a RE ‘= | | ISSN 1345-5834
Volume 20,Number1l June 2001

| a : BX ax an PA 40 2A ae we
| ; | 4 5 6 7 8

| 2 3 3

@e@ a6 £2e@ 26 6 Bt Ga © 646 Ge
9-19

44 00 68 4a Ca 4H oO ae 46 or MH Bas HH

20-33
os6 einai
Fe gs yy
BEV tS —
Pl i i: we
9 e >’ %
tA %
y
% oie j
\
yy
> i
We, &. i

FORMERLY THE JAPANESE JOURNAL OF HERPETOLOGY

Published by
THE HERPETOLOGICAL SOCIETY OF JAPAN
KYOTO

THE HERPETOLOGICAL SOCIETY OF JAPAN

Executive Council 2001

President: Masafumi MATSUI, Graduate School of Human and Environmental Studies, Kyoto
University, Yoshida Nihonmatsu-cho, Sakyo-ku, Kyoto 606-8501 Japan (fumi@zoo.zool.
kyoto-u.ac.jp)

Secretary: Tsutomu HIKIDA, Department of Zoology, Graduate School of Science, Kyoto
University, Kitashirakawa-Oiwake-cho, Sakyo-ku, Kyoto 606-8502 Japan (tom@zoo.zool.
kyoto-u.ac.jp)

Treasurer: Akira MORI, Department of Zoology, Graduate School of Science, Kyoto Univer-
sity, Kitashirakawa-Oiwake-cho, Sakyo-ku, Kyoto 606-8502 Japan (gappa@zoo.zool.
kyoto-u.ac.jp) |

Managing Editor: Hidetoshi OTA, Tropical Biosphere Research Center, University of the
Ryukyus, Nishihara, Okinawa, 903-0213 Japan (ota@sci.u-ryukyu.ac.jp)

Officers: Masami HASEGAWA (PX1M-HSGW@asahi-net.or.jp), Tamotsu KUSANO
(tamo@comp.metro-u.ac.jp), Showichi SENGOKU, Michihisa TORIBA (snake-a@
sunfield.:ne.jp), Kinji FUKUYAMA (fukuyama@hc.cc.keio.ac.jp) |

Main Office: Department of Zoology, Graduate School of Science, Kyoto University
Kitashirakawa-Oiwake-cho, Sakyo-ku, Kyoto 606-8502 Japan

Honorary Member: Hajime FUKADA

Current Herpetology.—The Current Herpetology publishes original research articles on
amphibians and reptiles. It is the official journal of the Herpetological Society of Japan and
is a continuation of the Acta Herpetologica Japonica (1964-1971) and the Japanese Journal
of Herpetology (1972-1999).

Herpetological Society of Japan.—The Herpetological Society of Japan was established in ~

1962. The society now publishes the Current Herpetology and Bulletin of the Herpetologi-
cal Society of Japan. Current Herpetology is printed solely in English and is international

in both scope of topics and range of contributors. Bulletin of the Herpetological Society of

Japan is printed exclusively in Japanese and publishes more regional topics and announce-
ments about the Herpetological Society of Japan.

Membership and Subscription—Membership (including subscription) is open to all per-
sons interested. Current annual fees are 5,000 yen for the regular members who receive
both Current Herpetology and Bulletin of the Herpetological Society of Japan. For those
who wish Current Herpetology only, there is a reduction to 3,000 yen. To subscribe Current
Herpetology and/or Bulletin of the Herpetological Society of Japan, please contact the
Treasurer. We accept VISA and Master Card. If you need to use bank check, please add
2,000 Japanese yen for handling charge.

All other correspondence regarding the society, including the availability and cost of soci-
ety publications should be directed to the office of the Secretary.

Cover Illustration: Karyotype of female Pelodiscus sinensis from Amamioshima. A photo-
graph taken by Hiroyuki Sato.

Current Herpetology 20(1): 1—10., June 2001
© 2001 by The Herpetological Society of Japan

Dispersal of Brown Frogs Rana japonica and R. ornativentris

in the Forests of the Tama Hills

SATOSHI OSAWA.’ AND TAKEHIKO KATSUNO

Laboratory of Landscape Science and Planning, College of Bioresource Sciences, Nihon

University, Kameino 1866, Fujisawa, Kanagawa 252-8510, JAPAN

Abstract: For the conservation of forest frogs, it is important to conserve not
only water sites for breeding but also forests for living in the non-breeding sea-
son. Therefore, the two species of brown frogs (Rana japonica and R. ornativen-
tris) common to Japan were surveyed for their dispersal capability and the
range of activity in the non-breeding season. The survey was conducted 16
times in forests on undeveloped land (about 28 ha) of the Tama Hills by the cap-
ture-and-mark method. The migrating distances were calculated from the sites
of capture and spawning. As a result, 90% of the individuals were within the
range of 200 to 270 m (R. japonica) and 330 to 390 m (R. ornativentris). The
maximum distance was about 500 m and both species were found capable of
migrating at least 500m. The migrating distances of recaptured individuals
indicated that both species have specific summer habitat. Yearlings especially
of both species grow rapidly between summer and fall in the fprests. R. orna-
tiventris was only rarely captured at the edge of the forest. The study demon-
strated that the forest is the main habitat and main dispersion route for R.
ornativentris in the non-breeding season. We discuss the conservation of both
species from the viewpoint of landscape ecology.

Key words: Rana japonica; Rana ornativentris; Mark-Recapture; Dispersal dis-
tance; Site fidelity; Landscape ecology

INTRODUCTION

According to recent reports, not only
locally rare species but also species that used
to be common in paddy fields are decreasing
in urban and other areas (e.g., RDRGK, 1995;
Hasegawa, 1996; Hasegawa et al., 2000). In
many cases, the decrease is attributable to the
loss of water sites for breeding and environ-
mental changes in wetlands. To conserve

( Corresponding author. Tel/Fax: +81—466—84-
3623.

E-mail address: oosawa@brs.nihon-u.ac.jp (S.
Osawa)

amphibians in various places, priority should be
given to the preservation and maintenance of
wetland breeding habitats. For forest frogs, it
is also necessary to conserve forests as habi-
tats in the non-breeding season. What is most
important is to maintain the meta-population by
the regional interconnection of habitats and
populations (Green, 1997).

In biota conservation, the patch-corridor-
matrix theory is generally used (Forman, 1995).
This concept is the key to the systematic estab-
lishment and conservation of a network of vege-
tation and ecosystems. However, the ecological
rationale of this concept is not always clear
in the forest setting and placement process in

Japan, although the distance between habitats
and the conditions of ecological corridors differ
depending on the guide species (Hioki and
Ide, 1997). Therefore, we need to accumulate
ecological knowledge about the migrating style
of the target frog species.

The past studies on the migrating habits of
amphibians in Japan, especially dispersion,
include demonstration surveys on Hynobius
tokyoensis (Hynobiidae) by Kusano (1984),
Bufo japonicus formosus (Bufonidae) by Yano
(1978), Okuno (1986), and Kusano et al. (1995),
and Rhacophorus arboreus (Rhacophoridae) by
Kusano (1998). However, there are almost no
studies on the dispersion of brown frogs, so we
studied two species of brown frogs (Rana
Japonica and R. ornativentris) which are widely
distributed in Japan.

These ranid species are terrestrial species of
40 to 70 mm (R. japonica) and 40 to 80 mm
(R. ornativentris) in snout-vent length (SVL).
Both species are widely found in the plains,
hills, and mountainous areas of Honshu,
Shikoku, and Kyushu (Maeda and Matsui,
1989). Rana japonica is not very numerous in
mountainous areas, but R. ornativentris 1s.
These species lay eggs in sunny, shallow waters
in early spring and live on the forest floor or in
grasslands surrounding breeding sites in the
non-breeding season. For some time after
metamorphosis and emergence, metamorphosed
juveniles can be seen on the verges of paddy
fields and in the surrounding grasslands. We
still know very little about their lives after dis-
persion (e.g., Hasegawa, 1995). This paper
reports the results of our survey on these two
species in isolated forests of the Tama Hills,
and centers on their dispersal capability and
range of activities in the non-breeding season.

STUDY SITE AND METHODS

The area of survey was Tsuzuki in Yoko-
hama City (35°32'N, 139°30'E) located at the
center of the Tama Hills. Despite develop-
ment of the surrounding area, valleys and for-
ests still remain (area: about 28 ha). This hilly
area is about 60 to 90 m in altitude and cov-

Current Herpetol. 20(1) 2001

ered by secondary forests (Quercus acutissima
and Q. serrata dominating), white oak forest
(Q. myrsinaefolia dominating), and cedar-
cypress plantations (Cryptomeria japonica and
Chamaecyparis obtusa dominating). The under-
growth is mainly Pleioblastus chino, Hedera
rhombea, Ophiopogon planiscapus, Oplis-
menus undulatifolius, and many other species
but the vegetation proportions and dominant

- species are not uniform.

Beside the forest is a valley (about-480 m
long and up to 35 m wide) that was once paddy
fields but is now a swamp. This valley is the
only breeding site surveyed in this area. The
valley bottom is mostly covered with common
reed (Phragmites communis) being left unculti-
vated for about 12 years. The dry part of the
lowland is covered with goldenrod (Solidago
altissima) and Miscanthus sacchariflorus. The
permanent water bodies in this lowland are two
former paddy fields (about 30 mx30 m, about
60 mx20 m), one pond (about 20 mx5 m), and
one swampy area with a colony of water fennel
(Oenanthe javanica). In the early spring of
1999 when this survey was conducted, about
280 egg masses of R. japonica and 180 egg
masses of R. ornativentris were found in the
four permanent water bodies.

We surveyed the forest by randomized walk-
ing (Crump and Scott, 1994) to look for brown
frogs, and by capture-and-mark method. The
surveyors walked through the forest at random
at the speed of about 2 km/h while stirring the
undergrowth with scoop nets to drive frogs out.
They immediately caught brown frogs jumping
out of the undergrowth with the nets, recorded
the frog data, and released the frogs after mark-
ing. Since the spawning sites are limited (only
four permanent water bodies) in this survey
area, the minimum migrating distances can be
calculated by linking the frog captured points
and spawning sites with straight lines. The
captured frogs were marked by cutting their
digits. No more than four digits were cut from
the four limbs of each frog. The survey was
conducted a total of 16 times in 1999: 14 times
at about one-week intervals in July to Septem-
ber 1999 and once each at the end of October

OSAWA & KATSUNO—DISPERSAL OF FROGS 3

and November. The forest was walked through
in the daytime. The average duration of the
walks was 5.5 hours/day until September and at
the end of October and 2.5 hours/day at the end
of November.

RESULTS

Although the survey was conducted 16 times
in total, no brown frogs were captured at the
end of November. From July to September
and at the end of October, 240 yearlings of R.
japonica and 154 yearlings of R. ornativen-
tris were captured. In addition, 52 adults of
R. japonica and 32 adults of R. ornativentris
were captured. In this survey, we use "adults"
to mean large individuals other than yearlings,
so they may perhaps have included juveniles
two years old. The data analysis does not
include individuals that jumped out of the under-
growth but could not be captured. The percent-
ages of individuals with some limbs or fingers
already missing at the first capture were about
4% for yearlings and adults of R. japonica and
about 5% for yearlings of R. ornativentris (0%
for adults). Therefore, the misreading of indi-
vidual markings at recapture seems low.

The growth of yearlings
First, to understand the growth of year-
lings, Fig. 1 shows the SVL data of individu-

als obtained by the survey. As shown in the
Figure, both species showed almost linear
increases from the beginning of July until the
end of October. However, the inclinations of
both regression lines show slow curves on and
after September. In this period (July to Octo-
ber), both increases are statistically significant
(Peason's correlation coefficient test, p<0.001).
After 9 September, some yearling individuals
of both species had nuptial pads on their
thumbs, a secondary male sexual character.

In summer especially, the SVL increased at
an almost constant rate as the season pro-
gressed. We calculated the inclinations of
the regression as representing straight lines.
The average daily growths (SVL) of both
species from 2 July to 30 September were
Y=0.0278X (r=0.908, N=233) for R. japon-
ica and Y=0.0302X (r=0.896, N=152) for R.
ornativentris. In the area of survey, the aver-
age daily growth rate of yearlings was about
1.03 mm/day for both species until the end of
September after emergence.

Recapture

The recaptured individuals were 27 year-
lings (recapture rate: 11.3%) and four adults
(recapture rate: 7.7%) of R. japonica and 11
yearlings (recapture rate: 7.1%) and three
adults (recapture rate: 9.4%) of R. ornative
tris (Table 1 and Fig. 2). At the first capture

SVL(mm)

8 Jul 22 Jul 5 Aug

19 Aug 2 Sep

© Rana japonica
@ Aana ornativentris

16 Sep 30 Sep 14 Oct 28 Oct

Fic. 1. The growth lines of yearlings. Regression lines were estimated with middle June (15 June) as the
starting point. The gray line indicates R. japonica: Y1*10+X?+0.0451X+1.460 (r=0.924, N=240). The
dark line indicates R. ornativentris: Y=-9x10-°X?+0.042X+1.595 (r=0.901, N=154).

in early and middle July (2, 10, and 18 July),
dispersal seems to be in progress for many
individuals, especially yearlings of R. japon-
ica (124.7492.3 m, mean+SD, N=12). In late
July and later, yearlings of R. japonica were
recaptured often (27.6437.0m, N=18) at
points not very far from the first point of cap-
ture. In contrast, yearlings of R. ornativen-
tris were recaptured at slightly distant points
(41.7+43.5 m, N=9). Not many adults of both
species were captured, but they were recap-
tured at almost identical points (R. japonica:
3.0+6.0 m, N=4; R. ornativentris: 0 m, N=3).
The number of days from capture to recap-
ture was 6 to a maximum of 112 days. The
estimated migrating distance per day was mostly
small (0—2.11 m/day), without counting year-
lings of R. japonica captured first in early and
middle July (6.67+8.8 m/day, N=12).

Despite the comparatively thorough walk-
through in this period, the recapture rate was
as low as 10%. In addition, only a total of four
individuals were recaptured twice in both spe-
cies. This indicates that both species are well
hidden under the forest floor of the woodland.

Distances from spawning sites

We plotted capture points on a map of the
survey area (Figs. 3, 4). The linear distance
of each capture point was measured from the
nearest spawning sites on maps of 1:2,500 to
obtain the migrating distance (indicating mini-

Current Herpetol. 20(1) 2001

mum distance). For yearlings, the distance was
surveyed in late July and later when the number

OX sr “°° stream
zs fo) 2 ‘a 2
° SN S206 BX.

eee , oe y
101 102 104 3
103 218 > A eis <o%

@
O 401
i ule <-40 61m
0223 oe 302
S 217_ S405 #3027 2D
0225 om
e (29°) 710m
6’ oO if =)
209 / 80n — A
Ss
cS
EZ
x 70m
a Od
Oe el enuses

stream—

70m 80m

FIG. 2. Migrations of frogs with capture and
recapture points. Open circles indicate R. japonica
and solid circles indicate R. ornativentris. Serial
numbers of 100 and 300 indicate adults, 200 and 400
indicate yearlings, and an asterisk (*) indicates indi-
viduals recaptured twice. The area inside the bold
lines was covered by forest with a closed canopy.
The area outside the bold lines was not covered by
forest. Open squares indicate houses and buildings.

Table 1. The range of distance(m) between the first capture and recapture.

Rana japonica

Rana ornativentris

Yearlings’ Yearlings Adults Yearlings" Yearlings Adults
(N=12) (N=18) (N=4) (N=2) (N=9) (N=3)
Distance
Mean 124.7 276 3.0 49.0 41.7 0.0
SD 225 37.0 6.0 — 43.5 —s
Range 0-245 0-123 0-12 30-68 0-120 0
Distance/day
Mean 6.67 ley 0.07 1.89 2A 0
SD 8.8 2 lull 0.14 — 2.63 =
Range 0-30.6 0-6.9 0-0.3 1.2-2.6 0-7-7 0

“The first capture in early and middle July. *"The first capture in late July and later.

OSAWA & KATSUNO—DISPERSAL OF FROGS 5

CM%
/61m
=
(eo)

ro)
°o (2 a
iF

FIG. 3. The dispersal of R. japonica in the forest.
Circles indicate the points where the frogs were cap-
tured. Solid circles indicate adult frogs and open cir-
_ cles indicate yearling frogs. Other symbols are the
same as in Fig. 2.

of individuals in the process of dispersal was
comparatively small. When an individual was
recaptured at more than one point, the greater
value was adopted. We show the migrating dis-
tances by the frequency distribution of 25m
individuals (Figs. 5, 6). A small number of ind-
viduals of both species were found about 500 m
away from the spawning sites but most of the
individuals were found within 250 to 300 m.

Table 2 shows the range of migrating dis-
tances. Rana japonica has an average migrat-
ing distance of 145 m (yearlings, N=158) and
114 m (adults, N=48). Rana ornativentris has
an average of 186m (yearlings, N=122) and
204 m (adults, N=29). Thus, 90% of the indi-
viduals found at this time were dispersed
within the range from 200 to 270 m (R. japon-
ica) and from 330 to 390 m (R. ornativentris).
75% of these were in the range from 150 to
200 m (R. japonica) and about 270 m (R. orna-
tiventris).

Site
100 150m

70m 80m / Lo
FIG. 4. The dispersal of R. ornativentris in the for-
est. Symbols are the same as in Fig. 3.

Distribution at the edge of woods

Lastly, we compare the number of capture
points at the edge of the forest with that in the
forest. The edge of the forest here refers to a
zone of about 4 to 5 m from where the grass in
the valley and the forest on the slope make
contact. The zone indicates where a colony of
grassy sun-loving plants invades the forest.
Yearlings found in early and middle July were
excluded but recaptured individuals were calcu-
lated as separate samples. Only a few individ-
uals of R. ornativentris were found at the edge
of the forest (yearlings: 0.02%, N=131; adults:
0.07%, N=32), but individuals of R. japonica
were found at a relatively high rate (yearlings:
19.2%, N=177; adults: 53.8%, N=52). More
than half of the adult capture points were at
the edge of the foreest. The x? test was per-
formed between the yearlings and adults of
each species. This test revealed significant
differences between the adults (y7=17.53,
df=1, p<0.001) and in yearlings (y7=21.12,
df=1, p<0.001) of both species.

Current Herpetol. 20(1) 2001

N of individuals

0 -100 -200 -300 -400 -500 -600
Distance (m)

FIG. 5. Distance of yearlings from breeding sites.

LC Rana Japonica

Rana ornativentris

N of individuals

0 -100 -200 -300 -400 -500 -600
Distance (m)

FIG. 6. Distance of adults from breeding sites.

OSAWA & KATSUNO—DISPERSAL OF FROGS

TABLE 2. The range of distance (m) from breeding sites.

Rana japonica

Yearlings (N=158) Adults (N=48)

Rana ornativentris

Yearlings (N=123)

Adults (N=29)

Mean 144.9 hI3:5 186.7 204.4

SD 104.6 89.7 les: 124.5

Max 459 459 559 443

Median 118 89 167 2

Three quarters 203 145 264 D3

90% value 274 200 330 392
DISCUSSION for breeding individuals in Kitakyushu, the

Life in the non-breeding season

We walked through the forest at random
while driving frogs out of the undergrowth for
capturing. However, the recapture rate was as
low as 10% or less. This indicates that the two
species have the habit of hiding under the
forest floor. Even when the same area was
checked at different times on the same day,
some individuals were captured by the second
trial due to contingency in this kind of survey.
The survey method needs to be improved to
raise the capture rate.

In 15 of the 16 censuses of the survey, exclud-
ing that in November, a total of 477 individuals
were captured, including 45 recaptured ones.
Especially recapturing the yearlings, we were
able to learn their growth curves for the period
of summer to fall. Tojo (1976) reported the
growth curve of R. japonica in paddy fields
area in Niigata, and the increase in SVL was
almost linear from June until late October. Our
study showed similar results. The average
daily growth (SVL) was about 1.03 mm/day for
both species in summer. The SVL rate of
increase during these four months was 2.29
times for R. japonica is (2 July: 22.7+2.1 mm,
N=9; 30 October: 51.9+3.3 mm, N=7), and
about 2.22 times for R. ornativentris (2 July:
24.5+2.9 mm, N=4; 30 October: 54.5 mm,
N=2). Thus yearlings of both species grow
quickly in forests as well as in paddy fields.
Kuramoto and Ishikawa (2000) reported that

SVL of R. japonica was 46.4 mm (males) and
51.0 mm (females), and the SVL of R. orna-
tiventris was 54.6 mm (males) and 72.9 mm
(females). In the Tama Hills (Fig. 1), yearlings
over 50.0 mm were often observed after the
middle of September (R. ornativentris) or late
October (R. japonica). Some yearlings of both
species have nuptial pads after September, so
individuals growing rapidly will probably join
in the next breeding season.

The recapture points were distant from the
capture points especially for the yearlings of R.
japonica captured in early and middle July
(Table 1). This indicates that many individuals
are still in the process of dispersal at that time
of the year. In late July and later, many individ-
uals were recaptured within 15—20 m, indicat-
ing that they have definite summer habitats after
dispersion. In North America, it is known that
some ranid species have a summer home range,
e.g., R. clamitans (Duellman and Tueb, 1986)
and R. sylvatica (Bellis, 1965). The migrating
distance per day was mostly small without R.
japonica of the first capture in early and middle
July, so these results suggest that our two ranid
species have a summer home range. In an
additional survey in July 2000 (Osawa and
Katsuno, unpublished data), we recaptured one
female R. ornativentris at the same point as in
the last survey. We suggest that female R. orna-
tiventris also perhaps exhibit site fidelity. Since
the number of samples is still small at present,
the scale of the summer home range for brown

frogs should be studied in more detail by further
studies.

