Title: Orientation of the gopher tortoise, Gopherus polyphemus (Daudin)
Full Citation
Permanent Link: http://ufdc.ufl.edu/UF00097759/00001
 Material Information
Title: Orientation of the gopher tortoise, Gopherus polyphemus (Daudin)
Physical Description: Book
Language: English
Creator: Gourley, Eugene Vincent, 1940-
Publisher: s.n.
Copyright Date: 1969
Genre: bibliography   ( marcgt )
non-fiction   ( marcgt )
 Record Information
Bibliographic ID: UF00097759
Volume ID: VID00001
Source Institution: University of Florida
Holding Location: University of Florida
Rights Management: All rights reserved by the source institution and holding location.
Resource Identifier: alephbibnum - 000985821
oclc - 17625109
notis - AEW2234


This item has the following downloads:

PDF ( 5 MBs ) ( PDF )

Full Text


Gopherus polyphemus (DAUDIN)






I am very grateful for t4L guidance and inspiration of

my chairman, Dr. Archie Carr.

Thanks are also due to the other members of my super-

visory committee, Dr. E.G.F. Sauer, Dr. Walter Auffenberg,

Dr. Brian McNab, and Dr. Donald Dewsbury, for their ideas

and criticism of the manuscript.

I am indebted to Dr. David Johnston, Dr. Frank flordlie,

and Dr. John Kaufmann, for the use of their equipment. I

would also like to thank the other faculty members and the

many graduate students who assisted with various phases of

this study.

My thanks to Mr. Paul Laessle for his advise on the

figures, to Mr.Kenneth Campbell for printing the photographs,

to Mr. Charles Harper and Miss Mary Glenn for proofreading

portions of the manuscript, and to Mrs. Mazelle Shadburn

for typing it.

This study was supported by the Zoology Department,

the Department of Biological Sciences, and the graduate

school of the University of Florida. Support was also pro-

vided by grants (NONR-580(12)) and (N.S.F. GB 3910) to

Dr. Archie Carr, Principal Investigator.



ACKNOULEDGIENTS. . . . . . . . . ... ii

LIST OF TABLES . . . . . . . . ... iv

LIST OF FIGURES. . . . . . . . . ... vi


I. Introduction . . . . . . . 1

II. Home Range . . . . . . . . 4

III. Short-Range Orientation. . . . ... .16

IV. Biological Cloc . . . . . .. 37

V. Lcng-Range Orientation . . . ... .64

VI. Summary . . . . . . . . .. 115

LITERATURE CITED . . . . . . . .. .. 119

BIOGRAPHICAL SKETCH. . . . . . . . ... 126


Table Page

1. Preference of tortoises to turn onto a straight
trail in the same direction in which they had
been forced previously on the straight trail
(positive choice). The data represents the
number of positive choices out of ten. ... 26

2. Comparison of preferences of tortoises to turn
onto a straight trail in the same direction that
they had been forced previously with nares
blocked with wax and unblocked . . ... .27

3. Preferences of tortoises to turn in the same
direction in which they had previously been
forced to turn on Y-shaped trails (positive
choice). The ranked data represent the
number of positive choices out of ten trails
on ten separate Y-trails per tortoise. .. 29

4. Results of tests in the T-maze. Positive refers
to the trials in which the tortoise chose the arm
leading to the home box (p. 23). Negative refers
to the trials in which the tortoise chose the other
arm. Except for the choices, the numbers repre-
sent the average number of responses per trial.
Probabilities refer to the expectation that the
numbers of responses for positive and negative
choices are equal. . . . ........ .30

5. Comparison of length of circadian period under
constant light based on time of onset of activity
by the two methods of mean-period length by the
classical method and by linear regression (see
p. 59) . . . . . . . . . . 55

6. Mean preferred directions; concentration around
mean direction, rc, and z statistic for signifi-
cance of tortoises tested in the arena through-
out the day. . . . . . . . ... .88

LIST OF TABLES (Continued)

Table Page

7. Comparison of preferred directions in open field
and arena tests. . . . . . . . . 90

8. Mean preferred directions; concentration around
mean direction, rc; and z statistic for signifi-
cance of tortoises tested in the arena at three
different times of day ............. 91



Figure Page

1. Relations of the size of tortoises and the maxi-
mum distances displaced from home burrows which
resulted in direct return. Line is least-squares
regression. . . . ... .. . ....... 8

2. Relations of the size of tortoises and the dis-
tance between the home burrow and the most dis-
tant other burrow (p. 9). Line is least-
squares regression . . . . . . ... 11

3. Activity box used for determining locomotor
activity patterns of tortoises. A, panel upon
which food and water was supplied; B, panels
which activated micro-switches; C, removable
cover; D, -ooden partition. . . . . ... 42

4. Mean per cent of daily activity each hour under
approximately 12 hours of dark alternating with
12 hours of light. Light on at approximately
0700 and off at approximately 1900. . . .. .46

5. The effect on the activity pattern of shifting
the phase of the artificial light-dark condi-
tions 6 hours. Vertical bars represent periods
during which the tortoises were active; shaded
areas, the periods of imposed darkness. To more
clearly illustrate the rhythmicity, the succeed-
ing day is shown above the numbered day. A shows
a retarded shift; B shows an advanced shift . 48

6. The effect on the activity pattern of a tor-
toise shifting the phase of the artificial
light-dark conditions 12 hours. Data presented
as explained in Figure 5. . . . . . ... 50


Figure Page

7. Activity patterns of tortoises under alternat-
ing light-dark conditions and under constant
light. Data presented as explained in Figure 5 52

8. The effect of temperature on the onset of activ-
ity under constant light. Solid line, repre-
senting onset of activity, to show parallel;
pattern; dashed line represents temperature. 57

9. Arena used for testing tortoise orientation
in the absence of landmarks. A, dimensions;
B, arena set up in field . . . . . 74

10. Preferred directions of travel shown by indivi-
dual tortoises taken from different geographic
populations. Tests were performed in an open
field. Arrows inside circles indicate mean
preferred directions; length of arrow indicates
strength of the concentration around the mean
direction, r. Then the length of the arrow
equals the radius perfect correlation of preferred
directions among trials is indicated ..... 82

11. Preferred directions of travel shown by indivi-
dual tortoises taken from the same geographic
population. Data presented as explained in
Figure 10. . . . . . . . . . 84

12. Lengths of paths travelled by tortoises and the
straight-line distances from starting point to
end point of each run. Trials were performed
in an open field. Points represent the mean
values obtained for an individual. Line is
least-squares regression . . . . .. 36

13. Comparison of original preferred directions of
tortoises in the arena during the morning testing
period and the preferred directions after the tor-
toises had been subjected to a phase-shift 6 hours
advanced. Open circles, original directions;


Figure Page

13. solid circles, directions after phase-shifting.
Solid arrows indicate original mean preferred
directions; dashed arrows indicate shifted mean
preferred directions; arrow lengths indicate re
values . . . . . . . .. . 94

14. Comparison of original preferred directions of
tortoises in the arena during the noon testing
period and the preferred directions after the
tortoises had been subjected to an advanced
phase-shift of 6 hours. Data presented as
explained for Figure 13.. . . . . .. 96





Animal orientation may be defined as the ability of animals

to find their way. Griffin (1952) classified it as follows:

landmark orientation: reliance upon landmarks in or near

familiar territory; compass orientation: maintenance of a cer-

tain direction by reference to celestial bodies; and navigation:

ability to choose the correct homeward direction in unfamiliar

territory. It is well 1knom that celestial cues provide the

information to an animal for maintenance of the direction in

the second case, and there is good evidence that the third

process occurs (Schmidt-Koenig, 1965). Thus it is obvious

that there must exist some ability to compensate for the change

in azimuth of these celestial bodies. This time-compensation

is possible because animals have internal clocks (Hoffnan, 1965).

Although much work has been done on orientation in animals

(Cold Spring Harbor Symposia XXV, 1960; Ergeb. der Biol.

XXVI, 1963; Animal Orientation and Navigation, 1967), many

problems, particularly in long-range orientation, remain unre-

solved. One of these is the question of the distribution and

character of celestial orientation among species, genera, and

classes of vertebrates. Much evidence has been obtained with

arthropods and birds, but little has been reported for the

lower vertebrate classes.

There are considerable data that turtles are capable

of landmark orientation (Ortleb and Sexton, 1964). There are

also some reports that turtles are capable of compass orienta-

tion and perhaps of navigation by means of celestial cues

(Gould, 1957, 1959; Fischer, 1964). However, I1nlen (1965)

could find no evidence to support either of these in Chrysemys

picta, one of the species which Could (1959) reported to be

capable of navigation.

