Pavan 2016 - Filogenia Taxonomia Integrativa Murcielagos

Molecular Phylogenetics and Evolution 103 (2016) 184–198
Contents lists available at ScienceDirect
Molecular Phylogenetics and Evolution

journal homepage: www.elsevier.com/locate/ympev
Integrating multiple evidences in taxonomy: species diversity and

phylogeny of mustached bats (Mormoopidae: Pteronotus)
Ana Carolina Pavan ⇑,1, Gabriel Marroig
Departamento de Genética e Biologia Evolutiva, IB/Universidade de São Paulo, 05508090, São Paulo, SP, Brazil
a r t i c l e i n f o a b s t r a c t
Article history: A phylogenetic systematic perspective is instrumental in recovering new species and their evolutionary
Received 8 December 2015 relationships. The advent of new technologies for molecular and morphological data acquisition and anal-
Revised 7 April 2016 ysis, allied to the integration of knowledge from different areas, such as ecology and population genetics,
Accepted 11 July 2016
allows for the emergence of more rigorous, accurate and complete scientific hypothesis on species diver-
Available online 12 July 2016
sity. Mustached bats (genus Pteronotus) are a good model for the application of this integrative approach.
They are a widely distributed and a morphologically homogeneous group, but comprising species with
Keywords:
remarkable differences in their echolocation strategy and feeding behavior. The latest systematic review
Integrative taxonomy
Evolutionary history
suggested six species with 17 subspecies in Pteronotus. Subsequent studies using discrete morphological
Euclidean distances characters supported the same arrangement. However, recent papers reported high levels of genetic
Phylogeny divergence among conspecific taxa followed by bioacoustic and geographic agreement, suggesting an
Species tree underestimated diversity in the genus. To date, no study merging genetic evidences and morphometric
Neotropical diversity variation along the entire geographic range of this group has been attempted. Based on a comprehensive
sampling including representatives of all current taxonomic units, we attempt to delimit species in
Pteronotus through the application of multiple methodologies and hierarchically distinct datasets. The
molecular approach includes six molecular markers from three genetic transmission systems; morpho-
logical investigations used 41 euclidean distances estimated through three-dimensional landmarks col-
lected from 1628 skulls. The phylogenetic analysis reveals a greater diversity than previously reported,
with a high correspondence among the genetic lineages and the currently recognized subspecies in the
genus. Discriminant analysis of variables describing size and shape of cranial bones support the rising
of the genetic groups to the specific status. Based on multiples evidences, we present an updated taxo-
nomic arrangement composed by 16 extant species and a new and more robust phylogenetic hypothesis
for the species included in the genus Pteronotus. Studies developed under such integrative taxonomic
approach are timely for a deeper and wider comprehension of Neotropical diversity, representing the first
step for answering broader questions on evolutionary and ecological aspects of Neotropical life history.
Ó 2016 Elsevier Inc. All rights reserved.
1. Introduction taxonomic issues in systematic studies (Dayrat, 2005; Will et al.,

2005; Schlick-Steiner et al., 2010) has also proven fruitful, allowing
The field of systematics has experiencing a major shift regard- the emergence of more rigorous, accurate and complete scientific
ing how to infer evolutionary patterns as well as recognize and hypothesis. This integrative approach is particularly important
delimit species. The introduction of molecular data in phylogenetic for biological groups whose species boundaries are not obviously
studies provided many conceptual challenges in the area, affecting correlated with morphological changes (Bickford et al., 2007;
not only methods for reconstructing phylogenies, but also the Adams et al., 2009), which are traditionally used in taxonomy as
relationship of systematics with other disciplines, such as phylo- main evidences for delimiting species (Wiens, 2007).
geography (Avise et al., 1987) and coalescent theory (Kingman, Bat diversity in the Neotropical region is steadily increasing due
1982). The integration of concepts from adjacent areas to elucidate to more frequent and powerful systematic investigations that have
been developed recently. According to Simmons (2005) the order
⇑ Corresponding author. Chiroptera harbors 18 families, 186 genera and around 1100 spe-
E-mail address: anapavan@usp.br (A.C. Pavan). cies, while this estimate was 10% smaller five years before
1
Current address: Departamento de Ciências Biológicas, ESALQ/Universidade de (1001 species sensu Hutson et al., 2001). Phylogenetic studies
São Paulo, 13418-900 Piracicaba, SP, Brazil.
http://dx.doi.org/10.1016/j.ympev.2016.07.011
1055-7903/Ó 2016 Elsevier Inc. All rights reserved.
A.C. Pavan, G. Marroig / Molecular Phylogenetics and Evolution 103 (2016) 184–198 185
with bats have been published in large numbers, with well- 2012). A brief synopsis on the natural history of mormoopid spe-
supported phylogenies being currently available for many groups. cies is provided by Simmons and Conway (2001) and Patton and
This increasing dataset relies mostly on molecular information, Gardner (2007).
which suggests that several bat taxa actually include more species The subsequent studies published for Mormoopidae confirmed
than previously believed or sometimes do not even represent independently the monophyly of the family and genera, although
monophyletic groups (Clare, 2011; Pavan et al., 2013; Velazco did not provide an unambiguous and well supported phylogeny
and Patterson, 2008, 2013; Parlos et al., 2014). These more refined for this group. The initial investigations on the genetic diversity
evolutionary hypotheses are offering important opportunities for of Mormoopidae reported divergent lineages within many Pterono-
reconstructing historical patterns of change in several bat lineages. tus species (Lewis-Oritt et al., 2001; Dávalos, 2006), suggesting an
Likewise, they are central to investigations in many related areas, underestimated diversity in this genus. Recently, more compre-
such as biogeography, ecology and evolution. hensive studies were performed, and results suggest a much more
The studies investigating the phylogenetic relationships within complex evolutionary history for the genus Pteronotus than previ-
Mormoopidae reflect such pursuit. The interest on this small ously described (Clare et al., 2011, 2013; Thoisy et al., 2014). Thus,
neotropical family of insectivorous bats increased rapidly in the additional species have been recognized in the subgenus Phyllodia,
last years, including studies with morphological (Simmons and elevated from the subspecies rank by such recent studies (P.
Conway, 2001), molecular (Lewis-Oritt et al., 2001; Van Den paraguanensis, Gutiérrez and Molinari, 2008; P. mesoamericanus,
Bussche and Weyandt, 2003) and combined analyses (Van Den Clare et al., 2013; P. pusillus and P. portoricensis, Thoisy et al., 2014).
Bussche et al., 2002; Dávalos, 2006). The taxonomy of Mormoopi- A deeper knowledge on the diversification pattern of this bat
dae is controversial and through the 20th century it was even clas- group has also being sought because of their echolocation system.
sified as one subfamily of Phyllostomidae (Chilonycterinae). Smith While the majority of mormoopids as well as all the remaining
(1972) raised it to family rank, recognizing within the group two Neotropical bat species are low duty cycle (LDC) echolocators,
genera – Mormoops and Pteronotus, three subgenera – Pteronotus, the extant representatives of subgenus Phyllodia (henceforth men-
Chilonycteris and Phyllodia (all in the genus Pteronotus), and eight tioned as P. parnellii complex) have evolved the high-duty cycle
species (two in Mormoops, six in Pteronotus), most of them with (HDC) echolocation, which is also present in the Old World families
associated geographical variation described as subspecies (Table 1). Rhinolophidae and Hipposideridae. HDC bats use long, narrowband
Two extinct species are also described in the family, Mormoops echolocation calls that improve their ability to detect, lock onto
magna and Pteronotus pristinus (Silva-Taboada, 1974), and a new and track flying prey, probably in areas of high clutter (for a review
genus and species is being described (Morgan and Czaplewski, about this topic see Fenton et al., 2012). The evolution of this
Table 1
Current taxonomic diversity in the family Mormoopidae according to the last systematic review (Smith, 1972) with updates provided by Simmons (2005).
Species Subspecies Type locality Geographic range

