BIOLOGY AND MIGRATORY BEHAVIOR
OF AGRAULIS VANILLA (L.)
RICHARD TERRANCE ARBOGAST
A DISSERTATION PRESENTED TO THE GRADUATE COUNCIL OF
THE UNIVERSITY OF FLORIDA
IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE
DEGREE OF DOCTOR OF PHILOSOPHY
UNIVERSITY OF FLORIDA
The writer wishes to express his thanks to the numerous
persons who contributed their assistance and suggestions during
the course of his graduate study.
Special thanks are due Dr. T. J. Walker, who first
a--oused the author's interest in insect migration and who served
as Chairman of the Supervisory Committee. His encouragement and
many suggestions during the conduction of the research and prep-
aration of the manuscript were invaluable contributions.
Much appreciation is also expressed to Dr. L. A. Hetrick,
Dr. D. B. Ward, Dr. D. H. Habeck, and Dr. E. G. F. Sauer who
served as members of the Supervisory Committee, and to Dr. C. D.
Monk who was originally a member of the committee.
Thanks also go to Dr. J. T. Creighton, former Head of
the Department of Entomology, for his assistance during the course
of the author's graduate work and to Mr. J. P. Ahrano for permis-
sion to use the field in which observations of the migration were
Finally, the writer would like to express his appreciation
to his wife for her patience during his years of graduate study
and for her assistancee in the preparation of this manuscript.
TABLE OF CONTENTS
ACKNCWLEDGMENTS . . . . . . . .
LIST OF TA3LES . . . . . . . . .
LIST OF FIGURES . . . . . ....
INTRODUCTION . . . . . . . . .
LIFE HISTORY . . . . . . . . .
Methods and Materials . . . . .
Rate of development . . . . .
Longevity of the adults . . . .
Reproductive development of the female .
Behavior of adults and larvae. . . .
The Egg and Oviposition . . . . .
The Larva . . . . . . ... .
The Pupa . . . .. . . . . .
The Adult . . . . . . ... .
Emergence . . .. . . . . .
Sex ratio . . . . . .. .
Reproductive development of the female
Mating behavior . . . . ...
Longevity,of the adults . . . .
DISTRIBUTION OF LARVAL FOOD PLANTS . . . .
POPULATIONS OF Anraulis vanilla IN THE VICINITY
GAINES3V LLE, FLORIDA . . . . . . .
Description of Observed Breeding Areas . .
Fluctuations in the Observed Populations .
THE FALL MIGRATION . . . . . . .
Methods and Materials . . . . .
Observational setup . . . . .
m father observations . . .....
Sp!ed and direction of flight. . . .
DCnsity of the migration . . . .
Reproductive maturity of the females and
Description of the Migration . . . .
Characteristics of the migratory flight.
Variations in migration density. . .
Reproductive maturity of the females and
ORIENTATION EXPERIMENTS. . . . . . . . . 58
Introduction . . . . . . . . . 53
Methods and Materials. . . . . . . . . 59
Results . . . . . . . . . . 66
DISCUSSTCN AND CONCLUSIONS . . . . . . ... 84
SU~vI ARY. . . . . . . . . . . ... . 91
LITERATURE CITED . . . . . . . . . . 93
BIOGRAPHICAL SKETCH ................... 97
LIST OF TABLES
Observed Number of Eggs and Larvae in Area 1
(.9o4-05) . . . . . . . . .
C..., ,rved Number of Eggs and Larvae in Area 2
( s19u4-65) . . . . . . . . . ..
O"' ,rved Number of Eggs and Larvae in Area 3
(1 .4-65 ). . . . . . . . . . .
OCis(rved Number of Eggs and Larvae in Area 4
(19u4-65) . . . . . . . . .
Density of Migration and Weather Observed 1300-
1400 EST (1964) . . . . . . . .
Tirme of Beginning of Migration on Various Days
Migration Densities Observed at Various Times on
September 23, 1964 . . . . . . . .
LIST OF FIGURES
1. Fourth instars of Agraulis vanilla nigrior
Michener. . . . . . . . . . ... 11
2. Fifth instar of Agraulis vanilla nigrior Michener. 12
3. Feeding rhythms of fifth instars. . . . . 15
4. Duration of the various instars at 23-240 C. (left)
and at 28.5-29.50 C. (right). . . . . . 16
5. Dorsal view of pupa of Agraulis vanilla nigrior
Michener. . . . . . . . . . .. 18
6. Lateral view of pupa of Agraulis vanilla nigrior
Michener . . . . . . . . . .. 19
7. Duration of pupa at 23-24 C. . . . . .. 20
8. Duration of pupa at 28.5-29.50 C. . . . .. .20
9. Time of emergence of the adult. .... . .. . 21
10. Recently emerged adult of ANraulis vanilla nigrior
Michener hanging from the pupal skin. . . . 23
11. Breeding areas in the vicinity of Gainesville,
Florida . . . . . . . . . . . 28
12. Setup for observing migrations. . . . . ... 37
13. Wind vane and anemometer used in migration studies. 38
14. Wind speed and direction dials used in migration
studies . . . . . . . . ... . . 39
15. Wind vane and anemometer mounted on tripod for
observations. . . . . . . . . .. .41
16. Convention used in measuring angles between track
and wind direction (6) and between track and
course (). . . . . . . . . ... .. 43
17. Tracks of migrants observed during the fall of
1963 (A) and 1964 (B) . . . . . . ... 47
18. Tracks of migrants observed at various times of day
on September 23, 1964 . . . . . . . 48
19. Tracks of migrants and mean wind speed and direction
observed between 1300 and 1400 EST on various days
in the fall of 1964 . . ... . . . . 49
20. Ground speed of migrants flying in calm air . . 51
21. Ground speed of migrants flying against a head wind
of 1-5 miles per hour (A) and with a tail wind of
1-4 miles per hour (B). . . . . . . ... 52
22. Air speed of migrants flying against a head wind of
1-5 miles per hour (A) and with a tail wind of 1-4
miles per hour (B). . . . . . . . . 53
23. Cage used for orientation tests . . . ... .61
24. Cutaway view of controlled photoperiod cabinet used
in clock resetting experiments. . . . . .. .65
25. Orientation of Individual A when tested between
0930 and 1030 EST,November 6, 1964, without reset-
ting the internal clock . . . . . ... 69
26. Orientation of Individual A when tested between
1440 and 1540 EST, November 6, 1964, without re-
setting the internal clock. . . . . . ... 70
27. Orientation of Individual B when tested between
0950 and 1050 EST, October 30, 1964, without reset-
ting the internal clock . . . . . ... 71
28. Orientation of Individual B when tested between
1410 and 1510 EST, November 5, 1964, without reset-
ting the internal clock . . . . . ... 72
29. Orientation of Individual C when tested between
1240 and 1340 EST, November 5, 1964, without reset-
ting the internal clock . . . . ... ... 73
30. Orientation of Individual D when tested between
1510 and 1610 EST, November 5, 1964, without reset-
ting the internal clock . . . . . ... 74
31. Orientation of Individual D when tested between
1350 and 1450 EST, November 10, 1964, after setting
the internal clock back 6 hours (5 days in out-of-
phase cycle) . . . . . . . . . 75
32. Orientation of Individual D when tested between
1310 and 1410 EST, November 11, 1964, after setting
the internal clock back 6 hours (6 days in out-of-
phase cycle). . . . . . . . . . 76
33. Orientation of Individual E when tested between
1030 and 1130 EST, November 9, 1964, without re-
setting the internal clock. . . . . . ... 77
34. Orientation of Individual E when tested between
1240 and 1340 EST, November 17, 1964, after setting
the internal clock back 6 hours (8 days in out-of-
phase cycle). . . . . . . . . . 78
35. Orientation of Individual F when tested between
0930 and 1030 EST, November 9, 1964, without re-
setting the internal clock. . . . . . ... 79
36. Orientation of Individual F when tested between
1430 and 1530 EST, November 16, 1964, after setting
the internal clock back 6 hours (7 days in out-of-
phase cycle . . . . . . . ... . . 80
37. Orientation of Individual G when tested between
1430 and 1530 EST, October 18, 1964, after setting
the internal clock back 6 hours (5 days in out-of-
phase cycle). . . . . . . . . .. .81
38. Orientation of Individual H when tested between
1530 and 1630 EST, October 20, 1964, after setting
the internal clock back 6 hours (3 days in out-of-
phase cycle) . . . . . . . . . . 82
39. Orientation of Individual I when tested between
1430 and 1530 EST, October 23, 1964, after setting
the internal clock back 6 hours (4 days in out-of-
phase cycle) . . . . . . . . . . 83
Migration has been defined by Schneider (1962) as a
prolonged escape movement in which there is a tendency to main-
tain a constant direction and which results in the permanent or
periodical abandonment of a habitat. Migration can be classi-
fied as dispersive, contractive, or collective according to the
spatial effect. Many insects migrate in this sense, and numerous
accounts of their migrations can be found in the literature. Re-
cent reviews of insect migration have been published by Schneider
(1962) and Williams (1957, 1958).
