¥

LIBRARY

Michigan State
University

fi

PLACE N RETURN BOX to remov- thb Mom from your "card.
TO AVOID FINES mum on or baton dd. duo.

DATE DUE DATE DUE DATE DUE
I ' .‘ ‘

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

—-——_—_—__—

A SYSTEMATIC AND BIOGEOGRAPHIC STUDY OF
THE BAT GENUS RHINOLOPHUS
(CHIROPTERA: RHINOLOPHIDAE)

By

Yining Luo

A DISSERTATION

Submitted to
Michigan State University
in partial fulfillment of the requirements
for the degree of

DOCTOR OF PHILOSOPHY

Department of Zoology

1995

ABSTRACT

A SYSTEMATIC AND BIOGEOGRAPHIC STUDY OF
THE BAT GENUS RHINOLOPH US
(CHIROPTERA: RHINOLOPHIDAE)

By

Yining Luo

The phylogenetic relationships among the Old World horseshoe bats, genus
Rhinolophus, are studied using morphometn'c and cladistic analyses. Twenty-seven skull
features and 15 skin features were measured from 1120 skull and 668 skin specimens,
representing 60 rhinolophid species. Principal components of both correlation matrix and
covariance matrix of the data were analyzed using multivariate procedures. The pattern of
species along principal components does not indicate the traditional views on the species
groups, but displays a separation of species of Africa and west Eurasia from those of
southeast Asia. Within species of southeast Asia, similarities exist among the members of
the traditional arcuatus, philippinensis, and pusillus groups.

the information contents and transformation series of 26 morphological characters
of Rhinolophus were examined for cladistic analyses by Wagner parsimony using PAUP
3.1.1. The most parsimonious cladograms strongly suggest four monophyletic groups: the
traditional philippinensis group, the traditional philippinensis group plus arcuatus group,
all rhinolophids of southeast Asia, and three African members of the traditional fumigatus

group. A monophyletic group consisting of southeast Asian members of the traditional

pusillus group is weakly suggested. Based on these monophyletic groups, a subgenen'c
taxonomy of Rhinolophus is proposed.

A cladistic biogeographic study of Rhinolophus in southeast Asia suggests a
progressive subdivision of the areas with distance from the continental Asia. The
Australian realm, as defined by Huxley’s line and Webber’s line of fauna] balance,
represents a monophyletic area group, whereas the Oriental realm represents a
paraphyletic area group. Both the phylogeny and the historical distribution of Rhinolophus

indicate an African origin of the genus.

ACKNOWLEDGMENTS

I thank the following persons and institutions for providing the specimens used in
this study: K. Koopman and American Museum of Natural History, D. A. Schlitter and
Carnegie Museum of Natural History, B. Patterson and Field Museum of Natural
History, M. Rutzmoser and Museum of Comparative Zoology, L. Gorden and National
Museum of Natural History. I thank J. Smith and J. Jankins for their help in laboratory
work. Particular thanks go to my major advisor, D. O. Straney. Without his enormous
help, this work would not be completed as is. I would like to thank the members of my
Committee, R. Snider, J. A. Holman, D. Hall, and G. Bush for their help and

encouragement.

iv

CONTENTS

INTRODUCTION ............................................ 1
MRPHOMETRIC AND PHYLOGENETIC STUDY OF

RHINOLOPH US .................................. 19

Material and Methods ................................... 19

Results ............................................... 29

MORPHOMETRIC ANALYSIS ......................... 29

A. Pooled skull and Skin data ....................... 29

B. Skull data set ............................... 38

C. Skin data set ............................... 41

D. Remarks on the principal component analysis .......... 44

E. Canoical discriminant analysis ...................... 47

CHARACTER ANALYSIS ............................. 55

CLADISTIC ANALYSIS .............................. 86

Analysis of Weighted Characters, Character 1 Ordered ...... 88

Analysis of Weighted Characters, Character 1 Unordered ..... 96

Analysis of Unweighted Characters, Character 1 Ordered .. . .100
Analysis of Unweighted Characters, Character 1 Unordered . . 104

The status of the fumigatus group ................... 105
Comparisons Between the Analyses .................. 107
Taxonomic Summary ........................... 111
Discussions .......................................... 1 17

HISTORICAL BIOGEOGRAPHY OF THE SOUTHEAST ASIAN

RHINOLOPH US .................................... 122
Introduction .................................. 122

Material and Methods ............................ 129

Results ..................................... 135

Discussion .................................... 139

APPENDIXCES ............................................ 145
BIBLIOGRAPHY ........................................... 1 6O

LIST OF TABLES

Table 1.1. The diagnostic features for the species groups of Rhinolophus used by Corbet
and Hill. (From Cortet and Hill, 1992). Symbols ‘+’, ‘-’ and ‘+/-’ indicate ‘present’,
‘absent’, and ‘may be present’ of the particular character state in the groups respectively.

Table 2.1: The descriptions and abbreviations of the skin measurements.

Table 2.2: Descriptions and abbreviations of the skull measurements. Numbers correspondent
to the labels in illustrations inFigures 2.1 and 2.2.

Table 3.1: The abbreviations and the traditional group identities of each species used in the
display of their principal components. f: ferrumequinum group; p: pusillus group; a:
wcuatus group; h: hipposideros group; 1: philippinensis (= Iuctus) group; m: fitmigatus (=
macrotis) group. The species groups were originally defined by Anderson (1905b, 1918),
and were modified by Tate and Archibold (1939), Koopman (1975), and Corbet and Hill
(1992).

Table 4.1. The character transformation series matrix used in the cladistic analysis.
Numeric numbers 1, 2, are correspondent to the states a, b, in the section of
character analysis respectively. Missing data are represented by ‘7’.

Table 4.2. Summary of taxonomic conclusions based on the monophyletic groups in Figure
4.28. No paraphyletic groups is recognized in this taxonomy. Monophyletic groups of species
are recognized at three different levels (supergroup, group, and subgroup). Those species that
can not be placed into a monophyletic group are included as ‘status uncertain’ at the
appropriate level.

Table 4.3. Diagnostic characters for the infiageneric taxa of Rhinolophus based on the
monophyletic groups illustrated in Figure 4.27. Species with phylogenetic relationships unclear
are treated as status uncertain at appropriate levels.

Table 4.4 : A comparison in the patterns of transformation between the weighted and
unweighted analyses, character 1 unordered, for each character. In each analysis, one
shortest cladogram, which has a topology identical to the majority consensus of that
analysis, is summarized. Shading indicate the characters with lower occurrence of
homoplasy ratio.

Table 5.1. The distribution of Rhinolophus species in the 11 areas of the southeast Asia region.
The abbreviations for the area names are: Cont = continental southeastern Asia including India,
southern China, and the adjacent major islands including Taiwan; IndC = Indochina including
Burma, Thailand, Cambodia, Vietnam and Laos; Maly = the Malay Peninsula; Sumt =

Sumatra; Bone = Borneo; Java = Java; SulT = Sulawesi and Timor, Mulk = the Maluku
Islands; Phil = the Philippine Islands; NewG = New Guinea; Aust = Australia

Table 5.2. Data matrix for rhinolophid distributions in the 11 areas of southeast Asia.
Characters 1-24 are based on the distributional data of individual Species (listed in Table
5.1). Characters 25-36, listed below, are based on components of relationships from the
majority consensus cladograms of all four cladistic analyses. All components that are
common to at least two analyses and are not in conflict with other cladograrns were
selected.

vii

LIST OF FIGURES

Figure 1.1: The world distribution of the bat family Rhinolophidae (shaded area). (From
Koopman, 1984.)

Figure 1.2: The features of the rhinolophid head, with emphasis on the noseleaves. (Alter
Lekagul and McNeely, 1977).

Figure 1.3: The morphology of the rhinolophid skull (From Rosevear, 1965).

Figure 1.4. the list of species groups proposed by Andersen (1905b, 1918). The group
names that are renamed by later researchers are indicated in the parenthesis.

Figure 1.5. The relationships among the species of the femmrequinum group proposed by
Andersen. Species of Afiica are indicated by ‘*’. (From Andersen 1905a, p120.)

Figure 1.6. The phylogenetic hypothesis proposed by Tate and Archibold (1939). The
simplex group and luctus group are presently referred to as the ferrumequinum group and
philrppinensis group respectively.

Figure 1.7. The 11 species groups proposed by Bogdanowicz (1992). (From
Bogdanowicz, 1992. The question marks indicate either that the species status is
questionable or that the assignment of the species into species group is not conclusive.)

Figure 1.8. The relationships among the 11 species groups proposed by Bogdanowicz
(1992). (After Bogdanowicz, 1992.)

Figure 2.1: Lateral views of cranium and lower jaw with the landmarks and the
measurements illustrated. Labels for measurements correspond to the measurements and
abbreviations listed in Table 2.1.

Figure 2.2: Ventral view of cranium with the landmarks and the measurements illustrated.
Labels for measurements correspond to the descriptions and abbreviations listed in Table
2.2.

Figure 3.1. Displays of species on PCl and PC2 of the correlation matrix from the pooled
skin and skull data. (a). The separation between the traditional species groups is not
clear. Species are symbolized in species group identity: f: fen-umequinum group; p:
pusillus group; a: arcuatus group; h: hipposideros group; I: philippinensis (= Iuctus)
group; m: fumigatus (= macrotis) group. (b) There is a approximate separation of species
associated to the geographic origins (dotted line). Species are symbolized in their
distribution: ‘ ’ = southeast Asia, ‘F’ = Afiica, and ‘E’ = west Eurasia.

viii

Figure 3.2. The display of traditional ferrumequinum species group in PC2 and PC3, fi'om
the correlation matrix of the pooled skin and skull data. There is extensive overlap
between species of this group and other groups. Species group abbreviations: f:
ferrumequinum group; p: pusillus group; a: arcuatus group; h: hipposideros group; I:
philippr’nensis (= Iuctus) group; m: fitmigatus (= macrotis) group.

Figure 3.3. The display of the traditional pusillus species group and arcuatus species
group in PC2 and PC3 from the correlation matrix of the pooled skin and skull data. There
is extensive overlap between species of the pusillus group and other groups, but only one
species of the firmigatus group is inside the amuatus group. Species group abbreviations:
f: femmrequinum group; p: pusillus group; a: ar'cuatus group; h: hipposideros group; I:
philippinensis (= luctus) group; m: fumigatus (= macrotis) group.

Figure 3.4. The display of traditional fitmigatus (= macrotis) group and philippinensis (=
Iuctus) group in PC2 and PC3 from the correlation matrix of the pooled Skin and skull
data. Extensive overlap exists between species of the fumigatus group and other groups.
Species group abbreviations: f: fermmequinum group; p: pusillus group; a: arcuatus
group; h: hipposideros group; 1: philippinensis (= luctus) group; m: fumigatus (=
macrotis) group.

Figure 3.5. Figure displays species on the PC2 and PC3 from the correlation matrix of
pooled skin and skull data. There is a non-overlapping separation of species clusters
associated with their geographic origins: one species cluster from Africa and west Eurasia,
and the other from southeast Asia. Species are symbolized in their distribution ‘A’ =
southeast Asia, ‘F’ = Africa, and ‘E’ = west Eurasia.

Figure 3.6. Figure displays species on the PC2 and PC3 fiom the covariance matrix of
pooled skin and skull data. Species of Africa and west Eurasian form a distinct cluster.
Species are symbolized in their distribution ‘A’ = southeast Asia, ‘F’ = Afiica, and ‘E’ =
west Eurasia.

Figure 3.7. Figure displays species on the PC2 and PC3 from the correlation matrix of
pooled skin and skull data. The traditional species groups are more distinct when only
those fi'om southeast Asia region is considered. Species are symbolized in their taxonomoc
group: f: ferrumequinum group; p: pusillus group; a: arcuatus group; h: hipposideros
group; 1: philippinensis (= Iuctus) group; m: firmigatus (= macrotis) group.

Figure 3.8. Figure displays species on PC2 and PC3 fi'om the correlation matrix of skull
data. Species of Afiica and west Eurasia are separated from the species of southeast Asia.
. Species are symbolized in their taxonomoc group: f: ferrumequinum group; p: pusillus
group; a: arcuatus group; h: hipposideros group; I: philzppinensis (= Iuctus) group; m:
fumigatus (= macrotis) group.

Figure 3.9. Figure displays species on PC2 and PC3 from the covariance matrix of skull
data. Species of Africa and west Eurasia are separated from the species of southeast Asia.

ix

 

] ‘l’lu. s

Two arrows indicate two misplaced species. Species are symbolized in their distribution:
‘A’ = southeast Asia, ‘F’ = Afiica, and ‘E’ = west Eurasia.

Figure 3.10. Figure displays species on the PC2 and PC3 from the correlation matrix of
skull data. Among the southeast Asian species, only members of the traditional mutants
group are close to each other. . Species are symbolized in their taxonomoc groups: f:
femmrequinum group; p: pusillus group; a: arcuatus group; h: hipposideros group; 1:
philippinensis (= Iuctus) group; m: fitmigatus (= macrotis) group.

Figure 3.11. Figure displays the first and second canonical variables (CAN 1 and CAN2)
for the southeast Asian species. Traditional species groups are used as a priori class.
Species are symbolized in their group identity: f: ferrumequinum group; p: pusillus
group; a: arcuatus group; h: hipposideros group; I: philippinensis (= Iuctus) group; m:
fianigatus (= macrotis) group.

Figure 3.12. Figure displays the first and second canonical variables (CAN 1 and CAN2)
for the southeast Asian species, Andersen’s macrotis group being merged with
philippinensis group. Traditional species groups are identified as a priori class. Species
are symbolized in their group identity: f: ferrumequinum group; p: pusillus group; a:
arcuatus group; b: hipposideros group; 1: philippinensis (= Iuctus) group; m: fitmigatus (=
macrotis) group.

Figure 3.13. Figure displays the first, second and fifth canonical variables (CAN 1, CAN2
and CAN 5) for all species of Rhinolophus. Traditional species groups are used as a priori
classes for southeast Asian species only, and all African and west Eurasian species are
assigned to a new class labeled ‘W’. Southeast Asian species are symbolized in their group
identity: f: fen'umequinum group; p: pusillus group; a: arcuatus group; h: hipposideros
group; 1: philippinensis (= lucnts) group; m: fitmigatus (= macrotis) group.

Figure 3.14. Figure displays of the third and fourth canonical variables (CAN 3 and
CAN4) for all species of Rhinolophus. Traditional species groups are used as a priori
class. Species are symbolized in their group identity: f: fenwnequinum group; p: pusillus
group; a: arcuatus group; h: hipposideros group; 1: philippinensis (= Iuctus) group; m:
fumigatus (= macrotis) group.

Figure 3.15. Figure displays the first, third and fourth canonical variables (CAN 1, CAN 3
and CAN4) for all rhinolophids. The fermmequinum group and filmigatus group are
separated on CAN 1 axis. Traditional species groups are used as a priori classes. Species
are symbolized in their group identity: f: ferrumequinum group; p: pusillus group; a:
arcuatus group; h: hipposideros group; I: philippinensis (= Iuctus) group; m: filmigatus (=
macrotis) group.

Figure 4.1: The Shape of the connecting process of the noseleaf (character 1) in lateral
view, pointed by arrow. (a) state a, height moderate and round (R. aflinis); (b) state b,
higher and shaper (R. macrotis); (c) state c, very low (R. Iuctus); (d) state d, hom-
shaped (R. pusillus); (e) state e, anterior base reaches the tip of the sella (R pearsom‘).

Figure 4.2. The shape of the sella (character 2), pointed by arrow "1“. (a) state a, narrow
(R alcyone); (b) state b, broader (R funigatus); (c) state c, with lappet, pointed by arrow
‘¢‘ (R. maclaudi). (From Rosevear, 1965).

Figure 4.3. Illistrations of the horseshoe (character 3). (a) state a, narrower (R cIivosus);
(b) state b, broader (R filmigatus). (F rom Rosevear, 1965).

Figure 4.4. Illustrations of the supplementary noseleaf (character 4). (a) state a, not
present (R. luctus); (b) state b, less developed (R pusillus); (c) state c, both sides meet at
the mid-line (R simulator). The dotted line indicates the horseshoe which usually covers
most part of the supplementary noseleaf.

Figure 4.5. Number of the lower lip grooves (character 5), pointed by arrow. (a). state a,
one groove; (b). state b, three grooves.

Figure 4.6. Illustrations of the front ear projection (character 6). (a) state a, not present;
(b) state b, present.

Figure 4.7. The shape of the lancet (character 7). (a) state a, hastate; (b) state b, nearly
triangle. Long hair, shown in B, are frequently present in the lancet. (fi'om Rosevear,
1965)

Figure 4.8. The insertion of the plagiopatatgium (character 10). (a) state a, at the ankle (R
megaphillus); (b) state b, above the ankle (R rufiis); (c). state c, near the tarsal-
metatarsal joint (R Iuctus). (After Rosevear, 1965).

Figure 4.9. The status of P2 (character 11), pointed by arrow. (a) state a, in the toothrow
(R ions) ; (b) state b, out of toothrow (R ferrumequinum); (c) state 0, absent (R
fumigates).

Figure 4.10. The shape of P2 (character 12), pointed by arrow. (a) state a, the length and
breadth bout equal (R mehelyi); (b) state b, length greater than breadth (R. clivosus); (c)
state c, length less than breadth (R ions).

Figure 4.11. The status of P3 (character 13), pointed by arrow. (a) state a, in the toothrow
(R macrotis); (b) state b, out of toothrow (R malayanus); (c) state c, absent (R

fitmigates).

Figure 4.12. The shapes of the stylarshelf in M3 (character 15). (a) state a (R qffinis); (b)
state b, the posterior v-shaped ridges (pointed by arrow) greatly reduced (R filmigarus).

Figure 4.13. Picture (R afiinis) illustrates character 13, the position of the posterior
margin of the palate pointed by an arrow. Label 0 through 3 correspond to the states a
through din the text.

Figure 4.14. The position of the anterior margin of the palate (character 17), pointed by
arrow, of R affinis. Label 0 through 2 correspond to the states a through c in the text.

Figure 4.15. The position of the front margin of anterior nasal swelling (character 18),
pointed by arrow, of R affinis. Label 0 through 2 correspond to the states a though c in
the text.

Figure 4.16. Pictures illustrate character 19, the length of median frontal nasal swellings.
(a) state a, small (R ajfinis); (b) state b, larger (R luctus).

Figure 4.17. The depth of the orbital constriction (charcter 20), pionted by arror. (a). state
a, Shallow (R lepidus); (b). state b, deep (R creaghi).

Figure 4.18. The shapes of the infraorbital canal and bar (character 21), pointed by
arrows. (a) state a, Size moderate (R affinis); (b) state b, infraorbital bar elongated (R
clivosus); (c) state c, canal lengthened and bar broader (R Iuctus).

Figure 4.19: The strict consensus cladograrn for the 24 most parsimonious cladograms
resulting from the weighted analysis, character I ordered. . The geographic location of the
species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ = Afiica, and ‘E’ =
west Eurasia.

Figure 4.20: The majority consensus cladograrn for the 1200 most parsimonious
cladograms resulting from the weighted analysis, character I ordered. The geographic
location of the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ =
Africa, and ‘E’ = west Eurasia. The numbers indicate the percentage of cladograms in
which this particular branching structure is present.

Figure 4.21: The strict consensus cladograrn for the 1300 most parsimonious cladograms
resulting from the weighted analysis, character 1 unordered. . The geographic location of
the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ = Africa, and
‘E’ = west Eurasia.

Figure 4.22: The majority consensus cladograrn for the 1200 most parsimonious
cladograms resulting from the weighted analysis, character 1 unordered. The geographic
location of the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ =
Afiica, and ‘E’ = west Eurasia. The numbers indicate the percentage ‘ of cladograms in
which this particular braching stucture is present.

Figure 4.23: The strict consensus cladograrn for the 1300 most parsimonious cladograms
resulting from the unweighted analysis, character 1 unordered. . The geographic location
of the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ = Afiica,
and ‘E’ = west Eurasia.

xii

Figure 4.24: The majority consensus cladograrn for the 1200 most parsimonious
cladograms resulting from the unweighted analysis, character I ordered. The geographic
location of the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ =
Afiica, and ‘E’ = west Eurasia. The numbers indicate the percentage of cladograms in
which this particular braching stucture is present.

Figure 4.25: The strict consensus cladograrn for the 1300 most parsimonious cladograms
resulting from the unweighted analysis, character 1 unordered. The geographic location of
the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ = Africa, and
‘E’ = west Eurasia.

Figure 4.26: The majority consensus cladograrn for the 1200 most parsimonious
cladograms resulting from the weighted analysis, character I ordered. The geographic
location of the species is indicated in the letters in parenthesis: ‘A’ = southeast Asia, ‘F’ =
Africa, and ‘E’ = west Eurasia. The numbers indicate the percentage of cladograms in
which this particular braching stucture is present.

Figure 4.27. Cladograms illustrate the consensus between the results from the weighted
and the unweighted analyses. (a) Results from the weighted analyses, African and west
Eurasian species branch from the base of the cladograms; (b) Results from the unweighted
analyses, Afiican and west Eurasian species constitute a monophyletic group; (c) In the
consensus cladograrn for (a) and (b), Afiican and west Eurasian species as well as the
monophyletic group of southeast Asian species branch from the multichotomous root.

Figure 4.28. The phylogenetic relationships within the genus Rhinolophus based on the
present study. The monophyletic groups (bold faced) strongly supported by my data set are
indicated by solid lines. A dotted line represents a set of species branching from that point;
relationships among these species are unresolved.

Figure 5.1: Southeast Asia. (After Hutchinson, 1989).

Figure 5.2. Fauna] boundaries suggested within the south-east Asia region. Line A, Huxley
(1868); Line B, Wallace (1860); Line C, Pelseneer (1904, Weber's line of fauna] balance);
Line D, Lydekker (1896); Line E, Gressitt (1956); Between line A and line D, Tate's
(1946) 'Wallacean region'; Between line C and line E, Gressitt's (1956) 'Papuan region'.
(After Holloway and Jardine, 1968).

Figure 5.3: The dendrogram calculated fi'om the coefficients of faunal dissimilarities
among the areas of southeast Asia for butterflies (After Holloway and Jardine, 1968).

Figure 5.4: The dendrogram calculated fi'om the coefficients of faunal dissimilarities
among the areas of southeast Asia for birds (After Holloway and Jardine, 1968).

Figure 5.5: Dendrogram computed fi'om the coefficients of fauna] dissimilarities between
the areas of southeast Asia for bats (After Holloway and Jardine, 1968).

xiii

Figure 5.6: The concensus dendrogram for the degrograrns based on the distributional
data of birds, butterflies and bats for southeast Asia (After Holloway and Jardine, 1968).

Figure 5.7. The two most parsimonious area cladograms computed fi'om the distributional
data of Rhinolophus in southeast Asia.

Figure 5.8. The consensus cladograrn for the two most parsimonious area cladograms
computed from the distributional data of Rhinolophus in southeast Asia.

Figure 5.9. One of the most parsimonious cladograms of southeast Asia based on rhinolophid

distributional data (Figure 5.7 a). When unrooted, it supports both Huxley’s line and the Line
of Fauna] Balance.

xiv

INTRODUCTION

Bats of the genus Rhinolophus, the horseshoe bats, constitute the only living genus of
the family Rhinolophidae (Marnmalia: Chiroptera). This genus contains about 60 species
distributed throughout the Old World tropical and warm temperate areas, including Afiica, the
southernEurasian confinerrttheislandsofsomheastAsiaNewGuineaandnonheI-nAustralia
(Corbet and Hill, 1981 and 1992; Honacki et al, 1982; Koopman, 1992)(Figure 1.1). While
some Species, such as R femimequinum and R cIivosus have transcontinental distribution,
many species, especially those in southeast Asia, are known only from very limited areas.

The Rhinolophidae is one of the oldest living bat families. R priscus is the earliest
representative of the family, known [Tom the Upper Eocene in France where it co-occurs with
bats of the families Hipposideridae, Vespertilionidae and Emballonuridae (Savage and Russell,
1983). Fewer than 20 species of fossil Rhinolophus have been described and their distribution
in time and space is spotty. The oldest rhinolophid fossil from Australia dates fiom the Middle
Mocene while the oldest fossil fiom Asia, the likely source of Australian species, is known
only fi'om the Pliocene (Koopman and Jones, 1970; Hall, 1989). Due to their habitats of
roosting in caves and hollow trees and foraging away fiom more common deposit sites such as
streams and lakes, bats are much less likely to be preserved as fossils than most other mammals
(Dawson and Krishtalda, 1984). Bats are the only mammalian order in which fewer fossil
species than living species are described. Our understanding of rhinolophid fossil history is

consequently very incomplete.

Sea Seneca gene Asa @833 SEESE beam as 2e a Susana Es; 2:. a ._ 03mm

83b—30J >I‘u Bab—go 5".

nguat ¢Oh<0lu1

53:80. 5 .2! Mn Hi
and...“ g 89. w

argon 20 wig

 

3
It is generally agreed that the genus Rhinolophus is a monophyletic group. Members of

the genus Share a unique external feature, the horseshoe-shaped noseleaf (Figure 1.2). The
noseleaf has three main parts. The anterior leaf, or horseshoe, is a horseshoe-shaped noseleaf
covering the upper lip and surrounding the nostrils; the sella is a thick median projection dorsal
to the nostrils; and the Lancet, or posterior leaf is the dorsal-most part of the noseleaf with a
tapered tip and two or three paired lateral ridges. Between the lancet and the sella, there is a
m m in the mid-sagitta] plane. Individual species display variations on this
common ground plan. Some species have lateral extensions of the sella, called lappets. Behind
the anterior leaf, some species have an additional piece of noseleafi the accessory leaf, that is
usually completely covered by the anterior noseleaf The intemarial septum between the nostrils
varies in size and shape. The skulls of Rhinolophus are readily distinguishable fi'om skulls of
other families by their nasal swellings and basal region (Figure 1.3). The nasal swellings are
inflated nasal bones giving support to the noseleaves. Four or six swellings are usually
recognizable and the anterior swellings are ofien higher than the posterior ones. The basal
region of Rhinolophus skulls is distinctive in the presence of a pair of large and exposed
cochlea The auditory bulla attaches to the anterior-lateral side of the cochlea. Two pairs of
upper incisors are present, but as both nasal and maxillary bones are deeply invaginated in the
front, the premaxillary bones connect the maxillary only with a narrow bend of cartilage at their
posterior end. Other features that define the genus include the absence of the tragus and the
presence of large antitragus on the ears, absence of the postorbital process, and absence of the
first phalanx in the second finger.

Students of bat phylogeny agree that hipposiderids, the Old World leaf-nosed bats, are

the closest living relatives of rhinolophids (Van Valen, 1979; Koopman, 1984). The

.952 £820:

23 Ewing 53¢ $96033: ofi no magma». a?» 68: vflmofioflfi 23° «033m 2? “a.“ 35mm

8028305
.8230: 83:3

 

 

Figure 1.3: The morphology of the rhinolophid Skull (From Rosevear, 1965).

6

family Hipposideridae contains about 60 species of nine genera; all have complex leaflike
outgrowths of skin on their muzzle. The noseleaves of the hipposiderids include an anterior
leaf, sometimes one or more accessory leaflets, and an erect transverse leaf. Although the
noseleaves of the two families are very different in shape, some researchers have homologized
the anterior leaf and erect transverse leaf of hipposiderids to the horseshoe and the lancet of
rhinolophids. Hipposiderids lack a sella The hipposiderids differ fi'om the rhinolophids also in
having two, instead of three, phalanges in each toes, in lacking P3 and in details of the
structures of the shoulder and girdles.

There are disagreements about the taxonomic relationship of hipposiderids and
rhinolophids. Some authors believe that Hipposideridae should be classified as a subfamily of
Rhinolophidae (Corbel, 1978; Ellennan and Morrison-Scott, 1966; Koopman, 1970, 1984 and
1992); other workers maintain that the two are distinct families (Miller, 1907, Walker, 1964;
Corbet and Hill, 1981 and 1992; Hayman and Hill, 1971). The debate is purely taxonomic and
depends on each author’s family concept. Nomenclatorical controversy should not obscure a
general agreement that hipposiderids are the sister group of rhinolophids.

Since Rhinolophidae is a monotypic family, I will refer to rhinolophids as a genus when
I discuss intrageneric phylogeny and as a family when I compare them with bats of other

families.

Previous studies on the systematics of rhinolophids
Among the earliest systematic studies of the genus, Andersen's work (1905a, 1905b,

1905c, l905d, 1905c, 1918) was the most important and has been the foundation for all

7
subsequent work. Based primarily on the shapes of nose-leaves, premolar dentition, nasal-

swellings, palate bridge of the cranium and the size of cochlea, be assigned the species to six
species groups: simplex group (renamed to fen'wnequiman group by later researchers;
indicated as ‘=ferrumequimmr ’ below), lepidus ( = pusillus) group, arcuatus group, macrotis
(= fianigatus ) group, hnpposiderm group and luctus ( = philippinensis) group (Figure 1.4).
With the exception of the arcuatus group, which is found only in southeast Asia, Andersen’s
species groups contain species distributed in Asia, Europe and Africa Andersen identified the
similarities between several Afiican and Asian species. He suggested that this resemblance
evidenced a close relationship and, in some cases, parallel evolution between the corresponding
species (Andersen, 1905a). Figure 1.5 displays Andersen’s view of relationships among species
of his fermmequiman group. For this group, as in the others, he concluded that southeast
Asian species almost always had more primitive features than species from Afiica. He
concluded that all the Ethiopian species of the genus are of Oriental origin.

In his studies, Andersen identified the following features in Rhinolophus as primitive
conditions: connecting process low; mental grooves three; front nasal swellings low; sagitta]
crest low; palatal bridge not shortened; P2 and P3 in the tooth row; basisphenoid not narrowed;
ternpora] fossa narrow; metacarpals about equal in length; ratio of 2nd to lst phalanges of the
third and the fourth fingers small. Unfortunately, Andersen did not indicate explicitly the
relationships among his species groups (Andersen 1905a, 1905b, 1905d).

Tate and Archbold (1939) revised the rhinolophid species of the Indo-Australian region
(Figure. 1.6). Although they used most of Andersen's characters for their group and subgroup

identification and their phylogenetic analysis, Tate and Archbold had reservations about the

simplex
(ferrumequinum) group

simplex
megaphyllus
keyensis
bomeensis
celebensis
malayanus
Virgo
nereis
stheno
simulator
denti
rouxi
thomasi
capensis
affinis
clivosus
darlingr'
ferrumequinum
deckenii

lepidus (pusillus) group
lepidus
acuminatus
pusillus
comutus
gracilis
subbadius
monoceros
blasii
landeri

euryale
mehelyi

midas (hipposideros) group
hipposideros

philippinensis (luctus) group
philippinensis
mitratus
macloudi
sedulus - -
mfoliatus
luclus

macrotis (fumigatus) group
macrotis
hirsulus
filmigatus
eloquens
hildebrandti
pearsoni

arcuatus group
arcuatus
submfils
inops
creaghi
coelophyllus
ewyotis

Incertae sedis
alcyone

Figure 1.4. the list of species groups proposed by Andersen (l905b, 1918). The group

names that are renamed by later researchers are indicated in the parenthesis.

ferrumequinum
darlingi *

aflinis

*

 

 

 

a I omasr'
capenSiS \
rouxi
. *
dentr
malayanus nerer's /stlreno
\ )/ simulator *
v: r go bomensis

megaphyllus

\, I
Slmp ex \

 

 

<57 pusillus group

Figure 1.5. The relationships among the species of the ferrumequinum group proposed

by Andersen. Species of Afiica are indicated by ‘*’. (From Andersen 1905a, p 120.)