Pteronotus parnellii P. p. parnellii Jamaica (unspecified locality) Cuba and Jamaica
P. p. pusillus Arroyo Salado, Dominican Republic Dominican Rep. and Haiti
P. p. gonavensisb En Café, La Gonave Island, Haiti Gonave Island, Haiti
P. p. portoricensis Cueva di Fari, Pueblo Viejo, Puerto Rico Puerto Rico
P. p. mexicanus San Blas, Nayarit, Mexico Sonora and Tamaulipas to Oaxaca and Veracruz,
Mexico
P. p. mesoamericanus Yepocapa, Chimaltenango, Guatemala Southern México and Pacific coast of Central
America to Panama
P. p. fuscus Las Quiguas, Puerto Cabello, Venezuela Caribbean coast of Colombia and Venezuela, north
of Orinoco river
P. p. rubiginosus Caiçara, Mato Grosso, Brazil From Honduras southward along the Caribbean
coast of Central America; Trinidad; Amazonian
lowlands and Guiana Shield, south of Venezuela,
Peru and Central Brazil
Pteronotus paraguanensisa – Pueblo Nuevo, Falcón, Venezuela Paraguana peninsula, Venezuela
Pteronotus personatus P. p. psilotis Tehuantepec, Oaxaca, Mexico Caribbean Coast of Honduras and El Salvador to
Sonora and Tamaulipas, Mexico
P. p. personatus São Vicente, Mato Grosso, Brazil Central Brazil to pacific coast of Costa Rica
Pteronotus davyi P. d. fulvus Las Peñas, Jalisco, Mexico Caribbean Coast of Honduras and El Salvador to
Sonora and Tamaulipas, Mexico
P. d. davyi Island of Trinidad Caribbean Coast of South America and Lesser
Antilles to south of Nicaragua
P. d. incae Suyo, Piura, Peru Peru
Pteronotus gymnonotus – Cuiabá, Mato Grosso, Brazil Central Brazil to Veracruz, Mexico
Pteronotus macleayi P. m. macleayi Guanabacoa, Habana, Cuba Cuba
P. m. griseus Phoenix Park, Westmoreland Parish, Jamaica Jamaica
Pteronotus quadridens P. q. quadridens Baracoa, Oriente Province, Cuba Cuba
P. q. fuliginosus Port au Prince, Haiti Jamaica, Haiti, Dominican Rep., Puerto Rico
Mormoops megalophylla M. m. megalophylla Parrás, Coahuila, Mexico USA to Honduras and El Salvador
M. m. tumidiceps Point Gourde Caves, Trinidad Colombia, Venezuela and Trinidad
M. m. intermedia Hatto, Curaçao, West Indies Dutch Islands
M. m. carteri La Paz, Carchi Province, Ecuador Ecuador
Mormoops blainvillei – Jamaica Greater Antilles
a
Subspecies described by Linares and Ojasti (1974) and recently ranked to the species level (Gutiérrez and Molinari, 2008).
b
Extinct.
186 A.C. Pavan, G. Marroig / Molecular Phylogenetics and Evolution 103 (2016) 184–198
system allowed a distinct foraging strategy by P. parnellii complex 2. Materials and methods
regarding its congeners, and involved many morphological and
physiological adaptations (Fenton et al., 2012; Mancina et al., 2.1. Molecular sampling
2012). Several studies point to a dual function of calls in resource
acquisition and communication in HDC bats, suggesting ecological Tissue samples of 411 individuals of Pteronotus were obtained
selection on frequency might lead to assortative mating and ulti- by loans from Brazilian and foreign research collections
mately reproductive isolation and speciation (Kingston et al., (Table S1). The molecular sampling represented 16 out of 18 cur-
2001; Kingston and Rossiter, 2004). This hypothesis led Clare rent nominal taxa within the genus (Smith, 1972) and a wide geo-
et al. (2013) to propose a similar process shaping diversity in main- graphic range (Fig. 1). Total genomic DNA was extracted from
land populations of P. parnellii complex and assume the existence muscle or liver tissues by a NaCl/SDS/Proteinase K protocol
of four different species in this taxon, based on molecular, morpho- (Bruford et al., 1992). Alternatively, for those samples provided in
metric and bioacoustic evidences. low quantities, DNA was extracted using a Qiagen DNeasy Blood &
Despite the number of studies on mormoopids, previous works Tissue Kit (Qiagen, Inc.) following the manufacturer’s protocol. Six
did not aim to produce an updated phylogenetic hypothesis for the molecular markers were selected for this study, representing three
genus Pteronotus as a whole, whose taxonomic diversity is doubtful different genetic transmission systems: the entire Cyt b gene
since its last systematic review (Smith, 1972). To date, no work (CYTB) and a fragment of the COI gene (COI) in the mitochondrial
merging multilocus genetic data and morphometric variation DNA; a non-coding fragment of the Dby gene in Y chromosome
along the entire geographic range of this group, sampling represen- (DBY); fragments of the autosomal genes STAT5A (non-coding),
tatives of all taxonomic units, was realized. The present study PRKC1 (non-coding) and RAG2 (coding). The amplification of
attempted to delimit species in Pteronotus through the application selected markers was performed via Polimerase Chain Reaction
of multiple methodologies and included a wide exploration of data (PCR) with primers and protocols described previously (Baker
at different hierarchical levels. The use of this integrative approach et al., 2000; Eick et al., 2005; Martins et al., 2007; Borisenko
allowed us to place the evidences presented by last studies into a et al., 2008; Lim et al., 2008; Pavan et al., 2013).
broad and detailed investigation, with direct taxonomic implica- Sequences were assembled and checked for quality using the
tions. For this, we started from the taxonomic arrangement pro- programs Phred, Phrap and Consed (Ewing and Green, 1998;
posed by Smith (1972) as our null hypothesis, since it represents Ewing et al., 1998; Gordon et al., 1998) or, alternatively, using Gen-
the most complete (geographically and taxonomically) study on eious v.7.1 (Biomatters). Coding regions (CYTB, COI and RAG2)
this genus. In addition, the proposal of a robust phylogenetic were aligned by eye, whereas the alignment of intron sequences
hypothesis for the genus Pteronotus is timely since it represents a (STAT5A, PRKC1, DBY) was carried out using the ClustalW tool
very interesting model for addressing broader evolutionary and available in MEGA 6 (Tamura et al., 2013). The mitochondrial data-
ecological questions in Neotropical Region. set was also analyzed for saturation with DAMBE5 (Xia, 2013). For
Fig. 1. Geographic distribution of molecular samples used in the present study. Taxonomic names displayed in the legend follow traditional classification (Smith, 1972;
Simmons, 2005).
autosomal genes analysis, the presence of heterozygous nucleotide

positions was represented in the dataset as ambiguous IUPAC
codes. For the introns PRKC1 and STAT5A, heterozygous individu-
als for the insertion/deletions (INDELS) of bases were also found,
and these samples were represented in the phylogenetic dataset
by the longer sequences of each individual.
2.2. Morphological sampling and cranial landmarks
The morphometric investigation was performed with 1628