In most of the migratory insects which have been care-
fully studied, the migratory direction is determined largely by
the prevailing wind. Swarms of the desert locust (Schistocerca
gregaria Forsk.) are carried downwind, and since the winds in
the lower few thousand feet of the atmosphere may be regarded
ultimately as blowing from areas of high pressure to areas of
low pressure, the downwind displacement results in movement into
areas of low pressure where abundant rainfall produces conditions
favorable for the reproduction of the locust (Rainey, 1951). The
migratory direction of the coccinellid Hippod,-.ia convergens
Gu6rin-Meneville in California is determined by the prevailing
winds at its flight level (Hagen, 1962). In the summer,.the
prevailing winds at this level are from the low lands toward
the mountains where aggregation occurs, and in the winter, they
are from the mountains to the low lands. The mosquito Aedos
taeniorhynchus Wiedemann migrates downwind (Provost, 1952,
In contrast to this, many migrating butterflies main-
tain, over long distances, a constant direction which is in-
fluenced little by wind, topography, or time of day (Williams,
1958). The determination of migratory direction appears to be
under the control of the insect itself, but the nature of the
underlying orientation mechanism has remained obscure. Only
two migratory butterflies have been studied in great detail
(Nielsen and Nielsen, 1950; Nielsen, 1961; Urquhart, 1960),
and these studies did not include an investigation of the
orientation mechanism. Recent work on the orientation of other
animals has suggested new approaches to this problem. (For
references, see Long Island Biological Association, 1960.) The
purpose of the research reported here, was to study the migra-
tory behavior of the gulf fritillary, Agraulis vanilla (L.),
against the background of its general biology and to investigate
the nature of the orientation mechanism involved in maintaining
the migratory direction.
The gulf fritillary is a member of the essentially neo-
tropical nymphalid subfamily Heliconiiae. Michener (1942) recog-
nizes eight subspecies as follows:
Agraulis vanilla vanilla (Linnaeus) occurs in northern
South America, Panama, and the southernmost of the Lesser
Agraulis vanilla insularis Maynard ranges throughout
the Bahamas and the Antilles except for the southern-
most Lesser Antilles.
Agraulis vanilla maculosa (Stichel) is found in northern
Argentina, Paraguay, southern Brazil, and Chile.
Aoraulis vanilla forbesi Michener occurs in the coastal
region of Peru.
Agraulis vanilla galapagensis Holland is restricted to
the Galapagos Islands.
Agraulis vanilla lucinia C. and R. Felder occurs on the
eastern side of the Andes in Ecuador, Peru, and Brazil.
Agraulis vanilla incarnata (Riley) is occasionally
found as far north as British Columbia but more com-
monly occurs in southwestern United States, Mexico,
and Central America.
Agraulis vanilla nigrior Michener is the subspecies of
southeastern United States but is occasionally found as
far north as New York.
The migratory habit has evolved in at least two of these sub-
species. Definite migrations of A. v. niorior to the north in
the spring and to the south in the fall have been observed in
Florida (Williams, 1958). Hayward (1962) reported an eastward
migration of A. v. maculosa at Tucaman, Argentina, on January
7, 1961. The research reported here was restricted to A. v.
niorior. This form intergrades with A. v. incarnata in Texas
and thence southward but is distinct from A. v. insularis.
The early stages of A. v. nigrior were described long
ago (Edwards, 1880; Scudder, 1889), but no detailed studies of
its biology have been made. More recently, detailed descrip-
tions of the early stages of A. v. vanilla have been published
(Beebe, Crane, and Fleming, 1960) and certain details of the
adult morphology have been described by Emsley (1963). The
biology and behavior of the larvae, pupae, and emerging adults
of A. v. vanilla were discussed by Alexander (1961a, 1961b)
and compared with those of other members of the subfamily
occurring in Trinidad.
Methods and Materials
Rate of development
Rearing to determine the rate of development was ini-
tiated with eggs obtained from females which had mated in cages.
These cages were 14 inches square and 24 inches high with sides
of aluminum screen and top and botton of 1 inch plywood. A 9
inch square opening in the top covered with a plate of glass
served as a door. One or more males and one or more females
which had been reared in captivity from larvae were confined
in a cage and placed outdoors in direct sunshine. Mating was
obtained in this manner using males 1-3 days after emergence
and females 0-3 days after emergence. Mating occurred in
both the morning and the afternoon and usually within 2 hours
after the butterflies were placed in the cage. Occasionally
no mating occurred even when the butterflies were left in the
cage all day, but frequently the same butterflies would mate
when placed in the cage the next day.
One to 3 days after mating, the females were confined
individually with cuttings of Passiflora incarnata L. This
was accomplished either by introducing a jar of water containing
the cuttings into the mating cage or by placing a cheesecloth
bag over cuttings contained in a flower pot full of damp sand
and then confining the butterfly in the bag. The latter
method was the most satisfactory. The confined butterflies
deposited many eggs on the side of the cage or on the cheese-
cloth bag. It was difficult to remove the eggs from the side
of the cage without damaging them, but those deposited on the
cheesecloth could be removed by cutting out the piece of material
to which they were attached.
The captive butterflies were fed once a day on sugar
water of variable concentration. A cotton ball was saturated
with the solution and placed in a small dish containing addi-
tional solution. The butterfly to be fed was grasped by the
wings and its tarsi touched to the cotton ball. Usually this
caused the butterfly to extend its proboscis and begin feeding.
Butterflies which refused to feed when touched to the cotton
ball could often be induced to feed by uncoiling the proboscis
and inserting its tip in the sugar water. Once feeding began,
the butterflies remained quietly on the cotton ball until
feeding was completed.
The time required to complete each stage of development
at each of two constant temperatures was determined by rearing
the insects in a constant temperature cabinet. In one case, the
temperature in the cabinet was maintained at 23-240 C. and in
the other case at 28.5-29.50 C. To determine the time required
for the eggs to hatch, 1-7 hour old eggs were placed in the cab-
inet in a covered petri dish and checked once a day for hatching.
To determine the duration of the remaining stages, the insects
were reared individually from egg to adult in pint fruit jars.
The caterpillars were kept supplied with one or two leaves of
the food plant. Each individual was examined once a day and
the dates of hatching, molting, pupation, and emergence were
The larvae suspended for pupation from the lids of the
jars or from the point just below the neck where the sides
curve inward. In both cases the point of attachment was too
smooth and many of the chrysalides fell. This difficulty was
overcome by placing a piece of coarse-textured paper (toweling)
beneath the lid and inserting a screen cylinder which extended
from the bottom to the top of the jar. The larvae crawled up
the cylinder and suspended successfully from the paper.
Longevity of the adults
In order to gain some idea of the adult life span,
butterflies which had been reared from eggs in pint fruit
jars at a temperature of 23-240 C. were confined individually
in cubic screen cages 4 inches on a side and left in the con-
stant temperature cabinet at the same temperature. They were
fed daily as described above.
Reproductive development of the female
The elapsed time between the emergence of a female and
the maturation of the eggs in the ovaries was determined. For
this purpose, larvae in various stages of development obtained
from the field and from eggs laid by females in'captivity were
reared together on food plant cuttings in screen cages in a
greenhouse. The larvae were crowded but were kept abundantly
supplied with food, and the adults which were produced were of
normal size. When the adults began to emerge, it became
apparent that all emergences occurred during the morning, so
the cages were checked each day at 1 hour intervals between
0600 and 1200 Eastern Standard Time. Each time, all the but-
terflies that had emerged during the preceding hour were re-
moved and the time of emergence recorded as the time midway
between the last two checks. Thus the time of emergence re-
corded for each individual was accurate within + 30 minutes.
The females were confined in pint fruit jars with screen lids
and left in the greenhouse for periods of 0-48 hours with 6
hour increments. They were then preserved by injecting them
with 10 per cent formalin and placing them in 3 per cent forma-
lin. The preserved butterflies were dissected and the ovaries
Behavior of adults and larvae
The behavior of larvae of all stages and of adults was
observed in the field and in screen cages. In order to observe
the manner in which periods of larval feeding alternate with
periods of rest and walking, fifth instars were brought into
the laboratory from the field, and each was placed on a bunch
of food plant cuttings in a screen cage. The larvae were
observed for 4 hours, and the times at which feeding began and
ended were recorded for each individual. These observations
were made in the afternoon of a cloudy day, and the cages were
located in front of a southwest window. The sun occasionally
shown through the clouds, and as a result, the light intensity
in the cages varied considerably. However, there was no cor-
relation between feeding activity and the light intensity.
The Eoa and Cviposition
The larvae feed upon Passiflora incarnata and probably
other plants of the genus, and the females oviposit upon and
in the vicinity of these plants. Ovipositing females fly low
over and through the vegetation, frequently pausing to hover
about individual plants. When the butterfly contacts the larval
food plant, it alights and deposits an egg. Actual contact with
the plant seems to be essential for the release of oviposition.
