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 research.
Phylogeny of the 
Dactyloa
 Clade of 
Anolis
 Lizards: New
Insights from Combining Morphological and Molecular Data
Author(s): Mar?a Del Rosario Casta?eda and KEVIN DE QUEIROZ
Source: Bulletin of the Museum of Comparative Zoology, 160(7):345-398.
2013.
Published By: Museum of Comparative Zoology, Harvard University
DOI: http://dx.doi.org/10.3099/0027-4100-160.7.345
URL: http://www.bioone.org/doi/full/10.3099/0027-4100-160.7.345
BioOne (www.bioone.org) is a nonprofit, online aggregation of core research in the
biological, ecological, and environmental sciences. BioOne provides a sustainable
online platform for over 170 journals and books published by nonprofit societies,
associations, museums, institutions, and presses.
Your use of this PDF, the BioOne Web site, and all posted and associated content
indicates your acceptance of BioOne?s Terms of Use, available at www.bioone.org/
page/terms_of_use.
Usage of BioOne 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.
PHYLOGENY OF THE DACTYLOA CLADE OF ANOLIS LIZARDS:
NEW INSIGHTS FROM COMBINING MORPHOLOGICAL AND
MOLECULAR DATA
MARI?A DEL ROSARIO CASTAN?EDA1,2,3 AND KEVIN DE QUEIROZ2
CONTENTS
Note Added in Proof ------- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - 345
Abstract ------- -- - - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - 346
Introduction -------- - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - 347
Current Taxonomy within Dactyloa - - -- - - - - - - - - - 349
Materials and Methods ------- - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - 351
Taxon and Character Sampling ------- - - - - -- - - - - - - - - - 351
Character Coding -------- - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - 352
Continuous characters ------- - - - -- - - - - - - - - - - - - - - - - - - - - -- 352
Polymorphic characters ------- -- - - - - - - - - - - - - - - - - - - - - -- 352
Comparison of coding methods ------- - - - - - - - -- 353
Morphological Data Sets and Phylogenetic
Analyses -------- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - 353
Combined Data Sets and Phylogenetic
Analyses -------- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - 354
Tests of Phylogenetic Hypotheses ------- - - - - - - - - - 355
Results ------- - - -- - - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - 356
Comparisons Between Coding Methods ---- 356
Phylogenetic Analyses -------- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - 356
Morphology-only data sets ------- - - - - - - - - - - - - - - - - - -- 356
Combined data sets ------- - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- 357
Tests of Phylogenetic Hypotheses ------- - - - - - - - - - 363
Discussion -------- - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - 364
Differences Among Coding Methods ------- - - - 365
Phylogeny of Dactyloa - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - 366
Previously Recognized Taxa ------- - - - - - - - - - - -- - - - - - - - - - 370
Proposed Taxonomy -------- - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - 371
Dactyloa - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- 374
aequatorialis series ------- - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- 376
latifrons series ------- - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- 377
Megaloa - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- 378
punctatus series ------- - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- 379
roquet series ------- - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - 381
heterodermus series ------- - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - 381
Phenacosaurus - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - 382
Incertae sedis - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - 384
Acknowledgments ------- - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - 385
Appendix I. Morphological character
descriptions -------- - - - - - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - 385
Literature Cited -------- - - - - - - - - - - - - - - - - -- - - - - - - - - - - - - - - - - - - - - -- - - - - - - 394
NOTE ADDED IN PROOF
Shortly after our paper was accepted, Nicholson
and colleagues published a phylogenetic analysis of
anoles and a proposal to divide Anolis into eight
genera (Nicholson, K. E., B. I. Crother, C. Guyer, and
J. M. Savage. 2012. It is time for a new classification of
anoles (Squamata: Dactyloidae). Zootaxa 3477: 1?
108). Here, we comment briefly on their study as it
pertains to the phylogeny and taxonomy of the
Dactyloa clade.
Despite not inferring Dactyloa to be monophyletic
in the tree used for their proposed taxonomy (i.e., the
consensus tree from the combined morphological and
molecular parsimony analysis; their fig. 5A, note
positions of Anolis bonairensis, A. chloris, A. per-
accae, and A. apollinaris), Nicholson et al. (2012)
recognized Dactyloa as one of their eight genera
without making reference to this inconsistency
(although Dactyloa was inferred to be monophyletic
in their molecular tree, fig. 4A). By contrast, our
combined data set supported the monophyly of
Dactyloa (Figs. 3, 4), and we have chosen to treat
Dactyloa as a subclade of Anolis rather than as a
separate genus in the interest of avoiding disruptive
and unnecessary name changes.
Some of our informally named series correspond,
with some differences in species composition, to the
species groups proposed by Nicholson et al. (2012). We
describe the differences below.
Our latifrons series corresponds to their latifrons
species group, except that in the tree purportedly used
for their taxonomy (fig. 5A), A. aequatorialis and A.
ventrimaculatus were inferred to be part of this species
1 Department of Biological Sciences, The George
Washington University, 2023 G Street NW, Washing-
ton, DC 20052.
2 Department of Vertebrate Zoology, National
Museum of Natural History, Smithsonian Institution,
MRC 162, Washington, DC 20560. Author for
correspondence (mcastanedaprada@fas.harvard.edu).
3 Address through April 2014: Museum of Compar-
ative Zoology, Harvard University, 26 Oxford Street,
Cambridge, Massachusetts 02138.
Bull. Mus. Comp. Zool., 160(7): 345?398, February, 2013 345
group (both species are absent from their molecular
tree, fig. 4A), although their classification (appendix
III) places both species in their punctata species group
with no explanation for this inconsistency. We inferred
these two species with strong support to be part of a
monophyletic aequatorialis series that is mutually
exclusive with respect to both the latifrons and
punctatus series. Additionally, we have tentatively
placed A. mirus and A. parilis in the aequatorialis
series based on their previous inclusion in the
traditional aequatorialis series (Williams, 1975; Ayala-
Varela and Velasco, 2010); the tentative assignment
reflects the current absence of these species from
explicit phylogenetic analyses. By contrast, Nicholson
et al. (2012) assigned A. mirus and A. parilis, neither of
which was included in any of their analyses, to their
latifrons species group without explanation. Finally, we
placed A. propinquus in the latifrons series based on its
hypothesized close relationship to A. apollinaris
(Williams, 1988). By contrast, Nicholson et al. (2012)
placed this species, which was not included in any of
their phylogenetic analyses, in their punctata species
group without explanation.
The combination of our aequatorialis and puncta-
tus series corresponds roughly to the punctata species
group in the classification of Nicholson et al. (2012,
appendix III). We inferred these two series to be
mutually exclusive clades (results further supported by
molecular data alone; Castan?eda and de Queiroz,
2011). Contradicting their own taxonomy, the tree of
Nicholson et al. (2012, fig. 5A) supports the separation
of the aequatorialis series, in that A. aequatorialis, A.
ventrimaculatus, A. chloris, and A. peraccae are not
inferred to be part of their punctata species group,
despite being referred to that group in their classifi-
cation (appendix III). Their tree does place A.
fasciatus in their punctata species group, whereas
our results indicate that this species is part of the
aequatorialis series. We have treated A. calimae and
A. cuscoensis as incertae sedis within Dactyloa based
on conflicting results for A. calimae (also found by
Castan?eda and de Queiroz, 2011) and the inferred
inclusion of A. cuscoensis by Poe et al. (2008) in clades
not inferred in our study. By contrast, Nicholson et al.
(2012) referred these two species to their punctata
species group, although neither species was included
in any of their phylogenetic analyses. Similarly, we
have treated A. laevis and A. phyllorhinus, species
formerly placed in the laevis series, as incertae sedis
based on their current absence from explicit phyloge-
netic analyses (although we consider it likely that A.
phyllorhinus belongs to the punctatus series). By
contrast, Nicholson et al. (2012) assigned both of these
species to the punctata species group, although
neither was included in any of their phylogenetic
analyses.
Our Phenacosaurus and our heterodermus series
both correspond approximately to the heterodema
species group of Nicholson et al. (2012), with the
exception that they included A. carlostoddi, A.
bellipeniculus, and A. neblininus. We consider these
three species as incertae sedis within Dactyloa based on
conflicting results in our analyses for A. carlostoddi and
A. neblininus and the absence from explicit phyloge-
netic analyses of A. bellipeniculus, as well as its
previously inferred close relationship to A. neblininus
(Myers and Donnelly, 1996).
Our roquet series corresponds approximately to
their roquet species group. However, in the tree
purportedly used for their taxonomy (their fig. 5A),
their roquet species group is not monophyletic: A.
bonairensis is inferred as sister to A. occultus outside of
Dactyloa (A. bonairensis is not included in their
molecular-only tree; fig. 4A). By contrast, we inferred
A. bonairensis to be part of a monophyletic roquet
series (Figs. 3, 4).
Our combined analyses are based on a sample of 60
of the 83 currently recognized species in the Dactyloa
clade, 40 of which were sampled for molecular data,
whereas the combined analysis of Nicholson et al.
(2012) is based on a sample of 31 Dactyloa species, 16
of which were sampled for molecular data (three others
were sampled for molecular data only). Additionally,
our molecular data consists of ,4,950 base positions
representing three gene regions and both mitochon-
drial and nuclear DNA, whereas theirs consists of
,1,500 base positions representing one of the two
mitochondrial gene regions used in our study. Because
our results are based on larger samples of Dactyloa
species (for both molecular and morphological data), as
well as larger samples of molecular data (with respect
to both numbers of bases and numbers of gene
fragments, and including both mitochondrial and
nuclear genes), and because many of their taxonomic
conclusions that differ from ours are either contradict-
ed by their own results or unsubstantiated, we do not
consider any of the differences between our phyloge-
netic results and taxonomic conclusions compared with
those in the study by Nicholson et al. (2012) to warrant
changes to our proposed taxonomy. In contrast to
Nicholson et al. (2012), we refrain from assigning some
species to series and treat some taxonomic assignments
as tentative because of contradictory results or poorly
supported inferences, and we present justifications for
all taxonomic decisions pertaining to species not
included in our analyses.
ABSTRACT. We present a phylogenetic analysis of the
Dactyloa clade of Anolis lizards, based on morpholog-
ical (66 characters of external morphology and
osteology) and molecular (,4,700 bases of mitochon-
drial and nuclear DNA) data. Our set of morphological
characters includes some that exhibit continuous
variation and others that exhibit polymorphism within
species; we explored different coding methods for
these classes of characters. We performed parsimony
and Bayesian analyses on morphology-only and com-
bined data sets. Additionally, we explicitly tested
hypotheses of monophyly of: 1) Dactyloa including
Phenacosaurus, 2) Dactyloa excluding Phenacosaurus
(as traditionally circumscribed), 3) taxa previously
ranked as series or species groups described based on
346 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
morphological characters, and 4) clades inferred from
molecular data. The morphological data alone did not
yield Dactyloa or any of the previously recognized
series described based on morphological characters;
only the Phenacosaurus clade (as delimited based on
molecular data) was inferred with the morphological
data, and only in the parsimony analysis. In contrast,
Dactyloa was inferred as monophyletic with the
combined data set, although topology tests failed to
reject the hypothesis of non-monophyly. Additionally,
five clades inferred based on molecular data (eastern,
latifrons, Phenacosaurus, roquet, and western) were
inferred with the combined data sets with variable
support and including additional species for which
molecular data were not available and which have
geographic distributions that conform to those of the
clades in which they were included. Of the previously
recognized taxa based on morphological characters,
only the roquet series, which corresponds in species
composition to the roquet clade, was inferred with the
combined data. Topology tests with the combined data
set rejected the monophyly of the aequatorialis,
latifrons (as traditionally circumscribed), and punctatus
series but not that of the tigrinus series and
Phenacosaurus (as traditionally circumscribed). Our
phylogenetic analyses and topology tests indicate that a
new taxonomy for Dactyloa is warranted; we therefore
present a revised taxonomy based on the results our
phylogenetic analyses and employing phylogenetic
definitions of taxon names.
Key words: Anolis, Character coding, Dactyloa,
Phylogeny, Taxonomy
INTRODUCTION
The Anolis clade, one of the most diverse
groups of vertebrates traditionally ranked as
a genus, is composed of 384 currently
recognized species (Uetz, 2012). This group
of lizards is primarily Neotropical in distri-
bution. Its members are characterized (with
a few exceptions) by the possession of
adhesive toe pads formed by laterally
expanded subdigital scales, called lamellae,
that are covered by microscopic setae, and
of extensible and often brightly colored
throat fans, called dewlaps, that are sup-
ported by elongated second ceratobran-
chials and occur in males and often in
females (Etheridge, 1959).
Based on Etheridge?s (1959) seminal
work on the phylogeny and taxonomy of
anoles, two large groups, traditionally
ranked as sections, were informally recog-
nized within Anolis based on the absence
(alpha section) or presence (beta section) of
transverse processes on the anterior auto-
tomic caudal vertebrae. Each section was
further subdivided into series and species
groups based on morphological characters
(Etheridge, 1959; Williams, 1976a,b). Sub-
sequent to the morphological studies of
Etheridge (1959) and Williams (1976a,b), a
wide variety of data have been brought to
bear on the phylogeny and taxonomy of
Anolis, including albumin immunology (e.g.,
Gorman et al., 1980b, 1984; Shochat and
Dessauer, 1981), allozymes (e.g., Gorman
and Kim, 1976; Gorman et al., 1980a;
Burnell and Hedges, 1990), behavior (e.g.,
Gorman, 1968), karyotypes (e.g., Gorman et
al., 1968, 1983; Gorman and Stamm, 1975),
and DNA sequences (e.g., Jackman et al.,
1999; Schneider et al., 2001; Glor et al.,
2003). Analyses of these data have provided
support for the monophyly of the beta
section and of several series and species
groups (e.g., Creer et al., 2001; Schneider et
al., 2001; Jackman et al., 2002; Nicholson,
2002). However, they have also indicated
that other groups, including the alpha
section, are not monophyletic. Additionally,
the phylogenetic relationships within and
among some groups are still disputed (e.g.,
Jackman et al., 1999; Nicholson, 2002; Poe,
2004).
Within the alpha section, Etheridge
(1959) recognized the latifrons series for
species with at least four postxiphisternal
chevrons attached to the bony dorsal ribs
and an arrow-shaped interclavicle (in which
the lateral processes of the interclavicle are
divergent from the proximal parts of the
clavicles). Etheridge?s (1959) latifrons series
was composed of all mainland alpha Anolis
(excluding Phenacosaurus; see below) along
with the species in the roquet series from
the southern Lesser Antilles, as well as
Anolis agassizi and A. gorgonae from the
Pacific islands of Malpelo and Gorgona,
respectively. The latifrons series of Ether-
idge (1959) corresponds to the genus
Dactyloa, one of five genera recognized by
Guyer and Savage (1987 [1986]) based on a
proposal to ??divide?? Anolis taxonomically.
Although the recognition of those genera is
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 347
controversial (Cannatella and de Queiroz,
1989; Williams, 1989; Poe and Iban?ez,
2007), some recent authors apply some of
the same names to clades within Anolis
regardless of rank and not necessarily with
identical composition (e.g., Nicholson, 2002;
Brandley and de Queiroz, 2004; de Queiroz
and Reeder, 2008). In the present study, we
use the name Dactyloa for the clade originat-
ing in the most recent common ancestor of
the species included in the genus Dactyloa by
Savage and Guyer (1989), which also includes
the anoles formerly assigned to the genus
Phenacosaurus according to the results of
recent phylogenetic analyses (e.g., Jackman et
al., 1999; Poe, 2004; Nicholson et al., 2005;
Castan?eda and de Queiroz, 2011). Currently,
there are 83 recognized species in the
Dactyloa clade, distributed among seven
subgroups (based on morphological charac-
ters) commonly assigned to the rank of series:
aequatorialis, laevis, latifrons, punctatus, Phe-
nacosaurus, roquet, and tigrinus (see ??Cur-
rent Taxonomy within Dactyloa,?? below).
Three phylogenetic analyses have includ-
ed more than 20% of the currently recog-
nized Dactyloa species: Poe (2004) included
28 species, Nicholson et al. (2005) included
17 species (13 of which were included in
Poe [2004]), and Castan?eda and de Queiroz
(2011) included 42 species (including 22 of
28 of Poe [2004] and 15 of 17 of Nicholson
et al. [2005]). In Poe?s (2004) combined
analysis of allozyme, karyotype, morpholog-
ical, and molecular data, Dactyloa (as
defined in the previous paragraph, the name
was not used by Poe) was inferred to be
monophyletic based on the arrow shape of
the interclavicle and the presence of a
splenial in the mandible. However, boot-
strap support for the clade (Poe, 2004, fig. 2,
node 352) was less than 50% and both
morphological characters supporting it are
reversals to ancestral conditions. In the
analyses of Nicholson et al. (2005, fig. 1)
and Castan?eda and de Queiroz (2011, fig.
1), based on molecular data, Dactyloa was
inferred with moderate to strong bootstrap
support ($80%) and Bayesian posterior
probabilities ($0.90).
Of the seven subgroups described within
Dactyloa based on morphological charac-
ters, only the roquet series has been
consistently inferred by several phylogenetic
analyses and has passed explicit statistical
tests of monophyly (Jackman et al., 1999;
Poe, 2004; Nicholson et al., 2005; Castan?eda
and de Queiroz, 2011). Phenacosaurus, as
traditionally circumscribed, was inferred as
monophyletic by Poe (2004) and Nicholson
et al. (2005), although only two and three
species of this group, respectively, were
included in their phylogenetic analyses. In
the analyses of Castan?eda and de Queiroz
(2011), which included six species of
Phenacosaurus, the group was inferred as
monophyletic with the exception of A.
neblininus. The remaining subgroups have
not been inferred in phylogenetic analyses
or passed explicit statistical tests of mono-
phyly, although the laevis series has not
been tested (Poe, 2004; Nicholson et al.,
2005; Castan?eda and de Queiroz, 2011). In
contrast to the poor support for the
traditional series, Castan?eda and de Queiroz
(2011) inferred five strongly supported
subclades within Dactyloa, which they
recognized informally as eastern, latifrons,
Phenacosaurus, roquet, and western clades.
Although some of these clades bear the
same names as species groups recognized by
Williams (1976b) and series recognized by
Savage and Guyer (1989), their composition
is not necessarily the same.
In this study, we describe and score 66
morphological characters (external and os-
teological) for 60 species of Dactyloa and 6
outgroup species (including non-Dactyloa
Anolis and non-Anolis Polychrotinae) to
resolve the phylogenetic relationships with-
in the Dactyloa clade. We analyze the
morphological characters alone and in
combination with ,4,720 bases of DNA
sequence data presented by Castan?eda and
de Queiroz (2011). We perform parsimony
and Bayesian analyses and examine differ-
ent coding methods for continuous and
polymorphic characters. We use tree topol-
ogy tests to test hypotheses of monophyly of:
1) Dactyloa including Phenacosaurus, 2)
348 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
Dactyloa excluding Phenacosaurus (as tra-
ditionally circumscribed), 3) the traditional-
ly recognized series delimited based on
morphological characters (Williams, 1976b;
Savage and Guyer, 1989), and 4) the clades
inferred based on molecular data (Casta-
n?eda and de Queiroz, 2011). Based on the
results of our analyses, we present a revised
taxonomy that is consistent with the current
knowledge of the phylogenetic relationships
within the Dactyloa clade.
Current Taxonomy within Dactyloa
Based on morphological characters, six
subgroups ranked as species groups by
Williams (1976b) and as series by Savage
and Guyer (1989) have been recognized
within Dactyloa: aequatorialis, laevis, lati-
frons, punctatus, roquet, and tigrinus. In
agreement with recent phylogenetic analyses
(e.g., Jackman et al., 1999; Poe, 2004;
Nicholson et al., 2005; Castan?eda and de
Queiroz, 2011), we recognize the group of
species previously identified as the genus
Phenacosaurus as an additional subgroup of
Dactyloa. Some Dactyloa species have not
been assigned to any of these subgroups; for
example, Anolis agassizi, A. anchicayae, A.
cuscoensis, and A. ibanezi were referred to
what is here recognized as the Dactyloa
clade, but with no series assignment (Ether-
idge, 1959; Poe et al., 2008, 2009a,b). Other
species have been assigned to subgroups, but
assignment was inconsistent. For example,
Anolis kunayalae was described as morpho-
logically similar to A. mirus and A. parilis,
both members of the aequatorialis series, but
assigned by the describing authors (Hulebak
et al., 2007) to the latifrons group sensu
stricto (5 latifrons species group of Williams,
1976b) which is equivalent to the latifrons
series of Savage and Guyer (1989); we
therefore consider the series assignment of
this species uncertain.
Aequatorialis series. The aequatorialis
series is currently composed of 13 species:
A. aequatorialis, A. anoriensis, A. antio-
quiae, A. eulaemus, A. fitchi, A. gemmosus,
A. maculigula, A. megalopithecus, A. mirus,
A. otongae, A. parilis, A. podocarpus, and A.
ventrimaculatus, which are characterized by
moderate to large body size (adult male
snout-to-vent length [SVL] 66?101 mm),
small head scales, smooth ventral scales,
uniform dorsal scalation, and in some
species narrow toe lamellae (Williams,
1976b; Williams and Acosta, 1996; Ayala-
Varela and Torres-Carvajal, 2010; Ayala-
Varela and Velasco, 2010; Velasco et al.,
2010). The species in the aequatorialis
series are distributed between 1,300 and
2,500 m above sea level in the Andes of
Colombia (western and central cordilleras)
and Ecuador (eastern and western slopes)
(Williams and Duellman, 1984; Ayala-Varela
and Torres-Carvajal, 2010; Velasco et al.,
2010; Ayala and Castro, unpublished).
Laevis series. The laevis series, composed
of A. laevis, A. phyllorhinus, and A.
proboscis, is characterized by the presence
of a soft, median protuberance from the
snout, called a proboscis (Williams, 1976b,
1979) or nose leaf (Peters and Orces, 1956).
Members of this series have a disjunct
geographic distribution: A. laevis is distri-
buted in the eastern foothills of the
Peruvian Andes, A. proboscis is found at
mid-elevations on the western slopes of the
Ecuadorian Andes, and A. phyllorhinus is
found in central Amazonia (Williams, 1979;
Rodrigues et al., 2002).
Latifrons series. The latifrons series is
composed of 12 species: A. apollinaris, A.
casildae, A. danieli, A. fraseri, A. frenatus, A.
insignis, A. latifrons, A. microtus, A. prin-
ceps, A. propinquus, A. purpurescens, and A.
squamulatus, which are characterized by
adult SVL . 100 mm, large dewlaps in adult
males (.500 mm2), expanded toe lamellae,
small head scales, smooth to weakly keeled
ventral scales, and uniform dorsal scalation
(Williams, 1976b; Savage and Talbot, 1978).
These species, also called the giant mainland
anoles (Dunn, 1937), are distributed in the
lowlands and premontane forests of Costa
Rica, western Panama, Colombia (western
cordillera), and Ecuador; in the northern
central lowlands of Venezuela; and in the
inter-Andean valleys of Colombia (Savage
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 349
and Talbot, 1978; Arosemena et al., 1991;
Ayala and Castro, unpublished).
Punctatus series. The punctatus series is
composed of 21 species: A. anatoloros, A.
boettgeri, A. calimae, A. caquetae, A. chloris,
A. chocorum, A. deltae, A. dissimilis, A.
fasciatus, A. festae, A. gorgonae, A. huilae,
A. jacare, A. nigrolineatus, A. peraccae, A.
philopunctatus, A. punctatus, A. santamar-
tae, A. soinii, A. transversalis, and A.
vaupesianus. The characters used to diag-
nose this series include adult SVL, 100 mm,
wide toe lamellae (compared with the narrow
lamellae observed in the aequatorialis se-
ries), small head scales, smooth to weakly
keeled ventral scales (except in A. punctatus
boulengeri, which has strongly keeled ven-
trals), uniform dorsal scalation, and in some
species a protuberant snout in males (Wil-
liams, 1976b, 1982). Species in the punctatus
series are distributed in the western lowlands
of Panama, Colombia, and Ecuador; the mid-
to high elevations of the Andes of Colombia
(including the Sierra Nevada de Santa
Marta), Venezuela, and Peru (eastern slope);
the Amazon region and the Orinoco delta
(Williams, 1982; Rodrigues, 1988; Poe and
Yan?ez-Miranda, 2008; Poe et al., 2008,
2009a,b; Ayala and Castro, unpublished).
Roquet series. The roquet series is
composed of 9 species: A. aeneus, A.
blanquillanus, A. bonairensis, A. extremus,
A. griseus, A. luciae, A. richardii, A. roquet,
and A. trinitatis. The monophyly of this
series is supported by karyological (Gorman
and Atkins, 1969), morphological (Lazell,
1972; Poe, 2004), and molecular data
(cytochrome b sequences, Giannassi et al.,
2000; ND2 sequences, Creer et al., 2001).
Six morphological synapomorphies support
the monophyly of this series: 1) greater
sexual size dimorphism, 2) an increase in
interparietal scale size relative to surround-
ing scales, 3) an increase in mean number of
postmental scales, 4) a straight (as opposed
to concave) posterior border of the mental
scale, 5) supraorbital semicircles in contact,
and 6) interparietal scale in contact with the
supraorbital semicircles (Poe, 2004). The
roquet series is distributed in the southern
Lesser Antilles, from Martinique south to
Grenada, and on the islands of La Blan-
quilla, Bonaire, Tobago, and Trinidad
(where A. aeneus and A. trinitatis have
been introduced; Gorman and Dessauer,
1965, 1966) and Guyana (where A. aeneus
has been introduced; Gorman and Des-
sauer, 1965; Gorman et al., 1971).
Tigrinus series. The tigrinus series is
composed of 9 species: A. lamari, A. menta,
A. nasofrontalis, A. paravertebralis, A. pseu-
dotigrinus, A. ruizii, A. solitarius, A. tigrinus,
and A. umbrivagus, and is characterized by
small body size (adult male SVL 5 40?
60 mm), large smooth head scales, a large
interparietal scale bordered by large scales
and usually in contact with the supraorbital
semicircles, and ventral scales smooth and
larger than dorsal scales. Some species
exhibit a parietal knob (a small projection
of the posteriormost end of the central ridge
of the Y-shaped parietal crests), externally
visible in some species on the occipital area
between the post-interparietal scales and
nape scales (Williams, 1976b, 1992). Species
in the tigrinus series are distributed in high
elevations of the Sierra Nevada de Santa
Marta (Colombia), the Andes of Colombia
(eastern cordillera) and Venezuela, and the
Atlantic forest of southeastern Brazil (Wil-
liams, 1992; Bernal Carlo and Roze, 2005).
