Citation
Systematic revision of the Neotropical fruit bats of the genus Sturnira

Material Information

Title:
Systematic revision of the Neotropical fruit bats of the genus Sturnira a molecular and morphological approach
Creator:
Iudica, Carlos Alberto, 1959-
Publication Date:
Language:
English
Physical Description:
xii, 284 leaves : ill. ; 29 cm.

Subjects

Subjects / Keywords:
Abbreviations ( jstor )
Bats ( jstor )
Datasets ( jstor )
Liver ( jstor )
Museums ( jstor )
Natural history ( jstor )
Phylogenetics ( jstor )
Species ( jstor )
Taxa ( jstor )
Teeth ( jstor )
Cladistic analysis ( lcsh )
Dissertations, Academic -- Zoology -- UF ( lcsh )
Sturnira -- Classification ( lcsh )
Zoology thesis, Ph. D ( lcsh )
Genre:
bibliography ( marcgt )
theses ( marcgt )
non-fiction ( marcgt )

Notes

Thesis:
Thesis (Ph. D.)--University of Florida, 2000.
Bibliography:
Includes bibliographical references (leaves 271-282).
General Note:
Printout.
General Note:
Vita.
Statement of Responsibility:
by Carlos Alberto Iudica.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
Copyright [name of dissertation author]. Permission granted to the University of Florida to digitize, archive and distribute this item for non-profit research and educational purposes. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder.
Resource Identifier:
025830167 ( ALEPH )
46714036 ( OCLC )

Full Text









SYSTEMATIC REVISION OF THE NEOTROPICAL FRUIT BATS OF THE
GENUS Sturnira: A MOLECULAR AND MORPHOLOGICAL APPROACH











By

CARLOS ALBERTO IUDICA


A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL
OF THE UNIVERSITY OF FLORIDA IN PARTIAL FULFILLMENT
OF THE REQUIREMENTS FOR THE DEGREE OF
DOCTOR OF PHILOSOPHY

UNIVERSITY OF FLORIDA


2000




























Copyright 2000

by

Carlos A. ludica


























This work is dedicated to Maria Eugenia Fullana Jornet, the most important
person in my life and to John Frederick Eisenberg, an eminent scientist whose
influential company will be missed after his retirement.













ACKNOWLEDGMENTS

This project would not have been possible without the help, support, and

encouragement of many people. Most significantly, I thank John F. Eisenberg for

suggesting the phylogeny of the entire genus Stumira as a dissertation subject.

W. Mark Whitten and Norris H. Williams guided me through all the lab work

involved in this project. I thank them also for their willingness to maintain endless

conversations on varied subjects, from science to history, from politics to life. For

their suggestions during the planning stage, for assisting on every aspect of my

work, and for helping me to process and interpret the data, I thank the members

of my committee: Brian Bowen, Colin Chapman, John F. Eisenberg

(Chairperson), Walter Judd, and Norris H. Williams. Although W. Mark Whitten

was not a formal member of my committee, his daily contributions were essential

to the development of this work and he deserves an honorary seat on my

committee. I thank Wesley E. Higgins, Juan Manuel Alvarez, Savita Shankar,

Ana "Sam" Bass, and Ginger A. M. Clark for their guidance on statistical and

computer analyses, and for their valuable help in the lab. I am grateful to W.

Mark Whitten and Norris H. Williams for allowing me to have unrestricted access

to their computers and to the molecular systematic laboratory at the Florida

Museum of Natural History.

Funding for this project was provided by several agencies, institutions, and

people. I thank The Lincoln Park Zoological Society for their "Scott Neotropical








Fund," the National Museum of Natural History and the American Museum of

Natural History for their "Collection Study Grants." Funding also was available

through travel awards from the Department of Zoology, the Graduate Student

Council, and the College of Liberal Arts and Sciences of the University of Florida.

The McLaughlin Dissertation Fellowship from the University of Florida partially

supported the last stages of this work.

Special thanks are extended to the following institutions and people who

kindly provided access to comparative specimens and tissue collections in their

care: American Museum of Natural History, New York (Nancy B. Simmons, Karl

F. Koopman, Brian Kraatz); Angelo State Natural History Collections, Angelo

State University, San Angelo (Robert Dowler); Carnegie Museum of Natural

History-Edward O'Neil Research Center, Pittsburgh (Timothy McCarthy, Sue

McLaren, Duane Schlichter, John Wible); Centro de Ecologia, Universidad

Nacional Aut6noma de M6xico, Distrito Federal (Rodrigo Medellin, Ricardo

Lopez-Wilches); Florida Museum of Natural History, University of Florida,

Gainesville (Charles A. Woods, Laurie Wilkins, Candace McCaffery); The Field

Museum, Chicago (Bruce D. Patterson, John Phelps); Museum of Natural

Science, Louisiana State University, Baton Rouge (Mark S. Hafner, Shannon K.

Allen, David Reed); Museum of Comparative Zoology, Harvard University,

Cambridge (Maria E. Rutzmoser, Terri McFadden, Gail Pinderhughes); Mus6um

d'histoire naturelle de la Ville de Gen6ve, Gen~ve; Museum of Southwestern

Biology, University of New Mexico, Albuquerque (Terri Yates, William Gannon);

Michigan State University, East Lansing; Museo de Historia Natural de la

Universidad Mayor de San Marcos, Lima (Irma Franke, Elena Vivar, Sergio








Solari); Museum of Vertebrate Zoology, University of California, Berkeley (James

L. Patton, Carla Cicero); Museum of Zoology, University of Michigan, Ann Arbor

(Phil Myers); United States National Museum of Natural History, Smithsonian

Institution, Washington (Linda Gordon, Alfred L. Gardner, Charles 0. Handley,

Jr., Jeffrey F. Jacobs, Don E. Wilson); Oklahoma Museum of Natural History,

University of Oklahoma, Norman (Michael Mares, Janet Braun); Royal Ontario

Museum, Centre for Biodiversity and Conservation Biology, Toronto (Mark

Engstrom, Lim Burton); Texas Cooperative Wildlife Collection, Department of

Wildlife & Fisheries Sciences,Texas A&M University, College Station (Duane

Schlitter, John W. Bickham); The Museum, Texas Tech University, Lubbock

(Nicky Ladkin, Robert Baker, Ricardo Monk); University of Nebraska State

Museum, Lincoln (Patricia Freeman, Thomas Labedz).

Many of my fellow students and colleagues opened their intellects and

invited me to fruitful discussions, creating the best academic environment at

school. Their friendship and support were a constant source of advice. I am

specially indebted to my friends for their comments and encouragement.

My deepest gratitude goes to my wife Maria, whose love, understanding,

patience, and support have been pivotal during these five years. She helped me

out in periods of great stress and encouraged me to pursue and continue my

work through her example of integrity and determination. I am truly grateful to

her.













PREFACE

During the course of my masters research (ludica 1994) 1 observed the

important ecological role that bats of the genus Sturnira have as seed dispersers

in northwestern Argentina. I became intrigued by the fact that among those fruit

bats there were three allied species with almost the same body size and (from

what was known) probably the same dietary requirements. This overview of the

ecological role of those bats in that particular rain forest provided the basic

information that I used as a preamble for the problem that I wanted to work on for

a Ph.D. dissertation. Together with Stumira lilium, two other putative species of

Stumira were supposed to occur sympatrically in the mountain rainforest of

northwestern Argentina. Tschudi (1844) described those two allied species as S.

erythromos and S. oporaphilum. Not much was known about their ecology, other

than what was assumed from studies of other related fruit bats and what was

known from S. lilium, a much more common, widespread, and better studied

species. S. erythromos is a well defined species with a restricted distribution on

the eastern slopes of the Andes from northwestern Argentina to Ecuador. The

other, S. oporaphilum represents a rather controversial taxon assignment (see

below on current status of Stumira), and little was known about its ecology.

I decided to analyze the status of these three co-occurring species in

northwestern Argentina in an effort to resolve their taxonomic status. My Ph.D.

advisor, Dr John F. Eisenberg, suggested that I embrace the entire problematic








genus instead of just doing a partial overview of three species of Sturnira.

Without realizing the challenge, I naively embarked in what was a new journey

for me. I then resolved to change my PhD project from being an ecological

description of resource partitioning among three sympatric fruit bat species into a

systematic review of the phylogeny of the entire genus Sturnira. In 1996 the

molecular systematics of bats was already an accepted taxonomic tool. I

decided to explore the evolutionary history of Sturnira by combining at that time

the most up-to-date techniques in molecular systematics with the traditional, low-

budget (and most "trustworthy" for many), morphological approach. The ultimate

goal was to obtain hard evidence to explain and describe the systematic status

and phylogenetic relationships among species of the genus Sturnira.













TABLE OF CONTENTS

Aae..


ACKNOW LEDGMENTS ........................................... iv

PR E FA C E ..................................................... vii
A BSTRAC T .................................................... xii

1. INTRO DUCTIO N .............................................. 1

Systematic Background and Current Status of Sturnira ................... 1

The Importance of a Phylogeny Based on DNA: Why Use the
Cytochrome b Gene? ........................................ 6

2. MATERIALS AND METHODS ................................... 16

M orphological Data .............................................. 16

M olecular Data ................................................. 18
DNA Extraction, Polymerase Chain Reaction (PCR),
and Sequencing .......................................... 18
DNA isolation and extraction ............................ 18
Gene amplification, PCR, and sequencing ................. 22
Search Strategies ......................................... 24

3. R ES U LTS ................................................... 35

All Available DNA Sequences and Selected Specimens ................. 35
The Large Data Set: 133 Specimens (4 Outgroup Species
and 129 Sturnira Specimens) ................................ 35
The Small Data Set: 37 Specimens (4 Outgroup Species
and 33 Stumira Specimens) .................................. 40
The DNA data set .................................... 40
The morphological data set ............................. 42
Combined data--DNA and morphological data sets .......... 43
Pairwise Distances ......................................... 45
Patterns of Character Evolution: Selected Cases ................. 46








DISCUSSIO N ...................................
The Corvira Complex .............................
The Ludovici Complex ............................
The Luisi Com plex ...............................
The Lilium Complex ..............................
The Entire "Well Defined Lingual Cusps" Group .........
The Remaining Species ...........................
Pairwise Distances ...............................
The Whole Picture: The Genus Stumira ...............
Final Rem arks ...................................

APPENDIX 1
INSTITUTIONAL NAMES ....................

APPENDIX 2
MORPHOLOGICAL CHARACTERS ............

APPENDIX 3
SPECIMENS MEASURED ....................

APPENDIX 4
QUALITATIVE CHARACTERS ................


APPENDIX 5
CHARACTER MATRIX ..........

APPENDIX 6
TISSUE SAMPLES .............


.......................... 247


.......................... 248


APPENDIX 7
LIST AND FORMULAE OF STOCK SOLUTIONS USED
IN THE DNA EXTRACTION OF TISSUE SAMPLES...


APPENDIX 8
FINAL SEQUENCES ..........


......... 261


............................ 262


REFERENCES ......................

BIOGRAPHICAL SKETCH .............


272

284


185
189
190
194
197
198
199
201
201
203


219


221


223


243













Abstract of Dissertation Presented to the Graduate School
of the University of Florida in Partial Fulfillment of the
Requirements for the Degree of Doctor of Philosophy
SYSTEMATIC REVISION OF THE NEOTROPICAL FRUIT BATS OF THE
GENUS Sturnira: A MOLECULAR AND MORPHOLOGICAL APPROACH

By

Carlos Alberto ludica

December 2000

Chairperson: John Frederick Eisenberg
Major Department: Zoology

The Neotropical fruit bat tribe Stenodermatini consists of 68 species in 19

genera, eleven of which are monotypic and only two genera contain ten or more

species. One of those two is the genus Stumira, which inhabits several

landscape types, and has a geographical range extending from M6xico to

Argentina.

Because a comprehensive revision of relationships within this diverse and

perplexing genus has not been attempted since 1961, and several new forms

subsequently have been discovered, I designed a project to describe the

diversity of this group and to resolve the systematic relationships among species

of the genus Stumira. This was done by sequencing the cytochrome b gene,

generating a matrix of morphological characters, and combining both data sets

into a phylogenetic analysis using a parsimony algorithm.