Mated females confined with food plant cuttings deposited an
egg only after actually touching the cuttings. Mated females
confined without cuttings did not oviposit. The stimulus which
releases the act of oviposition is probably provided by a
chemical which is characteristic of this genus of plants and
which is detected by gustatory organs in the tarsi or antennae
of the butterfly.
The position of the egg depends upon where the female
is able to gain a foothold after contacting the food plant.
Most eggs are deposited upon the upper surfaces of the leaves,
but some are deposited on the undersurfaces, stems, tendrils,
buds, and nearby objects. The butterfly curves the abdomen
ventrad until the oviducal pore contacts the object upon which
it has alighted, whereupon it deposits an egg. In this process,
the tip of the abdomen may be turned through more than 270
degrees. A female which lands on a leaf surface near the margin
of the leaf often curves the abdomen around the margin and de-
posits an egg on the opposite surface. It was noted previously
that females confined in cages with food plant cuttings deposit
many of their eggs on the side of the cage. This results from
the fact that the butterflies frequently land on the side of
the cage after fluttering about and coming in contact with the
cuttings. Ovipositing females were observed on a small plant-
ing of P. incarnata along the side of a building. Frequently
they would strike the side of the building and alight after
coming in contact with the plants. Each time this occurred,
they deposited an egg on the side of the building.
Observations of females in the field and in cages
indicate that there is always a period of flight after the
deposition of each egg. Several eggs were often found on the
same leaf, but this was probably the result of several visits.
In some cases, this was borne out by the fact that some of the
eggs were yellow, while others were reddish brown (see follow-
ing paragraph). This behavior distributes the eggs more evenly
over the available food supply. Furthermore, larvae in captiv-
ity were observed chewing into unhatched eggs, even in the pres-
ence of abundant food. The spacing of the eggs brought about by
the egg laying behavior of the female reduces losses of this sort.
The eggs are yellow when deposited, but become reddish
brown within 24 hours. Thirty-nine eggs kept at a temperature
of 23-24 C. hatched on the fifth day. Thirteen eggs kept at
a temperature of 28.5-29.50 C. hatched on the third day.
The larva is illustrated in Figures 1 and 2.
'iure 1. Fourth instars of s v.. iI _
Fifth instar of Agraulis vanilla nigrior
Upon hatching, the young larva consumes the ecgshcll
before beginning to feed upon the food plant, and newly hatched
larvae were frequently observed eating unhatched eggs which
they happened to encounter, even in the presence of abundant
food. The caterpillars feed mainly upon the leaves, but when
food becomes scarce, they feed upon the exterior portions of
the buds and fruit and strip the epidermis and cortex from the
stems. The first three instars generally feed away from the
leaf margins. Newly hatched larvae feeding upon thick tough
leaves eat only the epidermal cells of one surface and the
mesophyll cells, leaving the epidermis of the opposite surface
intact. Newly hatched larvae feeding upon thin tender leaves
and older larvae in general cut holes completely through the
leaves. The two final instars feed at the leaf margins. When
ample food is available, the portion of a leaf which is eaten
before it is abandoned varies considerably. The petioles are
almost always left but may be eaten, at least in part, when
food becomes scarce. Larvae were never observed eating one
another, even when kept in crowded cages without food. In tvo
instances, however, larvae kept under these conditions were
observed eating chrysalides.
Alexander (1961a) studied the feeding rhythms of 10
species of heliconiine butterflies in Trinidad, including
Agraulis vanilla vanilla (L.) and found that feeding periods
of about 20-30 minutes alternate with rest periods of similar
duration. Four fifth instars of Aoraulis vanilla nin ior
Michener observed feeding in the laboratory showed a similar
feeding rhythm. Feeding periods of about 10-60 minutes alter-
nated with periods of rest and walking of about 10-90 minutes
Frequently a larva cuts a girdle 2-3 mm. wide around
a stem. This girdle extends only through the epidermis and
cortex and does not kill the plant. Larvae were observed cut-
ting these girdles many times in the field and in cages, and
large numbers of plants-were found with healed wounds of this
kind. In one case which was timed, girdling of the stem re-
quired 10 minutes (Figure 3D). Alexander (1961a) found similar
behavior in the larvae of severalheliconiine butterflies but
apparently not in the larvae of A. v. vanilla. The caterpillars
of Heliconius melpomene (L.) and H. ricini (L.) chew furrows
across the midribs of leaves, while the caterpillars of Dryas
julia (Fabr.) and Dryadula phaetusa (L.) cut narrow channels
from the margins to the midribs. In all cases this behavior
is exhibited during a rest period and it is therefore unlikely
that the material is eaten solely for its nutritive value.
Alexander suggested that the furrowing and channeling behavior
might be a form of territory marking. Perhaps the girdling
behavior of A. v. nigrior also represents territory marking,
but no evidence was obtained to either prove or disprove this
The rate of development of the larva varies with tem-
perature (Figure 4). The mean duration of the larval stage
was 15.7 days at 23-240 C. and 11.5 days at 28.5-29.50 C.
These means are significantly different at the 1 per cent level.
I I I I I I I
B I I. I
i F i i it
Ij P s
Figure 3. Feeding rhythms of fifth instars. Black
represents resting and walking, white feeding, and
g girdling. Each division at the top is 15 minutes.
; 1 i i J
i IB -1 1
---. ,1, ....- .. --~I-. ;~-.-
i ; .+,
.: ..r -
i 3 RD 3 RD
o tr .
-" 0 -'-- .---
0 5.5 0 5.5
Figure 4. Duration of the various instars at
23-24 C. (left) and at 28.5-29.50 C. (right).
The molting and pupation behavior of A. v. vanilla
was described in detail by Alexander (1961b). This probably
does not differ in most respects from that of A. v. nigrior.
However, Alexander reported that A. v. vanilla almost in-
variably pupates on the stem, tendrils, flowers, or leaves
of its food plant. A. v. nigrior usually pupates on objects
at some distance from the food plant.
The pupa is illustrated in Figures 5 and 6.
Histograms of the duration of this stage are presented
in Figures 7 and 8. The mean duration was 11.7 days at 23-
240 C. and 7.5 days at 28.5-29.50 C. These means are signifi-
cantly different at the 1 per cent level.
Apparently most emergence from the chrysalid occurs
during the morning. Butterflies which were reared from larvae
in a greenhouse during September under the natural day-night
cycle emerged between 0600 and 1200 with maximum emergence oc-
curring between 0900 and 1000 (Figure 9). On the days during
which emergence occurred, morning civil twilight began at ap-
proximately 0550 and sunrise was at approximately 0610. Alexan-
der (1961b) noted a similar time of emergence for the heliconiine
butterflies Heliconius erato (L.) and H. melpomene in Trinidad.
Upon emergence, the butterfly hangs from the pupal skin
Figure 5. Dorsal view of pupa of Agraulis vanilla
Figure 6. Lateral view of pupa of Agraulis vanilla
S5. 0.5 5.5
Figure 7. Duration of pupa at 23-240 C.
I J . - -I- day
0 5 0.5 5.5
Figure 8. Duration of pupa at 28.5-29.50 C.
Figure 9. Time of emergence of the adult.
t ;me(E ST)
until the wings have expanded and dried (Figure 10). Expansion
of the wings requires from 3-4 minutes, and they are dry enough
for flight within an hour.
Of 265 butterflies which emerged in captivity 54 per
cent were males and 46 per cent females. This sex ratio is
not significantly different from 1:1 at the 5 per cent level.
Reproductive development of the female
Dissections of preserved females indicated that most
individuals reach reproductive maturity 12-18 hours after
emergence, although a few individuals are still not mature 48
hours after emergence. Mature eggs could be recognized in the
preserved specimens by their morphology. In these eggs, there
was a central dark yellow mass occupying no more than 3/4 the
volume of the egg, and the transparent chorion was very con-
spicuous. In the immature eggs, the entire volume of the egg
was occupied by cream colored or pale yellow material, and the
chorion, when it was present, was not so conspicuous. Length
proved to be an unreliable criterion for determining egg
maturity. Oviposited eggs which subsequently produced larvae
ranged in length from 1.30-1.60 mm. All the mature eggs fell
within this size range. However, in several cases immature
eggs 1.30 mm. long were found.
Females will mate within 24 hours after emergence and
will mate at least as long as 3 days after emergence as pointed
out in the-discussion of methods and materials. Males will
Figure 10. Recently emerged adult of Agraulis
vanilla nigrior Michener hanging from the pupal
mate at least as early as 1 day after emergence and as late as
3 days after emergence. No experiments were conducted to de-
termine the two extremes of age at which each sex will mate.