Phenacosaurus. Phenacosaurus is com-
posed of 11 species: A. bellipeniculus, A.
carlostoddi, A. euskalerriari, A. heteroder-
mus, A. inderenae, A. neblininus, A. nice-
fori, A. orcesi, A. tetarii, A. vanzolinii, and
A. williamsmittermeierorum. Earlier, Phe-
nacosaurus was considered a separate genus
from Anolis (Barbour, 1920) based on the
heterogeneous dorsal scalation (enlarged
round flat scales surrounded by smaller
scales and granules), the tail structure
(probably prehensile), an elevated rim of
head plates (casque), digits widely and
evenly dilated (such that their sides are
parallel), and a ??feebly developed?? dorso-
nuchal crest (Lazell, 1969). However, recent
phylogenetic analyses (Poe, 1998, 2004;
Jackman et al., 1999; Nicholson et al.,
2005; Castan?eda and de Queiroz, 2011)
350 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
inferred these species to be nested within
the clades composed of the species assigned
to both Anolis and Dactyloa; therefore, we
here consider Phenacosaurus another sub-
group of Dactyloa. Phenacosaurus species
are distributed in the Andean highlands
(between 1,300 and 3,000 m) of Colombia,
northern Ecuador, central Peru, and west-
ern Venezuela and the isolated tepuis of
southeastern Venezuela (Lazell, 1969;
Myers et al., 1993; Barros et al., 1996;
Myers and Donnelly, 1996; Williams et al.,
1996; Poe and Yan?ez-Miranda, 2007).
MATERIALS AND METHODS
Taxon and Character Sampling
Morphological data were collected for 60
species of Dactyloa, representing the sub-
groups aequatorialis, latifrons, laevis, Phe-
nacosaurus, punctatus, roquet, and tigrinus.
Anolis anoriensis (a recently described
species formerly considered part of A.
eulaemus) was treated as conspecific with
A. eulaemus given that the description of the
former was published after our data analy-
ses were performed. Six species were
included as outgroups: one non-Anolis
Polychrotinae (Polychrus marmoratus) and
five species representing different series of
non-Dactyloa Anolis (Anolis bimaculatus, A.
cuvieri, A. equestris, A. occultus, A. sagrei).
A total of 643 alcohol-preserved (66 species;
393 males, 250 females), 123 dry (49
species), and 10 cleared and stained (9
species) specimens were examined. Addi-
tional data were collected from radiographs
of 394 specimens (60 species). External
characters were scored for all 66 species,
and osteological characters were scored for
63 species (14 of which were only scored
from radiographs and thus lack data for all
cranial characters). The largest specimens
available were examined as a proxy for
including adult specimens only. All speci-
mens measured at least 70% of the maxi-
mum SVL reported in the literature for the
same sex and species (Williams and Acosta,
1996; Savage, 2002). A complete list of
specimens examined is given in the Supple-
mentary Appendix 1.1
Sixty-six morphological characters were
examined, including both continuous char-
acters (those that can be represented by real
numbers, e.g., tail length) and discrete
characters (those that can only be repre-
sented by integer values, including meristic
and presence/absence, e.g., number of
elongated superciliary scales). This data set
includes characters of external morphology
and osteology that have been previously
used in Anolis phylogenetic analyses, have
been regarded as diagnostic for Anolis
subgroups, or have been used historically
for species identification (Etheridge, 1959;
Williams, 1976b, 1989; de Queiroz, 1987;
Etheridge and de Queiroz, 1988; Frost and
Etheridge, 1989; Williams et al., 1995; Poe,
1998, 2004; Jackman et al., 1999; Brandley
and de Queiroz, 2004). Given that sexual
dimorphism occurs in many species of
Anolis (e.g., Schoener, 1969; Butler et al.,
2000, 2007), characters were scored for
both males and females and combined only
when t tests (for continuous characters) or
chi-square tests (for discrete characters)
revealed no significant difference between
the sexes or when tests could not be
performed because sample sizes were too
small. When significant differences were
found, only data from males were used.
However, given the small number of
specimens available as dry skeletons and
the absence of information on sex for
roughly one-third of them, data for charac-
ters examined on dry specimens were
combined without evaluating whether some
of the characters exhibit sexual dimorphism.
To ensure character independence, we
performed correlation tests between char-
acters. To remove the effects of correlation,
we estimated residuals by regressing each
variable against the correlated variable;
residuals were used in subsequent analyses.
In cases where a character was correlated
with several others (e.g., SVL, head length,
1 Supplementary material referenced in this paper is
available online at www.mcz.harvard.edu/Publications/.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 351
and head width), after residual estimation
between two of the variables a second
correlation test was performed to ensure
that the resulting residuals were not still
correlated with the other characters. A list
of the characters analyzed, including mea-
surement and coding details, is given in
Appendix I.
Character Coding
Continuous Characters. Continuous char-
acters were coded using two methods: the
gap-weighting method of Thiele (1993), and
Torres-Carvajal?s (2007) modification of
Wiens (2001) modification of Thiele?s meth-
od. In Thiele?s (1993) method, continuous
characters are coded into discrete values
while retaining information about order and
relative distance between states. Average
values per species were standardized by
calculating the natural logarithm (ln, or ln +
1 when zero average values were present) to
ensure equal variances; then each standard-
ized average value (x) was range-standard-
ized by applying the formula xs 5 [(x 2
min)/(max 2 min)] 3 (n 2 1), where min
and max are the minimum and maximum
among the standardized average values,
respectively, and n is the number of states
used. One hundred and one states (0?100,
n 5 101) were used in the parsimony
analyses (which allows capturing differences
of 0.01 between states) and six states (0?5, n
5 6) were used in the Bayesian analyses (in
that 6 is the maximum number of ordered
character states allowed in MrBayes
v.3.1.2). Use of the term (n 2 1) is a
modification of Thiele?s (1993) equation, in
which n was used incorrectly, because using
n will lead to the recognition of an
additional state (i.e., the total number of
states is one greater than the value of the
highest numbered state). Therefore, to
ensure a total of 101 (or 6) states (including
state ??0??), n 2 1 was used instead. Finally,
the resulting values were rounded to the
nearest integer and treated as states of a
multistate ordered character. Wiens (2001)
suggested a modification of Thiele?s method
using character state (step) matrices to
increase the number of character states
(then limited in PAUP* v.4.0b10 to 32 states
on 32-bit computers). In this method, the
term n of Thiele?s equation is replaced by
1,000 (the maximum cost in a step matrix in
PAUP*), and the difference between range-
standardized scores (xs) determines the cost
of transformation between the correspond-
ing states in the step matrix. Given that the
default cost of character state transforma-
tion in PAUP* is 1, this approach requires
weighting non-continuous characters by
1,000 to maintain equal weights among
characters. Torres-Carvajal (2007) suggest-
ed a modification of Wiens? approach in
which the term 1,000 in Wiens? equation is
replaced by 1; this practice results in step
matrices containing scores between 0 and 1
(rather than 0 and 1,000) and does not
involve reweighting the non-continuous
characters. Step matrices, in which trans-
formation costs between states are differ-
ences between these scores, were generated
in PAUP* as described by Torres-Carvajal
(2007).
Polymorphic Characters. Polymorphic
characters (including presence/absence and
meristic) were coded using two different
approaches: the MANOB approximation
(Manhattan distance, observed frequency
arrays) of the frequency parsimony method
described by Berlocher and Swofford
(1997), and the majority or modal condi-
tion. Berlocher and Swofford?s (1997) meth-
od was originally described for allele fre-
quency data (see also Swofford and Berlo-
cher, 1987) but has been applied to
polymorphic morphological characters
(Wiens, 2000; Brandley and de Queiroz,
2004; Torres-Carvajal, 2007). Under this
approach, each taxon with a unique combi-
nation of allele (character state) frequencies
is assigned a different character state, and
changes between states are assigned costs
equal to the Manhattan distances between
those states using step matrices, which are
analyzed under the parsimony criterion. In
the MANOB method, the reconstructed
hypothetical ancestors are required to have
352 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
a state (an array of allele [character state]
frequencies) from the pool of states ob-
served in the terminal taxa. Under the
alternative majority or modal method, a
polymorphic species is assigned the most
common state in the individuals examined.
The modal coding method was used despite
being outperformed by the frequency cod-
ing method (Wiens, 1995, 1998; Wiens and
Servedio, 1997) to perform Bayesian analy-
ses because, currently, the only models
available for morphological characters in
MrBayes do not allow unequal rates (anal-
ogous to differential parsimony costs)
among states. Cases with a 50:50 distribu-
tion of states were treated as partial
uncertainty; that is, from the subset of
states observed in a taxon, the software
assigns to the taxon the state that minimizes
length on a given tree.
Comparison of Coding Methods. To
compare alternative coding methods, we
estimated phylogenetic information content
using the g1 statistic (Fisher, 1930; Sokal
and Rohlf, 1995; Zar, 1999). The g1 statistic
measures the skewness of a distribution and
has been used to test for phylogenetic sig-
nal in data sets as part of the tree length
distribution skewness test (Hillis, 1991;
Huelsenbeck, 1991; Hillis and Huelsen-
beck, 1992). The test is based on the
observation that the shape of the distribu-
tion of tree lengths (for all possible trees, or
for a random subset when it is not feasible
to evaluate the lengths of all possible trees)
provides information about the presence of
phylogenetic signal in the data (Hillis,
1991). Data sets with phylogenetic signal
show a left-skewed distribution of lengths
(g1 , 0), which indicates that there are
fewer solutions near the best solution than
anywhere else in the distribution (Hillis and
Huelsenbeck, 1992). We evaluated phylo-
genetic signal in each of our data sets by
comparing the observed g1 values against a
distribution obtained from data with no
phylogenetic signal. To construct the null
distribution, we used Mesquite v.2.75
(Maddison and Maddison, 2011) to perform
1,000 random permutations (by shuffling
states within characters among taxa) of data
sets containing only those characters coded
with the method being tested. We calculat-
ed the g1 scores for the randomized
matrices by evaluating 10,000 random trees
in PAUP*. Coding methods whose g1 scores
fell outside the 95% confidence interval
were considered to have significant phylo-
genetic signal.
Besides testing for phylogenetic signal in
each data set, we evaluated differences in
amounts of phylogenetic signal among data
sets based on different coding methods. We
did this by directly comparing g1 values as
estimates of the amount of hierarchical
information recorded by each coding meth-
od. Values were compared between meth-
ods for coding continuous characters
(Thiele?s [1993] gap-weighting method with
101 character states versus Torres-Carvajal?s
[2007] version of the gap-weighting method
versus Thiele?s [1993] gap-weighting meth-
od with 6 character states) and between
methods for coding polymorphic characters
(frequency arrays using Manhattan distance
step matrices versus using modal condi-
tions). The g1 values were estimated in
PAUP* from data sets containing only those
characters coded with the method being
tested; g1 values for each method were
estimated from 10 samples of 500,000
random trees, and differences between the
g1 values were evaluated using t tests. To
further assess differences (or the lack
thereof) between the different coding
methods, we performed reciprocal topology
tests (Larson, 1998), in which for each data
set (or portion thereof), the optimal tree
inferred from that data set was compared
(using two-tailed Wilcoxon signed-ranks
tests) with the optimal trees inferred from
data sets based on alternative coding
methods.
Morphological Data Sets and
Phylogenetic Analyses
Based on the results of the statistical tests
comparing g1 values, we selected Torres-
Carvajal?s (2007) version of the gap-weight-
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 353
ing method for coding continuous charac-
ters and the frequency arrays using Man-
hattan distance step matrices for coding
polymorphic characters. This morphology-
only data set will be referred to as the
Torres-freq data set (See Supplementary
Appendix 2) and was used for all subsequent
parsimony analyses. A second morphology-
only data set was constructed with contin-
uous characters coded using Thiele?s (1993)
gap-weighting method with six character
states and polymorphic characters coded
using the modal condition. This data set,
hereafter referred to as Thiele6-mode (See
Supplementary Appendix 3), was specifical-
ly constructed to fulfill the requirements of
running analyses in MrBayes (a maximum of
six ordered character states and only rates
[analogous to costs] of 0 and 1). Both data
sets include 66 species and 66 characters
(33 external, 33 osteological; 20 continuous,
31 discrete polymorphic, and 15 discrete
non-polymorphic; Appendix I).
Parsimony analyses were performed on
the Torres-freq data set using PAUP*
(Swofford, 2002) with equal costs for state
transformations, except for continuous and
polymorphic characters (20 and 31 charac-
ters, respectively), in which differential
costs were implemented with step matrices
(see Supplementary Appendix 4), and for
discrete (non-continuous, non-polymorphic)
multistate ordered characters, which were
weighted so that the range of each character
equals 1. For each data set, a heuristic
search with 1,000 replicates of random
stepwise addition was performed, with all
additional options left on default settings.
Nodal support was assessed with non-
parametric bootstrap resampling (BS; Fel-
senstein, 1985) using 100 bootstrap pseu-
doreplicates and heuristic searches with 50
replicates of random stepwise addition
(remaining options were left on defaults)
for each bootstrap pseudoreplicate.
Bayesian analyses were performed on the
Thiele6-mode data set in MrBayes (Ron-
quist and Huelsenbeck, 2003) using the
Mkv and Mkv + rate variation (rv) models
for morphological data (Lewis, 2001). The
Mkv model is analogous to the Jukes Cantor
(JC) model of molecular sequence evolution,
which assumes equal state frequencies, equal
transformation rates between states, and
equal rates among characters. The Mkv +
rv model allows rate heterogeneity among
characters using the symmetric Dirichlet
(for multistate characters) and beta (for
binary characters) distributions, in which
one parameter determines the shape of the
distribution of rates. This approach is similar
to using the gamma (C) distribution to model
rate variation among sites in molecular
sequence data. Four independent runs?
each with four Markov chains, a random
starting tree, and default heating settings?
were run for 10 million generations. Trees
were sampled with a frequency of one every
1,000 generations. The first 25% of the trees
were discarded as the ??burn-in?? phase.
Stationarity in the post-burn-in sample was
confirmed following the same procedures
outlined in Castan?eda and de Queiroz
(2011). The maximum clade credibility tree
(i.e., the tree with the highest product of
posterior clade probabilities) was obtained
using the TreeAnnotator package of BEAST
v.1.6.2 (Drummond and Rambaut, 2007).
Bayesian posterior clade probabilities (PP)
were calculated based on the post-burn-in
sample of trees for all four independent runs
combined. Nodes with posterior probabili-
ties greater than 0.95 were considered
strongly supported, with the precaution that
PP might overestimate clade support, espe-
cially in short internodes (Suzuki et al., 2002;
Alfaro et al., 2003; Lewis et al., 2005). Bayes
Factors (Kass and Raftery, 1995; Pagel and
Meade, 2005) were used to compare the
results obtained with the Mkv and Mkv + rv
models for the morphological characters.
Combined Data Sets and
Phylogenetic Analyses
The same two morphological data sets used
in the morphology-only analyses were com-
bined with the DNA sequence data (40
species of Dactyloa and six outgroup species)
from Castan?eda and de Queiroz (2011). Two
354 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
species for which DNA sequence data were
available, Anolis sp1 and A. sp2, were
excluded from our combined data sets
because of the absence of morphological data.
DNA sequence data included three gene
regions: 1) the mitochondrial NADH dehy-
drogenase subunit II (ND2), five transfer
RNAs (tRNATrp, tRNAAla, tRNAAsn, tRNACys,
tRNATyr), and the origin for light strand
replication (OL, ,1,500 bases); 2) a fragment
of the mitochondrial cytochrome oxidase
subunit I (COI, ,650 bases), and 3) the
nuclear recombination activating gene (RAG-
1, ,2,800 bases). The combined data sets
included 60 species of Dactyloa, 20 of which
were missing molecular data, and 6 outgroup
species (for a total of 66 species). Hereafter,
the combined data sets will be referred to as
CombTorres-freq and CombThiele6-mode,
respectively. Parsimony analysis of the com-
bined data set was run under the same
conditions used for the morphology-only data
sets. For the Bayesian analysis, the data were
partitioned into morphological and molecular
data. The latter were further partitioned
(based on the results of Castan?eda and de
Queiroz [2011]) by gene region and, within
each region, by codon position and tRNAs
(ND2: four partitions; COI: three partitions,
RAG-1: three partitions). The model of
evolution for each molecular partition was
selected based on the Akaike Information
Criterion (AIC) as implemented in Modeltest
(Posada and Crandall, 1998) v.3.7. For the
morphological partition, the model of evolu-
tion was selected based on the Bayes factor
scores from the morphology-only analyses
(see above). Bayesian analyses were run under
the same conditions used for the morphology-
only analyses, except that the number of runs
was increased to five and the number of
generations was increased to 50 million to
ensure that stationarity and convergence
between chains were achieved.
Tests of Phylogenetic Hypotheses
We tested hypotheses concerning the
monophyly of: 1) Dactyloa including Phena-
cosaurus, 2) Dactyloa excluding Phenaco-
saurus (as traditionally circumscribed), 3)
subgroups described based on morphological
characters for which we had adequate taxon
samples: aequatorialis, latifrons (as tradition-
ally circumscribed), Phenacosaurus (as
traditionally circumscribed), punctatus, ro-
quet, and tigrinus, and 4) the clades inferred
by Castan?eda and de Queiroz (2011) based
on molecular data: eastern, latifrons, Phena-
cosaurus, roquet, and western. Given that we
obtained data for only one species of the
laevis series, no tests were performed
regarding the monophyly of that series.
Phylogenetic hypotheses were tested using
parsimony-based (Templeton, 1983) and
Bayesian (Larget and Simon, 1999; Huelsen-
beck et al., 2001) topological tests. The data
sets (morphology-only and combined) with
polymorphic characters coded using Torres-
Carvajal?s (2007) method and continuous
characters coded with frequency arrays using
Manhattan distance step matrices were
selected to perform the parsimony-based
tests. For the Bayesian tests, the data sets
(morphology-only and combined) with con-
tinuous characters coded using Thiele?s
(1993) method with six character states and
polymorphic characters coded using the
modal condition were used.
For the parsimony-based tests, optimal
trees resulting from parsimony analyses of
the morphology-only and combined data
sets were compared with optimal trees
resulting from parsimony analyses of the
same data sets incorporating each hypothe-
sis as a topological constraint (performed
using the same search conditions as for the
unconstrained analyses). In cases in which
the hypothesis of interest (e.g., previously
recognized taxa based on morphological
data or clades inferred based on molecular
data) was obtained in the optimal uncon-
strained trees, the alternative hypothesis of
non-monophyly was tested. Topologies cor-
responding to the hypotheses of interest
were constructed using MacClade (Maddi-
son and Maddison, 2001) v.4.07 and im-
ported into PAUP* as topological con-
straints. Monophyly constraints were used
for the hypotheses of previously recognized
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 355
taxa based on morphological characters,
because all species included in the tests
had been previously assigned to a group.
Backbone constraints were used for the
hypotheses of clades recognized based on
molecular data, because the species for
which only morphological data were avail-
able had not previously been assigned to any
of the clades. Wilcoxon signed-ranks (WSR)
tests (Templeton, 1983) were used to
determine whether the optimal uncon-
strained tree is significantly different from
the hypothesis corresponding to the con-
straint (Larson, 1998) and were performed
as two-tailed tests in PAUP*. In the
Bayesian tests, trees contained in the 95%
credible set of trees from the post-burn-in
sample for each data set were loaded into
PAUP* and filtered based on topological
constraints corresponding to each hypothe-
sis. Topologies that were not present within
the 95% credible set of trees (i.e., those that
resulted in no trees retained under a given
topological constraint) were considered
rejected by the test.
RESULTS
Comparisons Between Coding Methods
All the methods resulted in characters with
significant phylogenetic signal (as assessed by
g1) when compared with randomly permuted
data (P , 0.001 for all coding methods). Of
the methods used to code continuous char-
acters, Torres-Carvajal?s (2007) version of the
gap-weighting method resulted in the most
left-skewed distributions (g1 values most
negative; g1520.3016 0.004), indicating that
this method yielded characters that contain the
most phylogenetic information. The method of
Thiele (1993), with 101 character states,
followed (g15 20.269 6 0.003), and the same
method with 6 character states resulted in the
least skewed distributions (g15 20.253 6
0.003). Comparing the coding methods used
for polymorphic characters, the frequency
arrays using Manhattan distance step matrices
resulted in larger negative g1 values (g15
20.175 6 0.004) than did the modal condition
method (g15 20.156 6 0.004). Statistical (t)
tests indicated significant differences between
all coding methods for both continuous (P ,
0.001 for all comparisons) and polymorphic
(P , 0.001) characters in the amount of
phylogenetic information recorded.
Phylogenetic Analyses
In all Bayesian analyses, the average
standard deviation of split frequencies of
converging chains reached values lower than
0.06, and the potential scale reduction factor
(PSRF) of all runs combined reached 1.0 for
most parameters. Bayes factors (BF) favored
the Mkv + rv model over the Mkv model (BF
5 194.46) in the analyses of morphology-
only data sets, although the majority-rule
consensus tree inferred using the simpler
Mkv model was more resolved and contained
more moderately to strongly supported
nodes. With the Mkv model, 31 nodes were
resolved, 14 of which had moderate (PP $
0.75) to strong (PP $ 0.95) support (PP 5
0.77?0.99; tree not shown); with the Mkv + rv
model, 24 nodes were resolved, 12 of which
had moderate support (PP 5 0.76?0.94; tree
not shown). In the Bayesian analyses of the
combined data set, the five independent runs
did not all converge onto the same likelihood
values; instead, three runs converged onto a
lower negative natural log likelihood score, and
the remaining two converged onto a higher
score. However, the relationships among
species in the majority-rule topologies result-
ing from the two sets of runs were very similar,
differing only in one poorly supported node.
Nodal support and substitution model param-
eter values were also very similar, except for
the rate variation among sites (alpha) and the
rate multiplier (m) for several partitions. For
this reason, two of the five runs were discarded,
and only the three with lower negative natural
log likelihood scores were used for tree
estimation and hypotheses testing.
Morphology-only Data Sets. The parsi-
mony analysis (Torres-freq data set) yielded
a single fully resolved most parsimonious
tree of 466.63 steps (CI 5 0.22, RI 5 0.53;
Fig. 1). In this tree, Dactyloa is not inferred
356 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
to be monophyletic, and non-Dactyloa
Anolis outgroup species are located in three
different places (two within Dactyloa).
These results are poorly supported (BS 5
0%), and only two small, deeply nested
clades in the entire tree are moderately
supported (BS 5 75?81%). All species
previously placed in the genus Phenaco-
saurus were included in a clade (BS 5 0%)
that also contained A. microtus and A.
proboscis (traditionally placed in the lati-
frons and laevis series, respectively). The
Phenacosaurus clade, as defined in Casta-
n?eda and de Queiroz (2011), which includes
all the Phenacosaurus species sampled in
their study except A. neblininus, was
inferred with the addition of A. tetarii (for
which no molecular data are available) with
low nodal support (BS 5 44%). The
Bayesian analysis (Thiele6-mode data set)
under the Mkv + rv model, resulted in a
fully resolved but poorly supported maxi-
mum clade credibility tree (P PP 5 1.497
3 10212; Fig. 2). Dactyloa was not inferred
to be monophyletic, and non-Dactyloa
Anolis outgroup species were located in
four different places in the tree (all within
Dactyloa). Species previously placed in the
genus Phenacosaurus, except A. carlostoddi
and A. neblininus, formed a paraphyletic
group at the base of the tree. In both
parsimony and Bayesian analyses, neither
the series based on morphological charac-
ters nor the clades based on molecular data
(except the Phenacosaurus clade in the
parsimony analyses) were inferred.
Combined Data Sets. The parsimony
analysis (CombTorres-freq data set) yielded
a single fully resolved tree of 10,431.86 steps
(CI 5 0.30, RI 5 0.43; Fig. 3). The
Bayesian analysis (CombThiele6-mode data
set) resulted in a fully resolved maximum
clade credibility tree (P PP 5 1.449 3 1028;
Fig. 4). In both analyses, Dactyloa was
inferred to be monophyletic with low to
moderate support (BS 5 51%, PP 5 0.81).
Eleven unambiguous morphological syna-
pomorphies support the monophyly of Dac-
tyloa (Supplementary Appendix 5); however,
this interpretation should be made with
caution because it is most likely the result of
biased outgroup sampling. For example, three
out of five outgroup species have very large
body sizes (maximum male SVL . 123 mm;
Williams and Acosta, 1996) compared with
most Anolis species, and as a result, a decrease
in maximum male SVL is inferred as a
synapomorphy of Dactyloa. In the parsimony
analysis, the major clades inferred by Casta-
n?eda and de Queiroz (2011)?that is, eastern,
latifrons, Phenacosaurus, roquet, and west-
ern?were inferred with weak to strong nodal
support (BS 5 6?93%). Similarly, in the
Bayesian analysis, all five clades were in-
ferred with weak to strong nodal support
(0.15 , PP , 0.97). For the purpose of
assigning species to these clades (eastern,
latifrons, roquet, Phenacosaurus, and west-
ern), the clades were delimited using nodes
bounded by species for which molecular data
were available (e.g., the eastern clade is
defined as the clade originating with the last
common ancestor of a particular set of
species inferred from molecular data [Cas-
tan?eda and de Queiroz, 2011], thus excluding
species outside that node that are more
closely related to the eastern clade than to
any of the other four mutually exclusive
clades). In both parsimony and Bayesian
analyses, the same sets of additional species,
for which only morphological data were
available, were included in the western and
Phenacosaurus clades. In the case of the
latifrons and eastern clades, different sets of
additional species (for which only morpho-
logical characters were available) were in-
ferred in the parsimony and Bayesian
analyses. No additional species were includ-
ed in the roquet clade in either analysis. In
the following paragraphs, the species com-
position of each clade is detailed, with
daggers ({) indicating species lacking molec-
ular data. The synapomorphies that support
each clade, inferred based on the parsimony
analysis, are given in Supplementary Appen-
dix 5.
The western clade was inferred with weak
to moderate support in the parsimony and
Bayesian analyses (BS 5 25%, PP 5 0.82;
Figs. 3, 4) and is supported by nine
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 357
Figure 1. Most parsimonious tree inferred with the Torres-freq morphology-only data set (TL 5 466.63, CI 5 0.22, RI 5 0.53).
