ArticlePDF Available

Abstract and Figures

We describe a new species of Enyalius endemic to the Brazilian Cerrado, based on morphological and molecular data sets. In the face of uncertain taxonomy among museum specimens of Enyalius, we used a novel analytical approach based on Gaussian mixture modeling for species assignments. We also used a machine-learning classification procedure (random forests) to investigate morphological variation and identify species diagnostic characters. Phylogenetic and species delimitation analyses supported the distinction of the new species from its congeners. The new species is characterized by the fewest ventral scales and smallest snout–vent length in the genus. Moreover, we infer that this species diverged from its closest relative, E. bilineatus, in the late Miocene, presumably after colonization of Cerrado gallery forests by an Atlantic Forest ancestor, followed by ecological or geographical speciation linked to shrinkage or fragmentation of gallery forests associated with global cooling and increased aridity. Rapid conversion of natural habitats, the isolation of protected areas, and recent changes to the Brazilian Forest Code pose serious threats to the conservation of the new species described herein, and other gallery forest inhabitants. Key words: Brazil; Conservation; DNA barcoding; Integrative taxonomy; Molecular data; Morphological data
Content may be subject to copyright.
BioOne Complete ( is a full-text database of 200 subscribed and open-access
titles in the biological, ecological, and environmental sciences published by nonprofit societies,
associations, museums, institutions, and presses.
Your use of this PDF, the BioOne Complete website, and all posted and associated content indicates
your acceptance of BioOne’s Terms of Use, available at
Usage of BioOne Complete content is strictly limited to personal, educational, and non-commercial use.
Commercial inquiries or rights and permissions requests should be directed to the individual publisher
as copyright holder.
BioOne sees sustainable scholarly publishing as an inherently collaborative enterprise connecting authors, nonprofit
publishers, academic institutions, research libraries, and research funders in the common goal of maximizing access to critical
A New Species of Enyalius (Squamata, Leiosauridae) Endemic to the
Brazilian Cerrado
Authors: M. Florencia Breitman, Fabricius M.C.B. Domingos, Justin C. Bagley, Helga C.
Wiederhecker, Tayná B. Ferrari, et. al.
Source: Herpetologica, 74(4) : 355-369
Published By: Herpetologists' League
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
Herpetologica, 74(4), 2018, 355–369
Ó2018 by The Herpetologists’ League, Inc.
A New Species of Enyalius (Squamata, Leiosauridae) Endemic to the Brazilian Cerrado
Departamento de Zoologia, Universidade de Bras´
ılia, Bras´
ılia, DF 70910-900, Brazil
Instituto de Ciˆ
encias Biol´
ogicas e da Sa ´
ude, Universidade Federal de Mato Grosso, Pontal do Araguaia, MT 78698-000, Brazil
Departamento de Zoologia e Botˆ
anica, Universidade Estadual Paulista, Sa
˜o Jos´
e do Rio Preto, SP 15054-000, Brazil
Campus I, Universidade Cat´
olica de Bras´
ılia, ´
Aguas Claras, DF 71966-700, Brazil
Instituto Federal do Piau´
ı, Teresina, PI 64000-040, Brazil
ABSTRACT: We describe a new species of Enyalius endemic to the Brazilian Cerrado, based on morphological and molecular data sets. In the
face of uncertain taxonomy among museum specimens of Enyalius, we used a novel analytical approach based on Gaussian mixture modeling for
species assignments. We also used a machine-learning classification procedure (random forests) to investigate morphological variation and identify
species diagnostic characters. Phylogenetic and species delimitation analyses supported the distinction of the new species from its congeners. The
new species is characterized by the fewest ventral scales and smallest snout–vent length in the genus. Moreover, we infer that this species diverged
from its closest relative, E. bilineatus, in the late Miocene, presumably after colonization of Cerrado gallery forests by an Atlantic Forest ancestor,
followed by ecological or geographical speciation linked to shrinkage or fragmentation of gallery forests associated with global cooling and
increased aridity. Rapid conversion of natural habitats, the isolation of protected areas, and recent changes to the Brazilian Forest Code pose
serious threats to the conservation of the new species described herein, and other gallery forest inhabitants.
Key words: Brazil; Conservation; DNA barcoding; Integrative taxonomy; Molecular data; Morphological data
THE GENUS Enyalius (Wagler 1830) (Squamata, Leiosaur-
idae) includes small- to medium-sized lizards restricted to
Brazil (Etheridge 1969; Rodrigues et al. 2014) that are
diurnal, ombrophilous, semiarboreal, and insectivorous (Vitt
et al. 1996; Rautenberg and Laps 2010; Barreto-Lima and
Sousa 2011; Maia-Carneiro et al. 2016). Coloration patterns
are rarely exclusive to a given species, and are also highly
variable within and between sexes (Jackson 1978). Ten
species are currently recognized within Enyalius (Costa and
ernils 2014): E. brasiliensis,E. boulengeri,E. iheringii,E.
perditus, and E. pictus, mainly distributed in the Atlantic
Forest; E. erythroceneus in the Caatinga; E. leechii restricted
to the Amazon Forest; E. catenatus and E. bibronii in the
Atlantic Forest and Caatinga; and E. bilineatus, mainly in the
Atlantic Forest but also reported in the Cerrado. Recent
molecular evidence indicates that other undescribed lineages
might be present (Rodrigues et al. 2014).
The taxonomic history of Enyalius has been dynamic since
its origins (Etheridge 1969; Jackson 1978). Over the last
decade, the genus transitioned from having 6 recognized
species (and 3 subspecies) to include 10 species. This change
resulted from new species being described (Rodrigues et al.
2006), morphological evidence supporting the recognition of
E. catenatus,E. bibronii, and E. pictus as full species, and
molecular evidence supporting E. brasiliensis being recon-
sidered as a full species and the resurrection of E. boulengeri
(Rodrigues et al. 2014). Phylogenetic relationships among
Enyalius species were first proposed by Jackson (1978) using
morphology, and later modified by Frost et al. (2001) based
on a combination of morphological and mitochondrial DNA
data. The first comprehensive molecular phylogeny of the
genus was published only recently, however, based on one
nuclear and three mitochondrial markers analyzed with the
use of concatenation and species tree methods (Rodrigues et
al. 2014).
Although Enyalius bilineatus is known to occur primarily
in the Atlantic Forest, samples collected in the Cerrado have
also been referred to this species (e.g., Rodrigues et al. 2014;
Ledo and Colli 2016). Recent evidence indicates that these
samples might represent an undescribed species. Rodrigues
et al. (2014) inferred Cerrado samples as phylogenetically
sister to E. bilineatus, with an estimated time of divergence
from the most recent common ancestor (t
) around 6.86
million years ago (mya), in the late Miocene. Pairwise
cytochrome-bgenetic distances between E. bilineatus and its
sister molecular lineage (including samples from Bras´
Distrito Federal and Mariana, Minas Gerais) are 14.5%
(Vargas et al. 2015), which is a representative value among
different lizard lineages (Breitman et al. 2012). Moreover,
these two clades display marked differentiation, as they
occur in different biomes without significant distributional
overlap (Barreto-Lima 2012), their distributions are ex-
plained by unique combinations of bioclimatic predictors
(Barreto-Lima 2012), and lizards from Bras´
ılia seem to be
smaller than nominal Enyalius species, including samples of
E. bilineatus from other areas.
Systematists have the responsibility to address the ongoing
biodiversity crisis, which challenges us to publish accurate
species descriptions (Wilson 1985; Wheeler 2004; Agnarsson
and Kuntner 2007). Integrative taxonomy, in which several
independent lines of evidence are used for species
recognition (Dayrat 2005; Padial et al. 2010) under the
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
General Lineage Concept (species as independently evolving
lineages; de Queiroz 2007), has become a useful framework
to generate high-quality species descriptions (Pante et al.
2015). The goal of our study is to use molecular and
morphological data to delimit and describe a new species of
Enyalius lizard from the Cerrado in central Brazil. Given
that downstream morphological, phylogenetic, and species
delimitation analyses depend on the tip labels used (e.g., on
guide trees; Yang and Rannala 2010; Yang 2015), we also
develop a novel analytical framework to label samples
objectively while dealing with uncertainty in taxonomic
assignments and to identify traits robustly for species
Morphological Data
We studied 259 ethanol-preserved Enyalius samples
collected from throughout Brazil (deposited in different
museums), including representatives of all species and
putative candidate species (Rodrigues et al. 2014; Fig. 1;
also see the Supplemental Material 1 file, available online).
We included samples from or near type localities (Table 1) of
all species except for E. brasiliensis, as the only credible
records of this species come from Rio de Janeiro (Rodrigues
et al. 2014), ~1000 km north of the type locality in Santa
Catarina (Lesson 1830). Rodrigues et al. (2014) recovered
both E. catenatus and E. perditus as forming two polyphy-
letic and allopatrically distributed clades that diverged since
the Miocene–Pliocene boundary (BEAST t
,E. cate-
natus ¼14 mya; BEAST t
,E. perditus ¼4.2 mya).
Enyalius catenatus includes one clade distributed in Bahia
(E. catenatus 1 in Rodrigues et al. 2014: Fig. 1) and another
in Alagoas (E. catenatus 2). Given that the type locality of E.
catenatus is Jequi´
e, Bahia (Table 1), and assuming that
samples collected at or near the type locality are more likely
to belong to the nominal species, we considered the clade
from Bahia as representing E. catenatus and the clade from
Alagoas as an undescribed species. Similarly, E. perditus
includes one clade from Juquitiba and Sales ´
opolis, Sa
˜o Paulo
(E. perditus 1 in Rodrigues et al. 2014: Fig. 1), and another
from Sa
˜o Paulo and surrounding states (E. perditus 2). Given
that E. perditus 1 includes samples from the type locality
(Table 1), we assumed that it represents E. perditus, whereas
‘‘E. perditus 2’’ represents an undescribed species. We refer
to populations from Bras´
ılia and surrounding areas (identi-
fied as E. bilineatus by Rodrigues et al. 2014) as Enyalius sp.
and evaluate their specific status throughout the present
We assessed variation at 55 characters (24 meristic, 29
qualitative, and 2 morphometric characters; see Appendix
and Supplemental Material 2, available online). We took
measurements with rulers (60.1 mm) and made scale counts
and qualitative observations with stereomicroscopes. Scale
counts and terminology followed Smith (1946), Etheridge
(1969), Jackson (1978), and Rodrigues et al. (2006). Because
of variation in character names among previous publications,
however, we present a detailed list of character names and
states (Appendix). Our study included several observers and,
to minimize biases, we divided characters rather than
samples among observers. Each observer collected data on
one to a few characters, depending on their respective
complexities (e.g., one observer was responsible only for
collecting data on ventral scales). Because we could not
unambiguously determine sex by checking cloacal or tail size
and coloration patterns, we determined sex only when we
unequivocally identified ovaries or testicles by dissection. We
did not dissect individuals loaned from other collections, so
our sample size for sexed individuals was limited in all
species (except for Enyalius sp.). Because of these limita-
tions, and following previous papers describing Enyalius
species, we did not divide our data set by sex and
compensated for this shortcoming by including multiple
individuals per species from a range of sampling localities
(see Jackson 1978; Rodrigues et al. 2006).
After assembling an initial data set, we identified outliers
for each trait and the corresponding observer remeasured
specimens as needed, or noted if the measurement was a
true outlier. We detected univariate outliers by using box
plots and converting observations to Z-values, regarding
those with P(Z.jzj),0.0001 as outliers. We also evaluated
multivariate outliers by conducting a principal-components
analysis of morphological characters, converting scores on
the first two principal components to Z-values, and regarding
those with P(Z.jzj),0.0001 as outliers.
Taxonomic Assignment:
Semisupervised Gaussian Mixture Modeling
When identifying specimens of Enyalius, researchers are
often challenged by uncertainties resulting from uninforma-
tive original descriptions. To obtain an objectively labeled
data set for downstream morphological and genetic analyses
in the face of taxonomic uncertainty in the genus, we
performed a semisupervised classification analysis (Weston
et al. 2005). This analysis was based on Gaussian mixture
models (GMM) with the bgmm R package (Biecek et al.
