%. The decreasein nocturnal activity during the course of the experiments wasaccompanied by a parallel decrease in average body temperatureduring the night, and an increase in body temperature during thelight period.

In response to a 6 h phase delay, all individuals entrainedto the new LD cycle in a relatively uniform way, and within thetime frame reported for other species [e.g., 32,44]. However,during phase advance two interesting phenomena were ob-served: First, four out of the twelve individuals responded byfurther delaying their activity and body temperature rhythmsuntil they were entrained to the new LD cycle. Second, eight outof the twelve individuals advanced their activity rhythm asexpected, but the re-entrainment took them over twice as longas to re-entrain to the phase delay (6.12 vs. 15.12 days). Re-entrainment to phase advance taking much longer than to phasedelay was previously observed in other species [e.g., 45,46].

Since golden spiny mice are able to be active either diurnallyor nocturnally in their natural habitat, it is possible that the im-mediate masking effects of light or darkness on behavior [28] arenot strong in this species. In accord, it was previously suggestedthat changes in masking responses to light might be an essential

and integral component of switching between nocturnal anddiurnal activity [5]. This possible change in the masking re-sponse to light or dark may partially explain the difficulty indetermining activity onset and offset, the spontaneous shifts, andother observations of the current study. These results also sup-port our notion that non-photic cues play an important role in thedetermination of activity rhythms in this species.

In 1960, Aschoff suggested that nocturnal animals have taushorter than 24 h under DD conditions, and that tau will lengthenunder LL conditions. Diurnal animals, he further suggested,have tau longer than 24 h under DD conditions, and tau willshorten under LL conditions. However, it is now clear that al-though the rules generally hold for nocturnal species, diurnalspecies are much more variable [1,6]. In the current study wefound that golden spiny mice had tau longer than 24 h under LLconditions. Under DD conditions, tau of all individuals but onewas shorter than 24 h, and in all individuals but the same one itwas shorter than tau of the same individual under LL conditions.Therefore, golden spiny mice generally fit the nocturnal patternin this respect too, and provide yet another example of thediversity of the circadian system in diurnal mammals.

Almost all species can be classified as nocturnal, diurnal orcrepuscular, and each activity pattern is accompanied by phys-iological, behavioral and anatomical adaptations [47,48]. As aresult, evolutionary transitions between nocturnal and diurnallifestyles have been relatively infrequent, because they wouldinvolve a coordinated suite of changes not only in the temporalorganization of a wide range of behavioral patterns and phys-iological processes, but also in other physiological and anatom-ical adaptations for nocturnality or diurnality [8,48]. Hence,closely related species are generally active during the same partof the diel cycle [day or night, 7,47]. All species in the genusAcomys except the golden spiny mouse are nocturnal. While thegolden spiny mouse has evolved some adaptations to diurnalactivity such as dark skin pigmentation and a high concentrationof ascorbic acid in its eyes [49], both protective against the highultraviolet radiation during the day [49–52], it has also retainedtraits that would typify a nocturnal mammal: it has a similarpotential for non-shivering thermogenesis (NST) as does its noc-turnal congener, the common spiny mouse [53], which is exposedin winter to much lower ambient temperatures, and expends moreenergy on thermoregulation [53,54], suggesting that in terms ofNST the golden spiny mouse still displays its ancestry as a noc-turnal rodent. It also retains the retinal structure of a nocturnalmammal, i.e., its retina is composed of rod photoreceptors, and hasnot evolved to meet with their needs as a diurnal species [55].Moreover, our field studies suggest that in its natural habitat thegolden spiny mouse has underlying nocturnal activity and tem-perature rhythms [8,16,54], but is also able to be diurnally active.Based on the results of this and previous studies, we suggest thatthe circadian system of golden spiny mice is very flexible. It isoriginally a nocturnal rodent that has developed the ability to benocturnally or diurnally active, according to the environmentalconditions, or a nocturnal animal during evolutionary transitionsbetween nocturnal and diurnal lifestyles. This transition betweennocturnal and diurnal ways of life, infers that the golden spinymouse has a reduced response to the immediate masking effect of

573R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

light and/or dark on activity. This hypothesis should be furtherstudied.

The high flexibility and variability of the circadian rhythms ofthe golden spiny mice may enable it to better exploit its habitat,switching between diurnal and nocturnal activity patterns inorder to optimize its niche in the temporal axis [56].