Bootstrap support (BS) values are shown above branches; missing values above branches indicate BS 5 0%. The traditional
species groups/series based on morphological characters (see text for details) are differentiated by color. One major Dactyloa
subclade, of those described based on molecular data (see text for details), is indicated on the right.
358 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
Figure 2. Bayesian maximum clade credibility tree inferred with the Thiele6-mode morphology-only data set using the Mkv + rv
model. Bayesian posterior probabilities are shown above branches. The traditional species groups/series based on morphological
characters (see text for details) are differentiated by color.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 359
Figure 3. Most parsimonious tree inferred with the CombTorres-freq combined data set (TL 5 10,431.86, CI 5 0.30, RI 5 0.43).
Bootstrap support (BS) values are shown above branches; missing values above branches indicate BS 5 0%. Daggers ({)
following species names indicate the species for which only morphological data were available. The traditional species groups/
series based on morphological characters (see text for details) are differentiated by color. Major Dactyloa subclades described
based on molecular data (see text for details) are indicated on the right. The Dactyloa clade is indicated with a black dot on the
corresponding node.
360 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
morphological characters (Supplementary
Appendix 5). In both analyses, this clade is
composed of the same 10 species: A.
aequatorialis, A. antioquiae{, A. chloris, A.
eulaemus, A. fasciatus{, A. festae, A. gem-
mosus, A. megalopithecus{, A. peraccae, and
A. ventrimaculatus. Within this clade, the
topologies are largely congruent, with the
exception of the position of A. gemmosus
and the internal relationships within the
clade composed of A. chloris, A. fasciatus,
A. festae, and A. peraccae. Additionally, in
the parsimony analysis, A. boettgeri is
inferred with weak support as the sister
group of the western clade (BS 5 18%),
whereas in the Bayesian analysis, the sister
taxon of the western clade is the clade (A.
boettgeri, A. huilae) (PP 5 0.49).
The latifrons clade was inferred in the
parsimony analysis with low nodal support
(BS 5 20%; Fig. 3) and is supported by five
morphological characters (Supplementary
Appendix 5); it is composed of 13 species:
A. agassizi, A. apollinaris{, A. casildae, A.
chocorum, A. danieli, A. fraseri, A. frenatus,
A. insignis, A. maculigula, A. microtus, A.
latifrons{, A. princeps, and A. purpures-
cens{. In the Bayesian analysis, the latifrons
clade was inferred with low support (PP 5
0.46; Fig. 4) and is composed of the same
set of species as the parsimony analysis with
the addition of A. squamulatus. Three
mutually exclusive subclades were inferred
by both analyses: (A. agassizi (A. microtus,
A. insiginis)) (BS 5 32%, PP 5 0.61), (A.
casildae, A. maculigula) (BS 5 92%, PP 5
0.62), and (A. frenatus (A. latifrons, A.
princeps) (BS 5 80%, PP 5 0.72). In both
analyses, A. philopunctatus was inferred as
the sister taxon of the latifrons clade.
The eastern clade was inferred in the
parsimony analysis (Fig. 3) with low nodal
support (BS 5 6%) and is supported by
seven morphological characters (Supple-
mentary Appendix 5); it is composed of
eight species: A. anatoloros, A. carlostoddi{,
A. jacare, A. orcesi{, A. punctatus, A.
tigrinus, A. transversalis, and A. vaupesia-
nus{. In the Bayesian analysis (Fig. 4), the
eastern clade was inferred with low nodal
support (PP 5 0.15) with 11 species: A.
anatoloros, A. dissimilis{, A. jacare, A.
menta{, A. punctatus, A. ruizii{, A. santa-
martae{, A. solitarius{, A. tigrinus, A.
transversalis, and A. vaupesianus{. Despite
the differences in species composition, two
subclades were inferred in both analyses
(Figs. 3, 4): (A. transversalis (A. punctatus,
A. vaupesianus)) (BS 5 61%, PP 5 0.71)
and (A. anatoloros, A. jacare) (BS 5 69%,
PP 5 0.94).
The roquet clade was inferred with strong
support in both parsimony and Bayesian
analyses (BS 5 93%, PP 5 0.97; Figs. 3, 4)
and is supported by 16 morphological
characters (Supplementary Appendix 5). In
both analyses, this clade is composed of
the same eight species: A. aeneus, A. bo-
nairensis, A. extremus, A. griseus, A. luciae,
A. richardii, A. roquet, and A. trinitatis. The
topology within this clade is identical for
both phylogenetic analyses, with all nodes
moderately to strongly supported (BS $
72%, PP $ 0.96).
The Phenacosaurus clade was inferred
with moderate support in both parsimony
and Bayesian analyses (BS 5 87%; PP 5
0.84; Figs. 3, 4) and is supported by 19 mor-
phological characters (Supplementary Ap-
pendix 5). In both analyses, this clade is
composed of the same six species, A.
euskalerriari, A. heterodermus, A. inderenae,
A. nicefori, A. tetarii{, and A. vanzolinii, all
previously placed in the genus Phenaco-
saurus. The relationships within this clade
are identical between parsimony and Bayes-
ian analyses; however, in the parsimony
analysis, A. proboscis is inferred as its sister
group (BS 5 58%), whereas in the Bayesian
analysis, A. orcesi is inferred as its sister
group with moderate support (PP 5 0.87),
and A. proboscis is the sister group to that (A.
orcesi, Phenacosaurus) clade (PP 5 0.43).
In both parsimony and Bayesian analyses,
nine species, A. boettgeri{, A. calimae, A.
caquetae{, A. fitchi, A. huilae, A. neblininus,
A. philopunctatus{, A. podocarpus, and A.
proboscis{, were not included in any of the
five major clades within Dactyloa when those
clades are treated as originating in the last
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 361
Figure 4. Bayesian maximum clade credibility tree inferred with the CombThiele6-mode combined data set. Bayesian posterior
probabilities (PP) are shown above branches; asterisks (*) indicate PP 5 1.0. Daggers ({) following species names indicate the
species for which only morphological data were available. The traditional species groups/series based on morphological
characters (see text for details) are differentiated by color. Major Dactyloa subclades described based on molecular data (see text
for details) are indicated on the right. The Dactyloa clade is indicated with a black dot on the corresponding node.
362 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
common ancestors of the species for which
molecular data were available (see above).
However, A. boettgeri{, A. fitchi, A. huilae,
and A. podocarpus were consistently placed
closer to the western clade than to any of the
four other major clades; A. philopunctatus{
was consistently placed closer to the latifrons
clade, and A. proboscis{ was consistenty
placed closer to Phenacosaurus. Additionally,
the positions of A. carlostoddi{, A. dissi-
milis{, A. menta{, A. orcesi{, A, ruizii{, A.
santamartae{, A. solitarius{, and A. squamu-
latus{ were inconsistent (particularly relative
to the five major clades) between parsimony
and Bayesian analyses.
Tests of Phylogenetic Hypotheses
The WSR and Bayesian tests performed
on the morphology-only data sets (Torres-
freq and Thiele6-mode, respectively) yielded
very different results (Table 1): the WSR
test failed to reject the monophyly of
Dactyloa and each of the subgroups
previously described based on morpholog-
ical characters: aequatorialis, latifrons,
punctatus, roquet, tigrinus, and Phenaco-
saurus; in contrast, the Bayesian test
rejected the monophyly of Dactyloa and all
of the previously recognized subgroups
except the roquet series. Of the hypotheses
tested, only the monophyly of Dactyloa
excluding Phenacosaurus was rejected by
both the WSR and Bayesian tests. When
testing the clades inferred based on molecu-
lar data (Castan?eda and de Queiroz, 2011)
with the morphological data, contradictory
results between the parsimony and Bayesian
approaches were found again (Table 1):
monophyly of the eastern, latifrons, roquet,
and western clades was not rejected with the
TABLE 1. RESULTS OF THE WILCOXON SIGNED RANKS (WSR) AND BAYESIAN (B) TESTS OF PHYLOGENETIC HYPOTHESES OF PREVIOUSLY RECOGNIZED
TAXA BASED ON MORPHOLOGICAL CHARACTERS (TOP) AND OF CLADES INFERRED BASED ON MOLECULAR DATA (BOTTOM) ON THE BASIS OF MORPHOLOGICAL
DATA ONLY (TORRES-FREQ, THIELE6-MODE). FOR THE TORRES-FREQ DATA SET, DIFFERENCES BETWEEN TREE LENGTHS OF UNCONSTRAINED ANALYSES AND
THOSE CONSTRAINED TO CORRESPOND TO EACH TESTED HYPOTHESIS (DTL) AND WSR P-VALUES ARE GIVEN. FOR THE BAYESIAN TESTS OF THE THIELE6-
MODE DATA SET, THE PRESENCE (+) OR ABSENCE (2) OF THE ALTERNATIVE TOPOLOGY IN THE 95% CREDIBLE SET OF TREES IS SHOWN. SIGNIFICANT RESULTS
ARE INDICATED WITH AN ASTERISK (*).
Test Hypothesis
Dataset
Torres-freq Thiele6-mode
DTL WSR P-Value B
Traditional groups
Dactyloa 6.86 0.280 2*
Dactyloa excluding Phenacosaurusa 13.44 0.001* 2*
aequatorialis series 0.99 0.697 2*
latifrons seriesb 7.33 0.131 2*
punctatus series 9.32 0.332 2*
roquet series 0.84 0.851 +
tigrinus series 2.89 0.175 2*
Phenacosaurusa 2.33 0.629 2*
Groups based on molecular data
Eastern clade 2.58 0.468 2*
latifrons clade 8.42 0.145 2*
Phenacosaurus clade n/ac n/ac +
Phenacosaurus clade not monophyletic 0.44 0.827 n/ac
roquet clade 0.84 0.851 +
Western clade 1.05 0.969 2*
a As traditionally circumscribed, which includes (of the species sampled) A. carlostoddi, A. euskalerriari, A. heterodermus, A. inderenae, A. neblininus, A. nicefori, A.
orcesi, A. tetarii, and A. vanzolinii.
b As traditionally circumscribed, which includes (of the species sampled) A. apollinaris, A. casildae, A. danieli, A. fraseri, A. frenatus, A. insignis, A. latifrons, A.
microtus, A. princeps, A. purpurescens, and A. squamulatus.
c Not applicable: the hypothesis in question was present in the optimal (unconstrained) tree(s), so the alternative hypothesis (monophyly or non-monophyly) was
tested instead.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 363
WSR test, but it was rejected?except in the
case of the roquet series?with the Bayesian
test. The parsimony analysis indicated mono-
phyly of the Phenacosaurus clade, but the
WSR test failed to reject its non-monophyly;
conversely, the Bayesian analysis indicated
non-monophyly of the group, but the Bayes-
ian test failed to reject its monophyly.
With the combined data sets, the WSR
and Bayesian tests yielded mostly congruent
results concerning taxa recognized previously
on the basis of morphological characters
(Table 2). The non-monophyly of Dactyloa
(given that this hypothesis was inferred in the
optimal tree) and the monophyly of the
tigrinus series and Phenacosaurus were not
rejected by either test. In contrast, both tests
rejected the monophyly of the aequatorialis,
latifrons, and punctatus series. The non-
monophyly of the roquet series (a clade
inferred in both parsimony and Bayesian
optimal trees), was rejected by the WSR test,
but not by the Bayesian test. The monophyly
of Dactyloa excluding Phenacosaurus was
rejected by both the WSR and Bayesian tests.
The clades inferred based on molecular
data (Castan?eda and de Queiroz, 2011) were
also present in the parsimony and Bayesian
optimal trees of the combined data sets;
therefore, the non-monophyly of these groups
was tested. Results obtained with the WSR
and Bayesian tests differed in most cases
(Table 2): the WSR test rejected the hypoth-
esis of non-monophyly of the Phenacosaurus
clade and failed to reject the non-monophyly
of the eastern, latifrons, roquet, and western
clades. In contrast, the Bayesian tests rejected
the non-monophyly of all these clades.
DISCUSSION
The objectives of this study were to
reconstruct the phylogeny of the Dactyloa
clade based on morphological characters
alone and in combination with molecular
data, to explore different coding methods for
continuous and polymorphic characters that
TABLE 2. RESULTS OF THE WILCOXON SIGN RANKS (WSR) AND BAYESIAN (B) TESTS OF PHYLOGENETIC HYPOTHESES OF PREVIOUSLY RECOGNIZED
TAXA BASED ON MORPHOLOGICAL CHARACTERS (TOP) AND OF CLADES INFERRED BASED ON MOLECULAR DATA (BOTTOM) ON THE BASIS OF COMBINED
MORPHOLOGICAL AND MOLECULAR DATA (COMBTORRES-FREQ, COMBTHIELE6-MODE). FOR THE COMBTORRES-FREQ DATA SET, DIFFERENCES BETWEEN
TREE LENGTHS OF UNCONSTRAINED ANALYSES AND THOSE CONSTRAINED TO CORRESPOND TO EACH TESTED HYPOTHESIS (DTL) AND WSR P-VALUES ARE
GIVEN. FOR THE BAYESIAN TESTS OF THE COMBTHIELE6-MODE DATA SET, THE PRESENCE (+) OR ABSENCE (2) OF THE ALTERNATIVE TOPOLOGY IN THE 95%
OF CREDIBLE SET OF TREES IS SHOWN. SIGNIFICANT RESULTS ARE INDICATED WITH AN ASTERISK (*).
Test Hypothesis
Dataset
CombTorres-freq CombThiele6-mode
DTL WSR P-Value B
Groups based on morphological data
Dactyloa not monophyletic 1.47 0.712 +
Dactyloa excluding Phenacosaurusa 44.83 0.003* 2*
aequatorialis series 135.33 ,0.001* 2*
latifrons seriesb 83.91 ,0.001* 2*
punctatus series 170.43 ,0.001* 2*
roquet series not monophyletic 35.09 0.016* +
tigrinus series 4.82 0.336 +
Phenacosaurus groupa 21.35 0.253 +
Groups based on molecular data
Eastern clade not monophyletic 17.63 0.366 2*
latifrons clade not monophyletic 25.56 0.054 2*
Phenacosaurus clade not monophyletic 59.81 ,0.001* 2*
roquet clade not monophyletic 25.11 0.077 2*
Western clade not monophyletic 9.90 0.687 2*
a As traditionally circumscribed, which includes (of the species sampled) A. carlostoddi, A. euskalerriari, A. heterodermus, A. inderenae, A. neblininus, A. nicefori, A.
orcesi, A. tetarii, and A. vanzolinii.
b As traditionally circumscribed, which includes (of the species sampled) A. apollinaris, A. casildae, A. danieli, A. fraseri, A. frenatus, A. insignis, A. latifrons, A.
microtus, A. princeps, A. purpurescens, and A. squamulatus.
364 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
were part of our data set, and to test
hypotheses of monophyly of previously
described taxa. In the following paragraphs,
we discuss the advantages and disadvantages
of the different coding methods used, the
phylogenetic relationships inferred, and their
implications regarding previously recognized
taxa. Finally, based on our findings, we
propose a new taxonomy that recognizes
only monophyletic taxa and in which names
are defined following the rules of PhyloCode.
Differences Among Coding Methods
For continuous characters, the coding
method of Torres-Carvajal (2007) resulted
in characters containing the largest amount
of phylogenetic signal, followed by Thiele?s
(1993) method using 101 character states.
The main disadvantage of Torres-Carvajal?s
(2007) method is that it results in a significant
increase of computation time compared with
Thiele?s (1993) method (MRC, personal
observation), presumably because it uses
step matrices. Although Thiele?s (1993)
method discretizes continuous characters, it
maintains information on order and magni-
tude of change between states. Therefore, its
implementation using 101 character states
(i.e., allowing a 0.01 resolution between
states) might be sufficient to approximate a
continuous distribution (particularly if the
values from the continuous distribution are
estimated at a similar level of precision).
Despite significant differences in g1 values,
reciprocal WSR tests (Larson, 1998) indicate
that phylogenetic inferences between the
two methods (at least for the Dactyloa data
set) are not strongly in conflict: tests for
differences between the optimal trees result-
ing from only continuous characters under
each of the two coding methods (i.e., Thiele?s
[1993] gap-weighting method with 101
character states or Torres-Carvajal?s [2007]
step matrix modification of it) using the data
sets produced by each of the coding methods
were not statistically significant (P 5 0.872
using the Thiele-coded data set; P 5 0.391
using the Torres-coded data set). In contrast,
reciprocal tests between the optimal trees
resulting from only continuous characters
coded with Thiele?s (1993) method using 101
states compared with using 6 states were
statistically significant (P 5 0.036 using the
data set with 101 states; P 5 0.023 using the
data set with 6 states). Similarly, reciprocal
tests between the optimal trees resulting
from only continuous characters coded with
Torres-Carvajal?s method versus Thiele?s
(1993) method using six states were also
statistically different (P 5 0.040 using the
Torres-coded data set; P 5 0.046 using the
Thiele-coded data set). These results com-
bined with the results based on the g1
statistic indicate that a larger number of
character states significantly increases the
amount of phylogenetic information record-
ed by Thiele?s coding method. Moreover,
these findings suggest that Thiele?s (1993)
method, when implemented using a large
number of character states, may be an
effective alternative to fully continuous
coding methods. Despite performing more
poorly according to g1 values, it did not yield
a significantly different tree according to
reciprocal tests. At least in this case, the loss
of information appears to be small and is
compensated by lesser computational re-
quirements.
The polymorphic characters coded as
frequency arrays using Manhattan distance
step matrices were found, based on g1
values, to contain significantly more phylo-
genetic signal than those coded as standard
binary or multistate characters and scored
using modal conditions. Reciprocal tests
indicate that there are significant differenc-
es between the optimal trees resulting from
data sets including only polymorphic char-
acters coded using these two methods (P ,
0.001 using the frequency arrays?coded
data set; P , 0.001 using the standard
coding with modes data set). This result
further supports previous studies on empir-
ical (Wiens, 1995, 1998) and simulated
(Wiens and Servedio, 1997) data, showing
that methods that incorporate frequency
information outperform, based on accuracy
measurements, other methods for analyzing
polymorphic characters (including scoring
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 365
modal conditions). The advantages of fre-
quency methods include making use of more
phylogenetic information and reducing the
effects of sampling errors, in that the
probability of being mislead by the presence
or absence of states occurring at low frequen-
cies is reduced, which is particularly impor-
tant with small sample sizes (Swofford and
Berlocher, 1987; Wiens, 1995; Wiens and
Servedio, 1997). The main disadvantage of
some of the methods incorporating frequency
information is that they significantly increase
the computation time required because of the
use of step matrices (e.g., Wiens, 2000; MRC,
personal observation).
Phylogeny of Dactyloa
This study presents the phylogenetic
relationships of Dactyloa based on molecu-
lar data for 40 species and morphological
data for the same 40 species and 20
additional ones (for a total of 60 species).
This represents a substantial improvement
upon previous studies, which included a
maximum of 42 species (Castan?eda and de
Queiroz, 2011), 2 of which were not
included in the current study (see ??Materi-
als and Methods??) for molecular data only,
or a maximum of 28 species (Poe, 2004) for
multiple data sources. For the combined
data sets, Dactyloa was inferred to be
monophyletic, provided that it includes the
Phenacosaurus species, in agreement with
previous studies (Poe, 1998, 2004; Jackman
et al., 1999; Nicholson et al., 2005; Casta-
n?eda and de Queiroz, 2011), although its
monophyly was not strongly supported
according to topology tests employing con-
straint trees (the hypothesis of non-mono-
phyly was not rejected, though monophyly
of Dactyloa excluding Phenacosaurus was
rejected). Previous analyses of the entire
Anolis clade based on combined data, with a
similar set of characters and coding meth-
ods, inferred a fully resolved but poorly
supported Dactyloa (Poe, 2004, fig. 2),
although the analysis of only the morpho-
logical component of this data set did not
infer Dactyloa (Poe, 2004, fig. 5). In
contrast, previous analyses based solely on
molecular data strongly supported the
monophyly of Dactyloa (Nicholson et al.,
2005; Castan?eda and de Queiroz, 2011).
The considerable difference in nodal
support between the combined (morpho-
logical and molecular) and molecular-only
analyses could derive from intrinsic charac-
teristics of the morphological characters.
For example, the morphological characters
might be highly homoplastic, introducing
support for conflicting groupings within the
morphological data set, or they might have
low phylogenetic information content, allow-
ing multiple placements of taxa for which
only morphological data are avail-
able. Additionally, it is possible that conflicts
between the phylogenetic signal of the
morphological and molecular data sets result
in a reduction in nodal support. In our
analyses, we excluded characters commonly
used in studies of morphological conver-
gence (e.g., limb and tail length) to avoid this
potential bias (but see de Queiroz [1996,
2000] and Poe [2005] for the advantages of
including these characters in phylogenetic
reconstruction). Trees inferred with the
morphology-only data set showed a general
lack of support for any particular topology
(regardless of the coding method used),
which suggests that the morphological data
might not have sufficient phylogenetic signal
to produce any strong conflict with the
molecular data (and therefore strongly affect
nodal support). However, reciprocal WSR
topology tests comparing the tree inferred
from morphological data (Fig. 1) with that
inferred based on molecular data (Castan?eda
and de Queiroz, 2011, fig. 1A) indicated
strong disagreement between the two data
sets (P # 0.0001 for each case). Therefore,
the difference in nodal support between the
molecular and combined analyses would
seem to result, in this case, from multiple
placements of at least some of the species
that were scored for morphological charac-
ters only, as well as the larger total number of
species within the Dactyloa clade (so that
support is distributed among a larger num-
ber of nodes).
366 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
Castan?eda and de Queiroz (2011) in-
ferred, based on molecular data, five strongly
supported clades (informally named eastern,
latifrons, Phenacosaurus, roquet, and west-
ern) with coherent geographic distributions.
Based on the combined data, we inferred
those same five clades with the inclusion of
additional species for which only morpho-
logical data are available. In agreement with
the inferred relationships of those species, in
all cases, their geographic distributions lie
within or on the periphery of the clades in
which they were placed. The discussion that
follows concerns the composition and inter-
nal relationships of the five clades and is
based exclusively on the results obtained
with the combined analyses. Following
Castan?eda and de Queiroz (2011), we also
adopt (in the following discussion) delimita-
tions of the clades based on the species for
which molecular data were available.
The western clade, as delimited by
Castan?eda and de Queiroz (2011), included
seven species (A. aequatorialis, A. anorien-
sis, A. chloris, A. festae, A. gemmosus, A.
peraccae, and A. ventrimaculatus) distribut-
ed in the western and central cordilleras of
Colombia, the western slopes of the Ecua-
dorian Andes, and the pacific lowlands of
Colombia and Ecuador. In this study, the
western clade was inferred to include three
additional species (considering that we
treated A. anoriensis as conspecific with A.
eulaemus; see ??Materials and Methods??): A.
antioquiae, A. fasciatus, and A. megalopithe-
cus, distributed in the northernmost part of
the western cordillera in Colombia (A.
antioquiae and A. megalopithecus) and in
the Pacific lowlands of central Ecuador (A.
fasciatus). Two primary subclades were
inferred within the western clade. The first
includes four species, A. chloris, A. fascia-
tus, A. festae, and A. peraccae, all previously
placed in the punctatus series (Savage and
Guyer, 1989) or species group (Williams,
1976b), that have a humid forest distribu-
tion below 1,000 m above sea level and
small to moderate body size (max SVL 5 62,
72, 55, and 52 mm, respectively [Williams
and Acosta, 1996]). The second subclade
includes six species, A. aequatorialis, A.
antioquiae, A. eulaemus, A. gemmosus, A.
megalopithecus, and A. ventrimaculatus, all
previously placed in the aequatorialis series
(Savage and Guyer, 1989) or species group
(Williams, 1976b, 1985; Williams and Duell-
man, 1984; Rueda Almonacid, 1989), dis-
tributed from 1,500 to 2,000 m above sea
level and with moderate to large body size
(max SVL 5 92, 72 [MRC, personal
observation], 101, 66, 81, and 80 mm,
respectively [Williams and Acosta, 1996]).
In both parsimony and Bayesian optimal
trees (Figs. 3, 4), Anolis boettgeri, A. fitchi,
A. huilae, and A. podocarpus were inferred
closer to the western clade than to any of
the other five major clades. Castan?eda and
de Queiroz (2011) also inferred this close
relationship for the last three of those
species based on molecular data. Anolis
boettgeri and A. huilae were previously
included in the punctatus series (Williams,
1976b; Poe et al., 2008), whereas A. fitchi
and A. podocarpus were included in the
aequatorialis series (Williams, 1976b; Ayala-
Varela and Torres-Carvajal, 2010). The
geographic distributions of these four spe-
cies at mid to high elevations in the eastern
slopes of the Andes of Colombia (A. huilae),
Ecuador (A. fitchi, A. podocarpus), and
Peru (A. boettgeri), do not correspond with
the Pacific lowland and Colombian inter-
Andean valley distribution of the western
clade and suggest that a dispersal or
vicariance event was associated with the
branch separating the eastern and western
species (i.e., the one at the base of the
western clade).
The latifrons clade, as delimited by
Castan?eda and de Queiroz (2011), included
12 species (A. agassizi, A. casildae, A.
chocorum, A. danieli, A. fraseri, A. frenatus,
A. insignis, A. maculigula, A. microtus, A.
princeps, A. sp1, and A. sp2) distributed in
the Pacific lowlands of Costa Rica, Panama,
Colombia (including Malpelo island), and
Ecuador and in the Colombian inter-
Andean valleys below 1,000 m above sea
level (except A. danieli, which ranges from
1,700 to 2,200 m). Most of these species
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 367
were previously placed in the latifrons
species group (Williams, 1976b) or series
(Savage and Guyer, 1989), and in all except
A. chocorum, adult males reach a SVL
greater than 100 mm (large size was consid-
ered a diagnostic feature of the traditional
latifrons series [Williams, 1976b], also called
the giant mainland anoles [Dunn, 1937]). We
inferred the latifrons clade (excluding A. sp1
and A. sp2, which were not included in this
study, see ??Materials and Methods??), with the
inclusion of three additional species in the
parsimony analysis, A. apollinaris, A. latifrons,
and A. purpurescens, and further including A.
squamulatus in the Bayesian analysis. All four
potentially additional species were previously
included in the latifrons series or species group
(Williams, 1976b; Savage and Guyer, 1989)
and have Pacific lowland (A. latifrons and A.
purpurescens) or inter-Andean (A. apollinaris)
distributions, except A. squamulatus (see
below). In A. apollinaris, A. latifrons, and A.
squamulatus adult male maximum SVL ex-
ceeds 100 mm (106, 133, and 122 mm,
respectively; Williams and Acosta, 1996;
Ugueto et al., 2009); A. purpurescens, which
is known from a small number of specimens,
appears to be smaller (max SVL 5 78 mm;
Williams and Acosta, 1996). In both analyses,
A. philopunctatus is inferred as the sister group
of the latifrons clade. However, this species
was not considered as part of the latifrons clade
based on the delimitation of the clade using
species for which molecular data were available
(see above). This species was previously
included in the punctatus series and is
distributed in the Brazilian Amazon, and its
adult maximum SVL 5 73 mm (Rodrigues,
1988). Anolis squamulatus was previously
placed in the latifrons species group of
Williams (1976b) and series of Savage and
Guyer (1989) based on its large dewlap, small
head scales, uniform dorsal scales, and large
body size (max adult male SVL . 100 mm;
Williams and Acosta, 1996). Although inferred
as part of the latifrons clade in the Bayesian
analysis (Fig. 4), it was not in the parsimony
analysis (Fig. 3). Moreover, A. squamulatus is
the only species, of the sampled taxa, previ-
ously placed in the latifrons species group or
series that was not inferred as part of the
latifrons clade in the parsimony analysis.
