Syst. Biol. 53(2):193–215, 2004
Society of Systematic Biologists
ISSN: 1063-5157 print / 1076-836X online
Molecular Phylogeny, Historical Biogeography, and Divergence Time Estimates
for Swallowtail Butterﬂies of the Genus Papilio (Lepidoptera: Papilionidae)
EVGUENI V. ZAKHAROV,
MICHAEL S. CATERINO,
AND FELIX A. H. SPERLING
Department of Biological Sciences, University of Alberta, Edmonton, Alberta T6G 2E9, Canada; E-mail: email@example.com (E.V.Z.);
E-mail: firstname.lastname@example.org (F.A.H.S.)
Santa Barbara Museum of Natural History, 2559 Puesta del Sol Road, Santa Barbara, California 93105, USA
Abstract.— Swallowtail butterﬂies are recognized as model organisms in ecology, evolutionary biology, genetics, and con-
servation biology but present numerous unresolved phylogenetic problems. We inferred phylogenetic relationships for 51
of about 205 species of the genus Papilio (sensu lato) from 3.3-Kilobase (kb) sequences of mitochondrial and nuclear DNA
(2.3 kb of cytochrome oxidases I and II and 1.0 kb of elongation factor 1α). Congruent phylogenetic trees were recovered
within Papilio from analyses of combined data using maximum likelihood, Bayesian analysis, and maximum parsimony
bootstrap consensus. Several disagreements with the traditional classiﬁcation of Papilio were found. Five major previously
hypothesized subdivisions within Papilio were well supported: Heraclides, Pterourus, Chilasa, Papilio (sensu stricto), and Elep-
pone. Further studies are required to clarify relationships within traditional “Princeps,” which was paraphyletic. Several
biologically interesting characteristics of Papilio appear to have polyphyletic origins, including mimetic adults, larval host
associations, and larval morphology. Early diversiﬁcation within Papilio is estimated at 55–65 million years ago based on
a combination of biogeographic time constraints rather than fossils. This divergence time suggests that Papilio has slower
apparent substitution rates than do Drosophila and ﬁg-pollinating wasps and/or divergences corrected using best-ﬁt substi-
tution models are still being consistently underestimated. The amount of sequence divergence between Papilio subdivisions
is equivalent to divergences between genera in other tribes of the Papilionidae, and between genera of moths of the noctuid
subfamily Heliothinae. [Character evolution; fossils; mimicry; molecular systematics; swallowtail butterﬂies; substitution
The utility of DNA sequence comparisons for infer-
ring phylogenetic relationships has been demonstrated
thoroughly (e.g., Hasegawa and Yano, 1984; Gielly and
Taberlet, 1994; Simon et al., 1994). However, their use
for dating divergences has been much more controver-
sial. Early hopes for a molecular clock (Zuckerkandl
and Pauling, 1962, 1965) via evolution in accordance
with neutral theory (Kimura, 1983) have generally not
been satisﬁed. Rather, most studies have revealed sub-
stantial variation in rates of evolution across genes and
lineages (Britten, 1986; Gillespie, 1986; Avise, 1994; Li,
1997; Pawlowski et al., 1997; Page and Holmes, 1998;
Hebert et al., 2002; Soltis et al., 2002), and methods of
divergence time estimation must account for this varia-
tion (Rambaut and Bromham, 1998; Thorne et al., 1998;
Inferring absolute divergence dates for a given tree
requires an accurate phylogenetic reconstruction includ-
ing model-corrected branch lengths and reasonably well-
established calibration points for the group in ques-
tion. Accurate estimation of phylogeny remains perhaps
the greatest challenge for systematists. Among methods
commonly employed, maximum likelihood (ML) has
been a consistent and efﬁcient way to estimate phylo-
genies under a variety of simulated conditions where
maximum parsimony (MP) and distance methods are
expected to fail (Felsenstein, 1978, 1981; Huelsenbeck,
1995). ML has been remarkably resistant to variations in
models and model parameters (Yang, 1996). However,
robust ML analyses of large data sets are computation-
ally limited (Sanderson and Kim, 2000), whereas MP and
the various distance-based methods are less hindered by
large numbers of taxa (Hillis, 1996). Bayesian inference
of phylogeny has the beneﬁt of a parametric statistical
framework for analyzing DNA sequence data and, con-
trary to standard ML analysis, requires fewer computa-
tional resources because it does not necessarily attempt
to ﬁnd the globally optimal ML score (Huelsenbeck et al.,
2001). The estimation of branch support in Bayesian anal-
ysis accompanies tree estimation, thus eliminating the
need for separate time-intensive nonparametric boot-
strapping (Larget and Simon, 1999).
Calibrating local molecular clock(s) for a given phy-
logenetic hypothesis may be accomplished via fossils or
vicariance events, the latter chosen to represent one or
morenodes in a tree(Jacobs and Pilbeam, 1980; Smith and
Coss, 1984; Beerli et al., 1996). Calibration of a molecular
clock based on vicariance events presents a major prob-
lem for absolute dating when the selected event separates
previously continguous areas under a variety of scenar-
ios, e.g., gradual vicariances that occurred over extended
periods of time or multiple sequential events in which the
same areas were separated and reunited repeatedly over
time. Although fossils are often considered reliable for
calibration purposes, they pose other hidden dangers,
such as incorrect systematic placement. Fossils are also
constrained to the minimum age and cannot ﬁx dates of
internal nodes (Smith et al., 1992; Sanderson, 1997).
Here, we present a study of swallowtail butterﬂies of
the genus Papilio sensu lato (Lepidoptera: Papilionidae),
which includes about 206 species and represents more
than one third of all Papilionidae, which has about 551
recognized species (Ha ¨user et al., 2002). As one of the
most well-known and broadly studied insect groups,
swallowtails are recognized as model organisms in evo-
lutionary biology, ecology, genetics, and conservation bi-
ology (Collins and Morris, 1985; Scriber et al., 1995). Yet
despite numerous relevant recent studies (e.g., Ae, 1979;
Hancock, 1983; Igarashi, 1984; Miller, 1987; Tyler et al.,
1994; Scriber et al., 1995; Aubert et al., 1999; Caterino and
194 SYSTEMATIC BIOLOGY VOL.
Sperling, 1999; Reed and Sperling, 1999; Yagi et al., 1999;
Caterino et al., 2001), the phylogeny of Papilio is far from
resolved, and many coevolutionary and biogeographic
hypotheses founded on poorly supported phylogenetic
reconstructions hang in limbo. Answers to many of these
questions rely not only on accurate phylogenetic resolu-
tion but also on approximate dates for some of the more
signiﬁcant phylogenetic events (e.g., host switches, ori-
gins of mimicry). In this study, we made an attempt to
establish some of these dates. Divergence dates were es-
timated for nodes of the tree, and these dates were used
along with phylogenetic pattern to explore historical bio-
geography and the evolution of morphological and eco-
logical traits of Papilio.
A summary of major classiﬁcations of Papilio is pre-
sented in Table 1. Munroe (1961) divided Papilio into
TABLE 1. Classiﬁcation of Papilio by different authors.
Species group (Hancock, 1983) Munroe (1961) Hancock (1983) Igarashi (1984) Ha ¨user et al. (2002)
P. clytia (clytia) Section I Chilasa (Chilasa) Chilasa Chilasa (Chilasa)
P. agestor (epycides)-“”--“”--“”--“”-
P. veiovis -“”--“”--“”--“”-
P. laglazei -“”--“”--“”--“”-
P. elwesi Section II(A) Chilasa (Agehana) Agehana Chilasa (Agehana)
P. anactus (anactus)-“”- Eleppone Chilasa P. (Eleppone)
P. machaon (alexanor, indra, machaon, -“”- Papilio Papilio P. (Papilio)
hippocrates, polyxenes, zelicaon, hospiton)
P. xuthus (xuthus)-“”- Princeps (Sinoprinceps)-“”- P. (Sinoprinceps)
P. demolion -“”- Princeps (Menelaides) Papilio, Euchenor P. (Menelaides)
P. helenus (helenus)-“”--“”- Menelaides; Papilio -“”-
P. protenor (protenor, macilentus)-“”--“”- Menelaides -“”-
P. bootes -“”--“”--“”--“”-
P. memnon (memnon, rumanzovia)-“”--“”--“”--“”-
P. polytes (polytes)-“”--“”--“”--“”-
P. nephelus (nephelus)-“”--“”--“”--“”-
P. gambrisius (aegeus)-“”--“”--“”--“”-
P. amynthor -“”--“”--“”--“”-
P. fuscus (hipponous)-“”--“”- Achillides -“”-
P. paris (paris, bianor, maackii)-“”- Princeps (Princeps)-“”- P. (Achillides)
P. palinurus -“”--“”--“”--“”-
P. peranthus -“”--“”--“”--“”-
P. ulysses -“”--“”--“”--“”-
P. demoleus (demoleus, demodocus)-“”--“”- Papilio P. (Princeps)
P. menesteus -“”--“”--“”--“”-
P. delalandei (delalandei) Section II(B) -“”-?-“”-
P. phorcas (phorcas, dardanus, -“”--“”--“”--“”-
P. hesperus (nobilis)-“”--“”--“”--“”-
P. leucotaenia -“”--“”--“”--“”-
P. cynorta -“”- Princeps (Druryia)-“”- P. (Druryia)
P. nireus (oribazus, epiphorbas)-“”- Achillides -“”-
P. zalmoxis -“”--“”-?-“”-
P. antimachus -“”--“”--“”--“”-
P. r e x (rex)-“”--“”--“”--“”-
P zenobia -“”--“”--“”--“”-
P. glaucus (glaucus, canadensis, rutulus, Section III Pterourus (Pterourus) Papilio P. (Pterourus)
P. troilus (troilus, palamedes, pilumnus, -“”--“”- Pterourus -“”-
P. thoas (thoas, cresphontes, astyalus) Section IV(A) Heraclides Papilio P. (Heraclides)
P. torquatus (torquatus, hectorides) Section IV(B) ? -“”-
P. anchisiades (anchisiades, erostratus)-“”--“”- Chilasa -“”-
P. zagreus Section V Pterourus (Pyrrhosticta)? P. (Pterourus)
P. scamander (birchalli, scamander)-“”--“”--“”--“”-
P. homerus (garamas)-“”--“”--“”--“”-
Note: Species groups shown in bold are represented in this study (by the species listed in brackets).