10

pusillus group

Connecting process poi «..

 

   

  
   
    
 
 
       
 
  
  
  

simplex group
Connecting process of sella
rounded.

   
   
     
   
 

  

arcuatus group
Horseshoe median groove
broadened, nasal swelling
enlarged.

  

Generalized noseleaf, ear

moderate, metacarpals sub-

equal, palate not shortened,

P and P present and in
2

toothrow.

   
 
    
  
      
 

luctus group
Sella with lappets, post.
noseleaf and nasal swelling
enlarged

  

macrotis group
Sella, post. noseleaf and
ear greatly enlarged

Figure 1.6. The phylogenetic hypothesis proposed by Tate and Archibold (193 9).
The simplex group and luctus group are presently referred to as the ferrumequinum

group and philippinensis group respectively.

11

grounds for Andersen’s philippinensis group and macrotis group. In addition, Tate and
Archbold consider that the arcuatus group was closely related to the fermmequirmm and
pusillus groups, because the arcuatus group seemed to be an early branch from the
unspecialized forms of that complex. Among the characteristics Andersen utilized, Tate and
Archbold considered that noseleaf structures, particularly the connecting process, were more
reliable characters. They used this character as the primary basis of their subgeneric
classification. Later, Tate (1943) merged the macrotis group with the philippinensis group.

In their two review publications of the mammal collections from Paleoarctic and Indian
region and from south Africa in the British Museum of Natural History, Ellennan and
Morrison-Scott (1953, 1966) merged the arcuatus and philippinensis groups, recognizing a
total of four species groups in the genus. They did not explain the reason for this merger, only
claiming that this revision was in agreement with Tate’s conclusions. This is inaccurate; Tate
and Archbold (193 9) clearly indicated that the philippinensis group, which branched early in
generic evolution, was demonstrated by the coexistence of some primitive features, such as
very long palatal bridge, and some highly specialized features, such as the large noseleaf and
lappets, in this group.

Working primarily on Afiican bat faunas, Koopman revised Andersen’s species groups
of Rhinolophus in that region (1965, 1975, 1989). Although be retained all the traditional
species groups, Koopmn’s studies contained detailed descriptions and discussions of the
morphology and distribution of Rhinolophus in the region In his review on the biogeography
of Rhinolophidae, Koopman (1970) concluded that either Afiica or southern Asia could be its

region of origin.

12

 

26.5: to?
3338325

 

ema—
bo> 082er

 

38:88
>3ng
mo Swee—

058 co m:

.mfi: 32am

 

vows—~98
=2on

.. - .\+ E5835

 

Exam ”Beam =eEm

880on ems e=um

 

8:385
mmoooa
weuooqeou

 

. owe: he,
+ + + £2 aegis.

 

I I\+ I

+ + + own: 30> mam

 

 

 

um um Eel-3:5

managed .Iamea Swansea

aw ea:
aged 3 am

 

3% awn a: a

 

.bozuooamou 330% 05 5 88m $8535 33$th

«5.6 .2305 B as. e5 as? nausea. 283 1+. 2a ..+. ”38% ASE .am as eeoo

EBB .Em use 3950 .3 wow: grade—45a? 8.5% 860% 05 com 853$ oumeememv 2F .2 03¢.

 

13
Corbet and Hill (1992) revised the rhinolophids of the Indo-Malayan region.

Interestingly, except for adding some recently described species and moving R macrotr's from
Anderson’s macrotis group to the philippinensis group, Corbel and Hill endorsed all the
species groups initially proposed by Andersen (1905b). This is not surprising, since the
characters used for identifying species groups by Corbel and Hill were virtually the same as
used by Andersen (Table 1.1). Due to the morphological homogeneity of this genus and the
relatively obvious nature of the traditional characters, it would be surprising if any new result
would be considerably different without use of new characters or application of new methods.

The phenetic analysis by Bogdanowicz and Owen (1992) and Bogdanowicz (1992)
are based on quantitative characters. Multivariate morphometrics and quantitative character
analyses have been applied in the systematics and ecology of other bat families since the early
1970's (e.g. Findley, 1972; Freeman, 1981). In both papers, principal components were
calculated from quantitative (continuous) measurements of the skull and wings. Since the first
principal component is generally regarded as variation due to size differences between species,
which is not very informative about phylogenetic relationship, it was removed fi'om further
analysis. Clusters were computed from the remaining principal components which are
considered to represent the variation in Shape.

Although there is much in common between the results by Bogdanowicz and Owen
(1992) and by Bogdanowicz (1992), the latter is by far more interesting. In this second paper,
Bogdanowicz noticed two major groups associated with two major geographical regions: one
group associated with the Paleoarctic and Ethiopian regions and the other associated with the
Australian and Oriental regions, although in his subgeneric classification of the genus into 11

species groups the mm'or geographic groups are not presented (Figure 1.7 and 1.8).

14
Table 4.2. Summary of taxonomic conclusions based on the monophyletic groups in Figure
4.28. No paraphyletic groups is recognized in this taxonomy. Monophyletic groups of species
are recognized at three different levels (supergroup, group, and subgroup). Those species that
can not be placed into a monophyletic group are included as ‘status uncertain’ at the

appropriate level.

GENUS RHINOLOPH US R aflinis
qfi'im's subgenus R. nereis
philippinensis supergroup R simplex
philippinensrls group R stheno
R luctus R selebensis
R mfoliatus R megaphyllus
R sedulus R malayanus
R macrotis R rouxi
R marshelli R bomeensis
R rex R thomasi
R paradoxolophus ,
R philippinensis subgenus status uncertain
group status uncertain (All Afi'ican & west Eurasian species)
R arcuatus fumigatus group
R canuti R eloquens
R creaghi R fumigatus
R coelophyllus R hildebrandti
R euryotis group status uncertain
R inops R alcyone
R rufils R denti
R submfirs R euryale
R pearsoni R mehelyi
R ywrwrensis R landeri
pusillus group R blassi
R acuminatus R adami
R pusillus R clivosus
R 00an R ferrumequiman
R imaizumii R darlingi
R osgoodi R ccpensis
R subbadius R swimryi
R lepidus R simulator
R monoceros R hipposideros
group status uncertain
(southeast Asian species subgenus status uncertain
of the traditional R maclaudi

fenumequinum group)

15

rouxi group
egaphyllus group

, illus group
euryotis group
firmigatus group
\ ferrumequinum group
capensis group
ewyale group

hippsideros group

tnfliatus group

philippinensis group

Figure 1.8. The relationships among the 11 species groups proposed by
Bogdanowicz (1992). (After Bogdanowicz, 1992.)

l6
Phenetic similarities may be indicative of the true phylogeny when characters are

carefirlly selected and the assumptions associated to the clustering methods are met. I find three
reasons to doubt that Bogdanowicz’s studies are likely to reflect the phylogeny of the genus.
First, his measurement set disregarded some potentially informative qualitative traits, such as
noseleaf and dental morphology. The omission of such notable characters due to difiiculties of
measuring them may produce an incomplete picture of the overall morphological similarity and
difi‘e'ence among species. Second, because all variables are transformed to principal
components before being converted into similarity or dissimilarity indices in the study, it is very
diflicult to determine the specific morphological features that define or diagnose each cluster.
This in turn makes any analysis of clmracter transformation and the pattern of evolution
impossrble. Finally, the clustering algorithm Bogdanowicz used assumes that drift, rather than
selection, is the cause of evolutionary change (Bogdanowicz and Owert 1992). It is not evident
that this assumption is an appropriate one for Rhinolophus. The clusters resulting from the
distance analysis probably does not indicate the ancestor-descendant relationship, because the
joining points of the phenograrn only Show the relative degrees of similarity between
morphological groups. Bogdanowicz used a single species of hipposiderid (Aselliscus
trimspisatus) as an outgroup in his studies; the hipposiderids are a diverse family, and it is not
clear this one species is an adequate outgroup.

Genetic studies in the relationships of Rhinolophus species are limited. Chromosomal
and electrophoretic studies have been carried out on 21 species of Rhinolophus (Dulic and
Mutere, 1974; Zima, 1982; Ando et al, 1983; Harada et al, 1982; Harada et al, 1985;
Qumsiyeh et a], 1988). Although these studies provide usefirl information about the evolution

of the genus, which I will use later in this study, they cover too few Species and lack resolving

 

tr:

Pa

17
power to reconstruct the phylogeny of the genus on their own (Qurnsiyeh et aL 1988). Clearly,

more molecular and cytogenetic study of the genus is needed.

Study of the historical biogeography of Rhinolophus has advanced even less than the
study of phylogeny since Andersen’s early work (Andersen, 1905a and 1905b). Although
regionalbiogeographyofthegenushasbeendiscussedinanumberofareafaunal
investigations (Hayman and Hill, 1971; Koopman, 1966, 1975 and 1989; Lekagul and
Mmeely, 1977; Goodwin, 1979; DeBlase, 1980; Smithers, 1983; Heaney et a], 1987),
biogeographic review over the entire distribution of genus had not been undertaken until
Bogdanowicz and Owen (1992). Their biogeographic study focused only on the question of
whee Rhinolophus originated, and they supported Andersen’s hypothesis that the Oriental
region was the center of origin. No vacariance biogeographic study has been conducted on
overall or regional distribution of the genus.

In the present study, the search for the phylogeny of Rhinoloplms is taken in two steps.
First, I have performed a morphometric analysis with carefirlly selected new measurements on
skulls as well as the traditional ones in the skull and Skin. Emphasis is placed on the nasal
region and the basal region of the skull where considerable Shape variation occurs. Principal
component analysis and canonical discriminant analysis were conducted to find the pattern of
clusters and discover the characters which are most responsible for the clusters. Second,
traditional qualitative characters and new characters tested in the principal component analysis
were selected and analyzed for their phylogenetic information content. Alter constructing
transformation hypotheses for these characters, I performed cladistic analysis using Wagner

parsimony to find the phylogenetic relationship among the species of the genus. A hypothesis

18
of Rhinolophid phylogeny is proposed, and the subgeneric classification of the genus is revised
accordingly.

Based on phylogenetic analysis, I review the historical biogeography of the genus.
Emphasis is placed on southeast Asian species where great biogeographic interest exists.
Following early studies of this region, the southeast Asia region was divided into 11 areas and a
cladograrn of area relationships was computed using Rhinolophid distribution data. Finally, I
suggest that Africa, rather than southeast Asia, might have been the center of origin of the

genus.

MATERIALS AND METHODS

Morphological analysis

I recorded data from skins and skulls of specimens in the following museum
collections: Field Museum of Natural History (FMNH), National Museum of Natural
History (NMNH), American Museum of Natural History (AMNH), Carnegie Museum of
Natural History (CMNH), and Museum of Comparative Zoology (MCZ). The specimens
used are listed in Appendix 1.

I measured skin dimensions of 608 specimens representing 60 species. Fifteen
measurements were taken using digital calipers, except that car, tail and foot lengths
were c0pied from the specimen label recorded by the collector when they were present.
Table 2.1 lists of the Skin measurements and their abbreviations used in the
morphometric analysis. Only the right side, if available, of the body was measured to the
accuracy of one tenth of a millimeter.

The skulls of 1,112 specimens representing 60 species were examined. I
photographed each skull in three views: dorsal and ventral cranial views, and lateral
cranial views of the cranium and mandible together. The specimens were placed on the
top of a small piece of clay attached to a heavy metal base, and the horizontal level of the
specimen was judged visually from camera and side view. For the dorsal and ventral
cranial views of the cranium, the camera lens was centered at the middle of the Specimen
which was adjusted to be bilaterally symmetrical in the view finder. For the lateral views
of the cranium, the specimen was adjusted so that the tips of canines, last molars and

auditory bulla of both sides of the cranium overlap under the camera view.

19

20

Table 2.1: The descriptions and abbreviations of the skin measurements.

5321::

2Met
3Met
3M1P
3M2P
4Met
10. 4M1P
11. 4M2P
12. 5Met
13. SMIP
l4. 5M2P
15. EAR

99°89‘V‘PP’N

Length of forearm

Tail length

Length of foot

Length of tibia

Length of second metacarpal

Length of third metacarpal

Length of first phalanx of third metacarpal
lentth of second phalanx of third metacarpal
Length of fourth metacarpal

Length of first phalanx of fourth metacarpal

Length of second phalanx of fourth metacarpal

Length of fifth metcarpal

Length of first phalanx of fifth metacarpal
Length of second phalanx of fifth metacarpal
Length of ear

21
For the lateral view of the mandible, each specimen was adjusted so that the lower
canines and coronoid processes of both sides overlap. Pictures were enlarged to 3 by 5
inches and printed. Only the pictures of ventral and lateral views were actually used for
measurements, since some of the landmarks I planned to use on the dorsal view were too
obscure on the prints. Fortunately, most of these landmarks were available from the
other two views.

Forty landmarks on the ventral view of the cranium and lateral views of the
cranium and the mandible were selected. The landmarks were recorded as coordinates
using a Summagraphics digitizer. Twenty-seven measurements, either between pairs of
landmarks or fiom a landmark to a line defined by two other landmarks, were calculated
using a BASIC program. These measurements include traditional ones, such as dental
length and zygomatic arch width, as well as those that are very difficult to measure
directly with calipers and had not been analyzed before for Rhinolophus, such as size and
relative positions of cochlea and auditory bulla, or distance fiom a point to a line such as
the anterior-posterior distance from palatal bridge posterior margin to M3. The
landmarks selected and the measurements used in this study are illustrated in Figure 2.1
and 2.2. Descriptions and abbreviations for the measurements are listed in Table 2.2.

I decomposed some traditionally used, overall distance variables into several
regional distance variables to provide a more uniform coverage to the local structures
(Strauss and Bookstein, 1982). I replaced the basal length of the cranium with a series
of measurements including upper toothrow length, temporal fossa length, basal length

from fossa to cochlea, cochlea length and post-cochlea length. Some traditional

22

Table 2.2: Descriptions and abbreviations of the skull measurements. Numbers
correspondent to the labels in illustrations in Figures 2.1 and 2.2.

Ventral view of cranium:

255328

Vertical length of P2

Length of palate bridge

Distance between labial-most points of two M3
Width ofM’

length between two ptyrogoid processes

Length of temporal fossa
7. LSHF Length of sphenoid fossa
8. WZA Width between two lateral most points of zygomatic arches
9. WAB Vlfrdth of auditory bullar
10. LAB Length of auditory bullar
11. BL Basal length between sphenoid fossa and fi'ont tip of
the cochlea
12. BB Basal Breadth between cochlea
13. WCO Width of cochlea
14. PMP Distance from posterior margin of palate bridge to line
defined by caudal end of both M3
15. VLC Vertical length of cochlea
16. VLAB Vertical length of auditory bullar
17. PB Posterior brain case length from mastoid process to end of

cranium

Lateral views of cranium and mandible:

18. DH Height of mandibular ramus at lower canine

19. DL Dental length

20. LINF Length of infraorbital foramen

21. LOR Length of orbit fiom top of infiaorbital fiamen to most
restricted point of orbit region

22. HOR Height of orbit fi'om base of M3 to groove of

orbit

23. P4M3 Length of upper cheek tooth row

24. LBR Length fiom end of M3 to condyle fossa

25. HNS Height of nasal swelling

26. HCR Height of cranium

27. HOCC Height of occipital region

23

AN 035. E vow: 828383.“ v5 3582885 05 8 383:8 8588385 Low 233

.3352: 3:052:32: 05 23 Stages 2: 55 3%. 530— 98 :BESo mo m32> 383 ”mu oSmE

 

24

 

Figure 2.2: Ventral view of cranium with the landmarks and the easurements
illustrated. Labels for measurements correspond to the descriptions and
abbreviations listed in Table 2.2.

25
measurements, such as mandibular toothrow length, are omitted due to redundancy. The
three measurements between a point to a line, including the distance from posterior
margin of palate to posterior end of M3' post-cochlea cranial length, height of nasal
swelling, and nine measurements in the basal and auditory region covering detailed
structures of this region were not measured by Bogdanowicz and Owen (1992) nor have
they been analyzed before in any group of bats.

The SAS statistical package (Luginbuhl and Schlotzhauer, 1987) was used in the
morphometric analysis. For each species, specimen measurement means were calculated
to represent that species. A principal-components analysis was used to (1) find the
patterns of clusters based on morphological similarities; (2) find the variables that are
most responsible for the formation of the clusters. Univariate analyses were conducted
on measurements that showed high correlation coefficients with the informative principal
components. Variables most responsible for differentiating clusters were selected for
later cladistic analysis. The principal components were computed from both correlation
and covariance matrices, since variables with high variance are more strongly associated
to the first several components when a covariance matrix is analyzed (Luginbuhl and
Schlotzhauer, 1987). To better detect importance of significant characters, Skin and skull
data sets were analyzed separately as well as jointly. The SYSTAT software package
(Wilkinson et al, 1992) was used to plot the principal components.

A canonical discriminant analysis was performed to test the validity of traditional

species groups as well as those clusters revealed in the principal component analysis.

26
Meters for cladistic am

I selected 26 characters for phylogenetic analysis. They include both traditionally
used characters, characters newly identified fiom alcoholic specimens, and those
converted from the quantitative characters after morphometric analysis.

I used four hipposiderid genera (Asellia, Aselliscus, C Ioeotis and Hipposideros)
as outgroups for character analysis. In some cases, hipposiderids do not serve well as an
outgroup due to their specialization of specific characters. For example, all hipposiderids
have lost P3, which is a derived feature within bats (Van Valen 1979). For some other
characters, evidence fiom other bat families is informative. For example, the number of
caudal vertebrae is greater in hipposiderids than in Rhinolophids, and is even greater in
the earliest known bats. The conditions in the earliest bats helps confirm the direction of
character evolution. In these cases, more remotely related bat families were used as an
additional outgroup.

Identifying characters for cladistic analysis is a critical process. Because the
subsequent cladistic analysis of characters is, by itself, only a summary of information
contained within the data set (Neff, 1986; Bryant, 1989), characters selection largely
determines quality and reliability of cladistic results. Character identification requires
three steps:

1.) recognizing a morphological series of features between species that can be
hypothesized to be homologous;

2.) determining hypotheses of the polarity and transformations among the states
of a morphological series, utilizing the tools of outgroup analysis, paleontological

analysis developmental biology, etc;

27

3.) establishing the distribution of character states and character state
transformation among the taxa under study.

Throughout the process of character selection, I attempted to determine as
unambiguously as possible the order and polarity of each character. Recognition of
homologous states and structural transformation series in Rhinolophus is relatively
simple. The structures in different species of the genus usually have identical topology
and similar position. However, parsing the morphological series into distinct states and
proposing the transformation series demands more effort, since considerable
intraspecific variation exists in most characters.

Whenever possible, I determined the polarity and transformation matrix of the
states for each character. For some characters, two hypotheses in polarity were made
and both were used in the phylogenetic analysis. I used the outgroup distribution
criterion and ingroup commonality criterion (Watrous and Wheeler, 1981; Maddison et
al, 1984) to recognize primitive character states. The complex noseleaf of Rhinolophus is
unique among bats with which no known homologous structures in any outgroups may
be directly compared. I assumed that primitive states for characters on noseleaf are
typically the smaller, less developed and less prominent states in a series. This
assumption agrees with Hill’s descriptions on the primitive Hipposideros (Hill, 1963).

I am convinced that all characters should not be weighted equally, Since some
characters which are better studied, involve more evolutionary innovations, or are more
likely to be synapomorphic than others (Hecht and Edwards, 1977; Netf, 1986). Hecht
and Edwards' five weighting types were modified into four categories to fit the situation

in Rhinolophus. Each character was assigned to one of the four weighting groups. The

28

characters of weighting group one contain character states where transformation
involved simplification or reduction. This group was given the lowest weight. Weighting
group two contains features that were not reductiona] but which had relatively high
levels of intraspecific variation or where the boundaries between the states were, to some
extent, arguable. This group was given the next lowest weight. Weighting group three
contains characters that are relatively unique and innovative in nature, but where
distinctions between character states can not be recognized are clearly as the last group,
and the boundaries are still more or less arbitrarily determined. This group was given
higher weight than the previous two groups. Weighting group four comprises characters
that are evolutionary innovations where distinct states can be clearly recognized; these
characters are likely to be genealogically most informative and were given the highest
weight.

PAUP3.1.1 (Swofi‘ord, 1993) was used to compute the most parsimonious trees
for the genus, using the heuristic searching algorithm. It was assumed that the three
types of transformations (innovative, reversal, and parallel changes) are of equal
probability. When a character weighting scheme is applied, weighting group one receives
a weight of one unit, weighting group two receives a weight of two units, and so on.
Wagner parsimony was applied. The specimens examined for character analysis and

cladistic analysis are listed in Appendix 1.

RESULTS

MORPHOMETRIC ANALYSIS

I analyzed separately data from skin measurements, skull measurements and pooled
data of skins and skulls for a better identification of individual measurements. The first five
principal components (PC1-PC5) were computed fi'om each data set. In a preliminary
examination I observed no pattern beyond the PCS, therefore, I chose the first five principal
components for detailed analysis. My discussion will focus on the first three PCS, because PC4
and PCS account for relatively little variation. Two dimensional displays for various
combinations of principal components were made to examine patterns of morphological
Similarity among the species. I examined the correlation between the eigenvectors and the
original variables to determine the contribution of each original measurement to the species
distribution patterns. Finally, I conducted a canonical discriminant analysis to verify the
suggested species groups.
A The pooled skull and skin data set

The first principal component (PC1)ofthe covariance matrix accounts for 85.5% of
the total variation of the original variables, whereas in the correlation matrix it accounts for
77.2% of the total variation (Appendix 2.1 and 2.2). PC] has positive correlation coefficients
with all the original variables. In such situations, PCl is commonly interpreted as a size
component (Humphries et al, 1981). PC] is relatively more apparent as a size variable in the
correlation matrix where almost all variables have similar (between 0.1 to 0.2) correlation
coeflicients with PCl. PC] is less obvious a Size component in the covariance matrix, since

some of the fi'equently used size indicators (e. g. DL, LBR in the skull, and FA, 2MET, 3MET,

29

Table 3.1: The abbreviations and the traditional group identities of each species used in the

display of their principal components. f: ferrumequinum group; p: pusillus group; a:
arcrmtus group; h: hipposideros group; 1: philippinensis (= Iuctus) group; m: fumigatus
(= macrotis) group. The species groups were originally defined by Anderson (1905b,
1918), and were modified by Tate and Archibold (1939), Koopman (1975), and Corbet

and Hill (1992).

Abbr.

ac
ad
af
al
ar
bs
bt
b0
ca
CP
cv
ce
co
cr
da
dk
dt
eq
el
et
fe
fu
hl
hr
hp
im
in
ke
ld
1p
lt

Group species name

l—I'U'Ur'hm'U’J'l-‘BBWW'US’UMMW'OOJHMDJM'U'UW'UMM'U

acuminatus

blythi
bomensis
canuti
capemis
clivosus
coelophyllus
comutus
creaghi
darlingi
dekenii
denti
eloquens
ewyale
euryotis
fenumequinum
firmigatus
hildebrandti
111'an
hipposideros
imaizumii
inops
keyensis
landeri
lepidus
luctus

Abbr.

md
mr
ms
my
mg

me
05
pa
pe
ph
pu
re
ro
ru
rf
se
sh
sm
SP
st
sb
sr
sw
th
tf
vi

Group species
name

maclaudi
macrotis
marshellli
malayanus

megaphyllus
mehelyi

monoceros
nereis
osgoodi .
paradoxorous
pearsoni
philtppinensis
pusillus

rex

robinsoni
rouxi

mfi‘s

sedulus
shameli
simulator
simplex
stheno
subbadius

submfiis
swinnyi
thomasi
tnfoliatus
virgo
yrmanensis

BHI—‘l-hl'hDJ'UHrl'hr'hQ’l—‘ml'ht'hl-‘UHBH'U r-h'o'omer—ar—a

31
4MET, and SMET in the wing) have relatively low correlation with PC 1.

The other four PCS have both positive and negative correlation coefficients with the
original variables, indicating that these PCS represent contrasts between sets of measurements.
Generally such contrasts are interpreted as reflecting variation in shape (Humphries et al,
1981). In the covariance matrix, the four remaining PCs accounts for 7.7%, 2.3%, 1.7% and
1.1% of the total variance, respectively. In the correlation matrix, these same PCs accounts for
7.0%, 4.4%, 2.2% and 1.7% of the total variance. The variation not accounted for by the first
five PCS is much greater for the correlation matrix (7.3%) than for the covariance matrix
(1.6%), though the total variance represented in PC2 through PCS is also greater in the
correlation matrix (15.3%) than in the covariance matrix (12.8%) (Appendix 2.1 and 2.2).

Different original variables have high loadings on PC2 and PC3 in the two analyses.
For the covariance matrix, the variables having high positive loadings on PC2 are LPF (.23),
TEF (.21), HOC (.17) and 4MET (.10); those with high negative loadings are PAL (-.82),
VLIB (-.23). PC2 of the correlation matrix has high positive loadings for P4M3 (.49), DH
(.26) and 4MET (.20), and has high negative loadings for PAL (-.38), VLIB (-.23) and BB (-
.22). PC3 of the covariance matrix has great positive loadings for VLIB (.63) and LFC (.18),
and high negative loadings for WAB (-.59) and BB (-.39). For the correlation matrix those
variables having high positive loadings are 2MET (.46) and 3M1P (.30), and those having high
negative loadings are PB (-.30), LFC (-.28) and LIF (-.25).

The highly loaded original variables are from both the skin and skull, and are relatively
concentrated on the palate and basal regions of the skull. These variables include the length of
palate (PAL), length of upper cheek toothrow (P4M3), the length of cochlea (VLIB), length of

temporal fossa (TEF), basal breadth between the cochleae (BB), width of the auditory bulla

32
(W AB), the length from pterygoid fossa to cochlea, and length from mastoid process to the
posterior end of the skull. However, they are not restricted to a few particular structures.

No obvious pattern of species clustering can be found in the two-dimension plot of the
PCI by PC2 for the covariance matrix. In the PCl by PC2 plot ofthe correlation matrix, there
is an imperfect separation of species according to their geographic distribution: species fiom
Afiica and west Eurasia make an exclusive convex group and occupy the lower half of the
display (Figure 3.1). The separation is almost exclusively along the PC2 axis. No pattern of
species distribution is evident along the PCI axis, suggesting that size is not a major
difi‘erentiating factor in the subgeneric taxa of Rhinolophus. Because the length of the palate
bridge has greater absolute correlation coefficient with PC2 (-.82) than any other
measurements, three southeast Asian species of Andersen’s philippinensis group (R Iuctus, R
rex, and R macrotis) which have the longest palate bridges, are located in the negative side of
PC2 axis with the Afiican and west Eurasian species. Three Afiican and west Eurasian species
R simulator, R landeri and R alcyone and one southeast Asian species R osgoodi are also
misplaced.

In the displays of species on the display of PC2 by PC3 for the correlation matrix, none
of Andersen’s species groups can be clearly observed as distinct clusters, except that the
species of Andersen’s arcuatus group are situated close to each other with only two species of
Andersen’s ferrumequinum group distributed inside the arcuatus cluster. In Figures 3.2 to 3.4,
the distribution of the species of each traditional group is indicated by a convex hull. However,
the geographic pattern of taxa in these displays is more apparent than the species-groups

patterns. For the correlation matrix, a line can be drawn which separates the genus into two

 

 

 

 

 

 

 

”V if T l I
5‘ .0 O . "T
a .0
.o
'0 O . O m
.- O
B I O
l
. S .0 g o '0
I o
1 _ so ‘0 o m —<
'o
,2 ’c'o '0 ”a,
ti; ’— ’ o , _.
[.53] To a ‘0 .
‘:_\l' 'o r0 Co
.. o
O o n
l _ 1 2- to ”o _
'0 'o r
no '0 ° 0 r
p R o o o m
—.‘7’“ L
“J : r. M,
a: ' '
"rxj; IL_ 0 ()I ’n
_“J1> l»— o' —-+
...§':3; '9 J 1 1
- 1 2 «as a; 1 .2 .2 to
PC3 1
fix? ! T t l
l
.1 southeast Asran , , A _
" . A0 °
A n
specres ,0 . 0 ,
, .1 ,‘
Hm. ;"_ ‘0 A O ’ __
l A” A0 A0 ‘0
; O
1 7‘ ~°° 1’ O A '2
flo
A a 0 .......
.2 Am . _________
«.v A .. ,2’ . ----------- _
(v .n ‘0 . (I .......
will; A0 '.; ........... CD A
: ' A n .5.» """ r
n- .1 ......... r. ,,
— 1 Ir ..... ' ‘1‘ F n W o -‘
.i w n A 0 0 O W
. WWI " n o F
., l.— ——<
_ “S WI,
3 o '
‘ Afrrcan and west ., . :
_ Eurasran specres
— .JL 4— -<
Lu i” o A
-5 l 1 1 r
- 1 31; -2‘1 .21 1 2 ‘1‘! (Q)
P@ 1

(a)

(b)

Figure 3.1. Displays of species on PCI and PC2 of the correlation matrix from the
pooled skin and Skull data. (a). The separation between the traditional species groups
is not clear. Species are symbolized in Species group identity: f: ferrumequinum
group; p: pusillus group; a: arcuatus group; h: hipposideros group; 1: philippinensis
(= Iuctus) group; m: fumigatus (= macrotis) group. (b) There is a approximate
separation of species associated to the geographic origins (dotted line). Species are

symbolized in their distribution: ‘A’ =

Eurasia.

southeast Asia, ‘F’ = Afiica, and ‘W’ = west

 

 

34

 

 

PR|N3
J.
l

 

 

 

-5 -4 -3 -2 *1 0 1 2 3 4
PRINZ

Figure 3.2. The display of traditional ferrumequinum species group in PC2 and PC3,
from the correlation matrix of the pooled skin and skull data. There is extensive overlap
between species of this group and other groups. Species group abbreviations: f:
ferrumequinum group; p: pusillus group; a: arcuatus group; h: hipposideros group; 1:

philippinensis (= Iuctus) group; m: fumigatus (= macrotis) group.