adult specimens from 13 museum collections in Brazil and USA:
IEPA (Macapá-Brazil), MZUSP (São Paulo-Brazil), UFMG (Belo
Horizonte-Brazil), INPA (Manaus-Brazil), UFPE (Recife-Brazil),
UFPB (João Pessoa-Brazil), AMNH (New York-USA), TCWC (College
Station-USA), KUM (Lawrence-USA), NMNH (Washington-USA),
FMNH (Chicago-USA), LACM (Los Angeles-USA) and TTU
(Lubbock-USA). Sampling included localities along the geographic
range of all current species (Table S2). We used a Microscribe
MX digitizer (Microscribe, IL) to collect three-dimensional (3D)
coordinates from anatomical points (bones sutures or other dis-
crete cranial features; named henceforth landmarks) in Pteronotus
skulls. Twenty-two landmarks were established in the skull of
Pteronotus species (Fig. 2) and used to estimate 41 Euclidean dis-
tances (Table S4) describing size and shape within the genus. This
methodology for collecting data provides a more complete descrip-
tion on the skull variability than caliper measurements (Reig,
1996). The adoption of 3D coordinates instead of fixed
measurements in the skull may provide a better performance in
morphometric analysis since it yields a larger number of distances
to the morphometric dataset with similar time effort and associ-
ated error. In addition, the use of 3D data allows the storage of
skull coordinates for each specimen, which can be further used
to create new distances that prove to be useful for morphometric
analyses. Furthermore, the distances used here are meant to cap-
ture local developmental/functional processes while at the same
time represent the whole skull (Cheverud, 1982, 1996). Thus, traits
are not redundant and do not capture general growth process as
usual in the case of caliper measures since individual bones are
represented (Cheverud, 1996). All landmarks were collected twice
per specimen. Because asymmetry between the right and left sides
of the skull is not of interest in this study, measurements present in
both sides were treated as their average. Whenever one side of the
Fig. 2. Anatomical position (dorsal, ventral and lateral views) and description of the
skull was damaged, the other was considered as the mean. Normal-
22 landmarks collected in the skulls of Pteronotus.
ity and outlier analyses were also performed to check the quality of
the collected data. All subsequent analyses were carried out using
the average of repeated measurements.
albiventris at GenBank, these species were represented as missing
data for this locus in the five-gene and complete datasets.
2.3. Phylogenetic analyses Nucleotide substitution models that best explained the varia-
tion observed in each dataset were estimated by PartitionFinder
The phylogenetic analyses were initially performed for each v.1.1.0 (Lanfear et al., 2012). These models of molecular evolution
marker separately in order to observe congruence among topolo- allowed the setting of partitioned schemes in the datasets for sub-
gies. Four different datasets were then constructed for evaluating sequent multilocus analysis via Bayesian and Maximum Likelihood
the importance of different factors such as the genetic transmis- methods. Bayesian Inference (BI) was implemented in MrBayes
sion systems and the number of base pairs in the performance of 3.2.2 (Ronquist and Huelsenbeck, 2003) using four Markov chain
analysis: (1) mitochondrial dataset (mtDNA: CYTB and COI); (2) Monte Carlo (MCMC) in two independent runs at 10 million gener-
fast-evolving nuclear dataset (nDNA: introns PRKC1 and STAT5A); ations each for the mtDNA dataset, and 5 million generations for the
(3) five-gene dataset (CYTB, COI, PRKC1, STAT5A and DBY); (4) others. Sampling of chains occurred every 1000 generations and
complete dataset (CYTB, COI, PRKC1, STAT5A, DBY and RAG2). the first 25% of the sampled trees and estimated parameters were
Additional sequences of Pteronotus species from GenBank were discarded as burn-in. Stationarity of runs were checked in Tracer
included in the datasets when available (Table S1). GenBank v.1.6 (Rambaut et al., 2013) by examining the average standard
sequences from Peropteryx kappleri (Emballonuridae), Noctilio deviation of split frequencies (Ronquist et al., 2010). For Maximum
albiventris (Noctilionidae), Mormoops megalophylla (Mormoopidae), Likelihood (ML) analyses, five to eight independent searches with 5
Glossophaga soricina and Trachops cirrhosus (Phyllostomidae) were million generations were performed and compared for each data-
also used as outgroups in phylogenetic analysis. Since there are set in Garli 2.0 (Zwickl, 2006). The tree with the smaller likelihood
no available sequences of DBY gene for M. megalophylla and N. (Ln) value was kept as the best topology and used to plot the result
of 100 bootstrap replicates in each dataset. BI and ML analyses in the genus Pteronotus, even if they were not used in final data
were run on the CIPRES Science Gateway. Maximum parsimony analysis. Complete sequences of CYTB (1140 bp) and COI (651 bp)
(MP) analyses were performed in TNT v.1.1 (Goloboff et al., 2003) were obtained for 328 and 350 specimens, respectively, while
by 1000 replicates of heuristic searches using a combination of mtDNA dataset was represented by 314 individuals. Both mito-
tree-searching algorithms (sectorial searches, drift, ratchet and chondrial markers presented similar levels of genetic variation
tree-fusing mixed trees). Node supports in MP analysis were and non-significant results in the test of substitution saturation.
accessed by resampling via bootstrap and the values were plotted Due to the high number of available tissue samples, we obtained
in the majority consensus tree obtained by MP searches. a genetic variation overview by preliminary analysis with mtDNA
only. The remaining molecular markers were then sequenced for
2.4. Species tree a subset of the available samples, which was representative of
the genus phylogenetic diversity. Introns of PRKC1 (462 bp) and
In addition to phylogenetic analyses, a Bayesian framework for STAT5A (602 bp) were sequenced in 128 specimens and presented
species tree estimation from multilocus data was adopted. This a similar variation to that observed for the 64 males sequenced for
approach considers the incomplete lineage sorting as one of the DBY (510 bp). The nuclear exon RAG2 (804 bp) was sequenced in
major causes of gene tree heterogeneity and gene tree/species tree 27 specimens and, as expected, showed the smaller variation
conflicts (Heled and Drummond, 2010). The complete dataset was within data. The complete information about molecular variation
used for the species tree estimate, adopting independent nucleo- and evolutionary models of each analyzed dataset is available in
tide substitution models for each dataset partition according to Table S3.
the result obtained by Partition Finder. The analysis was performed
in *BEAST (StarBEAST), an extension within the software package 3.2. Phylogenetic inferences
BEAST v.1.8 (Drummond and Rambaut, 2007). The settings
included a run for 50 million generations, with MCMC sampling Phylogenetic analysis of each marker separately showed highly
every 5000 generations. A Yule speciation model was used as tree congruent, recovering similar patterns of relationship among spec-
prior, with an exponential mean of 0.8 (yule.birthRate). Substitu- imens. As a general result, molecular data points to the existence of
tion rate prior for each partition (ucld.mean) followed a lognormal four major clades (subsequently referred to as Clades 1–4) within
distribution model with mean 0.05 substitutions/site/million years the genus, regardless which dataset and searching method was
(subs/site/my) for mtDNA and mean 0.02 for introns, with a stan- adopted: Clade 1 comprises P. davyi and P. gymnonotus; Clade 2
dard deviation of 2. A total of 10,000 trees were generated and includes P. macleayi and P. quadridens; Clade 3 corresponds to P.
summarized to produce the topology with most credibility based personatus; Clade 4 encompasses all samples of P. parnellii. All phy-
on data, which was visualized in the software FigTree v.1.4 logenies confirmed the genus monophyly and presented the same
(Rambaut, 2012). arrangement of individuals within these clades. Analysis of mtDNA
and nDNA datasets alone did not generate topologies with highly
2.5. Morphometric analyses supported basal nodes in Pteronotus, with independent searching
methods exhibiting alternative relationships of clades. More robust
Exploratory analyses were done in the complete dataset in datasets (five-gene and complete), on the other hand, yielded a
order to observe the distribution of values for each variable and highly concordant result, depicting the same phylogenetic relation
deviations from normality. Principal Component Analysis (PCA) for the major clades with high support values in all tree nodes
was subsequently performed to investigate the distribution of each (Fig. 3).
current taxon (sensu Smith, 1972) within the total cranial space
variation of the genus. Then, Multivariate Analysis of Variance
3.2.1. mtDNA dataset
(MANOVA) and Discriminant Function Analysis (DFA) were applied
A total of 17 mitochondrial independent lineages were identi-
in a priori defined categories. These categories were established
fied within Pteronotus (Fig. S1). More than one lineage was found
according to results obtained by molecular data analyses. There-
in three Pteronotus taxa recognized as species by current taxon-
fore, the grouping hypothesis for morphometric analyses was
omy: P. parnellii (eight), P. personatus (four) and P. davyi (two).
based on the phylogeny found, which allowed the identification
The geographic range of these lineages generally corresponds to
of operational taxonomic units (OTUs) congruent with the evolu-
those of subspecies within each species (Smith, 1972). The
tionary history of the genus (evolutionary lineages, sensu de
observed genetic divergence among lineages, however, overcomes
Queiroz, 1998). MANOVA was performed in defined categories to
values currently described as intraspecific variation in Neotropical
identify the existence of sexual variation. In case of statistical sig-
bat species (Bradley and Baker, 2001; Baker and Bradley, 2006).
nificance (p < 0.05), variation between genders was controlled
The estimate of nucleotide distance among Pteronotus lineages
before dataset’s using in subsequent analysis. DFA was applied in
from the same mtDNA major clade varies from 3% up to 15%. In
each Pteronotus species group in order to test if cranial morpho-
some cases comparison between two genetic lineages of the same
metrics could discriminate specimens according to the grouping
current species (e.g. P. parnellii) presents higher levels of diver-
hypothesis based on molecular data. MANOVA analyses were per-
gence than estimates between two recognized species in the genus
formed in R software environment (R Development Core Team,
(e.g. P. davyi P. gymnonotus).
2013) while SYSTAT 11 (SYSTAT Software, Inc. 2004) was used
In Clade 1 (P. davyi and P. gymnonotus), mtDNA points to the
for PCA and DFA analyses.
existence of two paraphyletic and allopatric lineages within
P. davyi (Figs. S1a and S2a): one of them (N = 18) is distributed in
3. Results the Lesser Antilles (Santa Lucia, Dominica and Trinidad) and
coincides with the range of the South American subspecies
3.1. Molecular variation and evolutionary models P. d. davyi; the second (N = 10) occurs from Mexico to Honduras
and agrees with the distribution of the Central American sub-
From the total available tissue samples, 384 were sequenced species P. d. fulvus. Each genetic lineage includes one sample from
for, at least, one fragment of mtDNA (Table S1). This allowed the the type locality of the respective subspecies (Table 1). The mtDNA
assignment of these samples in the evolutionary lineages identified between these two lineages of P. davyi is 7% divergent, the
Fig. 3. Phylogenetic relationships in Pteronotus inferred from (a) five-genes and (b) complete datasets. Topologies of BI approach are presented in both cases and reflect
exactly the same relationships among terminals. Symbols on the nodes represent values of bayesian posterior probabilities (bpp), ML and MP bootstraps according to the
legend. For ML and MP, white = bootstrap frequencies < 50% or a relation not recovered by such method; gray = 50% < bootstrap frequencies < 70%; dark gray = 70%
< bootstrap frequencies < 90%; black = bootstrap frequencies P 90%. For BI, dark gray = 0.8 < bpp < 0.95; black = bpp P 0.95. Species names in black displayed in the center of
the figure correspond to those adopted by traditional taxonomy (Smith, 1972; Simmons, 2005). Terminals at the right tree were labeled with the correspondent subspecies
names of the geographic ranges presented in the left tree.
same distance observed among them and their sister-species few specimens from Guiana Shield (N = 4; obtained through COI or
P. gymnonotus (N = 44). CYTB GenBank sequences only) and the fourth lineage is repre-
In Clade 2, mitochondrial haplotypes were grouped into two sented by a unique sample from Guatemala (ROM 98438). These
highly divergent (13%) clades, which correspond to P. quadridens two last lineages have no available name in the current taxonomy
(N = 25) and P. macleayi (N = 7) samples. A discrete structuring in of P. personatus (Fig. S1b). Given their high mtDNA divergence from
the mitochondrial haplotypes was observed within each of these the two named lineages of P. personatus, they may represent new
clades: in P. macleayi we can find Cuban and Jamaican samples species, but the incomplete sampling prevented us from making
arranged separately for each other; in P. quadridens, specimens a deeper investigation.
from Dominican Republic and Puerto Rico are grouped together, Clade 4 (P. parnellii) is composed by eight different mitochon-
with Cuban and Jamaican haplotypes basal to them (data not drial lineages structured in three main clades (Figs. S1 and 3).