According to a survey also conducted in
Tokyo, most individuals of the eastern Japanese
common toad (Bufo japonicus formosus) that
metamorphose and emerge in May or June
determine their home ranges in the fall of the
same year, especially in October or later (Yano,
1978). Yearlings of R. japonica seem to deter-
mine their summer habitats by August or Sep-
tember, because the distances between capture
and recapture points are small (27.6 m, N=18)
in comparison with those of early and middle
July. In contrast, yearlings of R. ornativentris
seem to continue migrating slowly until fall
(49.0 m, N=2 and 41.7 m, N=9).

Dispersal capability

Concerning the dispersal of yearlings and
adults, most (about 90% of the total captured
frogs) were found within 240 to 310 m from the
spawning sites. Not many but some individuals
of both species were found about 500 m away
from their spawning sites. This shows that the
dispersal capability is at least 500 m. Since the
current survey was conducted on isolated forest
land in the Tama Hills, the dispersal capability
may actually be greater. According to a sur-
vey conducted on the Boso Hills of the Chiba
Peninsula covered with continuous forest, how-
ever, the migrating distances were 290 to
1000 m for R. japonica and 400 to 1840 m for
R. ornativentris at one basin of the Koito River
(Hasegawa, 1994), and within 200m for R.
Japonica and within 500 m for R. ornativentris
at the source of the Urajiro River (Hasegawa,
1998). In other words, it seems that the dis-
persal capability of both species is from several
hundreds meters to about one km. And the dis-
persal capability of R. ornativentris may per-
haps be even greater.

In this survey, the greatest percentages of
yearlings were found in the section (mode) of
75 to 100 m as shown in Fig. 5 and 6. This may
also be attributable to a decrease of density in
concentric dispersion because the survey time is
almost constant despite the distance from the
spawning sites. However, the number of cap-

Current Herpetol. 20(1) 2001

tured individuals does not always decrease in
reverse proportion to the distance. About 10
yearlings of R. ornativentris were captured in
the range from 25 to 275m. Therefore, the
reason for the quick decrease of individuals
captured beyond about 250 to 310 m is proba-
bly that the number of dispersed individuals
is actually small. We conclude that dispersal
of the two species is centered around the valley
lowland, because we could only find them in
low density at the far end of the continuous for-
est.

In the Tama Hills or similar terrain where
valleys exist among hills, the forest habitat for
the non-breeding season is close to a water
body for breeding. Under these environmental
conditions, many individuals do not disperse far,
but have summer habitats. On the other hand,
not many but some individuals of both species
were found dispersed about 500 m along the
ridge or in the neighboring valley catchment
basin beyond the bridge. Many species of
amphibians ensure genetic exchange with adja-
cent populations and colonization into new
breeding sites by the migration of juveniles
to maintain meta-populations (Beebee, 1996;
Berven and Grudzien, 1990). In the Tama Hills
where forested hills and wet valleys form a
complicated pattern, most yearlings stay around
the breeding sites but a small number travel
long distances for dispersal. This is how genetic
exchange with breeding populations has been
maintained in valley lowlands.

Although the area of the current survey was
limited to within the forest (including the forest
edge), both R. japonica and R. ornativentris are
widely dispersed in the forest and live there in
the non-breeding season. The survey showed
that the forest edge is used especially by adults
of R. japonica but rarely by R. ornativentris.
The survey results indicated that R. japonica
can live even in grassland alone, but that R.
ornativentris requires wide forests (e.g.,
Tomioka, 1990; Hasegawa, 1995). More year-
lings of R. japonica were found in the forest
than adults. This may probably be because
yearlings have an intrinsic desire for disper-
sive migrating, but the rationale is not secure

OSAWA & KATSUNO—DISPERSAL OF FROGS y)

yet. Yearlings and adults of R. ornativentris do
not use the forest edge much but how the edge of
the forest they need to penetrate needs further
study.

Implications for conservation

In the Tama Hills, habitats of the two spe-
cies of brown frog are decreasing rapidly
(Osawa and Katsuno, 1998; Sato, 1998).
Now these species are found in only about 20
places in the central and southern areas of the
Tama Hills. Urbanization is the greatest fac-
tor for this decline and is reducing not only
brown frogs but other amphibians as well.
Therefore, a regional-level strategy is neces-
sary for conserving amphibians, including meta-
populations in all the areas of the Tama Hills
(Osawa and Katsuno, 2000). For this conser-
vation, the population size and conditions of
each species should be checked to learn what
land units, such as habitat patches and eco-
logical corridors, they need. In this study, the
two types of brown frogs (R. japonica and R.
ornativentris) most numerous in Japan were
selected as the subjects of survey. The area of
survey was an isolated undeveloped patch of
land of about 28 ha consisting of valley low-
land for spawning and adjacent hilly forests.
About 200 to 300 egg masses of each species
were found in this area. This made us esti-
mate that undeveloped land under these con-
ditions can function stably as a patch where
both species can live. Further detailed studies
are certainly necessary on microhabitats in
the conserved valleys and forests, such as
vegetation, edges of forests, ridges, and
streams, to clarify the relationships between
the locations and structures and the status of
use. The lives of both species in the non-
breeding season are not clear yet. For the
conservation of common frogs, studies need
to be accumulated and linked at each regional
level, landscape unit level, microhabitat level,
and life history level. With the results of the
survey in mind, what should be noted in land-
scape and open space planning to conserve
both species in the Tama Hills is summarized
here.

(1) Both species of brown frogs occur widely
in the forests. Wide patches of forests should
be reserved in a radius of about 300 m from a
spawning site.

(2) Rana japonica was found frequently at
the edge of the forest. This indicates that the
edge has an important role for R. japonica.
For this species, not only the forest but also
the ecotope from grassland to forest should be
maintained adjacent to each other and a zone
of contact contact should be reserved.

(3) Not many R. ornativentris were found at
the edge of forests because the main habitat is
inside the forests. An ecological corridor link-
ing patches should be continuous forest of an
adequate width.

(4) The maximum distances of dispersion in
this survey were 475 m for R. japonica and
560m for R. ornativentris. Both species
showed the dispersal capabilities of about
500 m or more. One of the indexes for extend-
ing the distributions of both species in the
reserved forest and valley network may be
reserving spawning sites at about 500 m inter-
vals.

ACKNOWLEDGMENTS

We thank Takuya Obata, Junya Katano and
the members of the laboratory of Landscape
science and planning, Nihon Univ. for their
assistance in field surveys. We also thank K.
Fukuyama and M. Hasegawa for literature.

LITERATURE CITED

BEEBEE, T. J. C. 1996. Ecology and Conservation of
Amphibians. Chapman & Hall, London. 214 p.

BELLIS, E. D. 1965. Home range and movements of
the wood frog in a Northern Bog. Ecology 46: 90-
98.

BERVEN, K. A. and T. A. GRUDZIEN. 1990. Dispersal
in the wood frog (Rana sylvatica): Implications for
genetic population structure. Evolution 44(8):
2047-2056.

CRUMP, M. and J. N. SCoTT. 1994. Visual encounter
surveys. p. 84-92. In: W. R. Heyer, M. A. Don-
nelly, R. W. McDiarmid, L.-A. C. Hayek, and
M. S. Foster (eds.), Measuring and Monitoring

10

Current Herpetol. 20(1) 2001

Biological Diversity. Standard Methods for
Amphibians. Smithsonian Inst. Press, Washington,
DC:

DUELLMAN, W. E. and L. TUEB. 1986. Biology of
Amphibians. McGraw-Hill Publishing Co., New
York. 670 p.

FORMAN, R. T. T. 1995. Land Mosaics: The Ecology
of Landscapes and Regions. Cambridge University
Press, New York. 632 p.

GREEN, D. M. 1997. Perspectives on amphibian pop-
ulation declines: defining the problem and search-
ing for answers. p. 291-308. Jn: D. M. Green (ed.),
Amphibians in Decline: Canadian Studies of a Glo-
bal Problem. SSAR, Saint Louis.

HASEGAWA, M. 1994. Researches on amphibians and
reptiles essential for environmental assesment (5).
In: M. Numata (ed.), Contract Report on the
Research Protocol for Environmental Assessment
in the Region under Urban and Industrial Develop-
ment. p. 32-39. Division of Environ. Conserv. and
Policy, Chiba Pref. (in Japanese)

HASEGAWA, M. 1995. Brown frogs in valley-bottom
paddy fields. p. 105-112. Jn: T. Ohara and M.
Osawa (eds.), Natural History in the Southern
Kanto Plain. Asakura Shoten, Tokyo. (in Japanese)

HASEGAWA, M. 1996. Status and distribution of rep-
tiles and amphibians in Chiba city: Effects of
changes in the Yatsu valley habitat. p. 505—521. In:
M. Numata (ed.), Conservation of Biodiversity:
Surveys of Species, Communities and Ecosystems
in Chiba City. Shinzan-sha, Tokyo. (in Japanese
with English summary)

HASEGAWA, M. 1998. Dependency of frog communi-
ties on rice cultivation in the paddy fields. p. 53-
66. In: Y. Ezaki and T. Tanaka (eds.), Conservation
of Land-water Ecotone: Views from Community
Ecologists. Asakura Shoten, Tokyo. (in Japanese)

HASEGAWA, M., T. KUSANO, and K. FUKUYAMA.
2000. How have declining amphibian populations
been perceived by national, academic and regional
communities in Japan. J. Nat. Hist. Mus. Chiba,
Spec. Issue (3): 1-7. (in Japanese with English
summary)

HIOKI, Y. and K. IDE. 1997. A study on planning pro-
cess of ecological networks: Comparison of three
cases in the Netherlands. J. Jpn. Inst. Landscape
Architect. 60(5): 501-506. (in Japanese with
English summary)

KURAMOTO, M. and H. ISHIKAWA. 2000. Breeding
ecology of brown frogs in Yamada park, Kitak

ushu, Japan. Bull. Herpetol. Soc. Jpn. 2000(1):
7-18. (in Japanese with English abstract)

KUSANO, T. and K. MIYASITA. 1984. Dispersal of
the salamander, Hynobius nebulosus tokyoensis.
J. Herpetol. 18(3): 349-353.

KUSANO, T., K. MARUYAMA, and S. KANEKO.
1995. Post-breeding dispersal of the japanese
toad, Bufo japonicus formosus. J. Herpetol.
29(4): 633-638.

KUSANO, T. 1998. A radio-tracking study of post-
breeding dispersal of the treefrog, Rhacophorus
arboreus (Amphibia: Rhacophoridae). Jpn. J.
Herpetol. 17(3): 98-106. :

MAEDA, N. and M. MATsuI. 1989. Frogs and
Toads of Japan. Bun-ichi Sogo Shuppan, Tokyo.
198 p. (in Japanese with English abstract)

OKUNO, R. 1986. Studies on the natural history of
the japanese toad, Bufo japonicus japonicus. X.
Mating and spawning behaviors. Jpn. J. Ecol.
36(1): 11-18. (in Japanese with English synop-
sis)

OSAWA, S. and T. Katsuno. 1998. Discussion of
peculiarity in Yato (paddy field at bottomland
with sideslope) and conservation of frogs. J. Jpn.
Inst. Landscape Architect. 61(5): 529-534. (in
Japanese with English summary)

OSAWA, S. and T. KATSUNO. 2000. The discussion
of conservation and the habitat’s requirements of
schlegel’s green tree frog (Rhacophorus
schlegelii) on the south Tama-hills. J. Jpn. Inst.
Landscape Architect. 63(5): 495-500. (in Japa-
nese with English summary)

RDRGK (Red Data Research Group of Kanagawa).
1995. The red data species of vascular plants,
vertebrate animals and insects in Kanagawa Pre-
fecture. Res. Rep. Kanagawa Prefect. Mus. Hist.
(7): 133-136. (in Japanese)

SATO, D. 1998. Study on the ecological distribu-
tion of frogs in urban area. Unpubl. Master’s
Thesis, Yokohama University, Kanagawa. 27
p.+55 pls. (in Japanese)

ToJo, Y. 1976. The growth curve of Rana japon-
ica. Niigata Herpetol. J. 6: 18-19. (in Japanese)

TOMIOKA, K. 1990. Habitat segregation of Rana
Japonica and R. ornativentris of Gumma prefec-
ture and adjacent regions. Nippon Hepetol. J. 39:
21-28. (in Japanese)

YANO, M. 1978. Ecological studies of Bufo bufo
japonicus Schlegel. (111) movements. Rept. Inst.
Nat. Stud. (8): 121-133. (in Japanese)

Current Herpetology 20(1): 11-17., June 2001
© 2001 by The Herpetological Society of Japan

Female-biased Mortality and Its Consequence on Adult Sex
Ratio in the Freshwater Turtle Chinemys reevesii on an Island

TOSHIAKI TAKENAKA" AND MASAMI HASEGAWA2

! Department of Soil Zoology, Institute of Environmental Science and Technology, Yokohama
National University, Tokiwadai, Hodogaya-ku, Yokohama, 240-8501 JAPAN
“ Department of Ecological Science, Natural History Museum and Institute, Chiba, Aoba-cho,
Chiba, 260-8682 JAPAN

Abstract: A mark-recapture study of the fresh water turtle Chinemys reevesii
was conducted on a small (190 ha) island, Ishima Island, off the east coast of
Shikoku to learn the demographic characteristics of an insular turtle popula-
tion. Most of the Chinemys reevesii on the island were marked, yielding a reli-
able estimate of population size (1179 turtles per 15 ha aquatic habitats). The
estimated sex ratio of the turtles in the younger age class (3—6 yr) was essen-
tially equal, slightly skewed toward males in a higher age class, and significantly
male-biased in the older age class (>12 yr). A significantly female-biased sex
ratio among the turtles found dead was compatible with the male-biased sex
ratio in adult turtles. Possible physiological and ecological factors for change in
sex ratio with age are discussed with consideration of the insular environment.

Key words: Sex ratio; Mortality; Island population; Turtle; Chinemys

INTRODUCTION

Ecological peculiarities of insular popula-
tions of reptiles are well documented in the
recent literatures (Case, 1983; Hasegawa and
Moriguchi, 1989; Hasegawa, 1994). How-
ever, a few studies have dealt with the fresh-
water turtles on the insular environments.
Among such studies, Gibbons et al. (1979)
reported that the slider turtle Trachemys
scripta attains larger body sizes on Atlantic

i Corresponding author. Tel: +81—45—339-4357;
Fax: +8 1—45-—339-4375.

E-mail address: takenaka@kan.ynu.ac.jp (T. Tak-
enaka)

Coast barrier islands than populations on the
mainland. Yasukawa and Ota (1999) studied
geographic variations in morphological charac-
ters of Mauremys mutica and Cistoclemmys

flavomarginata on the islands of Yaeyama and

Taiwan, and in continental China. Tokumoto
and Yano (1998) studied an island population
of Reeves’ turtle (Chinemys reevesii) as a part
of an environmental assesment evaluating the
effects on insular biota of constructing a reser-
VOIr.

We studied the structure of a population of
Reeves’ turtle on a small island and compared
it with another island population (Tokumoto
and Yano, 1998) and a mainland population
(Yabe, 1994).

Current Herpetol. 20(1) 2001

me 1000 1500

m

Fic. 1. A map showing the study area. Ishima island is located ca. 6 km east of Kamoda Peninsula,
Shikoku. Shaded areas on the map of Ishima indicate swamps and ponds where the turtle survey was con-

ducted.

MATERIALS AND METHODS

Study area

Ishima (33°60'N, 134°40'E) is a small island
(190 ha) located ca. 6 km east of the Kamoda
Peninsula of Anan City, Tokushima Prefecture
(Fig. 1). The relatively recent separation of the
island by sea level change is suggested by the
shallow channel depth (<50 m). The island is
close proximity (Japan Environment Agency,
1982).

Potential aquatic habitats for the turtles have
a total area of 15 ha, consisting of two man-
made reservoirs (Otamegaya-dam and Kota-

megaya-dam), an irrigation pond, and two com-
plexes of marshes and shallow ponds formerly
used as rice paddies but now abandoned (Fig.
1). There is no river on the island.

Areas surrounding water bodies are exten-
sively covered by forest dominated by holm
oak (Quercus phillyraeoides) with scattered
patches of vegetable fields (Abe, 1990).

The fauna of Ishima has been little studied
except for birds (Japan Wild Bird Society,
1990). The impact of birds is not well known,
and during our study we did not see any birds
of prey on Ishima. The only known potential
predator on the turtles is the Japanese four-lined

TAKENAKA & HASEGAWA—TURTLE SEX RATIO its

snake (Elaphe quadrivirgata) which preys on
turtle eggs and hatchlings (Mori and Moriguchi,
1989). Mammalian carnivores such as the rac-
coon dog (Nyctereutes procyonoides), the bad-
ger (Meles meles) and the weasel (Mustela
itatsi) are absent from the island. Though feral
cats (Felis catus) are commonly seen around
the village and fishing port, their predation
impact on the turtle population is questionable.
We captured a small number of another turtle
species, Trachemys scripta elegans, which is a
potential competitor with C. reevesii.

Population survey

We visited Ishima 17 times from December
1998 to February 2000. For each survey, we
stayed 3-4 days on the island, for a total of 62
days. Turtles were captured by baited trap
method (Gibbons, 1990) in the open water
bodies (two reservoirs, an irrigation pond, and
open water within marshes), and by hand catch-
ing in the shallow portion of marshes and
ditches. The traps used in this study were
dome-shaped funnel traps, and fish scraps were
used as bait. The traps were set in the daytime
and left undisturbed overnight except when
captured turtles were removed for measure-
ment. The baits were replaced at these times.

All turtles captured were processed at the
points of capture to record species, sex, age,
body size (carapace length), body mass, and
degree of melanism (Yabe, 1994). Sex was
determined by examining the position of the
cloaca relative to the margin of the carapace by
stretching the tail straight back. Turtles with

the cloacal opening beyond the posterior mar-
gin of carapace were judged to be males, and
those with the cloacal opening within the mar-
gin of the carapace, females (Cagle, 1954), but
the tails of juvenile turtles could not be extended
to determine sex in many cases.

Straight line lengths of carapace and plastron
were measured to the nearest mm with a ver-
nier caliper, and body mass to the nearest gram
with a spring scale. Age was determined by
counting plastral growth annuli (Sexton, 1959).
All annuli were visible on individuals up to and
including 14-year males and 18-year females.
Large individuals with invisible plastron annul
were treated as unaged older turtles.

Turtles were assigned to three age classes (3—
6 yr, 7-12 yr, and 13 yr<), and the sex ratio was
compared among age classes (Table 1). On
Ishima, the lowest age of females with eggs
was seven years, and all females older than 12
years had eggs in reproductive seasons (Tak-
enaka and Hasegawa, unpublished data).

All turtles were individually marked by drill-
ing small holes (2 or 3 mm in diameter depend-
ing on size) in the marginal scutes of carapace
for individual recognition. We tried to capture
and mark all individual turtles in every aquatic
habitat on the island. Population size was esti-
mated by the Jolly-Seber method.

Carcass survey

During the preliminary survey for wintering
turtles in December 1998, several bleached
skeletons were discovered. Subsequently, we
paid special attention to collect these materials

Table 1. Shift in sex ratio by age classes of Reeves’ turtle on Ishima island. z value of the binominal test is
shown in the table with asterisk indicating significant deviation from sex ratio.

3-6 yr 7-12 yr >12 yr
Male 96 56 700
Female 21 45 139
Unknown 76 0 0
Sex ratio (male/all) 0.49 0.54 0.81
Z, 0.072 1.095 19.37
N 193

14

Current Herpetol. 20(1) 2001

TABLE 2. Regression equations for the relationship between the carapace length (Y) and the length of three
plastron bones (Xn) in 16 Reeves’ turtles from Ishima island.

Name of plastron bone Regression equation r p

Epiplastron (X,) Y=9.21X,411.2 0.871 <0.05
Hypoplastron (X>) Y=3.23X,131.4 0.857 <0.05
Xiphiplastron (X3) Y=7.04X,+5.09 0.911 <0.05

to obtain the size distribution of dead turtles.
When detected, we recorded substrate type,
condition of dead bodies, sex, and straight
length of carapace if possible. When broken
shells were found, measurements were taken on
the lengths of bones comprising the plastron
along the body axis from snout to vent. In
order to estimate carapace length from the plas-
tron bones, appropriate regression equations
were established based on measurements of
both carapace length and lengths of epi-, hypo-,
and xiphi-plastrons for the dead bodies with a
complete shell (Table 2).

RESULTS

We marked 1,123 individuals of C. reevesii.
The proportion of newly captured unmarked
turtles decreased with advance of the field sur-
vey to ca. 5% at the 13th survey and remained
constant thereafter (Fig. 2). The estimated pop-
ulation size of C. reevesii was 1,179 with a
95% confidence limit of 38.3 at the 17th sur-
vey.