A final successively weighted heuristic analysis produced a single most

parsimonious tree of 295 steps with a consistency index of 0.65 and a retention

index of 0.72. Seventy six percent of all clades had bootstrap support values

larger than 85%.

Stumira aratathomas S. lilium (sensu lato), S. parvidens, S. luisi, and S.

thomasi, constitute a clade of closely related taxa. Sturnira parvidens is a valid

species and the sister taxon to S. lilium, a relationship highly supported by

bootstrap percentage and decay values. Sturnira bogotensis is also a valid

species and forms a highly supported clade with S. erythromos. The highly

derived S. mordax forms a moderately supported clade with S. sp. A, another

highly derived taxon. Sturnira ludovici should only refer to specimens from

Ecuador and constitutes a highly supported clade with S. oporaphilum. Stumira

hondurensis, a valid species represented by only Central American specimens is

a sister taxon to the S. oporaphilum-S ludovici clade and its phylogenetic

relationship is highly supported. The subgenus Corvira is distinct from subgenus

Sturnira and should be recognized as such. Both species that comprise Corvira

have numerous apomorphies that separate them from the remaining species of

Sturnira. The molecular and morphological data confirm this distinctiveness.













CHAPTER 1
INTRODUCTION



Systematic Background and Current Status of Sturnira


The order Chiroptera includes two suborders, the Megachiroptera

represented by the Old World fruit- and nectar-feeding bats, all part of the family

Pteropodidae with 42 genera and 166 species; and the Microchiroptera, with 16

families, 135 genera, and 759 species (Kunz and Pierson 1994). Within

Microchiroptera, six families are exclusively Neotropical, with the most diverse

family (Phyllostomidae) containing 49 genera and 148 species. The

Phyllostomidae family can be divided into eight subfamilies, one of which

(Stenodermatinae), contains the only two genera represented by more than 10

species: Artibeus with 17 recognized species, and Stumira with two subgenera

and 12 recognized species (after Koopman 1993).

The type of Stumira lilium was originally described by Geoff roy Saint-

Hilaire (1810) as Phyllostoma lilium. Gray (1842) created the genus Sturnira

using S. lilium to typify the genus, but naming the species as S. spectrum, and

Gervais (1855) used the combined name of S. lilium for the first time. Dobson

(1878) offered a detailed description of morphological features of head, body and

hair. He explained that incisors in S. lilium are like those found in Carollia

brevicauda. The central upper incisors are unicuspidate, with oblique cusps.










The outer incisors are short and broad, with anterior surface concave and the

summits broadly conical. Either the lower incisors are equal or the middle

incisors are slightly larger with straight faintly notched cutting edges. The M1 is

larger in antero-posterior diameter at the base, but smaller in vertical extent. The

M3 is scarcely 1/3 the size of M1, with oval crown wider than longer. The pml is

larger than pm2 but equal in vertical extent. The m3 equals M3 but is longer than

wider. Molars are concave with an outer notched cutting edge. The author also

offered measurements from a male adult and stated that this bat inhabited the

Neotropical Region (tropical and subtropical areas), and reported distributional

records for specimens from Jamaica (probably Artibeus sp. or another

Stenodermidae), Honduras, several localities in Brazil (Pernambuco, Bahia, Rio

de Janeiro, St. Catarina, Minas Gerais, and somewhere in the Rio Napo, border

with Ecuador). Dobson's work offered very detailed, high quality figures of skulls

and faces of bats (see plate 28-5 and plate 30-4 in Dobson 1878).

Cabrera Latorre (1903) described general characteristics of the three

genera represented by one species each in Chile (Glossophaga, Stumira, and

Desmodus). Specifically for Sturnira lilium, dental formula, synomy,

morphological description, and selected measurements are offered. Cabrera

Latorre mentioned that this species inhabited Chile, Paraguay, Brazil, PerI,

Ecuador, and Central America including the Antilles (at least Jamaica). Elliot

(1904) produced a general description of external and cranial morphology of S.

lilium. Elliot was another author who used S. lilium to typify the entire genus for

which the distribution was then known from M6xico (Jalisco and Veracruz) to

Paraguay and Chile. Miller (1907) again used S. lilium to characterize the entire










genus Stumira, since at that moment, no other "forms" were recognized. Miller

used the "aberrant and highly specialized" dentition of Sturnira to justify the

existence of the subfamily Sturnirinae (Miller 1907: 148).

Thomas (1915) described for the first time Corvira bidens, referring this

new species to a new genus allied to Stumira (particularly to S. lilium). He based

his description on a "somewhat immature" (:311) individual whose size and

general appearance was very similar to S. lilium. The type specimen, an

immature male, BM 15.7.11.7 (original # 19) was collected in April 1914 by W.

Goodfellow at Baeza, Upper Coca River, Oriente of N. Ecuador, alt. 6,500'

(1,981 mts). The diferences between S. lilium and C. bidens mentioned by

Thomas are 1) only two lower incisors present, bicuspid; 2) other lower teeth

shorter antero-posteriorly than in Stumira; 3) muzzle and interorbital region

narrower; 4) angular process of mandible shorter; 5) upper incisors more

disproportionate than in Sturnira, outer ones smaller and narrower, inner pair

longer and slenderer, with small supplementary basal cusp postero-externally;

6) premolars and molars slightly separated from each other; 7) premolars evenly

spaced, oval shaped, transversely positioned, width about 0.75 of that in S.

lilium, and antero-posterior diameter only 0.50 of that in S. lilium; 8) M1

subtriangular, with rounded angles, more carnassial-shaped than in S. lilium; and

9) more triangular shape of ml and better developed cusps on some of the teeth

(a characteristic used by Thomas to assume that Corvira is less specialized than

Sturnira).

Several new entities subsequently have been discovered and described

(i.e., Davis 1980, Gardner and O'Neill 1971, Tamsitt and Valdivieso 1986), but a










revision of relationships within this diverse group was not undertaken until the

study of de la Torre (1961). His work was the first attempt to understand the

morphological variation and systematics of Stumira. The author recognized eight

species of Sturnira, but since his conclusions are based on a partial subset of the

genus, his mention here is only anecdotal.

Owen (1987) presented the second effort to describe the phylogeny of

Stumira (Figure 1). He recognized two subgenera (Stumira and Corvira) and

within the subgenus Stumira his data set revealed two major clades, one

represented by S. lilium and S. aratathomasi two very similar taxa, which share

several morphological features. The other clade was formed by the remaining

species (included in his analysis) within the subgenus Sturnira, but no resolution

of relationships among the species was offered. The subgenus Corvira, a sister

group of the subgenus Stumira, included S. nana and S. bidens. Pacheco and

Patterson (1991) addressed the same problem of species relationships with two

different tools, allozymic and morphological data sets. Their conclusions are

summarized in Figure 2. These authors also recognized Corvira as the sister

subgenus to Sturnira. Probably because of the small sample size they

suggested that S. sp. A may belong to a different subgenus, which was

hypothesized to be much closer to subgenus Sturnira than to the subgenus

Corvira. Based on their allozymic data set, they concluded that within the

subgenus Sturnira, S. erythromos and S. magna should be included in the same

clade, whereas based on morphological evidence, they suggested that S.

ludovici-S oporaphilum and S. bogotensis-S erythromos "seem to be sister










pairs" (Pacheco and Patterson 1991: 117), and that S. lilium, S. luisi, and S.

thomash form a common lineage.

Koopman (1993) circumscribed the Family Phyllostomidae and recognized

48 genera divided into 8 subfamilies. He described Sturnira with two subgenera

(Sturnira and Corvira) and 12 species. Although he recognized S. bogotensis as

a species, apparently not accepting Pacheco and Patterson's (1991)

conclusions, he suggested that the correct name should be S. oporaphilum, with

S. bogotensis a synonym of S. oporaphilum. It is not clear where the

geographical limits of S. ludovici are and eventually he placed S. hondurensis in

synomy, probably following Hershkovitz (1949) and also reflecting on the largely

accepted current concepts based on superficial morphological characters or

simply based on distribution (Baker and Greer 1962; Carter and Jones 1978;

Dalquest 1953; de la Torre 1952; Goodwin 1953; Jones and Owen 1986; Jones

et al. 1988; Lukens and Davis 1957; Ramfrez-Pulido et al. 1996; Starrett and de

la Torre 1964).

Koopman (1994) put Stumira Gray 1842 in the tribe Sturnirini Miller 1907,

as part of the subfamily Stenodermatinae Gervais 1855. He recognized two

subgenera, subgenus Stumira Gray 1842 with 10 species, and subgenus Corvira

Thomas 1915 with two species. The author gave a series of apomorphic

features to separate the different clades (Figure 3) and listed the known

subspecies. No phylogenetic relationships are depicted but he suggested two

major clades based on degree of development of lingual cusps on lower first

molars (the "serrated" and "not serrated" groups of de la Torre (1961) may have

had some influence on this decision). He did not recognize S. hondurensis or S.










bogotensis as separate species, but he added a bit more to our knowledge with

the listed subspecies. For S. ludovici, he now recognized two subspecies (S.

ludovici hondurensis and S. L occidentalis) with distribution north of Panamd and

S. L ludovici present in northern South America.

McKenna and Bell (1997, with Koopman as a contributor) divided the

family Phyllostomidae into four subfamilies. One of them, the subfamily

Stenodermatinae, was divided into two tribes (Carolliini and Stenodermatini).

The tribe Stenodermatini in turn was divided into two subtribes, subtribe

Stenodermatina and subtribe Sturnirina, with one genus Sturnira Gray 1842,

including subgenus Corvira Thomas 1915, subgenus Stumirops Goodwin 1938

and subgenus Sturnira Gray 1842. In this work, Koopman suggested that

Sturnira is much more closely related to Uroderma, Vampyressa and Centurio

(same tribe, different subtribe) than to Carollia (a different tribe).


The Importance of a Phylogeny Based on DNA: Why Use the Cytochrome b
Gene?


The phylogeny of mitochondrial DNA (mtDNA) sequences should track the

historical relationships of the different populations under study. An important

distinction to make here is that between a gene tree and a species tree or

organismal phylogeny (sensu Pamilo and Nei 1988; Avise 1989). A gene tree is

the phylogeny of haplotypes (unit of inheritance for haploid genotypes) for a

specific segment of DNA, whereas an organismal or species tree reflects the

evolutionary history of a group or lineage characterized by ancestral-descendant

relationships (Nei 1987).










A question emerges about which DNA type (nuclear or mtDNA) has the

evolutionary rate that will provide resolution at the desired level. Since the

introduction of PCR (Polymerase Chain Reaction, Saiki et al. 1988) for amplifying

specific segments of DNA, mtDNA sequences have become a favorite genetic

marker for both macro- and microevolutionary studies (Avise 1994; Palumbi

1996; Patton et al. 1996; Smith and Patton 1993; Wendel and Doyle 1998).

Although vertebrate mtDNA evolves on average about five to ten times faster

than does most single-copy nuclear DNA, no generalization is valid in terms of

rates of evolution. "The pace of mtDNA evolution among animal groups may be

linked causally to differences in metabolic rates and/or to the generation length

and body size differences with which metabolic rate is negatively correlated"

(Avise 1994: 103). Many times, the use of mtDNA has proved to be especially

versatile for interspecific studies, whereas works including nuclear loci has

received far less attention in this kind of investigation (Hillis et al. 1996).

The mitochondrial genome of vertebrates is a circular molecule comprising

13 protein coding genes involved in oxidative phosphorylation, two ribosomal

RNA genes, and 22 transfer RNA genes, as well as the control region (a

noncoding section responsible for replication and transcription). The entire

molecule is approximately 17,000 base pairs (bp) long in most mammals

(Arnason et al. 1995; Anderson et al. 1982; Janke et al. 1994).

The cytochrome b gene (cyt b) is one of the best known of the proteins

associated with the mitochondrial oxidative phosphotylation system (Hatefi 1985)

and the knowledge of the structure and function relationships in this protein

enhances the utility of this gene for evolutionary investigations (Irwin et al. 1991).