shown). Even so, the overall nucleotide divergence within these These clades are present in all performed analysis, but relation-
Antillean species is smaller than 1%, suggesting they are genetically ships among them are different in MP, BI and ML topologies. The
cohesive groups. overall K2P nucleotide distance between genetic lineages from dif-
In Clade 3 (P. personatus) four divergent lineages (4–9%) were ferent clades ranges from 11.5 to 14.6% while K2P divergences
identified through mitochondrial data (Fig. S1). The first lineage from 3 to 8.9% are observed among lineages within the same clade.
(N = 52) is widely distributed in Brazil and Guiana Shield Mitochondrial lineages found in P. parnellii also seem to present
(Fig. S2b), including specimens from Brazilian localities close to geographic ranges (Fig. S2) overlapping with subspecies currently
the type locality of P. p. personatus in Mato Grosso state (Smith, recognized within the taxon (sensu Smith, 1972), although some
1972), probably corresponding to this subspecies. The second lin- disparities will be further discussed.
eage (N = 8) encompasses specimens from Mexico (sequenced at The first clade (4-A) includes three allopatric lineages dis-
the present study) and Guatemala (available only as mtDNA Gen- tributed in the islands of Jamaica (N = 6), Dominican Republic
Bank sequences), which agrees with the geographic range of the (N = 6) and Puerto Rico (N = 7), which correspond to the distribu-
subspecies P. p. psilotis (Fig. S2b). As further evidence, this lineage tion of the Antillean subspecies P. p. parnellii, P. p. pusillus and P.
includes haplotypes from Oaxaca (Mexico), the type locality of P. p. p. portoricensis, respectively (Smith, 1972). The second clade (4-B)
psilotis (Smith, 1972). The third lineage of P. personatus comprises a is composed by three parapatric lineages spread in Western
Mexico (N = 17), Southern Mexico and Central America (N = 38), suggesting lower evolutionary rates for the introns when com-
and Northern South America including Lesser Antilles (N = 26), pared to the mitochondrial markers. Lacking of reciprocal mono-
showing very similar geographic ranges, in this order, to the sub- phyly among Pteronotus lineages diagnosed by mtDNA is
species P. p. mexicanus, P. p. mesoamericanus and P. p. fuscus. In therefore expected in nuclear data for recent diversification events
addition, each of the lineages in Clade 4-B includes one or more (Maddison, 1997), such as those within Clades 4-B and 4-C. Also,
samples close to the type locality of the correspondent subspecies nDNA dataset contains only 1053 bp, which are less informative
(Table 1). The third clade (4-C) is represented by two South Amer- than the 1791 bp of the mtDNA dataset. Evidences for the separa-
ican sympatric lineages, occurring in Brazil and Guiana Shield. Our tion noticed in mtDNA are present as lineage-specific SNP’s and
dataset contains several localities where these lineages were sam- INDEL’s in nuclear data for these lineages (data not shown).
pled together, i.e., in syntopy (Fig. S2c). One of them (N = 106)
extends further southward in Brazilian Cerrado, including all sam-
3.2.3. Five-gene and complete datasets
ples from Mato Grosso state, the type locality of the subspecies P. p.
Two distinct multilocus analysis, using five markers (62 speci-
rubiginosus (Smith, 1972), while the other (N = 27) encompasses
mens, 3368 base pairs) and the complete dataset (32 specimens,
only Amazonian localities and represent an unnamed taxon, here-
4172 base pairs), were performed and compared as regards their
after mentioned Pteronotus sp1 (corresponding to P. sp3 sensu Clare
robustness in solving the basal relationships within the genus
et al., 2013; Thoisy et al., 2014).
(Fig. 3). Both datasets resulted in the same topology through ML
Some differences between subspecies and genetic lineages
and BI methods, with some nodes displaying better support values
ranges deserve consideration. Smith (1972) originally assigned P.
with the inclusion of the sixth maker (RAG2) in the analysis
parnellii specimens from Trinidad to the South American sub-
(Fig. 3b). The basal position of Clade 4 (P. parnellii samples) in
species P. p. rubiginosus. However, the molecular results undoubt-
the genus and a closer relationship between Clades 1 and 2 were
edly place Trinidad and Saint Vicent and Grenadines samples in
strongly supported. The MP search also found a similar relationship
the same clade of Venezuelan samples (Figs. 3 and S1), which cor-
among lineages, though the support values for basal nodes were
responds to the subspecies P. p. fuscus. Another difference is the
still below 70%.
contact zone postulated by Smith (1972) between P. p. mesoamer-
Regarding intraclade diversification, all lineages previously
icanus and P. p. rubiginosus along Central America. According to the
identified through mtDNA were strongly supported in the five-
author, a zone of intergradation would exist in eastern Honduras,
gene dataset analysis, which sampled two or more specimens for
southward more or less along the continental divide. Therefore,
most groups (Fig. 3a). The more robust analyses suggest the taxon
populations of P. parnellii in Central America southward Honduras
P. davyi (henceforth referred as P. davyi complex) as a mono-
should be assigned to P. p. mesoamericanus when located in the
phyletic group comprising two divergent lineages in Clade 1
Pacific versant or to P. p. rubiginosus if occurring in the Atlantic
(Fig. 3). However, support values for the sister relationship
(Caribbean) versant (see Table 1). According to our results, how-
between these two lineages (P. d. davyi and P. d. fulvus) were not
ever, all specimens from Central America (including a large sam-
conclusive. The close relationship among Antillean species P.
pling across Honduras) belong to the same genetic lineage
quadridens and P. macleayi in Clade 2 remained stable along all
(Fig. S2), which was identified under the name P. p. mesoameri-
phylogenetic inferences. In Clade 3 (P. personatus complex), the
canus, similarly to the findings of Clare et al. (2013). These evi-
Mexican lineage (P. p. psilotis) exhibited a basal position, with the
dences led us to limit the north distribution of P. p. rubiginosus
South American (P. p. personatus) and the Guatemalan specimen
group to Guiana Shield and Southern Venezuela.
being sister groups in all resulted topologies. Within Clade 4 (P.
parnellii complex), ML and BI results agreed with a sister relation-
3.2.2. nDNA dataset
ship between Antillean (Clade 4-A) and Central American (Clade 4-
Introns presented a very congruent result with mtDNA dataset,
B) clades, leaving South American lineages (Clade 4-C) in a basal
although a smaller phylogenetic structure within each of the
position within the P. parnellii complex. Relationships among
Pteronotus major clades is observed (Fig. S3). In Clade 1, both
lineages of Clade 4-B (P. p. mesoamericanus, P. p. mexicanus and
P. davyi and P. gymnonotus are recovered as monophyletic groups.
P. p. fuscus) were resolved and well supported only when the larger
Within P. davyi, one of the mitochondrial lineages (P. d. davyi) is
dataset was analyzed, which may be consequence of the recent
placed as a nested clade within the other lineage (P. d. fulvus) for
splitting within this group.
all phylogenetic analyses (ML, BI and MP trees). This same nested
pattern is observed in ML and BI phylogenies for P. personatus
(Clade 3): Mexican samples (P. p. psilotis) are grouped as a highly 3.3. Species tree
supported clade inside the major clade containing remaining sam-
ples of P. personatus (Brazil, Guiana Shield, Guatemala - Fig. S3a). For performing this analysis, each divergent lineage identified
MP tree, however, displays the two mitochondrial lineages equiv- in the complete dataset topology (Fig. 3) was considered a different
alent to P. p. psilotis and P. p. personatus as monophyletic groups species under the multispecies coalescent model, designating a
(Fig. S3b). Regarding the two unnamed mitochondrial lineages group of individuals that have no history of breeding with individ-
within Clade 3, the specimen ROM98938 from Guatemala was uals outside that group (Heled and Drummond, 2010). Three of
placed together with P. p. personatus samples for nuclear sequences these lineages are equivalent to current Pteronotus species accord-
while the Guiana Shield specimens could not be included in nDNA ing to Fig. 3: P. gymnonotus, P. quadridens and P. macleayi. Eleven
analyses since only mtDNA sequences were available for them at lineages received names from correspondent subspecies in the tra-
the GenBank. The Antillean species composing Clade 2 (P. quadri- ditional taxonomy of P. parnellii, P. davyi and P. personatus com-
dens and P. macleayi) were recovered as reciprocal monophyletic plexes (Fig. 3b) but were ranked to the specific status given their
groups, reinforcing the pattern observed for mtDNA dataset. For higher distinctiveness in phylogenetic analysis. One last lineage
P. parnellii (Clade 4) the three main mtDNA clades above described has no available name in the current taxonomy and was referred
(A, B and C) were recovered by the nDNA dataset in all analyses, as Pteronotus sp1 (one South American lineage in P. parnellii com-
but monophyly of mitochondrial lineages was not always observed plex). The lineage identified in Clade 3 (P. personatus complex) by
(Fig. S3b). a unique specimen from Guatemala was not included in the species
In general, the overall nucleotide divergence among Pteronotus tree approach since we do not have satisfactory evidence it repre-
nDNA lineages is one order of magnitude smaller than in mtDNA, sents a distinct species. The topology resulted from this analysis
Fig. 4. Species tree for the genus Pteronotus estimated from the complete dataset, depicting the relationships among major clades according to the multispecies coalescent
approach. Names of terminals incorporate results of Fig. 3 and correspond to our new proposed hypothesis of species diversity in the genus Pteronotus (for more details, see
text).
(Fig. 4) presented the same relationship among lineages of genus different ratios of rostrum and braincase along the skulls morpho-
Pteronotus than phylogenetic analysis above described. types. The contrast between these two PCs established two main
groups in Pteronotus: the first group is composed by P. parnellii
complex specimens, which presents the higher scores in PC1; the
3.4. Morphometrics
second group includes all the remaining species, with smaller
PC1 scores than P. parnellii. The smaller Pteronotus species are dif-
The morphometric dataset was divided in 17 categories (or
ferentiated along the PC2, exhibiting a continuum of variation
grouping hypothesis) to be tested, corresponding to the currently
which departs from a small braincase e elongated rostrum in P.
described subspecies within the genus Pteronotus (Smith, 1972).
macleayi to a robust braincase and shorter and wider rostrum in
Variables means and standard deviations in each category are pre-
P. gymnonotus. It is noticeable that the total skull variation occu-
sented in Table S4. Although the same number of categories in
pied by P. parnellii complex in the principal component space is
morphometric analyses was tested in relation to the genetic lin-
equivalent to several species in the second group of Pteronotus.
eages found, they are not strictly equivalent. Some of the genetic
lineages identified (e.g. the unnamed mtDNA lineages of P. person-
atus and P. parnellii complexes) could not be established as mor-
3.4.2. Sexual variation
phological categories because they overlapped geographically the
Since we missed information on the geographic distribution/
lineages assigned to current subspecies. Since we lacked informa-
contact zones of the genetic lineages within species complexes,
tion on skull diagnostic characters for each of these lineages (this
we probably incurred in some errors during morphological groups
was not a goal of the present study), the assortment among them
delimitation, which primarily followed the available information
was not possible. Therefore, specimens belonging to such lineages,
on subspecies geographic range (Smith, 1972). In order to avoid
if sampled for morphometric data, were classified in the subspecies
false positive tests of sexual dimorphisms due to the existence of
recognized for that given locality according to the species taxon-
interspecific variation within some categories, the MANOVA tests
omy. On the other hand, some taxonomic units not identified as
were performed twice. The first test evaluated groups covering
independent lineages (subspecies of P. quadridens and P. macleayi)
the complete geographic range of the correspondent subspecies.
or not included in the molecular component (P. d. incae) were
Once such groups returned significant differences (p < 0.05)
tested in the morphometric dataset as distinct groups.
between genders, a second test with geographically restricted
groups, defined from one or some adjacent localities undoubtedly
3.4.1. Principal component analysis assigned to the same genetic lineage, was performed to confirm
The PCA included all Pteronotus specimens sampled and result. The second MANOVA test resulted in non-significant differ-
allowed a general outline of the skull morphological variation ences for some groups initially considered dimorphic (Table S5),
within the genus (Fig. S4). The first and second principal compo- which allowed the inclusion of both male’s and female’s original
nents (respectively, PC1 and PC2) explained 95% of the total varia- data for DFA. For the categories with significant results, we
tion of this group. PC1 represents a size vector while PC2 describes adjusted the female’s values adding the observed main difference
for each variable between males and females (Marroig and