A terrestrial turtle that seems to be appropriate for

orientation study is the gopher tortoise, Go perus polyphemus,

found in the southeastern United States. It makes burrows in

which it dwells. These burrows provide relatively constant

conditions of temperature and humidity, and are goals to which

the tortoise regularly returns after foraging. Because these

goals are small targets, the tortoises must have a precise

means of finding them. This situation does not hold for other

terrestrial or pond turtles, in which the homing drive may be

only to regain contact with a somewhat diffused '.ome range.

Other advantages offered by the gopher tortoise as a subject

for orientation studies are that it is slow-moving, and thus

easier to track accurately; and that it is diurnal, which

permits the elimination of nocturnal celestial cues from


The aims of the present study have been as follows:

1. To establish the size of the home range in a

population, in order to restrict the field of

search for the cues which the tortoises may use

for short-range orientation.

2. To determine some of the short-range cues used.

3. To determine the presence and accuracy of the

internal clocks of the gopher tortoise.

4. To determine the existence and nature of any

celestial orientation mechanism in this species.



There are several reasons for determining the home

range of a population of gopher tortoises prior to studying

the orientation mechanism itself. If it is found that the

animal moves only short distances in its daily or seasonal

wanderings then cues that are effective only over great dis-

tances can be discounted and attention focused upon the

local cues. Second, if the home range can be clearly de-

teLrmined, then the set of stimuli p.ctentially available to

the tortoise for orientation can better be determined.

Third, allowance can clearly be made for the fact that the

tortoise may use more complex forms of orientation outside

its home range but may rely upon simpler forms of orientation

within its home range.

Home range is defined by Durt (1943) as "the area,

usually around a home site, over which the animal normally

travels in search of food." Homing cues are probably more

plentiful inside the home range than outside it. In any case,

successful return to the burrow must require fairly specific

guidance information.


A common method of determining the limits of a home

range is to mark and release the animal, then to recapture

it repeatedly. Because the gopher tortoise lives in a long,

deep burrow, and when disturbed might remain in the burrow

for at least four weeks, this method was not used. Instead,

each of nineteen inhabited burrows at the Gainesville Munici-

pal Airport (Alachua Co., T 9S, R 20E, S 24) was excavated,

leaving the mouth and first four feet undisturbed. The

tortoise was displaced a short distance from the mouth and

allowed to wander for a period up to thirty minutes. Each

tortoise was displaced six to ten times. Successful return

to the home burro-w was tal-en to mean that the tortoise was

familiar with the intermediate area. The maximum displace-

ment that still resulted in successful homing was regarded

as an indicator of the size of the home range.


In setting up the tests it was foreseen that three

results might be obtained. First, the tortoise could return

to its home burrow. Second, it could move to another burrow,

Third, the tortoise might wander but not go to any burrow,

even though its movements took it near burrows that appeared

to be suitable. All three of these results were obtained,

although under somewhat different circumstances.

The first possibility occurred when the displacement

from the home burrow was 2.3 to 30 meters. The maximum dis-

placement for an individual that resulted in a return to the

home burrow varied directly with its age, determined by

carapace length (Figure 1). This distance corresponded pre-

cisely with the maximum distance between the home burrow and

the feeding areas. The feeding areas were apparent because

of the closely-cropped vegetation. The tortoises had clearly

delineated trails between their feeding areas and the home


The second possibility was obtained when a tortoise

was displaced just beyond the feeding area. Tortoises with

carapace lengths less than 12 cm proved to have no other

burrows beyond the feeding area, but the larger tortoises

Figure 1. Relations of the size of tortoises anc the r.iaximum
distances displaced from home burrows which result-
ed in direct return. Line is least-squares re-

fa a0 0
^/ o a
c^ o

5 10 15 20 25


I )


I -~

headed toward one of a number of other burrows. These burrows

were from 6.7 to 100 meters from the home burrow. The dis-

tance from the home burrow to the most distant other burrow

was directly related to the age, expressed as carapace length

(Figure 2). There were few trails between the feeding areas

and these other burrows. Such trails were much less distinct

than the trails near the home burrow.

Thenr the tortoise was displaced further from the home

burrow than the most distant secondary burrow, the third

possibility was realized. Small tortoises without other

burrows gave this result when displaced beyond their feeding


Figure 2. Relations of the size of tortoises and the
distance between the home burro.- and the most
distant other burrow (p. 9 ). Line is least-
squares regression.


I0 -

lO O



10 15 20 25 30

^ -

< /
0 40- 6 /

1AJ *

0 /



From the data, the dimensions of the home range of

G. polyphemus may be estimated by several methods. The

direct homing distance, which correlated well with the dis-

tance between the outer edge of the feeding areas and the

home burrow, may be used as an indication of either the

radius or the diameter of the home range, depending upon

whether the home burrow was centrally or eccentrically

located. This gave a home range area of 14.2 m2 for the

smallest tortoises and 6447 m2 for the largest tortoises,

providing that the burrow was centrally located. In the

case of eccentrically located burrows, the home range areas

varied from 4 m2 to 1659 m2. If, on the other hand, the

distance to the most distant other burrow is used as the

criterion, then with a central burrow the home range is from

4 m2 to 31,400 m2.

Oliver (1954) stated that G. polyphemus seldom

wanders further than 100 feet from the burrow, but this does

not help to resolve the question of which criterion to use.

The home range of G. agassizii varies from 10 to 100 acres

in southwestern Utah. Differences in the size of the home

ranges of these two species are undoubtedly attributable to

1) differences between the species, 2) differences in distri-

bution of the food supply, and 3) to differences in the

methods by which ;loodbury and Hardy (1943) and I made the

estimates involved. Their data does suggest that direct

homing distance underestimates the size of the home range,

and that calculations using the maximum distance to another

burrow give a more valid area.

Further evidence of the value of distance to other

burrows as a criterion of the size of the home range is

suggested by.the data of Blair (1951) on Peromyscus polionotus

leucocephalus. The home range of this mouse normally contained

20 burrows. A mouse would rely upon five or six of these to

escape predators. A parallel situation is obvious in the

gopher tortoise. Blair (1951) also found that the animal was

likely to be recaptured more often in certain parts of its

home range than others, suggesting that not all parts of its

home range are equally utilized. The areas used more often

were generally near the present home site.

It could be argued that the tortoise spends most of

its time above ground near the home burrow, and only occasion-

ally wanders to the extremes of its home range. The trail

systems are well maintained near the burrow, but few obvious

trails are found between the home burrow and the secondary

burrows. If this interpretation is correct then several

questions are raised. Uhich tortoise constructed the secon-

dary burrows? Does a tortoise construct several burrows in

succession and then abandon them in favor of other burrows

which are built nearby? Or are these alternate burrows the

work of tortoises that made them and then died or emigrated

during courtship, or at some other time?

The following circumstantial observations support

the idea that a tortoise will construct a number of burrows.

In one burrow with a length of only 4 feet, a 20.3 cm tortoise

was found. This burrow was extremely short for a tortoise of

that size in that area. Also, there was little dung in the

burrow and no arthropods were present. In another case a

tortoise was taken from a long burrow (18 feet) in which there

were few arthropods, and in which the grass in the mound piled

at the mouth was still green, indicating that the burrow had

been recently constructed.

On the other hand, there is also considerable evidence

that tortoises will remain in their burrows for considerable

lengths of time. Of 16 tortoises that were returned to their

home burrows after these had been excavated, four were found

in these same burrows eight months later. In each case the

tortoise had extended its burrow; beyond its original length.

Three other tortoises were recovered in freshly dug burrows--

each of which had little or no dung at the bottom. Two tor-

toises were recovered from burrows which had been regarded

as secondary burrows during the displacement tests. The

burrowing history of the other seven tortoises was not known.

Why would a tortoise normally abandon a burrow?

Environmental conditions affect the habitability of the

burrow. If the surrounding food supply decreased below a

certain threshold the tortoise would have to emigrate. Also,

if a tortoise wandered far afield during courtship and was

not strongly motivated to return, or was unable to regain its

home range, the burrow would be abandoned.

From the above i conclude that the home range of the

gopher tortoise is relatively small. The larger tortoises are

familiar with the positions of a number of secondary burrows,

but it is not clear whether these burrows are the product of

one tortoise or of several. Obviously, much work needs to be

done on even some of the fundamental aspects of the life history

of this species.



The most readily observed feature of the home range

of GoTherus polyphemus, besides the burrow, is the trail

system converging on most of the burrows. The trails

appear to be of primary importance in the short-range orien-

tation of the gopher tortoise. The fundamental question as

to how the original trail-blazing orientation was achieved

still remains obscure. Once laid, the trails may provide

visual, olfactory, and tactile cues.

In general, turtles have a good sense of vision (Walls,

1942),although they are probably not as sensitive as birds

and mammals. They are able to discriminate form (Casteel,

1911) and color (Quaranta, 1952). Furthermore, most studies

on turtle orientation have regarded visual cues as most im-

portant (Ortleb and Sexton, 1964).