The cytochrome b gene has been used extensively in molecular phylogenetic

studies (i.e., Conroy and Cook 2000; Lim and Engstrom 1998; McKnight 1995;

Patton et al. 1996; Peppers and Bradley 2000; Sullivan et al. 1995; Sullivan

1996; Wright et al. 1999), and most specifically, mitochondrial cytochrome b

gene has proven useful in deriving phylogenetic hypotheses of relationships of

species within genera such as Artibeus, Dermanura, Phyllostomus, Chiroderma,

Rhinophylla, and Carol/ia (Baker et al. 1994; Van Den Bussche and Baker 1993;

Van Den Bussche et al. 1993, 1998; Wright et al. 1999). As a protein-coding

gene, cyt b displays very little length polymorphism, which facilitates the

alignment of sequences from different groups of organisms. Homologous DNA

sequences permit comparisons among widely divergent taxa. Because of the

widespread use of cyt b and its status as a universal metric, results of a particular

study can be compared meaningfully to a larger body of work (Meyer 1994b).

There are several recently published studies on systematic issues of Neotropical

bats choosing the entire or partial cyt b sequences instead of a portion of the

nuclear genome as a species-level sampling of variation (i.e., Baker et al. 1994;

Lim and Engstrom 1998; Sudman et al. 1994; Wright et al. 1999). By using the

same molecular marker the data can be compared directly, taxonomic

hypotheses can be falsified, and the results of this study can be put in a broader

context.

Probably representing one of the most extensive sampling efforts for this

kind of study, a combined total of over 600 museum specimens encompassing

the whole geographical range of the genus were used in this project to address

the major purpose of this research, which is to study the diversity and








9

evolutionary history of the genus Stumira. This was done by a) sequencing cyt b

gene, b) generating a matrix of qualitative characters, and c) combining both data

sets into a comprehensive phylogenetic analysis.



























Figure 1. Phylogeny of Stumira proposed by Owen (1987) showing the two
subgenera (Stumira and Corvira) and 12 species included in his analyses. The
bar on the right indicates an apomorphy that does not coincide with proposed
grouping.









11








S. lilium r




S. aratathomasi (well defined
lingual cusps)


S. luisi

S. thomasi

S. tildae

,S. bogotensis

S. erythromos
S. magna

S. ludovici

S. mordax



, S. nana


bidens




























Figure 2. Phylogenies of Stumira proposed by Pacheco and Patterson (1991), based on: A) discrete morphological
characters, and B) electrophoretic analysis of frozen tissues.












S. aratathomasi
S. magna
S. bogotensis
S. erythromos
S. oporaphilum
S. ludovici
S. lilium
S. luisi
S. thomasi
Stumira -S. tildae
S. mordax






S. spA



S. nana
Corvira

S. bidens


Stumira


Stumira


S. oporaphilum

S. luisi

S. ilium


-S. erythromos







-S. magna








bidens


Stumira




























Figure 3. Phylogenetic relationships of the genus Stumira proposed by
Koopman (1994) based on synapomorphies (described). S. thomasi vulcanensis
and S. thomasi thomasi have been added after Genoways (1998).











































Stumira
(uropatagium
virtually absent,
crown of molars
with distinct
longitudinal
grooves, cusps
strictly lateral)


Stumira
(outer lower
incisors well-
developed and
functional,
zygomatic arch
always complete)


(well defined
lingual cusps)


(absence of
vertical notched
that defined
lingual cusps)


S. lilium


S. lilium parvidens
S. lilium lilium
S. lilium angeli
S. lilium zygomaticus
S. lilium paulsoni
S. lilum luciae


IS. lilium serotinus (after
Genoways 1998)

S. luisi
S. thomasi vulcanensis
S. thomasi (after Genoways 1998)
S. thomasi thomasi
(after Genoways 1998)
S. tildae

S. aratathomasi


S. erythromos

S. oporaphilurr&S, oporaphilum bogotensis
IS. oporaphilum oporaphilum

S. ludovici occidentalis
S. ludovici S. ludovici hondurensis

S. ludovici ludovici

S. mordax

S. magna


S. nana

S. bidens


Corvira
(outer lower
incisors vestigial
or absent,
zygomatic arch
weak or
incomplete)













CHAPTER 2
MATERIALS AND METHODS



Morphological Data


More than 500 voucher specimens were obtained as loaned material from

20 museum and scientific collections (see Appendix 1 for abbreviations,

affiliations, and names of institutions and agencies providing specimens for

examination). Twenty-five cranial and postcranial quantitative measurements

were taken from alcohol-preserved specimens, study skins, and skulls of 16

species from the genus Sturnira for the traditional morphometric analysis (see

Appendix 2 and Figures 4 and 5 for abbreviations and description of each

continuous morphological character). The choice of quantitative characters was

based on a suit of dimensions and diagnostic characters used by other

investigators (Arita 1990; Barone 1966; Bogdanowicz et al. 1997; Davis 1964; de

la Torre 1961; Freeman 1981; Legendre 1984; McLellan 1984; Pacheco and

Patterson 1992). Specimens measured for quantitative morphometric analysis

totaling 412 individuals (224 females and 188 males) are listed in Appendix 3. All

continuous characters were recorded to the closest 0.01 mm with a digital caliper

(Fowler Caliper Ultra-Cal Mark Ill, Fowler Co., Newton, MA) connected through a

cable to an interface (Smartcable) and to a PC loaded with the appropriate

software (Wedge for Windows version 1.1 c, 1996 TAL Technologies, Inc.) to








17

automatically collect and store the morphological data (using Microsoft Excel 97

for Windows). Data were arranged by species and within species by sex for

further analyses. Values were log-transformed to stabilize variance (Hutcheson

et al. 1995; Mauk et al. 1999). Morphological differences between males and

females were tested within species by multivariate analysis of variance

(MANOVA) on all quantitative characters (Manly 1994). The analysis was

conducted using SPSS version 8.0. The purpose of this analysis was to assess

the possibility of pooling males and females within species to increase sample

sizes. A power analysis was also conducted to reduce the chance of type II error

(acceptance of the null hypothesis of no difference between males and females)

in species with inadequate sample sizes. Only tests with a power of 80% or

higher were used to test for sexual dimorphism. These tests showed that

significant sexual dimorphism does occur in Sturnira, and therefore males and

females were not pooled, with all subsequent analyses being conducted on

specimens of a single sex. Because females generally had higher sample sizes,

they were used instead of males in subsequent analyses. After all

measurements were taken and analyzed, it was decided to use only the length of

the forearm for this project. The rest of the other quantitative characters present

a considerable number of individuals with missing values and/or showed too

much intrataxon variation. It has long been accepted that the forearm is a good

indicator of absolute body size in bats (Rails 1976).

When breaking more or less continuous variation into states Stevens

(1991) recommended that character states be delimited by carefully analyzed

discontinuities (and not necessarily by looking for absolute gaps) in the character










variation, thus "avoiding arbitrary decisions when delimiting states" (Kron and

Judd 1997: 481). Considering that we have small, medium, and large species of

Stumira, forearm length was tranformed into length classes (0-35 mm,

35-53 mm, and 53 or more mm) to be included in the morphological matrix as

discrete states.

For the qualitative multistate osteological analysis, I used selected

individuals of every species that represent, for each character, all possible states.

The 47 characters used in the present study (see Appendix 4 and Figure 6 for

description of each qualitative character and their states) were scored for four

outgroup species (from 6 individuals) and sixteen ingroup species (from 412

individuals) for phylogenetic analysis and the resulting data matrix (Appendix 5)

generated on MacClade version 3.08a (Maddison and Maddison 1999) was

analyzed using PAUP* version 4.Ob4a (Swofford 2000). Details of the analyses

and search strategies are described later in this chapter.



Molecular Data


DNA Extraction, Polymerase Chain Reaction (PCR), and Sequencing


DNA isolation and extraction

In 1996 1 asked different mammal collections the status of their Stumira

holdings. I was particularly interested not just in the dry or alcohol-preserved

specimens, but in frozen or ethanol-preserved (95% EtOH) tissue samples that

matched vouchers. Among the 223 available samples (Appendix 6) total

genomic DNA was extracted from 169 frozen and fifteen ethanol-preserved










tissue samples (heart, liver, or kidney). The first consecutive 75 processed

samples (60 frozen and fifteen ethanol-preserved tissue samples) were extracted

using a modified phenol-chloroform extraction technique (Hillis et al. 1996), which

consisted of the following steps: tissue was homogenized in 1000 AiL STE buffer

(for this and other stock solution abbreviations, see Appendix 7) with a ground

glass homogenizer. Tissue homogenate was tranfered to a 1.5 mL

microcentrifuge tube, in which 50 AL 20% SDS had been added previously. The

mix was gently shaken. 700 AL buffered phenol was added and gently but

thoroughly mixed and incubated at room temperature for five minutes. The mix

was centrifuged for 5 min (12,000 g), while the second set of tubes was prepared

for the next step. The aqueous (top) layer was removed with a pasteur pipette

and transfered to a new 1.5 mL labeled tube, trying not to transfer the cellular

debris at the phenol/STE interface. The aqueous phase was re-extracted with

phenol as previously described.

Then 700 AL PCI was added, gently but thoroughly mixed, incubated at

room temperature for 5 minutes, and centrifuged for 5 min (12,000 g). The

aqueous (top) layer was removed with a pasteur pipette and the mix was

transfered to a new 1.5 mL tube. The aqueous phase was re-extracted with PCI

as described above.

The 700 ,L Cl was added, mixed gently, and incubated at room

temperature for 5 min. The mix was then centrifuged for 5 min (12,000 g),

aqueous phase transfered to a new tube and re-extracted with Cl as explained

above. About 50 AL (1/10 volume) of 3 M sodium acetate was added and the

tube was filled with cold (-200 C) 95% ethanol and shaken. At this point (and only








20
at this point) some stringy DNA may be observable. The sample was incubated

on ice or in a freezer (-200 C) for 10 to 20 min or overnight, after which the tube

was centrifuged for 5 to 10 min (12,000 g). This yielded a small white pellet of

DNA at the bottom of the tube in most cases. The ethanol was decanted and the

DNA pellet (attached at the botton of the tube) was washed with 0.5 mL of 70%

ethanol to remove any remaining salts. The sample was spun for 5 min (12,000

g). The ethanol was decanted and the DNA pellet was dryed with a vacuum

centrifuge or by inverting the tubes on a rack for about 1 hour. The dryed pellet

was resuspended in 50 to 250 jL of warm TE buffer (depending on the size of

the pellet) for up to12 hours.

The remaining 109 frozen tissue samples (of the original 184 frozen or

ethanol-preserved) were digested and their DNA was extracted using Qiagen

DNeasyTM Tissue extraction Kit (Qiagen # 04-1999, Qiagen, Inc., Valencia, CA),

following the manufacturer's recommendations (DNeasyTM Tissue Kit Handbook,

1999: 16-18), with minor modifications. Each sample was incubated at 55 QC in

180 gL of buffer ATL and 20 14L of proteinase K (20 mg/mL) with occasional

vortexing or gentle shaking. Digestion continued for 1 to 3 hours until the tissues

were completely lysed. When digestion was complete, 200 /L of AL buffer was

added, the mixture was vortexed and incubated at 70 QC for 10 min. Two

hundred ML of 100% ethanol were added, the sample was vortexed, and the pH

was checked using color pHast indicator strips (EM-Reagent, cat. no. 9583). The

pH was adjusted to pH 6.5 to 7.0 with 0.25 M HCI; a pH greater than 7.0 will

prevent adsorption of DNA onto the silica membrane of the Qiagen column. The

pH-adjusted sample was passed through the Qiagen column; the column was










washed with 500 gL of AWl buffer, 500 /L of AW2 buffer and then spun dry.

DNA was eluted twice from the column in 40 to 50 pL of AE buffer (10 mM TRIS,

pH 8.0); column and buffer were incubated for 1 to 5 min at 65 9C prior to

centrifugation to increase yield of DNA from the membrane.

DNA was extracted from 8 dry specimens (bone and/or skin) and from 6

formalin-preserved voucher specimens. The difficulties of obtaining usable DNA

from such specimens are well-known (McArthur and Koop 1999), yet such

specimens form a potentially valuable source of data, especially for populations

of animals that are not represented in museum collections as frozen or ethanol-

preserved vouchers. During this study, I developed extraction protocols that

have routinely yielded amplifiable DNA from up to 66 year old museum skins,

bones, and formalin-fixed tissues. These techniques, which are modifications of

Qiagen DNeasy Tissue Extraction procotols, should be directly applicable to

other mammal specimens.

