outgrowth does the attraction begin to However, the ectopic In branches going Received March 23, 1994

influence a fiber's direction of out- on unusual paths straight ahead to

growth? As long as a fiber grows out distant receptor areas (Figs. 3, 5; see
within the ln, it assumedly does not fol- also Fig. 11 in [2], Figs. 2, 3 in [3]) sug-
1. Roth, A.: Naturwissenschaften 73, 264
low the attraction of a certain specific gest that some sort of attraction is in- (1986)
organ area (made free of organs by volved and that in ontogeny all segmen- 2. Roth, A.: J. Comp. Neurol. 328, 473
experimental denervation) or even of a tal In branches (and the vr [3]) are also (1993)
single organ site (as developed in the formed in response to this attraction. 3. Roth, A.: Naturwissenschaften 81, 89
normal maturation of the specimen). There is a further indication for the (1994)
Outgrowing fibers may even enter the vr attraction hypothesis: In the experi- 4. Sperry, R. W.: Proc. Nat. Acad. Sci. USA
or a certain segmental dorsal or ventral ments of Fig. 5A 24 and 18 organs were 50, 703 (1963); Stephenson, R. S.: Exp.
segmental branch without being already induced by vr fibers, having grown to Neurol. 64, 174 (1979); Lance-Jones, C.,
attracted by one free receptor site in the rows IV and V within 4 months. About Landmesser, L.: J. Physiol. (Lond.) 302,
581 (1980); Stuermer, C. A. O.: J. Neu-
special innervation area of this branch. the same number of organs is induced in rosci. 8, 4513 (1988); Kuang, R. Z., Kalil,
They are, rather, directed to the one or unoperated animals by outgrowing vr K.: J. Comp. Neurol. 302, 461 (1990);
other branch by stochastic events occur- fibers in the normal vr-innervated areas Denburg, J. L., Caldwell, R. T.: ibid. M5,
ring within the In at the moment the out- within the same time space. This means 364 (1992)
growing fiber tip passes the bifurcation that all of the new fibers had grown to 5. Diamond, J., Cooper, E., Turner, C.,
site. rows IV and V. Thus they cannot have McIntre, L.: Science 193, 371 (1976); Pur-
Possibly, there is even no attraction at spread unspecifically in any direction. yes, D., Lichtman, J. W.: Principles of
all. Up to now it cannot be excluded that Neural Development. Sunderland, Mass.:
The experiments comply with the "Principles Sinauer 1985
the fibers at first spread out unspecifi-
of animal care", publication No. 85-23, re-
cally in any direction and only the fibers vised 1985 of the National Institutes of Health
arriving at an unoccupied site induce an and also with the laws of the respective coun-
organ and survive, the others becoming try in which the experiments were performed
retracted, detoured, or degenerated. (Deutsches Tierschutzgesetz).

Naturwissenschaften 81, 272-275 (1994) © Springer-Verlag 1994 their reproductive cycles with the envi-
ronment [11, 18].
We looked for activity cycles in another
Sleep Arrhythmia in the Eusocial Naked Mole-Rat social, subterranean rodent, the naked
mole-rat (Heterocephalus gIaber, Bathy-
J. Davis-Walton ergidae). These animals breed during all
Psychology Department, McMaster University, Hamilton, Ontario L8S 4K1, seasons, independent of photoperiod
Canada [3], and matings typically occur among
members of the same colony [25]. Naked
R W. Sherman mole-rats open burrows to the surface
Neurobiology & Behavior, Cornell University, Ithaca, New York 14853, U S A briefly and at irregular intervals; they do
so to "volcano" [5], i.e., to eject
material loosened during tunnel excava-
Most homeotherms exhibit endogenous (Spalax ehrenberghi, Spalacidae) [22], tions. Volcanoing occurs primarily dur-
sleep-wake cycles with circadian periods which are solitary, and Damaraland ing rainy seasons and when the soil sur-
[1]. The internal pacemaker that regu- mole-rats (Cryptomys damarensis, face is cool [4]. Since H. glaber breed
lates these activity rhythms is entrained Bathyergidae) [18], which are colonial, year round and do not appear to syn-
to the 24-h day by "zeitgebers", such as are more active when the lights are on chronize volcanoing primarily with the
light, that reliably indicate the environ- than when they are off and exhibit free- light cycle, the factors hypothesized to
mental state. Most species that have running rhythms in constant darkness. S. maintain circadian rhythmicity in other
been studied are regularly exposed to ehrenbergi have hypertrophied brain subterranean mammals are absent. Thus
the light cycle in nature, and sleep-wake pathways to enable entrainment, we predicted that H. glaber should not
rhythms can be entrained with light although the species is otherwise blind exhibit circadian sleep-wake cycles. This
alone [6, 19]. [9]. In these mole-rats, circadian study is an attempt to test our hypothe-
Recently it was discovered that two sub- rhythmicity may function partly to sis.