The physiological adaptations of the golden spiny mouse toboth diurnal and nocturnal activity, or its ability to switch from anocturnal to a diurnal activity pattern, can be explained in thelight of its zoogeographical history. Shkolnik (1966) suggestedthat since the species Acomys was originally a savanna rodent oftropical Africa [57], it developed adaptations to high ambienttemperatures that preceded its adaptation to water shortage.These adaptations to heat demanded high rates of evaporativewater loss for thermoregulation [58]. Later on, when the spe-cies expanded its range to dry deserts, it developed adaptationsto water shortage, manifested in a high ability to conserve waterin the kidney [58]. We suggest that these adaptations enabledthe golden spiny mouse to switch its activity to the more ther-moregulatory challenging diurnal niche upon encounteringunfavorable conditions during the night (i.e., low ambient tem-peratures in the Sinai desert, competition with the common spinymouse in the Judean desert).

The inability to assign the golden spiny mouse to a specificactivity phase was reinforced in the current study. We found ahigh variation in activity and body temperature rhythms, andconcluded that as far as rhythm characteristics are concerned, itis unclear whether the golden spinymouse should be categorizedas diurnal or nocturnal rodent.We also suggest that this species isnot very responsive to the immediate masking effect of light ordark, and that during its zoogeographical history, it also selec-tively developed plasticity in the mechanisms that determinedaily activity patterns. This finding also reinforces the hypo-thesis that, since the common ancestors to all mammals werenocturnal, the circadian system of present-day nocturnal mam-mals is relatively fixed, while that of diurnal mammals, whichevolved independently at various times [1,6,7], is more likely tobe flexible [1,6].

Acknowledgments

We wish to dedicate this paper to the late Prof. A. Shkolnik,whose insights and support keep inspiring our work. We thankProf. Laura Smale and Prof. Tamar Dayan for their insightfulcomments on a previous version of this MS and Nomi Paz forediting the MS. This research was supported by a BSF grant(2003048).

References

[1] Smale L, Lee T, Nunez AA. Mammalian diurnality: some facts and gaps.J Biol Rhythms 2003;18:356–66.

[2] Lee MT. Growing evidence that some aspects of SCN function differ innocturnal and diurnal rodents. Am J Physiol 2004;286:R814–5.

[3] Dardente H, Menet JS, Challet E, Tournier BB, Pevet P, Masson-Pevet M.Daily and circadian expression of neuropeptides in the suprachiasmaticnuclei of nocturnal and diurnal rodents. Brain Res Mol Brain Res2004;124:143–51.

[4] Novak CM, Albers HE. Novel phase effects of GABAA receptoractivation in the suprachiasmatic nucleus of a diurnal rodent. Am JPhysiol 2004;286:R820–5.

[5] Redlin U, Mrosovsky N. Nocturnal activity in a diurnal rodent (Arvicanthisniloticus): the importance of masking. J Biol Rhythms 2004;19:58–67.

[6] Johnson CH, Elliot J, Foster R, Honma KI, Kronauer R. Fundamentalproperties of circadian rhythms. In: Dunlap JC, Loros JJ, DeCoursey PJ,editors. Chornobiology, biological timekeeping. Sunderland, MA: SinauerAssociates Inc., Publishers; 2004. p. 27–65.

[7] Roll U, Dayan T, Kronfeld-Schor N. On the role of phylogeny indetermining activity patterns of rodents. Evol Ecol (accepted forpublication).

[8] Kronfeld-Schor N, Dayan T. Partitioning of time as an ecological resource.Ann Rev Ecolog Evol Syst 2003;34:153–81.

[9] Garcia-Allegue R, Lax P, Madariaga AM, Madrid JA. Locomotor andfeeding activity rhythms in a light-entrained diurnal rodent, Octodondegus. Am J Physiol 1999;277:R523–31.

[10] Challet E, Pitrosky B, Sicard B, Malan A, Pevet P. Circadian organizationin a diurnal rodent, Arvicanthis ansorgei Thomas 1910: Chronotypes,responses to constant lighting conditions, and photoperiodic changes.J Biol Rhythms 2002;17:52–64, 19:58–67.

[11] Mendelssohn H, Yom-Tov Y. Mammalia of Israel. The Israel Academy ofSciences and Humanities, Jerusalem, Israel; 1999.

[12] Shkolnik, A. Studies in the comparative biology of Israel's two species ofspiny mice (genus Acomys). PhD thesis (in Hebrew, English summary)Hebrew Univ., Jerusalem; 1966.

