MASARYK UNIVERSITY Faculty of Science

DISSERTATION THESIS

Brno 2008 Peter Vallo MASARYK UNIVERSITY Faculty of Science

Peter Vallo

MOLECULAR GENETIC STUDIES IN AFRICAN LEAF­NOSED BATS (HIPPOSIDERIDAE)

Dissertation thesis

Supervisor: doc. RNDr. Petr Koubek, CSc. Brno 2008 Bibliographic identification

First name and surname of author: Peter Vallo

Title of dissertation thesis: Molecular genetic studies in African leaf­nosed bats (Hipposideridae)

Title of dissertation thesis in Czech: Molekulárně genetické studie afrických pavrápencovitých (Hipposideridae)

Study program: Biology

Field of study: Zoology

Supervisor: doc. RNDr. Petr Koubek, CSc.

Year of defence: 2008

Keywords: mitochondrial DNA, cytochrome b, taxonomy, phylogeny, Hipposideros caffer, Hipposideros ruber, Triaenops persicus

Keywords in Czech: mitochondriální DNA, cytochrom b, taxonomie, fylogeneze, Hipposideros caffer, Hipposideros ruber, Triaenops persicus © Peter Vallo, Masaryk University, 2008 Contribution of Ph.D. student Peter Vallo to papers presented in the thesis

Paper A Vallo et al. in press. Variation of mitochondrial DNA in the Hipposideros caffer complex (Chiroptera: Hipposideridae) and its taxonomic implications: co­authored the original idea of the paper, took part in field sampling of bats in Senegal and Morocco, processed samples from isolation of DNA to completion of DNA sequences (ca. 50% of samples in the paper), performed molecular phylogenetic analysis, wrote the first version of manuscript and prepared its final version

Paper B Vallo et al. manuscript. Genetic structure of H. cf ruber from southeastern Senegal inferred from sequences of the mitochondrial cytochrome b gene: co­authored the original idea of the paper, took part in field sampling of bats in Senegal, processed samples from isolation of DNA to completion of DNA sequences, performed molecular phylogenetic analysis, wrote the first version of manuscript and prepared its final version

Paper C Benda and Vallo submitted. Taxonomy of the genus Triaenops (Mammalia: Chiroptera: Hipposideridae) with description of a new species from southern Arabia: processed samples from isolation of DNA to completion of DNA sequences, performed molecular phylogenetic analysis, wrote the molecular phylogenetic parts of manuscript and contributed to its final version

doc RNDr. Petr Koubek, CSc. supervisor Acknowledgements

I would like thank my supervisor Dr. Petr Koubek for supervising my PhD study and for the opportunity to be a part of the research team investigating the vertebrate fauna of Senegal. Further I would like to thank Dr. Petr Benda for tutoring my existence in the world of bat science and for fruitful cooperation. I cordially thank Dr. Natália Martínková for introducing me into molecular techniques and phylogenetic analysis. I also thank Prof. Jiří Gaisler for provision of bat literature. For opportunities to get experience in non­bat research, I thank Dr.'s Steven Monfort, William McShea, Krzysztof Schmidt and Leandro Silveira. The Foundation Nadání Josefa, Marie a Zdeňky Hlávkových kindly supported my stay in Brazil, the last non­bat experience (so far...). I thank Maťo and Lenka for friendship and social well­being, and Adam for yet another social well­being and friendship. My thanks further go to Jarka and Mirka for five years of inspirative coexistence in one office, and to people from the lab in Studenec for kind attitude. Last but not least, I would like to express thousand thanks to my family and Vlaďka for never­ending support, help and love.

Dissertation research was carried out at the Institute of Vertebrate Biology, v.v.i., Academy of Science of the Czech Republic, in Brno. Laboratory work was done at the molecular genetic laboratory of the IVB AS CR, v.v.i., in Studenec. My dissertation related research activities were supported by the grant nr. IAA6093404 by the Grant Agency of the AS CR. Abstract

Molecular methods using DNA sequence data have been widely used in solving systematic relationships in bats and have radically changed perception of this mammalian order. The family of leaf­nosed bats (Hipposideridae) has not yet been thoroughly studied by the molecular genetic approach. Suspicion of artificial intrafamilial grouping and proved existence of high cryptic diversity encourage use of this approach. This pays especially for leaf­nosed bats of Afrotropics, as only fragmentary information on their taxonomy and phylogeny has been published to date. This dissertation thesis presents results of analysis of sequences of mitochondrial cytochrome b gene in cryptic forms of Hipposideros caffer complex, and in species Triaenops persicus. Genetic structure revealed within the H. caffer complex challenged a hypothesis of two cryptic species, H. caffer and H. ruber. Six distinct lineages of potentially specific statuses were recovered. Two currently recognised subspecies of H. caffer can be considered two distinct species, H. caffer and H. tephrus. Taxonomic assignation of three lineages of H. ruber and one lineage comprising either H. caffer or H. ruber morphotypes has yet to be confirmed. Two sympatric genetic forms of the West African lineage, originally assigned to H. ruber, were found in southeastern Senegal. Morphological differences between these two forms were slight. A large range of echolocation frequencies detected in a local population, however, had no relevance to this distinction, but was related to sex in the more numerous form. The two Senegalese forms are apparently paraphyletic, because phylogeny revealed sister relationship between the less numerous form and a form from Benin. This indicates that another cryptic species may exist in Senegal. Revision of the Afro­Arabian T. persicus revealed several distinct morphotypes from Africa and the Middle East with significant genetic divergences among most of them. Therefore, a split of the current T. persicus rank into three species was proposed: T. afer in Africa, and T. persicus and T. parvus spec. nov. in the Middle East. Due to considerable genetic and morphologic differences among continental and island forms of Triaenops, two Madagascan species T. furculus and T. auritus were separated into a new genus Paratriaenops gen. nov. Abstrakt

Molekulární metody s využitím studia sekvencí DNA jsou velmi často využívány k řešení systematických vztahů u netopýrů a významným způsobem mění pohled na členění tohoto savčího řádu. Pavrápencovití (Hipposideridae) jsou jednou z čeledí, která dosud nebyla těmito metodami komplexně studována. Podezření na umělé vnitřní členění čeledi a prokázaná existence vysoké kryptické diverzity k využití těchto metod přímo vybízí. To platí především o pavrápencovitých „Afrotropů”, o jejichž taxonomii i fylogenezi byly dosud publikovány jen dílčí informace. Předložená disertace shrnuje výsledky získané analýzou sekvencí mitochondriálního genu pro cytochrom b u kryptických forem komplexu Hipposideros caffer a druhu Triaenops persicus. Genetická struktura zjištěná v komplexu H. caffer popřela hypotézu dvou kryptických druhů H. caffer a H. ruber. Bylo identifikováno šest rozdílných linií, které mají charakter druhu. Dva dosud platné poddruhy H. caffer je možné považovat za dva odlišné druhy, H. caffer a H. tephrus. Taxonomickou příslušnost tří linií H. ruber a jedné linie obsahující oba morfotypy H. caffer i H. ruber je ještě potřeba potvrdit. Dvě sympatrické genetické formy západoafrické linie původně určené jako H. ruber byly zjištěny v jihovýchodním Senegalu. Morfologicky se lišily jen nepatrně. Poměrně velký rozsah echolokačních frekvencí zjištený u lokální populace však s tímto rozdělením nesouvisel, ale měl vztah k pohlaví u početnejší formy. Senegalské formy jsou zřejmě parafyletické, protože fylogeneze odhalila sesterský vztah mezi méně početnou formou a formou z Beninu. To naznačuje možnost existence dalšího kryptického druhu v Senegalu. Revize afro­arabského druhu T. persicus odhalila několik rozdílných morfotypů z Afriky a Středního Východu s významnými genetickými rozdíly mezi většinou z nich. Proto bylo navrženo rozdělení T. persicus do tří druhů: T. afer v Africe, a T. persicus a T. parvus spec. nov.. na Středním Východě. Kvůli značným genetickým a morfologickým rozdílům mezi kontinentálními a ostrovními formami rodu Triaenops byly madagaskarské druhy T. furculus a T. auritus vyčleněny do rodu Paratriaenops gen. nov. Contents

Introduction...... 9

DNA sequence data...... 9

Systematic relationships of Chiroptera...... 11

Family Hipposideridae...... 13

Aims of the thesis...... 17

Summary of results...... 18

References...... 22

Paper A...... 29

Paper B...... 56

Paper C...... 77

Curriculum vitae...... 128 Introduction

Many generations of naturalists have researched biological forms with the aim of describing and explaining their diversity. Classification of biological objects in the sense of Linnaeus (1758) had gradually changed into systematics, where classification of organisms considers their evolutionary relatedness in the framework of phylogeny, a hierarchical structure interrelating all biological forms into the tree of life (Hillis et al. 1996, Avise 2004). Human­created approximations, dependent on current state of knowledge, began to follow objective approaches based on shared traits of extant and fossil organisms in the second half of the twentieth century (e.g. Hennig 1950). Methods of phylogenetic reconstruction have been refined during subsequent decades, as more powerful computing machines enabled utilisation of more sophisticated algorithms, and phylogenetic inference has permeated into almost every field of biology. Traditional approaches for estimation of phylogeny involved comparisons of phenotypic data from morphology or other organismal characteristics amenable to observation. Along with advances in methods of phylogenetic inference, molecular biology has been rapidly developing. Examination of molecular structure of proteins and nucleic acids was soon implemented into estimation of phylogeny and molecular data became a valuable source of information for systematical studies. The real boom of utilization of molecular markers in phylogenetic inference was caused by invention of the polymerase chain reaction (PCR; Mullis et al. 1986, Mullis and Faloona 1987), which enabled large­ scale investigation of the direct source of genetic information, the sequences of DNA. Around the turn of the twentieth century, analysis of DNA sequences became a standard for phylogenetic inference and today systematicists routinely employ PCR to amplify desired fragments of DNA, sequence these on automated sequencers, and analyze many kilobases of sequence information to reveal relationships that would otherwise remain hidden.

DNA sequence data Phylogenetic inferrence is based on a central concept of correspondence of compared characters due to common ancestry. This correspondence is called homology (Fitch 1970). More specifically, the compared characters should descend directly from a

9 speciation event, in other words, be orthologous (Fitch 1970). Although, according to the 'tree of life' concept all organisms share a common ancestor (Woese and Fox 1977), similarity in molecular characters can be eroded due to accumulation of too much variation over time. This issue is particularly obvious in sequences of nucleic acids, which are the most straightforward source of genetic information for estimation of phylogeny (Salemi and Vandamme 2003, Avise 2004). Each nucleotide position in a DNA sequence is considered a character, which can attain only four character states, i.e. nucleotides adenine, cytosine, guanine and thymine. Homologous positions in DNA obviously provide only information on identity or difference of character states but no evidence for step by step changes. Thus, similarity of sequences can result from events such as multiple or parallel substitutions, creating the effect called homoplasy. The principal merit of DNA sequences lies in combination of large number of nucleotide sites with limited number of character states (Avise 2004). Due to a considerable degree of selective neutrality according to Kimura (1968), assuming nucleotide subsitutions to be random events, phylogenetic inference from sequence data can be considered a statistical problem (Salemi and Vandamme 2003). As such it can tackle homoplasy, which obscures phylogenetic information. An adequate level of mutation rate at particular segment of DNA can help to retain this information. Markers with different mutation rate are therefore chosen for research on various taxonomic levels. The majority of molecular systematic studies have been based on analysis of mitochondrial DNA (mtDNA) (Avise 2004). MtDNA markers have several advantageous properties, which make them suitable for phylogenetic inference. They are largely clonal, maternally inherited and generally nonrecombining (Macaulay et al. 1999). Rapid evolution enables recovery of multiple haplotypes that can easily be aligned according to homologous positions and organized into phylogeny. Considerable variation in evolutionary rates of different segments (Lopez et al. 1997) makes mtDNA sequences informative across a large span of evolutionary time (Hillis et al. 1996, Avise 2004). Due to variation existing among individuals both within and between populations, mtDNA has proven to be also an effective marker for assessment of population structure (Harrison 1989). Studies investigating phylogenetic relationships among species and closely related genera require characters that evolve at a sufficient rate to detect changes that have

10 occurred over short evolutionary periods. Over the last two decades, cytochrome b gene (cytb) has become one of the most studied and best understood mitochondrial genes (Bradley and Baker 2001, Avise 2004, Baker and Bradley 2006). Cytochrome b is one of the proteins related to the mitochondrial oxidative phosphorylation system and is the only gene in this complex encoded by the mitochondrial genome (Irwin et al. 1991). Cytb has been a widely utilized genetic marker to estimate phylogenies on various systematic levels in mammals (Van Den Bussche et al. 1998). This was largely facilitated by publication of universal primers utilizable for most mammalian groups (Kocher and Irwin 1989, Irwin et al. 1991).

Systematic relationships of Chiroptera

Molecular efforts can be most enlightening when they concentrate on problematic areas where traditional phylogenetic appraisals have been inconclusive. Entangling relationships among extant orders of placental mammals, which exhibit high diversification in physical forms, has been amongst the most interesting issues of systematics (Waddell et al. 1999). With application of molecular analysis in late 1990s, alterations in evolutionary positions of some orders and lack of support of the existence of others changed the standard classification of mammals (Simpson 1945). Remarkable conclusion of analysis of multiple gene sequences was that about one third of all placental mammals in the world originated from a common ancestor that inhabited the African continent about 75 million years ago (Murphy et al. 2001, Scally et al. 2001, Springer et al. 1997, Stanhope et al. 1998, Waddell et al. 1999). This superordinal group called Afrotheria comprises creatures as diverse as e.g. elephants, aardvarks and elephant shrews. Three other primary superordinal clades were further revealed in placental mammals: Xenarthra, traditionally grouping sloths, anteaters and armadillos, Glires, containing rodents and lagomorphs, as a sister taxon to Euarchonta, containing primates, flying lemurs and tree shrews; and the remaining orders of placental mammals into Laurasiatheria (Nikaido et al. 2000, 2001) with bats exhibiting close relationship to perissodactyls, carnivores and pholidots (Nishihara et al. 2006). This points to a remarkable phenotypical plasticity of mammals and parallel radiations, from which similar biological forms originated and were mistakenly considered as phylogenetically close, e.g. former order Insectivora (Madsen et

11 al. 2001). Conventional view based on morphology that placed the bats to Archonta as a sister group to flying lemurs (Simmons and Geisler 1998) has been largely abandoned. Molecular data, however, opened another interesting discussion concerning relationships within the order Chiroptera. According to the traditional view of bat systematics based on morphology, the order Chiroptera forms a monophyletic group comprising two suborders, Megachiroptera, containing only one family, Pteropodidae, and Microchiroptera, comprising all other bat families (Koopman 1994, Simmons and Geisler 1998). This classification has been challenged by an alternative hypothesis that disagreed with the traditional placement of Megachiroptera. This hypothesis, known as “flying primate” hypothesis, argued that megabats were a lineage of primate radiation, and that this relationship was supported by neurological data (Pettigrew 1986, Pettigrew et al. 1989). Later phylogenetic analyses, however, stated that neurological traits of megabats and primates are not a result of shared evolutionary history but rather highly convergent due to common ecology (Legendre and Lapointe 1995). Molecular approach using DNA sequences supported the traditional monophyly of bats (Ammerman and Hillis 1992, Simmons 1994, Van Den Bussche 1998), despite criticism stating monophyly of bats was confounded by DNA base compositional bias in the megabat and microbat lineages due to metabolic constraint associated with flight (Pettigrew 1994, Pettigrew and Kirsch 1995, Pettigrew and Kirsch 1998). However, molecular data brought another contrasting hypothesis that challenged traditional division to Megachiroptera and Microchiroptera. This new hypothesis suggested that megabats are a lineage derived from microbats, and are related to Rhinolophoidea (Pettigrew and Kirsch 1995, Kirsch 1996, Hutcheon et al. 1998). Morphology obviously supports monophyly of microbats, since all possess a complex laryngeal echolocation system (Simmons and Geisler 1998). The contradicting hypothesis of microbat paraphyly was confirmed by several recent studies using multiple nuclear gene sequences (Teeling et al. 2000, 2002, 2005, Springer et al. 2001). According to Simmons (2005), monophyletic Chiroptera are nowadays systematically divided into Yinpterochiroptera (Springer et al. 2001), grouping megabats and Rhinolophoidea, and Yangochiroptera (Simmons and Geisler 1998), containing all other microbats. Along with disputes on ordinal level, molecular data have been intensively used in assays aiming at relationships among bat species. Most species are defined by morphology,

12 however, only forms diagnosable by our sensory perception can by defined by this morphospecies approach (Mayr 1942, Mayr 1996). In microbats, morphologically similar species, i.e. cryptic species (Mayr 1942, Lincoln et al. 1998), can often go undetected, because their diagnostic features exist beyond the range of human senses. Thus, the degree of bat species diversity is still uncertain due to existence of morphologically cryptic but acoustically divergent species, although bats surely count among the most extensively studied mammals (Mayer and Helversen 2001b, Jones and Barlow 2004). Examination of molecular data demonstrated that many morphologically cohesive populations harbour genetically distinct species. In European bat fauna, the number of species recently increased by about 20% (Mayer and Helversen 2001b, Ibáñez et al. 2006). One of the most illustrious cases of long undetected cryptic variation is a common species Pipistrellus pipistrellus (Schreber, 1774), which has traditionally been considered to be monotypic in Europe. However, discovery of two distinct phonic types stimulated molecular analysis using mtDNA sequences, which subsequently confirmed existence of two highly divergent genetic lineages acknowledged as two distinct species P. pipistrellus and P. pygmaeus (Leach, 1825) (Barrat et al. 1997). Further molecular phylogenetic research showed more cryptic diversity in the P. pipistrellus complex existing in the Mediterranean (Hulva et al. 2004, 2007, Benda et al. 2004). This complex of forms is nowadays perhaps the most comprehensive model for cryptic variation in the Palaearctic bats (Hulva et al. 2004). Similarly cryptic species appeared within other species groups that also became the major subject of molecular studies, e. g. European Myotis or Plecotus species (Kiefer et al. 2002, Mayer and von Helversen 2001a, b, Spitzenberger et al. 2001).

Family Hipposideridae

The family of leaf­nosed bats, Hipposideridae Lydekker, 1891, is a typical Palaeotropical group of bats that occurs throughout sub­Saharan Africa including Madagascar, with marginal distribution in the Maghreb. In Asia, southern regions of the Middle East, southern and southeastern Asia, including majority of the Indomalayan islands, and northern parts of Australia (Koopman 1994, Nowak 1999, Simmons 2005). It comprises of nine genera: Anthops Thomas, 1888, Aselliscus Tate, 1941, Asellia Gray, 1838, Cloeotis Thomas, 1901, Coelops Blyth, 1848, Hipposideros Gray, 1831,

13 Paracoelops Dorst, 1947, Rhinonycteris Gray, 1847, and Triaenops Dobson, 1871. Although they display a wide range of variation in morphology, their common and most significant feature is a discoidal transversally split nasal structure, which operates in constant frequency nasal echolocation. In the past, this family was regarded a part of Rhinolophidae (Koopman 1994, Simmons and Geisler 1998), the family of horseshoe bats. Nowadays, Hipposideridae and Rhinolophidae are considered separate families (Bogdanowicz and Owen 1998, Simmons 2005). Genera Anthops, Paracoelops and Rhinonycteris are monophyletic. Species Anthops ornatus, Paracoelops megalotis, Rhinonycteris aurantius (Gray, 1845) are rather enigmatic taxa living in restricted distribution areas in Solomon Islands, Vietnam and northern Australia, respectively. Aselliscus and Coelops include two species each, and they are distributed in large parts of southeastern Asia. Four other genera inhabit Africa. Genus Cloeotis, with only one species C. percivali, is distributed across a rather large area of southern and eastern Africa. Asellia and Triaenops have Afro­Arabian distribution. Asellia occurs throughout the arid and desert areas of Africa north of Equator and reaches the Middle East and Arabian peninsula to Pakistan. Triaenops occurs predominantly in eastern Africa and in the Middle East, represented by T. persicus Dobson, 1871, and in Madagascar, where three other species live (Koopman 1994, Simmons 2005, Ranivo and Goodman 2006). Hipposideros is the most species­rich genus within Hipposideridae. Since the list of bat species by Simmons (2005), three new species have been described from southeastern Asia (Thabah et al. 2006, Guillen­Servent and Francis 2006, Bates et al. 2007), which raised the number of extant species of the genus currently recognised to seventy. Family Hipposideridae has long attracted attention of systematicists, as the intrafamilial relationships are obscure and subject to controversies. Especially, systematics of the genus Hipposideros has been complicated by morphological similarity of many of its species on one side and highly divergent appearance of specialized forms on the other side. Taxonomic revision by Tate (1941) was the first comprehensive work on the genus, although primarily it has dealt with Asian species and only briefly mentioned their African congeners. Therein suggested eleven supraspecific groups were further reduced to seven by Hill (1963). Hill's (1963) organisation of the genus is generally accepted in subsequent systematic studies until today (e.g. Rosevear 1956, Koopman 1994, Simmons 2005, Van Cakenberghe and Seamark 2008), despite suspicions on its intuitiveness and doubts on

14 reflection of real phylogeny (Bogdanowicz and Owen 1998, Fahr and Ebigbo 2003). These concerns were partially confirmed by Bogdanowicz and Owen (1998) who analysed extensive set of morphological traits on several hundreds of specimens, but unfortunately, they were not able to provide an unambiguous alternative to current systematics of the genus and family. Molecular approach seems to suggest such a solution. Up to date, the molecular approach to systematics of the genus reveals artificial relationships of several African species to supraspecific groups traditionally grouped with morphologically similar Asian species (Vallo et al. in press). Several studies on Hipposideridae have employed molecular phylogenetic methods but only few have targeted phylogenetic relationships in the family. Studies by Wang et al. (2003) and Li et al. (2007) contributed to settling relationships in Chinese species of genera Aselliscus, Coelops and Hipposideros. Rather exceptional were the studies by Russel et al. (2007, 2008) who used a synthetic approach of coalescent, population genetic and traditional phylogenetic methods to assess biogeographical history of Madagascan Triaenops species. Although their study does not deal with systematics, parts of their results were considered in taxonomic revision of Madagascan Triaenops based on morphology (Ranivo and Goodman 2006). A different issue, the phenomenon of cryptic species, has driven molecular analyses of Hipposideros. Leaf­nosed bats appear to exhibit particularly high level of cryptic diversity (Pye 1972, Jones et al. 1993, Francis et al. 1999, Kingston et al. 2001, Thabah et al. 2006, Guillén­Servent and Francis 2006). Frequent occurrence of cryptic species in the family as a whole is probably related to acoustic divergence, as narrowband echolocation calls are a typical feature of Hipposideridae, and they are considered acoustic signatures reliably identifying species (Jones and Barlow 2004). Several cryptic species have been revealed in southeast Asia in the last decade: e. g. H. rotalis and H. orbiculus (Francis et al. 1999), H. khasiana (Thabah et al. 2006), H. khakhouayensis (Guillén­Servent and Francis 2006), H. boeadi (Bates et al. 2007). Their specific rank has often been supported by molecular analyses of mitochondrial DNA (Thabah et al. 2006, Guillén­Servent and Francis 2006). On the other hand, cryptic diversity in African representatives of the genus still remains elusive. Cryptic forms have so far been acknowledged in two taxa. The first one, H. commersoni, has been divided into several species in the current bat species list (Simmons 2005). However, this division has not yet been supported by any recent study on

15 cryptic species within this taxon, except remarks on existence of distinct phonic types by Pye (1972). The other well known example of crypticity is H. caffer (Sundevall, 1846), one of the most abundant mammals of Africa (Brosset 1984). Hill (1963) regarded H. caffer as only one species but currently two distinct species, H. caffer and H. ruber (Noack, 1893) are recognized instead (Hayman and Hill 1971, Koopman 1994, Simmons 2005). H. caffer is generally smaller than H. ruber (Hayman and Hill 1971, Koopman 1994) and has a distinctively higher frequency of echolocation call (Fenton 1986, Heller 1992, Jones et al. 1993). Despite these apparent interspecific boundaries, phylogenetic relationships within the H. caffer complex remain unresolved, and existence of further cryptic species seems probable (Simmons 2005).

16 Aims of the thesis

The general aims of this thesis were to reconstruct phylogenenetic relationships in selected bat species of the family Hipposideridae using mitochondrial DNA sequences and amend to their current systematics, based so far largely on traditional morphology. Particularly, following specific aims were set:

● to resolve phylogeny and identify main taxonomically important groups in the complex of forms related to the species Hipposideros caffer, of which only H. ruber is curently recognized as a valid species

● to assess genetic diversity in Senegalese population of the West African form of the H. caffer complex, assigned to H. ruber

● to reconstruct phylogenetic relationships among morphotypes of Triaenops persicus and to revise its systematic position within genus and family

17 Summary of results

Presented thesis is composed of three partial studies, results of which were composed into original scientific papers. One paper was accepted for publication in Acta Chiropterologica, another has been submitted to Zootaxa and the last paper is a manuscript prepared for submission to an impacted journal. Paper A presents molecular analysis of cryptic forms of H. caffer, a pan­African bat of controversial taxonomic status. Paper B explores in more detail genetic structure of H. cf ruber from Senegal, which has been pinpointed in Paper A. Paper C revises genetic and morphological variance of Triaenops persicus in its Afro­Arabian distributon range.

Paper A

Variation of mitochondrial DNA in the Hipposideros caffer complex (Chiroptera: Hipposideridae) and its taxonomic implications Peter Vallo, Antonio Guillén­Servent, Petr Benda, Debra B. Pires and Petr Koubek accepted for publication in Acta Chiropterologica

Aim of this study was to reveal relationshisp among complex of forms in the Afrotropical leaf­nosed bat Hipposideros caffer. This taxon has been traditionally regarded as two recognized cryptic species, H. caffer and H. ruber, that differ by their size and echolocation calls. Recent concerns assume that morphology might not reflect true phylogenetic relationships. Forty­nine bats of the complex, keyed into two morphospecies, were analysed based on sequences of the mitochondrial cytochrome b gene. Revealed phylogeny challenged the hypothesis of two cryptic species. Instead of two monophyletic lineages expected to represent the two species of the complex, four distinct lineages diverged from an unresolved polytomy together with three other recognised species. Based on topology of the phylogenetic tree and genetic divergencies among revealed phylogroups can be suggested that at least five species exist within the H. caffer complex. A lineage of H. caffer morphospecies consised of two monophyletic lineages representing current subspecies of H. caffer. These can be considered two distinct species, the nominotypical H.

18 caffer restricted to Southern Africa and H. tephrus inhabiting northern parts of distribution range. Individual assigned to either H. caffer or H. ruber from West and East Africa grouped into another lineages. Because some individuals were sampled close to type locality of H. ruber, this lineage might represent H. ruber s. str.. Taxonomical assignation remains uncertain and should be confirmed by further sampling and comparison with type material. The third main lineage contained two Central African forms of H. ruber morphospecies, which occur in sympatry around the Gulf of Guinea. Whether they represent separate species has yet to be approved by other means of evidence, as gene flow may exist between divergent but incompletely sorted maternal ineages. The last lineage comprised West African bats falling within the H. ruber morphotype. These bats represent another distinct species but obviously does not conform to H. r. guineensis, traditionally assumed to occur in West Africa. Although, names for the Central African forms are readily available, taxonomic assignation for West African forms has to be carefully checked.

Paper B

Genetic structure of H. cf ruber from southeastern Senegal inferred from sequences of the mitochondrial cytochrome b gene Peter Vallo, Petr Benda, Natália Martínková, Peter Kaňuch, Petr Koubek and Jaroslav Červený manuscript in preparation

In this study, a population from Senegal originally assigned to H. ruber is studied using sequences of mitochondrial cytochrome b gene. According to results presented in paper A, the specific determination is actually not exact, as H. ruber s. str. is probably absent from West Africa. Two sympatric genetic forms were revealed to occur in southeastern Senegal. We attempted to survey phylogenetic relationships among fifty­two captured bats from several localities in the Niokolo Koba National Park in order to assess genetic diversity and identify status of the two mitochondrial lineages. Skull morphology was also studied in detail. A local population in locality Dar Salam, from which the divergent haplotype from paper A originated, was also explored in echolocation calls. Differences in echolocation

19 frequencies can eventually suggest presence of potential cryptic species. Detailed multidimensional analysis of skull metric dimensions indicated slight morphological differences between the two forms. A survey on a local population yielded a large range of echolocation frequencies, which were hypothesized to be related to genetic distiction. This was not proved because all sampled bats were shown to belong to only one, more numerous, genetic form. Instead, sex­dependent bias in frequencies was shown in this form. Phylogenetic comparison to other West African samples revealed sister relationship between the less numerous form and a form from Benin. Thus, the two Senegalese forms are apparently paraphyletic. Current situation may have resulted from a secondary contact of allopatric populations resulting from changes of desert and forested ares in recent age. These two forms may well be considered cryptic species, as their genetic divergence was comparable to divergences in other cryptic species of bats. However, absence of gene flow between them has to be checked to prove their reproductive isolation.

Paper C

Taxonomy of the genus Triaenops (Mammalia: Chiroptera: Hipposideridae) with description of a new species from southern Arabia Petr Benda and Peter Vallo submitted to Zootaxa

This paper is a contribution to taxonomic revision of the genus Triaenops, which has been considered monospecific in its Afro­Arabian range as the species T. persicus. Three other species of the genus have been recognised to inhabit Madagascar, T. rufus, T. furculus and T. auritus. Representative samples of T. persicus from East Africa and the Middle East were analused using both morphologic and molecular genetic approaches. Morphologic comparison showed several distinct morphotypes in the set, three of them in the Middle East and one in Africa. The Middle Eastern morphotypes differed mainly in size, while the allopatric African form showed differences in skull shape. Molecular analysis based on sequences of the cytochrome b gene revealed significant divergences among most of the morphotypes. Two allopatric morphotypes from the Middle East, although morphologically different, exhibited minimal genetic differences. The third Middle Eastern

20 morphotype, the smallest in examined set and occurring in sympatry with one of the large morphotype, showed considerable genetic divergence. Therefore, a split of the current T. persicus rank into three species was proposed. Form occurring in Africa was ranked as T. afer and forms living in the Middle East were assigned as T. persicus and T. parvus spec. nov.. Molecular genetic analysis also proved a relatively close relationship of the Madagascan T. rufus to Arabian T. persicus, what suggests a northern way of colonisation of Madagascar, independent of the African populations. Due to a considerable genetic divergence comparable to intergeneric distances in Hipposideridae and morphologic differences from the continental forms of Triaenops, generic status is proposed for the pair of Madagascan species T. furculus and T. auritus, as Paratriaenops gen. nov..