2012), as automated in GaussClust (Bagley 2017), and used
all 55 morphological characters. In semisupervised classifi-
cation, a subset of observations in a data set may be labeled
(classified with certainty), while no information is provided
about the labels (classes) of the remaining observations (e.g.,
because of dubious taxonomic IDs). The modeling approach
then uses the knowledge of the labeled data to improve
TABLE 1.—Type localities of Enyalius species according to original
descriptions. All localities are in Brazil, and correct spellings or more precise
definitions are given in parentheses.
Species Type locality
E. bibronii Boulenger 1885 ‘‘Bahia’’
E. bilineatus Dum´
eril and
Bibron 1837
‘‘Br ´
E. boulengeri Etheridge 1969 ‘‘Espirito Santo, Brazil’’ (¼Esp´
ırito Santo)
E. brasiliensis (Lesson 1830) ‘‘Sainte-Catherine du Br ´
esil’’ (¼Santa
E. catenatus (Wied 1821) ‘‘Cabe¸ca do Boi’’ (¼Jequi ´
e, Bahia;
Bokermann 1957; Moraes 2009, 2011)
E. erythroceneus (Rodrigues
et al. 2006)
‘‘Fazenda Cara´
ıbas, Mucugˆ
e, Bahia’’
E. iheringii Boulenger 1885 ‘‘Rio Grande do Sul’’
E. leechii (Boulenger 1885) ‘‘Santarem’’ (¼Santar´
em, Para
E. perditus Jackson 1978 ‘‘Esta¸ca
˜o Biol´
ogica de Borac´
ıpio de Sales´
opolis, Estado de
˜o Paulo, Brazil’’
E. pictus (Schinz 1822) ‘‘Mucur´
ı, southeastern Bah´
ıa, Brazil’’
(¼Mucuri, Bahia)
356 Herpetologica 74(4), 2018
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
prediction of class membership (i.e., species assignment) of
unlabeled observations. The goal of this analysis was not to
perform a test of the validity of the described or undescribed
species. Rather, this framework objectively labels, for
downstream analyses, the individuals that could belong to
described or undescribed taxa but (1) lacked appropriate
diagnostic characters that could be matched to nominal
species, (2) exhibited geographical overlap with other species
(sympatry), or (3) lacked detailed locality data, or (4) lacked
genetic data or were not from a locality for which genetic
data were available for other individuals; (Supplemental
Material 3, available online).
We started with a highly conservative and robust
classification by providing unequivocal labels for 214
individuals (Set A) based on two sets of criteria. (1) For
described species, we considered the simultaneous occur-
rence of ID labels from museum curators, proximity to type
locality, and four easily distinguished morphological char-
acters (EDMC) regarded as diagnostic (keeled vs. smooth
subdigital lamella, keeled vs. smooth ventral scales, small
vs. enlarged subocular scale, and presence vs. absence of
dorsal crest; Etheridge 1969; Rodrigues et al. 2006). (2) For
undescribed species, we compared the geographical
sampling localities with the geographical distribution of
the corresponding molecular lineage in Rodrigues et al.
(2014), in addition to remoteness to known type localities,
and variation at EDMC. Next, we subjected the unequiv-
ocally labeled specimens (Set A) along with the remaining,
unlabeled specimens (Set B) to the semisupervised
classification analysis with the use of bgmm (criteria
presented in Supplemental Material 3). Unlabeled individ-
uals that were then assigned to a species/lineage with high
FIG. 1.—Distribution map of Brazilian biomes showing morphological sampling. The star indicates the type locality of Enyalius capetinga sp. nov.
(Reserva Ecol´
ogica do IBGE, Bras´
ılia, Distrito Federal, Brazil; 15.94678S, 47.86868W; 1154 m above sea level). A color version of this figure is available
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
probabilities by bgmm were double-checked and excluded
from further analysis if a mistake could have been made,
e.g., in cases of damaged specimens (Supplemental
Material 4, available online). In downstream analyses, we
only included the labeled specimens (Set A) plus a subset of
the unlabeled individuals (Set B) that were subsequently
classified with posterior probabilities .0.95 by our semi-
supervised GMM, and met strict criteria used to decide
whether or not a sample was known (Supplemental
Materials 2, 4).
Finding Morphological Differences:
Guided Regularized Random Forest
After semisupervised GMM analyses, we arrived at a
highly conservative and accurate data set including 228
individuals and 45 morphological characters (excluding
‘‘scales along the tail,’’ which had a large amount of missing
data, as well as monomorphic characters; see Appendix and
Supplemental Material 2). We assessed morphological
variation among species and identified the best predictors
of species with the use of a guided regularized random forest
(GRRF) analysis (Supplemental Material 5, available online).
GRRF is an enhanced regularized random forest method
where importance scores of a previous run on a complete
training data set are used to guide the predictor selection
process (Deng 2013). Importance scores are estimated by
penalizing the selection of new predictors for splitting the
decision tree when the gain (i.e., mean decrease in Gini
impurity; Mingers 1989) is similar to that of features used in
previous splits (Breiman 2001). Machine learning algo-
rithms, such as support vector machines and GRRF, have
been used in recent systematics papers (Murphy et al. 2016)
and have been shown to outperform traditional linear
methods (Domingos et al. 2014), while also avoiding their
limitations (Quinn and Keough 2002). In addition, meristic
data sets frequently do not meet linear model assumptions
(Houle 1992); thus, applying linear classification methods to
meristic data could lead to erroneous results (Cutler et al.
Prior to the GRRF analyses, we imputed 81 missing
values (0.8% of the total data set) with the use of multivariate
imputation by chained equation in the mice R package (van
Buuren and Groothuis-Oudshoorn 2011), with 100 multiple
imputations. To determine predictor importance in discrim-
inating among Enyalius lineages/species, we implemented
GRRF analyses in the RRF package (Deng 2013) on each
imputed matrix, and calculated the average importance of
each predictor across all runs. For each of the 100 imputed
matrices, we also calculated GRRF model accuracy based on
50 replicates of fivefold cross-validation (Breiman and
Spector 1992; Kohavi 1995), sequentially increasing the
number of predictors based on their importance (Supple-
mental Material 5). Finally, to identify the best predictors of
the new species, we conducted three GRRF analyses
comparing the new species: (1) against E. bilineatus only
(the closest relative of Enyalius sp.), (2) against all other
species except E. bilineatus, and (3) against all species except
E. bilineatus and E. leechii (E. leechii has a distinctively high
number of vertebral scales). Analyses were performed in R
v3.3.1 (R Core Team 2017).
Molecular Data
We combined new and previously published genetic data
(Supplemental Materials 6 and 7, available online) including
79 cytochrome-bsequences (cyt-b; 470 bp), 73 NADH
dehydrogenase subunit four sequences (ND4; 817 bp), and
71 sequences of a nuclear protein-coding gene, the c-mos
proto-oncogene (503 bp), for all described and candidate
species. We selected these markers because they are
phylogenetically informative for Enyalius (Rodrigues et al.
2014). Novel sequences included 14 individuals of Enyalius
sp. from Bras´
ılia and surrounding areas, as well as 2
individuals of E. bibronii from Crato, Ceara
´. Remaining
data from Rodrigues et al. (2014) were downloaded from
GenBank. We treated samples identified as E. bilineatus in
Rodrigues et al. (2014), but collected in Bras´
ılia, as Enyalius
sp., and we treated samples identified as E. catenatus 1and
E. perditus 1 in Rodrigues et al. (2014: Fig. 1) as belonging
to the nominal species. We used samples from specimens of
Anisolepis grillii,A. longicauda, and Urostrophus vautieri
(all Iguania:Leiosauridae) as outgroups.
We extracted genomic DNA with the use of Qiagent
DNeasyt96 Tissue Kit, following the manufacturer’s
protocol. We amplified cyt-b, ND4, and c-mos with the
primers and PCR protocols given in Rodrigues et al. (2014).
We also sequenced a mitochondrial COI fragment (629 bp)
for the holotype of the new species (CHUNB 74591) with
the primers and PCR protocols described in Nagy et al.
(2012). Sequencing was conducted on an ABI 3130xl
sequencer and in-house protocols. We edited sequences
with Geneious v9.1.7 (Kearse et al. 2012) and aligned them
with PASTA v1.6.4 (Mirarab et al. 2015) with MAFFT used
as the alignment algorithm, OPAL as merger, RAxML as the
tree estimator (applying the GTRþGmodel),andan
iteration limit setting of three. We assigned a barcode to
the holotype of the new species and attached metadata to the
sequence following the Barcode of Life Data System
(BOLD) manual (Ratnasingham and Hebert 2007).
Phylogenetic and Species Delimitation Analyses
Prior to phylogenetic analyses, we estimated a partitioning
scheme with the use of PartitionFinder v2.1.1 (Lanfear et al.
2017), while separating all genes by codon position and
estimating partitions under the GTRþG family of models.
Models were selected based on the Akaike information
criteria with the greedy algorithm (Lanfear et al. 2012, 2014).
We ran maximum-likelihood phylogenetic analyses in
RAxML v8.2.8 (Stamatakis 2014) with the use of partitions
(Supplemental Material 8, available online), and estimated
nodal support with 100 rapid bootstraps. We considered
nodal support significant or moderate for bootstrap values
over 95% (Felsenstein and Kishino 1993) and 70% (Hillis
and Bull 1993), respectively. Bayesian inference analysis was
performed in ExaBayes v1.5 (Aberer et al. 2014) with the
same partition schemes and linking branch lengths across
partitions. We performed two independent runs with four
chains each, starting from a parsimony tree, for at least one
million generations (sampling every 500th) until the average
standard deviation of split frequencies was below 0.01,
indicating good convergence. A minimum acceptable
effective sample size of 200 was reached for all parameters,
and we checked the potential scale reduction factor (~1.0)
358 Herpetologica 74(4), 2018
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
with the use of the postProcParam and extractBips utilities
distributed with ExaBayes. We considered Bayesian poste-
rior probabilities .0.95 as significant (Huelsenbeck and
Ronquist 2001). All of the above phylogenetic analyses used
tip labels (individually or in aggregate) derived from the
semisupervised GMM analyses in bgmm.
We tested species limits among Enyalius by conducting
joint Bayesian species delimitation and species tree estima-
tion in BPP v3.3 (Yang 2015). This program uses the
multispecies coalescent model and accounts for incomplete
lineage sorting attributable to ancestral polymorphism and
gene tree-species tree conflict (Yang and Rannala 2010,
2014). Individuals were assigned to species hypotheses
(putative new and nominal species) based on their placement
in the Bayesian and ML phylogenetic trees, and on results
from the semisupervised GMM analysis. We ran BPP for 5 3
generations, sampling every 5 generations, and discard-
ing the first 10,000 steps as burn-in. To account for
potentially different speciation histories, we ran BPP with
different gamma priors on population sizes (hs) and age of
the root of the species tree (s
), as follows: (1) large
population size and deep divergence, ~G(2, 1000) for both
priors; (2) small population size and shallow divergence,
~G(1, 10) for both priors; (3) large population size and
shallow divergence, ~G(2, 2000) for hs, ~G(1, 10) for ss; (4)
small population size and deep divergence, ~G(1, 10) for hs,
~G(2, 2000) ss; and (5) intermediate values of ~G(2, 100)
for both priors. Other divergence time parameters were
assigned a Dirichlet prior (Yang and Rannala 2010: Equation
2). We also used both available cleandata options (one using
the data as is, and the other omitting gaps or ambiguity
characters). We checked for convergence by running BPP
twice for each prior and cleandata option, starting from
different random trees. All runs returned very similar results;
thus, we report results from eight runs (four for each
cleandata option) with ~G(2, 1000) used for both priors,
which represent realistic biological choices for the species
(Sturaro and da Silva 2010; Barreto-Lima and Sousa 2011;
Rodrigues et al. 2014).