ﬁve sections, which he did not designate as subgenera
because of the lack of any simple system of adult char-
acters. Hancock (1983), in his explicitly cladistic estima-
tion of relationships within Papilio, recognized six gen-
era based on phylogenetic evidence and their inferred
evolutionary antiquity, but that phylogeny still suffered
from lack of character justiﬁcation. Another classiﬁcation
was proposed by Igarashi (1984) based on morphology of
immature stages, but this work was not complete in rep-
resenting all of Hancock’s genera. Igarashi recognized
seven genera, with numerous discrepancies between this
treatment and that of Hancock (1983). Hancock’s clas-
siﬁcation was criticized by Miller (1987), who did not
ﬁnd justiﬁcation for elevating Papilio subdivisions to the
generic level. However, some elevation is seen in the lat-
est widely available checklist of swallowtail butterﬂies
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
(Ha ¨user et al., 2002), where Chilasa is treated as a
Successful hand-pairing of swallowtail butterﬂies by
Clarke (1952) gave rise to studies of experimental hy-
bridization in Papilio (e.g., Clarke and Sheppard, 1955,
1956, 1957; Ae, 1960, 1962, 1990; Remington, 1960). Data
on egg viability, sex ratio in hybrid progeny, and fertility
’s were summarized in the form of a biological in-
compatibility index used to evaluate Papilio relationships
(Ae, 1979, 1995; Scriber, 1995a).
Partly because of the limitations of traditional ap-
proaches and with the development of new systematic
techniques, the classiﬁcation of Papilio has received sig-
niﬁcant attention in recent years. Relationships among
species within the P. machaon and P. glaucus-troilus
species groups were studied based on allozyme varia-
tion (Sperling, 1987; Hagen and Scriber, 1991). Restriction
fragment length polymorphism of mitochondrial DNA
(mtDNA) was used to compare taxa within the same
species groups in later studies (Sperling, 1991, 1993a,
1993b; Sperling and Harrison, 1994, Tyler et al., 1994).
Phylogenetic relationships within Papilio have also been
analyzed using DNA sequences of a variety of genes
(28S, cyt b, EF-1α, ND1, ND5, COI, COII), but these stud-
ies have been conﬁned to single species groups (Vane-
Wright et al., 1999) or local geographic areas (Yagi et al.,
1999) or have included limited sampling across Papilio
subdivisions (Aubert et al., 1999; Caterino and Sperling,
1999; Reed and Sperling, 1999).
Although Munroe’s (1961) hypothesis of relationships
within Papilio has been widely used for classiﬁcation, nu-
merous details have been challenged. We focused on ﬁve
areas that include species that are taxonomically prob-
lematic key species for testing biogeographic hypothe-
ses (e.g., P. anactus), evolution of mimicry (e.g., Chilasa,
P. nobilis), and host-plant associations (e.g., P. alexanor),
and species that are rare and insufﬁciently studied (e.g.,
P. esperanza). Prior to the discovery of its larva, the enig-
matic Mexican swallowtail, Papilio esperanza, was var-
iously placed in four different species groups in two
Papilio subgenera (Pterourus or Heraclides). Larval char-
acters suggest yet a different species group of Pterourus
(Beutelspacher, 1975; Hancock, 1983; Tyler et al., 1994),
and the issue remains unresolved. The systematic posi-
tion of Papilio anactus is similarly unclear. It is presently
placed in a monotypic subgenus (Eleppone), with rela-
tionships somewhere “between the subgenera Heraclides
and Chilasa” (Hancock, 1970: 53). A third species, Papilio
nobilis, has wavered between the phorcas and hesperus
groups (Munroe, 1961, and Hancock, 1983, respectively).
Its placement is important for the understanding of
mimicry evolution in the P. phorcas group (Vane-Wright
et al., 1999). Fourth, the monophyly and rank of the
danaid-mimicking Chilasa are uncertain. Munroe (1961)
split its members among two Papilio subunits, but Han-
cock placed them together in a single genus, considered
to be the sister taxon of Eleppone. We also revisited the
question of Papilio alexanor. Although the relationships of
this odd European Apiaceae feeder have been examined
several times (Aubert et al., 1999; Caterino and Sperling,
1999; Reed and Sperling, 1999), no strong resolution has
been obtained, and denser sampling was used to help
reﬁne its placement.
ATERIALS AND METHODS
We added sequences for 31 species of Papilio sensu
lato to the 23 available previously (Caterino and
Sperling, 1999; Reed and Sperling, 1999; Caterino et al.,
2001). Overall, we include sequences for 54 specimens
from 51 Papilio species, which represent almost 25% of all
species in this large genus. Species were selected to give
more complete representation of major species groups
and subgenera within Papilio. Representatives of 26 of
42 species groups (sensu Hancock, 1983) were included.
The sampled species represent all genera recognized by
Munroe (1961), Hancock (1983), and Ha¨user et al., (2002)
and all except two genera (Agehana and Euchenor) rec-
ognized by Igarashi (1984). Outgroups were 16 species
of Papilionidae from tribes other than Papilionini, and
Pieris napi from the putative sister family, Pieridae.
Newly sampled taxa, sources of material, and GenBank
accession numbers for new materials are given in
We sequenced about 2.3 kilobases (kb) of the mitochon-
drial genes cytochrome oxidase subunit I (COI), tRNA-
leucine, and cytochrome oxidase subunit II (COII) and
about 1.0 kb of the nuclear protein-coding gene elonga-
tion factor 1 alpha (EF-1α). Phylogenetic utility of both
COI-COII and EF-1α has been widely demonstrated (e.g.,
Simon et al., 1994; Cho et al., 1995; Mitchell et al., 1997),
and a substantial database of lepidopteran sequences al-
ready exists for these genes (Sperling, 2003). These se-
quences are valuable sources for studies of the evolution
of these genes and for reconstruction of the global phy-
logeny for Lepidoptera. We followed previous work on
swallowtail phylogeny that utilized both COI-COII and
EF-1α (Caterino and Sperling, 1999; Reed and Sperling,
1999; Caterino et al., 2001) to expand the amount of ana-
lyzed data available for global analyses.
Total genomic DNA was extracted using a Qiagen
DNeasy tissue kit. Polymerase chain reactions (PCRs)
were performed with a Biometra TGradient thermal cy-
cler using a hot start in which Taq Polymerase was added
at the end of an initial 2-min denaturation at 94
step was followed by 35 cycles of 1 min at 94
C, 1 min at
C (depending on primer combinations), and 1 min
C and then a 7-min ﬁnal extension at 72
products were cleaned using the Qiagen QIAquick PCR
puriﬁcation kit when only a single DNA band was visi-
ble in a gel or using a combination of gel separation and
subsequent puriﬁcation with the Qiagen QIAEX II gel
extraction kit when more than one band was observed.
Sequencing reactions were carried out using a DYEnamic
196 SYSTEMATIC BIOLOGY VOL.
TABLE 2. List of new material examined in present study.
Locality COI-COII EF-1α
Papilio (Princeps) bianor Taiwan: Taipei AY457572 AY457603
P. (Pr.) maackii Japan: Gifu Pref. AY457573 AY457604
P. (Pr.) paris China: Guangzhou AY457574 AY457605
P. (Pr.) helenus Japan: Gifu Pref. AY457575 AY457619
P. (Pr.) hipponous Philippines: Marinduque AY457576 AY457620
P. (Pr.) macilentus Japan: Mt. Takao AY457577 AY457622
P. (Pr.) memnon Japan: Gifu Pref. AY457578 AY457623
P. (Pr.) nephelus Malaysia: Penang AY457579 AY457624
P. (Pr.) polytes Malaysia: Penang AY457580 AY457627
P. (Pr.) protenor Japan: Aichi Pref. AY457581 AY457628
P. (Pr.) rumanzovia Philippines AY457582 AY457631
P. (Pr.) aegeus Australia: Queensland AY457583 AY457611
P. (Pr.) dardanus Madagascar: Fianarantsoa AY457584 AY457612
P. (Pr.) delalandei Madagascar: Fianarantsoa AY457585 AY457613
P. (Pr.) rex (two specimens) Kenya: Kakamega AY457586 AY457630
Kenya: Kakamega AY457587 AY457629
P. (Pr.) demodocus Madagascar: Radriarmasy AY457588 AY457614
P. (Pr.) epiphorbas Madagascar: Radriarmasy AY457589 AY457615
P. (Pr.) nobilis
Kenya: Nairobi AY457590 AY457625
P. (Pr.) oribazus Madagascar: Fianarantsoa AY457591 AY457626
P. (Eleppone) anactus Australia: Queensland AY457592 AY457608
P. (Papilio) hippocrates Japan: Gifu Pref. AY457593 AY457621
P. (Chilasa) clytia Malaysia: Penang AY457594 AY457606
P. (Ch.) epycides Taiwan: Taoyuan: Gapyi AY457595 AY457607
P. (Pterourus) birchalli Costa Rica: Guatuso de Alajuela AY457596 AY457610
P. (Pt.) esperanza Mexico: Oxaca AY457597 AY457617
P. (Heraclides) astyalus Brazil: Campinas AY457598 AY457609
P. (H.) erostratus EI Salvador AY457599 AY457616
P. (H.) hectorides Brazil: Campinas AY457600 AY457618
P. (H.) thoas French Guiana: Pointe Macouria AY457601 AY457632
P. (H.) torquatus Brazil: Campinas AY457602 AY457633
Subgeneric names correspond to genera of Hancock (1983).
Same specimen used by Vane-Wright et al. (1999).
ET terminator cycle sequencing kit (Amersham Pharma-
cia Biotech, Cleveland, OH). Sequenced products were
ﬁltered through Sephadex-packed columns, dried, resus-
pended, and fractionated on an ABI 377 automated se-
quencer. Part of the sequences were obtained following
molecular procedures described previously (Caterino
et al., 2001). All fragments were sequenced in both di-
rections. Nucleotide sequences of the primers have been
described previously (Caterino and Sperling, 1999; Reed
and Sperling, 1999; Caterino et al., 2001). Sequences were
assembled into contiguous arrays using Sequencher 4.1
(GeneCode Corp., Ann Arbor, MI).
Alignment of Sequence Data
Sequences of COI-COII genes were aligned against the
published sequence from Drosophila yakuba (Clary and
Wolstenholme, 1985) with multiple sequence alignment
using ClustalX 1.81 (Thompson et al., 1997) with the de-
fault settings (gap opening = 10, gap extension = 0.20)
followed by adjustment by eye. EF-1α gene sequences
did not contain any introns and were aligned against
sequence from Bombyx mori (Kamiie et al., 1993) by eye
with the aid of SeAl 2.0 (Rambault, 2002). To determine
codon positions, we used MacClade 4.0 (Maddison and
4.0b8-b10 (Swofford, 1998) was used for all par-
simony, ML, bootstrap, and decay analyses. To get an ac-
curate estimation of phylogenetic relationships in Papilio
sensu lato, we followed the following strategies.