Of 1,123 C. reevesii captured, 833 were
males, 214 were females, and the remaining 76
were juveniles of undetermined sex. The mean
carapace lengths and weights of the old indi-
viduals whose ages were unknown were
131.9 mm (1SD=14.29, range=88.0—164.6 mm),
353.1 g (ISD=95.96, range=136—670 g) in
males (N=617), and 165.5 mm (SD=20.76,
range=114.7—209.4 mm), 743.3 g (SD=238.86,
range=250—1450 g) in females (N=148). In
smaller juveniles, it was difficult to distinguish
sex by the tail. However, we could determine
three-year-old juvenile males. So all turtles of
undetermined sex older than three years were
assumed to be females. Estimated sex ratio of

Proportion of newly marked turtles

0 3 6 9 12 15 18
Order of field survey (Times)

FIG. 2. Sequential change in the proportion of
newly marked turtles in relation to the order of field
survey. Numbers on the curve are the number of
captured turtles.

the turtles was essentially equal in the youngest
age class (3-6 yr) (males:females = 0.49:0.51;
binominal test, z=0.072, p>0.05), slightly
skewed toward males in the middle age class
(0.54:0.46; z=1.095, p>0.05), and significantly
male-biased in the turtles older than 12 yrs
(0.83:0.17; z=19.37, p<0.001) (Table 1).
Thirty-nine dead bodies identified as C.
reevesii were collected in 1998-1999. Two
were found dead on the ground in summer, and
the other 37 were collected on the muddy bot-
tom (>70 cm from the water surface) of marshes
and ponds together with wintering but living
individuals. Of 39 dead bodies, eight were rel-
atively fresh and their sex was easily deter-
mined (six females and two males). The
remaining 31 bodies were bleached shells or
clusters of broken plastral shell bones, and their
carapace length and sex were estimated indi-
rectly with the regression equations between

TAKENAKA & HASEGAWA—TURTLE SEX RATIO 15

50

CL] Living individuals
Female dead

Sex unknown dead

40

30

20

Number of female turtles

|

50 100 150 200

Yi

Carapace length (mm)

Fic. 3. Frequency distribution histograms of the
carapace lengths of female turtles caught on Ishima
island (N=214 living and 39 dead turtles).

size of plastral bones and carapace length
(Table 2).

Twenty-six of 31 undetermined bodies were
assigned as old females (Fig. 3), because their
estimated carapace lengths exceeded the maxi-
mum recorded carapace length of male turtles
on Ishima (178 mm: Takenaka and Hasegawa,
unpublished data). Consequently, at least 32
~ (82.1%) of 39 dead individuals were assigned
as females. The sex ratio of dead bodies was
significantly female-biased (binominal test,
z=4.00, p<0.01), and was significantly different
from that of living individuals older than seven
years (two-tailed chi-square test; y°=85.0, df=1,
p<0.001).

DISCUSSION

The nearly complete census of C. reevesii on
Ishima made demographical analysis truly real-
istic.

The male-biased sex ratio in the older age
class was thus actual but not superficial. The
population structure of C. reevesii on Ishima
is different from that reported for the main-
land population (Yabe, 1994). Yabe (1994) sug-
gested that female C. reevesii in the mainland
aquatic habitats exhibited higher survivorship
and lived longer than males. A field study of
another island population of C. reevesii revealed
a population structure similar to our results; the
sex ratio of C. reevesii on an island off the

northern coast of Yamaguchi Prefecture was
found to be male-biased (Tokumoto and Yano,
1998). A pattern of male-biased sex ratio might
be a general tendency of insular C. reevesii pop-
ulations.

Gibbons (1990) discussed four demographic
factors that influence actual sex ratios within a
turtle population: (1) sex ratios of hatchlings, (2)
differential mortality of the sexes, (3) differen-
tial emigration and immigration rates of the
sexes, and (4) differences in age at maturity of
the sexes. First, we can eliminate the factor of
differential rates of emigration and immigra-
tion, because the population of our study was
completely confined to the island. Because we
did not investigate breeding of the population,
we can not discuss age at maturity as a factor
influencing sex ratio of mature and reproduc-
tively active individuals. Instead, sex ratios of
comparable age classes were analyzed indepen-
dently to see how sex ratios change with age. If
hatchling sex ratios were equal, younger turtles
should have an equal sex ratio. Estimated sex
ratio among the turtles in the age class of 3—
6 yrs was equal. Slightly male-biased but sta-
tistically equal sex ratios among the turtles at
7-12 yr suggests an equal sex ratio not only in
the younger age classes but also at the time of
hatching (Table 1).

Finally, we discuss differential mortality
between sexes as a prime factor influencing
adult sex ratio within this insular turtle popula-
tion of C. reevesii. The female-biased sex ratio
of dead bodies indicates higher mortality in
adult females, which in turn cause the adult sex
ratio to be male-biased. Because there are no
native mammalian carnivores capable of killing
and predating adult turtles, natural mortality of
the turtle population on Ishima could not be
from predation but from natural death by sim-
ple aging or death caused by nutritional defi-
ciency or disease.

Most dead turtles were larger than the mini-
mum size of mature females (carapace length
>170 mm: Yabe, 1994) and were found at the
sites similar to those of the wintering indi-
viduals, suggesting that females died while
overwintering. One possible reason for such

16

Current Herpetol. 20(1) 2001

mortality pattern is that some mature females
were unable to recover nutritionally from repro-
duction before winter. Because female turtles in
general are known to expend more energy on
reproduction than males (Congdon and Tinkle,
1982) and because female C. reevesii have larger
body size (Yabe, 1994), depletion of energy or
nutrition would be more serious in females than
in males.

Systematic differences in adult mortality
between sexes are rare in turtle populations
(Gibbons, 1990; Iverson, 1991), but consistently
higher female mortality is observed in the desert
tortoise Gopherus berlandieri (Hellgren et al.,
2000). They proposed that female G berlandieri
experienced calcium deficiency by an interac-
tion between the calcium cost of egg laying and
the consumption of cactus fruit. High concentra-
tions of calctum oxalates are contained in cactus
fruit. Calcium oxalates reduce calcium avail-
ability in the diet, and this intensifies bone
resorption for egg production. In G berlandieri,
calcium deficiency is hypothesized ultimately to
cause higher female mortality from increased
susceptibility to pathogens or from calcium defi-
ciency itself (Hellgren et al., 2000).

Similar physiological mechanisms might
operate in female C. reevesii on Ishima. Means
of carapace lengths and weights of the old indi-
viduals were markedly smaller than those of
mainland populations (Yabe, 1994). The same
phenomenon has also been shown in males.
This suggests the influence of nutritional defi-
ciency. Monitoring both physiological condi-
tions and demography of female turtles is
needed to test the proposed physiological mech-
anisms causing higher female mortality. How-
ever, environmental conditions peculiar to
islands should also be considered.

Demographically, the high population density
of C. reevesii on Ishima would reduce food
intake per individual turtle. Approximately
1200 individual turtles are living in a relatively
limited area of aquatic habitats on Ishima.

Though we do not have any comparable fig-
ures for the turtle density or biomass in Japan,
the population density of C. reevesii on Ishima
(ca. 80 turtles/ha in aquatic habitats) is compa-

rable to the highest reported densities of fresh-
water turtles of comparable body size (41.8—
88/ha for Trachemys scripta, and 39.9-89.5/ha
for Chrysemys picta) in North American wet-
lands (Congdon et al., 1986).

Higher density and change in size composi-
tion of insular turtle populations are also
known for T. scripta (Gibbons et al., 1979;
Congdon et al., 1986). However, the effects of
different demographic environments on turtle
life history have not yet been studied ade-
quately. A comparable population study of C.
reevesii in both island and mainland habitats
now being conducted by us will offer a promis-
ing opportunity to pursue this field of turtle
biology.

ACKNOWLEDGMENTS

This study was financially supported by a
grant of the research project on conservation
methods of subtropical island ecosystems to
M. Hasegawa. We are very thankful to Y.
Kosuge and R. Kobayashi for their assistance
in our field work.

LITERATURE CITED

ABE, K. 1990. Flora of Tokushima-Prefecture. Cen-
tral Study of Education Publication, Inc. 580 p.
(in Japanese)

CAGLE, F. R. 1954. Observetion on the life cycles
of painted turtle (genus Chrysemys). Am. Midl.
Natur. 52(1): 225-235.

CASE, T. J. 1983. The reptiles: ecology. p. 159-209.
In: T. J. Case and M. L. Cody (eds.), Island Bio-
geography in the Sea of Cortez. Univ. California
Press, Berkeley.

CONGDON, J. D. AND D. W. TINKLE. 1982. Repro-
ductive energetics of the painted turtle (Chryse-
mys picta). Herpetologica 38(2): 228-237.

CONGDON, J. D., J. L. GREENE, AND J. W. GIBBONS.
1986. Biomass of freshwater turtles: A geo-
graphic comparison. Am. Midl. Natur. 115(1):
165-173.

GIBBONS, J. W. 1990. Sex ratios and their signifi-
cance among turtle populations. p. 171-182. Jn:
J. W. Gibbons (eds.), Life History and Ecology of
the Slider Turtle. Smithsonian Inst. Press, Wash-
ington, D. C.

GIBBONS, J. W., G. H. KEATON, J. P. SCHUBAUER, J.
L. GREENE, D. H. BENNET, J. R. MCAULIFFE,

TAKENAKA & HASEGAWA—TURTLE SEX RATIO 17

AND R. R. SCHRITZ. 1979. Unusual population
size structure in freshwater turtles on barrier
islands. Georgia. J. Sci. 37: 155-159.

HASEGAWA, M. 1994. Insular radiation in life his-
tory of the lizard, Eumeces okadae on the Izu
Islands. Copeia 1994(3): 732—747.

HASEGAWA, M. AND H. MORIGUCHI. 1989. Geo-
graphic variation in food habits, body size and life
history traits of the snakes on the Izu Islands. p.
414-432. Yn: M. Matsui, T. Hikida, and R. C.
Goris (eds.), Current Herpetology in East Asia.
Herpetol. Soc. Jpn., Kyoto.

HELLGREN, E. C., R. T. KAZMAIER, D. C. RUTHVEN
Ill, AND D. R. SYNATZSKE. 2000. Variation in
tortoise life history: Demography of Gopherus
berlandieri. Ecology 81(5): 1297-1310.

IVERSON, J. B. 1991. Patterns of survivorship in tur-
tles. Can. J. Zool. 69: 385-391.

Japan Enviroment Agancy. 1982. Actual Vegetation
Map (1) Ishima, Tokushima Prefecture. (in Japa-
nese)

Japanese Wild Birds Society. 1987. A Catalog of
Bird in Tokushima Prefecture. 228 p. (in Japanese)

Mort, A. AND H. MORIGUCHI. 1988. Food habits of
the snakes in Japan: A critical review. Snake 20:
98-113.

SEXTON, O. J. 1959. A method of estimating the age
of painted turtle for use in demographic studies.
Ecology 40(7): 716-718.

TOKUMOTO, T. AND H. YANO. 1998. On turtles of
Mishima Island, Hagi-City, Yamaguchi Prefec-
ture. Seibutu- Yamaguchi 25: 17—24. (in Japanese)

YABE, T. 1994. Population structure and male mela-
nism in the reeves turtle, Chinemys reevesii. Jpn.
J. Herpetol. 15(4): 131-137.

YASUKAWA, Y. AND H. OTA. 1999. Geographic
variation and biogeography of the geoemydine
turtles (Testudines: Bataguridae) of the Ryukyu
Archipelago, Japan. p. 271-297. In: H. Ota (ed.),
Tropical Island Herpetofauna: Origin, Current
Diversity, and Conservation. Elsevier, Amster-
dam.

tra ao 7

~

Current Herpetology 20(1): 19-25., June 2001
© 2001 by The Herpetological Society of Japan

Karyotype of the Chinese soft-shelled turtle, Pelodiscus
sinensis, from Japan and Taiwan, with chromosomal data for
Dogania subplana

HIROYUKI SATO! AND HIDETOSHI OTA2™

! Graduate School of Science and Engineering, University of the Ryukyus, Nishihara,
Okinawa, 903—0213 JAPAN
* Tropical Biosphere Research Center, University of the Ryukyus, Nishihara, Okinawa, 903-
0213 JAPAN

Abst[ract: The male and female karyotypes were examined for 10 specimens of
the Chinese soft-shelled turtle, Pelodiscus sinensis, from the main islands of
Japan and Taiwan by the bone marrow air-dry technique. One specimen of
Dogania subplana was also examined karyologically. Although both P. sinensis
and D. subplana had 2n=66 chromosomes, they showed an interspecific varia-
tion in chromosome morphology. No karyotypical differences were evident
between the Japanese and Taiwanese samples of P. sinensis, or within either of
them. Comparisons of the present results with previously reported trionychine
karyotypes indicate that there are some interspecific and intraspecific varia-
tions in centromeric positions of several macrochromosome pairs in this sub-

family.

Key words: Turtle; Chromosomes; Japan; Tatwan; Karyotype

INTRODUCTION

The Chinese soft-shelled turtle, Pelodiscus
sinensis (Wiegmann, 1835), is widely distrib-
uted in the eastern part of the Eurasian conti-
nent and adjacent islands, i.e., an area from
eastern Russia to northern Vietnam, including
the main islands of Japan, Taiwan, Hong Kong,
and Macao (type locality) (Iverson, 1992).
Based on an allozyme study, Sato and Ota
(1999) reported genetic differentiations between
populations of this species from the main islands

: Corresponding author. Tel: +81—98—895-8937;
Fax: +8 1—98—895-8966.
E-mail address: ota@sci.u-ryukyu.ac.jp (H. Ota)

of Japan, and Taiwan and Hong Kong. Oguma
(1936, 1937), in the initial description of the
karyotype of P. sinensis, reported that male and
female turtles from southern Japan (probably
Kyushu) had 2n=64 chromosomes in a sper-
matogonium and 63 in an oogonium, respec-
tively. He argued that the difference in diploid
number between male and female cells repre-
sents the ZO type of sex chromosome hetero-
morphism. Later, Suzuki (1950) also reported
64 and 63 chromosomes, respectively, for male
and female soft-shelled turtles from Korea.
More recently, however, Bickham et al. (1983)
claimed that all trionychid specimens exam-
ined by them, including a male P. sinensis
from an unknown locality, had 2n=66 chrom

20

Current Herpetol. 20(1) 2001

somes without any heteromorphism. They,
while ignoring Suzuki’s (1950) report, assumed
that the difference between their results and the
results of Oguma’s (1936, 1937) investigation
was derived from miscounting of chromosomes
by the latter. Rong and Li (1984) also reported
the diploid chromosome number in P. sinensis
from continental China to be 66. However, no
authors subsequent to Oguma (1936, 1937) and
Suzuki (1950) examined Japanese and Korean
populations of P. sinensis karyologically.

Considering that both Oguma (1936, 1937)
and Suzuki (1950) adopted the classical par-
affin sectioning, a method which sometimes
makes it difficult to discriminate small chromo-
some elements correctly (Gorman, 1973), the
claim by Bickham et al. (1983), who employed
a more advanced cell culturing with leukocyte
cells for chromosome preparation, sounds plau-
sible. However, there is another possibility that
the differences between results in those works
actually reflect the differentiation between the
Japanese-Korean and other populations as was
demonstrated by allozyme analysis (Sato and
Ota, 1999).

Thus, we re-examined the karyotype of P.
sinensis from Japan and Taiwan by the bone
marrow air-dry method, because this method
yields better resolution of metaphase cells than
the gonadal sectioning (Sharma, 1980). A spec-
imen of the Malayan soft-shelled turtle, Doga-
nia subplana, one of the closest relatives of P
sinensis (see Meylan, 1987), was also exam-
ined.

MATERIALS AND METHODS

A total of 10 hatchlings of P. sinensis and
one hatchling of D. subplana were used. Of
these specimens, five, four, and one P. sinen-
sis were from Amamioshima Island of the
Ryukyu Archipelago, Taiwan, and Minamidai-
tojima Island, respectively, whereas the sam-
ple of D. subplana was purchased from a pet
dealer and its locality was thus unknown. Sam-
ples of P. sinensis from Amamioshima and
Minamidaitojima are regarded as representing
populations from the mainislands of Japan

and Taiwan, respectively, since current popula-
tions on those two Islands of the Ryukyus are
known to have been derived from recent artifi-
cial introductions from these main islands (Sato
et al., 1997; Sato and Ota, 1999). Voucher
specimens are deposited in the herpetological
collection of the Department of Zoology, Kyoto
University (KUZ), as KUZ 48712-48716 (speci-
mens from Amamioshima), 48717 (Minamidai-
toujima), 48735, 48737-48739 (Taiwan), and
47851 (D. subplana). |

All specimens were injected with colchicine
solution (0.1 mg/ml) into their body cavities at
0.05 mg/g body weight. They were euthanized
with pentobarbital solution about 12 hr after
the colchicine injection. Bone marrow cells,
extracted from a femur of each specimen, were
subjected to hypotonic treatment in KCI solu-
tion (0.05 mol/l) for 40 min, and then were
rinsed and fixed in Carnois solution (glacial
acetic acid:methanol, 1:3). Chromosome prep-
arations made by the air-dry method were
soaked in 3% Giemsa solution for 30 min and
then were observed microscopically and pho-
tographed. The karyotype was determined for
each individual on the basis of at least five
well-spread metaphase cells. Terminology for
chromosomal descriptions follows Green and
Sessions (1991).

Specimens karyotyped were sexed by his-
tological examination of gonadal sections.
Gonads, removed from each specimen along
with an adjacent portion of kidney, were infil-
trated and embedded in paraffin, and then were
sectioned at 10 um and stained with hematox-
ylin and eosin solutions. Sexual identification
follows the criteria used by Yntema and Mros-
ovsky (1980).

RESULTS

Microscopical observations of gonadal sec-
tions explicitly showed sexual features for
each specimen. In male gonads, no cortex was
observed, and cells in the medullary region
were organized into seminiferous tubules (Fig.
1A). In female gonads, on the other hand, a rel-
atively thick cortex covered the ventral surface

SATO & OTA—KARYOTYPE OF SOFT-SHELLED TOURTLE 2)|

FIG. 1. Histological sections of gonads from male (A: KUZ 48737) and female (B: KUZ 48739) hatchlings
of Pelodiscus sinensis. K, kidney; S, cells organized into seminiferous tubules; C, cortex; and M, unorganized
medulla. Horizontai scale bars represent 50 tm.

22

Current Herpetol. 20(1) 2001

TABLE 1. Diploid chromosome numbers and morphology of macrochromosomes (Nos. 1—8) in trionychine
turtles reported so far. Populations on Amamioshima and Minamidaitojima are known to have been derived
from artificial introductions from the main islands of Japan and Taiwan, respectively (see text).

Sample size
Tissues (male/ Diploid Morphology of chromosome pairs 1-8,
Species used Locality female) number 1 2 3 4 5 6 7 8 Sources
Pelodiscus ene Oe ; 6463" Oguma
sinensis ® P (M/F) (1936,1937)
64/63 Suzuki
9 Seg: Die erm see eee cag oe
gonads Korea (M/F) (1950)
leukocyte oe Bickham et al.
eae unknown 1 (1/-) 66 M,M,SM,T ,T ,ST, M, M (1983)
pone Me NGhiag ? 66 M,M,SM,SM,SM, T, SM***,sm*** Rong and Lt
marrow (1984)
bene ical 5 (3/2) 66 M,M,SM,ST, ST,ST, SM, SM __ This study
marrow _oshima
pone Taiwan 4 (2/2) 66 M,M,SM,ST, ST,ST, SM, SM __ This study
marrow
Bone «7 naa 1(1/-) 66 M,M,SM,ST, ST,ST, SM, SM __ This study
marrow _ daitojima
Apalone leukocyte ok eK Bickham et al.
spiniferus cells Ge) CON Eee to erage Me (1983)
blood and Stock
gonads REEDS OG Tres eiee eae tee (1972)
leukocyte ok Bickham et al.
A. ferox aelle 1 (1/-) 66 M,M,SM,T ,T ,STI, M, M (1983)
; blood and Stock
A. muticus sons 7 (4/3) OOS eas me (1972)
Amyda carti- = , 66 a Gorman
lagineus Da ares ae (1973)
Dogania blood and Stock
subplana gonads ee OO ini ore area a (1972)
bone
1 (-/1) 66 M,M.SM,ST, ST, ST, T, T This study
marrow

2K 28

: M, metacentric; SM, submetacentric; ST, subtelocentric; T, telocenttric.
: Originally Bickham et al. (1983) described telocentric as acrocentric.
: Rong and Li (1984) did not explicitly describe the morphology of the seventh and eighth

macrochromosome pairs. However, based on the photograph of arranged chromosomes given in
that paper, we tentatively considered these pairs to be submetacentric.

of the gonadal

unorganized medulla (Fig. 1B).

were identified as females (Table 1).

body containing visible
oocytes, and the medullary region contained
By these cri-
teria, three specimens from Amamioshima, two
from Taiwan and the one from Minamidaitojima
were identified as males, and the remainder,
including the representative of D. subplana,

No sex chromosome heteromorphisms, either
in chromosome number or in morphology, were
evident in any metaphase cells examined at the
Giemsa level. We consider macrochromosomes
to be so large in size that centromeric posi-
tions can be unambiguously determined even
in spreads of mediocre quality. The five speci-
mens of P. sinensis from Amamioshima had a

SATO & OTA—KARYOTYPE OF SOFT-SHELLED TOURTLE 28

PA 4A AA ne nn
l 2 3 4 5 6 7 8
A SP at Ga 046 4% B24 4 CH HE et He

9-19

4h 60 O04 4m te GH OH HH HH OEM HOO ewe
20-33

80 AB GA wn an

I 2 3 4 5 6 7 8

RB ab hae Bem BO OH HH Be 69 4 BA om
9-19

Re PW BUS 8D OR Hs tM BH HD em 4% HE
20-33

| 2 3 4 D 6 7 8

¢C oe mah 2° 2 wh ade OH CM BH ee
= HH ——. 9-19

one wots ae tw @h me *& -“@ «f° 88 Fe ae ee

20-33

FIG. 2. Karyotypes of trionychine turtles obtained from the bone marrow cells. A, female Pe/odiscus sinen-
sis from Amamioshima (KUZ 48716; 2n=66). B, male P. sinensis from Taiwan (KUZ 48737; 2n=66). C,
female Dogania subplana (KUZ 47851; 2n=66). Scale-bar represents 10 um.