terranean mammals, which in nature are ensure that burrows are not opened and Naked mole-rats are endemic to arid
exposed to sunlight only on the rare surface foraging does not occur at times regions of northeastern Africa [16]. They
occasions when they burrow to the sur- when predators are likely to be active are nearly hairless and are poikilo-
face, nonetheless exhibit endogenous [18, 20]. S. ehrenbergi and C. damarensis thermic [7, 14, 30]. Their main living bur-
activity cycles and entrain to light. In the are apparently seasonal breeders, so cir- rows and nests lie so deep underground
laboratory, Mediterranean mole-rats cadian rhythms may also synchronize (->50 cm) that temperature and light
272 Naturwissenschaften 81 (1994) © Springer-Verlag 1994
levels are virtually invariant [4]. Naked that required maintenance of muscle K constant dark
mole-rats are eusocial: they live in large tone. Reduced response to stimuli meant
colonies (mean 75-80/colony; range 2 to that the animal did not respond with
>295), within which only 1 female and voluntary motion to being walked on,
1-3 males reproduce. Like social insect nudged, groomed, or otherwise touched I~ 0 5 .... 1'0 1~- 2~
workers, nonbreeding mole-rats forage, by a colony mate. Animals exhibiting ~15 t
maintain, and defend the colony's exten- REM-associated muscle twitches were ~ 10
sive burrow network [15, 17, 27]. also classified as asleep. ~ f
We studied two captive H. glaber colo- Mole-rats sleep in piles several animals =~
".~, 0 5 10 15 20
nies, designated K (n=19 individuals) deep, making it impossible for a single ,,
and TT (n=28). Both colonies were cap- camera to image all the individuals. Thus 'r- r r c o n s t a n t dark
tured in southeastern Kenya; they were we needed to determine how many ~ 2025
shipped to Cornell University in March mole-rats were missed on the video _~ 15
1981. Each colony was housed in a sepa- tapes. Over a 15-day period, 75 observa- ,m~ 10 5
rate room, within its own artificial tunnel tions of K colony's nest box were made ~ 0 5 10 15 20
system, constructed of 20-25 m of trans- using the monitor, and the number of "~ 2o 25
parent Plexiglas tubing (4.5 cm inside animals visible was recorded. Immedi- 15
diameter) interspersed with clear Plexi- ately after each count, an investigator $ t0
entered the colony room and counted the "~ 05i
glas boxes (20 x 20 × 12 cm; three for
5 10 15 20
K colony, six for TT) which served as number of animals in the nest box. The z
Time [h]
either nesting, food, or toilet chambers results were analyzed for their goodness
[17]. To simulate the mole-rats' natural of fit to several possible models charac- Fig. 1. Number of naked mole-rats exhibiting
environment, colony rooms were main- terizing the discrepancy between video sleep behaviors in two lab colonies over 48 h.