21 References Ammerman, L. K., and D. M. Hillis. 1992. A Molecular Test of Bat Relationships: Monophyly or Diphyly?. Systematic Biology 41:222–232. Avise, J. C. 2004. Molecular markers, natural history, and evolution, 2nd ed. Sinauer Associates, Inc., Sunderland, Massachusetts. Baker, R. J., and R. D. Bradley. 2006. Speciation in mammals and the genetic species concept. Journal of Mammalogy 87:643–662. Barrat, E.M., R. Deaville, T. M. Burland, M. W. Bruford, G. Jones, P. A. Racey, R. K. Wayne 1997. DNA answers the call of pipistrelle bat species. Nature 387: 138–139. Bates, P. J. J., S. J. Rossiter, A. Suyanto, and T. Kingston. 2007. A new species of Hipposideros (Chiroptera: Hipposideridae) from Sulawesi. Acta Chiropterologica 9:13– 26. Benda, P., P. Hulva, and J. Gaisler. 2004. Systematic status of African populations of Pipistrellus pipistrellus complex (Chiroptera: Vespertilionidae), with a description of a new species from Cyrenaica, Libya. Acta Chiropterologica, 6:193–217. Bradley, R. D., and R. J. Baker. 2001. A test of the genetic species concept: cytochrome­b sequences and mammals. Journal of Mammalogy 82: 960–973. Brosset, A. 1984. Chiroptères d'altitude du Mont Nimba (Guinée). Description d'une espèce nouvelle, Hipposideros lamottei. Mammalia 48: 545–555. Bogdanowicz, W., and R. D. Owen. 1998. In the Minotaur`s labyrinth: phylogeny of the bat family Hipposideridae. In Bat biology and conservation (T. H. Kunz and P. A. Racey, eds.). Smithsonian Institution Press, Washington, D.C. Fahr, J., and N. M. Ebigbo. 2003. A conservation assessment of the bats of the Simandou Range, Guinea, with the first record of Myotis welwitschii (Gray, 1866) from West Africa. Acta Chiropterologica, 5:125–141. Fenton, M. B. 1986. Hipposideros caffer (Chiroptera : Hipposideridae) in Zimbabwe: morphology and echolocation calls. Journal of Zoology, London 210: 347–353 Fitch, W. M. 1970. Distinguishing homologous from analogous proteins. Systematic Zoology 19:99–113. Francis, C. M., D. Kock, and J. Habersetzer. 1999. Sibling species of Hipposideros ridleyi (Mammalia, Chiroptera, Hipposideridae). Senckenbergiana Biologica 79:255–270. Guillén­Servent, A., C. Ibáñez, and J. Juste. 2000. Variation in the echolocation calls of

22 Hipposideros ruber in the Gulf of Guinea: An exploration of the adaptive meaning of the constant frequency value in Rhinolophoid bats. Journal of Evolutionary Biology 13:70–80. Guillén­Servent, A. and C. M. Francis. 2006. A new species of bat of the Hipposideros bicolor group (Chiroptera: Hipposideridae) from Central Laos, with evidence of convergent evolution with Sundaic taxa. Acta Chiropterologica, 8:39–61. Harrison, R. G. 1989. Animal mitochondrial DNA as a generic marker in population and evolutionary biology. Trends in Ecology and Evolution 4:6–11. Hayman, R. W. and J. E. Hill. 1971. Part 2. Order Chiroptera. In The mammals of Africa: an identification manual (Meester, J., and H. W. Setzer, eds.). Smithsonian Institution Press, Washington, D.C. Heller, K.­G. 1992. The echolocation calls of Hipposideros ruber and Hipposideros caffer. In Prague Studies in Mammalogy (Horáček, I. and V. Vohralík, eds.). Charles University Press, Prague. Hennig, W. 1950. Grundzüge einer Theorie der phylogenetischen Systematik. Deutscher Centralverlag, Berlin. Hill, J. E. 1963. A revision of the genus Hipposideros. Bulletin of the British Museum (Natural History), Zoological Series 11:1–129. Hill, J. E. 1982. A review of the leaf­nosed bats Rhinonycteris, Cleotis and Triaenops (Chiroptera: Hipposideridae). Bonner Zoologische Beiträge 33:165–186. Hillis, D. M., C. Moritz, and B. K. Mable.1996. Molecular Systematics. Sinauer Associates, Inc., Sunderland, Massachusetts. Hoofer, S. R., and R. A. Van Den Bussche. 2003. Molecular phylogenetics of the chiropteran family Vespertilionidae. Acta Chiropterologica 5(supplement):1–63. Hulva, P., I. Horáček, P.P. Strelkov, and P. Benda. 2004. Molecular architecture of Pipistrellus pipistrellus/Pipistrellus pygmaeus complex (Chiroptera: Vespertilionidae): further cryptic species and Mediterranean origin of the divergence. Molecular Phylogenetics and Evolution 32:1023–1035. Hulva, P., P. Benda, V. Hanák, A. Evin, and I. Horáček. 2007. New mitochondrial lineages within the Pipistrellus pipistrellus complex from Mediterranean Europe. Folia Zoologica 56:378–388. Hutcheon, J. M., Kirsch, J. A. W., Pettigrew, J .D. 1998. Base­compositional biases and

23 the bat problem III. The question of microchiropteran monophyly. Transaction of the Royal Society of London B 353:607–617. Ibáñez, C., J. L. García­Mudarra, M. Ruedi, B. Stadelmann, and J. Juste. 2006. The Iberian contribution to cryptic diversity in European bats. Acta Chiropterologica 8:277–297. Irwin, D. M., T. D. Kocher, and A. C. Wilson. 1991. Evolution of the cytochrome b gene of mammals. Journal of Molecular Evolution 32:128–144. Jones, G., M. Morton, P. M. Hughes, and R. M. Budden. 1993. Echolocation, flight morphology and foraging strategies of some West African hipposiderid bats. Journal of Zoology, London 230:385–400. Jones, G., and K. E. Barlow. 2004. Cryptic species in echolocating bats. In Thomas, J. A., C. F. Moss, and M. Vaters (eds) Echolocation in bats and dolphins. university of Chicago Press, Chicago, Illinois. Kiefer, A., F. Mayer, J. Kosuch, O. von Helversen, and M. Veith. 2002. Conflicting molecular phylogenies of European long­eared bats (Plecotus) can be explained by cryptic diversity. Molecular Phylogenetics and Evolution 25:557–566. Kimura, M. 1968. Evolutionary rate at the molecular level. Nature 217:624–626. Kingston, T., M. C. Lara, G. Jones, Z. Akbar, T. H. Kunz, and C. J. Schneider. 2001. Acoustic divergence in two cryptic Hipposideros species: a role for social selection? Proceedings of the Royal Society of London 268:1381–1386. Kirsch, J. A. W. 1996. Bats are monophyletic, megabats are monophyletic; but are microbats also? Bat Research News 36:78. Kocher, D. T., W. K. Thomas, A. Meyer, and D. M. Irwin. 1989. Dynamics of mitochondrial DNA evolution in mammals: amplification and sequencing with conserved primers. Proceedings of the National Academy of Sciences 86:6169–6200. Koopman, K. F. 1994. Chiroptera: systematics. In Handbuch der Zoologie, Vol. 8, Mammalia, Part 60. Walter de Gruyter, Berlin. Legendre, P., and F.­J. Lapointe. 1995. Matching behavioral evolution to brain morphology. Brain, Behavior and Evolution 45:110–121. Lincoln, R., G. Boxshall, P. Clark 1998. A dictionary of ecology, evolution and systematics, 2nd ed. Cambridge University Press, Cambridge. Linnaeus, C. 1758. Systema Naturae per Regna Tria Naturae, Secundum Classes, Ordines, Genera, Species cum Caracteribus, Differentiis, Synonymis, Locis. Editio Decima,

24 Reformata, Vol. I., Regnum Animale. Laurentii Salvii, Stockholm. Lopez, J. V., M. Culver, J. C. Stephens, W. E. Johnson, and S. J. O’Brien. 1997. Rates of nuclear and cytoplasmic mitochondrial DNA sequence divergence in mammals. Molecular Biology and Evolution 14:277–286. Macaulay, V., M. Richards, and B. Sykes.1999. Mitochondrial DNA recombination – no need to panic. Proceedings of the Royal Society of London Series B 266:2037–2039. Madsen, O., Scally, M., Douady, Ch. J., Kao, D. J., DeBry, R. W., Adkins, R., Amrine, H. M., Stanhope, M. J., de Jong, W. W., and Springer, M. S. 2001. Parallel adaptive radiations in two major clades of placental mammals. Nature 409:610–614. Mayer, F., and O. von Helversen. 2001a. Sympatric distribution of two cryptic bat species across Europe. Biological Journal of the Linnean Society 74:365–374. Mayer, F., and O. von Helversen. 2001b. Cryptic diversity in European bats. Proceeding of the Royal Society of London B 268:1825–1832. Mayr, E. 1942. Systematics and origin of species. Columbia University Press, New York. Mayr, E. 1996. What is a species, and what is not?. Philosophy of Science 63:262­277. Mullis, K., F. Faloona, S. Scharf, R. Saiki, G. Horn, and H. Erlich. 1986. Specific enzymatic amplification of DNA in vitro: The polymerase chain reaction. Cold Spring Harbor Symposium on Quantitative Biology 51:263–273. Mullis, K., and F. Faloona. 1987. Specific synthesis of DNA in vitro via a polymerase catalyzed chain reaction. Methods in Enzymology 155:335–350. Murphy, W. J., E. Eizirik, W. E. Johnson, Y. P. Zhang, O. A. Ryder, and S. J. O’Brien. 2001. Molecular phylogenetics and the origins of placental mammals. Nature 409: 614– 618. Nikaido, M., M. Harada, Y. Cao, M. Hasegava, and N. Okada. 2000. Monophyletic origin of the order Chiroptera and its phylogenetic position among mammalia, as inferred from the complete sequence of the mitochondrial DNA of a Japanese megabat, the Ryuku flying fox (Pteropus dasymallus). Journal of Molecular Evolution 51:318–328. Nikaido, M., K. Kawai, Y. Cao., M. Harada, S. Tomita, N. Okada, and M. Hasegawa. 2001. Maximum likelihood analysis of the complete mitochondrial genomes of eutherians and a reevaluation of the phylogeny of bats and insectivores. Journal of Molecular Evolution 53:508–516. Nishihara, H., M. Hasegawa, and N. Okada. 2006. Pegasoferae, an unexpected mammalian

25 clade revealed by tracking ancient retroposon insertions. Proceedings of the National Academy of Science 103: 9929–9934. Nowak, R. M. 1999. Walker's mammals of the world, 6th ed. Johns Hopkins University Press, Baltimore. Pettigrew, J. D. 1986. Flying primates? Megabats have the advanced pathways from eye to midbrain. Science 231: 1304–1306. Pettigrew, J. D., B. G. Jammieson, S. K. Robson, L. S. Hall, K. I. McAnally, H. M. Cooper. 1989. Phylogenetic relations between microbats, megabats and primates (Mammalia: Chiroptera and Primates). Philosophic Transaction of the Royal Society of London B 325: 489–559. Pettigrew, J. D., 1994. Flying DNA. Current Biology 4:277–280. Pettigrew, J. D., and J. A. W. Kirsch. 1995. Flying primates revisited – DNA hybridization with fractionated, GC­enriched DNA. South African Journal of Science 91:477–482. Pettigrew, J. D., and J. A. W. Kirsch. 1998. Base­compositional biases and the bat problem. I. DNA­hybridization melting curves based on AT­ and GC­enriched tracers. Philosophic Transaction of the Royal Society of London B 353:369–379. Pye, J. D. 1972. Bimodal distribution of constant frequencies in some hipposiderid bats (Mammalia: Hipposideridae). Journal of Zoology, London 166:323–335. Ranivo, J., and S. M. Goodman. 2006. Révision taxinomique des Triaenops malgaches (Mammalia: Chiroptera: Hipposideridae). Zoosystema 28:963–985. Rosevear, D. R. 1965. The bats of West Africa. British Museum of Natural History, London, xvii + 418 pp. Russel, A. L., J. Ranivo, P. Palkovacs, S. M. Goodman, and A. D. Yoder. 2007. Working at the interference of phylogenetic and population genetics: a biogeographical analysis of Triaenops spp. (Chiroptera: Hipposideridae). Molecular Ecology 16:839–851. Russell, A. L., S. M. Goodman, and M. P. Cox. 2008. Coalescent analyses support multiple mainland­to­island dispersals in the evolution of Malagasy Triaenops bats (Chiroptera: Hipposideridae). Journal of Biogeography 35:995–1003. Salemi, M., and A.­M. Vandamme. 2003. The phylogenetic handbook. Cambridge University Press, Cambridge. Scally, M., O. Madsen, C. J. Douady, W. W. de Jong, M. J. Stanhope, and M. S. Springer. 2001. Molecular evidence for the major clades of placental mammals. Journal of

26 Mammalian Evolution 8:239–277. Simmons N. B. 1998. A reappraisal of inrefamilial relationship of bats. In: Kunz, T. H., and P. A. Racey (eds), Bats: phylogeny, morphology, echolocation, and conservation biology. Smithsonian Institution Press, Washington. pp. 3–26 Simmons, N. B. 2005. Order Chiroptera. In Mammal species of the world: a taxonomic and geographic reference, 3rd ed (D. E. Wilson and D. M. Reeder, eds.). Johns Hopkins University Press, Baltimore, Maryland. Simmons, N., and J. H. Geisler. 1998. Phylogenetic relationships of Icaronycteris, Archaeonycteris, Hassianycteris, and Palaeochiropteryx to extant bat lineages, with comments on the evolution of echolocation and foraging strategies in Microchiroptera. Bulletin of the American Museum of Natural History 235:1–182. Simpson, G. G. 1945. The principles of classification and a classification of mammals. Bulletin of the American Museum of Natural History 85:1–350. Spitzenberger, F, J. Piálek, and E. Haring. 2001. Systematics of the genus Plecotus (Mammalia, Vespertilionidae) in Austria based on morphometric and molecular investigations. Folia Zoologica 50:161–172 Springer, M.S., G. C. Cleven, O. Madsen, W. W. de Jong, V. G. Waddell, H. M. Amrine, and M. J. Stanhope. 1997. Endemic African mammals shake the phylogenetic tree. Nature 388:61–64. Springer, M. S., E. C. Teeling, O. Madsen, M. J. Stanhope, and W. W. de Jong. 2001. Integrated fossil and molecular data reconstruct bat echolocation. Proceedings of National Academy of Sciences 98:6241–6246. Stanhope, M. J., O. Madsen, V. G. Waddell, G. C. Cleven, W. W. de Jong, and M. S. Springer. 1998. Highly congruent molecular support for a diverse superordinal clade of endemic African mammals. Molecular Phylogenetics and Evolution 9:501–508. Tate, G. H. H. 1941. Results of the Archbold Expeditions. No. 35. A review of the genus Hipposideros with special reference to Indo­Australian species. Bulletin of the American Museum of Natural History 78: 353–393. Teeling, E. C., M. Scally, D. J. Kao, M. L. Romagnoll, M. S. Springer, and M. J. Stanhope. 2000. Molecular evidence regarding the origin of echolocation and flight in bats. Nature 403:188–192. Teeling, E. C., O. Madsen, R. A. Van Den Bussche, W. W. de Jong, M. J. Stanhope, and

27 M. S. Springer. 2002. Microbat paraphyly and convergent evolution of a key innovation in Old World rhinolophid microbats. Proceedings of the National Academy of Sciences 99:1431– 1436. Teeling, E. C., M. S. Springer, O. Madsen, P. Bates, S. J. O’Brien, and W. J. Murphy. 2005. A molecular phylogeny for bats illuminates biogeography and the fossil record. Science 307:580­584. Thabah, A., S. J. Rossiter, T. Kingston, S. Zhang, S. Parsons, K. Mya Mya, A. Zubaid, and G. Jones. 2006. Genetic divergence and echolocation call frequency in cryptic species of Hipposideros larvatus s.l. (Chiroptera: Hipposideridae) from the Indo­Malayan region. Biological Journal of the Linnean Society 88: 119–130. Vallo, P., A. Guillén­Servent, P. Benda, D. B. Pires, and P. Koubek. 2008. Variation of mitochondrial DNA in the Hipposideros caffer complex (Chiroptera: Hipposideridae) and its taxonomic implications. Acta Chiropterologica in press. Van Cakenberghe, V., and E. C. J. Seamark. 2008. African Chiroptera Report 2008. African Chiroptera Project, Pretoria. Van Den Bussche, R.A., R. J. Baker, J. P. Huelsenbeck, and D. M. Hillis. 1998. Base compositional bias and phylogenetic analyses: a test of the “flying DNA” hypothesis. Molecular Phylogenetics and Evolution 13:408–416. Waddell, P. J., Y. Cao, M. Hasegawa, and D. P. Mindell. 1999. Assessing the Cretaceous superordinal divergence times within birds and placental mammals by using whole mitochondrial protein sequences and an extended statistical framework. Systematic Biology 48:119–137. Wang, H., B. Liang, J. Feng, L. Sheng, and S. Zhang. 2003. Molecular phylogenetic of Hipposiderids (Chiroptera: Hipposideridae) and Rhinolophids (Chiroptera: Rhinolophidae) in China based on mitochondrial cytochrome b sequences. Folia Zoologica 52:259–268. Woese, C. R., and G. E. Fox. 1977. Phylogenetic structure of the prokaryotic domain: The primary kingdoms. Proceedings of the National Academy of Sciences 74:5088–5090.

28 Variation of mitochondrial DNA in the Hipposideros caffer complex (Chiroptera: Hipposideridae) and its taxonomic implications

1,2 3 4,5 6 1 PETER VALLO , ANTONIO GUILLÉN­SERVENT , PETR BENDA , DEBRA B. PIRES , and PETR KOUBEK

1Institute of Vertebrate Biology, v. v. i., Academy of Sciences of the Czech Republic, Květná 8, 603 65 Brno, Czech Republic; E­mail: [email protected]* 2Institute of Botany and Zoology, Faculty of Science, Masaryk University, Kotlářská 2, 611 37 Brno, Czech Republic 3Instituto de Ecología, A. C., km 2.5 Ctra. Antigua a Coatepec #351, Congregación el Haya, 91070 Xalapa, Veracruz, México 4Department of Zoology, National Museum (Natural History), Václavské náměstí 68, 115 79 Praha 1, Czech Republic 5Department of Zoology, Faculty of Science, Charles University, Viničná 7, 128 44 Praha 2, Czech Republic 6Department of Life Sciences, University of California, 621 Charles E. Young Drive South, Los Angeles, California 90095–1606, USA

*Corresponding author: Peter Vallo, Institute of Vertebrate Biology AS CR, v.v.i., Květná 8, 603 65 Brno, Czech Republic; phone +420543422543, fax +420543211346; e­mail [email protected]

Key words: Africa, Hipposideros caffer, H. ruber, leaf­nosed bats, cryptic species, cytochrome b, molecular systematics, phylogeny

Running title: Variation of mtDNA in the Hipposideros caffer complex

Abstract The Afrotropical leaf­nosed bat Hipposideros caffer has been traditionally regarded as a complex of populations, currently pertaining to two recognized cryptic species, H. caffer and H. ruber. Extent of distribution and morphological variation of these bats has raised concerns on whether the current perception of the complex reflects true phylogenetic

29 relationships and taxonomic diversity. Our phylogenetic analysis of nucleotide sequences of the mitochondrial cytochrome b gene challenged the hypothesis of two cryptic species. Instead of the two reciprocally monophyletic lineages expected, corresponding to the two species, we recovered four distinct lineages with deep internal divergences. Two sister clades within a lineage of bats of H. caffer represent respectively the nominotypical form H. c. caffer, restricted to Southern Africa, and H. c. tephrus, inhabiting Maghreb and the Arabian peninsula. Geographical isolation and deep genetic divergence suggest species statuses for these two forms. Another lineage is comprising specimens of both morphotypes from West and East Africa. It probably represents a distinct species but its taxonomic assignation remains obscure. A Central African lineage of H. ruber comprises two sister clades, which become sympatric in Cameroon. Their status has to be clarified with additional evidence, since nuclear gene flow may be taking place. A further divergent lineage with H. ruber morphotype, most probably representing another distinct species, is restricted to West Africa. Although all three genetic forms of H. ruber may correspond to named taxa, their proper taxonomic assignation has to be assessed by comparison with type material.

Introduction Flight and echolocation constraint bat morphology in a way that parallel and convergent evolution can be widespread among species that use similar ecological niches (Norberg 1994, Ruedi and Mayer 2001). This may make taxonomy difficult, since species evolutionarily related and ecologically similar may lack conspicuous morphological characters useful to discriminate among them. Genetic data may be fundamental in these cases, allowing the identification of deeply divergent lineages that may represent evolutionary independent units (Bradley and Baker 2001). Extensive use of molecular phylogenetic methods has contributed in recent years to reveal many new cryptic forms of bats within traditionally recognized species. The recent increase of about 20% in the number of species of the European fauna of bats, probably the best known Chiropteran fauna in the World, is an illuminating example of the utility of the genetic data to screen bat hidden taxonomic diversity (Mayer and Helversen 2001, Ibáñez et al. 2006, Mayer et al. 2007). The genus Hipposideros (Gray, 1831), the largest in the Palaeotropical family of

30 the leaf­nosed bats, Hipposideridae, has traditionally had a difficult taxonomy due to the extreme morphological similarity of many of its members (Hill 1963, Bogdanowicz and Owen 1998). This morphological similarity suggests that crypticity might be particularly common. Systematic biologists have so far paid most attention to the Southeast Asian species of Hipposideros, where a high number of cryptic forms have been revealed recently, often by using molecular data (e.g. Francis et al. 1999, Kingston et al. 2001, Guillén­Servent and Francis 2006, Thabah et al. 2006, Bates et al. 2007). African leaf­ nosed bats have not been surveyed using molecular methods. However, the taxonomy of many African forms is still unclear, and cryptic forms may eventually appear. Particularly interesting in relation to cryptic diversity is the group of forms related to the species H. caffer (Sundevall, 1846). Bats in the Hipposideros caffer complex (= H. caffer s. l.) are among the most abundant mammals in Africa (Brosset 1984). According to Hill (1963) these bats belong to the African lineage of the H. galeritus subgroup within the H. bicolor group, together with several other similarly looking African species: H. beatus (Allen, 1917), H. fuliginosus (Temminck, 1853), H. curtus (Allen, 1921), and H. lamottei (Brosset, 1984). Rosevear (1965) and Koopman (1975) considered the group caffer­ruber­fuliginosus­beatus as a closely related complex of morphologically similar species with uncertain systematics and phylogenetic relationships. H. caffer was still regarded as only one species in Hill's (1963) revision of the genus Hipposideros, where the large size differences among populations were regarded as largely clinal. Hollister (1918) suggested that bats under the name of H. caffer belonged to two different species in East Africa, but this was generally ignored by subsequent authors. However, after Hayman and Hill (1971), two cryptic species are generally recognized within the complex: H. caffer originally described from Natal near Durban, South Africa, and H. ruber (Noack, 1893) originally described from the region of Ngerengere river, Tanzania. As currently understood, H. caffer ranges in arid ecosystems and seasonally dry open forests throughout most Africa and south­western Arabia, being absent from the core of the rainforest belt. H. ruber, is restricted to the rainforest belt and the wet forested savannas of sub­Saharan Africa and still absent from the drier eastern and southernmost regions of the continent. Both forms are present in a wide zone of sympatry in the open forests and savannahs surrounding the rainforest belt (Hayman and Hill 1971, Simmons 2005). The two forms can generally be discriminated throughout much of their

31 range by their size (e.g. forearm or condylocanine lenght), H. caffer being smaller than H. ruber (Hayman and Hill 1971, Koopman 1975), and by differences in the form of nasal inflations (Lawrence 1964, Kock 1969). Echolocation frequency also seems to be a distinctive feature; when recorded in sympatry, the smaller form generally has a distinctively higher frequency of the constant frequency element of the calls than the larger form (Fenton 1986, Heller 1992, Jones et al. 1993). Nevertheless, species determination is not so straightforward, since these apparently clear interspecific boundaries in morphology and echolocation become less obvious when confronted with many described forms of various geographic origin. Six forms of the complex are currently recognized as subspecies of the two respective species by Simmons (2005): H. c. caffer, H. c. tephrus (Cabrera, 1906), H. c. nanus (Allen, 1917), H. c. angolensis (Seabra, 1898), H. r. ruber, and H. r. guineensis (Andersen, 1906). In the past, however, H. c. nanus was either synonymised with H. beatus (Hill 1963) or with H. c. tephrus (Lawrence 1964), and H. r. ruber was either substituted by (Hill 1963, Koopman 1975) or supplemented with (Hayman and Hill 1971) two other subspecies, H. r. centralis Andersen, 1906 and H. r. niapu (Allen, 1917). The large morphological variability reflected in the numerous forms described within the two recognized species and the uncertain relationships among them allow the suspicion that still more cryptic species exist within the H. caffer complex (Koopman 1994, Fahr and Ebigbo 2003, Simmons 2005). The systematics and taxonomy of this complex of forms would definitely benefit from the molecular­genetic approach, which could contribute to reveal the phylogenetic relationships and focus further work with morphological data. In this study we took the first step towards decrypting this complex through phylogenetic analysis of sequences of the mitochondrial gene for cytochrome b, a marker that has proved its suitability for exploring intra­ and interspecific relationships in many bats (Bradley and Baked 2001). We assessed the genetic structure within the H. caffer complex, and compared its relationship to current taxonomic and systematic classification. In other words, we asked if two reciprocally monophyletic lineages corresponding to the two morphotypes currently recognized as species exist, or whether more cryptic diversity may be present within the complex.

Material and methods Samples of Hipposideros were obtained from various regions of the African

32 continent, including the Maghreb, and from southwestern Arabia (Appendix 1, Figure 1). Besides forms of the H. caffer complex, several other African congeneric species, i.e. H. cyclops (Temminck, 1853), H. gigas (Wagner, 1845), H. jonesi Hayman, 1947, H. abae Allen, 1917, and the closely related H. beatus Andersen, 1906 and H. fuliginosus (Temminck, 1853), were included to improve reconstruction and interpretation of the phylogeny. Either tissue samples from collected specimens or wing membrane biopsies were used for molecular analysis. The nominal species were initially determined on basis of morphological traits according to Hayman and Hill (1971) and Koopman (1975), and their nomenclature followed Simmons (2005). In order to obtain a preliminary idea of the correspondence between genetic lineages and morphology, we measured with a mechanical caliper the forearm length (FL), condylocanine length (CCL) and zygomatic width (ZW) of the available voucher specimens genotyped for the cytochrome b gene. We also obtained measurements for some other specimens that were assigned to lineages by other means. In particular, this applied to a set of nine specimens from Cameroon, previously genotyped for the mitochondrial control region by one of the authors (DBP). These specimens could be unambiguously matched with the cytochrome b lineages revealed here, since some specimens were sequenced for both genes and the mitochondrial genome is inherited as a single unit without recombination (Avise 2004). Measurements of the genotyped South African specimens were also unavailable, thus, we used measurements from three specimens from the same region to represent this form. Following Hayman and Hill (1971) and Koopman (1975), specimens were initially assigned to the H. caffer morphotype when their FL was less than 48 mm and their CCL less than 15.5 mm, while those with larger measurements were assigned to the H. ruber morphotype. We used a plot of ZW against CCL in order to visualize the morphological variation within and among lineages. Total genomic DNA was extracted from alcohol preserved tissue with DNeasy Tissue Kit (Qiagen) or Puregene Tissue Kit (Gentra Systems) according to the manufacturer`s protocols. Complete cytochrome b gene (cytb) was amplified by polymerase chain reaction using universal primers L14724 and H15915 (Irwin et al. 1991) or their slightly modified versions L14724ag and H15915ag (Guillén­Servent and Francis 2006). Each 50 μ l reaction contained 0.8 μ M of each primer, 0.2 mM dNTP, 1U of HotMaster Taq DNA polymerase (Eppendorf) or Taq DNA polymerase (Promega),

33 corresponding buffer, and ca. 50­100 ng template DNA. PCR was performed in 35 cycles of denaturation for 40 s at 94°C, annealing for 40 s at 50°C, extension for 90 s at 65°C (HotStart polymerase) or 72°C (Promega Taq), preceded by 3 min initial denaturation at 94°C and terminated by 5 min final extension at 65°C or 72°C. Obtained product ca. 1.2 kb long was purified with QIAquick PCR Purification Kit (Qiagen) or GeneClean Purification Kit (Bio 101) and sequenced in both directions with the same primers using BigDye Terminator sequencing chemistry (Applied Biosystems) on ABI 3730xl and 310 sequencers. Sequences were assembled, checked by eye, and edited in Sequencher v.4.6 (Gene Codes) and BioEdit v.7.0.1 (Hall 1999), and submitted to the GenBank database under accession numbers EU934446–EU934485 and EUWaiting–EUWaiting (Appendix 1). Alignment of sequences was created in BioEdit v.7.0.1. Only haplotypes were included in our analysis. Redundant sequences were not further regarded because of unequal geographic sampling. A sequence of Rhinolophus landeri (P. Vallo, unpublished) was used as an outgroup for rooting phylogenetic trees. Phylogenetic relationships were first inferred under maximum parsimony (MP) criterion using PAUP* v.4.10b (Sinauer Associates). Due to strong bias in transitions, which is a usual feature of vertebrate mitochondrial genes, transversions were weighed 5:1 against transitions, corresponding to the inverse of the transition bias calculated empirically by maximum likelihood with PAUP* onto an initial MP tree. Heuristic search was conducted 100 times with random addition of sequences and a tree bisection­reconnection (TBR) swapping algorithm for rearrangement of branches. Confidence of the resulting topology was assessed by 1000­ replicate bootstrapping (Felsenstein 1985). Phylogeny was further inferred under maximum likelihood (ML) using PhyML (Guindon and Gascuel 2003), which enabled heuristic search of 100 sequence addition replicates and TBR swapping in a reasonably short time. A Tamura­Nei model of evolution (Tamura and Nei 1993) with gamma distributed rate across sites (Yang 1996) (TN93+ Γ ) was used as the best model determined by Modeltest 3.7 (Posada and Crandall 1998) under the Akaike Information Criterion (AIC). Tree support was assessed by 1000 bootstrap replicates. Phylogenetic relationships were also inferred under a Bayesian approach using MrBayes 3.1.2 (Ronquist and Huelsenbeck 2003) in a combined analysis of sequence partitions corresponding to the three codon positions of cytb. General time­reversible model (Tavaré 1986) with a

34 proportion of invariant sites and gamma distributed among­site rate variation (GTR + I + Γ) was selected as the appropriate model for each partition under the AIC as implemented in MrModeltest 2.2 (Nylander 2004). The tree used for calculation of likelihoods under the different models was the shortest obtained under weighted parsimony criterion with transversions weighted 5 times over transitions. Model parameters were unlinked so that each partition had its own set of independent estimates, and the three partitions were allowed to evolve under different rates. Two simultaneous Bayesian analyses were run for 106 generations with one cold and three incrementally heated Metropolis­coupled Markov chains with default heating values starting from random trees. Trees were sampled every 100 generations. First 10000 generations, as determined by checking the convergence plot of likelihood values, were discarded as burn­in. Results were summarized as a 50% majority rule consensus tree from both runs with posterior probabilities as estimators of confidence (Ronquist and Huelsenbeck 2003). Sequence divergence was expressed as percentual Kimura (1980) two­parameter (K2P) genetic distance.