35

 

 

PRIN3
I

 

 

 

 

 

_5. l L l l l l I l

-5 -4 -3 -2 -1 O 1 2 3 4
PRIN2

Figure 3.3. The display of the traditional pusillus species group and arcuatus species
group in PC2 and PC3 from the correlation matrix of the pooled skin and skull data.
There is extensive overlap between species of the pusillus group and other groups, but
only one species of the filmigatus group is inside the arcuatus group. Species group
abbreviations: f: ferrumequinum group; p: pusillus group; a: arcuatus group; h:
hipposideros group; 1: philippinensis (= Iuctus) group; In: filmigatus (= macrotis)

group.

36

 

PRIN3

 

 

 

 

Figure 3.4. The display of traditional fumigatus (= macrotis) group and philippinensis
(= Iuctus) group in PC2 and PC3 from the correlation matrix of the pooled skin and
skull data. Extensive overlap exists between species of the filmigatus group and other
groups. Species group abbreviations: f: ferrumequinum group; p: pusillus group; a:
arcuatus group; h: hipposideros group; 1: philippinensis (= Iuctus) group; m:

filmigatus (= macrotis) group.

37

 

3 T I T l I l l l
#460
We
2 ~ _
3
WO F0 0
3:;
1 '— '0 '0 0‘ —

   
 

African and west 1:: o
0 *— Eurasian species

 

 

‘ o A A
‘oo‘ oonozp‘ O‘OOA
fl 0‘ ° 0
2 0A 0A
_ —‘l -—— A 0A *1
m A°oA 0A
a.
-2 . southeast Asian species c
..A
-3 g -
-4 p E
A.
_5 1 1 1 1 L 1 1 1 E-

 

-5 -4 -3 -2 -1 O 1 2 3 4
PRIN2

Figure 3.5. Figure displays species on the PC2 and PC3 from the correlation matrix
of pooled skin and skull data. There is a non-overlapping separation of species
clusters associated with their geographic origins: one species cluster from Africa and
west Eurasia, and the other from southeast Asia. Species are symbolized in their

distribution ‘A’ - southeast Asia, ‘F’ = Africa, and ‘E’ = west Eurasia.

3 8

groups according to their geographical origin without misplacement or overlap (Figure 3.5).
Both PC2 and PC3 contribute to this separation. The PC2 axis shows that the African and
west Eurasian species have relatively long palatal bridges and short upper cheek toothrows,
and the PC3 axis indicates that these species have relatively long second metacarpals and long
first phalanges of the third finger, and short posterior basal areas in the skull. In the display of
PC2 and PC3 from the covariance matrix the genus also appears to form two groups. African
and west Eurasian species (with the exception of R maclaudr) form a cluster in the negative
sides of both axes, and the southeast Asian species are distributed in the positive side of both
axes (Figure 3.6). PC3 in this analysis is the main distinguishing component along which the
southeast Asian species show longer cochlea and shorter auditory bullae than Afiican and west
Eurasian species. None of the displays containing PC4 or PCS show any additional patterns of
clustering.

The fermmequimon and pusillus species groups have species in both geographical
clusters of Figures 3.5. The extent of these two species groups are evident in the plot of PC2
versus PC3. However, when only the southeast Asian members of each species group are
examined, relative closeness between species of the same group becomes apparent in four
traditional species groups as shown in Figure 3.7. This structure is not apparent among the
African and west Eurasian species. The members of the traditional species groups in southeast
Asia are noticeably more differentiated from each other than are their relatives in Africa and
west Eurasia. The species in the latter two areas are more homogeneous in skull and wing
shape variables summarized by PC2 and PC3 then are the southeast Asian species.

B. The skull data set alone

39

 

  
 

 

 

 

8 1 1 1 1 1 1
A0
A0
6 1- _
‘9
A A
A0 ° °
‘ A ‘0 . .
0 southeast Asran specres
A o A
2 ~ ‘0 -
A0

(‘0 10% A ° ‘° A A
0 ° ‘5» * °
0. Q ° A

o — 3» ° a“ r a

A ._w o 2 A
.. A
w ._ n ‘
A0 0 ‘
Africa and west _ O 0 A
Eruasian species
-6 1 1 1 J 4 1 J

-15 -1O -5 O 5 1O 15 20

Figure 3.6. Figure displays species on the PC2 and PC3 from the covariance matrix of
pooled skin and skull data. Species of Africa and west Eurasian form a distinct
cluster. Species are symbolized in their distribution ‘A’ = southeast Asia, ‘F’ =
Africa, and ‘E"= west Eurasia.

4O

 

   

 

 

 

3 I I I I I I I I
African and west 13.,
2 Eurasian species
1 a.
O — a
2
E '1 ”
n.
-2 _ a
philippr’nenszk group _
southeast Asian species
.—4 ~ ~
I
_5 1 1 1 1 1 1 1 1
-5 -4 -i -2 -1 O 1 2 1 4
PRINZ

Figure 3.7. Figure displays species on the PC2 and PC3 from the correlation matrix
of pooled skin and skull data. The traditional species groups are more distinct when
only those from southeast Asia region is considered. Species are symbolized in their
taxonomoc group: f: ferrumequinum group; p: pusillus group; a: arcuatus group; b:
hipposideros group; 1: philippinensis (= Iuctus) group; m: filmigatus (= macrotis)
group.

41

PC] for the correlation matrix accounts for less total variation (71.6%) than PC] of the
covariance matrix (.90). In the covariance matrix, the correlation coeficients of the original
variables with PCI vary from 0.0 (PZ) to .62 (DL), but commonly used size measurements
(eg. DL, WZA and LBR) are highly correlated with PC]. In the correlation matrix, the same
correlation coefficients are nearly uniformly positive (Appendix 2). As was the case for the
conrbineddataseLPCl fortheskulldatasetisasizefactor.

The measurements PAL, BB, LAB and PB have high correlation with PC2 and PC3 as
inthecombineddataset.ButtheskulldatasetshowsthatPMP,WZAandLIFaremajor
shape variables as well.

The displays of PC2 by PC3 for both correlation matrix and covariance matrix show a
good separation between the African and west Eurasian species on the one hand and southeast
Asian species on the other, although R hipposideros and R adami are misplaced in the
covariance matrix and R hipposideros and R simulator are misplaced in the correlation matrix
(Figure 3.8 and 3.9). For both matrices, separation occurs primarily along the PC3 axis, which
indicates that Afiican and west Eurasian species have relatively broader zygorrmtic arches,
shorter distance between the posterior margin and posterior end of M3, shorter mandible
lengths, longer lengths fi'om pterygoid fossa to cochlea, and smaller P2.

W

PC] of both covariance matrix and correlation matrix account for about the same
percent of total variation (86.1% and 85.5%). The high loadings of commonly used size
measurements such as FA (.43), 3MET (.28), 4MET (.32) and MET (.34) on PCI for the
covariance matrix, and uniform positive loadings on the same component in the correlation

matrix suggest that PC] is a size component (Appendix 2).

42

 

 

 

 

 

2 I I I I I
‘0 A .
1 _ “° ' southeast Asian species ~
67
(0
Ci} 0
Q
_ 1 _ F African and west
w 0 w' 0 Eurasian species
W 0
w .
-2 1 1 L 1
_2 - 1 O 1 2 I 4
PC2

Figure 3.8. Figure displays species on PC2 and PC3 from the correlation matrix of
skull data. Species of Africa and west Eurasia are separated from the species of
southeast Asia. . Species are symbolized in their taxonomoc group: f: ferrumequinum
group; p: pusillus group; a: arcuatus group; h: hipposideros group; I: philippinensis

(= Iuctus) group; m: fumigatus (= macrotis) group.

43

 

*0
Ac
2 *— An A0A T“
A0
‘33 ‘0 0A
A, M‘:
1 1— ‘ A90 _
A”AC A: 0
“a . .
~» southeast Asran specres

Q, fi) (111 l--
Zn '1

 

   
 

 

 

1
: w“
-2 l r,
1 African and west
1 w , Eurasian species
l W
_; ' 1 l 1 1
‘4 "2 2 4 “ 1 G 5

Figure 3.9. Figure displays species on PC2 and PC3 from the covariance matrix of
skull data. Species of Africa and west Eurasia are separated from the species of
southeast Asia. Two arrows indicate two misplaced species. Species are symbolized

in their distribution: ‘A’ = southeast Asia, ‘F’ = Africa, and ‘W’ = west Eurasia.

44

The display of species in PC2 by PC3 for the covariance matrix shows a geographic
pattern where Afiican and west Eurasian species are separated from southeast Asian species,
with only three species misplaced. This pattern of separation is move evident in the similar
display of PC2 by PC3 for the correlation matrix where only one species is misplaced. The
species plot of PC2 by PC3 for the correlation matrix also shows some pattern of traditional
species groups among southeast Asian species: the species ofthe arcuatus group and most
species of the pusillus group are close to each other, but both groups are overiapped with the
femanequimmr group which is more scattered (Figure 3.10).

PC2 and PC3 of the skin data reveal several important morphological features which
are not disclosed in the pooled data set. The high correlation of these origirml variables with
PC2 and PC3 shows that the Afiican and west Eurasian species have longer tails, longer
second phalanges of the fourth finger but shorter first phalanges of the fourth finger, longer
second phalanges of the third finger but shorter third metacarpals, and smaller ears (Appendix
2).

D. Remarks on the Principal Commnent Mses

Among the traditional species groups, only the arcuatus group is distinct in these
analyses. The species of the philippinensis group span a broader range but do not have
extensive overlap with other groups. All other four species groups are not completely
distinguishable in any of the three analyses. However, there is a pattern of species separation
associated with the two major geographic regions: the Afiica and west Eurasian region and the
southeast Asian region. When only the southeast Asian species are considered, two additional
traditional species groups, the pusillus group and the ferrumequinum group become

distinguishable. The arcuatus group is distinct because it contains only southeast Asian species.

45

 

   

 

 

   

 

 

2 I 1 F 1
. I
1 e a
I.
southeast $1.
on
0')
E 11.3) F a
a:
n.
"'7 p.
-1 ._ W
l
1 Africa and west
Eurasian species
-2 l 1 1 1
-3 -2 -1 In 1 2

Figure 3.10. Figure displays species on the PC2 and PC3 from the correlation matrix
of skull data. Among the southeast Asian species, only members of the traditional
arcuatus group are close to each other. . Species are symbolized in their taxonomoc
groups: f: ferrumequinum group; p: pusillus group; a: arcuatus group; h:
hipposideros group; 1: philippinensis (= Iuctus) group; m: firmigatus (= macrotr's)

group.

46

Considering that all of the species groups recognized have some overlap with other groups and
gaps between these groups are small or nonexistent, these analyses indicate that the traditional
species groups are only weakly differentiated.

Inalldatasetsandanalysesofbothcorrelationandcovariancematrices, PCl canbe
interpretedasasizecomponent. ItisonlyonthePC3 andPC2 axesthatsomepatternsof
clusters are seen No patterns of distribution are found on PC4 and subsequent PCs. In all
principal components on these axes species distributions are rather continuous.

Considering that the morphometric data do not contain characters from the noseleaf
which is the most important basis for establishing traditioml species groups, the relative
distinctiveness of these groups within the southeast Asia region is significant. It is also
significantthatthesarnepattemismanifestedindependentlyintheskullmeasurementsandin
external measurements as well as overall morphology.

Based on the principal component analysis, there is not enough evidence to decide
whether the traditional hipposideros species group, which contains R hipposideros only,
should be considered as a distinct group. Although observations shows that R hipposideras
has proportionally larger cochlea, the major shape components do not demonstrate that the
difference between R hipposideros and other species with this feature constitutes an important
part of overall generic morphological variation of the genus.

The traditional fen-umequinum group is recognizable in the correlation matrix of
pooled skin and skull data set, but in the displays of other analyses, it has more extensive
overlap with the arcuatus and pusillus species. Both Andersen (1905a) and Tate (1939)

observed that the fen-umequinum group has less specialized features than other groups (e.g.,

47
modest-sized ear, cochlea and nasal swellings). VVrthout characters from the noseleaf, skull and
skin measurements do not provide enough evidence to unite the species of this group.

Andersen’s macrotis group originally contained four Afiican species and three
southeast Asian species. This group was merged with the phildvpinensis group by Tate (1939)
and reinstated by Corbet and Hill (1992). Corbet and Hill renamed it the fianigarus group since
R macrotis had been moved to the philippinem'is group. Species of this traditional group are
scattered on the PC2 and PC3 in all analyses and, therefore, this group is not confirmed.

Based on the correlation coefiiciencies of original variables with PC2 and PC3 of the
three analyses, the measurements contributing most to observed clusters are: length of palate
bridge (PAL), position of posterior margin of the palate bridge (PMP), length of upper cheek
tooth row (P4M3) and width between zygomatic arches (ZAW) in skull measurements; and the
tail length (TL), ear size (EAR), length of second phalanx of the third finger (3M2P), length of
first and second phalanges of the fourth finger (4M1P, 4M2P) in the skin measurements.

The genus Rhinolophus is well known for its homogeneity in morphology. The results
of this study confirms this. There is little structure to the phenetic similarity Rhinolophus
species. Consequently, morphometric data used in this study seem insufiicient for
reconstructing the phylogeny of this genus. More phylogenetically informative qualitative
characters are necessary for this purpose.

E. The Canonical Discriminant Analyses:

A canonical discriminant analysis was used to test the validity of the traditional species

groups and the species clusters revealed in the principal component analysis of the pooled skin

and skull data set. Three separate tests were conducted. In the first, the traditional species

48

 

 

 

 

4 l l l l
M M .
3 — a
2 _ a
g
(£3 1 F 4
Q 9 L
. .~ .. LL
'1’: PR? ., L, 3"
I) >- F . P ° .H
F ,
at???"
_ 1 ._ A AA “ a
A :A .
-2 . 1 1 1
-2 -1 <01 1 2 I
(322318431

Figure 3.11. Figure displays the first and second canonical variables (CAN 1 and
CAN2) for the southeast Asian species. Traditional species groups are used as a
priori class. Species are symbolized in their group identities: f: ferrumequinum
group; p: pusillus group; a: arcuatus group; h: hipposideros group; 1:
philippinensis (= Iuctus) group; m: firmigatus (= macrotis) group.

49

 

 

 

 

 

2 l l
”a... L .
L 41,,
1 Ir r
l
1 ”3. 1»
g] . 9": '5
(“E KO) T T
(Q)
. f Ii.
. 85.?”
_ 1 it
; M ,.
T A“ ‘A .
1
!
1
_ 2 i 1 1
-2 - 1 13> 1 2

@A N 1

Figure 3.12. Figure displays the first and second canonical variables (CAN 1 and
CAN2) for the southeast Asian species, Andersen’s macrotis group being merged
with philippinensis group. Traditional species groups are identified as a priori class.
Species are symbolized in their group identities: f: ferrumequinum group; p: pusillus
group; a: arcuatus group; h: hipposideros group; 1: phierpinensis (= Iuctus) group; m:
firmigatus (= macrotis) group.

50
groups were identified as a priori classes and the canonical discriminant analysis was done
within the southeast Asian species. The display of the first two canonical variables shows five
very distinctive species groups and the separation between the five species groups is perfect
(Figure 3.11). I then merged the macrotis with the philippinemz‘s group as Tate (1939)
suggested. The distinctiveness of the four species groups are equally apparent in this analysis
(Figure 3.12). The test strongly confirmed the existence of these groups.

Inthesecondarralysis,IassignedallthespeciesofAfiicaandwestEurasiatoanewa
priori class, and the species of the southeast Asia remained in their traditional species groups.
Figure 3.13 and 14 shows five distinct clusters on the CANl by CAN2 plot, and the
fernanequimm group and pusillus group are well separated in CANS (Figure 3.13). The
display not only shows an unequivocal separations between groups, but it also shows a larger
gap between the species group of Afiican and west Eurasian species and species groups of
southeast Asian than among species groups of the southeast Asian region.

In the final test, the traditional species groups were use as a priori classes for the entire
genus and five canonical variables are computed. Figure 3.14 and 3.15 show six non-
overlapping species groups, though the species of individual species group are not as
concentrated as the species groups including the southeast Asian species only. It seems that the
traditional species groups may be confirmed by this analysis.

Canonical discriminant analysis of the pooled skin and skull data set is able to
distinguish species grouped in all of the analyses. While this analysis did confirm that southeast
Asian species groups can be distinguished with these data, this fornr of analysis also readily
distinguished between groups whose distinctness was not apparent in the principal component

analysis.

51

    

 

 

 

 

1:4}
5:372
i; r 1
~‘ I \v
1 1 r r
\r
a 1 v o
‘1 Q ll
11412) ‘2. ‘ <3
(,5 U .11 I 41:; y
259 ' ‘ l
.4 1 1 ‘ § 1 I
BE , 4? ° 1 '
ry~ ‘ . .
((1; 4') ’ 1 «I ,1 r
I l \—
i I
I}: Q ‘ "
p? ‘ l
(:17, v ‘ 'r
P
l p
1 1 ‘ '
v 1
L ‘7' 1 v- l
N: v ‘1 l ,
1r ‘ l 15:1. "
C4?) ? Tl ‘

Figure 3.13. Figure displays the first, second and fifth canonical variables (CAN 1,
CAN2 and CAN 5) for all species of Rhinolophus. Traditional species groups are used
as a priori classes for southeast Asian species only, and all Afiican and west Eurasian
species are assigned to a new class labeled ‘W’. Southeast Asian species are
symbolized in their group identity: f: fiarrumequinum group; p: pusillus group; a:
arcuatus group; h: hipposideros group; 1: philippinensis (= Iuctus) group; m:

fumigatus (= macroris) group.

52

 

 

 

 

1 (Q I I I
. H
_ F ..
5 — _.
F . F ~
F 9
F5. f .
. .
. F ,
F. ”F. u . 9 FE
V . L
z W M “FE: ” EL
(ll *—- A _, . L L __4
1‘15?) A» L
' A . A A P
. . p
P am PP P o P
. P
, P
n3 .. P P
_. _ . P _,
_ 1 ((3’) L_ a 1 1 l
- 1 £11.!) -7 17.11 5:: 11:11

"(a A c. n 43>
I'LI‘ J1 \))
\1'5’ um m K;

Figure 3.14. Figure displays of the third and fourth canonical variables (CAN3 and
CAN4) for all species of Rhinolophus. Traditional species groups are used as a priori
class. Species are symbolized in their group identity: f: ferrumequinum group; p:
pusillus group; a: arcuatus group; b: hipposideros group; I: philippinensis (= Iuctus)

group; m: firmigatus (= macrotis) group.

53

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

1 199 ’ .. r ”
\- »
fin;
1 .»
[2.2, “ H
«J, 9
L- I 1
l ’ p.‘
1 I Q r
#1»
's
I '~ ”W q
( V, K l 1?"
/ 1? ; ax
” L ' .
.{J '
gas 1
1:745) 1 l * ‘
'1‘: A! J <
I [-3) 9 ! i ‘* 1*
Q ,
Q l ‘
" l
' 1
L 9‘ , if
\( 4‘) \. ' i
\\ l
1537,. \. y
. , --—" a“
.2. (Q . ~ (73‘,
.._—L/' ‘ ”I x J
:;,_ \~\ (
:1) 1' 3, "x _ ),_,..« ) ,
/ CZ ' .4 ’11 ~ (It.
/\ d) I E

Figure 3.15. Figure displays the first, third and fourth canonical variables (CANl ,
CAN3 and CAN4) for all rhinolophids. The ferrumequinum group and firmigatus
group are separated on CAN 1 axis. Traditional species groups are used as a priori
classes. Species are symbolized in their group identity: f: ferrumequinum group;
p: pusillus group; a: arcuatus group; h: hipposideros group; 1: philippinensis (=

Iuctus) group; m: firmigatus (= macrotis) group.

54

I conclude that, while canonical discriminant analysis does confirm the results from
principal components, the ability of this technique to difi‘erentiate among groups not
distinguishable in the principal component analysis makes this result suspect. In canonical
discriminant analysis, the number of observations should be at least five times as large as the
number of the variables to receive an unbiased and consistent result (Kalayeh and Landgrebe,
1983). When the observation/variable ratio is small, arbitrary separations between a priori
groups can be nrade (Ness, 1979; Dubes and Jain, 1991). In my data the observation/variable
ratio is less than two. My results suggest this form of analysis is overly powerful at
discriminating groups in my data, and is unlikely to be trustworthy for exploring the taxonomic

structure in Rhinolophus.

CHARACTER ANALYSIS

The transformation series of states for each character were determined by methods
described which follow. In general, I first considered the states that have been commonly
recognized in systematic studies of this genus, and used outgroup analysis to determine
polarity. When the outgroup criteria did not provide sufficient evidence, in-group commonality
criteria (Eldredge, 1979) was applied to determine polarity. In some cases, boundaries are not
clear between character states previous identified. This usually occurred for regional
Rhinolophus fauna but not the entire genus. I combined those states that were ambiguous so
tlrattherernainingstateswerereasonablydistinct. Incaseswheretherewasdisagreementin
evolutionary direction changes between states, I either took a position when there was enough
evidence to do so, or left the states as unordered. I placed each character in one of the four
weighting groups described in the Material and Methods section, with group one receiving the
lowest weight and group four the highest

1. Shape of the Connecting Process of Sell_a_ (ConPr). I recognized 5 states for this
character (Figure 4.1): a, moderate height and rounded, anterior base not reaching the summit
of the sella (e. g. R fenwnequinum and R climsus); b, higher and sharper, anterior base not
reaching summit of the sella, as represented by R macrotis; c. very low, the anterior base
distant fi'om the summit of the sella (e.g. R luctus and R mfoliatus); d dorsal edge with a
slmrp angle or hom-shaped, the anterior edge forming a notch where it connects to the sella
(e. g R acuminatus and R comutus); and e. dorsal edge low and round, anterior base reaching

the tip of the sella (e. g. R pearsoni and R arcuatus). State e was described by

55

(a) E (c) - (e)
@

(b) /

Figure 4.1: The shape of the connecting process of the noseleaf (character 1) in lateral

view, pointed by arrow. (a) state a, height moderate and round (R aflinis); (b) state

b, higher and shaper (R. macrotis); (c) state c, very low (R Iuctus); ((1) state d,

hom-shaped (R pusillus); (e) state e, anterior base reaches the tip of the sella (R

pearsoni).

57
Andersen (1905d) for the arcuatus groups as “strongly arcuate, almost semicircular in outline
and starting from the very summit of the sella”. However, few subsequent researchers have
utilized this as a character state.

The connecting process shape of the sella is one of the characters used for group
idmtification in this genus. It is the primary distinguishing feature for the pusillus group
(Corbet and Hill, 1992). Both Anderson (1905a) and Tate (1939) indicated that the
connecting process of ancestral rhinolophids was most likely generalized, being low and not
pointed. But since the horseshoe of all living species are beyond this hypothetical primitive
stage, identifying this primitive state does little to resolve transformations among the observed
character states. Andersen and Tate’s presumed primitive condition does not include the
relationship between the anterior base of the connecting process and the sella. In state e the tip
of the sella is continuous with the connecting process. In states a - d the connecting process is
distinctly separate from the tip of the sella. State e may be the result of an anterior and dorsal
extension of the connecting process, in which case it is likely derived; or it may be the result of
a relatively short sella, in which case it could be primitive. In both cases, this distinction
between state e and other states may indicate a departure of state e from all other states.

The states of this character were treated in two ways; both assume that none of the
observed states fi'om a to e are primitive. The first treatment assumed that transformations
among the states were indeterminable and treated the states as unordered. This avoided
incorrect ordering at the expense of abandoning some usefirl information. The second approach
adopted Tate's view (1939) of transformations indicated in his phylogenetic hypothesis of the
genus, assuming an additional primitive state p, with an order illustrated in the following

diagram:

58
Tate's view (193 9) of transformations indicated in his phylogenetic hypothesis of the genus,

assuring an additional primitive state p, with an order illustrated in the following diagram:

P

Tate’s view of the transformation series for character 1,
indicated in his phylogenetic hypothesis of the genus (Tate and
Archibald, 1939).

Because this portion of the noseleaf is innovative in nature and is unique to rhinolophids, this
character was placed in weighting group four.

2. Sella Sham (Sella). The shape of the sella exhibits three states (Figure 4.2): a,
narrow, without a lateral process (lappet) (e. g. R pusillus and R rouxr); b. broad without a
lappet (e. g. R macrotis and R inops); c. broad with a lappet (e. g. R Iuctus and R maclaucb).

This character has also been used previously for species group identification. While the
existence of a lappet is readily distinguishable, some sella shape differences between species are
difficult to characterize or assign to a transformation series. I have chosen to ignore subtle sella
shape difi’erences and focus on obvious differences. Because the sella is a structure unique to
rhinolophids, this character was placed in weighting group four. I assumed the sella arose in a
simple, narrow form which broadened and gained a lappet later in evolution. This hypothesized

sequences fiom simple to complex yields a transformation series for the three states of a->b-

>c, with state 0 representing the primitive condition.

59

 

Figure 4.2. The shape of the sella (character 2), pointed by arrow ‘IO‘. (a) state a, narrow
(R alcyone); (b) state b, broader (R jimigatus); (c) state c, with lappet, pointed by arrow

‘-=0‘ (R maclaudi). (From Rosevear, 1965).

 

Figure 4.3. Illistrations of the horseshoe (character 3). (a) state a, narrower (R clivosus); (b)

state b, broader (R firmigams). (From Rosevear, 1965).

6O

 

Figure 4.4. Illustrations of the supplemartary noseleaf (character 4). (a) state a, not present
(R Iuctus); (b) state b, less developed (R pusillus); (c) state c, both sides meet at the mid-

line (R simulator). The dotted line indicates the horseshoe which usually covers most part of

the supplementary noselea£

 

Figure 4.5. Number of the lower lip grooves (character 5), pointed by arrow. (a). state a, one

groove; (b). state b, three grooves.

61

\ .

(a) (b) 9

Figure 4.6. Illustrations of the fiont ear projection (character 6). (a) state a, not present; (b)

state b, present.

   

(a) (b)

Figure 4.7. The shape of the lancet (character 7). (a) state a, hastate; (b) state b, nearly

triangle. Long hair, shown in B, are frequently present in the lancet. (from Rosevear, 1965).

62

3. Shim of Anterior Nose-leaf or Ho_rsgsh_oe (ANL). I recognized three states for this
character (Figure 4.3): a, narrow (e. g R euryale and R simulate); b. expanded in one
direction, either laterally (e. g. R afiinis) or ventrally (e. g. R filmigatus); c. significantly
expanded in both lateral and ventral directions (e. g. R Iuctus).

State b possibly represents a mixture of stages in the noseleaf development; it may not
be uniquely intermediate between states a and c. I could not unambiguously subdivide state b
or advance a reliable argument for the sequence between character states. States of this
character were treated as unordered. Given the possible heterogeneous nature of states b, and
variation within and between species in more subtle details of horseshoe shape, I placed this
character in weighting group two.

Horseshoe size has been used in previous taxonomic studies of this genus. Recent
studies demonstrate a correlation between horseshoe size and fiequencies of ultrasonic pulses
bats emit during echolocation (Bogdanowicz, 1992). Some of the variation in horseshoe size
between species probably is based in differences in echolocation firnction and may not reflect
phylogenetic relationships. Until we can distinguish how echolocation behavior has evolved in
this genus, it will be difficult to interpret horseshoe size as phylogenetic information. I did not
atterrrpted to analyze horseshoe size here.

4. Supplementary Leaflet Beneath the Horseshoe (SupLeaf) (Figure 4.4). The
supplementary noselea£ located beneath the horseshoe, displays three states: a, not present or
not easily identifiable (e. g. R mfoliatus and R dentr); b, distinct but with a wide median gap
between the two pieces (e. g. R macrotis and R makomus); 0. pieces of both sides meet or
almost meet at rnidventral line (e.g. R comutus and R stheno). This character has been used in

previous studies for distinguishing species of particular geographic regions but not species

63

groups. Vlfrthin each state, differences exist in breadth of the supplementary noseleaf, although
previous workers appear to have ignored this variation It is likely that additional states could
be recognized, but at present I am not confident ofthe criteria that would be used. Hill (1963)
argues that the outgroup genus HWros, with lateral leaflets that are positionally identical
to the Rhinoloplms supplementary noseleaf, lacked lateral leaflets in its primitive condition. It
therefore seems reasonable to identify state a as primitive. Although state c seems more
advanced. thereisnoevidencethatstatebisanecessaryintermediate stage. No ordercanbe
determined, except that state a is primitive. I placed this character in weighting group three for
the reason that it is part of the noseleaf, an innovative complex structure.

5. Number of Mental Grooves or Lower Lip Grooves (LLG). There are two states
(Figure 4.5): a, one median groove (e.g. R hildebramlti and R firmigatus); b, three grooves
(e.g. R arcuatus and R mops). Previous studies of this genus have considered this to be a
good character because it is unique and invariant within species. For these reasons, I assigned
it to weighting group four.

Although this character was described and discussed by Andersen (l905a) and used in
regional keys, it was never used to diagnose species groups. Probably because the distribution
states of this character is not in conformity with other characters previous taxonomists treated
as fundamental (e.g. the noseleat). Contrary to Andersen's (1905a, p.107) view, outgroup
examination indicates that only Rhinolophus displays three lower lip grooves. It is most
reasonable to consider a as a primitive state within this genus.

6. Front Ear Projection (EarPr) (Figure 4.6). There are two character states: a, not
present (e.g. R maclaudi and R landeri); and b, present (e. g R malayanus and R clear

subbadz'us). Although generally very small, this projection is identifiable when present. It is not

64
whether this projection, located at the basal, rostral-proximal side of the pinna in Rhinolophus,
is homologous to the tragus of other bat fanrilies. No such projection has been found in closely
related bat families (e. g. Hipposideridae and Nycteridae), and I hypothesize that this feature is
an innovation and unique to rhinolophids. Considering that some variation exists in some
species (e. g. 20% absent and 80% present in R lepicbas), this character is assigned to weighting
group three. The primitive state is a.

7. Shim of the Lancet (Lancet) (Figure 4.7). The shape of the lancet encompasses two
broad categories: a, hastate, abruptly narrowed in its distal half (e.g. R lepidw and R
clivums); and b, not hastate, triangular in shape, distal end blunt (e.g. R thamasi and R
philippinensis). Despite its frequent usage in regional keys, this character more or less forms a
continuum among the entire genus. Intraspecific variation is also present. This character is
therefore assigned to weighting group two. I can not establish unambiguously the primitive
state of this character. A new state p is introduced as a hypothetical primitive state and its
relationship with a and b is unordered.

8. Number of Ear Ridges (Eang). The ears of Rhinolophus posses ridges parallel to
ear width I have partitionw variation in this character into two states: a, 10 or more ridges
(e.g. R blassi and R clivosus); b, 9 or fewer (R lepirbls and R euryotr's). This character has
not been used in earlier work. The character state in each species was determined by examining
as many specimens as possrble, usually 10 or more. The most frequent number of ear ridges
among the specimens examined was used to assign each species to a character state. Division

of states is arbitrary and there is some intraspecific variations. In addition, outgroup

65
testing does not provide evidence for determining the ancestor state, since both states are
present in Hipposideridae, the assumed sister group of Rhinolophus. For these reasons, the
character is assigned to weighting group one and treated as unordered.