No courtship was observed in the field, and that ob-
served in cages was probably greatly abbreviated. The confined
butterflies alternately rested on the sides of the cage and
fluttered against the sides and top of the cage. When a
fluttering male approached a resting female, the female usually
responded by elevating the abdomen and vibrating the wings with
the hind wings opened 90 degrees and the fore wings opened about
45 degrees. Then the male would land beside the female facing
the same direction, vibrate his wings, and thrust at the female's
abdomen with the end ,of his abdomen. During this process, the
female would continue to vibrate her wings in the described
position. When the ends of the abdomens came in contact, the
male would grasp the end of the female's abdomen with his claspers
and then turn 180 degrees to assume the copulatory position.
Occasionally mating occurred even when the female showed no
response to the male. When both the male and the female were
fluttering, the female would land on the side of the cage, and
the process would proceed as described. Copulation lasted from
45-60 minutes. A detailed study of the mating behavior including
the roles played by motion, color, size, shape, pattern, and odor
remains to be done. Crane (1955) has made such a study of the
related butterfly Heliconius erato, and many of her conclusions
will probably be found to apply equally well to Agraulis vanilla.
Longevity of the adults
Nine males and nine females kept in a constant tem-
perature cabinet at 23-24o C. lived from 14-27 days after
emergence. The mean life span was 18.4 days with a standard
deviation of 3.0 days. There was no significant difference
between the mean longevity of the males and females at the
10 per cent level. Since these butterflies were unmated and
were not subjected to the same stresses normally encountered
in nature, the observed life span is perhaps slightly longer
than that of the species in nature.
DISTRIBUTION OF LARVAL FOOD PLANTS
Presumably any species of Passiflora can serve as a
food plant for Agraulis vanilla nigrior, but only P. incarnata
was examined for eggs and larvae. Only two species of Passi-
flora are widely distributed in the United States (Killip, 1938).
Passiflora incarnata ranges from Virginia to Missouri and south
to Florida and Texas but has been introduced farther north.
Passiflora lutea L. ranges from Pennsylvania to Illinois and
Kansas and southward to Florida and Texas.
Seven species of Passiflora have been recorded from
Florida (Killip, 1938), but no detailed account of their distri-
bution within the state has been published. A general idea of
their distribution was obtained from specimens in the herbarium
of the University of Florida Agricultural Experiment Station.
Apparently P. sexflora Juss., P. pallens Poepp. ex Mast., P.
multiflora L., and P. foetida L. are restricted to the extreme
southern portion of the state. Passiflora lutea occurs in
northern penninsular Florida and in West Florida. Passiflora
suberosa L. ranges throughout peninsular Florida. Passiflora
incarnata occurs throughout the state but is apparently less
common in the south and west. With the exception of P. incar-
nata, which occurs largely in old fields and along roadsides,
these plants appear to occur largely in wooded areas.
POPULATIONS OF Agraulis vanilla IN THE VICINITY
OF GAINESVILLE, FLORIDA
Description of Observed Breeding Areas
Adult gulf fritillaries can be found in almost any open
situation. However, at least in north central Florida, their
reproduction is limited to rather small widely scattered areas
by the patchy nature of the food plant distribution. Four such
areas in the vicinity of Gainesville, Florida, with growths of
Passiflora incarnata were selected for observation (Figure 11).
Area 1 measured approximately 120 x 220 feet and lay in an old
field with growths of broomsedge (Andropogon glomeratus /Walt./
BSP) and Blackberry (Rubus sp.). Late in the summer the food
plants were almost hidden by a dense growth of ragweed (Ambro-
sia artemisiifolia L.). Area 2 measured approximately 50 x 70
feet and was also in an old field with growths of broomsedge
and blackberry, but ragweed was absent. Area 3 measured approxi-
mately 90 x 200 feet. The dominant plants were broomsedge,
blackberry, groundsel-tree (Baccharis halimifolia L.), elder
(Sambucus canadensis L.), and shining sumac (Rhus copallinum
L.). The growth of Passiflora in this area was extremely dense.
Area 4 measured approximately 90 x 140 feet and lay in a pine
flatwoods. Longleaf pine (Pinus palustris Mill.) was the domi-
nant plant. There were widely scattered shrubs including
US 441 FIo 24
A ir-p rt
GAINED SV I LLE
Fl Ia 24 N
y^ \a 21Prairie I
Figure 11. Breeding areas in the vicinity of
waxmyrtle (Myrica cerifera L.), live oak (Quercus virginiana
Mill.), water oak (Quercus nigra L.), Daubentonia punicea (Cav.)
DC, groundsel-tree, and shining sumac. Blackberry was a dominant
plant of the ground cover.
Fluctuations in the Observed Populations
It was not possible to obtain an estimate of the actual
number of eggs and larvae present in a given breeding area at a
given time. However, it was possible to obtain an index of this
number and thereby observe fluctuations. This was done by arbi-
trarily choosing 75 plants in such a way that all portions of
the breeding area were represented and counting the number of
eggs and larvae on the terminals and first six leaves. Later
in the season, when the plant population began to decline, it
was sometimes not possible to find 75 plants. In this case,
as many plants as could be found were examined.
The number of plants examined and the number of eggs
and larvae found at various times between July, 1964, and May,
1965, are presented for each of the four observed breeding areas
in Tables 1-4. In all cases except Area 4, the number of eggs
and larvae increased slightly during July and then remained
fairly constant until the end of August when there was a rapid
increase. Area 4 was apparently unoccupied until the latter
part of August when large numbers of eggs and larvae appeared
there. In all cases this rapid increase in numbers caused a
marked reduction in the available food. The available food was
reduced further in Area 1 by large numbers of blister beetles
OBSERVED NUMBER OF EGGS AND LARVAE IN AREA
Larval Instar Eggs
Date Plants Eggs and
1 2 3 4 5 Larvae
2May 75 0 0 0 2 1 2 5
1 2 5
2 May 75 0
0 0 2
OBSERVED NUMBER OF
EGGS AND LARVAE IN AREA 2 (1964-65).
Larval Instar Eggs
Date Plants Eggs and
1 2 3 4 5 Larvae
8 Jul 75 0 0 0 0 1 0 1
16 Jul 75 0 0 0 0 0 2 2
22 Jul 75 1 0 2 3 3 3 12
4 Aug 75 5 2 6 3 7 10 33
12 Aug 75 5 0 0 1 3 9 18
19 Aug 75 17 1 0 2 3 1 24
26 Aug 75 33 8 2 0 2 1 46
6 Sep 75 36 13 7 0 4 15 75
24 Sep 25 13 16 12 6 1 9 57
7 Oct 5 3 2 1 0 2 1 9
2 Nov 13 3 1 2 0 2 0 8
18 Nov 10 10 29 5 3 3 5 55
2 Dec 13 0 2 2 4 8 0 16
20 Dec 4 0 0 0 0 0 0 0
14 Mar 0 0 0 0 0 0 0 0
3 Apr 24 0 0 0 0 0 0 0
24 Apr 75 0 0 0 0 0 1 1
OBSERVED NUMBER OF EGGS AND LARVAE IN AREA 3 (1964-65)
Larval Instar Eggs
Date Plants Eggs and
1 2 3 4 5 Larvae
24Apr 75 1 0 0 0 0 2 3
0 2 3
24 Apr 75 1
0 0 0
OBSERVED NUMBER OF EGGS AND LARVAE IN AREA 4 (1964-65)
Larval Instar Eggs
Date Plants Eggs and
1 2 3 4 5 Larvae
3 Jul 75 0 0 0 0 0 0 0
18 Jul 75 0 0 0 0 0 0 0
27 Jul 75 0 0 0 0 0 0 0
6 Aug 75 0 0 0 0 0 0 0
24 Aug 75 32 1 1 1 1 3 39
2 Sep 75 7 7 8 15 16 14 67
24 Sep 0 0 0 0 0 0 0 0
14 Mar 4 0 0 0 0 0 0 0
3 Apr 75 0 0 0 0 0 0 0
2 May 75 0 0 1 3 0 0 4
(Meloidae) which seemed to be particularly attracted to Passi-
flora. In Area 3, where the growth of Passiflora was very dense,
the food supply remained abundant in spite of the reduction, and
the destroyed plants were soon replaced by new growth. In the
other areas there was only a small amount of new growth, and it
soon became impossible to find 75 plants. The reduction in food
plants was especially marked in Area 4 where it was impossible
to find any plants 31 days after eggs and larvae were first dis-
covered there. The insect was therefore absent from this site
until the following spring. In the other localities, the number
of eggs and larvae began declining in October. Area 2 showed
a second peak in mid-November and then declined again. On Decem-
ber 1, a minimum temperature of 300 F. was recorded at Gaines-
ville Municipal Airport (U.S. Weather Bureau, 1964). This tem-
perature was taken in an instrument shelter 5 feet above the
ground, and it is reasonable to expect that the temperature in
the open near the ground was slightly lower due to the loss of
heat by radiation. The terminals and young leaves of exposed
plants in areas 1 and 3 were damaged by this low temperature,
while those growing under shrubs were not. No damage occurred
to the plants in area 2, probably due to the moderating in-
fluence of the nearby water (Figure 11). The larvae were appar-
ently not injured by this frost. No further freezing tempera-
tures occurred until January when minimum temperatures below
320 F. were recorded at the airport on 8 days and below 280 F.
on 3 days (U.S. Weather Bureau, 1965). This was sufficiently
low to kill all the plants. New shoots emerged in March. Ap-
parently the first eggs were deposited early in April.