Results Field sampling and museum loans rendered a collection of 49 tissue samples of the H. caffer complex originating from localities widely distributed throughout the African continent and the Arabian Peninsula. A large part of the distribution of the complex was covered, although some wide areas remained unsampled (especially the southwestern savannah areas; Figure 1). Based on morphology, twenty specimens were assigned to H. caffer and twenty­nine to H. ruber, respectively (Appendix 1, Fig. 2). Morphotype assignment was not congruent in several occasions. In those cases, CCL and geographical origin of the specimen were considered for discrimination (e.g. specimens from Swaziland, Yemen, and Senegal). Three loose groups of specimens could be visually identified in the ZW­CCL bivariate graph. A group of four specimens from Senegal and one from Yemen, all with the H. caffer morphotype, showed the lower values of both ZW and CCL. A second group of bats showed intermediate values of ZW (9.2–9.6 mm) and CCL (15.0– 15.5 mm). This group included specimens from Ivory Coast, Senegal, South Africa and Tanzania assigned to the H. caffer morphotype. However it also included some specimens with CCL smaller than 15.5 mm from the region of Gulf of Guinea, where H. caffer is absent, that only differed in size from other syntopic specimens morphologically assigned

35 to H. ruber. Some specimens from Morocco, South Africa and Yemen showed measurements intermediate between these two groups. A large group of specimens from diverse origin (Benin, Cameroon, Democratic Republic of Congo, islands of Bioko, Príncipe and São Tomé, Ivory Coast, mainland Equatorial Guinea, Malawi, and Senegal) showed CCL values above 15.8 mm and ZW values above 9.7 mm. Two specimens from Ivory Coast and the Democratic Republic of Congo were considerably larger, with CCL values above 17 mm (Appendix 1, Fig. 2). Our dataset for molecular phylogenetic analysis included sixty­three haplotypes (1140 bp) of leaf­nosed bats, of which forty­nine belonged to the H. caffer complex (Appendix 1). In the alignment of sequences, 446 characters were variable and 403 parsimony informative. Overall base composition A=0.284, C=0.318, G=0.142, T=0.256 was similar to other species of hipposiderid bats with a low proportion of guanine residues and no significant differences among taxa (χ2=33.875, d.f.=189, P=1). Substitutions occurred mostly at third codon positions (76.7%), followed by first (19.5%) and second (3.8%) codon positions. Saturation plot of p distances against K2P distances showed a deflection from linearity on the third codon transitions only. Sequence divergence among Hipposideros species excluding the H. caffer complex ranged from 10.2% to 20.2% (Table 1) and reached up to 5.1% within species. Divergences between specimens in the H. caffer complex and other species ranged from 7.9% to 20.4%. Within the H. caffer complex divergences were up to 11.3% (Table 1). Transversion weighted parsimony rendered sixteen shortest trees of 1431 steps, with a retention index of 0.792 and a consistency index of 0.433. Maximum likelihood and Bayesian analysis yielded trees with topology similar to MP trees with most of the major clades well supported by bootstrap and posterior probability values (Figure 3). Within the major clades, additional diversity was observed. H. cyclops and H. gigas joined in one clade branching out from the most basal node of the phylogenetic trees, showing a deep split of 16.8–17.0% sequence divergence between the two species. The H.cyclops­H.gigas clade differed at 15.4–20.4% sequence divergence from its sister clade, which comprised all the remaining African Hipposideros. H. jonesi diverged from the next node, having as sister a large clade comprising all haplotypes of the H. caffer complex and the closely related species H. beatus and H. fuliginosus. H. abae was also included in this clade. This crown clade split in seven lineages with high support under the three phylogenetic

36 optimization criteria assayed (weighted parsimony bootstrap > 92%, ML bootstrap > 89%, Bayesian posterior probability > 0.99). Phylogenetic relationships among the seven main lineages within this large clade were not resolved under any of the three methods. Three of these seven lineages, corresponded to the species H. abae, H. beatus and H. fuliginosus, while the other four contained haplotypes of specimens keying morphologically within the H. caffer complex (Figure 3). H. abae showed the largest genetic distance (9.6–12.8%) to other lineages in the crown group, followed closely by H. beatus (9.3–12.7%). Distances among the remaining 5 lineages, including H. fuliginosus, were very similar (7.4–11.3%). Haplotypes of H. fuliginosus showed distances to the H. caffer complex lineages intermediate within this range (7.9–10.2%; Table 1). Most haplotypes corresponding to specimens morphologically keyed as H. caffer were clustered in lineage A, which differed by 7.4–11.0% from other lineages containing haplotypes of other specimens keyed either as H. caffer or H. ruber. This lineage comprised two subclades 7.4–9.2% apart in genetic distance. Subclade A1 contained haplotypes from South Africa, Swaziland and Mozambique, while the other subclade A2 grouped haplotypes from Morocco, Senegal, and Yemen. Several haplotypes of specimens originally assigned to H. caffer grouped together with haplotypes of specimens keyed as H. ruber into lineage B. This lineage showed consistently slightly larger genetic distances (7.5–11.3%) to other lineages within the complex that these other lineages showed among them. Lineage B split in two clades 5.0–7.0% apart in genetic distance. B1 clade had a West African distribution, with haplotypes from Senegal and Benin. B2 spread in East Africa, containing haplotypes from Kenya, Tanzania and Malawi (Appendix 1 and 2, Figures 1 and 3). Specimens determined as H. ruber clustered in two main lineages. Lineage C consisted of two reciprocally monophyletic well supported subclades representing two groups of haplotypes from Central Africa, differing at 4.2–6.6% genetic distance (Figure 3). Two easternmost haplotypes, from Uganda and Democratic Republic of Congo, belonged in the subclade C1 with several haplotypes from Cameroon. A western haplotype from Ivory Coast, which had been morphologically assigned to H. caffer (Figure 2), also clustered within this subclade. More compact subclade C2 contained haplotypes from Cameroon, mainland Equatorial Guinea, and the three main islands of the Gulf of Guinea (Appendix 1). Within this lineage, haplotypes from São Tomé and Príncipe aggregate in a

37 distinct clade differing 3.4–4.5% from the mainland haplotypes. The last lineage D of specimens morphologically keyed as H. ruber appeared as paraphyletic to lineage C in all the analyses. A genetic distance of 7.5–9.7% separated these two lineages. Lineage D contained haplotypes from West Africa grouped in three clades corresponding to their geographic origin in Senegal, Benin and Ivory Coast (Appendix 1, Table 1, Figures 1 and 3).

Discussion The Hipposideros caffer complex is currently understood as consisting of two species, H. caffer and H. ruber. This taxonomic structure anticipated the expectation of two reciprocally monophyletic mitochondrial lineages, each corresponding to one of these two recognized species. Instead of that, our phylogenetic analyses revealed four different lineages with haplotypes of the H. caffer complex splitting off from an unresolved polytomy involving three other lineages corresponding to the species H. abae, H. beatus, and H. fuliginosus. Furthermore, three of the four lineages detected in the H. caffer complex showed deep internal divergences associated with geographical phylogroups. These results suggest that previous perceptions of this complex underestimated the diversity of these bats and that the traditional division into two cryptic species, H. caffer and H. ruber, does not reflect its real complexity throughout the African continent. Although our sampling did not cover the whole ranges of both species, nor were all topotypes of taxonomically important forms included, the results suggest at least a partial re­evaluation of the phylogeny and taxonomy of the complex. Genetic divergence in mitochondrial genes varies widely among species. Due to the matrilineal nature of the inheritance of the mitochondrial genome, relatively deep divergences do not necessarily correspond to species boundaries, since significant nuclear gene flow may occur among very divergent mitochondrial phylogroups (Avise 2004). However, among species average divergences in mitochondrial genes are larger than maximum intraspecific divergences. A recent comprehensive study by Baker and Bradley (2006) documented a large interval of 2.3–14.7% uncorrected genetic distance between sister species of bats, and distances up to 5.9% within species. Asian cryptic species of Hipposideros show interspecific divergences from as low as 3.9% (Guillén­Servent and Francis 2006) to as high as 13.4% (Thabah et al. 2006), naturally including the

38 intermediate value of 6.5% (Kingston et al. 2001). Wang et al. (2003) documented K2P sequence divergences of 8.6–13.4% among Asian species of Hipposideros. K2P divergences among lineages of bats in the H. caffer complex and their closest relatives H. beatus and H. fuliginosus ranged 7.9–11.3%, being in the medium­upper interval of the range of divergences found among species in the genus. These deeply divergent lineages probably represent different cryptic species, however their taxonomic status should be further clarified by other lines of evidence. Since description of H. caffer in 1846, many forms have been added under this name. Except for H. ruber, they either keep a subspecific status or were synonymised with one of the presently recognized subspecies. Our results indicate that the South African lineage A1 of H. caffer, comprising haplotypes from South Africa, Swaziland and Mozambique, represents the nominotypical form of the H. caffer complex. It is actually the only natural unit in our dataset, for which the species name H. caffer is legitimate. The lineage A1 most probably corresponds to a species­level taxon associated with the type of H. c. caffer, and remains restricted to southern Africa. Thus, traditional assignment of bats from West and northeastern Africa to subspecies H. c. caffer (Hill 1963, Hayman and Hill 1971, Kock 1969, Koopman 1975, 1994, Rosevear 1965) probably does not reflect real phylogenetic relationship to this form. However, assignation of bats from regions not covered by our sampling where H. caffer is supposedly present, e.g. Zimbabwe and Botswana (Koopman 1994, Taylor 2001), has yet to be tested. The sister lineage to this South African clade, denoted as A2, corresponds to the currently recognized subspecies H. caffer tephrus, and comprises two groups of haplotypes from Morocco and Senegal, and Yemen, respectively. The morphological assignation of populations of small­sized caffer­like bats from Senegal and Yemen to the subspecies H. c. tephrus by Adam and Hubert (1972), and Nader (1982), respectively, proved right in terms of their close phylogenetic relationship with the true tephrus form, originally described from Morocco. A separate species status for the two sister lineages caffer (A1) and tephrus (A2), is supported by the large sequence divergence of 7.4–9.1% between them and the wide geographical separation of more than 3000 km with no presence of either lineage in the equatorial African belt. Western and eastern populations of H. c. tephrus show a rather high sequence divergence of 3.4–4.0%, what may be due to a long geographical isolation. However, this divergence could alternatively be due to isolation by distance, since the

39 tephrus form could be continuously present in the poorly surveyed sub­Saharan dry forests and savannahs north of the rainforest belt. A broader sampling from these areas, including e.g. Sudan, where occurrence of H. c. tephrus is stated by Kock (1969) and Koopman (1975), would therefore be necessary to understand the processes underlying the genetic pattern. Another set of specimens of the H. caffer morphotype in our dataset joined together with some specimens keyed as H. ruber in the lineage denoted as B. This lineage differed by 7.5–11.3% from other lineages of the H. caffer complex. Several specimens of this lineage lie at the interspecific boundary between H. caffer and H. ruber when considering the supposedly discriminating condylocanine length threshold of 15.5 mm (Koopman 1975), while other fall well inside the H. ruber range. The geographical distribution and the intermediate size of the specimens in this lineage suggest that they represent yet another cryptic species within the H. caffer complex. West African (B1) and East African (B2) subclades within this lineage differ by 5.0–7.0%, a divergence that could be explained by long term geographical isolation if the lineage is absent from the intervening areas or by isolation by distance in the case that the lineage is present between the Sahara and the northern rim of the Central African rainforest belt. Because the original description of H. ruber was based on Tanzanian specimens collected between the localities of our lineage B2 samples from Zanzibar and Kenya, and considering that some specimens in this lineage were morphologically keyed as H. ruber, there is a possibility that lineage B corresponds taxonomically to the name of H. ruber. However, a question still remains on the correct name of this form, since specimens identified as H. caffer and H. ruber have been reported from elsewhere in Kenya, Tanzania and Malawi (Hollister 1918, Koopman 1994, Simmons 2005), and it is also possible that three forms of the H. caffer complex coexist in East Africa. This was the case of the bats from Senegal, where our analyses revealed sympatric presence of three forms of the complex (lineages A2, B, and D). A more extensive sampling of molecular haplotypes and a morphological revision of all type material available from this region is necessary to reliably answer this question. A question remains as well on the relationships of Hipposideros caffer angolensis, a form not represented in our samples and distributed in a wide area from Southern Gabon through Angola into the African Southwest (Hill 1963, Koopman 1994), to the two lineages (A1 and B2) present to the south of the rainforest belt. It would be also of much importance to assess relationships

40 of the nominotypical H. caffer to the other currently recognized subspecies, H. c. nanus, from northeastern Democratic Republic of Congo, which was neither included into our analysis. Even more interesting and complicated is the situation of the bats assigned to H. ruber in our dataset. The main lineage of specimens with this morphotype, dubbed C, was associated with the Central African rainforest belt. It contains two sublineages, C1, with samples coming from Ivory Coast, Uganda, northwestern Democratic Republic of Congo and Cameroon, and C2, with specimens coming from the islands of the Gulf of Guinea, Cameroon and mainland Equatorial Guinea. The genetic divergence between these two sublineages was substantial, 4.2–6.6%, although much lower than the divergence among the four main lineages of the H. caffer complex. Bats of these two lineages seem to be morphologically undistinguishable in the areas of sympatry in the mainland bordering the Gulf of Guinea. Specimens with haplotypes in the two clades are found sharing the same roosts at least in some areas of Cameroon (Debra Pires, unpublished). These issues cast doubts on the idea of these two clades might represent different cryptic species, although this hypothesis cannot be ruled out until further lines of evidence are examined (i.e. nuclear molecular data, echolocation frequency). However, it is quite possible that the two mitochondrial lineages originated in allopatry and became only recently sympatric in Cameroon. Nuclear gene flow may then have been re­established, such as it has been recently described for deeply divergent mitochondrial lineages in North American bat Myotis lucifugus (Lausen et al. 2008). Genetic divergence of 3.4–4.5% within the lineage C2 further suggests that the form on the islands of São Tomé and Príncipe has been isolated from the populations on the mainland for quite a long time. The last main lineage D comprised haplotypes of specimens of the H. ruber morphotype collected in forested areas of West Africa. The large divergence to other main clades (7.5–11.0%) and the sympatric occurrence together with two other morphologically distinct forms of the H. caffer complex in Senegal (lineages A2 and B1) support the ranking of this lineage as a distinct species. Lineage D is as well sympatric with lineage C1 in Ivory Coast, and with lineage B1 in Benin. Interestingly, specimens of lineage C1, showing H. ruber morphotype in the Congo basin and the Gulf of Guinea, show a H. caffer appearance in Ivory Coast. On the other hand, specimens of lineage B1 show a H. ruber appearance in Benin. This suggests that character displacement processes might

41 operate among species­level taxa of the H. caffer complex when they become sympatric. Clades from Senegal, Benin and Ivory Coast show substantial genetic differentiation (3.2– 4.4%). Two distinct forms of H. caffer s. l. were recognized from Senegal already by Adam and Hubert (1972). Lineage D mensurally corresponds with the larger form mentioned by the authors, which is currently assigned to H. r. guineensis (Simmons 2005). The name H. r. guineensis has traditionally been assigned to large bats of the H. caffer complex throughout West Africa (e. g. Adam and Huber 1972, Jones et al. 1993, Koopman 1975). However, H. r. guineensis was originally described from northeastern Gabon. Thus, this name might probably apply to the H. ruber lineage C2 of the Gulf of Guinea. If the sister lineage C1 from the Congo basin proves to be a species level taxon it might deserve the name H. r. centralis. This name was originally used for a form described from Uganda (Hill 1963, Hayman and Hill 1971), and was later subsumed by Koopman (1994) under the nominotypical subspecies H. r. ruber together with its junior synonym, H. r. niapu, described from the northeastern Democratic Republic of Congo. Interesting is as well the affinity of the specimen keyed as H. caffer from Ivory Coast to the lineage C1, which suggests the presence of this lineage throughout the rainforest belt from West to Central Africa. This seems plausible based on the close phylogenetic relationships of the geographically rather distant haplotypes, but further extensive sampling would be necessary to validate this assumption. Thus, the phylogenetic analyses of the mitochondrial molecular data combined with the geographical distribution of the samples and the simple morphometric assessment suggest the existence of at least 5 species­level taxa within the H. caffer complex, two corresponding to the H. caffer morphotype, two corresponding to the H. ruber morphotype, and one including specimens from both morphotypes. Besides, a deep divergence within the Ivory Coast­Congo basin lineage with H. ruber morphotype remains to be examined for evidence of nuclear gene flow. These suggested taxonomic implications can be confirmed only after more extensively genetic sampling throughout Africa, further morphological comparisons, and phylogenetic analyses of nuclear gene molecular data. Proper assignment of names will require the inclusion of the types or the respective topotypes in these analyses. Besides the new perception of the H. caffer complex, our analysis brought also some new information on intrageneric phylogeny of African representatives of the genus

42 Hipposideros. Current understanding of the taxonomic structure of Hipposideros still dates from Hill´s (1963) revision in a time before formal phylogenetic inference was used. Hill´s (1963) views have been followed in subsequent taxonomic revisions up to present times (e.g. Hayman and Hill 1971, Koopman 1994, Simmons 2005). The morphology based phylogenetic analyses of Bogdanowicz and Owen (1998) challenged Hill’s (1963) suprageneric grouping. However this study raised doubt on morphological characters being reliable for inferring phylogeny in this group of bats, and did not offer a well­founded alternative to the systematics of the genus. Although molecular methods have been employed for systematics of Hipposideros, the only study aimed at higher intrageneric phylogeny was that by Wang et al. (2003), focused on the Chinese species. Unfortunately, their limited sampling does not allow inferring phylogenetic relationships in a broader intrageneric perspective. Some of our results contradict the traditionally accepted intrageneric relationships in Hipposideros. First, H. abae appears in our phylogeny as very closely related to H. beatus¸ H. fuliginosus and the H. caffer complex. Although cytochrome b data were unable to resolve the relationships among these taxa, this phylogenetic position of H. abae is rather unexpected, since this species was considered by Hill (1963) a member of the distant H. speoris group of species, together with some Asian forms. These results make the H. bicolor group and the H. galeritus species sub­group sensu Hill (1963), and the Hipposideros caffer species group sensu Rosevear (1965; i. e. H. beatus, H. fuliginosus and H. caffer s.l.), paraphyletic, and the H. speoris group polyphyletic. Interesting are as well the phylogenetic positions of H. cyclops and H. gigas, the latter being a West African form taxonomically segregated from H. commersoni Geoffroy, 1813, in recent times (Simmons 2005). The clustering of H. cyclops and H. gigas in a clade originating from a basal node in our phylogenetic tree suggests an early differentiation. This conflicts with Hill´s (1963) idea of these species belonging to very distant phylogenetic lineages recently diverged from other groups of Hipposideros. The apparent close relationship among the two species also challenges the assignation of H. gigas and H. commersoni s.l., to the mostly Asian H. diadema group. Genetic differences of H. cyclops and H. gigas to other Hipposideros in this study exceed the intergeneric divergence of 17.2% between Hipposideros and Aselliscus reported by Wang et al. (2003), suggesting that the genus might be well split in several genera. Evidently, much more work is needed

43 to get out of the labyrinth in the systematic of the round leaf bats.

Acknowledgements We thank mainly Jaroslav Červený, Adam Konečný, and Josef Bryja but also other colleagues from the Institute of Vertebrate Biology AS CR, Brno, v. v. i. for assistance with the field work in Senegal. Further we are indebted to Carlos Ibáñez and Javier Juste (Doñana Biological Station, Sevilla, Spain), Peter Taylor (Durban Museum, South Africa), Julian Kerbis, John D. Phelps and Bruce Patterson (Field Museum of Natural History, Chicago), Judith Eger (Royal Ontario Museum), Richard Monk, James Gardner and Robert Baker (Texas Tech University Museum), and Teresa Kearney (Transvaal Museum, South Africa), David Jacobs and Samantha Stoffberg (University of Cape Town, South Africa), and Ana Rainho (University of Lisbon, Portugal) for providing tissue samples. Charles Francis, Suzanne McLaren, Teresa Kearney and Paul Velazco provided measurements or scaled photographs of some museum specimens. Field work and bat sampling in Senegal was approved by the Direction des Parcs Nationaux du Sénégal, Dakar. Samples of H. fuliginosus were taken by Ana Rainho under licence issued by the Instituto da Biodiversidade e Áreas Protegidas, Guinea­Bissau. Alejandro Espinosa facilitated the use of lab space and equipment to AGS, and Cristina Bárcenas helped in the electrophoresis of sequencing products at the DNA core of the Instituto de Ecología, A.C, Xalapa, México. PV also thanks Natália Martínková (IVB AS CR, Brno, v. v. i.) for introduction into molecular­genetic methods and helpful discussions on data analysis. The research was supported by grant IAA6093404 by the Grant Agency of the Academy of Sciences of the Czech Republic, grant MK00002327201 by the Ministry of Culture of the Czech Republic, and grant 39709 by the Mexican Consejo Nacional de Ciencia y Tecnología.

Literature cited

ADAM, F., and B. HUBERT. 1972. Les chiroptères nouveaux pour le Sénégal. Mammalia, 36: 59–70.

nd AVISE, J. C. 2004. Molecular markers, natural history, and evolution, 2 edition. Sinauer Associates, Sunderland, Massachusetts, 684 pp.

BAKER, R. J., and R. D. BRADLEY. 2006. Speciation in mammals and the genetic species concept. Journal of Mammalogy, 87: 643–662.

44 BATES, P. J. J., S. J. ROSSITER, A. SUYANTO, and T. KINGSTON. 2007. A new species of Hipposideros (Chiroptera: Hipposideridae) from Sulawesi. Acta Chiropterologica, 9: 13–26.

BRADLEY, R. D., and R. J. BAKER. 2001. A test of the genetic species concept: cytochrome­b sequences and mammals. Journal of Mammalogy, 82: 960–973.

BROSSET, A. 1984. Chiroptères d'altitude du Mont Nimba (Guinée). Description d'une espèce nouvelle, Hipposideros lamottei. Mammalia, 48: 545–555.

BOGDANOWICZ, W., and R. D. OWEN. 1998. In the Minotaur's labyrinth: phylogeny of the bat

family Hipposideridae. Pp. 27­42, in Bat biology and conservation (T. H. KUNZ and P.

A. RACEY, eds.). Smithsonian Institution Press, Washington, D.C., xiv + 365 pp.

FAHR, J., and N. M. EBIGBO. 2003. A conservation assessment of the bats of the Simandou Range, Guinea, with the first record of Myotis welwitschii (Gray, 1866) from West Africa. Acta Chiropterologica, 5: 125–141.

FELSENSTEIN, J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution, 39: 783–791.

FENTON, M. B. 1986. Hipposideros caffer (Chiroptera : Hipposideridae) in Zimbabwe: morphology and echolocation calls. Journal of Zoology, London, 210: 347–353

FRANCIS, C. M., D. KOCK, and J. HABERSETZER. 1999. Sibling species of Hipposideros ridleyi (Mammalia, Chiroptera, Hipposideridae). Senckenbergiana Biologica, 79: 255–270.

GUINDON S., and O. GASCUEL. 2003. PhyML ­ A simple, fast, and accurate algorithm to estimate large phylogenies by maximum likelihood. Systematic Biology, 52: 696–704.

GUILLÉN­SERVENT, A. and C. M. FRANCIS. 2006. A new species of bat of the Hipposideros bicolor group (Chiroptera: Hipposideridae) from Central Laos, with evidence of convergent evolution with Sundaic taxa. Acta Chiropterologica, 8: 39–61.

HALL, T. A. 1999. BioEdit: a user­friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symposium Series, 41: 95–98.

HAYMAN, R. W., and J. E. HILL. 1971. Part 2. Order Chiroptera. Pp. 1–73, in The mammals

of Africa: an identification manual (MEESTER, J., and H. W. SETZER, eds.). Smithsonian Institution Press, Washington, D.C.

HELLER, K.­G. 1992. The echolocation calls of Hipposideros ruber and Hipposideros

caffer. Pp. 75–77, in Prague Studies in Mammalogy (HORÁČEK, I. and V. VOHRALÍK, eds.). Charles University Press, Prague.

45 HILL, J. E. 1963. A revision of the genus Hipposideros. Bulletin of the British Museum (Natural History), Zoological Series, 11: 1–129.

HOLLISTER, N. 1918. East African mammals in the United States National Museum. Part I. Insectivora, Chiroptera, and Carnivora. Bulletin of the United States National Museum, 99: 1– 194.

IBÁÑEZ, C., J. L. GARCÍA­MUDARRA, M. RUEDI, B. STADELMANN, and J. JUSTE. 2006. The Iberian contribution to cryptic diversity in European bats. Acta Chiropterologica, 8: 277–297.

IRWIN, D. M., T. D. KOCHER, and A. C. WILSON. 1991. Evolution of the cytochrome b gene of mammals. Journal of Molecular Evolution, 32: 128–144.

JONES, G., MORTON, M., HUGHES, P. M. and BUDDEN, R. M., 1993. Echolocation, flight morphology and foraging strategies of some West African hipposiderid bats. Journal of Zoology, London, 230: 385–400.

KIMURA, M. 1980. A simple method for estimating evolutionary rates of base substitutions through comparative studies of nucleotide sequences. Journal of Molecular Evolution, 16: 111–120.

KINGSTON, T., M. C. LARA, G. JONES, Z. AKBAR, T. H. KUNZ, and C. J. SCHNEIDER. 2001. Acoustic divergence in two cryptic Hipposideros species: a role for social selection? Proceedings of the Royal Society of London, 268: 1381–1386.

KOCK, D. 1969. Die Fledermaus­Fauna des Sudan (Mammalia, Chiroptera). Abhandlungen der Senckenbergischen Naturforschenden Gesellschaft, 521: 1–238.

KOOPMAN, K. F. 1975. Bats of the Sudan. Bulletin of the American Museum of Natural History, 154: 353–444.

KOOPMAN, K. F. 1994. Chiroptera: systematics. Pp. 1–217 in Handbuch der Zoologie, Vol. 8, Mammalia, Part 60. Walter de Gruyter, Berlin, vii + 217 pp.

LAWRENCE, B. 1964. Notes on the horseshoe bats Hipposideros caffer, ruber and beatus. Breviora, 207: 1–5.

LAUSEN, C. L., I. DELISLE, R. M. R. BARCLAY, and C. STROBECK. 2008. Beyond mtDNA: nuclear gene flow suggests taxonomic over­splitting in the little brown bat (Myotis lucifugus). Canadian Journal of Zoology, 86: 700–713.

LIM, B. K., M. D. ENGSTROM, J. W. BICKHAM, and J. C. PATTON. 2007. Molecular phylogeny of New World sheath­tailed bats (Emballonuridae: Diclidurini) based on loci from the four genetic transmission systems in mammals. Biological Journal of the Linnean

46 Society, 93: 189–209.

MAYER, F. and O. VON HELVERSEN. 2001. Cryptic diversity in European bats. Proceedings of the Royal Society of London, 268: 1825–1832.

MAYER, F., C. DIETZ, and A. KIEFER. 2007. Molecular species identification boosts bat diversity. Frontiers in Zoology, 4: 4.

NADER I. A. 1982: New distributional records of bats from the Kingdom of Saudi Arabia (Mammalia: Chiroptera). Journal of Zoology, London, 198: 69–82.

NORBERG, U. M. 1994. Wing design, flight morphology and habitat use in bats. Pp. 205–239

in Ecological morphology: (WAINWRIGHT, P. C., and S. M. REILLY, eds. ). Chicago University Press. Chicago. 367 pp.

NYLANDER, M. A. A. 2004. MrModeltest v2.2. Program distributed by the author. Evolutionary Biology Centre, Uppsala University (http://www.ebc.uu.se/systzoo/staff/nylander.html).

POSADA, D., and K. A. CRANDALL 1998. Modeltest: testing the model of DNA substitution. Bioinformatics, 14: 817–818.

PYE, J. D. 1972. Bimodal distribution of constant frequencies in some hipposiderid bats (Mammalia: Hipposideridae). Journal of Zoology, London 166:323–335.

RONQUIST, F. R., and J. P. HUELSENBECK. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics, 19: 1572–1574.

ROSEVEAR, D. R. 1965. The bats of West Africa. British Museum of Natural History, London, xvii + 418 pp.

RUEDI, M., and F. MAYER. 2001. Molecular systematics of bats of the genus Myotis (Vespertilionidae) suggests deterministic ecomorphological convergences. Molecular Phylogenetics and Evolution, 21: 436–448.

SIMMONS, N. B. 2005. Order Chiroptera. Pp. 312–529 in Mammal species of the world: a

taxonomic and geographic reference, 3rd edition (WILSON, D. E., and D. M. REEDER, eds.). Johns Hopkins University Press, Baltimore, Maryland, 2142 pp.

TAMURA, K., and M. NEI. 1993. Estimation of the number of nucleotide substitutions in the control region of mitochondrial DNA in humans and chimpanzees. Molecular Biology and Evolution, 10: 512–526.