24

Current Herpetol. 20(1) 2001

karyotype consisting of 2n=66 chromosomes.
Of these, the largest eight pairs were macro-
chromosomes, of which the first and second
pairs were metacentric, the third submetacen-
tric, pairs 4—6 subtelocentric, and pairs 7 and 8
submetacentric. The remaining 25 pairs were
microchromosomes with centromeres most
likely being located terminally (Fig. 2A).
Karyotypes of the Minamidaitojima male, and
the four specimens from Taiwan, appeared to
be identical with the karyotype of the Ama-
mioshima sample described above (Fig. 2B).
The karyotype of the female D. subplana,
though also being composed of 2n=66 chro-
mosomes, differed from the P. sinensis karyo-
type in macrochromosome morphology: in the
D. subplana karyotype, pairs 7and 8 were
telocentric instead of submetacentric (Fig.
Ze)

DISCUSSION

All specimens examined in the present study,
including both male and female representatives
for populations from the main islands of Japan,
had 2n=66, an apparently homologous karyo-
type. Such a chromosomal arrangement, while
being consistent with those of P. sinensis of
unknown locality and from continental China
reported by Bickham et al. (1983) and Rong
and Li (1984), substantially differs from those
reported by Oguma (1936, 1937) and Suzuki
(1950) for samples from the main islands of
Japan and Korea, respectively (Table 1).
Considering the technical disadvantage of the
gonadal sectioning methods (Gorman, 1973;
Sharma, 1980: see above), it is highly likely
that such differences did not reflect actual
intraspecific variation, but were due to mis-
counting of chromosomes by the previous
authors.

It is generally considered that chromosome
morphology in the cryptodiran turtles is quite
invariable, with most confamilial and consub-
familial species possessing invariable karyo-
types (e.g., Bickham and Carr, 1983; Stock,
1972). With respect to the Trionychinae,
Bickham et al. (1983) reported that three spe-

cies of two genera examined by them (P. sin-
ensis, Apalone spiniferus, and A. ferox) had
almost identical karyotypes (Table 1). They
thus predicted an extremely low chromosomal
variability in the subfamily. Our results, how-
ever, partially negate this prediction, because
they show that: (1) the karyotypes of P. sinen-
sis and D. subplana differ from each other in
centromeric positions in macrochromosome
pairs 7 and 8 (Figs. 2A, B, and C); (2) the
karyotypes of P. sinensis and D. subplana
obtained in the present study differ from those
of other Trionychinae species (Apalone spin-
iferus, and A. ferox) reported by Bickham et
al. (1983) in morphology of macrochromo-
some pairs 4, 5, 7, and 8 (Table 1); and (3) the
karyotype of P. sinensis from the present study
differs from conspecific karyotypes reported
by Bickham et al. (1983) and Rong and Li
(1984) in centromeric position in macrochro-
mosome pairs 4, 5, 6, 7, and 8 (Table 1). Of
these, (1) and (2) indicate the presence of
karyotypic variation at an intergeneric level in
this subfamily, whereas (3) strongly suggests
the presence of an intraspecific chromosomal
variation in P. sinensis. Thus, we suspect that
in the subfamily Trionychinae, the morphol-
ogy of macrochromosomes is more variable
than was expected by Bickham et al. (1983),
although the diploid, macrochromosome, and
microchromosome numbers are highly conser-
vative. Further extensive surveys of karyo-
types in other trionychine species are needed
to verify this prediction.

Very few authors have studied chelonian
karyotypes by the bone marrow air-dry method
(Kamezaki, 1989). In the present study, how-
ever, we successfully obtained many hemato-
poietic metaphase cells by this method. Even
so, preliminary attempts at karyotyping using
the same method for adult P. sinensis (n=4,
CL larger than 150 mm) yielded no metaphase
cells at all (Ota, unpublished data). It is thus
probable that adult turtles tend to reduce
active cells in the bone marrow, and that in
turtles the bone marrow air-dry method is
most effective for juveniles.

SATO & OTA—KARYOTYPE OF SOFT-SHELLED TOURTLE 25

ACKNOWLEDGMENTS

We thank Mr. T. Kanmura and Dr. Y.
Yasukawa for provision of materials studied
here, Dr. E. Hirose for technical advice for
gonadal sections, Miss. S. Yamashiro for
helping with the laboratory work, and Prof.
E. Zhao, Dr. S.-L. Chen, and Prof. K. Adler
for literature. This research was partially
supported by a grant from the Association for
the Preservation of Aquatic Resources, Japan
(to H. Ota).

LITERATURE CITED

BICKHAM, J. W. AND J. L. CARR. 1983. Taxonomy
and phylogeny of the higher categories of cryptodi-
ran turtles based on a cladistic analysis of chromo-
somal data. Copeia 1983(4): 918-932.

BICKHAM, J. W., J. J. BULL, AND J. M. LEGLER. 1983.
Karyotypes and evolutionary relationships of tri-
onychoid turtles. Cytologia 48(2): 177-183.

GORMAN, G. C. 1973. The chromosomes of the Rept-
lila, a cytotaxonomic interpretation. p. 349-424. In:
A. B. Chiarelli and E. Capanna (eds.), Cytotaxon-
omy and Vertebrate Evolution. Academic Press,
New York.

GREEN, D. M. AND S. K. SESSIONS 1991. Nomencla-
ture for chromosomes. p. 431-432. Jn: D. M. Green
and S. K. Sessions (eds.), Amphibian Cytogenetics
and Evolution Academic Press, San Diego.

IVERSON, J. B. 1992. A Revised Checklist with Dis-
tribution Maps of the Turtles of the World. Pri-
vately printed, Richimond, Indiana. 363 p.

KAMEZAKI, N. 1989. Karyotype of the loggerhead
turtle, Caretta caretta, from Japan. Zool. Sci. 6(2):
421-422.

MEYLAN, P. A. 1987. The phylogenetic relationships
of soft-shelled turtles (family Trionychidae). Bull.
Am. Mus. Nat. Hist. 186(1): 1-101.

OGUMA, K. 1936. Sexual difference of chromosomes
in the soft-shelled turtle. Jpn. J. Genet. 12: 59-61.

OGUMA, K. 1937. Studies on sauropsid chromo-
somes. III. The chromosomes of the soft-shelled
turtle, Amyda japonica (Temminck and Schleg.),
as additional proof of female heterogamety in the
Reptilia. J. Genet. 34(2): 247-264.

RONG, S.-B. AND X.-W. LI. 1984. The karyotype of
Amyda sinense. Zool. Res. 5 (suppl. 3): 29-31. (in
Chinese with English summary)

SATO, H. AND H. OTA. 1999. False biogeographical
pattern derived from artificial animal transporta-
tions: A case of the soft-shelled turtle, Pelodiscus
sinensis, in the Ryukyu Archipelago, Japan. p.
317-334. In: H. Ota (eds.), Tropical Island Her-
petofuna: Origin, Current Diversity, and Conser-
vation. Elsevier, Amsterdam.

SATO, H., T. YOSHINO, AND H. OTA. 1997. Origin
and distribution of the Chinese soft-shelled turtle,
Pelodiscus sinensis (Reptilia: Tesudines), in
islands of Okinawa Prefecture, Japan. Biol. Mag.
Okinawa (35): 19-26. (In Japanese with English
abstract)

SHARMA, A. K. 1980. Chromosome Techniques:
Theory and Practice. Third Edition. Butter Worths,
London. 711 p.

STock, A. D. 1972. Karyological relationships in
turtles (Reptilia: Chelonia). Can. J. Genet. Cytol.
15(4): 859-868.

SUZUKI, K. 1950. Studies on the chromosomes of
Korean soft-shelled turtle (Amyda_ maackii,
Brandt) with special reference to the sex chromo-
somes. Jpn. J. Genet. 25(5/6): 222. (in Japanese)

YNTEMA, C. L. AND N. MROSOVSKY. 1980. Sexual
differentiation in hatchling loggerheads (Caretta
caretta) incubated at different controlled tempera-
tures. Herpetologica 36(1): 33-36.

‘Oi utente
> ne
ag oat ih

Current Herpetology 20(1): 27—31., June 2001
© 2001 by The Herpetological Society of Japan

Wider Head Shape in Larval Salamanders (Hynobius
retardatus) Induced by Conspecific Visual and Chemical Cues

YUKIHIRO KOHMATSU*

Center for Ecological Research, Kyoto University, Hiranocho, Kamitanakami, Otsu, Shiga,
520-2113 JAPAN

Abstract: The role of visual and chemical cues of conspecifics in the growth of
head width was experimentally tested in Hynobius retardatus larvae reared
under group and solitary conditions. In the former, the head width of the larvae
increased proportionally in the presence of visual, chemical, and visual plus
chemical cues, whereas such a morphological change was not apparent with
similar cue treatments in solitary larvae. It was concluded that both visual and
chemical cues from other conspecific individuals induced the wider head shape

in H. retardatus larvae.

Key words: Cannibalism; Hynobius retardatus; Salamander; Visual and chemical

cues

INTRODUCTION

Cannibalistic polyphenism has been reported
for the larvae of several salamander species
(see review in Crump, 1992). Such cannibalis-
tic morphs, which are flexibly induced after
typical morphs in having greater head size and
mouth width, and elongated vomerine teeth
(Powers, 1903; Rose and Armentrout, 1976;
Lannoo and Bachmann, 1984; Collins et al.,
1993). Since the prey size of larval salamanders
is generally gape-limited (Kusano et al., 1985;
Ohdachi, 1994), cannibalistic morphs can
achieve advantageous foraging of larger prey
items over typical morphs in the same popula-
tion (Reilly et al., 1992). However, larval sala-
manders are expected to flexibly develop the

' Corresponding author. Tel: +81—77—549-8200;
Fax: +81—77-549-8201.

E-mail address: kohmatsu@ecology.kyoto-u.ac.jp
(Y. Kohmatsu)

cannibalistic morph assessing the conspecific
density precisely to achieve an adaptive com-
promise, because such a morph is not always
advantageous in intra-specific competition when
they can not eat conspecifics in very low-density
conditions. Although the characteristic mouth-
part morphology of cannibalistic morphs is
considered to be less adapted to foraging on
small benthic invertebrates than that of typi-
cal morphs, present understanding of direct
environmental stimuli responsible for morpho-
logical changes in larval salamanders 1s still
insufficient.

Previous studies found that the development
of cannibalistic morphs is triggered by various
stimuli, e.g., diet (Walls et al., 1993), kinship
(Pfennig and Collins, 1993), and density of (Col-
lins and Cheek, 1983) and direct contact among
conspecific individuals (Hoffman and Pfennig,
1999). The flexible development of cannibal-
istic morphs was also reported for Hynobius
retardatus (Wakahara 1995). Moreover, larvae

28

Current Herpetol. 20(1) 2001

were found to enlarge the head width when
perceiving conspecific cues under high den-
sity conditions without actual cannibalism or
direct contact among larvae (Nishihara 1996a,
b). Cues inducing the enlargement of head
width were not analyzed in those studies. In
the present study, the role of visual and chemi-
cal cues in determining the development of
head width in H. retardatus larvae was exam-
ined experimentally.

MATERIALS AND METHODS

Six clutches (1 clutch=2 egg sacs) of H.
retardatus eggs (N=96) were removed from the
sacs and transferred to separate plastic contain-
ers (3 cm diameter, 2.5 cm height) containing
3 ml of well water. These were kept in a labo-
ratory at the Tomakomai Experimental Forest
of Hokkaido University (TOEF) under a 14-h

Density treatments

light: 10-h dark regime at room temperature
(17—20°C) until hatching. Larvae were assigned
to one of the following experimental treatments
within a day of hatching.

The experiments performed from 14 June to
14 July 1997, included (1) visual cues, (2) chem-
ical cues, and (3) visual plus chemical cues, plus
a control (4), so as to test the effects of the
cues from conspecific larvae on morphologi-
cal change of head shape (Fig. 1). Each larva
was kept separately in a clear plastic cup
(6 cm diameter, 5 cm height) held in a con-
tainer (18x12 cm area, 5 cm height), the cups
being painted (with insoluble black ink) but not
perforated (controls), painted and perforated
(chemical cues), neither painted nor perfo-
rated (visual cues), or perforated but not painted
(chemical plus visual cues). Perforations were
arranged in 10 rows of 15 holes (1 mm diame-
ter each). For each cue treatment, two larval

container

Boas:
<P
=
aN
Group cup Solitary
sega

Conspecifc cues treatments

Visual cue Chemical cue

Control

Visual +
Chemical cue

FIG. 1. Schematic representation of the experimental treatments for the effects of density and conspecific

cues.

KOHMATSU—HEAD MORPHOLOGY IN SALAMANDER 29

densities were set; viz. six (hereafter group
treatment) or a single individual (solitary treat-
ment). The group and solitary treatments cor-
responded, respectively, to 300 and 50
individuals m™, both being within the range
observed in natural habitats in Hokkaido (Y.
Kohmatsu, unpublished data). Each treatment
was replicated, using two and 12 containers
respectively, 48 larvae being tested in total for
each of the group and solitary treatments. In
the group treatment, to equalize sibship effects
among the cue treatments, each of the six cups
in the same container was stocked with a larva
from each of the six egg sacs initially collected.
To standardize experimental procedures, both
the cups and the containers were filled with
well water (4cm deep). In the laboratory, a
14-h light: 10-h dark regime and room tem-
perature of 17—20°C were maintained during
the experimental period. On day 7 of the
experiment, larval feeding with equal quantities
(10 mg wet mass) of frozen chironomid larvae
was initiated, and continued three times a week
thereafter. Each container was cleaned three
times a week and refilled with fresh well water.

To assess the effects of the treatments on the
relative head width, snout-vent length (SVL:
distance from anterior tip of snout to posterior
margin of vent) and head width (HW: maxi-
mum width across dorsal surface of head) were
measured to the nearest 0.01 mm on day 30 of
the experiment. This time period (30 days) was
considered sufficient for the development of
morphological changes (see Nishihara, 1996a;
Nishihara-Takahashi, 1999). Because the rela-
tive HW of larval H. retardatus varies within a
population (Wakahara, 1995; Nishihara, 1996a),
the residuals of a simple liner regression of
HW on SVL were examined so as to assess the
enlargement of HW. The HW of larvae was
significantly correlated with SVL (r’=0.509,
p<0.001 for log, transformed data) at the end
of the experiment.

For all analyses in the group treatments,
individuals in different containers (N=12)
were pooled for each cue treatment, because
the container effects were not significant in
the preliminary analyses of SVL (t-tests,

t=0.53, p=0.60, for log,) transformed data) and
HW (t-tests, t=1.05, p=0.30, for log, trans-
formed data). The effects of each conspecific
cue on the residual of HW were compared
between group and solitary treatments using t-
tests. Subsequently, the effects of each conspe-
cific cue on the residual of HW in both group
and solitary treatments were analyzed sepa-
rately using two-way ANOVAs with visual and
chemical cues as main factors. Log,, transfor-
mations were made for exact values in order to
standardize variances and improve normality, if
necessary, to satisfy the assumptions of the
ANOVA model. All statistical tests were
two-tailed and used an alpha value of 0.05 for
statistical significance. All analyses were con-
ducted using the statistical package StatView
(ver. J 5.0, SAS Institute Inc., Carolina, USA).

RESULTS

The effect of each conspecific cue on the
residual of HW differed between the solitary
and group treatments (chemical cues: t,.=4.97,
p<0.01; visual cues: t,.=3.00, p<0.01; chemical
plus visual cues t,.=4.45, p<0.01) but not
controls (t,.=1.15, p=0.26; Fig. 2). In the
group treatment, two-way ANOVA revealed
that chemical cues (F, 44=18.76, p<0.01) and
chemicalxvisual cues interaction (F3 44=10.79,
p<0.01) effects were significant, but the effect
of visual cues was not significant (Fj 44=2.51,
p<0.12). In contrast, the effects of all the
cues were not significant in the solitary treat-
ment (F34,=0.41-1.70, p>0.20 by two-way
ANOVA).

DISCUSSION

The development of relative HW in H. retar-
datus larvae differed between the group and
solitary treatments. The presence of any one of
the conspecife cues (visual, chemical, or visual
plus chemical) resulted in a greater HW pro-
portion than in the solitary condition (Fig. 2).
Certainly a larger HW was induced in the larvae
in the group by visual cues, although the effect
of visual cues was relatively weak. Moreover,

30

Current Herpetol. 20(1) 2001

Residuals of head width

Conspecific cues

FIG. 2. Effects of conspecific larval cues and density on the residuals of head width (the deviations from
expected HW for larvae of given SVL). Data expressed as meanstSE (N=12). Open and closed circles denote
minimum and maximum values, respectively.

there was no additional effect of both chemical
and visual cues on development of relative HW
(Fig. 2). I assumed that these results were
caused by individual variation in the degree of
relative HW development among the larvae. In
fact, there was one larva with a much larger
relative HW in the chemical cue treatment (Fig.
2). In contrast, the enlargement of HW of the
larvae in visual cue treatment was considered
to be small. For this reason, the effect of visual
cues was not detected when the effect of these

cues in the group was examined by two-way
ANOVA.

Even under solitary conditions, the concen-
tration of larval-originated chemicals in both the
visual and control treatments could have been
higher than in the chemical and visual plus
chemical treatments. In addition, dissolved oxy-
gen concentration in the former may have been
lower than in the latter owing to the oxygen
consumption occurring in a closed cup. There-
fore, it can be concluded that the cues from

KOHMATSU—HEAD MORPHOLOGY IN SALAMANDER 31

other conspecific individuals induced the devel-
opment of large head shape in the H. retardatus
larvae. In this experiment, the enlargement of
HW was induced without any direct contact
between individuals, and this is consistent with
the findings of Nishihara (1996a, b). The
present experiment showed that both conspe-
cific visual and chemical cues induced the mor-
phological change, and larvae could detect the
presence of conspecifics on the basis of either
cue. However, the relative reliability of such
cues may be changed by several environmen-
tal factors. For instance, the concentration of
chemical cues very likely varies with fluctua-
tions in water volume and/or water flow in
local habitats, as well as with larval density. In
contrast, although visual cues are less subject to
the above environmental factors, they would
differ according to water turbidity and density
of aquatic vegetation. This is because these
environmental factors can vary greatly among
the temporary ponds used for breeding by H.
retardatus. The variability and fluctuation of
such environmental conditions in temporary
ponds would provide background for the evolu-
tion of the ability to perceive multiple cues in
larval H. retardatus.

ACKNOWLEDGMENTS

I am grateful to S. Nakano for critical com-
ments on an early draft of the manuscript, and
express my sincere thanks to Y. Yasui, H. Mit-
suhashi, H. Miyasaka, and Y. Kawaguchi for
their logistic support during the study. This
research was partly supported by the Japan
Ministry of Education, Science, Sport, and Cul-
ture (Grants OONP1501 and 11440224 to S.
Nakano).

LITERATURE CITED

COUMINS, J; BAND J. E. CHEEK: 1983.. Effect of
food and density on development of typical and
cannibalistic salamander larvae in Ambystoma
tigrinum nebulosum. Amer. Zool. 23: 77-84.

COLLINS, J. P., K. E. ZERBA, AND M. J. SREDL. 1993.
Shaping intraspcific variation: development, ecol-

ogy and the evolution of morphology and life
history variation in tiger salamanders. Genetica
89: 167-183.

CRUMP, M. L. 1992. Cannibalism in amphibians.
p: 256-276. In? M.-A’ Elger and" BJ. respi
(eds.), Cannibalism: Ecology and Evolution
among Diverse Taxa. Oxford Univ. Press,
Oxford.

HOFFMAN, E. A. AND D. W. PFENNIG. 1999. Proxi-
mate causes of cannibalistic polyphenism in lar-
val tiger salamanders. Ecology 80: 1076-1080.

KUSANO, T., H. KUSANO, AND K. MIYASHITA.
1985. Size-related cannibalism among larval
Hynobius nebulosus. Copeia 1985: 472-476.

LANNOO, M. J. AND M. D. BACHMANN. 1984.
Aspects of cannibalism morphs in a population
of Ambystoma t. tigrinum larvae. Am. Midl. Nat.
112: 103-109.

NISHIHARA, A. 1996a. Effects of density on growth
of head size in larvae of the salamander Hyno-
bius retardatus. Copeia 1996: 478-483.

NISHIHARA, A. 1996b. High density induces a
large head in larval Hynobius retardatus from a
low density population. Jpn. J. Herpetol. 16:
134-136.

NISHIHARA-TAKAHASHI, A. 1999. Faster growth of
head size of pre-feeding larvae in a cannibalistic
population of the salamander Hynobius retarda-
tus. Zool. Sci. 16: 303-307.

OHDACHI, S. 1994. Growth, metamorphosis and
gape-limited cannibalism and predation on tad-
poles in larvae of salamanders Hynobius retarda-
tus. Zool. Sci. 11: 127-131.

PFENNIG, D. W. AND J. P. COLLINS. 1993. Kinship
affects mophologenesis in cannibalistic sala-
manders. Nature 362: 836-838.

Powers, J. H. 1907. Morphological variation and
its causes in Ambystoma tigrinum. Stud. Univ.
Nebraska 7: 197-274.

REILLY, S. M., G. V. LAUDER, AND J. P. COLLINS.
1992. Performance consequence of a trophic
polymorphism: feeding behavior in typical and
cannibal phenotypes of Ambystoma tigrinum.
Copeia 1992: 672-679.

ROSE, F. L. AND D. ARMENTROUT. 1976. Adaptive
strategies of Ambystoma tigrinum inhabiting the
Llano Estacado of West Texas. J. Anim. Ecol.
45: 713-729.