These observations indicate considerable overlapping
of the generations. This is to be expected since the oviposi-
tion period for each female is relatively long, and the time
required for the insect to pass from egg to adult is relatively
There were seldom more than a few adults present in a
breeding area when observations were made regardless of time
of day and size of larval population. Apparently the adults
spend only brief periods in the breeding areas for oviposition
and possibly for mating.
THE FALL MIGRATION
Methods and Materials
Migrations of the gulf fritillary were observed during
October, 1963, and September, October, and November, 1964, in a
large open field near Gainesville, Florida. A circle 50 feet
in diameter was laid out in this field. The circle was marked
by bottles 5 inches high and 1.5 inches in diameter buried up
to their necks at 20 degree intervals around the circumference.
Before each observation period, a stake 4 feet long was in-
serted in each of the bottles. The stakes were labeled with
large black numbers beginning with 1 at north and continuing
clockwise to 18. After each observation period, the stakes
were removed and the bottles capped. The circle is shown set
up for observations in Figure 12. Observations were made from
stations 8 feet outside the circle. There was one station
directly outside each stake, and they were used in random order
with a different station for each day of observations.
Wind speed and direction were measured by means of a
cup anemometer and wind vane connected to remote wind speed and
direction indicators (Nassau Windmaster, Model No. 409, Science
Associates, Princeton, New Jersey, Figures 13-14). The anemom-
Figure 12. Setup for observing migrations.
Figure 13. Wind vane and anemometer used in migration studies.
Figure 14. Wind speed and direction dials used in migration
eter and wind vane were mounted at a height of 5 feet on a
tripod placed 25 feet outside the circle and 20 degrees clock-
wise from the observation station (Figure 15). The remote in-
dicators were at the observation position. Wind observations
were made every half hour in 1963 and every 5 minutes in 1964.
Sky cover and weather conditions were recorded every hour un-
less a major change occurred during the course of an hour.
Sky cover was classified according to the fraction of the
celestial dome covered by clouds as follows:
Clear -- less than 1/10
Scattered -- 1/10-5/10
Broken -- 6/10-9/10
Overcast -- more than 9/10
The term thin was applied to any of the above when the sun was
clearly visible through the clouds. Hourly temperatures were
obtained from U.S. Weather Bureau records for Gainesville Mu-
nicipal Airport which is approximately 7 miles from the obser-
vation site. The temperatures were taken in a standard in-
strument shelter at a height of 5 feet and were probably rep-
resentative of air temperatures over the entire Gainesville
Speed and direction of flight
In describing the flight speed and direction of migrat-
ing butterflies, it is convenient to employ the terminology of
aircraft navigation. The following terms are used:
Track -- the migrant's direction relative to the ground.
Ground speed -- the migrant's speed relative to the ground.
Figure 15. Wind vane and anemometer mounted on
tripod for observations.
Course -- the direction in which the migrant is heading.
Air speed -- the migrant's speed relative to the air.
Wind direction -- the direction from which the wind blows.
Wind speed -- speed of the wind.
The track of each migrant crossing the circle was deter-
mined by recording the numbers of the stakes between which it
entered and left the circle. It was assumed that the points of
entrance and exit were midway between the stakes through which
the migrant passed. A line through the center of the circle
parallel to the line through these two points gives the track
within + 10 degrees. Ground speed was determined by measuring
with a stop watch the time required for a migrant to cross the
circle and noting the points of entrance and exit as described
above. The wind speed and direction read at the time the mi-
grant left the circle were assumed to represent the wind speed
and direction as it was crossing the circle. The ground speed,
g, in miles per hour is given by the expression
g = 34.1 t
where t is the time in seconds required for the migrant to
cross the circle and co is the angle subtending the chord of
the circle which represents the path of the butterfly through
In calculating air speed and course, angles were
measured counter-clockwise from 0 to 180 degrees and clock-
wise from 0 to -180 degrees with respect to the vector rep-
resenting the track, using the tail of the vector as the origin
(Figure 16). The air speed, v, is given in miles per hour by
Wind or Course
-180 I,* 0O Track
Wind or Courie
Figure 16. Convention used in measuring angles
between track and wind direction (e) and between
track and course (0).
v = g w cos 8,
where w is the wind speed in miles per hour and 9 is the angle
between the wind direction and the track. The true bearing of
the course, c, is given by the expression
c = a 0,
where a is the true bearing of the track, 0 is the angle between
the course and the track, and
w sin 6
sin 0 = -
Density of the migration
The density of the migration is expressed as the number
of migrants per mile per hour crossing a northeast to southwest
line through Gainesville, Florida. This quantity was estimated
by determining the number of migrants crossing the circle in
one hour. Since the point of entrance or exit was inadvertently
missed for some of the butterflies, this estimate was not as
accurate as it might have been. Of those migrants whose track
was determined, 97.8 per cent had track bearings between 100
and 180 degrees. It is apparent from the geometry of the circle
that any migrant crossing the circle and having a track bearing
within this range must cross a northeast to southwest line 66.4
feet long with its midpoint at the center of the circle (assum-
ing that all migrants enter and leave the circle at points mid-
way between two stakes). Since some of the migrants had track
bearings outside this range, the assumption that all migrants
passing through the circle cross this line results in a positive
error in the density estimate. On the other hand, this line
extends 8.2 feet beyond the circle on either side. Therefore,
the assumption that only those migrants passing through the
circle cross the line results in a negative error in the density
estimate. Since both of these errors are probably small, and
since one at least partially cancels the other, a fair estimate
of the migration density can be obtained by multiplying the num-
ber of migrants crossing the circle in one hour by 79.5 (the
number of times 66.4 feet is contained in 1 mile).
A better estimate could have been obtained if the
points of entrance and exit had been noted for every butterfly
crossing the circle. Then it would have been possible in mak-
ing the estimate to consider only those migrants which crossed
the northeast to southwest diameter of the circle, and both
types of error would have been eliminated. Furthermore, in
discussing migration density, it would be desirable to separate
those butterflies which are flying southward from those which
are flying northward. This could be done if the track of every
migrant crossing the circle were known.
Reproductive maturity of the females and sex ratio
Migrating gulf fritillaries were captured with a butter-
fly net to determine the sex ratio among them. Some of the
females were preserved as described previously and later dis-
sected to determine their reproductive maturity and if they had
Description of the Migration
Characteristics of the migratory flight
Migrating gulf fritillaries fly at a height of 3-6
feet over open terrain, and upon encountering an obstacle,
such as a building or a wooded area, they fly up and over it
without changing their direction. In general, the flight is
very persistent, but occasionally they pause briefly to feed
at flowers. The direction of most individuals lies between
110 and 160 degrees (Figure 17) and does not vary with time
of day (Figure 18). While the path of a migrant over the earth
may be influenced by the wind, the migratory direction is not
determined by this factor (Figure 19). Instead, it appears
to be under the control of the insect itself.
The direction of a flying animal relative to the
earth's surface is determined by its motion through the air and
by the motion of the air itself. If the animal maintains a
constant course, fluctuations in the crosswind will cause
fluctuations in its track. To maintain a constant track, it
must alter its course to compensate for these fluctuations or,
in other words, correct for wind drift. The crosswinds ob-
served during this study were too light to determine if migrat-
ing gulf fritillaries make this correction.
The ground speed of a flying animal is a function of
the energy it expends per unit time and of the wind component
along its course. If the animal expends a constant amount of
energy, its ground speed will be less with a head wind than with
I i 4 i
__________- = 10
Figure 17. Tracks of migrants observed during
the fall of 1963 (A) and 1964 (B). The numbers
in the circles represent the total number of
migrants observed in each case.
17 D. 10
Figure 18. Tracks of migrants observed at various times
of day on September 23, 1964. (A) 0800-0900 EST, (B) 1000-
1100 EST, (C) 1200-1300 EST, (D) 1300-1400 EST, (E) .1400-
1500-EST, (F) 1600-1700 EST. The numbers in the circles
represent the total number of migrants observed in each
Figure 19. 'Tracks of migrants and mean wind speed and
direction observed between 1300 and 1400 EST on various
days in the fall of 1964. Wind direction is indicated
in each case by the direction of the arrow. Each full
barb in the tail represents 2 miles per hour of wind
speed. The numbers in the circles represent the total
number of migrants observed in each case.
a tail wind, while its air speed will be the same in both cases.