9. Number of Free TaiVertebrae (Tail). Tail length has been a frequently used
character in regional studies of Rhinolophus species. The variation in tail length is made more
discretebycountingthenumberofvertebraeinthetailwhicharefieefromthesacrum. The
number of caudal vertebrae has not been used in previous systematic studies of the genus.
Perhaps there is irrtraspecific variations in counts and ambiguity due to vertebrae that are
reduced in size or partly attached to the sacrum. To recognize variation present in caudal
vertebrae counts, I have recognized only two states: a, five or six caudal vertebrae (eg R
fermmequiman and R. erayale); b, fewer than five caudal vertebrae (R inops and R
subbadius). The state for each species was determined as the most common count found in
examining as many specimens as possible (usually 10 or more). Vertebrae that were greatly
reduced in size, or partly connected to the sacrum were counted as 0.5 vertebrae.

Outgroup analysis supported the hypothesis that five to six caudal vertebrae is the
primitive condition in Rhinolophus. The eight species in four genera examined as outgroups
possessed five or six caudal vertebrae. The earliest-known fossil bat ( Icaronycteris index,
Jepsen, 1970) had seven tail vertebrae. I hypothesized that state a is primitive (contrary to
Andersen’s views [1905a, p. 107]), and that evolution within the genus has resulted in a
numbers, and likely pattern of reduction of vertebral numbers over time, I considered this
reduction of caudal vertebrae number. Because there is intraspecific variation in vertebral

character to be in weighting class two.

66

10. Insertion of Plagiopatagium (1' ailMem). The wing membrane inserts at three
difl’erent points along the leg in species of Rhinolophus: (Figure 4.8): a, at the ankle (e.g. R
megaphillus, R bomensis and most other species); b, along the lower leg, 5 mm or more
above ankle (e.g. R. creaghi and R rufirs); and c, at or close to the tarsal-metatarsal joint (R
luclus and R maclardr). Character states of this may be related to body size or feeding habits
(Straney, 1984) and in Rhinolophus may display homoplasy. Consequently, I placed this
character in weighting group two. All three states appear in other bat families; with state a by
far the most common in outgroups. Parsimoniously, I assume state a is primitive and the other
two state are independemly derived: c<- a -> b.

11. Anterior Upfl Premolar (P2)(Figure 4.9). P2 in bats displays a common set of
character states: a, in the toothrow, b, small and displaced out of the toothrow; and 6, absent.
It is generally agreed (Slaughter, 1970) and confirmed by outgroup tests, that state a is
primitive and c is most derived. Intraspecific variation is common. Even within the same
individual, it sometimes happens that the two P2 are in different states. The character state for
each species was decided by majority rule afler examining many specimens, the technique used
by Andersen and other researchers. Because I hypothesize P2 has become reduced and lost
over time, and intraspecific variability is present. My criterion placed this character in weighting
group one.

12. Shap_e of the Anterior Lower Premolar (P2)(Figure 4.10). Variation in the shape of
P2 falls into three categories: a, length (rostral-caudal distance) about equal to width (labial-
lingual distance) (e. g. R macrotis and R Iuctus); b, length conspicuously greater (20% or

more) than width (R arcuatus and R inops); and c, width conspicuously greater than length

67

      
 

WI

' .113" ll.
; , v‘ Lil/1 mil...
“5.5! §‘¥¢,’ -

   

(C)

Figure 4.8. The insertion of the plagiopatatgium (character 10). (a) state a, at the ankle (R
megaphr‘llus); (b) state b, above the ankle (R rufils); (c). state c, near the tarsal-metatarsal

joint (R. Iuctus). (After Rosevear, 1965).

68

 

(a) ' ’ (b)

Figure 4.9. The status of P2 (character 11), pointed by arrow. (a) state a, in the toothrow
(R ions) ; (b) state b, out of toothrow (R fenwnequr'num); (c) state 6, absent (R

firmigates).

 

Figure 4.10. The shape of P2 (character 12), pointed by arrow. (a) state a, the length and
breadth bout equal (R meheri); (b) state b, length greater than breadth (R cIivosus); (c)
state c, length less than breadth (R ions).

69

(R euryale and R fianigarus). The shape of P2 has sometimes been referred to in species
descriptions, but has not been used in previous systematic analyses of Rhinolophus. In fact,
there is much less intraspecific variation in this feature than in the frequently-used position of P2
(see above). Considering the conservative nature of rhinolophid dental morphology, distinct
shape modifications in P2 justifiably places this character in weighting group four. Although
there seems to be some relationship between the shape of P2 and compressedness of the cheek
toothrow (which is considered a derived feature by Andersen [l905a] ), gaps sometimes are
found at one or both ends of shortened P2 Thus, a compressed cheektooth row and short P2
are not necessarily functionally related. All four genera and eight species of hipposiderids
examined as outgroup displays a square shape of P2. I hypothesize that state a is primitive and
statesb andcare independently derived: b <-a-> c.

13. Middle Lower Premolar (P3) (Figure 4.11). This tooth displays three states: a, in
the toothrow, b, small and displaced out of the toothrow; 6, absent. The discussion for
character 11 applies to this one. The states are ordered as a -> b -> c, with state a primitive. I
placed this character in weighting group one.

14. Cingpla of Lower Molars (Mm). The cingula of the lower molars displays two
levels of development: a, wealdy developed. as in most species; and b, strongly developed (e. g.
R fianigatus and R yrmrmem'is). Because molar morphology is homogeneous in Rhinolophus,
the distinction made by this feature is significant. State a is by far most fiequent in both
ingroup and outgroup species; it is reasonable to hypothesize state a as primitive. The
boundary between the two states is sometimes ambiguous and I have placed this character in
weighting group two.

15. Sglarshelf Shelf of M3 : The size of the stylar shelf (the posterior V—shaped triangle

70

 

Figure 4.11. The status of P3 (character 13), pointed by arrow. (a) state a, in the toothrow
(R macrotis); (b) state b, out of toothrow (R malayanus); (0) state 0, absent (R

firmigates).

71

of the W-shaped outer shelf) in the last upper molar (M3) varies among species of
Rhinoloplms, it is described by two states (Figure 4.12): a, moderately reduced. as in most
species; b, greatly reduced, with the anterior ridge of the posterior V less than 2/3 the length of
the posterior ridge of the anterior V (R ccpensis and R WW3). It is generally agreed
(Slaughter, 1970) that the reduction of styleshelf is a derived feature within bats and I
hypothesizethat stateaisprimitive. Becausethischaracterrepresentsareduction ofM3 and it
isacommontrendinotherfarrriliesofbats,thischaracterwasassignedtoweightinggroup
one.

16. Posterior Edge of Pa_lpt_e_:: The length of the palate is a character widely used in
previous taxonomic studies of Rhinolophus. The palate varies in length largely due to the depth
of the median ernargination of its edge. Two species can have the same average palate length,
but differ in location of posterior and anterior ernargination edges. I recognized two characters
that capture the details of palatal morphology in Rhinolophus: position, relative to the
toothrow, of the posterior edge of the palate (this character), and position of the anterior edge
of the palatal ernargination (character 17).

The posterior edge of palate displays four character states (Figure 4.13): the posterior
edge lies a, next to the metastyle of M2; b, between metastyle and metacone of M2; 0, between
metacone and mesostyle of M2; and d anterior to mesostyle of M2. Outgroup analysis
indicated that longer, shallowly emarginated palates (state a) are primitive within Rhinolophus.
Consequently, I hypothesized that state a is primitive and that the four states were connected in
a linear transfonnational sequence: a -> b -> c -> (1

Some intraspecific variation is present for both this character and character 17. I

assigned states to species afler examining as many specimens as possible (usually 10 or more)

 

 

72

 

(b)

Figure 4.12. The shapes of the stylarshelf in M3 (character 15). (a) state a (R
a mis); (b) state b, the posterior v-shaped ridges (pointed by arrow) greatly
reduced (R fumigatus).

 

 

73

 

Figure 4.13. Picture (R affinis) illustrates character 13, the position of the posterior
margin of the palate pointed by an arrow. Label 0 through 3 correspond to the states a
through d in the text.

74
and using the state most common within a species. Because of this variation, and the arbitrary
nature of the landmarks chosen to recognize character states, I placed this character in
weighting group two.

17. A_nterior Mggin ofPalate (AMP) (Figure 4.14). This character has been used in a
key to Afiican Rhinolophus by Koopman (1975) in which he recognized two states. For
species of the entire genus I recognized three states for this character: anterior nrargin of palate
emargination located a, anterior to the protocone of P‘; b, between protocone of P‘ and
mesostyle of M‘; and c, at or posterior to the mesostyle of M'. As discussed in character 16,
this character is assigned to weighting group two and state a is assumed to be primitive, with
transformation series a->b->c.

18. Front Margin of Anterior Nasal Swelling (FMNS) (Figure 4.15). The anterior
margin of the anterior nasal swelling lies at three different positions relative to the toothrow in
Rhinolophus: a, at or anterior to the parastyle of M‘; b, between parastyle and mesostyle of
M'; and c, at or posterior to mesostyle of M'. This character has not been utilized in previous
systematic research within the genus. It becomes obvious once photographs of lateral views of
the skills are examined. There appears to be some correlation between position of the front
margin of this nasal swelling and that of the palate bridge. I do not believe this possible
correlation is important for two reasons: first, I know of no functional explanation for such a
correlation; second, while it is reasonable to assume that a shallower palatal emargination is a
primitive state, it is not at all evident that the relative anterior position of this nasal swelling is
primitive. Because no clear homologous structure is present in other closely related bat

families, I applied the in-group commonality rule (W atrous and Wheeler, 1981) and

75

 

Figure 4.14. The position of the anterior margin of the palate (character 17), pointed by
arrow, of R. aflim's. Label 0 through 2 correspond to the states a through c in the text.

 

Figure 4.15. The position of the front margin of anterior nasal swelling (character 18),
pointed by arrow, of R. ajfim‘s. Label 0 through 2 correspond to the states a though c
in the text.

76
hypothesized that state a is the primitive state: a -> b -> c. The arbitrary boundary between
states placed this character in weighting group two.

19. ngth of Medi_an Frontal Nasal Swellings (LNS) (Figure 4.16). The length of the
median fi'ontal nasal swelling varies between species of Rhinoloplrus. A convenient way to
characterize this variation is with the following two character states: a, small or less than the
combined length of M1 and M2 located beneath it in lateral view, as is the case of most species;
andb, largewithlengthgreaterthanthecombinedlengthofMl ansz (e.g R IuctusandR
sedulus). Except for a few species with intermediate size (e. g. R clivosus), most species fit
relatively easily into these two size categories. The landmarks defining states of this character
are arbitrarily chosen for convenience. This character appears to be positively correlated with
noseleaf size, although firnctional reasons for this are unclear. For these reasons, I assigred this
character to weighting group two. State a is assumed to be primitive for two reasons: first, it is
the state found in most species; second, the nasal swelling is a feature unique to Rhinolophas
and it is parsimonious to assume it arose as a small feature.

20. Depth of Orbital Constriction (OrbC) (Figure 4.17). Variation in the orbital
constriction displays two states: a, shallow as in most species; and b, very deep, the depth
greater than half of the front nasal swelling height (e. g. R blassi and R creaghr). State a is
assumed to be primitive on the basis of its frequency of occurrence both within and outside of
the genus. The arbitrary state boundary placed this character in weighting group two.

21. Infiaorbital Canal and Ba_r (InfOrb) (Figure 4.18). The shape of infraorbital canal
and bar vary together and fall into three categories: a, canal nearly round and infiaorbital bar
short and narrow (e.g. R rouxi and R bomensis); b, canal heightened, bar elongated and thin

(e. g. R clivosus and R eloquens); and c, canal round but moved anteriorly with the bar

77

 

Figure 4.16. Pictures illustrate character 19, the length of median frontal nasal
swellings. (a) state a, small (R. affinis); (b) state b, larger (R. Iuctus).

78

 

(b)

Figure 4.17. The depth of the orbital constriction (charcter 20), pionted by arror. (a).
state a, shallow (R. lepidus); (b). state b, deep (R creaghr').

79

 

Figure 4.18. The shapes of the infraorbital canal and bar (character 21), pointed by
arrows. (a) state a, size moderate (R. aflim's); (b) state b, infraorbital bar elongated (R.
clivosus); (c) state c, canal lengthened and bar broader (R Iuctus).

 

 

80

greatly broadened (e. g. R Iuctus and R mfoliatus). It is not entirely clear which is the
primitive state. Because it is most common in both ingroup and outgroup, state a is most likely
to be the primitive state with an order ofb <- a -> c. Except for a few species (e.g. R inops
andR erayon's), mostcanbeassignedastatewithoutdificulty.1hischaracterwasplaced in
weighting group three.

22. Shape of Intemairal Region (IntNa). Two states were identified: a, not expanded,
(e.g. R pusillus and R afinis); and b, expanded (e.g. R m’folratus and R WW). This
character is used traditionally to separate the otherwise similar arcuatus group from the
firmigams group of species. I recognize such separation of two character states despite the
somewhat arbitraryboundarybetween states. Forthesarnereasonsldiscrrssed incharactersl
through 4 on noseleaves (noseleaf structures are highly informative, and small feature size is
prinritive), I hypothesized, that state a is primitive, and assigned this character to weighting
group three.
Continuous characters

Characters 23 to 26 are external measurements of specimens, adjusted by body size.
These continuous characters are coded into discrete states using gap-coding (Michevich and
Johnson, 1976). The means and standard deviations of each variable (a ratio in this case) for
each species were calculated. The pooled standard deviation (Sd) for each variable was also
computed. For each variable, species means were sorted and gaps greater than (Sd * C) were
identified between successive species. Species on the two sides of these gaps belong to two
distinct character states. C was chosen to set the overlap between species on different sides of

the gap to a predetermined level. When C = 0.25, the percent of overlap between two species

81
separated by a gap is 45%; when C = 0.5, the overlap is 40.1%; when C = 1, the overlap is
30.9% and when C = 2, the overlap is 15.9% (Archie, 1985).

Thismethod, thediscrete-statecodingmethodsingeneraLisassociatedwithaweak
assumption that gaps stand for significant evolutionary steps that occur less often than
evolutionary changes between gaps. (Felsenstein, 1988). There is no a pn'ori reason to believe
this assumption models correctly the pattern of rhinolophid evolution The gap-coding method
alsohasthedisadvantagethatgapsterrdtobecomelessnumerouswhenthenumberofspecies
involved becomes larger. The present study contains a large enough number of species to
suspect that gaps used here are conservative estimates of the true, ‘evolutionary’ gaps. Since I
chose C between 0.1 and 0.25, which was necessary to recognize 4 or 5 distinct states for each
character, overlapbetween speciesseparatedbyagapmaybemorethan50%. Iutilizedgap
coding to transform continuous variables, viewed by previous workers as important indicators
of species difl’erences, into discrete states that can be compared and analyzed with previously
described characters. I am more concerned with the comparability gap coding transform makes
possible than with special assumptions about the way continuous character evolve in
Rhinolophw. Due to the significant overlap evolution in between species separated by gaps, I
placed character 23 - 26 in weighting group two.

23. BMW Ear Size [Ear/Fa). Ear size was adjusted using forearm length as an
estimator of the body size. Wrth C = 0.15, four states were identified among Rhinolophus
species, state a representing the smallest ratio. This character was used because relatively larger
ears (and antitragus) have been used by previous taxonomists to distinguish the philippinensis
and arcuatus groups. While it is generally accepted (and confirmed by outgroup comparison)

that relatively larger ears is a derived feature in rhinolophids, it is doubtfirl whether the smallest

82
ratio is the primitive state within the genus. I hypothesized that (by far) the most common state,
state b, isprimitive, yieldingapolarity ofa <- b-> c-> d

24. Relative Phalangeal Length of Digit 3 (1/2F3). The relative lengths of the third digit
first and second phalanges have been widely used in previous studies of Rhinolophus. VVrth C =
0.11, four states were identified with state a representing the smallest ratio. Both ingroup and
outgroup examinations demonstrated that b is the most common state. I hypothesize that state
bisprirrritive, andthepolarityisa<-b-> c-> d Bothareductioninsecond phalanx length
and an elongation of the third resulted in an increase of this ratio.

25. Relative Phalangeal length of Digit 4 (1/2F4). The relative lengths of the fourth
digit first and second phalanges has been used in regional keys for Rhimlophus, and is a
diagnostic character for several species (e. g. R mehelyi and R ewyale). Variation in the
relative phalangeal length of the fourth digit is greater than that of the third digit due to a more
remarkable reduction in the first phalanx. “nth C = 0.25, five states were recognized with state
a representing the smallest ratio. Because state b was the most common state both in this genus
and in outgroup, I hypothesized that state b is primitive and the order was assumed a <- b ->
c -> d -> e.

26. ReLative Length of the Third and Forth Metamal (M3/M4). The ratio of the third
and the forth metacarpal has been used in regional species diagnosis of Rhinolophus and in
recognizing relationships within species groups. VVrth C = 0.1, four states were recognized,
state a representing the smallest ratio. It is generally agreed that state d (two metacarpals being
about equal in length) is the primitive state, and that the decrease of this ratio is due primarily

to a shortening of the third metacarpal (Andersen, l905a). The present data also showed that

83
state dis the most common in the genus. For these reasons, I hypothesized a polarity of a <-
b <- c <- d for this character.
Table 4.1 summarizes the distribution of the character states described above. The

cladistic analysis is based on this character analysis.

84

Table 4.1. The character transformation series matrix used in the cladistic analysis.

in the section of character

are correspondent to the states a, b,

Numeric numbers I, 2,

analysis respectively. Missing data are represented by ‘?’.

12 3 4 5 6 7 8 910111213141516171819202122 23242526

Spp. \ character#

412220111202100121001202224
224?1027721720702000002252

Racunrimrtus
Radami
Raflinis

R alcyon

1332111111210031100102224

41431111120010002100102352

l

1031001012224

5233211112021
41311010120010002100002243

R arcuatus

R blassi

100002223
1001012224

1322121120210021
52432121111021
1 110100201

1
l

R bomensis

13
10102200102243

R canuti

l

l
I
l

R. capensis

100002224

1

1021002
120012200102232

1321111

R celebensis

R. clivosus

1

131010021

101012224

100002224

1
1

11021102
1
524321101101

52322121

R. coelophyllus
R. comutus
R. creaghi

R. darlingi

R. denti

1020002

132121

1

4

1001012224

103
1.0012100002242

4l312010020010012200102242

1

10101211

113

1

10102443

1

2

242102102212011

l

4

R. eloqens
R. euryale
R. eruyotis

12010010010002200002453

1

52421121

1

001111224

1
l

103

11021

13111002110001

100102232

R ferrumequinum l 1

10102343

120121
131020020010012000002242

123210100211

1

R. hildebrandti
R. hipposideros
R. inops

1

1001112224

103
1010020010012100102252

1021

524321211

1
1321

33411111

4

l

l
l

4
4

R. landeri
R. lepisus
R luctus

100002224

10021

111200

I

121322

100002224

13001002001

13210012021002]

1

R. irnaizumii

11214233

1

1

3342102103000100

R maclaudi

10213224

224221211200000001

R. macrotis

1222111120110031001102224

334?2?20?30000000010213224

l

R malayanus

Rmarshalli

101102224

13212112001002]
41112010021010102201102554

l

I

R megaphyllus

R. mehelyi

1100002224

1112001002

1321
1232121

1

4

Rmonoceros

Rnereis

1000002222

1021003

I

l

85

Figure 4.1. (Continued)

12 3 4 5 6 7 8 91011121314151617181920212223242526

Spp. \ character#

R osgoodi

100002224

132120010200001

1
Rparadoxolophus32472720770210000010212224

4

120201020100102222

52421121

R.pearsoni

32422120120200010011013224

R philippinensis
R. pusillus
R rex

R rouxi

1020200021100002224

33412120130200000011212224

12201

4 1

1120110122100002224

1
1

13221
524121

1

11212214

1

1

123

1021

1
1

1
l
1
l

R. rufus

1001012224

1 0 2 1 1 0 2
1 2 0 2 1 0 0 3
1 3 0 2 1 0 0 3

5243212
333

R. shameli

R. sedulus

12221

1 1

1

0
1000202224

1

111
3211

1

l

1
1
1
1

l
1
1

4

R. simplex

100102242
1032201002224

10001

3321200200

R. simulator

R.stheno

13211111021
1321111

52422121

1

100002223

0

1
1

1020003

R. subbadius ‘
R. subrufus
R. swinny

4

1

1
102252

120210022100001224

13001

1012

103 1

12021

13321101210101

100

1 3

122121

1

1

R. thomasi

1212221

101001

1 1

1
13212102021003

1

3341

R trifliatus
R. vergo

1000002223

1

1
524

1

111102223

12112021012
1000020000000000002224

1

1
1

R. yunanensis

1

1

0

0018er

CLADISTIC ANALYSIS

I performed a cladistic analysis of the characters described above for 56 species of
Rhinolophus. This analysis was based on characters that were, for the most part, ordered into
transformation series that could be weighted for thdr likely content of phylogenetic
information Several characters, though, were assigned transformation series that were
uncertain Character 1, in particular, was assigned a transformation series based more on
traditional arguments than on compelling evidence. I believed the this was one of the most
informative characters. Since the manner of relating its character states was likely to influence
the structure of a phylogenetic hypothesis based on them, I performed the cladistic analysis in
two ways: treating character 1 as unordered; and as ordered into the transformation series
described in the previous section. Additionally, the use of character weights in phylogenetic
analysis is controversial (Sneath and Socal, 1973; Sharkey, 1989). To judge the effect of
weighting characters on the resulting cladograms, I performed the analysis both with and
without character weighting. In all, four sets of analyses were performed: (a) characters
weighted and character I ordered into a transformation series; (b) characters weighted and
character 1 unordered; (c) characters unweighted and character I ordered; and ((1) characters
unweighted and character 1 unordered. A monophyletic group present in all analyses represents
a phylogenetic hypothesis better supported by the data than those present only in some, but not
all, analyses (Straney, 1981). The outgroup used is a hypothetical organism with the primitive
state for all characters except for characters whose polarities are unclear, where an additional

hypothetical primitive state is assigned to it.

86

87

ForeachsetofanalyseslusedthePAUP 3.1.1 program, asdescrrbedintheMaterials
and Methods section above. This program calculates the shortest cladograrn for a set of input
data, cladograrn length being measured by the total number of evolutionary steps required by a
particular hypothesis of phylogenetic relationships‘. Because of the relatively large number of
species examined in this study (56 species), the practical limitations of the programs required
use of the heuristic search procedure to search for the shortest cladogram or cladograms for
that data setz. Some of my analyses resulted in thousands of cladograms of equivalent shortest
length Available computer memory limited the number of cladograms that could be stored to
about 1300. Consequently, only the first 1300 shortest cladograms identified by the program
could be saved and analyzed. This limitation introduces a possible source of inaccuracy, as the
heuristic search procedure is sensitive to the order of samples in the input data matrix (Geske,
1992). To minimize the effect of this limitation I repeated the computer analysis for the same
data set and each time rearranged the species positions in the data matrix. The number ofsteps
in the shortest cladograms derived from differently rearranged data matrices was invariable. I
firrther computed strict consensus cladograms (Swofford and Begle, 1993) for each set of
1,300 shortest cladograms. The discrepancies between these strict consensus cladograms were
very small and insignificant. To show variation among the shortest cladograms I also computed
the consensus cladograrn under mm'ority rule for each set. Particular relationship was

preserved if it was common to fifty percent or more of the shortest cladograms.

 

1Whenthrechar'aetersarreweighted, however, thetransformationsoocurring indifferentcharactersare
themselves weighted For example, ifone transfomration occurred in a character having a weight oftwo, then
that change aooomrts for 2 units of length

2The algorithm for exhaustive search of shortest tree is computationally intractable. Such searching
algorithms are not practical for large data sets regardless of the computer system and the program (Garey
and Johnson. 1991). ’

88

The use of heuristic search and consensus cladograms has become a standard
approach to cladistic analysis of large data sets (Swofl‘ord and Begle, 1993; Maddison and
Maddison, 1992). Three additional considerations justify the use of heuristic search and
consensusmethodsinthepresent study. First, adatasetwithmorethanZOtaxa(morethan20
species in the present case) can not be analyzed using exhaustive search procedure (Swofl‘ord
and Begle, 1993). The computational feasibility of heuristic search procedure with more than
20taxaisachievedattheexpenseofoptimal results. Thereisnoguarantee, whentheheuristic
algorithm is used, that the true, most parsimonious cladograrn is included in the set of shortest
cladograms. As a result, in many cases the phylogenetic relationship for a relatively large group
is computationally an approximation Second, as the sample set of the shortest cladograms
taken from all possible shortest cladograms was quite large (1300). The possibility that this
sample set is unrepresentative of the total set should be relatively small. Indeed, the 1300
shortest cladograms produced by repeated analyses of the (same) reordered data matrix show
very little, and often no, discrepancy in the topology of the consensus cladograms they entail.
Finally, because strict consensus, extracts only the relationships common to all shortest
cladograms produced by a given analysis, it provides a very conservative estimate of
relationships.

Analysis of Weighted Characters, Character 1 Ordered

The strict consensus and majority consensus cladograms for this analysis are presented

 

in Figure 4.19 and 4.20, respectively. The species names are followed by symbols for their
geographic distribution in parenthesis (A = ‘southeast Asia’, F = ‘Afiica’, E = ‘western Eurasia

and Afiica’). The species’ identity in the traditional species groups is the same as in Table 3.1.

89

outgroup
acuminatus(A)
pusillus( (AA
cornutus( )
imaizumiAA)
os oodi(
adius(A)
Iepidus(A)
monoceros(A)
aflinis(A)
arcuatuAs(A)
canuti(A )
creaghi(A?
coelloph lus(A)
euryotiszA)
inops(A)
Iuctus(A)
tritoliatus(A)
sedulus 4A2)
maclau
macrotis(A)
marshalli(A)

paraJoxolophusM)
philippinensis(A)
pearsoni(A )

unanensisA
¥ufus(A) ( )

shameli(A)

subrufus(Al
borneensrs A)

celebensis(A)
nereis(A)
virgo(A)
simplex(A)
stheno(A
thomaSI( )
aphyllus sA)
ma ayanus(A
rouxi(A)
swinnyi(F)
simulator(EF)
terrumequinum(EFA)
capensis?)
clivosus( F)
loquens( 2
mm atus
hilde ran tiZF)
darlingi(F)
hipposideros(EF)
alcyone(F)
denti(F)
muemalqEF
elyiE )
landerI ’éF) F)
blasii(F)
adF)amr(

Figure 4.19: The strict consensus cladograrn for the 24 most parsimonious cladograms
resulting from the weighted analysis, character I ordered. . The geographic location of
the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ = Africa, and
‘E’ = west Eurasia.

90

outgroup
acurninaxrsM)
SI us
grnutug(A)
imaizumii(A)
osgoodi(A)
subbadius(A)
lepIdus(A)
monoceros(A)
arcuatus(A)
canuti(_A
creaghr( I
coellophy lus(A)
euryotis(A)
pearsoni(A)
yunanensrs(A)
Inops(A)
rutus(A)
subruiusXA)
shameII( )
luctus(A)
triioliatus(A)
sedulus}?
maclau I( 1
macrotis(A
marshalli(A)

raw?)
para oxolophus(A)
ggilippinensisM)
rneensis(A)
thomaSI(A)
attinis(A)
nereis(A)
virgo(A)
srmplex(A)
stheno(A)
celebegsif(A)A
m a us
mglzaygnjirsms )
rouxi(A)
swinnyi(F)
simulator(EF)
ferrumequrnum(EFA)
alo uens(F
turn atus 2
hiide ran ti F)
capensis'(:F)
clivosus )
darlingi( )
hipposideros(EF)
alcyone(F)
denti(lF)(E
eu ae
merhelyiéER
landerI( F)
blasii(EF)
adamI(F)

 

Figure 4.20: The majority consensus cladograrn for the 1200 most parsimonious
cladograms resulting from the weighted analysis, character I ordered. The geographic
location of the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ =
Africa, and ‘E’ = west Eurasia. The numbers indicate the percentage of cladograms in
which this particular branching structure is present.

 

 

Mr

denti

fem”
group
of thr
(19051

by Sir

recon;
osgoa;
traditio
which i
monOpl
Mach”),
sFifties o
sPatties in

R ”10001

91

One of the most obvious patterns present in this cladograrn is the species branching
fi'om Afiica and western Eurasia near the root of the cladograrn While all species distributed in
southeast Asia plus one Afiican species (R mackmdr) constitute a monophyletic group
(synapomorphies in characters 8, 16, 24, 25), as a sister group of the Afiican species R
wimryi. Among Afiican and western Eurasian rhinolophids, seven species, R alcyone, R
denti,R. euryale, R mehelyi, R. landeri, R blassi, andR adamr'formagroupthatisseparated
basally fi'om the remaining species. Except for R. adami which has been placed in the
femlmequimmr group, all these species are members of the traditional pusillus group. This
group of seven species is defined by synapomorphies in characters 1, 3 and 24. Another group
of three species, R eloquens, R fianigatus and R hildebrandti, recognized by Andersen
(1905b) as African members of the firmigatus group, also form a monophyletic group defined
by synapomorphies in characters 2, 4, 19, 24 and 26. The remaining Afiican and western
Eruasian species are resolved into a series of dichotomous relationships.

Two major monophyletic groups are apparent within the southeast Asian clade in this
reconstruction. The first includes R. acraninarus, R pusillus, R cornutus, R. imaizumii, R
oagoodi, R subbadius, R Iepidw, and R monoceros, all southeast Asian members of the
traditional pusillus group. The only synapomorphies for these species is state d in character 1,
which has been the primary feature defining the traditional pusillus group. The second major
monophyletic group of southeast Asian species includes R Iuctus, R trifoliams, R sedulus, R
maclaudi, R macrotis, R marshalli, R rex, R paradoxolophus, and R philippinensis, all
species of the traditional philippinensis group. In this clade, R macrotis differs from all other
species in its higher and relatively more acute connecting process (character 1). Three species,

R macrotis, R paradoxolophus, and R philippinensis are more primitive by absence of sella

92

lappets (character 2). The synapomorphies invariably present in all species of this clade are
enlargednasal swellings (statecincharacters 19) and expended sella(statec in character 2). It
isveryinterestingthatthiscladealsoincludesR maclaudi, theonlyAfiican speciesfoundin
this monophyletic group of an otherwise exclusively southeast Asian species group. R
maclaudi clearly presents character states in characters 1, 2, 19 which are synapomorphies for
thisclade. Apparently,theseremarkable similaritiesbetweenR maclaudiandthesoutheast
Asian species of the traditional philippinensis group placed this Afiican species with the
southeast Asian species, even though R maclaudi does not possess derived states in character
5 and 9 which are synapomorphies for all southeast Asian rhinolophids.