WAKAHARA, M. 1995. Cannibalism and the result-
ing dimorphism in larvae of a salamander Hyno-
bius retardatus, inhabited in Hokkaido, Japan.
Zool. Sci. 12: 467-473.

WALLS, S. C., S. S. BELANGER, AND A. R.
BLAUSTEIN. 1993. Morphological variation in a
larval salamander: dietary induction of plasticity
in head shape. Oecologia 96: 162-168.

Current Herpetology 20(1): 33—37., June 2001
© 2001 by The Herpetological Society of Japan

Absence of Lines of Arrested Growth in Overwintered
Tadpoles of the American Bullfrog, Rana catesbeiana
(Amphibia, Anura)

WICHASE KHONSUE AND MASAFUMI MATSUI’

Graduate School of Human and Environmental Studies, Kyoto University, Yoshida
Nihonmatsu-cho, Sakyo-ku, Kyoto 606—8501 Japan

Abstract: In order to ascertain formation of lines of arrested growth (LAGs) in
anuran larvae, cross-sections of long bones in overwintered tadpoles and juve-
niles of Rana catesbeiana were observed. However, no LAGs were evident in
these individuals, although we could observe clear LAGs in adults treated in the
same way. The result contrasts to those reported for urodelan larvae. We dis-
cuss the causes of this phenomenon on the assumption that the absence of LAGs
in tadpoles is common to other anurans as well.

Key words: Overwintered tadpoles; Rana catesbeiana; LAG formation; Skeleto-

chronology

INTRODUCTION

In a study of Hynobius kimurae, Misawa and
Matsui (1999) found lines of arrested growth
(LAGs) formed in limb bones of overwintered
larvae. However, no studies have reported the
occurrence of LAGs in overwintered anuran
tadpoles, although formation of LAGs in bones
of many postmetamorphic anuran species has
been well documented (e.g., Castanet and Smi-
rina, 1990; Esteban et al., 1996).

Rana catesbeiana normally hibernates in lar-
val form for at least one winter both in its
original habitat (eastern USA: e.g., Viparina
and Just, 1975) and in introduced regions
(e.g., Japan: Maeda and Matsui, 1999). Meta-

; Corresponding author. Tel: +81—75—753-6846;
Fax: +81—75—753-289 1

E-mail address: fumi@zoo.zool.kyoto-u.ac.jp
(M. Matsui)

morphs of this species also often hibernate in
the water, under mud at the bottom of ponds
(Maeda and Matsui, 1999). Thus, this species
offers a good opportunity to examine the condi-
tion of LAGs in the larval stage as compared to
that after metamorphosis in anurans.

MATERIALS AND METHODS

Forty-three overwintered tadpoles (stage 39—
46 of Gosner, 1960) and three metamorphs were
collected between 25 and 30 July 1998 from
a permanent pond called Bungaike, Kizu-cho,
Kyoto Prefecture. For comparisons, we also
collected a metamorphosed juvenile on 10 Sep-
tember 1998 and a gravid female on 23 June
1998 from around the same pond. Whole
bodies in the case of tadpoles and small juve-
niles or digital bones in larger individuals were
first fixed in 10% formalin, and preserved in
70% ethanol until study. Fifty-nine toe and

34

Current Herpetol. 20(1) 2001

Fic. 1. Phalangeal cross section of overwintered Rana catesbeiana tadpoles; (A) stage 39 (Gosner, 1960)
with 39.8 mm body length (BL) collected on 25 July 1998, (B) stage 40 with 43.7 mm BL collected on 25 July
1998, (C) stage 41 with 38.4 mm BL collected on 25 July 1998, (D) stage 42 with 42.3 mm BL collected on 25
July 1998, (E) stage 43 with 41.4 mm BL collected on 28 July 1998, (F) stage 44 with 39.7 mm BL collected
on 30 July 1998, (G) stage 45 with 41.0 mm BL collected on 28 July 1998, (H) stage 46 with 43.2 mm BL col-
lected on 30 July 1998. All sections show no lines of arrested growth in bones. MC=marrow cavity. Scale

bar=50 um.

femoral bones and three finger bones were
removed from the tadpoles and metamorphs,
respectively, and were decalcified in 5% nitric
acid for 120 min. They were washed in run-
ning water for one night before and after this
process. Then the bones were sectioned on a

freezing microtome and stained with hematox-
ylin (Mayer’s acid hemalum) for 30 min. We
selected sections which included the diaphysis
region, and mounted them in glycerin for micro-
scopic examination.

KHONSUE & MATSUI—ABSENCE OF LAGS IN TADPOLES 35

RESULTS

The overwintered tadpoles varied in average
body length (mean+2SE; in mm) from 39.9
(N=1) at stage 39 to 43.340.96 (N=11) at stage
46 (Table 1). Completely metamorphosed fro-
glets varied in snout-vent length (SVL) from
46.3 to 57.3 mm (N=3). Cross sections of dig-
ital bones in all individuals of overwintered
tadpoles (Figs. 1A—H) and completely meta-
morphosed froglets exhibited incomplete ossifi-
cation with many holes. There were no LAGs
in any of these sections. Although a metamor-
phosed line (ML: Hemelaar, 1985) was found
in an individual that was collected shortly after
emergence from hibernation, we did not find
any lines in any other individuals. A juvenile
of 86.1 mm SVL collected in early September
also lacked LAGs (Fig. 2A) and showed a very
rapid remodeling of the finger bone. Minimum

diameter of the resorption line (RL, irregular,
inner surface of the finger bone: Hemelaar and
van Gelder, 1980) in this juvenile was always

TABLE 1. Developmental stage (Gosner, 1960),
number of individuals (N), and body length
(meant2SE in mm) of overwintered Rana
catesbeiana tadpoles.

Stage N Body length
39 1 39.8

40 5 41.3+2.89
4] 2 40.6

42 ) 41.9+3.95
43 6 41.01.77
a+ 6 41.7+1.84
45 d 42.1+1.28
46 i) 43 .3+40.96

Fic. 2. Phalangeal cross section of Rana catesbeiana; (A) juvenile with 86.1 mm SVL collected on 10 Sep-
tember 1998 showing no lines of arrested growth (LAGs), but the resorption line (RL), (B) adult female with
143.8 mm SVL collected on 23 June 1998 showing two LAGs besides RL. MC=marrow cavity. Scale

bar=200 um.

36

Current Herpetol. 20(1) 2001

larger than outer diameter of fingers in froglets
which had just completed metamorphosis. On
the other hand, two clear LAGs were observed
in the adult frog (Fig. 2B).

DISCUSSION

From the time of capture and body size, it is
obvious that the tadpoles and froglets we exam-
ined had gone through at least one overwinter-
ing. In some anuran species hitherto studied
(e.g., Rana tagoi and R. sakuraii: Kusano et
al., 1995a, b; R. nigromaculata: Khonsue et al.,
unpublished data), formation of LAGs occurs
fairly late after the emergence of frogs from
hibernation. However, the dates of collection
of our samples (25-30 July and 10 September
1998) seem to be late enough for the forma-
tion of a new ring, and the larvae and juve-
niles examined must have had a chance to
increase bony material after overwintering,
because, unlike adults, they did not need to
expend energy for special activities such as
reproduction.

Previous skeletochronological studies on
amphibians usually resulted in detection of
clear LAGs in bones of both juveniles and
adults (e.g. Hemelaar, 1985; see review in Cas-
tanet and Smirina, 1990). Similarly, formation
of LAGs has been confirmed in overwintered
larvae in at least one species of urodela (Mis-
awa and Matsui, 1999). Thus, the absence of
LAGs in overwintered tadpoles of R. cates-
beiana is surprising. This discovery, though
whether specific to R. catesbeiana or common
to other anuran species is not yet determined,
suggests a different process of LAG forma-
tion between the orders Urodela and Anura.
Although only data for adults are available,
Typhlonectes natans, a member of the third
amphibian order, Gymnophiona, also exhibits
LAGs (Measey and Wilkinson, 1998). Spe-
cies of this genus spend a totally aquatic life
even after metamorphosis (Taylor, 1968), and
prove that life in the water by itself does not
prevent the formation of LAGs. Why are there
no LAGs in bones of overwintered R. catesbe-
iana tadpoles or juveniles, unlike in urodeles?

One possible reason may be related to the
physiological differences between these two
orders. In urodelan larvae, as is clear from the
rapid development of forelimbs long before
metamorphosis, limbs are more important for
locomotion than in anuran larvae. Hence cal-
cium deposition, which is essential in the for-
mation of LAGs in limb bones, would occur
much earlier in urodelan larvae than in anuran
larvae.

In Rana saharica from a desert of Morocco,
Esteban et al. (1999) found that some froglets
lacked LAGs, unlike adults. They considered
that the froglets may have hibernated before
the start of the ossification process which was
vital to LAG formation.

The pond where our tadpoles were collected
is filled with deep water (deepest point >3 m)
all year round, and the temperatures at the bot-
tom of this pond, where the tadpoles must
have stayed throughout hibernation, might be
more constant than in the surface stratum or
on the surrounding banks. Such a stability of
the ambient temperature might have a role in
blocking the formation of LAGs.

However, metamorphs of R. catesbeiana are
also reported to usually hibernate in the water
(Maeda and Matsui, 1999), and we assume
this might have been the case with our samples.
They still showed the formation of LAGs.
Therefore, as suggested by Esteban et al.
(1999), the absence of LAGs in overwintered
tadpoles and juveniles of R. catesbeiana might
be affected by age-dependent timing of the
strat of hibernation and ossification. We need
further ecological. ethological, and histologi-
cal studies to solve this problem.

Finally, what has become clear from this
study is that the number of LAGs found in
older juveniles and adults of R. catesbeiana
does not represent their actual age from the
time of egg laying but their age after meta-
morphosis.

ACKNOWLEDGEMENTS

We thank T. Hirai for instruction of the sam-
pling site. H. Ota provided valuable comments

KHONSUE & MATSUI—ABSENCE OF LAGS IN TADPOLES 37)

on the manuscript and R. C. Goris kindly cor-
rected verbal errors.

LITERATURE CITED

CASTANET, J. AND E. SMIRINA. 1990. Introduction
to the skeletochronological method in amphibians
and reptiles. Annal. Sci. Nat. Zool. 11: 191-196.

ESTEBAN, M., M. GARCIA-PARIS, AND J. CASTANET.
1996. Use of bone histology in estimating the age
of frogs (Rana perezi) from a warm temperate
climate area. Can. J. Zool. 74: 1914-1921.

ESTEBAN, M., M. GARCIA-PARIS, D. BUCKLEY, AND
J. CASTANET. 1999. Bone growth and age in
Rana saharica, a water frog living in a desert
environment. Annal. Zool. Frennici 36: 53-62.

GOSNER, K. L. 1960. A simplified table for staging
anuran embryos and larvae with notes on identifi-
cation. Herpetologica 16: 183-190.

HEMELAAR, A. 1985. An improved method to esti-
mate the number of year rings resorbed in phalan-
ges of Bufo bufo (L.) and its application to
populations from different latitudes and altitudes.
Amphibia-Reptilia 6(4): 323-341.

HEMELAAR, A. S. M. AND J. J. VAN GELDER. 1980.
Annual growth rings in phalanges of Bufo bufo
(Anura, Amphibia) from the Netherlands and

their use for age determination. Neth. J. Zool. 30:
129-135.

KUSANO, T., K. FUKUYAMA, AND N. MIYASHITA.
1995a. Age determination of the stream frog,
Rana sakuraii, by skeletochronology. J. Herpetol.
29(4): 625-628.

KUSANO, T., K. FUKUYAMA, AND N. MIYASHITA.
1995b. Body size and age determination by skel-
etochronology of the brown frog, Rana tagoi
tagoi, in southwestern Kanto. Jpn. J. Herpetol.
16(2): 29-34.

MAEDA, N. AND M. MATSUI. 1999. Frogs and Toads
of Japan, Rev. Ed. Bun-ichi Sogo Shuppan,
Tokyo. pp. 100-107.

MEASEY, G. J. AND M. WILKINSON. 1998. Lines of
arrested growth in the caecilian, Typhlonectes
natans (Amphibia: Gymnophiona). Amphibia-
Reptilia 19(1): 91-95.

MISAWA, Y. AND M. MATSUI. 1999. Age determina-
tion by skeletochronology of the Japanese sala-
mander Hynobius kimurae (Amphibia, Urodela).
Zool. Sci. 16(5): 845-851.

TAYLOR, E. H. 1968. The Caecilians of the World.
Univ. Kansas Press, Lawrence, Kansas. v—
Vilit+848 p.

VIPARINA, S. AND J. J. JUST. 1975. The life period,
growth and differentiation of Rana catesbeiana
larvae occurring in nature. Copeia 1975(1):
103-109.

Current Herpetology 20(1): 39-49., June 2001

© 2001 by

The Herpetological Society of Japan

Preliminary Observations on Chemical Preference,

Antipredator Responses, and Prey-Handling Behavior of
Juvenile Letoheterodon madagascariensis (Colubridae)

AKIRA MORI AND Koyl TANAKA

Department of Zoology, Graduate School of Science, Kyoto University, Sakyo, Kyoto, 606—

8502 JAPAN

Abstract: Innate responses to prey chemicals, antipredator responses, and
prey-handling behavior of a Madagascan colubrid snake, Leioheterodon mada-
gascariensis, were experimentally examined. In a chemical test, ingestively
naive hatchlings flicked their tongues frequently to the chemicals prepared
from the animal taxa included in their natural diets, suggesting the presence of
innate chemical prey preference. In a second test, three different types of stim-
uli were presented to elicit antipredator responses of the hatchlings. In a non-
moving stimulus session, only a single snake struck and exhibited characteristic
displays such as body flattening, neck flattening, head elevation, and jerk. Ina
moving stimulus session, either no specific responses or a simple flight response
was exhibited. In a tactile stimulus session, the above characteristic displays
were frequently exhibited. Among these, lateral neck tilting posture accompa-
nying head elevation and neck flattening was unique to L. madagascariensis. In
a third test, five types of prey animals were offered to juveniles to examine the
effects of prey size and type on prey-handling behavior, but no such effects were
detected. Direction of ingestion seemed to depend on initial bite position. Con-
stricting behavior was observed only in a single trial. All but one prey were
swallowed alive. This inflexibility of prey-handling methods in the juvenile
snakes may reflect the characteristics of generalist feeders, which require onto-
genetic experiences to handle prey efficiently. Although L. madagascariensis is
considered to have well-developed Duvernoy’s glands with enlarged, posterior
maxillary teeth, the gland secretion did not seem to cause rapid death of prey.

Key words: Madagascar; Leioheterodon madagascariensis; Chemical preference;
Antipredator behavior; Prey-handling behavior

INTRODUCTION

Madagascar is a biologically unique island

acer eesonding author Tel: +81-75-753-4075: characterized by an abundance of endemic
Fax: +81—75-753-4113 taxa. Currently, 17 genera of colubrid snakes

E-mail
(A. Mori)

address: gappa@zoo.zool.kyoto-u.ac.jp are known from Madagascar, all of which are
endemic to this island and adjacent islets

40

(Randriamahazo, 1999). Because of this pecu-
liarity, most previous studies on these snakes
have focused on their taxonomy and biogeogra-
phy, and little is yet known of behavioral and
ecological aspects of the Madagascan snakes.

Leioheterodon madagascariensis is a rela-
tively large (SVL>100 cm), common colubrid
distributed throughout Madagascar (Glaw and
Vences, 1994). Available information suggests
that this species is a generalist predator feed-
ing on a variety of vertebrates (Conant, 1938;
Campbell and Murphy, 1977; Groves and
Groves, 1978; Preston-Mafham, 1991; von
Dathe and Dedekind, 1996; Mori and Ran-
driamahazo, in press). It is also reported that L.
madagascariensis is quite aggressive (Bry-
goo, 1982; von Dathe and Dedekind, 1996)
and is potentially capable of envenomation
through enlarged, posterior, maxillary teeth
equipped with distinct Duvernoy’s glands
(Domergue and Richaud, 1971; Domergue,
1989; Mori, 2000). However, no systematic
behavioral observations have been conducted
for this species.

We experimentally examined three behav-
ioral features of the juvenile L. madagascar-
iensis: innate responses to prey chemicals,
antipredator responses, and _ prey-handling
behavior. These behavioral features have been
well studied in many other colubrid species, but
no comparable studies have been made for
Malagasy snakes. Innate chemical preference
to various animal taxa was examined using
ingestively naive hatchlings. Characteristics
of antipredator responses and prey-handling
behavior are described, and the effects of stimu-
lus and prey types on these behaviors were
investigated. Basic information on reproduc-
tive traits is also provided.

ANIMALS AND GENERAL METHODS

A female L. madagascariensis (SVL=1293
mm, body mass=975 g), collected in the dry for-
est at the Jardin Botanique A of Ampijoroa,
northwestern Madagascar on 21 October 1999,
laid ten eggs in a snake bag on 1 November.
The eggs were removed from the bag and mea-

Current Herpetol. 20(1) 2001

TABLE |. Measurements of eggs of Leioheterodon
madagascariensis. The shells of the lower five eggs
were soft and yellowish. These eggs were preserved
in formalin within five days after oviposition.
Maternal body mass after oviposition was 685 g.

Length*width (mm) Mass (g)
54.2x29.4 26.5
54.2x29.5 26.9
59.7x27.9 26.1
53.2x30.0 DS
53.9x28.7 25.0
48.8x27.6 22.0
49.8x26.5 21.4
51.0x29.7 20.6
45.1x33.3 i)
45.0x29.2 17.6

sured immediately (Table 1). Except for
three soft yellowish eggs, which were appar-
ently abnormal and thus were preserved in for-
malin immediately, the eggs were kept in a
plastic box filled with wet sand at natural
temperatures in Ampijoroa. On 4 November
two more eggs turned yellowish and started
to smell, and were preserved in formalin. The
remaining five eggs were transported to our
laboratory in Japan on 13 November, where
they were kept in an environmental chamber at
approximately 30°C.

The eggs hatched from 26 to 29 January
2000. The hatchlings were measured (Table
2) and individually housed in a plastic cage
(295x185x165 mm) provided with a paper
floor covering, shelter, and water dish at room
temperature between 25 and 30°C.

TABLE 2. Measurements of hatchlings of
Leioheterodon madagascariensis. SVL: snout-vent
length, TL: tail length, BM: body mass.

SVL (mm) TL (mm) BM (g) Sex
527, 70 19.9 m
334 70 19.2 m
B52 68 18.7 m
525 69 19.7 f
320 64 18.9 f

MORI & TANAKA—BEHAVIOR OF SNAKES

4]

Chemical and antipredator response tests were
conducted on 18 and 20 February, respec-
tively. No foods were offered before these
experiments. From 22 February, small mice
(Mus musculus, 1.5—5.6 g) were provided basi-
cally twice a week. The prey-handling test was
conducted from 20 July 2000. The snakes were
fasted for a week before the prey-handling test.
Because of the small sample size, no statistical
tests were conducted.

CHMICAL TEST

It has been reported that in many colubrids
ingestively naive snakes respond to chemical
cues from species-typical prey with increased
tongue flicking, indicating the presence of
innate chemical prey preference (see Ford and
Burghardt, 1993 for review). The first test
was conducted to investigate responses of
newly hatched, ingestively naive L. madagas-
cariensis to chemicals from various groups of
potential prey and to examine the occurrence of
innate chemical prey preference in this species.

Methods

Preparation and presentation of chemical
stimuli followed a well-established procedure
employed for testing chemical discrimination by
the vomeronasal organ in Squamata (Burghardt,
1970; Cooper, 1998). Cotton swabs bearing
the chemical stimuli were presented to the
snakes. The animals used for chemical stim-
uli represented a variety of taxa. They
included earthworms (Oposthopora), slugs
(Stylommatophora), fish (Cyprinidae, Caras-
sius auratus; Adrianichthyidae, Oryzias lati-
pes), frogs (Rhacophoridae, Rhacophorus vy.
viridis; Ranidae, Rana narina), newts (Sala-
mandridae, Cynops ensicauda), lizards (Oplu-
ridae, Oplurus c. cuvieri; Scincidae, Eumeces
elegans), snakes (Colubridae, Elaphe quadriv-
irgata, Rhabdophis tigrinus), birds (Ploceidae,
Lonchura striata), and mammals (Muridae,
baby and adult Mus musculus). All of these
animals except for O. c. cuvieri and M. muscu-
lus are not sympatric with L. madagascarien-
sis. Although it is ideal to use sympatric species

to examine the chemical responses (see Coo-
per et al., 2000), unavailability of appropriate
animals obliged us to use allopatric species.
Nonetheless, our aim was to examine the
responses of the snake to a given major taxon,
not to a specific animal, and we believe that the
above treatment did not substantially affect
the results at that level. Distilled water and
cologne (Fresh Floral, Mandam Co. Ltd,
diluted to ca. 33%) were used as controls for
the experimental procedure and for detectable
but biologically irrelevant odors, respectively.

Before the experiment, the cages of the
snakes were moved onto the testing table, and
water bowls, shelters, and paper floor cover-
ings were removed. After more than 10 min,
the tip of a cotton swab, which was either rolled
over the external surface of the animals or
dipped into the control fluids just before each
test, was presented 1—2 cm in front of the snout
of the snake. The number of tongue flicks
directed to the swab was counted for 60 sec
after the first tongue flick was observed. If no
tongue flicks were made 30 sec after the pre-
sentation of the swab, the tip of the swab was
gently touched to the snout of the snakes. If the
snake did not flick its tongue for 60 sec, the
trial was terminated. A new stimulus was used
for each snake.

The order of the presentation of the stimuli
was randomized and counterbalanced. To
make the blind test, the second author pre-
sented swabs to each snake without being
informed of the source of the chemicals on
each swab.