Samples from dried specimens included bones (2 ribs or 2 to 3 mm of the

diaphysis of a wing bone) or 3 x 2 mm of dry skin with hair. Formalin-preserved

tissue samples included 4 x 3 mm pieces of liver or skin with muscle. All tissue

was chopped into small (<0.5mm) pieces; bone was crushed within a folded

piece of a 10 x 10 cm weighing paper (Fisher Scientific Co., cat. No. 09-898-12B)

with a pair of needle-nose pliers. Chopped or crushed samples were placed in

1.5 mL tubes and washed in 250 4L of phosphate-buffered saline (Sambrook et

al. 1989) for 10 min at 55 'C with occasional vortexing. The sample was spun

briefly and the wash solution decanted. This wash was repeated 3 to 5 times.










The repeated washes act to rehydrate the tissues and remove PCR inhibitors

and residual fixatives.

Digestion and extraction was performed with Qiagen DNeasy Tissue

extraction Kit (Qiagen # 04-1999, Qiagen, Inc., Valencia, CA), with modifications

from the previously explained protocol as follows: after incubation at 55 QC with

occasional vortexing or gentle shaking, digestion continued for 24 to 72 hours

(instead of only 1 to 3 hours) until the tissues were completely lysed. During the

digestion, fresh 20 ,L aliquots of proteinase K were added every ten to twelve

hours until lysis was complete.



Gene amplification, PCR, and sequencing

Sequencing a portion of the cytochrome b gene (ca. 850 bp total length)

was done by amplifying four overlapping regions of ca. 250 bp each. The

outermost primers were those of PA bo et al. (1988) and Edwards et al. (1991),

called here Cytb8F and Cytb7R respectively. Attempts to amplify the entire 850

bp region in one amplification from dry (bone and/or skin) or wet (formalin-

preserved) voucher specimens were unsuccessful, probably due to the highly

degraded template DNA. Consequently, internal primer pairs were designed "ad

hoc" from consensus sequences of Sturnira obtained from fresh tissue samples

of six different species. Those primers were named: bat1 97F (5'-AGC CAC CGC

ATT CAA CTC HG-3'), bat2lOR (5'-CCG TAG TTT ACA TCT CGG CAR-3'),

bat380F (5'-TTC GCC GTC ATA GCC ACA-3'), bat420R (5'-TGG TGA TGA CTG

TTG CTC CYC-3'), bat570F (5'-CCT MCT TCC CTT TAT CGT AG-3'), and

bat620R (5'-CYG GGT CTG ATG GGA TYC C-3').








23

Hot-start PCR reactions (50 gL) for DNA samples from frozen or ethanol-

preserved tissue samples contained Sigma 1OX buffer (5 gL), 25 mM MgCl2

(7 jL), 10 mM dNTPs (1 4L), 1 4L of each primer (10 picomoles), 15 4L of a

saturated aqueous betaine solution, 15 ML water, 1 to 5 pL template (depending

on DNA purity and concentration), and 1 unit of Sigma Taq polymerase. The

thermocycler program consisted of an initial denaturation at 94 -C (3 min)

followed by 35 cycles of 94 1C (1 min), 52 1C (1 min), 72 2C (1 min), ending with

a final elongation at 72 QC for 3 min. Products were visualized on a 1% agarose

electrophoresed gel stained with ethidium bromide. The PCR products were

cleaned using Qiagen Qiaquick columns (QlAquickTM Spin Handbook 1999:19).

Products were cycle sequenced using PE/Applied Biosystems Big Dye-terminator

mix according to manufacturer's protocols, except that cycle sequencing

reactions were scaled down to 5 /L. Both strands were sequenced to assure

accuracy in base calling. Labeled products were analyzed with an automated

DNA sequencer (Applied Biosystems, Inc. model 373a and 377, ABI, Foster City,

California) at the DNA Sequencing Core Laboratory (DSEQ) at the University of

Florida. Negative (template-free PCR reactions) and positive (known DNA

samples) controls were used to test for contamination. For bone, skin, or

formalin-preserved voucher specimens the thermocycler had a variation of its

program, with 39 cycles at 94 QC for 1 min, instead of 35 cycles.

Sequences are read by a laser scanner as a series of dye-labeled

fragments on an acrylamide gel. Thus, the raw data consists of

electropherograms, with fluorescent peaks interpreted by support software as

either A, T, G, or C. Data are stored as both electropherograms and sequence










files, so that ambiguities can be checked against the original scanner output.

Because the data are computer-resident at all times, errors introduced in data

entry and manual transfer are effectively eliminated. Sequence files from the

automated sequencer can be converted directly to a format appropriate for

phylogenetic and statistical analyses. Individual DNA (complementary and

overlapping) sequences were edited and assembled using "Sequence

NavigatorTM" version 1.0.1 and "AutoAssemblerTM" version 1.3.0 (ABI software

packages, Foster City, California) on an Apple iMac computer. Sequences were

aligned manually using an iMac computer with "Sequence NavigatorM" version

1.0.1 and by final visual inspection in PAUP* version 4.Ob4a (Swofford 2000)

using an "ad hoc" font (BKGCuclc) developed by Ron Sogin (Sogin personal

communication). Gaps were coded as missing data. The end of matrices were

trimmed to exclude sequencing artifacts. The final aligned sequences are

available from the author (Appendix 8), and representatives will be submitted to

GenBank upon acceptance of corresponding manuscripts.


Search Strategies

Four complete sets of phylogenetic analyses were conducted: (1) an

analysis including 133 taxa and 778 bp of cytochrome b; (2) an analysis of all

discrete morphological characters, including 47 characters and 37 taxa (4

outgroups and 33 specimens of Stumira); (3) a molecular analysis of the same

37 taxa of the previous analysis and 778 molecular characters; and (4) a

combined analysis including 37 taxa and 825 characters (the combined 778

molecular and 47 morphological characters).










The first analysis was designed to provide a starting point by evaluating

relationships based solely on molecular data from all sequences available for this

study. The second analysis was designed in the same way with the idea of

evaluating relationships, but using morphological characters present in selected

taxa. The third analysis was based on the same selected taxa of the previous

one but including only molecular data and was aimed at obtaining phylogenetic

relations that may or may not be congruent with the morphological analysis. The

fourth analysis represents the principal goal of this study--a morphological

character-molecular data combined analysis. The main purpose was to evaluate

the effects that both data sets may have on the outcome of a phylogenetic

analysis. Many authors recently have stressed the importance of combining

independent sources of data and offered a clear rationale for merging different

data sets (Graham et al. 1998; Huelsenbeck et al. 1996; Sanderson and

Donoghue 1989; Soltis et al. 1998; Wiens 1998).

Although one only needs one outgroup to root a phylogenetic tree (Nixon

and Carpenter 1993), two or more outgroups are frequently used in cladistic

analyses to test for ingroup monophyly and to establish character polarity

(Maddison et al. 1984; Simmons and Geisler 1998). For outgroups I chose

Uroderma bilobatum, Vampyressa pusilla, and Centurio senex within the tribe

Stenodermatini, and for representatives from another subfamily, Carolliinae, I

sequenced Carolia perspicillata and scored and measured for inclusion in the

morphological data set Carollia brevicauda, Carollia perspicillata, and Carollia

subrufa.








26
PAUP* version 4.Ob4a (Swofford 2000) was used for cladistic parsimony

analyses with the following search strategies: each matrix (three separate and

the combined morphology and DNA) was subjected to 1,000 replicates of

random taxon entry additions, with the multiple trees (MULTREES) option, using

subtree pruning and regrafting (SPR) swapping, holding ten trees per replicate to

minimize time spent swapping on suboptimal islands. The shortest trees from

this search were used as starting trees and up to 10,000 trees (tree limit due to

computer memory limitations) were swapped to completion using SPR. All Fitch

(equally weighted) trees were saved to file.

Because successive weighting de-emphasizes patterns created by

homoplasious sites, and thus gives weight to sites that are more consistent (see

Lled6 et al. 1998 for convincing reasons for using successive weighting), I

applied reweighting strategies using all shortest trees, with a base weight of 1,

based on rescaled consistency index (RC) and best fit on any of the trees using

the menu command in PAUP*. Each round of analyses consisted of 10

replicates of random taxon entry, MULTREES on, SPR swapping, holding 10

trees per replicate. The shortest trees collected in each of these ten replicates

were used as starting trees to collect all the shortest successively weighted (SW)

trees, which were swapped to completion. Reweighting schemes were repeated

until tree length, consistency index (Cl), and retention index (RI) remained the

same in two successive rounds as calculated in PAUP*. Other weighting

schemes were also used to estimate phylogenetic relationships among taxa

(weighting transversions over transitions by 1:0, 3:1, 5:1, and 10:1) and results

were compared against hypotheses generated with the SW scheme. The










confidence of each clade was evaluated through 1,000 replicates of heuristic

bootstrap analysis (Felsenstein 1985) on both SW and Fitch trees using SPR

swapping, MULTREES on, and holding only 10 trees per replicate. Support of

clades also was evaluated using the total support or decay index (Bremer 1988,

1994). To measure tree stability in terms of supported resolution, each matrix

was subjected to 1,000 replicates of random taxon entry additions, MULTREES

on, and SPR swapping. Tree scores were recorded and trees were saved (as

rooted trees) to file. Autodecay version 4.01 (Eriksson 1998) was run under the

following PAUP* search parameters as follow: addseq=random, nrep=100,

rseed=1, Nchuck=5, ChuckScore=100. The resulting file was executed in PAUP*

and posteriorly the decay value was extracted for each node in Autodecay. Final

decay values were viewed, edited, saved, and printed with TreeView version 1.5

(Page 1996).

Based on the assumption that the trees from the combined (SW) analyses

are the best estimate of the phylogeny (in part due to higher overall support

values), it will make sense in a study like this one to follow and determine the

patterns of character evolution of each meaningful qualitative multistate

osteological character (estimated using MacClade version 3.08a, Maddison and

Maddison 1999) on one of these trees, rather than using trees produced from

analyses of separate data sets. However, it is important to mention here that

although I combined molecular and morphological data, I did not find a topology

where relationships among different species (or species groups) of Sturnira were

highly supported and resolved at all levels. Therefore, the lack of a unique, final,

and meaningful tree on which to base my analysis of character evolution at this










stage would force me to make only a preliminary interpretation at best, and my

results should be regarded as such.

Quantitative pairwise comparisons among all taxa were made for the

cytochrome b gene. Those comparisons included percentage of sequence

divergence within and between taxa. Sequence-divergence values were

obtained using the two-parameter model of Kimura (1980) using PAUP* version

4.0b4a (Swofford 2000). Evolutionary relationships among haplotypes were

assessed with a neighbor-joining tree (Saitou and Nei 1987) using PAUP* with

the option rates for variable sites at a gamma 0.5 and ties broken randomly.

Support for nodes in dendrograms was assessed with bootstrap resampling of

the neighbor-joining tree using 100 replicates.

Using the single most parsimonious tree from a cladistic analysis of a

combined cytochrome b gene and a qualitative multistate osteological data set

(successively weighted parsimony) of 37 individuals representing Sturnira

species and related outgroup taxa, I investigated the patterns of character

evolution of each qualitative character using MacClade version 3.08a (Maddison

and Maddison 1999). Data set from the character matrix (Appendix 5) containing

all character states present in four outgroup species and sixteen ingroup species

of the genus Sturnira was loaded into MacClade and invoking "GO TO TREE

WINDOW," "GET TREE FILE," and "TRACE CHARACTER" commands it was

possible to trace each character mapped on the SW-combined tree. A brief

description of each meaninful tree is provided in the Results section.




























Figure 4. Abbreviations and the recording points of 18 cranial quantitative
measurements (see Appendix 2 for description of each continuous morphological
character) taken from skulls of 15 species (recognized here) of bats of the genus
Stumira, A) lateral view of skull, B) lateral view of mandible, C) dorsal view of
mandible, D) ventral view of skull, E) dorsal view of skull, F) detail of ventral view
of right upper molars.










A
.......... .. .. .








CIL:
GLS








A
.S S
I S


S S





Cl l


B








LD

BEM









. --... -.......... .. .. . a



:BFM
S 5








55m A IF




























Figure 5. Abbreviations and the recording points of external facial and body
measurements referred to in text (see Appendix 2 for description of each
continuous morphological character) taken from alcoholic specimens and study
skins of 15 species (recognized here) of bats of the genus Sturnira. A) nose-leaf
of Stumira tildae, B) left wing and body of Stumira erythromos.