Morphological Results
The classification derived from semisupervised GMM
analysis assigned 17 of the unlabeled individuals (unknowns)
to a species with high posterior probability (P.0.95) and, in
seven cases, the assignment matched their original field/
curator IDs (Supplemental Materials 3–5). We identified 3
individuals, of the 17 unknowns classified with high posterior
probability, as potentially being erroneously classified (see
Supplemental Materials 3–5). The remaining 14 individuals
were labeled according to the bgmm results. Our final matrix
included 45 morphological characters measured for 228
individuals of Enyalius (Appendix and Supplemental Mate-
rials 2–5). Means 61 SD, and ranges, for predictors and
lineages are presented in Supplemental Material 2.
Across lineages, the GRRF results (Fig. 2) revealed a
cross-validation error ranging from 0.65–0.1 (corresponding
to using the single best predictor to using all predictors). The
GRRF model based on the five best predictors (ventrals,
vertebrals, suboculars, scales around the tail, and paraver-
tebrals) had an accuracy of ~70%, based on 5000 replicates
of fivefold cross-validation (Fig. 2). The GRRF model
comparing the new species with pooled samples of all other
species except E. bilineatus identified ventrals, supraciliaries,
paravertebrals, gulars (from mental scale to gular fold), and
suboculars as the best predictors. Likewise, the model
comparing the new species against E. bilineatus identified
the number of ventral scales, shape of lateral scales, number
of dorsolateral tibials, and tail length as the four best
predictors (details in Supplemental Material 9, available
online). Overall, our GRRF results highlight several
dimensions of morphological evidence supporting Enyalius
sp. as a new species. Preliminary GRRF analysis also
indicated that E. catenatus 2andE. perditus 2are
morphologically different from other Enyalius species (E.
catenatus 2: number of dorsolateral tibials, number of
vertebral and midbody scales; E. perditus 2: number of
ventral, paravertebral, and midbody scales).
Phylogenetic Analyses and Species Delimitation Results
Our maximum-likelihood and Bayesian inference recon-
structions showed congruent results (Fig. 3; Supplemental
Material 10, available online). We recovered individuals
from each nominal and putative new species in distinct
clades with significant to moderate nodal support, and with
an overall similar topology (cf. Rodrigues et al. 2014). Minor
differences between the Bayesian and ML trees were only
observed in the placement of individuals within species. Our
species delimitation analyses inferred nominal and putative
new Enyalius species as different entities with posterior
probabilities above 0.93 in cleandata runs, and above 0.99 in
no cleandata runs. No other species hypothesis had a
posterior probability of more than 0.05, and the hypothesis
of 16 full species (including outgroups) had the highest
posterior probability across runs. Finally, the coalescent
species tree estimated by BPP recovered E. leechii and E.
erythroceneus in a different position compared to ML and
Bayesian concatenated estimations (Supplemental Material
10). These differences had no influence on the species
delimitation results, but they should be addressed in future
research that incorporates additional molecular markers and
broader geographical sampling. Overall, we found strong
molecular support for Enyalius sp. being a new species,
because BPP inferred the corresponding samples as forming
a distinct species with posterior probability of 1 in all runs.
This species was also strongly supported as sister to E.
bilineatus across the BPP coalescent species tree, as well as
Bayesian and maximum- likelihood gene trees.
Enyalius capetinga sp. nov.
(Figs. 4–6)
Enyalius bilineatus: Rodrigues et al. (2006: 11–24), Ro-
drigues et al. (2014: 137–146), Ledo and Colli (2016: 98–
Enyalius aff. bilineatus: Nogueira et al. (2009: 83–96).
Holotype.—CHUNB 74591 (Figs. 4, 5), an adult male
collected by Ana Cec´
ılia Holler Del Prette and Anderson
Kennedy Soares de Lima on 10 June 2016 at Reserva
ogica do IBGE (Instituto Brasileiro de Geografia e
ılia, Distrito Federal, Brazil
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
(15.94678S, 47.86868W; 1154 m above sea level [asl]; in all
cases, datum ¼WGS84).
Paratypes.—CHUNB 8173, 8906, 29290, 29292, and
29295 adult males, 23847 juvenile male, and CHUNB 8168,
8182, 29312, 52396, 52407 adult females; all from Bras´
Distrito Federal, Brazil (14.7572 S, 47.9761 W [1154 m asl]
except for CHUNB 52497 from 15.9467 S, 47.8686 W [1154
m asl]); collected by Guarino Rinaldi Colli and students of
Universidade de Bras´
ılia, except for CHUNB 52407
collected by Roger Maia Dias Ledo. Paratypes collected
between 1994 and 2008. Specifically, CHUNB 8173 on 9
May 1994, CHUNB 8906 between 1994 and 1998, CHUNB
29290, 29292, 29295, and 29312 between 2001 and 2003,
CHUNB 23847 in 2001, CHUNB 8168 on 15 June 1994,
CHUNB 8182 on 22 August 1991, CHUNB 52396 on 23
April 2008, and CHUNB 52407 on 20 May 2008.
Diagnosis.Enyalius capetinga differs from E. bilinea-
tus mainly in having fewer ventral (34.45 62.25 [29–40] vs.
43.2 64.06 [35–49] in E. bilineatus) and dorsolateral tibial
scales (10.38 61.29 [8–13] vs. 12.52 60.59 [12–14] in E.
bilineatus), and a higher number of midbody (52.96 64.34
[42–64] vs. 45.8 63.64 [39–52] in E. bilineatus)and
vertebral scales (68.8 64.94 [55–82] vs. 63.56 64.88 [52–
70] in E. bilineatus; Table 2; Supplemental Materials 2 and
10). Also, all samples of E. capetinga present granular
lateral scales, whereas only 32% of E. bilineatus samples
present this character state, and the remaining individuals
(68%) present non-granular scales. Enyalius capetinga
differs from all other species of the genus (excluding E.
bilineatus; see character summary in Table 2 and Supple-
mental Materials 2, 10) mainly in having the smallest count
of ventral scales (34.45 62.25), supraciliaries (6.59 6
1.64), paravertebral scales (63.84 65.96), dorsolateral
tibials (10.38 61.29), supralabials (7.86 61.03), and scales
from mental to gular fold (34.34 64.04). Likewise, the new
species is distinguished from all others in the genus by
having keels on most dorsal scales. Enyalius capetinga
differs from most other Enyalius species in lacking contact
between nasal and postrostral scales (absent in 89% of the
samples), which is common in most other species (exclud-
ing E. bilineatus and E. pictus); having an enlarged
subocular scale (present in 96% of E. capetinga samples)
that is smaller in the majority (.70%) of individuals of
other species (except E. bilineatus,E. pictus,andE.
erythroceneus). Enyalius capetinga differs from E. leechii
in having a dorsal crest and smooth fourth toe lamellae,
from E. bibronii in having keeled ventral scales, and from
E. boulengeri and E. brasiliensis in having smooth
infracarpals (Supplemental Material 2).
Descriptionofholotype.—Adult male; snout–vent
length (SVL) 68 mm; tail length (complete, not regenerated)
185 mm; axilla–groin distance 30.1 mm. Distance between
anterior edge of the auditory meatus and posterior edge of
eye 6.8 mm; external auditory meatus conspicuous, higher
(2.2 mm) than wide (1.6 mm). Head 15.7 mm in length (from
anterior border of auditory meatus to center of rostral scale),
FIG. 2.—Guided regularized random forest (GRRF) results for Enyalius morphological data. (A) Prediction error of GRRF models based on increasing
number of predictors ranked by importance, based on 50 replicates of fivefold cross-validation for each of 100 multiply imputed data sets. (B) Importance of
morphological predictors in correctly assigning individual lizards to species of Enyalius based on mean decrease in Gini accuracy of GRRF models, based on
50 replicates of fivefold cross-validation for each of 100 multiply imputed data sets. The higher the mean decrease in Gini accuracy, the higher the predictor
importance. Only the 10 most important predictors are presented. (C) Variation in number of ventral and vertebral scales, the two best predictors of
differences among species/lineages of Enyalius (to improve visualization, E. leechii is not included in the plot because it is highly distinct from other
360 Herpetologica 74(4), 2018
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
FIG. 3.—Phylogenetic relationships among Enyalius species as estimated by Bayesian inference and maximum-likelihood analyses. Posterior speciation
probabilities estimated in BPP for each species/lineage are given at right. Numbers along nodes denote posterior probabilities for the Bayesian analysis and
bootstrap scores from the maximum-likelihood analysis. A color version of this figure is available online. Scale indicates rate of base substitutions per site.
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
11.8 mm wide (at temporal region), and 9.8 mm high (at
parietal region). Snout length 7.4 mm (from posterior corner
of eye to center of rostral scale). Eye length 4.1 mm; eye
(from anterior edge) to nostril distance 5.6 mm. Forelimb
length 19.1 mm (excluding hand); hindlimb length 30 mm
(excluding foot); hand length 15.1 mm (wrist to tip of fourth
finger excluding claw); foot length 29.9 mm (ankle to tip of
fourth toe excluding claw).
Dorsal head scales irregular in size and shape (quadran-
gular or rhomboidal), juxtaposed, smooth, decreasing in size
and becoming granular in the temporal region and towards
the body. Rostral rectangular, wider (3.2 mm) than high (1.0
mm), contacting six postrostrals that separate nasal from
rostral. Nasals rounded, surrounded by 10–11 scales; nostril
occupies most of nasal, laterally oriented. Six quadrangular
internasals. Nineteen scales between rostral and occipital
scales. Interparietal smooth and rounded, surrounded by
eight smaller scales; pineal eye in the center of the scale but
not very evident. Parietals similar in size to interparietal,
slightly bulged and irregularly shaped. Circumorbital semi-
circles (10–11 scales) defined by smaller smooth scales,
separated by one big scale at the closest point. Supraoculars
in 6–7 rows, quadrangular, similar in size to frontal,
becoming smaller towards the superciliary ridge. Canthal
ridge straight, with one rounded and one enlarged scale
(quadrangular, bulged) that connects the superciliaries with
the nasal scale.
Loreals in 6–7 series, juxtaposed, smooth or slightly
bulged, mostly quadrangular. Loreals variable in size;
smaller ones (half the size of large ones) near the nasal
and supralabial areas. Eight elongated and juxtaposed
superciliaries; the three anterior ones much more elongated
FIG. 4.—Squamation of the Enyalius capetinga holotype (CHUNB 74591, adult male): (A) dorsal view of head, (B) lateral view of head, (C) ventral view of
head, and (D) cloacal region.
FIG. 5.—Photograph of the Enyalius capetinga holotype (CHUNB 74591,
adult male). Inset shows ventral aspect of specimen (photo credits: M.F.
Breitman and C.J.S. Morais). A color version of this figure is available online.
362 Herpetologica 74(4), 2018
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
(2.2 mm) than the posterior ones (0.7 mm). Preocular and
subocular enlarged (2.3 and 3.3 mm, respectively), longer
than wide (0.7 mm), keeled along their upper border. Eight
smaller quadrangular postoculars. Ocular region covered
with granular scales, except for quadrangular, juxtaposed
palpebrals. Nine supralabial scales, separated from subocular
by one row of juxtaposed scales. Temporals larger than
granular occipitals, juxtaposed, quadrangular, somewhat
bulged. Nine and eight infralabials on the right and left
sides of holotype, respectively. Lateral gulars juxtaposed,
slightly smaller than ventral gulars. Anterior auriculars
similar to adjacent posterior temporals; posterior auriculars
smaller, granular, and similar to scales on sides of neck.
External auditory meatus conspicuous, higher (2.4 mm) than
wide (1.4 mm).