To maintain a constant ground speed, it must alter its energy
output to allow for the effect of wind, and since it is a
function of the energy expenditure, the air speed will vary.
Within the range of observed winds, it appears that each
gulf fritillary expends a more or less constant amount of energy,
but this amount varies from individual to individual. The
ground speed of 23 individuals flying in calm air ranged from
6.9-14.6 miles per hour with a mean of 10.4 (Figure 20), while
that of 11 individuals flying against a head wind of 1-5 miles
per hour ranged from 6.7-11.4 with a mean of 8.4 (Figure 21A),
and that of 9 individuals flying with a tail wind of 1-4 miles
per hour ranged from 8.8-15.6 with a mean of 12.1 (Figure 21B).
The means are significantly different at the 5 per cent level.
The air speed of the headwind group ranged from 8.6-14.6 with
a mean of 11.2 (Figure 22A), and that of the tailwind group
ranged from 8.8-13.3 with a mean of 11.0 (Figure 22B). The
means are not significantly different at the 20 per cent level.
It follows that the ground speed will decrease as the head wind
increases until the butterfly must increase its energy output,
land, or be carried backwards. Which of these alternatives
actually occurs was not determined. The head winds did not
reach this magnitude during the course of the present study,
and they seldom do at the flight level of the migrants.
Variations in migration density
The migration densities determined between 1300 and
1400 Eastern Standard Time on various dates during the course
per cent frequency
0 s peed(MPM)
0.55 10.55 20.55
Figure 20. Ground speed of migrants
flying in calm air.
per cen t Irqq u n c y.
rd (M Pn)
per cent Irequeny
Figure 21. Ground speed of migrants flying against a
head wind of 1-5 miles per. hour (A) and with a tail wind
of 1-4 miles per hour (B).
petr cent Itrqu n cy
ip..d (M PH)
per cenl I requeny
- peed (M P)
Figure 22. Air speed of migrants flying against
a head wind of 1-5 miles per hour (A)'and with a
tail wind of 1-4 miles per hour (B).
of the 1964 migration are presented in Table 5. The density
varies considerably from day to day, but in general, the migra-
tion is heavier during the first half of the migratory period.
The greatest density observed would be classified as extremely
thin according to the scheme of Williams (1958). Wind speed
and direction within the limits observed do not seem to affect
the number of migrants flying. It appears that cloud cover
does not influence the migration density unless the sky is
overcast, and then the migration ceases. On four occasions,
observations were begun within 1.5 hours after sunrise (Table
6). On three of these days, the day's migration had not yet
started when observations were begun. The sky condition was
either clear or scattered on all of these days. These data
suggest that the time at which the migrants begin flying could
be determined by either temperature or light intensity. On
September 23, 1964, the sky was clear and the wind was light
and variable all day, and observations were made throughout the
day. The migratory activity ceased quite abruptly at 1715
EST (about 1.25 hours before sunset). Based on the observations
of this day, the migration density does not seem to vary in a
regular manner with time of day (Table 7).
Reproductive maturity of the females and sex ratio
Of 43 migrants captured between September 22 and Novem-
ber 7, 1964, 72 per cent were females and 28 per cent were males.
This sex ratio is significantly different from 1:1 at the 0.5
per cent level. Since there is no reason to believe that fe-
males are more easily captured than males, it appears that while
DENSITY OF MIGRATION AND WEATHER OBSERVED
1300-1400 EST (1964)
Date Crossing Density Sky Cover
Circle Speed Direction
20 Sep 9 718 Scattered 00-15 360-060
23 Sep 12 954 Clear 00 -------
25 Sep 7 556 Thin Broken 00-07 030-110
29 Sep 3 238 Broken 00-08 100-170
5 Oct O O Overcast 06-19 250-280
7 Oct 1 80 Scattered 05-12 330-060
9 Oct 10 795 Scattered 00-08 360-090
23 Oct 2 159 Clear 03-07 340-110
25 Oct 1 80 Broken 02-09 050-100
7 Nov 3 238 Scattered 00-08 090-190
18 Nov 2 159 Broken 00-06 330-230
TIME OF BEGINNING OF MIGRATION ON VARIOUS DAYS
Began Observation First Migrant
(EST) Time (EST) Temp. (OF.) Time (EST) Temp. (OF.)
11 Oct 63 0630 0700 53 0930 64-69
16 Oct 63 0632 0800 66 0930 70-73
18 Oct 63 0634 0800 58 0930 65-72
23 Sep 64 0617 0800 67 0815 67-75
MIGRATION DENSITIES OBSERVED AT VARIOUS TIMES ON
SEPTEMBER 23, 1964
Time (EST) Crossing Density
0800-1000 11 874
1000-1100 11 874
1200-1300 19 1,510
1300-1400 12 954
1400-1500 16 1,272
1600-1700 11 874
the sex ratio is essentially 1:1 in the total population of
Aqraulis vanilla, the females outnumber the males among the
migrants. Six migrant females were preserved and dissected.
Of these, five had spermatophores in the bursa. Mature eggs
were present in the oviducts of three, and in some, the
abdomen was partially empty, suggesting that they had already
The ability to maintain a constant compass direction
by referring to the sun and compensating for its apparent move-
ment was first demonstrated in bees by von Frisch (1950) and
in birds by Kramer (1950). Since that time, this ability has
been found in many animals including fish (Hasler, Horrall,
Wisby, and Braemer, 1958), amphibians (Ferguson, 1963), reptiles
(Gould, 1957), and arthropods (Birukow, 1956; Papi, 1955; Pardi
and Papi, 1952), and it has been dealt with in a recent sym-
posium on biological clocks (Long Island Biological Association,
1960) and in a more recent symposium on animal orientation
(Autrum, 1963). An animal which is maintaining a constant
course by means of this time-compensated sun compass changes
its angle with respect to the sun at a rate which is equal in
magnitude and opposite in direction to the angular velocity of
the sun. This mechanism requires a clock furnishing the exact
local time and a knowledge of the sun's angular velocity.
The internal or biological clock which furnishes the
local time can be reset, at least in many cases, by subjecting
the animals to a light-dark cycle which is out of phase with
the natural cycle (Birukow, 1960; Braemer, 1960; Hoffmann, 1960;
Pardi and Grassi, 1955; Schmidt-Koenig, 1960), and the amount by
which the phase of the clock is shifted depends upon the magni-
tude of the phase shift in the light-dark cycle. Animals in
which the phase of the internal clock has been shifted show a
corresponding phase shift in the orientation rhythm. If, for
example, an animal which has been trained under the natural
sun to search for food in a given direction is subjected for
several days to a light-dark cycle in which the light period
begins and ends 6 hours later than in the natural cycle, it
will search for food 90 degrees to the right of the training
direction when tested again under the natural sun. This
phenomenon provides one means of demonstrating a time-compen-
sated sun compass in an orienting animal.
Frequently, animals which are orienting by means of a
time-compensated sun compass will recognize a fixed light
source as the sun and will change their direction during the
course of the day by changing their angle with respect to this
fixed light source (Birukow, 1960; Braemer, 1960; Kramer, 1952).
The rate of change of this angle is such that a constant direc-
tion would be maintained if the artificial sun were moving with
the angular velocity of the natural sun. This phenomenon pro-
vides another means of demonstrating the existence of a time-
compensated sun compass.
Methods and Materials
The field observations suggested that the orientation
mechanism underlying the unidirectional migratory flight of the
gulf fritillary could be a time-compensated sun compass, so
two series of experiments were performed to test this hypothesis.
In one series, orientation tests were made in a room which was
completely dark except for the light provided by a 150 watt
flood light serving as an artificial sun. In the other series,
orientation tests were made on the roof of a three story build-
ing when the sun was clearly visible. The butterflies used in
these tests were captured between September 30 and November 7,
1964, at the site used for observations of the migration. They
were taken with an insect net while in flight or while pausing
to feed at flowers.
The butterflies were tested individually in an octa-
gonal cage with screen sides 24 inches wide and 60.5 inches high,
a screen top, and a plywood floor (Figure 23). The top of the
cage was divided into eight sectors by four diameters. The
butterflies were introduced into the cage by placing them under
an opaque container in the center of the floor and then raising
the container to the top by means of a string extending to the
outside. After the container was raised, they usually remained
on the floor for a short time and then flew to one of the sides.
Periods of rest, when the butterflies sat motionless with the
wings held over the back, alternated with periods of activity
consisting of opening and closing the wings while remaining
stationary or while walking, and of flying about and into the
sides of the cage. The ratio of rest to activity varied con-
siderably from individual to individual and from one time to
another in the same individual. Each butterfly was scored by
recording its position by sector every 10 seconds during periods
Figure 23. Cage used for orientation tests.
of flight for one hour beginning at the time the container
was raised. If an individual was not scored 25 or more times
during the first half hour, it was considered inactive, and
the test was discontinued.