Results

There was considerable individual variation
in the responses to the chemicals: one snake
flicked its tongue at least once for all chemi-
cal stimuli, whereas another snake flicked its
tongue to only three stimuli (two frog species
and Oplurus chemicals). However, the mean
numbers of tongue flicks showed a variation
among the stimuli (Fig. 1). Snakes flicked
their tongues frequently to the chemicals of
frogs and reptiles and moderately to those of
birds and mammals. Only a few tongue flicks

42

Current Herpetol. 20(1) 2001

Number of tongue flicks

water

cologne 3
invertebrate
fish

__

oO)
O
=

ne
lizard f
snake

Chemical cues

FIG. 1. Mean number of tongue flicks made by hatchling Leioheterodon madagascariensis to cotton swabs

bearing various chemical cues. Bars indicate 1SE.

were observed to the chemicals of cologne,
invertebrates, fish, and newts. The mean of
tongue flicks to the distilled water was moder-
ately high (8.8), but this was apparently due to
a high response of one individual: if this indi-
vidual was excluded from the calculation, the
mean tongue flick to the distilled water dropped
to 3.5. No snakes attacked the stimuli.

Discussion

Available information on natural diets of L.
madagascariensis suggests that this species is a
generalist feeder, eating frogs, lizards, birds, and
mammals (Preston-Mafham, 1991; Mori and
Randriamahazo, in press). Chemicals from
these animals elicited moderate to high tongue
flick responses. It is suggested, therefore, that
naive hatchlings can recognize, by chemical

means, the food items utilized by this species in
nature. In several species of snakes, newborns
are known to respond to prey chemicals not
only with increased tongue flicking but also by
open-mouthed attack on the swab (Ford and
Burghardt, 1993). The absence of attack in L.
madagascariensis suggests that other stimul1,
such as visual ones, may be indispensable to
elicit prey attack in this species (Ford and
Burghardt, 1993).

ANTIPREDATOR RESPONSE TEST

The aim of this experiment was to examine
antipredator responses of neonate L. madagas-
cariensis to various threatening stimuli and
characterize the antipredator behaviors of this
species.

MORI & TANAKA—BEHAVIOR OF SNAKES

43

Methods

To elicit antipredator responses of the snakes,
we used standardized methods developed for
assessing levels of antipredator reactions of
snakes (for non-tactile stimulus, Herzog and
Burghardt [1986] and Herzog et al. [1989]; for
tactile stimulus, Mori et al. [1996] and Mori
and Burghardt [2000]). A snake was gently
removed from its home cage and introduced
into an arena (442930 cm) at an ambient tem-
perature of 24—25°C. After leaving the snake
undisturbed for five minutes, the experimenter
slowly brought a forefinger to within | to 2 cm
of the snake’s snout and held it stationary for
60 sec. This is referred to as a nonmoving
stimulus session. If the snake crawled away
during the test, the experimenter followed it,
keeping the extended finger in front of the
snake. Then the snake was left undisturbed for
60 sec. The moving stimulus session began
when the experimenter again extended the fore-
finger to within 1 to 2 cm of the snake’s snout.
This time he moved the finger back and forth at
the rate of approximately three to four oscilla-
tions per second throughout the 60 sec period.
As with the nonmoving stimulus session, the
experimenter kept the finger in front of the
snake. The snake was then given another 60 sec
undisturbed period. Following this period, the
tactile stimulus session, which lasted 60 sec,
began. During this session the snake’s body
(excluding head and tail) was gently pinned

every three sec for a total of 20 times, with a
long metal snake hook. On the tip of the
hook, a 50x15 mm polyproplylene plate was
attached so that the snake was pinned down
by the plate.

In the first two sessions, we recorded the
occurrences of “strike”, “bite”, and “flight”,
which are the main behavioral variables previ-
ously used in comparable studies. When the
“strike” occurred, its frequency was recorded as
well. Other characteristic behaviors observed,
such as “body flattening” and “neck flattening”
were also recorded. In the tactile stimulus ses-
sion, behavior of the snake in response to each
stimulus was observed and recorded. All ses-
sions were videotaped, and videotape analysis
was done to record the above variables.

Results

Only a single snake struck the nonmoving
stimulus (twice) and exhibited characteristic dis-
plays such as body flattening, neck flattening,
head elevation, and jerk (see below). Other
snakes in the nonmoving stimulus session either
showed no specific responses or simply fled
(two snakes each). Toward the moving stimu-
lus, two snakes exhibited no specific responses,
and the other three showed a simple flight
response.

In response to the tactile stimulus, one
snake simply fled in response to all 20 stimuli.
The other four snakes exhibited dorsoventral

a

Fic. 2. Hatchling Leioheterodon madagascariensis exhibiting characteristic displays against artificial threat-
ening stimuli. (a) Neck flattening accompanying slight head elevation and body flattening. (b) Neck flattening
with prominent elevation and tilting of the anterior body so that the dorsal surface of the neck is directed

toward the source of the stimulus.

44

Current Herpetol. 20(1) 2001

flattening of the neck (neck flattening) and/or
flattening of the body posterior to the neck (body
flattening; Fig. 2a), occasionally accompanied
by lifting the head and neck region high above
the substrate (head elevation). During head ele-
vation, the snake often tilted the neck so that its
dorsal surface was directed to the source of stim-
ulus (Fig. 2b). From this posture, the snake fre-
quently showed strikes with its mouth open in
response to the stimulus (strike), but actual bit-
ing rarely occurred. In some cases, the snake
irregularly formed circular or S-shaped loops
with its neck and body flattened, and wriggled
violently and intermittently in response to the
physical contact of the stimulus (jerk). Average
frequencies of these responses for the four
snakes were 18.25 (neck flattening), 16.75 (body
flattening), 9.0 (head elevation), 10.5 (strike),
3.75 (jerk), and 0.75 (flee).

Discussion

Body flattening and head elevation accompa-
nying neck flattening are both common anti-
predator displays in snakes (Greene, 1988).
Neck tilting during head elevation seems to be a
characteristic display of L. madagascariensis.
Antipredator responses similar to jerk of L.
madagascariensis are known in other colubrids
(e.g., Rhabdophis tigrinus, Mori et al., 1996) and
members of other ophidian families such as the
Elapidae (e.g., Maticora intestinalis, Mori and
Hikida, 1991).

A suite of antipredator displays (neck flatten-
ing, head elevation, body flattening, and strike)
were apparently more easily elicited by physical
contact than by non-contact threatening stimuli.
Similar results were observed in an Asian colu-
brid, Rhabdophis tigrinus (Mori et al., 1996).
However, in the present test possible contribu-
tion of presentation order to the results can not
be precluded because the tactile stimulus session
was always conducted after non-tactile stimulus
Sessions.

PREY-HANDLING TEST

It has been demonstrated that as a conse-
quence of behavioral adaptation, snakes change

prey-handling behavior according to prey size
and type: larger prey animals tend to be con-
stricted to death, and then swallowed head first
more frequently (Loop and Bailey, 1972; Mori,
1991); mammalian prey tends to be constricted
to death prior to swallowing more frequently
than ectothermic animals such as frogs (Gregory
et al., 1980; Mori 1991); and certain prey ani-
mals are always swallowed head first regardless
of their size (Voris et al., 1978; Mori 1998). The
degree of such behavioral flexibility is, to some
extent, species-specific (Halloy and Burghardt,
1990; Mori, 1996, 1997, 1998), and it has been
suggested that dietary specialists change their
behavior more efficiently than dietary general-
ists, especially when they are young (Drum-
mond, 1983; Halloy and Burghardt, 1990; Mori,
1993, 1994, 1995, 1996).

The third test was conducted to examine the ~
effects of prey size and type on prey-handling
behavior of the juvenile L. madagascariensis
that had been fed only one type of prey. Possible
roles of Duvernoy’s gland secretion in feeding
were also examined.

Methods

Five types of prey animals were used: small
mice (baby Mus musculus one to three days after
birth, 1.7—2.0 g), large mice (juvenile M. muscu-
lus, 9.5-12.2 g), small frogs (metamorphosed
Hyla japonica, 0.3—0.6 g), medium frogs (juve-
nile Rana nigromaculata, |1.8—2.3 g), and large
frogs (adult R. nigromaculata, 12.0—18.9 g).
Approximately one hour prior to the experiment,
water bowls, shelters, and paper floor coverings
were removed from the cages, and the ceiling
was replaced with a transparent acrylic board.
Each trial began by gently introducing a prey
into the cage. Feeding behavior of the snakes
was recorded with an 8mm video camera-
recorder until the prey was completely swal-
lowed. If the snakes did not attack the prey
within 20 min, the trial was terminated.

For the index of prey size, we used relative
prey width (1.e., prey head width/snake head
width), because width is generally considered
a better indicator of the prey size than length
or mass when evaluating the prey-handling

MORI & TANAKA—BEHAVIOR OF SNAKES

efficiency of gape-limited predators (Pough
and Groves, 1983; Mori, 1998). The following
variables were recorded for each trial by ana-
lyzing the videotapes. Bite position: the site
where the snake first seized the prey. Direction
of prey ingestion: the prey was eventually
swallowed “head first” or “hind legs (rump)
first”. Holding duration: time in sec from the
initial seizure of the prey to the first “lateral
jaw walking movements” by the snake in any
direction. Manipulating duration: time in sec
from the first lateral jaw walking movements
to the commencement of swallowing. The
commencement of swallowing was defined as
the first jaw walking movements over the body
of the prey in the direction in which the prey
was subsequently swallowed. Swallowing
duration: time in sec from the commencement
of swallowing to the moment at which the prey
was no longer visible externally. Condition of
prey at swallowing: prey that made any move-
ments, including breathing, during swallowing
were considered “‘alive’’, otherwise “dead’’. Pat-
tern of constricting or coiling behavior during
prey-handling was also recorded, whenever
observed.

Results

One individual did not attempt to eat small
and large mice, and another did not eat a
large mouse. All individuals readily struck
and attempted to swallow the frogs regardless
of their size. In all cases, snakes struck the
prey immediately after the prey showed some
kind of movement (locomotion, head turn,
body adjustment, etc). Attempts were made to
swallow two large frogs, but they were eventu-
ally regurgitated probably because the snakes
were unable to engulf them completely due to
their gape limitation (see below).

Irrespective of prey type, the direction of
ingestion seemed to depend on the initial bite
position: prey animals were swallowed head
first when the bite position was located on the
anterior part of the body including the head,
and they were swallowed hind legs first when
the bite position was located on the posterior
part of the body including the hind legs (Table

45

TABLE 3. Relationships between the initial bite
position and the direction of prey ingestion by
juvenile Leioheterodon madagascariensis. Numerals
indicate the numbers of prey animals.

Bite position

Direction of Anterior Posterior Hind
ingestion Head body body leg
Head first 7 l l l
Hind legs first 0 0 i 5

3). No apparent tendency for head first inges-
tion was recognized even for large mice and
frogs. In trials with small and medium sized
prey seven out of 14 were swallowed head first,
and three out of eight large prey were swal-
lowed head first. Except for one large frog, all
prey animals were swallowed alive.

Constricting behavior was observed only in
a single trial with a large mouse. Two seconds
after a snake seized the rump of a mouse,
inducing a vigorous struggling by the prey, the
snake attempted to coil around the prey by typ-
ical wrapping and winding movements (Green-
wald, 1978). However, the configuration of the
coil was irregular and unstable, and the snake
was not able to successfully coil around the
prey. The prey continued to struggle, and the
snake then tried to reconstrict the prey two
more times, eventually holding three neat coils
around the prey. However, the prey was not
killed by constriction and was swallowed alive
and rump first. Large mice and frogs in the
other trials also struggled vigorously during the
holding phase, but the snakes never attempted
to constrict them.

Holding duration was quite short, and
snakes began lateral jaw walking movements
soon after the initial seizure (holding dura-
tion, x =2 sec, range=0 to 9). In all but one trial
with small and medium sized frogs snakes
began to swallow without manipulation phase.
Manipulating duration was also short for small
mice (x =8.3 sec, range=2 to 23) and varied
from 0 to 492 sec for large frogs and mice
(x =147 sec). Swallowing duration considerably
increased when relative prey width exceeded 0.8

Current Herpetol. 20(1) 2001

46

Frog-head first
Frog-hind leg first
Frog-hind leg first (regurgitate)

_— 10000

O

ed)

<2

< 1000

2

ad

O

=

O 100

©)

£

=

Oo 10

©

S

op)

1
0.2 0.4 0.6

A Mouse-head first
4 Mouse-rump first

0.8 1.0 Nee 1.4

Relative prey width

Fic. 3. Relationship between relative prey width (prey head width/snake head width) and swallowing dura-
tion in feeding trials of juvenile Leioheterodon madagascariensis. Direction of prey ingestion (head first or

hind leg [or rump] first) is also shown.

(Fig. 3). For prey animals with relative prey
width<0.8, direction of ingestion did not seem
to affect the swallowing duration. For larger
prey (relative prey width>0.8), however, head
first ingestion was likely to enable the snakes
to swallow prey quicker than hind leg first
ingestion. In fact, two snakes that tried to
swallow large frogs (relative prey width=1 .33
and 1.30) hind leg first eventually abandoned
the swallowing attempts after 42.4-min and
79.3-min, respectively: the snakes engulfed the
frogs over their pectoral girdles but not their
forelegs, and voluntarily regurgitated the frogs.
In both cases the snakes initially began to swal-
low the frog from one of the hind legs letting
the other stretch far anteriorly so that the tip
of the latter leg reached anterior to the snout
of the frog while swallowing. The snakes, left
undisturbed for additional 10 min after regur-

gitation, flicked their tongues toward the frogs
frequently, but they never tried to swallow the
regurgitated frogs (one dead and one alive)
again. The live frog eventually died approxi-
mately 4 hours after regurgitation. Extensive
bleeding during the swallowing phase was
observed in three trials with large frogs.

Discussion

Virtually no effects of prey size and type on
prey-handling behavior were detected in the
juvenile L. madagascariensis. Adaptive func-
tion of head first ingestion of (large) prey has
been considered to reduce swallowing dura-
tion by minimizing the resistance caused by
the prey’s appendages (Diefenbach and Emslie,
1971; Mori, 1991). Direction of prey inges-
tion in L. madagascariensis seems to basi-
cally depend on initial bite position. As a

MORI & TANAKA—BEHAVIOR OF SNAKES

47

result, swallowing duration was prolonged
when large prey was ingested hind leg first,
and in the extreme cases, prey animals were
eventually regurgitated after a prolonged swal-
lowing attempt.

Selective advantage of constriction has been
considered to prevent prey from retaliating or
to reduce total feeding duration by killing or
restraining the prey (Loop and Bailey, 1972;
de Queiroz, 1984; Mori, 1991). In the present
observations, except for a single large mouse,
all prey were simply swallowed without any
constricting attempt even though most of the
large frogs and mice struggled vigorously dur-
ing the handling phase. The single case of
neat coiling indicates that L. madagascarien-
sis 18 potentially capable of constriction from
the viewpoint of morphological body plan,
either to kill or simply restrain the prey. The
ability of efficient constriction in L. mada-
gascariensis may develop ontogenetically as
in other generalist semi-constrictors such as
Elaphe quadrivirgata (Mori, 1994, 1996).

The functional roles of Duvernoy’s glands
during feeding have been disputed and tested
in several colubrid snakes (Kardong, 1982;
Hayes et al., 1993; Rodriguez-Robles, 1994).
After the initial seizure of prey, some snakes
hold it with the jaws for a while presumably
until the glands’ secretion takes effect (Broad-
ley, 1983; Jayne et al., 1988; Mori, 1998), and
other snakes repeatedly open and close the jaws
without shifting the position of bite (forceful
chewing) probably to inject more secretion
(Jansen and Foehring, 1983; Rodriguez-Robles,
1992; Thomas and Leal, 1993). In some cases,
prey animals are killed during handling (Jayne
et al., 1988; Mori, 1997), whereas in other
cases they are only weakened and incapaci-
tated (Mori, 1998). Some prey animals are
swallowed alive without any sign of envenom-
ation (Mori, 1997) or killed by constriction
without any apparent assistance of the secre-
tion (Rochelle and Kardong, 1993).

In L. madagascariensis initial holding dura-
tion was quite short, and no chewing-like
behavior was observed. Although a single
frog was recorded as “swallowed dead” on the

basis of our definition, it is highly likely that
the frog was still alive when the swallowing
process was initiated because its body did not
seem to lose tension throughout the process.
Thus, virtually no prey would have been
killed before the swallowing phase. How-
ever, the eventual death of the regurgitated
frogs suggests that Duvernoy’s secretion was
delivered during prey-handling. These
results suggest that Duvernoy’s secretion of
L. madagascariensis is insufficient to bring
about rapid death of prey, but may be effec-
tive to incapacitate large prey when feeding
duration is prolonged. Obviously L. mada-
gascariensis offers an additional example of
snakes with both well-developed Duver-
noy’s glands and potential for prey constric-
tion during feeding (Shine, 1985; Rochelle
and Kardong, 1993; Mori, 1998).

CONCLUSION

Our results showed that L. madagascarien-
sis 18 an ideal species for studies on various
aspects of snake behavior. This snake is a
dietary generalist, responds to chemical stim-
uli from various types of natural prey, has a
rich repertoire of distinct antipredator dis-
plays, and possesses two representative prey
subduing mechanisms of snakes (i.e., con-
striction and venom injection). Apparently
inefficient prey-handling by the juvenile
snakes suggests that this species is a good
subject to study the ontogenetic development
of prey-handling behavior as a representative
of generalist feeders (see Mori, 1996). Fur-
ther examination of these behavioral proper-
ties with greater sample size, coupled with
comparisons involving snakes from other
regions of the world, might shed light on the
unique features of the behavioral evolution in
Malagasy snakes, as well as on the ubiqui-
tous behavioral patterns in snakes.

ACKNOWLEDGMENTS

We thank S. Yamagishi for giving us an
opportunity to conduct this study and the

48

Current Herpetol. 20(1) 2001

staff of Parc Botanique et Zoologique de
Tsimbazaza for their assistance with the field
study. M. Toda, R. Yamamoto, J. Motokawa,
H. Matsubara, and H. J. A. R. Randriama-
hazo helped us with obtaining prey animals.
This study was partially supported by a
Grant-in-Aid from the Monbusho under an
International Scientific Research Program
(Field Research, no. 11691183).

LITERATURE CITED

BROADLEY, D. G. 1983. FitzSimons’ Snakes of
Southern Africa. (Revised ed.), Delta Book,
Cape Town. 376 p.

BrYGOO, E. R. 1982. Les ophidiens de Madagas-
car. Mem. Inst. Butantan 46: 19-58.

BURGHARDT, G. M. 1970. Chemical perception in
reptiles. p. 241-308. In: J. W. Johnston, JR., D.
G. Moulton and A. Turk (eds.), Advances in
Chemoreception, Vol. 1. Communication by
Chemical Signals, Appleton-Century-Crofts,
New York.

CAMPBELL, J. A. AND J. B. MURPHY. 1977. Mis-
cellaneous notes on the reproductive biology of
reptiles. I. Two colubrid snake species from the
Malagasy Republic, Leioheterodon madagas-
cariensis and Madagascariophis colubrina
(Reptilia, Serpentes, Colubridae). J. Herpetol.
11(2): 225-228.

CONANT, R. 1938. A note on eggs and young of
Leioheterodon madagascariensis (Duméril &
Bibron). Zoologica 23: 389-393.

COOPER, W. E., JR. 1998. Evaluation of swab and
related tests as a bioassay for assessing
responses by Squamata reptiles to chemical
stimuli. J. Chem. Ecol. 24(5): 841-866.

COOPER, W. E., JR., G. M. BURGHARDT, AND W. S.
BROWN. 2000. Behavioural responses by
hatchling racers (Coluber constrictor) from two
geographically distinct populations to chemical
stimuli from potential prey and predators.
Amphibia-Reptilia 21(1): 103-115.

DE QUEIROZ, A. 1984. Effects of prey type on the
prey-handling behavior of the bullsnake, Pituo-
phis melanoleucus. J. Herpetol. 18(3): 333-
336;

DIEFENBACH, C. O. AND S. G. EMSLIE. 1971.
Cues influencing the direction of prey ingestion
of the Japanese snake, Elaphe climacophora
(Colubridae, Serpentes). Herpetologica 27(4):
461-466.

DOMERGUE, C. A. 1989. Un serpent venimeux de
Madagascar observation de deux cas de morsure

par Madagascarophis (Colubridaé opisthoglyphe).
Arch. Inst. Pasteur Madagascar, 56(1): 299-311.

DOMERGUE, C. A. AND J. RICHAUD. 1971. Activité
hémolytique des sécrétions des glandes de Duver-
noy chez Lioheterodon (Colubridé aglyphe). Arch.
Inst. Pasteur Madagascar, 40(1): 145-148.

DRUMMOND, H. 1983. Aquatic foraging in garter
snakes: a comparison of specialists and generalists.
Behaviour 86(1): 1-30.

FORD, N. AND G. M. BURGHARDT. 1993. Perceptional
mechanisms and the behavioral ecology of snakes.
p. 117-164. In: R. A. Seigel and J. T. Collins (eds.),
Snakes. Ecology & Behavior. McGraw-Hill, New
York.

GLAW, F. AND M. VENCES. 1994. A Fieldguide to the
Amphibians and Reptiles of Madagascar. (2nd ed.),
Zoologisches Forschungsinstitut und Museum
Alexander Koenig, Bonn. 480 p.

GREENE, H. W. 1988. Antipredator mechanisms in
reptiles. p. 1-152. Jn: C. Gans and R. B. Huey
(eds.), Biology of the Reptilia, Volume 16, Ecology
B, Defense and Life History. Alan R. Liss, Inc,
New York.