Although olfaction has been shown to be important in

the orientation of other lower vertebrates (fish: Wisby and

Hasler, 1954, Groot, 1965; salamanders: Twitty, 1959), no

experiment has yet shown that this sense is important in

turtle orientation. However, a number of reports suggest

that olfaction may be important in the life of many species

of turtles: Poliaocv (1930) conditioned three European pond

turtles, Dnys orbicularis, and Boycott and Guillery (1962)

conditioned red-eared turtles, Pseudgm-s scripta cleans, to

discriminate between organic compounds. Allard (1949)

attempted and failed to demonstrate that box turtles,

Terranene carolina, could discriminate between burlap-wrapped

rocks and fish. Ortleb and Sexton (1964) were unable to show

that the painted turtle, Chrysemys picta, could distinguish

between "aromatic" water, i.e., water in which plants were

grown, and plain tap water.

Eglis (1962) described head-bobbing movements in tor-

toises when they were presented with a novel food item. He

had no other reasonable hypothesis for this behavior and

attributed it to an olfactory response. Auffenberg (1965,

1966) described similar head-bobbing in three species of

tortoises ( Geochelone carbonaria, Geochelone denticulata,

and Gopherus go~Ihemus) during courtship. He attributed

this to a visual recognition cue evolved from the original

responses to olfactory stimuli. He further demonstrated
the importance of olfaction in tortoise courtship by rubbing
the importance of olfaction in tortoise courtship by rubbing

cloacal secretions on the rear of a skeletonized shell and

eliciting attempts at mounting by a male. Based on experi-

mental evidence, Weaver (1967) concluded that in Gopherus

berlandieri head-bobbing was primarily an olfactory response.

Electrophysiological recordings from the nasal epithelium of

G. polyphemus (Tucker, 1963) demonstrated that stimulation

by amyl acetate, benzyl amine, butyric acid, and geranoil

elicited neural responses. These observations clearly show

the potential importance of olfaction in the life of turtles.

Any short-range orientation by the gopher tortoise

probably involves the use of visual or olfactory cues. Be-

cause the trails surrounding the burrows can provide such

cues, I regarded the trails as the primary source of short-

range orientation information to the tortoise.

Ilethods and Materials

Experiments with Artificial Trails

In an attempt to determine whether running on famil-

iar trails would account for the successful short range

orientation of the gopher tortoise, I made test trails out-

side the home range of the experimental subjects. Before a

tortoise establishes a distinct trail system around the bur-

row, it must be able to return to the burrow, and if the

trails are the important devices by which the return is

made, the tortoise must be able to follow cues which are

not obvious to a human observer. To test this possibility,

trails which were discernable to me only as strips of bent-

down grass were formed by dragging a sandbag in a series of

straight lines, each 60 feet long, through a grassy field.

Each of ten tortoises was tested on one of these trails.

The trail systems surrounding most burrows were

much more distinct than the artificial trails formed by

dragging the sandbag. To simulate a well-established trail

a series of straight paths 60 feet long were cut with a

grass clipper in the grassy field. Each of ten tortoises

was tested on a single trail. The natural trail systems of

the tortoise consist of intersecting paths, rather than of

a series of straight trails. To approximate this condition

in the experiment artificial trails were cut at right angles

to the original cut trails.

Another series of cut trails were made to investigate

behavior on complex trail systems in another way. These were

Y-shaped, with each arm 2 feet long. The floors of two of

the arms were covered with Benchkote, a paper that is absorbent

on one side and has a plastic coating on the other side. This

permitted the manipulation of possible odor trails by moving

the Benchkote. Ten of these trails were provided for tests

with each of twelve tortoises. The starting positions and

the arm down which the tortoise was to be forced to run were

predetermined and all tortoises followed the same sequence of

starting positions and turns.


Each tortoise was placed at one end of a straight

trail and the distance it travelled before it completely

left the trail was recorded. For the indistinct trails,

where there was potential directional information from the

warp of the grass, each tortoise was started at both ends of

the trail.

Each tortoise w7as placed at the end of the trails cut

at right angles to the straight trail, and was permitted to

move down this trail and turn onto the straight trail. The

tortoise was tested once on each of ten of these intersecting

trails to determine whether it would follo-w the original

straight trail. After testing the tortoise was forced to

run the length of the straight trails five more times. Two

days elapsed between the forced runs dono the straight trail

and the testing on the side trails. After fifteen forced runs

and two tests had been recorded the nares of five animals were

blocked with Carbowax 3000 and Carbowax 1500 (in the ratio of

3:1). The other five animals were used as a control. Both

groups were then retested.

In the tests on the Y-trails the tortoises were

started at the uncovered branch and were forced to turn right

or left on the Benclhhote according to a predetermined sequence.

After five of these forced runs a tortoise was permitted to

choose between the two directions. A turn in the direction

of the forced run was recorded as a positive response. This

same procedure was repeated for a total of twenty forced

runs and four test runs. Tortoises were not tested immedi-

ately after the forced runs; an interval of a day elapsed

between the two types of runs.

iWhen a series of tests was completed the Benchkote

was reversed and the tortoises were retested. Then the

original orientation of the Benchkote flooring was restored

and each tortoise was forced five more times. The next day

the tortoises were divided into two groups. The nares of

one group Tere blocked with Carbowax. The other group was

used as a control. Both groups were then retested.

Laboratory experiments


To investigate the possibility that olfactory stimuli

provide orientation information, the tortoises were tested in

a T maze, the inside of which had been coated with white fiber-

glass resin. The runway was 24 inches long and choice arms

extended 12 inches to either side. All passageways were 11

inches wide, thus permitting the tortoises to turn around, and

all had walls 3 inches high. An aluminum door separating the

starting box (8" x 11") from the rest of the maze was kept

closed until the start of each trial, at which time it was

removed. The maze was roofed with 1/4 inch mesh hardware

cloth, covered with opaque plastic tape to minimize visual

cues that might confuse the turtles.

Each choice am led into a wooden box with the same

dimensions as the starting box. One of these boxes served

as a home for each tortoise at least a week before starting

the tests; the other was a dummy box in which a tortoise had

never been kept. To the insides of both boxes several coats

of white fiberglass resin were applied to prevent penetration

of odorants into the wooden floor. In the home box food and

water were provided at all times, except Curing the tests.


The home box and each neutral box were attached to

the ends of the choice arms. Zach tortoise tested was

placed in the starting box with the aluminum door in place.

The position of the home box was randomly alternated from

side to side between tests. The tortoise was subjected to

the heat generated by a 150 Y infra-red bulb hung over the

starting box. The length of time the tortoise was retained

in the starting box varied from 2 to 15 minutes, depending

on the ambient temperature. When the animal attempted to

escape, the door was removed and the trial begun.

Continuous observations were made during each trial.

Records were made of the choice taTken, of the time spent at

the choice point, of any head-bobbing behavior, of putting

the snout down on the floor, and of any unusual behavior

patterns. A choice was not recorded until the tortoise had

moved down one of the maze arms far enough so that it was no

longer visible from the starting box. ITo time limit was

placed on a trial, because the tortoises often showed con-

siderable hesitation at the choice point, not related to

correctness of choice.

After each trial the tortoise was placed in its home

box until the next run. In a few cases I varied the procedure,

placed a tortoise in a neutral box after it had made a negative

choice, but this had no obvious effect on subsequent tests.

Between trials the T maze was washed with Alconox solution

and thoroughly rinsed with distilled water to minimize

extraneous odors. Before each trial the home was wiped

out with paper towels to eliminate humidity and visual marks

of dung or food. The neutral box was also wiped out with

paper towels to control for odorants on the toweling itself.

To maximize the amount of any odorant deposited by the

tortoise, the floors of the choice arms were covered with

Benchkote. It was assumed that the tortoise deposited a

scent trail as it walked over the Benchkote. This "scented"

Benchkote was subsequently placed in the positive choice

arm, i.e., on the same side as the home box.



Artificial Trail Experiments

The tortoises placed on the indistinct trails (p. 19)

invariably followed them for short distances before wandering

off. Although the distances travelled on the trails were so

small that they could be attributed to chance, no statistical

test of their significance was found to be appropriate. The

distances travelled down the grain of bent grass (mean = 5.7 ft.

+ S.E. = 1.5 ft.) were not significantly different from those

travelled against the grain (mean = 3.2 ft + S.E. = 1.1 ft),

although the tortoises appeared to follow the trails somewhat

longer in the direction of the bend of the grass. The tor-

toises stayed on the distinct trails for significantly greater

distances (mean=22.5 ft + S.E. 7.0 ft) than they did on the

indistinct trails (t-test, P < .05).

The tortoises placed on the side trails, after being

forced down the straight trails, turned in the same direction

onto the straight trails as they had been forced, with fre-

quencies significantly different from random (z=3.70, P<

.0001). It is clear that this significance applies to the

behavior of half of the group of tortoises, the other half

chose randomly (Table 1). WThen the external nares were

blocked, random choices were made (Table 2).