In the artificial sun experiments, the light was directly
opposite the center of one sector. In the outdoor experiments,
the center of one sector was aligned with true north so that each
sector represented 45 degrees centered about one of the points
of an eight point compass. The scores were analyzed using a
modification of the method outlined by Papi and Tongiorgi (1963).
The direction of the mean vector, 9, and its length, r, were
calculated from the distribution of n scores for each individual.
The sectors were numbered clockwise from 0-7 beginning with the
position of the light, or with north. The direction and length
of the mean vector are given by the expressions
Sni sin i .600
tan 6 = and
Sni cos i 3600
i o 8
Sni sin i 3600\ + ni cos i 360
i o 'O
r = is te n r of s i t i s T
where ni is the number of scores in the ith sector. The direc-
tion of the vector is expressed in degrees measured clockwise
from the position of the light, or from north. The length is a
measure of the dispersion of the scores and varies from 1-0
as the dispersion increases. In the present study, individuals
showing values of r less than 0.7 were considered to be dis-
Artificial sun experiments were performed on two
migrants. The butterflies were confined individually in cubic
screen cages 4 inches on a side and kept in a greenhouse under
the natural light-dark cycle until they were tested. Observa-
tions were made for one hour every other hour between 0800 EST
and 1700 EST in one case and between 0900 EST and 1600 EST in
the other. The light was located 6 feet south of the center
of the cage at a height of 6 feet. Observations were made from
behind the light.
Outdoor experiments were conducted between October 18
and November 17, 1964, at Gainesville, Florida. Observations
were made from four stations about 3 feet outside the cage, one
at each of the four major compass directions, and each station
was used for 15 minutes during each test. The butterflies used
in these experiments were treated in three different ways. Be-
tween tests, the first group was kept under a light-dark cycle
which was the same as the natural cycle at the time of year
the tests were made. For at least 3 days prior to testing and
between tests, the second group was subjected to a light-dark
cycle in which the light period began and ended 6 hours later
than in the natural cycle. The third group was kept under the
in-phase cycle until one or more tests had been made, then sub-
jected to the out-of-phase cycle for at least 3 days and tested
again. When they were not being tested, the butterflies were
confined in cubic screen cages 4 inches on a side. They were
fed once a day as described previously.
The desired light-dark cycles were maintained in con-
trolled photoperiod cabinets in which the temperature was
held constant. One of these cabinets is illustrated in Figure
24. It is constructed of inch plywood on a frame of x 3/4
inch wooden strips and is lined, except for the top, with
inch cane-fiber insulation board (Celotex). The dimensions are
26 x 22 x 34 inches on the outside and 24 x 20 x 33-3/4
inches on the inside. Two 24 inch base fluorescent fixtures
are fastened inside the top and connected to a time switch
(Sears, Roebuck and Co., Model number 5870) on the outside.
Lighting is provided by two 20 watt daylight fluorescent bulbs,
and the interior of the cabinet is painted white to obtain
maximum brightness. The cabinet is partitioned by two panes
of double strength glass supported at distances of 111 and
17 inches from the top by frames of x 3/4 inch wooden
strips. Thus the cabinet is divided into a chamber containing
the lights, a dead air space between the panes of glass, and a
chamber for housing the butterflies. The purpose of the dead
air space is to keep to a minimum temperature fluctuations
in the housing chamber caused by the light-dark cycle. Access
to the housing chamber is provided by a 12 inch high door
across the entire front of the cabinet. A strip of inch
plywood 1 inches wide along the top of the door covers the
crack between the door and the front of the cabinet to exclude
light. Both cabinets were placed in a constant temperature
room, and air was circulated through the housing chambers by
means of a blower. Air entered through a 2 inch diameter hose
Figure 24. Cutaway view of controlled photoperiod cabinet
used in clock resetting experiments. (A) time switch, (B)
light chamber, (C) dead air space, (D) air inlet, (E) hous-
in the back of each cabinet and left through a 2 inch diameter
hose in the side. Both hoses were curved to exclude light.
The temperature in the cabinets varied from 24-250 C. This
fluctuation was the same as that elsewhere in the constant tem-
perature room and was not associated with the light-dark cycle.
The artificial sun experiments failed to demonstrate
the existence of a time-compensated sun compass. The test
butterflies directed their flight activity toward the light
throughout the day. The overall illumination in the room was
rather low, and it is possible that under these conditions,
the compass orientation is replaced by a simple positive photo-
taxis. Better results might have been obtained by providing
diffuse light in addition to the light provided by the artifi-
The results of the outdoor tests are summarized in
Figures 25-39. In these figures, north is at the top, and the
small circle represents the position of the sun at.the mid-
point of the test period. Each small dot represents a single
score, the solid arrow represents the direction and length of
the mean vector, and the dashed arrow represents the subjective
direction; i.e., the direction the butterfly would be flying if
it were maintaining the observed angle to the sun 6 hours before
the time of testing. In many cases, the butterfly being tested
was inactive, and in nine cases it was active but disoriented.
This was probably the result of the repeated handling to which
the insects were subjected.
Nine butterflies were active and oriented in one or more
tests. When an individual was tested before being subjected to
the phase-shifted cycle, it usually flew in the migratory direc-
tion whether it was tested in the morning or afternoon (Figures
27, 28, 29, 30, 33, 35). In one case, however, the orientation
was reversed (Figures 25-26). This reversal can be accounted
for if the model proposed by Mittelstaedt (1960) for the control
system of time-compensated sun orientation is accepted.
Some individuals which had experienced the phase shift
flew approximately 90 degrees to the right of the migratory
direction (Figures 36, 38, 39), while one flew approximately
90 degrees to the right of the reversed direction (Figures
31-32). If it is assumed that under the experimental condi-
tions, the orientation is sometimes turned 180 degrees, as
appears to be the case, these results provide good evidence
for time-compensated sun orientation.
In the case of Individuals E and G (Figures 34, 37),
the mean vector fell in the northwest quadrant after subjection
to the phase shift. In both cases, however, there appeared to
be a conflict between the reversed migratory direction and the
direction imposed by the phase shift. Individual E flew to the
northeast for the first 45 minutes of the test period and then
to the northwest. Individual G flew to the northeast for the
first 10 minutes and then to the northwest. No explanation for
this directional conflict is immediately apparent.
The results of these experiments suggest that the mi-
gratory direction might be maintained by means of a time-
compensated sun compass. While they are too inconsistent to
be conclusive, they are suggestive enough to warrant further
investigation along these lines.
Figure 25. Orientation of Individual A when tested
between 0930 and 1030 EST, November 6, 1964, without
resetting the internal clock. -e = 295. degrees,
r = 0.844.
Figure 26. Orientation of Individual A when tested
between 1440 and 1540 EST, November 6, 1964, without
resetting the internal clock. 9 = 296 degrees,
r = 0.902.
Figure 27. Orientation of Individual B when tested
between 0950 and 1050 EST, October 30, 1964, without
resetting the internal clock. 8 = 163 degrees, r = 0.919.
Figure 28. Orientation of Individual B when tested
between 1410 and 1510 EST, November 5, 1964, without
resetting the internal clock. e = 167 degrees,
r = 0.716.
Figure 29. Orientation of Individual C when tested
between 1240 and 1340 EST, November 5, 1964, without
resetting the internal clock. e = 156 degrees,
r = 0.825.
S \ .........
Figure 30. Orientation of Individual D when tested
between 1510 and 1610 EST, November 5, 1964, without
resetting the internal clock. 9 = 111 degrees,
r = 0.746.
Figure 31. Orientation of Individual D when tested
between 1350 and 1450 EST, November 10, 1964, after
setting the internal clock back 6 hours (5 days in
out-of-phase cycle). 8 = 67 degrees, r = 0.853.
Figure 32. Orientation of Individual D-when tested
between 1310 and 1410 EST, November 11, 1964, after
setting the internal clock back 6 hours (6 days in
out-of-phase cycle). G = 61 degrees, r = 0.801.
Figure 33. Orientation of Individual E when tested
between 1030 and 1130 EST, 'November 9, 1964, without
resetting the internal clock. Q = 151 degrees, r = 0.904.
Figure 34. Orientation of Individual E when tested
between 1240 and 1340 EST, November 17, 1964, after
setting the internal clock back 6 hours (8 days in
out-of-phase cycle). 6 = 352 degrees, r = 0.700.
Figure 35. Orientation of Individual F when tested
between 0930 and 1030 EST, November 9,. 1964, without
resetting the internal clock. 8 = 138 degrees,
r = 0.730.
Figure 36. Orientation of Individual Fwhen tested
between 1430 and 1530 EST, November 16, 1964, after
setting the internal clock back 6 hours (7 days in
out-of-phase cycle). 0 = 218 degrees, r = 0.741.
Figure 37. Orientation of Individual G when tested
between 1430 and 1530 EST, October 18, 1964, after
setting the internal clock back 6 hours'(5 days in out-
of-phase cycle). 9 = 331 degrees, r = 0.743.