TAVARÉ, S. 1986. Some probabilistic and statistical problems in the analysis of DNA sequences. American Mathematical Society: Lectures on Mathematics in the Life

47 Sciences, 17: 57–86.

TAYLOR, P. J. 2000. Bats of Southern Africa: Guide to biology, identification, and conservation. University of Natal, Pietermaritzburg. 206 pp.

THABAH, A., S. J. ROSSITER, T. KINGSTON, S. ZHANG, S. PARSONS, K. MYA MYA, A. ZUBAID, and

G. JONES. 2006. Genetic divergence and echolocation call frequency in cryptic species of Hipposideros larvatus s.l. (Chiroptera: Hipposideridae) from the Indo­Malayan region. Biological Journal of the Linnean Society, 88: 119–130.

WANG, H., B. LIANG, J. FENG, L. SHENG, and S. ZHANG. 2003. Molecular phylogenetic of Hipposiderids (Chiroptera: Hipposideridae) and Rhinolophids (Chiroptera: Rhinolophidae) in China based on mitochondrial cytochrome b sequences. Folia Zoologica, 52: 259–268.

YANG, Z. 1996. Among­site variation and its impact on phylogenetic analysis. Trends in Ecology and Evolution, 11: 367–372.

48 Fig. 1. Geographical distribution of analysed specimens of the H. caffer complex. Different colours corresponding to main phylogenetic lineages are consistent throughout all figures. Type localities of described forms of H. caffer are designated: Rhinolophus caffer Sundevall, 1846; Phyllorrhina gracilis Peters, 1852; Phyllorrhina bicornis von Heuglin, 1861; Phyllorhina rubra Noack, 1893; Phyllorhina angolensis Seabra, 1898; Hipposiderus tephrus Cabrera, 1906; Hipposiderus caffer centralis Andersen, 1906; Hipposiderus caffer guineensis Andersen, 1906; Hipposideros caffer niapu Allen, 1917; Hipposideros nanus Allen, 1917; Hipposideros caffer var. aurantiaca De Beaux, 1924; Hipposideros braima Monard, 1939.

49 Fig. 2. Graph of condylocanine length (CCL) vs. zygomatic width (ZW) for the set of measured specimens for which the mitochondrial lineage could be identified. Colours correspond to molecular lineages, as in Figure 1. Dashed line represents the threshold value for discrimination between H. caffer and H. ruber (Koopman 1975).

50 Fig. 3. Bayesian 50%­majority rule consensus phylogram with support for major nodes. Posterior probabilities of Bayesian analysis is indicated above branches, bootstrap support of MP and ML analyses below branches. Nodes with values equal or more than 0.95 and 70% are considered well supported.

51 Table 1. Percentual sequence divergences among main phylogenetic lineages based on pairwise K2P genetic distances. For each pairwise comparison, minimal and maximal values are given. Appendix 1. List of specimens included in the study, with geographical locations, sample collection numbers, Genbank accession numbers and morphometric data. Abbreviations in haplotype names respond to countries of origin. Acronyms under lineage indicate the molecular lineages identified in the study as they are described in the results section of the text. FL: forearm length, CCL: condylocanine length, ZW: zygomatic width. Institution acronyms in the sample/voucher column correspond to: DBP: Debra B. Pires, University of California, Los Angeles, California, USA; DM: Durban Museum, Durban, South Africa; DSJ: David S. Jacobs, University of Cape Town, Cape Town, South Africa; EBD: Estación Biológica de Doñana, Sevilla, Spain; FMNH: Field Museum of Natural History, Chicago, Illinois, USA; NMP: National Museum, Prague, Czech Republic; ROM: Royal Ontario Museum, Toronto, Canada; IVB: Institute of Vertebrate Biology AS CR, v. v. i., Brno, Czech Republic; TK: The Museum of Texas Tech University, Lubbock, Texas, USA; TM: Transvaal Museum, Pretoria, South Africa. haplotype lineage sample/voucher locality coordinates accesion # FL [mm] CCL [mm] ZW [mm] caffer SAf1 A1 DSJ HC2 South Africa, Mpumalanga, Kruger NP ­ EU934452 ­ ­ ­ caffer SAf2 A1 DM 4586 South Africa, KwaZulu­Natal, Mkuzi Reserve 27°37' S 32°14' E waiting ­ ­ ­ caffer Swa1 A1 DM 7920 Swaziland, Kubuta ­ EU934458 48.1 ­ ­ caffer Swa2 A1 DM 8431 Swaziland, Hlane NP ­ EU934459 48.4 ­ ­ caffer Moz A1 TM 48051 Mozambique, Gerhard's cave 21°31' S 35°06' E EU934451 ­ ­ ­ ­ A1 TM 1047 South Africa, KwaZulu­Natal, Durban ­ ­ ­ 15.1 9.4 ­ A1 TM 30136 South Africa, KwaZulu­Natal, Ixopo ­ ­ 47.0 15.0 9.1 ­ A1 TM 36368 South Africa, KwaZulu­Natal, Babanango ­ ­ 47.0 14.9 9.1 caffer Mor1 A2 EBD COI318 Morocco, 30 km NW from Agadir 30°38’ N 09°50’ W waiting ­ ­ ­ caffer Mor2 A2 NMP pb3833 Morocco, Sidi Binzarne 30°04' N 09°40 W EU934449 47.7 14.8 8.9 caffer Mor3 A2 NMP pb3834 Morocco, Sidi Binzarne 30°04' N 09°40' W EU934450 46.3 14.8 9.0 caffer Sen2 A2 IVB S235 Senegal, Niokolo Koba NP, Simenti 13°01' N 13°17' W EU934454 43.9 14.0 8.7 caffer Sen3 A2 IVB S1217 Senegal, Niokolo Koba NP, Simenti 13°01' N 13°17' W EU934455 42.5 13.8 8.7 caffer Sen4 A2 IVB S748 Senegal, Niokolo Koba NP, Badi 13°08' N 13°13' W EU934456 44.6 14.2 8.8 caffer Sen5 A2 IVB S818 Senegal, Niokolo Koba NP, Dindefelo 12°21' N 12°19' W EU934457 45.0 13.9 8.9 caffer Yem1 A2 NMP pb2911 Yemen, Sana, Sana’a village 15°17' N 44°10' E EU934463 47.4 14.3 8.8 caffer Yem2 A2 NMP pb2913 Yemen, Sana, Sana’a village 15°17' N 44°10' E EU934464 48.1 14.8 9.0 caffer Sen1 B IVB S862 Senegal, Niokolo Koba NP, Dindefelo 12°21' N 12°19' W EU934453 48.4 15.1 9.3 ruber Ben1 B NMP 91848 Benin, Awaya, E of Dassa 07°47' N 02°16' E EU934474 52.5 16.2 10.3 ruber Ben2 B NMP 91849 Benin, Awaya, E of Dassa 07°47' N 02°16' E EU934475 54.3 16.5 10.5 caffer Tan1 B NMP pb2624 Tanzania, Zanzibar, Slave cave 06°00' S 39°11' E EU934460 47.8 15.3 9.5 caffer Ken B TK 33199 Kenya, Coastal Province, Kwale District ­ waiting ­ 15.0 9.5 ruber Mal B NMP Mw155 Malawi, Mulanje, Chipoka village 16°02' S 35°30' E EU934477 54.0 16.4 10.1 caffer ICo C1 ROM 100516 Ivory Coast, Sibabli 06°56' N 07°13' E waiting ­ 15.1 9.6 ruber DRC C1 FMNH 149408 DR Congo, Ituri, 2 km W of Epulu ­ waiting 53.7 17.3 10.7 ruber Uga C1 FMNH 137629 Uganda, Masaka, Bugala Island 18°11’ S 32°17’ E waiting 51.4 16.1 ­ ruber Cam1 C2 DBP 99­117 Cameroon, Dikome Balue 04°55' N 09°15' E waiting ­ ­ ­ ruber Cam2 C1 DBP 150 Cameroon, Linte 05°24' N 11°42' E waiting ­ ­ ­ ruber Cam3 C2 DBP 265 Cameroon, Kribi 02°46' N 09°53' E waiting ­ ­ ­ ruber Cam4 C2 DPB 267 Cameroon, Kribi 02°46' N 09°53' E waiting ­ ­ ­ ruber Cam5 C1 DBP 342 Cameroon, Mindourou 03°30' N 12°49' E waiting ­ ­ ­ ruber Cam6 C1 DBP 348 Cameroon, Mindourou 03°30' N 12°49' E waiting ­ ­ ­ ruber Cam7 C2 DBP 356 Cameroon, Mindourou 03°30' N 12°49' E waiting ­ ­ ­ ruber Cam8 C1 DBP pw2 Cameroon, Kousse 04°27' N 11°34' E waiting ­ ­ ­ ruber Cam9 C1 DBP 105 Cameroon, Dja Reserve 03°11' N 12°49' E waiting ­ ­ ­ ­ C1 DBP 99-123 Cameroon, Dja Reserve 03°11' N 12°49' E ­ 49.1 16.2 10.2 ­ C1 DBP 99-129 Cameroon, Dja Reserve 03°11' N 12°49' E ­ 50.4 15.9 10.6 ­ C1 DBP 99-125 Cameroon, Dja Reserve 03°11' N 12°49' E ­ 51.1 16.4 10.8 ­ C1 DBP 163 Cameroon, Linte 05°24' N 11°42' E ­ 50.7 16.1 9.9 ­ C1 DBP 170 Cameroon, Linte 05°24' N 11°42' E ­ 51.6 16.3 10.2 ­ C1 DBP 350 Cameroon, Mindourou 03°30' N 12°49' E ­ 48.7 16.4 10.5 ruber EGu1 C2 EBD RM815 Equatorial Guinea, Río Muni 01°52’ N 09°46’ E waiting 50.9 15.3 9.6 ruber EGu2 C2 EBD RM816 Equatorial Guinea, Río Muni 01°52’ N 09°46’ E waiting 49.5 15.9 10.3 ruber EGu3 C2 EBD M815 Equatorial Guinea, Bioco Island, Malabo 03°45’ N 08°43’ E waiting 52.4 16.5 10.7 ruber STo C2 EBD ST813 São Tomé e Principe, São Tomé Island 00°13’ N 06°43’ E waiting 50.0 16.1 9.9 ruber Pri C2 EBD P811 São Tomé e Principe, Principe Island ­ waiting 49.0 16.0 10.1 ­ C2 DBP 99­124 Cameroon, Dja Reserve 03°11' N 12°49' E ­ 50.05 16.1 10.5 ­ C2 DBP 265 Cameroon, Kribi 02°46' N 09°53' E ­ 50.15 16.3 10.6 ­ C2 DBP 337 Cameroon, Mindourou 03°30' N 12°49' E ­ 48.53 15.8 10.6 ruber Sen1 D IVB S119 Senegal, Niokolo Koba NP, Lengue 13°03' N 13°05' W EU934478 49.8 16.6 10.6 ruber Sen2 D IVB S1374 Senegal, Niokolo Koba NP, Dar Salam 13°15' N 13°13' W EU934479 48.3 16.5 10.8 ruber Sen3 D IVB S1400 Senegal, Niokolo Koba NP, Dar Salam 13°15' N 13°13' W EU934480 48.3 16.4 10.5 ruber Sen4 D IVB S272 Senegal, Niokolo Koba NP, Simenti 13°01' N 13°17' W EU934481 47.0 16.4 10.2 ruber Sen5 D IVB S273 Senegal, Niokolo Koba NP, Simenti 13°01' N 13°17' W EU934482 48.5 16.6 10.3 ruber Sen6 D IVB S275 Senegal, Niokolo Koba NP, Simenti 13°01' N 13°17' W EU934483 47.4 16.4 10.7 ruber Sen7 D IVB S285 Senegal, Niokolo Koba NP, Simenti 13°01' N 13°17' W EU934484 46.8 16.5 10.7 abae Sen abae IVB S822 Senegal, Niokolo Koba NP, Dindefelo 12°21' N 12°19' W EU934448 ­ ­ ­ abae Ben1 abae NMP 91850 Benin, Awaya, E of Dassa 07°47' N 02°16' E EU934446 ­ ­ ­ abae Ben2 abae NMP 91851 Benin, Awaya, E of Dassa 07°47' N 02°16' E EU934447 ­ ­ ­ beatus ICo beatus ROM 100579 Ivory Coast, Sibabli 06°56' N 07°13' E waiting ­ ­ ­ beatus DRC beatus FMNH 149406 DR Congo, Ituri, 2 km W of Epulu ­ waiting ­ ­ ­ cyclops Sen1 cyclops IVB S02­61 Senegal, Niokolo Koba NP, Assirik 12°53' N 12°43' W EU934465 ­ ­ ­ cyclops Sen2 cyclops IVB S747 Senegal, Niokolo Koba NP, Badi 13°08' N 13°13' W EU934466 ­ ­ ­ fuliginosus GBi1 fuliginosus AR GB.5.Afia Guinea­Bissau, Afia 11°20' N 13°54' W EU934467 ­ ­ ­ fuliginosus GBi2 fuliginosus AR GB.2.Cunfa Guinea­Bissau, Cunfa 11°07' N 14°56' W EU934468 ­ ­ ­ gigas Sen1 gigas IVB S1032 Senegal, Niokolo Koba NP, Dar Salam 13°15' N 13°13' W EU934469 ­ ­ ­ gigas Sen2 gigas IVB S1404 Senegal, Niokolo Koba NP, Dar Salam 13°15' N 13°13' W EU934470 ­ ­ ­ jonesi Sen1 jonesi IVB S804 Senegal, Niokolo Koba NP, Dindefelo 12°21' N 12°19' W EU934472 ­ ­ ­ jonesi Sen2 jonesi IVB S854 Senegal, Niokolo Koba NP, Dindefelo 12°21' N 12°19' W EU934473 ­ ­ ­ jonesi Ben jonesi NMP 91842 Benin, Awaya, E of Dassa 07°47' N 02°16' E EU934471 ­ ­ ­ Genetic diversity in morphologically uniform Hipposideros cf. ruber (Chiroptera: Hipposideridae) in southeastern Senegal inferred from sequences of mitochondrial cytochrome b gene

Peter Vallo1,2, Petr Benda3,4, Natália Martínková1, Peter Kaňuch1,5, Petr Koubek1, and Jaroslav Červený1, 6

1 Institute of Vertebrate Biology, v.v.i., Academy of Sciences of the Czech Republic, Květná 8, Brno Czech Republic; E­mail: [email protected] 2 Institute of Botany and Zoology, Faculty of Science, Masaryk University, Kotlářská 2, 611 37 Brno, Czech Republic 3 Department of Zoology, National Museum (Natural History), Václavské náměstí 68, 115 79 Praha 1, Czech Republic 4 Department of Zoology, Charles University, Viničná 7, 128 44 Praha 2, Czech Republic 5 Institute of Forest Ecology, Slovak Academy of Sciences, Zvolen, Slovak Republic 6 Department of Forest Protection and Game Management, Faculty of Forestry and Environment, Czech University of Agriculture, Prague, Czech Republic

Keywords: cytochrome b, Hipposideros caffer complex, cryptic species, systematics

Running title: Genetic diversity in Hipposideros cf. ruber in Senegal

Abstract Two sympatric genetic forms of bats assigned to H. ruber morphospecies, were previously shown to occur in southeastern Senegal. We therefore surveyed genetic diversity in population of the Niokolo Koba National Park using sequences of mitochondrial cytochrome b gene in order to identify status of the two mitochondrial lineages. Skull morphology was also studied in detail. A local population, where both forms were suspected to occur, was also explored in echolocation calls to identify potential cryptic species. Detailed multidimensional analysis of skull metric dimensions indicated slight morphological differences between the two genetic forms. A large range of echolocation frequencies identified in local population was not related to the genetic distiction, as all

56 bats sampled for echolocation belonged to one, more abundant, genetic form. Sex­ dependent bias in frequencies was shown in this form. Phylogenetic comparison to other West African samples revealed sister relationship between the less numerous form and a form from Benin. Based on genetic divergence, the two genetic forms from Senegal may be considered cryptic species. However, absence of gene flow between them has to be checked to prove their reproductive isolation.

Introduction Genetic differenes in living organisms may substantially precede phenotypic differences, making genetically distinct forms undetectable by traditional morphological means (Yoder et al. 2000). Such forms are believed to belong to the same species until some other evidence shows that they represent independent evolutionary units. Existence of these so called cryptic species (Mayr 1996, Lincoln et al. 1998) has recently been revealed in many taxonomic groups (Avise 2004, Bickford et al. 2006). Although careful examination of morphology, ecology or behaviour sometimes helped to discover cryptic species, it was the employment of molecular genetic methods in the last two decades that has enormously contributed to revelation of cryptic forms within traditionally recognized taxa (Avise 2004, Bickford et al. 2006). Crypticity appears to be a rather common phenomenon in bat systematics, and molecular data have played an important role for direct recognition of cryptic forms or justification of forms differing in other, not fully testifying traits (Jones 1997, Mayer and von Helvesen 2001, Baker and Bradley 2006, Ibáñez et al. 2006). Especially, differences in echolocation have often induced research of seemingly uniform bat species, be it a traditional example of the Pipistrellus pipistrellus/P. pygmaeus complex (Barrat et al. 1997) or a recent discovery of cryptic species within South African Scotophilus dinganii (Jacobs et al. 2006). The genus Hipposideros Gray, 1831, the largest group within the Palaeotropical family Hipposideridae has been particularly rich on phonic forms that were proved as cryptic species, often with the help of molecular analyses (Francis et al. 1999, Kingston et al. 2001, Guillén­Servent and Francis 2006, Thabah et al. 2006). Several other phonic forms within Hipposideros have been regarded as cryptic species, although no molecular justification has been attempted to date. Besides phonic types of H. commersoni revealed by Pye (1972), which have been recently considered distinct species (Simmons 2005), H.

57 caffer (Sundevall, 1846) constitutes the most known cryptic complex, whose two recognized species H. caffer and H. ruber are presumably separable by echolocation (Pye 1972, Fenton 1986, Heller 1992, Jones et al. 1993). The first molecular phylogenetic study on the H. caffer complex (Vallo et al. in press), however, showed that H. caffer complex consists of several evolutionary units that can be regarded as separate species. Metric characters that have served for specific discrimination between H. caffer and H. ruber (Noack, 1893) (Hayman and Hill 1971, Koopman 1975) highly overlapped, which probably makes also the traditional interpretation of echolocation frequencies for specific distinction of little use. Our field excursions in the Niokolo Koba National Park (NKNP), southeastern Senegal, yielded a large catch of leaf­nosed bats originally assigned to H. ruber by their morphology. These bats actually represent a distinct phylogenetic lineage restricted to West Africa, from Senegal to Benin, which is probably not closely related to true H. ruber s. str., primarily described from Tanzania (Vallo et al. in press). The captured bats from NPNK showed no obvious morphological differences in external traits but in the subsample used in Vallo et al. (in press) one haplotype significantly diverged from the main mitochondrial lineage of the Senegalese population. In this study, we therefore researched phylogenetic relationships in H. cf. ruber using sequences of the mitochondrial gene for cytochrome b in order to reveal genetic diversity in the population of the NKNP and systematic importance of the two mitochondrial lineages. The divergent haplotype originated from the village of Dar Salam, situated at the north­western border of the NKNP. Echolocation calls in the local population were also explored in order to detect differences in echolocation frequencies, which might indicate presence of potential cryptic species. Thus, we also attempted to identify whether the diversity of echolocation frequencies has a relationship to genetic distinction into two sympatric forms.

Material and Methods Sampling Bats were netted at six localities in the Niokolo Koba National Park (NKNP), SE Senegal, in 2004–2007 (Figure 1, Appendix 1). Captured specimens were weighted and external measurements were recorded. A subset of specimens was collected and fixed in ethanol for further research. Tissue samples (spleen) were taken from collected specimens. Skulls

58 were extracted from ethanol preserved vouchers for morphological analysis. All biological material was deposited at the Institute of Vertebrate Biology AS CR, v.v.i., Brno, Czech Republic. For comparison we included into the analysis additional samples of West African H. cf. ruber from Benin and Ivory Coast, and H. caffer (Sundevall, 1846) from South Africa (Appendix 1).

Morphological analysis Collected specimens were examined in the same way as in previous studies (e. g. Benda and Vallo in press). Fifteen skull dimensions were measured using mechanical calliper: LCr = greatest length of skull incl. praemaxillae, LOc = occipitocanine length, LCc = condylocanine length, LaZ = zygomatic width, LaI = width of interorbital constriction, LaInf = rostral width between foramina infraorbitalia, LaN = neurocranium width, LaM = mastoidal width, ANc = neurocranium height, LBT = largest horizontal length of tympanic bulla, CC = rostral width between canines (incl.), M3M3 = rostral width between third upper molars (incl.), CM3 = length of upper tooth­row between CM3 (incl.), LMd = condylar length of mandible, ACo = height of coronoid process, CM3 = length of lower tooth­row between CM3 (incl.). Analysis of variance and canonical discrimination analysis was used to reveal morphologic differences in our dataset. Statistical analyses were performed using the Statistica 6.0 software.

Analysis of echolocation calls Echolocation calls of individuals captured in the target locality Dar Salam were recorded in time expansion and heterodyne mode with the ultrasound bat­detector Pettersson D240x (Pettersson Elektronik AB, Uppsala) linked to the Sony MiniDisc MZ, at a sampling rate of 120 kHz. Animals were flown in an open­air field volary made of a mosquito net with internal dimensions 2x2x2 m, and recordings were obtained during their free approaching to detector microphone. Calls with maximum of energy were analysed in time­expanded sequences (10×) by BatSound 3.31 software (Pettersson Elektronik AB, Uppsala). Power spectra were used to derive CF frequency. This parameter was measured from four consecutive randomly selected calls for each individual.

59 DNA processing Total genomic DNA was extracted from alcohol preserved tissue with DNeasy Tissue Kit (Qiagen) according to manufacturer’s protocol. Complete cytochrome b gene (cytb) was amplified using universal primers L14724 and H15915 (Irwin et al. 1991). Each 50 μ l PCR reaction volume contained 0.8 μ M of each primer, 0.2 mM dNTP, 0.2 μ l (1U) of HotMaster Taq DNA polymerase (Eppendorf), 5 μ l of 10x buffer, and 2–5 μ l of extracted DNA. Initial denaturation at 94 °C for 3 min was followed by 35 cycles of denaturation for 40 s at 94 °C, annealing for 40 s at 50 °C, and extension for 90 s at 65 °C. Final extension at 65 °C lasted for 5 min.Obtained PCR products were purified with QIAquick PCR Purification Kit (Qiagen) and sequenced commercially in both directions with the same primers using BigDye Terminator sequencing chemistry (Applied Biosystems) on ABI 3730xl sequencer. Sequences were assembled and edited in Sequencher 4.6 (Gene Codes). Sequences were submitted to the GenBank database and are available under accession numbers EUwaiting­EUwaiting (Appendix 1).

Phylogenetic analysis Sequences were aligned in BioEdit 7.0 (Hall 1999). Polymorphism within the sequence dataset was assessed using DnaSP 4.0 (Rozas et al. 2000), and a median­joining network was constructed in Network 4.2 (Fluxus Technologies). Based on this initial analysis, the sequence dataset was reduced to haplotypes representing main haplogroups to facilitate reconstruction of phylogenetic trees. Additional sequences downloaded from the GenBank including outgroup sequence of H. caffer were added to the final alignment (Lim et al. 2008, Vallo et al. in press; Appendix 1). Phylogenetic trees were reconstructed in PAUP* 4.10b (Sinauer Associates) using maximum parsimony (MP) and maximum likelihood (ML). In both methods, tree space was heuristically searched with tree bisection­reconnection swapping algorithm on 100 random sequence additions. Transversion weighted model with a significant proportion of invariant sites (TVM+I) was used in ML method. This model was suggested by program Modeltest 3.7 (Posada and Crandall 1998) under Akaike Information Criterion (AIC). Reliability of branching pattern was assessed by bootstrapping using 1000 (MP) and 100 (ML) replicates, respectively. Phylogeny was further inferred using Bayesian method (BA) in program MrBayes 3.1.2 (Ronquist & Huelsenbeck 2003) under general time

60 reversible model (GTR+Γ ) with gamma distributed rates among sites (Tavaré et al. 1986, Yang 1993). This model was suggested by program MrModeltest 2.3 (Nylander 2002) under AIC. BA was carried out in two independent simultaneous Metropolis­coupled MCMC runs of four chains running for 106 generations, sampled every 100 generation, starting from a random trees. First 2500 generations were discarded as burn­in. Results of BA were summarized in the 50% majority rule consensus tree with cumulative posterior probabilities representing confidence estimates of topology. Templeton test (Templeton 1983) and Shimodaira­Hasegawa test (SH test; Shimodaira and Hasegawa 1999) were used to compare alternative topologies with best topologies obtained under MP and ML criterion, respectively. Sequence divergences were based on pairwise Kimura two­parameter genetic distances (K2P; Kimura 1980).

Results Fifty­two specimens of leaf­nosed bats of H. ruber morphotype from NKNP were included in the analysis (Appendix 1). In this dataset, twenty­nine haplotypes of cytb (1140 bp) were identified, resulting in a high haplotype diversity H = 0.966. Most of the haplotypes differed with only a few substitutions, which meant a low nucleotide diversity π = 0.009. Median­joining network showed diversification into two main groups within Senegalese H. cf. ruber, denoted here as A and B (Figure 1). Twenty­seven haplotypes representing forty­seven specimens were comprised in the group A, only two haplotypes from five specimens built the group B. The main groups A and B diverged 2.4–3.6%. Within the group A, two loose subgroups, A1 and A2, could be recognized. Subgroup A1 comprised 11 haplotypes (20 specimens), subgroup B 16 haplotypes (27 specimens). Genetic divergence between the two subgroups A1 and A2 ranged 0.4–1.3%, within subgroups 0.1–0.6% and 0.1–0.8% in A1 and A2, respectively. This genetic structure was not related to distribution of sampling sites in the NKNP, as haplotypes of main groups and subgroups were mixed among localities (Appendix 1). Bats of the main groups A and B occurred syntopically in Dar Salam, Simenti and Lengue. Bats of both subgroups A1 and A2 were present in all localities except Assirik, from which only one specimen was analyzed. From the roosting Dar Salam population, 16 specimens were recorded for their echolocation calls. Their calls showed a continuous range of the ending frequencies from 131 kHz to 139 kHz (Figure X, Appendix 1). Specimens with known echolocation

61 frequencies were clustered exclusively in the group A. Distribution of echolocation frequencies did not reflect the division into A1 and A2 subgroups, as frequency ranges in A1 and A2 overlapped (A1: 132–139 kHz, A2 131–138 kHz). Frequencies were however related to sex, males (134–139 kHz) echolocated higher than females (130–135 kHz) (Mann­Whitney test, U = 4, P = 0.004). This trend was obvious in both subgroups A1 (males 134–139 kHz, females 132–135 kHz) and A2 (males 134–138 kHz, females 130– 131 kHz). Morphological analysis of skull dimensions did not show any distinct morphotypes within the examined set of samples. Measurements of two main haplogroups A and B mostly overlapped (Table 1) ; they differed only in CM3 (ANOVA; F = 5.813, d.f. = 48, P = 0.020) and slighly in ANc (ANOVA; F = 3.779, d.f. = 48, P = 0.058). No difference was observed between subgroups A1 and A2. Size differences between sexes were not revealed within main groups A and B. No differences were found also within subgroup A1. Within subgroup A2, sexes differed in dimension LaZ (ANOVA; F = 4.967, d.f. = 24, P = 0.035). Discrimination analysis of skull dimension showed variables LCo, LCc, LaZ, LaI, CC,

ANc, CM3 and CM3 as the most important for discrimination between groups A and B along the 1st canonic variable (CV1) explaining 71.04% of variance (Figure 2). Subgroups A1 and A2 were indicated as identical. Due to a high number of haplotypes within the group A, only a representative subsample of six haplotypes was included into the dataset to facilitate assessment of phylogenetic relationships. Thus, the final ingroup haplotype dataset for the cytb consisted of 11 sequences; eight from Senegal, two from Ivory Coast and one from Benin. Heuristic search under MP criterion yielded one best tree 182 steps long. This tree showed a supported sister relationship between lineage B (= group B) from Senegal and lineage from Benin, and sister relationship between lineage A (= group A) from Senegal and lineage from Ivory Coast, which was not supported. A tree 183 steps long was also found during heuristic search, in which the lineage from Ivory Coast was paraphyletic to the other three lineages. This topology did not differ significantly from the MP tree (Templeton test; diff. length = 1, z = ­0.2582; P = 0.7963). Both ML tree and BA consensus tree revealed the same sister relationship between lineage A from Senegal and lineage from Ivory Coast but it was supported neither by bootstrap nor by posterior probability. In both trees the lineage from Benin did not cluster with lineage B from Senegal as it did in MP tree. Instead, it clustered

62 as sister to lineage A from Senegal and lineage from Ivory Coast, although this position was also not supported. Alternative topologies corresponding with 182­ and 183­step trees rendered under MP criterion did not differ significantly from the ML tree (SH test; diff. – lnL = 0.26372, P = 0.836 and diff. –lnL = 1.60391, P = 0.552, respectively). Topology assumed monophyly of the two lineages from Senegal differed significantly from the ML tree diff. (–lnL = 10.99482, P = 0.024), thus the alternative hypothesis of sister relationship between the two lineages was rejected. Genetic divergence among the four lineages in the phylogenetic trees ranged 2.0–5.4%. Senegalese lineage A differed 3.2–4.1% from lineage from Benin and 4.0–4.3% from lineage from Ivory Coast. Senegalese lineage B differed 2.0­2.1% from lineage from Benin and 5.1–5.2% from lineage from Ivory Coast. Lineages from Benin and Ivory Coast differed 5.3–5.4%.