Cheverud, 2004) and this corrected dataset was henceforth used
in DFA.
3.4.3. Discriminant analysis

DFA was performed in each of the major clades found in the
molecular results. This multivariate analysis provides a description
of differences among a priori groups through the extraction of clas-
sification (discriminant) functions from the raw morphometric
data. It generates tables showing the success classification rates
of specimens in each of the defined groups, i.e., the power of all dis-
criminant functions in assigning specimens into discrete morpho-
logical categories. Small differences in the original and jackknifed
classification matrices values are expected when discriminant
functions (DFs) allow robust sorting of specimens in the a priori
groups (Klecka, 1980). Together with molecular data, results
described by such analyses were our guidance for species delimita-
tion in the genus Pteronotus. In general, the extracted functions
showed a satisfactory to high discrimination among the majority
of the morphological groups established.
The DFA performed among the four categories established in
Clade 1 (davyi, fulvus, incae and gymnonotus) was highly significant
(Wilk’s lambda = 0.016, df = 123, 1208, p < 0.001). The three indi-
vidual discriminant functions (DF1 to 3) extracted from the data
represented 92%, 6% and 2% of the total variation, respectively.
The correlations between DF scores and skull measurements indi-
cate the DF1 as a size factor because all significant values are pos-
itive, whereas DF2 represents the ratio of the zygomatic
components in the total cranial length (data not shown). Post
hoc classification of cases in the defined categories presented accu-
racy in the classification rate superior to 85% both in original and
jackknifed matrices (Fig. 5; Table S6). The morphometric differen-
tiation among categories defined within Clade 1 is also represented
in Fig. 5 by the first two DF space plot. We observe a clear separa-
tion of categories fulvus and gymnonotus, while davyi and incae
show a slight overlap in their DF scores, suggesting a larger similar-
ity between them. It is important to highlight, however, that inter- Fig. 5. [above] Rates of correct classification of specimens in the morphological
pretation of the groups’ morphological distinctiveness based solely categories defined for the DFA within Clade 1; [below] Plot of Clade 1 morphological
groups scores against first two discriminant functions (DF1 and DF2).
on their position in the plot may be inaccurate, since there is a
third DF that also contributes to classification rates in the DFA (4
groups tested = 3 DFs). If we repeat the DFA without the species morphological groups for DFA analyses. Therefore, all specimens
P. gymnonotus, for example, position of the three groups from the measured from Brazil to Costa Rica were included in the category
P. davyi complex in the DF1 DF2 plot changes due to the decreas- personatus and the same was done for psilotis regarding specimens
ing in DF1 variation axis caused by removal of the gymnonotus from Honduras to Mexico (Smith, 1972). The DFA resulted in one
group, which has a larger size. highly significant discriminant function (Wilk’s lambda = 0.103,
Within Clade 2 (P. macleayi and P. quadridens), we considered df = 41, 228, p < 0.001), which describes the size difference
the two subspecies currently recognized within each species between categories. Both original and jackknifed classification
(Table 1) as independent categories in the DFA in order to matrices show correct assignments of specimens to the categories
investigate their morphological differentiation (Smith, 1972). In close to 100% (Fig. 6; Table S8), suggesting a robust separation of
both cases the extracted discriminant function (DF1) was signifi- P. personatus complex in two morphologically distinct groups.
cant (P. macleayi: Wilk’s lambda = 0.074, df = 33, 10, p = 0.015; For the morphometric analysis in Clade 4, specimens of
P. quadridens: Wilk’s lambda = 0.134, df = 41, 35, p < 0.001) and P. parnellii complex were split into seven categories, three of them
represents size variation between the subspecies. Fig. 6 shows in Caribbean Islands - parnellii, portoricensis and pusillus – and four
the success classification rates and the differences in DF1 values in the continental portion – mexicanus, mesoamericanus, fuscus
exhibited by the two morphological categories within each species. and rubiginosus. Group’s delimitation followed primarily the
Post hoc classification of cases in the defined categories (sub- geographic distribution observed for the genetic lineages, with
species) was higher in P. quadridens than in P. macleayi (Fig. 6; their contacts zones inferred from information provided by the
Table S7). In both cases, however, the values reported by jackknifed subspecies geographic range (Smith, 1972). As occurred in the
classification matrices were much smaller than the original P. personatus complex, the unnamed South American lineage of
matrices, suggesting an unsatisfactory assignment of data to the P. parnellii complex could not be sorted from its sympatric sister-
categories, likely due to the low sampling values. group rubiginosus by cranial characters. The unique measured
The DFA within Clade 3 (P. personatus complex) was performed specimens surely classified as Pteronotus sp1 were those also
between two categories: personatus and psilotis. As said previously, sequenced (n = 11). Since we probably have many other specimens
we had no information regarding geographic range and cranial belonging to the Pteronotus sp1 lineage hidden within the cranial
diagnostic characters for testing the unnamed genetic mtDNA sampling of P. p rubiginosus, allocating just the sequenced speci-
lineages from Guiana Shield and Guatemala as independent mens of Pteronotus sp1 in a separate morphological group would
Fig. 6. [above] Rates of correct classification of specimens in the morphological categories defined for the DFA within P. quadridens and P. macleayi (Clade 2) and P. personatus
complex (Clade 3); [below] Range of DF1 values exhibited by specimens of each morphological category. Numbers above bars represent canonical scores of group means. Note
that, despite their distinctiveness in DF1 axis, jackknifed classification rates of P. quadridens and P. macleayi groups were low, probably due to closer values in the group’s
variables means and small sample size.
not make sense. Therefore, all specimens from South America intermediate phenotype between mesoamericanus and rubiginosus
southward Orinoco River were grouped in the category rubiginosus presented by this group, but still satisfactory (Table S9).
and considerations regarding the DFA results will be make further. Additional exploratory DFAs including just the insular or conti-
The six discriminant functions extracted from DFA were signif- nental lineages were performed to evaluate the morphological dis-
icant (Wilk’s lambda = 0.017, df = 287, 4912, p < 0.001), with the tinctiveness of some specimens (Fig. S5). For both analyses, the
first three explaining 70%, 17% and 5% of total variation, respec- specimens investigated were not defined as a priori groups in
tively. The correlations between DF scores and skull variables sug- DFA. This enabled us checking their position in the DF1 DF2
gest DF1 as a size vector, which allows a major discrimination space without assuming the hypothesis they represent indepen-
between insular and continental lineages (Fig. 7). The DF2 dent morphological groups, since this would orientate the analysis
describes the relative proportion of facial components in the total in finding discriminant functions maximizing their differences.
cranial size, representing an important differentiation axis along This is an important detail to be mentioned because the multivari-
continental lineages. The post hoc classification scores of speci- ate approach attempts to maximize differences just among the a
mens in the defined categories were in general smaller than in priori defined groups. The DFA comprising only insular lineages
the other Pteronotus clades, although the original and jackknife included one measured specimen of P. p. gonavensis - a subspecies
classification matrices consistently presented close values for most described from fossil specimens found in La Gonave Island, Haiti -
groups (Fig. 7; Table S9). The greater uncertainty observed in the to verify its position in the DF space generated by the lineages cur-
classification results regarded the assignment of specimens to rently distributed in the Greater Antilles. The analysis included
portoricensis, which was considered unsatisfactory for our criteria only 37 distances since the specimen of P. p. gonavensis is damaged.
(difference larger than 10% between classification rates of original The result shows gonavensis close to specimens of pusillus, suggest-
and jackknifed matrices). This group presented a high overlap of its ing a high cranial similarity between these two taxonomic units.
first two DFs with parnellii, suggesting these insular categories are The DFA of mainland lineages analyzed specimens of Pteronotus
very similar morphological units (Fig. 7). Post hoc classification of sp1 (n = 11) and P. paraguanensis (n = 2), plotting their scores
cases to fuscus by jackknife resampling method was lower than within the DF space created by the four mainland categories. The
those of remaining continental groups, probably due to the result show P. paraguanensis specimens apart from the remaining
supports within a phylogeny (Knowles and Carstens, 2007; Corl