tained at 26-28 °C and 50-65 % relative observations and direct counts. The Colonies had been maintained under constant
humidity. Several lamps equipped with model which assumed that a constant conditions of temperature, humidity, and dim
25-W red bulbs provided localized illu- number of animals was hidden from view red lights for >10 y. For analyses, 100, 1 min
fit the data best (/)<0.05; mean number long sub-samples, selected by a random num-
mination and heat (except during certain
ber generator, were chosen from the video-
experiments). Each mole-rat was tat- of missed animals/observation = 3.5 __+
tapes for each 24-h period in each colony, and
tooed uniquely (on its back) for indi- 2.4). Thus the results should not be the number of mole-rats meeting all three of
vidual identification. biased by our use of videotapes instead our sleep criteria was scored for each sub-
Data on sleep-wake cycles were gathered of direct observations to determine how sample. Data were transcribed from video-
from September 1990 to September many mole-rats were asleep. tapes, and smoothed for presentation using a
1992. Colony members slept primarily in We began our study by videotaping the 3-point moving average
the nest box, although occasionally they animals under the constant temperature,
also basked under the lights. We posi- humidity, and low (red) light conditions
tioned a video camera (Sony AVC-3450) they had experienced in the lab for over 02]- ~-01 ~ I ~ Ob..... d~-
equipped with a wide-angle lens directly a decade. Each colony was videotaped -=. / \ I~-- Expected }

over each colony's nest box so that the for five separate 24-h periods. Analysis ~ 01
entire box was visible. The V C R (Sam- of the tapes indicated that neither colony ~..~00L- ~
sung VR2310) and a television monitor exhibited regular sleep-wake cycles (see
5 10 15 20
were placed outside the colony room, Fig. 1). Interestingly, however, the colo- 0 Number Asleeto
enabling us to change tapes and make nies' sleep-wake patterns suggested
"live" observations without disturbing some synchrony of sleep and wakeful- Fig. 2. Frequency distribution of the number
the animals. Recording was done in real ness (Fig. 2). Analysis of the distribution of naked mole-rats exhibiting sleep behaviors
time; tapes were changed every 6 h. of active mole-rats was significantly non- (O) and the expected frequency with which
We developed a bioassay to identify random (p<0.05), and in a direction that number would have exhibited sleep be-
haviors if individuals' activity cycles were ran-
sleeping mole-rats on the tapes, based indicating that colony mates tend to arise
dom (O). The number of animals awake in
on several behaviors that are consistent and retire simultaneously. each 1-min interval was compared with ran-
and reliable indicators of sleep in other The lack of colony-level circadian dom (i. e., binomial) expectation, using the
mammals [8, 10, 29]. A mole-rat was rhythmicity indicates either that indi- Kolmogorov-Smirnov test for goodness of fit.
considered asleep if it showed (i) lack of vidual mole-rats do not have circadian Analyses were done on data from the K
motion, (ii) reduced muscle tone, and sleep-wake cycles or that cycles of colony colony control and entrainment conditions
(iii) reduced response to stimuli. Lack of mates are out of phase with each other. and the TT colony constant and entrainment
motion meant that the animal made _<1 To discriminate these alternatives, we conditions (see Fig. 4). Each condition was
movement/min, excluding muscle twitch- studied the behavior of (marked) individ- analyzed separately. These data are from the
K colony entrainment condition. The signifi-
es; the animal was considered awake if it uals. Two video recorders were used
cant discrepancy between expectation and
stood or performed any other movement simultaneously, one trained on the nest observation indicates that there is intra-colony
requiring the entire body for its coor- box and the other on the single exit tun- synchrony of sleep and wakefulness. Data
dination. Reduced muscle tone meant nel from the nest box. This enabled us to from the other condition in K colony and both
that the mole-rat was not standing, hold- determine, at any moment, which ani- conditions in TF colony were similarly non-
ing its head up, or resting in a position mals were in the nest, and by elimina- random
Naturwissenschaften 81 (1994) © Springer-Verlag 1994 273
tion, which ones were active in the tun- masking effects because naked mole-rats consecutive 24-h periods. Tapes were
nels. Each colony was taped for three behave abnormally when isolated. analyzed as before (see Fig. 1). Results
consecutive 24-h periods. There was considerable variation in of the entrainment experiment were
Only 2 of the 47 animals (4 %) in our sleep times and their temporal distribu- surprising: even after 25 days of expo-
sample exhibited significant cycles tion among colony mates (Fig. 3), and sure to a 12:12 LD cycle, there was still
(p<0.05), about what would be expected the differences were related to breeding no indication of a sleep-wake rhythm in
by chance (Fig. 3). Despite these results, status and size. Breeding females were either study colony (Fig. 4).