32









A B
BOS
....... ......... ......... .......... .........










I '2R


w

INW
................................................
BOH




























Figure 6. Terminology used in this work refering descriptions on topography of
the molar tooth. A) upper left molars, B) lower left first molar. Ml: first upper
molar tooth, M2: second upper molar tooth, ml: first lower molar tooth.










LABIAL


metacone



j


ANTERIOR


POSTERIOR





d A


/ \
protocone hypocone


LINGUAL


paraconid


LINGUAL


metaconid

i


entoconid


protoconid

ANTERIOR


POSTERIOR


B


MB/AL hypoconid


hypoconid


LABIAL













CHAPTER 3
RESULTS


For each separate and combined cladistic analysis, Table 1 presents the

number of aligned positions in the matrix, the number of variable sites, the

number of phylogenetically informative sites, and the percentage of sites that are

variable. For each portion of every analysis (Fitch and/or successively

weighted), I report the number of trees, number of steps, consistency index (Cl),

retention index (RI), and the average number of changes per variable site (tree

length divided by the number of variable sites). The following descriptions are

based on combinations of both equally weighted and successively weighted (SW)

parsimony analyses. The topologies obtained using different weighting schemes

resulted in very similar hypothesized relationships among taxa, and therefore I

derived my descriptions using only SW schemes. When mentioning polytomies

on the description of my results (and later on the discussion) I will be referring to

real polytomies or else to weakly supported dichotomies that I purposely

colapsed into a polytomy in an attempt to be more conservative.


All Available DNA Sequences and Selected Specimens


The Large Data Set: 133 Specimens (4 Outgroup Species and 129 Sturnira
Specimens)

After trimming ambiguous sequences at each end of the DNA fragment, a

total of 778 bp was usable and compared for all taxa. Of those 778 positions,








36

299 (38%) represent variable sites and within that, 231 sites (30% of all positions

or 77% of all variable sites) were phylogenetically informative. This matrix does

not contain any indels. The following sixteen species recognized here were

represented by different number of individuals per species (given within

parentheses): S. aratathomasi (1), S. bidens (2), S. bogotensis (3), S.

erythromos (9), S. hondurensis (20), S. lilium (48), S. ludovici (2), S. luisi (4), S.

magna (7), S. mordax (1), S. nana (3), S. oporaphi/um (6), S. thomasi (3), S.

tildae (16), and two undescribed species made available to me through the

courtesy of Timothy J. McCarthy and Luis Albuja (here denoted as S. sp. A (2)

and S. sp. B (2), respectively).

Equally weighted heuristic searches (Fitch criterion) yielded 910 equally

parsimonious trees of 1021 steps (Cl: 0.41 and RI: 0.85). Figure 7 shows one of

those 910 equally parsimonious trees. Terminal branches that defined species

such as S. bogotensis, S. magna, S. sp. B, S. erythromos, S. tildae, S. sp. A,

southeastern and northern neotropical specimens of S. lilium, Ecuadorean,

Panam.-Costa Rican, and Guatemala-Honduran specimens of S. ludovici-

hondurensis, S. bidens, and S. nana received very strong support (bootstrap of

100 and two-digits decay values). Stumira oporaphilum, S. hondurensis, the

entire northern region clade S. Iudovici-hondurensis, S. mordax, and

northeastern South American and Panam6-Costa Rican specimens of S. lilium,

S. luisi, S. thomasi and S. aratathomasi received only moderate to no support,

for many different reasons, which will be explained below. Clearly, S. bidens and

S. nana stand alone as the sister group of the remaining species of Sturnira.

Details of the diversity of sequences and specimens used in this analysis are










depicted in Figures 8 through 14. From these seven trees one can notice the

occurrence of a large polytomy (indicated by asterisks) among species (deep

level), which may reflect the inability of the cytochrome b gene to resolve

phylogenetic relationships between species other than S. bidens and S. nana.

Figure 8 shows a detailed account of each specimen used in this analysis

for S. magna, S. bogotensis, and S. mordax. More individual tissue samples of

S. mordax were not available for this analyses due to permit-related problems.

The position of the "mordaX' branch is not completely resolved. All specimens of

S. bogotensis form a monophyletic group as well as all of S. magna. However,

there are two major groups within the "magna clade". The individual of S. magna

from Pern (on the top line of the clade) is from southern Per, whereas the other

S. magna from Peru is in fact from the amazonian border between PerO and

Ecuador. This may indicate geographical partitioning of populations.

Several conclusions can be drawn from Figure 9. All specimens of S.

oporaphilum form a monophyletic clade highly supported by bootstrap and decay

values. Some degree of geographical partitioning is evident. The same is true

for all specimens of S. ludovici from Ecuador (country of the type locality, see

Anthony 1924). The node that groups together S. oporaphilum and S. ludovici

from Ecuador is moderately supported. The remainding 20 specimens

representing S. ludovici S. hondurensis from Guatemala, El Salvador,

Honduras, Panama, and Costa Rica clustered into two major clades. The node

that groups together these two clades is well supported. One of those clades

(specimens from Panamd and Costa Rica) is a well defined, highly supported

clade. The other, including all specimens from Guatemala, El Salvador, and









Honduras is also well defined, highly supported, and includes two separated

clades, one well defined and supported (bootstrap value of 99 and decay value of

5) and the other moderately supported (bootstrap value of 71 and decay value of

2). A common node between the clade formed by S. oporaphilum and S. ludovici

from Ecuador and all Central American specimens is moderately supported.

Figure 10 shows all specimens used for S. sp. B and S. erythromos. Both

are represented by well defined, monophyletic clades related to one another.

The position of both clades compared with other clades is not resolved

(polytomy).

Figure 11 shows the clade formed by specimens of S. tildae with a series

of nested internal clades, with three main groups: a terminal clade represented

by 12 specimens from Bolivia, Ecuador, Venezuela, Guyana, Suriname and

French Guiana, a sister group to the previous clade represented by three

individual sequences from Trinidad and Tobago, and one individual from Brazil

standing as a sister clade of the two most internal clades aforementioned.

Sturnira tildae is part of an unresolved polytomy formed by all previously

mentioned clades representing S. magna, S. bogotensis, S. mordax, S.

oporaphilum, all S. ludovici, S. sp. A and B, and S. erythromos. However,

internally each of those clades display high bootstrap support and decay values.

Figures 12 and 13 depict a group of highly supported clades represented

by a complex of species (S. thomasi, S. luisi, S. lilium, and S. aratathomas). Its

relative position within the genus Sturnira is not well resolved and the monophyly

of the group is only moderately supported. In fact, Figure 12 displays several

subclades with surprisingly high support values. The entire group of clades










depicted in the upper portion of this figure is branching off a node with a

bootstrap value of 92, which contains five internal nodes with bootstrap values

above 85%. Although weakly supported, there is another cluster of populations

(lower clade) that are intriguing because they may the sister group to this

previously mentioned group of clades that it became more divergent

morphologically (perhaps an ancestor-descendent line). In Figure 13, two clades

emerge with high resolution and support values, those represented by specimens

of S. lilium from Brazil, Bolivia, and Paraguay (type locality for S. lilium lilium),

and those specimens of S. filium from Costa Rica, Honduras, El Salvador,

Guatemala, and M6xico (type locality for S. lilium parvides). The position of the

specimen representing S. aratathomasi is not resolved but is part of this complex

of species (S. thomasi, S. luisi, S. lilium, and S. aratathomas).

The clades of S. nana and S. bidens are shown in Figure 14. All

specimens belonging to the subgenus Sturnira are the sister group of S. nana

and all subgenus Sturnira and S. nana are supported as the sister group of S.

bidens. The node between the entire genus Stumira and the outgroup species is

relatively well supported.

By performing successively weighted (SW) analyses I expected to

decrease the effects of highly homoplasious characters. I also expected the

number of trees to decrease "because those created by characters that changed

frequently are eliminated as less parsimonious" (Lied6 et al, 1998: 24). However,

the SW heuristic search produced 1291 equally parsimonious trees of 337 steps

(CI: 0.62 and RI: 0.90). Figure 15 shows one of those 1291 equally

parsimonious trees. The topology of the Fitch and SW trees was slightly










different. While on the Fitch trees the entire subgenus Sturnira was part of a

large unresolved polytomy sister to S. nana, on these SW trees (shown in Figure

15) the large polytomy was partially resolved. Within the subgenus Sturnira the

complex of species S. thomasi, S. luisi, S. lilium, and S. aratathomasi form an

unresolved polytomy that is sister to another polytomy composed of specimens

from S. magna, S. bogotensis, S. sp. A and B, S. oporaphilum, S. ludovici-

hondurensis, S. erythromos, S. mordax and S. tildae. Within the most internal

polytomy, S. oporaphilum and S. ludovici from Ecuador are sister taxa. Stumira

ludovici-hondurensis from Guatemala, El Salvador, and Honduras are sister taxa

of S. ludovici from Panam-Costa Rica. And S. oporaphilum-S. ludovici from

Ecuador form a sister clade of S. ludovici-hondurensis from Guatemala, El

Salvador, and Honduras, and S. ludovici from PanamA-Costa Rica. Twelve

nodes in Figure 15 had improved resolution and support after SW analysis.

The Small Data Set: 37 Specimens (4 Outgroup Species and 33 Sturnira

Specimens)

The DNA data set

From a total of 778 positions in the matrix, 266 (34%) represent variable

sites and within that total, 213 (27% of all positions or 80% of all variable sites)

were phylogenetically informative. This matrix is a subset of the previous DNA

data set (with 133 samples) generated with the sole purpose to have a

comparable data matrix with the same number of individuals used on the

qualitative multistate osteological analysis. With the exception of S. sp. B, which

is represented by one specimen, all other species of Sturnira are represented

here (and in the subsequent data sets of this section) by two individuals.









Equally weighted heuristic search (Fitch criterion) yielded two equally

parsimonious trees (or two equally parsimonious hypotheses) of 782 steps (CI:

0.48 and RI: 0.64) shown in Figures 16 and 17. Figure 16 shows very strong

support (bootstrap and decay values) for terminal branches of S. aratathomasi,

S. luisi, S. thomasi, S. lilium, S. magna, S. bogotensis, S. erythromos, S. mordax,

S. sp. A, S. tildae, S. ludovici, S. oporaphilum, S. hondurensis, S. nana, and S.

bidens. The "S. lilium S. aratathomasi complex" remains part of a large

unresolved polytomy at a deep level between clades. The "S. ludovici complex"

however, forms a moderately supported clade (73% Fitch bootstrap) including S.

ludovici, S. oporaphilum, and S. hondurensis. The clade formed by S. magna

and S. bogotensis is weakly supported as well as the clade formed by specimens

of S. ilium from North and South America. Sturnira nana forms a sister clade to

all species of the subgenus Sturnira and this group is a sister group of S. bidens.

The entire genus Stumira seems to be monophyletic with good support (87%

Fitch bootstrap and a decay value of 6). The position of S. sp. B is not stable or

resolved. Figure 17 does not provide notable differences in support or resolution.

A successively weighted heuristic analysis did produce a topology only

slightly different from the previous equally weighted Fitch trees (Figures 16 and

17). It produced a single most parsimonious tree of 260 steps (CI: 0.70 and RI:

0.75) shown in Figure 18. The bootstrap analysis of this SW data set yielded

higher level of support (75% of all clades in bootstrap consensus had values

larger than 85%) than did the same unweighted data set or the SW DNA data set

with 133 samples. It did partially resolve the polytomy that included the entire

subgenus Stumira into three main clusters of species. However, the










phylogenetic position of those three groups is not resolved. One of them, the

"lilium-aratathomasi complex" displays weak bootstrap support. Another, a group

of species including S. magna, S. bogotensis, S. erythromos (these three species

form a weakly supported clade), S. mordax, S. sp. A (these two species also

form a weakly supported clade), S. sp. B, and S. tildae display weak bootstrap

support as well. The third group, the "ludovici complex" shows high bootstrap

values across all nodes. It improved the resolution (with weak bootstrap support)

of five nodes and weakly resolved five additional nodes (see arrowheads on

Figure 18).