Mental small, pentagonal, wider (2.2 mm) than high (1
mm). Thirty-seven gulars between mental and collar (1–27
gradually shift from rounded to slightly keeled; remaining
scales mucronated and imbricated). First row of postmentals
with two scales, followed by row of six scales; postmentals
juxtaposed, elliptical, slightly bulged until transforming into
gulars. Gulars slightly imbricated, keeled and mucronated,
similar in shape to ventrals but slightly larger towards collar
fold. Smaller granular scales hidden between pledges of
gular fold with very small granules filling the space among
them. Collar continues dorsolaterally, forming a prehumeral
Dorsal crest formed by 77 enlarged vertebral scales (~13
0.7 mm), pyramidal in the occipital region (scales 1–24) and
becoming keeled, mucronated, elongated, and slightly
imbricated at the forelimb level until the hindlimbs (scales
25–50), where they became quadrangular and slightly keeled
from there to the first 1/10th (17 mm) of the tail length.
Nuchal and lateral scales small (,0.2 mm), granular.
Paravertebral scales (counted three rows right of middor-
sum), slightly imbricated, quadrangular (~0.5 30.5 mm),
somewhat keeled; 125 paravertebral scales between occiput
and anterior insertion of hindlimb; 75 paravertebral scales
between posterior insertion of forelimb and anterior
insertion of hindlimb. Thirty-seven rows of ventrals (from
posterior insertion of forelimb to anterior insertion of
hindlimb; 53 rows from gular fold to cloaca) at least four
times larger (1.2 31.3 mm) than dorsals, flat, keeled, slightly
mucronated but rounded. One hundred six scales around
FIG. 6.—Pattern polymorphism in Enyalius capetinga sp. nov. Drawing in upper-right panel shows the four components of dorsal coloration patterns
observed (illustration credit: Mariana G. Zatz).
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
midbody. Hemipenis mostly everted; no distinguishable
femoral or preanal pores.
Suprabrachials and antebrachials larger than dorsal and
crest scales, imbricated, rhomboidal, keeled, becoming
granular near the axilla. Infrabrachial scales small, granular.
Supracarpals strongly imbricated, rhomboidal, keeled, slight-
ly mucronated. Infracarpals imbricate, elongated, rounded.
Lamellae of the hands quadrangular, strongly imbricated,
nonmucronated, not keeled. Infradigital lamellae of the left
hand in each finger (in parentheses: counts for the right hand
when different from left hand): I:8, II:12, III:18 (20), IV:19
(missing finger), V:12. Claws robust, curved, sharp, light
brown. Suprafemorals larger than dorsal and crest scales,
rhomboidal, imbricated, varying in size and shape, small and
smooth near the body and the infrafemorals, bigger and
keeled toward the knee. Infrafemorals small, granular.
Tibials rhomboidal, imbricated, keeled, slightly mucronated,
smaller and granular toward the foot. Supratarsals strongly
imbricated, rhomboidal, keeled, slightly mucronated. In-
fratarsals juxtaposed, rounded. Lamellae of the feet qua-
drangular, strongly imbricated, nonmucronated, not keeled.
Infradigital lamellae of the left feet in each finger (in
parentheses: counts for the right feet when different from
left): I:8 (missing), II:12 (13), III:20 (18), IV:26—missing a
section (30), V:15 (missing). Claws robust, curved, sharp,
light brown.
Tail complete, nonregenerated, 2.72 times longer than
SVL. Dorsal and lateral caudal scales keeled, smaller,
juxtaposed in anterior portion of tail, increasing in size and
becoming imbricated towards tail tip. Ventral caudal scales
keeled and mucronated, like caudal posterior scales.
Variation.—In Enyalius capetinga, we found sexual
dimorphism in tail length (females ¼170.40 630.31 mm,
120–211 mm; males ¼183.45 618.03 mm, 154–221 mm),
SVL (females ¼71.33 69.03 mm, 55–88 mm; males ¼
68.76 65.27 mm, 61–80 mm), and the ratio of tail length to
SVL (females ¼2.39 60.33, 1.51–2.92; males ¼2.67 6
0.15, 2.31–3.00). No other characters presented statistically
significant sexual dimorphism (Supplemental Material 2).
With the use of .200 lizards collected in and around
ılia, Zatz (2002) identified four components of coloration
patterns in E. capetinga: (1) dorsal lozenges along vertebral
line (reaching tip of tail; present in 86.5% of individuals); (2)
yellow lateral circles (varying in size and number; 73.4%); (3)
lateral bars (complete or punctuated, perpendicular to body
axis, between paravertebral and ventral region; 45.7%); and
(4) white paravertebral stripes (35.2%; Fig. 6). The more
common coloration combinations were: lozenges þcircles
(34.1%), followed by lozenges þbars þcircles (24.7%; as in
the holotype), and lozenges þstripes þbars (14.2%).
Although no exclusive patterns were associated with sex,
the patterns of lozenges þbars þstripes and stripes were
found in females more than twice as often as in males; and
the patterns lozenges þbars þcircles, lozenges þcircles,
lozenges þstripes, and circles were more than twice as
common in males than in females. Coloration patterns are
not associated with sexual maturity, climate seasonality, or
sex. Immature individuals are statistically more likely to
present lateral bars than mature individuals. Tail coloration
presents lozenges and perpendicular bars in all individuals
(Zatz 2002).
FIG. 7.—Satellite view of the type locality for Enyalius capetinga sp. nov. (dotted pin symbol), within a gallery forest of the Cerrado biome. A color version
of this figure is available on line.
364 Herpetologica 74(4), 2018
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
Coloration of holotype in life.—Dorsal background
gray; ventral background white (Fig. 5). Dorsum of head
brownish (relative to the laterals and rest of body), with
black, thin, uneven, transversal lines; small scattered dark
spots on remaining gray scales. Two anterior lines between
rostral and superciliaries, less defined and thinner than three
posterior ones that reach superciliaries. Temporal region
with smaller black line. Sides of head change gradually from
dorsal coloration to ventral coloration. Two black lines (like
posterior lines of dorsal head) between eye and last
supralabial and anterior auricular scales. Suboculars with
black spots. Ventral head coloration white, with some
scattered black spots.
Coloration of temporal area like anterior region; black,
thin, uneven lines form V-like shape that transitions into
larger, darker, thicker lines (almost twice as large as lines on
head), surrounded by white scales defining eight transversal
lozenges along the body. Dorsolateral parts of body (between
neck and forelimbs) gray, with several pairs of thinner
punctuated lines continuing, at the uppermost end, to the
dorsal lozenges and fading away at the bottom, where
coloration shifts to ventral white. Sides of body with 8–9
conspicuous, bright, yellow, large (each including 6–30
scales) spots, concentrated in the anterior region, and several
white/yellowish smaller (1–2 scales) spots in remaining area.
Venter white, almost immaculate, with black spots on a few
Tail dorsally gray, with diamondlike black blotches that
along the first third of the tail transforming into grayish
marks that dissolve into a gray/brownish tail. Limbs dorsally
variegated, gray, with some black marks that form lines.
Ventrally, limbs and tail white with some grayish spots.
Coloration of holotype in preservative.—After 6 mo in
preservative, coloration turns dark gray (darker than in life).
Even though all brownish and yellow colors turn gray, dorsal
lozenges on the body, limbs, and tail remain noticeable.
Dorsolateral coloration homogeneously gray with black lines;
yellow and white spots faded and undistinguishable. Venter
white; ventral parts of limbs and tail darker gray/brownish.
Etymology.—The specific epithet of this species is a
Portuguese word that refers to Capetinga Creek, located in
Fazenda ´
Agua Limpa, property of Universidade de Bras´
~7 km SW of the type locality, where GRC collected the first
specimens of Enyalius capetinga in November 1987.
Capetinga is formed by the combination of the Tupinamba
language (Tupi-Guarani) words ‘‘kapi’’’ (grass) and ‘‘tinga’’
(dry/white), meaning ‘‘dry grass.’’ We also chose this name
for the resemblance with ‘‘capeta,’’ a Portuguese word for
‘‘devil’’ or ‘‘ bugbear.’’ Sometimes, while conducting our
investigation and dealing with the complex interactions of
academic life, we felt as if the capeta was coming to get us!
Distribution.Enyalius capetinga is endemic to the
Brazilian Cerrado, and is known from Bras´
ılia, Distrito
Federal and nearby localities in Minas Gerais (Paracatu and
ı; Fig. 1).
Barcoding of holotype.—BOLD records (sequence and
specimen metadata) can be accessed at http://www.
Natural history.Enyalius capetinga inhabits primarily
gallery forests, but some individuals were collected in nearby
areas of cerrado sensu stricto and cerrada
˜o. Cerrado sensu
TABLE 2.—Values of the most informative diagnostic characters among species of Enyalius. Values for meristic variables are presented as means 61 SD followed by range in parentheses. The character
‘‘Dorsal scales shape’’ is qualitative and values are presented as a ratio of frequencies, with relative frequencies listed in parentheses after absolute frequencies. Ranges of samples sizes are listed in
parentheses after species names. Studied variables, data, and summary statistics of all variables are presented in the Appendix and in Supplemental Material 2.
Ventral scales Supraciliaries Paravertebral scales Gular scales Supralabials Dorsolateral tibials Dorsal scales shape
E. bibronii (22–23) 48.64 64.6 (43–65) 10.09 60.95 (8–12) 83.14 66.68 (73–98) 39.04 64.17 (32–49) 10.43 60.79 (9–12) 12.35 61.07 (10–15) 4 (0.17):13 (0.57):6 (0.26)
E. bilineatus (25) 43.2 64.06 (35–49) 7.48 62.31 (5–16) 62.04 66.82 (51–73) 36.76 63.99 (29–44) 7.68 60.85 (6–9) 12.52 60.59 (12–14) 0:0:25 (1)
E. boulengeri (7) 43.57 64.96 (35–49) 11 61.53 (10–14) 89.14 610.92 (75–108) 50.14 63.58 (44–56) 10.29 60.76 (9–11) 12.43 61.27 (11–15) 0:7 (1):0
E. brasiliensis (10–11) 45.1 64.43 (38–53) 10.6 60.84 (9–12) 90.18 67.59 (75–98) 50.18 63.43 (46–56) 10.09 60.7 (9–11) 13.64 61.29 (12–16) 0:11 (1):0
E. catenatus 2 (5) 43.2 66.42 (38–54) 9.2 61.79 (8–12) 87.6 63.21 (83–91) 41.8 63.7 (39–48) 10.4 61.52 (8–12) 15.6 62.51 (14–20) 0:5 (1):0
E. catenatus (11) 44.45 63.93 (36–50) 11.64 61.03 (10–13) 88.27 67.67 (78–102) 46.64 64.03 (42–55) 11.18 61.6 (8–14) 17.18 62.6 (14–20) 0:11 (1):0
E. erythroceneus (6) 44.33 61.75 (42–47) 8.67 61.51 (7–11) 79.17 63.82 (75–84) 38.33 63.33 (33–42) 9.5 61.22 (8–11) 11.67 61.37 (10–13) 0:6 (1):0
E. iheringii (22–23) 50.23 63.85 (42–60) 13.7 61.99 (10–17) 75.3 68.34 (59–102) 44.09 62.66 (39–49) 11.09 61.12 (8–13) 13.04 61.22 (11–15) 7 (0.3):11 (0.48):5 (0.22)
E. leechii (22) 39.45 62.7 (35–44) 11.23 61.31 (8–13) 84.73 65.74 (74–94) 55.05 65.13 (47–67) 11.18 60.73 (10–12) 17 63.31 (12–22) 0:17 (0.77):5 (0.23)
E. perditus (18) 55.22 63.08 (51–62) 10.33 61.14 (8–12) 101 66.18 (90–114) 51.44 63.33 (46–58) 9.67 61.08 (8–12) 13 61.5 (11–18) 0:18 (1):0
E. perditus 2 (15) 54.47 64.21 (45–60) 10.4 61.18 (8–12) 93.73 612.19 (72–114) 48.6 64.22 (43–56) 9.6 61.12 (9–13) 12.73 61.03 (12–15) 0:14 (0.93):1 (0.07)
E. pictus (6) 46.33 65.43 (39–55) 10.33 61.37 (8–12) 92.17 63.97 (85–97) 46.5 63.94 (40–52) 11.33 61.03 (10–13) 18.17 63.6 (13–23) 0:5 (0.83):1 (0.17)
E. capetinga (56) 34.45 62.25 (29–40) 6.59 61.64 (4–14) 63.84 65.96 (51–77) 34.34 64.04 (28–47) 7.86 61.03 (6–12) 10.38 61.29 (8–13) 0:2 (0.04):54 (0.96)
States of the characters are listed in the following order: mostly smooth, mostly granular, and mostly keeled.