Discussion Analysis of DNA sequences showed that two distinct lineages differing at 2.4–3.6% sequence divergence exist sympatrically in southeastern Senegal. Only five specimens from our collection were proved to belong to the less frequent lineage (B). Haplotypes of both lineages were discovered in three localities of the NKNP. Absence of the lineage B in two other localities may have been due to limited sampling. Local population in the locality Dar Salam yielded specimens of both lineages. It was therefore researched in more detail regarding echolocation calls, which can indicate presence of cryptic species (Jones 1997, Mayer and von Helversen 2001). Identified range of echolocation frequencies 130– 139 kHz was comparable to a range discovered in cryptic forms of H. bicolor (128–144 kHz; Kingston et al. 2001), and absolute difference of values 9 kHz exceeded difference in cryptic forms of H. ridley (7 kHz, 65–72 kHz; Francis et al. 1999). Genetic analysis however showed that all sixteen specimens with known echolocation frequencies belonged to the more abundant lineage (A) and no internal grouping with respect to echolocation frequency was found within this lineage. Instead, distribution of echolocation frequencies appeared to be sex­related, because males echolocated significantly higher than females. This might be explained by relationship of echolocation frequency and sexual dimorphism, shown in H. ruber from Equatorial Guinea (Guillén­Servent et al. 2000), which implies higher echolocation frequency for males as the smaller­sized sex. Guillén­Servent et al.

63 (2000) however state that intraspecific shifts of CF values may be independent of sexual dimorphism in body size in hipposiderids. Lack of sex­related size differences in our sampled population evidently supports their notion. Evidently, much broader sampling and detailed analysis of both genetic and morphologic traits has to be undertaken to clarify social and ecological meaning of sexual dimorphism in echolocation. Silent bands within the echolocation range and bimodal distribution of peak frequencies are good indicators of potential crypticity in hipposiderids (Pye 1972, Jones et al. 1993). Absence of these features could have cast doubt on presence of cryptic forms in the sampled Dar Salam population. Also the identified range 130–139 kHz might have been considered to lie within intraspecific variation, as similar ranges 132–138 kHz were found in population of H. ruber from Irangi, Democratic Republic of Congo (Heller 1992). However, these bats might have belonged to two distinct mitochondrial lineages showed to exist in Central Africa (Vallo et al. in press), which naturally remained undetected by Heller (1992). Interesting is that bats from DR Congo and Senegal used the same frequency range, although the former were considerably larger (forearm length ca. 55 mm in DR Congo and ca. 48 mm in Senegal). Guillén­Servent et al. (2000), who found similar values in populations in the Gulf of Guinea but in much narrower intrapopulational ranges, considers the low frequences around 135 kHz to be an adaptation to forest environment. Bats determined as H. ruber from Gambia exhibited also similar range of echolocation frequencies, 121–136 kHz (Jones et al. 1993). These bats may have belonged to the same mitochondrial lineages like the bats in Senegal, as sampled regions are geographically close, and size of bats was similar. Anyway, it has been proved that the large range of echolocation frequencies was not related to existence of two distinct mitochondrial lineages. Morphologic analysis found only slight evidence on differentiation in skull dimensions but practically bats of these two lineages remain undistinguishable by means standardly used for systematic determination. It seems plausible that the two lineages originated in allopatry as their sister relationship was not supported. Moreover, the lineage B was showed to be closely related to the lineage from Benin by maximum parsimony, and also in terms of sequence divergence. Sympatric occurrence of the two lineages from Senegal may have resulted from a secondary contact of originally isolated lineages caused by climatic changes in Quaternary and their influence on sahel and rainforest environment (Moritz et al. 2000, Plana 2004). The shifts in

64 savannah and forest areas are hypothesized to have influenced phylogeographic pattern e.g. in West African mice of genera Praomys and Hylomyscus (Lecompte et al. 2005, Nicolas et al. 2006, Nicolas et al. in press). These woodland­bound mice were forced to patchy forest refugia separated by dry savannahs, where divergent lineages evolved. Such allopatric species secondarily met in restored forest regions. Bats, as mammals capable of powered flight, might have responded in quite a different way than terrestrial mammals, as they can overcome environmental barriers like rivers and stretches of unsuitable habitat rather easily. Broader sampling throughout West Africa is needed to fully understand the complicated phylogeographic pattern indicated by our analysis. Although geographic distance between Senegal and Benin is roughly twice as much as from either of the countries to Ivory Coast, the lineage from Ivory Coast was the most divergent, 4.0–5.4%, from the other lineages and even suggested basal in the phylogenetic tree. Judged solely on values of sequence divergence, the two lineages from Senegal could be considered separate species based on known limits of divergence between cryptic species (Bradely and Baker 2001, Baker and Bradley 2006). Similar values of interspecific divergences, 2.4–3.9%, have been found between South African cryptic forms of Scotophilus dinganii (A. Smith 1833) (Jacobs et al. 2006). Also between Southeast Asian species Hipposideros khakhouayensis and H. rotalis the sequence divergence was similar 3.9–4.1% (Guillén­Servent and Francis 2006). Sequence divergence values however have to be interpreted with caution in deciding upon specific status of suspected forms. Laussen et al. (2008) clearly showed that divergence in mitochondrial sequences misleaded recent assignation of lineages within North American Myotis lucifugus into distinct species, as they discovered a substantial nuclear gene flow among these lineages. Similarly, Bilgin et al. (2008) found mitochondrial differentiation in Myotis capaccinii, which was not reflected in nuclear genetic markers. It is therefore important to prove reproductive isolation of the two genetically distinct forms in Senegal to confirm their potential specific statuses. An attempt to use a paternally inherited gene as a counterweight to maternally inherited cytb (Trujillo 2003, Lim et al. 2008) did not bring any alternative explanation for relationship between the two forms from Senegal. Sequences of 1.2 kb portion of male zinc finger protein (zfy; Page et al. 1987) amplified according to Trujillo (2003) failed to distinguish between them (data not shown). This uniformity may indicate reproductive connection but it could also be an artefact resulting from a low mutational rate within the

65 particular paternal gene.

Acknowledgements We thank Adam Konečný, Josef Bryja and other colleagues from the Institute of Vertebrate Biology AS CR, v.v.i., Brno, for their assistance in bat capturing in Senegal. Field work in the NKNP and collecting of bat specimens was approved by the Direction des Parcs Nationaux du Sénégal, Dakar. Further we thank Burton Lim (Royal Ontario Museum) for providing tissue samples of bats from Ivory Coast. Research was supported by the Grant Agency of the Academy of Sciences of the Czech Republic (IAA6093404) and Ministry of Culture of the Czech Republic (MK00002327201).

Literature Avise, J. C. 2004. Molecular markers, natural history, and evolution, 2nd edition. Sinauer Associates, Sunderland, Massachusetts, 684 pp. Baker, R. J., and R. D. Bradley. 2006. Speciation in mammals and the genetic species concept. Journal of Mammalogy 87:643–662. Barrat, E.M., R. Deaville, T. M. Burland, M. W. Bruford, G. Jones, P. A. Racey, and R. K. Wayne. 1997. DNA answers the call of pipistrelle bat species. Nature 387:138–139. Benda, P., and P. Vallo. Taxonomic revision of the genus Triaenops (Mammalia: Chiroptera: Hipposideridae) with description of a new species from southern Arabia and definitions of new genus and tribe Zootaxa. Acta Chiropterologica. in press. Bickford D., D. J. Lohman, N. S. Sodhi, P. K.L. Ng, R. Meier, K. Winker, K. K. Ingram and I. Das. 2007. Cryptic species as a window on diversity and conservation. Trends in Ecology and Evolution 22:148–155. Bilgin, R., A. Karataş, E. Çoraman, J. Morales. 2008. The mitochondrial and nuclear genetic structure of Myotis capaccinii (Chiroptera: Vespertilionidae) in the Eurasian transition, and its taxonomic implications. Zoologica Scripta 37:253–262. Bradley, R. D., and R. J. Baker. 2001. A test of the genetic species concept: cytochrome­b sequences and mammals. Journal of Mammalogy 82:960–973. Fenton, M. B. 1986. Hipposideros caffer (Chiroptera : Hipposideridae) in Zimbabwe: morphology and echolocation calls. Journal of Zoology, London 210:347–353. Francis, C. M., D. Kock, and J. Habersetzer. 1999. Sibling species of Hipposideros ridleyi

66 (Mammalia, Chiroptera, Hipposideridae). Senckenbergiana Biologica 79:255–270. Guillén­Servent, A., C. Ibáñez, and J. Juste. 2000. Variation in the echolocation calls of Hipposideros ruber in the Gulf of Guinea: An exploration of the adaptive meaning of the constant frequency value in Rhinolophoid bats. Journal of Evolutionary Biology 13:70–80. Guillén­Servent, A., and C. M. Francis. 2006. A new species of bat of the Hipposideros bicolor group (Chiroptera: Hipposideridae) from Central Laos, with evidence of convergent evolution with Sundaic taxa. Acta Chiropterologica 8:39–61. Hall, T. A. 1999. BioEdit: a user­friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symposium Series 41:95–98. Hayman, R. W., and J. E. Hill. 1971. Part 2. Order Chiroptera. Pp. 1–73. In: The mammals of Africa: an identification manual (Meester, J., and H. W. Setzer, eds.). Smithsonian Institution Press, Washington D.C. Heller, K.­G. 1992. The echolocation calls of Hipposideros ruber and Hipposideros caffer. Pp. 75–77. In: Prague Studies in Mammalogy (Horáček, I., and V. Vohralík, eds.). Charles University Press, Prague. Ibáñez, C., J. L. García­Mudarra, M. Ruedi, B. Stadelmann, and J. Juste. 2006. The Iberian contribution to cryptic diversity in european bats. Acta Chiropterologica 8:277–297. Irwin, D. M., T. D. Kocher, and A. C. Wilson. 1991. Evolution of the cytochrome b gene of mammals. Journal of Molecular Evolution 32:128–144. Jacobs, D. S., G. N. Eick, M. C. Schoeman, and C. A. Mathee. 2006. Cryptic species in an insectivorous bat, Scotophilus dinganii. Journal of Mammalogy 87:161–170. Jones, G., M. Morton, P. M. Hughes, and R. M. Budden. 1993. Echolocation, flight morphology and foraging strategies of some West African hipposiderid bats. Journal of Zoology, London 230:385–400. Jones, G. 1997. Acoustic signals and speciation: the roles of natural and sexual selection in the evolution of cryptic species. Advanced Study in Behaviour 26:317–354. Kimura, M. 1980. A simple method for estimating evolutionary rates of base substitutions through comparative studies of nucleotide sequences. Journal of Molecular Evolution 16:111–120. Kingston, T., M. C. Lara, G. Jones, Z. Akbar, T. H. Kunz, and C. J. Schneider. 2001. Acoustic divergence in two cryptic Hipposideros species: a role for social selection?

67 Proceedings of the Royal Society of London 268: 1381­1386. Koopman, K. F. 1975. Bats of the Sudan. Bulletin of the American Museum of Natural History 154:353–444. Lausen, C. L., I. Delisle, R.M.R. Barclay, and C. Strobeck. 2008. Beyond mtDNA: nuclear gene flow suggests taxonomic over­splitting in the little brown bat (Myotis lucifugus). Canadian Journal of Zoology 86:700–713. Lecompte, E., C. Denys, L. Granjon 2005. Confrontation of morphological and molecular data: The Praomys group (Rodentia, Murinae) as a case of adaptive convergences and morphological stasis. Molecular Phylogenetics and Evolution 37:899–919. Lim, B. K., M. D. Engstrom, J. W. Bickham, and J. C. Patton. 2007. Molecular phylogeny of New World sheath­tailed bats (Emballonuridae: Diclidurini) based on loci from the four genetic transmission systems in mammals. Biological Journal of the Linnean Society 93:189–209. Lincoln, R., G. Boxshall, and P. Clark. 1998. A Dictionary of Ecology, Evolution and Systematics, 2nd ed. Cambridge University Press, Cambridge, United Kingdom. Mayr, E. 1996. What is a species, and what is not? Philosophy of Science 63:262–277. Mayer, F., and O. von Helversen. 2001. Cryptic diversity in European bats. Proceedings of the Royal Society of London B 268:1825–1832. Moritz, C., J. L. Patton, C. J. Schneider, T. B. Smith 2000. Diversification of rainforest faunas: An integrated molecular approach. Annual Review of Ecology and Systematics 31:533–563. Nicolas, V., S. Quérouil, E. Verheyen, W. Verheyen, J. F. Mboumba, M. Dillen, M. Colyn 2006. Mitochondrial phylogeny of African wood mice, genus Hylomyscus (Rodentia, Muridae): Implications for their taxonomy and biogeography. Molecular Phylogenetics and Evolution 38:779–793. Nicolas V., J. Bryja, B. Akpatou, A. Konečný, E. Lecompte, M. Colyn, A. Lalis, A. Couloux, C. Denys and L. Granjon in press. Compared phylogeography of two forest rodent sibling species (Praomys rostratus and P. tullbergi) in West Africa: Differential reactions to past forest fragmentation. Molecular Ecology. Page, D. C., R. Mosher, E. M. Simpson, E. M. C. Fisher, G. Mardon, J. Pollack, B. McGillivray, A. de la Chapelle, and L. G. Brown. 1987. The sex­determining region of the human Y chromosome encodes a finger protein. Cell 51:1091–1104.

68 Plana, V. 2004. Mechanisms and tempo of evolution in the African Guineo–Congolian rainforest. Philosophical transactions of the Royal Society of London B 359:1585–1594. Posada, D., and Crandall, K. A. 1998. Modeltest: testing the model of DNA substitution. Bioinformatics, 14:817–818. Pye, J. D. 1972. Bimodal distribution of constant frequencies in some hipposiderid bats (Mammalia: Hipposideridae). Journal of Zoology, London 166:323–335. Ronquist, J., and J. J. Huelsenbeck. 2003. MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics 19:1572–1574. Rozas, J., J.C. Sanchez­DelBarrio, X. Messeguer, and R. Rozas. 2003. DNASP, DNA polymorphism analysis by coalescent and other methods. Bioinformatics 19:2496–2497. Shimodaira, H., and M. Hasegawa. 1999. Multiple comparisons of log­likelihoods with applications to phylogenetic inference. Molecular Biology and Evolution 16:1114– 1116. Simmons, N. 2005. Order Chiroptera. Pp. 312–529. In: Mammal species of the world: a taxonomic and geographic reference (Wilson, D. E. and D. M. Reeder, eds.). 3rd ed. Johns Hopkins University Press, Baltimore, Maryland. Tavaré, S. 1986. Some probabilistic and statistical problems in the analysis of DNA sequences. American Mathematical Society: Lectures on Mathematics in the Life Sciences 17:57–86. Templeton, A.R. 1983. Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the humans and apes. Evolution 37:221–244. Thabah, A., S. J. Rossiter, T. Kingston, S. Zhang, S. Parsons, K. Mya Mya, A. Zubaid, and G. Jones. 2006. Genetic divergence and echolocation call frequency in cryptic species of Hipposideros larvatus s.l. (Chiroptera: Hipposideridae) from the Indo­Malayan region. Biological Journal of the Linnean Society 88:119–130. Trujillo, R. G. 2005. Phylogenetics of the genus Scotophilus (Chiroptera: Vespertilionidae): perspectives from paternally and maternally inherited genomes with emphasis on African species. PhD dissertation, Texas A&M University. Vallo, P., A. Guillén­Servent, P. Benda, D.B. Pires, and P. Koubek. 2008. Variation of mitochondrial DNA reveals high cryptic diversity in Hipposideros caffer complex (Chiroptera: Hipposideridae). Acta Chiropterologica 10. In press. Yang, Z. 1996. Among­site variation and its impact on phylogenetic analysis. Trends in

69 Ecology and Evolution 11:367–372. Yoder, A. D., R. M. Rasoloarison, S. M. Goodman, J. A. Irwin, S. Atsalis, M. J. Ravosa, J. H. Ganzhorn. 2000. Remarkable species diversity in Malagasy mouse lemurs (Primates, Microcebus). Proceedings of the National Academy of Sciences USA 97:1325–1330.

70 Figure 1. Distribution of localities sampled in Niokolo Koba NP, southeastern Senegal.

71 Figure 2. Median­joining network of 29 haplotypes found in the population of NKNP. Size of circles is proportional to frequency of particular haplotypes. Colours correspond to haplogroups A1 (blue), A2 (red) and B (green), and are consistent throughout all figures. Figure 3. Plot of main canonical variables CV1 and CV2 from discriminant analysis of 15 skull dimensions. Colour identification as in Figure 1.

73 Figure 4. Bayesian 50% majority rule consensus tree depicting phylogenetic relationships among West African lineages of H. cf ruber. Bootstrap support for MP and ML >70% (above branches) and posterior probabilities of BA >0.95 (below braches) are considered significant. Colours as in Figure 2.

74 Table 1. Skull dimensions (in millimetres) of the examined specimens. See Methods for abbreviations of dimensions.

haplogroup A1 haplogroup A2 haplogroup B n M min max SD n M min max SD n M min max SD LAt 20 47.59 44.2 49.8 1.354 27 48.10 46.6 49.4 0.829 5 47.08 46.3 48.3 0.792 LCr 19 18.84 18.27 19.24 0.305 26 18.98 18.44 19.62 0.283 5 18.77 18.28 19.11 0.338 LCo 19 18.58 17.78 19.12 0.358 26 18.75 18.21 19.23 0.219 5 18.56 18.14 18.86 0.264 LCc 19 16.28 15.83 16.66 0.278 26 16.39 16.03 16.83 0.195 5 16.23 16.03 16.37 0.149 LaZ 19 10.46 10.02 10.83 0.242 26 10.55 10.11 10.86 0.202 5 10.57 10.28 10.75 0.195 LaI 19 2.86 2.59 3.08 0.130 26 2.89 2.64 3.09 0.114 5 2.82 2.60 2.98 0.143 LaInf 19 5.04 4.85 5.25 0.103 26 5.09 4.81 5.42 0.120 5 5.06 4.98 5.18 0.088 LaN 19 8.31 7.97 8.55 0.153 26 8.32 7.75 8.68 0.193 5 8.27 7.85 8.73 0.370 LaM 19 9.85 9.52 10.04 0.147 26 9.93 9.63 10.28 0.154 5 9.80 9.69 9.97 0.121 ANc 19 5.81 5.29 6.33 0.237 26 5.88 5.49 6.38 0.212 5 6.07 5.80 6.67 0.353 CC 19 4.98 4.73 5.24 0.161 26 4.98 4.74 5.19 0.115 5 4.93 4.75 5.06 0.138 M3M3 19 7.16 6.68 7.42 0.195 26 7.14 6.82 7.42 0.131 5 7.11 6.82 7.25 0.168 CM3 19 7.07 6.81 7.29 0.147 26 7.09 6.78 7.26 0.105 5 6.95 6.84 7.05 0.076 LMd 19 12.32 11.82 12.73 0.246 26 12.43 12.17 12.74 0.142 5 12.41 12.31 12.51 0.091 ACo 19 3.01 2.74 3.29 0.157 26 3.07 2.76 3.31 0.128 5 3.03 2.93 3.23 0.117

CM3 19 7.64 7.37 7.89 0.165 26 7.66 7.38 7.93 0.118 5 7.62 7.55 7.68 0.048 LBT 19 3.33 3.07 3.48 0.100 26 3.37 3.23 3.55 0.084 5 3.34 3.28 3.48 0.080

75 Appendix 1. List of material included in this paper.

sample haplotype haplogroup echo [kHz] country locality GenBank # source of sequence IVB S8 Hap_1 A1 – Senegal Assirik EUwaiting this study IVB S95 Hap_9 A2 – Senegal Simenti – this study IVB S112 Hap_2 A2 – Senegal Lengue EUwaiting this study IVB S119 Hap_3 A1 – Senegal Lengue EU934478 Vallo et al. 2008 IVB S132 Hap_4 B – Senegal Lengue EUwaiting this study IVB S139 Hap_6 A2 – Senegal Lengue EUwaiting this study IVB S218 Hap_8 A2 – Senegal Simenti EUwaiting this study IVB S253 Hap_6 A2 – Senegal Simenti EUwaiting this study IVB S272 Hap_9 A2 – Senegal Simenti EU934481 Vallo et al. 2008 IVB S273 Hap_10 A1 – Senegal Simenti EU934482 Vallo et al. 2008 IVB S275 Hap_11 A2 – Senegal Simenti EU934483 Vallo et al. 2008 IVB S278 Hap_5 A2 – Senegal Simenti EUwaiting this study IVB S280 Hap_12 A2 – Senegal Simenti EUwaiting this study IVB S281 Hap_7 B – Senegal Simenti EUwaiting this study IVB S283 Hap_13 A2 – Senegal Simenti EUwaiting this study IVB S285 Hap_14 A1 – Senegal Simenti EU934484 Vallo et al. 2008 IVB S286 Hap_7 B – Senegal Simenti – this study IVB S290 Hap_15 A2 – Senegal Simenti EUwaiting this study IVB S291 Hap_11 A2 – Senegal Simenti – this study IVB S341 Hap_16 A2 – Senegal Simenti EUwaiting this study IVB S342 Hap_9 A2 – Senegal Simenti – this study IVB S362 Hap_17 A2 – Senegal Simenti EUwaiting this study IVB S403 Hap_7 B – Senegal Dar Salam – this study IVB S695 Hap_18 A2 – Senegal Dar Salam EUwaiting this study IVB S701 Hap_14 A1 – Senegal Dar Salam – this study IVB S702 Hap_19 A2 – Senegal Dar Salam EUwaiting this study IVB S803 Hap_5 A2 – Senegal Lengue – this study IVB S819 Hap_3 A1 – Senegal Dindefelo – this study IVB S820 Hap_20 A1 – Senegal Dindefelo EU934485 Vallo et al. 2008 IVB S821 Hap_21 A1 – Senegal Dindefelo EUwaiting this study IVB S825 Hap_20 A1 – Senegal Dindefelo – this study IVB S899 Hap_5 A2 – Senegal Dar Salam – this study IVB S900 Hap_22 A1 – Senegal Dar Salam EUwaiting this study IVB S1374 Hap_5 A2 – Senegal Dar Salam EU934479 Vallo et al. 2008 IVB S1377 Hap_27 A2 – Senegal Dar Salam EUwaiting this study IVB S1400 Hap_7 B – Senegal Dar Salam EU934480 Vallo et al. 2008 IVB S1538 Hap_28 A2 138 Senegal Dar Salam EUwaiting this study IVB S1539 Hap_11 A2 137 Senegal Dar Salam – this study IVB S1540 Hap_9 A2 134 Senegal Dar Salam – this study IVB S1541 Hap_14 A1 139 Senegal Dar Salam – this study IVB S1551 Hap_23 A1 132 Senegal Dar Salam EUwaiting this study IVB S1554 Hap_23 A1 136 Senegal Dar Salam – this study IVB S1555 Hap_24 A1 137 Senegal Dar Salam EUwaiting this study IVB S1561 Hap_29 A1 133 Senegal Dar Salam EUwaiting this study IVB S1654 Hap_24 A1 133 Senegal Dar Salam – this study IVB S1655 Hap_24 A1 136 Senegal Dar Salam – this study IVB S1657 Hap_25 A1 134 Senegal Dar Salam EUwaiting this study IVB S1660 Hap_11 A2 131 Senegal Dar Salam – this study IVB S1662 Hap_5 A2 131 Senegal Dar Salam – this study IVB S1663 Hap_14 A1 135 Senegal Dar Salam – this study IVB S1664 Hap_14 A1 135 Senegal Dar Salam – this study IVB S1665 Hap_26 A2 130 Senegal Dar Salam EUwaiting this study NMP 91879 Hap_pb53 – – Benin Tagayé EU934476 Vallo et al. 2008 ROM 100546 Hap_r46 – – Ivory Coast Sibabli EF584226 Lim et al. 2008 ROM 100534 Hap_r33 – – Ivory Coast Sibabli EUwaiting this study DSJ HC2 H. caffer – – South Africa Kruger NP EU934452 Vallo et al. 2008

76 Taxonomic revision of the genus Triaenops (Mammalia: Chiroptera: Hipposideridae) with description of a new species from southern Arabia and definitions of new genus and tribe

PETR BENDA1,2 & PETER VALLO3,4

1 Department of Zoology, National Museum (Natural History), Václavské nám. 68, 115 79 Praha 1, Czech Republic. E­mail: [email protected] 2 Department of Zoology, Faculty of Science, Charles University, Viničná 7, 128 44 Praha 2, Czech Republic 3 Institute of Vertebrate Zoology, AS CR, v.v.i., Květná 8, 603 65 Brno, Czech Republic. E­ mail: [email protected] 4 Institute of Botany and Zoology, Masaryk University, Kotlářská 2, 611 27 Brno, Czech Republic

Abstract

The genus Triaenops has been considered monospecific in its African nad Middle Eastern range (T. persicus), while three other species have been recognised to inhabit Madagascar (T. ‘rufus’, T. furculus and T. auritus). We analysed representative samples of T. persicus from East Africa and the Middle East using both morphologic and molecular genetic approaches and compared them with most of the available type material. Morphologic comparison showed four distinct morphotypes in the set, one of them in Africa, the others in the Middle East. The Middle Eastern morphotypes differed mainly in size, while the allopatric African form showed differences in skull shape. Two of three Arabian morphotypes occur in sympatry. Cytochrome b gene based molecular genetic analysis revealed significant divergences (K2P distance 6.4–8.1% in complete cyt b sequence) among most of the morphotypes. Therefore, we propose a split of the current T. persicus rank into three species: T. afer in Africa, and T. persicus and T. parvus sp. nov. in the Middle East. Results of molecular genetic analysis also indicated relative close proximity of the Madagascan T. ‘rufus’ to Arabian T. persicus, what suggests northern way of the colonisation of Madagascar, independently of the African populations. Due to

77 a considerable genetic distance (21.6–26.2% in 731 bp sequence of cyt b) and morphologic differences from the continental forms of Triaenops as well as from Madagascan T. ‘rufus’, we propose generic status (Paratriaenops gen. nov.) for the pair of Madagascan species, T. furculus and T. auritus. The genera Triaenops and Paratriaenops gen. nov. we separated from other hipposiderid bats into Triaenopini trib. nov. regarding their rather isolate position within the family Hipposideridae Lydekker, 1891.

Key words: Triaenops, Triaenops parvus sp. nov., Paratriaenops gen. nov., Triaenopini trib. nov., morphologic analysis, genetic analysis, cytochrome b, Middle East, Afrotropics

Introduction

The hipposiderid genus Triaenops Dobson, 1871 is most known for a typical structure of its noseleaf. Its most distinctive features are four tall pointed processes on the strongly cellularised posterior leaf, three of them forming a trident­like structure on its caudal margin, in combination with the strap­like projection extending forward from the internarial region of the anterior leaf (Dobson 1878; Dorst 1948; Hill 1982). Its distribution range covers mainly the Afrotropics and only marginally the southern Palaearctic. The genus occurs from Iran and Pakistan through southern Arabia to East Africa, from Eritrea and Somalia to Zimbabwe and Mozambique, and to Madagascar and some islands of western Indian Ocean (Harrison 1955, 1963, 1972; Dalquest 1965; Funaioli & Lanza 1968; Kingdon 1974; Largen et al. 1974; DeBlase 1980; Kock & Felten 1980; Harrison & Bates 1991; Happold & Hapold 1998; Cotterill 2001; Pearch et al. 2001; Taylor 2005; Ranivo & Goodman 2006; etc.). Rather isolated records were reported from south­western Congo (Brazzaville) and north­western Angola (Aellen & Brosset 1968; Crawford­Cabral 1989). Within the genus Triaenops, four species are currently recognised (Simmons 2005) and a description of another new species from south­western Seychelles is presently in press (Goodman & Ranivo, in press). Three species are mentioned to inhabit western and northwestern portions of Madagascar (Simmons 2005; Ranivo & Goodman 2006; Russell et al. 2007): T. rufus Milne­Edwards, 1881, T. furculus Trouessart, 1906 and T. auritus Grandidier, 1912. Since the name T. rufus (= T. humbloti Milne­Edwards, 1881) was just recently found unavailable for designation of any Madagascan population of Triaenops

78 (Goodman & Ranivo, submitted), we hereafter use a name form T. ‘rufus’ to name the respecive species while the new name is formally designated (Goodman & Ranivo, submitted). However, from the extensive belt of savannas of East Africa as well as from Congo and southern parts of the Middle East, only one species is reported, Triaenops persicus Dobson, 1871 (Hill 1982; Koopman 1993, 1994; Duff & Lawson 2004; Simmons 2005). Within the rank of the latter species, four names were proposed and three of them were accepted as those of separate subspecies (Hill 1982; Simmons 2005). T. afer Peters, 1877, described and for a long time considered as a separate species (Dobson 1878; Trouessart 1904; Miller 1907; Allen 1939; Tate 1941; Dorst 1948; Aellen 1957; Harrison 1961, 1963), is currently regarded a subspecies of T. persicus inhabiting East African part of range (Aellen & Brosset 1968; Funaioli & Lanza, 1968; Hayman & Hill 1971; Kingdon 1974; Largen et al. 1974; Corbet 1978; Hill 1982; Aggundey & Schlitter 1984; Koopman 1994; etc.). Some authors (Harrison 1964; Aellen & Brosset 1968; Corbet 1978; Hill 1982; Nader 1990; Harrison & Bates 1991; Koopman 1994; Al­Jumaily, 1998) assigned to this subspecies also individuals found in the former Aden Protectorate (= SW Yemen) (cf. Yerbury & Thomas 1895), however, such opinion does not conform with some older authors (e.g., Thomas 1900; Miller 1907; Dorst 1948; Ellerman & Morrison­Scott 1951). T. p. persicus is reported to inhabit the Middle East, including Pakistan, Iran, United Arab Emirates, Oman and possibly Yemen; the subspecies T. p. macdonaldi Harrison, 1955, described from U. A. E., is considered a junior synonym of the former name by the most of recent authors (DeBlase 1980; Hill 1982; Koopman 1994; Simmons 2005; contra Harrison 1955, 1956, 1964; Atallah & Harrison 1967; Nader 1990; Harrison & Bates 1991). The geographically well isolated Congolese population of T. persicus was described as a separate subspecies, T. p. majusculus Aellen et Brosset, 1968. Hill (1982) and Koopman (1994) regarded also the population of Uganda belonging to this subspecies. Hill (1982) discussed a possible subspecific position of the Madagascan form T. ‘rufus’ under T. persicus, this opinion was not accepted by modern authors (Peterson et al. 1995; Garbutt 1999; Duff & Lawson 2004; Simmons 2005; Ranivo & Goodman 2006; Russell et al. 2007; Goodman & Ranivo, submitted), with an exception of Koopman (1993, 1994). The particular subspecies of T. persicus were separated by minute differences in pelage coloration and body size (Hill 1982). Indeed, a clinal trend to increase of body size from

79 north­east to south­west is evident within this species rank. T. p. persicus was reported to be on average the smallest and T. p. majusculus the largest form among its subspecies; moreover, the Arabian populations of T. persicus were reported to demostrate the largest size variation among all the subspecies (Hill 1982; Harrison & Bates 1991). Intrageneric taxonomy of Triaenops has been reviewed only several times (Table 1), and from two to five species have been recognised within this genus. Here, we present results of analysis of Triaenops persicus (sensu e.g. Simmons 2005 = T. persicus s.l.) samples from rather northern part of its distribution range, mostly newly collected ones, conducted with the aim of defining the intraspecific variation of this variable species and to checking the validity of the current intraspecific, intrageneric and partly also intrafamilial taxonomy.