The morphological data set

This matrix has only 47 characters; however, all 47 sites are variable

(100% of the total) and 46 (98% of the total) represent phylogenetically

informative sites. As in the previous data set, with the exception of S. sp. B,

which is represented by one specimen, all other putative species of Sturnira are

represented here by two individuals, which indicates the number of individuals

per species needed to show all possible character states displayed in one

species.

Equally weighted heuristic search (Fitch criterion) yielded two equally

parsimonious trees (or two equally parsimonious hypotheses) of 257 steps (Cl:

0.28 and RI: 0.62). Those two slightly different topologies (or hypotheses)

resulting from this analysis are shown in Figures 19 and 20. Overall support is

weak and only S. aratathomasi, S. magna, S. sp. A, S. nana, S. bidens, S.

oporaphylum, S. hondurensis, S. thomasi, S. luisi and S. lilium have weak to well










supported clades. In both hypotheses Uroderma bilobatum and Vampyressa

pusilla clustered together, and with Carollia spp. are integrated with the large

unresolved polytomy that includes all specimens of the genus Sturnira.

A successively weighted heuristic analysis did produce a topology slightly

different from the previous equally weighted Fitch trees depicted in Figures 19

and 20. It produced a single most parsimonious tree of 42 steps (CI: 0.39 and

RI: 0.74) that is shown in Figure 21. Within the general large polytomy, this

analysis indicates an internal cluster (weakly supported) composed of S. nana, S.

bidens (these two species form a weakly supported clade), S. oporaphilum, S.

bogotensis, S. erythromos, S. sp. A, S. mordax, S. hondurensis, S. ludovici, and

S. magna. The remainder of the tree is similar to the previous Figures 19 and 20

with the exception of a weakly supported branch sister to the entire polytomy

represented by an individual of S. lilium from El Salvador. Centurio senex is

sister taxon to the rest of the specimens included in this analysis. In nine

instances the bootstrap analysis of this SW data set yielded a higher level of

support than did the same equally weighted data set. Overall, it did not produce

a highly supported hypothesis (only a third of the clades have more than 85%

bootstrap support) and it did not resolve most of the polytomies.


Combined data--DNA and morphological data sets

The combined data matrix has a total of 825 positions with 313 (38%)

variable sites and within those variable sites, 259 (31 % of all positions or 83%

within variable sites) were phylogenetically informative. This matrix was

generated by running in PAUP* a copy of the previous DNA data set (with 37








44

taxa) and a copy of the qualitative multistate osteological data set (also with the

same 37 taxa).

Equally weighted heuristic search of this combined matrix under Fitch

criterion yielded four equally parsimonious trees of 1078 steps (CI: 0.41 and RI:

0.61). These four trees are shown in Figures 22, 23, 24 and 25. Overall, all

topologies are highly congruent, with similar high support values for almost all

terminal branches. In all four hypotheses, strong support is shown for the entire

Sturnira clade as a monophyletic group clearly separated from all species of the

outgroup clade. The outgroup formed by Vampyressa pusilla, Uroderma

bilobatum, Centurio senex, and Carollia perspicillata is well defined and highly

supported too. An internal, large polytomy is also present in all trees and the

phylogenetic positions of clades along 5 nodes (see arrowheads pointing at those

nodes on Figures 22, 23, 24 and 25) are not resolved.

All trees have a weakly supported but consistent clade constituted by S.

nana and S. bidens, sister group to the rest of the species of Stumira. Within

subgenus Stumira is a well supported clade of S. ludovici, S. oporaphilum, and S.

hondurensis. All trees displayed a poorly resolved clade constituted by highly

supported internal clades that comprised S. lilium (specimens from North and

South America), S. luisi and S. thomasi, and S. aratathomasi. The four last

mentioned species in turn made a sister clade of S. tildae, a clade that also

shows weak support values on the common node. In all four trees, S.

erythromos and S. bogotensis are together in a weakly supported clade. All

species of the subgenus Sturnira always form a clade, which shows little

resolution basally.









A successively weighted heuristic analysis produced a single most

parsimonious tree of 295 steps (CI: 0.65 and RI: 0.72) shown in Figure 26. Its

topology is similar in many ways to the four previous trees, but the bootstrap

analysis of this SW combined data set yielded much higher level of support (76%

of all clades in bootstrap consensus with values larger than 85%) than did any of

the single data sets or the four previous equally weighted Fitch trees. Seven

nodes showed improved support values and the polytomy on the clade formed by

S. mordax and S. sp. A (seen on Figures 22, 23, 24, and 25) is resolved,

although with weak support values. Sturnira tildae still is the sister taxon of the

aratathomasi-lilium-luisi-thomasi group but without statistical support. The final

result of a combined, successively weighted data set is a highly supported set of

clades with improved resolution at all levels within Sturnira and even between

outgroup species and Sturnira clades.


Pairwise Distances

Mean percentage of sequence divergence for all pairwise comparisons

revealed a wide range of nucleotide divergence values among representatives of

the genus Sturnira (Table 2). Mean sequence divergence between subgenera

was 11.7% with the greatest value of 16.4% between S. bidens and S. magna

and the lowest value of 9.4% between S. nana and S. sp. B. Mean sequence

divergence within the subgenus Corvira was 9.7%. The subgenus Sturnira was

divided in the two clades defined by de la Torre (1959) and Koopman (1994) as

species with "well defined lingual cusps on lower molars" and species with

"vestigial or absence of vertical notches that defined lingual cusps" (see Figure








46

74). Mean sequence divergence within species with "well defined lingual cusps

on lower molars" was 5.5% with the greatest value of 11.2% between S. tildae

and S. lilium from M6xico and the lowest value of 0.1 % between S. lilium from

Pern and S. lilium from Suriname. Mean sequence divergence within species

with "vestigial or absence of vertical notches that defined lingual cusps" was

7.7% with the greatest value of 11.2% between S. erythromos and S. ludovici

from Guatemala and the lowest value of 0.4% between S. ludovicifrom

Guatemala and S. ludovici from Honduras. Mean sequence divergence between

species of the subgenus Stumira (species with "well defined lingual cusps on

lower molars" and species with "vestigial or absence of vertical notchs that

defined lingual cusps") was 8.9% with the greatest value of 13.5% between S.

magna and S. lilium from M6xico and the lowest value of 6.0% between S.

mordax and S. Ilum from Suriname.


Patterns of Character Evolution: Selected Cases

Each of the 47 qualitative multistate osteological characters was plotted

and investigated on the single most parsimonious tree (295 steps, Cl: 0.65, RI:

0.72) from a cladistic analysis of the combined (DNA and morphology) data set

(successively weighted parsimony) of 37 individuals representing Sturnira

species and related outgroup taxa. Trying only to highlight synapomorphies that

may assist in the definition or characterization of one or a group of species, three

trees suggest evidence to separate (or link) elements of the outgroup to elements

of the ingroup. Figure 29 shows Centurio senex as the only taxon having a flat

distal edge on the upper inner incisors. The state "pointed" seems to be the










plesiomorphic condition for the entire clade and the rounded state evolved at

least five times within Stumira. Figure 31 depicts character number five (see

Appendix 4 for definition), in which state "2 lobes with secondary postero-external

basal cusp" seems to be an autapomorphy for the "Corvird' clade (S. nana and

S. bidens). "Two lobes" seem to represent the plesiomorphic condition, and "no

lobes (or pointed distal edge of upper inner incisors)" is expressed once in the

outgroup (Carollia spp.) and twice within the ingroup (S. aratathomasi and S.

hondurensis). Figure 43 displays all states for character number 17, in which the

condition "horizontal surface of ml is tilted towards its labial side" is only

expressed in S. sp. B. The same surface tilted towards its lingual side seems to

be the plesiomorphic condition for this character and having a horizontal surface

appears several times scattered throughout Sturnira species, but it is well

established on individuals of the three species forming the "ludovici clade."

Mapping the length of the forearm on the same tree (Figure 46) reveals

that the ancestral form of the entire clade was of medium size and that based

only on size, S. nana is easily identifiable as the smallest taxon of the Stumira

group, and S. aratathomasi and S. magna are the largest ones. Character

number 22 distinguishes S. magna from all other species (Figure 48). All other

Stumira species seems to share the plesiomorphic condition for this character,

with the exception of S. sp. B, which shares equal relative length of M1 and M2

with three of the four taxa of the outgroup. Only S. aratathomasi and S. sp. A

display a relatively long rostrum (autapomorphic for character number 33) within

Sturnira species (Figure 59). Carollia spp. are the only taxa within the outgroup

displaying this state. A relatively short rostrum seems to be the ancentral








48

condition. Figure 60 shows that a squarish condition of the nasal aperture may

be plesiomorphic and that S. oporaphilum displays a trapezoidall" condition and

S. aratathomasian elongated one. In Figure 67 we can see another

autapomorphy useful to separate one of the outgroups (Carollia spp.) from all

others. The plesiomorphic condition for this character seems to be a smaller

relative size of M2 when compare with M1. Another apomorphy (in this case for

S. aratathomas) is shown in Figure 68. No occipital crest seems to be the

ancestral condition for the entire clade. All individuals of S. luisi scored for

character number 46 (Figure 72) presented the state "presence of a divided cusp

on metaconid of m2." Some individuals of S. lilium, S. tildae, and S. hondurensis

display the same condition as well. No division on the cusp of m2 seems to be

the ancestral condition for the character. The distribution of other characters are

shown in Figures 27 through 73.











Table 1. Values and statistics from PAUP* analyses of separate and combined
data matrices. Cl: consistency index, RI: retention index, SW: successively
weighted.

DNA Morphology DNA Combined


Number of specimens

Number of included
positions in matrix

Number of variable sitesa

Number of phylogenetically
informative sites


Number of trees (Fitch)

Number of steps

Cl

RI

Average number of changes
per variable sited

Number of trees (SW)

Number of steps (SW)

Cl (SW)

RI (SW)

Number of clades in
bootstrap consensus
with >85% supporte


133


778


299(38%) 47(100%)


231
(77b, 30c)

910

1021

0.41

0.85


3.4


1291

337

0.62

0.90


46
(98, 98)


?57


778


37


825


266(34%) 313(38%)


213
(80, 27)


782


0.28

0.62


0.48

0.64


260


0.39

0.74


0.70

0.75


34 (51%) 5 (33%)


259
(83, 31)

4

1078

0.41

0.61


3.4

1

295

0.65

0.72


24 (75%) 22 (76%)


a Number of parsimony-informative characters + number of parsimony-
uninformative characters, b Percentage within variable sites, c Percentage of all
positions in matrix, d Number of steps/number of variable sites, Calculated from
SW bootstraped trees.









50









Table 2. Percentage of sequence divergence for all pairwise comparisons using the Kimura 2-parameter algorithm (1980) in PAUP* (Kimura 2-parameter

distance matrix, rates assumed to follow gamma distribution with shape parameter = 0.5). Haplotype numbers follow: 1 Centurio senex, 2 Carolia

perspicillata, 3 Uroderma bilobatum, 4 Vampyressa pusilla, 5 (S.)tumira aratathomasi 1, 6 S. aratathomasi 2, 7 S. magna 1, 8 S. magna 2, 9 S. ludovici 1, 10

S. ludovici 2, 11 S. hondurensis 1, 12 S. hondurensis 2, 13 S. lilium 1, 14 S. lilium 2, 15 S. lilium 39, 16 S. li/um_ 1, 17 S. luisi 1, 18 S. luisi 2, 19 S. thomasi 1,

20 S. thomasi 2, 21 S. mordax 1, 22 S. mordax 2, 23 S. sp. A 1, 24 S. sp. A 2, 25 S. sp. B, 26 S. erythromos 1, 27 S. erythromos 2, 28 S. bogotensis 1, 29 S.

bogotensis 2, 30 S. oporaphilum 1, 31 S. oporaphilum 2, 32 S. tildae 1, 33 S. tildae 2, 34 S. nana 1, 35 S. nana 2, 36 S. bidens 1, 37 S. bidens 2.