Eleven other species join the philippinends group to form a larger monophyletic unit.
Phylogenetic relationships among these species were not resolved in the consensus cladograrn.
Nine of these species, constituting the traditional arcuatus species group, are R arcuatus, R
canuti, R creaghi, R coelophyllus, R erawtr’s, R mops, R 110515, R shameli, and R
submfils. The other two species , R pearsom' and R yuncmensis, are Asian members ofthe
traditional firmigatus group. The phylogenetic reconstruction implied by Figure 4.19 indicates
that the hypothetical ancestor of this larger monophyletic group had synapomorphies in
characters 2, 3, 20, and 22.

The remaining southeast Asian species, which together constitute the southeast Asian
members of the traditional ferrumequinum group, are situated at the base of the southeast
Asian clade. Relationships among these 11 species are not resolved in the consensus
cladograrn. Under this reconstruction, even the southeast Asian members of the
femrmequirmm group are paraphyletic. \Vrthout additional shared derived characters, the

shared shape of the connecting process alone (the state used traditionally to define this species

93
group) did not provide sufficient evidence to unite these species. Their shared shape in
connecting process is not a true synapomorphy. The synapomorphies defining the clade of all
southeast Asian species are characters 5, 8, 9, and 16.

In general, then, the strict consensus of this data set yields a phylogenetic hypothesis
that recognizes six major monophyletic groups: (a) southeast Asian species of the traditional
pusillus group; (b) Afiican and western Eurasian species of the pusillus group; (c) species of
the traditional philippinensis group; ((1) species of the traditional philippinensis and arcuatus
groups; (e) all species from southeast Asia plus R maclaudi from Afiica; and (t) Afiican
species of the traditional fiam’gatus group. The majority consensus cladograrn (Figure 4.20)
indicates that two additional monophyletic groups are supported by a majority of, but not all,
shortest cladograms in the analysis. The majority consensus recognizes the traditional arcuatus
group as a monophyletic group, and groups several species of the traditional femrmequimmr
group together as a monophyletic group. The presence of these two monophyletic groups in
the majority consensus, but not in the strict consensus, suggests that the data provide weaker
support for a phylogenetic hypothesis that recognizes these as monophyletic.

The shortest cladograms have a consistency index (the ratio of the length of
innovative transformation length to total length of transformation) of 0.228. This means
that on average there are nearly 3 .5 convergence or reversals after each original character
transformation. This low consistency index indicates that a considerable number characters
used in this analysis are relatively unstable. Each of the characters 3, 7, 9, 15, 20 and 21,
in particular have homoplasy ratio of six or greater. The phylogenetic relationships based

on these characters should be carefirlly examined.

94

outgroup
acuminatus(A)
pusil.lus(A)
aflInIs(A)
arcuatus(A)
canuti(A)

cceealgpih Aline“)

euryotisA

inops(A))
luctus(A)
maclaudI(F
macrotis(A
philippinensis(A)

rex A)

para oxolo hus(A)

marshalli(A A1>
triioliatus((A)
sedulus( (A)
yunanensrs(A)

pearsoni(A)

\ rutus(A)

‘ shameli( (AA
subrulus( )
borneensis(A)
celebensis(A)
cornutus A)
Iepidus(A
imaizumii(A)
osgoodi(A)
monoceros(A)
nereis A)
virgo()
simp plexgk)
stheno()
subbadius(A)
thomasi(A)
maiaranusM)
rouxrA
megaphyllus(A)
simulator(EF)
swinnyi(F)
alcy IcyoneF)
blasii(E)
landeri(EF)
denti(F)

eurKaIe(EF
me alyi(E )
capensisg)
clIvosus
elo quens(
tum atus( )
hilde randti(F)
terrlumequinuMEFA)
dar "19:51) EF)
hippos eros(
adami(F)

Figure 4.21: The strict consensus cladograrn for the 1300 most parsimonious cladograms
resulting fi'om the weighted analysis, character 1 unordered. . The geographic location of
the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ = Afiica, and
‘E’ = west Eurasia.

95

 

outgroup
acuminatus(A)
pusillus( (AA
cornutus( )
imaizumii(A)
osgoodi(A)
subbadius(A)

1m 1:

1
2E

78 I d A
—_1: :Ighggérgs(A)
13%

’ZE'E

 

 

aiiinisiA 1
nereIs A
virgo()
simplex(A)
stheno(A)
celebensrsSA)
arcuatus(A
canuti(A)
creaghi(A
euryotis( )
90 "“I’ps‘r’

ru us( )
88 —£ subruius(A)
shameli(Afl
coellophy us(A)
luctus(A)
maclaudI(F)
macrotis(A)
phiIippinensis(A)
rex(A
para hoxolophusM)
marshal IIA
triioliaIus(A)
sedulus(A)
yunanensis(A)
pearsoni(A)
borneensis(A)
tmhomasi(A)

h "us A
misfiaygniism )

1CD

 

 

 

 

 

 

 

 

 

18

 

 

 

 

 

 

 

 

 

 

 

1m

 

F13

rouxi(A)
simulator(EF)

 

 

 

 

 

 

LandiagSEF)
[— enti
100 \Q:e mugrzaleiEF
eri(E )
capensis?)
clivosus()
ierrumequinum(EFA)
eloquens(F)
‘00 l___Ium atus(
l-—-hi|de randti()F)
darlingi(F) EF
hipposideros( )
adami(F)

 

 

 

 

 

 

 

 

 

 

Figure 4.22: The majority consensus cladograrn for the 1200 most parsimonious
cladograms resulting from the weighted analysis, character 1 unordered. The geographic
location of the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ =
Africa, and ‘E’ = west Eurasia. The numbers indicate the percentage of cladograms in
which this particular braching stucture is present.

96

Analysis of Weighted Characters, Character 1 Unordered

This analysis differs from the preceding one by removing the transformation series for
character 1 and treating this character as unordered. The strict and majority consensus
cladograms fi'om this analysis are presented in Figures 4.21 and 4.22, respectively. The
topologies of these consensus cladograms difi’er fi'om those of the previous analysis in
important ways, indicating the important role of character 1 in delineating monophyletic groups
within the genus.

The monophyletic group containing the traditional philippinensis and wcuarus groups
Previously analyzed remains in the consensus cladograrn for the present study. \Vrthin this
group, two species (R pearsoni and R yunanensr's) are placed with the traditional
philippinensis group species rather than the arcuatus group. More markedly, the traditional
pusillus group, clearly monophyletic in the previous analysis, is less consistently present in the
cladograms produced by the present analysis. This group is not present in the strict consensus
cladograrn of Figure 4.22, although it is present in the majority consensus (Figure 4.22). This
outcome is likely due to the decreased number of steps needed to change between certain
states of character 1 fi'om multiple to single step. A change between state c (e.g. R
philippinensis) and state d (e. g. R pusillus) of character 1 in an ordered analysis adds a length
of 16 units to the cladograrn. An unordered analysis adds only four units to the cladogram,
which makes a group primarily defined by character 1 less stable. As shown in the previous
analysis, only two synapomorphies for the southeast Asian member of pusillus group were

characters 1 and 7, in which character 7 is in the second lowest weighting group. The features

97
of the traditional pusillus group are otherwise relatively primitive. A small clade, including
three species of fen-unrequimrm group (R nereis, R Virgo and R simplex), is present.

Both consensus cladograms for this analysis indicated a more dichotomous pattern of
relationship for Afiican and western Eurasian species than did the previous analysis. This was
primarily due to the disintegration of the clade consisting of the Afiican and west Eurasian
members of the traditional pusillus group. This clade, defined by synapomorphies in characters
1, 3, 9 and 25 and containing 7 species in Figure 4.19, was reduced to a much smaller clade of
only 3 species in the present analysis. When the hypothesized transformation series for
character 1 was applied (ordered), a transformation from state a (represented by
ferrumequinum group) to state d (represented by pusillus group) required two steps; when no
particular transformation series for character 1 was assumed or unordered, the same
transformation is achieved in one step. The species of the traditional pusillus group have
moved from the base of the cladograrn to more derived positions among the African and west
Eurasian species in which a reversal of character 1 occurred. Three species of the traditional
firmigatus group, R eloquens, R fumigatus and R hildebrwrdtr', form a monophyletic group
as they did in the previous analysis. No monophyletic groups that were identified in this
analysis were not found in the previous analysis. The characters responsible for this
dichotomous branching pattern near the base of the cladograrn in the present analysis include
characters 1, 4, 7, 11, 12, 16, 18, 21, and 25.

Overall the unordered analysis resulted in less resolved consensus cladograms than the
ordered analysis. The strict consensus identifies three major monophyletic groups: a clade for
all members of the traditional philippinensis group, a clade for all members of the traditional

wcuatus and philippinensis groups, and a clade for all southeast Asian species plus R

98

outgroup
megaphyllus(A)
nereis A)

virgo( )
simplex(A
pearsonI( )
pehilippinensism)

luctus(A)
maclaudILF;
macrotIs A
sedulus( )
triioliatus(A)
paradoxo hus(A)
marshaIIi A
—— arcuatus A
canuti(A)
inops(A)
ruius(A)
subruius A)
shameli( 3
coellophy Ius(A)
creag hi( (AA
euryotis( )
yunanensis(A)
rou'xi(A) A
maa anu
ailinigm)“ )
borneensis(A)
celebensis(A)
stheno(A)
thomasi(A)
monoceros(A)
lepidus(A)
pusillus(A)
acuminatus(A)
imaizumii(A)
subbadius(A)
cornutusXA
os oodiE (é):
me 91Yi(EF
eu ae
alcVone(( )
capensis
clivosus(
hiIdebrandti(F)
eloq unens(2
IumIgatus( )
hipposideros(EF)

 

adami I(
simulator(EF)
swinnyi( F

darlingi(F)
ferrdumeguinum(EFA)
land on

denti((F

blasii( F)

Figure 4.23: The strict consensus cladograrn for the 1300 most parsimonious cladograms
resulting from the unweighted analysis, character 1 unordered. . The geographic location
of the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ = Afiica,
and ‘E’ = west Eurasia.

99

 

59

muetgaph'; "us(A)

 

pearsonI(A)

 

95

yunanensis(A)

 

philip inenSIs(A)

 

rex(A
luctus(A
sedulus A)
triloliatus A)
maclaudi )

 

marshalli A)

 

 

paradoxo phus(A)

 

 

81

Is

 

1w

 

 

 

 

 

97

81

LE

macrotis(A)
arcuatus(A)
area h AA)
canu i()

84 .— inops(A)

 

 

 

'—- euryotis(A)

Ioo rufus(A)

 

subrulusAA)
shameli(

 

ooellophyllus(A)
rouxi A)
malayanus(A)
l——borneensis(A)

1‘33:

 

 

100

 

100

92

 

 

100

E

 

thomasi(A)
nereisAA)
VIrgo(
simplex)(A)
alfinis(A
stheno(

 

celebenSIs(A)
monoceros(A)
Iepidus(A)
pusillus(A)
acuminatus(A)

 

 

imaizumii(A)

 

corn utus(A)
subbadius(A)

 

 

 

100

 

Figure 4.24: The majority consensus cladograrn for the 1200 most parsimonious
cladograms resulting from the unweighted analysis, character I ordered. The geographic
location of the species is indicated by letters in parentheses: ‘A’
Africa, and ‘E’ = west Eurasia. The numbers indicate the percentage of cladograms in

 

 

 

osgoodi(A
mehelyi(E )
ewaled?)
a on
Ianderi(éF§
denti(F)

 

 

capensis?)
clivosus
simulator( F)
swinnyi(F)
lerrumequinumLEFA)
hipposideros(E
adami(
darlingi( )
hildebrandti(F)
eloquens(F
turmgatus( )

 

which this particular braching stucture is present.

blasiI(EF)

= southeast Asia, ‘F’ =

100
maclaudi. The majority consensus cladograrn identified two additional monophyletic groups,
one for eight species ofthetraditional southeast Asianmemberofthepusr’llusgroup andthe
other for six species of the traditional southeast Asian member of the fenumequiman group.
The synapomorphies for both clades were the same as those in consensus cladograms from
weighted and ordered analysis.
The consistency index for the most parsimonious cladograms of this analysis was 0.23,

slightlyhigherthanthatinthepreviousanalysis.

Analysis of Unweighted Characters, Character 1 Ordered

The strict and majority consensus cladograms from this analysis are presented in
Figures 4.23 and 4.24, respectively. When characters were unweighted, the resulting strict
consensus cladograrn indicated a clear division between southeast Asian species and Afiican
and western Eurasian species of the genus; both form distinct, monophyletic groups. The
monophyletic Afiican and western Eurasian clade in this analysis, a paraphyletic group in the
previous two analyses (above), is defined by synapomorphies in characters 13, 17, 25, and 26,
while the monophyletic group of southeast Asian species are related by synapomorphies in
character 1, 4, 5, 6, 7, 8. As happened in the weighted analyses, the Afiican species R
maclaudi is found in the southeast Asian clade.

In the southeast Asian clade, Figure 4.23 shows the outlines of the traditional
fermmequimm, arcuatus, pusillus and philippinensis species groups though virtually no
pattern of relationship is resolved within each group. The cladograrn indicates that the
traditional philippinensis group is the most derived. The philippinensis group and arcuatus

group together constitute a larger clade defined by the same set of synapomorphies, character

101
1, 2, and 3, as in the weighted analyses. These two groups are fiuther joined by 10 species of
the traditional fen-unrequinum group, making a more inclusive clade, defined only by
synapomorphy in character 1. Finally, members of the traditional pusillus group are found at
the base ofthe southeast Asian clade.

The present study recognized a monophyletic group of 10 species within the Afiican
and western Eurasian clade, R. clhnsus, R simulator, R swimorr’, R fermmequiman, R
hnpposiderm, R adami, R darlingi, R hildebrandti, R eloquens, and R fiam’gatus, which
was not found in previous, weighted analyses. The synapomorphies for this clade were
characters 4 and 11. This clade, joined by one more species, R capensis, forms a larger
monophyletic group defined by synapomorphies in character 1 and 12, including all the African
and west Eurasian species of the traditional fenwnequinum and fimrigatus group. The
monophyletic group of R eloquens, R firmigatus and R hikiebrwxlti, found in the previous
two analyses, is also present in Figure 4.23. However, the monophyletic group found in the
first analysis (Figure 4.19), consisting of all Afiican and west Eurasian species of the traditional
pusillus group, was not present. These species branch from the base of the clade, with
otherwise unresolved relationships.

The mq'ority consensus cladograrn displays the relationships within the traditional
philippinensis and arcuatus groups; these two groups are resolved into sister groups. Two
more monophyletic groups, one for 7 southeast Asian species of the fen-umequimrm group
and another for 5 African and west Eurasian species of the pusillus group are present in the
majority consensus cladograrn

The most significant difference of this analysis from the weighted analysis is the

presence of a clade for all the Afiican and west Eurasian rhinolophids. Most of the

102

outgroup
acuminatus(A)
affinis(A)
arcuatus(A)
canuti(A)
coellophxllusm)
creaghi(2
euryotIs( )
inops(A)
rufus(A)
subruqu A)
shamelI( )
luctus(A)
maclaudI(
macrotis(A .
phIlepInensrsM)
para oxolophus(A)
rex(A)
marshalliulke
trItolIatus( )
sedulus(A)
yunanensrs(A)
’arsoni(A)

rneenSIs(A)
celebensisSA)
cornutus(A
Iepidus(A)
monoceros(A)
imaizumii(A
malayanus A
megaphyllus A)
nereis(A)
Virgsll(A)A

usr us
gouxi(A)( )
simplex(A)
stheno( )
subbadius(A)
thomasr(A)
osgoodi(A)
alcyone F)
swmn I
blasiiéF)’
capensis?)
clivosus( )
ferrumeguinuMEFA)
darlin I( )

' . ' denti( )
eloquens(F
lumIgatus( )
adamI(F
hildebrandti(F)
eurgale(EF
me elyj E )
hlppOSl eros(EF)
landeri(E
simulator( F)

Figure 4.25: The strict consensus cladograrn for the 1300 most parsimonious cladograms
resulting from the unweighted analysis, character 1 unordered. The geographic location
of the species is indicated by letters in parentheses: ‘A’ = southeast Asia, ‘F’ = Africa,
and ‘E’ = west Eurasia.

103

 

68

_;’.5___:

 

77

 

 

 

100

 

 

100
ME

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

92

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

 

outgroup
acuminatus(A)
pusillus(A)
——aflinis(A)
celebensis(A)
nereis(A)
virgo(A)
simplex(A)
stheno(
subbadius(A)
Iepidus(A)
monoceros(A)
arcuatus(A)
creaghAA)
canuti
auryotlszA)
inops(A)
shameli(A2
coellophy lus(A)
rufus(A)
subrulus(A)
luctus(A)
maclaudI(F;
macrotis A
philip inensis(A)
para oxolophus(A)
marshalli(A)
rex(A)
triloliatus(A)
sedulus(A)
yunanenSIs(A)
pearsoni(A)
malayanus(A)

megiaphyllusM)
borneensis(A)
thomasi(A
cornutus( K
imaizumii(A
osgoodi(A)
alcyone(F)
swinnyi(F
simulator EF)
clivosus( )
Ierrume guinum(EFA)
darlingi( (a?
eloq uens
Iumqlgatus( )
adami( F)
hildebrandti( FE
hipposideros( F)
landen(EH
denti(F).
capensis F
flrzalq

IiéEF)

l—[:blasii(

Figure 4.26: The majority consensus cladograrn for the 1200 most parsimonious
cladograms resulted from the weighted analysis, character I ordered. The geographic
location of the species is indicated in the letters in the parenthesis: ‘A’ = southeast Asia,

‘F’ = Afiica, and ‘E’

at which this particular braching stucture is present.

= west Eurasia. The numbers indicate the percentage of cladograms

104

synapomorphies for this clade, character 13, 17, 25, and 26, had low weight in the previous,
weighted analyses. When characters were unweighted, the relatively large number of shared
derived characters constitute strong evidence to support this monophyletic group. In contrast,
two monophyletic groups in the first analysis (Figure 4.19), consisting of the traditional pusillus
group species from two difi‘erent regions, both defined by a single synapomorphy in high
weight character 1, disappeared in this analysis.

The consistency index for the shortest cladograms in this analysis is 0.222, slightly
lower than those in the two previous analyses. The total number of transformations implied by
the phylogenetic hypotheses is higher in the weighted analysis (245) than in the present analysis
(23 8). But by reducing the number of convergent and reversal transformations in high
weighting characters, phylogenetic reconstruction of the weighted analyses had higher

consistent indices.

Analysis of Unweighted Characters, Character 1 Unordered
Figures 4.25 and 4.26 present strict and majority consensus cladograms, respectively,
for analyses of unweighted characters, with character 1 not ordered by a transformation series.
As was the case in the previous unweighted analysis, Afiican and western Eurasian species
form a clearly monophyletic clade. This clade was defined by synapomorphies in characters 7,
13, 25, and 26. The southeast Asian clade, on the other hand, is defined by synapomorphies in
characters 4, 5, 6, 7, and 12. Character 1 did not play a role in the major division of the genus
in either unweighted analyses.
In the strict consensus cladograrn from this analysis, an additional monophyletic group,

not found in the previous unweighted analysis, was apparent. This group contains all the

105
species of the traditional wcuatus group, defined by synapomorphies in characters 10 and 26.

Together with another monophyletic group of the traditional philippinensis group, they form a
larger monophyletic group recognized by the other three analyses. As with the previous
unweighted analysis, there was very little pattern of relationships among Afiican and western
Europeanspeciesinthestrictconsensuscladogramofthisanalysis. Thefourspecies, R
eloquens, R fianigatus, R hildebrandti and R adami form a monophyletic group defined by
synapomorphiesincharacters4and11. Thefirsttlueespeciesconstituteamonophyletic
group in the previous three analyses.

The majority consensus cladograrn fi'om this analysis was similar to that from the
previous unweighted analysis in the relationships displayed for the southeast Asian species.
Within southeast Asian species, an additional monophyletic group containing 11 species
(definwbythesynapomorphiesincharacternwaspresent. Amongthe 11 species, sixspecies
of the traditional femanequinum group form a smaller clade, which was also present in the
majority consensus of all three previous analyses; the other five species were fi'om the
traditional pusillus group. The relationships among the Afiican and west Eurasian clade
resulting from this analysis did not agree well with those from the previous three analyses. R
simulator, a species of the traditional fenwnequimnn group, was found closely related to two
species of the pusillus group (R denti and R landeri). This clade has a single synapomorphy
in character 26.

The consistency index for the most parsimonious cladograms in this study was 0.220,

being slightly lower than other three analyses.

106
The Status of the fiam‘gatus Group.

Two of the traditional species groups recognized within Rhinolophus have not been
discussed in the foregoing descriptions of the cladistic analyses. One, the hipposideros species
group of Andersen (1905b), is monotypic, containing only the species R hipposideros. This
species ‘group’ was trivially present in all of the analyses, because the cladograms do not
reflect the degree of specialization any particular species may reach The other species group,
Corbet and Hill's (1992) fianigatus group, fonneriy the macrotis group of Andersen (1918),
deserves more discussion

Andersen (1918) and Corbet and Hill (1992) diagnosed thefimrigatus group based on
sella shape and connecting process (characters 1 and 2), the margins of the palate (characters
16 and 17), and ear size (character 23). Although these characters were included in the
cladistic analysis, none of the cladograms indicated a monophyletic group of these 5 species (R
eloquens, R firmigatus, R hildebramlti, R pearsoni and R yrmanensis). Bogdanowicz
(1992) further divided the firmigatus group, separating R pearsom’ and R yunanensis (Asian
species) as the pearsoni species group distinct from a restricted filmigarus group containing the
three Afiican species only. This view was supported by my analysis, since these three Afiican
species (R eloquens, R fianigarus, R hikiebrandtr) fiom a monophyletic group in all of the
consensus cladograms. However, the justification for a distinct group containing R pearsoni
andR Wm was not as evident. Although these two species are very close in all
consensus cladograms, they did not appear to be sister species in all the consensus cladograms.
Their relationships with other species are also sensitive to the change of assumptions. In the
strict consensus cladograrn of the weighted analysis where character 1 is ordered, these two

species are within the monophyletic group containing the traditional arcuatus and

107
philippinensis groups, sharing derived characters 2, 3, 20 and 22. In the strict consensus

cladograrnsofthetwo analyseswherecharacter l isunordered, theyaresisterspeciesofthe
monophyletic group containing traditional the philippinengs group only, sharing derived
characters 5 and 21 with them. There were some suggestive evidencethatR pearsom' andR
Wing's are distinct from the traditional warm and philippinensis groups. Based on the
relationships that are common to all the strict consensus cladograms, these two species are
members of the arcuatus + philippinensis (+ R pearsoni and R yumnsis) clade but not
within the philippinensis clade. In the absence of evidence that would firrther clarify their

relationships. I treated these two species as unresolved within the former clade.

Comparisons Between the Analyses

Thefoursetsofdadisficanalysesdifi‘erinchmacterweighfingandwhetherornot
character 1 was represented by a particular transformation series. The substantial differences
between these assumptions could have produced totally difi‘erent patterns of relationship in the
resulting consensus cladograms. That many of the same monophyletic groups appeared in
most, if not all, analyses was therefore surprising. It is necessary to examine the details of the
monophyletic groups present in each analysis to reach an appropriate phylogenetic hypothesis
for the genus Rhinolophus.

Four monophyletic groups were consistently present in the strict consensus cladograms
of all four analyses. The first included species of the traditional philippinensis group. The
second was the first clade plus species of the traditional arcuatus group and two Asian species
of the traditional firmigatus group. The third group contains all the species from southeast

Asian plus the African R maclaudi. The final clade contains the three species, R eloquens,

108
R fiam’gatus, and R hildebrandti. Because these groups are present in all cladograms

produced in this study, despite very different assumptions involved, I concluded that those four
monophyletic groups are very strongly supported by the data set.

One monophyletic group, containing all Afiican and west Eurasian species (except the
Afiican species of R maclaudr), supported by a relatively large number of synapomorphies
(characters 7, 13, 25, and 26), was present in the strict consensus cladograms fi'om the
unweighted analyses but not in those from the weighted analyses. This difference poses a
question about the basic phylogenetic division of genus: whether the group of southeast Asian
species were derived from the group of Afiican and west Eurasian species, or these two are
sister groups. I decided that the southeast Asian group was derived from the Afiican and west
Eurasian species for two reasons. First, all of the synapomorphies that define the Afiican and
west Eurasian clade are relatively low in information content (discussed in Character Analysis)
and were placed in weighting groups one (characters 13, 25, 26) or weighting group two
(characters 7). The groups defined by these characters, therefore, were less reliable. Second, a
consensus cladograrn for the results fi'om both the weighted analyses and unweighted analyses
(Figure 4.27) would place all the Afiican and west Eurasian species as well as the monophyletic
group of southeast Asian species at the root. Although the relationships among Afiican and
west Eurasian rhinolophids remain unresolved, the cladograrn clearly suggested that the species
of southeast Asia were derived from the ancestors in the Afiica and west Eurasia

Relationships patterns among the Afiican and western Eurasian species differ greatly
among the analyses. The relationships among these species were not resolvable with the
current data set. At the very least, to determine a reasonable hypothesis of relationship among

these species would require deciding whether characters should be weighted, and whether the

109

African & west southeast Asian
Eurasian species species
l‘ s
Afriean & WESI southeast Asian
Eurasian species species
African & west southeast Asian
Eurasian species species

(C)

 

Figure 4.27. Cladograms illustrate the consensus between the results from the
weighted and the unweighted analyses. (a) Results from the weighted analyses,
Afi'ican and west Eurasian species branch from the base of the cladograms; (b) Results
fi'om the unweighted analyses, Afiican and west Eurasian species constitute a
monophyletic group; (c) In the consensus cladograrn for (a) and (b), Afiican and west
Eurasian species as well as the monophyletic group of southeast Asian species branch

fi'om the multichotomous root.

110

proposed transformation series for character 1 is really appropriate. With the current data set,
the relationships among the African and western Eurasian species are very sensitive to how
these questions are resolved. I concluded that the present data set does not support an
unambiguous hypothesis for the relationships of these species. Two more significant
monophyletic groups are unique to the weighted and character 1 ordered analysis. They are the
species of the traditional pusillus group from southeast Asian and those fi'om Afiica and west
Eurasia, respectively. Character 1 is virtually the only synapomorphy for both groups.

Inconsistencies due to different assumptions about southeast Asian species are less
sever, two monophyletic groups, one for all species of the traditional philippinensis group,
another for all species of traditional philippinensis group plus arcuatus group, were present in
all consensus cladograms. The monophyletic group for the southeast Asian members of the
traditional pusillus group was present in the strict consensus cladograrn from the weighted and
ordered analysis and in the majority consensus cladograrn from the weighted and unordered
analysis but is not present in unweighted analyses. Because this inconsistency was about
resolution rather than conflict, this monophyletic group should be accepted based on the
present data. Another group containing six southeast Asian members of the traditional
femanequimrm group (R. qfi'inis, R. nereis, R. Virgo, R. simplex, R. stheno, and R. celebensis)
was present in the majority consensus cladograms of all analyses but not present in any of the
strict consensus cladograrn. Because both synapomorphies of this group (characters 7 of
weighting group two and character 10 of weighting group one) were of lower information
content, I considered that this group was unreliable.

The African species of R. maclaudi is in the traditional philippinensis group clade in all

the consensus cladograms and it was placed in the philippinensis group by most previous

111

researchers (Andersen, 1918; Koopman, 1975). Nevertheless, it has primitive features in
character 5 (with one lower lip groove) and character 9 (with more than five caudal vertebrae)
which resembles other Afiican and west Eurasian species. A more serious question is how it
occurs so distant fi‘om all other species of that group. There is no indication of such
distributional pattern in other groups of the genus. Considering the marked rate of homoplasy
in the morphology of the genus revealed by this study, a convergent evolution of species
acquiring features characteristic of the philippinensis group can not be entirely ruled out.

\Vrthout further morphological and distributional evidence about the this group, I find the

status of R maclaudi can not be concluded at this time.

Taxonomic Summary

I present the summary cladograrn in Figure 4.28 to indicate the monophyletic groups
strongly supported by my data set. This cladograrn includes all monophyletic groups present in
all strict consensus cladograms from the four armlyses, plus the clade containing southeast
Asian species of the traditional pusillus group present in strict consensus cladograrn in Figure
4.20 (weighted and character I ordered) and majority consensus cladograrn in Figure 4.23
(weighted and character 1 unordered). The species at the base of the southeast Asian clade, all
belonging to the traditional femmrequinum group, are not resolved into a clade in any
consensus cladograrn and, are represented as an unresolved group. This cladogram does not
resolve the relationships of all of the species of Rhinolophus. Instead, it draws attention to
those members of the genus whose phylogenetic relationships are supported well enough in this

analysis to merit taxonomic recognition at this time.

112

philippinensis group (the
traditional philippinensis group)

(the traditional arcuatus and
SE Asian species of fitmigatus
group, unresolved)

pusillus group (SE. Asian
species of the traditional

pusillus group)

. (SE. Asian species of the
' - traditional ferrumequinum
group, unresolved) .

 

fumigatus group (Afiican
species of the traditional
firmigatus group)

. (All Afiican and west Eurasian
species, unresolved)

Figure 4.28. The phylogenetic relationships within the genus Rhinolophus based on the
present study. The monophyletic groups (bold faced) strongly supported by my data set are
indicated by solid lines A dotted line represents a set of species branching fi'om that point;
relationships among these species are rmresolved

113

Table 4.2. Summary of taxonomic conclusions based on the monophyletic groups in Figure
4.28. No paraphyletic groups is recognized in this taxonomy. Monophyletic groups of
species are recognized at three different levels (supergroup, group, and subgroup). Those
species that can not be placed into a monophyletic group are included as ‘status uncertain’ at

the appropriate level.