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
stricto is the most common vegetation physiognomy of
Cerrado, characterized by the dominance of trees and
shrubs, 3–8 m tall, with more than 30% of canopy cover, with
a fair amount of herbaceous vegetation among them. The
˜o is denser, with 30–90% of canopy cover, dominated
by 8–12 m tall trees and very reduced herbaceous vegetation
(Oliveira-Filho and Ratter 2002). The new species is diurnal,
oviparous, ombrophilous, and feeds primarily on arthropods
(Zatz 2002). Females reach sexual maturity at a larger size
than males (females ¼71 mm SVL, males ¼57.82 mm SVL)
and breeding occurs from September to February (Zatz
2002). The distributional range of the new species is
characterized by the Aw climate in K ¨
oppen’s classification,
with marked rainfall seasonality: a dry, cold season from May
to September, and a wet, hot season from October to April
(Alvares et al. 2013). The annual mean temperature in
ılia is 20.68C (16.1–26.68C) and the annual mean
precipitation is 1540.6 mm, of which 1413 mm (91.7%)
occurs in the wet season (Ramos et al. 2009).
Accurate species descriptions are necessary when
addressing the ongoing biodiversity crisis, and for
supporting other disciplines that rely on accurate taxon-
omy (Wilson 1985; Wheeler 2004; Drew 2011). We set out
to test the hypothesis that populations of Enyalius lizards
from central Brazil identified by previous molecular and
ecological studies (Barreto-Lima 2012; Rodrigues et al.
2014; Vargas et al. 2015) constituted a new species, by
generating and analyzing morphological data (100% novel;
55 variables for 259 individuals) and molecular data
(~20% novel; three genes for ~75 individuals) for the
genus. Our approach—combining semisupervised Gauss-
ian mixture modeling followed by machine-learning
classification—required a greater investment of time than
alpha taxonomy methods, but minimized the problem of
overconfidence in museum identifications and produced
more detailed species hypotheses (Sangster and Luksen-
burg 2015). Our species delimitation results supported the
recognition of E. brasiliensis and E. boulengeri,establish-
ing that E. perditus 2andE. catenatus 2 (sensu Rodrigues
etal.2014),andE. capetinga are independently evolving
lineages. Although GRRF analysis indicated that samples
of E. catenatus 2andE. perditus 2werealsomorpholog-
ically different from other congeners; these species await
formal description.
Enyalius capetinga is endemic to the Cerrado and an
inhabitant of gallery forests. The Cerrado plateaus harbor
the headwaters of major South American rivers; therefore,
gallery forests form an extensive network through which
many Amazonian and Atlantic Forest species penetrate
deeply within Cerrado (Redford and Fonseca 1986; Oliveira-
Filho and Ratter 1995; Silva 1996; Costa 2003). In general,
gallery forests of the central and southern Cerrado have
stronger affinities with the Atlantic Forest than with
Amazonia, presumably because of their shared higher
elevations and lower temperatures (Oliveira-Filho and
Ratter 1995; Silva 1996). Gallery forests also provide routes
for biotic exchange between Amazonian and Atlantic Forest
during climatic fluctuations (Ledo and Colli 2017). Consid-
ering that an Atlantic Forest distribution is ancestral for
Enyalius (Rodrigues et al. 2014), E. capetinga probably
diverged from an Atlantic Forest ancestor that colonized
Cerrado gallery forests and underwent geographical or
ecological speciation.
The divergence between E. capetinga and its closest
relative, E. bilineatus, is estimated to have occurred in the
late Miocene at ~6.86 mya (confidence interval ¼5.09–
8.69; Rodrigues et al. 2014). Following a warm climate
optimum in the middle Miocene, the late Miocene was
characterized by global cooling and aridity driven by
Andean orogeny, with pronounced effects in South America
(Zachos et al. 2001, 2008; Armijo et al. 2015). This was
accompanied by replacement of forests by open habitats in
South America (Hynek et al. 2012; Le Roux 2012; Palazzesi
and Barreda 2012; Pound et al. 2012; Nie et al. 2016).
Gallery forests can exhibit similar dynamics to upland
tropical forests during climate fluctuations, and their
borders can retreat or advance over adjacent savannas
(Silva et al. 2008). Therefore, under unfavorable conditions,
gallery forests might shrink or even become fragmented,
with profound effects on their environmental conditions
caused by climate, soil, vegetation, and fire feedbacks
(Hoffmann et al. 2002; Beckage et al. 2009; Staver and
Levin 2012; MacDermott et al. 2017). We hypothesize that
this resulted in the differentiation of E. capetinga from its
ancestral clade through range fragmentation and/or a
steepened ecological gradient between the Atlantic Forest
and Cerrado gallery forests.
Enyalius capetinga has the smallest SVL of any species of
Enyalius, and this could reflect a correlation between
morphology and the use of more open vegetation or
ground-level habitats relative to its congeners, as observed
among Greater Antillean anoles (Losos 2009). Given the
availability of a recent phylogenetic hypothesis (Rodrigues et
al. 2014), future studies should explore the effects of habitat
shifts on the evolution of morphological and ecological traits
(e.g., behavioral, reproductive, and thermal ecology) among
species of Enyalius.
The known populations of Enyalius capetinga are all
under some level of threat. Even though the Distrito Federal
populations occur inside protected areas, they are facing
increased pressures including the complete isolation of these
areas because of the expansion of urban areas, which leads to
increasing levels of ecological disturbance by fires, pollution,
local climate change, logging, and introduced species
(UNESCO 2002; Fran¸coso et al. 2015). Outside of protected
areas (e.g., populations in Paracatu and Una´
ı), these
conditions are exacerbated by the rapid conversion of
natural Cerrado habitats into crops and pastureland (Jepson
et al. 2010; Sano et al. 2010), and by recent changes in the
Brazilian Forest Code that reduce the protection of gallery
forests (Ledo and Colli 2016). The future seems bleak for E.
capetinga and many other Cerrado endemics that depend on
gallery forests; thus, the need to understand the biology and
evolutionary history of these species cannot be understated,
in order to provide improved taxonomic and ecological
foundations for conservation.
Acknowledgments.—We thank A.S. Arruda Camara Cabral and N.
Cazzaniga for discussion and references clarifying the derivation of the word
‘‘capetinga,’’ and L.J. ´
Avila for help with scale nomenclature. We thank J.
Lozier for providing office space during an internship at the University of
Alabama. We thank the following collections and their curators for loaning
366 Herpetologica 74(4), 2018
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
specimens: MZUSP (Museu de Zoologia da Universidade de Sa
˜o Paulo),
MCP (Museu de Ciˆ
encias e Tecnologia da Pontif´
ıcia Universidade Cat´
do Rio Grande do Sul), ZUEC (Museu de Zoologia da Universidade de
Campinas), MNRJ (Museu Nacional da Universidade Federal do Rio de
Janeiro), MPEG (Museu Paraense Em´
ılio Goeldi), MZUFV (Museu de
Zoologia Joa
˜o Moojen da Universidade Federal de Vi¸cosa), and MCNR
˜o de Herpetologia do Museu de Ciˆ
encias Naturais da Pontif´
Universidade Cat´
olica de Minas Gerais). This research was supported by the
following postdoctoral fellowships: Programa Nacional de P ´
CAPES (MFB), Partnerships for Enhanced Engagement in Research
encia Sem Fronteiras Young Talent
Fellow award (JCB), and a PROCAD award (HW). GRC thanks CAPES,
CNPq, Funda¸ca
˜o de Apoio `
a Pesquisa do Distrito Federal (FAPDF), and
the USAID’s PEER program under cooperative agreement AID-OAA-A-11-
00012 for financial support.
Supplemental material associated with this article can be
found online at
Aberer, A.J., K. Kobert, and A. Stamatakis. 2014. ExaBayes: Massively
parallel Bayesian tree inference for the whole-genome era. Molecular
Biology and Evolution 31:2553–2556.
Agnarsson, I., and M. Kuntner. 2007. Taxonomy in a changing world:
Seeking solutions for a science in crisis. Systematic Biology 56:531–539.
Alvares, C.A., J.L. Stape, P.C. Sentelhas, J.L.D. Goncalves, and G. Sparovek.
2013. K¨
oppen’s climate classification map for Brazil. Meteorologische
Zeitschrift 22:711–728.
Armijo, R., R. Lacassin, A. Coudurier-Curveur, and D. Carrizo. 2015.
Coupled tectonic evolution of Andean orogeny and global climate. Earth-
Science Reviews 143:1–35.
Bagley, J.C. 2017. justincbagley/GaussClust: GaussClust, Version 0.1.0 [Data
Set]. Zenodo. Available via Archived
by WebCite at on 5 July 2017.
Barreto-Lima, A.F. 2012. Distribui¸ca
˜o, Nicho Potencial e Ecologia
ogica do Gˆ
enero Enyalius (Squamata, Leiosauridae): Testes de
oteses para Lagartos de Florestas Continentais Brasileiras. Ph.D.
dissertation, Universidade Federal do Rio Grande do Sul, Brazil.
Barreto-Lima, A.F., and B.M. Sousa. 2011. Feeding ecology and sexual
dimorphism of Enyalius perditus in an Atlantic forest, Brazil. Herpeto-
logical Bulletin 118:1–9.
Beckage, B., W.J. Platt, and L.J. Gross. 2009. Vegetation, fire, and
feedbacks: A disturbance-mediated model of savannas. American
Naturalist 174:805–818.
Biecek, P., E. Szczurek, M. Vingron, and J. Tiuryn. 2012. The R package
bgmm: Mixture modeling with uncertain knowledge. Journal of Statistical
Software 47:1–31.
Bokermann, W.C.A. 1957. Atualiza¸ca
˜o do itinerario da viagem do Principe de
Wied ao Brasil (1815–1817). Arquivos de Zoologia (Sa
˜o Paulo) 10:209–251.
Boulenger, G.A. 1885. A list of reptiles and batrachians from the province
Rio Grande do Sul, Brazil, sent to the Natural History Museum by Dr. H.
von Ihering. The Annals and Magazine of Natural History 15:191–196.
Breiman, L. 2001. Random forests. Machine Learning 45:5–32.
Breiman, L., and P. Spector. 1992. Submodel selection and evaluation in
regression. The X-random case. International Statistical Review 60:291–
Breitman, M.F., L.J. Avila, J.W. Sites, Jr., and M. Morando. 2012. How
lizards survived blizzards: Phylogeography of the Liolaemus lineomacu-
latus group (Liolaemidae) reveals multiple breaks and refugia in southern
Patagonia and their concordance with other codistributed taxa. Molecular
Ecology 21:6068–6085.
Costa, H.C., and R.S. B´
ernils. 2014. R´
epteis brasileiros: Lista de esp ´
2015. Herpetologia Brasileira 4:75–93.
Costa, L.P. 2003. The historical bridge between the Amazon and the Atlantic
Forest of Brazil: A study of molecular phylogeography with small
mammals. Journal of Biogeography 30:71–86.
Cutler, D.R., T.C. Edwards, K.H. Beard, A. Cutler, and K.T. Hess. 2007.
Random forests for classification in ecology. Ecology 88:2783–2792.
Dayrat, B. 2005. Towards integrative taxonomy. Biological Journal of the
Linnean Society 85:407–415.
de Queiroz, K. 2007. Species concepts and species delimitation. Systematic
Biology 56:879–886.