1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37


24.9 -
24.0 24.0 -
23.7 27.0 18.4 -
21.7 22.6 21.2 24.1
21.7 22.6 21.2 24.1
20.9 23.5 21.5 21.6 1
20.7 21.0 21.5 22.4 1
17.9 20.4 18.9 21.2 1
17.8 20.1 19.2 21.5 1
19.5 20.7 21.5 19.5 1
19.6 20.9 22.4 21.5 1
22.0 20.1 20.3 22.7 1
22.0 22.4 21.5 24.5
22.6 19.6 19.8 23.0
22.1 21.8 21.2 24.7
21.6 21.2 20.3 22.1
21.6 21.2 20.3 22.1
21.6 21.9 20.1 21.9
21.6 21.9 20.1 21.9
18.9 19.7 20.2 22.2
18.9 19.7 20.2 22.2
20.8 20.1 20.8 22.0
20.4 19.6 19.7 21.2
18.8 21.0 20.0 20.9
23.9 22.0 20.5 22.0
23.5 22.5 19.6 20.4 1
20.5 22.8 19.7 23.8 1
20.0 21.8 19.7 23.3 1
19.3 20.5 19.7 22.5 1
18.6 20.3 19.7 21.9 1
21.3 22.5 21.4 22.6
21.8 21.3 20.9 22.0
21.9 18.6 19.7 17.7 1
22.3 19.6 19.5 19.0 1
20.1 19.4 19.4 20.9 1
23.5 20.1 20.8 21.7 1


4.9 -
4.5 0.6
5.5 6.7
6.7 6.8
7.0 7.4
6.9 7.8
6.2 7.4
6.5 8.4
6.3 8.2
6.3 7.3
6.5 7.3
10.2 9.7
9.8 10.3
11.1 11.4
12.0 12.4


7.3
7.1 1.0 -
9.4 9.5 8.7 -
9.6 9.7 8.9 1.2 -
8.0 8.1 7.2 9.5 8.9 -
8.2 8.8 7.8 9.3 8.7 0.6 -
11.2 12.8 12.5 10.7 10.7 10.1 10.1 -
10.4 12.8 11.8 10.8 10.6 10.3 9.7 0.1
13.0 12.7 11.8 12.7 12.1 13.3 12.7 9.8 9.5 -
14.3 13.5 12.6 13.7 13.0 13.7 13.5 9.7 9.8 2.3 -




























Figure 7. One of 910 equally parsimonious trees of 1021 steps (Cl: 0.41, RI:
0.85) from a cladistic analysis of cytochrome b gene data set (Fitch parsimony) of
133 individuals representing Sturnira species and related outgroup taxa. "S."
before each specific epithet stands for Stumira. Bootstrap percentage (of 1000
replicates) and decay index (Bremer support) values are listed above each
branch. Values below branches are Fitch lengths. Asterisks indicate polytomies
(unresolved nodes). Values within parentheses are number of individuals
representing each species. Bars on the right indicate the figure number that
shows detailed information on each species clade displaying all available
individuals.










73/3 96/10


bilobatum
impyressa pusilla
,;Carollia perspicillata


Honduras (8)


Costa Rica (4)


.Iuisi Panama (14)


lilium paulsoni


100/12 S.Iilium Paraguay (10)

61/2
7
100/14 S.lilium Mexico (13)
15 */0
........ S.aratathomasi (1)


I0L.

0
















-





V0)


0)
e
oI.







0)






0::





























Figure 8. Numerical values, abbreviations, and symbols are as indicated in
figure 7. Detailed information of phylogenetic relationships of all specimens
available for S. magna and S. bogotensis.










96/10
20
Centurio senex


*/0
4



*/0
4
4


Uroderma bilobatum
Vampyressa pusilla
Carollia perspicillata


70/3
11


16 S.magna Peru
S.magna Ecuador
100115 S.magna Ecuador
20 56 S.magna Ecuador
80 S.magna Ecuador

1 4 S.magna Peru
S.magna Ecuador

100115 8 S.bogotensis Peru
2 S.bogotensis Peru
*/ ra S.bogotensis Peru
*" S.mordax
'74/2 1 8 S.oporaphilum (6)
6 100/13 S.ludovici Ecuador (2)
16
99/5 S.hondurensis Honduras (8)
6


8 8/3


100/11
12
71/2 S.ludovici Guatemala (8)
36


100/9 S.ludovici Costa Rica (4)
100/14
15 S.sp. B (2)


4F 100/12
20 S.erythromos (9)

100/17 S.tildae (16)
100/14
17 S.sp. A (2)
94 S.thomasi Guadeloupe (5)

55/1
92/0 61/1:

9206 1/2 S.Iuisi Panama (14)

S.lilium Suriname (13)


100/12
19


S.lilium Paraguay (10)


52/2
5



100/15
100/12 26
22 2 S.bidens (2)


61/2
7
100/14 S.lilium Mexico (13)
15 */0 S.aratathomasi (1)
S.nana (3) 32


83/51


73/3
29


72/2
14


56/5
12






























Figure 9. Numerical values, abbreviations, and symbols are as indicated in
figure 7. Detailed information of phylogenetic relationships of all specimens
available for S. oporaphilum, S. ludovici, and S. hondurensis.










73/3
29


*/0
4




1
4


7013
11


100/14
15
100/12
20
100/17
9A


96/10
20t Uroderma bilobatum
Centurio senex Vampyressa pusilla
Carollia perspicillata 100/15 Smagna (7)
*/1 20
100/15 S.bogotensis (3)
21 oprplu
S.mordax (1) 4 S.oporaphilu
S.oporaphilua
9616 4 S.oporaphfl
8 2 2 S.oporaphi
74/2 2 9 S.oporaph
6 2 S.oporaphilum Per
100113 S.ludovici E
*/0 16 Sbudovici Ecu;
A


S.ludovici Guatemala
S.Iudovici Guatemala
S.Iudovici Guatemala
99/5 S.ludovici El Salvador
6 S.hondurensis Honduras
100/11 S.hondurensis Honduras
12 S.ludovici El Salvador
3 S.hondurensis Honduras
2 S.ludovici Guatemala
S.Iudovici Guatemala
2 S.ludovici Guatemala
S.hondurensis Honduras
2 2S.hondurensis Honduras
71/24 2 S.ludovici El Salvador
3 4S.ludovici Guatemala
6 3 S.Iudovici Guatemala
2S ludovici Costa Rica
10019 1 -S.ludovici Costa Rica
11 S.Iudovici Panama
S.sp. B (2) S.ludovici Costa Rica

S.erythromos (9)

S.tildae (16)


100/14 A
17 S.sp.A(2)
55/1 94/4S.thomasi Guadeloupe (5)

92/0 3 6/1S.luisi Panama (14)
6


100/15
100/12 26 S.bidens(2)
22


61/2


S.lilium Suriname (13)
100/12 S.lilium Paraguay (10)
19


150/0 S lilium Mexico (13)
15 () 32 S.aratathomasi (1)
S.nana (3) 32


83/5,


Peru
n Bolivia
rm Bolivia
Hlum Bolivia
Hum Bolivia

cuador
ador


72/2
14


56/5
12



19


52/2
5


"lt


8f






























Figure 10. Numerical values, abbreviations, and symbols are as indicated in
figure 7. Detailed information of phylogenetic relationships of all specimens
available for S. sp. B and S. erythromos.










96/10
20,


Centurio senex


70/3
11


73/3
29


88/3
6


71/2 S.hondurensis Guatemala (8)
3
100/9
1 S hondurensis Costa Rica (4)
11


52/2
5


100/15
100/12 26 S.bidens(2)
22


100/12 S.lilium Paraguay (10)
61/2 19
7 100/14 S.lilium Mexico (13)
15 */0 S.aratathomasi (1)
S.nana (3) 32


Uroderma bilobatum
Vampyressa pusilla
100/15 Carollia perspicillata
20 S.magna (7)
100/15
21 S.bogotensis (3)

S.mordax (1)

74/2 96/6 S.oporaphilum (6)
6 100/13 S.ludovici Ecuador (2)
16
99/5 S.hondurensis Honduras (8)
6
100/11
12


83/5
4


100/14 -lS. sp. B Ecuador
15 S. sp. B Ecuador
4
5 S.erythromos Peru
15 S.erythromos Peru
1 S.erythromos Peru
2 3S.erythromos Peru
1 2 S.erythromos Argentina

100112 S.erythromos Bolivia
20 1 eythromos Bolivia
S.erythromos Bolivia
S.erythromos Peru


100/17 S.tildae (16)
100/14 24
17 S.sp. A (2)

55/1 945S.thomasi Guadeloupe (5)
92/03, 61/1S.luisi Panama (14)
6
"14 S.lilium Suriname (13)


72/2
14


56/5
12






























Figure 11. Numerical values, abbreviations, and symbols are as indicated in
figure 7. Detailed information of phylogenetic relationships of all specimens
available for S. tildae and S. sp. A.










96/10
20,


Centurio


10011
24


72/2
14


10011 ln


83/5


17 2 S. sp. A Ecuador

55/1 94/4 S.thomasi Guadeloupe (5)
92/03 6/ S.luisi Panama (14)
6
S.lilium Suriname (13)


56/5
12


52/2
5

100/15
100/12 26 S.bidens (2)
22


61/2


100/12 S
19 S.lilium Paraguay (10)


7 100/14 S.lilium Mexico (13)
15 */0 S.aratathomasi (1)
S.nana (3) 32


73/3
29


Uroderma bilobatum
senex Vampyressa pusilla
100/15 Carollia perspicillata
1 00/15 S.magna (7)
7 100/15 S.bogotensis (3)
21
S.mordax (1)
74/2 96 S.oporaphilum (6)
6 100/13 S.ludovici Ecuador (2)
70/31695
11 10016 965 S.hondurensis Honduras (8)
11 100/11 6

88/3 12 71/2 S.hondurensis Guatemala (8)
6 100/9 3
100/14 1 S.hondurensis Costa Rica (4)
15 S.sp.B (2)
100/12
20 S.erythromos (9)
S.tildae Guyana
2:1 Stildae Suriname

S.tildae Guyana

1 S.tildae French Guiana
2 Stldae Ecuador
14 S.tildae Ecuador
1 S.tildae Ecuador
1 S.tildae Bolivia
2 1 S.tildae Bolivia
1

2 2S.tldae Venezuela
1 S.tildae Guyana
7 21 S.tildae Guyana
2 S.tildae Trinidad
2 S.fidae Trinidad & Tobago
S.tildae Trinidad & Tobago
4 S.tildae Brazil
S. sp. A Ecuador






























Figure 12. Numerical values, abbreviations, and symbols are as indicated in
figure 7. Detailed information of phylogenetic relationships of all specimens
available for S. lilium, S. thomasi, and S. luisi.












S.thomasi Guadeloupe

9915 S.thomasi Guadeloupe
5
94/4 63/1 6 S.lum angeli Dominica
5 1 S.lilium zygomaticus Martinique

S.thomasi vulcanensis Montserrat
1
S.luisi Ecuador

66/1 S.luisi Panama
1 S./uisi Panama
55/1 S.luisi Panama
3
2 S.i/um Panama
2 2 S.lilium Panama
1 1 S./ilium Guyana

2 S./l/urn Guyana
S.i/ium Guyana
1 1 S.lilium Panama
92/0 86/2 4 S.ilum Panama
6 3 S.lnum Costa Rica

99/5 1 S.flium serotinus Grenada
6 S.lilium paulsoni St. Vincent

80/1 2 S.Ilum Suriname
3 1S.liium French Guiana
4 S.ilum Tobago
4
S.illum Peru
1
S.ilium Bolivia
1 S.fflium Ecuador

5411 2 S.lilium Ecuador
S.liliumn Peru
to figure 13 2
55/1 1 S.llum Ecuador
1 65/1 S.1ilium Peru
1 S.lilium Peru





























Figure 13. Numerical values, abbreviations, and symbols are as indicated in
figure 7. Detailed information of phylogenetic relationships of all specimens
available for S. lilium and S. aratathomasi.















Ato figure 12


100/12
19


61/2
7


100/14
15


3 S.lilium Brazil
4 S.ilium Paraguay
S.Iilium Brazil

2 S.lillum Brazil
94/0 S.lilium Brazil
6
52/0 S.lIilum Bolivia
11 Slilium Paraguay

S.ifflum Bolivia
S.lilium Paraguay

6 S.lilium Brazil
S.lilium Guatemala
S.lilium Guatemala

S.lilium Guatemala
2 S.lilium Honduras
S./ilium El Salvador
1 1 S.ilium Mexico

2 S.ilium Mexico
S. liuium El Salvador
1
1 S.ilium El Salvador

SS.lilium Costa Rica
S.lilium Guatemala

1 S.lilium Guatemala
S.lilium Mexico
S.aratathomasi Colombia


52/2
5





























Figure 14. Numerical values, abbreviations, and symbols are as indicated in
figure 7. Detailed information of phylogenetic relationships of all specimens
available for S. nana and S. bidens.