CENU S RHINOLOPHUS
afiinis wbgenus R qfiim‘s
philippinemrls mpergroup R nereis
philippinensis group R simplex
R luctus R stheno
R mfoliams R selebensrls
R sedulus R megaphyllus
R macrotis R malayamrs
R mshelli R rouxi
R rex R bomeemrls
R paradaxolophus R thomasi
R philippinemrs
group status uncertain subgenus status rmcertain
R arcuatus (All Afiican & west Eurasian species)
R canuti fumigatus group
R creaghi R eloquens
R coelophyllus R jimrigatus
R euryotzs R hildebrandti
R inops group status rmcertain
R rufils R alcyone
R submfirs R denti
R pearsoni R euryale
R manensis R mehelyi
pusillus group R landeri
R acuminatus R blassi
R pusillus R adami
R comums R clivasus
R imaizwnii R femanequimtm
R osgoadi R darlingi
R subbadius R capensis
R lepidus R sm'rmyi
R monoceras R simulator
group status rmcertain R luppasideras
(southeast Asian species
of the traditional subgenus status rmcertain

fenwnequinum group) R maclaudi

1 14

Table 4.2 summarizes my taxonomic conclusions based on the monophyletic groups of
Figure 4.28, and the diagnosis for these monophyletic groups is presented in Table 4.3.
Monophyletic groups of species are recognized at the species group, supergroup, and subgenus
levels. Only the four monophyletic groups which are strongly indicated in all the analyses, and
the pusillus group which is indicated in two weighted analyses, are assigned group (subgenus,
group, and subgroup) names. The decision to recognize the pusillus group does not affect the
relationships among the other monophyletic groups. I chose not to recognize paraphyletic
groups in this taxonomy. Those species that can not be placed into a monophyletic group are
included as ‘status uncertain’ at appropriate levels. While this approach results in an unusual
number of ‘status uncertain’ designations, it does draw attention to the parts of the taxonomy
that require firrther clarification.

Among the five designated monophyletic groups, the philippinensis group contains the
same species as the traditional philippinensis group (Andersen, 1905b, l905f; Tate, 1943;
Corbet and Hill, 1992), referred to as the Iuctus group Andersen, (1918); Ellerman et al,
(1953); Koopman, (1975). Although the species name of R Iuctus Temminck, 1835 predates
the species name of R philzppinensis Waterhouse, 1843, the latter name was the first to be
used for this species group. By Article 23 (Principle of Priority) of the International Code of
Zoological Nomenclature (Ride et al, 1985, referred to as the Code in this section),
philippinensis is the valid name of the group. The pusillus group contains the southeast Asian
member of the traditional pusillus species group. Since this monophyletic group includes the
nominaltypical species, R pusillus, of the traditional species group, by Article 37
(Nominotypical taxa) of the Code, this nominotypical group retains the group name. The

monophyletic filmigatus group of Afiican species retains the traditional group name for the

115

 

 

 

 

 

s; 52: ~88.
come—m stow .«o women—Ea
.3an ”5.03... 322 .m Because .8 sex .m
”39 502 Ben: mo «2 R553 flowed—co
55 comes $2.5 333 .N 389:. «mean; 332 .m
Hows—:0 meuooeeoo .uoeoveofi
332m mnwubnee EB mam A A use. 8.523 new e=om A

98..” sateen: season:
98..» fineneihmaem 53825 9.533% 33% 93..» «Seumfiax. 33% 95..»
«38m anew anew 9.8m

 

602896 common @3585 .m
Econ 53.588 33.0 .N
”3:335 wee—owe: com—35 v5 a=om .—

 

 

 

 

 

 

 

989895
macaw—2...». umueuzmmémam.
. m2 8 Fame
Ewan:— eofloamon 515 352 3033—2 333% 333 .v
”523 55 83on 5&5— Nm .m
”gm .85 530m 3385» 3380 .N 3:033
330on a: $32 ooh; A
assume» Manage 53.00:: 33% 3:033.

 

 

«40>». 03.2833 8 53.00:: menu 8 @885 Pa 320:: 82332 awesome—Ea .23 860mm .56
23mm 5 Banana 338m one—Becca 05 no n83 haemoNoSquo 8g econometfi 2: 8m €808.20 osmoewaa 5+ 033.

116
same reason. Both the philippinensis supergroup and qfiinis arbgenus of southeast Asian
species are new taxa. The philippinensis supergroup contains a single designated group; by
Article 36 (Principle of Coordination) in the Code, it is appropriate to name the group after its
only designated subgroup. I choose R qfi‘im‘s Horsfield, 1823, the eariiest designated nominal
species of the group, as the name of the subgenus. Table 4.3 presents diagnostic characters for
the infi'ageneric taxa I recognize.

This taxonomy of Rhimloplms, though leaving much for firture studies, clearly
indicates the basic phylogenetic relationships and patterns of character evolution within the
genus. This taxonomy difi‘ers fi'om the traditional taxonomy of the genus in three significant
aspects. First, this taxonomy identifies a monophyletic group (as a subgenus) consisting of all
the southeast Asian species, while leaving the taxonomic status of the remaining species as
largely unsolved. Because I used a much larger collection of characters than has been used in
the previous phylogenetic analysis, I was able to detect considerable homoplasy in a broad
range of characters including some widely used in the past (e. g, the shape of connecting
process and the shape of the sella). I found that those species groups defined by these
characters are polyphyletic (e. g., the traditional pusillus group and firmigatus group). Second,
this taxonomy does not recognize the traditional femmrequinum group as a valid species
group. The present phylogenetic study indicates that this traditional species group is
paraphyletic, representing a collection of species that arise at different points in the phylogeny.
The relationship among species of the traditional fenumequimrm remains unresolved. Third,
only monophyletic groups are recognized as taxa in this taxonomy, leaving unresolved groups

as status uncertain.

1 17

My phylogenetic hypothesis of the relationships among the southeast Asian species is
similar to that of Andersen (19053, 1905d, 1905c). However, the present hypothesis differs
from that of Tate, since I consider the traditional wcuatus group closely related to the
traditional philippinensis group whereas Tate viewed the former group to be closer to his
pusilhas and fen'umequiman groups.

Therecognitionofthe subgenusforall southeastAsianspeciesalsodistinguishesthis
taxonomy from the one proposed by Bogdanowicz (1992). Although the separation of species
from the two mq'or geographic regions is to a degree indicated in his phenetic analysis,
Bogdanowicz did not recognize the southeast Asian rhinolophids as a monophyletic group and
did not present them as a distinct taxon in his taxonomy. Furthermore, by recognizing the
monophyletic groups at difi‘erent taxonomic levels, this taxonomy presents a clear view of
firndarnental intrageneric relationships. In Bogdanowicz’s taxonomy the relationships between
his 11 species groups unresolved. In the underlying phylogenetic hypotheses, the present study
hypothesizes that the philippinensis group is a most derived monophyletic group, whereas in
their phenograms (Bogdanowicz and Owen, 1992; Bogdanowicz,1992) this group of species is
divided into two distantly related groups and one of them is the earliest branch of the genus.
Finally, this taxonomy provides a diagnosis for each designated taxon and the hypothesis of
character evolution of the genus, both of which are not available for his taxonomy.

DISCUSSIONS

The consistency indices (CI) are rather low (from 0.22 to 0.23) for all the shortest
cladograms computed. This means that the ratio of convergent and reversive transformation to
the innovative transformations is more than four to one for the characters used in the present

study of Rhinolophus. The differences in the CIs between the four analyses are very

118

Table 4.4 : A comparison in the patterns of transformation between the weighted
and unweighted analyses, character 1 unordered, for each character. In each analysis,
one shortest cladogram, which has a topology identical to the majority consensus of
that analysis, is summarized. Shading indicate the characters with lower occurrence

of homoplasy ratio.

weight— number of number
ters ing of transformation homoplasy in transformation in homoplasy in
group transformation in first analysis first analysis third analysis third analysis

 

119

small, but the ratio of convergent and reversive to the innovative transformations varies greatly
among characters. Table 4.4 shows the minimal number of necessary transformation
(without homoplasy), the actual number of transformations based on the phylogenetic
hypotheses, and the ratio of homoplasy to minimal transformation present in each of the
four analyses. Characters 1 and 2 have relatively low rate of homoplasy in both weighted and
unweighted analyses (between 1 and 2.5). This agrees with the assumption that these
characters are more informative due to their conservativeness. Four of the characters converted
from continuous measurements (characters 22, 23, 24, and 25) also display little homoplasy.
They were assigned low weight because the boundaries between the states of these characters
are relatively arbitrary. Characters 3, 7, 9, and 20 have very high rate of homoplasy; they
contributed less reliable evidence about the phylogeny of the genus.

The fact that some highly weighted characters are less consistent with the shortest trees
does not constitute a compelling reason for a character weight change, since a review of
character analysis after cladistic analysis does not convince me to change the weighting criteria.
However, the low consistency index does reiterate an early recognition that there are many
convergent and reversive changes in the rhinolophid morphology.

Despite a high rate of homoplasy, certain patterns of character evolution can be seen
from the phylogenetic hypothesis in Figure 4.27. The hypothesized ancestors of Rhinolophus
most likely had a small sella and anterior noseleaf, a connecting process of state a, one lower
lip groove, 5 to 6 caudal vertebrae, P2 width greater than length, and posterior palatal
margin not rostral to M3. The sella and anterior noseleaf become broader and the
intemairal region expanded in more derived groups (e. g. in the philippinensis group of the

present taxonomy); both the ear and antitragus are expanded and the nasal swellings are

120

enlarged in the most derived group (the philippinensis subgroup). The connecting process
(character 1) was derived independently in the Africa and west Eurasian species and in
southeast Asian species, though only the latter developed all five states of this character.
State e of character 1 did not evolve in Afi'ican species, and it is unclear whether the state
c has evolved in Afiican species, since the phylogenetic position of R maclaudi is still
questionable. Another trend in character evolution is the reduction of the first! second
phalangeal ratio in the third and fourth fingers, which reached the most derived state, state
d, in some of the African and west Eurasian species (e. g. R mehelyi, and the fumigatus
group). The evolutionary significance of most of these morphological changes within
Rhinolophus is still not clear.

Based on their karyotypic studies, Harada et al (1985) classified the genus into three
groups based on the number of chromosomes. The first group included R creaghi, R
aaanr'natus, R connrtus, R imaizwnii, R malayanus, R coelophyllus, R pusillus, R afinis,
R stheno and R marshalli, all with 2n = 62 including 30 acrocentric autosome pairs. The
second group comprised R euryale, R blassi, R mehelyi, R darlingi, R denti, R
ferrumequinum and R hildebrandti, with 2n = 58 including 25 acrocentric pairs and two
metacentric pairs. The third group included R hipposideros, R luctus and R mmnsis, all
with some large metacentric autosome pairs and 2n = 32. The last group is most similar to the
karyotype of Hipposideros. Furthermore, considerable variation in chromosome number was
found within the three subspecies of R Iuctus.

The first two karyotypic groups correspond to the two major geographic groups
discussed in this study. My results suggest that the 2N = 62 karyotype may represent a derived

karyotype, since it occurs in the qfiinis subgenus. If so, I would expect this karyotype to be

121

found in other members of the group (or serve as the ancestral karyotype for others that might
be found there). The 2N = 58 karyotype may represent a more primitive karyotype since it is
present in species that are scattered across the phylogeny. My results do not support the notion
that the 2N = 32 karyotype is primitive to the genus. The three Rhinolophus with this
karyotype are not cleariy related in my phylogenetic hypothesis, nor are they located near the
baseofthecladogram.1predictthatthekaryotypesofthese specieswillbefound tobe
convergerrtly similar to each other and to Hipposideros. While the karyotypic data available is
incomplete, it offers tantalizing suggestions about the complexity of generic chromosomal
evolution in the genus.

Considering the rate of homoplasy in the morphological characters used in this study, a
more conclusive view of the phylogeny and systematics of Rhinolophus may require more

molecular and cytogenetic technology data.

THE HISTORICAL BIOGEOGRAPHY OF

SOUTHEAST ASIAN RHINOLOPHUS

INTRODUCTION

I concluded in a previous section that the species of Rhinolophus occurring in
southeast Asia constitute a monophyletic group. To use Sclater’s classic biogeographic terms,
this distribution covers most of the Oriental realm, northern part of the Australian realm and a
small southeastern portion of the Paleoarctic realm regions (Holloway and Jardine, 1968). This
general area inhabited by Rhinolophus in southeast Asia has fi'equently been referred to as the
Indo-Australian region (Tate, 1939) and the Indo-Malay region (Koopman, 1989; Corbet and
Hill, 1992). Figure 5.1 shows the southeast Asian region.

Southeast Asia has been of great biogeographic interests since Alfi'ed Wallace’s (1860)
publication which demonstrated the strikingly discontinuous faunas present on adjacent islands
in the Malay Archipelago. Wallace recognized these discontinuities by what is now referred to
as Wallace's line. A somewhat different line was proposed by Huxley (1868). Biogeographic
studies of diverse animal and plant groups have been carried out in this region, resulting in
various different proposals for where a line should be drawn to delimit the Oriental biota from
the Australian biota (George, 1981). Some of these lines are illustrated in Figure 5.2.

There have been two basic approaches to recognizing the biotic regions in southeast
Asia. One approach has been to draw a single line separating the two regions. Among them

Weber’s line, originally proposed by Pelseneer (1904) and often called the 'line of faunal

122

123

as: aasfismsee sea. seesaw .2 came ,

 

 

 

 

 

 

 

 

 

 

        
     

  

  

 

 

— q _ J
00-. no: I .09 0g
r o. 853‘ £35. 328
.8: 8o 0
3.. em. 1 . lo
% u :3
E 4
Tea .00
.9;
10°
3 z r om. - .2
K r
a
T00 % .I.
6 .61
9m
s. 0 r
0423‘!
a Q s
o \
.. Q o.
1.0. .
i v . . a. Q 00.1
r ”A a e P& \ o A .3. ..... m. ~
0 L / .Nr A mmdwxuamidxuw
.o r o .h a. 3’1!
r n. H c m
. r .. . n
P .....S/oz<4_<x_.n.. .91
t 223 0 N
:f. 0......- .... s. .. ...
\ Cu ... .o . . m
a. J
jam 9. t .
53.2...
. . .. n.
.01? Q enun— cOrM. ‘- 00”. ~ ~ 04.... 2 d m m fl 6... Sou cap.

 

 

 

124

Figure 5.2. Faunal boundaries suggested within the south-east Asia region. Line A, Huxley
(1868); Line B, Wallace (1860); Line C, Pelseneer (1904, Weber's line of faunal balance);
Line D, Lydekker (1896); Line E, Gressitt (1956); Between line A and line D, .Tate's
(1946) 'Wallacean region'; Between line C and line B, Gressitt’s (1956) 'Papuan region'.

(After Holloway and Jardine, 1968).

125

.2 sea

 

 

  

VA

8282.

 

 

126

balance', has been the most commonly accepted. The other approach, in contrast, has been to
recognize a broad transitional zone between the two biotas. Tate (1946), based on the
distribution of bats, proposed the 'Wallacean region' which includes Sulawesi, the Philippine
Islands and the Maluku Islands. Gressitt (1956) used data on the distribution of the insect order
Coleoptera to propose the 'Papuan region' which includes the Maluku Islands, New Guinea and
northern Australia.

Biogeogaphers have identified the geomorphological areas that may represent the
boundary between the Oriental and Australian biotas. The southeast Asian continental shelf
extends offshore to include islands separated by water gaps less than 200 meter deep. This
area, the Sunda Shelf, represents an area where land (or fiesh water) organisms may have been
distnbuted during periods of low sea levels. In a similar manner, the Australian continental
Shelf (the Sahul Shelf) extends off the shore of that continent to include islands that were likely
connected to Australia by land during times of low sea levels. The two continental shelves are
separated by deep water areas that may mark the historical limits of the two biotas. Studies of
organisms with limited vagility confirm that the boundary between the two continental shelves
predicts well the limits of the two biotas (Drandsfield, 1981; Cranbrook, 1981).

Very vagile organisms, however, pose a problem for recognizing biota boundaries.
Bats, birds, and butterflies, for example, can easily fly across water barriers that would limit the
dispersal of other organisms. Biogeogaphers have been interested in studying the patterns of
vagile organisms such as bats to determine how they deviate from patterns obtained from study
of less vagile species. Holloway and Jardine (1968), for example, examined the biogeogaphic

patterns of bats, butterflies, and birds in southeast Asia. They used a phenetic analysis of fauna]

127

Continental
___{::::Ma1aya
Sumatra

Borneo

 

 

 

 

 

 

 

 

Java

 

 

 

Indochina

[ Sulawesi

Philippines
Maluku
New Guinea

 

 

 

 

 

 

2L

 

 

 

Australia

Figure 5.3: The dendrogram calculated from the coeflicients of faunal dissimilarities among
the areas of southeast Asia for butterflies (Alter Holloway and Jardine, 1968).

Continental A
‘ Sumatra

Borneo

 

 

 

 

 

 

 

 

Java

 

 

 

Indochina

r———-—-Sulawesi
L-—-——-Philippines

Australia

I Maluku

New Guinea

 

 

 

 

 

 

 

Figure 5.4: The dendrogram calculated from the coeficients of faunal dissimilarities among
the areas of southeast Asia for birds (After Holloway and Jardine, 1968).

128

Sulawesi & Timor

Maluku
New Guinea

Australia
Philippines
Borneo
Java
Malay
Sumatra

Indochina

Continental Asia

(a)

Sulawesi &. Timor
Maluku

New Guinea
Philippines
Australia

Borneo

Java

Malay

Sumatra
Indochina

Continental Asia

0))

Figure 5.7. The two most parsimonious area cladograms computed from the distributional

data of Rhinolophus in southeast Asia.

129

similarity between the continents and major islands of this region to identify geographically
coherent areas based on shared species. The dendrograms in Figures 5.3, 5.4 and 5.5
summarize their conclusions on what the distribution of butterflies, birds and bats implies about
the historical breakup (vicariance) of southeast Asian habitats. Nelson and Platnick (1980) have
further analyzed Holloway and Jardine’s results, using vicariance biogeogaphic methods, to
produce area cladogams for southeast Asia They conclude that the patterns displayed by birds
and butterflies are concordant but difl‘erent from the pattern displayed by bats. The consensus
area cladogam for all three goups is uninformative (Figure 5.6).

As interesting as these studies of southeast Asian biogeography might be, they are
gernerally not based on phylogenetic analysis of the organisms that comprise the biota In resent
years, several studies have appeared that use cladistic techniques to examine biogeogaphic
patterns in this region (Wiley, 1988b). Because bats have been viewed as a biogeogaphically
anomalous part of the southeast Asia biota (e. g, Nelson and Platnick, 1981), it is of interest to
determine to what extent that view changes if it is based on cladistic methods. Because a
monophyletic subgoup of the genus Rhinolophus is distributed throughout southeast Asia, and
is represented there by numerous species, it is a natural choice to reexamine chiropteran
biogeogaphy in this region from a phylogenetic perspective.

METHODS

This cladistic analysis of the biogeography of southeast Asian rhinolophids is based on
the results of my generic phylogenetic analysis. In this study, geographic areas are the
operational taxonomic units (OTUs) and distributional data of species are the characters. The
presence of a species or a monophyletic goup in two or more areas constitutes a shared

‘derived’ character relating the areas. The data matrix, in which the columns are area OTUs

130

and the rows are distributional characters, is analyzed in the same way as phylogenetic analysis,
and the results are cladogams. The most parsimonious area cladogarn resulting from this
analysis hypothesizes the historical biogeogaphical relationships among the areas based on the
phylogeny of organisms that inhabit them (Nelson and Platnick, 1980; Cracrafl, 1988; Wiley,
1988a).

I used the same area identifications as Holloway and Jardine (1968) used in their
biogeogaphic study of bats, except that three small areas they recognized were joined into an
single adjacent larger area because the distributional data of Rhinolophus in these small areas
was too linrited. The 11 areas I used and their abbreviations are: (1) continental southeastern
Asia including India, southem China, and the adjacent major islands including Taiwan (Cont);
(2) Indochina including Burma, Thailand, Cambodia, Vietnam and Laos (Indc); (3) the Malay
Peninsula (Maly); (4) Sumatra (Sumt); (5) Borneo (Bone); (6) Java (Java); (7) Sulawesi and
Timor (SulT); (8) the Maluku Islands (Mulk); (9) the Philippine Islands (Phil); (10) New
Guinea (NewG); (11) Australia (Aust). Some areas include adjacent smaller islands. Thirty-five
species of Rhinolophus are reported in the southeast Asian region. Table 5.1 is the
distnbufional data of Rhinolophus in these 11 areas, based on data in Corbet and Hill (1992)
and Honacki et al (1982).

Each species present in two or more areas represents a distributional character. Areas
where this species occurs were assigned the ‘derived’ character state for this character and the
areas without this species were assigned the ‘prinritive’ state. A species present in only a single
area was not informative about the relationships between the areas; this information was not

included in the character set.

131

Table 5.1. The distribution of Rhinolophus species in the 11 areas of the southeast Asia region
The abbreviations for the area names are: Cont = continental southeastern Asia including India,
southern China and the adjacent major islands including Taiwan; IndC = Indochina including
Burma, Thailand, Cambodia, Wetnamandlaos;Maly=theMalayPeninsula; Sumt=
Sumatra; Bone = Borneo; Java = Java; SulT = Sulawesi and Timor, Mulk = the Maluku
Islands; Phil=thePhilippineIslands; NewG=NewGuinea; Aust=Australia

 

Cont IndC Maly Sumt Bone Java SulT Mulk Phil NewG Aust

R_ acuminatus + + + + + +

R_ aflinis + + + + + + +

R. anderseni +
R armatus + + + + +
R bomeensis + + + +

R canuti
R celebensis + + +

R coelophyllus + +

R creaghi + +

R eruyotis + + +
R inops +

R lepidus +
R luctus +
R macrotis +
R malayanus

R rnarshalli

R megaphillus

R monoceros +
R nereis + +

R osgoodi +

R paradoxolophus +

R philippinensis + + + +
R pusillus + + + + + +

R rouxi + +

R rufus +
R sedulus + + -

R sharneli +

R simplex +

R_ stheno + + + +

R subbadius + +

R subrufus +
R thomasi + +

R trifoliatus + + + + + +

R virgo +
R yunanensis + +

Area total species 13 19 14 10 10 ll 7 4 8
Endemic grecies 2 2 0 0 0 1 0 0 6 0 0

 

+
+

++++++
++++

+

+

+

+
+
+
+
+

u
N

132

Table 5.2. Data matrix for rhinolophid distributions in the 11 areas of southeast Asia.
Characters 1-24 are based on the distributional data of individual species (listed in Table
5.1). Characters 25-36, listed below, are based on components of relationships from the
majority consensus cladograms of all four cladistic analyses. All components that are
common to at least two analyses and are not in conflict with other cladograms were

selected.

25: R. nereis + R virgo

26: R nereis + R virgo + R simplex

27: R nereis + R virgo + R simplex + R aflinis

28: R nereis + R virgo + R simplex + R aflinis + R celebensis
29: R acuminatus + R pusillus

30: R rouxi + R malayanus

31: R. creaghi + R canuti

32: R creaghi + R canuti + R arcuatus

33: R creaghi + R canuti + R arcuatus + R eruyotis

34: R creaghi + R canuti + R arcuatus + R euryotis + R shameli
35: R creaghi + R canuti + R arcuatus + R euryotis + R shameli + R yunanensis

36: R macrotis}- R philippinensis

The abbreviations for the area names are: Cont = continental southeastern Asia including
India, southern China, and the adjacent major islands including Taiwan; IndC = Indochina
inchrding Burma, Thailand, Cambodia, Vietnam and Laos; Maly = the Malay Peninsula; Surnt
= Sumatra; Bone = Borneo; Java = Java; SulT = Sulawesi and Timor; Mulk = the Maluku
Islands; Phil = the Philippine Islands; NewG = New Guinea; Aust = Australia.

133

Areas

Cont IndC Maly Sumt Bone Java SulT Mulk Phil

Species

Newg Aust

O

1 R acuminatus
2 R aflinis
3 R arcuatus

4 R bomeensis

5 R canuti

O
O

6 R celebensis

O

7 R coelophylllus
8 R creaghi
9 R euryotis
10 R lepidus

ll Rluctus

O

12 R macrotis

O
0

13 R malayanus

14 R megaphillus

15 R nereis

16 R philippinensis O
17 R pusillus

18 R rouxi

0

19 R sedulus
20 R stheno

1

21 R subbadius

22 R thomasi

23 R trifoliatus

1
1

24 R yunanensis

25

26
27
28
29
30
31
32
33
34
35
36

134

Each pair of sister species or monophyletic groups, if they together occupy more areas
than either of them does alone, also defines a distributional character. Cladistic biogeography
assumes that the combined distribution of sister species or groups indicates the distribution of
their immediate common ancestors. The areas they occupy were assigned the derived state, and
other areas were assigned the primitive state of this character. But if the distribution of one of
the sister species or group completely covered that of the other species or group, or the two
sistergroupstogetheroccupiedalltheareas,thenthecombinedareasofthispairdonot
constitute an informative character since this distributional information is redundant.
Successively more inclusive monophyletic groups were treated in the same way to identify
additional characters.

The phylogenetic data of Rhinolophus was based on the majority consensus
cladograms from all four cladistic analyses (section ‘Cladistic Analysis’). I used majority
consensus cladograms because they had the necessary resolution in species relationships for
area analysis. Only those components which were common to at least two analyses and were
not in conflict with other analyses were selected to construct the distributional characters.

I consider the relationships selected in this way to be strongly supported by the data because
they were invariant to modification in the assumptions of cladistic analysis, although some of
the relationships were not present in all the shortest cladograms.

A total of 36 area characters were defined. Among them 24 characters were based on
the distributional data of individual species, including all the species that occurred in two or
more areas. The other 12 characters were based on the distributional data of monophyletic
groups. The characters of the second type were not as numerous as I expected because, as

sister species were joined into more inclusive monophyletic groups, the monophyletic groups

13 5
soon became large enough to occur in the entire region. Any monophyletic group distributed in
all areas would not define a new area character.

In addition to information about the shared occurrence of species or monophyletic
groups, the total number of species and the number endemic species for each area were
counted. The resulting distributional data matrix (Table 5.2) was analyzed using PAUP version
3.1.1 (Swofi‘ord and Begle, 1993) to identify most parsimonious area cladograms.

Because the phylogenetic analysis in the previous section has clearly indicated that the
genus Rhinolophus was originated in Afiican and west European region, the earliest
rhinolophids of southeast Asian are likely to occur in continental Asia. Accordingly, I rooted
the area cladograrn at continental Asia in my area analysis. However, in comparing the various
proposed lines dividing the Oriental and the Australian biotas, I treated the cladogram as
unrooted. This treatment simulated the traditional research in which only regional distributional

similarities or dissimilarities between areas were considered.

RESULTS

Two most parsimonious area cladograms were identified fi'om cladistic analysis (Figure
5.7). These two cladograms are similar, differing only in the positions of Australia and the
Philippines about which area is closer to the Maluku and New Guinea group. The strict
consensus cladograrn computed fi'om these two cladograms contains a trichotomous node,
leaving the relationships among Australia, the Philippines, and the area group of Maluku and
New Guinea unresolved (Figure 5.8).

The branching pattern of the consensus area cladograrn suggests a progressive

subdivision of areas with distance from continental Asia. The partitions of the southeast

136

Sulawesi & Timor

Maluku
New Guinea

Australia
Philippines
Borneo
Java
Malay
Sumatra

Indochina

Continental Asia

(8)

Sulawesi & Timor

Maluku

New Guinea
Philippines
Australia
Borneo

Java

Malay
Sumatra
Indochina

Continental Asia

03)

Figure 5.7. The two most parsimonious area cladograms computed from the distributional
data of Rhinolophus in southeast Asia.

137

Sulawesi &
Timor

Maluku
New Guinea
Philippines
Australia
Borneo

Java

Malay

Sumatra

Indochina

Continental
Asia

Figure 5.8. The consensus cladogram for the two most parsimonious area cladograms

computed from the distributional data of Rhinolophus in southeast Asia.

138

Asian biota occurred first between continental Asia and Indochina, and successively took place
southeastwards. Each of the partitions separates one area fiom all areas located farther from
thecontinentalAsia TlreareasatflreeastardofdreregionMalukuandNewGuinea, where
the most recent vicariant events is inferred, have the most derived rhinolophid biotas.

This pattern parallels the species diversity of pattern the genus (Table 5.2). Species
diversityishighestinlndochina, decliningsteadilyinthesoutheastAsianislandswithdistance
fi‘om Indochina, being lowest in New Guinea and Australia. The lower level of species diversity
in continental Asia can be explained by subtropical temperature conditions there that may be
less conducive to these bats than the tropical monsoon forests of Indochina The reduction of
diversity towards more remote southeast Asian islands may be explained by the islands'
distances fi'om the continent. Island biogeography has demonstrated, in other organisms, that
species diversity is negatively related to the distance of an island fiom the mainland, and
positively related to the size of the island (MacArthur and Wilson, 1967; Lomolino, 1994).

As the partitioning of biota occurred progressively towards one direction,
geographically adjacent areas are generally closely related. Both continental Asia and Indochina
are situated at the root of the cladograrn. Two pairs of adjacent areas, one for the Malay
Peninsula and Sumatra, and the other for Maluku Islands and New Guinea, appear to be sister
areas. Four eastrnost areas, the Philippines, Maluku, New Guinea, and Australia, form a
monophyletic area group. These four areas, together with Sulawesi, constitute a larger
monophyletic area group of five areas.

The consensus area cladograrn from the distributional data of Rhinolophus indicates
that the Oriental realm, as defined by all proposed lines in Figure 5.2, is a paraphyletic area

group. In contrast, the Australian realm, as defined by the Huxley’s line (1864) or by the

139
Weber’s line of faunal balance, is a monophyletic area group. In both most parsimonious area
cladograms Australia and New Guinea are closer to the Philippine Islands than to Sulawesi.
Forthisreason,aseparationoftheAustralianrealmfi'omtheOrientalrealmbyanyother
proposed line (see Figure 5.2) would result in both realms paraphyletic or polyphyletic.

The area cladograrn of the present study does not support the ‘Wallacean’ zone
suggested by Tate (1946) or the 'Papuan zone' suggested by Gressitt (1956). The ‘Wallacean’
zone represents a polyphyletic area group in the rhinolophid area cladograrn; the ‘Papuan
zone’, consisting of Maluku and New Guinea, on the other hand, is a most derived area group.
However, since the present study indicates that the Australia realm is a derived monophyletic
group, a sister area of the Australian realm may be close to what a ‘transitional zone’ would
suggest. Such ‘transitional zone’ would have been evolved fi'om the Oriental realm with the
Australian realm but lacks the synapomorphies that the Australian realm has. A possible
candidate for such a transitional zone is Sulawesi. In the consensus area cladogram, Sulawesi is
a sister area to the monophyletic group containing the Philippines, Australia, Maluku, and New
Guinea. In a strict cladistic point of view, though, a sister group does not suggest a transitional
zone.

DISCUSSION

I did not attempt a cladistic analysis of biogeography over the entire distribution of
Rhinolophus. In my phylogenetic hypothesis of the genus, the relationships among the Afiican
and western Eurasian species are unresolved. Because the rhinolophids of southeast Asia have
been cleariy identified as a monophyletic group, a cladistic biogeographic analysis of

Rhinolophus for that region is not only interesting but achievable.

140

“With 24 out of 36 area characters based on data of individual species distribution, the
area character data inherit a substantial portion of information present in the traditional
similarity matrix used by Holloway and Jardine (1968). The critical data in the present analysis
are the 12 phylogenetically based area characters. Relationships between the species that
inhabit these areas are likely to preserve certain distributional patterns of the past which are not
visible in individual species distributions.