However, the geographic distribution of A.
squamulatus, in the cloud forests of the
northern Venezuelan Andes, does not corre-
spond to the Pacific lowland and Colombian
inter-Andean valley distribution of the latifrons
clade or to the Pacific mid and low elevations in
Colombia and Ecuador distribution of the
western clade (to which it was inferred as being
closer in the parsimony analysis). Instead, it
corresponds more closely to the geographic
distribution of the eastern clade. Either of
these alternative relationships of A. squamula-
tus (within or closer to either the western or
the eastern clades) would require the conver-
gent evolution of large body size with members
of the latifrons clade. In agreement with
previous studies suggesting the close relation-
ship between A. frenatus, A. latifrons, and A.
princeps and including the possibility that
those three taxa represent a single species
(Savage and Talbot, 1978; Williams, 1988;
Castan?eda and de Queiroz, 2011), we inferred
the clade ((A. frenatus (A. latifrons, A.
princeps)) in both parsimony and Bayesian
analyses.
The eastern clade, as delimited by
Castan?eda and de Queiroz (2011), included
five species (A. anatoloros, A. jacare, A.
punctatus, A. transversalis, and A. tigrinus)
distributed in the northern portion of the
eastern cordillera of Colombia into the
Venezuelan Andes and the Amazon region.
In our parsimony results, the eastern clade
was inferred to include three additional
species: A. carlostoddi, A. orcesi, and A.
vaupesianus, whereas in the Bayesian anal-
ysis, it was inferred to include six additional
species: A. dissimilis, A. menta, A. ruizii, A.
santamartae, A. solitarius, and A. vaupesia-
nus. All of the potential additional species
have an eastern Andean and Amazonian
distribution. They occur in Amazonia (A.
vaupesianus, A. dissimilis), the eastern
slopes of the northern Andes of Ecuador
(A. orcesi), the eastern cordillera of Colom-
bia (A. ruizii), the Sierra Nevada de Santa
Marta in Colombia (A. menta, A. santamar-
tae, A. solitarius), and the Chimanta? tepui in
368 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
Venezuela (A. carlostoddi). Within the
eastern clade, two subclades were inferred
in both analyses: the first includes three
species, A. anatoloros, A. jacare, and A.
tigrinus, with mostly Andean, high-elevation
distributions and smaller body size (max
male SVL 5 55?68 mm; Williams and
Acosta, 1996; Ugueto et al., 2007); the
second subclade includes three species, A.
punctatus, A. transversalis, and A. vaupe-
sianus, with Amazonian, low-elevation dis-
tributions, and larger size (max male SVL 5
76?82 mm; Williams and Acosta, 1996). It is
not surprising that A. vaupesianus (distrib-
uted in the Vaupes and Amazonas depart-
ments in Colombia and known only from
the type series) was consistently inferred as
the sister taxon of A. punctatus (with a
broad Amazonian distribution). These two
species were described as close relatives
that differ primarily in the size and degree
of keeling of the ventral scales (smaller and
weakly keeled in A. vaupesianus versus larger
and strongly keeled in A. punctatus), the
dewlap coloration in preservative (black skin
with white scales in A. vaupesianus versus
light skin with small dark spots and purplish
scales in A. punctatus) and the dorsal color
pattern of preserved specimens (in A.
vaupesianus, ??brown, strongly blotched with
darker, dorsal blotches tending to form
transverse series across the back?? versus an
unpatterned dark dorsum in A. punctatus)
(Williams, 1982: 8). The unique color pattern
of A. vaupesianus is only found in one of the
paratypes (UTA 6850), whereas the colora-
tion of the other specimens in the type series
is not particularly different from that of A.
punctatus (Williams, 1982: 8?9). Given the
small differences separating these two spe-
cies, a more comprehensive sampling of A.
punctatus in Colombia (currently ,10 spec-
imens are known) and the collection of
additional molecular data (particularly for
A. vaupesianus) could clarify whether these
two taxa are conspecific as well as whether
the characters used to distinguish them
represent extremes in a continuous distribu-
tion or are based on an atypical specimen. In
contrast to the likely close relationship
between A. punctatus and A. vaupesianus,
the relationships of A. carlostoddi and A.
orcesi (parsimony) or A. dissimilis, A. menta,
A. santamartae, and A. solitarius (Bayesian)
to A. tigrinus are less clear. The former two
species were previously placed in Phenaco-
saurus and thus not considered closely
related to A. tigrinus. In contrast, two of
the latter species, A. menta and A. solitarius,
were placed in the tigrinus series (Williams,
1976b; Ayala et al., 1984) (the other two, A.
dissimilis and A. santamartae, were placed in
the punctatus series). In both cases, the
putative clade formed by all three or all six
species has low support, and given the low
resolving power of the morphological data set
and the fact that molecular data are available
for neither A. carlostoddi and A. orcesi nor A.
dissimilis, A. menta, A. santamartae, A.
ruizii, and A. solitarius, the inferred rela-
tionships are questionable.
The roquet clade, as delimited by Casta-
n?eda and de Queiroz (2011), included eight
species (A. aeneus, A. bonairensis, A.
extremus, A. griseus, A. luciae, A. richardii,
A. roquet, and A. trinitatis), distributed in
the southern Lesser Antilles, from Martini-
que to Grenada, as well as the islands of
Bonaire and Tobago (with introduced pop-
ulations in Trinidad and Guyana [Gorman
and Dessauer, 1965, 1966; Gorman et al.,
1971]). This clade corresponds to the
previously described roquet species group
or series (Underwood, 1959; Gorman and
Atkins, 1967, 1969; Lazell, 1972; Williams,
1976a; Savage and Guyer, 1989; Creer et al.,
2001). One additional species, A. blanquilla-
nus, from the island of La Blanquilla, was
previously referred to the roquet series
(Williams, 1976a) but was not included in
our analyses; however, its inclusion in the
roquet clade is supported by allozyme data
(Yang et al., 1974; Creer et al., 2001). Poe
(2004) inferred the roquet clade (BS 5 74%,
fig. 2), supported by six morphological
characters (see ??Current Taxonomy within
Dactyloa??); two of those, greater sexual size
dimorphism and an increase in the number of
postmental scales, were also inferred as
synapomorphies for the clade in this study.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 369
The Phenacosaurus clade, as delimited by
Castan?eda and de Queiroz (2011), included
five species (A. euskalerriari, A. heteroder-
mus, A. inderenae, A. nicefori, and A.
vanzolinii) distributed in the Andean re-
gions of Colombia, Ecuador, and Venezuela,
all of which were previously placed in the
genus Phenacosaurus (Lazell, 1969; Barros
et al., 1996). In this study, the Phenaco-
saurus clade was inferred with the addition
of A. tetarii, a species that was previously
placed in the genus Phenacosaurus (Barros
et al., 1996) and whose geographic distribu-
tion (Venezuelan Andes) conforms to that of
the clade. Three other species that were
previously referred to Phenacosaurus, A.
carlostoddi, A. orcesi, and A. neblininus,
were not inferred to be part of this clade in
the parsimony analysis. However, in the
Bayesian analysis, A. orcesi was inferred as
the sister group of the Phenacosaurus clade
with moderate support (PP 5 0.87; Fig. 4).
In contrast, in the parsimony analysis, A.
orcesi was placed in the eastern clade (as
was A. carlostoddi), although with weak
support (BS 5 6%, 24%, and 19%). In both
parsimony and Bayesian analyses, A. pro-
boscis was inferred as sister group of the
Phenacosaurus clade (or of the Phenaco-
saurs clade plus A. orcesi) with weak
support (BS 5 58%, PP 5 0.43), a
relationship also inferred by Poe (2004, fig.
2). A close relationship between A. probos-
cis and species traditionally referred to
Phenacosaurus was also inferred by Poe et
al. (2009b, 2012). The geographic distribu-
tions of both species lie on the periphery of
that of the Phenacosaurus clade: A. orcesi is
distributed along the eastern slopes of the
northern Andes of Ecuador, whereas A.
proboscis is distributed along the western
slopes of the northern Andes of Ecuador.
The more deeply nested species within the
Phenacosaurus clade (A. heterodermus, A.
inderenae, A. tetarii, and A. vanzolinii) are
larger in size (max male SVL576, 98, 86,
and 104 mm, respectively [Williams and
Acosta, 1996]) and correspond to the
heterodermus group of Williams et al.
(1996); the two earlier diverging lineages
(A. euskalerriari and A. nicefori) are smaller
in size (max male SVL 5 53 and 63 mm,
respectively [Williams and Acosta, 1996]).
Except for A. euskalerriari, all of the species
in the Phenacosaurus clade have heteroge-
neous dorsal scales, suggesting that this
condition originated in the ancestral lineage
of A. heterodermus and A. nicefori after it
diverged from that of A. euskalerriari.
Anolis carlostoddi, A. orcesi, and A. nebli-
ninus, which were previously placed in the
genus Phenacosaurus but were not inferred
in this study to be part of the Phenacosaurus
clade (although A. orcesi was inferred to be
closely related in the Bayesian analysis), also
lack heterogeneous dorsal scales.
Seventeen species (A. boettgeri, A. cali-
mae, A. caquetae, A. carlostoddi, A. dissim-
ilis, A. fitchi, A. huilae, A. menta, A.
neblininus, A. orcesi, A. philopunctatus, A.
podocarpus, A. proboscis, A. ruizii, A.
santamartae, A. solitarius, and A. squamula-
tus) were not consistently placed in any of
the five mutually exclusive clades just
discussed, either because their positions did
not satisfy the criterion based on the last
common ancestor of the species for which
molecular data were available or because
they differed across phylogenetic methods
(and were commonly poorly supported).
However, 6 of the 17 species (A. boettgeri,
A. fitchi, A. huilae, A. philopunctatus, A.
podocarpus, and A. proboscis) were each
consistently placed closer to one of the five
clades than to the others. For those species
whose relationships differed among analyses
(A. calimae, A. caquetae, A. carlostoddi, A.
dissimilis, A. menta, A. neblininus, A. orcesi,
A. ruizii, A. santamartae, A. solitarius, and A.
squamulatus), 9 out of 11 of which currently
lack molecular data (all but A. calimae and A.
neblininus), more data will be necessary to
clarify their relationships within Dactyloa.
Previously Recognized Taxa
None of the traditionally recognized sub-
groups of Dactyloa based on morphological
characters (aequatorialis, latifrons, Phenaco-
saurus, punctatus, roquet, and tigrinus) were
370 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
inferred in the optimal trees inferred from
either the morphology-only or the combined
data sets except the roquet series in the
combined analyses. The parsimony-based
topology tests (WSR) using the morpholo-
gy-only data set failed to reject the mono-
phyly of Dactyloa or any of the previously
described subgroups (Table 1); therefore,
despite the morphological data not support-
ing any of these groups when analyzed under
parsimony, it is also unable to reject any of
them using parsimony-based tests. In con-
trast, the Bayesian tests using the morphol-
ogy-only data set rejected the hypotheses of
monophyly of Dactyloa and all traditionally
recognized series except the roquet series.
With the combined data sets and both
parsimony and Bayesian tests (2), the aequa-
torialis, latifrons, and punctatus series were
also strongly rejected, but the tigrinus series
and Phenacosaurus were not rejected.
Contradictory evidence and absence of
support for the series described based on
morphological characters is consistent with
previous suggestions that morphological
characters used for series delimitation might
show a high degree of convergence and
parallelism (Williams, 1976b: 260) and that
some series were described only for conve-
nience (Williams, 1979: 10). Furthermore,
traditional series delimitation did not dis-
tinguish clearly between ancestral and
derived conditions, and the former are not
indicative of close phylogenetic relation-
ships. Differences between the results of
WSR and Bayesian tests might reflect the
conservativeness of the WSR?the require-
ment of a stronger signal to reject a given
hypothesis (Lee, 2000)?because Bayesian
tests often rejected hypotheses when WSR
tests did not, but WSR tests rarely rejected
hypotheses not rejected by Bayesian tests.
Differences between the data sets (i.e.,
resulting from alternative coding methods)
do not appear to be the reason for the
different results between the two tests. For
one thing, the Bayesian tests rejected more
of the hypotheses despite using the data set
(Thiele6-mode) that contained the least
phylogenetic signal. For another, WSR tests
performed on the Thiele6-mode data set
(the data set used for the Bayesian tests)
yielded the same qualitative results as with
the Torres-freq data set (results not shown).
Proposed Taxonomy
Our results indicate that a revised taxon-
omy for Dactyloa is warranted. Optimal
phylogenetic trees and topology tests indi-
cate that most of the previously recognized
taxa within Dactyloa based on morphological
characters and traditionally ranked as species
groups or series are not monophyletic.
Moreover, our previously published results
based on molecular data (Castan?eda and de
Queiroz, 2011) indicate the existence of five
well-supported major subclades, and the
results of the combined analyses of morpho-
logical and molecular data in the present
study both corroborate the monophyly and
clarify the composition of those subclades.
Here, we propose a revised taxonomy based
on the results of our phylogenetic analyses,
including names that are defined explicitly in
terms of phylogenetic relationships (de
Queiroz and Gauthier, 1990, 1992).
The optimal topologies obtained from the
different (parsimony versus Bayesian) meth-
ods are in substantial but not complete
agreement. We considered it inappropriate
to select one topology over the other
because each topology has advantages and
disadvantages: The parsimony tree is based
on a character coding method that incorpo-
rates more phylogenetic information,
whereas the Bayesian tree is based on more
realistic evolutionary models. Therefore, we
used a consensus tree as the basis for our
proposed taxonomy (Fig. 5). Specifically, we
used a pruned and regrafted2 consensus tree
(Gordon, 1980; Finden and Gordon, 1985;
Bryant, 2003) derived from the most
parsimonious tree (Fig. 3) and the maxi-
mum clade credibility tree (Fig. 4) for the
combined morphological and molecular
2 The term ??grafted?? seems more appropriate here,
given that the branch has not been grafted before;
however, we use ??regrafted?? because it has been
commonly used in the literature.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 371
Figure 5. Pruned and regrafted consensus tree based on the most parsimonious tree (inferred with the CombTorres-freq data
set; Fig. 3) and the Bayesian Maximum Clade Credibility tree (inferred with the CombThiele6-mode data set; Fig. 4) with the
taxonomy proposed in this study. Daggers ({) following species names indicate the species for which only morphological data
were available. Regrafted branches are indicated by a break near their bases. See text for details concerning alternative prunings
and regraftings 1) within the clade composed of A. chloris, A. fasciatus, A. festae, and A. peraccae; 2) of A. purpurescens versus
A. danieli; and 3) of members of the punctatus series. The Dactyloa clade is indicated with a black dot on the corresponding node.
Five mutually exclusive, informally named clades (??series??) within Dactyloa are distinguished by color, with lighter versions of the
372 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
data sets (i.e., CombTorres-freq and
CombThiele6-mode matrices).
Because a few of the species (e.g., A.
carlostoddi, A. orcesi, A. squamulatus) have
very different relationships on the primary
trees, the strict and semistrict consensus
trees had little resolution, and we therefore
originally intended to use an Adams consen-
sus tree (for a review of consensus methods,
see Swofford [1991]). However, the Adams
consensus tree contained unexpected groups
that we felt could not be justified on the basis
of the primary trees (e.g., A. squamulatus
closer to the western clade than to the
latifrons clade, which is contradicted by the
Bayesian tree; Fig. 4), and it turned out that
the pruned and regrafted consensus tree
exhibited the desired properties we had
incorrectly attributed to the Adams consen-
sus method (i.e., placement of species with
conflicting relationships within the smallest
possible clade in the consensus tree for
which there is complete agreement concern-
ing their higher level relationships in the
primary trees).
To generate the pruned and regrafted
consensus tree, we first produced agree-
ment subtrees (also called common pruned
trees; Finden and Gordon, 1985) using
PAUP* with identical topologies that result
from removing (pruning) the same set of
taxa from the primary trees (Finden and
Gordon, 1985). We obtained 12 largest
agreement subtrees for the 60 Dactyloa
species (including only one outgroup spe-
cies, Polychrus marmoratus, to root the
trees), which differed only in the inclusion
of all possible combinations of two species
from the clade composed of A. festae, A.
chloris, A. fasciatus, and A. peraccae (six
possible combinations) and the inclusion of
either A. danieli or A. purpurescens. We
arbitrarily selected one of the 12 largest
agreement subtrees (the one including A.
fasciatus, A. peraccae, and A. purpurescens)
as the base tree for the regrafting process.
In the second step, we manually reattached
each previously pruned species or set of
species to the node representing the most
recent common ancestor of its alternative
placements on the primary trees. Because
the position of the entire eastern clade and
its possible relatives differed between the
two primary trees (closer to the latifrons
clade in the parsimony tree versus closer to
the roquet clade in the Bayesian tree), all of
those species were excluded from the
largest agreement subtrees. To determine
whether those species should be reattached
singly or in sets, we determined the largest
agreement subtree for those species (A.
anatoloros, A. caquetae, A. carlostoddi, A.
dissimilis, A. jacare, A. menta, A. orcesi,
A. punctatus, A. ruizii, A. santamartae, A.
solitarius, A. tigrinus, A. transversalis, A.
vaupesianus), plus one representative each
of the western, latiforns, Phenacosaurus,
and roquet clades; we also included Poly-
chrus marmoratus to root the trees. We
obtained four largest agreement subtrees
representing two different topologies for the
set of species making up the eastern clade
and its possible relatives (the other two
topologies differed only in whether the
representative of the western or the roquet
clade was included). Both topologies in-
cluded (A. transversalis (A. punctatus, A.
vaupesianus)), but one included (A. tigrinus
(A. anatoloros, A. jacare)) and A. caquetae,
whereas the other included ((A. dissimilis,
A. santamartae) (A. menta, A. solitarius)).
We selected the former for grafting because
those groups were consistent, after pruning,
with the primary trees, whereas the latter
depended on pruning the representative of
the latifrons clade, which we intend to be a
fixed point of reference (i.e., not a candidate
for pruning) in this secondary analysis
(given that several members of the latifrons
clade were present in all of the primary
r
same hue indicating tentative assignment to the clade represented by that hue. These informally named clades have the same
name as some of the groups traditionally ranked as series within Anolis; however, species composition is not necessarily identical.
A black bar across a branch indicates an apomorphy used to define the clade name above the bar.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 373
agreement subtrees). The resulting pruned
and regrafted consensus tree (Fig. 5) serves
as the basis for our taxonomy.
In all but two cases, we have selected
preexisting names that have been applied
traditionally to groups of species approxi-
mating, to one degree or another, the clades
to which we apply them. Three of the
selected names are similar in appearance to
genus names; however, they are here tied to
clades rather than to the rank of genus and
are implicitly ranked below the genus level
(given that we use the name Anolis in the
binomina of all of the species included in
the named clades). Five of the selected
names combine the species name (epithet)
of the first described species (e.g., ??roquet??)
with the name of a rank (i.e., ??series??).
Because those names violate the rule stating
that clade names must be single words
beginning with a capital letter (ICPN,
Article 17.1), they are treated here as
informal names. They are nevertheless
given explicit phylogenetic definitions as
guides for applying the names in the context
of future phylogenetic hypotheses. Despite
being given explicit phylogenetic defini-
tions, the names in question are compatible
with traditional nomenclature, in that they
have been defined so as to ensure that they
will always refer to mutually exclusive taxa
(see de Queiroz and Donoghue, 2013), as
would names associated with the rank of
series under traditional nomenclature. As a
consequence, the ??series?? names are ap-
plied to clades that are more inclusive than
the five clades based on the species for
which molecular data were available that
formed the basis of our discussion in the
??Phylogeny of Dactyloa?? section (above),
though those five less inclusive clades form
the cores of the ??series?? clades. The number
of Dactyloa subclades named in our taxono-
my exceeds five, the number of well-
supported clades based on molecular data
(Castan?eda and de Queiroz, 2011) and
corroborated in this study, because in two
cases, we considered it useful to name
additional clades associated with the origins
of distinctive apomorphies.
Dactyloa Wagler 1830, converted clade
name
Definition (branch-modified node-
based): The crown clade originating in the
most recent common ancestor of Anolis
punctatus Daudin 1802 and all extant
species that share a more recent common
ancestor with A. punctatus than with A.
bimaculatus (Sparrman 1784), A. cuvieri
Merrem 1820, A. equestris Merrem 1820, A.
occultus Williams and Rivero 1965, and A.
sagrei Dume?ril and Bibron 1837. Refer-
ence phylogeny: Figure 5, this study (see
also Poe, 2004, figs. 2?4). Inferred com-
position: Anolis aeneus Gray 1840, A.
aequatorialis Werner 1894, A. agassizi
Stejneger 1900, A. anatoloros Ugueto,
Rivas, Barros, Sa?nchez-Pacheco and Gar-
c??a-Pe?rez 2007, A. anchicayae Poe, Velasco,
Miyata and Williams 2009 (inclusion based
on inferred close relationship to A. peraccae
following Poe et al., 2009b), A. anoriensis
Velasco, Gutie?rrez-Ca?rdenas and Quintero-
Angel 2010 (inclusion based on inferred
close relationship to A. aequatorialis follow-
ing Castan?eda and de Queiroz, 2011), A.
antioquiae Williams 1985, A. apollinaris
Boulenger 1919, A. bellipeniculus (Myers
and Donnelly 1996) (inclusion based on
inferred close relationship to A. neblininus
following Myers and Donnelly, 1996), A.
blanquillanus Hummelinck 1940 (inclusion
based on inferred close relationship to A.
bonairensis following Yang et al., 1974), A.
boettgeri Boulenger 1911, A. bonairensis
Ruthven 1923, A. calimae Ayala, Harris and
Williams 1983, A. caquetae Williams 1974,
A. carlostoddi (Williams, Praderio and
Gorzula 1996), A. casildae Arosemena,
Iba?n?ez, and de Sousa 1991, A. chloris
Boulenger 1898, A. chocorum Williams
and Duellman 1967, A. cuscoensis Poe,
Yan?ez-Miranda and Lehr 2008 (inclusion
based on the results of Poe et al., 2008), A.
danieli Williams 1988, A. deltae Williams
1974 (inclusion based on inferred close
relationship to A. dissimilis following Wil-
liams, 1974), A. dissimilis Williams 1965, A.
eulaemus Boulenger 1908, A. euskalerriari
(Barros, Williams, and Viloria 1996), A.