and Ellegren, 2013).
4.2. Diversification pattern in the genus Pteronotus
The phylogenetic inference points to a close relationship

between the species P. gymnonotus and P. davyi complex within
Clade 1, and P. macleayi and P. quadridens within Clade 2, which
were recovered as sister groups in all performed analysis. Addi-
tionally, phylogenetic analyses suggests that Clades 1 and 2 have
the most recent common ancestor in the genus, followed by a clo-
ser relationship with Clade 3, while Clade 4 is the sister group to all
other clades. The multispecies coalescent model also produced the
same topology as result, providing independent evidence for such
evolutionary pattern within the genus. The basal position of
P. parnellii complex is already reported by previous molecular
investigations (Lewis-Oritt et al., 2001; Van Den Bussche and
Weyandt, 2003; Dávalos, 2006) and the relationship among
Pteronotus species presented here received strong node supports
in the final analysis. In accordance with molecular data, the general
pattern of cranial variation provided by PCA places all lineages of
P. parnellii complex in a group detached from the remaining
species, mostly due to its larger size (PC1). This dichotomy
between P. parnellii complex and the other Pteronotus species also
receives support from other biological aspects, such as morphology
(Smith, 1972; Simmons and Conway, 2001) and echolocation calls
(Novick, 1963; Fenton et al., 2012; Mancina et al., 2012).
Regarding taxonomy, our phylogeny supports the subgenus
Pteronotus (Clade 1) and Phyllodia (Clade 4) as natural arrange-
ments, whereas points to subgenus Chilonycteris (Clade 2 + Clade
3) as an artificial group. This subgenus was described by Smith
Fig. 7. [above] Rates of correct classification of specimens in the morphological (1972) including the species P. macleayi, P. quadridens and
categories defined for the DFA within Clade 4; [below] Plot of Clade 4 morphological P. personatus. Simmons and Conway (2001) corroborated the group
groups scores against first two discriminant functions (DF1 and DF2). Additional monophyly based on morphological data, although presented a
plots within Clade 4 are presented in Fig. S5. weak node support for the clade. Molecular evidences, however,
does not support this arrangement (Lewis-Oritt et al., 2001; Van
Den Bussche et al., 2002; Van Den Bussche and Weyandt, 2003;
groups, with their DF scores closer to mexicanus and fuscus. present work) and we recommend the allocation of P. personatus
Pteronotus sp1, on the other hand, is completely overlapped with complex (Clade 3) in a new subgenus.
rubiginosus in this DF1 DF2 space.
4.3. Integrative taxonomy
4. Discussion
Our results also confirm previous reports on the underesti-
4.1. Performance of phylogenetic methods and molecular datasets mated diversity within the genus (Lewis-Oritt et al., 2001;
Dávalos, 2006; Clare et al., 2011, 2013; Thoisy et al., 2014). Three
The results obtained by probabilistic methods of phylogenetic of the major clades herein described include species complexes,
inference (BI and ML) presented a more similar result than the entities receiving the same name under current taxonomy but har-
MP search. This fact may be consequence of the distinct searching boring two or more evolutionary lineages: P. parnellii, P. personatus
strategies adopted by these approaches, where MP search may pre- and P. davyi. One impressive result is the equivalence of several
sent a lower performance due to the incomplete lineage sorting in current subspecies described in Pteronotus with a mitochondrial
one or more markers or because of the presence of short branches lineage. These lineages, in turn, could be discriminated by skull
deep in the tree (Kubatko and Degnan, 2007). In general, results morphometrics, suggesting most taxonomic units previously iden-
obtained by the different datasets were highly congruent in estab- tified as subspecies by Smith (1972) actually deserve the specific
lishing four major clades within Pteronotus as well as the genetic status. The existence of discontinuities in morphometric data,
lineages they encompass. Although nDNA alone did not strongly combined with other criteria such as geography and biological
support all lineages identified by mtDNA, molecular data investi- aspects, were considered solid evidences when integrated to the
gated through multiloci approaches (with 5 or 6 genes) clearly molecular results for delimiting species in this bat group following
recover the same evolutionary lineages. Regarding the relationship the unified proposal of the General Lineage Concept of Species (de
pattern among these major clades, more robust datasets were Queiroz, 1998). Below we discuss the diversity within each of the
necessary to present a confident and accurate hypothesis on clades identified in molecular results based on the available evi-
the evolutionary history of this group. The topologies found by dences (summarized in Table S10), providing a taxonomic update.
the five-gene and complete datasets are in accordance with Clade 1 – Corresponds to the subgenus Pteronotus described by
relationships presented by the mtDNA dataset, but higher support Smith (1972), being currently composed by P. davyi complex and
values are observed in the deep tree nodes. This result highlights P. gymnonotus. The phylogenetic results exhibit all P. gymnonotus
the importance of independent loci, with distinct evolutionary specimens in a shallow and weakly-structured monophyletic
rates, for depicting the basal relationships and adequate node group. The analysis of skull variation also indicates a single
morphological unit within this taxon. The pattern of star-like tree hypothesis could not be verified in deep due to sample scarcity
presented by P. gymnonotus, with short internal branches and low for two of these lineages, one in the Guiana Shield and other in
nucleotide divergences, suggests a rapid population growth or a Guatemala. We were able to identify the existence of two discrete
recent origin for this species (Nordborg, 2001; Wakeley, 2004). units within this complex through multiloci and morphological
For P. davyi complex, molecular results show two allopatric, recip- data (Table S10). The two well sampled lineages form mono-
rocally monophyletic groups corresponding to the geographic phyletic and genetically cohesive groups whose geographic ranges
range of current subspecies P. d. davyi and P. d. fulvus. Cranial overlap those of the two described subspecies within the taxon tra-
morphometrics strengthens this result, pointing to divergent ditionally classified as P. personatus (Smith, 1972). These groups
morphogroups for these lineages. The third subspecies, P. d. incae, are corroborated by cranial morphometric analysis, which strongly
shows morphometric distinctiveness and geographic isolation, supports the separation of P. personatus complex in two entities
suggesting this taxon may represent a third species within this (Fig. 6). Based on these results, we recommend the elevation of
complex, but its absence from the phylogenetic investigation subspecies P. p. personatus and P. p psilotis to the species level.
prevents us for making such recognition. Therefore, Clade 3 corresponds to a new subgenus in Pteronotus
Therefore, we assume the existence of three species within encompassing at least two allopatric species. P. personatus com-
Clade 1: P. davyi, P. fulvus and P. gymnonotus (Tables S10). Morpho- prises populations from Mato Grosso state in Brazil northward to
metric and geographic evidences suggest the isolated population of Costa Rica, while P. psilotis includes individuals from Honduras to
P. davyi complex in Peru (P. d. incae) is closer to P. davyi, so we con- Mexico (Table S11). These geographic ranges essentially follow
sider incae as a geographic variation of the specie P. davyi until its subspecies distribution proposed by Smith (1972), although we
phylogenetic position and genetic distinctiveness is investigated. provide new occurrence records of P. personatus in Brazil
The relationship among species within Clade 1 was not well (Fig. S2b). Additional data is nonetheless necessary to better inves-
resolved in our phylogenetic analysis: mtDNA suggests a sister tigate the diversity within this complex as well as to infer species
relationship between P. davyi and P. gymnonotus, though the final ranges in Central and northern South America.
molecular dataset points to the closer relation between P. davyi Clade 4 – Corresponds to the subgenus Phyllodia (Smith, 1972)
and P. fulvus with low support values. The proposed geographic and is composed by distinct populations of P. parnellii complex,
range for P. gymnonotus is being updated by including several many of them already recognized as valid species by previous stud-
Brazilian localities (Fig. S2a, Table S11); ranges of P. davyi and ies (Gutiérrez and Molinari, 2008; Clare et al., 2013; Thoisy et al.,
P. fulvus are provisionally being kept equivalent to the ranges pro- 2014). Our phylogenetic results point to three divergent clades
posed by Smith (1972). While P. davyi and P. fulvus apparently pre- within this complex, recovered by all datasets. Molecular data sug-
sent allopatric distributions regarding each other, P. gymnonotus gests a closer relationship between Clades 4-A (Caribbean Islands)
occurs sympatrically with the other two. Curiously, we have found and 4-B (Central and northern South America) while Clade 4-C
one specimen of P. gymnonotus with introgressed mtDNA from (exclusively South American samples) diverged first during group
P. fulvus (ROM 98442), two clearly established species. More sam- diversification. This process apparently occurred in the same time
pling in Central America is therefore necessary to explore the exis- frame, in a rapid radiation event, given the short internal branch
tence of a contact zone between P. davyi and P. fulvus, as well as to splitting Clade 4-C from the other two (see Fig.3).
better understand interspecific interactions of P. gymnonotus with Mitochondrial data points to eight distinct lineages in
its sister-groups. P. parnellii complex, with nucleotide average distances between 3
Clade 2 – Comprises the Antillean species P. quadridens and P. and 15%. This result exceeds diversity levels previously reported
macleayi and should be synonymized to the subgenus Chilonycteris, (Clare et al., 2011, 2013; Thoisy et al., 2014) and shows conspecific
which currently also includes P. personatus (Smith, 1972). These lineages of Clade 4 diverging more than species currently described
two species are displayed as monophyletic groups in all molecular (Bradley and Baker, 2001; Baker and Bradley, 2006). The nDNA
markers investigated. Both P. macleayi and P. quadridens exhibit a reveals a more discrete structure within P. parnellii complex, where
geographical structure in the mitochondrial haplotypes, but the mtDNA lineages are not always recovered as monophyletic groups.
correspondence with subspecies currently described occurs just Nevertheless, this result is expected, since nuclear genes have lar-
for P. macleayi. Additionally, divergence values observed between ger effective population sizes (Ne) and consequently longer time
such geographic groups are small when compared to the inter- scales of the coalescent process (Nordborg, 2001; Wakeley, 2004;
specific variation reported for bats (Bradley and Baker, 2001). Degnan and Rosenberg, 2009). Given its reduced Ne and relatively
The morphometric results indicate the existence of size differences rapid rate of evolution, lineage sorting in mtDNA occurs faster than
between groups within both species, but larger sampling are nec- in nDNA. Consequently, diverging patterns in this molecule are
essary to better investigate their distinctiveness. Overall, results anticipated as evidence of speciation (Rosenberg and Nordborg,
from molecular and morphometric approaches suggest the varia- 2002; Knowles and Carstens, 2007; Weisrock et al., 2010), even
tion observed is compatible with population structuring. The enough time has not elapsed to generate the diagnostic phyloge-
diversity in Clade 2 is hence kept to two species, P. quadridens netic pattern of distinct evolutionary paths, i.e. reciprocal mono-
and P. macleayi, with almost sympatric ranges along the Caribbean phyly (de Queiroz, 1998), in multiple loci.
islands (Table S11). The skull variation within this complex is characterized by a
The genetic and morphological cohesion observed for the two marked division between Caribbean islands and American conti-
species of Clade 2 is contrasting given their long branches in the nent, as highlighted by PCA (Fig. S4). Insular lineages are smaller,
phylogeny, which stand for a relatively old age of these lineages. with size increasing gradually in the continent, from mexicanus
The similarity among intraspecific mitochondrial haplotypes in in the north of the geographic range, to rubiginosus in the south,
P. macleayi and P. quadridens may reflect a recent history of gene where individuals exhibit the largest cranial sizes of P. parnellii
flow among populations occupying different islands within each complex. For the insular groups (Clade 4-A), the single distinct
species range. The morphological similarities, on the other hand, morphological category was pusillus. The cranial similarity
can be consequence of adaptations to the insular habitat or derived between parnellii and portoricensis pointed by our results (Fig. 7)
from a short period of drift after gene flow was interrupted. was previously reported by Smith (1972). Notwithstanding, the
Clade 3 – Mitochondrial data suggests a much more complex author justified the recognition of different subspecies for
pattern in this group than currently recognized by taxonomy, with Jamaican (parnellii) and Puerto Rican (portoricensis) populations
P. personatus complex comprising four distinct lineages. This due to their geographic isolation by P. p. pusillus, a notably smaller
population inhabiting Haiti and Dominican Republic. In addition, undoubtedly places it within the fuscus clade. The cause of such
molecular evidence also strongly suggests that parnellii and discrepancy in the two mitochondrial markers regarding this indi-
portoricensis represent independent lineages, not being even vidual was not investigated by us since it does not weaken our
sister-groups (Figs. 3 and 4). Thus we recommend the elevation interpretations. In general, lineages of Clade 4-B appear to present
of the taxonomic status of all current insular populations of parapatric distributions regarding each other, although additional
P. parnellii complex to the species level, namely P. parnellii (sensu data is necessary to explore contact zones among them.
strictu), P. pusillus and P. portoricensis. The taxonomic status of the Mormoopids are gregarious and obligatory cave dwellers, which
Cuban population remains uncertain since we did not include sam- probably restrict their dispersion to new habitats presenting this
ples from this island in our analysis. Following Smith (1972), this ecological requirement. Thus, geography in the case of Clade 4-B
population would belong to the species P. parnellii (sensu strictu) possibly account as a factor limiting gene flow between distant
together with Jamaica. However, it may also represent another evo- populations. Besides, distinct echolocation frequencies are
lutionary lineage since geographic isolation of the Greater Antilles reported for many localities within the continental range of Clade
apparently acted as an effective barrier to gene flow, originating 4 (Novick, 1963; Gutiérrez and Molinari, 2008; Clare et al., 2013;
island-specific lineages within this complex. Regarding the sub- Thoisy et al., 2014), which has been shown to have a dual function
species P. p. gonavensis, the availability of just one complete skull in the species ecological interactions, ultimately facilitating speci-
prevented us from making further interpretations. The DFA suggest ation in other HDC bats (Kingston and Rossiter, 2004). Therefore,
this fossil specimen is very similar, but a little smaller, to P. pusillus adaptation to the echolocation strategy seems to be the key for
(Fig. S5a). The geographic overlap and morphological resemblance understanding the diversification process within this complex
between these two entities might be evidences that they actually (Clare et al., 2013). Following this hypothesis, diverging lineages
belong to the same species, sampled in different moments of its in Clade 4 may have kept very similar cranial features during their
evolutionary history. Based on this, we provisionally consider evolutionary history given the anatomical and physiological
P. p. gonavensis as a temporal variation of P. pusillus until a deeper requirements necessary to perform HDC strategy, and this would
investigation with this taxon is performed. explain their morphometric similarity.
The DFA comprising just the mainland lineages shows mexi- Based on our results and additional information regarding the
canus and rubiginosus as discrete, non overlapped morphological group biology (Table S10), we recognize five species in mainland
units. The specimens from the geographically intermediate groups, P. parnellii complex: P. mexicanus, P. mesoamericanus, P. fuscus,
mesoamericanus and fuscus, are more similar between each other P. rubiginosus and Pteronotus sp1, which will be further described.
and with those in the tip groups, and consequently were misclas- Tentative geographic ranges for these species are presented in
sified more frequently. Nevertheless, errors in the allocation of Table S11, but sampling gaps precluded us from making a complete
specimens to the defined morphological categories might be analysis of their distribution. Our results corroborate the work of
occurred, due to the lacking of knowledge regarding the contact Clare et al. (2013) in suggesting the existence of four species of
zones between adjacent lineages. This means that specimens actu- P. parnellii in the continent from Southern Mexico to Guiana, and
ally belonging to the group fuscus may be originally assigned by us we further include a fifth species in Western Mexico. Another
to mesoamericanus or rubiginosus because we do not have full infor- species, P. paraguanensis, was not well investigated through our
mation on the lineage’s geographic ranges. These errors may have dataset due to the low sample size, but the morphometric
caused a decreasing of the successful classification rates in DFA. In distinctiveness of the two measured specimens in our DFA led us
addition, the two sympatric lineages in South America were to maintain the specific status proposed by Gutiérrez and
included in just one morphological category, rubiginosus. Since pre- Molinari (2008). In total, we thereby recommend the recognition
vious works report a morphometric discrimination between these of nine species within P. parnellii complex, three of them in
lineages (Clare et al., 2013; Thoisy et al., 2014), the allocation of all Caribbean Islands and six in the American continent. However,
specimens into one single group probably inflated its morphologi- the inclusion of molecular data is crucial to fully investigate the
cal variation and contributed for a larger overlapping between phylogenetic position and taxonomic status of P. paraguanensis.
rubiginosus and other groups. Taking these facts into account, the Finally, a deeper knowledge on the acoustic variation along the
morphometric differentiation among mainland groups of complex range is needed and will certainly contribute to elucidate
P. parnellii complex is probably higher than results in Fig. 7 show. the secondary contact zones in this group.
The available data suggests all mainland lineages of P. parnellii
complex may represent valid species. The two lineages of Clade 4.4. Species classification and taxonomic arrangement
4-C (rubiginosus and Pteronotus sp1) are represented by clades dis-
tributed in South America, with specimens collected syntopically Most of the Pteronotus groups currently recognized at the sub-
in several localities in the Guiana Shield and Brazilian Amazon species rank in taxonomy fulfilled species criteria, possessing a
(Fig. S2c). Persistency of reciprocal monophyly between sympatric range of diagnostic features including cranial morphometric differ-
sister-groups constitutes strong evidence that they are in indepen- ences, mtDNA reciprocal monophyly, multiple polymorphisms in
dent evolutionary paths (Coyne and Orr, 2004). Also, previous nuclear sequences and differences in phonic types (mean constant
studies report some morphometric differences and distinct peak frequency). Based on an integrative approach, we propose a new
frequencies in the echolocation patterns of these two lineages taxonomic arrangement for this genus, composed by 16 extant
(Clare et al., 2013; Thoisy et al., 2014). The three lineages from and one fossil species (Table 2). Some of the taxa herein listed were
Clade 4-C (mexicanus, mesoamericanus and fuscus) occur in a wide suggested as valid species by previous works (Dávalos, 2006; Clare
geographic range from Mexico to northern South America. They et al., 2013; Thoisy et al., 2014), which did not depict, however, a
show the smallest between-group mtDNA divergences within the full phylogenetic framework of the group.
genus (3–5%), suggesting they represent recently diverged lin- The present work represents the more complete investigation
eages. We have observed a direct, strong relation between speci- performed in the genus Pteronotus including multiple and indepen-
men’s geographic origin and mitochondrial clades for the large dent sources of evidence (maternal, paternal and autosomal
sampling within this clade (N = 78), demonstrating its higher geo- genetic systems and morphometrics). The morphological diagnosis
graphic structuring. We found one single specimen (TK128442) of the species herein recognized are however beyond our scope,
presenting a questionable position in our mitochondrial phylogeny since it requires the identification of discrete characters represen-
due its divergent CYTB haplotype, though its COI haplotype tatives of each taxon. The cranial variation we reported was just
Table 2 throughout group diversification, despite the deep genetic diver-