MP analyses.—To check the data for nucleotide bias
among taxa, we used the test of homogeneity of base
composition implemented in PAUP* using all characters
and using parsimony-informative sites alone. To reveal
possible incongruence among different genes, we
performed an incongruence length difference (ILD) test
(Farris et al., 1994) referred to in PAUP* as a partition
homogeneity test. We implemented the test under
parsimony with 100 random addition sequences of taxa
and used 500 replicates to generate the null distribu-
tion. The utility of the ILD test has been questioned
(Graham et al., 1998; Barker and Lutzoni, 2002; Darlu
and Lecointre, 2002). Therefore, to determine whether
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
the data partitions carried substantially different phy-
logenetic signals, we also analyzed combined and
partitioned data separately.
To evaluate the effect of different outgroup and in-
group combinations, we performed several MP analyses
for the combined data set: (1) with Pieris napi as an out-
group, and the 70 remaining taxa as the ingroup; (2) with
Pieris napi excluded from the analysis, two sequences
from Baronia brevicornis as an outgroup, and the 68 re-
maining taxa in the ingroup; and (3) with only Papilio
sensu lato in the ingroup, two outgroups (Pachliopta nep-
tunus and Eurytides marcellus) used in previous studies
(Caterino and Sperling, 1999; Reed and Sperling, 1999),
and all other taxa excluded from the analyses.
All parsimony analyses utilized heuristic searches:
starting trees determined by 100 random taxon addi-
tions, tree bisection–reconnection (TBR) branch swap-
ping, gaps treated as missing data, multistate charac-
ters treated as polymorphisms, and all characters equally
weighted. Robustness of the parsimony hypothesis was
tested with bootstrap analyses (Felsenstein, 1985) with
500 repetitions and 10 random taxon additions but oth-
erwise under the same conditions as for initial parsi-
mony searches. Constraint searches were carried out for
each data set to determine the number of additional
steps needed to accommodate alternative phylogenetic
hypotheses (Bremer, 1988), and decay indices were ex-
tracted using the program Autodecay (Eriksson, 1998).
ML analyses.—Prior to ML phylogenetic reconstruc-
tion, we applied a hierarchical likelihood ratio test to de-
termine how well competing substitution models ﬁt the
data. We tested models ranging from the simple Jukes–
Cantor model to the most parameter-rich general time
reversible (GTR) model. Using the program Modeltest
(Posada and Crandall, 1998), we calculated the test statis-
tic δ = 2 log , where is the ratio of the likelihood of the
null model divided by the likelihood of the alternative
model (Huelsenbeck and Crandall, 1997). Tests were per-
formed for both combined and partitioned data sets. All
model parameters were estimated from corresponding
(partition) MP trees and then ﬁxed during ML searches.
MP trees were used as starting trees for TBR branch
swapping under the best model supported by Model-
test. To obtain a measure of support for ML trees, we ran
1,000 bootstrap replicates under minimum-evolution cri-
teria based on the same model used in the ML analysis.
Bayesian analyses.—Bayesian phylogenetic analyses
were conducted for combined and partitioned data sets
with MrBayes 3.0 (Huelsenbeck and Ronquist, 2001) un-
der the same model that was selected for ML (GTR model
with gamma shape parameter and proportion of invari-
able sites: GTR++I). Speciﬁc nucleotide substitution
model parameter values were estimated as part of the
analysis, and each gene in the combined analysis was
allowed to have its own estimates. We ran four chains si-
multaneously, three heated and one cold. Each Markov
chain was started from a random tree and run for 10
erations, sampling the chains every 100th cycle. The log-
likelihood scores of sample points were plotted against
generation time to determine when the chain became sta-
tionary. All sample points prior to reaching stationarity
(2,000–3,000 trees) were discarded as burn-in samples. To
reduce bias in our results, we ran each partitioned analy-
sis twice and combined analyses three times, each begin-
ning with different starting trees, and compared their ap-
parent stationarity levels for convergence (Huelsenbeck
and Bollback, 2001). Data remaining after discarding
burn-in samples were used to generate a majority rule
consensus tree, where percentage of samples recover-
ing any particular clade represented the clade’s posterior
probability (Huelsenbeck and Ronquist, 2001). Probabil-
ities of ≥95% were considered indicative of signiﬁcant
support. The mean, variance, and 95% credibility inter-
val were calculated from the set of substitution param-
eters. Because only a single outgroup taxon is allowed
in MrBayes, we used Eurytides marcellus (Graphiini) to
root the trees. For dating analyses, the produced basal tri-
chotomy had to be resolved, and E. marcellus was pruned
from the treesleaving Pachliopta neptunus (Troidini) as the
only sister taxon to the ingroup (Papilio).
Selected morphological and ecological traits (includ-
ing host-plant associations, mimicry, presence of irides-
cent patches on wings, female secretion deposited on egg
surface, shape of larval minute body hairs, resting larval
behavior, and number of crochet rows on larval prolegs)
were scored based on literature or personal observations.
Character states were optimized on the ML phylogeny
using the program Mesquite (Maddison and Maddison,
2003). Both MP and ML character optimizations were
applied to reconstruct the ancestral states. Characters
were treated as unordered in MP reconstructions. ML
optimizations were done using the Markov k-state one-
parameter model (Lewis, 2001). Because of methodolog-
ical restrictions, ML optimization of larval feeding habits
was done only for primary host plants, and polymorphic
traits (use of multiple plant families) were reconstructed
using MP optimizations.
Other uncertainties with reconstruction of the ances-
tral states are related to incomplete information about
resting larval behavior and proleg structures. Optimiza-
tion of larval feeding habits may be tentative in the
absence of a well-supported phylogeny for Papilio out-
groups. The tribe Troidini was believed to be the sister
group for Papilionini, i.e., Papilio sensu Miller (1987), and
the tribe Leptocircini was considered their sister group
(Hancock, 1983; Caterino et al., 2001). However, some
studies have suggested a basal position of the genus Me-
andrusa within Papilionini (Aubert et al., 1999). Although
we were not able to sample this genus, we performed al-
ternative character optimizations placing Meandrusa as a
sister taxon of Papilio to test different hypothesis. Differ-
ences in ancestral state reconstructions due to choice of
sister taxa are outlined in the discussion.
Divergence Time Estimation
The hypothesis of rate constancy among taxa was
tested by comparing the likelihoods of the data, using a
198 SYSTEMATIC BIOLOGY VOL.
likelihood ratio test (Felsenstein, 1988), given the ML tree
topology under the best model selected with and without
the constraint of a molecular clock. We tested clocklike
behavior in both combined and partitioned data sets. To
estimate divergence times, semiparametric rate smooth-
ing using a penalized likelihood approach was applied
to the inferred phylogeny reconstruction with the aid
of the computer program r8s (Sanderson, 2002). Penal-
ized likelihood combines the likelihood term for a sat-
urated model with a different rate on every branch and
the nonparametric penalty function that keeps those rate
estimates from varying excessively across the tree. The
relative contribution of the two terms is controlled by a
smoothing parameter. Cross validation provides an ob-
jective method for model selection and choice of optimal
smoothing value (Sanderson, 2002).
To obtain SDs for estimated divergence times, the data
set was bootstrapped 100 times using the seqboot module
from PHYLIP 3.6 (Felsenstein, 1989), and branch lengths
were reestimated for each node under the constrained
initial topology in PAUP
. The dating analyses were then
repeated for each tree, and node statistics were summa-
rized using the program r8s.
Calibration of the molecular clock was not an easy
task because fossil data are scarce for swallowtail but-
terﬂies. Praepapilio colorado and P. glacialis, known from
the middle Eocene and dated at about 48 million years
ago (MYA), are two of the oldest known undisputed
butterﬂy fossils (Durdon and Rose, 1978). Resembling
Baronia brevicornis, these two species were considered the
most primitive swallowtails and were placed into their
own subfamily Praepapilioninae. However, their afﬁn-
ity to swallowtails has been disputed (Scott, 1986). Other
unidentiﬁed papilionids have been found in Europe
(dated at 24 MYA) and Japan (dated at 1.6 MYA)
(Emmel et al., 1992).
We tried to estimate divergence dates based on molec-
ular clock rates of 0.02 substitutions per site per million
years, as calibrated for COI and COII of other insects
(DeSalle et al., 1987; Brower, 1994). Thus, with an av-
erage number of 0.336 substitutions per site (calculated
under GTR++I model) between ingroup taxa and out-
groups, the estimated age of the subfamily Papilioninae
was approximately 16.8 ± 2.7 MY. The origin of the fam-
ily Papilionidae, estimated under the same assumption,
was calculated to be no more than 26 ± 7 MYA based on
the average number of 0.52 substitutions per site. Neither
standard rates of molecular evolution nor fossil dates
for Papilionidae accord well with the present day geo-
graphic distribution of swallowtails. Patterns exhibited
by the group strongly suggest the effects of continental
drift (Holloway and Nielsen, 1999), which indicate much
Although at least 18 species of the family Papilionidae
have been recorded as migrating (Williams, 1930), few
butterﬂies are capable of passing over 1,000-mile water
barriers (Drake and Gatehouse, 1995). Because the frag-
mentation of Gondwanaland was completed well before
48 MYA (Dietz and Holden, 1970), the best explanation
for the worldwide distribution of Papilio is an older age
for the group. The hypothesis of an older age for Papilio is
supported by a recent study by Gaunt and Miles (2002),
who presented an insect molecular clock calibration esti-
mating the date for the most recent common ancestor of
Papilio and its sister taxa (Pachliopta neptunus and Eury-
tides marcellus) as 82.5–89.1 (MYA).We used these dates as
an external calibration point to date the root (i.e., the most
recent common ancestor) of the Papilio tree. Several ad-
ditional age constraints for the internal nodes that could
be assumed from phylogenetic pattern and geological
history were also applied.
Papilio can be considered to fall into two major clades:
one (with few exceptions), restricted to the New World
and the other comprising mostly Old World species.