GREENWALD, O. E. 1978. Kinematics and time rela-
tions of prey capture by gopher snakes. Copeia
1978(2): 263-268.

GREGORY, P. T., J. M. MACARTNEY, AND D. H.
RIVARD. 1980. Small mammal predation and prey
handling behavior by the garter snake Thamnophis
elegans. Herpetologica 36(1): 87-93.

GROVES, F. AND J. D. GROVES. 1978. Fatal toad poi-
soning in snakes. Herpetol. Rev. 9(1): 19-20.

HAYES, W. K., P. LAViN-MURCIO, AND K. V. KAR-
DONG. 1993. Delivery of Duvernoy’s secretion
into prey by the brown tree snake, Boiga irregu-
laris (Serpentes: Colubridae). Toxicon 31(7): 881—
887.

HALLOY, M. AND G. M. BURGHARDT. 1990. Ontog-
eny of fish capture and ingestion in four species of
garter snakes (Thamnophis). Behaviour 112(3/4):
299-318.

HERZOG, H. A., JR. AND G M. BURGHARDT. 1986.
Development of antipredator responses in snakes:
I. Defensive and open-field behaviors in newborns
and adults of three species of garter snakes (Tham-
nophis melanogaster, T. sirtalis, T: butleri). J.
Comp. Psychol. 100(4): 372-379.

HERZOG, H. A., JR, B. B. BOWERS, AND G M.
BURGHARDT. 1989. Stimulus control of antipreda-
tor behavior in newborn and juvenile ga ter snakes
(Thamnophis). J. Comp. Psychol. 103(3): 233-242.

JANSEN, D. W. AND R. C. FOEHRING. 1983. The
mechanism of venom secretion from Duvernoy’s
gland of the snake Thamnophis sirtalis. J. Morphol.
175(3): 271-277.

JAYNE, B. C., H. K. VORIS, AND K. B. HEANG. 1988.
Diet, feeding behavior, growth, and numbers of a

MORI & TANAKA—BEHAVIOR OF SNAKES

49

population of Cerberus rynchops (Serpentes:
Homalopsinae) in Malaysia. Fieldiana Zoology,
New Ser. (50): 1-15.

KARDONG, K. V. 1982. The evolution of the venom
apparatus in snakes from colubrids to viperids &
elapids. Mem. Inst. Butantan 16: 105-118.

Loop, M. S. AND L. G BAILEY. 1972. The effect of
relative prey size on the ingestion behavior of
rodent-eating snakes. Psychon. Sci. 28(3): 167-169.

Mor!, A. 1991. Effects of prey size and type on
prey-handling behavior in Elaphe quadrivirgata.
J. Herpetol. 25(2): 160-166.

Mor!I, A. 1993. Prey handling behavior of neonatal
rat snakes, Elaphe taeniura and E. dione (Colu-
bridae). Jpn. J. Herpetol. 15(2): 59-63.

Mor!, A. 1994. Prey handling behavior of newly
hatched snakes in two species of the genus Elaphe
with comparison to adult behavior. Ethology
97(3): 198-214.

MorI, A. 1995. Prey handling behavior of young rat
snake, Elaphe taeniura (Squamata: Reptilia).
Mem. Fac. Sci. Kyoto Univ. Ser. B. 16: 43-47.

Mor], A. 1996. A comparative study of the devel-
opment of prey handling behaviour in young rat
snakes, Elaphe quadrivirgata and E. climacoph-
ora. Herpetologica 52(3): 313-322.

Morl!, A. 1997. A comparison of predatory behavior
of newly hatched Rhabdophis tigrinus (Serpentes:
Colubridae) on frogs and fish. Jpn. J. Herpetol.
17(2): 39-45.

Mor!, A. 1998. Prey-handling behavior of three
species of homalopsine snakes: features associ-
ated with piscivory and Duvernoy’s glands. J.
Herpetol. 32(1): 40-50.

Mor, A. 2000. A case of envenomation by the
Madagascan colubrid snake, Leioheterodon
modestus. Snake 29(1/2):(in press)

Mor!, A. AND G. M. BURGHARDT. 2000. Does prey
matter? Geographic differences in antipredator
responses in hatchlings of a Japanese colubrid
snake, Rhabdophis tigrinus. J. Comp. Psychol.
114(4):408-413

MorI, A. AND T. HIKIDA. 1991. Notes on defensive
behavior of the Asian elapid, Maticora intestina-
lis. Snake 23(2): 107-109.

Mor], A. AND H. J. A. R. RANDRIAMAHAZO. (in
press) Leioheterodon madagascariensis (Mada-
gascar Menarana Snake). Diet. Herpetol. Rev.

Mor], A., D. LAYNE, AND G. M. BURGHARDT.
1996. Description and preliminary analysis of
antipredator behavior of a colubrid snake hav-
ing nuchal glands, Rhabdophis tigrinus tigri-
nus. Jpn J. Herpetol. 16(3): 94-107.

PouGH, F. H. AND J. D. GROVES. 1983. Special-
izations of the body form and food habits of
snakes. Amer. Zool. 23(2): 443-454.

PRESTON-MAFHAM, K. 1991. Madagascar, A Nat-
ural History. Facts On File, Oxford. 224 p.

RANDRIAMAHAZO, H. 1999. Reptiles in Mada-
gascar. p. 167-212. In: Yamagishi, S. (ed.),
The Animals of Madagascar: Its Wonderful
Adaptive Radiation. Shokabo, Tokyo. (in Japa-
nese)

ROCHELLE, M. J. AND K. V. KARDONG. 1993.
Constriction versus envenomation in prey cap-
ture by the brown tree snake, Boiga irregularis
(Squamata: Colubridae). Herpetologica 49(3):
301-304.

RODRiGUEZ-ROBLES, J. A. 1992. Notes on the
feeding behavior of the Puerto Rican racer,
Alsophis portoricensis (Serpentes: Colubridae).
J. Herpetol. 26(1): 100-102.

RODRiGUEZ-ROBLES, J. A. 1994. Are the Duver-
noy’s gland secretions of colubrid snakes venom?
J. Herpetol. 28(3): 388-390.

SHINE, R. 1985. Prey constriction by venomous
snakes: a review, and new data on Australian spe-
cies. Copeia 1985(4): 1067-1071.

THOMAS, R. AND M. LEAL. 1993. Feeding enveno-
mation by Arrhyton exiguum (Serpentes: Colu-
bridae). J. Herpetol. 27(1): 107-109.

VorIs, H. K., H. H. VorIS, AND L. B. LIAT. 1978.
The food and feeding behavior of a marine
snake, Enhydrina_ schistosa (Hydrophiidae).
Copeia 1978(1): 134-146.

VON DATHE, F. AND K. DEDEKIND. 1996. Pflege
und Vermehtung von Madagaskar-Hakennattern
(Leioheterodon madagascariensis Duméril &
Bibron, 1854) im Tierpark Berlin-Friedrichsfelde.
Zool. Garten N. F. (66): 69-76.

Current Herpetology 20(1): 51—60., June 2001
© 2001 by The Herpetological Society of Japan

Microhabitat Choice of Tadpoles of Seven Anuran Species

NINGAPPA C. HIRAGOND AND SRINIVAS K. SAIDAPUR“*

Department of Zoology, Karnatak University, Dharwad-580 003, INDIA

Abstract: The habitat choice of tadpoles of seven anuran species with respect to
light and dark phases was studied using a choice tank with a simulated pond
edge. In each trial, which lasted for 12 hrs, twenty freshly collected tadpoles
(Gosner stage 32-38) of a given species were used. After introducing tadpoles
into the choice tank, their number at the surface, up the water column, and on
the substrate was recorded at half-hour intervals. Four trials each for day and
night were conducted for each species. Tadpoles of Rana curtipes, Rana
cyanophlyctis, and Rana temporalis, which possess a ventral mouth, predomi-
nantly occupied the substrate zone, whereas those of Bufo melanostictus,
Polypedates maculatus, and Rana tigrina, which have an antero-ventral mouth,
utilized both substrate and column zones but with a clear preference for the
substrate zone. Within the substrate zone, the number of tadpoles of B. mel-
anostictus, P. maculatus, R. temporalis, and R. tigrina was greater at night than
in the daytime, whereas the opposite was true for R. curtipes. Of these five spe-
cies response for day-night changes was prominent in P. maculatus. It is unclear
whether variation in the number of tadpoles with respect to day and night is
due to changes in the light intensity or temperature or oxygen levels. The tad-
poles of R. cyanophlyctis did not show any response to day-night changes. On
the other hand, the tadpoles of Microhyla ornata, which possess an antero-dor-
sally placed and highly simplified mouth (devoid of teeth) preferred the surface
(64%) and column (28%) zones in both light and dark phases. The tadpoles of
M. ornata are not easily spotted despite their surface occupancy due to their
transparent appearance. The present study shows the existence of diversity in
the microhabitat choice of the anuran tadpoles that correlates well with their
morphological characteristics.

Key words: Habitat choice; Tadpoles; Anurans; Bufo melanostictus; Microhyla
ornata, Polypedates maculatus; Rana cyanophlyctis; Rana curtipes; Rana tempora-
lis; Rana tigrina.

ephemeral pond ecosystems. In southern India
following southwest monsoon rains, many anu-
Anuran larvae are the major components of ran species co-breed in ephemeral ponds and
puddles (Saidapur, 1989). Hence, tadpoles of
different species that live together are subjected
to severe competition for food and space and to
predation pressures due to aquatic insects as
well as carnivorous tadpoles of other species.

INTRODUCTION

“Corresponding author. Tel/Fax: +91—0836-—
448047.

E-mail address: saidapur@satyam.net.in/said-
apur@hotmail.com (S. K. Saidapur)

Sy

Current Herpetol. 20(1) 2001

Some species of tadpoles also exhibit cannibal-
ism. Among many possible ways, one way to
avoid such competition is to select proper
microhabitat vis-a-vis stratification within a
given water body. There are a few studies
that deal with the background preference of
tadpoles for specific types of substrates such
as featureless or square-patterned substrates
(Wiens, 1972), smooth rocks (Altig and
Broodie, 1992), and dark or white background
(Moriya et al., 1996). However, studies on the
microhabitat choice of anuran larvae are scanty
(Hoff et al., 1999). The experimental studies
on microhabitat choice of tadpoles with respect
to surface, column, and substrate zones are lim-
ited to Ranidella signifera and Litoria ewingi
(Peterson et al., 1992).

It is generally considered that morphological
features of tadpoles are correlated with their
habitat choice. Thus, the tadpoles of differ-
ent anuran species living sympatrically may
exhibit diversity in their microhabitat choice.
The present study was undertaken to experi-
mentally investigate the microhabitat choice of
seven anuran species using a specially designed
choice tank with a simulated pond edge. Of
these, five species, viz., Bufo melanostictus,
Microhyla ornata, Polypedates maculatus, Rana
cyanophlyctis and R. tigrina often live together
in ephemeral ponds or puddles. The tadpoles
of the remaining two species (Rana curtipes
and R. temporalis) coexist in gently flowing

streams as well as in isolated pockets of still
water alongside the streams.

MATERIALS AND METHODS

Subjects

Tadpoles of R. cyanophlyctis, R. tigrina
(Ranidae), P. maculatus (Rhacophoridae), B.
melanostictus (Bufonidae), and M. ornata (Hyl-
idae) were collected from ephemeral ponds
or puddles around Dharwad city (15°17'N,
75°03'E), and those of R. curtipes and R. tem-
poralis (Ranidae) from a stream 75 km from
Dharwad (15°17'N, 74°03'E). The tadpoles
were acclimated to laboratory conditions for
a day before experiments. In all experiments
tadpoles of stages 32-38 (Gosner, 1960) were
used. Tadpoles of B. melanostictus, P. macula-
tus, R. cyanophlyctis, R. curtipes, and R. tem-
poralis were fed boiled spinach, and those of R. ~
tigrina were fed with Chironomus larvae or
small fresh water prawns ad-libitum before the
commencement of experiments. Tadpoles of
M. ornata fed on suspended matter present in
the aquarium water. In all trials, no food was
provided in the choice tank.

Design of the Choice Tank

An all-glass aquarium (1806060 cm) with
a simulated pond edge, created using an asbes-
tos sheet, was used as the choice tank (Fig. 1).
The aquarium was marked outside so as to

60 i——————

b = Tal = Sa Olcini= aa See ee J

Fic. 1. A front view of the choice tank, the simulated pond edge and the different zones (surface=sections 1—
3; column=sections 4—6; substratum=sections 7-9). AS= Asbestos sheet.

HIRAGOND & SAIDAPUR—HABITAT CHOICE OF TADPOLES a3

divide it into 3 zones: surface, water column,
and substrate. The three zones were subdivided
into sections to represent different areas of the
aquarium. Sections 1—3, 4—6, and 7—9 repre-
sent surface, column, and substrate zones,
respectively. The choice tank was kept in a
place which had a roof and walls on only two
sides. Therefore, it was exposed to ambient
temperature and light cycles.

Experimental Design

Experiments were conducted during the day
(0600 h to 1800 h) and at night (1800h to
0600 h). A red bulb (15-watt) fitted 1 m above
the aquarium was used to enable us to record
the position of tadpoles in the darkness. Twenty
tadpoles of a given species were introduced into
the choice tank and allowed 30 minutes to posi-
tion themselves in the preferred zone of the
tank. Then the number of individuals found in
each section was recorded at half-hour inter-
vals. A given set of tadpoles, numbering 20,
was used only once, either for the day or for the
night recordings. The aquarium was cleaned,
and water was replenished between the trials.
A total of four trials were made, separately for
day and night recordings, for each species. The
number of tadpoles found in sections 1—3, 4-6,
and 7—9 was pooled to determine the total num-
ber of tadpoles occupying surface, column, and
substratum zones, respectively, for each record-
ing.

Statistical analysis

The mean of four trials of each observation
for day or night were used for statistical analy-
ses. A chi-square test [rxc model; df, (24-1)(3-
1)=46] was used to examine whether the distri-
bution of tadpoles was homogeneous within a
given zone throughout the day or night phase
(24 observations). To examine the zone prefer-
ence of a particular species during light or
dark phase, a chi-square test with 2 df was
performed. This test was to determine whether
the distribution of tadpoles in the three zones
(surface, column, and substrate) was uniform
(1:1:1). Further, if the distribution was homo-
geneous, the mean of 24 observations for day

or night was computed and used for chi-square
analysis. On the other hand, if the distribution
of tadpoles was heterogeneous for a particular
species, the 24 observations individually for day
or night, were used to determine location pref-
erence. In all cases, the null hypothesis was
rejected at the 0.05 level. The microhabitat
preference of each species was also ascertained
by checking observed frequencies. When
tadpoles occupied a particular zone in a sig-
nificantly large number, Friedman two-way
ANOVA was performed to examine whether
they preferred a particular section of the zone.
Mann-Whitney U test was used to examine the
difference in the number of tadpoles between
day and night with respect to each zone.
Minitab and SPSS (version 6.3.1 for Windows)
statistical packages were used for data analysis.

RESULTS

Bufo melanostictus

These tadpoles generally adhered to the
walls of the aquarium when at the surface or
in the column. On the substrate they were
found either resting or scraping. The toad tad-
poles were distributed homogeneously in each
zone throughout the light phase (¥74,=59.86,
p>0.05) but not in the dark hours of the night
(¥746=70.52, p<0.05). The location preference
of tadpoles for the three zones differed signifi-
cantly in light (y*,=2389.28, p<0.05) as well
as in dark (y*,>6, p<0.05) phases. The occu-
pancy of the tadpoles in the substrate zone
(Table 1) was always significantly greater than
that of the other two zones (Fig. 2). The tad-
poles preferred to occupy the deep section of
the substrate (section 9) especially during the
light phase (Friedman y7,=41.33, p<0.05)
while in darkness their number increased in
section 8 (Friedman y7,=14.96, p<0.05, Table
2). In the column zone the number of tad-
poles was significantly greater in the day than
in the night (U=145.0, p<0.05). On the sub-
strate the number of tadpoles was larger dur-
ing the night than during the day (U=132.0,
p<0.05, Fig. 2). In the surface zone, there
was no difference in the number of tadpoles in

54

Current Herpetol. 20(1) 2001

TABLE |. The natural habitat, some morphological features, and the distribution pattern of tadpoles during day and
night in the choice tank, as shown by the distribution of tadpoles in percent.

Day Night
Species Habitat/Color/mouth Pattern” Sur. Col. Sub. Pattern” Sur. Col. Sub.
eee mele Puddles/Black/Anteroventral Homogeneous 3.8010.4085.75 Heterogeneous 2.70 8.05 89.40
anostictus
DEE IOS I Heterogeneous 62.65 30.90 6.60 Heterogeneous 62.2524.25 9.70
ornata Anterodorsal

3) Polypedates Puddles/Light green or muddy

maculatus dark/Anteroventral

4) Rana Streams/Black or reddish
curtipes brown/Ventral

5) Rana Puddles/Light or dark brown/

cynophlyctis Ventral

6) Rana Streams/Muddy green, yellow-
temporalis ish or olive brown/Ventral

7) Rana Puddles/Olive or muddy brown/
tigrina Anteroventral

Homogeneous

Homogeneous

Heterogeneous 23.55 31.0045.45 Homogeneous 7.8016.3075.90

Heterogeneous 5.60 13.7580.70 Heterogeneous 10.85 17.8071.40

0.99 7.2491.83 Homogeneous 1.47 7.03 91.57

Heterogeneous 0.55 9.2090.35 Heterogeneous 0.75 5.2094.10

2.55 7.1090.35 Homogeneous 0.55 3.65 95.80

* Based on the chi-square test for homogeneity (r<c model and df 46) with reference to distribution of tadpoles
within surface, column, or substrate zones, respectively, during the study period (24 observations). Sur=Surface;

Col=Column; Sub=Substratum.

TABLE 2. Number of tadpoles (Mean+SE) found in different sections (A—C) of their preferred zones of the choice
tank during day and night. A-C refer to sections 7—9 and 1—3,respectively, for tadpoles that preferred substrate zone

and surface zone.

Day Night
Species A B Cc A B C
Tadpoles Showing Preference for Substrate zone
Bufo melanostictus 1.3140.16 6.24+0.39 9.60+0.49 5.13+0.35 6.65+0.26 6.09+0.44
Polypedates maculatus 075820.10" — 12590309 6.92+0.37 4.47+0.30 2.98+0.17 7.73+0.40
Rana curtipes 2.3340.25 2.4140.14 11.39+0.49 2,9340)27 9 92-342020 9.01+0.32
Rana cyanophlyctis 1:8920:16 3.652022 -12782+0:31 3.90+0.28 2.83+0.14 11.58+0.32
Rana temporalis 2.02+0.25 4.0940.17 11.96+0.41 4.58+0.33 4.61+0.19 9.61+0.34
Rana tigrina 1.35+0.14 2.3640.15 14.35+0.27 4.96+0.25 4.16+0.13 10.04+0.27
Tadpoles Showing Preference for Surface zone
Microhyla ornata 4.8140.50 1.64+0.15 6.08+0.30 8.29+0.45 = 2.11+0.18 2.75+0.26

relation to day and night (U=195.0, p>0.05,
Fig. 2).

Microhyla ornata

These tadpoles swam for prolonged periods.
They remained stationary for brief periods by
fluttering their posterior tail. On the substrate,
they remained either static or moved slowly.

The distribution of tadpoles within each of the
three zones of the choice tank differed during
both day (y74=197.79, p<0.05) and night
(X746=72.28, p<0.05). There was also a sig-
nificant difference in occupancy of tadpoles
between the three zones during day as well as
night (y7,>6, p<0.05). A greater number of tad-
poles occupied the surface zone (Table 1) in

HIRAGOND & SAIDAPUR—HABITAT CHOICE OF TADPOLES 35

[__] Day
night

B. melanostictus US b@

PF. 7acwilaris

Number of Tadpoles (Mean + SE)

SUR COL SUB

FIG. 2. The distribution of tadpoles of B. melanostictus, M. ornata, and P. maculatus, in the choice tank dur-
ing day and night. Dissimilar alphabetical superscripts on bars in a given zone indicate a significant difference
between day and night. Open and closed circles indicate significantly high number on the substrate (P. macula-
tus and B. melanostictus) and at the surface (M. ornata) compared to other zones during day and night, respec-
tively.

comparison to the other two zones regardless of | also varied (day: Friedman y*,=29.19, p<0.05;
light or darkness (Fig. 2). The number of tad- night: Friedman y*,;=35.50, p<0.05). In the
poles in different sections of the surface zone night, tadpoles preferred section 1 as compared

56

Current Herpetol. 20(1) 2001

to section 3, but during the day the preference
was reversed (Table 2). During the night, the
tadpole number on the substrate increased com-
pared to day time (U=115.5, p<0.05). The dif-
ference in the number of tadpoles between day
and night was not statistically significant at the
surface (U=239.5, p>0.05) or column (U=194.5,
p>0.05) zones.

Polypedates maculatus

These tadpoles moved actively in the choice
tank. The distribution of tadpoles in each of the
three zones was heterogeneous during the day
time (y74,=171.44, p<0.05) but it was homog-
enous during the night hours (¥*,,=60.04,
p>0.05). Analysis of location preference of tad-
poles with respect to day (y7,>6, p<0.05) and
night (y7,=2991.43, p<0.05) revealed that there
is a significant difference in the mean number of
tadpoles occupying the surface, column, or sub-
strate (Fig. 2). In general, a greater number of
tadpoles always occupied the substrate (Table 1),
especially section 9 (Tale 2), regardless of light
and darkness (Friedman y*,;=48.00, p<0.05 and
¥*,=35.04, p<0.05 respectively, Fig. 2). At the
surface (U=76.0, p<0.05) and column zones
(U=4.0, p<0.05) the mean number of tadpoles
was higher during the day compared to night.
This was reflected in a corresponding decline in
number of tadpoles on the substrate (U=0,
p<0.05) during the day time (Fig. 2).