[13] Shkolnik A. Diurnal activity in a small desert rodent. Int J Biometeorol1971;15:115–20.

[14] Haim A, Borut A. Size and activity of cold resistant population of thegolden spiny mouse (Acomys russatus: Muridae). Mammalia 1975;39:605–11.

[15] Kronfeld N, Dayan T, Zisapel N, Haim A. Co-existing populations ofAcomys cahirinus and A. russatus: a preliminary report. Isr J Zool1994;40:177–83.

[16] Elvert R, Kronfeld N, Dayan T, Haim A, Zisapel N, Heldmaier G.Telemetric field studies of body temperature and activity rhythms ofAcomys russatus and A. cahirinus in the Judean Desert of Israel. Oecologia1999;119:482–92.

[17] Kronfeld-Schor N, Dayan T, Elvert R, Haim A, Zisapel N, Heldmaier G.On the use of the time axis for ecological separation: diel rhythms as anevolutionary constraint. Am Nat 2001;158:451–7.

[18] Scott D, Dunstone N. Environmental determinants of desert-living rodentcommunity structure in the north-east Badia region of Jordan. J Zool Lond2000;251:481–94.

[19] Shargal E. Population biology and ecophysiology of coexisting Acomyscahirinus and Acomys russatus. MSc thesis, Tel Aviv University, Israel (inHebrew, English summary) 1997.

[20] Gutman R, Dayan T. Temporal partitioning between spiny mouse species:an experimental field study. Ecology 2005;86(1):164–73.

[21] Haim A, Borut A. Thermoregulation and nonshivering thermogenesis asfactors limiting distribution of golden spiny mouse (Acomys russatus). Isr JMed Sci 1976;12:896.

[22] Heldmaier G, Steinlechner S, Ruf T, Wiesinger H, Klingenspor M.Photoperiod and thermoregulation in vertebrates — body-temperaturerhythms and thermogenic acclimation. J Biol Rhythms 1989;4:251–65.

[23] Refinetti R, Menaker M. The circadian-rhythm of body-temperature.Physiol Behav 1992;51:613–37.

[24] Mistlberger RE, Skene DJ. Social influences on mammalian circadianrhythms: animal and human studies. Biol Rev Camb Philos Soc2004;79:533–56.

[25] Klein DC, Moore RY, Reppert SM. Suprachiasmatic nucleus: the mind'sclock. New York: Oxford University Press; 1991.

[26] Deacon S, Arendt J. Adapting to phase shifts, I. An experimental model forjet lag and shift work. Physiol Behav 1996;59:665–73.

[27] Kas MJH, Edgar DM. A nonphotic stimulus inverts the diurnal–nocturnalphase preference in Octodon degus. J Neurosci 1999;19:328–33.

[28] Mrosovsky N. Masking: history, definition, and measurement. ChronobiolInt 1999;16(4):415–29.

574 R. Cohen, N. Kronfeld-Schor / Physiology & Behavior 87 (2006) 563–574

[29] Friedman D, Haim A, Zisapel N. Temporal segregation in coexisting spinymice (genus Acomys): role of photoperiod and heterospecific odor. PhysiolBehav 1997;62:407–11.

[30] Blanchong JA, McElhinny TL, Mahoney MM, Smale L. Nocturnal anddiurnal rhythms in the unstriped Nile rat, Arvicanthis niloticus. J BiolRhythms 1999;14:364–77.

[31] Mrosovsky N. Locomotor activity and non-photic influences on circadianclocks. Biol Rev 1996;71:343–72.

[32] Katona C, Smale L. Wheel-running rhythms in Arvicanthis niloticus.Physiol Behav 1997;61:365–72.

[33] Mohawk JA, Lee TM. Restraint stress delays reentrainment in male andfemale diurnal and nocturnal rodents. J Biol Rhythms 2005;20:245–56.

[34] Fluxman S, Haim A. Daily rhythms of body temperature in Acomysrussatus — the response to chemical signals released by Acomyscahirinus. Chronobiol Int 1992;110:159–64.

[35] Rubal A, Choshniak I, Haim A. Daily rhythms of metabolic-rate and body-temperature of 2 murids from extremely different habitats. Chronobiol Int1992;9:341–9.

[36] Johnston PG, Zucker I. Lability and diversity of circadian-rhythms ofcotton rats Sigmodon hispidus. Am J Physiol 1983;244:R338–46.