Table 1. Preference of tortoises to turn onto a straight
trail in the same direction in which they had
been forced previously on the straight trail
(positive choice). The data represent the
number of positive choices out of ten.

Subject Number of positive z score
choices after ten
forced runs


AC 3

JJ 1

SF 1

A 15

AC 1










A 19





P < 0.001


Table 2. Comparison of preferences of tcrtoises to turn
onto a straight trail in the same direction that
they had been forced previously with nares blocked
with wax and unblocked.

Nares Blocked

Subject NIumber of positive z score
choices after fifteen
forced runs







P > 0.70

Nares Unblocked

LO 2.85

LO 2.85

LO 2.85

8 1.53

4 -0.32
z 9.81

z 3.10

P < 0.001

SF 1

AC 3


A 19

A 15

JJ 1



AC 1


On the Y-trails the results were similar to those

obtained on the side trails (Table 3). There was a gradual
improvement in the turning of the tortoise toward the train-

ing direction with a greater number of forced trials. Again,

half of the subjects showed this improvement while the choices

of the other half was not different from random. Blocking

the nose with wax destroyed the direction preference (z=0.99,

P >.i0). Reversing the Benchkote which covered the floors

of the choice arms, however, did not cause the tortoises to

reverse their preferred choices. Instead, they turned non-

randomly (z=1.79, P <.05) in the same direction as previously.

Laboratory Experiments

Analysis of results in the T maze tests showed no

significant preference for the home box (Table 4). Behavioral

differences were observed between trials resulting in the

choice of the home box and those which did not. These dif-

ferences related mainly to the amount of head-bobbing (mean

for positive trials = 7.0, for negative trials = 3.2) and

lowering the snout on the substrate (mean for positive trials

= 5.6, for negative trials = 2.2). There was also some dif-

ference in the amount of time spent at the choice point

(mean for positive trials 20.9 min., for negative trials

= 3.3 min.).

Table 3. Preferences of tortoises to turn in the same
direction in which they had previously been
forced to turn on Y-shaped trails (positive
choice). lThe ranked data represent the
number of positive choices out of ten trails
on ten separate Y-trails per tortoise.

















of forced













runs before testing

15 20































<0.01 <0.01



















0 I
a P 4
>i 4 U4

Ml 01 C0 a

0 0d 0

0 L r. 0 -
4 a) a 4-

Lo 0 0 W
fi ,Ci O 0

( j-, P 4 >
0 4- 4- 4-o)

U 0 P 0
O4 4i o
o0 0 0

SCo M-H 0

t '0 02 o20
P 0 4 4-
4 -HC 0 t4
o r P, 0

4 PH 4-4 P
)04 a U)

(4 0 (c t 4 ) f,
g .C o 0U *H
4 4 C) p
S[-P u .4J 14
0 U 41 -1

44 0 O 0
40 4-40 n

4r L a 4 -4 c:
u 4v1 & 9

z cz n Q r-i
1nr O 1-J 0 c
10 4 0( q1

0-4W 10
CO O4 04

40J 14 P

4r-44 O U



In '
0 H




to, a
0 4 H

* *
CO 0

Ui H--
03 r-

in 1-

4o 0
0 ,-a to ro 0
0 0 -4 r4- 0

(m o

o 4

yD si

in o

oi o


in Qo -4 n r-H
o o r-4 o

O N 0
* *-
in o


4- 0 "4
Oc v-A 0

In .-i

0 0 0 Un 0
u)n Or O 0 C
* * *

Ln Ln 00 r-

Ln 0t CcO

H <4








o I



Li 0




The results of the tests on the "indistinct" trails

suggest that these were not very effective orientation

stimuli. Perhaps the weak trails lacked some factor that

a tortoise recognizes as being indicative of a trail. Re-

peated testing of a tortoise on the same trail resulted in

a slight improvement in trail-following, perhaps because the

trails became increasingly more distinct. Although there was

a slight improvement in trail-following when the tortoise was

going with the direction of the bend of the grass, rather

than against it, there was no significant difference (t

test, P >.05) because of the short distances travelled.

This suggests that weak tactile or visual stimulation is of

little use to the tortoise for orientation.

The cut trails, on the other hand, were readily

followed over significantly greater distances (t test, P<.05)

distances than on the indistinct trails. This suggests that

strong tactile or visual stimuli can direct the tortoises

along the path of least resistance. This may explain the

formation of the distinct trails near the burrows.

The short side trails cut into the straight cut

trails were made to see whether tortoises which had wandered

down the straight trails would turn onto one of these trails

after being placed on the side trails. They did indeed turn
consistently onto the trails suggesting that the features

of the trails themselves stimulated the tortoise directly.

It is unlikely that they utilized a kinesthetic pattern for

this. An interesting side result was obtained. In a signi-

ficant number of cases (z = 3.70, P<.0001) tortoises turned

down the straight trail in the same direction in which they

had headed when tested on the straight trail alone. With

very little experience the tortoises had learned to travel

in a particular direction. Cues that may have controlled

this behavior could have been either a scent trail with

directional properties deposited by the tortoises during the

straight-trail tests, or landmarks such as clumps of grass

or the tree-line. Still another possibility was that the

straight trails chanced to run in a direction matching an

innate direction preference (e.g., sun compass). The latter

possibility seems unlikely because of the pronounced vari-

ability of direction preference among individual tortoises

(p. 80).

To test the possibility that the tortoises use odor

trails, the nares of five specimens were blocked with wax.

This resulted in the loss of directional preference estab-

lished in earlier s-eries. Control animals with the nares un-

plugged retained and improved their preference for the di-

rection they had travelled down the straight trail. Much

individual variation existed within each of these two groups.

The rigorous analysis with the z test indicated that

only half of the tortoises showed this preference; they did

it so strongly that one could get the impression that, on

the average, the entire group was demonstrating a direction

preference. Still another test was necessary.

The experiment was repeated on Y-trails. Each tor-

toise was started from a different position on each of a

series of these trails to minimize any possible use of a

solar compass in the directional choice. The tortoises

turned in the correct direction significantly greater (z =

2.92, P <.002) than random expectation. Again, only half of

the animals showed a strong directional preference; with their

nares blocked with wax their preference was lost. Thus, the

tortoises showed comparable results on the two types of trails.

However, when the Benchkote that covered the floor of the

choice arms of the Y-trails was changed to the negative run-

way, the tortoises did not reverse their preferred directions,

but retained a preference for the original choice. This

suggests that blocking the nares with wax, rather than merely

preventing olfactory input, upset the tortoises' behavior.

These data suggest that for short-range orientation visual

cues are more important to the tortoise than olfaction. The

tortoises seem to learn the landmarks at each choice point

after only a few experiences. The observation that only

half of the subjects showed a direction preference in the

experiments is attributed to individual variation in learning

ability and motivation.

The lack of evidence for the use of olfaction in

short-range orientation, as revealed in the preceding tests,

does not exclude the possibility that it is used by Gopherus.

The T maze tests were designed to eliminate all stimuli except

olfactory ones. The lack of preference for the home box in the

T maze parallels the findings of Boycott and Guillery (1962),

but the differences in head-bobbing and snout-lowering be-

tween successful and unsuccessful trials suggests that olfaction

may be involved. The time spent at the choice point was also

longer in two cases for the successful trials, although the

rates of bobbing and lowering were actually less than for the

unsuccessful trials, suggesting that these responses may not

be accurate indicators of olfactory input or that tortoises

unsuccessful in picking up the odors kept trying but showed

the response only when they could perceive the odors. These

data suggest that, although the tortoise may be capable of

perceiving olfactory stimuli which they themselves have pro-

duced, either the experimental conditions were not appropriate

to elicit this behavior or the tortoises do not rely upon

olfactory cues for directional information.

The tortoises in the present study may not have de-

posited the usual scent cues, although dung was occasionally

deposited. .lhen this was left for a subsequent trial by that

tortoise, most of the tortoises showed no reaction to its

presence. One appeared to avoid the side with the dung on

it, but this avoidance was not seen the next day. Possibly

dehydration or decomposition of an aversive odorant altered

the stimulus properties of the dung. Rehydration with dis-

tilled water did not restore the original aversion.

The tortoises also urinated on the Benchlote during

some of the trials. One tortoise, with its nose blocked with

wax, followed the urine-soaked trails ten times in ten runs.

The result was not conclusive because the urine made dark

spots on the floor. It seems likely that the tortoise fol-

lowed the visual cues.

Although the number of tortoises tested in each of

these experiments was not large, the conclusion is inescap-

able that visual cues provide the primary directional informa-

tion to the tortoise. These may be rapidly learned under

novel conditions. Although only slight evidence that olfac-

tion was used as a directional stimulus, the possibility

remains. Perhaps with appropriate training techniques the

tortoises can subsequently be show to utilize olfactory

stimuli as supplementary short-range cues.