it is conceivable that individual mole-rats asleep during only 3 0 - 4 0 % of the To be sure that the mole-rat's lack of cir-
have endogenous sleep-wake cycles sampled intervals, whereas breeding cadian rhythmicity was not an artifact
which are masked by the presence and males and the larger nonbreeders were induced by building disturbances, we
activities of colony mates. Unfortu- asleep during 45-65 %, and small non- made seismographic recordings while
nately, we were unable to test for such breeders during 50-70 % of the sampled
K colony control
intervals. These results accord with pre-
Time vious reports [23, 24] that the breeding
12:30 00:30 12:30 female is the most active colony mem-
Individuals I,~,,,,~,,,,I,,,~,,wl~lll
1{-3 Day 1
ber. By frequently patrolling the tunnel
system and aggressively shoving certain ~ 05101520253035,4045
(Queen) Day 2
Day 3 "- K colony light entrainment
colony mates, the breeding female en-
I,,,,,,,,,i,l,,,,i,,,,,d sures that colony chores are performed
Day 1 ~ ~o
K-17 Day 2
Day 3 n I I • i HI I • I
around the clock and also maintains her
~ 5
position of reproductive dominance.
.~ o
Day 1 i i i i i To see if naked mole-rats can be E ~ 0 5 10 15 20 25 30 35 40 45
J~ 15
K-F Day 2 ~ii mm n ai I• i entrained to a light cycle, we conducted
Day 3 I I sa li I I
10
a two-part experiment. The first part
12:30 00:30 12:30 served as a control. In the K colony
room a light timer was set for a 12:12 0 5 10 15 20 25 30 35 40 45
T T - + Day 1
(Queen) Day 2 light:dark (LD) cycle, but the lights Time [hi
Day 3 were removed. For 25 consecutive days
IIIIIIIl[lll[lllllllllll| Fig. 4. Sleep-wake patterns of naked mole-rats
Day 1 the animals heard the timer click "on" at
TT-Z Day 2
in a lab colony recorded immediately after 25
Day 3 ~ I B 3 T ~ r ~ 3 1 T ~ I T ~
0300 and "off" at 1500, but experienced days of constant (dim red) light conditions
no change in illumination (i.e., three (control), and 25 days of entrainment to a
illlililillllillllllllll[
Day 1 25-W red bulbs were always on in the 12:12 LD cycle (entrainment). Light bars
TT-17 D a y 2 in uulntmuunl n aim room). K colony was fed once per day, at represent times when lights were on, dark bars
Day3 TT~'I--f~IB~Fg~
randomly determined times, to avoid when they were off. Time 0 is 0300, when the
Fig. 3. The sleep-wake patterns of individual any possibility of food entrainment. On timer simply clicked (control) or the lights
naked mole-rats in two lab colonies. Dark the 26th-30th days K colony was video- actually went on (entrainment). For presenta-
bars represent times when each animal was taped. tion, data were smoothed using a 3-point mov-
awake, light bars when it was asleep. Data ing average. Only data from K colony are
We used TT colony as a control for the
from three different days are shown for each shown. Tapes were analyzed as in Fig. 1, i. e.,
effects of low illumination during the by randomly choosing 100, 1-min sub-samples,
colony. Individuals K-3 (41.0 g) and TT-+
(58.6 g) were the breeding females of their 'dark' phase of the 'light' cycle. We did and for each counting the mole-rats that
respective colonies, K-17 (45.5 g) and K-F this by removing the red lights, so the exhibited sleep behaviors i-iii (see text). Since
(39.8 g) were large nonbreeders in K colony, colony was placed in total darkness. it was impossible to tape T r colony during the
and TT-Z (33.9 g) and TT-17 (24.7 g) were Since it was not possible to tape TT dark periods, only samples from light periods
small nonbreeders in TT colony. Tapes were colony in the dark, we used 4 days of the were scorable (approximately 50 samples per
analyzed by sampling 1 min every 15, and tapes taken under constant dim red illu- 24 h). After the first day of taping, TT colony
identifying all mole-rats that exhibited sleep mination (see Fig. 1) for activity level changed nest boxes, so all tapes for hours
behaviors i-ill. Sleep-wake cycles of each indi- 24-56 were unusable. To compensate, TT was
comparisons.