Deng, H. 2013. Guided random forest in the RRF package. arXiv
Domingos, F.M.C.B., R.J. Bosque, J. Cassimiro, G.R. Colli, M.T. Rodrigues,
M.G. Santos, and L.B. Beheregaray. 2014. Out of the deep: Cryptic
speciation in a Neotropical gecko (Squamata, Phyllodactylidae) revealed
by species delimitation methods. Molecular Phylogenetics and Evolution
Drew, L.W. 2011. Are we losing the science of taxonomy? BioScience
eril, A.M.C., and G. Bibron. 1837. Erp ´
etologie G´
erale ou Histoire
Naturelle Compl`
ete des Reptiles. Tome Quatri `
eme. Librairie Ency-
edique de Roret, France.
Etheridge, R. 1969. A review of the iguanid lizard genus Enyalius. Bulletin
of the British Museum (Natural History) 18:231–260.
Felsenstein, J., and H. Kishino. 1993. Is there something wrong with the
bootstrap on phylogenies? A reply to Hillis and Bull. Systematic Biology
Fran¸coso, R.D., R. Branda
˜o, C.C. Nogueira, Y.B. Salmona, R.B. Machado,
and G.R. Colli. 2015. Habitat loss and the effectiveness of protected areas
in the Cerrado Biodiversity Hotspot. Natureza & Conserva¸ca
˜o 13:35–40.
Frost, D.R., R. Etheridge, D. Janies, and T.A. Titus. 2001. Total evidence,
sequence alignment, evolution of polychrotid lizards, and a reclassifica-
tion of the Iguania (Squamata: Iguania). American Museum Novitates
Hillis, D.M., and J.J. Bull. 1993. An empirical test of bootstrapping as a
method for assessing confidence in phylogenetic analysis. Systematic
Biology 42:182–192.
Hoffmann, W.A., W. Schroeder, and R.B. Jackson. 2002. Positive feedbacks
of fire, climate, and vegetation and the conversion of tropical savanna.
Geophysical Research Letters 29:4.
Houle, D. 1992. Comparing evolvability and variability of quantitative traits.
Genetics 130:195–204.
Huelsenbeck, J.P., and F. Ronquist. 2001. MRBAYES: Bayesian inference of
phylogenetic trees. Bioinformatics 17:754–755.
Hynek, S.A., B.H. Passey, J.L. Prado, F.H. Brown, T.E. Cerling, and J.
Quade. 2012. Small mammal carbon isotope ecology across the Miocene–
Pliocene boundary, northwestern Argentina. Earth and Planetary Science
Letters 321:177–188.
Jackson, J.F. 1978. Differentiation in the genera Enyalius and Strobilurus
(Iguanidae): Implications for Pleistocene climatic changes in eastern
Brazil. Arquivos de Zoologia (Sa
˜o Paulo) 30:1–79.
Jepson, W., C. Brannstrom, and A. Filippi. 2010. Access regimes and
regional land change in the Brazilian Cerrado, 1972–2002. Annals of the
Association of American Geographers 100:87–111.
Kearse, M., R. Moir, A. Wilson . . . A. Drummond. 2012. Geneious Basic: An
integrated and extendable desktop software platform for the organization
and analysis of sequence data. Bioinformatics 28:1647–1649.
Kohavi, R. 1995. A study of cross-validation and bootstrap for accuracy
estimation and model selection. Pp. 1137–1143 in Proceedings of the
14th International Joint Conference on Artificial Intelligence, Volume 2
(IJCAI95, ed.). Morgan Kaufmann Publishers, Inc., Canada.
Lanfear, R., B. Calcott, S.Y.W. Ho, and S. Guindon. 2012. PartitionFinder:
Combined selection of partitioning schemes and substitution models for
phylogenetic analyses. Molecular Biology and Evolution 29:1695–1701.
Lanfear, R., B. Calcott, D. Kainer, C. Mayer, and A. Stamatakis. 2014.
Selecting optimal partitioning schemes for phylogenomic datasets. BMC
Evolutionary Biology 14:14.
Lanfear, R., P.B. Frandsen, A.M. Wright, T. Senfeld, and B. Calcott. 2017.
PartitionFinder 2: New methods for selecting partitioned models of
evolution for molecular and morphological phylogenetic analyses.
Molecular Biology and Evolution 34:772–773.
Le Roux, J.P. 2012. A review of Tertiary climate changes in southern South
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
America and the Antarctic Peninsula. Part 2: Continental conditions.
Sedimentary Geology 247:21–38.
Ledo, R.M.D., and G.R. Colli. 2016. Silent death: The new Brazilian Forest
Code does not protect lizard assemblages in Cerrado riparian forests.
South American Journal of Herpetology 11:98–109.
Ledo, R.M.D., and G.R. Colli. 2017. The historical connections between the
Amazon and the Atlantic Forest revisited. Journal of Biogeography
Lesson, R.P. 1830. Description de quelques reptiles nouveaux ou peu
connus. Pp. 1–65 in Voyage Autour du Monde, Ex ´
e par Ordre du
Roi, sur la Corvette de Sa Majest ´
e, La Coquille, Pendant les Ann´
1822, 1823, 1824 et 1825. Zoologie, Tome Second. (M.L.I. Duperrey,
ed.). Arthur Bertrand, France.
Losos, J.B. 2009. Lizards in an Evolutionary Tree: Ecology and Adaptive
Radiation of Anoles. University of California Press, USA.
MacDermott, H.J., R.J. Fensham, Q. Hua, and D. Bowman. 2017.
Vegetation, fire, and soil feedbacks of dynamic boundaries between
rainforest, savanna and grassland. Austral Ecology 42:154–164.
Maia-Carneiro, T., T. Motta-Tavares, D. Vrcibradic, M.C. Kiefer, T.A.
Dorigo, C.F.D. Rocha, and M. Van Sluys. 2016. Feeding habits of
Enyalius perditus (Squamata: Leiosauridae) in an Atlantic Forest
remnant in southeastern Brazil. Phyllomedusa 15:21–27.
Mingers, J. 1989. An empirical comparison of selection measures for
decision-tree induction. Machine Learning 3:319–342.
Mirarab, S., N. Nguyen, S. Guo, L.S. Wang, J. Kim, and T. Warnow. 2015.
PASTA: Ultra-large multiple sequence alignment for nucleotide and
amino-acid sequences. Journal of Computational Biology 22:377–386.
Moraes, P.L.R. 2009. The Brazilian herbarium of Maximilian, Prince of
Wied. Neodiversity 4:16–51.
Moraes, P.L.R. 2011. Notes and lectotypification of names based on Brazilian
collections of Prince Maximilian of Wied. Kew Bulletin 66:493–503.
Murphy, J.C., M.J. Jowers, R.M. Lehtinen, S.P. Charles, G.R. Colli, A.K.
Peres, C.R. Hendry, and R.A. Pyron. 2016. Cryptic, sympatric diversity in
tegu lizards of the Tupinambis teguixin group (Squamata, Sauria,
Teiidae) and the description of three new species. PloS One
11:e0158542. DOI:
Nagy, Z.T., G. Sonet, F. Glaw, and M. Vences. 2012. First large-scale DNA
barcoding assessment of reptiles in the biodiversity hotspot of
Madagascar, based on newly designed COI primers. PloS One
7:e34506. DOI:
Nie, Z.L., V.A. Funk, Y. Meng, T. Deng, H. Sun, and J. Wen. 2016. Recent
assembly of the global herbaceous flora: Evidence from the paper daisies
(Asteraceae: Gnaphalieae). New Phytologist 209:1795–1806.
Nogueira, C., G.R. Colli, and M. Martins. 2009. Local richness and
distribution of the lizard fauna in natural habitat mosaics of the Brazilian
Cerrado. Austral Ecology 34:83–96.
Oliveira-Filho, A.T., and J.A. Ratter. 1995. A study of the origin of central
Brazilian forests by the analysis of plant species distribution patterns.
Edinburgh Journal of Botany 52:141–194.
Oliveira-Filho, A.T., and J.A. Ratter. 2002. Vegetation physiognomies and
woody flora of the Cerrado biome. Pp. 91–120 in The Cerrados of Brazil:
Ecology and Natural History of a Neotropical Savanna (P.S. Oliveira and
R.J. Marquis, eds.). Columbia University Press, USA.
Padial, J.M., A. Miralles, I. De la Riva, and M. Vences. 2010. The integrative
future of taxonomy. Frontiers in Zoology 7:16.
Palazzesi, L., and V. Barreda. 2012. Fossil pollen records reveal a late rise of
open-habitat ecosystems in Patagonia. Nature Communications 3:1294.
Pante, E., C. Schoelinck, and N. Puillandre. 2015. From integrative
taxonomy to species description: One step beyond. Systematic Biology
Pound, M.J., A.M. Haywood, U. Salzmann, and J.B. Riding. 2012. Global
vegetation dynamics and latitudinal temperature gradients during the
mid to late Miocene (15.97–5.33 Ma). Earth-Science Reviews 112:1–22.
Quinn, G.P., and M.J. Keough. 2002. Experimental Design and Data
Analysis for Biologists. Cambridge University Press, UK.
R Core Team. 2017. R: A Language and Environment for Statistical
Computing, Version v3.3.1. Available at R
Foundation for Statistical Computing, Austria.
Ramos, A.M., L.A.R. dos Santos, and L.T.G. Fortes. 2009. Normais
ogicas do Brasil 1961–1990. INMET, Brazil.
Ratnasingham, S., and P.D.N. Hebert. 2007. BOLD: The Barcode of Life
Data System. Molecular Ecology Notes 7:355–364.
Rautenberg, R., and R.R. Laps. 2010. Natural history of the lizard Enyalius
iheringii (Squamata, Leiosauridae) in southern Brazilian Atlantic forest.
Iheringia S´
erie Zoologia 100:287–290.
Redford, K.H., and G.A.B. Fonseca. 1986. The role of gallery forests in the
zoogeography of the Cerrado non-volant mammalian fauna. Biotropica
Rodrigues, M.T., M.A. Freitas, T.F.S. Silva, and C.E.V. Bertolotto. 2006. A
new species of lizard genus Enyalius (Squamata, Leiosauridae) from the
highlands of Chapada Diamantina, state of Bahia, Brazil, with a key to
species. Phyllomedusa 5:11–24.
Rodrigues, M.T., C.E.V. Bertolotto, R.C. Amaro, Y. Yonenaga-Yassuda,
E.M.X. Freire, and K.C.M. Pellegrino. 2014. Molecular phylogeny,
species limits, and biogeography of the Brazilian endemic lizard genus
Enyalius (Squamata: Leiosauridae): An example of the historical
relationship between Atlantic Forests and Amazonia. Molecular Phylo-
genetics and Evolution 81:137–146.
Sangster, G., and J.A. Luksenburg. 2015. Declining rates of species
described per taxonomist: Slowdown of progress or a side-effect of
improved quality in taxonomy? Systematic Biology 64:144–151.
Sano, E.E., R. Rosa, J.L.S. Brito, and L.G. Ferreira. 2010. Land cover
mapping of the tropical savanna region in Brazil. Environmental
Monitoring and Assessment 166:113–124.
Schinz, H.R. 1822. Das Thierreich eingetheilt nach dem Bau der Thiere: als
Grundlage ihrer Naturgeschichte und der vergleichenden Anatomie.
Zweiter Band. Reptilien, Fische, Weichthiere, Ringelw ¨urmer. J.G.
Cotta’schen Buchhandlung, Germany.
Silva, J.M.C. 1996. Distribution of Amazonian and Atlantic birds in gallery
forests of the Cerrado region, South America. Ornitologia Neotropical
Silva, L.C.R., L. Sternberg, M. Haridasan, W.A. Hoffmann, F. Miralles-
Wilhelm, and A.C. Franco. 2008. Expansion of gallery forests into central
Brazilian savannas. Global Change Biology 14:2108–2118.
Smith, H.M. 1946. Handbook of Lizards: Lizards of the United States and of
Canada. Comstock Publishing Company, USA.