Figure 38. Orientation of Individual H when tested
between 1530 and 1630 EST, October 20,1. 964, after
setting the internal clock back 6 hours (3 days in
out-of-phase cycle). 8 = 258 degrees, r = 0.903.
Figure 39. Orientation of Individual I when tested
between 1430 and 1530 EST, October 23, 1964, after
setting the internal clock back 6 hours (4 days in
out-of-phase cycle). 6 = 202 degrees, r = 0.807.
DISCUSSION AND CONCLUSIONS
Southwood (1960) presented evidence supporting the
hypothesis that in the course of evolution a low level of
migratory activity has been associated with the colonization
of permanent habitats and a high level closely correlated
with the adoption of temporary ones. The prime evolutionary
advantage of migratory movement is the colonization of new
habitats and of previously vacated ones. The observations
made on the gulf fritillary during the course of this study
lend additional support to his hypothesis.
The habitat of Agraulis vanilla nigrior is temporary
in two respects. First, its relatively small and widely
scattered breeding areas are frequently destroyed by the feeding
of the larvae or by the depredations of other insects. Second,
winter temperatures are too low for its survival over the
greater portion of its range. The gulf fritillary is essen-
tially a tropical insect and probably lacks a cold-hardy
stage which would enable it to overwinter in these colder areas.
At the present time, no precise information is available con-
cerning the cold-hardiness of the various stages or what effect
various conditioning factors might have upon this cold-hardiness.
Turner (1963) claimed that this insect overwintered for three
consecutive winters in west-central Missouri. He based this
cla:. on the fact that the species was quite abundant on
P. --ifora in this vicinity during three successive summers.
This idea was challenged by Howe (1965) who pointed out that
fr-rles which have traveled from much farther south reach the
latitude of Missouri and Kansas during June and July. He
also reported that 37 chrysalides kept outdoors in a screen
cage during late October in Kansas were all destroyed during
a single freezing night.
The observations made during the course of the research
reported here suggest that the species is not able to pass the
winter even in the vicinity of Gainesville, Florida. If the
insect successfully overwintered in this locality, it would
probably appear in substantial numbers with the return of
favorable conditions. Conditions appear to be favorable for
its development by the first of April, but at this time of
year it is very scarce and remains so until midsummer. This
suggests that the breeding areas in this part of the state are
repopulated by females arriving from farther south.
The coastal areas of southern Florida appear to be
suitable for the development of the gulf fritillary throughout
the year. These localities are less subject to frost than the
inland areas due to the moderating influence of water. During
the 20 years between 1937 and 1957, less than 25 hours of tem-
peratures less than 320 F. were recorded for an area 0-15 miles
wide running along the coast from Palm Beach County to Lee
County (Federal-State Frost Warning Service, 1958). Further-
more, field observations showed that both the larvae and the
food plants can survive brief exposures to freezing tempera-
tures. It therefore appears very likely that the gulf frit-
illary continues to breed throughout the winter in these
localities and possibly in other warm areas along the Gulf
At least some individuals leave these breeding sites
in the spring and move northward. Apparently the insect is
able to maintain only a low population density during the
winter, as the northward migration is much sparser than the
southward one and is too thin for making measurements of direc-
tion and density by means of the technique described earlier.
The migrating females apparently lay eggs enroute as patches
of the food plant are encountered but do not deposit more than
a few eggs in any one locality before continuing their migra-
tory flight. This movement could account for the first
appearance of eggs and larvae in the vicinity of Gainesville,
Florida, about the first of April and for the appearance of
the insect as far north as Kansas and Missouri in June or July.
Since the breeding areas are widely scattered, only a few of
the females passing through a given latitude will encounter
suitable oviposition sites there. Furthermore, since the
butterflies are widely separated in time and space, we would
not expect all the available breeding sites in a given area
to be occupied at the same time. The observations made in the
vicinity of Gainesville show that this is the case.
It appears that this migration continues until late
summer with the individuals produced in a given locality
leaving that locality and moving northward as did their female
(a,in ,erhapb male) parents. This seems to be the only possible
ex:planation for the low population density which prevailed at
Ga:n-.sville through most of the summer in spite of abundant
foo( a ._ favorable weather. No evidence was found which sug-
co.sts that predators, parasites, or disease played a significant
role in preventing a population buildup.
At least in some years, the insect reaches the latitude
of New York. How far each individual travels before it ceases
its migratory flight and leaves the northward expansion of the
range to its offspring is a question which can be answered only
by a tagging program such as was carried out on the monarch
Late in the summer, some of the butterflies begin moving
southward. This reversal of migratory direction perhaps appears
first in the northernmost segment of the population and spreads
southward as the season advances. This movement results in the
abandonment of the greater portion of the range. The observa-
tions made at Gainesville suggest that not all individuals take
part in this migration, but it is unlikely that the descendents
of those remaining behind survive the winter. The fact that at
least some (and probably many) of the migrant females have mated,
have mature eggs in their ovaries, and have partially empty ab-
domens, strongly suggests that they oviposit enroute. The sud-
den increase in the size of the egg and larval populations in the
vicinity of Gainesville late in th-. summer could be accounted for
by the passage through the area of large numbers of ovipositing
females produced in the large northern portion of the range.
The estimates of migration density indicate that the southward
migration was already at or beyond its peak on September 20,
1964, when the first observation was made. This could also
account for the sudden occupancy in August of Area 4 by large
numbers of eggs and larvae after it had remained empty during
most of the summer. The fact that few adults were ever present
in the breeding areas at any one time lends further support to
the hypothesis that the eggs were deposited by migrant females.
The decline of the populations at most of the breeding sites
could be explained by the destruction of food plants, but it
cannot be explained in this manner for Area 3. The decrease in
all the populations (except in Area 4) is probably related in
part to a decrease in the migration density.
The individuals produced from many of the eggs deposited
enroute probably reach maturity and move southward themselves
before the occurrence of freezing temperatures. Again, the
distance traveled by each individual remains to be determined
by a tagging program.
The flight direction of the fall migrants passing through
Gainesville, Florida, is predominately southeastward. This di-
rection may be affected somewhat by the wind but is not deter-
mined by the wind. From the observations presented here, it
must be concluded that the direction of the displacement of the
population is largely under the control of the insects them-
selves. If we suppose, as is probably the case, that the flight
continues in this direction, the migrants will eventually reach
t'"e coastal areas of southern Florida. The fact that A. v.
r i r' does not intorcrade with the Bahaman and Antillian
zu: s-.cies A. v. insularis strongly suggests that the migration
<' ,. not extend beyond the Florida Keys.
Tle, pattern of migratory activity described here is
hi -hLy adaptive. The northward migration in the spring results
in L:i colonization of new habitats and prevents extensive
population buildups in the rather small breeding areas which
would result in food shortage and an increase in the rate of
parasitism and disease. The northward direction has a dis-
tinct evolutionary advantage in that the insects are more likely
to find unoccupied oviposition sites in that direction. The
descendents of those individuals which migrate southward in
the fall survive the winter and repopulate the northern breeding
areas the following summer. The descendents of those which do
not migrate perish.
Several important problems concerning the migratory
behavior of the gulf fritillary remain to be solved. The
orientation experiments described here indicate that a time-
compensated sun compass may be the mechanism underlying the
oriented flight, but the results were too inconsistent to prove
this definitely. The best approach to this problem is probably
the performance of artificial sun experiments in which diffuse
ligc:tin, is provided in addition to the light provided by the
artificial sun. Experimental work is needed to determine what
initiates the spring and fall migrations and how the migratory
direction is determined. The most likely initiating factor
appears to be either temperature or photoperiod or a combination
o- the two, but this remains to be proved. Both migratory di-
rections may be genetically determined, one being manifested
un:dor one set of conditions and the other under another set
of conditions, or the direction may be determined in another
manner as it is in the great southern white (Nielsen, 1961).
Answers to all of these questions must await future research.
The larvae of Agraulis vanilla nigrior Michener feed
upon Passiflora incarnata L. and probably other plants of the
ccnus, and the adults oviposit upon and in the vicinity of
thise plants. The insect develops very rapidly, reaching the
adult stage in approximately 28 days at 23-24o C. and in
approximately 22 days at 28.5-29.50 C. Most adults emerge
from the chrysalides during the morning. The females usually
reach reproductive maturity 12-18 hours after emergence, and
the average adult life span is about 18 days under laboratory
The adults can be found in almost any open situation,
but because of the patchy nature of the food plant distribution,
the insect can breed only in small widely scattered areas. The
food plants in these small areas are frequently destroyed by
the larvae or by the depredations of other insects. Winter
temperatures are too low for its survival over the greater
portion of its range, but it is probably able to breed through-
out the winter in the coastal areas of southern Florida and in
other warm areas along the Gulf Coast.
The insect has evolved a migratory habit which has
adapted it for the utilization of habitats which are frequently
rendered unsuitable for its survival by the depletion of the
food suply or by low temperatures. Observations of the fall