73/3 96/10


na bilobatum
Vampyr
perspicillata


pusilla


aphilum (6)
S.ludovici Ecuador (2)
5 S.hondurensis Honduras (8)


o.ludovici Guatemala (8)
dovici Costa Rica (4)


.tildae (16)


--1-S.thomasi Guadeloupe (5)
55/1
3 61/1S.luisi Panama (14)
92/0 2
6
S.lilium Suriname (13)

100/12 _Slilium Paraguay (10)

61/2
7
100/14 S.lilium Mexico (13)
15


I S.nana Peru
110115 2 S.nana Peru
26 2 S.nana Peru

7 S.bidens Ecuador
10 S.bidens Peru


100/12
22






























Figure 15. One of 1291 equally parsimonious trees of 337 steps (Cl: 0.62, RI:
0.90) from a cladistic analysis of cytochrome b gene data set (successively
weighted parsimony) of 133 individuals representing Stumira species and related
outgroup taxa. Numerical values, abbreviations, and symbols are as indicated in
figure 7. Arrowheads show nodes where resolution (and support values) had
improved after successively weighted analysis.









68

'93 IkJ99

SUroderma bilobatum
Centurio senex Carollia perspicillVampyressa pusilla

k 100lO S.nmagna (7)

T 2 100 S.bogotensis (3)
84 S.oporaphilum (6)
1 ,2 S.ludovici Ecuador (2)
90 "9 S.hondurensis Honduras (8)



3 S.ludovici Guatemala (8)

7S. 1S.Iudovici Costa Rica (4)
fS~mordax (1) 3i
59 t100 2 _S.sp. B (2)


59S.erythromos (9)

8 S.tildae (16)
100 .S.sp. A(2)

90 S.thomasi Northern Lesser Antilles (5)
2


93 92 66 S.luisi Panama (14)

5 2


S.lilum Suriname (13)



88 2
100 S.lilium Paraguay (10)
6
1- 55

100 S.1diium Mexico (13)
S.aratathomasi (1)
9 8S.bidens (2)






























Figures 16. One of two (first hypothesis) equally parsimonious trees (782 steps,
Cl: 0.48, RI: 0.64) from a cladistic analysis of cytochrome b gene data set (Fitch
parsimony) of 37 individuals representing Stumira species and related outgroup
taxa. Numerical values, abbreviations, and symbols are as indicated in figure 7.
Bars on the right indicate tentative "complex" of species.











Centurio senex In
a
Uroderma bilobatum
.0
Vampyressa pusilla

Carollia perspicillata 0
100/231 S. aratathomasi I x
32 IS
3 aratathomasi 2 E
0
100/5 s. luisi l .
7 S.lis
97/6 '*uii2E
9100/6 IS. thomasi '
7S. thomasi 2
100/14 IS. lilium I
21 lilium 39 E
100/15 IS. ilium 114
16 IS. flium 2
100/16S. magna 1
51/ TO LS. magna 2

100/17 -s. bogotensis 1
18 S. bogotensis 2

*11_. [100/1 S. erythromos 1
31 14 i---S. erythromos 2

18 -S. sp. B
10011 S. mordax I
*162 13 1IS. mordax 2
6 100/15 --S- sp. A 1
I 17 S. sp. A 2

100118F r-. tildae 1
24 S. tildae 2
100/13 I S. ludovici 1
l x

1/S. ludovici 2 a.
5 E
100/9 S. oporaphilum 1 0
S. oporaphilum 2 .2
0
100/11 S. hondurensis 1 V
17 S. hondurensis 2 -

S. nana 1 0
* nana 2 E
0
1

ns 2
0
U


71/3





























Figures 17. One of two (second hypothesis) equally parsimonious trees (782
steps, CI: 0.48, RI: 0.64) from a cladistic analysis of cytochrome b gene data set
(Fitch parsimony) of 37 individuals representing Stumira species and related
outgroup taxa. Numerical values, abbreviations, and symbols are as indicated in
figure 7. Bars on the right indicate tentative "complex" of species.











Centurio senex


0.
-Uroderma bilobatum
.0
Vampyressa pusilla
rollia perspicillata 0
100/231S. aratathomasi 1
32 S. aratathomasi 2 C.
E
S. luisi 1 0
S. luisi 2
E
S. thomasi 1 0
S. thomasi 2
S.flum 1
lilium 39 E
Is. lilium 114
IS. lilium 2 S m I
100/16aga 1
20 .S. magna 2
100/17 r-S. bogotensis 1

18 LS. bogotensis 2

-S. erythromos 1
-S. eiythromos 2
. sp. B
mordax 1
mordax 2
S. sp. A 1
S. sp. A 2
-S. tildae 1
-S. tildae 2
S. ludovici 1
x
- S. ludovici 2 0.
E
S. oporaphilum I 0
-S. oporaphilum 2
0
S. hondurensis 1 "
-S. hondurensis 2 -
x
CL

E

0
0
U


-S. bidens 2





























Figure 18. The single most parsimonious tree of 260 steps (Cl: 0.70, RI: 0.75)
from a cladistic analysis of cytochrome b gene data set (successively weighted
parsimony) of 37 individuals representing Stumira species and related outgroup
taxa. Numerical values, abbreviations, and symbols are as indicated in figure 7.
Bars on the right indicate tentative "complex" of species. Arrowheads show
nodes where resolution (and support values) had improved after successively
weighted analysis.











Centurio senex
0.
Uroderma bilobatum
0
Vampyressa pusilla
Carollia perspicillata 0

100 S. aratathomasi 1 x
,59 S. aratathomasi 2
56 0
1100 IS. luisi 1I(
2 S. luisi 2
4 2 100 S. thomasi 1
2 S. thomasi 2
100 S. lilium 1
6 S.ilium 39 E
9 100 S. li/ium114
5 100 S i lum2 S .'magnalI

75 6 .magna 2
100 S. bogotensis 1
1 r0 c
6 ES. bogotensis 2
_6 100 S. erythromos 1

100 S. erythromos 2

6 S. sp. B

100 S. mordax 1
70 3 IS. mordax 2
2 0 lo rS. sp. A 1
5 L-S. sp. A 2

100 S. tildae 1
7 S. tildae 2
100 L S. ludovici 1
87 S. ludovici 2 -.
100 [--. oporaphilum 1 0

S. oporaphilum 2
3 >
l00 r S. hondurensis 1
4 --L S. hondurensis 2 --
S. nana 1 .2
0.
nana 2 E


ns 2 "
0
0






























Figures 19. One of two (first hypothesis) equally parsimonious trees (257 steps,
Cl: 0.28, RI: 0.62) from a cladistic analysis of a qualitative multistate osteological
data set (Fitch parsimony) of 37 individuals representing Stumira species and
related outgroup taxa. Numerical values, abbreviations, and symbols are as
indicated in figure 7.











/Carollia spp.
4100/11 rS. aratathomasi 1
15 LS. aratathomasi 2

57/2 Uroderma bilobatum
4 Vampyressa pusilla ,,F


magna 1
-S. magna 2


.sp. A I
S. sp. A 2


bidens 2


*1 S. ludovici 1
4 L...S. ludovici 2

-S. hondurensis I
S. hondurensis 2

E s. erythromos I
S. bogotensis 1
E S. eiythromos 2
S. bogotensis 2
4 95/3 rS. oporaphilum 1
8 L.. oporaphilum 2

sp. B


168L2S. thomasi 1
3 L. S. thomasi 2

f2_ -S. liliumlI
Z S. luisilI

llium39






























Figures 20. One of two (second hypothesis) equally parsimonious trees (257
steps, Cl: 0.28, RI: 0.62) from a cladistic analysis of a qualitative multistate
osteological data set (Fitch parsimony) of 37 individuals representing Stumira
species and related outgroup taxa. Numerical values, abbreviations, and
symbols are as indicated in figure 7.













aratathomasi I
aratathomasi 2
S. magna I
6 S. magna 2


sp. A 1


sp. A2
nana 1


bidens 2


sp. B
rS. erythromos I


oporaphilum 1
oporaphilum 2


-S. hondurensis 1
. hondurensis 2


S. lilium I
luisi 1
lilium39


.S. tildae 2
tildae 1






























Figure 21. The single most parsimonious tree of 42 steps (Cl: 0.39, RI: 0.74)
from a cladistic analysis of a qualitative multistate osteological data set
(successively weighted parsimony) of 37 individuals representing Stumira
species and related outgroup taxa. Numerical values, abbreviations, and
symbols are as indicated in figure 7. Arrowheads show nodes where resolution
(and support values) had improved after successively weighted analysis.






























-S. bogotensis 2
94 -S. oporaphilum I
1L S. oporaphilum 2






























Figures 22. One of four (first hypothesis) equally parsimonious trees (1078
steps, Cl: 0.41, RI: 0.61) from a cladistic analysis of a combined cytochrome b
gene data set and a qualitative multistate osteological data set (Fitch parsimony)
of 37 individuals representing Stumira species and related outgroup taxa.
Numerical values, abbreviations, and symbols are as indicated in figure 7.
Arrowheads are pointed at unresolved nodes of well-supported internal clades.











Centuro senex Uroderma bilobatum






























Figures 23. One of four (second hypothesis) equally parsimonious trees (1078
steps, CI: 0.41, RI: 0.61) from a cladistic analysis of a combined cytochrome b
gene data set and a qualitative multistate osteological data set (Fitch parsimony)
of 37 individuals representing Sturnira species and related outgroup taxa.
Numerical values, abbreviations, and symbols are as indicated in figure 7.
Arrowheads are pointed at unresolved nodes of well-supported internal clades.












Uroderma bilobatum
i p Vampyressa pusilla
100p a A rS. aratathomasi 1
40j S. aratathomasi 2






























Figures 24. One of four (third hypothesis) equally parsimonious trees (1078
steps, Cl: 0.41, RI: 0.61) from a cladistic analysis of a combined cytochrome b
gene data set and a qualitative multistate osteological data set (Fitch parsimony)
of 37 individuals representing Stumira species and related outgroup taxa.
Numerical values, abbreviations, and symbols are as indicated in figure 7.
Arrowheads are pointed at unresolved nodes of well-supported internal clades.











Uroderma bilobatum
Vampyressa pusilla
i prspiioai S* aratathomasi I
40 S. aratathomasi 2


78/423 L S. lilium 39


17 S. flium 114




9 [. thomasi I
10 S. thomasi 2


-S. bogotensis 1
-S. bogotensis 2
ludovici I
- S. ludovici 2
3. oporaphilum I
-S. oporaphilum 2
-S. hondurensis 1
-S. hondurensis 2





























Figures 25. One of four (fourth hypothesis) equally parsimonious trees (1078
steps, Cl: 0.41, RI: 0.61) from a cladistic analysis of a combined cytochrome b
gene data set and a qualitative multistate osteological data set (Fitch parsimony)
of 37 individuals representing Stumira species and related outgroup taxa.
Numerical values, abbreviations, and symbols are as indicated in figure 7.
Arrowheads are pointed at unresolved nodes of well-supported internal clades.










67/1


bilobatum


100/34 [S. aratathomasi 1
40 LS. aratathomasi 2
100/18 S /ilum 1
23 ES. lilium 39
11I 100/13 S. li/ium 2
17 1--- S. lilium 114

9 8/5 S. luisi I

9 oo rS. thomasi I
10 s. thomasi 2


bogotensis I
. bogotensis 2


-S. sp. A 1
-S. sp. A2
S. magna 1


ludovici 2


% oporaphilum I
-S. oporaphilum 2
-S. hondurensis I
S. hondurensis 2


bidens I
-S. bidens 2




Full Text
xml version 1.0 encoding UTF-8
REPORT xmlns http:www.fcla.edudlsmddaitss xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.fcla.edudlsmddaitssdaitssReport.xsd
INGEST IEID EYFGG96AO_BG5G23 INGEST_TIME 2014-06-16T22:03:15Z PACKAGE AA00022121_00001
AGREEMENT_INFO ACCOUNT UF PROJECT UFDC
FILES