Changes in taxonomic arrangement proposed for the Family Mormoopidae by the gence distinguishing them. A deeper and more detailed study on
present study when compared to traditional classification (Smith, 1972; Simmons,
2005).
the taxonomic units proposed is thought necessary, including
information of bioacoustics, morphology and geographical varia-
Smith (1972)/Simmons (2005) Present study tion, to better understand the evolution of this interesting
Family Mormoopidae Family Mormoopidae Neotropical bat group. More refined hypotheses of taxa evolution-
Genus Mormoops Genus Mormoops ary history and diversity, like the one above shown, represent the
Mormoops blainvillei Mormoops blainvillei
Mormoops megalophylla Mormoops megalophylla
first step for answering broader questions on evolutionary and
Mormoops m. megalophylla Mormoops m. megalophylla ecological aspects of Neotropical life history.
Mormoops m. intermedia Mormoops m. intermedia
Mormoops m. tumidiceps Mormoops m. tumidiceps Acknowledgements
Mormoops m. carteri Mormoops m. carteri
Mormoops magna b Mormoops magna b
Genus Pteronotus Genus Pteronotus This research was supported by the Brazilian agencies Fundação
Subgenus Pteronotus Subgenus Pteronotus de Amparo à Pesquisa do Estado de São Paulo (FAPESP grants
Pteronotus davyi Pteronotus davyi 2009/54731-0 and 2011/14295-7) and Coordenação de Aperfeiçoa-
Pteronotus d. davyi Pteronotus d. davyi
mento de Pessoal de Nível Superior (CAPES PDSE grant 4179-11-0).
Pteronotus d. fulvus Pteronotus d. incae
Pteronotus d. incae Pteronotus fulvus Financial Support was also provided by a grant from the Smithso-
Pteronotus gymnonotus Pteronotus gymnonotus nian Institution and we thank Don E. Wilson for his kind assistance
Subgenus Chilonycteris Subgenus Chilonycteris on that. We are very grateful to curators and staff who provided
Pteronotus quadridens Pteronotus quadridens access to specimens in scientific collections: AMNH (N. Simmons,
Pteronotus q. quadridens Pteronotus q. quadridens
E. Westwig, N. Duncan); FMNH (B. Patterson); Smithsonian Institu-
Pteronotus q. fuliginosus Pteronotus q. fuliginosus
Pteronotus macleayi Pteronotus macleayi tion (D. Wilson, K. Helgen, D. Lunde, S. Peurach); TTU (R. Baker, H.
Pteronotus m. macleayi Pteronotus m. macleayi Garner, K. Macdonald); TCWC (B. Marks, J. Light); KUNHM (R.
Pteronotus m. griseus Pteronotus m. griseus Timm); LACM (J. Dines); IEPA (I. Castro, C. Silva), INPA (M.N. da
Pteronotus personatus New Subgenus
Silva, M. Borges, S. Farias); UFPE (D. Astua, P.P. Alves); UFPB (P.
Pteronotus p. personatus Pteronotus personatus
Pteronotus p. psilotis Pteronotus psilotis Estrela, A. Feijó); UFMG (A. Paglia, L. Moras, M. Bento); MZUSP
Subgenus Phyllodia Subgenus Phyllodia (M. de Vivo, J. Gualda). Many thanks to the researchers and institu-
Pteronotus parnellii Pteronotus parnellii tions who loaned us tissues and made this work possible: V. C.
Pteronotus p. parnellii Pteronotus portoricensis Tavares, A. C. Martins, A. Césari, C. Aires, A. Feijó, P. Rocha,
Pteronotus p. portoricensis Pteronotus pusillus
F. Martins, L. Pessoa, M. Oprea, M. Marcos, F. Santos, T. Oliveira,
Pteronotus p. pusillus Pteronotus p. gonavensisb
Pteronotus p. gonavensisb Pteronotus mexicanus E. Bernard, B. Thoisy, F. Catzeflis, C. Bernabé, NSRL-TTU (R. Baker),
Pteronotus p. mexicanus Pteronotus mesoamericanus ROM (B.K. Lim), FMNH (B. Patterson) and AMCC-AMNH (L. Davalos,
Pteronotus p. mesoamericanus Pteronotus fuscus A. Tejedor, N. Simmons). We also thank Albert Ditchfield, Paúl
Pteronotus p. fuscus Pteronotus paraguanensisa
Velazco and two anonymous reviewers for reading early drafts of
Pteronotus p. paraguanensis Pteronotus rubiginosus
Pteronotus p. rubiginosus Pteronotus sp1
the manuscript and making helpful suggestions for its
Pteronotus pristinus b Pteronotus pristinus b improvement.
a
Taxon not investigated in the present study. Taxonomic status provided by
Gutiérrez and Molinari (2008). Appendix A. Supplementary material
b
Extinct.
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.ympev.2016.07.
quantitative, precluding us from providing such information. 011.
While this information is not available, we propose species identi-
fication in the genus Pteronotus follow provisionally the geographic
References
criteria (Table S11), which delimits satisfactorily most of taxo-
nomic units of the genus but a few contact zones previously Adams, D.C., Berns, C.M., Kozak, K.H., Wiens, J.J., 2009. Are rates of species
discussed. diversification correlated with rates of morphological evolution? Proc. R. Soc. B:
Biol. Sci. 276, 2729–2738.
Avise, J.C., Arnold, J., Ball, R.M., Bermingham, E., Lamb, T., Neigel, J.E., Reeb, C.A.,
Saunders, N.C., 1987. Intraspecific phylogeography the mitochondrial DNA
5. Conclusions bridge between population genetics and systematics. Ann. Rev. Ecol. Syst. 18,
489–522.
Baker, R., Bradley, R., 2006. Speciation in mammals and the Genetic Species
With this study the family Mormoopidae, a small Neotropical Concept. J. Mammal. 87, 643–662.
group of insectivores bats, has its diversity increased from six spe- Baker, R., Porter, C., Patton, J., Van Den Bussche, R., 2000. Systematics of bats of the
cies to more than twice this number. Integrated analysis of molec- family phyllostomidae based on RAG2 DNA sequences. Occas. Pap. Tex. Tech
Univ. Mus. 202, 1–16.
ular and morphometric data support the recognition of almost all Bickford, D., Lohman, D.J., Sodhi, N.S., Ng, P.K.L., Meier, R., Winker, K., Ingram, K.K.,
nominal taxa in the genus Pteronotus as valid species. The taxa Das, I., 2007. Cryptic species as a window on diversity and conservation. Trends
P. parnellii, P. davyi and P. personatus represent species complexes, Ecol. Evolut. 22, 148–155.
Borisenko, A.V., Lim, B.K., Ivanova, N.V., Hanner, R.H., Hebert, P.D.N., 2008. DNA
as suggested by previous works, encompassing nine, two and two barcoding in surveys of small mammal communities: a field study in Suriname.
species respectively. We propose a new taxonomic arrangement Mol. Ecol. Resour. 8, 471–479.
for the genus Pteronotus composed by 16 extant and one fossil Bradley, R.D., Baker, R.J., 2001. A test of the genetic species concept: cytochrome-b
sequences and mammals. J. Mammal. 82, 960–973.
species.
Bruford, M.W., Hanotte, O., Brookfield, J.F.Y., Burke, T., 1992. Single-locus and
Our work comprised a geographically broad sampling and large multilocus DNA fingerprinting. In: Hoelzel, A.R. (Ed.), Molecular Genetic
exploration of data within Pteronotus, demonstrating how integra- Analyses of Populations. A Practical Approach. IRL Press, Oxford, pp. 225–269.
tive analysis of multiples sources of evidences contributes to the Cheverud, J.M., 1982. Phenotypic, genetic and environmental morphological
integration in the cranium. Evolution 36, 499–516.
study of Neotropical bat diversity. Species of Pteronotus appear to Cheverud, J.M., 1996. Developmental integration and the evolution of pleiotropy.
have kept their cranial and external morphology relatively similar Am. Zool. 36, 44–50.
Clare, E.L., 2011. Cryptic species? Patterns of maternal and paternal gene flow in 8 Morgan, G.S., Czaplewski, N.J., 2012. Evolutionary history of the Neotropical
neotropical bats. PLoS ONE 6, e21460. Chiroptera: the fossil record. In: Gunnell, G.F., Simmons, N.B. (Eds.),
Clare, E.L., Adams, A.M., Maya-Simões, A.Z., Eger, J.L., Hebert, P.D., Brock, F.M., 2013. Evolutionary History of Bats: Fossils, Molecules and Morphology. Cambridge
Diversification and reproductive isolation: cryptic species in the only New University Press, New York, USA, pp. 105–161.
World high-duty cycle bat, Pteronotus parnellii. BMC Evol. Biol. 13, 26. Nordborg, M., 2001. Coalescent Theory. Handbook of Statistical Genetics.
Clare, E.L., Lim, B.K., Fenton, M.B., Hebert, P.D.N., 2011. Neotropical bats: estimating Novick, A., 1963. Orientation in neotropical bats. II. Phyllostomatidae and
species diversity with DNA barcodes. PLoS ONE 6, e22648. desmodontidae. J. Mammal. 44, 44–56.
Corl, A., Ellegren, H., 2013. Sampling strategies for species trees: the effects on Parlos, J.A., Timm, R.M., Swier, V.J., Zeballos, H., Baker, R.J., 2014. Evaluation of
phylogenetic inference of the number of genes, number of individuals, and paraphyletic assemblages within Lonchophyllinae, with description of a new
whether loci are mitochondrial, sex-linked, or autosomal. Mol. Phylogenet. Evol. tribe and genus. Occas. Pap. Tex. Tech Univ. Mus., 320
67, 358–366. Patton, J.L., Gardner, A.L., 2007. Family Mormoopidae. In: Gardner, A.L. (Ed.),
Coyne, J.A., Orr, H.A., 2004. Speciation. W.H. Freeman. Mammals of South America. University of Chicago Press, Chicago, Illinois, pp.
Dayrat, B., 2005. Towards integrative taxonomy. Biol. J. Linnean Soc. 85, 407–415. 376–384.
de Queiroz, K., 1998. The general lineage concept of species, species criteria, and the Pavan, A.C., Martins, F.M., Morgante, J.S., 2013. Evolutionary history of bulldog bats
process of speciation: a conceptual unification and terminological (genus Noctilio): recent diversification and the role of the Caribbean in
recommendations. In: Howard, D.J., Berlocher, S.H. (Eds.), Endless Forms. Neotropical biogeography. Biol. J. Linnean Soc. 108, 210–224.