374 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
extremus Garman 1887, A. fasciatus Bou-
lenger 1885, A. festae Peracca 1904, A. fitchi
Williams and Duellman 1984, A. fraseri
Gu?nther 1859, A. frenatus Cope 1899, A.
gemmosus O?Shaughnessy 1875, A. gorgo-
nae Barbour 1905 (inclusion based on
inferred close relationship to A. andianus
[a synonym of A. gemmosus according to
Williams and Duellman (1984)] following
Barbour, 1905), A. griseus Garman 1887, A.
heterodermus Dume?ril 1851, A. huilae
Williams 1982, A. ibanezi Poe, Latella, Ryan
and Schaad 2009 (inclusion based on
inferred close relationship to A. chocorum
following Poe et al., 2009a), A. inderenae
(Rueda and Herna?ndez-Camacho 1988), A.
insignis Cope 1871, A. jacare Boulenger
1903, A. kunayalae Hulebak, Poe, Iba?n?ez
and Williams 2007 (inclusion based on
inferred close relationship to A. mirus
following Hulebak et al., 2007), A. laevis
(Cope 1876) (inclusion based on inferred
close relationship to A. proboscis following
Williams, 1979), A. lamari Williams 1992
(inclusion based on inferred close relation-
ship to A. tigrinus following Williams,
1992), A. latifrons Berthold 1846, A. luciae
Garman 1887, A. maculigula Williams 1984,
A. megalopithecus Rueda Almonacid 1989,
A. menta Ayala, Harris and Williams 1984,
A. microtus Cope 1871, A. mirus Williams
1963 (inclusion based on inferred close
relationship to A. fraseri following Williams,
1963), A. nasofrontalis Amaral 1933 (inclu-
sion based on inferred close relationship to
A. tigrinus following Amaral, 1933), A.
neblininus (Myers, Williams and McDiar-
mid 1993), A. nicefori (Dunn 1944), A.
nigrolineatus Williams 1965, A. orcesi (La-
zell 1969), A. otongae Ayala-Varela and
Velasco 2010 (inclusion based on inferred
close relationship to A. gemmosus following
Ayala-Varela and Velasco, 2010), A. para-
vertebralis Bernal Carlo and Roze 2005
(inclusion based on inferred close relation-
ship to A. solitarius following Bernal Carlo
and Roze, 2005), A. parilis Williams 1975
(inclusion based on inferred close relation-
ship to A. mirus following Williams, 1975),
A. peraccae Boulenger 1898, A. philopunc-
tatus Rodrigues 1988, A. phyllorhinus
Myers and Carvalho 1945 (inclusion based
on inferred close relationship to A. puncta-
tus following Myers and Carvalho, 1945), A.
podocarpus Ayala-Varela and Torres-Carva-
jal 2010, A. princeps Boulenger 1902, A.
proboscis Peters and Orce?s 1956, A. pro-
pinquus Williams 1984 (inclusion based on
inferred close relationship to A. apollinaris
following Williams, 1988), A. pseudotigrinus
Amaral 1933 (inclusion based on inferred
close relationship to A. tigrinus following
Amaral, 1933), A. punctatus Daudin 1802,
A. purpurescens Cope 1899, A. richardii
Dume?ril and Bibron 1837, A. roquet (Bon-
naterre 1789), A. ruizii Rueda and Williams
1986, A. santamartae Williams 1982, A.
soinii Poe and Yan?ez-Miranda 2008 (inclu-
sion based on inferred close relationship to
A. transversalis following Poe and Yan?ez-
Miranda, 2008), A. solitarius Ruthven 1916,
A.squamulatus Peters 1863, A. tetarii (Bar-
ros, Williams, and Viloria 1996), A. tigrinus
Peters 1863, A. transversalis Dume?ril 1851,
A. trinitatis Reinhardt and Lu?tken 1862, A.
umbrivagus Bernal Carlo and Roze 2005
(inclusion based on inferred close relation-
ship to A. solitarius following Bernal Carlo
and Roze, 2005), A. vanzolinii (Williams,
Orce?s, Matheus, and Bleiweiss 1996), A.
vaupesianus Williams 1982, A. ventrimacu-
latus Boulenger 1911, and A. williamsmit-
termeierorum Poe and Yan?ez-Miranda 2007
(inclusion based on inferred close relation-
ship to A. orcesi following Poe and Yan?ez-
Miranda, 2007). Comments: The name
Dactyloa was previously used by Savage
and Guyer (1989) for a taxon ranked as a
genus containing all species referred to the
Dactyloa clade in this study except those
species formerly assigned to the genus
Phenacosaurus Barbour 1920 (A. bellipeni-
culus, A. carlostoddi, A. euskalerriari, A.
heterodermus, A. inderenae, A. neblininus,
A. nicefori, A. orcesi, A. tetarii, A. vanzoli-
nii). Here, the name Dactyloa is not
associated with the rank of genus (it is
implicitly associated with a lower rank) so
that the binomina of the included species
retain the prenomen (genus name) Anolis.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 375
The following named clades are subclades
of Dactyloa.
aequatorialis series Savage and Guyer
1989, informal clade name
Definition (branch-modified node-
based): The crown clade originating in the
most recent common ancestor of Anolis
aequatorialis Werner 1894 and all extant
species that share a more recent common
ancestor with A. aequatorialis than with A.
latifrons Berthold 1846, A. punctatus Dau-
din 1802, A. roquet (Bonnaterre 1789), and
A. heterodermus Dume?ril 1851. Reference
phylogeny: Figure 5, this study. Inferred
composition: Anolis aequatorialis Werner
1894, A. anoriensis Velasco, Gutie?rrez-
Ca?rdenas and Quintero-Angel 2010 (see
??Comments??), A. antioquiae Williams
1985, A. boettgeri Boulenger 1911 (see
??Comments??), A. chloris Boulenger 1898,
A. eulaemus Boulenger 1908, A. fasciatus
Boulenger 1885, A. festae Peracca 1904, A.
fitchi Williams and Duellman 1984 (see
??Comments??), A. gemmosus O?Shaughnessy
1875, A. huilae Williams 1982 (see ??Com-
ments??), A. megalopithecus Rueda Almona-
cid 1989, A. peraccae Boulenger 1898, A.
podocarpus Ayala-Varela and Torres-Carva-
jal 2010 (see ??Comments??), and A. ventri-
maculatus Boulenger 1911. Other species
that may belong to the aequatorialis series
are A. anchicayae Poe, Velasco, Miyata and
Williams 2009, A. mirus Williams 1963, A.
otongae Ayala-Varela and Velasco 2010, and
A. parilis Williams 1975 (see ??Comments??).
Comments: Referral of A. anoriensis to this
clade is based on the results of Castan?eda
and de Queiroz (2011). In the context of
the reference phylogeny, six species not
traditionally included in the aequatorialis
series (e.g., Savage and Guyer, 1989) are
included in the aequatorialis series as
conceptualized here: A. boettgeri, A.
chloris, A. fasciatus, A. festae, A. huilae,
and A. peraccae. There is strong evidence
supporting inclusion of A. chloris, A. festae,
and A. peraccae (Castan?eda and de
Queiroz, 2011), and A. fasciatus is placed
consistently in a subclade with those three
species (Figs. 3, 4). In contrast, because of
inconsistent placement of A. huilae be-
tween analyses (Fig. 3 versus Fig. 4), be-
cause of weak support for the inclusion of
both A. boettgeri and A. huilae and because
molecular data are currently lacking for A.
boettgeri, inclusion of those two species in
the aequatorialis series should be consid-
ered tentative. Additionally, A. maculigula,
a species included in the traditional cir-
cumscription of the aequatorialis series, is
excluded based on strong evidence sup-
porting its inclusion in the latifrons series
(see below). The inclusion of A. fitchi and
A. podocarpus, traditionally included in the
aequatorialis series (Savage and Guyer,
1989), should also be considered tentative
given the weak support for the relevant
relationships (Figs. 3, 4). Although the
parsimony analysis (Fig. 3) places A. squa-
mulatus in the aequatorialis series, we have
tentatively retained that species in
the latifrons series based on the results of
the Bayesian analysis (Fig. 4) and its large
body size (see ??Comments?? on the latifrons
series). The inclusion of A. mirus, A.
otongae, and A. parilis, previously consid-
ered members of the aequatorialis series
(Williams, 1975; Ayala-Varela and Velasco,
2010), should also be considered tentative
given the current absence of these species
from explicit phylogenetic analyses. Anolis
anchicayae, inferred as closely related to A.
peraccae in a recent phylogenetic analysis
(Poe et al., 2009b), should also be consid-
ered tentatively included in the aequator-
ialis series given that in that analysis these
two species were not inferred as close
relatives of A. aequatorialis.
The aequatorialis series as conceptualized
here corresponds approximately to the
western clade of Castan?eda and de Queiroz
(2011). However, the aequatorialis series is
more inclusive than the western clade of
Castan?eda and de Queiroz (2011) in that it
appears to include A. boettgeri, A. fitchi, A.
huilae, and A. podocarpus and might also
include some species currently considered
incertae sedis within Dactyloa or absent
from explicit phylogenetic analyses if they
376 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
are found to be more closely related to A.
aequatorialis than to A. latifrons, A. punc-
tatus, A. roquet, and A. heterodermus.
Additionally, species in the western clade
of Castan?eda and de Queiroz (2011) are
characterized by having a cohesive western
Andean geographic distribution, but A.
boettgeri, A. fitchi, A. huilae, and A.
podocarpus, and possibly other species in
the more inclusive aequatorialis series, do
not conform to this geographic pattern (see
??Phylogeny of Dactyloa?? above for more
details about the geographic distributions of
the four species mentioned).
latifrons series Gorman and Dessauer
1966, informal clade name
Definition (branch-modified node-
based): The crown clade originating in the
most recent common ancestor of Anolis
latifrons Berthold 1846 and all extant
species that share a more recent common
ancestor with A. latifrons than with Anolis
aequatorialis Werner 1894, A. punctatus
Daudin 1802, A. roquet (Bonnaterre 1789),
and A. heterodermus Dume?ril 1851. Refer-
ence phylogeny: Figure 5, this study.
Inferred composition: Anolis agassizi
Stejneger 1900, A. apollinaris Boulenger
1919, A. casildae Arosemena, Iba?n?ez, and
de Sousa 1991, A. chocorum Williams and
Duellman 1967, A. danieli Williams 1988, A.
fraseri Gu?nther 1859, A. frenatus Cope
1899, A. insignis Cope 1871, A. kunayalae
Hulebak, Poe, Iba?n?ez and Williams 2007, A.
latifrons Berthold 1846, A. maculigula
Williams 1984, A. microtus Cope 1871, A.
princeps Boulenger 1902, A. philopunctatus
Rodrigues 1988 (see ??Comments??), A.
purpurescens Cope 1899, and A. squamula-
tus Peters 1863 (see ??Comments??). Other
species that may belong to the latifrons
series are A. ibanezi Poe, Latella, Ryan and
Schaad 2009, and A. propinquus Williams
1984 (see ??Comments??). Comments: Re-
ferral of A. kunayalae to this clade is based
on the results of Nicholson et al. (2005, fig.
1, where A. kunayalae corresponds to
Nicholson et al.?s ??New Species 1?? [Hulebak
et al., 2007]). In the context of the reference
phylogeny, four species not traditionally
included in the latifrons series (e.g., Savage
and Guyer, 1989) are included in the
latifrons series as conceptualized here: A.
agassizi, A. chocorum, A. maculigula, and A.
philopunctatus. There is strong evidence
supporting inclusion of the former three
species (Castan?eda and de Queiroz, 2011).
In contrast, because of weak support for the
relationship of A. philopunctatus (Figs. 3, 4)
and because of a current lack of molecular
data, its inclusion in the latifrons series
should be considered tentative. Similarly,
the inclusion of A. squamulatus, tradition-
ally included in the latifrons series (e.g.,
Savage and Guyer, 1989) should be consid-
ered tentative given the weak and inconsis-
tent support for the relevant relationships
(Fig. 3 versus Fig. 4) and the current lack of
molecular data. We have tentatively includ-
ed A. squamulatus in the latifrons series,
where it was placed in the Bayesian tree
(Fig. 4), rather than in the aequatorialis
series, where is was placed in the parsimony
tree (Fig. 3) or in the punctatus series, with
which it agrees best in terms of geographic
distribution (see ??Phylogeny of Dactyloa,??
above) because it shares the derived char-
acter of large body size with members of the
latifrons series. Anolis propinquus has
traditionally been considered part of the
latifrons series (Williams, 1988); however,
the inclusion of this species should be
considered tentative given its current ab-
sence from explicit phylogenetic analyses.
Inclusion of Anolis ibanezi, inferred as
closely related to A. chocorum in a recent
phylogenetic analysis (Poe et al., 2009a),
should also be considered tentative given
that the phylogenetic tree was not shown by
Poe et al. (2009a); thus, the placement of
these two species with respect to the
latifrons series as conceptualized here is
uncertain.
The first use of the name ??latifrons
series?? appears to have been in Etheridge?s
(1959) dissertation; however, that use does
not qualify as published according to the
ICPN (Article 4.2). Moreover, Etheridge
used the name for a more inclusive taxon
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 377
approximating the clade to which the name
Dactyloa is applied here. The oldest pub-
lished use of the name ??latifrons series??
appears to be that of Gorman and Dessauer
(1966), who also used the name for the
more inclusive clade. As delimited here, the
latifrons series more closely approximates
the taxon called the laticeps group by Cope
(1899; which appears to be a lapsus because
the species is elsewhere [p. 7] referred to by
the correct name A. latifrons), the giant
mainland anoles or squamulatus-latifrons
group by Dunn (1937), the latifrons species
group by Williams (1988), the latifrons
series by Savage and Guyer (1989), and
the latifrons clade of Castan?eda and de
Queiroz (2011). However, the latifrons
series as conceptualized here is potentially
more inclusive than the latifrons clade of
Castan?eda and de Queiroz (2011), in that it
appears to include A. philopunctatus and
might also include some species currently
considered incertae sedis within Dactyloa or
absent from explicit phylogenetic analyses if
they are found to be more closely related to
the A. latifrons series than to A. aequator-
ialis, A. punctatus, A. roquet, and A.
heterodermus.
Megaloa Castan?eda and de Queiroz, new
clade name
Definition (apomorphy-based): The
clade originating in the ancestor in which
a maximum SVL . 100 mm in males,
synapomorphic with that of Anolis latifrons
Berthold 1846, originated. Reference phy-
logeny: Figure 5, this study. Inferred
composition: Anolis agassizi Stejneger
1900, A. apollinaris Boulenger 1919, A.
casildae Arosemena, Iba?n?ez, and de Sousa
1991, A. chocorum Williams and Duellman
1967 (see ??Comments??), A. danieli Williams
1988, A. fraseri Gu?nther 1859, A. frenatus
Cope 1899, A. insignis Cope 1871, A.
latifrons Berthold 1846, A. maculigula
Williams 1984, A. microtus Cope 1871, A.
princeps Boulenger 1902, and A. purpur-
escens Cope 1899 (see ??Comments??).
Another species that might belong to
Megaloa is A. squamulatus Peters 1863 (see
??Comments??). Etymology: Derived from
the Greek Mega (large) + loa (the last part
of the name Dactyloa) in reference to the
large body size of the members of this
subclade of Dactyloa.3 Comments: Two
supraspecific names are based on species in
this clade: Diaphoranolis Barbour 1923 (type
species 5 D. brooksi 5 Anolis insignis
according to Etheridge [1959] and Savage
and Talbot [1978]) and Mariguana Dunn
1939 (type species 5 Anolis agassizi). These
names were applied to taxa ranked as genera
and separated from Anolis based on differ-
ences in dorsal scalation (juxtaposed pave-
ment-like scales in A. insignis [Barbour,
1923], and tiny non-imbricating granules
interspersed with larger, single, obtusely
keeled scales in A. agassizi [Dunn, 1939;
Etheridge, 1959]) and dewlap morphology
(supposedly nonextensible in A. insignis
[Barbour, 1923] and poorly developed in A.
agassizi [Dunn, 1939]). Given that we are
emphasizing the associations of names with
clades, rather than with categorical ranks,
and that neither of these names has been
associated with the clade of mainland anoles
with large body size (if they have been
associated with clades at all, those clades are
subclades of the large size clade), it is more
appropriate to create a new name for this
clade than to use either Diaphoranolis or
Mariguana (which remain available for
smaller clades including their type species).
Therefore, we created a name that refers
etymologically to the large size character (see
??Etymology??).
In the context of the reference phylogeny,
A. chocorum and A. purpurescens are
included in Megaloa despite not being
known to possess the synapomorphy of the
clade. In the case of A. chocorum, smaller
size is parsimoniously interpreted as a
reversal. In the case of A. purpurescens,
the only two known male specimens have
3 The component loa is not intended to have any
other meaning beyond reference to Dactyloa because it
contains parts of both of the Greek words on which the
name Dactyloa is based (daktylos, finger + oa, hem,
border; in reference to the toe pads).
378 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
SVLs of 74 and 78 mm and have been
considered juveniles (Williams 1988; MRC,
personal observation), which suggests that
adults may reach a body size larger than
100 mm.
Megaloa corresponds closely to the lati-
ceps group of Cope (1899; which appears
to be a lapsus because the species is
elsewhere [Cope, 1899: 7] referred to by
the correct name, A. latifrons), the giant
mainland anoles or squamulatus-latifrons
group of Dunn (1937), the latifrons species
group of Williams (1988), the latifrons series
of Savage and Guyer (1989), and the
latifrons clade of Castan?eda and de Queiroz
(2011). However, it should be noted that
Megaloa as conceptualized here is less
inclusive than the latifrons series as con-
ceptualized here, in excluding species that
are more closely related to A. latifrons than
to A. aequatorialis, A. punctatus, A. roquet,
and A. heterodermus but branched from the
lineage leading to A. latifrons before large
size evolved (currently, there is only one
known species, A. philopunctatus, that is
considered to belong to the latifrons series
but not to Megaloa [Fig. 5]). If A. squamu-
latus (which exhibits large body size) is part
of the latifrons series, then it is also likely
part of Megaloa, although it might not be
part of either clade (see ??Comments?? on the
latifrons series).
punctatus series Guyer and Savage 1987
(??1986??), informal clade name
Definition (branch-modified node-
based): The crown clade originating in the
most recent common ancestor of A. punc-
tatus Daudin 1802 and all extant species
that share a more recent common ancestor
with A. punctatus than with Anolis aequa-
torialis Werner 1894, A. latifrons Berthold
1846, A. roquet (Bonnaterre 1789), and A.
heterodermus Dume?ril 1851. Reference
phylogeny: Figure 5, this study. Inferred
composition: Anolis anatoloros Ugueto,
Rivas, Barros, Sa?nchez-Pacheco and Gar-
c??a-Pe?rez 2007, A. caquetae Williams 1974
(see ??Comments??), A. dissimilis Williams
1965 (see ??Comments??), A. jacare Boulenger
1903, A. menta Ayala, Harris and Williams
1984 (see ??Comments??), A. punctatus Dau-
din 1802, A. ruizii Rueda and Williams 1986
(see ??Comments??), A. santamartae Williams
1982 (see ??Comments??), A. solitarius Ruth-
ven 1916 (see ??Comments??), A. tigrinus
Peters 1863, A. transversalis Dume?ril 1851,
and A. vaupesianus Williams 1982. Other
species that might belong to the punctatus
series are A. deltae Williams 1974, A.
gorgonae Barbour 1905, A. lamari Williams
1992, A. nasofrontalis Amaral 1933, A.
paravertebralis Bernal Carlo and Roze
2005, A. pseudotigrinus Amaral 1933, A.
soinii Poe and Yan?ez-Miranda 2008,
and A. umbrivagus Bernal Carlo and Roze
2005 (see ??Comments??). Comments: In the
context of the reference phylogeny, eight
species traditionally associated with the
punctatus series are excluded (A. boettgeri,
A. chloris, A. chocorum, A. fasciatus, A.
festae, A huilae, A. peraccae, A. philopuncta-
tus), and four species not traditionally
associated with the punctatus series (all
placed in the tigrinus series, see below) are
included (A. menta, A. ruizii, A. solitarius, A.
tigrinus). Anolis tigrinus and its previously
hypothesized relatives have traditionally
been included in the tigrinus series (e.g.,
Williams, 1976b, 1992; Savage and Guyer,
1989); however, strong evidence supports A.
tigrinus as nested within the punctatus series
(Castan?eda and de Queiroz, 2011). Williams
(1992) noted problems in distinguishing the
punctatus and tigrinus series (referred to by
him as species groups) and raised the
possibility that the tigrinus series is an
ecomorphic subgroup of the punctatus series
(he considered members of the tigrinus
series to be representatives of the twig
ecomorph, whereas he classified at least
some members of the punctatus series as
trunk-crown anoles). Our results support this
hypothesis and we therefore consider A.
tigrinus and its relatives part of the punctatus
series rather than a separate tigrinus series,
although a clade containing A. tigrinus and
all species closer to it than to A. punctatus
could be recognized as a sub-series or a
species group within the punctatus series.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 379
Strong evidence also supports the inclusion
of A. chloris, A. fasciatus, A. festae, and A.
peraccae in the aequatorialis series and A.
chocorum in the latifrons series (see ??Com-
ments?? on the aequatorialis and the latifrons
series, above) and therefore the exclusion of
those species from the punctatus series. In
contrast, because of inconsistent placement
or weak support for the relevant relationships
between analyses (Fig. 3 versus Fig. 4), and
because of a current lack of molecular data
for most of the species (all except A. huilae),
exclusion of A. boettgeri, A. huilae, and A.
philopunctatus from the punctatus series and
inclusion of A. menta, A. ruizii, and A.
solitarius in that series should be considered
tentative. For similar reasons, inclusion of A.
caquetae, A. dissimilis, and A. santamartae,
all traditionally included in the punctatus
series (Williams, 1965, 1974, 1982), should
also be considered tentative. Anolis calimae,
another species traditionally referred to the
punctatus series (Ayala et al., 1983), was
inferred as closely related to species that we
tentatively refer to the punctatus series in the
parsimony tree (Fig. 3) but not in the
Bayesian tree (Fig. 4); because of this and
additional contradictory results regarding the
relationship between A. calimae and the
punctatus series (Castan?eda and de Queiroz,
2011), A. calimae is here considered incertae
sedis (which does not rule out inclusion in the
punctatus series). The long branch leading to
this species and its ambiguous relationships
are consistent with Williams? (1983) conclu-
sion that this species has no evident close
relatives. The geographic distribution of A.
calimae in the central portion of the western
Cordillera of Colombia between 1,300 and
1,800 m (Ayala et al., 1983) does not
correspond to the eastern distribution of the
punctatus series but instead corresponds
more closely to the distribution of the
aequatorialis or the heterodermus series.
Although A. carlostoddi and A. orcesi
were placed in the punctatus series in the
parsimony tree (Fig. 3), they were not
placed there in the Bayesian tree (Fig. 4).
We have considered A. carlostoddi incertae
sedis within Dactyloa based on the deep
level of disagreement concerning its place-
ment between analyses (Figs. 3, 4), as
indicated by its basal regrafted position on
the pruned and regrafted consensus tree
(Fig. 5). The eastern distribution of A.
carlostoddi suggests that it may be part of
the punctatus series, as does that of A.
neblininus, another species that we consider
incertae sedis but which is grouped in the
parsimony tree with species that we tenta-
tively refer to the punctatus series (A.
menta, A. solitarius, and A. ruizii). By
contrast, we have tentatively assigned A.
orcesi to the heterodermus series (and
Phenacosaurus) based on moderate support
from the Bayesian analysis for its inclusion
(Fig. 4) as well as its possession of charac-
ters of the twig ecomorph (Losos, 2009).
The distribution of A. orcesi on the eastern
slopes of the northern Andes of Ecuador is
consistent with referral to the heterodermus
series, although it is also compatible with
referral to the punctatus series (see ??Com-
ments?? section on the heterodermus series
below). The inclusion of A. deltae, A.
gorgonae, and A. soinii, traditionally con-
sidered members of the punctatus series
(Williams and Duellman, 1967; Williams,
1974; Poe and Yan?ez-Miranda, 2008), and
A. lamari, A. nasofrontalis, A. paraverteb-
ralis, A. pseudotigrinus, and A. umbrivagus,
traditionally considered members of the
tigrinus series (Williams, 1992; Bernal Carlo
and Roze, 2005), should also be considered
tentative given their current absence from
explicit phylogenetic analyses.
The punctatus series as conceptualized
here is more inclusive than the eastern clade
of Castan?eda and de Queiroz (2011) in that it
includes species that are more closely related
to A. punctatus than to Anolis aequatorialis,
A. latifrons, A. roquet, and A. heterodermus,
but that diverged before the last common
ancestor of the members of the eastern clade
and might also include some species cur-
rently considered incertae sedis within Dac-
tyloa or absent from explicit phylogenetic
analyses. Additionally, species in the eastern
clade of Castan?eda and de Queiroz (2011)
are characterized by having a cohesive
380 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
eastern Andean and Amazonian geographic
distribution, and it is possible that some
species in the punctatus series do not
conform to this geographic pattern. All of
the species here tentatively referred to the
punctatus series (A. caquetae, A. dissimilis,
A. menta, A. ruizii, A. santamartae, A.
solitarius) are outside of the eastern clade
in the parsimony tree (Fig. 3), and one of
them (A. caquetae) is outside of the eastern
clade in the Bayesian tree (Fig. 4), although
all have eastern geographic distributions (see
??Phylogeny of Dactyloa?? above for details).
roquet series Williams 1976a, informal
clade name
Definition (branch-modified node-
based): The crown clade originating in the
most recent common ancestor of Anolis
roquet (Bonnaterre 1789) and all extant
species that share a more recent common
ancestor with A. roquet than with A.
aequatorialis Werner 1894, A. latifrons
Berthold 1846, A. punctatus Daudin 1802,
and A. heterodermus Dume?ril 1851. Refer-
ence phylogeny: Figure 5, this study.
Inferred composition: Anolis aeneus Gray
1840, A. blanquillanus Hummelinck 1940
(see ??Comments??), A. bonairensis Ruthven
1923, A. extremus Garman 1887, A. griseus
Garman 1887, A. luciae Garman 1887, A.
richardii Dume?ril and Bibron 1837, A.
roquet (Bonnaterre 1789), and A. trinitatis
Reinhardt and Lutken 1862. Comments:
Referral of A. blanquillanus to this clade is
based on the results of Yang et al. (1974)
and Creer et al. (2001). The roquet series as
conceptualized here corresponds exactly in
known composition to the roquet group,
species group, series, and clade of previous
authors (Underwood, 1959; Gorman and
Atkins, 1967, 1969; Lazell, 1972; Williams,
1976a; Savage and Guyer, 1989; Creer et al.,
2001; Castan?eda and de Queiroz, 2011).
Nevertheless, the roquet series as concep-
tualized here is potentially more inclusive
than the roquet clade of Castan?eda and de
Queiroz (2011) in that it could include some
species currently considered incertae sedis
within Dactyloa or absent from explicit
phylogenetic analyses if they are found to
be more closely related to the roquet clade
than to the four other mutually exclusive
Dactyloa subclades inferred by Castan?eda
and de Queiroz (2011).
heterodermus series Castan?eda and de
Queiroz, new informal clade name
Definition (branch-modified node-
based): The crown clade originating in the
most recent common ancestor of Anolis
heterodermus Dume?ril 1851 and all extant
species that share a more recent common
ancestor with A. heterodermus than with
Anolis aequatorialis Werner 1894, A. lati-
frons Berthold 1846, A. punctatus Daudin
1802, and A. roquet (Bonnaterre 1789).
Reference phylogeny: Figure 5, this
study. Inferred composition: Anolis eu-
skalerriari (Barros, Williams, and Viloria
1996), A. heterodermus Dume?ril 1851, A.
inderenae (Rueda and Herna?ndez-Camacho
1988), A. nicefori (Dunn 1944), A. orcesi
(Lazell 1969) (see ??Comments??), A. probos-
cis Peters and Orce?s 1956 (see ??Com-
ments??), A. tetarii (Barros, Williams, and
Viloria 1996), and A. vanzolinii (Williams,
Orce?s, Matheus, and Bleiweiss 1996). An-
other species that might belong to the
heterodermus series is A. williamsmitter-
meierorum Poe and Yan?ez-Miranda 2007
(see ??Comments??). Comments: Inclusion
of A. orcesi, a species traditionally included
in Phenacosaurus (Lazell, 1969), should be
considered tentative given the inconsistent
placement of this species between analyses
(Fig. 3 versus Fig. 4) and because molecu-
lar data are currently lacking. We have
included A. orcesi in the heterodermus
series, where it was placed in the Bayesian
tree (Fig. 4), rather than in the punctatus
series, where it was placed in the parsimony
tree (Fig. 3), because of the stronger sup-
port obtained for that relationship as well as
its sharing of characters of the twig eco-
morph with other species traditionally re-
ferred to Phenacosaurus (Losos, 2009). The
geographic distribution of A. orcesi along the
eastern slopes of the northern Ecuadorean
Andes corresponds with the distribution of
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 381
the heterodermus series, but also with that of
the punctatus series. Anolis williamsmitter-
meierorum, previously considered closely
related to A. orcesi (Williams and Mitterme-
ier, 1991; Poe and Yan?ez-Miranda, 2007), is
tentatively referred to Phenacosaurus given
the current absence of this species from
explicit phylogenetic analyses.