Material and Methods

We analysed representative samples of museum specimens of T. persicus s.l. from East Africa (Ethiopia, Tanzania), Madagascar and the Middle East (Yemen) using morphologic and molecular genetic approaches. This material was compared with type specimens of the genus Triaenops; viz. ZMB syntypes of Triaenops persicus Dobson, 1871 (type locality: Shiraz, Persia); ZMB holotype of Triaenops afer Peters, 1877 (t.l.: Mombaça [= Mombasa, Kenya]; see Turni & Kock 2008); MNHN type series of Triaenops rufus Milne­Edwards, 1881 (t.l.: Madagascar [= east coast of Madagascar sensu e.g. Hill 1982, but apparently incorrect, see Goodman & Ranivo, submitted]); MNHN type series of Triaenops humbloti Milne­Edwards, 1881 (t.l.: Madagascar [= east coast of Madagascar sensu e.g. Hill 1982, but apparently incorrect, see Goodman & Ranivo, submitted]); MNHN type series of Triaenops furcula Trouessart, 1906 (t.l.: Grotte de Sarondrana, W Madagascar); and MNHN type series of Triaenops persicus majusculus Aellen et Brosset, 1968 (t.l.: Grotte de Doumboula, Loudima (Kouilou), Congo). For material used in morphologic analysis see Appendix 1, for that in genetic analysis see Appendix 2. For morphologic comparisons, the museum specimens were examined in the same way as described in our previous studies (e.g. Benda et al. 2004a, b). For morphologic analysis, we used mainly the skull metric dimensions in order to describe morphologic trends in

80 particular populations rather than an individual variation. The specimens were measured in a standard way with the use of mechanical or optical calipers. The evaluated external, cranial and bacular measurements are listed in Abbreviations. With exception of the MNHN and ZMB specimens, the external dimensions were taken from freshly collected material. Bacula were extracted in 6% solution of KOH and coloured with the alizarin red. Statistical analyses were performed using the Statistica 6.0 software. In the genetic analysis, we used a subset of museum specimens of Triaenops persicus from Ethiopia and Yemen, along with specimens of another two African hipposiderids Cloeotis percivali Thomas, 1901 and Asellia tridens (Geoffroy, 1913), and three African rhinolophid bats Rhinolophus alcyone Temminck, 1853, R. fumigatus Rüppell, 1842 and R. landeri Martin, 1838. From the GenBank database we retrieved sequences of East African (Tanzanian) T. persicus and Madagascan T. ‘rufus’, T. furculus and T. auritus (Russell et al. 2007), sequences of Hipposideros abae Allen, 1917, H. caffer (Sundevall, 1846) and H. jonesi Hayman, 1947 (Vallo et al. in press), and Aselliscus stoliczkanus (Dobson, 1871), A. tricuspidatus (Temminck, 1835) and Coelops frithii Blyth, 1848 (Li et al. 2007). Sequences of vespertilionid bats Vespertilio murinus (Linnaeus, 1758), Myotis nattereri (Kuhl, 1817) and Myotis schaubi Kormos, 1934 (Ruedi & Mayer 2001), which were used as outgroup, were also taken from the GenBank. Sequences for phylogenetic analysis were obtained by standard laboratory procedures. Genomic DNA was extracted from alcohol preserved tissue samples with DNA Blood and Tissue Kit (Qiagen) following manufacturer’s protocol. Complete sequence of the mitochondrial gene for cytochrome b (cyt b) was PCR amplified using primers F1 (modified; 5’­CCACGACCAATGACAYGAAAA­3’) and R1 (5’­ CCTTTTCTGGTTTACAAGACCAG­3’) from Sakai et al. (2003) in 50 μ l reaction volume containing 800 μ M dNTP, 200 μ M of each primer, 1U of HotMaster Taq DNA polymerase with appropriate 10× buffer (Eppendorf), and 2–5 μ l of extracted DNA. Reaction conditions were 3 min initial denaturation at 94 °C, 35 cycles of 40 s denaturation at 94 °C, 40 s annealing at 50 °C and 90 s extension at 65 °C, and 5 min final extension at 65 °C. Products were purified using QIAquick PCR Purification Kit (Qiagen), and sequenced commercially in both directions on ABI 3730XL sequencing machine with the same primers and BigDye Terminator Sequencing Kit (Applied Biosystems). Two ca. 800 bp long partially overlapping fragments were obtained and subsequently assembled in

81 Sequencher (GeneCodes) into complete sequences of cyt b (1140 bp). Final sequences were submitted to the GenBank database under accesion numbers EU798748–EU798758 and EUwaiting­EUwaiting. Sequences were aligned in BioEdit 7.0 (Hall 1999). Alignment of 1140 bp was built from newly obtained sequences of Triaenops persicus and Cloeotis percivali, and was used for assessment of genetic variation. Sequences of Triaenops species retrieved from the GenBank were then added to the new haplotypes and the alignment was trimmed to 731 bp, which was the length of the GenBank Triaenops sequences. This 731 bp alignment was used for inferring phylogenetic relationships in the genus Triaenops. After this analysis, Triaenops sequences were reduced to one of each phylogroup and sequences of the other species were added to the 731 bp alignment, with the intention of inferring phylogenetic position of Triaenops species within the family Hipposideridae. Phylogenetic analyses were done in programs PAUP* 4.10b (Sinauer Associates) and MrBayes 3.1.2 (Ronquist & Huelsenbeck 2003). The 731 bp alignment of all Triaenops species was analysed using maximum parsimony (MP), neighbour­joining (NJ) and maximum likelihood (ML) methods. MP and ML trees were heuristically searched with 100 additions of sequences and tree bisection­reconnection branch­swapping algorithm. HKY+I+Γ model of evolution was used in computation of ML tree, as suggested by the program Modeltest 3.7 (Posada & Crandall 1998) under AIC criterion. NJ tree was calculated under Kimura’s two­parameter substitution model (K2P; Kimura 1980). Support for topologies of MP and NJ trees was checked by 1000× bootstrapping, ML tree support was checked by 300× bootstrapping of 20 sequence additions only. Analysis of the extended dataset started also with MP method. From the MP tree, transition to transversion ratio was estimated and used to weight transversions 5 times against transitions in weighted MP (wMP) analysis. Both MP and wMP trees were searched as in the first step, including support of topology for wMP. Phylogeny was further inferred using maximum likelihood (ML) and Bayesian (BA) methods. ML tree was heuristically searched with 100 random additions of sequences and TBR swapping algorithm under TVM+I+Γ model of evolution, which was suggested under the AIC criterion in Modeltest 3.7. Support for its topology was assessed by 300 bootstrap replicates of 20 random sequence additions only. Bayesian analysis was carried out as a combined analysis of three partitions, corresponding to three codon positions of cyt b.

82 Each partition had its own rate of evolution and independent estimates of model parameters. Suitable models of evolution, SYM+Γ for the 1st, HKY+I for the 2nd and GTR+I+Γ for the 3rd codon position, were selected with under the AIC criterion help of the program MrModeltest 2.3 (Nylander 2002). Two simultaneous Metropolis­coupled MCMC runs of four chains were run for 1000 000 generations, sampled each 100 generation, and burning was set to 25%. Genetic divergences among haplotypes were based on K2P distances, which is a considered a ‘standard’ measure for comparison with other studies on bats (Bradley & Baker 2001). For testing of alternative topologies, Templeton (Templeton 1983) and Shimodaira­ Hasegawa (SH; Shimodaira & Hasegawa 1999) tests were conducted for MP and ML method, respectively, using RELL resampling algorithm and 1000 replicates in SH test. Relevant constraints were used in heuristic searches of trees under the same conditions as the unconstrained ones. Approximate dates of divergencies were estimated from a linearized tree (Takezaki et al. 1995), computed under ML criterion with molecular clock enforced. The assumption of clock­like evolution for the dataset was tested with likelihood ratio test between trees with and without molecular clock. Calibration of the molecular clock was based on the split of Rhinolophidae and Hipposideridae set approximately to 40 MA (= Mega Annum), according to estimation range of 43–37 MA (Remy et al. 1987; Simmons & Geisler 1998)

Abbreviations

COLLECTIONS. BCSU – Biological Collection of the Sana’a University, Sana’a, Yemen; DNSM – Durban Natural Science Museum, Durban, South Africa; IVB – Institute of Vertebrate Biology AS CR, Brno, Czech Republic; MNHN – National Museum of Natural History, Paris, France; NMP – National Museum (Natural History), Prague, Czech Republic; ZMB – Zoological Museum, Humboldt University, Berlin, Germany.

MEASUREMENTS. External: LC = head and body length; LCd = tail length; LAt = forearm length; LA = auricle length; LaFE = horseshoe width; G = body weight. Cranial: LCr = greatest length of skull incl. praemaxillae; LOc = occipitocanine length of skull; LCc = condylocanine length of skull; LaZ = zygomatic width; LaI = width of interorbital constriction; LaN = neurocranium width; LaM = mastoidal width of skull; ANc =

83 neurocranium height; LBT = largest horizontal length of tympanic bulla; CC = rostral width between canines (incl.); M3M3 = rostral width between third upper molars (incl.); CM3 = length of upper tooth­row between CM3 (incl.); LMd = condylar length of mandible; ACo = height of coronoid process; CM3 = length of lower tooth­row between

CM3 (incl.). Bacular: LBc = total length of baculum; LBcB = basal length of baculum (i.e. without proximal appendices); LaMin = least width of baculum diaphysis; LaProx = largest width of proximal epiphysis; LaDist = largest width of distal epiphysis (across arms); LArBc1 = length of the longer arm; LArBc2 = length of the shorter arm; AnBc = angle of bacular arms.

OTHER ABBREVIATIONS. A = alcoholic preparation; f = female; M = mean; m = male; min, max = dimension range margins; S = skull; SD = standard deviation.

Results

Morphologic comparison Analysis of body and skull dimensions showed several more or less distinct morphotypes within the examined set of samples. According to a mere comparison of skull dimensions, three size types appeared among the examined geographical samples of specimens, however, they mostly overlapped in their measurement ranges (Fig. 1; Table 2); (1) small­ sized bats from Madagascar (LAt 42.5–50.4 mm; LOc 16.9–18.7 mm; CM3 5.9–6.5 mm) composed of two nominate species, T. furculus and T. ‘rufus’, (2) large­sized bats from Africa (LAt 52.7–56.6 mm; LOc 17.9–20.5 mm; CM3 6.3–7.5 mm), and (3) the Middle Eastern bats with an extreme size variation stretching over the ranges of the two precedent groups (LAt 44.7–57.3 mm; LOc 16.3–20.8 mm; CM3 5.8–7.7 mm). The Madagascan and African size types are of a rather little size variation, covering with their ranges ca. one and two thirds of the extent of size range of the Middle Eastern size type, respectively. The bats of the African size type showed relatively short and wide rostrum (CM3/LOc 0.34–0.36 [M 0.353]; CC/LOc 0.24–0.28 [M 0.266]; CC/CM3 0.67–0.79 [M 0.754]) and relatively and absolutely rather large tympanic bullae (LBT/LOc 0.15–0.17 [M 0.159]). Into the dimensional range of the African morphotype fell well the dimensions and ratios of the type specimen of T. afer Peters, 1877 from Kenya as well as of the type specimens

84 of T. persicus majusculus Aellen et Brosset, 1968 from Congo (Figs. 1 and 2; Tables 2 and 3). Some specimens from the latter type series showed rather larger forearm length (up to 59.5 mm), however, average length in that series was 56.0 mm, i.e. the value lesser than the average value in the African group as a whole (Table 2). The bats of the Madagascan size type showed relatively short but rather narrow rostrum (CM3/LOc 0.34–0.37 [M 0.353]; CC/LOc 0.23–0.26 [M 0.252]; CC/CM3 0.67–0.76 [M 0.714]) and also relatively and absolutely large tympanic bullae (LBT/LOc 0.15–0.18 [M 0.162]). However, the samples (type series) of T. furculus showed relatively longer and on average also narrower rostrum than those of T. ‘rufus’. Within the Middle Eastern set were present bats with both relatively short and rather narrow rostrum (the specimens absolutely smaller in size) and also with relatively long and rather wide rostrum (the specimens absolutely larger in size) (CM3/LOc 0.34–0.37 [M 0.360]; CC/LOc 0.24–0.28 [M 0.256]; CC/CM3 0.67–0.75 [M 0.712]; Fig. 2; Table 2); this group of samples as a whole showed relatively small tympanic bullae (LBT/LOc 0.14– 0.17 [M 0.157]). The Middle Eastern group was, however, represented by specimens of three size groups according to their geographic origin with neither or very small dimensional overlap, respectively (Fig. 2, Table 2). (1) group of six individuals collected in western Yemen (NMP 92275–92279, BCSU pb3123) were of the largest skull size within the whole set of compared Triaenops bats (LAt 54.7–57.3 mm; LOc 19.2–20.8 mm; CM3 7.0–7.7 mm); this group overlapped in longitudinal skull dimensions with the largest individuals of the African morphotype (Fig. 2; Table 2). (2) group of medium­sized to large specimens (NMP 92253–92263, 92266, 92271, 92273, BCSU pb3037, pb3038) from south­eastern Yemen (LAt 48.0–55.1 mm; LOc 17.7–19.9 mm; CM3 6.4–7.3 mm; Table 2) conformed in size with the syntypes of T. persicus Dobson, 1871 from Iran (Table 3) and also with published dimensions of T. persicus from the Middle East (see Harrison 1955, 1964; DeBlase 1980; Hill 1982; Harrison & Bates 1991; etc.). To the range of dimensional overlap of these medium­sized bats with the largest ones fitted the dimensions of the type specimens of T. rufus Milne­Edwards, 1881 and T. humbloti Milne­Edwards, 1881 (LAt 51.5–56.1 mm; LOc 19.4–20.1 mm; CM3 7.1–7.4 mm). (3) group of small individuals coming from the south­eastern part of Yemen (NMP 92264, 92265, 92267–92270, 92272, 92274, BCSU pb3009, pb3010), i.e. an area of sympatry with the medium­sized bats,

85 demostrated absolutely smallest dimensions within the compared set of bats (LAt 44.7– 48.1 mm; LOc 16.4–17.4 mm; CM3 5.8–6.2 mm) (Table 2). While the latter group of the smallest specimens (hereafter called as the morphotype A of the Middle Eastern samples) showed relatively short and narrow rostrum (CM3/LOc 0.34–0.36 [M 0.351]; CC/LOc 0.24–0.26 [M 0.252]; CC/CM3 0.69–0.74 [M 0.718]) and relatively very large tympanic bullae (LBT/LOc 0.15–0.17 [M 0.162]) although absolutely the smallest ones (Table 2), the group of medium­sized bats from south­eastern Yemen (morphotype B) and large specimens from western Yemen (morphotype C) exhibited relatively smaller bullae (LBT/LOc in morphotype B: 0.14–0.16 [M 0.155]; in C: 0.15– 0.16 [M 0.156]) and relatively long and wide rostrum (CM3/LOc in morphotype B: 0.35– 0.37 [M 0.363]; in morphotype C: 0.36–0.37 [M 0.366]; CC/LOc in B: 0.24–0.27 [M 0.256]; in C: 0.25–0.28 [M 0.264]; CC/CM3 in B: 0.67–0.74 [M 0.706]; in C: 0.69– 0.75 [M 0.720]). To be summarised, the Middle Eastern samples were composed of at least two clearly distinct morphotypes differing in size, rostrum shape and relative size of bulla, A vs. B+C, where later B and C differed in size. Size exclusivity of the skull morphotype A among the Middle Eastern bats was confirmed also by principal component analysis based on nine most variable skull dimensions (see below for their selection); the first principal component (representing some 89.89% of the whole metric variance) clearly separated the morphotype A (PC1<1.2) from the common cluster of remaining two morphotypes B+C (PC1>–0.5) according to skull size expressed by the large skull dimensions (not figured). Mutual positions of Triaenops skull morphotypes defined above according to the absolute size of skull, the relative size of tympanic bullae and to the shape of rostrum as well as their affinities to the examined type material was showed by discriminant function analyses (Figs. 3 and 4). The analysis of the whole set of examined skulls selected nine most variable dimensions (LCr, LOc, LaI, LaM, ANc, CC, CM3, LMd, ACo; CV1=57.68% of variance; CV2=25.50%; Fig. 3) analysis of these selected dimensions clearly separated the most differing samples (CV1>8), the type series of Triaenops furculus from Madagsacar, apart from all other samples (CV1<5). In the common cluster of the remaining samples, there was possible to distinguish three groups of specimens; (1) a group (CV1<–0.1;5.0>; CV2<–1.3) composed of small individuals of T. persicus from the Middle East (morphotype A) and of T. ‘rufus’ from Madagascar; (2) a group (CV1<–

86 3.4;0.4>; CV2<–3.0;1.6>) composed of African specimens from Ethiopia and Tanzania as well as the type specimens of T. afer and T. persicus majusculus; and (3) a group (CV1 <– 4.3;–0.3>; CV2 <–0.9;4.7>) composed of remaining Middle Eastern samples (morphotypes B and C) and type series of T. persicus, T. rufus and T. humbloti (Fig. 3). The discriminant analysis of all 15 skull measurements of the whole set of examined skulls with exception of those of T. furculus (separated as most different by the previous analysis) clustered four groups of samples (CV1=57.23% of variance; CV2=26.28%; Fig. 4). Similarly as the previous analysis, it indicated the group of African samples (CV1 <– 1.8;–2.4>; CV2 <0;3.8>), but the rest of specimens clearly clustered according to their geographic origin and also to belonging to the above defined skull morphotypes; these groups almost did not overlap. Group of T. ‘rufus’ from Madagascar (CV1 <2.9;6.6>; CV2 <–1.0;2.9>), a close up group of smallest individuals (morphotype A) from south­eastern Yemen (CV1 <1.5;3.4>; CV2 <–4.0;–2.4>), and two groups of larger individuals partly overlapping with each other from the Middle East (western and south­eastern Yemen, morphotypes B and C) together with group of the type specimens of T. persicus, T. rufus and T. humbloti (CV1 <–6.0;–0.6>; CV2 <–2.5;2.4>). Although Middle Eastern bats of the morphotype C were on average the largest ones according to the first canonical variable (CV1), they widely overlapped in the first two variables with the cluster of the specimens of morphotype B. Bats of the four morphotypes of Triaenops coming from northern part of the genus range (samples of African bats from Ethiopia and of three Middle Eastern morphotypes A, B, C from Yemen) were additionally examined for noseleaf, baculum, and coloration variation. Noseleaf was in the compared samples of the identical form, differing only in size, which, however, depended on the body size of the respective specimen (Table 2). Small individual variation was found only in noseleaf pigmentation (see below). Examination of bacula extracted from the examined specimens (two bacula per each skull morphotype) showed nearly uniform shape of bone, an elongated stick (length 1.5– 2.1 mm) extended to broad pyramid in proximal epiphysis and bifurcated at distal epiphysis (Fig. 5). Besides the slight differences in size, we found some minute differences in baculum shape. Most distinct bacula came from the Ethiopian bats showing slightly more rubust diaphysis (relative width of diaphysis 0.12 and 0.16), longer and robust distal arms (relative length of arm 0.27–0.29) and more robust proximal epiphysis (relative width

87 of the basis 0.44 and 0.48) than in other samples. Other distinct baculum shape was demostrated by the samples of the SE Yemeni morphotype A, in which it was gracile bone (relative width of diaphysis 0.08 in both bones) with short arms (relative length of arm 0.17–0.20) and narrow proximal epiphysis (relative width of the basis 0.23 and 0.31); in both bones was also observable distinct proximal projection (possibly ossified distal part of the erectile body), which was present only in one from the rest of examined bacula. Bats of the Yemeni morphotypes B and C exhibited similar structures of baculum, by its shape and relative dimensions lying in between the above characterised two baculum morphotypes (Fig. 5, Table 4). Principal component analysis of eight bacular dimensions clearly separated three clusters of samples conforming with the above mentioned three groups (PC1=57.22% of variance; PC2=18.98%); (1) a pair of African samples (PC1<1; PC2<0), (2) a pair of Yemeni samples of the morphotype A (PC1>1; PC2<0) and (3) common cluster of the Yemeni morphotypes B and C (PC2>0) (not figured). Pelage coloration of the compared samples from Ethiopia and Yemen exhibited wide variation mostly depending on the sample size, with an exception of the Yemeni morphotype A. In the latter bats the coloration was uniformly beige or pale brownish­grey above without any tinge of reddish or rusty colours (which was present in some individuals of all the remaining morphotypes), very pale beige to pale greyish­brown below and with a pale (in alcohol fixed specimens, i.e. unpigmented) to pale greyish­brown coloured noseleaf (see Fig. 6 for face coloration of two syntopically collected individuals of south­ eastern Yemeni morphotypes A and B). Most bright pelage was found in Ethiopian bats, in which it was deep greyish­brown, dark brown or reddish­brown above, pale beige to brown below, with pale (unpigmented) to greyish­brown noseleaf. In the most numerous samples of the SE Yemeni morphotype B was the dorsal pelage from pale greyish­brown over reddish­brown to dark greyish­brown, ventral pelage beige, pale grey or pale rusty to greyish­brown and/or deep grey, with pale grey (almost unpigmented) to brown or dark grey noseleaf. In the western Yemeni morphotype C the dorsal pelage was greyish brown to dark reddish­brown, ventral pelage pale grey to dark greyish­brown, and noseleaf pale beige (unpigmented) or dark greyish­brown. Wing membranes were found in all samples dark brown, without any well observable distinctions of the tinge.

88 Genetic comparison We analysed 20 samples of T. persicus and obtained 17 complete sequences of cyt b (1140 bp). From three samples, only an initial portion of cyt b ca. 600 bp long could be recovered but these matched to other complete sequences obtained (Appendix 2). The obtained sequences corresponded to 11 Triaenops haplotypes, two unique haplotypes were recovered from Cloeotis samples. Genetic divergences among Triaenops haplotypes ranged 0.1–8.1%, among Triaenops and Cloeotis 22.4–24.9% (Table 5). Bats of the two Middle Eastern Triaenops skull morphotypes B and C showed minute genetic distance of 0.0–0.2% to each other (i.e. an identical haplotype, ME1, was found in both the morphotypes, see Appendix 2), while genetic difference among these two sample sets and those of the Middle Eastern skull morphotype A ranged 6.4–6.7%. The East African group of samples differed from all three Middle Eastern morphotypes at 7.1–8.1%. After appending sequences of Triaenops from the GenBank and trimming to 731 bp, the new Triaenops haplotypes shrinked to eight and Cloeotis to one (Appendix 2). The 731 bp dataset thus contained 17 ingroup sequences. MP analysis revealed 12 shortest trees 667 steps long that differed in arrangement of well supported monophyletic clade from continental Africa and the Middle East (Fig. 7). NJ and ML (–lnL=3724.69385) trees basically agreed with most MP trees and showed the same well supported monophyletic clades, except that in the ML tree, T. ‘rufus’ haplotypes did not form a monophyletic clade and were even paraphyletic to other African and Middle Eastern haplotypes. This position of T. 'rufus' haplotypes, however, was not supported by bootstrap. These clades corresponded to the above defined grouping of the Middle Eastern skull morphotypes A, B and C of T. persicus, East African morphotype of T. persicus comprising Ethiopian and Tanzanian haplotypes, and three other Triaenops species from Madagascar. From the basal node, Cloeotis percivali diverged as the first taxon differing 22.5–26.9% from the Triaenops haplotypes. A deep split divided the Triaenops haplotypes into two main lineages differing 21.6–26.2%. One lineage represented the Madagascan sister species T. furculus and T. auritus, which differed at 4.1–4.6%. The other lineage comprised four clades: East African T. persicus, Middle Eastern T. persicus morphotype A, morphotypes B+C and the Madagascan T. ‘rufus’. Within the East African clade, Ethiopian haplotypes differed 1.1–1.4% from Tanzanian ones. Genetic divergences among the Middle Eastern morphotypes A and B+C of T. persicus, and Madagascan T. ‘rufus’ ranged 6.8–8.4%

89 (Table 6). Relationships among the four clades remained unresolved by bootstrap analysis, although MP and NJ trees suggested affinity of T. ‘rufus’ to the Middle Eastern clades A and B+C. Therefore, this hypothesis was tested against hypothesis suggested by the ML tree. Also two other alternative hypotheses assuming affinity of T. ‘rufus’ to the African clade, and paraphyly of monophyletic T. ‘rufus’ clade to other African and Middle Eastern clades. SH test proved that monophyly of the Middle Eastern haplotypes and T. ‘rufus’ was not significantly different from the ML tree (diff. –lnL=0.61436, d.f.=17, P= 0.726), and could not be rejected. The other two alternative hypotheses differed significantly from the ML tree (diff. –lnL=30.95783, P=0.004; diff. –lnL=32.28474, P=0.007), and also from the alternative tree assuming monophyly of the Middle Eastern haplotypes and T. ‘rufus’ (diff. –lnL=30.34347, P=0.004; diff. –lnL=31.67038, P=0.004). The extended 731 bp dataset of hipposiderids and rhinolophids contained 18 ingroup sequences, of which three haplotypes of T. persicus represented the morphotypes/phylogroups from the Middle East A, B+C and East Africa from previous analysis. In the alignment, 336 positions were variable, 289 of them parsimony informative. Approximately 22% of substitutions occurred at 1st, 8% at 2nd, and 70% at 3rd codon positions. Base composition did not differ among ingroup sequences (χ2=24.437, d.f.=48, P=0.998) and mean values for base frequencies were A=0.272, C=0.306, G=0.153, and T=0.268. Weighted MP yielded two most parsimonial trees with length of 2832 steps. These two trees topologically differed in the position of Triaenops clade, which was sister either to other hipposiderids or to rhinolophids, without bootstrap support, though. Highly supported were two lineages of Triaenops but not their sister relationship. Cloeotis percivali clustered with other hipposiderids instead of Triaenops but its position also was not supported. ML tree (–lnL=5890.55931) and BA consensus tree exhibited basically the same topology, differing in the position of Cloeotis percivali, which clustered as a sister to Madagascan Triaenops in ML tree and as sister to all Triaenops in BA tree. However, ML and BA trees were congruent with wMP trees in grouping rhinolophids, African and Middle Eastern Triaenops, Madagascan Triaenops, and other hipposiderids into monophyletic clades (Fig. 8). Monophyly of the two Triaenops lineages was not supported by bootstrap or posterior probability, as well as the sister position of C. percivali to Triaenops species. Similarly as in wMP trees, relationships of Triaenops to other hipposiderids and rhinolophids remained unresolved. Alternative phylogeny, which

90 considered sister position of Cloeotis to Triaenops within Hipposideridae, did not differ significantly from wMP (Templeton test; diff. length=8, z=–0.5252, P=0.5994) and ML (SH test; diff. –lnL=0.60680, P=0.342) trees and thus could not be rejected. Because we could not reject the traditional phylogeny of Hipposideridae, we kept on it in rough assessment of divergence times in molecular dating of phylogeny. Clock­like phylogenetic tree was computed under topological constraints of assuming monophyly of the genus Triaenops, monophyly of Triaenops and Cloeotis, and monophyly of Hipposideridae. Vespertilionid outgroup taxa were excluded from this clock­like tree, as these negativelly affected stationarity of base frequencies (χ2= 82.825173, d.f.=57, P=0.014). Likelihood­ratio test of ML tree with (–lnL=4881.69411) and without molecular clock (–lnL=4870.69325) could not reject the molecular clock assumption (diff. – lnL=11.00086, d.f.=15, P=0.1077) under HKY+I+Γ model of evolution. According to the assumed monophyly of Hipposideridae, three rhinolophids were used for rooting the tree. However, the most basal branch was collapsed to the root and the topology remained unresolved with three lineages emanating from root: (1) Rhinolophidae, (2) Triaenops and Cloeotis, and (3) other Hipposideridae. Estimates of approximate minimal dates of splits among lineages are visualised in the linearised tree (Fig. 9).

Discussion

The combination of results of the above morphologic and molecular genetic analyses proved existence of six distinct evolutionary units within the genus Triaenops (sensu Simmons 2005 = Triaenops s.l.). They differed well in size and skull morphology and in genetic traits as well as in geographic distribution. Largest distance, both in morphology and genetics, was present between the pair of Madagascan species T. furculus and T. auritus and the whole remaining content of the genus. These two distant groups were formerly distinguished as two species by Hill (1982) and Koopman (1993, 1994), differing in skull structure and shape, ear shape and significantly also in structure of the noseleaf (Dorst 1948; Hayman & Hill 1971; Hill 1982; Koopman 1994; Ranivo & Goodman 2006). However, within these ‘species’ hidden deeper distinctions were found of clearly different and more fine morphologic value than Hill (1982) described as sufficient for specific level.