This fact, along with the suggested center of the ori-
gin for Papilio in the North America–Europe landmass
(Hancock, 1983), gives reason to assume that initial di-
versiﬁcation was related to disconnection between the
present Nearctic and Palearctic. This event was almost
completed by the end of the Cretaceous about 65 MYA
(Dietz and Holden, 1970). The latest direct land connec-
tion between Europe and eastern North America, known
as the Thulean route, was apparently available into the
Miocene and persisted up to as recently as approximately
20 MYA (Tiffney, 1985). However, because climatic cool-
ing after the Eocene/Oligocene boundary apparently
rendered the Thulean route unsuitable for many warm
temperature–adapted groups, the vicariance event may
have been approximately 35 MYA (Noonan, 1988) or
even earlier (Chapco et al., 2001). Based on this geological
time frame, we constrained the maximum and minimum
time of the initial split within Papilio to 65 and 35 MYA,
The distribution of P. anactus, which is conﬁned to
Australia, and its common ancestry with African and
Malagasy species groups of P. phorcas, P. r e x , and P.
delalandei suggest that the species may have evolved
during a long period of isolation, since separation of
Australia from the remains of Gondwanaland. Australia
had close contact with Antarctica before approximately
45 MYA when its northward drifting speed increased
rapidly (Raven and Axelrod, 1972). Australia is thought
to have been linked with Antarctica by an island chain
before approximately 35 MYA, when circumpolar cur-
rents became established and triggered glaciation of
Antarctica (Cook, 1990; Li and Powell, 2001). We used
the later date (35 MYA) to constrain the minimum age
of the common ancestor for P. anactus and its sister
The next calibration point was selected for another
species with a highly restricted distribution, P. hospiton,
which is endemic to the islands Corsica and Sardinia. The
separation to the Corsica–Sardinia microplate from the
Iberian Peninsula is consistently dated at approximately
29 MYA (Alvarez, 1972), although this date seems to be
too old and rather implausible to explain endemism of
P. hospiton. A more recent event, recession of the Mediter-
ranean sea for about 1500 years and successive reisola-
tion of the islands by the waters from the Atlantic ocean,
took place about 5 MYA (Hsu, 1972) and provides a more
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
reasonable date to constrain the split between P. hospiton
and P. machaon.
Papilio memnon and P. rumanzovia are sister species with
allopatric distributions (Collins and Morris, 1985). Papilio
memnon occurs widely in southeast Asia, from India to
southern Japan and Indonesia, but not in the Philippines.
Papilio rumanzovia has a more restricted geographic range
and occurs in the Philippines (except Palawan) and on
Batu, Talaud, and Sangihe islands. Around 5 MYA, the
Philippine platform approached its present position in
the immediate vicinity of East Malaysia and Indonesia
(Hall, 1996, 2001), allowing the ancestral species to ex-
pand its range and eventually evolve into two distinct
species. Based on this assumption, we constrained the
maximum age of the P. memnon–P. rumanzovia split at
To provide some more recent divergences, we con-
strained the minimum age of the common ancestor of
P. hippocrates and P. machaon to 0.01 MY. Known from
Japan, P. hippocrates is generally regarded as a subspecies
of P. machaon but was accepted as a full species by
Hancock (1983). Speciation in the P. machaon complex has
been attributed to Pleistocene climatic changes (Sperling,
1987; Sperling and Harrison, 1994). Pleistocene glaciation
was similarly implicated as a driving force in the specia-
tion of Parnassius stubbendorﬁ (which inhabits a large area
of continental Asia) and P. glacialis (which occurs only
in the Japan archipelago) in East Asia (Yagi et al., 2001).
Papilio hippocrates may have been isolated in Japan before
the last glacial maximum, which affected connections be-
tween the mainland and the Japan archipelago approxi-
mately 0.01 MYA (Oshima, 1990; Matsui et al., 1998). To
constrain a very recent age of siblings with close to zero
amount of sequence divergence, we arbitrarily ﬁxed the
maximum age of the P. r e x clade at 100 years.
To obtain estimates for a possible range of divergence
dates, the selected calibration points were applied in dif-
ferent combinations. We ﬁxed the time of the tree root at
82.5 MYA (ﬁrst) and 89.1 MYA (second), with no other
constraints enforced. Then we applied all discussed con-
straints for internal nodes without constraining the time
of the root. We then applied all age constraints at the
Alignment and Data Description
Alignment of EF-1α sequences was unambiguous be-
cause of the absence of indels. Total length of the aligned
EF-1α region was 995 bp. Some length differences were
found in the mtDNA sequence, primarily in the COI
initiation region (Caterino and Sperling, 1999). Seven-
teen sites here were deleted from the alignment, includ-
ing six nucleotides corresponding to the ﬁrst two codon
positions of the COI gene (Clary and Wolstenholme,
1985; Caterino and Sperling, 1999). As Caterino and
Sperling (1999) noted, a few taxa demonstrated 1-bp
indels (mostly phylogenetically uninformative) in the
tRNA-leucine gene. There were also a 3-bp insertion
immediately following the COI termination codon in
P. zelicaon and an insertion of one codon (AAT: argi-
nine) in P. dardanus between positions equivalent to 3,471
and 3,472 of Drosophila yakuba. We found that P. orib-
azus had a 3-bp insertion (AAA: lysine) in the same
position as in P. dardanus. The ﬁnal aligned sequences
included 2,293 nucleotides for the mtDNA partition
and 3,288 nucleotides in the combined data set. Previ-
ously published EF-1α sequence for Atrophaneura alci-
nous (Caterino et al., 2001) was determined to be incorrect
at the 5
end and has been updated for this study and in
No base composition heterogeneity was found for any
gene partition among ingroup taxa (COI: χ
= 61.6, df =
159, P = 1.0; COII: χ
= 56.6, df = 159, P = 1.0; EF-1α:
= 30.7, df = 159, P = 1.0) and across all species in-
cluding outgroup taxa (COI: χ
= 103.2, df = 213, P =
1.0; COII: χ
= 74.6, df = 213, P = 1.0; EF-1α: χ
df = 213, P = 0.99). However, after all noninformative
characters were excluded, base composition heterogene-
ity was revealed in the COI partition both for ingroup
= 237.9, df = 159, P < 0.005) and for the full
set of taxa (χ
= 390.7, df = 213, P < 0.005). However,
analysis of the corresponding MP tree did not reveal any
grouping of species with similar nucleotide frequencies.
Analysis of partitioned and combined data using LogDet
distances resulted in trees with topologies almost iden-
tical to those of the corresponding ML trees.
Based on the results of the ILD test, partitions of the
data into COI, COII, and EF-1α were homogeneous (sum
of gene tree length = 8,407; P = 0.114). COI and COII
gene partitions alone were homogeneous (sum of gene
tree length = 6,789; P = 0.492) and partitions between
COI and COII together and EF-1α were also homoge-
neous (sum of tree lengths = 8,505; P = 0.774). We ana-
lyzed both partitioned and combined data to gain insight
into any distinctively different phylogenetic results due
to data partitions and to obtain a phylogenetic recon-
struction based on a maximum number of informative
characters. In all cases, when all the sequences were com-
bined for the analysis resolution and node support in
the tree improved markedly. Many other studies have
demonstrated improved resolution and increased boot-
strap supports in combined analyses (e.g., Baker and
DeSalle, 1997; Vogler and Welsh, 1997; Crespi et al., 1998;
Remsen and DeSalle, 1998; Klompen et al., 2000; Chapco
et al., 2001).
MP analyses.—The total number of informative char-
acters for the combined data set with all outgroups was
1,129 (34.3%), with 847 sites in the mtDNA partition and
282 sites in the EF-1α data set. Number of informative
characters was greatest for third-codon positions (601
sites [71% of all informative characters] for COI-COII and
260 sites [92.2%] for EF-1α). With all outgroups excluded,
number of informative characters was reduced to 933
(28.4%), with 725 informative sites in the mtDNA parti-
tion and 208 sites in the EF-1α data set. Proportion of
200 SYSTEMATIC BIOLOGY VOL.
informative characters for third-codon positions in-
creased and was 75% (547 sites) of all informative char-
acters for COI-COII and 94% (196 sites) for EF-1α.
Parsimony analysis of the combined data set resulted
in a single tree (8,474 steps; consistency index [CI] =
0.242; retention index [RI] = 0.492) shown in Figure 1. We
used this topology as an initial estimate of the phylogeny
and compared it with those obtained under alternative
optimality criteria. The comparison of all topologies re-
vealed that the monophyly of subclades corresponding
to Hancock’s (1983) genera and subgenera was mostly
consistent among trees recovered from differentdata par-
titions. Thus, we labeled those clades as individual nodes
and illustrated alternative topologies in the form of sim-
pliﬁed trees (Fig. 2). Because the issue of phylogenetic re-
lationships outside of Papilio is beyond the scope of this
study, all outgroup taxa were excluded from the illustra-
tion of these simpliﬁed trees. The number of outgroups
used in the phylogenetic analyses also was reduced be-
cause the topology within the Papilio subtree was not
affected by the exclusion of distant outgroups (data not
shown). Two outgroups, Eurytides marcellus and Pach-
liopta neptunus, were used for further analyses. There are
few disagreements between phylogenies inferred from
mtDNA and nuclear partitions, but the conﬂicts are pri-
marily in nodes that are weakly supported in the com-
bined MP tree and in bootstrap analyses (e.g., basal po-
sition of Heraclides in COI + COII and combined tree
vs. a sister relationship with the P. alexanor + Chilasa +
ML analyses.—Modeltest supported use of the GTR
model (Lanave et al., 1984), with invariable sites (I) and
gamma-distributed rates (), as the best ﬁt for all data
sets (including the combined data set) except for a re-
duced EF-1α data set (Papilio sensu lato plus two out-
group taxa), for which the best model was TrN (Tamura
and Nei, 1993) + I +. Estimated parameters used in ML
analyses are given in Table 3. The C-T substitution rate
was substantially higher than other substitution types,
and substitution rates for A-C and C-G were 2–2.5 times
higher for the 71-taxon data set within the COI-COII
Heuristic searches performed for three data sets un-
der estimated parameters produced the trees shown in
Figure 2. Analyses based on only mtDNA data produced
a tree with the same relationships as illustrated in the MP
bootstrap consensus tree in Princeps and Papilio sensu
stricto but with alternative groupings in the Heraclides +
Pterourus + Chilasa clade. Analysis from the EF-1α data
set resulted in a tree with several polytomies.
The ML analysis for combined data revealed a sin-
gle tree with a negative log-likelihood score (–ln L)of
31023.619 that was identical to the combined MP boot-
strap consensus tree (Fig. 2). In the expanded ML tree
for combined data that shows all species (Fig. 3), group-
ing within the major terminal subclades (corresponding
to species groups and subgenera) is usually congruent
with the relationships inferred by the MP combined data
tree (Fig. 1). However, relationships among many major
clades remain weakly supported.