Although Gould (1957,1959) suggested that turtles

utilize the sun for orientation, he did not examine their

internal chronometer. If a mechanism of celestial orienta-

tion is to be functional for a long time over a long distance,

the organism must compensate for the characteristically and

continually changing angle between the sun's azimuth and the

goal direction. The procedure of resetting the internal

clock by changing the beginning and end of the light-dark

cycle has become a useful technique to demonstrate the inter-

action between the internal chronometer and solar orientation.

Most phase-shifting work has been done with invertebrates,

birds, and mammals, but Birukow et al. (1963) obtained similar

deviations from the trained direction in three species of

lizards, Lacerta viridis, L. sicula, and L. muralis, when

these were phase-shifted.

Because so many organisms, including mammals, birds,

lizards, and insects, have been shown to have an internal

clock (Hoffman, 1965) Pittendrigh (1960) maintains that

circadian rhythms are ubiquitous in living systems. The

presence of a circadian rhythm is demonstrated by placing

an animal under constant environmental conditions and ob-

serving whether an activity pattern whose frequency approxi-

mates that of the solar day is maintained. Results obtained

from a wide variety of organisms have been very consistent.

The circadian rule states that a diurnal animal in constant

light will show a shorter circadian frequency than it will in

complete darkness, the reverse being true for nocturnal

animals (Aschoff, 1960). With few exceptions, much evidence

supports this generalization (Hoffman, 1965). The presence

of an internal clock does not imply that it is functional

in orientation. In fact, the inaccuracy of these internal

clocks has often been cited as evidence that the current

theories of celestial orientation are not sufficient to ex-

plain the ability of some organisms to return home from

relatively short distances (Adler,.1963a, 1963b; Meyer, 1964;

Walraff, 1960, 1967).

The purposes of this phase of the study were:

1. To learn whether the gopher tortoise possesses an

internal clock.

2. If so, to determine its accuracy.


3. To provide a basis for phase-shifting in order

to investigate the possibility of celestial


Methods and Materials

The method to determine the existence of a circadian

rhythm was to record the locomotor activity of tortoises

under various conditions. The daily patterns were deter-

mined in automatically monitored activity boxes in which

light and temperature were kept under control.

Activity Box

Two wooden boxes (37 x 31 x 16 inches) with nine

wooden floor panels (each 10 x 12 inches) were used. Only

one of the panels, serving as a feeding platform, was

rigidly mounted on wooden blocks. The other panels were

supported on small springs and connected to adjustable micro-

switches. When the tortoise depressed a panel the micro-

switch was closed; this information was fed into an Ester-

line-Angus Event Recorder. A wooden partition (Figure 3)

prevented the tortoise from depressing more than two panels

simultaneously. The partition also directed the tortoise

to cross the panels in a regular sequence.

A wooden cover was placed above six of the panels

(Figure 3) to determine the effect of the presence of a

refuge from the imposed light condition on the activity

Figure 3. Activity box used for determining locomotor
activity patterns of tortoises. A, pan l upon
which food and water was supplied; B, panels
which activated micro-switches; C, rem-ovable
cover; D, ,wocden partition.




pattern. Harker (1960, p. 354) suggested that a clearer

rhythm is found -when such a dark shelter is provided.


Each freshly-caught tortoise was kept for seven

days in an activity box under twelve hours of light alter-

nating with twelve hours of dark, in order to acclimate it

to the experimental situation. Records were kept of the

total activity and of the pattern of activity during this


Each tortoise was then subjected to one or more of

the following conditions: constant light, constant dark, or

a regimen in which the light-dark pattern was phase-shifted

either six or twelve hours. The illumination in all cases

came from a 100 W light bulb approximately 2 feet above the

floor of the activity box. The incident light 4 inches above

the surface was approximately 350 lux. The temperature was

kept constant for each individual tortoise and did not vary

more than from 26 C to 29 C in all experiments. Each tortoise

remained under the constant conditions for at least seven days,

or until the activity pattern had changed in response to a

phase-shifted light-dark cycle. o


Under the alternating twelve hours of light and

twelve hours of dark, single individuals showed fairly regular

activity patterns, but there was considerable variation be-

tween individuals, with from two to four peaks of activity

(Figure 4). Usually the tortoise became inactive near

solar noon.

The tortoises subjected to a six-hour phase-shift

adjusted their activity patterns within one or two days

(Figure 5). A twelve-hour phase-shift required approximately

four days of adjustment, although there were conspicuous

changes on the first day (Figure 6)..

In constant light each individual showed a progres-

sive shift of onset of activity in the direction expected

for a diurnal animal: each successive activity period began

earlier (Figure 7). The length of the period is subject to

various interpretations depending on the method of analysis

(p. 59), but the differences are not great (Table 5). Offset

of activity was also determined, and similar results were

obtained. An individual exposed to constant light and

fluctuating temperatures did not show a regular shift of

onset of activity. Instead the onset appeared to be related

to the temperature fluctuations (Figure 3).

Figure 4. Nean per cent of Caily activity each hour vnc'ee
approximately 12 hours of dark r._tern:tinr ;ith
12 hours of light. Light cn at approximately
0700 anc off at approxirimtely 1900.


6 12 l 2 2


15 15

10 10

13 i15
1 s I
6 S! 21 4 12 II 24


6 122 Ht o


The effect on the activity pattern of shifting the
phase of the artificial light-dark conditions 6
hours. Vertical bars represent periods during
which the tortoises were active; shaded areas,
the periods of imposed 'arkness. To more clearly
illustrate the rhythmicity, the succeeding day
is sho-rw above the numbered day. A shows a
retarded shift; B shows an advanced shift.

Figure 5.

.... ......



Figure 6. The effect on the activity pattern of a tortoise
shifting the phase of the artificial light-dark
conditions 12 hours. Data presented as explained
for Figure 5.


Figure 7. Activity patterns of tortoises unoer alternating
light-dar! conditions anc' under constant light.
Data presented as explained in Figule 5.

.. 1 1

[ *I I I

o I aIe


" .0 U

I 1 '

.I -I ,
SI -

il[ *I


1 I I :

i i I1
'_ _1 "'

I I 1I

C n l






S -I




Figure 7. (Continued)

. I .- ,, ;:- "

'' !
:- i : I .

;- .;:- : I: "II;
S '.
- .*- -, II '

*- I i

.. II

'* .. I '- '" ,.


I -

I. *

I i.l

'1. .I ;*



.. 1

. g"

I.. * *
I. .

SI -*-.

-S.:' 1
*" : '" ;, i


Table 5. Comparison of length of circadian period under constant
light based on time of onset of activity by the
two methods of mean-period length by the classical
method and by linear regression (see p. 59).

Classical Linear Regression
Condition Tortoise n mean S.E. n mean Sb
(hrs) (min.) (hrs) (min)

Cover off A 1 10 21:27.9 47.3 11 21:21.5 12.7
box A 2 10 22:14.0 57.8 11 22:29.9 15.3

BUC1 18 23:27.7 1:17.9 19 23:00.3 12.5

BUC2 8 24:32.4 1:38.3 9 23:43.2 33.2

Cover on P 1 16 22:28.3 19.3 17 22:21.5 3.9
box P 2 11 22:55.2 26.8 12 22:51.8 5.7

C 1 13 23:35.5 42.2 14 23:25.0 12.0

BC1 18 23:03.9 67.5 19 22:55.9 9.3

Figure 8. The effect of temperature on the onset of activity
unc'er constant light. Solid line, representing
onset of activity, to sho7 parallel pattern;
dashed line represents temperature.

; I \ c
S---- -- ,o 26 Q

0 ---0--O 29

S- 30

3 6 9 12 15


Comparison of the onset of the tortoises' activity

with the onset of light under twelve hours of light alter-

nating with twelve hours of dark indicates that the rhythmic

behavior is entrained by light onset, although some indivi-

duals remain active well into the dark period, and some

"anticipate" the onset by an hour. This might have adaptive

significance for the poikilothernous tortoise, whose activity

is dependent upon relatively high ambient temperatures. The

inactivity at midday may reflect an adaptive avoidance of

the extremes of heat that prevail at that time during the

summer. This midday inactivity was evident in tortoises ob-

served in the laboratory throughout the year.

More pertinent to the demonstration of an internal

clock in the gopher tortoise is the evidence that tortoises

maintain circadian activity patterns under constant light

and temperature conditions. The period of the activity

cycle is less than twenty-four hours as would be expected

for a diurnal animal. The mean period length was calculated

according to the classical method

t = t2-tl) + (t3-t2) + ... (t-tn-l)

where t is the mean period and tl, t2, t3, and tn refer to the

time of initiation of activity. However, I concur with the

argument of Lowe et al. (1]67) that regression analysis re-

sults in a more precise estimate of the natural period.