91 Extraordinary genetic distance between these two groups exceed the intergeneric distance among other hipposiderids (e.g. 17.2% between Hipposideros Gray, 1831 and Aselliscus Tate, 1941; see Wang et al. 2003) and even interfamilial distance among rhinolophids and hipposiderids and overlap the range of distances between the presumably sister genera Cloeotis and Triaenops s.l. Such a considerable distance as well as the double categorial type of morphologic differences, justify us to separate the Madagascan forms (besides T. ‘rufus’) into a separate genus. Such a new genus, however, shows most similarities with the genus Triaenops s.str. and both these genera evidently compose a natural evolution unit, being a sister lineage to remaing hipposiderid taxa. For this unit we therefore propose a new tribe (see Taxonomic part below), while other hipposiderid taxa we consider members of the tribe Hipposiderini Lydekker, 1891 (with an exception of the genus Cloeotis Thomas, 1901; for resolving its position within the family, genetic analysis using a marker with a lower mutational rate should be done). In the Taxonomic part, also morphologic and genetic differences among the taxa mentioned are specified in details. The assessment of phylogenetic relationships of Triaenops s.l. to other members of the family Hipposideridae brought additional interesting results. Although none of the methods used could fully resolve the phylogeny, our results indicate that the family Hipposideridae is not a monophyletic group as already suggested by e.g. Hulva & Horáček (2002) or Hoofer & Van Den Bussche (2003). A rather compact lineage comprising genera Asellia, Aselliscus, Coelops and Hipposideros stays separately from the genera Triaenops s.l. and Cloeotis, which form a loosely defined phylogroup showing larger genetic divergencies to the other members of Hipposideridae than are divergences among the other hipposiderids to Rhinolophidae. In contrast, a close relationship of two distinct lineages of Triaenops s.l. and Cloeotis could not be confirmed by bootstrapping in any of the phylogenetic methods used. Similarly, resolution at the basal node of phylogeny remained obscure suggesting trichotomic evolution of the family Rhinolophidae, consisting of the lineages; (1) rhinolophids, (2) a lineage comprising Triaenops s.l. and Cloeotis (delimited mostly on morphologic traits, see e.g. Hill 1982), and (3) other hipposiderids. Such a weak resolution can be undoubtedly influenced by high saturation in cyt b sequences at large genetic distances. Nevertheless, it may also indicate a rapid radiation of the respective forms, as wMP and ML methods can handle the effect of saturation by adopting a proper weighting scheme and model of nucleotide substitution. Testing of best hypotheses resulting from

92 different phylogenetic methods against currently accepted systematic perception of Hipposideridae provided an ambiguous solution. The alternative hypothesis assuming monophyly of the genus Triaenops s.l. and a sister relationship of Triaenops and Cloeotis within monophyletic Hipposideridae could not be rejected based on our limited sequence data. Genetic markers with a lower mutational rate should be employed to obtain a definite resolution of this issue. Within the current Triaenops persicus content (sensu Simmons 2005), three evolutionary units were proved. The first unit is represented by Yemeni bats of the morphotype A, extremely small individuals (absolutely smallest within the examined set of Triaenops), living in sympatry and even syntopy with bats of the morphotype B in south­ eastern Yemen, which are medium­sized to large. The morphotype C coming from westernmost Yemen was characterised by the largest size among compared bats, however, in the most of characters (skull structure, baculum) it was close to or just overlapped with the morphotype B. These two morphotypes (B+C), differing in size but neither or almost imperceptibly differing in the examined genetic traits (four haplotypes differing in one substitution from each other, i.e. in 0.1%), represent the second unit. The types of T. persicus, T. rufus and T. humbloti fell also into ranges of dimensions of this unit. On the other hand, the sympatric morphotypes A and B, besides their size and morphologic differences, diverged in 6.4–6.7% of the complete sequence of cyt b gene. Such a value lies within the range of interspecific genetic divergencies referred for Hipposideridae and other bat families (Baker & Bradley 2006; Vallo et al. 2008), thus these two units (morphotypes A and B+C) should be considered separate species. The third phylogenetic unit is composed of the fourth morphotype, found in the African samples (Ethiopian and Tanzanian specimens and the types of T. afer from Kenya and T. persicus majulusculus from Congo­Brazzaville), differing in the structure of skull and baculum from the Middle Eastern morphotypes A, B and C and markedly in size from the Middle Eastern morphotype A and Madagascan T. ‘rufus’. This last unit also diverges in genetic traits from the Yemeni group of morphotypes (7.1–8.1% at 1140 bp and 7.4–8.7% at 731 bp of cyt b, respectively), i.e. by a larger distances than the sympatric A and B morphotypes. This situation clearly suggests that all three here defined particular phylogenetic units currently enclosed into the species rank of T. persicus (Simmons 2005) represent three separate species.

93 From the area of the Middle East and Africa, five names of the genus Triaenops are presumably available; T. persicus Dobson, 1871 (type locality: Shiraz, Iran), T. afer Peters, 1877 (t.l.: Mombasa, Kenya), T. rufus Milne­Edwards, 1881 (t.l. unknown [east coast of Madagsacar sensu e.g. Hill 1982, but apparently incorrect, see Goodman & Ranivo, submitted]), T. humbloti Milne­Edwards, 1881 (t.l. unknown [east coast of Madagascar sensu e.g. Hill 1982, but apparently incorrect, see Goodman & Ranivo, submitted]), and T. persicus macdonaldi Harrison, 1955 (t.l.: Al Ain, U. A. E.). Bats of the African morphotype from our set corresponded in their traits with those of the holotype of T. afer; haplotypes of the Ethiopian samples were shown to be closest to the Tanzanian ones (sensu Russell et al. 2007), i.e. to bats from an area more distant from Ethiopia than is the Kenyan coast of the Indian Ocean, the type locality of T. afer. The type series of T. p. majusculus did not show any remarkable difference from other examined African samples and in statistic analyses it was placed among other bats from Africa. It suggests that all African populations belong to one form and therefore, we consider the name T. afer appropriate for African Triaenops populations including those formerly assigned as separate subspecies majusculus, which name represent a junior synonym of afer. Separate position of afer is in accordance with previously mentioned opinions of various authors, however, we proved for these populations a separate species status based mainly on genetic traits. Such a result is identical with the original and several older taxonomic opinions (Peters 1877; Dobson 1878; Trouessart 1904; Miller 1907; Allen 1939; Tate 1941; Dorst 1948; Aellen 1957; Harrison 1961, 1963; etc.). So, Triaenops afer Peters, 1877 is the only proved member of the genus occurring in continental Africa. Two names originated from the Middle East, persicus and macdonaldi, (Hill 1982; Simmons 2005) as well as two names suggested to originate in Middle East and/or Somalia, rufus and humbloti, (Goodman & Ranivo, submitted) seem to be all appropriate for the species above designed as the ‘second unit’ within Triaenops, composed by the Middle Eastern morphotypes B and C. Since the above analyses proved a close proximity of this species and the pair of syntypes of T. persicus from Iran, there is well­founded to consider this name for this larger Middle Eastern species. The types of rufus and humbloti were shown by our morphologic analysis as to be most close to the morphotype C originated in western Yemen, and therefore, we suggest the origin of these types in Aden (south­western Yemen) as already proposed by Goodman & Ranivo (submitted). The

94 origin in Somalia is less probable since in continental Africa such a morphotype (nor any close one) was not find, although its presence there is not possible to exclude completely due to geographical proximity of these areas. Anyway, the synonymy of the names rufus and humbloti with persicus as already suggested by Goodman & Ranivo (submitted) seems to be proved sufficiently in our analysis. The name macdonaldi was proposed by Harrison (1955) for the populations of south­ eastern Arabia, from the oasis of Buraimi on the present border of Oman and U. A. E. as a form of similar size as T. persicus from Iran (LAt 47.1–51.6 mm; LCc 16.2–17.2 mm; CM3 6.3–6.6 mm [Harrison 1955: 903]; comp. Table 2, Middle Eastern morphotype B), but of a slightly paler pelage colour. Since the pelage coloration, both its tinge and intensity, was shown to be extremely variable within Triaenops, we regard this name to be a junior synonym of the name T. persicus. This opinion is also more convenient from the biogeographical point of view; the Irani and Pakistani populations seem to be only small projections of Arabian centre of the range of this form across the Strait of Hormuz. Validity of this subspecies was doubted already by DeBlase (1980), who examined and compared both type series (of persicus and macdonaldi) in details, and was accepted by the subsequent authors (Hill 1982; Koopman 1994; Simmons 2005). Anyway, if the Omani populations really differ from the Irani ones as tentatively suggested by Harrison & Bates (1991), this difference has never been expected on the species level and moreover, the name macdonaldi – although we did not have an opportunity to examine its type series – is absolutely not applicable for the smaller Yemeni species, referable here to the morphotype A. This form, characterised by very small body size, cannot be attributed to the name macdonaldi as its type series in size fully conform with Irani persicus (Harrison 1955; DeBlase 1980; Hill 1980) as well as with our Yemeni morphotype B. Therefore, we propose for the newly recognised species of the morphotype A from south­eastern Yemen a new name, see Taxonomic part. The area of eastern Yemen belongs to the most arid parts of the range inhabited by the genus Triaenops, from the ecological point of view it is rather startling a sympatric presence of two species there; in other more fertile parts of genus range (Triaenops s.str.), only monospecific populations are known. From the above comparison remains clear that the western Yemeni populations of T. persicus formerly assigned to the African form afer (for the first time suggested by

95 Harrison 1964) is a part of the Middle Eastern form T. persicus s.str. (in the sense of the present review), although their representatives are larger than those of the typical T. persicus (of the morphotype B). However, this solely size difference could be explained by cline shift of the size characters along the southern Arabian coast. Although the geographic distance between the collection areas comprises nearly 1000 km and the size variation ranges of both forms overlaps only minutely, the gene flow among them seems to be present as in both areas were found identical haplotypes in 1140 bp of a mitochondrial genome. The topologies obtained by all methods exhibited rather low bootstrap supports for mutual positions of the six distinct clades of Triaenops, obtained from the analysis of 731 bp portion of cytochrome b. Especially, position of T. ‘rufus’ appeared questionable after comparison of MP and NJ trees, suggesting sister position of T. ‘rufus’ to the Middle Eastern clades, to ML tree, which did not corroborate monophyly of T. ‘rufus’ and placed T. ‘rufus’ haplotypes at the base of the Afro­Arabian lineage. According to Russell et al. (2007, 2008), T. ‘rufus’ is a sister taxon to African haplotypes group (= T. afer, see below), and represents a result of second colonisation event to Madagascar from Africa, following the first colonisation resulting in the pair of other currently recognised Madagascan species T. auritus and T. furculus (here separated to a new genus, see below). Testing of alternative hypotheses assuming either sister relationship of T. ‘rufus’ and the Middle Eastern forms or basal position of T. ‘rufus’ in the Afro­Arabian lineage, however, suggested that T. ‘rufus’ is more related to the Middle Eastern populations. Thus, colonisation of Madagascar may have occurred via a northern way from north­eastern Africa or Arabian Peninsula, rather than from direct colonisation from neighbouring African regions. Nevertheless, independent additional evidence from other molecular markers should be included to fully resolve true geographic origin of Madagascan T. ‘rufus’, as should more samples from other regions be included into the dataset analysed.

Taxonomic part

Triaenops parvus sp. nov.

HOLOTYPE. Adult male (NMP 92270 [S+A]), Hawf, Yemen, 15 October 2005, leg. P. Benda.

96 PARATYPES (7). Four adult males, three adult females (NMP 92264, 92265, 92267, 92269 [S+A], 92268 [A], BCSU field Nos. pb3009, pb3010 [S+A]), Hawf, Yemen, 14 October 2005, leg. P. Benda.

TYPE LOCALITY. Republic of Yemen, Province of Al Mahra, oasis of Hawf (easternmost edge of the country), 16° 39’ N, 53° 03’ E, 410 m a. s. l.

DESCRIPTION AND DIAGNOSIS. Least representative of the genus Triaenops Dobson, 1871 s.str. (= T. persicus, T. ‘rufus’, T. afer, and T. parvus sp. nov.). It is in most respects very similar to other species of the genus Triaenops s.str., including the structure and relative size of noseleaf (Figs. 6 and 11). In body and skull size, T. parvus sp. nov. clearly differs from Triaenops persicus (Fig. 8) and T. afer, but overlapping dimensionally with T. ‘rufus’ (Fig. 1). Forearm length 44.7–48.1 mm, occipitocanine length of skull 16.3–17.4 mm, length of the upper tooth­row 5.8–6.2 mm. T. parvus sp. nov. shares with T. ‘rufus’ the shape of rostrum, it is relatively short and narrow, and in this character differs from T. afer (with broad and short rostrum) and T. persicus (with broad and long rostrum). T. parvus sp. nov. has relatively high braincase (character shared with T. afer and T. persicus and differring from T. ‘rufus’). T. parvus sp. nov. has relatively large tympanic bullae (character shared with T. ‘rufus’), their large horizontal diameters represent 15–17% of the occipitocanine length of skull, although absolutely they are comparatively small (2.6–2.9 mm). From T. persicus s.str. living in sympatry with T. parvus sp. nov., the latter form differs by less dorsally prominent posterior nasal swellings and much less pronounced sagittal crest in skull (Fig. 10). T. parvus sp. nov. is in size similar to members of the genus Paratriaenops gen. nov., from which it differs by larger wings (forearms relatively longer) and completely in the shape of rostrum and structure of noseleaf (see Fig. 11 and the description of Paratriaenops gen. nov. below). Baculum of T. parvus sp. nov. is a long gracile bone roughly 1.5 mm long, with broad basal epiphysis and bifurcated distal epiphysis; it has a relatively very narrow diaphysis (ca. 8% of the baculum length) with relatively short arms at its distal epiphysis (length of arm represent ca. 17–20% of the baculum length) and relatively narrow proximal epiphysis (width of the basis 23 and 31% of the baculum length). In two examined bones, there were

97 distinct proximal projections in their bases, possibly representing an ossified distal part of the erectile penial body, however, this character is hardly to state as typical for T. parvus sp. nov. without examination of a sufficiently numerous series of bacula. The coloration of dorsal pelage is in T. parvus sp. nov. beige or pale brownish­grey (without reddish or rusty tinges), ventral pelage is very pale beige to pale greyish­brown. Noseleaf is unpigmented to pale greyish­brown. Wing membranes are dark brown. Genetics. Within the genus Triaenops s.str. (except for T. ‘rufus’, i.e. 11 haplotypes from T. parvus sp. nov., T. persicus and T. afer; see Appendix 3), T. parvus sp. nov. showed unique base positions within the complete mitochondrial gene for cytochrome b (1140 bp) at 39 sites: 231, 405, 408, 423, 462, 585, 609, 685, 711, 753, 759, 813, 816, 960 (A→G), 42, 180, 285, 312, 569, 644, 789, 924, 969, 993 (C→T), 18, 129, 138, 640, 898, 907, 1105, 1131 (G→A), 351, 456, 498, 858, 979 (T→ C), 696 (C/A → T), and 750 (G/C → A). Triaenops parvus sp. nov. shares identical unique base positions within the complete mitochondrial gene for cytochrome b (1140 bp) with T. persicus Dobson, 1871 at 41 sites (Appendix 3): 168, 171, 352, 486, 552, 576, 697, 720, 864, 873, 888, 915, 996, 1023 (A), 5, 54, 135, 207, 354, 396, 432, 459, 558, 561, 636, 708, 717, 732, 906, 939, 999 (C), 111, 429, 483, 984 (G), 87, 186, 291, 724, 744, 819 (T); and with T. afer Peters, 1877 at 28 sites (Appendix 3): 93, 117, 234, 297, 450, 861, 897, 1069 (A), 309, 321, 473, 478, 633, 718, 846, 891, 948, 990 (C), 369, 480, 1026 (G), 261, 286, 327, 579, 666, 672, and 840 (T). Within the 731 bp partial sequence of the mitochondrial gene for cytochrome b, Triaenops parvus sp. nov. shares identical unique base positions with T. ‘rufus’ at three sites only (Appendix 4): 138 (A), 231 and 711 (G).

DIMENSIONS OF THE HOLOTYPE. See Table 1.

MITOCHONDRIAL SEQUENCE OF THE HOLOTYPE (complete sequence of the mitochondrial gene for cytochrome b; GenBank Accessite Number EU798754; haplotype ME8 [Appendix 2], 5’ end). atg acc aac ata cga aaa tcc cac cca cta ttc aaa att att aac gac tca ttc gta gac ctc cca gcc cca tcc agc atc tca tct tga tga aac ttt ggc tca cta ctg ggc gta tgc tta gca gta cag atc tta act ggc cta ttc cta gcc ata cac tac aca gca gac aca gct acc gct ttc caa tca gtc acc cat atc tgc cga gac gtt aat tac ggt tgg gta ctg cgc tat ctc cac gcc aac gga gct tcc ata ttc ttc atc tgc cta ttt tta cat gta gga cgt ggc atc tac tat gga tcc tac aca ttt aca gaa aca tga aac att ggc atc atc ctc cta ttc gcg gtg ata gca aca gca ttc atg ggc tat gtc cta cca tgg ggg cag ata tcc ttc tgg ggg gcg acc gtc

98 att act aac tta cta tcc gcc atc ccg tac atc gga aca agc ctg gtg gaa tga gta tga ggc ggc ttc tca gta gac aaa gcc act cta aca cga ttt ttc gcc cta cac ttc cta ctc ccc ttc atc atc gta gcc cta gtt atg gtg cac ctc tta ttc cta cac gaa acg gga tcc aac aac cca aca gga atc ccc tca aat gtg gac ata atc ccg ttc cac cct tat tat aca atc aaa gac gtc ctc ggc ctt atc cta ata atc atg gct ctc cta tct tta gta ctc ttt tca cca gat tta cta ggg gac ccg gat aac tac acc cca gcc aac cca cta aat aca ccc cca cat att aaa cca gag tgg tat ttc ctc ttt gcc tac gcc att cta cgc tca att ccc aac aaa cta gga ggc gta gta gcc tta gta tta tcc atc cta atc ctt gcc atc atc cca cta cta cat aca tca aaa caa cgc agc atg acc ttc cga cca ctg agc cag tgt cta ttt tga ctc ctg gta gcc gat cta gcc aca ctc acc tga atc gga gga caa ccg gtt gaa cac cca ttt atc atc atc ggc caa ata gcc tca att atc tac ttc tta atc atc cta gta ctc ctc cca cta aca agt atc gca gaa aac cgc cta tta aaa tga aga.

DERIVATIO NOMINIS. The name parvus (= small in Latin) reflects the extraordiary small size of the species representatives, the main character which distinguishes the new species from all other species within Triaenops sensu lato.

DISTRIBUTION. Triaenops parvus sp. nov. is known from three sites in the easternmost part of Yemen, all in the province of Al Mahra; Hawf, Damqawt, and 25 km WSW of Sayhut, distant for ca. 270 km from each other at maximum.

Paratriaenops gen. nov.

TYPE SPECIES. Triaenops furcula Trouessart, 1906: Bulletin du Museum d’Histoire Naturelle, Paris, 7, 446.

DESCRIPTION. Medium­sized bats, forearm length 42–51 mm, greatest length of skull 15.9– 18.8 mm, condylocanine length of skull 14.1–16.2 mm (Ranivo & Goodman 2006). Ears large, internal border of ear is not notched. Noseleaf (Fig. 11b). Noseleaf relatively simple and large, bearing three long trident­like posterior projections and a medial process. Anterior leaf lacks lateral supplementary leaflets; the internarial projection (leaflet) is narrow, forked in mesial direction, its lateral margins are parallel and its mesial projections are broad and nearly pointed. Lateral margins of the posterior leaf are parallel or slightly convex; the posterior leaf composed of

99 eleven cells, five cells surrounding the caudal margin of the intermediate leaf; their dividing septa are thin, most lateral cells basally without septa. Posterior median cell very large, wider than the base of medial posterior projection and almost as wide as the intermediate leaf, sagitally incompletely divided by a low septum. Medial process of the intermediate leaf is small and laterally flattened. The posterior projections are long, almost as long as the anterior leaf; the medial projection wider than the lateral ones, which are slightly shorter. The projections extend across the whole width of caudal margin of the posterior leaf. Lateral margins of the projection bases extend ventrally to form the lateral walls of the adjacent cells. Skull (Ranivo & Goodman 2006: 972, Fig. 3A, B; 973, Fig. 4A, B). Skull is typical with dorsally projecting and posteriorly tapered nasal swellings, their anterior margins are nearly vertical. In the interorbital region, a deep post­nasal concavity is present and in frontal region, a low sagittal crest. In dorsal view, nasal swellings are triangular­shaped with extreme short anterior celullae and extensive posterior celullae, in the mesio­distal direction as twice longer as the anterior ones. Dorsal margin of nasal openings mesially stretches to a level of tips of the second upper premolars (P4). Interorbital constriction is relatively narrow (mostly below 12% of the occipitocanine length of skull). Premaxillae are mesio­distally relatively short, shorter than the palate, sphaenoidalia as broad as the interorbital part of frontalia. Zygomata bear high postorbital processes. Bullae tympanicae are mediolaterally narrow. Genetics. Paratriaenops gen. nov. showed unique base positions in 731 bp partial sequence of the mitochondrial gene for cytochrome b at 72 sites (9.8% of the sequence, 29.0% of the variable sites; Appendix 4; haplotypes of the NCBI Accessite Numbers DQ005787, DQ005795, DQ005843, and DQ005849) within the group of close genera Triaenops s.str. (12 haplotypes), Cloeotis (one haplotype) and Paratriaenops gen. nov. (four haplotypes): 330, 402, 630 (A→C), 258, 617 (A→G), 336, 624 (A→T), 63, 183, 201, 555, 694 (C→A), 120, 125, 150, 156, 174, 198, 276, 303, 323, 355, 365, 384, 417, 420, 573, 597, 660, 700 (C→T), 67, 387 (G→A), 331 (G→C), 712 (G → C/T), 39, 345, 441, 534, 669, 670 (T→ C), 522 (T → A/C), 492 (A/C→G), 12, 195, 687 (A/C→ T), 138, 147, 171, 333, 429, 645, 657, 720 (A/G→ C), 66 (A/G → T), 480, 582, 675 (A/G → C/T), 57, 105, 594, 729 (A/T→ C), 48, 264, 357, 501, 579 (C/T→ A), 228 (C/T → G), 399 (A/C/G → T), 234, 297 (A/ G/T→ C), and 87, 141 (C/G/T → A).

100 Paratriaenops gen. nov. shares identical unique base positions with Triaenops Dobson, 1871 at 47 sites (6.4% of the sequence, 18.9% of the variable sites; Appendix 4) of the examined part of cyt b: 27, 213, 294, 324, 328, 375, 381, 466, 471, 472, 474, 475, 507, 525, 580, 612, 705 (A), 6, 69, 75, 190, 244, 246, 252, 280, 318, 342, 358, 453, 465, 468, 477, 537, 540, 541, 549, 564, 688, 693 (C), 127, 232, 476, 643 (G), and 99, 136, 222, 393 (T); and with Cloeotis Thomas, 1901 at 29 sites (4.0% of the sequence, 11.7% of the variable sites; Appendix 4): 55, 114, 124, 132, 219, 237, 300, 348, 364, 483, 574, 615, 690, 699, 714 (A), 177, 192, 204, 315, 369, 438, 585, 592, 642, 710 (C), and 81, 96, 178, 713 (T).

DIFFERENTIAL DIAGNOSIS. Paratriaenops gen. nov. is very similar to Triaenops Dobson, 1871 and Cloeotis Thomas, 1901, from both the genera it differs mainly in the shape and morphology of the noseleaf (Fig. 11); from both genera it differs by lacking of lateral supplementary leaflets; from Triaenops it differs by its narrow internarial projection forked in mesial direction (character shared with Cloeotis, in which is rather diamond­shaped). Paratriaenops gen. nov. has relatively longest trident­like pointed processes on the posterior leaf, being as long as or even longer than the anterior leaf. The medial process of the intermediate leaf is in Paratriaenops gen. nov. smaller than in Triaenops Dobson, 1871. Skull in Paratriaenops gen. nov. has triangular­shaped nasal swellings (when looked dorsally) with extreme short anterior celullae (in mesio­distal direction) and extensive posterior celullae; in Triaenops are nasal swellinds broad and rather rectangular, anterior and posterior celullae are equally long mesio­distally (see Ranivo & Goodman 2006: 973, Fig. 4). In lateral view, the skull of Paratriaenops gen. nov. has a deep post­nasal concavity and dorsally prominent nasal swellings, rather similar to state in the genus Rhinolophus Lacépède, 1799, and completely differing from that in Triaenops Dobson, 1871. Paratriaenops gen. nov. differs from Cloeotis Thomas, 1901 having dorsal vertical processes on zygomata (sharing with Triaenops Dobson, 1871, and also with somer other hipposiderids); Cloeotis has relatively much smaller and more rounded ears.

DERIVATIO NOMINIS. The name refers to close similarity of Paratriaenops gen. nov. with the genus Triaenops Dobson, 1871; Greek prefix para­ means beside, next to. Masculinum.

101 CONTENT. Paratriaenops gen. nov. contains two named species, Triaenops furcula Trouessart, 1906 [= Paratriaenops furculus comb. nov.] and Triaenops aurita Grandidier, 1912 [= Paratriaenops auritus comb. nov.], and a newly recognised species from Seychelles, previously included in the rank of T. furculus (Goodman & Ranivo, in press).

DISTRIBUTION. Western and northern parts of Madagascar and southwestern islands of Seyechelles (Aldabra, Picard, Cosmoledo) (Hayman & Hill 1971; Hill 1982; Russell et al. 2007; Goodman & Ranivo, in press).

AFFILIATION. Although substantially distant for the generic level, Paratriaenops gen. nov. is systematically positioned close to the genus Triaenops Dobson, 1871. According to the above genetic analyses, this pair of genera is a sister group to the most of the remaining content of the family Hipposideridae Lydekker, 1891 (see above). For these closely related genera we here propose a new tribe within that family:

Triaenopini trib. nov.

TYPE GENUS. Triaenops Dobson, 1871: Journal of the Asiatic Society of Bengal, 40, 455.

DESCRIPTION. Hipposiderid bats with a noseleaf bearing four tall pointed projections on its strongly cellularised posterior leaf, three of them forming a trident­like structure on its caudal margin. A strap­like projection extending forward from the internarial region is typical for the anterior leaf (Fig. 11a, b).

CONTENT. Triaenops Dobson, 1871 and Paratriaenops gen. nov. Most probably, Triaenopini trib. nov. also includes geneticly and mainly morphologicly close related genus Cloeotis Thomas, 1901, however, for its inclusion, more robust genetic evidence should be brought that we are able for the moment.

Conclusions

The above presented revision we summarise into the following review of the classification of Triaenops (sensu Simmons 2005):

102 Triaenopini trib. nov. Triaenops Dobson, 1871 Triaenops persicus Dobson, 1871 (SE Middle East from SW Yemen to S Iran and Pakistan) = T. rufus Milne­Edwards, 1881 = T. humbloti Milne­Edwards, 1881 = T. persicus macdonaldi Harrison, 1955 Triaenops afer Peters, 1877 (East Africa from Eritrea to Mozambique, SW Congo, NW Angola) = T. persicus majusculus Aellen et Brosset, 1968 Triaenops parvus sp. nov. (SE Yemen) Triaenops sp. [formerly named T. rufus] (Goodman et Ranivo, submitted) (Madagascar) Paratriaenops gen. nov. Paratriaenops furculus (Trouessart, 1906) comb. nov. (Madagascar) Paratriaenops auritus (Grandidier, 1912) comb. nov. (Madagascar) Paratriaenops sp. (Goodman et Ranivo, in press) (SW Seychelles)

Acknowledgements

We thank Robert Asher and Hendrik Turni (ZMB) and Cécile Callou and Allowen Evin (MNHN) for the accessing of the type material of the nominal Triaenops taxa under their care for examination. We thank Peter J. Taylor (DNSM) for the kind providing of the Cloeotis percivali samples from DNSM and Steven M. Goodman for the sending of his unpublished results. We acknowledge grant supports of the Grant Agency of Academy of Sciences of the Czech Republic (# IAA6093404) and Ministry of Culture of the Czech Republic (# MK00002327201; DE06P04OMG008).

References

Aellen, V. (1957) Les chiroptères africains du Musée Zoologique de Strasbourg. Revue Suisse de Zoologie, 64, 189–214.