Bayesian analyses.—Both independent analyses for
COI-COII reached stationarity well before 300,000 cy-
cles. We discarded 3,000 trees as burn-in samples and
computed the consensus tree from the remaining 14,000
trees. The branching pattern is identical to that of the ML
tree and resembles the MP tree based on only mtDNA
data (Fig. 2), where Heraclides (see Fig. 1) is a sister group
to the rest of Papilio. Also, as in the ML tree, Pterourus
appeared to be paraphyletic with respect to Papilio alex-
anor + Chilasa, but there is low support for this relation-
ship. Many nodes on the tree are well supported, and the
overall posterior probability of the tree was 0.93. Of the
53 total nodes, 32 had a signiﬁcance level of 1.0 and 42
had support >0.9.
Bayesian analyses performed with two independent
runs on the EF-1α data subset yielded only 2,000 burn-
in trees per run. A consensus tree was constructed from
the remaining 16,000 trees and resulted in a resolution
very similar to that of the tree from the corresponding
partitioned ML analysis (Fig. 2). The overall probabil-
ity of the tree was only 0.85, and many nodes reﬂecting
phylogenetic relationships between major lineages in the
Princeps + Papilio (sensu stricto) clade had lower poste-
Three independent analyses for the combined data set
converged on similar log-likelihood scores and reached
stationarity before generation 300,000. For the COI-COII
data, the initial 3,000 trees from each analysis were dis-
carded. A consensus tree was constructed from the com-
bined set of 21,000 trees (Fig. 3). The branching pattern of
the tree is completely identical to that of the MP bootstrap
consensus and ML trees from combined analyses. The
average posterior probability for the inferred phylogeny
was 0.97. Thirty-eight ingroup nodes (of 53) had poste-
rior probabilities of 1.0, and 46 nodes were supported
with signiﬁcance level >0.9. Bayesian analyses supplied
higher values for posterior probabilities compared with
bootstrap supports from ML and MP analyses.
Divergence Time Estimation
Likelihood ratio tests rejected clocklike behavior of se-
quences (P < 0.001) for combined and partitioned (COI-
COII vs. EF-1α) data and also for full (71 taxa) and re-
duced (56 taxa) taxon sets. Tests for all six permutations
of partitions (three) versus taxon sets (two) gave P val-
ues of 0.001. Application of the selected calibration
points in the penalized likelihood procedure provided
us with a range of dates (with SDs). Initial results were
obtained with the default settings for dating analyses in
the r8s program, with the cross-validation function en-
forced. The rate smoothing parameter with the lowest
cross-validation scores was selected, and the dating pro-
cedure was then repeated. Age estimates (with SDs) for
all internal nodes numbered in Figure 4 are shown in
Character evolution optimized on the phylogeny of
Papilio is illustrated in Figure 4. Based on the inferred
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
FIGURE 1. Maximum parsimony tree from combined equally weighted data. Numbers above branches indicate bootstrap support (values
>50 shown). Bremer support (decay index) is given under branches for nodes within Papilio sensu lato. Circled letters label subclades and nodes
to indicate terminal nodes on trees in Figure 2.
202 SYSTEMATIC BIOLOGY VOL.
FIGURE 2. Summarized tree topologies inferred from MP, ML, and Bayesian analyses for partitioned and combined equally weighted data
sets with two outgroup taxa and only ingroup topologies shown. Terminal node names correspond to subclades and nodes labeled in Figure 1.
Lowercase letters indicate nodes and subclades different from those on the tree in Figure 1. A = subgenera Druryia and Princeps (a = Papilio rex;
b = P. phorcas;c= P. dardanus and P. constantinus;d= P. delalandei); B = Eleppone anactus;C= subgenus Menelaides;D= P. demoleus species group;
E = subgenus Achillides;F= P. xuthus;G= subgenus Papilio;H= P. oribazus and P. epiphorbas;I= P. nobilis;J= subgenus Pterourus (a = Papilio
glaucus species group; b = P. garamas and P. birchalli;c= P. scamander;d= P. troilus species group; e = P. esperanza); K = subgenus Chilasa (a = P.
clytia;b= P. epycides); L = P. alexanor;M= subgenus Heraclides (a = P. torquatus and P. anchisiades species groups plus P. astyalus;b= P. thoas and
P. cresphontes). When more than one MP tree was found in a particular analysis, topology of the strict consensus tree is shown. Asterisks beside
taxa on the EF-1α MP consensus tree indicate that all nodes within these clades are collapsed with three exceptions: clade G (machaon gorganus +
machaon oregonius); clade A has an unresolved tritomy (dardanus + phorcas + constantinus); and clade J has an unresolved tritomy (glaucus +
canadensis + multicaudatus). Asterisks beside taxa on the EF-1α ML tree indicate that these clades have polytomies: clade G (indra + hippocrates +
hospiton) and clade C (nephelus + macilentus + the remaining species of Menelaides, except helenus). Node A with two asterisks represents and
alternative position of P. phorcas as sister taxon to P. dardanus + P. constantinus. The MP bootstrap consensus tree has a branching pattern identical
to that of the combined ML tree but different from that of the single most-parsimonious tree.
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
TABLE 3. Substitution model parameters
estimated under ML from combined and partitioned data sets.
Base frequencies Substitution rates
Data set Model A C G T A-C A-G A-T C-G C-T G-T
mtDNA, 71 taxa GTR++I 0.3775 0.0746 0.1092 0.4387 42.6477 23.0684 13.2155 12.3900 494.9504 1.0000 0.3733 0.4277
EF-1α, 71 taxa GTR++I 0.2817 0.2340 0.2200 0.2643 0.7154 6.4703 2.2787 0.8428 9.4152 1.0000 1.3979 0.6431
Combined 71 taxa GTR++I 0.3290 0.1079 0.1417 0.4214 13.1789 18.6198 13.5734 10.0175 136.6209 1.0000 0.5502 0.5048
mtDNA, 56 taxa GTR++I 0.3525 0.870 0.1145 0.4460 20.5719 19.4641 9.2873 4.9981 292.5973 1.0000 0.4016 0.4637
EF-1α, 56 taxa TrN++I 0.2688 0.2486 0.2343 0.2483 1.0000 5.6763 1.0000 1.0000 9.5215 1.0000 1.2584 0.6503
Combined 56 taxa GTR++I 0.3022 0.1376 0.1562 0.4039 6.2882 15.9335 11.3400 5.0694 84.5150 1.0000 0.5856 0.5258
= α, estimated shape parameter; I = proportion of invariable sites.
phylogeny, the available morphological and ecological
data, and Pachliopta neptunus (a member of Troidini) as
the sister taxon, MP reconstructions indicated that the
ancestral states for the selected characters were larval
feeding on Rutaceae, nonmimetic wing pattern, lack of
iridescent patches on wings, lack of female secretion de-
posits on egg surface, straight larval minute body hairs,
one row of crochets on larval prolegs, and lack of a silk
pad on the larval resting site (Table 5).
When a different outgroup was chosen, no substan-
tial differences were observed for reconstruction of an-
cestral states of any characters except larval host plant.
When Meandrusa was placed as the closest Papilio out-
group, an uncertain MP reconstruction of the ancestral
host plant was obtained at the root of Papilio (node 88),
indicating equal probabilities for larval feeding on Ru-
taceae and Lauraceae. In ML reconstructions, with Mean-
drusa as a sister taxon for Papilio, all changes except larval
hosts at nodes 88 and 101 (Fig. 4) had the same tran-
sitions but with slightly higher or equal log-likelihood
support (data not shown). Inclusion of Meandrusa de-
creased the likelihood (from 97.3% to 87.1% out of to-
tal likelihood) that Rutaceae feeding was the primitive
trait for Papilio larvae. As the ancestral larval host, Lau-
raceae had less than 0.6% of total likelihood. Inclusion
of Meandrusa increased this estimate up to 11.6%. Un-
certainty remains for the ancestral state of node 101,
where none of the reconstructions gave a strong indi-
cation for larvae feeding on any one of Rutaceae, Api-
aceae, or Lauraceae as the primitive trait. Incomplete
data did not allow ML character optimizations for pres-
ence or absence of female secretion on eggs, shape of lar-
val body hairs, number of crochet rows on larval prolegs,
and presence or absence of a silk pad on larval resting
No previous molecular phylogenetic studies on Pa-
pilio have resulted in a single robust or relatively com-
prehensive phylogenetic reconstruction. The improved
resolution obtained here allowed us to revisit several in-
adequately resolved issues. In congruence with previous
studies of Papilio phylogeny (Aubert et al., 1999; Caterino
et al., 1999; Reed and Sperling, 1999), our results in the
ML and Bayesian analyses on combined data show that
the genus Papilio comprises two lineages (Fig. 5): one in-
cludes Hancock’s genera (here treated as subgenera) Her-
aclides, Pterourus, and Chilasa and Papilio alexanor, and the
other consists of Princeps, Papilio (sensu stricto), and the
monotypic Eleppone. A potential alternative basal topol-
ogy, with Heraclides as the ﬁrst clade to diverge from the
remainder of Papilio sensu lato (Fig. 1), suggests the ex-
istence of three major Papilio lineages: Heraclides, Pter-
ourus + Chilasa + Papilio alexanor, and Princeps + Papilio
(sensu stricto) + Eleppone. However, this branching pat-
tern has weaker support in most analyses, and therefore
we relied on the relationships obtained from the com-
bined MP bootstrap, ML, and Bayesian analyses.
Our most signiﬁcant taxonomic ﬁnding is the strong
placement of Papilio esperanza. Several other researchers,
although recognizing some uncertainty, concluded that
this species was a member of Heraclides (Hancock, 1983).
We ﬁnd strong support for the countervailing position,
that the species is instead related to (and in our analysis
basal to) the troilus species group (see Collins and Morris,
Another signiﬁcant result is the strongly supported
placement of P. anactus at the base of part of the Prin-
ceps lineage. This species had previously been placed
in its own subgenus, close to Chilasa (in the other ma-
jor clade of Papilionini) (Hancock, 1979). The placement
was based in part on shared mimicry, although the spe-
ciﬁc mimetic patterns differ as do larval morphology and
host plants. Larvae of Chilasa develop on Lauraceae and
Magnoliaceae, whereas those of P. anactus, like members
of Princeps, feed on Rutaceae (Hancock, 1979).