The classical method results in a range of 21:37.9 hours

to 24:32.4 hours with a mean of 22:58.1. By the regression

method the range was 21:21.5 to 23:43.2 with a mean of 22:47.1.

However, analysis of covariance (Snedecor and Cochran, 1967)

indicates that the regressions obtained from tortoises examined

with the cover absent cannot be pooled. The tortoises with

the cover present can be pooled (F test, P<.05). The period

length calculated by either analysis supports Aschoff's (1960)

circadian rule for diurnal animals, and also agrees with

values obtained with other animals--23 to 26 hours (Dinning,

1.64). Similar values were obtained -when the offset rather

than the initiation of activity was used as the criterion

for the limit of the activity period.

The variances of the pooled data under the two con-

ditions, cover present and absent, are significantly differ-

ent (F test, P <.05), the variance obtained with the cover

off being greater, thus supporting Harker's (1960, p. 354)

suggestion that providing a refuge from constant light re-

sults in a more distinct rhythm. These data suggest that

the gopher tortoise may behavorially affect the rhythm. IHow-

ever, the tortoise's activity rhythm is more variable than

has been reported for other animals, e.g., Glauco2ys volans,

DeCoursey (1960), which had a minimum standard deviation of

only two minutes (by the classical method). The standard

error of the mean calculated by the classical method was

26.8 minutes; the minimum standard deviation of the slope by

the regression method was 3.9 minutes.

Aschoff's (1960) rule further predicts that under

constant dark conditions the period of a diurnal animal is

lengthened, as compared with that developed under constant

light. Actually, in the single animal which showed rhythmic-

ity under these conditions the period was 22:34.4 hours, which

is not a statistically significant deviation from the results

obtained under constant light. However, only two tortoises

were subjected to this condition; a larger sample might show

the expected results. In constant dark the onset and offset

of the animals' activity was not clear cut and rhythmic

pattern was lost. One tortoise rapidly became inactive, the

other was active in bursts throughout the testing period.

Therefore, little valid comparison can be made between the

results obtained and those predicted by Aschoff's circadian

rule. However, if upon subsequent examination of a larger

sample this discrepancy remains, it may be attributed to

the fossorial habit of this species. At present there is

no evidence that the internal clock of the gopher tortoise

is different from that found in any other species of diurnal


It seems evident that under natural patterns of

light and dark the tortoises adjust their locomotor patterns

to the natural conditions. The factors which keep an animal

in phase with the diurnally fluctuating conditions are called

entraining agents; most important are light and temperature

(Bruce, 1960). Under alternating light and dark conditions

it is clear that light-dark cycles are important entraining

agents in this species as well. Adjustments to phase-shifts

further support this idea. It only took one or two experi-

ences of a six-hour phase-shifted light pattern before the

tortoises were in phase with the shift. A twelve-hour shift

required several days to reach the same result of a new

steady-state. That these were not merely immediate responses

to the onset of light is indicated by the "anticipatory"

activity prior to the onset of light.

iTe experiments demonstrated that temperature fluctua-

tions have an effect on the periodicity of the tortoise.

During a single test made over several days in constant light,

the temperature fluctuated between 25 C and 30 C; the onset of

the tortoise's activity fluctuated correspondingly, beginning

earlier when the temperature was high and later in a lower

temperature. A similar result was reported by Hoffman (1968)

who found that lizards were entrained by temperature fluctua-

tions as small as 0.9 C.

If Hoffman's (1965) suggestion that a rigid or very

slowly shifting clock is necessary for bicoordinate naviga-

tion is valid, as seems self-evident, then the internal

chronometer of the gopher tortoise would appear to be poorly

suited for such position-finding. On the other hand, a

a rapidly shifting clock such as that found in these tor-

toises would appear adequate for compass orientation. Hoffman

(1954), for example, found that starlings started to shift

in response to a six-hour phase-shift after one day but had

not completed the shift until the thirteenth day. This is a

much slower shift than that demonstrated by the gopher tor-


Among students of animal navigation there is some

opinion that the accuracy of the internal chronometer has

not been shown to be sufficient for effective bicoordinate

navigation (Adler, 1963a, 1963b; Meyer, 1964; Walraff, 1960,


1967). The accuracy of the clock is usually indicated by

the standard deviation from the mean frequency. As mentioned

above, the internal clock of the gopher tortoise is less

accurate than that of some other organisms which would

also argue against the possibility that bicoordinate

navigation is used in travel orientation.



Recent investigations have revealed that many species

of animals are capable of orienting even when they are not

in direct sensory contact with familiar fixed landmarks.

Most of the work to date has been concerned with migratory

birds, although fish and reptiles also have been considered

(Matthevs, 1955; Hasler, 1964; Carr, 1967).

Some species of turtles, especially marine forms,

are migratory and capable of long-range orientation, but

the nature of this orientation is unknown. Little experi-

mental evidence has been obtained to support any of the

various hypotheses formulated to explain the mechanism of

long-range, open-sea orientation. One of the major obstacles

to such experimentation is the difficulty of accurately

tracl-ing individuals over open ocean for long periods of

time. In turtles several investigators have tried to solve

the homing-orientation problem by examining hatchling sea

turtles, fresh-water or terrestrial species. There have

been numerous observations of turtles returning to their


home ponds or to particular portions of the pond (Cagle,

1944), to their burrows (Uoodbury and Hardy, 1943), or to

the same corner of a room (Mertens, 1953). Although this

homing phenomenon may or may not be directly related to the

problem of sea turtle navigation, its investigation should

provide clues to the mechanism of turtle orientation.

The major hypotheses to explain turtle orientation

involve reference to fixed landmarks along the travel path,

especially horizon effects (Ortleb and Sexton, 1964); ol-

faction; and celestial orientation. Other types of guidance

such as magnetic orientation and inertial control have been

suggested for birds, but the evidence of their existence

remains controversial and no attempt to show their utiliza-

tion in lower vertebrates has been made.

Landmarks are usually regarded as short-range stimuli,

but they may interfere with other orientation stimuli (Schmidt-

Koenig, 1965).

Olfactory orientation may, theoretically, be a long-

range cue under certain conditions--for example, where a

prevailing current of wind or water from the goal maintains

a gradient. This may account for the island-finding orienta-

tion of green turtles, Chelonia mydas (Carr, 1967). Another

situation in which olfaction may be utilized for long-range

orientation is where scent trails are deposited by the animal,

which can then home by retracing its trail. However, a

terrestrial or semi-aquatic animal is still able to orient

in spite of living in an area where the currents are not

constant and displaced beyond its scent trails.

Celestial orientation has been suggested several

times as possible means of turtle orientation. Fischer (1964)

found that hatchling green turtles, C. mydas, possess a time-

compensated sun compass and perhaps are capable of true solar

navigation, which is defined by Schmidt-Koenig (1965) as a

phenomenon in which the compass direction is goal-related.

Carr (1962, 1963, 1964) suggests that adult green turtles

utilize some form of celestial orientation and implies that

more than a compass sense is involved. Threnfeld and Koch

(1S67) found that green turtles are extremely myopic in air,

which argues against the utilization of stars for navigation.

Gould (1957, 1959) obtained evidence that the

terrestrial box turtle, Terrapene c. carolina, and the aquatic

painted turtle, Chryseys picta, are capable of true celestial

navigation. zflen's (1965) re-examination of C. picta pro-

vided no evidence that this species was capable of celestial


Gopher tortoises, GoQherus polyphemus, would seem to

be good subjects to test for celestial guidance capacity.

First, they are terrestrial, which permits the accurate

tracking of their paths. Second, they live in permanent or

semi-permanent burrows, so that the goal to which they may

be assumed to return is very limited in area. Third, they

are diurnally active, which obviates the necessity of deter-

mining their possible use of stellar or lunar orientation.

The limited range of their travels suggests that a capacity

for bicoorcinate position finding would be altogether super-

flucus, and physiologically unachieveable. It does not,

however, preclude the possible use of a sun-compass sense.

The purpose of this phase of the study is to deter-

mine whether gopher tortoises are capable of solar orienta-

tion, or if so, to determine the nature of such an orienta-

tion whether a cohipass sense or true navigation is involved.

Methods and Materials

Open Field jests

Source and Care of the Experimental Animals

Twenty-three tortoises were used in these tests.

Twelve were from the Gainesville Municipal Airport (Alachua

County, T 9S R 20E, S 24), located approximately seven miles

from the open field in which they were tested. The remain-

ing tortoises were from several Florida localities. To

avoid the presence of landmarks with which the tortoise

was familiar, no tortoise whose home burrow was less than

five miles from the test site was used.