vidual in each colony were initially analyzed taped for a 5th 24-h period. Numbers of mole-
Following the control taping, the rats that were asleep and awake at randomly
using a chi-square periodogram [28], run on a
Macintosh Plus computer, to determine the entrainment condition began: fight bulbs selected rimes were compared using McNe-
closest fit rhythm. Activity period lengths of were replaced in their sockets and both mar's test [13]. There was no significant differ-
each colony member were checked at 15-min colonies experienced a real 12:12 LD ence in activity patterns between control and
intervals from 20 to 30 h. Data from each indi- cycle, again with feeding times random- light entrainment conditions on any day for
vidual were then analyzed using Rayleigh's ized. During the light phase, light levels either colony (0.80>p>0.20 for all tests).
test [2], based on a 24-h period, because the were 220 lx for K colony and 110 lx for Because the same mole-rat was likely to be
periodogram analysis indicated that 24 h best TT. Similar fight levels were sufficient to counted as asleep in more than one sub-
fit the data. The individuals shown here were sampling period, and since these animals are
entrain activity cycles in Mediterranean
chosen to illustrate intra-colony differences in colonial, our data are not really "independ-
mole-rats [22]. During the dark phase, K ent" samples, as McNemar's test requires.
sleep patterns among mole-rats differing in
body size and breeding status. None of the colony was again dimly illuminated and Strict lack of independence should not bias
animals shown exhibited significant cycles. TT experienced total darkness. the outcome, however, assuming that colony
Note how infrequently the breeding females After 25 days of exposure to the light mates are similarly entrainable (or non-
slept, relative to other colony members cycle, each colony was taped for four entrainable) to light
274 Naturwissensehaften 81 (1994) © Springer-Verlag 1994
simultaneously videotaping TT colony. thus absent. Moreover, it is probably 11. Elliot, J. A., Goldman, B. D., in: Neu-
We used an MEQ-800 single-channel advantageous for colonies to have active roendocrinology of Reproduction, p. 377
analog seismograph with time/amplitu- foragers, maintenance workers, and (N. T. Adler, ed.). New York: Plenum
de/filtering control, 2-Hz geophones, vigilant defenders around the clock, 1981
12. Heffner, R. S., Heffner, H. E.: J. Comp.
and a timing of 1 ppm (gain=60 dB, fil- throughout the year. Although the
Neurol. 331,418 (1993)
ter=10 Hz high frequency). To analyze results of this study do not eliminate the 13. Hinkle, D. E., Wiersma, W., Jurs, S. G.:
the data (n = 107 randomly chosen points possibility that naked mole-rats may Applied Statistics for the Behavioral
over 4 days), we devised a "disturbance have other types of endogenous cycles in Sciences. Boston: Houghton Mifflin 1979
index" ( = number of times seismic dis- physiology (e. g., hormones) or behavior 14. Jarvis, J. U. M.: Bull. Carnegie Mus. Nat.
turbances occurred when "x" mole-rats (e. g., ultradian cycles), the data indicate Hist. 6, 81 (1978)
were asleep/total number of times "x" that H. glaber is the only known wild 15. Jarvis, J. U. M.: Science 212, 571 (1981)
animals were observed sleeping in that mammal that does not regularly exhibit 16. Jarvis, J. U. M., O'Riain, M. J., Bennett,
sub-sample). Spearman rank correla- circadian sleep-wake cycles. N.C., Sherman, R W.: Trends Ecol. Evol.
9, 47 (1994)
tions revealed no significant relation-
17. Lacey, E. A., Sherman, E W., in: The
ships between the number of mole-rats We thank G. D. Harned, C. D. Hop- Biology of the Naked Mole-Rat, p. 275 (R
that were awake and this index in either kins, H. C. Howland, and the Cornell W. Sherman, J. U. M. Jarvis, R. D. Alex-
colony (all p > 0 . 2 ) . Geology Department for equipment ander, eds.). Princeton: Princeton Univ.