Several ﬁndings add further support to recent sug-
gested taxonomic changes. In agreement with Scriber
et al. (1991) and Caterino and Sperling (1999), we found
Pterourus sensu Hancock (1983) to be paraphyletic with
respect to Pyrrhosticta. Monophyly of Heraclides, and rela-
tionships within it, were strongly supported in all anal-
yses. There is some evidence against division of Hera-
clides into four subgenera, Heraclides, Calaides, Troilides
and Priamides, as suggested in Tyler et al. (1994). At least
for Troilides, we found relatively strong support for pa-
raphyly. However, additional samples are needed. Our
results support the previously weak placement of Pa-
pilio alexanor near Pterourus, as found by Caterino and
Sperling (1999) and Reed and Sperling (1999). Previous
suggestions that this species belongs near the P. machaon
species group (Munroe, 1961; Hancock, 1983) can now
204 SYSTEMATIC BIOLOGY VOL.
FIGURE 3. Phylogenetic tree inferred by ML from the combined data set with a reduced number of outgroups. The identical topology
with similar branch lengths resulted from Bayesian analysis of the same data. Numbers above branches indicate bootstrap support counted in
minimum evolution analysis under the same model of substitution used in ML analysis (values >50 shown). Numbers under the nodes indicate
Bayesian posterior probabilities. Vertical lines beside taxon names correspond to Papilio sections sensu Munroe (1961).
be considered refuted, and the resemblance of P. alexanor
and P. machaon is considered the result of convergence
due to similar host plants utilized by larvae.
Only two species represented the subgenus Chilasa in
our study. One important species group, elwesi, could not
be sampled. Nevertheless, monophyly of Chilasa sensu
stricto, represented by C. epycides and C. clytia, has strong
support. Contrary to Hancock’s conclusions, Chilasa ap-
pears to be the sister taxon of Pterourus. Although sup-
port for this particular result is only moderate, Chilasa
clearly belongs to the same lineage as Pterourus and Pa-
pilio alexanor, as indicated by strong bootstrap support in
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
FIGURE 4. Calibration points and evolution of ecological and morphological traits in Papilio. Penalized likelihood procedure (Sanderson,
2002) was applied to the combined Bayesian tree using constrained age of the nodes indicated by arrows. Dating information for all internal nodes
(shown in black circles) is provided in Table 4. Roman numerals correspond to Hancock’s genera: I = Pterourus;II= Heraclides; III = Eleppone;
IV = Chilasa;V= Papilio;VI= Princeps. Branches corresponding to major Rutaceae feeding groups are heavier (Tyler et al., 1994; Scriber et al.,
1995). Independent origin of Apiaceae feeding is shown as red bars on the tree. Branches for mimetic species are shown in blue (if only female is
mimetic) or green (if both sexes are mimetic) (Hancock, 1983). Presence of metallic iridescent scales on large areas of the wings is shown as mauve
diamonds (Munroe, 1961). Blue triangles beside taxon names indicate that minute body hairs in last-instar larva are bent backwards, whereas
they are straight in other species (Igarashi, 1984). Green squares beside taxon names indicate the ability of larvae to spin a silk pad on leaf upper
surface (where data are available; Hagen, 1999). Open squares indicate that larvae rest on twigs or stems of the host plant without spinning
a silk pad. Numerals inside squares indicate number of crochet rows, if known (Hagen, 1999). Yellow circles indicate species that have eggs
with visible granular female secretion attached to the chorion (Igarashi, 1984; Scriber et al., 1995). Presence of species in zoogeographic regions
(Collins and Morris, 1985) is shown on the right of the tree: OR = Oriental; AU = Australasian; ET = Ethiopian; PA = Palearctic; HA = Holarctic;
NA = Nearctic; NT = Neotropical. Circled M indicates probable phylogenetic position of Meandrusa.
206 SYSTEMATIC BIOLOGY VOL.
TABLE 4. Age estimates (±SD) in millions of years of internal nodes in Papilio tree using calibration points shown in Figure 4 and model-
corrected branch lengths.
Combined data set
Root and six nodes constrained
Node Root ﬁxed at Root ﬁxed at Root relaxed, six Root and six nodes
(Fig. 4) 82.5 MY 89.1 MY nodes constrained constrained COI+COII EF-1α
56 4.73 ± 2.62 5.11 ± 2.81 5.00 ± 1.67 4.70 ± 1.65 5.00 ± 2.40 4.38 ± 8.50
57 13.36 ± 6.55 14.50 ± 7.04 14.52 ± 5.46 13.27 ± 5.62 15.99 ± 7.90 11.11 ± 9.10
58 15.14 ± 7.20 16.43 ± 7.75 16.49 ± 6.05 15.03 ± 6.23 18.18 ± 8.76
59 13.17 ± 6.33 14.30 ± 6.79 14.35 ± 5.49 13.07 ± 5.53 16.09 ± 7.36 10.01 ± 9.68
60 17.46 ± 8.36 18.94 ± 9.06 19.05 ± 7.28 17.33 ± 7.37 21.31 ± 10.33
61 20.22 ± 9.44 16.50 ± 10.18 22.10 ± 8.26 20.07 ± 8.48 12.00 ± 8.19
62 15.20 ± 7.72 21.94 ± 8.39 16.63 ± 6.78 15.08 ± 7.11 20.16 ± 10.11 7.35 ± 9.57
63 23.04 ± 10.44 25.01 ± 11.34 25.24 ± 9.29 22.87 ± 9.46 25.24 ± 12.94 20.22 ± 9.14
64 32.77 ± 13.49 35.60 ± 14.65 36.04 ± 12.83 32.51 ± 12.86 31.67 ± 15.53 40.09 ± 11.43
65 13.88 ± 6.99 15.07 ± 7.55 15.26 ± 6.50 13.77 ± 6.64 13.69 ± 6.91 16.88 ± 7.67
66 35.43 ± 14.35 38.50 ± 15.58 39.01 ± 13.82 35.15 ± 13.71 33.26 ± 16.71
67 10.42 ± 6.00 11.35 ± 6.47 11.52 ± 5.91 10.34 ± 5.79 11.53 ± 6.35 9.66 ± 8.39
68 13.12 ± 7.01 14.28 ± 7.58 14.49 ± 6.71 13.02 ± 6.73 13.59 ± 7.97 13.19 ± 10.43
69 37.47 ± 15.28 40.72 ± 16.56 41.28 ± 14.56 37.17 ± 14.70 36.02 ± 18.70
70 14.69 ± 7.34 15.98 ± 7.75 16.20 ± 7.10 14.58 ± 6.99 14.88 ± 7.86 11.87 ± 8.79
71 29.77 ± 12.36 32.36 ± 13.35 32.80 ± 12.01 29.53 ± 11.92 28.36 ± 14.28 34.79 ± 10.17
72 40.37 ± 15.99 43.86 ± 17.38 44.51 ± 15.32 40.04 ± 15.58 39.88 ± 20.97 45.10 ± 9.39
73 0.52 ± 0.72 0.56 ± 0.77 0.36 ± 1.20 0.35 ± 1.65 0.20 ± 0.27 4.36 ± 5.92
74 3.08 ± 2.55 3.35 ± 2.78 1.97 ± 2.74 1.94 ± 2.73 1.50 ± 1.18
75 8.73 ± 5.61 9.50 ± 6.19 5.00 ± 5.46 5.00 ± 5.62 5.00 ± 3.37
76 10.96 ± 6.69 11.91 ± 7.28 8.31 ± 6.42 7.90 ± 6.72 7.90 ± 5.80 12.44 ± 7.02
77 5.86 ± 4.35 6.37 ± 4.64 4.61 ± 4.26 4.37 ± 4.34 4.45 ± 4.00 7.51 ± 5.37
78 18.02 ± 9.26 19.59 ± 10.04 16.58 ± 8.85 15.33 ± 9.10 17.76 ± 10.36 15.64 ± 7.98
79 33.60 ± 13.79 36.53 ± 15.00 35.70 ± 13.13 32.27 ± 13.39 35.48 ± 18.42 33.58 ± 9.84
80 42.92 ± 16.83 46.61 ± 18.28 47.86 ± 16.18 42.98 ± 16.53 41.87 ± 22.30 49.67 ± 7.41
81 6.12 ± 3.53 6.62 ± 3.73 6.76 ± 3.42 6.11 ± 3.47 6.93 ± 4.27 4.43 ± 6.08
82 20.04 ± 8.97 21.75 ± 9.70 22.21 ± 8.77 19.98 ± 8.95 23.05 ± 11.82 19.95 ± 6.98
83 23.69 ± 10.63 25.71 ± 11.60 26.26 ± 10.22 23.61 ± 10.49 27.87 ± 14.38 21.11 ± 7.42
84 29.55 ± 12.39 32.10 ± 13.60 32.79 ± 11.82 29.45 ± 12.09 32.84 ± 16.51 29.96 ± 8.46
85 0.26 ± 1.53 0.29 ± 3.80 0.00 ± 0.35 0.00 ± 0.35 10
86 33.67 ± 13.75 36.58 ± 15.07 37.42 ± 13.07 33.62 ± 13.33 33.52 ± 18.29 38.90 ± 8.36
87 40.50 ± 16.05 43.99 ± 17.49 45.10 ± 15.40 40.52 ± 15.70 40.18 ± 21.52 47.08 ± 10.73
88 55.75 ± 20.07 60.42 ± 21.67 65.00 ± 20.06 57.89 ± 20.12 56.05 ± 27.35 65.00 ± 6.37
89 5.08 ± 3.90 5.64 ± 4.22 6.17 ± 3.86 5.34 ± 3.93 5.65 ± 3.65 8.25 ± 14.71
90 9.71 ± 5.77 10.74 ± 6.29 11.70 ± 5.83 10.18 ± 5.98 11.67 ± 6.51
91 13.66 ± 7.53 15.00 ± 8.16 16.27 ± 7.66 14.26 ± 7.76 16.83 ± 9.15 10.04 ± 5.73
92 31.56 ± 12.47 34.26 ± 13.50 36.79 ± 12.94 32.76 ± 13.06 36.61 ± 16.40 20.85 ± 8.80
93 14.29 ± 6.51 15.44 ± 7.01 16.52 ± 6.66 14.80 ± 6.91 20.14 ± 9.19 4.63 ± 3.91
94 23.00 ± 9.58 24.92 ± 10.31 26.72 ± 9.83 23.86 ± 10.17 25.83 ± 11.83 16.59 ± 7.13
95 38.73 ± 14.60 42.01 ± 15.76 45.08 ± 15.10 40.19 ± 15.01 45.49 ± 19.78 25.91 ± 9.65
96 14.43 ± 6.47 15.65 ± 7.06 16.80 ± 6.74 14.98 ± 6.75 20.54 ± 9.42 5.75 ± 5.10
97 23.94 ± 9.52 25.95 ± 10.23 27.83 ± 9.75 24.83 ± 9.86 29.11 ± 12.84 14.75 ± 7.50
98 32.97 ± 12.54 35.74 ± 13.54 38.34 ± 13.01 34.21 ± 12.94 39.56 ± 16.70 21.95 ± 8.84
99 41.44 ± 15.50 44.95 ± 16.76 48.24 ± 15.96 43.00 ± 15.74 46.70 ± 21.08 29.94 ± 10.55
100 33.29 ± 12.88 36.13 ± 13.97 38.79 ± 13.26 34.56 ± 13.36 37.09 ± 16.39 25.21 ± 10.32
101 45.31 ± 16.61 49.14 ± 18.00 52.75 ± 16.88 47.02 ± 16.77 49.58 ± 22.51 34.28 ± 9.96
102 4.15 ± 2.94 4.46 ± 2.95 4.76 ± 2.80 4.29 ± 2.76 4.68 ± 3.27 3.13 ± 6.08
103 15.93 ± 7.77 17.27 ± 8.49 18.52 ± 8.25 16.53 ± 7.90 17.41 ± 9.28 15.60 ± 7.82
104 19.62 ± 8.97 21.30 ± 9.73 22.86 ± 9.33 20.36 ± 9.02 21.60 ± 11.11 17.55 ± 8.75
105 35.35 ± 13.73 38.42 ± 14.92 41.27 ± 14.17 36.71 ± 13.92 38.91 ± 18.35 28.32 ± 10.01
106 40.68 ± 15.68 44.19 ± 16.87 47.46 ± 15.76 42.23 ± 15.76 43.24 ± 20.95 33.01 ± 9.50
107 16.37 ± 8.97 17.82 ± 9.05 19.18 ± 8.55 17.01 ± 8.81 18.13 ± 10.52 12.25 ± 8.71
108 50.94 ± 18.50 55.24 ± 19.94 59.34 ± 18.51 52.87 ± 18.54 54.37 ± 25.07 42.24 ± 9.12
109 82.50 ± 0.00 89.10 ± 0.00 102.78 ± 11.59 89.10 ± 2.29 89.10 ± 1.17 89.10 ± 2.15
No age estimation could be obtained for nodes that were collapsed in partitioned phylogenies.