Rana curtipes

Tadpoles of R. curtipes moved actively
throughout the choice tank. They frequently
dashed to the surface for aeration but quickly
returned to the substrate. Distribution of tad-
poles within the three zones varied during
both light (y74,=196.05, p<0.05) and dark
(%°46=97.75, p<0.05) phases. Likewise, there
was a significant variation in the number of
tadpoles between the three zones in both phases
(x7.>6, p<0.05). The mean number of tadpoles
occupying the substrate zone was greater in both
light and dark phases as compared to other
zones (Fig. 3, Table 1). Within this zone a sig-
nificantly greater number of tadpoles occupied
section 9, the deep substrate, during both day

(Friedman ¥*,=37.53, p<0.05) and night hours
(Friedman y*,=38.08, p<0.05). In surface and
column zones the number of tadpoles was very
low throughout the day but it increased during
the dark phase (surface: U=110.5, p<0.05, col-
umn: U=186.5, p<0.05) due to migration of
some tadpoles from the substrate zone during the
night. This resulted in a corresponding decrease
in their number in the substrate zone at night
(U=141.0, p<0.05, Fig. 3).

Rana cyanophlyctis

In the choice tank, the tadpoles occasionally
made a rapid vertical swim to the surface but
suddenly returned to the column or substratum.
The distribution of tadpoles within a given
zone was homogeneous (Table 1) at different
times of the day (y74,=54.25, p>0.05) as well as
night (¥74;=38.37, p>0.05). A greater number
of tadpoles occupied the substrate zone (Fig.
3) compared to the other two zones in both
light (y7,=2961.76, p<0.05) and dark
(¥°2=2938.62, p<0.05,) phases. Furthermore,
irrespective of light or darkness a greater num-
ber of tadpoles were found in the deep sub-
strate zone (Friedman y*,=46.08, p<0.05 for
day, 77,=39.00, p<0.05 for night, Fig. 3) than
in sections 7—8 of the substrate zone (Table
2). Light and dark phases did not influence
the number of tadpoles occupying surface
(U=211.0, p>0.05), column (U=277.0, p>0.05)
or the substrate (U=247.5, p>0.05) zones (Fig.
3).

Rana temporalis

These tadpoles remained motionless for pro-
longed periods on the substrate. Occasionally,
they moved to the surface but quickly descended
back to the substrate zone. The distribution
of tadpoles within the surface, column, and
substrate zones varied during the day time
(¥746=142.64, p<0.05) as well as in the night
(¥746=71.28, p<0.05). In both day and night
hours a greater number of tadpoles occupied
the substrate zone (Table 1) in comparison to
other zones (y7,>6, p<0.5, Fig. 3). Furthermore,
within the substrate zone a greater number of
tadpoles preferred section 9 to sections 7 and 8

HIRAGOND & SAIDAPUR—HABITAT CHOICE OF TADPOLES 37)

20
i. CUrtLDeS

Number of Tadpoles (Mean + SE)

OQ

SUR

COL

SUB

ES) wes;
Night

R. cyanophlyctiso @

ho LEOTIUNG ae

SUR

COL SUB

Fic. 3. The distribution of R. curtipes, R. cyanophlyctis, R. temporalis, and R. tigrina tadpoles in the choice
tank during day and night. Dissimilar alphabetical superscripts on bars in a given zone indicate a significant
difference between day and night. Open and closed circles indicate significantly high number on the substrate
compared to other zones during day and night, respectively.

(Table 2) of the choice tank (day: Friedman
x7,=43.66, p<0.05, night: Friedman y?,=36.40,
p<0.05). The surface zone occupancy of tad-
poles did not vary with regard to light or
darkness (U=269.0, p>0.05). In the column
zone, the number of tadpoles was relatively
greater during the day than in the night hours
(U=133.5, p<0.05) due to the movement of
some individuals from the substrate (Fig. 3)

where the tadpole number decreased (U=150.5,
p<0.05).

Rana tigrina

These tadpoles occasionally took a vertical
swim to the surface but suddenly descended to
the substrate. The tadpoles of R. tigrina were
less active in the night as compared to the
day. The distribution of tadpoles in a given

58

zone was homogeneous during both light
(4 246=62.75, p>0.05) and dark (y%24=39.55,
p>0.05) phases. However, there was a signifi-
cant variation in the number of tadpoles occu-
pying the three zones. A greater number of
tadpoles occupied the substrate zone (Table 1)
during both day (y*,=2816.13, p<0.05) and
night hours (y7,=3372.10, p<0.05, Fig. 3). The
tadpoles always preferred section 9 (Table 2),
the deep substrate (day: Friedman y,=45.66,
p<0.05 and night: Friedman y~;=37.17, p<0.05).
During the day time, the number of tadpoles was
greater at the surface (U=105.0, p<0.05) and col-
umn (U=97.0, p<0.05) smaller on the substrate
(U=54.5, p<0.05) as compared to night time
recordings (Fig. 3).

DISCUSSION

Microhabitat selection is an important strategy
of anuran larvae as it plays a key role in ensuring
their survival (Peterson et al., 1992) and growth
(Hoff et al., 1999). Our study showed that the
specially designed aquarium with a simulated
pond edge was a useful model to experimentally
examine the microhabitat choice of tadpoles in
the laboratory. In habitat choice experiments it
is typically assumed that the entire study area
is available for all individuals in a trial. This
is especially true when the relative amount of
different microhabitat zones remains stable
throughout the study period. Therefore each
position of the animal may be regarded as one
independent choice (Hjermann, 2000). Of the
seven species studied (B. melanostictus, P. mac-
ulatus, R. curtipes, R. cyanophlyctis, R. tempo-
ralis, and R. tigrina), six of them showed clear
preference for the substrate zone during both
light and dark phases. These species used the
surface and column zones to a very small extent
with the exception of P. maculatus, which used
both zones to an appreciable extent.

Anuran tadpoles are not stationary creatures
and as such they keep moving continuously after
intermittent periods of rest. In choosing a micro-
habitat the tadpoles may adopt different strate-
gies. A tadpole may perceive the habitat within
a certain radius prior to each move, and then

Current Herpetol. 20(1) 2001

select a point based on that perception. Also, it
is possible that the tadpoles move randomly in
all directions until they reach a habitat that con-
firms to their needs, and if a useful habitat is not
encountered, their random movement continues
for a long time. Thus, in the process of select-
ing a microhabitat, a fluctuation in the number
of tadpoles due even to a transitory movement
(during one or two occasions out of 24 obser-
vations) from one zone to another may con-
tribute to a statistically significant difference.
This explains the situations where small differ-
ences between day and night or between dif-
ferent sections of a given zone resulted in a
statistical significance. In fact, a variation in the
number of tadpoles seen at the surface and/or in
the column zone in case of B. melanostictus, R.
temporalis, R. curtipes, and R. tigrina 1s indica-
tive of this phenomenon. Yet, the observed pat-
terns are better indicators of habitat selection.
Accordingly, the present observations clearly
indicate that all the six species exhibit a prefer-
ence for the substrate over the other two zones.

The available studies on the behavioral
responses to changes in light intensity and tem-
perature indicate that tadpoles of some species
show responses to proximate factors, whereas
others do not (Beiswenger, 1977; Hoff et al.,
1999; Ultsch et al., 1999). The present study
shows that the preference of tadpoles of B. mel-
anostictus, P. maculatus, R. temporalis, and R.
tigrina for the substrate zone is greater in night
hours than in the day time, unlike in R. curtipes
where the opposite is true. Of these five spe-
cies the response of P. maculatus for day-night
changes was prominent. However, whether
variation in the number of tadpoles with respect
to day and night is due to a change in the light
intensity, temperature, or oxygen levels is not
clear. Tadpoles of R. cyanophilyctis that pre-
ferred the substrate did not show any response to
day-night changes.

The tadpoles of M. ornata showed a clear
preference for surface and column zones with
a minimum occupation of the substrate zone at _
all times of the day and night. Thus, a fluctua-
tion in the proximate factors such as the ambi-
ent light and temperature had no appreciable

HIRAGOND & SAIDAPUR—HABITAT CHOICE OF TADPOLES SY

effect on the distribution pattern of M. ornata
tadpoles as in R. cyanophlyctis. The mean per-
centage figures for M. ornata do not differ sig-
nificantly with respect to day and night for a
given zone (Table 1). Within the surface zone,
they preferred section 3 during the day and sec-
tion 1 during the night, i.e. deep and surface
areas, respectively. This might indicate that tad-
poles of M. ornata move towards shore during
night and away from it during the day.

The effects of light and/or temperature on
microhabitat choice of the seven species are
apparently not the same as reported for other
anuran species (Beiswenger, 1977; Duellman
and Trueb, 1986; Ultsch et al., 1999). In the
present study changes in the ambient tempera-
ture, light intensity, or oxygen levels were not
measured. Therefore, specially designed experi-
ments are needed to elucidate the influence of
proximate factors, if any, on the microhabitat
choice of the seven anuran tadpoles.

The present study provides some clues regard-
ing a possible correlation between morphologi-
cal features and microhabitat selection since
tadpoles with diverse features have been used.
The position of the mouth is anteroventral in B.
melanostictus, P. maculatus, and R. tigrina; ven-
tral in R. cyanophlyctis, R. curtipes (Hiragond et
al., 2001), and R. temporalis (Hiragond and
Saidapur, 1999) and anterodorsal in M. ornata.
A ventrally positioned mouth is typically suited
for foraging on the substrate while an anteroven-
tral mouth may help in feeding on the substrate
as well as in the column. In general, the tadpole
species with a ventral mouth showed a micro-
habitat choice of substrate and those with an
antero-ventral mouth selected the substrate and
to some extent the column zone. Transient
movements of tadpoles to other zones for brief
periods results in overlapping use of preferred
and other zones. Hence, occasionally, we find
bottom dwellers (tadpoles with ventral mouth) in
the column and those preferring column zones
(tadpoles with antero-ventral mouth) on the sub-
strate.

In tadpoles of the six species showing prefer-
ence for the substrate zone, the body is typically
flattened dorsolaterally. The coloration of their

bodies matches well the color of the substrate
of the natural habitat from where they were col-
lected. The highly simplified mouth (devoid of
denticles) of M. ornata placed in an anterodor-
sal position is typical of surface or suspension
feeders. The tadpoles of M. ornata due to their
transparent appearance are not easily spotted
despite their surface or column occupancy.
Also, the ventral fin of M ornata is more
robust than that of the other six species. Such a
ventral fin is needed for surface dwellers.
Thus, the microhabitat choice of surface zone
by M. ornata tadpole is well correlated with its
morphological characteristics. In summary, the
present study documents existence of a diver-
sity in the microhabitat choice of anuran tad-
poles and shows that there is a good correlation
between morphological characteristics and
microhabitat selection.

ACKNOWLEDGEMENTS

The work was supported by the Department
of Science and Technology (DST), New Delhi
(grant No. SP/SO/C-022/95). We thank Dr. A.
P. Gore, Dr. S. A. Paranjape, Department of
Statistics, University of Pune, Puna and Dr. I.
D. Shetty, Department of Statistics, Karnatak
University, Dharwad for their help in statistical
analysis of the data. NCH is thankful to DST
for a Junior Research Fellowship.

LITERATURE CITED

ALTIG, R. AND E. D. BRODIE, JR. 1972. Laboratory
behavior of Ascaphus truei tadpoles. J. Herpetol.
6: 21-24.

BEISWENGER, R. E. 1977. Diel patterns of aggrega-
tive behavior in tadpoles of Bufo americanus, in
relation to light and temperature. Ecology 58: 98—
108.

DUELLMAN, W. E. AND L. TRUEB. 1986. Biology of
Amphibians. McGraw-Hill, New York. 670 p.

GOSNER, K. L. 1960. A simplified table for staging
anuran embryos and larvae with notes on identifi-
cation. Herpetologica 16: 183-190.

HIRAGOND, N. C. AND S. K. SAIDAPUR. 1999.
Description of tadpole Rana temporalis from
south India. Curr. Sci. 76: 442-444.

HIRAGOND, N. C., B. A. SHANBHAG, AND S. K.
SAIDAPUR. 2001. Description of the tadpole of a

60

stream breeding frog, Rana curtipes. J. Herpetol.
35: 166-168.

HJERMANN, 2000. Analyzing habitat selection in ani-
mals without well defined home ranges. Ecology
81: 1462-1468.

Horr, K. S., A. R. BLAUSTEIN, R. W. MCDIARMID,
AND R. ALTIG. 1999. Behavior: Interactions and
their consequences. p. 215-239. In: R. W. McDi-
armid and R. Altig (Eds.), Tadpoles: The Biology
of Anuran Larvae. The Univ. Chicago Press, Chi-
cago.

MortyA, T., K. MIYASHITA, AND K. ASAMI. 1996.
Preference for background color of the Xenopus
laevis tadpoles. J. Exp. Zool. 276: 335-344.

Current Herpetol. 20(1) 2001

PETERSON, A. G, C. M. BULL, AND L. M. WHEELER.
1992. Habitat choice and predator avoidance in
tadpoles. J. Herpetol. 26: 142-146.

SAIDAPUR, S. K. 1989. Reproductive Cycles of
Indian Vertebrates. Allied Publishers, New
Delhi.

WIENS, J. A. 1972. Anuran habitat selection: Early
experience and substrate selection in Rana cas-
cadae tadpoles. Anim. Behav. 20: 218-220.

ULTSCH, G R., D. F. BRADFORD, AND J. FREDA.
1999. Physiology. Coping with environment. p.
189-214. In: R. W. McDiarmid and R. Altig
(eds.), Tadpoles: The Biology of Anuran Larvae.
The Univ. Chicago Press, Chicago.

61

Sri Lanka: December, 2001

wai a
NGRESS OF HERPETOLOGY

We are happy to inform you that the Fourth World Congress of Herpetology will be held
from 2nd to 9th December 2001, in the BMICH, Colombo Sri Lanka. This is the first signifi-
cant herpetology event of the new millennium and the organizing committee is determined to
make this the most memorable herpetological forum of the century.

The Registration-Full participant US $ 350, Students US $ 250, and Accompany persons US
$ 200. A surcharge of US $ 100 will be applied after 1st September 2001.

Deadline for abstracts-31 August 2001.

Registration entitlements for delegates-In addition to attendance at all formal sessions, con-
gress registration will be entitle you to:

Airport transfers (both ways)

Daily shuttle service from hotel to congress venue

Conference kit, which includes: Label badge, Special 4WCH pen, Special 4WCH note pad,
4WCH picture post cards, Sri Lanka herp stickers, Welcome cocktail, Two coffee breaks with
snacks on each day of the congress, Lunch on all congress days, Congress banquet, Full day
professional excursion, Abstract book and the programme and many more.

We have reserved special discount rate in leading five star hotel - shared room with break-
fast is only US $ 35 on wards.

We will be mailing you the congress Brochure and Registrations forms.

Sri Lanka is a herpetological hotspot in the world. In addition to upward 150 species of
amphibians, Sri Lanka is also home to diverse reptiles: Five species of marine turtles, chelo-
nians, crocodiles, varanids, agamids, skinks, lacertids, chameleons, geckos, and snakes (from
primitive uropeltids to modern vipers) inhabit our small island of 64,742 square km in area.

I look forward to welcoming you in Sri Lanka in December 2001.

REGISTER TODAY!

Pre-registration & information: Congress organizer/director:
4WCH Promotions Office, Anslem de Silva,
95 Cotta Road, Faculty of Medicine,
Colombo 8, University of Peradeniya,
SRI LANKA. Peradeniya,

SRI LANKA.
E-mail: admin@4wch.com Fax: (+94 8) 389 106

Internet: http://www.4wch.com E-mail: director@4wch.com

62

Instructions to Contributors

Manuscripts in English should be typed on sturdy typing paper of A4 size (28 x 21 cm).
Typing should be double-spaced with 2.5 cm margins on all sides. Words should not be broken
at the end of a line. Full papers should not exceed 12 printed pages, and short communications
should not exceed 2 printed pages, Longer papers may be printed, on condition that the author
pays the excess page charges in full. Each of the following divisions should be begun on a sep-
arate page: cover page, title page, abstract, main text, references, each table, figure legends,
each figure. The cover page should show the title, name of author and addresses to whom
communications should be directed with his or her telephone and fax numbers, and E-mail
address, the date of submission, the number of pages of main text, the number of figure leg-
ends, number of figures, and number of tables.

The title page should show the title, the name(s) of the author(s), and their affiliation(s) and
addresses. In the case of several authors, indicate the one to whom communications should be
directed, his or her telephone and fax numbers, and E-mail address. The abstract page should
contain an abstract of about 200 words and five keywords. The main text should consist of an
Introduction, Materials and Methods, Results, Discussion, and Acknowledgements. Format
for text, citations, and literature cited should follow those used in the most recent issue of the
journal. Short reports should follow the format of full papers, including the use of subtitles for
Materials and Methods, etc. Each table should be on a separate page. Line drawings should be
such that they can be used directly for printing. Figures larger than 28 x 21 cm cannot be
accepted. When several drawings or photographs are to be reproduced as one figure, they
should be mounted on cardboard in the desired arrangement. Figure legends should be collec-
tively typed on a separate page. Authors should consult a recent issue of the journal regarding
style.

After the acceptance of the manuscript, the author is requested to send original figures and a
floppy disk containing exactly the same contents as the final version of the manuscript.

Contributors should send three clear copies of all material (text, figures, tables) to the Man-
aging Editor. Originals should be retained until notification of acceptance is received. Copies
of photographs and line cuts should be as clear as the originals. All papers will be refereed by
competent persons.

Reprints can be ordered in multiples of 50. The number of desired reprints and whether cov-
ers are desired should be indicated in red on the first page of the galley proofs. Only the first
set of the galley proofs will be sent to the author. Proofs should be corrected promptly and
returned without delay. Only typographical errors should be corrected.

CURRENT HERPETOLOGY

MANAGING EDITOR

Hidetoshi OTA

: Tropical Biosphere Research Center, University of the Ryukyus, Nishihara,
Okinawa, 903-0213 JAPAN
(ota@sci.u-ryukyu.ac.jp)

ASSOCIATE EDITORS

aig ADLER, Department of Neurobiology and Behavior, Cornell University, Seeley G.
_ Mudd Hall, Ithaca, New York 14853-2702, USA (kka4@cornell.edu)

\aron M. BAUER, Department of Biology, Villanova University, 800 Lancaster Avenue, Vill-
a anova, PA 19085 USA (aaron.bauer@villanova.edu)

lya S. DAREVSKY, Zoological Institute, Russian Academy of Sciences, St.Petersburg
~ 199034 RUSSIA (Darevsky@herpet.zin.ras.spb.ru)

_ Indraneil DAS, Institute of Biodiversity and Environmental Conservation, Universiti Malaysia
_ Sarawak, 94300, Kota Samarahan, Sarawak, MALAYSIA (idas@mailhost.unimas.my)
Richard C. GORIS, Hatsuyama 1-7-13, Miyamae-ku, Kawasaki 216-0026 JAPAN (goris@

__ twics.com)
e Ivan INEICH, Laboratoire des Reptiles et Amphibiens, Museum National d'Histoire Naturelle,
ae 25 rue Cuvier 75005 Paris, FRANCE (ineich@cimrs1.mnhn.fr)
rs Ulrich JOGER, Hessisches Landesmuseum Darmstadt, Zoologische Abteilung, Friedensplatz
: ee 1, D-64283 Darmstadt, GERMANY (u.joger@hlmd.tu-darmstadt.de)

ee Tamotsu KUSANO, Department of Biological Science, Graduate School of Science, Tokyo
< e + Metropolitan University, Minami-ohsawa, Hachioji, Tokyo 192-0397 JAPAN (tamo@
a comp.metro-u.ac.jp)

== Masafumi MATSUI Graduate School of Human and Environmental Studies, Kyoto University,
‘Yoshida Nihonmatsu-cho, Sakyo-ku, Kyoto 606-8501 Japan (fumi@zoo.zool.kyoto-
u.ac.jp)
Colin McCARTHY, Department of Zoology, The Natural History Museum, Cromwell road,
& _ London SW7 5BD, UK (cjm@nhm.ac.uk)

_ Michihisa TORIBA, Japan snake Institute, Yabuzuka-honmachi, Nitta-gun, Gunma 379-2300
a JAPAN (snake-a@sunfield.ne.jp)

: St
ft

if a ‘ees
fa hates
ad Ae Ke
‘ Moe
rv
gr BF eh

Home Page of THE HERPETOLOGICAL SOCIETY OF JAPAN
http://zoo.zool.kyoto-u.ac.jp/~herp/

DATE OF PUBLICATION
Current Herpetology, Vol. 19, No. 2,
was mailed 26 December 2000.

- 6 ae aie

i. rab
a. 7 . INSTITUTION LIBRARIES

9088 012

CONTENTS

Originals

Dispersal of brown frogs Rana japonica and R. ornativentris VS Oe aa
in the forests of.the’Tama Hills <=. "3772-7 Satoshi Osawa and Take 0 Kat :

Y “Karyotype of the Chinese soft-shelled turtle, Pelodiscus sinensis,
from Japan and Taiwan, with chromosomal data for Dogania subplana

‘C’ Absence of lines of arrested growth in overwintered tadpoles of the American |
catesbeiana (Amphibia, Anura) vv" Wichase Khonsue and Mas:

FUTURE MEETING
Niigata University, Niigata, Japan, 10-11 November 2001
(Kunio Sekiya, Chair)