The interpretation of the area cladograms in the present study is greatly afl‘ected by the
assumption about where the first ancestors of southeast Asian rhinolophids occurred. The
panbiogeographic approach assumes that the earliest southeast rhinolophids inhabited the entire
region (Craw, 1988). Under this assumption the hypothetical area relationships will be
presented in an unrooted area cladograrn which is ambiguous to many specific questions
regarding area relationships ( e. g. one of the most parsimonious cladograms, shown in Figure
5.6a, when unrooted, may support both Huxley’s line and the Line of Fauna] Balance, as
illustrated in Figure 5.9). An alternative approach assumes particular areas as the most likely
earliest distribution, suggesting a specific pattern of the regional biota fiagmentation. My
decision to root the cladograms at continental Asia, assuming that continental Asian is the most
likely earliest habitat of southeast Asian rhinolophids, was based on my conclusion that the
earliest members of the genus inhabited Afiica before they emerged in southeast Asia.

Andersen (1 905a) noted that many rhinolophid species in Afiica were closely related to
species in southeast Asia. Andersen believed that in each of these groups of species the Oriental
form displayed more primitive features. He concluded that the Oriental region was the site of

generic origin, and multiple dispersal of rhinolophids from southeast Asia to Afiica had

141

 

 

continental Sulawesi
ASia & Timor
Indochina Maluku
Malay New Guinea
Sumatra Australia
Java Philippines
Borneo Huxley’s line
(3)
Asia
Continental
Indochina
Malay Maluku
Sumatra New Guinea
Java .
Australia
Borneo
Sulawesi .
& Timor
, , , Web’s of faunal balance
Philippines

(b)

Figure 5.9. One of the most parsimonious cladograms of southeast Asia based on
rhinolophid distributional data (Figure 5.7 a). When unrooted, it supports both
Huxley’s line and the Line of Faunal Balance.

142

occurred. Koopman (1970) was more cautious. He concluded that either Afiica or southern
Asia could be the center of origin for this genus. Bogdanowicz and Owen (1992) and
Bogdanowicz (1992), based on their multivariate morphometric study of the genus, questioned
the close relationships among corresponding forms between Afiiea and southeast Asia
proposedbyAndersen. Theymaintainedthatthecenteroforiginofthegermswassoutheast
Asia and used two arguments to support their conclusion. First, Bogdanowicz argued that
rhinolophids of southeast Asia region were phenetically most diverse. This is not a compelling
argument. In determining the geographic origin of a group, morphological diversity should be
defined in phylogenetic terms. Presence of a greater number of living species in southeast Asia,
or presence in these species of more derived character states does not necessarily indicate that
area was the site of the earliest distribution of the genus. Furthermore, relative morphological
diversity for a group of organisms in a particular region can be afl‘ected by many geographical
and ecological factors. In this particular region, repeated isolation and reconnection of
southeast Asia islands associated with the rise and fall of sea levels may have played an
important role in the divergent rhinolophid evolution.

Second, Bogdanowicz stated that no Rhinolophus species has been found in
Madagascar. In contrast, there are five species of Rhinolophus in the Japanese Islands.
Bogdanowicz argues that the chance of rhinolophids dispersing fi’om a continental region to its
ofishore islands may be proportional to the length of time they were present on that continent.
Therefore, the absence of rhinolophids in Madagascar may indicate that they have been in
Afiica for less time than in southeast Asia. He neglects the fact that some ofi‘shore islands are

separated from the continent by a shallow continental shelf. These islands had extensive land

143

connection with the continent duringthe periodswherrthe sealevel fell. Animals could disperse
fiomthecontinenttotheseislandsthroughlandconnections.

According to Vail and Mitchell (1979), the fall ofsea levels to more than 200 meters
belowpresentlevelhasoccurredthreetimessincetheOligocene. Thefirstwasinthelate
Oligocene about 29 million years ago when sea level was about 250 meters lower than present.
ThesecondperiodwasneartheendoftheMiocenewhensealevelfell about200meters
belowpresentlevel. ThethirdperiodwasinthebeginningofthePleistocene whensealevels
fell to about 200 meter below present level. The Korea Strait and Yellow Sea which separate
theJapaneseIslandsformtheAsiancontinent, arebothlessthan200metersdeep. Soarethe
seas isolating Taiwan, Sumatra, Borneo and Java Although Sulawesi, the Philippine Islands
andNewGuineaaredetachedfiomtheAsiancontinental shelfbyseasofmorethan200
meterstheyarelinkedwiththeAsiancontinental shelfbymanysrnallislands. Australiaand
NewGuineaareconnectedby Sahul shelfwhichislessthanZOOmetersindepth. Incontrast,
the Mozambique Channel which isolates Madagascar fi'om the Afiican continent is more than
3,000 meters deep in its stretch of greater than 235 km (Brenan, 1972).

The fossil occurrence of Rhinolophus species supports an Afiican origin of the genus.
The earliest Rhinolophus is known at Robiacian of the Upper Eocene in France. Also
appearing first in the same formation were representatives of the bat families Hipposideridae,
Emballonuridae and Vespertilionidae (Savage and Russell, 1983). These fossils represent the
oldest extant bat families we know of. On the other hand, the earliest rhinolophids found in
southeast Asia are Pliocene in age (Savage and Russell, 1983). These fossil discoveries, as well
as fossil bats belonging to the family Palaeochropterygidae found in the Lower and Middle

Eocene of Europe, suggest an extensive family level divergence of bats during the middle of

144

the Eocene. \Vrthout further evidence, it is more parsimonious to conclude that modern family
divergence occurred first in west Eurasia or Afiica. To suppose that rhinolophid ancestors of
moved to southeast Asia, where the family then arose, and new family members migrated back
to west Europe is an unnecessarily complex hypothesis. The fact that the earliest rhinolophids
and their close relative, Hippasideros spp. , were discovered in the same formation supports the
hypothesis that rhinolophids evolved from common ancestors close to that region.

It has been proposed that more advanced vertebrate orders and families originated in
areas characterized by large geographic size, heterogeneous topography, warm and relatively
steady temperatures, and maximum species diversity (Briggs, 1984; Dariington, 1958).
Although most early fossil species of Rhinolophus were unearthed from Europe, this region
does not meet the conditions suggested above. However, Europe is close to the Afiican
continent. All the living species of Rhinolophus in Europe are present in Africa It is possible
that the rhinolophid faunas of these two continents have had easy communication since early in
rhinolophid evolution. While both Africa and southeast Asia are in the tropical region, Africa is
much larger than southeast Asia in land area. The Indian subcontinent had been isolated from
both Afiica and Eurasian continents since Early Cretaceous times and did not join the Eurasian
plate until Early Eocene times (Briggs, 1989). The discovery of several early bat families in
west Europe may indicate that the origin of these bat families is somewhere in the general
vicinity of the fossil sites. The best candidate for the center of origin is tropical Afiica. It is
likely that a group of early rhinolophids migrated to eastern Asia and then to southeast Asia
after Late Eocene times. These immigrants became the ancestors of all present rhinolophid

species of that region.

APPENDIX 1

Appendix 1.

Eigenvalues of the Correlation Matrix for pooled skin and skull data

assesses

BB
WCO
PMP
VLAB
PB
DH
DL
LINF
LOR
HOR
P4M3
LBR
HNS
HCR
HOCC
FA

LT
2Met
3Met
3M1P
3M2P
4Met
4M1P
4M2P

5M1P
5M2P
EAR

PC 1
PC2
PC3
PC4
PCS

Eigenvalue
3 1.6678

2.8730
1.8394
0.9169
0.6934

Difference
28.7948

1.0336
0.9226
0.2235

Proportion

0.772386
0.070073
0.044864
0.022363
0.016912

Cumulative

0.772386
0.842459
0.887323
0.909686
0.926598

Eigenvectors for pooled skin and skull data

PCl

0.172418
0.109400
0.162211
0.168653
0.155555
0.169703
0.167963
0.162510
0.159313
0.173056
0.147886
0.149697
0.173658
0.154513
0.146259
0.133495
0.155997
0.166992
0.153470
0.157654
0.158598
0.074910
0.157694
0.157232
0.157377
0.171319
0.157678
0.164202
0.173948
0.169151
0.102772
0.130181
0.121998
0.169543
0.159412
0.170793
0.174779
0.154596
0.172724
0.150633
0.156812

PC2
-.083247
-.376085
-.079260
0.029704
0.064086
-.028797
-.019548
-. 163413
-. 152750
-.084008
-.020179
-.223336
-.064537
-. 179875
-.232026
-.200016
0.257842
0. 165475
-.042989
-.024207
0.091240
0.487515
0. 120396
-.027819
-.03 1913
0. 136549
-.099449
0.034843
0.086503
0.048470
0.097246
-. 1 12146
-. 108303
0. 105244
0.201372
0. 147164
0.026485
0. 182958
0.068675
-. 151323
0. 185484

145

PC3
-.0401 17
0.003675
-.015707
-.058913
0.021484
-.079167
-.092259
-.0153 54
0. 189751
-.024395
-.277334
0.216808
-.023871
-.01 1526
-. 165428
-.303237
0.01 1993
0.079035
-.247919
0. 173877
-. 129096
-.090402
-.097730
-. 194245
0.185730
-.019199
0.166521
0. 181913
0.019445
0. 102121
0.465489
-.244813
0.296024
0.028385
-.022670
0.005296
0.027528
-. 170745
-.00833 1
0. 1241 12
0. 104708

PC4
-.023871
-.255750
-.284690
-.219661
-. 177906
0.012044
0.049479
-. 194893
-.03 3099
-.053937
-.075955
-.079916
-.058046
-.253997
0.070126
0. 163660
-. 126961
-.076298
0.339418
-.078762
0. 107009
-.215033
0. 170772
0.31 1 190
0.016893
0.005108
-.009323
-.034848
0.025176
0.005459
0.282857
0.036792
0.338833
-.061263
0. 157464
-.01 1604
0.048558
0.047629
0.065893
0.214585
-.086978

PCS

-. 124842
0. 128854
-.083 366
-.020514
-.351054
-. 152928
-. 185398
0. 109304
-.058756
-.092963
0.067196
-.052380
-.076075
0.084077
-.081830
-.084484
0.075170
0.071835
-.034871
-.034122
0.039947
0. 152157
0.013017
0.032402
-.223995
0.028054
0. 194144
0.240700
0.047405
0.212480
0.053452
0.599065
0.229818
0.026027
-. 161098
0.038829
0.040217
-.036945
-. 108945
-. 156713
-.078796

PAL

LPT
TEF
LSHF
WZA
WAB
LAB
BL
BB
WCO
PMP
VLC
VLAB
PB
DH
DL
LINF

HOR
P4M3
LBR
HNS
HCR
HOCC
FA

2Met
3Met
3M1P
3M2P
4Met
4M1P
4M2P
5Met
5M1P
5M2P
EAR

146

Eigenvalues of the Covariance Matrix for pooled skin and skull data

Eigenvalue Difference Proportion Cumulative

PCl 301.554 274.436 0.854695 0.854695

PC2 27.119 18.946 0.076862 0.931557

PC3 8.172 2.130 0.023163 0.954720

PC4 6.042 1.900 0.017126 0.971846

PCS 4.143 0.011742 0.983588

Eigenvectors for pooled skin and skull data

PCl PC2 PC3 PC4 PCS
0.427171 0.138884 -.006055 -.103191 -. 179621
0.279135 -.820383 -.022807 -.420030 0.031011
0. 129803 -.002387 0.006303 0.000956 0.249672
0.222023 0.151601 0.022316 -.055488 0.524562
0.030818 0.031388 -.009520 -.014099 0.030517
0.293281 0.208402 0.090323 -.253996 -.260799
0.276359 0.225703 0.131221 -.249152 -.328737
0.166513 -.075781 -.052433 0.077341 0.275196
0.276712 -.039657 -.585712 0.388955 -. 163227
0.317651 0.089441 -.015351 -.033877 -.022903
0.093535 0.056973 0.187981 -.225879 0.054846
0.164808 -. 103041 -.398292 0.210823 -.110767
0.341564 0.142149 -.057450 -.044121 0.027353
0.152527 -. 107637 -.001925 0.074744 0.366343
0.284149 -.234686 0.626665 0.636720 -. 127412
0.000020 0.006808 0.044724 -.004527 0.011480
0.029617 -.007122 0.082531 0.036843 -.03 3621
0.015491 0.029456 -.003447 -.013971 0.056562
0.065836 0.090218 -.051316 -.021181 0.156842
0.029250 0.016785 0.070297 0.024068 -.042669
0.019358 0.007809 -.032990 0.017429 0.024704
0.026326 0.027774 0.027835 0.023040 0.024010
0.009339 0.075096 0.018613 -.042883 0.155333
0.014869 0.021814 0.006932 -.000818 0.006819
0.024373 0.017809 0.038639 0.020813 -.014912
0.027615 0.009511 -.030425 0.029557 -.012436
0.138000 0.175018 0.033671 0.044094 0.275876
0.021468 -.007831 -.023292 0.000334 0.018353
0.032023 0.019015 -.053734 0.033227 0.068681
0.042457 0.042998 -.002678 0.024002 0.062385
0.007249 0.011792 -.039148 0.018211 0.007149
0.027470 -.006274 0.044977 -.007597 0.044210
0.015478 0.000017 -.035292 0.005859 -.038744
0.034888 0.036070 -.012491 0.005158 0.069851
0.055394 0.100559 0.046933 0.022899 0.071356
0.056720 0.074534 -.003652 -.008974 0.1 10922
0.055252 0.035594 -.007278 0.044363 0.061603
0.029090 0.042546 0.055189 0.000702 0.055357
0.047814 0.042473 0.011225 0.019577 0.027394
0.024961 -.005790 -.008093 0.041172 -.034141
0.021570 0.032357 -.021238 0.002602 0.047342

147

Eigenvalues of the Correlation Matrix for skull data

Eigenvalue Difference Proportion Cumulative

PC 1 19.3322 16.4191 0.716007 0.716007

PC2 2.9131 1.1508 0.107893 0.823899

PC3 1.7623 1.1855 0.065271 0.889170

PC4 0.5768 0.0532 0.021362 0.910532

PC5 0.5235 0.019390 0.929923

Eigenvectors for skull data
PCl PC2 PC3 PC4 PC5

P2 0.009179 0.415359 0.391855 0.131644 0.468400
PAL 0.148541 0.379935 -. 165256 0.076674 0.186574
M3 0.210370 -. 135304 0.188235 -. 104874 -.040451
WM3 0.219435 -. 120614 0.048376 -.068119 -.035555
LPT 0.180877 -.084540 0.365858 0.151442 0.075115
TEF 0.217341 -. 126597 -.080247 -.020207 0.016237
LSHF 0.175958 0.276401 -. 164580 -.239902 0.103061
WZA 0.222428 -.079249 -.046037 -.026750 -.0551 10
WAB 0.175435 0.262821 -.1 17214 0.028417 -.089472
LAB 0.195812 0.259159 -.038921 0.073811 -.049812
BL 0.198548 -.092717 -. 191659 0.079955 0.072128
BB 0.145993 -.263847 -.330174 -.255487 0.327620
WCO 0.210154 0.075651 0.038124 -.210022 -.035638
PMP 0.118714 -.215496 0.543588 -. 196547 -.1 19163
VLC 0.205497 0.092993 0.077554 0.100176 -.063990
VLAB 0.191459 0.258466 -.043733 0.249588 -.097418
PB 0.198089 -. 105206 -.224834 -. 193158 0.264186
DH 0.207777 -. 129494 0.058281 0.005960 0.156400
DL 0.224207 -.013410 0.056064 0.015884 0.043478
LINF 0.136774 -.270597 -. 172971 0.742782 -. 106497
LOR 0.158542 0.233621 -.071250 -.217513 -.659519
HOR 0.215988 -.094500 0.031 123 0.003797 -.099984
P4M3 0.223480 -.057426 0.058355 0.006167 -.040874
LBR 0.222934 0.008333 -.073397 0.046122 -.033217
HN S 0.206121 0.091284 0.202106 0.074482 0.039103
HCR 0.222773 0.018588 -.024679 -.005238 0.115746
HOCC 0.207672 -. 159877 0.051933 -.046481 0.033146

PC 1
PC2
PC3
PC4
PC5

P2
PAL
M3
WM3
LPT
TEF
LSHF
WZA
WAB
LAB
BL
BB
WCO
PMP
VLC
VLAB
PB
DH
DL
LIN F
LOR
HOR
P4M3
LBR
HN S
HCR
HOCC

148

Eigenvalues of the Covariance Matrix for skull data

Eigenvalue Difference Proportion

19.0069 18.1599 0.900603

0.8470 0.4302 0.040135

0.4169 0.2133 0.019752

0.2036 0.0575 0.009646

0.1461 0.006922

Eigenvectors for skull data

PCl PC2 PC3 PC4
0.000066 0.098295 0.194460 -.044111
0.098809 0.543833 0.025836 -. 194663
0.076232 -.098997 0.072820 0.050225
0.303104 -.251654 -.075799 0.177502
0.049636 -.070836 0.143034 -.027437
0.203941 -. 121266 -.268494 -.029665
0.050712 0.187641 0.004688 -.014028
0.409631 -. 120652 -.396100 0.269466
0.072276 0.237217 0.005220 -.009809
0.110873 0.313789 0.093354 0.008567
0.079485 -.002290 -.177160 -.047223
0.028784 -.052631 -. 157202 -.042350
0.1 13467 0.086922 0.092509 0.083209
0.078624 -.385170 0.553803 0.190585
0.065010 0.051937 0.033462 0.071895
0.094873 0.269371 0.046182 0.051082
0.106582 -.008067 -.269528 -. 191774
0.106173 -.096452 0.000927 -. 149642
0.615513 0.021842 0.312321 -.276641
0.038669 -.081645 -. 140076 -.095937
0.098389 0.306155 0.039306 0.788146
0.152446 -.080203 -.015337 0.041119
0.255446 -.082392 0.069473 0.027845
0.232025 0.107344 -. 160786 -.03 8530
0.130262 0.082608 0.306489 -.096184
0.198609 0.094811 -.029875 -. 121372
0.098873 -.1 11734 -.023634 -.049602

Cumulative

0.900603
0.940738
0.960490
0.970136
0.977058

PCS
-.025751
-.257108
0.015220
-.021 176
0.038633
-.006642
0.125452
0.070962
0.308227
0.356226
-.08 1392
-.010142
0.302898
0.209492
0.102953
0.064888
0.380480
0.187939
-.394067
-. 136892
-. 131604
0.004925
-.056648
-.043887
0.160635
0.335269
0.127937

PC 1
PC2
PC3
PC4
PC5

FA
TL
FT
LT
2Met
3Met
3M1P
3M2P
4Met
4M1P
4M2P
5Met
5M1P

5M2P
EAR

Eigenvalues of the Correlation Matrix for skin data

Eigenvalue

12.8459
0.8235
0.4688
0.2698
0.2214

PC 1
0.275035
0.206276
0.261894

0.261612
0.266481
0.262670
0.268462
0.257280
0.276293
0.237179
0.250366
0.275339
0.264162
0.257894
0.243163

149

Difference Proportion
12.0224 0.856395
0.3547 0.054899
0.1990 0.031253
0.0484 0.017989
0.014761
Eigenvectors for skin data
PC2 PC3 PC4
-. 121487 -.071431 0.099358
0.604998 0.420025 0.356650
0.009019 0.041509 -.423584
-198163 -048781 -453206
-.268150 -.041156 0.189945
-.306383 -.027219 0.226376
0.151158 0.052295 -. 197703
0.177362 -.481256 0.097662
-.093600 -.070184 0.024855
-.370892 0.500884 0.179606
0.314792 -.449177 0.146815
-. 134645 -.O93647 0.013465
-.047769 -.026577 0.332882
0.232801 0.134638 -.405753
0.199409 0.304013 -.110013

Cumulative

0.856395
0.911294
0.942546
0.960535
0.975297

PC5
-.079332
0.204390
0.1 10276
. 159603
-.1 16367
-. 179791
0.183863
-.045640
-.072220
0.244728
0.046789
-.008840
0.240891
0.153806
-.826730

FA

LT
2Met
3Met
3M1P
3M2P
4Met
4M1P
4M2P
5Met
5M1P
5M2P
EAR

PC1
PC2
PC3
PC4
PC5

Eigenvalues of the Covariance Matrix for skin data

Eigenvalue
291.159
25.567
7.845
5.985
3.581

PC1
0.436917
0.286216
0.131921
0.225196
0.297813
0.280391
0.170217
0.282318
0.324035
0.095237
0.168455
0.348338
0.101482
0.155578
0.289912

150

Difference Proportion Cumulative

265.591 0.861871 0.861871

17.722 0.075683 0.937554

1 .860 0.023224 0.960778

2.404 0.017716 0.978494

0.010600 0.989094

Eigenvectors for skin data
PC2 PC3 PC4 PC5

-. 177662 0.007483 -.086968 -.112129
0.838009 0.032346 -.425519 -.058424
-.000564 -.011854 0.006849 0.289219
-. 164422 -.019462 -.045728 0.619027
-.231497 -.114306 -.252160 -.236267
-.247885 -.157035 -.248075 -.311316
0.068958 0.059092 0.083375 0.323803
0.018436 0.593296 0.386236 -. 190846
-.112374 0.010306 -.025399 0.017004
-.062720 -. 194659 -.223 802 0.088361
0.090546 0.403284 0.208864 -.1 14035
-. 168407 0.053094 -.035630 0.080587
-.011961 0.042441 -.059414 -.023 549
0.105969 0.004239 0.081638 0.423889
0.232636 -.632500 0.650353 -. 144137

APPENDIX 2

Appendix 2.

LIST OF SPECIMENS USED

1. R. acuminatus - skull (44): Java (3), USNM 456262, 156351, 155791; Sumatra (9),
USNM 141012, 141014, 141015,141340, 141341, 141344, 141346, 241241, 241242;
Borneo (2), USNM 292390, 449972; Philippine (8), USNM 477613, 477615-477620,
477623; Siam (5), USNM 84493, 254766, 254768, 254770, 355561; Thailand (17),
AMNH 88016-88032; skin (17): Tailand (17), AMNH 88016-88032;

2. R. adami - skull (1):Cameroon (1), CMNH 13178; skin (1):Cameroon (1), CMNH 13178;

3. R qfiinis - skull (75):Burma (3), USNM 279204, 279205, 18456; China (42), USNM
238849-238851, CMNH 88033-88036, 88551-88553, 88555, 88557, 92146, 92140-
92142, 92143, FMNH 33806-33813, 33818, 33819, 75996-77999, 76001-76003, 76005-
76012, 76015, 33924, 33922, 33923; Malysia (4), USNM 481057, FMNH 64089, 87345,
87351; Siam (3), 83538, 83571, 83540; Borneo (3), USNM 152045, 154402, 154406;
Vietnam (9), USNM 320630, FMNH 32143-32146, 32149-32152; Assam (4), FMNH
75956, 75962, 82639, 82641; Borneo (7), FMNH 44154, 47076-47081; skin (27): China
(15),CMNH 88033-88036, , 9214a92142, 92146, FMNH 33813-33816, 33818, 33819;
Borrrio (6), 47076-47081; Siam (6), 76007-76009, 76011, 76012, 76015;

4. R alcyon - skull (27):Camerom (8), USNM 511918, 511919, AMNH 236298, 206955,
86880, CMNH 58295-58297, 41000; Ghana (1), USNM 414973; Sierra Leone (18),
USNM 546967-546977, 546979, 546980, 546982-546984, 546986, 546987; skin (7):
CMNH Carnerom (7) 58295-58297, 41000; AMNH 86880, 206955, 236298;

5. R alticolus - skull (17):Camerom (5), CMNH 5701, 42308, 58311, 58312, 58320; S.
Afiica (12), CMNH 58298-58309; skin (5):Camerom (5), CMNH 5701, 42308, 58311,
58312,58320;

6. arcuatus - skull (24):Philippine (24), USNM 101093, 101964, 175798, 175803, 175817,
175820, 175824, 303960, 303952, 303965, 304354, 304355, 304357, 304359, 459451,
459452, 573282, 573283, AMNH 241805, 241807, 187134, 187136-187138; FMNH
140671-140678; skin (10):Phi1ippine (10), FMNH 61229-61233, MCZ 35106-35108,
35110035111;

7. blassi - skull (30):Burma (1), USNM 327990; Ethiopia (1), AMNH 48077; Palestine (2),
AMNH 54413, 54414; Yugoslavia (2), AMNH 239591, 239591; Turkmenia (1), AMNH
245355; Fordan (6), CMNH 78840-78845; Afghanistan (11), FMNH 102271-102275,
102277-102281, 102369; Iran (16), FMNH 96608-96610, 96612, 11169, 11174,

151

152

96580-96584, 96563, 96566, 96567, 96570, 96572; skin (19): Afghan (19), CMNH
78841-78845, FMNH 102271-102275, 102277-102281, 102369;

8. R. blylh - Skull (26):Burma (10), USNM 279206-279209, 279212-279214, 279216-

9.

279218; China (15), USNM 238855, 238857-238860, 238862, 260045, 279350, 238156,
238158-238160, 294812, AMNH 58311, 58461; Siam (1), USNM 296498; skin (10):
China (10), MCZ 20291-20294, 7515-7517, 58293, 58464, 58474;

R bomeensis - skull (21):Natuna (4), USNM 140751-140752, 107755, FMNH 640950;
Indonasia (4), USNM 521820, 145611, 145612, 145699; Malaysia (1), USNM 449973;
Bomeo (1), AMNH 106844; Timor (4), AMNH 153511, 237759, 237760, 237777; Java
(Bali) (2), AMNH 107887, 107888; Noesa (2), AMNH 107958, 107959; Sulawesi (3),
102231, 102360, 102246; skin (10): Bomeo(8), AMNH 102230-102233, 102360,
102366, AMNH 103918, MCZ 36081; Celebes (1) AMNH 106844; Malaya (1), USNM
449973;

10. R. ccpensis - skull (9):Afi'ica (9), USNM 342583-342588, CMNH 46787-46789; Skin

()zAfi'ica (10), CMNH 46787-46789, MCZ 17899, 37226, 37227, 37049-3 7051;

11. R celebensis - skull (3):Sulawesi (3), USNM 217464, 219379, 219383; skin (2): USNM

217463, 217464;

12. R chaseni - skull (8):VletNam (8), USNM 357010-357013, 357094, 357096, 357257,

357351; Skin (6):VietNam (6), USNM 357010-357013, 357257, 357258;

13. R. clilnsus - skull (66):Afiica (36), USNM 381538-381541, 381544-381550, 381552-

381555, CMNH 40669-40672, 46793, 46796, 93167, 93169, 52642-52650, FMNH
38137-38140; Egypt (12), USNM 282406, 282478, 312514, CMNH 42356, 78854,
78855, FMNH 78780, 78783, 78821, 78822, 123219, 123220; Liberia (1), AMNH
265710; Libya (1), CMNH 78856; Sudan (7), FMNH 77648, 77649, 78470, 78474,
108146-108148; Ethiopia (3), FMNH 28775, 28777, 79289; Kenya (6), FMNH 67897-
67902; skin (11): Africa (11) CMNH 40669-40672, 46793, 46796, 93167, 93169, 78854,
78855, 62356;

14. R coelophyllus - skull (16):Siam (6), USNM 267260, 296824, 296825, 296827, 296828,

356305; Japan (2), USNM 278722, 278723; Borneo (2), USNM 198947, 198948;
Malaysia (3), AMNH 216856, 216857, 216861; Thailand (3), CMNH 88037-88039; skin
(6):Thailand (3), CMNH 88037-88039; ; Malaysia (3), AMNH 216856, 216857, 216861;

15. R cornutus - skull (2):Japan (2), AMNH 244343, FMNH 73678; skin (7): Japan (7)

USNM 728722, 728722, 23691, 23692, 23694-23696;

16. R creaghi - skull (7):Timor (2), AMNH 237784, 237787; Borneo (5), FMNH 47071-

47075; 64094; Skin(11):Timor(6), AMNH 237784-237787, 237801, 237802; Borneo (5),
FMNH 47071-47075;

153

17. R. darlingi - skull (13):Bechwana (3), USNM 382645, 365203, 470252; S. Afiica (9),
AMNH 257157-257161, 257163, CMNH 93168, 93170, 40675; Tanganyika (1), AMNH
188272; skin 0: Africa (8) MCZ 34093-34097, CMNH 93168, 93170, 40675;

18. R denti - skull (23):Bechwana (21), USNM 322855-322862, 322864, 322869-322871,
322874, 322876-322882, 322890; 3. Afiica(1),CMNH 36015; Carneroom (1), CMNH
58313; skin (10): Bechwana (6) USNM 322875-322878, 322883, 367683; Ivory Coast (2)
FMNH 105206, 105325; Afiica (2) CMNH 93171, 58313;

19. R. deckeni - skull (3):Kenya (2), USNM 247386, FMNH 48827; Tanganyida (1), AMNH
208341;

20. R. eloquens - skull (26): Zaire (1), AMNH 82392; Kenya (12), CMNH 10704, 97940-
97943, 79746-97949, 102165, FMNH 67916, 67917; Sudan (4), FMNH 56291, 66665-
66667; S. Africa (5), 97932-97935, 102164; Zaire (4), FMNH 25600, 67497, 68063,
68068; skin (12): Kenya (12), CMNH 93171, 10704, 97940-97943, 7974697949,
102165;

21. R. euryale - skull (26):France (1), USNM 38351; Greece (1), USNM 153596; Italy (4),
USNM 105790, 105792, 86586, 86588; Spain (1), USNM 260652; Czechoslovakia (1),
USNM 540777; Morocco (4), USNM 476274, 476267-47629; Algeria (4), CMNH
78857-78859, 78869; Iran (9), FMNH 96540, 96544, 96545, 11170-11173, 11175,
11176; Lebanon (1), FMNH 99556; skin (20): Algeria (5), CMNH 78857-78859, 78869,
89477; Iran (15), FMNH 96540-96542, 96544, 96547-96551, 11170-11173, 11175,
11176;

22. R eruyotis - skull (24):Indonasia:Moloccas (2), USNM 543263, 543264; Sulawesi (6),

. USNM 501515-501517, 501519—501521; Bismarcks (1), AMNH 195249; New Guinea
(9), AMNH 109956, 109957, 101940, 101941, 195248, 157400, 158462, 158470,
190270; Mulaccos (2), ANINH 54432, FMNH 34051; Sulawesi (4), AMNH 196475,
102236, 102239, 102241; skin (15): Sulawesi (8), AMNH 196475, 196476, 102236-
102239, 102241, 102241; Malayasia (7), USNM 198371-198373, 198375-198378;