the combined ML tree and the high posterior probability
of the corresponding clade in the Bayesian phylogeny.
Another signiﬁcant ﬁnding is the support for para-
phyly of Princeps with respect to Papilio sensu stricto
and Eleppone. Hancock (1983) divided Princeps into four
subgeneric lineages. Only one, Menelaides, is supported
here as a monophyletic taxonomic unit; a second group,
Sinoprinceps, is represented here by a single species.
Hancock’s other two subgenera, Druryia and Princeps
sensu stricto, were found to be paraphyletic. Some clades
within Princeps, however, are strongly supported: (1)
Menelaides (sensu Hancock, 1983); (2) the P. demoleus
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
TABLE 5. Reconstructed ancestral states and major gains and losses
(excluding character changes in individual terminal taxa) for selected
characters in Papilio using Pachliopta neptunus (Troidini) as its sister
Node (Fig. 4) MP ML
Host plant families
88 Rutaceae (Rutaceae or Rutaceae (46.97; 0.973)
101 Rutaceae or Apiaceae Apiaceae (48.05; 0.33)
Lauraceae (48.07; 0.33)
Rutaceae (48.09; 0.32)
99 Lauraceae Lauraceae (46.95; 0.991)
91 Rosaceae, Oleaceae, Rosaceae (46.96; 0.988)
78 Apiaceae Apiaceae (46.96; 0.988)
Mimicry in imago
88 Nonmimetic Nonmimetic (50.55; 0.992)
104 Female Female limited (50.83; 0.75),
nonmimetic (52.55; 0.18)
102 Both sexes Both sexes (50.63; 0.917)
100 Both sexes Both sexes (50.63; 0.912)
94 Female Female limited (50.83; 0.748),
nonmimetic (52.26; 0.171)
85 Both sexes Both sexes (50.62; 0.92)
81 Female Female limited (50.63; 0.911)
58 Female or both sexes Both sexes (51.49; 0.386),
female limited (51.52; 0.374),
nonmimetic (51.97; 0.240)
56 Female Female limited (50.60; 0.942)
Iridescent patches on wings
88 Absent Absent (13.59; 0.999)
70 Present Present (13.62; 0.973)
100 Present Present (13.62; 0.972)
Female secretion on eggs
88 Not present Not computable (missing data)
Larval body hair
88 Straight Not computable (missing data)
Number of crochet rows on larval prolegs
88 1 Not computable (missing data)
Silk pad on a resting site
88 Not present Not computable (missing data)
78 Not present
Estimates of marginal ancestral state probabilities are given as character state
(negative log likelihood; proportion of a total likelihood).
Alternative transition when Meandrusa was placed as a sister taxon to Papilio.
species group; (3) Achillides (sensu Shimogori, 1997); (4) P.
nireus species group; and (5) the ((rex, phorcas)delalandei)
species groups from subdivisions Druryia and Princeps.
In congruence with another study (Vane-Wright et al.,
1999), P. nobilis is strongly associated with this latter
group. The present results offer the strongest support
yet for relationships among the three species of the phor-
cas group; our ﬁnding of ((dardanus, constantinus)phorcas)
contrasts with the ((dardanus, phorcas)constantinus) reso-
lution suggested by Vane-Wright et al. (1999).
Molecular Clock Predates Fossil Records of Papilionidae
The estimate of 26 ± 7 MYA for the origin of the family
Papilionidae calculated from standard rates of 20 × 10
substitutions per site per year is substantially less than
the minimum age of 48 MY indicated by fossils. One ex-
planation for this mismatch is that despite efforts to cor-
rect them, divergences have been substantially underes-
timated because of high saturation in relatively rapidly
evolving COI and COII genes. Standard substitution
rates have generally been applied to recently evolved
taxa separated by sequence divergence of <5–7% (Harri-
son and Bogdanowicz, 1995). The approximately twofold
difference in GTR+I+-corrected relative substitution
rates in COI-COII sequences between the data sets with
and without basal species of Papilionidae (Table 3) may
indicate that at least the mtDNA sequences are badly
saturated and that even the best-ﬁt model gives cor-
rected divergences that are a substantial underestimate.
At the same time, proportionally less difference in rates
was observed in EF-1α sequences for data with complete
versus reduced numbers of outgroup taxa. Substitution
rates at third codon positions are roughly 18–28 times
greater in COI and COII than in EF-1α, which is compa-
rable to estimates from previous studies (e.g., Reed and
Sperling, 1999). Nevertheless, the selected substitution
model was designed to incorporate signiﬁcant rate dif-
ferences among taxa, and thus the divergence estimates
should have accounted for some of this variation.
Our proposed dating suggests slow rates of evolution
in mtDNA in the genus Papilio. Substitution rates for the
mtDNA genes COI and COII and the nuclear gene EF-
1α are 7.8–10.2 × 10
and 1.32–2.0 × 10
per site per
year, respectively. These rates are 2–4 (for mtDNA) and
up to 30 (for EF-1α) times slower than the “standard”
substitution rates for COI in Drosophila (20.0–29.0 × 10
substitutions per site per year; Beckenbach et al., 1993)
and ﬁg-pollinating wasps (19.0 × 10
; Machado et al.,
2001). However, there is now abundant evidence for a
diversity of mtDNA rates in insects. Low transforma-
tion rates have been reported for carabid beetles (2.8 ×
substitutions per site per year; Su et al., 1998), Halo-
bates sea skaters (4–7 × 10
per year; Andersen et al.,
2000), Bacillus phasmids (3–4.7 × 10
per year; Manto-
vani et al., 2001), and Limnoporus water-striders (4–4.5 ×
per year; Sperling et al., 1997). Accelerated substi-
tution rates have been reported for the honeybee Apis
mellifera (approximately 34.0-49.0 × 10
; Crozier et al.,
1989) and parasitic lice of the genus Dennyus (up to 90 ×
; Page et al., 1998).
Our estimate of an age for major lineages within
Papilio of around 30–50 MY (with the age of the split
of Papilio from Troidini up to 100 MY) would also in-
crease the time of the origin of the family Papilionidae to
208 SYSTEMATIC BIOLOGY VOL.
FIGURE 5. Phylogenetic relationships of species groups and major subdivisions of Papilio sensu lato: Left: tree according to Hancock
(1983); right: tree based on data presented here. Roman numerals correspond to Hancock’s genera: I = Pterourus;II= Heraclides; III = Eleppone;
IV = Chilasa;V= Papilio;VI= Princeps. Letters refer to Hancock’s subgenera: a = Pyrrhosticta;b= Pterourus;c= Sinoprinceps;d= Druryia;
e = Princeps;f= Menelaides. Single asterisk indicates inclusion of Papilio alexanor as per Hancock (1983). Double asterisks indicate exclusion
of P. alexanor based on our data.
the lower Cretaceous, substantially predating the earli-
est fossils. However, this type of observation is increas-
ingly common in studies that attempt to ascertain dates
of taxonomic origin based on molecular data (Pellmyr
and Leebens-Mack, 1999; Machado et al., 2001). Because
of their fragile nature, delicate-wing butterﬂies are rarely
preserved as fossils. Although at least 44 butterﬂy fos-
sils are known worldwide (Emmel et al., 1992), very few
swallowtail fossils have been reported, with the oldest
putative swallowtail, Praepapilio, dated as 48 MY old,
and few other Papilio representatives from more recent
periods. The richest fossil butterﬂy fauna in the world
is known from 36–34-MY Florissant shales, and this as-
semblage indicates that all major families of butterﬂies
had already evolved well before this date (Emmel et al.,
1992). Some fossils from the Lower Oligocene have been
described and placed within extant genera of the fam-
ily Nymphalidae, e.g., Vanessa amerindica, which has a
strong afﬁnity to V. indica (see Emmel et al., 1992). Thus,
at least some butterﬂies very similar to modern genera in
Rhopalocera were already present at least 35 MYA. The
family Papilionidae is certainly among the oldest Papil-
ionoidea, so the origin of swallowtail butterﬂies in the
Cretaceous seems plausible.
Previous authors have outlined relatively concrete
biogeographic scenarios for Papilionini (Munroe, 1961;
Hancock, 1983; Miller, 1987; Scriber, 1995b). The ancestor
of the Papilionini, from a center of origin in eastern North
America, is considered to have dispersed ﬁrst to the
Palearctic and later to South America before the end of
the Cretaceous, leaving Pterourus in North America. The
Palearctic ancestor spread to China (and subsequently to
Southeast Asia, Africa, and Australia) as Princeps, leav-
ing Papilio sensu stricto in western Asia. The South Amer-
ican ancestor spread via Gondwanaland to Australia as
Eleppone and from there to Southeast Asia as Chilasa, leav-
ing Heraclides in South America.