Anolis proboscis, A. laevis, and A. phyl-
lorhinus were previously placed in the laevis
species group or series, a group character-
ized by the presence of a nose leaf
(Williams, 1979). Here we consider the
relationships of A. laevis and A. phyllorhi-
nus to be uncertain (see ??Comments??
section on Phenacosaurus below). The
geographic distribution of A. laevis in the
eastern foothills of the Peruvian Andes does
not suggest a close relationship with A.
proboscis but is consistent with referral to
the heterodermus series (as well as the
aequatorialis and punctatus series). In
contrast, the geographic distribution of A.
phyllorhinus in central Amazonia (Williams,
1979; Rodrigues et al., 2002) suggests
neither a close relationship to A. proboscis
nor inclusion in the heterodermus series but
instead suggests inclusion in the punctatus
series as proposed by Rodrigues et al.
(2002), Ya?nez-Mun?oz et al. (2010), and
Poe et al. (2012). If A. laevis and one or
both other species form a clade within the
heterodermus series, then that clade could
be recognized as the laevis species group
(see also comments on Scytomycterus,
below). If A. laevis and one or both other
species are closely related to members of
the punctatus series, as has been hypothe-
sized previously (Williams, 1965, 1979),
then they should be included within the
punctatus series (perhaps as the laevis
species group). However, if A. laevis and
one or both of the other species lie outside
of the five clades whose names incorporate
the term ??series?? as defined here, then it
would be appropriate to include them in a
separate laevis series (defined as the most
inclusive crown clade containing A. laevis
but not Anolis aequatorialis, A. latifrons, A.
punctatus, A. roquet, and A. heterodermus).
Regardless of whether a separate laevis series
is to be recognized, if A. laevis forms a clade
with either or both A. phyllorhinus and A.
proboscis that can be diagnosed by the nose-
leaf synapomorphy (but see Ya?nez-Mun?oz et
al., 2010), the name Scytomycterus Cope
1876 (derived from the Greek Skytos, skin or
leather, + mykteros, nose; type species 5 A.
laevis) would be an appropriate name for that
clade. However, if A. phyllorhinus or A.
proboscis form a clade but are not closely
related to A. laevis, which differs from the
other two species in having only a rudimen-
tary nose leaf (Williams, 1979), the name
Scytomycterus is not appropriate for that
clade (given that the type is A. laevis);
therefore, if that clade is to be named, a
new name would be appropriate.
The heterodermus series as conceptual-
ized here corresponds closely to the Phena-
cosaurus clade of Castan?eda and de Queiroz
(2011) and Phenacosaurus as conceptual-
ized here. However, it should be noted that
the heterodermus series as conceptualized
here is potentially more inclusive than
Phenacosaurus as conceptualized here (see
below), in that it might include some species
currently considered incertae sedis within
Dactyloa or absent from explicit phyloge-
netic analyses if they are found to be more
closely related to A. heterodermus than to
members of the other four well-supported
clades but branched from the lineage
leading to A. heterodermus before the twig
morphology evolved (currently, all species
assigned to the heterodermus series are also
referred to Phenacosaurus).
Phenacosaurus Barbour 1920, converted
clade name
Definition (apomorphy-based): The
clade originating in the ancestor in which
the combination of morphological charac-
ters of the twig ecomorph (long pointed
snout; forelimbs, hindlimbs, and tail short in
proportion to body size), synapomorphic
with that in Anolis heterodermus Dume?ril
1851, originated. Reference phylogeny:
Figure 5, this study. Inferred composition:
Anolis euskalerriari (Barros, Williams, and
382 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
Viloria 1996), A. heterodermus Dume?ril
1851, A. inderenae (Rueda and Herna?ndez-
Camacho 1988), A. nicefori (Dunn 1944), A.
orcesi (Lazell 1969) (see ??Comments??), A.
proboscis Peters and Orce?s 1956 (see ??Com-
ments??), A. tetarii (Barros, Williams, and
Viloria 1996), and A. vanzolinii (Williams,
Orce?s, Matheus, and Bleiweiss 1996). An-
other species that might belong to Phenaco-
saurus is A. williamsmittermeierorum Poe
and Yan?ez-Miranda 2007 (see ??Comments??).
Comments: Phenacosaurus was originally
proposed (Barbour, 1920) as the name of a
genus separate from Anolis. However, the
addition of subsequently discovered species
(e.g., Dunn, 1944; Lazell, 1969; Rueda and
Herna?ndez-Camacho, 1988; Myers et al.,
1993; Barros et al., 1996; Williams et al.,
1996) has decreased the morphological gap
between the two taxa, and several phyloge-
netic studies (e.g., Jackman et al., 1999; Poe,
2004; Nicholson et al., 2005; Castan?eda and
de Queiroz, 2011; this study) have inferred
Phenacosaurus to be nested within Anolis, so
that recognizing Phenacosaurus as a genus
would render Anolis paraphyletic. We there-
fore use the name Phenacosaurus for a
subclade of Anolis that is not associated with
the rank of genus (it is implicitly associated
with a lower rank). All species in the
Phenacosaurus clade have been considered
twig anoles (Losos, 2009), an ecomorpholo-
gical category characterized by long pointed
snouts, few toepad lamellae, short limbs, and
short, often prehensile, tails. Because several
of those characters were used in the original
diagnosis of Phenacosaurus (Barbour, 1920),
we have defined that name as referring to the
clade of twig anoles that includes its type
species (A. heterodermus).
In the context of the reference phylogeny,
one species not traditionally referred to
Phenacosaurus is included (A. proboscis).
Molecular data are currently lacking for A.
proboscis, but this species was consistently
placed with other species referred to Phena-
cosaurus (see also Poe et al., 2009b, 2012).
Anolis proboscis possesses the morphological
features characteristic of the twig ecomorph:
long pointed snout, forelimbs, hindlimbs,
and tail short in proportion to body size.
Moreover, ecological data indicates A. pro-
boscis should be classified as a twig anole
(Losos et al., 2012; Poe et al., 2012). Two
species traditionally referred to Phenaco-
saurus, Anolis carlostoddi and A. neblininus,
were placed inconsistently between analyses
(Fig. 3 versus Fig. 4), but in neither case
were they placed within Phenacosaurus.
Because molecular data are currently lacking
for A. carlostoddi and because some analyses
based on molecular data suggest inclusion of
A. neblininus in Phenacosaurus (Castan?eda
and de Queiroz, 2011, fig. 2C), neither
species can be confidently excluded. Both
species are here considered incertae sedis
within Dactyloa. Inclusion of A. orcesi, a
species traditionally included in Phenaco-
saurus (Lazell, 1969), should be considered
tentative given the inconsistent placement of
this species between analyses (Fig. 3 versus
Fig. 4) and because molecular data are
currently lacking. We have included A. orcesi
in Phenacosaurus, where it was placed in the
Bayesian tree (Fig. 4), rather than in the
punctatus series, where it was placed in the
parsimony tree (Fig. 3), because of the
stronger support obtained for that relation-
ship as well as its sharing of characters of the
twig ecomorph with other species tradition-
ally referred to Phenacosaurus (Losos, 2009).
Anolis williamsmittermeierorum, previously
considered closely related to A. orcesi
(Williams and Mittermeier, 1991; Poe and
Yan?ez-Miranda, 2007), is tentatively referred
to Phenacosaurus given the current absence
of this species from explicit phylogenetic
analyses. Anolis bellipeniculus is tentatively
excluded from Phenacosaurus based on its
previously hypothesized close relationship to
A. neblininus (Myers and Donnelly, 1996)
and is here considered to be of uncertain
position within Dactyloa (see ??Incertae
sedis,?? below).
Williams (1979) hypothesized that A.
laevis and A. phyllorhinus are closely related
to A. proboscis, which is here included in
Phenacosaurus; however, that relationship
has been questioned by Ya?nez-Mun?oz et al.
(2010) and Poe et al. (2012), and therefore
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 383
we consider A. laevis and A. phyllorhinus to
be incertae sedis within Dactyloa. Little is
known about A. laevis, which is known only
from the type specimen, now in poor
condition (Williams, 1979). However, as
illustrated in Williams (1979, fig. 1), this
specimen does not possess a nose leaf but
only a protruding rostral scale, which is
questionably homologous with the ample
appendages of the other species. Moreover,
the geographic distribution of A. laevis in the
eastern foothills of the Peruvian Andes does
not suggest a close relationship with A.
proboscis, and although it is consistent with
referral to the heterodermus series, it is also
consistent with referral to the aequatorialis
and punctatus series. Anolis phyllorhinus is
better known, and although it possesses a
true nose leaf, which is both similar to and
different from that of A. proboscis, the
information in Williams (1979) and Rodri-
gues et al. (2002) suggest that A. phyllorhi-
nus is a trunk-crown rather than a twig anole
(e.g., green coloration, moderate snout and
limb lengths, long tail, high lamella counts,
relatively large perch diameters, upward
flight behavior, and high degree of similarity
to A. punctatus, which has been classified as
a trunk-crown anole [Williams, 1992]).
Moreover, the geographic distribution of A.
phyllorhinus in central Amazonia (Williams,
1979; Rodrigues et al., 2002) suggests neither
a close relationship to A. proboscis nor
inclusion in the heterodermus series but
instead suggests inclusion in the punctatus
series. Although we think that Ya?nez-Mun?oz
et al. (2010) are likely correct in assigning A.
phyllorhinus in the punctatus series, the
inclusion of neither A. phyllorhinus nor A.
laevis in an explicit phylogenetic analysis
leads us to treat both species as incertae sedis
within Dactyloa.
Phenacosaurus as conceptualized here is
more inclusive than the Phenacosaurus clade
of Castan?eda and de Queiroz (2011) in that it
contains species (e.g., A. proboscis and
possibly A. orcesi, see below) that share the
twig ecomorph synapomorphy with A. hetero-
dermus but lie outside of the smallest
clade containing A. heterodermus and A.
euskalerriari. Phenacosaurus as conceptual-
ized here is less inclusive than the hetero-
dermus series as conceptualized, in excluding
species that are more closely related to A.
heterodermus than to members of the other
four clades recognized here whose names
include the term ??series,?? but branched from
the lineage leading to A. heterodermus before
the twig morphology evolved (although cur-
rently all known species referred to Phenaco-
saurus are also referred to the heterodermus
series).
Incertae sedis
The placement of the following species
could not be resolved because of lack of
data or because of conflicting results
between analyses, and we therefore defer
assigning them to any of the above de-
scribed clades until more definitive evi-
dence is available: A. calimae Ayala, Harris
and Williams 1983, A. carlostoddi (Williams,
Praderio and Gorzula 1996), A. laevis (Cope
1876), A. neblininus (Myers, Williams and
McDiarmid 1993), and A. phyllorhinus
Myers and Carvalho 1945. Possible rela-
tionships of these species have been dis-
cussed under ??Comments?? on the punctatus
series (A. calimae, A. carlostoddi, and A.
neblininus), the heterodermus series (A.
laevis and A. phyllorhinus), and Phenaco-
saurus (A. carlostoddi, A. laevis, A. neblini-
nus and A. phyllorhinus). We also consider
the position of A. bellipeniculus (Myers and
Donnelly 1996) to be uncertain within
Dactyloa given its hypothesized close rela-
tionship to A. neblininus (Myers and
Donnelly, 1996) and the uncertain relation-
ships of that species. The eastern distribu-
tion of A. bellipeniculus on the isolated
Cerro Yav?? tepui of southeastern Venezuela
(Myers and Donnelly, 1996) suggests that it
may be part of the punctatus series.
Similarly, the placement of A. cuscoensis
Poe, Yan?ez-Miranda and Lehr 2008 is
considered unresolved, because although
this species has been included in an explicit
phylogenetic analysis (Poe et al., 2008), its
hypothesized relationships are incongruent
384 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
with the clades recognized here. The
geographic distribution of this species along
the eastern slopes of the southern Peruvian
Andes (Poe et al., 2008) is congruent with
the distributions of the heterodermus,
punctatus, and aequatorialis series (if ex-
tended to the south). Although many
species in the aequatorialis series have
western distributions, the earliest branching
species within the clade (including A.
boettgeri, which was considered closely
related to A. cuscoensis) inhabit the eastern
slopes of the Andes.
According to the definitions presented
above, some species might not belong to any
of the five clades whose names incorporate
the term ??series??; specifically, any species or
clade that is sister to a clade composed of
two or more of the five clades whose names
include the term ??series?? would not be a
member of any of those clades. If strong
support were to be found for such relation-
ships, new ??series?? names could be pro-
posed for the corresponding species or
clades, although such names might be
judged unnecessary for ??series?? composed
of single species. Currently, however, most
known species of Dactyloa are at least
tentatively referable to one of the five
mutually exclusive ??series?? clades, and even
those species that are the best candidates
for not being members of those clades (i.e.,
the species that we consider incertae sedis)
might belong to them.
ACKNOWLEDGMENTS
This research was partially funded by The
George Washington University and the Ernst
Mayr Travel Grant in Animal Systematics.
For access to herpetological collections, the
first author thanks (in Colombia) Mauricio
Alvarez and Diego Perico (Instituto Hum-
boldt), Hermano Roque Casallas and Arturo
Rodr??guez (Museo La Salle), Fernando
Castro (Universidad del Valle), John Lynch
and John Jairo Mueses-Cisneros (Instutito de
Ciencias Naturales, Universidad Nacional),
Vivian Pa?ez and Paul D. Gutie?rrez (Museo
de Herpetolog??a?Universidad de Antioquia);
(in Venezuela) Tito Barros and Gilson Rivas
(Museo de Biolog??a?Universidad del Zulia),
Celsa Cen?aris (Museo de Historia Natural La
Salle); (in Ecuador) Ana Almenda?riz (Uni-
versidad Polite?cnica del Ecuador), Luis
Coloma (formerly at Universidad Cato?lica
de Quito), Mario Ya?nez-Mun?oz (Museo
Ecuatoriano de Ciencias); and (in the United
States) James Hanken, Jonathan Losos, and
Jose Rosado (Museum of Comparative Zool-
ogy) and Jeff Seigel (Los Angeles County
Museum). Steve Gotte, Ken Tighe, Rob
Wilson, and Addison Wynn (U.S. National
Museum of Natural History) provided help
with the clearing and staining of specimens,
radiographs, specimen loans, and other
collection-related issues. James Clark pro-
vided comments on earlier versions that
resulted in significant improvements. Omar
Torres-Carvajal provided FREQPARS files
and help with the coding methods.
APPENDIX I
Morphological Character Descriptions
Description of morphological characters used in
phylogenetic analyses. Ranges of species means (for
continuous characters) correspond to values before
data transformation and coding. Results of correlation
tests (R2 and P values) are shown.
External Characters, Examined on Alcohol-Pre-
served Specimens
1. Maximum male snout-to-vent length (SVL; Wil-
liams et al., 1995, character 35). Measured with a
1-mm precision ruler from the tip of the snout to
the anterior lip of the cloacal opening. Continu-
ous character. Range: 41?170 mm.
2. Ratio of maximum female SVL to maximum male
SVL (Poe, 1998, character 11), both measured
with a 1-mm precision ruler. This character was
not correlated with SVL (R2 5 0.03, P 5 0.25).
Continuous character. Range: 0.64?1.35.
3. Length of head (Poe, 2004, character 4), measured
with 0.01-mm precision calipers from the tip of the
snout to the anterior edge of the ear opening. This
character was correlated with SVL (R2 5 0.94, P,
0.001) and head width (R2 5 0.97, P , 0.001). To
correct for size, head length mean values were
natural log transformed and regressed on natural
log?transformed SVL mean values. Residuals were
subsequently used. Continuous character. Range:
10.68?47.16 mm.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 385
4. Width of head (Poe, 2004, character 5), measured
with 0.01-mm precision calipers at the widest
part of the head?usually the corners of the
mouth. This character was correlated with SVL
(R2 5 0.94, P , 0.001) and head length (R2 5
0.97, P , 0.001). To correct for size, head width
mean values were natural log transformed and
regressed on natural log?transformed SVL mean
values. Residuals were subsequently used. Con-
tinuous character. Range: 5.00?29.46 mm.
5. Height of ear (Poe, 2004, character 6), measured
between the internal borders with 0.01-mm
precision calipers. This character was correlated
with SVL (R2 5 0.58, P , 0.001), head length (R2
5 0.46, P , 0.001), and head width (R2 5 0.56, P
, 0.001). To correct for size, ear height mean
values were natural log transformed and re-
gressed on natural log?transformed SVL mean
values. Residuals were subsequently used. Con-
tinuous character. Range: 0.60?4.84 mm.
6. Interparietal scale length (modified from Poe, 2004,
character 7), measured with 0.01-mm precision
calipers from the anterior to posterior edges of the
scale. The interparietal scale is defined as the scale
overlying the parietal foramen (Peters, 1964).
Located in the parietal area, this scale is typically
of larger size than surrounding scales and exhibits
an area of clear skin above the parietal eye. In some
species, no clear skin area is observed, but a scale
appears to be homologous to the interparietal based
on position, shape, and size. These scales were
measured as interparietals. When scale edges were
not parallel to each other, the distance between the
most anterior to the most posterior points on the
scale were measured. This character was correlated
with SVL (R2 5 0.07, P 5 0.03), head length (R2 5
0.09, P 5 0.002), and head width (R2 5 0.10, P 5
0.01). To correct for size, interparietal length mean
values were natural log transformed and regressed
on natural log-transformed SVL mean values.
Residuals were subsequently used. Continuous
character. Range: 0.55?3.86 mm.
7. Mean number of dorsal scales in 5% of SVL (Poe,
2004, character 19). The equivalent of 5% of SVL
was set on 0.01-mm precision calipers, and the
number of scales contained in this length was
counted three times (using the average as the
final count) lateral to the dorsal midline at the
level of the forelimbs. This character is an
estimate of dorsal scale size. Continuous charac-
ter. Range: 4.20?18.27.
8. Mean number of ventral scales in 5% of SVL
(Poe, 2004, character 20). The equivalent of 5%
of SVL was set on 0.01-mm precision calipers,
and the number of scales contained in this length
was counted three times (using the average as
final count) lateral to the ventral midline in
middle and posterior areas of the body. This
character is an estimate of ventral scale size.
Continuous character. Range: 5.08?13.80.
9. Mean number of scales between the second
canthals (Williams et al., 1995, character 2).
Minimum count between left and right second
canthals, excluding canthal scales. This character
was not correlated with SVL (R2 5 0.01, P 5
0.38), head length (R2 , 0.001, P 5 0.93), or
head width (R2 5 0.003, P 5 0.64), thus no
correction for size was applied. Continuous
character. Range: 2.00?17.50.
10. Mean number of postrostral scales (Williams et
al., 1995, character 3). Postrostrals are all scales
in contact with (posterior to) the rostral scale,
between supralabials. This character was corre-
lated with SVL (R2 5 0.08, P 5 0.02) and head
width (R2 5 0.07, P 5 0.03), but not head length
(R2 5 0.05, P 5 0.06). To correct for size, mean
numbers of postrostral scales were natural log
transformed and regressed on natural log?trans-
formed SVL mean values. Residuals were subse-
quently used. Continuous character. Range:
2.88?8.60.
11. Mean number of scales between supraorbital
semicircles (Williams et al., 1995, character 6).
Minimum count between left and right supraor-
bital seimicircles. This character was correlated
with SVL (R2 5 0.20, P , 0.001), head length (R2
5 0.16, P , 0.001), and head width (R2 5 0.18, P
, 0.001). To correct for size, the mean numbers of
scales between supraorbital semicircles were ln(x
+ 1) transformed and regressed on natural log?
transformed SVL mean values. Residuals were
subsequently used. The ln(x + 1) transformation
was used because this character contains mean
zero values. Continuous character. Range: 0?5.50.
12. Mean number of loreal rows (Williams et al., 1995,
character 10). Loreal scales cover the area
between canthals, supralabials, and subocular
scales. Rows were counted as the minimum
number of scales, in a straight line, from the first
or second canthal to the sublabial scales on the
right side of the head, unless the area was
damaged, and then the left side was scored. This
character was correlated with SVL (R2 5 0.12, P5
0.01), head length (R2 5 0.06, P5 0.05), and head
width (R2 5 0.10, P 5 0.01). To correct for size,
mean numbers of loreal rows were natural log
transformed and regressed on natural log?trans-
formed SVL mean values. Residuals were subse-
quently used. Continuous character. Range: 1.00?
10.20.
13. Mean number of supralabial scales to below the
center of the eye (Williams et al., 1995, character
16), counted from the rostral (not included) to
the midpoint of the eye. More than half the scale
had to be anterior to the center of the eye to be
included in the count. This character was
correlated with SVL (R2 5 0.12, P 5 0.01), head
length (R2 5 0.15, P , 0.001), and head width (R2
5 0.09, P 5 0.02). To correct for size, mean
numbers of supralabial scales were natural log
386 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
transformed and regressed on natural log?trans-
formed SVL mean values. Residuals were subse-
quently used. Continuous character. Range:
5.00?11.00.
14. Mean number of postmental scales (Williams et
al., 1995, character 17). Postmentals are all scales
in contact with (posterior to) the mental scale
between the infralabials (i.e., including the
anteriormost sublabial scale on left and right
sides). This character was correlated with SVL
(R2 5 0.06, P 5 0.04) and head width (R2 5 0.07,
P 5 0.03), but not head length (R2 5 0.03,
P 5 0.21). To correct for size, mean number of
postmental scales were natural log transformed
and regressed on natural log?transformed SVL
mean values. Residuals were subsequently used.
Continuous character. Range: 2.63?10.13.
15. Mean number of sublabial scales (Williams et al.,
1995, character 18; Poe, 2004, character 44).
Sublabial scales are abruptly enlarged scales
(more than twice the size) located medial and
parallel to the infralabials and posterior to the
mental. This character was correlated with SVL
(R2 5 0.13, P , 0.001), head length (R2 5 0.07, P
5 0.03), and head width (R2 5 0.11, P 5 0.01).
To correct for size, the mean number of sublabial
scales were ln(x + 1) transformed and regressed
on natural log?transformed SVL mean values.
Residuals were subsequently used. The ln(x + 1)
transformation was used because this character
contains mean zero values. Continuous character.
Range: 0?7.00.
16. Mean number of scales between the interparietal
scale and the supraorbital semicircles (Williams
et al., 1995, character 13; Poe, 2004, character
46). The minimum number of scales between the
interparietal scale and the supraorbital semicir-
cles was counted. This character was correlated
with SVL (R2 5 0.11, P 5 0.01), head length (R2
5 0.07, P 5 0.03), and head width (R2 5 0.09, P
5 0.02). To correct for size, the mean number of
scales between interparietal and supraorbital
semicircles were ln(x + 1) transformed and
regressed on natural log?transformed SVL mean
values. Residuals were subsequently used. The
ln(x + 1) transformation was used because this
character contains mean zero values. Continuous
character. Range: 0?7.25.
17. Number of elongated superciliary scales (Wil-
liams et al., 1995, character 8). Superciliaries are
scales along the dorsal rim of the orbit, and
elongation occurs toward the posterior end of the
orbit. Left and right sides were scored separately.
States: (0) 0, (1) 1, (2) 2, (3) 3. Polymorphic
character. Ordered.
18. Number of scales between subocular and supra-
labial scales (Williams et al., 1995, character 15;
Poe, 2004, character. 28). The minimum number
of scales was recorded for each specimen. States:
(0) 0, (1) 1, (2) 2. Polymorphic character.
Ordered.
19. Number of ventral scales posteriorly bordering
one scale (modified from Poe, 2004, character
14). Middle and posterior ventral areas were
examined. Ventral scales may be bordered
posteriorly by two scales (0), by two and three
scales (1), or by three scales (2). Polymorphic
character. Ordered.
20. Shape of the base of the tail (modified from
Williams et al., 1995, character 30; Poe, 2004,
character 15). On each specimen, at the point
where the knee would reach the tail if the leg were
folded back, the height and width of the tail was
measured, and then the ratio of width/height was
calculated. States: (0) tail round, for ratios larger
than 1; (1) tail laterally compressed, for ratios
smaller than 1. Polymorphic character.
21. Toepad overlap (Williams et al., 1995, character
27; Poe, 2004, character 9). The toepad under
phalanges III and IV may project distally under
phalanx II (0) or not project distally (1), or the
toepad may be completely absent (2). Polymor-
phic character. Ordered.
22. Male dewlap extension (Williams et al., 1995,
character 33; Poe, 2004, character 16). On the
ventral side, the posterior extension of the
unfolded dewlap is examined. Four states were
considered: posterior extension past the arm
insertion (0), posterior extension to arm insertion
(1), shorter than arm extension (2), dewlap absent
(3). Polymorphic character. Ordered.
23. Female dewlap extension (Williams et al., 1995,
character 34; Poe, 2004, character 17). Measure-
ment and coding as in male dewlap extension.
This character is not correlated with male dewlap
extension (R2 5 0.210, P 5 0.136); therefore, it
was considered a separate character. Polymor-
phic character. Ordered.
24. Size of scales in supraocular discs (modified from
Poe, 2004, character 41). Three different states
were considered: (0) scales vary continuously in
size, in which a few scales are slightly larger (less
than twice the size) than the others, showing
gradual reduction in size; (1) one to three
abruptly enlarged scales (more than twice the
size) with all other scales of smaller size; and (2)
all scales about equal in size. Polymorphic
character. Unordered.
25. Dewlap scales (modified from Poe, 2004, charac-
ter 21). The scales on the dewlap may be in rows of
single scales (0); in double rows (1) or have
scattered scales covering the entire dewlap (2). In
some specimens, most rows were either single or
double, with a few rows exhibiting the alternative
condition. In such cases, the most common
condition was scored for the specimen. Polymor-
phic character. Unordered.
26. Width of mental relative to rostral (modified from
Poe, 2004, character 27). In ventral view, the mental
scale may be broader than the rostral (0), the rostral
scale may be broader than the mental (1), or both
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 387
scales may show the same width (2). Polymorphic
character. Unordered.
27. Enlarged postanal scales in males (Williams et al.,
1995, character 32). Postanal scales may be: (0)
absent, (1) present, as a pair of significantly
enlarged (more than four times the surrounding
scales), (2) present, as a series of more than two
scales slightly enlarged (less than twice the size of
surrounding scales). Polymorphic character. Un-
ordered.
28. Presence or absence of tail crest in males
(Williams et al., 1995, character 31). The tail crest
in males may be: (0) absent, (1) present as a series
of enlarged, but not elevated, serrated scales, or
(2) present as the result of enlarged neural spines.