Species and Speciation. Oxford University Press, Oxford, England, UK, p. 470. R Development Core Team, 2013. R: A Language and Environment for Statistical
Degnan, J.H., Rosenberg, N.A., 2009. Gene tree discordance, phylogenetic inference Computing. R Foundation for Statistical Computing, Vienna, Austria.
and the multispecies coalescent. Trends Ecol. Evolut. 24, 332–340. Rambaut, A., 2012. FigTree. Institute of Evolutionary Biology, University of
Drummond, A.J., Rambaut, A., 2007. BEAST: Bayesian evolutionary analysis by Edinburgh.
sampling trees. BMC Evol. Biol., 7 Rambaut, A., Suchard, M., Drummond, A., 2013. Tracer. Institute of Evolutionary
Dávalos, L.M., 2006. The geography of diversification in the mormoopids Biology, University of Edinburgh.
(Chiroptera: Mormoopidae). Biol. J. Linnean Soc. 88, 101–118. Reig, S., 1996. Correspondence between interlandmark distances and caliper
Eick, G.N., Jacobs, D.S., Matthee, C.A., 2005. A nuclear DNA phylogenetic perspective measurements. In: Marcus, L., Corti, M., Loy, A., Naylor, G.P., Slice, D. (Eds.),
on the evolution of echolocation and historical biogeography of extant bats Advances in Morphometrics. Springer, US, pp. 371–385.
(chiroptera). Mol. Biol. Evol. 22, 1869–1886. Ronquist, F., Huelsenbeck, J.P., 2003. MrBayes 3: Bayesian phylogenetic inference
Ewing, B., Green, P., 1998. Basecalling of automated sequencer traces using Phred II. under mixed models. Bioinformatics 19, 1572–1574.
Error probabilities. Genome Res. 8, 186–194. Ronquist, F., Huelsenbeck, J.P., van der Mark, P., 2010. MrBayes 3.2 Manual.
Ewing, B., Hillier, L., Wendl, M., Green, P., 1998. Basecalling of automated sequencer Rosenberg, N.A., Nordborg, M., 2002. Genealogical trees, coalescent theory and the
traces using phred. I. Accuracy assessment. Genome Res. 8, 175–185. analysis of genetic polymorphisms. Nature 3, 380–390.
Fenton, M.B., Faure, P.A., Ratcliffe, J.M., 2012. Evolution of duty cycle echolocation in Schlick-Steiner, B.C., Steiner, F.M., Seifert, B., Stauffer, C., Christian, E., Crozier, R.H.,
bats. J. Exp. Biol. 215, 2935–2944. 2010. Integrative taxonomy: a multisource approach to exploring biodiversity.
Goloboff, P., Farris, S., Nixon, K., 2003. TNT (Tree Analysis Using New Technology). Annu. Rev. Entomol. 55, 421–438.
Available at: <http://www.lillo.org.ar/phylogeny/tnt/>. Silva-Taboada, G., 1974. Fossil Chiroptera from cave deposits in central Cuba, with
Gordon, D., Abajian, C., Green, P., 1998. Consed: a graphical tool for sequence descriptions of two new species (genera Pteronotus and Mormoops) and the first
finishing. Genome Res. 8, 195–202. West Indian record of Mormoops megalophylla. Acta Zool. Cracov. 19, 34–73.
Gutiérrez, E.E., Molinari, J., 2008. Morphometrics and taxonomy of bats of the genus Simmons, N.B., 2005. Order chiroptera. In: Wilson, D.E., Reeder, D.M. (Eds.),
Pteronotus (subgenus Phyllodia) in Venezuela. J. Mammal. 89, 292–305. Mammal Species of the World: A Taxonomic and Geographic Reference. Johns
Heled, J., Drummond, A.J., 2010. Bayesian inference of species trees from multilocus Hopkins University Press, Baltimore, Maryland.
data. Mol. Biol. Evol. 27, 570–580. Simmons, N.B., Conway, T.M., 2001. Phylogenetic relationships of mormoopid bats
Hutson, A.M., Mickleburgh, S.P., Racey, P.A., 2001. Microchiropteran Bats: Global (Chiroptera: Mormoopidae) based on morphological dataBull. Am. Mus. Nat.
Status Survey and Conservation Action Plan. IUCN/SSC Chiroptera Specialist Hist. 258, 97.
Group., IUCN, Gland, Switzerland and Cambridge, UK. Smith, J.D., 1972. Systematics of the Chiropteran Family Mormoopidae, vol. 56.
Kingman, J.F.C., 1982. On the genealogy of large populations. J. Appl. Probab. 19A, Miscellaneous Publication, Mus. Nat. Hist., Univ. Kansas, pp. 1–132.
27–43. Tamura, K., Stecher, G., Peterson, D., Filipski, A., Kumar, S., 2013. MEGA6: molecular
Kingston, T., Lara, M.C., Jones, G., Akbar, Z., Kunz, T.H., Schneider, C.J., 2001. Acoustic evolutionary genetics analysis version 6.0. Mol. Biol. Evol. 30, 2725–2729.
divergence in two cryptic Hipposideros species: a role for social selection? Proc. Thoisy, B.D., Pavan, A.C., Delaval, M., Lavergne, A., Luglia, T., Pineau, K., Ruedi, M.,
R. Soc. B: Biol. Sci. 268, 1381–1386. Rufray, V., Catzeflis, F., 2014. Cryptic diversity in common mustached bats
Kingston, T., Rossiter, S.J., 2004. Harmonic-hopping in Wallacea’s bats. Nature 429, Pteronotus cf. parnellii (Mormoopidae) in French Guiana and Brazilian Amapa.
654–657. Acta Chiropterol 16, 1–13.
Klecka, W.R., 1980. Discriminant Analysis. Sage (ed. 19). Van Den Bussche, R.A., Hoofer, S.R., Simmons, N.B., 2002. Phylogenetic relationships
Knowles, L.L., Carstens, B.C., 2007. Delimiting species without monophyletic gene of mormoopid bats using mitochondrial gene sequences and morphology. J.
trees. Syst. Biol. 56, 887–895. Mammal. 83, 40–48.
Kubatko, L.S., Degnan, J.H., 2007. Inconsistency of phylogenetic estimates from Van Den Bussche, R.A., Weyandt, S.E., 2003. Mitochondrial and nuclear DNA
concatenated data under coalescence. Syst. Biol. 56, 17–24. sequence data provide resolution to sister-group relationships within
Lanfear, R., Calcott, B., Ho, S.Y.W., Guindon, S., 2012. PartitionFinder: combined Pteronotus (Chiroptera: Mormoopidae). Acta Chiropterol. 5, 1–13.
selection of partitioning schemes and substitution models for phylogenetic Velazco, P.M., Patterson, B.D., 2008. Phylogenetics and biogeography of the broad-
analyses. Mol. Biol. Evol. 29, 1695–1701. nosed bats, genus Platyrrhinus (Chiroptera: Phyllostomidae). Mol. Phylogenet.
Lewis-Oritt, N., Porter, C.A., Baker, R.J., 2001. Molecular systematics of the family Evol. 49, 749–759.
mormoopidae (chiroptera) based on cytochrome b and recombination Velazco, P.M., Patterson, B.D., 2013. Diversification of the yellow-shouldered bats,
activating gene 2 sequences. Mol. Phylogenet. Evol. 20, 426–436. genus Sturnira (Chiroptera, Phyllostomidae), in the New World tropics. Mol.
Lim, B.K., Engstrom, M.D., Bickham, J.W., Patton, J.C., 2008. Molecular phylogeny of Phylogenet. Evol. 68, 683–698.
New World sheath-tailed bats (Emballonuridae: Diclidurini) based on loci from Wakeley, J., 2004. Inferences about the structure and history of populations:
the four genetic transmission systems in mammals. Biol. J. Linnean Soc. 93, coalescents and intraspecific phylogeography. In: Singh, R.S., Uyenoyama, M.K.
189–209. (Eds.), The Evolution of Population Biology. Cambridge University Press.
Linares, O.J., Ojasti, J., 1974. Una nueva subespecie del murciélago Pteronotus Weisrock, D.W., Rasoloarison, R.M., Fiorentino, I., Ralison, J.M., Goodman, S.M.,
parnellii, en las cuevas de la Península de Paraguaná, Venezuela (Chiroptera: Kappeler, P.M., Yoder, A.D., 2010. Delimiting species without nuclear
Mormoopidae). Bol. Soc. Venezolana Espeleología 5, 73–78. monophyly in Madagascar’s mouse lemurs. PLoS ONE 5, e9883.
Maddison, W.P., 1997. Gene Trees in Species Trees. Syst. Biol. 46, 523–536. Wiens, J.J., 2007. Species delimitation: new approaches for discovering diversity.
Mancina, C.A., García-Rivera, L., Miller, B.W., 2012. Wing morphology, echolocation, Syst. Biol. 56, 875–878.
and resource partitioning in syntopic Cuban mormoopid bats. J. Mammal. 93, Will, K.W., Mishler, B.D., Wheeler, Q.D., 2005. The perils of DNA barcoding and the
1308–1317. need for integrative taxonomy. Syst. Biol. 54, 844–851.
Marroig, G., Cheverud, J.M., 2004. Cranial evolution in sakis (Pithecia, Platyrrhini) I: Xia, X., 2013. DAMBE5: a comprehensive software package for data analysis in
interspecific differentiation and allometric patterns. Am. J. Phys. Anthropol. 125, molecular biology and evolution. Mol. Biol. Evol. 30, 1720–1728.
266–278. Zwickl, D.J., 2006. GARLI: Genetic Algorithm Approaches for the Phylogenetic
Martins, F.M., Ditchfield, A.D., Meyer, D., Morgante, J.S., 2007. Mitochondrial DNA Analysis of Large Biological Sequence Datasets Under the Maximum Likelihood
phylogeography reveals marked population structure in the common vampire Criterion. The University of Texas at Austin, Austin, Texas.
bat, Desmodus rotundus (Phyllostomidae). J. Zool. Syst. Evol. 45, 372–378.

Pavan 2016 - Filogenia Taxonomia Integrativa Murcielagos

Uploaded by

Copyright:

Available Formats

Pavan 2016 - Filogenia Taxonomia Integrativa Murcielagos

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Pavan 2016 - Filogenia Taxonomia Integrativa Murcielagos

Uploaded by

Copyright:

Available Formats

Molecular Phylogenetics and Evolution 103 (2016) 184–198

Contents lists available at ScienceDirect

Molecular Phylogenetics and Evolution

Integrating multiple evidences in taxonomy: species diversity and

1. Introduction taxonomic issues in systematic studies (Dayrat, 2005; Will et al.,

Species Subspecies Type locality Geographic range

autosomal genes analysis, the presence of heterozygous nucleotide

2.2. Morphological sampling and cranial landmarks

The morphometric investigation was performed with 1628

for each variable between males and females (Marroig and

3.4.3. Discriminant analysis

supports within a phylogeny (Knowles and Carstens, 2007; Corl

4.2. Diversification pattern in the genus Pteronotus

The phylogenetic inference points to a close relationship

Table 2 throughout group diversification, despite the deep genetic diver-

You might also like