Stamatakis, A. 2014. RAxML version 8: A tool for phylogenetic analysis and
post-analysis of large phylogenies. Bioinformatics 30:1312–1313.
Staver, A.C., and S.A. Levin. 2012. Integrating theoretical climate and fire
effects on savanna and forest systems. American Naturalist 180:211–224.
Sturaro, M.J., and V.X. da Silva. 2010. Natural history of the lizard Enyalius
perditus (Squamata: Leiosauridae) from an Atlantic forest remnant in
southeastern Brazil. Journal of Natural History 44:1225–1238.
UNESCO (United Nations Educational, Scientific and Cultural Organiza-
tion). 2002. Vegeta¸ca
˜o do Distrito Federal: Tempo e Espa¸co. UNESCO,
van Buuren, S., and K. Groothuis-Oudshoorn. 2011. mice: Multivariate
imputation by chained equations in R. Journal of Statistical Software
Vargas, S.M., M.d.O. Santos, I.A. Novelli, . . . B.M. de Sousa. 2015. Genetic
diversity and structure of two species of Enyalius (Squamata: Leiosaur-
idae) from Neotropical biodiversity hotspots. Phyllomedusa 14:99–111.
Vitt, L.J., T.C.S. Avila-Pires, and P.A. Zani. 1996. Observations on the
ecology of the rare Amazonian lizard, Enyalius leechii (Polychrotidae).
Herpetological Natural History 4:77–82.
Wagler, J. 1830. Nat ¨urliches System der Amphibien, mit vorangehender
Classification der S¨
augethiere und V¨
ogel. Ein Beitrag zur vergleichenden
Zoologie. J.G. Cotta’schen Buchhandlung, Germany.
Weston, J., C. Leslie, E. Ie, D.Y. Zhou, A. Elisseeff, and W.S. Noble. 2005.
Semi-supervised protein classification using cluster kernels. Bioinfor-
matics 21:3241–3247.
Wheeler, Q.D. 2004. Taxonomic triage and the poverty of phylogeny.
Philosophical Transactions of the Royal Society of London Series B—
Biological Sciences 359:571–583.
Wied-Neuwied, M. 1821. Reise nach Brasilien in den Jahren 1815 bis 1817.
Zweiter Band. Heinrich Ludwig Br¨
onner, Germany.
Wilson, E.O. 1985. The biological diversity crisis. BioScience 35:700–706.
Yang, Z. 2015. The BPP program for species tree estimation and species
delimitation. Current Zoology 61:854–865.
Yang, Z., and B. Rannala. 2010. Bayesian species delimitation using
multilocus sequence data. Proceedings of the National Academy of
Sciences 107:9264–9269.
Yang, Z., and B. Rannala. 2014. Unguided species delimitation using DNA
sequence data from multiple loci. Molecular Biology and Evolution
Zachos, J., M. Pagani, L. Sloan, E. Thomas, and K. Billups. 2001. Trends,
rhythms, and aberrations in global climate 65 Ma to present. Science
Zachos, J.C., G.R. Dickens, and R.E. Zeebe. 2008. An early Cenozoic
368 Herpetologica 74(4), 2018
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
perspective on greenhouse warming and carbon-cycle dynamics. Nature
Zatz, M.G. 2002. O Polimorfismo Croma
´tico e sua Manuten¸ca
sp. (Squamata: Leiosauridae) no Cerrado do Brasil Central. Master’s
thesis, Universidade de Bras´
ılia, Brazil.
Accepted on 8 August 2018
Associate Editor: Matthew Fujita
List of the morphological characters used in comparisons among species
of Enyalius, including their acronyms in parentheses. Where necessary, we
list clarifications and different states of the character in question. Asterisks
indicate characters and states that were monomorphic for our Enyalius
Qualitative Characters
1. Canthus rostralis (CRo), shape: (0) incomplete, (1) complete.
2. Nasal (N), shape: (0) asymmetric, *(1) symmetric.
3. Supraoculars (SpO), form: *(0) smooth, (1) conic or pyramidal, (2)
4. Auriculars (A), pattern: (0) homogeneous, (1) heterogeneous.
5. Intercanthals (IntC), form of scales between nostrils and anterior region
of the eyes: *(0) smooth, (1) pyramidal, (2) keeled.
6. Canthal ridge scales (CRiS), form: *(0) smooth, (1) not smooth.
7. Supracarpals (SpC), form: (0) smooth, (1) keeled.
8. Infracarpals (InC), form: (0) smooth, (1) keeled.
9. Infrabrachials (InB), form: (0) granular, (1) keeled, (2) mixed.
10. Suprafemorals (SpF), form: *(0) smooth, (1) keeled, (2) predominantly
11. Infrafemorals (InF), form: (0) smooth, (1) keeled.
12. Infratibials (InT), form: *(0) smooth, (1) keeled.
13. Dorsal crest scales (DCS), form: (0) pyramidal, (1) spinelike shape, (2)
keeled, (3) conic.
14. Tail crest scales (TCS), from the base until 1/3 of tail length: *(0) absent,
(1) present.
15. Paravertebral scales (PVS), form: (0) pyramidal keeled, (1) elongated
keeled, (2) mixed keeled, (3) conic not keeled, (4) elongated not keeled.
16. Dorsal scales (DS), form of dorsal scales excluding crest scales form: (0)
mostly smooth, (1) mostly granular, (2) mostly keeled.
17. Lateral scales (LS), form: (0) granular, (1) conic or pyramidal, (2)
18. Ventral scales (VnS), form: (0) smooth, (1) keeled.
19. Nasal/postrostral contact (NPRC): (0) absent, (1) present.
20. Subocular (SbO), size: (0) enlarged, (1) small.
21. Gular fold (GF), defined by smaller scales inside fold: (0) absent, (1)
22. External gular fold scales (EGFS), form: (0) smooth, (1) keeled.
23. Dorsal head scales (DHS), form: (0) smooth, (1) conic, (2) keeled.
24. Mental (M), form: (0) flat, (1) triangular.
25. Supratarsal keels (SpTK), type of keels: *(0) unicarinated, (1) multi-
26. Fourth toe lamellae (4TL), form: (0) smooth, (1) keeled.
27. Fourth finger lamellae (4FL), form: (0) smooth, (1) keeled.
28. Dorsal crest (DC): (0) absent, (1) present.
Meristic Characters
29. Paravertebral scales (nPVS), number of scales between posterior edge of
forelimb and anterior edge of hindlimb insertions.
30. Midbody scales (nMS), one half the number of scales around the
31. Vertebral scales (nVrS), number of scales between interparietal scale
and posterior edge of hindlimb insertions.
32. Ventral scales (nVnS), number of scales between posterior edge of
forelimb and anterior edge of hindlimb insertions.
33. Infralabials (nInL), number.
34. Postmentals (nPM), number.
35. Contact postmentals (ncPM), number of scales in contact with
postmental scales.
36. Gular scales (nGS), number of scales between mental scale and gular
37. Canthus rostralis (nCRo), number of scales located in the canthus
38. Between nasal and supralabials (nbNSpL), minimum number of scales
between nasal and supralabial scales.
39. Loreal series (nLS), minimum count.
40. Contact interparietal (ncIp), number of scales in contact with
41. Between circumorbital arcs (nbCoA), minimum number of scales.
42. Supralabials (nSpL), number.
43. Postrostrals (nPR), number.
44. Suboculars (nSbO), number.
45. Supraciliaries (nSpC), number.
46. Between subocular and supralabials (nbSbOSpL), number of scales.
47. Fourth toe lamellae (n4TL), number.
48. Fourth finger lamellae (n4FL), number.
49. Sulcated fourth finger lamellae (n4FL1), number.
50. Dorsolateral tibials (nDlT), number.
51. Around the tail (nArT), number of scales around the tail at the first ring
after the cloaca.
52. Along the tail (nAlT), number of scales between cloaca and tip of the
Morphometric Characters
53. Tail length (TL), length between cloacae and tip of the tail.
54. Snout–vent length (SVL), ventral length between mental and cloacae.
Downloaded From: on 1/15/2019
Terms of Use: Access provided by University of Oklahoma
... However, for some highly polymorphic and taxonomic convoluted groups, additional evidence and analyses are frequently needed to provide an improved framework for taxonomic decisions (Breitman et al., 2018;Leaché et al., 2009;Pante et al., 2015). Recent methods and theoretical studies have also indicated combined analyses of morphological and molecular evidence as a more complete approach to assessing the phylogenetic relationship of squamates (Hsiang et al., 2015;Murphy et al., 2016;Pyron, 2015Pyron, , 2017Reeder et al., 2015;Wiens, 2004;Wüster et al., 2018;Zaher et al., 2019). ...
Despite the recent advances in the systematics of snakes, the diversity of several Neotropical groups of species remains poorly understood. The lack of studies focused on the phylogenetic relationship within most of the 20 tribes of Dipsadidae precludes a better understanding of the evolution of this diverse family. Here, we present a comprehensive phylogenetic analysis of Tachymenini, a heterogeneous tribe that comprises 36 viviparous species of dipsadids, grouped in seven genera based on morphological similarities. The tribe is widely distributed throughout South America presenting very distinctive phenotypes, habitats, and behaviours. The phylogenetic relationship among tachymenins is a well‐recognized challenge regarding the systematics of the South American dipsadids. The similar morphotype of some generalist species, combined with the very derived morphology of some strict specialists, creates a complexity of traits that has prevented the comprehension of the systematics of the group. To address such a challenge, we combine molecular (six loci) and morphological (70 characters) datasets in an integrative phylogenetic approach. The resultant phylogenetic trees indicate, with strong support, that three of the seven current recognized genera (Tachymenis, Tomodon, and Thamnodynastes) are non‐monophyletic and, consequently, we propose a new systematic arrangement for Tachymenini. We revalidate two genera, Dryophylax and Mesotes, and we describe three additional monotypic genera, Apographon gen. n., Tachymenoides gen. n., and Zonateres gen. n. to accommodate Tomodon orestes, Tachymenis affinis, and Thamnodynastes lanei, respectively. We also include Tomodon ocellatus and Pseudotomodon trigonatus in Tachymenis and describe a new genus, Galvarinus gen. n., to accommodate the Tachymenis chilensis species group. Furthermore, we also provide an evolutionary scenario for the speciation events based on a time‐calibrated tree, commenting on the diversification and origin of the tribe, and on the probable existence of undescribed species of Mesotes and Dryophylax.
... A denominada "taxonomia inte-grativa", que propõe linhas de evidências independentes (i.e., morfológicas, genéticas, comportamentais, fisiológicas, etc.) para considerar o conceito de espécie como linhagens evolutivas independentes, não se restringindo apenas a caracteres morfológicos, como na época de Saint-Hilaire. Assim, ao se analisar dados usando uma vasta gama de métodos e depositar conclusões na congruência observada nos resultados, diversas novas espécies têm sido descritas para o Cerrado nos últimos anos (e.g., Recoder et al., 2014;Breitman et al., 2018). ...
Full-text available
Auguste Saint-Hilaire (1779-1853), além de botânico e naturalista, foi fruto de um tempo do período histórico no qual estava inserido. Portanto, para entender melhor o papel que esse naturalista francês teve para o desenvolvimento das ciências naturais no Brasil, especificamente do estado de Goiás, é preciso contextualizar a época em que Saint-Hilaire viajou por aqui. O estado da arte das ciências naturais daquela época nos permite comparar com o nosso conhecimento atual sobre o meio ambiente, e como a ecologia se desenvolveu nos últimos duzentos anos. Uma melhor percepção deste panorama nos permite hoje planejar e delinear ações conservacionistas e de preservação de espécies e ecossistemas. Sob o prisma desse contexto histórico, é possível sublinhar os percalços, limitações e vislumbres lúcidos e quase proféticos que o naturalista teve em sua época sobre os ecossistemas brasileiros que conheceu.