Why are the activities of naked mole-rats loans, H. S. Porte for stimulating our Press 1991
apparently independent of ambient light interest in this project, and M. Daly, N. 18. Lovegrove, B. G., Heldmaier, G., Ruf,
cycles? This question can be considered Mrosovsky, B. Rusak, L. Smale, I. T.: S. Afr. J. Zool. 28, 46 (1993)
from both proximate and ultimate per- Zucker, and several anonymous review- 19. Moore-Ede, M., Czeisler, C. M.,
spectives [26]. Regarding the former, the ers for helpful comments on earlier Richardson, G. S.: New Engl. J. Med.
animals do have eyes and can sense light, drafts of the manuscript. 309, 469 (1987)
20. Nevo, E., Guttman, R., Haber, M., Erez,
although their visual acuity is poor [12]. E.: J. Mammal. 63, 453 (1982)
Their pineal gland is atrophic, however, Received August 4, 1993 and March 10, 21. Quay, W. B.: J. Neural Transm. 52, 107
and the smallest in absolute size of any 1994 (1981)
known rodent [21]; neither their melato- 22. Rado, R., Gev, H., Goldman, B. D., Ter-
nin titer nor the neurophysiology of their 1. Aschoff, J. (ed.), in: Handbook of Be- kel, J., in: Photobiology, p. 581 (E. Riklis,
SCN has been investigated. It seems pos- havioral Neurobiology, Vol. 4, p. 81. New ed.). New York: Plenum 1991
sible that naked mole-rats do not possess York: Plenum 1981 23. Reeve, H. K.: Nature 358, 147 (1992)
the necessary physiological mechanisms 2. Batschelet, E.: Circular Statistics in Biol- 24. Reeve, H. K., Sherman, E W., in: The
to enable them to entrain to light. ogy. Toronto: Academic Press 1981 Biology of the Naked Mole-Rat, p. 337 (R
3. Brett, R. A., in: The Biology of the W. Sherman, J. U. M. Jarvis, R. D. Alex-
From an ultimate perspective, naked
Naked Mole-Rat, p. 97 (E W. Sherman, J. ander, eds.). Princeton: Princeton Univ.
mole-rats may have lost any entrainment U. M. Jarvis, R. D. Alexander, eds.). Press 1991
capabilities because, deep under the Princeton: Princeton Univ. Press 1991 25. Reeve, H. K., Westneat, D. E, Noon, W.
ground, light-cycle zeitgebers are neither 4. Brett, R. A.: ibid., p. 137 A., Sherman, E W., Aquadro, C. E: Proc.
readily available nor ecologically rel- 5. Braude, S. H.: ibid., p. 185 Nat. Aead. Sci. 87, 2496 (1990)
evant to them. In nature, naked mole- 6. Bruce, V. G.: Cold Spring Harbor Symp. 26. Sherman, R W.: Anim. Behav. 36, 616
rats forage exclusively below ground, Quant. Biol. 25, 29 (1961) (1988)
breed the year round, mate within the 7. Buffenstein, R., Yahav, S.: J. Thermal 27. Sherman, P. W., Jarvis, J. U. M., Braude,
colony, and open their burrows to eject Biol. 16,227 (1991) S. H.: Sci. Am. 267, 72 (1992)
8. Carroll, D. A., Denenberg, V. H., Tho- 28. Sokolove, E G., Bushell, W. N.: J. Theor.
loose soil only when the surface tem-
man, E. B.: Phys. Beh. 54, 269 (1993) Biol. 72, 131 (1978)
perature is cool and predatory snakes are 9. Cooper, H. M., Herbin, M., Nevo, E.: 29. Van Twyver, H.: Physiol. Behav. 4, 901
least active. The ecological factors Nature 361, 156 (1993) (1969)
thought to maintain circadian rhythmic- 10. Dement, W., Kleitman, N.: EEG C1. 30. Yahav, S., Buffenstein, R., Jarvis, J. U.
ity in other mole-rats [11, 18, 20] are Neut. 9, 673 (1957) M., Mitchell, D.: S. Aft. J. Sci. 85, 340
(1989)

Naturwissenschaften 81 (1994) © Springer-Verlag 1994 275