The tortoises were either housed for six weeks in

terraria and wooden pens in the laboratory or in an outdoor

pen 11.5 miles south-southeast of the test field. Light

for the laboratory animals was provided either through a

north-facing window or by fluorescent lighting which was

automatically turned on and off at approximately sunrise

and sunset. The tortoises were permitted to feed during

each testing period and also provided with fresh lettuce,

grapes, melons, and water. There was no significant change

in the weight of the tortoises during these tests, nor was

there any detectable difference in behavior between those

held indoors and those kept under more natural conditions.

Description of Test Field

All twenty-three tortoises were released and studied

in an open field of closely cropped grass, on the University

of Florida Experimental Station grounds (Alachua Co., T 10S,

R 20E, S 1). The field measured 245 yards in a north-south

direction and 65 yards in the east-west direction. The

northern edge of the field was bounded by a dirt road. A

twelve-foot-wide strip of grass on the other side of the

road separated it from a large plowed field. The road could

not be seen by tortoises located in the center of the field

because of a slight rise at this end of the field. The

eastern edge of the field was bordered by a row of two- to

four-foot fruit tree saplings, which marked the edge of an

adjoining orchard. To the south the field widened

slightly. There was a clump of trees to the southeast in

which a trailer park was located. From the trailer park a

line of longleaf pine trees ran almost all the way across

the southern end of the field. Toward the southwestern

corner there was a gap in this row of trees. Approximately

50 yards beyond this first row of trees was a second which

almost filled this gap on the horizon. A dense border of

shrubs separated the pines from the edge of the cut grass

in the field. The western edge of the field consisted of an

irregular row of live-oak and longleaf pines; this was the

edge of a mixed oak-pine woods which extended to the west.

There was a relatively dense growth of shrubs along this

border, but under the trees were few plants near the ground.


All of the tortoises were transported to the test

field by automobile in closed styrofoam ice chests. They

were carried to the three release points in the field in

these chests. The twenty-three tortoises were released

singly from the release points, which were in the middle of

the field, about 50 feet apart. The first three tortoises

were allowed to move unimpeded for periods of 10 to 90

minutes, depending on the activity:of the tortoise. To

avoid excessive heating and yet permit the tortoise to make

a usefully long run, 30 minutes appeared to be the optimum

period for each trial. Accordingly, the remaining twenty

tortoises were observed for one-half hour. I moved slowly

at least 15 feet behind the tortoise, and my presence did

not appear to disturb the experimental animal. The initial

direction upon release was chosen at random by me, so that

each of the cardinal compass points were faced at least once

during each individual's test.

The paths of travel were tracked in three ways. First,

the tortoises were followed and their trails plotted. This

was possible because of their slow gait and because they de-

pressed the grass as they moved over it. The distance travelled

was then calculated by means of a properly calibrated map

reader. Second, the position of the tortoise was marked

every 5 minutes by pushing a nail into the ground. After the

test the directions could be determined by sighting on the

nails with a compass, with an accuracy of approximately -2.

The distances from the starting point were paced. The two

methods resulted in angular differences within o8, and the

linear distances within 5 feet.

The third method used was that of the thread trailer

of Breder (1927) and Stickel (1950). Although this method

worked well and the results agreed with those obtained with

the other two methods, it required two persons to untangle

the thread and measure it in a short time.

iTe mean direction for each test was determined by the

polar angle from the starting point to the position of the

tortoise at the end of the test period.

Arena Tests

Elimination of all fixed landmarks should result in
disruption of any direction tendency not based upon a sun

compass sense. Such tests were carried out in an orientation



The arena had a diameter of 20 feet and consisted of

twelve panels, each measuring 42 inches high and 60 inches

long. The height was sufficient to exclude from view of

the tortoises any trees or other features of the outside

landscape. The panels were of black polyethylene sheeting

mounted on a wooden framework (Figure 9 ). To minimize glare

the polyethylene was sprayed with a dull black paint. The

panels were connected by strips of Velcro, a nylon zipper

material, which permitted rapid assembly and dismantling of

the arena after each day's tests. A regular dodecagon was

fonred by the panels constituting the arena. The panels were

individually numbered and interchangeable so that they could be

randomly placed in the arena wall to eliminate cues from

individual panels.

Bearing and cooling effects during the tests caused

wrinkles to form in the polyethylene, but this was approxi-

Figure 9. Arena used for testing tortoise orientation in
the absence of larn-arks. A, diraensions; 3,
arena set un in field.


53.5 ft

-- 20--- 5ft

indicate the approximate location where turn takes

effect. For the higher Reynolds number case, the local

Nusselt numbers near the inlet region of the present study

deviate only slightly from the Hornbeck's results.

For all the cases studied for the heat exchanger, a

large Nusselt number is found near the inlet, the turning

point, and in the reattachment region and low Nusselt

number occurs in the recirculation region.

It is clear that once the flow is deflected into the

annulus, a redevelopment of flow occurs. In the annulus

region, a fully developed condition is reached earlier for

fluid of low Peclet number, see figures 18 through 20. In

order to check on the computations of the present work,

Lundberg and coauthors' results [24] are used. As shown in

Table 4, Lundberg's results for fully developed velocity

and temperature profiles in an annulus appear to be in good

agreement with the present investigation that is computed

at Pr=0.7 and Re=100. If a linear interpolation based on

Lundberg's results at r+=0.5 and 1.0 can serve as a guide,

then the error is only 1.5%.

VI.6 Mean Nusselt Number

Figure 21 presents mean Nusselt number as a function

of Reynolds number with Prandtl number as a parameter.

This figure shows that mean Nusselt number increases with

both the Reynolds and Prandtl numbers. A correlation of



co o

z I C

Z a

0 0



O t4 1 ' r'
.' g O

"- C -
-J -

SLn a N Lc
o: i CD C
K (C

a i

C' f
(Cn a.

'4 C


C~~i C CC
C~ ~~~ IC'C C CU
C'.'~~ C' C' -

The tortoises from the previous arena tests which had

shown a significant preferred direction were subjected to a

phase-shift six hours advanced. Light onset was approximately

0045 LST and light offset was approximately 1410 LST. The

light source was a 150 i sun lamp. Three other tortoises

that had shown a preferred direction in the arena were also

brought into the laboratory but were kept in phase with local

solar time by an open window and fluorescent lamps which went

on at approximately natural sunrise.

One of the animals subjected to the phase shift was

kept in the previously described activity box (p.40) to

verify that the locomotor activity pattern shifted. The tor-

toises were all kept under these artificial conditions for at

least seven days before beginning of tests in the arena, to

make certain that not only the locomotor activity internal

but also the oriantational clock had been shifted. Others

(e.g., Hoffman, 1960) have found that it takes longer for

the orientaticnal clock to be shifted than it takes the

activity clock.

These tortoises were then tested in the arena as

described above each tortoise was tested eight times during

the morning (0900 to 1030 LST). Upon completion of each day's

testing they wera returned to the laboratory. Each was then

tested eight times in the arena at midday (1130 to 1300 LST).

All tortoises required at least two days to complete testing

for each time of day.


The directions which an individual chose each time

were compared to determine whether a non-random distribution

resulted. Such a distribution indicated that a particular

direction was preferred. Mean angles and their angular

deviations were determined by methods described by Batschelet

(1965), as follows:

Co =L .here n is the ntuber of observations

y Sin

r =\2 2 w y2here r is a measure of the concentration

around the mean direction.

z = nr2 where z is a statistic for non-random

direction preference. The critical

values for z have been calculated by

Greenwood and Durand (1955) and are

given by Batschelet (1965)

x = cos of mean angle; = sin of mean angle.
r r

Although repeated tests of the same individual are

not totally independent, the movement of the arena to

different parts of the field and the use of several release

points in the open field minimizes the possibility of learn-

ing, and the z statistic is used with the assumption that

the runs are independent, even though this condition is not

rigidly met.

ahen the preferred directions between individuals are

compared, only the mean direction is used, rather than the

individual tests. Thus, the statistics are not biased by an

artificially inflated sample size.

For comparison of preferred directions and homeward

directions the F rest of Watson and iilliams (1956) as sum-

marized by Batschelet (1965) was used. The null hypothesis

is that there was no difference between the preferred direc-

tion and the home direction.

The arena tests were analyzed similarly but because

the actual degree headings \:ere not made, a correction for

grouping data was made. Thle Gilroy (1965) correction factor

for grouping data into twelve sectors is 1.0115 (Batschelet,

1965). This gives a corrected r, rc, which better indicates

the concentration around the mean direction than does an

uncorrec-ted r.

Results of Field and Arena Tests

Open Field Tests

The tortoises released in the field showed no tend-

ency to head in the same direction (Figure 10). Tortoises

from the same geographic population (Figure 11) also showed

no uniformity. They were neither motivated nor able to head

in a direction which would take them back to their home burrows.

However, as is indicated by the high r values, and

the z statistics at the 0.05 level of significance, thirteen

of the twenty-three (56.5%) tortoises repeatedly chose the

same individual non-random directions from the release point.