23. R. ferrumequinum - skull (56):Frace (3), USNM 154221, 154525, 154526; Italy (2),
USNM 38198, 38343; Spain (2), USNM 172123, 172129; Japan (2), USNM 291737,
291739; Morocco (2), USNM 476307, 470606; HEAT? (4), USNM 182665, 182667,
162499, 182668; Kenya (15), USNM 350880-350882, 350884-350887, 436583-436586,
436596, 436598, 436599, 436602; Jordan (15), CMNH 78876, 78877, 62112-62115,
62120-62124, 78872-78875; China (1), CMNH 92167; Afghanistan (10), FMNH 102370-
102379; skin (15): Jordan (15), CMNH 62112-62115, 6211962124, 78872, 78874,
78876,78877,92167;

154

24. R. fiam’gatus - skull (30): Kenya (17), USNM 350889-350894, 436519-436522, 436531,
436533436536, 436538, 436540436542; Namibia (3), CMNH 93172, 93173, 61476;
Algeria (10), CMNH 97930, 97931, 97950, 97951, 67971, 93174, 94985-94987, 98531;
skin (13): Namibia (2), CMNH 93172, 93173; Algeria (11), CMNH 61476, 97930,
97931, 97950, 97951, 67971, 93174, 94985-94987, 98531;

25. R. guineensis - skull (3):Lrberia (3), AMNH 257046, 265719, 265738;

26. R. hildebrandti - skull (17):ABA (1), AMNH 49102; Mozambique (7), AMNH 245158,
216206, 216208-216212; Tanganyika (1), AMNH 161308; Zimbabwe (2), AMNH
213048, 213049; Kenya (1), AMNH 161917; S. Africa (1), CMNH 93175; Sudan (3),
FMNH 78196, 79553, 79554; s. Afiica (1), FMNH 95148; skin (10): S. Afiica (1),
CMNH 93175; Sudan (4), FMNH 78196, 79553, 79554, 95148, 95149; Afiica (5) MCZ
22790,22791,38923,38982,43764;

27. R. hipposideros - skull (34):Iran (6), USNM 350138, FMNH 96667, 111183, 111184,
111188, 111190; France (1), 172121; Germany (3), USNM 152530, 67540, 67541; Spain
(2), USNM 172122, 172126; Italy (3), USNM 38347, 152527, 38192; Switzerland (3),
USNM 121183, 124393, FMNH 44123; Morocco (1), USNM 476320, 476321; Austria
(1), AMNH 150439; Germany (1), AMNH 217131; Poland (1), AMNH 212186; Georgia
(1), AMNH 245359; Algeria (1), CMNH 62111; Poland (1), CMNH 45292; Jordan (1),
CMNH 78905; Afghanistan (6), FMNH 102410402413, 102415, 102424; Egypt (1),
FMNH 74476; Lebanon (1), FMNH 9956; skin (18): Africa (4), CMNH 45292, 62111,
78905, 78906; Afghanistan (7) FMNH 102409, 102411-102415, 102424; Iran (7), FMNH
96656, 96657, 111183-111185, 111189-111190;

28. R hirsutus Philippine (1), USNM 125487;

29. R imaizumii - skull (1):Japan (1), AMNH 241142; skin ()1 Japan (7), AMNH 241137-
241143;

30. R inops - SkulliPhilippine (21), USNM 125314, 458607-458612, 459494, 573289,
458580-458587, 458590-458592, 458594; Skin (12): USNM 458604, 458606, 458609-
458612, 574818-574823;

31. R keyensis - skull (2): Moluccas (2), AMNH 222739, 222741; skin (3): Moluccas (3),
AMNH 222739-222741;

32. R. landeri - skull (42): Gambia (7), USNM 379388, 412004412006, 412009, 412011,
412016; Nigeria (4), USNM 379508, 379513, 402708, 402710; Mozambique (4), USNM
365181, 365184, 3665187, 365191; West Africa (1), 185330; Ghana (1), AMNH 237419;
Kenya (1), AMNH 114476; Congo (1), AMNH 49132; Botswana (2), AMNH 89174,
89175; Cameroom (13), CMNH16064, 16067, 16069-16073, 16077, 42309, 42310,
58314-58316, 59318; Kenya (1), CMNH 97952; Gabon (2), CMNH 90800, 90801; India
(2), CMNH 92230-92234; Sudan (3), FMNH 67323, 79546, 79547; Ivery Coast (1),

155

FMNH 105236; skin (122): Carneroom (19), CMNH 7443, 16064-16073, 58314-59318,
4230942311; Gabon (2), CMNH 90800, 90801; Kenya (1), CMNH 97952;

33. R lepidus - skull (27):China (1), AMNH 84384; India (12), AMNH 236216, 208837,

174287, 247284, 216889, 216896, 216897, FMNH 82654, 82652, 82653, 82647, 82649,
82650; SN (6), CMNH 92235-92240; Afghanistan (8), FMNH 102283-12286, 102288,
102416-102418; skin (17): Afghanistan (8), FMNH 102283-12286, 102288, 102416-
102418, India (4), FMNH 82651-82654; SN (5), CMNH 92230-92234;

34. R Iucrus - skull (33):Bomeo (4), USNM 292387, 292388, 300837, 300838; Taiwan (8),

USNM 358199-358202, 332843-332845, 294141, 358198; Siam (2), USNM 296829,
296830; Thailand (3), USNM 528271, AMNH 167933, CMNH 88040; Indochina (1),
AMNH 87311; Java (1), AMNH 107853; Borneo (3), AMNH 106834-106836; Malaysia
(7), CMNH 88041-88046, 98681; India (3), FMNH 82646, 48497, 85046; Indochina (1),
FMNH 46539; skin (9): India (8), FMNH 82646, 48497, 85046, 99466, 46539, 73005,
76016, 98681; Thailand (1), CMNH 88040;

35. R. maclaudi - skull (4):Liberia (1), AMNH 265708; Uganda (1), AMNH 245634; Uganda

(2), FMNH wtsS95.jek1966; skin (1): Uganda (1), AMNH 245634;

36. R. macrotis - skull (13):India (2), USNM 399303, FMNH 47403; Malaysia (4), AMNH

243057, 216864, 84382, 57161; Indochina (4), FMNH 32142, 32127, 32215, 38992A;
China (1), FMNH 33892; Vietnam (2), FMNH 38992, 32142; skin (10): China (3),
AMNH 84888, 56894, 56897; Malaya (2), 216864, 216870; Indochina (5), FMNH 32142,
32127,32215,33892,47403;

37. R malajwrus - skull OzThailand (), USNM 528272-528277; Vietnam 0, USNM 260043;

Indochina O, AMNH 87300-87302, 216875, FMNH 32117, 32119, 32121, 32126, 32139,
32217, 32218, 32225, 33768; skin ()2 Thailand (16), FMNH 32117, 32119, 32121, 32126,
32138, 32139, 32217, 32218, 32225, 33768; CMNH 88041-88046;

38. R megtphyllus - Skull (15):Australia (10), AMNH 194238, 160288, 154594, 183446,

39.

154626, 183514, 162663, 154627-154629, FMNH 64398; Papua (5), AMNH 157391,
158652-158654, 158674; skin (10): Australia (10), FWH 60851, 60852, 60853, 64398,
MCZ 29087, 27928, 27930-27933;

R. mehelyi - skull ():Egypt 0, USNM 312517, 312518, FMNH 79082, 79088, 79089;
Sardinia 0, USNM 86536; Morocco 0, USNM 476213, 476215, 476223, 476232-
476234; Dagestan (USSR), AMNH 245361; Azerbaijan (USSR), AMNH 245360; Tmisia
(), AMNH 217132; Algeria 0, CMNH 78884, 78885, 78893-78897; Libya 0, CMNH
78898-78904; Sudan 0, CMNH 89625-89632; Iran 0, FMNH 111129411132, 96614,
9661696619, 96621;

156

40. R milutilus - skull (2): Indonesia (1), USNM 101770; Siam (1), USNM 254763; skin (7):
Malaya (2), AMNH 234060; Thailand (4), USNM 528278, 528279, 528280, 528281;
Siam (2);USNM 260606, 260607;

41.R monoceros - Skull (22):Taiwan (22), USNM 294142, 294143, 330053, 332849,
332850, 358144, 358145, 358151, 358152, 38154, 38155, 358174-358179, 358181,
358193-358196; Skin (11): Taiwan (11), 215770, 215777, 215784, 215786; USNM
330052, 332850, 358149-358153;

42. R. nereis - skull (1):Phi1ippine (1), USNM 101714; skin (1):Phi1ippine (1), USNM 101714;

43- R W - skull (4):China (4). AMNH 45046, 44547, FMNH 33295, 33689; skin 0:
China 0, 4505245055, 45074, 45078, 45838, 45089, FMNH 33295, 33297;

44. R pearsom' - skull (13):China (4), AMNH 84862, 58282, FMNH 33839, 33840; Thailand
(3), AMNH 250003, 250004, 167935; Malaysia (1), AMNH 234063; Burma (1), AMNH
112910; Assam (4), FMNH 75963, 75966, 75968, 75969; India (3), FMNH 82643-82645;

skin (8): China (3), AMNH 84862, 58281, MCZ 249339; Thailand (3), AMNH 250003,
250004, 167935; Burma (2), AMNH 112908, 112910;

45. R philoppinensis - skull (4): Philippine (1), USNM 459469; Australia (2), AMNH
157069, 157071; Sulawesi (1), AMNH 102348; Skin (11): Sulawesi (9), 35007-35009,
35098, 35099, AMNH 102348-102351; Negris Id. (2), USNM 459496, 459497;

46. R pusillus - Skull ()zThailand O, USNM 528278; China 0, AMNH 56910, 56922, 57156,
57160, 58294, CMNH 88047-88049, 33829, 33831; Vietnam 0, FMNH 32220; R
refirlgens - skull 0: Thailand 0, USNM 528280;

47. R rex - skull (3):China (3), AMNH 84381, 56893, FMNH 39548; Skin (4): China (4),
AMNH 84891, 56893, 56970, MCZ 20286;

48. R robinsom‘ - Skull (2)2Malaysia (1), AMNH 236201; Thailand (1), CMNH 88050; skin
(2): Malaysia (1), AMNH 236201; Thailand (1), CMNH 88050;

49. R rouxi - skull (38):Sri1anda (1), USNM 540556; China (25), USNM 238834, 238836,
238838-238844, 279352, 279353, AMNH 84848, 84855, 84857, 84859, 44682, 44694,
60217, 56935, 60225, CMNH 92144, 92145, FMNH 33823, 33824, 33896; India (3),
CMNH 92241, FMNH 82634, 82635; Assam (8), FMNH 75958, 75959, 75965, 76063-
76026, 76013; Ceylon (2), FMNH 99465, 35376; skin (10): India (4), CMNH 92241,
FMNH 82634, 82635, 32632, 32633; Ceylon (3), FMNH 99465, 35375, 35376; China
(3),CMNH 92144, 92145, 92241;

50. R rufirs - skull (10): Philippine (10), USNM 303953, 458613458616, 573588, FMNH
61220-61222, 49275; skin (11): Pilippine (11), MCZ 35086, 3508845090, 35092, 35166,
35167, FMNH 6122061222, 49275;

157

51. R sedulus - skull (6): Malaysia (6), USNM 115494, 449975, 449976, AMNH 235576,
247289, FMNH 87277; skin (8): Burma (5), 106801, 235576, 234788-234790; Malaya
(3), AMNH 234088-234090;

52. R shameli - skull (1):Thailand (1), USNM 528285-528287;
53. R slhesrris -skull(1):Gabon(1),F1VINH 73827;

54. R simplex - skull (3): Lesser Sundas 10, AMNH 54861; skin (3): Lesser Sundas (3),
AMNH 54861, 54862, 54868;

55. R. simulator - skull (33):Mommbique (5), USNM 365193, 365198-365201; S. Afiica
(18), USNM 376755, 368607, AMNH 168149-168155, CMNH 4607746082, 46084-
46086, 40686; Liberia (1), AMNH 265746; S. Africa (9), AMNH 168241, 245213,
245214, 257165-257170; skin (1 1): CMNH 4607640686;

56. R stheno - skull (10):Thailand (2), USNM 528288, 88051; Vietnam (2), USNM 320629,

320631; Malaysia (6), AMNH 235577, 216905, 216909, 216929, 216931, FMNH 64090;

skin (11): Malaysia (11), AMNH 216921, 216923-216928, 216930, 216931-216934,
CMNH 88051;

57. R. subbadius - skull (7): India (1), USNM 398802; Vietnam (3), FMNH 32216, 32209,
32266; Assam (3), FMNH 76018-76020; skin 0: India (1), USNM 398802; Vietnam (3),
FMNH 32216, 32219, 32226; Assam (3), FMNH 85062, 75976,76019;

58. R submfils - skull (12):Philippine (8), USNM 303901, 303903, 303907, 303908, 125315,
573286-573288, AMNH (4), 241804, 241808, 241810, FMNH 49274; skin (8): Philippine
(8), FMNH 49274, MCZ 35010-35014, 35094-35096;

59. R Mmryi - skull (6): Afiica (3), USNM 344268, CMNH 93176, 36971, 98532; Botswana
(2), AMNH 207416, 115827; Mozambique (1), USNM 365202; skin (6): Afiica (6),
USNM 344268, 365202, 368608, CMNH 93176, 36971, 98532;

60. R thomasi - Skull (15): China (9), USNM 258019, 260044, FMNH 33680, 33286-33290;
Burma (3), 142553, 142554, AMNH 115567; Indochina (3), FMNH 32140, 32141,
32231; sldn 0: Indochina (3), FMNH 32140, 32141, 32231; Burma (1), AMNH 115567;
China 0, AMNH 45040, 45041, 45059, 45075;

61. R mfoliatus - skull (25):Bomeo (12), USNM 142384, 153962, 198951, 449977449979,
AMNH 106838, 103825, 103826, FMNH 8241-, 33029, 64092; Sumatra (2), USNM
141091, 143323; Malaysia (8), AMNH 216937, USNM 283687, 481059481061, FMNH
87274, 87275, 64091; Siam (3), USNM 86787, 83537, 258950; skin (10): Borneo (10),

158

AMNH 106837, 106838, 103825, 103826, 106242, 106243, 103875, 216937; FMNH
8241-, 33029;

62. R virgo - skull (16):Phi1ippine (15), USNM 463869, 463873463875, 477624, 477627-
477629, 477633477636, 477638, 477685, 483692, AMMH 207522; Palawan (Philippine)
(1), FMNH 63633; skin (10): Philippine (10), USNM 303954-303958, 303961, FMNH
63632-63634, MCZ 35017;

63. R yrmtmensis - skull (1): Thailand (1), USNM 528298; skin (3): Thailand (3), AMNH
167934, 67937, USNM 528289;

LIST OF REFERENCES

LIST OF REFERENCES

Andersen, K1905a On somebats ofthe genuserinolophus, with remarksoftheirrmrtual
aflinities, and descriptions of twenty-six new forms. Proc. zoo]. Soc. Lond, 2275-145.

Andersen, K 1905b. A list ofthe species and subspecies ofthe genus Rhinoloplms, with some
notes on their geographic distribution Ann. Mag. Nat. Hist, (7), 16:648-662.

Andersen, K. 1905c. Further descriptions of new Rhinolophi from Afiica. Ann. Mag. Nat
Hist, (7), 15:70-76.

Andersen, K 1905d. On the bats of the Rhinolophus arcuatus group, with description of five
new forms. Ann. Mag. Nat Hist, (7), 16:281-288.

Andersen, K 1905c. On the bats of the Rhinolophus macrotis group, with descriptions of two
new forms. Ann. Mag. Nat Hist, (7), 16:289-292.

Andersen, K. l905f. On the bats of the Rhinolophusphilippinensis group, with descriptions of
five new species. Ann. Mag Nat. Hist, (7), 16:243-257.

Andersen, K. 1918. Diagnosis of new bats of the families Rhinolophidae and Megadermatidae.
Ann. Mag. Nat Hist, (9), 21374-384.

Ando, K, Yasazuma, F., Tagawa, T. and Ushida, TA 1983. Further study on the karyotypic
evolution of the genus Rhinolophus (Mammalia: Chiroptera). Caryologia, 36:101-1 1 1.

Archie, J .W. 1985. Methods for coding variable morphological features for numerical
taxonomic analysis. Syst Zool, 34(2): 326-345.

Bogdanowicz, W. and Owen, R D. 1992. Phylogenetic analysis of the bat family
Rhinolophidae. Z 2001. Syst. Evolut. Forsch., 30:142-160.

Bogdanowicz, W. 1992. Phenetic relationships among bats of the family Rhinolophidae. Acta
Iberiol, 37(3): 213-240.

Brenon, P. 1972. The geology of Madagascar. In Biogeogrqrhy and ecology in Madagascar.
eds. Battistini, B. and Richard-Vindard, G., Dr. W. Junk B.V., Publishers, The Hague.

Briggs, J .C. 1984. Center of origin in biogeography. Biogeographical Monographs 1. Univ. of
Leeds Printing Service.

159

160

Briggs, J.C. 1987. Biogeogrcphy and plate tectonics. Elsevier, Amsterdam, Oxford, New
York and Tokyo.

Bryant, HN. 1989. An evaluation of cladistic and character analysis as hypothetico-deductive
procedures, and the consequences for character weighting. Syst. Zool, 38(3): 214-227.

Corbet, GB. and Hill, J.E. 1981. A world list of mammalicm species. British Mus. (Nat. Hist),
London, Comstock

Corbet, GB. and Hill, J .E. 1992. fire mammals of the Irdonralcnmr region. Nat. Hist. Mus.
Pub., Oxford Univ. Press. Oxford.

Cracrafi, J. 1988. Deep-history biogeography: retrieving the historical pattern of evolving
continental biotas. Syst. 2001. 37:221-236.

Cranbrook, E. 1981. The vertebrate faunas. In Wallace's line and plate tectonics. ed.
Whitrnore, T.C., pp. 57-69. Clarendon Press, Oxford.

Craw, R 1988 Continuing the synthesis between panbiogeography, phylogenetic systematics
and geology as illustrated by empirical studies on the biogeography of New Zealand
and the Chatharn Islands. .8361 Zool, 37(3):291-310.

Darlington, P.J.Jr. 1958. Area, climate, and evolution. Evolution, 13:488-510.

Dawson, MR and Krishtalda, L. 1984. Fossil history of the families of resent mammals. In
Orders and families of recent mammals of the world eds. Anderson, S. and Jones,
J .K.Jr., pp11-58. John Wiley & Sons. New York

DeBlase, AF. 1980. The bats of Iran: systematics, distribution, ecology. Fielditma Zool.,
No. 4. pp1-424.

Dransfield, J. 1981. Palms and Wallace’s line. In Wallace's line and plate tectonics. ed.
Whitrnore, T.C., pp. 43-56. Clarendon Press, Oxford.

Dubes, RC. and Jain, AK. 1991. Pattern recognition. Notes for CPS808, Computer Science
Department, Michigan State University, East Lansing, Michigan. pp 1-322.

Duncan, T. and Stressy, TR 1984. Cladistics: Perspective on the Reconstruction of
Evolutionwy History. Columbia University Press. New York.

Dulic, B. and Mutere, FA 1974. The chromosomes of two bats from East Afiica:
Rhinolophus clivosus Cretzchma 1928 and Hipposideros cqfi’er (Sundevall, 1946).
Period Biol, 76:31-34.

161

Eldredge, N. 1979. Cladisrn and common sense. In Phylogenetic analysis and paleontology.
eds. Cracrafi. J and Eldredge, N, pp165-198. Columbia University Press, New York

Ellennan, J .R, Monison-Scott, T.C.S. and Hayman, RW. 1953. Checklist of southern
African marrrrnals. British Mus. (Nat. Hist), London.

Ellennan, J .R and Morrison-Scott, T.C.S. 1966. Checklist of Palaearctic and Indian
nrwrrmals. British Mus. (Nat. Hist), London.

Farris, IS 1969. A successive approximations approach to character weighting. Syst. 2001.,
18:374-385.

Felsenstein, J. 1988. Phylogenies and Quantitative characters. Syst. Zool. 45:445-471.

Findley, IS 1972. Phenetic relationships among bats of the genus Myotis. Syst Zool, 21 :31-
52.

Freeman, P.W.1981. A multivariate study of the family Molossidae (Marnmalia, Chiroptera):
morphology, ecology and evolution. Fieldiana Zool., n.s., no.7, vii + 173p.

Geske, J .G. 1992. Theory of Algorithms. Course notes for CPS834. Department of Computer
Sciences, Michigan State University, East Lansing, Michigan. pp 1-259.

George, W., 1981. Wallace and his line. In Wallace's line and plate tectonics. ed. Whitrnore,
T.C., pp. 3-8. Clarendon Press, Oxford.

Garey, MR. and Johnson, D.S. 1991. Computers and intractability: A guide to the theory
of NP-completeness. W.H. Freeman and Company, New York.

Goodwin, RE. 1979. The bats of Timor: Systematics and ecology. Bull. Amer. Mus. Nat.
Hist, 163(2): 77-122

Gressitt, IL 1956. Some distribution patterns of Pacific Island faunae. Syst. Zool. 5:11-32.

Hall, LS. 1989. Rhinolophidae. In Fauna of Australia. Vol. 1. Mammalia eds. Walton, D.W.
and Richardson, B.J., pp. 857-863. Govern. Pub. Service. Canberra.

Harada, M., Minemwa, M., Takada, S., Yenbutra, S., Nunparkdee, P. and Ohtarli, S. 1982.
Karyological analyses of 12 species of bats fiom Thailand. Caryologia, 35: 269-278.

Harada, M., Yenbutra, S., Yosida, TH. and Takada, S. 1985. Cytogenetical study of
Rhinolophus bats (Chiroptera, Marnmalia) from Thailand. Proc. Jpn Acad. Ser. B.,
61:455-458.

Hayman, RW. and Hill, J.E. 1971. Part 2, Order Chiroptera In The mammals of Afiica:

162

cm identification Manual. eds. Meester, J. and Setzer, RW., pp1-73.
Smithsonian Institution Press. Washington, DC.

Heaney, LR, Gonzales, PC. and Alcala, AC. 1987. An annotated checklist of the taxonomic
and conservation status of land mammals in the Philippines. Silliman J., 34:32-66.

Hecht, MK. and Edwards, J.L. 1977. The methodology of phylogenetic inference above the
species level. In Major patterns in vertebrate evolution. eds. Hecht, MK, Goody,
PC and Hecht, BM, pp 3-51. NATO Advanced Study Institute Series A Volume
14. Plenum Press. New York and London

Hill, J.E. 1963. A revision of the genus HWderw. Bull Brit Mus (Nat. Hist.) 2001., 11:1-
129.

Holloway, JD. and Jardine, N. 1968. Two approaches to zoogeography: a study based on the
distributionsofbutterflies, birds, andbatsintheIndo—Australianarea. Proc. Linn.
Soc. London, 179:153-188.

Honacki, 1H, Kinman, KB and Koeppl, J.W. (eds) 1982. Mammal species of the world: A
taxonomic and geogrcphic reference. Allen Press and the Assoc. Syst. Collec.,
Lawrence, USA

Hutchinson, CS. 1989. Geological Evolution of South-east Asia. Clarendon Press. Oxford

Humphries, J. M. ,Bookstein, F. L. ,Chemofl; B. ,Smith, F. R, Elder, KL. and Poss, S. F. 1981.
Multivariate discrimination by shape m relation to size Syst. 2001. ,30(2): 291-308.

Huxley, TH. 1868. On the classification and distribution of the Alectoromorphae and
Heteromorphae. Proc. zoo]. Soc. Lond. 1868:294-319.

Jepson, G.L. 1970. Bat origins and evolution In Biology of bats. ed. Wrmsatt, WA Acad.
Press. New York and London

Kalayeh, HM. and Landgrebe, DA 1983. Predicting the required number of training samples.
IEEE Transactions on Pattern Analysis and Machine Intelligence. Vol. Pam-5, No 6.
pp. 664-667.

Koopman, RF. 1965. Status of forms descnbed or recorded by J .A. Allen in “American
Museum Congo Expedition collection of bats”. Am. Mus. Novit., 2219: 1-34.

Koopman, KR 1966. Taxonomic and distributional notes on southern African bats Palm,
Occ. Papers Dept. Game and Fisheries, Zambia, no. 4.

Koopman, KR 1970. Zoogeography of bats. In About bats: A chiropteran biology
symposium. eds. Slaughter, BB. and Walton, D.W., pp 29-50. Southern

l 63
Methodist Univ. Press. Dallas.

Koopman, KR 1975. Bats of Sudan Bull Amer. Nus. Nat. 154: 353-444.

Koopman, RF. 1984. Bats. In Orders and families of recent mammals of the world. eds.
Anderson, S. and Jones, J.KJr., pp. 145-186. John Wiley & Sons. New York.

Koopnmn, KR 1989. Distributional patterns of Indo-Malayan bats (Mammalia: Chiroptera).
Amer. Mus. Novit. 294211-19.

Koopman, KR 1992. Order Chiroptera. In Mammal wecies of the world: A taxonomic and
geographic reference. eds. Wilson, DE. and Reeder, D.M_ Smithsonian Institution
Press. Washington and London.

Koopman, K.F. and Jones, J .K., Jr. 1970. Classification of bats. In About bats: A
chiropteran biology symposium. eds. Slaughter, B.H. and Walton, D.W., pp 22-28.
Southern

Methodist Univ. Press. Dallas.

Lekagul, B. and McNeely, 1A 1977. Mammals of Thailand. Sahakambhat, Bangkok.
ppl-758.

Lomolino, M.V. 1994. Species richness of mammals inhabiting near-shore archipelagoes: area,
isolation, and immigration filters. .1. Mcnnm. 75(1):39-49.

Lydekker, R 1896. A geographical history of mammals. Cambridge University Press.

Luginbuhl, RC. and Schlotzhauer, SD. 1987. SAS/STAT Guide for Personal Computers,
Version 6 Edition. SAS institute Inc. Cary, NC.

MacArthur, RH, and Wilson, EO 1967. fire theory of island biogeography. Princeton Univ.
Press, Princeton, New Jersey.

Maddison, W.P., Donoghue, MI. and Maddison, DR 1984. Outgroup analysis and
parsimony. Sjrst. Zool. 33(1):83-103.

Maddison, WP. and Maddison, DR 1992 MacClade: Analysis of Phylogeny and Character
Evolution, version 3.0. Sinauer Associates, Sunderland, Mass.

Michevich, M. F. and Johnson, MP. 1976. Congruence between morphological and allozyme
data in evolutionary inference and character evolution. Syst. Zool. 25:260-270.

Miller, G.S., Jr. 1907. The families and genera of bats. Bull. U.S. Natl. Mus. 57: 1-282.

Nefl; NA 1986. A rational basis for apriori character weighting. Syst. 2001. 35:1 10-123.

164

Nelson, G. and Platnick, NJ. 1981. Systematics and biogeography: Cladistics and vacariance.
Columbia Univ. Press. New York

Ness, H.V. 1979. On the effects of dimension in discrirnanant analysis for unequal covariance
populations. Technometrics 21(1):]19-127.

Pelseneer, P. 1904. La ligne de Weber, linite zoologique de l’Asie et de l’Australie. Bull. Acad.
r. Belg, Cl Sci. 1904:1001-1022.

Ride, W.D.L., Sabrosky, C.W., Bernardi, G. and Melville, RV. 1985. International Code of
Zoological Nomenclature. Third edition. International Trust for Zoological
Nomenclature, British Museum (NH), London; University of California Press,
Berkeley and Los Angeles.

Qurnsiyeh, MB., Owen, RD. and Chesser, RK 1988. Difl‘erential rates of genic and
chromosomal evolution in bats of the family Rhinolophidae. Genome, 30:326-335.

Rosevear, DR 1965. The bats of west Afi'ica. Brit. Mus. (Nat. Hist). London.

Savage, D.E. an Russell, DE. 1983. Mammalian palaeoftmnas of the world. Addison and
Wesley Publication. London.

Sharkey, MJ. 1989. A hypothesis-independent method of character weighting for cladistic
analysis. Cladistics. 5:63-86.

Slaughter, B.H. 1970. Evolutionary trends of Chiropetan dentition. In About bats: A
chiropteran biology symposium. eds. Slaughter, B.H. and Walton, D.W., pp 51-
83. Southern Methodist Univ. Press. Dallas.

Smithers, RH.N. 1983. The mammals of southern African subregr'on. Univ. Pretoria, pp 1-
736.

Sneath, P.H.A and Sokal, RR. 1973. Numerical taxonomy: The principles and practice of
mrmerical classification W.H. Freeman, San Francisco.

Straney, DO. 1981. The stream of heredity: genetics in the study of phylogeny. In Mammalian
population genetics. eds. Smith, MH. and Joule, J., pp. 100-138. Univ. Georgia
Press, Athens, Georgia

Straney, DO. 1984. The nasal bones of Chiroderma (Phylostomatidae). J. Mamm. 65(1): 163-
165.

Strauss, ER and Bookstain, FL. 1982. The truss: body form reconstructions in
morphometrics Syst. 2001., 31(2):113-135.

165

Swofl‘ord, DL, and Begle, DP. 1993. PAUP: Phylogenetic Analysis Using Parsimony. User's
Manual. Smithonian Institution

Tate, G.HH and Archbold, R 1939. Results of the Archbold expeditions. No. 24. Oriental
Rhinolophus, with special reference to material from the Archbold collections. Am
Mus Novit. 1036:1-12.

Tate, G.H.H 1943. Results of the Archbold expeditions. No.49. Further notes on the
Rhinolophus philippinensis group (Chiroptera). Amer. Mus. Novit. no. 1219.

Tate, G.H.H. 1946. Geographical distribution of the bats in the Australasian Archipelago.
Amer. Mus Novit. 1323:1-21.

Vail, PR and Mitchum, RM, Jr. 1979. Global cycles of relative change of sea level from
seismic stratigraphy. Geological and geoplprsical investigations of continental
margins, American association of petroleum geologists memoir, 292496-72.

Van Valen, L., 1979. The evolution of bats. Evolutionary Theory. 4: 103-121.

Walker, E.P. 1964. Mammals of the world Johns Hopkings Press. Baltimore. pp. 1-1500.

Wallace, AR 1860. On the zoological geography of the Malay Archipelago. J. Linn. Soc.
(Zool), 42172-184.

Watrous, LE. and Wheeler, QB. 1981. The out-group comparison method of character
analysis. Syst. 2001. 30(1): 1-1 1.

Wilkinson, L., Hill, M., Miceli, S., Birkenbeuel. G. and Vang E. 1992. Systat: Graphics.
Systat, inc. Evanston.

Wiley, E.O. 1988a Parsimony analysis and vicariance biogeography. Syst. 2001.,
37(3):271-290.

Wiley, E.O. 1988b. Vacariance biogeography. In Annual Review of Ecology and systematics.
Vol. 19. eds. Johnston, RF., Frank, RW. and Michener, C.D., pp.513-542. Annual
Reviews Inc. Palo Alto, California

Zirna, J. 1982. Karyotypes of three species of horseshoe bats fi'om Czechoslovakia. Lynx
(Prague). 21:121-124.

"‘lllllllilllllllli