The presence of a crest is associated with sex and
age of the specimen; therefore, when intraspecific
variation was observed (presence and absence) the
species was coded as present. However, states 1
and 2 were never observed in the same species.
Unordered.
29. Heterogeneous flank scales (modified from Wil-
liams et al., 1995, character 23). Heterogeneous
scales may be (0) absent, (1) very large and
separated from one another by many scales of
much smaller size, (2) a mosaic of scales of
different sizes but not very different in size from
one another, or (3) of average size surrounded by
granular-minute scales. Polymorphic character.
Unordered.
30. Mental scale (Poe, 2004, character 26). The
mental scale may be partially divided (0), in which
a longitudinal split begins from the posterior edge
of the mental but does not reach the anterior edge,
or completely divided (1), in which the split is
complete. Polymorphic character.
31. Frontal depression (Poe, 2004, character 45). A
depression around the frontal area may be absent
(0), in which case the dorsal surface of the snout
is flat, or present (1). Polymorphic character.
32. Presence or absence of an externally visible
parietal eye (Estes et al., 1988, character 26).
The parietal eye, when visible externally, is
located within the interparietal scale (see char-
acter 6). States: absent (0), present (1). Polymor-
phic character.
33. Keeling of dorsal, ventral, supradigital and head
scales (Williams et al., 1995, characters 20, 25, 29,
1; Poe 2004, character 40). Dorsal, ventral, and
supradigital scales may be smooth (S) or keeled
(K); head scales may in addition be rugose (R) or
have pustules (P). The four apparently indepen-
dent characters were combined as one after
correlation was found between ventral, supradigi-
tal, and head keeling with dorsal keeling (R2 5
0.201, P , 0.0001; R2 5 0.635, P , 0.0001; R2 5
0.456, P , 0.0001, respectively). The condition
present in the majority of the scales was reported.
Dorsal scale keeling was scored excluding mid-
dorsal scales because these often differ from the
remaining dorsals (e.g., some species exhibit
smooth dorsal scales, but a double row of keeled
middorsal scales). Weakly keeled specimens were
coded as keeled. Rugose refers to multiple, less
pronounced keels or bent ridges (these two
conditions were commonly found combined in
one scale); with pustules refers to multiple
granular projections scattered on the scale. States
(for dorsals, ventrals, supradigitals, head scales):
KKKK (0), KKKR (1), KKKS (2), KKSP (3),
KSKK (4), KSKS (5), SSKR (6), SSKS (7), SSSS
(8), SSSR (9). Modal condition coded. Unordered.
Osteological Characters Examined on Dry, Cleared,
and Stained Specimens and/or Radiographs.
34. Shape of parietal crests (Etheridge, 1959, fig. 9;
Cannatella and de Queiroz, 1989, characters 6,
7; Williams, 1989, character 7, modified from
Poe, 1998, character 87). Three different states
were considered: (0) trapezoid?shape: lateral
borders of the crest reach the occipital crest
directly (i.e., do not touch each other before
occipital crest contact); (1) V-shape: lateral
borders of the crest join at the point of contact
with the occipital crest, and there is no extension
beyond the point of contact; (2) Y-shape: lateral
borders of the crest join before occipital crest
contact and extend posteriorly beyond the point
of contact (i.e., a unified crest extends toward or
beyond the occipital). Etheridge (1959) showed
that this character exhibits ontogenetic variation,
from a U/trapezoid shape seen in early stages to
an intermediate V-shape, to a Y-shaped crest
seen in adult stages. To compensate for the
absence of sex and SVL information to confirm
adulthood in some specimens, the most devel-
oped state observed (following Etheridge?s
ontogenetic sequence) was scored for each
species. Ordered.
35. Presence or absence of crenulation along lateral
edges of parietal (Poe, 1998, character 88). The
parietal may exhibit irregular (crenulated) or
smooth lateral edges. States: absent (0), present
(1). Polymorphic character.
36. Extension of the parietal roof (modified from Poe,
2004, character 59). In anoles, a parietal casque has
been defined as the shelf-like posterolateral
extension of the parietal roof over the supratem-
poral processes of the parietal. Poe (2004) coded
the presence or absence of the casque, but we
found variation in the length of the extension. The
roof extension may be large, almost completely
covering the supratemporal processes and some-
times extending beyond the posteriormost margin,
or the extension could be small, leaving more of the
supratemporal process uncovered and not reaching
the posterior margin of the parietal. States: not
extended (0; e.g., A. chocorum, MCZ 115732);
present and small, not reaching posteriormost
margin of supratemporal processes (1; e.g., A.
fitchi, MCZ 178084); present and large, reaching or
388 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
extending beyond the posteriormost margin of
supratemporal processes (2; e.g., A. heterodermus,
MCZ 110138) (Fig. 6). The largest extension of the
parietal roof observed among all individuals was
scored for each species. Ordered.
37. Parietal foramen (Etheridge, 1959; Williams, 1989,
character 5). The parietal foramen may be located
completely within the parietal (0) or may be at the
fronto-parietal suture (1). Cases in which the
foramen is located within the parietal but connect-
ed to the fronto-parietal by a suture were coded as
0. Absence of the parietal foramen was coded as ?,
instead of as a third state, given that the information
on the presence or absence of an externally visible
parietal eye was coded as a separate character
(character 32) from alcohol-preserved specimens.
Polymorphic character.
38. Fronto-parietal suture (this study). The fronto-
parietal suture may form a straight transverse line
(i.e., perpendicular to the longitudinal axis of the
body) (0; e.g., A. danieli, MCZ 164894) or it may
exhibit a posteriorly convex curve in the center
because of extension of the frontal bone into the
parietal bone (1; e.g., A. eulaemus, MCZ 158390)
(Fig. 7). Significant variation was observed in the
latter state (from a slight to a substantial
protuberance), but this variation was not quan-
tified. Polymorphic character.
39. Presence or absence of postfrontal (Etheridge
and de Queiroz, 1988, character 6; Poe, 1998,
Figure 6. Dorsal views of the skulls of three anoles illustrating differences in the extension of the parietal roof (character 36). (a)
Anolis chocorum, MCZ 115732 (state 0); (b) Anolis fitchi, MCZ 178084 (state 1); (c) Anolis heterodermus, MCZ 110138 (state 2).
Scale bar 5 5 mm. Abbreviations: par, parietal; st-par, supratemporal processes of the parietal; pm-par, posterior margin
of parietal.
Figure 7. Dorsal views of the skulls of two anoles illustrating differences in the fronto-parietal suture (character 38). (a) Anolis
danieli, MCZ 164894 (state 0); (b) Anolis eulaemus, MCZ 158390 (state 1). Scale bar 5 5 mm. Abbreviations: fr, frontal;
par, parietal.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 389
character 92; 2004, character 62). The postfrontal
bone is located in the posterodorsal margin of the
orbit between (or overlapping) the frontal and
the postorbital. States: absent (0), present (1).
Polymorphic character.
40. Frontal (Poe, 1998, character 94; 2004, character
64). The anterior suture of the frontal may be in
contact only with nasals (0), may be separated
from nasals by an open gap (1), or may be in
contact with both the premaxilla and nasal (2).
State 1 includes cases where the gap was along
the entire suture or, most commonly, in the
center only, allowing partial lateral contact
between frontal and nasals. Posterior extension
of the premaxilla, sufficient to potentially contact
the frontal (i.e., if the gap were absent), was
never observed along with the open gap.
Polymorphic character. Unordered.
41. Prefrontals (Poe, 1998, character 93, fig. 4).
Prefrontals may be in contact with nasals (0) or
may be separated from nasals by the contact
between frontal and maxilla (1). Any contact
between prefrontal and nasal was scored as 0.
Differences between left and right sides were
observed in some specimens; therefore, each side
was treated separately for frequency calculations.
Polymorphic character.
42. Posterior extension of maxilla (Poe, 1998, charac-
ter 103, fig. 8). Different landmarks have been
used as boundaries to quantify the posterior
extension of the maxilla (e.g., Estes et al., 1988,
character 27; Frost and Etheridge, 1989, character
3). Following Poe (1998), the posterior edge of the
ectopterygoid was used to delimit two different
states: (0) maxilla does not extend posteriorly
beyond the posterior edge of ectopterygoid
(including cases in which it extends to that level)
or (1) maxilla extends beyond the posterior edge of
ectopterygoid. Differences between left and right
sides were observed in some specimens; therefore,
each side was treated independently for frequency
calculations. Polymorphic character.
43. Mean number of premaxillary teeth (de Queiroz,
1987, characters 43, 44). This character was not
correlated with SVL (R2 5 0.01, P 5 0.46), head
length (R2 5 0.013, P 5 0.39), or head width (R2
5 0.009, P 5 0.47); thus, no correction for size
was applied. Range: 6?13. Continuous character.
44. Presence or absence of pterygoid teeth (Ether-
idge, 1959; Poe, 1998, character 101). Pterygoid
teeth are found along the edge facing the
pyriform recess, either clumped or in a single
row. States: absent (0), present (1). Polymorphic
character.
45. Presence or absence of contact between jugal and
squamosal (Frost and Etheridge, 1989, character
8). The jugal and squamosal bones may be in
contact along the ventral edge of the temporal bar,
or they may be separated by the postorbital bone.
In some specimens, differences between the left
and right sides were found; therefore, each side
was treated separately for frequency calculations.
States: absence (0), presence (1). Polymorphic
character.
46. Shape of posteroventral corner of jugal (modified
from Poe, 2004, character 69). Poe (2004)
recognized two states of this character: postero-
ventral corner of jugal is anterior to the posterior
edge of jugal (in species where the posterior edge
of the jugal shows a straight or convex border) or
is posterior to the posterior edge of the jugal (in
species where the posterior edge of jugal shows a
concave border). However, we found these two
character states not to be mutually exclusive;
therefore, the states were modified as follows:
posterior border of the jugal concave, with a
sharp (pointed) posteroventral corner (0), or
posterior border straight or convex, with a
rounded posteroventral corner (1). Differences
between left and right sides were observed in
some specimens; therefore, each side was treated
independently for frequency calculations. Poly-
morphic character.
47. Presence or absence of contact between parietal
and epipterygoid (Poe, 1998, character 99). The
epipterygoid extends from the palate toward the
skull roof and may or may not reach the parietal.
In some species, the most distal portion of the
epipterygoid is cartilaginous and often lost during
skull preparation, rendering the structure not in
contact with the parietal. Cases in which the
absence of contact is an artifact of preparation
could not be distinguished from those in which
the epipterygoid (with or without cartilaginous
portion) is short enough not to be in contact with
the parietal. All cases with no contact were coded
as absence. No intraspecific variation was ob-
served. States: absent (0), present (1).
48. Supraoccipital cresting (Poe, 1998, character 105,
fig. 9; 2004, character 55). The supraoccipital may
show: (0) a single medial process (called processus
ascendens; e.g., A. heterodermus, MCZ 110133);
(1) a medial process in addition to two distinct and
smaller lateral processes (not always ossified; e.g.,
A. chloris, MCZ 101290) or (2) a continuous (e.g.,
A. agassizi, MCZ 18088) or partially continuous
crest (showing two lateral processes with a distinct
crest between them) running along the edge of the
osseus labyrinth (Fig. 8). Significant ontogenetic
variation was observed within each one of the
states, but a sequence linking all three states was
not observed; therefore, the modal condition was
scored for each species. Unordered.
49. Contact between parietal and supraoccipital (this
study). The parietal may be widely separated from
the supraoccipital, leaving free space between the
two on either side of the processus ascendens (0;
e.g., A. princeps, MCZ 147444), or may be in
contact (or almost in contact) with the supraoc-
cipital, leaving no open space in between (1; e.g.,
A. ventrimaculatus, MCZ 127711) (Fig. 9). Poly-
morphic character.
390 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
50. Extension of the supratemporal processes of the
parietal (this study). In some species (e.g., A.
agassizi, MCZ 27120), the supratemporal pro-
cesses of the parietal extend dorsally forming a
vertical flange (1); in others (e.g., A. heteroder-
mus, MCZ 110133), the supratemporal processes
of the parietal do not extend (0) (Fig. 10).
Significant variation was observed in the height
of the extension, but it was not quantified.
Additionally, ontogenetic variation was observed
within some species; therefore, the most devel-
oped state (i.e., supratemporal processes extend-
ed) was scored for the species if it was observed
in any specimens. States: supratemporal process-
es of the parietal not extended (0), extended (1).
51. Presence or absence of the quadrate lateral shelf
(Poe, 1998, character 106, fig. 10). The quadrate
lateral shelf is the lateral extension of the external
edge of the quadrate. Ontogenetic variation was
observed within some species; thus, the devel-
oped state (i.e., presence of quadrate lateral
shelf) was scored for the species when observed.
States: absent (0), present (1).
52. Presence or absence of angular process of
prearticular (de Queiroz, 1987, character 41, fig.
28; Poe, 1998, character 110, fig. 11). This process
is located on the medial side of the retroarticular
process of the prearticular and has a fin-like or
rounded shape. Presence was coded as a signifi-
cant extension beyond an imaginary line along the
medial edge (in dorsal view) of the prearticular.
Absent and rudimentary processes were coded as
absent. Differences in size of the process were
observed, but were not quantified. Poe (1998,
2004) called this structure angular process of the
articular, but the articular is an endochondral
(rather than dermal) bone that results from the
ossification of the posterior end of Meckel?s
Figure 8. Posterior views of the skulls of three anoles illustrating differences in the supraoccipital cresting (character 48). (a)
Anolis heterodermus, MCZ 110133 (state 0); (b) Anolis chloris, MCZ 101290 (state 1); (c) Anolis agassizi, MCZ 18088 (state 2).
Scale bar 5 5 mm. Abbreviations: pa, processus ascendens; soc, supraoccipital.
Figure 9. Posterior views of the skulls of two anoles illustrating differences in the contact between the parietal and supraoccipital
(character 49). (a) Anolis princeps, MCZ 147444 (state 0); (b) A. ventrimaculatus, MCZ 127711 (state 1). Scale bar 5 5 mm.
Abbreviations: par, parietal; soc, supraoccipital.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 391
cartilage and forms the articular condyle, but
neither the retroarticular nor the angular process
(de Queiroz, 1987). No intraspecific variation was
observed. States: absent (0), present (1).
53. Position of posteriormost tooth with respect to
the combined alveolar-mylohyoid foramen (camf;
modified from de Queiroz, 1987, characters 34,
35; Poe, 1998, character 109; 2004, character 81).
Etheridge (1959) reported that in some iguanids
(e.g., anoles) the anterior mylohyoid foramen
(amf, usually located within the splenial) is united
with the anterior inferior alveolar foramen (aiaf,
located between the dentary and splenial),
resulting in a single foramen. In the present
study, the single foramen is called the combined
alveolar-mylohyoid foramen (camf). Poe (1998,
2004) compared the position of the posteriormost
tooth to the amf, which is the same as this
character. We compared the position of the
posterior edge of the posteriormost tooth to the
camf, and considered three states: posteriormost
tooth is anterior to camf (0), overlaps with camf
(1), posteriormost tooth is posterior to camf (2).
Left and right mandibles were coded separately.
Polymorphic character. Unordered.
54. Shape of the posterior suture of dentary (Poe,
1998, character 111, fig. 12). In lateral view, the
suture of the dentary with the surangular may
have a distinctly pronged (i.e., with two process-
es) or a blunt, undifferentiated shape. No
intraspecific variation was observed. States:
pronged (0), blunt (1).
55. Position of posterior suture of dentary, relative to
mandibular fossa (Poe, 1998, character 112).
Given the possible shape of this suture (blunt
or pronged), the anteriormost aspect of the
posterior border is the point used for comparison.
States: posterior border of dentary is anterior to
mandibular fossa (0) or within mandibular fossa
(1). Polymorphic character.
56. Position of surangular foramen (Frost and
Etheridge, 1988, character 19, fig. 3; Poe, 1998,
character 115, fig. 13). The surangular foramen
(on the lateral surface of the mandible; same as
Poe?s [2004] supra-angular foramen) may be
located entirely within the surangular (0) or be
partially bordered by the dentary (1). Differences
between left and right sides were observed in
some specimens; therefore, each side was treated
separately for frequency calculations. Polymor-
phic character.
57. Presence or absence of splenial bone (Etheridge,
1959; Poe, 2004, character 85). States: absent (0),
present as anteromedial sliver (1), or present and
large, as in Polychrus and other non-anole
iguanids (2). No intraspecific variation was
observed. Ordered.
58. Presence or absence of angular bone (Etheridge,
1959). States: absent (0), present (1). No
intraspecific variation was observed.
59. Overlap between clavicles and lateral processes of
interclavicle (modified from Etheridge, 1959).
Etheridge (1959) described two different types of
interclavicles in anoles: arrow-shaped (in which the
lateral processes of the interclavicle are caudolat-
erally directed and only medially overlapped by the
clavicle) or T-shaped (in which the lateral process-
es are laterally directed and broadly overlapped by
the clavicle). The two components of the inter-
clavicle shape, as described by Etheridge (1959),
can vary independently; therefore, this character
was divided into two. The first was quantified as
Figure 10. Posterior views of the skulls of two anoles illustrating differences in the extension of the supratemporal processes of
the parietal (character 50). (a) Anolis heterodermus, MCZ 110133 (state 0); (b) Anolis agassizi, MCZ 27120 (state 1). Scale bar 5
5 mm. Abbreviations: st-par, supratemporal processes of the parietal.
392 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
the fraction of the length of the lateral process of
the interclavicle in direct contact with (i.e.,
overlapped by) the clavicle. The total length of
the lateral process was measured as a straight line
from the midline of the interclavicle (an imaginary
line along the long axis of the median [posterior]
process) to the tip of the lateral process (Fig. 11).
The overlapped distance was measured along the
same straight line. Length measurements were
made on photographs of dry or clear and stained
interclavicles using the software MacMorph (Spen-
cer and Spencer, 1993). Two measurements were
made on each side (left and right sides separately)
and used to calculate the average per species.
Continuous character. Range: 0.36?0.96.
60. Angle between the median (posterior) process
and the lateral process of the interclavicle (this is
the second character derived from the arrow-
shaped and T-shaped conditions of Etheridge
[1959] described in the previous character). The
angle was measured between the long axis of the
median process and that of the lateral process (as
described in the previous character; Fig. 12) on
photographs of dry or cleared and stained
interclavicles, using the software MacMorph
(Spencer and Spencer, 1993). Two measure-
ments were made on each side (left and right
sides separately) and used to calculate the
average per species. Continuous character.
Range: 44.9?64.5.
61. Postxiphisternal inscriptional rib formula (Ether-
idge, 1959). The postxiphisternal inscriptional ribs
are the cartilaginous ventral rib elements located
caudal to the xiphisternum. The first number in
the formula refers to the number of such ribs
attached to the (ossified) dorsal ribs; the second
refers to the number of floating (unattached)
postxiphisternal inscriptional ribs caudal to the
attached ones. The modal condition was scored for
each species. States: (0) 2:2, (1) 3:1, (2) 4:0, (3) 4:1,
(4) 5:0, (5) 5:1, (6) 5:2, (7) 8:1. This character was
ordered using a step matrix (modified from
Jackman et al., 1999), in which the gain or loss of
a rib or a connection (from attached to floating or
vice versa) costs one step.
62. Number of presacral vertebrae (Etheridge,
1959). The presacral vertebrae are all vertebrate
anterior to the sacrum. States: (0) 22, (1) 23, (2)
24, (3) 25, (4) 27. Polymorphic character.
Ordered.
63. Number of lumbar vertebrae (Etheridge, 1959).
The lumbar vertebrae are post-thoracic vertebrae
(i.e., those that are not attached directly or
indirectly to the sternum) that bear no ribs.
States: (0) 1, (1) 2, (2) 3, (3) 4, (4) 5. Polymorphic
character. Ordered.
64. Type of caudal vertebrae (Etheridge, 1959).
Caudal vertebrae may be of the alpha type (0),
in which the transverse processes are caudolat-
erally or laterally directed and present only on
the most anterior vertebrae (7?15), or the beta
type (1), in which transverse processes are
present much farther posteriorly in the caudal
sequence, where they are directed craniolater-
ally. No instraspecific variation was observed.
65. Caudal autotomy septa (Etheridge, 1959). Autot-
omy septa are observed in radiographs as unossi-
Figure 11. Ventral view of the pectoral girdle illustrating
details on measurements of the overlap between clavicles and
the lateral processes of the interclavicle (character 59). Scale
bar 5 5 mm. Abbreviations: cl, clavicle; icl, interclavicle; od,
overlapped distance; tl, total length of lateral process
of interclavicle.
Figure 12. Ventral view of the pectoral girdle illustrating
details on measurements of the angle between the median
(posterior) process and the lateral process of the interclavicle
(character 60). Abbreviations: lp-icl, lateral process of inter-
clavicle; mp-icl, medial process of interclavicle.
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 393
fied areas in the vertebrae anterior, posterior, or
through the transverse process. The anteriormost
autotomy septum usually coincides with a change
in the condition of the transverse process (e.g.,
disappearance, change in direction, or appearance
of a second pair; Etheridge, 1959). This character
exhibits ontogenetic variation, as in some species,
progressive fusion of the septa occurs from caudal
to cranial with age (Etheridge, 1959). To account
for this variation, three states were recognized:
septa absent in all specimens, representing those
species that do not have (at any life stage) autotomy
(0), septa present in some specimens and absent in
others, representing those species that progres-
sively lose autotomy with age (1), or septa present
in all specimens, representing those species that
retain autotomy throughout life (provided that
large individuals were examined) (2). This coding
approach is strongly biased by sample size but was
the best found to incorporate information on the
gradual change of the character. Ordered.
66. Mean number of caudal vertebrae bearing
transverse processes (Etheridge, 1959). Trans-
verse processes are always present on the
anteriormost caudal vertebrae and progressively
decrease in size posteriorly until absent. Contin-
uous character. Range: 6.8?32.2.
LITERATURE CITED
ALFARO, M. E., S. ZOLLER, AND F. LUTZONI. 2003. Bayes
or bootstrap? A simulation study comparing the
performance of Bayesian Markov chain Monte
Carlo sampling and bootstrapping in assessing
phylogenetic confidence. Molecular Biology and
Evolution 20: 255?266.
AROSEMENA, F. A., R. IBAN?EZ D., AND F. DE SOUSA. 1991.
Una nueva especies de Anolis (Squamata: Iguani-
dae) del grupo latifrons de Fortuna, Panama?.
Revista de Biolog??a Tropical 39: 255?262.
AYALA, S., D. HARRIS, AND E. E. WILLIAMS. 1983. New or
problematic Anolis from Colombia. I. Anolis
calimae, new species, from the cloud forest of
western Colombia. Breviora 475: 1?11.
AYALA, S. C., AND F. CASTRO. Familia Iguanidae?II
Anolinos. In S. C. Ayala and F. Castro (eds)
Saurios de Colombia. Unpublished manuscript.
AYALA, S. C., D. M. HARRIS, AND E. E. WILLIAMS. 1984.
Anolis menta, sp. n. (Sauria, Iguanidae), a new
tigrinus group anole from the west side of the
Santa Marta Mountains, Colombia. Pape?is Avulsos
de Zoologia 35: 135?145.
AYALA-VARELA, F. P., AND O. TORRES-CARVAJAL. 2010. A
new species of dactyloid anole (Iguanidae, Poly-
chrotinae, Anolis) from the southeastern slopes of
the Andes of Ecuador. ZooKeys 53: 59?73.
AYALA-VARELA, F. P., AND J. A. VELASCO. 2010. A new
species of dactyloid anole (Squamata: Iguanidae)
from the western Andes of Ecuador. Zootaxa 2577:
46?56.
BARBOUR, T. 1920. A note on Xiphocercus. Proceedings
of the New England Zoological Club 7: 61?63.
BARBOUR, T. 1923. Notes on reptiles and amphibians
from Panama?. Occasional Papers of the Museum of
Zoology, University of Michigan 129: 1?16.
BARROS, T., E. E. WILLIAMS, AND A. VILORIA. 1996. The
genus Phenacosaurus (Squamata: Iguania) in western
Venezuela: Phenacosaurus tetarii, new species,
Phenacosaurus euskalerriari, new species, and Phe-
nacosaurus nicefori Dunn, 1944. Breviora 504: 1?30.
BERLOCHER, S. H., AND D. L. SWOFFORD. 1997.
Searching for phylogenetic trees under the fre-
quency parsimony criterion: an approximation
using generalized parsimony. Systematic Biology
46: 211?215.
BERNAL CARLO, A., AND J. A. ROZE. 2005. Lizards of the
genus Anolis (Reptilia: Polychrotidae) from Sierra
Nevada de Santa Marta, Colombia, with descrip-
tion of two new species. Novedades Colombianas 8:
9?26.
BRANDLEY, M. C., AND K. DE QUEIROZ. 2004. Phylogeny,
ecomorphological evolution, and historical bioge-
ography of the Anolis cristatellus group. Herpeto-
logical Monographs 18: 90?126.
BRYANT, D. 2003. A classification of consensus methods
for phylogenetics, pp. 1?21. In M. Janowitz, F. J.
Lapointe, F. McMorris, B. Mirkin, and F. Roberts
(eds.), BioConsensus. Piscataway, New Jersey:
American Mathematical Society Publications?DI-
MACS Series in Discrete Mathematics and Theo-
retical Computer Science.
BURNELL, K. L., AND S. B. HEDGES. 1990. Relationships
of West Indian Anolis (Sauria: Iguanidae): an
approach using slow-evolving protein loci. Carib-
bean Journal of Sciences 26: 7?30.
BUTLER, M. A., S. A. SAWYER, AND J. B. LOSOS. 2007.
Sexual dimorphism and adaptive radiation in Anolis
lizards. Nature 447: 202?205.
BUTLER, M. A., T. W. SCHOENER, AND J. B. LOSOS. 2000.
The relationship between sexual size dimorphism
and habitat use in Greater Antillean Anolis lizards.
Evolution 54: 259?272.
CANNATELLA, D., AND K. DE QUEIROZ. 1989. Phyloge-
netic systematics of the Anoles: is a new taxonomy
warranted? Systematic Zoology 38: 57?69.
CASTAN?EDA, M. d. R., AND K. DE QUEIROZ. 2011.
Phylogenetic relationships of the Dactyloa clade
of Anolis lizards based on nuclear and mitochon-
drial DNA sequence data. Molecular Phylogenetics
and Evolution 61: 784?800.
COPE, E. D. 1899. Contributions to the herpetology of
New Grenada and Argentina, with descriptions of
new forms. Philadelphia Museums Science Bulletin
1: 1?21.