[37] Labyak SE, Lee TM, Goel N. Rhythm chronotypes in a diurnal rodent,Octodon degus. Am J Physiol 1997;273:R1058–66.

[38] Ben-Shlomo R, Ritte U, Nevo E. Activity pattern and rhythm in thesubterranean mole-rat superspecies Spalax eherenbergi. Behav Genet1995;25:239–45.

[39] Goldman GD, Goldman SL, Riccio AP, Terkel J. Circadian patterns oflocomotor activity and body temperature in blind mole-rats, Spalaxehrenbergi. J Biol Rhythms 1997;12:348–61.

[40] Riccio AP, Goldman BD. Circadian rhythms of locomotor activity in nakedmole-rats (Heterochephalus glaber). Physiol Behav 2000;71:1–13.

[41] Begall S, Daan S, Burda H, Overkamp GJF. Activity patterns in asubterranean social rodent, Spalacopus cyanus (Octodontidae). J Mammal2002;83:153–8.

[42] Oosthuizen MK, Cooper HM, Bennet N. Circadian rhythms of locomotoractivity in solitary and social species of African mole-rats (family:Bathyergidae). J Biol Rhythms 2003;18:481–90.

[43] Hart L, Bennett NC, Malpaux B, Chimimba CT, Oosthuizen MK. Thechronobiology of the Natal mole-rat, Cryptomys hottentotus natalesis.Physiol Behav 2004;82:563–9.

[44] Weinert D, Sturm J, Waterhouse J. Different behavior of the circadianrhythms of activity and body temperature during resynchronizationfollowing an advance of the LD cycle. Biol Rhythm Res 2002;33:187–97.

[45] Aschoff J, Hoffmann K, Pohl H, Wever R. Re-entrainment of circadianrhythms after phase-shifts of the Zeitgeber. Chronobiologia 1975;2:23–78.

[46] Ruby NF, Heller HC. Phase shift magnitude and direction determinewhether Siberian hamsters reentrain to the photocycle. J Biol Rhythms1998;13:506–17.

[47] Daan S. Adaptive daily strategies in behavior. In: Aschoff J, editor.Handbook of behavioral neurobiology: biological rhythms, vol. 4. NewYork: Plenum Press; 1981. p. 275–98.

[48] DeCoursey PJ. The behavioural ecology and evolution of biological timingsystems. In: Dunlap JC, Loros JJ, DeCoursey PJ, editors. Chornobiology,biological timekeeping. Sunderland, MA: Sinauer Associates Inc.,Publishers; 2004. p. 27–65.

[49] Koskela TK, Reiss GR, Brubaker RF, Ellefson RD. Is the highconcentration of ascorbic acid in the eye an adaptation to intense solarirradiation? Invest Ophthalmol Vis Sci 1989;30:2265–7.

[50] Chaplin G. Geographic distribution of environmental factors influencinghuman skin coloration. Am J Phys Anthropol 2004;125:292–302.

[51] Jablonski NG. The evolution of human skin and skin color. Annu RevAnthropol 2004;33:585–623.

[52] Diamond J. Geography and skin color. Nature 2005;435:283–4.[53] Kronfeld-Schor N, Haim A, Dayan T, Zisapel N, Klingenspor M,

Heldmaier G. Seasonal thermogenic acclimation of diurnally andnocturnally active desert spiny mice. Physiol Biochem Zool 2000;73:37–44.

[54] Kronfeld-Schor N, Shargal E, Haim A, Dayan A, Zisapel T, Heldmaier N.Temporal partitioning among diurnally and nocturnally active desert spinymice: energy and water turnover costs. J Therm Biol 2001;26:139–42.

[55] Kronfeld-Schor N, Dayan T, Jones ME, Kremer I, Mandelik Y, WollbergM, et al. Retinal structure and foraging microhabitat use of the goldenspiny mouse. J Mammal 2001;82:1016–25.

[56] Kronfeld-Schor N, Dayan T. Partitioning of time as an ecological resource.Ann Rev Ecolog Evol Syst 2003;34:153–81.

[57] Denys C, Gautun JC, Tranier M, Volobouev V. Systematics of the spinymice (Acomys: Muroidae): molecular and biochemical patterns. Isr J Zool1994;40:215–46.

[58] Shkolnik A, Borut A. Temperature and water relations in two species ofspiny mice (Acomys). J Mammal 1969;50:245–55.