This individual preferred direction is not homeward directed,

but the data nevertheless indicate that the individuals were


Further evidence that the tortoises were oriented is

that the paths taken by each individual were relatively

straight. If the straight line distance between the release

point and the end point of each run (best path) is plotted

against the actual path taken by tortoises (Figure 12), a

linear relation becomes evident, which would not be expected

in the absence of orientation. The mean ratio of best path

to actual path is 1.6:1.

Preferred directions of travel shown by indivi-
dual tortoises taken from different geographic
populations. Tests were perfonred in an open
field. Arrows inside circles indicate mea-
preferred directions; length of arrow in.dicat3e
strer-gth of the concentration arcurndi the ~oean
direction, r. When the length of the ar -ow
equals the rncius perfect correlation of pre-
ferred directions ar.ion trial s s indicated.

Figure 10.

H1 P1








How 1



Preferred directions of travel shoun by individual
tortoises taken from the same geographic population
Data presented as explained in Figure 10.

Figure 11.

A2 A4


0 0
A6 A?

A 14 A 71
All A l?


Figure 12. .Lngths of paths travelled by tortoises ane the
straight-line c(istances from starting, point to
end point cf each run. Trails were performed
in an open fieldC. Points represent the mean
values obtained for an inclivicdel. Line is
least-square regression.


"- 24C


cv 160



o Q



-0 3

oL--. -L 1 I I I
40 80 120 160 200 240

Arena Tests

In the arena nineteen of thirty-seven (51.4%) showed

a particular direction preference throughout the day (Table 6).

There was no tendency for the group to head in the same direc-

tion, even among tortoises from the same population. The pre-

ferred directions had no relation to the homeward direction,

although there was some similarity between those taken by the

same individuals in the open field tests (Table 7). There

was considerable variation in the tortoises' demonstration

of a direction preference at different times of day, although

the mean directions which were shown ;were very consistent

(Table 8).

After the internal clocks of the tortoises which

showed a direction preference in the arena were advanced 6

hou-rs, a significant change in preferred direction was noted.

Preferred directions of the tortoises before and after ph'ase-

shifting the internal clocks were compared, but only if the

individual oriented at the same solar time under both conditions

(Figures 13,14). The mean_ directional shift frona the original

preferred direction for six tortoises, in the morning, was

1210 co-ntercloc!kise. Four of the tortoises -which had not

shown a preferred direction at noon did so after being phase-

shifted. The controls showed no significant change in

Table 6. IHean preferred directions; concentration arcuInrc
mean direction, rc, an' z statistic for signifi-
cance of tortoises tested in the arena throughout
the day.

Tortoise r z Angle

67AL5 0.4795 4.1379** 2790

67A22 0.0769 0.1066 940

67A19 0.2924 1.3679 2940

67AC1 0.0169 0.0046 300

67AC2 0.1565 0.4406 161

67AC3 0.4065 2.9738* 540

67AC4 0.3746 2.1045 100

67ifH1 0.3351 2.0763 150

67SF2 0,3098 1.7274 2620

67SWV1 0.2312 0.621 850

67F1 0.5730 5. 9090'. 3CC0

67X1 0.5594 5.6329** 356

67GC1 0.1318 0.5286 3350

68A1 0.7485 11.2040** 182

68A2 0.2515 1.3660 1400

63A4 0.6843 9.3640** 167

68A5 0.3780 2.8560 3070

68A6 0.2133 0.9140 2380

Table 6 (Continued)

Tortoise rc

68A7 0.5958

68A8 0.2050

68A9 0.0942

68A10 0.6973

683JJ1 0.6685

68M1 0.5050

68M2 0.7367

68S1 0.4733

63SF1 0.1989

68L1 0.3070

63HC1 0.6558

68MI 0.4966

68CJ1 0.6170

68HAW1 0.5836

68RS1 0.5918

68ACP 0.4020

68RCG 0.3568

6301 0.5646

68N1 0.2575

* indicates significance
**indicates significance
preferred direction.

z Angle




















at the 0.05 level.
at the 0.01 level for





















Table 7. Comparison of preferred directions in open field
and arena tests.

Tortoise Open Field Arena Remarks
Heading (deg) Heading (deg)

---- o orientation under
either set of conditions.
232** Same direction preference
under both sets of condi-
298** Orientation, but shift
in preference
260* "




Orientation in arena only.

Orientation in field only.

* indicated 0.05 level of significance;
**indicates 0.01 level.

68 LI

68 01

68 RS

68 ACP

68 HAW 1





68 M I

63 HC1

68 Nl


O, cQ C DO i -1 L .-- c o o Cu Lfl N '!i CO '1n EG 1o c-; C o
CON to - -1 -1 C) k- 0 E-L n U) r r-4 I-4 L CG L) -4 C-
<1 -1 r-4 1 -4 CN r-, M (i CM .' C4 N -i M C' ,-o


a0 O 0 C En -7 N

C" C: r-4 ) 1N N

-t -X -
-X S
C0 r-- C' ,--I Ln rP CQ CM CO r-4 C) c ) C0 C)o In c) o-G c C' if- i'-. -" C C--
" -- N N <-' CO in o L' uCl -S uE cO --
N C C LO u NI- C) C C- 0 '- C C - -
,-4 r ,-4 CM -N CM (r- C] O C M ( N -4 un

4 I

. -4
dO J

O 00
o Ci

C )


o i
P 0

.i 0

J.- ('1

CU '-4
r H Ci

** 40

o LC

o -Jii
4-i .

+)J -4 >

0 0
C 5

C) 0

v ,'.

C00000C 00

oN '-i in r- CO o 0 -X) 01t CI
CM iii C-- CMC -C; m En> ni> c CMin o
--4 LO CC N r 1 -c --4 CO c') En Eli r-


Co c) 7 Ec -C N n C Li U7 0 C-40 C 7 Co ) (C CI N LT n c
r C o ( D- C o 0 ('c o C O n -E r-i. co o o r- C'J D3 CN CNN
r-4 r-4-) 01 C .-4 -4 (' r- 4 '1 CN --4 r-4 C ,--

I7 U) ICl I L) r -- IN r-
0) O CN C CO LI r-1 4 NC

r-1 I-: Co C O O C( o. C

OC C c] Co Co r-4 0

; -4 CLO CO C
G-' 0 I' 0")
* *

-,t CI- -h r3o co 0" on r-@ vL - -

mc c- r- CM Ci LNn o- o Cin pI Ti co
S'-01 0 n In C 01 CO -O C, 0" '

C C r-4 C0 <)O Tn ,- CN ,-4 0 CON

r- c.3 q LI ) -- * 0 o, QD %.0
U- CO -n -i rP- u) N 11 I \n n 0 C-
N CiM Q' C, -0 -o -T -l 0C 03 -C *o L 0 CI E3 <-
* . . . . . .

D .'-' O 0 d O C ,; 0 C O(D CD 0 O c C; (,D O0 0 CI) (C -

U- C1 L-f 0 0 IJ1C UL Lt iM L u) 0 1 : L co)r 01 C t O LI- Lr;
0 1- Crn Ci in H- L0 CN LC) C ) r. C u- Cl c- -C o co Cr-
nM 01 OC; rC) rE-nEn l CM i CM' 00 E El -I

Sc -X Sc
c'- I. c c 01n u o-- n -iC- OC c?) r --c c
'kc (C' E o) (D oi En n in ; n En cE i- 41 k,0 L-
cc CO co "s,- 0 -1 o "I co -.o LT ?C -41 o, cn
c'- --O co -4 Cr- N -n EC01 C1 O C- C' C

,- 0 o0 v C 0 'l .- Ic r--4 Co )-1 C
C- CO c0 o i Cr. r. --I Cr- n L cl-- CC '
o C<'c, or i co CM i o [r) o o' r-0 CO 0

C7 ( C)C CD 0 C) C0 o C O C En C D C C (D

' Cr c N cE no 4- r- eM N r
c: < : < - < -i r-- 4 _) < <- 4>
r--. r r--- r-- --- rC - -- C- CO 0C1 cO
C CC En En CC EL En En I0 E o En E o l) CC EO I O Q

r- . '1, (7; ,-n
O 0O I"co D L-a

E C C- co C-.- Ci -
3 C r-' -'q -,-, CM

C0 C0 C) CM C) En

EnO C -O Ejn ( -I
L 0 01 N CO

iD CC CM oj CD 01
ICM En iiEn Ein En

L. kc r0- 0 C-1 r-4

^. 13 0 y3 y"> y

,-4 CO P-

C) -F' L'-

rl- r -

co -i r-4
En '0 H

Sci in E

o Lun -J-

,-4 CC C-- C--
E r-E 01- UC-
* C E n

University of Florida Home Page
© 2004 - 2010 University of Florida George A. Smathers Libraries.
All rights reserved.

Acceptable Use, Copyright, and Disclaimer Statement
Last updated October 10, 2010 - - mvs