CREER, D. A., K. DE QUEIROZ, T. R. JACKMAN, J. B.
LOSOS, AND A. LARSON. 2001. Systematics of the
Anolis roquet series of the southern Lesser Antilles.
Journal of Herpetology 35: 428?441.
DE QUEIROZ, K. 1987. Phylogenetic systematics of
Iguanine lizards. A comparative osteological study.
394 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
University of California Publications in Zoology
118: 1?203.
DE QUEIROZ, K. 1996. Including the characters of interest
during tree reconstruction and the problems of
circularity and bias in studies of character evolution.
American Naturalist 148: 700?708.
DE QUEIROZ, K. 2000. Logical problems associated with
including and excluding characters during tree
reconstruction and their implications for the study
of morphological character evolution, pp. 192?212.
In J. J. Wiens (ed.), Phylogenetic Analysis of
Morphological Data. Washington, DC: Smithso-
nian Institution Press.
DE QUEIROZ, K., AND M. DONOGHUE. 2013. Phylogenetic
nomenclature, hierarchical information, and test-
ability. Systematic Biology. doi: 10.1093/sysbio/
sys054
DE QUEIROZ, K., AND J. GAUTHIER. 1990. Phylogeny as a
central principle in taxonomy: phylogenetic defini-
tions of taxon names. Systematic Zoology 39: 307?
322.
DE QUEIROZ, K., AND J. GAUTHIER. 1992. Phylogenetic
taxonomy. Annual Review of Ecology and System-
atics 23: 449?480.
DE QUEIROZ, K., AND T. REEDER. 2008. Squamata?
Lizards, pp. 24?45. In B. I. Crother (ed.), Scientific
and Standard English Names of Amphibians and
Reptiles of North America North of Mexico, with
Comments Regarding Confidence in Our Under-
standing. Society for the Study of Amphibians and
Reptiles, Herpetological Circular No. 37.
DRUMMOND, A. J., AND A. RAMBAUT. 2007. BEAST:
Bayesian evolutionary analysis by sampling trees.
BMC Evolutionary Biology 7: 214.
DUNN, E. R. 1937. The giant mainland Anoles.
Proceedings of the New England Zoological Club
16: 5?9.
DUNN, E. R. 1939. Zoological results of the George
Vanderbilt South Pacific Expedition of 1937. Part
III?The lizards of Malpelo Island, Colombia.
Notulae Naturae 4: 1?3.
DUNN, E. R. 1944. The lizard genus Phenacosaurus.
Caldasia 3: 57?62.
ESTES, R., K. DE QUEIROZ, AND J. GAUTHIER. 1988.
Phylogenetic relationships within Squamata, pp.
119?281. In R. Estes and G. Pregill (eds.),
Phylogenetic Relationships of the Lizard Fami-
lies?Essays Commemorating Charles L. Camp.
Stanford, California: Stanford Univ. Press.
ETHERIDGE, R. 1959. The relationships of the Anoles
(Reptilia: Sauria: Iguanidae): an interpretation
based on skeletal morphology. Ph.D. Thesis. Ann
Arbor, Michigan: Univ. of Michigan.
ETHERIDGE, R., AND K. DE QUEIROZ. 1988. A phylogeny
of Iguanidae, pp. 283?367. In R. Estes and G.
Pregill (eds.), Phylogenetic Relationships of the
Lizard Families?Essays Commemorating Charles
L. Camp. Stanford, California: Stanford Univ. Press.
FELSENSTEIN, J. 1985. Confidence limits on phylogenies:
an approach using the bootstrap. Evolution 39:
783?791.
FINDEN, C. R., AND A. D. GORDON. 1985. Obtaining
common pruned trees. Journal of Classification 2:
255?276.
FISHER, R. A. 1930. The moments of the distribution for
normal samples of measures of departure from
normality. Proceedings of the Royal Society A, 130:
16?28.
FROST, D. R., AND R. ETHERIDGE. 1989. A phylogenetic
analysis and taxonomy of iguanian lizards (Reptilia:
Squamata). Miscellaneous Publications University
of Kansas Natural History Museum 81: 1?65.
GIANNASI, N., R. S. THORPE, AND A. MALHOTRA. 2000. A
phylogenetic analysis of body size evolution in the
Anolis roquet group (Sauria: Iguanidae): character
displacement or size assortment? Molecular Ecol-
ogy 9: 193?202.
GLOR, R. E., J. J. KOLBE, R. POWELL, A. LARSON, AND J.
B. LOSOS. 2003. Phylogenetic analysis of ecological
and morphological diversification in Hispaniolan
trunk-ground anoles (Anolis cybotes group). Evo-
lution 57: 2383?2397.
GORDON, A. D. 1980. On the assessment and compar-
ison of classifications, pp. 149?160. In R. Tomas-
sone (ed.), Analyse de Donnees et Informatique. Le
Chesnay, France: INRIA.
GORMAN, G. C. 1968. The relationships of Anolis of the
roquet species group (Sauria: Iguanidae). III.
Comparative study of display behavior. Breviora
284: 1?31.
GORMAN, G. C., AND L. ATKINS. 1967. The relationships
of Anolis of the roquet species group (Sauria:
Iguanidae). II. Comparative chromosome cytology.
Systematic Zoology 16: 137?143.
GORMAN, G. C., AND L. ATKINS. 1969. The zoogeography
of Lesser Antillean Anolis lizards?an analysis
based upon chromosomes and lactic dehydroge-
nases. Bulletin of the Museum of Comparative
Zoology 138: 53?80.
GORMAN, G. C., D. G. BUTH, M. SOULE?, AND S. Y. YANG.
1980a. The relationships of the Anolis cristatellus
group: electrophoretic analysis. Journal of Herpe-
tology 14: 269?278.
GORMAN, G. C., D. G. BUTH, M. SOULE?, AND S. Y. YANG.
1983. The relationships of the Puerto Rican Anolis:
electrophoretic and karyotypic studies, pp. 626?
642. In A. G. J. Rhodin and K. Miyata (eds.),
Advances in Herpetology and Evolutionary Biolo-
gy. Cambridge, Massachusetts: Museum of Com-
parative Zoology.
GORMAN, G. C., D. G. BUTH, AND J. S. WYLES. 1980b.
Anolis lizards of the eastern Caribbean: a case
study in evolution. III. A cladistic analysis of
albumin immunological data, and the definitions
of species groups. Systematic Zoology 29: 143?158.
GORMAN, G. C., AND H. C. DESSAUER. 1965. Hemoglobin
and transferring electrophoresis and relationships
of island populations of Anolis lizards. Science 150:
1454?1455.
GORMAN, G. C., AND H. C. DESSAUER. 1966. The
relationships of Anolis of the roquet species group
(Sauria: Iguanidae). I. Electrophoretic comparison
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 395
of blood proteins. Comparative Biochemistry and
Physiology 19: 845?853.
GORMAN, G. C., AND Y. J. KIM. 1976. Anolis lizards of the
eastern Caribbean: a case study in evolution. II.
Genetic relationships and genetic variation of the
bimaculatus group. Systematic Zoology 20: 167?185.
GORMAN, G. C., P. LICHT, H. C. DESSAUER, AND J. O.
BOOS. 1971. Reproductive failure among the
hybridizing Anolis lizards of Trinidad. Systematic
Zoology 20: 1?18.
GORMAN, G. C., C. S. LIEB, AND R. H. HARWOOD. 1984.
The relationships of Anolis gadovi: albumin
immunological evidence. Caribbean Journal of
Science 20: 145?152.
GORMAN, G. C., AND R. B. STAMM. 1975. The Anolis
lizards of Mona, Redonda, and La Blanquilla:
chromosomes, relationships, and natural history
notes. Journal of Herpetology 9: 197?205.
GORMAN, G. C., R. THOMAS, AND L. ATKINS. 1968. Intra-
and interspecific chromosome variation in Anolis
cristatellus and its closest relatives. Breviora 293:
1?12.
GUYER, C., AND J. M. SAVAGE. 1987 [1986]. Cladistic
relationships among anoles (Sauria: Iguanidae).
Systematic Zoology 35: 509?531.
HILLIS, D. M. 1991. Discriminating between phyloge-
netic signal and random noise in DNA sequences,
pp. 278?294. In M. M. Miyamoto and J. Cracraft
(eds.), Phylogenetic Analyses of DNA Sequences.
New York: Oxford Univ. Press.
HILLIS, D. M., AND J. P. HUELSENBECK. 1992. Signal,
noise and reliability in molecular phylogenetic
analyses. Journal of Heredity 83: 189?195.
HUELSENBECK, J. P. 1991. Tree-length distribution
skewness: an indicator of phylogenetic information.
Systematic Zoology 40: 257?270.
HUELSENBECK, J. P., F. RONQUIST, R. NIELSEN, AND J. P.
BOLLBACK. 2001. Bayesian inference of phylogeny
and its impact on evolutionary biology. Science
294: 2310?2314.
HULEBAK, E., S. POE, R. IBA?N?EZ, AND E. E. WILLIAMS.
2007. A striking new species of Anolis lizard
(Squamata, Iguania) from Panama. Phyllomedusa
6: 5?10.
JACKMAN, T. R., D. J. IRSCHICK, K. DE QUEIROZ, J. B.
LOSOS, AND A. LARSON. 2002. Molecular phyloge-
netic perspective on evolution of lizards of the
Anolis grahami series. Journal of Experimental
Zoology 294: 1?16.
JACKMAN, T. R., A. LARSON, K. DE QUEIROZ, AND J. B.
LOSOS. 1999. Phylogenetic relationships and tempo
of early diversification in Anolis lizards. Systematic
Biology 48: 254?285.
KASS, R. E., AND A. E. RAFTERY. 1995. Bayes factors.
Journal of the American Statistical Association 90:
773?795.
LARGET, B., AND D. L. SIMON. 1999. Markov chain
Monte Carlo algorithms for the Bayesian analysis
of phylogenetic trees. Molecular Biology and
Evolution 16: 750?759.
LARSON, A. 1998. The comparison of morphological and
molecular data in phylogenetic systematics, pp.
275?296. In R. DeSalle and B. Schierwater (eds.),
Molecular Approaches to Ecology and Evolution.
Basel, Switzerland: Birkha?user Verlag.
LAZELL, J. D., JR. 1969. The genus Phenacosaurus
(Sauria: Iguanidae). Breviora 325: 1?24.
LAZELL, J. D., JR. 1972. The anoles (Sauria: Iguanidae)
of the Lesser Antilles. Bulletin of the Museum of
Comparative Zoology 143: 1?115.
LEE, M. S. Y. 2000. Tree robustness and clade
significance. Systematic Biology 49: 829?836.
LEWIS, P. O. 2001. A likelihood approach to estimating
phylogeny from discrete morphological character
data. Systematic Biology 50: 913?925.
LEWIS, P. O., M. T. HOLDER, AND K. E. HOLSINGER.
2005. Polytomies and Bayesian phylogenetic infer-
ence. Systematic Biology 54: 241?253.
LOSOS, J. B. 2009. Lizards in an Evolutionary Tree:
Ecology and Adaptive Radiation of Anoles. Berke-
ley and Los Angeles, California: Univ. of California
Press.
LOSOS, J. B., M. L. WOOLLEY, D. L. MAHLER, O. TORRES-
CARVAJAL, K. E. CRANDELL, E. W. SCHAAD, A. E.
NARVA?EZ, F. AYALA-VARELA, AND A. HERREL. 2012.
Notes on the natural history of the little-known
Ecuadorian horned anole, Anolis proboscis. Bre-
viora 531: 1?17.
MADDISON, D. R., AND W. P. MADDISON. 2001.
MacClade: Analysis of Phylogeny and Character
Evolution. Version 4.03. Sunderland, Massachu-
setts: Sinauer Associates.
MADDISON, W. P., AND D. R. MADDISON. 2011.
Mesquite: A Modular System for Evolutionary
Analysis. Version 2.75. Available from: http://
mesquiteproject.org.
MYERS, C. W., AND M. A. DONNELLY. 1996. A new
herpetofauna from Cerro Yav??, Venezuela: first
results of the Robert G. Goelet American Muse-
um?TERRAMAR expedition to the northwestern
Tepuis. American Museum Novitates 3172: 1?56.
MYERS, C. W., E. E. WILLIAMS, AND R. W. MCDIARMID.
1993. A new anoline lizard (Phenacosaurus) from the
highland of Cerro de la Neblina, southern Vene-
zuela. American Museum Novitates 3070: 1?15.
NICHOLSON, K. E. 2002. Phylogenetic analysis and a test
of the current infrageneric classification of Norops
(Beta Anolis). Herpetological Monographs 16: 93?
120.
NICHOLSON, K. E., R. E. GLOR, J. J. KOLBE, A. LARSON, S.
B. HEDGES, AND J. B. LOSOS. 2005. Mainland
colonization by island lizards. Journal of Biogeog-
raphy 32: 929?938.
PAGEL, M., AND A. MEADE. 2005. Mixture models in
phylogenetic inference, pp. 121?139. In O. Gascuel
(ed.), Mathematics of Evolution and Phylogeny.
Oxford: Oxford Univ. Press.
PETERS, J. A. 1964. Dictionary of Herpetology: A Brief
and Meaningful Definition of Words and Terms
Used in Herpetology. New York: Hafner VII.
396 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7
PETERS, J. A., AND V. G. ORCES. 1956. A third leaf-nosed
species of the lizard genus Anolis from South
America. Breviora 62: 1?8.
POE, S. 1998. Skull characters and the cladistic
relationships of the Hispaniolan dwarf twig Anolis.
Herpetological Monographs 12: 192?236.
POE, S. 2004. Phylogeny of anoles. Herpetological
Monographs 18: 37?89.
POE, S. 2005. A study of the utility of convergent
characters for phylogeny reconstruction: do eco-
morphological characters track evolutionary history
in Anolis lizards? Zoology 108: 337?343.
POE, S., F. AYALA, I. M. LATELLA, T. L. KENNEDY, J. A.
CHRISTENSEN, L. N. GRAY, N. J. BLEA, B. M. ARMIJO,
AND E. W. SCHAAD. 2012. Morphology, phylogeny,
and behavior of Anolis proboscis. Breviora 530: 1?11.
POE, S., AND R. IBA?N?EZ. 2007. A new species of Anolis
lizard from the cordillera de Talamanca of western
Panama. Journal of Herpetology 41: 263?270.
POE, S., I. M. LATELLA, M. J. RYAN, AND E. W. SCHAAD.
2009a. A new species of Anolis lizard (Squamata,
Iguania) from Panama. Phyllomedusa 8: 81?87.
POE, S., J. VELASCO, K. MIYATA, AND E. E. WILLIAMS.
2009b. Descriptions of two nomen nudum species
of Anolis lizard from northwestern South America.
Breviora 516: 1?16.
POE, S., AND C. YAN?EZ-MIRANDA. 2007. A new species of
phenacosaur Anolis from Peru. Herpetologica 63:
219?223.
POE, S., AND C. YAN?EZ-MIRANDA. 2008. Another new
species of green Anolis (Squamata: Iguania) from
the eastern Andes of Peru. Journal of Herpetology
42: 564?571.
POE, S., C. YAN?EZ-MIRANDA, AND E. LEHR. 2008. Notes on
variation in Anolis boettgeri Boulenger 1911, assess-
ment of the status of Anolis albimaculatus Henle and
Ehlr 1991, and description of a new species of Anolis
(Squamata: Iguania) similar to Anolis boettgeri.
Journal of Herpetology 42: 251?259.
POSADA, J. L., AND K. A. CRANDALL. 1998. Modeltest:
testing the model of DNA substitution. Version 3.7
(June 2005). Bioinformatics 14: 817?818.
RODRIGUES, M. T. 1988. A new anole of the punctatus
group from central Amazonia (Sauria: Iguanidae).
Pape?is Avulsos de Zoologia 36: 333?336.
RODRIGUES, M. T., V. XAVIER, G. SKUK, AND D. PAVAN.
2002. New specimens of Anolis phyllorhinus
(Squamata, Polychrotidae): the first female of the
species and of proboscid anoles. Pape?is Avulsos De
Zoologia 42: 363?380.
RONQUIST, F., AND J. P. HUELSENBECK. 2003. MrBayes 3:
Bayesian phylogenetic inference under mixed
models. Bioinformatics 19: 1572?1574.
RUEDA, J. V., AND J. I. HERNA?NDEZ CAMACHO. 1988.
Phenacosaurus inderenae (Sauria: Iguanidae),
nueva especie gigante, proveniente de la cordillera
oriental de Colombia. Trianea 2: 339?350.
RUEDA ALMONACID, J. V. 1989. Un nuevo y extraordi-
nario saurio de color rojo (Iguanidae: Anolis) para
la Cordillera Occidental de Colombia. Trianea 3:
85?92.
SAVAGE, J. M. 2002. The Amphibians and Reptiles of
Costa Rica. Herpetofauna between Two Conti-
nents, between Two Seas. Chicago: Univ. of
Chicago Press.
SAVAGE, J. M., AND C. GUYER. 1989. Infrageneric
classification and species composition of the anole
genera, Anolis, Ctenonotus, Dactyloa, Norops, and
Semiurus (Sauria: Iguanidae). Amphibia-Reptilia
10: 105?116.
SAVAGE, J. M., AND J. J. TALBOT. 1978. The giant anoline
lizards of Costa Rica and western Panama. Copeia
1978: 480?492.
SCHNEIDER, C. J., J. B. LOSOS, AND K. DE QUEIROZ. 2001.
Evolutionary relationships of the Anolis bimacula-
tus group from the northern Lesser Antilles.
Journal of Herpetology 35: 1?12.
SCHOENER, T. W. 1969. Size patterns in West Indian
Anolis lizards: I. Size and species diversity.
Systematic Zoology 18: 386?401.
SHOCHAT, D., AND H. C. DESSAUER. 1981. Comparative
study of albumins of Anolis lizards of the
Caribbean islands. Comparative Biochemistry and
Physiology 68A: 67?73.
SOKAL, R. R., AND F. J. ROHLF. 1995. Biometry: The
Principles and Practice of Statistics in Biological
Research. 3rd ed. San Francisco: W.H. Freeman.
SPENCER, M. A., AND G. S. SPENCER. 1993. MacMorph
Image Analysis Software, Version 2.1. Stony Brook,
New York.
SUZUKI, Y., G. V. GLAZKO, AND M. NEI. 2002. Over-
credibility of molecular phylogenies obtained by
Bayesian phylogenetics. Proceedings of the Nation-
al Academy of Science 99: 16138?16143.
SWOFFORD, D. L. 1991. When are phylogeny estimates
from molecular and morphological data incongru-
ent? pp. 295?333. In M. M. Miyamoto and J.
Cracraft (eds.), Phylogenetic Analysis of DNA
Sequences. New York: Oxford Univ. Press.
SWOFFORD, D. L. 2002. PAUP*. Phylogenetic Analysis
Using Parsimony (* and Other Methods). Version
4. Sunderland, Massachusetts: Sinauer Associates.
SWOFFORD, D. L., AND S. H. BERLOCHER. 1987. Inferring
evolutionary trees from gene frequency data under
the principle of maximum parsimony. Systematic
Zoology 36: 293?325.
TEMPLETON, A. R. 1983. Phylogenetic inference from
restriction endonuclease cleavage site maps with
particular reference to the evolution of humans
and apes. Evolution 37: 221?244.
THIELE, K. 1993. The holy grail of the perfect character:
the cladistic treatment of morphometric data.
Cladistics 9: 275?304.
TORRES-CARVAJAL, O. 2007. Phylogeny and biogeogra-
phy of a large radiation of Andean lizards (Iguania,
Stenocercus). Zoologica Scripta 36: 311?326.
UETZ, P. (ED.) [Internet]. The Reptile Database [cited
2012 Mar 10]. Available from: http://www.reptile-
database.org/.
UGUETO, G. N., G. RIVAS, T. BARROS, AND E. N. SMITH.
2009. A revision of the Venezuelan anoles II:
redescription of Anolis squamulatus Peters 1863
PHYLOGENY OF THE DACTYLOA N Castan?eda and de Queiroz 397
and Anolis tigrinus Peters 1863 (Reptilia: Poly-
chrotidae). Caribbean Journal of Science 45: 30?
51.
UGUETO, G. N., G. RIVAS-FUENMAYOR, T. BARROS, S. J.
SA?NCHEZ-PACHECO, AND J. E. GARCI?A-PE?REZ. 2007.
A revision of the Venezuelan Anoles I: a new Anolis
species from the Andes of Venezuela with the
redescription of Anolis jacare Boulenger 1903
(Reptilia: Polychrotidae) and the clarification of
the status of Anolis nigropunctatus Williams 1974.
Zootaxa 1501: 1?30.
UNDERWOOD, G. 1959. The anoles of the eastern
Caribbean (Sauria, Iguanidae): revisionary notes.
Bulletin of the Museum of Comparative Zoology
121(part III): 191?226.
VELASCO, J. A., P. D. A. GUTIE?RREZ-CA?RDENAS, AND A.
QUINTERO-ANGEL. 2010. A new species of Anolis of
the aequatorialis group (Squamata: Iguania) from
the central Andes of Colombia. Herpetological
Journal 20: 231?236.
WIENS, J. J. 1995. Polymorphic characters in phyloge-
netic systematics. Systematic Biology 44: 482?500.
WIENS, J. J. 1998. Testing phylogenetic methods with
tree congruence: phylogenetic analysis of polymor-
phic morphological characters in phrynosomatid
lizards. Systematic Biology 47: 427?444.
WIENS, J. J. 2000. Coding morphological variation
within species and higher taxa for phylogenetic
analysis, pp. 115?145. In J. J. Wiens (ed.),
Phylogenetic Analysis of Morphological Data.
Washington, DC: Smithsonian Institution Press.
WIENS, J. J. 2001. Character analysis in morphological
phylogenetics: problems and solutions. Systematic
Biology 50: 689?699.
WIENS, J. J., AND M. R. SERVEDIO. 1997. Accuracy of
phylogenetic analysis including and excluding
polymorphic characters. Systematic Biology, 46:
332?345.
WILLIAMS, E. E. 1965. South American Anolis (Sauria,
Iguanidae): two new species of the punctatus
group. Breviora 233: 1?15.
WILLIAMS, E. E. 1974. South American Anolis: three
new species related to Anolis nigrolineatus and A.
dissimilis. Breviora 422: 1?15.
WILLIAMS, E. E. 1975. South American Anolis: Anolis
parilis, new species, near A. mirus Williams.
Breviora 434: 1?8.
WILLIAMS, E. E. 1976a. West Indian anoles: a taxonomic
and evolutionary summary. 1. Introduction and a
species list. Breviora 440: 1?21.
WILLIAMS, E. E. 1976b. South American anoles: the
species groups. Pape?is Avulsos de Zoologia 29:
259?268.
WILLIAMS, E. E. 1979. South American anoles: the
species groups. 2. The proboscis anoles (Anolis
laevis group). Breviora 449: 1?19.
WILLIAMS, E. E. 1982. Three new species of the Anolis
punctatus complex from Amazonian and inter-
Andean Colombia, with comments on the eastern
members of the punctatus species group. Breviora
467: 1?38.
WILLIAMS, E. E. 1985. New or problematic Anolis from
Colombia. IV. Anolis antioquiae, new species of the
Anolis eulaemus subgroup from western Colombia.
Breviora 482: 1?9.
WILLIAMS, E. E. 1988. New or problematic Anolis from
Colombia. V. Anolis danieli, a new species of the
latifrons species group and a reassessment Anolis
apollinaris Boulengeri, 1919. Breviora 489: 1?25.
WILLIAMS, E. E. 1989. A critique of Guyer and Savage
(1986): cladistic relationships among anoles
(Sauria: Iguanidae): are the data available to
reclassify anoles? pp. 434?478. In C. A. Woods
(ed.), Biogeography of the West Indies: Past,
Present and Future. Gainsville, Florida: Sandhill
Crane Press.
WILLIAMS, E. E. 1992. New or problematic Anolis from
Colombia. VII. Anolis lamari, a new anole from the
cordillera oriental of Colombia, with a discussion of
tigrinus and punctatus species groups boundaries.
Breviora 495: 1?24.
WILLIAMS, E. E., AND N. ACOSTA B. 1996. The Anolis
Handlist. Version 1.0b0. Cambridge, Massachu-
setts: Museum of Comparative Zoology.
WILLIAMS, E. E., AND W. E. DUELLMAN. 1967. Anolis
chocorum, a new punctatus-like anole from Darien,
Panama (Sauria, Iguanidae). Breviora 256: 1?12.
WILLIAMS, E. E., AND W. E. DUELLMAN. 1984. Anolis
fitchi, a new species of the Anolis aequatorialis
group from Ecuador and Colombia, pp. 257?266.
In R. A. Siegler, L. E. Hunt, J. L. Knight, L.
Malaret, and N. L. Zuschlag (eds.), Vertebrate
Ecology and Systematics?A Tribute to Henry S.
Fitch. Lawrence, Kansas: University of Kansas,
Museum of Natural History Special Publication
No. 10.
WILLIAMS, E. E., AND R. MITTERMEIER. 1991. A Peruvian
phenacosaur (Squamata: Iguania). Breviora 492:
1?16.
WILLIAMS, E. E., M. J. PRADERIO, AND S. GORZULA. 1996.
A phenacosaur from Chimanta? Tepui, Venezuela.
Breviora 506: 1?15.
WILLIAMS, E. E., H. RAND, A. S. RAND, AND R. J.
O?HARA. 1995. A computer approach to the
comparison and identification of species in difficult
taxonomic groups. Breviora 502: 1?47.
YA?NEZ-MUN?OZ, M. H., M. A. URGILE?S, M. ALTAMIRANO-
BENAVIDES, AND S. R. CA?CERES. 2010. Redescrip-
cio?n de Anolis proboscis Peters & Orce?s (Reptilia:
Polychrotidae), con el descubrimiento de las
hembras de la especie y comentarios sobre su
distribucio?n y taxonom??a. Avances 2: B7?B15.
YANG, S. Y., M. SOULE?, AND G. C. GORMAN. 1974. Anolis
lizards of the eastern Caribbean: a case study in
evolution. I. Genetic relationships, phylogeny, and
colonization sequence of the roquet group. Sys-
tematic Zoology 23: 387?399.
ZAR, J. H. 1999. Biostatistical Analysis. 4th Ed. Upper
Saddle River, New Jersey: Prentice Hall.
398 Bulletin of the Museum of Comparative Zoology, Vol. 160, No. 7