... For instance, several gecko species have been described with more than a dozen authors, such as Cyrtodactylus phuocbinhensis Nguyen et al. 2013, Cyrtodactylus taynguyenensis Nguyen et al. 2013, Cyrtodactylus puhuensis Nguyen et al. 2014, and Cnemaspis bidongensis Grismer et al. 2014, each with 14 authors. None of these approach the reptile species with the highest number of authors though, which is the leiosaurid Enyalius capetinga Breitman et al. 2018, with 27 authors. Many of the most prolific gecko describers (Table 1) have worked together, thus, for example, almost all the descriptions by Perry Wood and Evan Quah were co-authored by Lee Grismer. ...
1935 gecko species (and 224 subspecies) were known in December 2019 in seven families and 124 genera. These nearly 2000 species were described by ~950 individuals of whom more than 100 described more than 10 gecko species each. Most gecko species were discovered during the past 40 years. The primary type specimens of all currently recognized geckos (including subspecies) are distributed over 161 collections worldwide, with 20 collections having about two thirds of all primary types. The primary type specimens of about 40 gecko taxa have been lost or unknown. The phylogeny of geckos is well studied, with DNA sequences being available for ~76% of all geckos (compared to ~63% in other reptiles) and morphological characters now being collected in databases. Geographically, geckos occur on five continents and many islands but are most species-rich in Australasia (which also houses the greatest diversity of family-level taxa), Southeast Asia, Africa, Madagascar, and the West Indies. Among countries, Australia has the highest number of geckos (241 species), with India, Madagascar, and Malaysia being the only other countries with more than 100 described species each. As expected, when correcting for land area, countries outside the tropics have fewer geckos.
We describe two new endemic Hypostomus species from central Brazil, which were previously identified as genetically distinct lineages in a recent genomic study that recommended their testing and potential description based on morphological data. A machine learning classification procedure (random forest) was used to investigate morphological variation and identify putatively diagnostic characters for these candidate species, and revealed that each is morphologically distinct. The new species Hypostomus cafuringa is characterized by small size, dark spots under a light background color, deeper caudal peduncle, and shorter first ray of the pectoral‐fin and base of the dorsal‐fin when compared to congeneric species from the region. Hypostomus cafuringa is known from the headwaters of the Maranhão River, upper Tocantins River basin, Distrito Federal, Brazil. The second new species, Hypostomus crulsi, is characterized as having dark spots under a light background color, absence of plates along the abdomen region, shorter first ray of the pelvic‐fin, shorter first ray of the pectoral‐fin, and smaller body size. Hypostomus crulsi is known from the headwaters of the São Bartolomeu River, upper Paraná River basin, Distrito Federal, Brazil. The rapid conversion of natural habitats for agricultural development and the isolation of protected areas represents a serious threat to the continued existence of these two newly described and endemic species, which warrant conservation assessment. This article is protected by copyright. All rights reserved.
Full-text available
Popular species names have always been based on human relationships with things around them, usually reflecting species' external morphology , behavior or even the habitat they inhabit. In Brazil, the high number of popular names, in many cases for the same species, makes it difficult to comprehensively recognize these names, hampering communication between everyone interested in reptiles. This study presents a compilation of the popular names for the species of reptiles occurring in Brazil based on literature data. We listed 1264 popular names, 25 for Amphisbaenia, 29 for Crocodylia, 137 for Testudines, 301 for "Lizards" and 772 for Snakes. All Testudines and
Full-text available
Popular species names have always been based on human relationships with things around them, usually reflecting species' external morphology , behavior or even the habitat they inhabit. In Brazil, the high number of popular names, in many cases for the same species, makes it difficult to comprehensively recognize these names, hampering communication between everyone interested in reptiles. This study presents a compilation of the popular names for the species of reptiles occurring in Brazil based on literature data. We listed 1264 popular names, 25 for Amphisbaenia, 29 for Crocodylia, 137 for Testudines, 301 for "Lizards" and 772 for Snakes. All Testudines and Crocodylia have popular names, however we found no popular names for three (4%) species of Amphisbaenia, 66 (24%) “Lizards” and 60 (15%) Snakes. Therefore, 15% of the 802 reptile species currently known in the country still have no popular names. Also, all species of Testudines and Crocodylia have exclusive popular names, nevertheless we found no exclusive names for 68 (92%) species of Amphisbaenia, 191 (69%) species of “Lizards” and 218 (53%) species of Snakes. This inventory is currently the most comprehensive effort to list the popular names of reptile species in Brazil.
Full-text available
Enyalius boulengeri Etheridge, 1969 occurs in the Atlantic Forest, in the states of Espírito Santo and Minas Gerais, in southeastern Brazil (Costa and Bérnils, 2018), and its conservation status has not yet been assessed, since basic information on its biology is still scarce or non-existent in the literature. We here describe for the first time a copulation of E. boulengeri under natural conditions, in a protected area in the state of Espírito Santo. In addition, we discuss the courtship and copulation behaviours of some other Enyalius species.
Full-text available
The new version of the Brazilian Forest Code prescribes a 30 m forest buffer around small width streams (less than 10 m width) as Areas of Permanent Preservation (APPs), to ensure water supply, erosion control, and biodiversity conservation. We tested the effectiveness of the prescribed buffer in maintaining lizard assemblages in riparian forests associated with small width streams in the Cerrado of central Brazil. We used a capture-recapture study in three riparian forests: two in protected areas, and one deforested until the limit imposed by law. We captured lizards using pitfall traps with drift fences during a period of 12-14 months in each forest. We recorded 11 lizard species during the sampling period. The reduced riparian forest had lower abundance, richness, evenness, and phylogenetic diversity compared to the preserved habitats. This forest also lacked forest specialist species (e.g., Enyalius bilineatus), bearing an assemblage composed exclusively by two generalist and synanthropic species (Tropidurus torquatus and Ameiva ameiva). Our results indicate that the new Brazilian Forest Code is inadequate to ensure the protection of lizard assemblages in Cerrado riparian forests, and should take into account larger riparian buffers associated with small rivers, and also consider other factors such as the environmental quality around areas of permanent protection along riparian forests.
Full-text available
Tegus of the genera Tupinambis and Salvator are the largest Neotropical lizards and the most exploited clade of Neotropical reptiles. For three decades more than 34 million tegu skins were in trade, about 1.02 million per year. The genus Tupinambis is distributed in South America east of the Andes, and currently contains four recognized species, three of which are found only in Brazil. However, the type species of the genus, T. teguixin, is known from Bolivia, Brazil, Colombia, Ecuador, French Guyana, Guyana, Peru, Suriname, Trinidad and Tobago, and Venezuela (including the Isla de Margarita). Here we present molecular and morphological evidence that this species is genetically divergent across its range and identify four distinct clades some of which are sympatric. The occurrence of cryptic sympat-ric species undoubtedly exacerbated the nomenclatural problems of the past. We discuss the species supported by molecular and morphological evidence and increase the number of species in the genus Tupinambis to seven. The four members of the T. teguixin group continue to be confused with Salvator merianae, despite having a distinctly different morphology and reproductive mode. All members of the genus Tupinambis are CITES Appendix II. Yet, they continue to be heavily exploited, under studied, and confused in the minds of the public, conservationists, and scientists.
Full-text available
In the present study, we aimed to provide information regarding feeding habits and other ecological aspects of Enyalius perditus in an Atlantic Forest remnant in the state of Rio de Janeiro, Brazil. Enyalius perditus individuals in Rio de Janeiro fed predominantly upon arthropods such as caterpillars, isopods, and spiders, as is typical of lizards in the genus. There were no pronounced variations in the main types of prey consumed between this and other populations of E. perditus studied in spite of potential differences in local food supplies between environments. The individuals consumed fragments of dead leaves, ingested prey commonly found in leaf litter (e.g., Isopoda and Coleoptera larvae), and were sometimes found on the ground or captured in pit-fall traps during fieldwork, which suggests that these lizards use the forest floor leaf litter for foraging and other activities, although they also use perches above ground level. Our study provides additional information and reinforces the idea of the use of both the vegetation and the leaf litter of forest floor by lizards of the genus Enyalius.
Full-text available
Studies on the natural history of the lizard Enyalius iheringii Boulenger, 1885, as well as other tropical lizards, are rare. In this study, some aspects of the natural history of this endemic species from the Atlantic forest are reported in areas of Vale do Itajai, state of Santa Catarina, Brazil. Twenty individuals were found, of which 18 were collected. Most of them were found over the vegetation (n= 17) and on the ground (n= 3). The main defensive strategy displayed was camouflage (n= 16). Jumping (n= 1), jumping and running (n= 1) and running (n= 2) were also observed in some individuals. When handled, lizards exhibited mouth wide open, hissing, and occasionally biting, as well as color change in males. Regarding its diet, the numerically most important prey was beetles (Coleoptera), followed by Lepidoptera larvae. Beetles, lepidopteran larvae and spiders were the most frequent food items. Males and females did not differ in size. Three sexually mature females (100-113 mm SVL) were found in December and January.
Full-text available
Classical supervised learning enjoys the luxury of accessing the true known labels for the observations in a modeled dataset. Real life, however, poses an abundance of problems, where the labels are only partially defined, i.e., are uncertain and given only for a subset of observations. Such partial labels can occur regardless of the knowledge source. For example, an experimental assessment of labels may have limited capacity and is prone to measurement errors. Also expert knowledge is often restricted to a specialized area and is thus unlikely to provide trustworthy labels for all observations in the dataset. Partially supervised mixture modeling is able to process such sparse and imprecise input. Here, we present an R package called bgmm, which implements two partially supervised mixture modeling methods: soft-label and belief-based modeling. For completeness, we equipped the package also with the functionality of unsupervised, semi- and fully supervised mixture modeling. On real data we present the usage of bgmm for basic model-fitting in all modeling variants. The package can be applied also to selection of the best-fitting from a set of models with different component numbers or constraints on their structures. This functionality is presented on an artificial dataset, which can be simulated in bgmm from a distribution defined by a given model.
At fine spatial scales, savanna-rainforest-grassland boundary dynamics are thought to be mediated by the interplay between fire, vegetation and soil feedbacks. These processes were investigated by quantifying tree species composition, the light environment, quantities and flammability of fuels, bark thickness, and soil conditions across stable and dynamic rainforest boundaries that adjoin grassland and eucalypt savanna in the highlands of the Bunya Mountains, southeast Queensland, Australia. The size class distribution of savanna and rainforest stems was indicative of the encroachment of rainforest species into savanna and grassland. Increasing dominance of rainforest trees corresponds to an increase in woody canopy cover, the dominance of litter fuels (woody debris and leaf), and decline in grass occurrence. There is marked difference in litter and grass fuel flammability and this result is largely an influence of strongly dissimilar fuel bulk densities. Relative bark thickness, a measure of stem fire resistance, was found to be generally greater in savanna species when compared to that of rainforest species, with notable exceptions being the conifers Araucaria bidwillii and Araucaria cunninghamii. A transect study of soil nutrients across one dynamic rainforest – grassland boundary indicated the mass of carbon and nitrogen, but not phosphorus, increased across the successional gradient. Soil carbon turnover time is shortest in stable rainforest, intermediate in dynamic rainforest and longest in grassland highlighting nutrient cycling differentiation. We conclude that the general absence of fire in the Bunya Mountains, due to a divergence from traditional Aboriginal burning practices, has allowed for the encroachment of fire-sensitive rainforest species into the flammable biomes of this landscape. Rainforest invasion is likely to have reduced fire risk via changes to fuel composition and microclimatic conditions, and this feedback will be reinforced by altered nutrient cycling. The mechanics of the feedbacks here identified are discussed in terms of landscape change theory.
PartitionFinder 2 is a program for automatically selecting best-fit partitioning schemes and models of evolution for phylogenetic analyses. PartitionFinder 2 is substantially faster and more efficient than version 1, and incorporates many new methods and features. These include the ability to analyze morphological datasets, new methods to analyze genome-scale datasets, new output formats to facilitate interoperability with downstream software, and many new models of molecular evolution. PartitionFinder 2 is freely available under an open source license and works on Windows, OSX, and Linux operating systems. It can be downloaded from The source code is available at