103 Aellen, V. & Brosset, A. (1968) Chiroptères du sud du Congo (Brazzaville). Revue Suisse de Zoologie, 75, 435–458. Aggundey, I. R. & Schlitter, D. A. (1984) Annotated checklist of the mammals of Kenya. I. Chiroptera. Annals of Carnegie Museum, 53, 119–161. Al­Jumaily, M. M. (1998) Review of the mammals of the Republic of Yemen. Fauna of Arabia, 17, 477–502. Allen, G. M. (1939) A checklist of African mammals. Bulletin of the Museum of Comparative Zoölogy at Harvard College, 83, 1–763. Atallah, S. I. & Harrison, D. L. (1967) New records of rodents, bats and insectivores from the Arabian Peninsula. Journal of Zoology, London, 153, 311–319. Baker, R. J. & Bradley, R. D. (2006) Speciation in mammals and the genetic species concept. Journal of Mammalogy, 87, 643–662. Benda, P., Hulva, P. & Gaisler, J. (2004a) Systematic status of African populations of Pipistrellus pipistrellus complex (Chiroptera: Vespertilionidae), with a description of a new species from Cyrenaica, Libya. Acta Chiropterologica, 6, 193–217. Benda, P., Kiefer, A., Hanák, V. & Veith, M. (2004b) Systematic status of African populations of long­eared bats, genus Plecotus (Mammalia: Chiroptera). Folia Zoologica, 53 (Monograph 1), 1–47. Bradley, R. D., & Baker, R. J. (2001) A test of the genetic species concept: cytochrome­b sequences and mammals. Journal of Mammalogy, 82, 960–973. Corbet, G. B. (1978) The Mammals of the Palaearctic Region: a taxonomic review. British Museum (Natural History) and Cornell University Press, London and Ithaca, 314 pp. Cotterill, F. P. D. (2001) New distribution records of leaf­nosed bats (Microchiroptera: Hipposideridae) in Zimbabwe. Arnoldia Zimbabwe, 10, 199–210. Crawford­Cabral, J. (1989) A list of Angolan Chiroptera with notes on their distribution. Garcia de Orta, Serie de Zoologia, 13, 7–48. Dalquest, W. W. (1965) Mammals from the Save River, Mozambique, with descriptions of two new bats. Journal of Mammalogy, 46, 254–264. DeBlase, A. F. (1980) The bats of Iran: systematics, distribution, ecology. Fieldiana: Zoology, New Series, 4, xvii+424 pp. Dobson, G. E. (1871) On a new genus and species of Rhinolophidae, with description of a new species of Vesperus, and notes on some other species of insectivorous bats from

104 Persia. Journal of the Asiatic Society of Bengal, 40, 455–461. Dobson, G. E. (1878) Catalogue of the Chiroptera in the collection of the British Museum. British Museum, London, xlii+567 pp, xxx pls. Dorst, J. (1948) Les chiropteres du genre Triaenops Dobson (Hipposidérinés). Mammalia, 12, 15–21. Duff, A. & Lawson, A. (2004) Mammals of the World. A Checklist. A & C Black, London, 312 pp. Ellerman, J. R. & Morrison­Scott, T. C. S. (1951) Checklist of Palaearctic and Indian Mammals 1758 to 1946. Trustees of the British Museum, London, 810 pp. Funaioli, U. & Lanza, B. (1968) On some bats from Somalia. Monitore Zoologico Italiano (n. s.), 2 (Supplement), 199–202. Garbutt, N. (1999) Mammals of Madagascar. Pica Press, The Banks, Mountfield, 320 pp. Goodman, S. M. & Ranivo, J. (in press) A new species of Triaenops (Mammalia, Chiroptera, Hipposideridae) from Aldabra Atoll, Picard Island (Seychelles). Zoosystema. Goodman, S. M. & Ranivo, J. (submitted) The geographical origin of the type specimens of Triaenops humbloti and T. rufus (Chiroptera: Hipposideridae) reputed to be from Madagascar and the description of a replacement species name. Mammalia. Hall, T. A. (1999) BioEdit: a user­friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symposium Series, 41, 95–98. Happold, M. & Happold, D. C. D. (1998) New distributional records for bats and other mammals of Malawi. Nyala, 20, 17–24. Harrison, D. L. (1955) On a collection of mammals from Oman, Arabia, with the description of two new bats. The Annals and Magazine of Natural History, Series 12, 8, 897–910. Harrison, D. L. (1956) A key to the identification of the bats (Chiroptera) of the Arabian peninsula. Proceedings of the Zoological Society of London, 127, 447–452. Harrison, D. L. (1961) A checklist of bats (Chiroptera) of Kenya Colony. Journal of the East Africa Natural History Society, 23, 286–295. Harrison, D. L. (1963) On the occurrence of the leaf­nosed bat Triaenops afer Peters, 1877, in Mozambique. Durban Museum Novitates, 7, 71–72. Harrison, D. L. (1964) The Mammals of Arabia. Volume I. Indroduction, Insectivora –

105 Chiroptera – Primates. Ernest Benn Limited, London, xx+192 pp. Harrison, D. L. & Bates, P. J. J. (1991) The Mammals of Arabia. Second Edition. Harrison Zoological Museum, Sevenoaks, 354 pp. Hayman, R. W. & Hill, J. E. (1971) Part 2. Order Chiroptera. In: Meester, J. & Setzer, H. W. (Eds.), The Mammals of Africa. An Identification Manual. Smithsonian Institution Press, Washington, pp. 1–73. Hill, J. E. (1982) A review of the leaf­nosed bats Rhinonycteris, Cleotis and Triaenops (Chiroptera: Hipposideridae). Bonner Zoologische Beiträge, 33, 165–186. Hoofer, S. R. & Van Den Bussche, R. A. (2003) Molecular phylogenetics of the chiropteran family Vespertilionidae. Acta Chiropterologica, 5 (Supplement), 1–63. Hulva, P. & Horáček, I. (2002) Craseonycteris thonglongyai (Chiroptera: Craseonycteridae) is a rhinolophoid: molecular evidence from cytochrome b. Acta Chiropterologica, 4, 107–120. Kimura, M. (1980) A simple method for estimating evolutionary rates of base substitutions through comparative studies of nucleotide sequences. Journal of Molecular Evolution, 16, 111–120. Kingdon, J. (1974) East African Mammals. An Atlas of Evolution in Africa. Volume II, Part A (Insectivores and Bats). Academic Press, London, xi+341+Al pp. Kock, D. & Felten, H. (1980) Zwei Fledermäuse neu für Pakistan (Mammalia: Chiroptera). Senckenbergiana Biologica, 61, 1–9. Koopman, K. F. (1993) Order Chiroptera. In: Wilson, D. E. & Reeder, D. M. (Eds.), Mammal Species of the World. A Taxonomic and Geographic Reference. Smithsonian Institution Press, Washington and London, pp. 137–241. Koopman, K. F. (1994) Chiroptera: Systematics. In: Niethammer, J., Schliemann, H. & Starck, D. (Eds.), Handbuch der Zoologie. Band VIII. Mammalia. Teilband 60. Walter de Gruyter, Berlin and New York, pp. 1–217. Largen, M. J., Kock, D. & Yalden, D. W. (1974) Catalogue of the mammals of Ethiopia. 1. Chiroptera. Monitore Zoologico Italiano (n. s.), 16, 221–298. Li, G., Liang, B., Wang, Y., Zhao, H., Helgen, K. M., Lin, L., Jones, G. & Zhang, S. (2007) Echolocation calls, diet, and phylogenetic relationships of Stoliczka’s trident bat, Aselliscus stoliczkanus (Hipposideridae). Journal of Mammalogy, 88, 736–744. Nader, I. A. (1990) Checklist of the mammals of Arabia. Fauna of Saudi Arabia, 11, 329–

106 381. Nylander, M. A. A. (2002) MrModeltest. Version 2.2. Evolutionary Biology Centre, Uppsala University Available from: http://www.ebc.uu.se/systzoo/staff/nylander.html. Pearch, M. J., Bates, P. J. J. & Magin, C. (2001) A review of the small mammal fauna of Djibouti and the results of a recent survey. Mammalia, 65, 387–409. Peters, W. (1877) Mittheilung über eine kleine Sammlung von Säugethieren, welche der Reisende Hr. J. M. Hildebrandt aus Mombaca in Ostafrica eingesand hat. Monatsberichte der Königlichen Preußischen Akademie der Wissenschaften zu Berlin, 1876, 912–914. Peterson, R. L., Eger, J. L. & Mitchell, L. (1995) Faune de Madagascar. 84. Chiroptères. Muséum national d’Histoire naturelle, Paris, 208 pp. Posada, D. & Crandall, K. A. (1998) Modeltest: testing the model of DNA substitution. Bioinformatics, 14, 817–818. Ranivo, J. & Goodman, S. M. (2006) Révision taxinomique des Triaenops malgaches (Mammalia, Chiroptera, Hipposideridae). Zoosystema, 28, 963–985. Remy, J. A., Crochet, J.­Y., Sigé, B., Sudre, J., de Bonis, L., Vianey­Liaud, M., Godinot, M., Hartenberger, J.­L., Lange­Badré, B. & Comte, B. (1987) Biochronologie des phosphorites du Quercy: Mise à jour des listes fauniques e nouveaux gisements de mammifères fossiles. Münchener Geowissenschaftliche Abhandlungen, Reihe A, Geologie und Palaeontologie, 10, 169–188. Ronquist, J. & J. J. Huelsenbeck (2003) MrBayes 3: Bayesian phylogenetic inference under mixed models. Bioinformatics, 19, 1572–1574. Ruedi, M. & Mayer, F. (2001) Molecular systematics of bats of the genus Myotis (Vespertilionidae) suggests deterministic ecomorphologic convergences. Molecular Phylogenetics and Evolution, 21, 436–448. Russel, A. L., Ranivo, J., Palkovacs, P., Goodman, S. M. & Yoder, A. D. (2007) Working at the interference of phylogenetic and population genetics: a biogeographical analysis of Triaenops spp. (Chiroptera: Hipposideridae). Molecular Ecology, 16, 839–851. Russell, A. L., Goodman, S. M. & Cox, M. P. (2008) Coalescent analyses support multiple mainland­to­island dispersals in the evolution of Malagasy Triaenops bats (Chiroptera: Hipposideridae). Journal of Biogeography, 35, 995–1003. Sakai, T., Kikkawa, Y., Tsuchiya, K., Harada, M., Kanoe, M., Yoshiyuki, M. &

107 Yonekawa, H. (2003) Molecular phylogeny of Japanese Rhinolophidae based on variations in the complete sequence of the mitochondrial cytochrome b gene. Genes and Genetic Systems, 78, 179–189. Shimodaira, H. & Hasegawa, M. (1999) Multiple comparisons of log­likelihoods with applications to phylogenetic inference. Molecular Biology and Evolution, 16, 1114– 1116. Simmons, N. (2005) Order Chiroptera. In: Wilson, D. E. & Reeder, D. M. (Eds.), Mammals Species of the World. A Taxonomic and Geographic Reference. Third Edition. Volume 1. The John Hopkins University Press, Baltimore, pp. 312–529. Simmons, N. B. & Geisler, J. H. (1998) Phylogenetic relationship of Icaronycteris, Archaeonycteris, Hassianycteris, and Palaeochiropteryx to extant bat lineages, with comment on the evolution of echolocation and foraging strategies in Microchiroptera. Bulletin of American Museum of Natural History, 235, 1–185. Takezaki, N., Rzhetsky, A. & Nei, M. (1995) Phylogenetic test of the molecular clock and linearized trees. Molecular Biology and Evolution, 12, 823–833. Tate, G. H. H. (1941) Results of the Archbold Expeditions. No. 36. Remarks on some Old World leaf­nosed bats. American Museum Novitates, 1140, 1–11. Taylor, P. J. (2005) Order Chiroptera. In: Skinner, J. D. & Chimimba, C. T. (Eds.), The Mammals of the Southern African Subregion. Third Edition. Cambridge University Press, Cambridge, pp. 256–352. Templeton, A. R. (1983) Phylogenetic inference from restriction endonuclease cleavage site maps with particular reference to the humans and apes. Evolution, 37, 221–244. Thomas, O. (1900) On the mammals obtained in South­western Arabia by Messrs. Percival and Dodson. Proceedings of the Zoological Society of London, 1900, 95–104. Thomas, O. (1901) Some new African bats (including one from the Azores) and a new Galago. The Annals and Magazine of Natural History, (7) 8, 27–34. Trouessart, E.­L. (1904) Catalogus Mammalium tam viventum quam fossilium. Quinquennale supplementum. R. Friedländer & Sohn, Berolini [Berlin], iv+929 pp. Trouessart, E.­L. (1906) Description de mammifères nouveaux d’Afrique et de Madagascar. Bulletin du Museum d’Histoire Naturelle, Paris, 7, 443–447. Turni, H. & Kock, D. (2008) Type specimens of bats (Chiroptera: Mammalia) in the collections of the Museum für Naturkunde, Berlin. Zootaxa, 1869, 1–82.

108 Yerbury, J. W. & Thomas, O. (1895) On the mammals of Aden. Proceedings of the Zoological Society of London, 1895, 542–555. Vallo, P., Guillén­Servent, A., Benda, P., Pires, D. B. & Koubek, P. (2008) Variation of mitochondrial DNA reveals high cryptic diversity in Hipposideros caffer complex (Chiroptera: Hipposideridae). Acta Chiropterologica, 10, in press. Wang, H., Liang, B., Feng, J., Sheng, L. & Zhang, S. (2003) Molecular phylogenetic of hipposiderids (Chiroptera: Hipposideridae) and rhinolophids (Chiroptera: Rhinolophidae) in China based on mitochondrial cytochrome b sequences. Folia Zoologica, 52, 259–268.

109 Appendix 1 List of the material examined in morphologic analysis

Triaenops afer Peters, 1877

Central African Republic: 2 m, 1 f (MNHN 1985­1198, 1985­1199, 1985­1366 [S+A]), La Maboké, leg. R. Pujol. – Congo (Brazzaville): 3 m, 2 f (MNHN 1968­412 [S+A], holotype of Triaenops persicus majusculus Aellen et Brosset, 1968; MNHN 1985­1348a, 1985­1348b, 1985­1349, 1985­1497 [S+A]), Grotte de Doumboula, Loudima (Kouilou), 19 June 1964, leg. J. P. Adam; – 2 f (MNHN 1985­1350, 1985­1351 [S+A]), Grotte de Meya­Nzouari (Kouilou), 22 November 1966, leg. J. P. Adam. – Ethiopia: 8 m, 6 f (NMP 92150–92152, 92161, 92163–92167 [S+A], 92153, 92160, 92162, pb2497, pb2521 [A]), Sof Omar Caves, 2 and 3 May 2003, leg. P. Benda & J. Obuch. – Kenya: 1 m (ZMB 5074 [S+A], holotype of Triaenops afer Peters, 1877), Mombaca, leg. J. M. Hildebrandt. – Tanzania: 6 inds. s.i. (MNHN 1911­730/3–5, 8, 10 [S+A]), Tanga, Grotte de Kulumuzi, 1909, coll. M. Alluaud.

Triaenops parvus sp. nov.

Yemen: 1 m (NMP 92272 [S+A]), Damqawt, 16 October 2005, leg. P. Benda; – 5 m, 3 f (NMP 92270 [S+A], holotype of Triaenops parvus sp. nov.; BCSU pb3009, pb3010 [S+A], NMP 92264, 92265, 92267, 92269 [S+A], 92268 [A]), Hawf, 14 and 15 October 2005, leg. P. Benda; – 1 m (NMP 92274 [S+A]), 25 km WSW of Sayhut, 17 October 2005, leg. P. Benda.

Triaenops persicus Dobson, 1871

Iran: 1 m, 1 f (ZMB 4370/1–2 [S+A], syntypes of Triaenops persicus Dobson, 1871), Shiraz. – Yemen: 1 m, 1 f (NMP 92271, 92273 [S+A]), Damqawt, 16 October 2005, leg. P. Benda; – 9 m, 5 f (NMP 92253, 92254, 92256–92262, 92266 [S+A], 92255, 92263 [A], BCSU pb3037, pb3038 [S+A]), Hawf, 12, 14 and 15 October 2005, leg. P. Benda; – 2 m, 1 f (NMP 92275, 92276 [S+A], BCSU pb3123 [S+A]), Jebel Bura, W of Riqab, 30 October 2005, leg. P. Benda; – 1 f (NMP 92277 [S+A]), Wadi Tuban, Kadamat al’Abali, 24 October 2007, leg. P. Benda & A. Reiter; – 1 m (NMP 92279 [S+A]), Wadi Zabid, ca. 10 km SE of Al Mawkir, 30 October 2007, leg. P. Benda & A. Reiter; – 1 f (NMP 92278 [A]), Wadi Zabid, ca. 15 km SE of Al Mawkir, 29 October 2007, leg. P. Benda & A. Reiter. – Yemen/Somalia (?): 4 inds. s.i. (MNHN 1997­1854 [S+A], holotype of Triaenops rufus Milne­Edwards, 1881; MNHN 1997­1856, 1997­1857 [S+A], 1997­1857 [A]), Madagascar [incorrect locality], 1880, leg. L. Humblot; – 3 m, 3 f, 2 inds. s.i. (MNHN 1962­2659 [S], holotype of Triaenops humbloti Milne­Edwards, 1881; MNHN 1985­836–1985­842 [A]), Madagascar, cote est [incorrect locality], 1880, leg. L. Humblot.

Triaenops ‘rufus’ [= Triaenops sp. nov. by Goddman & Ranivo, submitted]

Madagascar: 1 m, 1 f (MNHN 1947­861, 1947­862 [S+B]), Lac Tsimanompetsoa, 20 February 1930, leg. Mission F. A. A.; – 2 inds. s.i. (MNHN 1996­352, 1996­353 [S+B]), Tsaratanana, 16° 46’ N, 47° 40’ E, November 1966; – 2 m, 1 f (MNHN

110 1985­487–1985­489 [S+A]), Madagascar, February 1959, leg. A. Robinson; – 1 ind. s.i. (MNHN 1947­312 [S+A]), Madagascar, October 1938, leg. R. Decasy; – 1 m, 2 f (MNHN 1985­480–1985­482 [S+A]), Madagascar, September 1952, leg. R. Paulian.

Paratriaenops furculus (Trouessart, 1906) comb. nov.

Madagascar: 9 m, 4 f (MNHN 1912­40 [A], holotype of Triaenops furcula Trouessart, 1906; MNHN 1912­40b, 1912­40c [S+A], 1997­1859, 1997­1864–1997­1866 [S+A], 1997­1860–1997­1863, 1997­1867 [A]), Grotte de Sarondrana, 19 May 1898, leg. G. Grandidier.

111 Appendix 2 List of the material used in the genetic analysis

112 Appendix 3 Polymorphic sites identified in the complete cyt b (1140 bp) sequenced in Triaenops Dobson, 1871 s.str.

113 Appendix 4 Polymorphic sites identified in the partial cyt b (731 bp) sequenced in Triaenopini trib. nov., including Cloeotis Thomas, 1901

114 115 Fig. 1. Bivariate plot of compared Triaenops samples: occipitocanine length (LOc) against rostral length of the upper tooth­row (CM3).

Fig. 2. Bivariate plot of compared Triaenops samples: relative width of rostrum (rostral width across upper canines vs. occipitocanine length – CC/LOc) against relative length of rostrum (length of the upper tooth­row vs. occipitocanine length – CM3/LOc).

116 Fig. 3. Bivariate plot of compared Triaenops samples: results of discriminant analysis of nine skull dimensions of the whole compared set of specimens (see text for details).

Fig. 4. Bivariate plot of compared Triaenops samples: results of discriminant analysis of all skull dimensions of the whole compared set with an exception of Triaenops furculus (see text for details).

117 Fig. 5. Baculum preparations of the Triaenops morphotypes from northern part of distribution range (see text for details). Explanations: a – Sf Omar Caves, Ethiopia, NMP 92164; b – Sof Omar Caves, Ethiopia, NMP 92166; c – Wadi Zabid, W Yemen [morphotype C], NMP 92279; d – Jebel Bura, W Yemen [morphotype C], NMP 92275; e – Hawf, SE Yemen [morphotype B], NMP 92262; f – Damqawt, SE Yemen [morphotype B], NMP 92271; g – Hawf, SE Yemen [morphotype A], NMP 92264; f – Sayhut, SE Yemen [morphotype A], NMP 92274. Scale bar = 1 mm.

Fig. 6. Faces of two Triaenops morphotypes from Hawf, eastern Yemen: left = morphotype A [= Triaenops parvus sp. nov.], right = morphotype B [= Triaenops persicus s.str.].

118 Fig. 7. One of the maximum parsimonial trees showing phylogenetic relationships within the genus Triaenops Dobson, 1871. Nodal support from 1000 bootstrap pseudoreplicates under MP and NJ methods, respectively, is indicated above and below branches. Labelling of Triaenops haplotypes follows Appendix 2, in brackets the morphotype designation used throughout Results.

119 Fig. 8. Bayesian consensus tree showing phylogenetic relationship of the genus Triaenops to other Hipposideridae and to the sister family Rhinolophidae. Nodal support expressed as Bayesian posterior probabilities is indicated above branches, bootstrap values from 1000 and 300 pseudoreplicates under weighted maximum parsimony and maximum likelihood methods, respectively, are indicated below branches. Labelling of Triaenops haplotypes follows Appendix 2, in brackets the morphotype designation used throughout Results.

120

Fig. 9. Clock­like ML tree constructed under constraints reflecting current assumptions on phylogeny of hipposiderid bats. The tree is calibrated according to basal split of Rhinolophidae/Hipposideridae, set to approximately 40 MA. Labelling of Triaenops haplotypes follows Appendix 2, in brackets the morphotype designation used throughout Results.

121 Fig. 10. Skulls of two Triaenops morphotypes from Hawf, south­eastern Yemen: above = morphotype A, female, NMP 92267 [= Triaenops parvus sp. nov.]; below = morphotype B, male, NMP 92254 [= Triaenops persicus s.str.]. Scale bar = 5 mm.

Fig. 11. Noseleafs of three close related genera of trident bats (after Hill 1982); a – Triaenops Dobson, 1871; b – Paratriaenops gen. nov.; c – Cloeotis Thomas, 1901. Scale bars = 2 mm.

122 Table 1. Review of the published opinions on the taxonomic content of the genus Triaenops Dobson, 1871. In parentheses are subspecies of the preceding species, in brackets are taxa separated into genus other than Triaenops. Question mark denotes taxonomic position not expressed properly by the respective author.

author species (subspecies) Dorst (1948) furculus ‘rufus’ persi afer humbloti cus Aellen & Brosset furculus ‘rufus’ persicus (persicus, macdonaldi, afer, (1968) majusculus) Hayman & Hill (1971) furculus ‘rufus’ persicus humbloti Hill (1982) furculus ?‘rufus’ persicus (persicus, afer, majusculus, ?‘rufus’) Koopman (1994) furculus persicus (persicus, afer, majusculus, ‘rufus’) Ranivo & Goodman furculus auritus ‘rufus’ persi (2006) cus present revision [furculus] [auritus] ‘rufus’ persi afer parvus sp. nov. cus

123 Table 2. Body and skull dimensions (in millimetres) of the examined samples. External dimensions other than forearm length were taken only from Middle Eastern samples. See Abbreviations for explanation of dimension abbreviations.

124 125 Table 3. Forearm and skull dimensions (in millimetres) of the examined holotype (syntype in T. persicus) specimens. The holotype of T. furcula represents alcohlic specimen with skull not extracted – for cranial measurements of the paratype series of T. furcula see Table 1. See Abbreviations for explanation of dimension abbreviations. * two alcoholic specimens are associated with the holotype skull (one of them should be a paratype, see also Goodman & Ranivo, submitted).

Table 4. Dimensions (in millimetres) of examined baculum preparations (see text for details and Fig. 5). See Abbreviations for explanation of dimension abbreviations.

126 Table 5. Percentual genetic distances among lineages of Triaenops Dobson, 1871 and Cloeotis Thomas, 1901 computed under Kimura’s two­parameter model of evolution (K2P; Kimura 1980) based on complete sequences (1140 bp) of cyt b (for the naming of lineages see text).

Table 6. Percentual genetic distances among morphotypes of T. persicus and other Triaenops species computed under Kimura’s two­parameter model of evolution (K2P; Kimura 1980) based on partial sequences (731 bp) of cyt b (for the naming of lineages see text).

127 Curriculum vitae

Personal data born 16 April 1979 in Skalica, Slovakia

Education 2003– Ph.D. student of Zoology, Faculty of Science, Masaryk University, Brno; dissertation research carried out at Institute of Vertebrate Biology, v. v. i., Academy of Sciences of the Czech Republic

1998–2003 Faculty of Natural Sciences, Comenius University, Bratislava, SR  Master degree (Mgr.) in Biology, specialization Ecology, diploma thesis: Habitat preference of the harvest mouse (Micromys minutus) in conditions of the National Reserve Šúr [in Slovak], May 2003  Bachelor degree (Bc.) in Biology, specialization Ecology, July 2001

Employment Feb–May08 assistant, Faculty of Sport Studies, MU Jan08– research assistant, Department of Mammal Ecology, IVB AS CR Jan07– aquarium technician, Department of Fish Ecology, IVB AS CR May–Dec06 research assistant, Department of Mammal Ecology, IVB AS CR

Study and working stays Apr05 volunteer stay in wild cat monitoring project in Ilha Grande National Park, Paraná, Brazil, partially supervised by the Institute of Ecological Research, São Paulo, Brazil Feb–Apr05 volunteer stay in research project on carnivore ecology in Emas National Park, Jaguar Conservation Fund, Mineiros, Goiás, Brazil Aug–Sep04 volunteer stay in research project on Eurasian lynx ecology, Mammal Research Institute PAS, Bialowieza, Poland Aug–Nov03 internship stay in research project on white­tailed deer ecology, Smithsonian Institution, Conservation & Research Center, Front Royal, Virginia, USA Oct01–Jan02 semester study stay at the University Vienna, Austria

Schoolings and training courses  Course on Scientific Work, AS CR, Brno; October 2005  Summer School in Biodiversity and Ecology, Mammal Research Institute, Polish Academy of Sciences, Bialowieza, Poland; May 2004

Stipends and awards  travel stipend award by the Hlavka Foundation, Prague, CR, for the stay in Brazil, 2005  scholarship award by the intergovernmental program “Action Austria­Slovakia” for the study stay in Vienna, Austria, 2001

Professional activities member of the Slovak Zoological Society

128 Impacted publications Vallo, P., A. Guillén­Servent, P. Benda, D. B. Pires, P. Koubek (in press). Variation of mitochondrial DNA in the Hipposideros caffer complex (Chiroptera: Hipposideridae) and its taxonomic implications. Modrý D., K. J. Petrželková, K. Pomajbíková, T. Tokiwa, J. Křížek, S. Imai, P. Vallo, I. Profousová, J. Šlapeta (in press). The occurrence and ape­to­ape transmission of the entodiniomorphid ciliate Troglodytella abrassarti in captive gorillas. Journal of Eukaryotic Microbiology.

Non­impacted publications Vallo, P. (in press). Proceedings of the 4th International Summer School on Computational Biology 2008. Benda, P., C. Dietz, M. Andreas, J. Hotový, R. K. Lučan, A. Maltby, K. Meakin, J. Truscott, P. Vallo 2008. Bats (Mammalia: Chiroptera) of the Eastern Mediterranean and Middle East. Part 6. Bats of Sinai (Egypt) with some taxonomic, ecological and echolocation data on that fauna. Acta Societatis Zoologicae Bohemicae 72:1–103.

Abstracts from international conferences Vallo, P., P. Benda, A. Guillén­Servent, J. Červený, D. B. Pires, P. Koubek 2008. Mitochondrial phylogeny of Hipposideros caffer complex and implications for taxonomy of its cryptic forms. In Hutson, A. M. and P. H. C. Lina (eds) Book of abstracts from XIth European Bat Research Symposium, Cluj­Napoca, Romania Benda, P., M. Andreas, C. Dietz, R. K. Lučan, P. Vallo 2008. On the new records of Barbastella leucomelas from Sinai, Egypt. In Hutson, A. M. and P. H. C. Lina (eds) Book of abstracts from XIth European Bat Research Symposium, Cluj­Napoca, Romania Vallo, P., N. Martínková, P. Koubek 2007. Genetic variability of Moravian polecats: a pilot study using mitochondrial DNA. In Book of abstracts from the 25th International Mustelid Colloqium, Třeboň, Czech Republic Barančeková M., P. Vallo, J. Prokešová, P. Koubek 2007. Genetic roots of sika deer in the Czech Republic inferred from mitochondrial gene for cytochrome b. In Book of abstracts from the 1st International Conference on genus Cervus, Fiera di Primiero, Italy Prokešová J., P. Vallo, M. Barančeková, P. Koubek 2007. Genetic variability of red and sika deer populations in the Czech Republic based on mitochondrial DNA. In Book of abstracts from the 1st International Conference on genus Cervus, Fiera di Primiero, Italy Vallo, P., P. Benda, J. Červený, N. Martínková, P. Koubek 2007. Cryptic diversity in Hipposideros caffer group inferred from cytochrome b gene. In Espinosa et al. (eds) Book of abstracts from the XVI. International Bat Research Conference, Merida, Mexico Vallo, P., P. Benda, A. Reiter 2007. Molecular phylogeny amends systematics of the Rhinolophus ferrumequinum/clivosus complex. In Espinosa et al.(eds) Book of abstracts from the XVI. International Bat Research Conference, Merida, Mexico Benda P. & P. Vallo 2007. Taxonomic revision of the genus Triaenops in the Afro­Arabic region. In Espinosa et al. (eds) Book of abstracts from the XVI. International Bat Research Conference, Merida, Mexico

129 Benda, P., P. Vallo, A. Reiter 2007. Notes on systematics of the Rhinolophus ferrumequinum/clivosus group (Chiroptera) in Africa. In Book of abstracts from the 10th African Small Mammal Symposium, Abomey­Calavi, Benin Benda P. & P. Vallo 2007. Taxonomic revision of the genus Triaenops (Chiroptera) in the Afro­Arabic region. In Book of abstracts from the 10th African Small Mammal Symposium, Abomey­Calavi, Benin

Abstracts from national conferences Vallo, P., P. Benda, P. Koubek 2008. Molecular phylogeny of roundleaf bats (Hipposideridae) [in Czech]. In Book of abstracts from the conference „Zoologické dny ‘08“, České Budějovice Vallo, P., P. Benda, A. Reiter 2007. Taxonomic relationships of the horseshoe bats of the Rhinolophus ferrumequinum/clivosus complex: shall we solve the riddle on the DNA level? [in Czech]. In Bryja J., J. Zukal, and Z. Řehák (eds) Book of abstracts from the conference „ Zoologické dny ‘07“, Brno Vallo, P., P. Benda, J. Červený, N. Martínková, P. Koubek 2007. Cryptic diversity of roundleaf bats of the Hipposideros caffer complex [in Czech]. In Bryja J., J. Zukal, and Z. Řehák (eds) Book of abstracts from the conference „Zoologické dny ‘07“, Brno Vallo, P. 2006. Senegalese roundleaf bats (Hipposideros spp.) of the caffer­ruber group in view of molecular systematics [in Slovak]. In Stloukal E. (ed.) Book of abstracts from the conference „12. Feriancove dni 2006“, Bratislava Vallo, P. 2006. Intra­ and interspecific variability of roundleaf bats (Chiroptera: Hipposideridae) of south­eastern Senegal based on mitochondrial gene for cytochrome b [in Slovak]. In Halgoš J. (ed.) Book of peer­reviewed contributions from the Student Scientific Conference 2006, Bratislava Vallo, P. 2006. How to trap a harvest mouse? [in Slovak]. In Bryja J., and J. Zukal (eds) Book of abstracts from the conference „Zoologické dny ‘06“, Brno Vallo, P. 2003. Habitat preference of the harvest mouse (Micromys minutus) in the National Reserve Šúr [in Slovak]. Student Scientific Conference 2003, Bratislava Vallo, P. 2002. Preference of environment by the harvest mouse (Micromys minutus) in the NR Šúr (preliminary results) [in Slovak]. In Kováč V., and D. Némethová (eds) Book of abstracts from the Congress of Slovak Zoologists ‘02, Smolenice Sedilek, I. & P. Vallo 2001. Spatial activity of the harvest mouse (Micromys minutus) in conditions of the NR Šúr [in Slovak]. In Kautman J. (ed.) Book of abstracts from the zoological conference „Feriancove dni 2001“, Bratislava

Other presentations Vallo, P. 2006. Phylogenetic relationships of Senegalese roundleaf bats (Chiroptera, Hipposideridae): first results based on sequencies of cytB gene [in Slovak]. Seminar of the Department of population biology, IVB AS CR, Studenec Vallo, P. 2006. Genetic variability of senegalese bats of the families Hipposideridae and Rhinolophidae [in Slovak]. Zoological seminar, Faculty of Science, MU, Brno Vallo, P. 2006. Ecology of the harvest mouse [in Slovak]. Vertebratological seminar. Faculty of Science, MU Vallo, P. 2006. Use of camera­traps in research of felids [in Slovak]. Vertebratological seminar. Faculty of Science, MU

130