The inferred pattern of phylogenetic relationships
within Papilio as found here at least partly conﬂicts with
these previously proposed vicariance hypotheses. The
major conﬂict stems from the origin of Papilio alexanor,
subgenus Chilasa, and subgenus Pterourus from a com-
mon ancestor and the distant relationship between New
World branches of Papilio and the Old World P. (Elep-
pone) anactus. The origin of Chilasa from the common an-
cestor with Eleppone via Gondwanaland to Australia and
Asia, as was suggested based on a traditional view of Pa-
pilio phylogeny (Scriber, 1995b), is not supported by our
The best currently available evidence indicates that the
common ancestor of Papilio apparently produced two de-
scendant lineages in the North America–Europe block
about 55–65 MYA (see Fig. 4, Table 4), well before the
complete disjunction of the two continents. This scenario
is not dependent on the use of this vicariance event in
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
establishing divergence rates, because several other cal-
ibration points gave similar dates. One lineage was the
ancestor of Papilio (sensu stricto), Princeps, and Eleppone,
and another was the common ancestor of Heraclides, Pter-
ourus, Chilasa, and Papilio alexanor. Before the end of the
Eocene (around 53–59 MYA, or even 42 MYA according to
EF-1α data), New World swallowtails were established in
South America, and modern Heraclides evolved. Around
60 MYA, North and South America were temporarily
and loosely connected by an island arc (Donnelly, 1988),
providing a possible avenue for this dispersal. Diver-
siﬁcation of P. alexanor, Chilasa, and Pterourus is dated
within 30–48 MY, at the time of favorable climatic con-
ditions across the Thulean route (Bowen et al., 2002).
Chilasa may have then spread to Asia soon after the
Turgai Strait, which separated Europe and Asia un-
til 45 MYA, had dried, opening the Turgai route from
Europe to Asia (Kurt´en, 1971). The ancestral lineage that
was left in North America evolved Pterourus millions of
years before the Thulean route was severed by climate
changes around 35 MYA (Noonan, 1988).
The common ancestor of Old World swallowtails
might have produced three major lineages. The oldest
one may have spread to Africa and Madagascar some-
time between 42 and 50 MYA and left as descendants
some extant groups currently placed in Princeps. Based
on our phylogenetic results, this lineage should some-
how have contributed to the origin of the Australian
endemic Papilio (Eleppone) anactus. Traditional views on
the origin of this species (Hancock, 1979) suggested in-
stead a South American ancestry, with migration through
Antarctica. Eleppone could have reached the Australian
continent through the Tasmanian land bridge around
40—47 MYA. That hypothesis requires some connec-
tion between Africa–Madagascar and Antarctica around
that time. A possible explanation comes from a late
Cretaceous connection between Australia/Antarctica
and Indo-Madagascar through the Kergu´elen Plateau
(Sampson et al., 1998; Hay et al., 1999) and Eocene–
Oligocene uplift of archipelagos in the Indian Ocean and
Mozambique Strait that apparently served as stepping
stones for faunal exchange (McKenzie and Sclater, 1973;
Haq et al., 1988).
A second lineage of Old World Princeps-like swal-
lowtails underlies extant Asian Sinoprinceps and Ho-
larctic Papilio (sensu stricto), which apparently di-
verged around 33–36 MYA. Speciation in Papilio (sensu
stricto) has been attributed to several subsequent disper-
sals across Beringia before and during the Pleistocene
The third lineage of Old World swallowtails pro-
duced a diverse group of modern Princeps, which would
have been distributed throughout southern Eurasia,
Africa, and Australia sometime between 32 and 41 MYA.
Around that time India collided with Eurasia, and
Australia established a connection with Southeast Asia
via intervening islands (Raven and Axelrod, 1972).
Geographic proximity of the continents would favor nu-
merous dispersals, resulting in the present day mixed
distribution pattern with numerous distinctive popula-
tions appearing to be taxonomically isolated (Wallace,
Evolution of Morphological and Ecological Traits
Species of the genus Papilio are highly diverse in adult
wing pattern and immature stage morphology. They also
have highly varied larval host plant utilization, ranging
from restricted use of one or two host species to broad
polyphagy of up to eight different plant families (see
Bossart and Scriber, 1995). However, about 80% of species
of Papilio (Eleppone and almost all species in Heraclides
and Princeps) are primarily Rutaceae feeders. The main
hosts of most Papilio (sensu stricto) species are Apiaceae.
Larvae of Chilasa usually develop on Lauraceae. Basal
species of Pterourus also feed on members of the families
Lauraceae and Magnoliaceae, but some species utilize
Rosaceae and other hosts (see Tyler et al., 1994; Scriber
et al., 1995). The lack of a stable Papilio phylogeny and its
relationships with other swallowtails hindered previous
attempts to resolve host use evolution and to determine
ancestral host plant relationships (Miller, 1987).
Larvae of Pachliopta neptunus and all other Troidini,
a generally accepted sister group of Papilionini, de-
velop on species of Aristolochiaceae, whereas the pri-
mary hosts for Leptocircini (the sister group for Troi-
dini and Papilionini) are Annonaceae. Some species of
Leptocircini also develop on Lauraceae, Hernandiaceae,
and Rosaceae. However, according to a recently pub-
lished molecular phylogeny of the genus Graphium
(Makita et al., 2003), basal lineages of Leptocircini
are those whose larva develop on Hernandiaceae and
Rosaceae. Thus, none of the presumed close relatives of
the genus Papilio indicate Lauraceae as the ancestral host
Based on the possible sister relationship of Papilio
and Meandrusa, whose larvae develop on Lauraceae
(Igarashi, 1984), it was hypothesized that use of these
plants should be regarded as a primitive character
(Aubert et al., 1999). Based on this assumption, there
were at least two shifts to Rutaceae, once in Heraclides
and once in the Old World lineage of Papilio, with sub-
sequent shifts in the P. machaon group to Apiaceae and
in some African Princeps back to Lauraceae or close rel-
atives (Ackery et al., 1995). However, the reliability of
morphological characters that were used to justify the
afﬁnity of Meandrusa to the tribe Papilionini is question-
able (Ha ¨user, 1993), and molecular evidence for this re-
lationship (Aubert et al., 1999) is tentative.
The phylogenetic relationships within Papilio obtained
here (Fig. 4) suggest that ancestral feeding on Rutaceae
is at least equally plausible based on MP character op-
timization when Meandrusa was included as the sister
taxon of Papilio. Reconstruction of the ancestral larval
feeding habits under the ML model supported Rutaceae
feeding as a primitive trait even when Meandrusa was in-
cluded in character optimization. We have estimated the
ﬁrst split within Papilio at approximately 55–65 MYA, ﬁt-
ting the proposed age of the initial diversiﬁcation within
Rutales at about 67 MYA (Magall ´on et al., 1999). Under
210 SYSTEMATIC BIOLOGY VOL.
this hypothesis, there would be only one major shift to
Lauraceae, in the stem lineage of Chilasa and Pterourus
(after the divergence of P. alexanor). Another shift to Lau-
raceae in the African P. hesperus group (Ackery et al.,
1995) is not associated with major diversiﬁcation and can
be compared with the shift from Rutaceae to Fabaceae
feeding in the Australian subspecies of P. demoleus, P. d .
sthenelus (Braby, 2000). Another observation supporting
Rutaceae as the ancestral host is that even those species
whose larvae normally develop on members of other
plant families possess the ability to feed on Rutaceae,
both in the wild and in captivity (Tyler et al., 1994).
The pattern of relationships within Papilio and the ex-
clusion of P. alexanor from the P. machaon group suggest at
least three independent origins of Apiaceae feeding. This
shift may be associated with changes in larval coloration.
The distinctive larval coloration of P. alexanor appears to
be convergent with that of the machaon group (Caterino
and Sperling, 1999). Some populations of P. demodocus in
South Africa have switched from Rutaceae to Apiaceae
and have evolved a similarly cryptic pattern (Clarke
et al., 1963), whereas other P. demodocus on Apiaceae
have retained the green Princeps-like larval coloration. A
shift to Apiaceae feeding is seen in some populations of
P. paeon, a member of the P. thoas group (subgenus
Heraclides) that usually develops on Rutaceae (Tyler et al.,
1994; Sperling and Feeny, 1995); however, there appears
to be no change in larval color pattern in these pop-
ulations (Walker, 1882). Evolution of host plant use in
Papilio also has been affected by recent speciation in tem-
perate latitudes. Papilio machaon, with its Holarctic dis-
tribution and larvae feeding primarily on Apiaceae, has
further evolved to use Asteraceae in western and north-
ern North America and parts of central Asia (Sperling,
1990). Nearctic species of the P. glaucus group breed on
Lauraceae, Rutaceae, Tiliaceae, or Magnoliaceae in the
southern parts of their ranges but develop to the north
on a wide range of temperate deciduous species in the
Betulaceae, Oleaceae, Salicaceae, and Rosaceae (Bossart
and Scriber, 1995).
About half of all Papilio species exhibit mimetic wing
patterns (Fig. 4). However, nonmimetic wing pattern is
very likely the ancestral state for Papilio, and the di-
rection of evolution among mimetic forms is unclear.
Mimetic phenotypes have evolved independently in Her-
aclides, Chilasa, Princeps (Menelaides), Eleppone, and Papilio
sensu stricto and at least twice in Pterourus and African
Princeps. Most Princeps (except several African repre-
sentatives) and Papilio (sensu stricto) are nonmimetic.
Many examples are restricted to female-limited Batesian
mimicry, and many of the mimetic forms are polymor-
phic. Among the most famous is the African P. dardanus,
which may have three sympatric forms of females that
mimic different species of the genus Amauris (Danainae)
(Clarke and Sheppard, 1960). In Southeast Asia, P. mem-
non also mimics three or more different models of troi-
dine Papilionidae (Clarke and Sheppard, 1971). For the
most part, the mimetic phenotype in both sexes seems
to be a derived character state with males occasionally
developing mimetic wing patterns in species with orig-
inally female-limited mimicry. However, both sexes in
Chilasa and Eleppone appear to have evolved mimetic
phenotypes directly from their respective sister non-
mimetic species. This may also be true of P. r e x , but phy-
logenetic study of more African mimetic species of Pa-
pilio (apparent relatives of the phorcas, rex, and delalandei
groups) is needed to explore this possibility.
Many other morphological characteristics of Papilio
may have polyphyletic origins (Fig. 4). Some of these
have been used as signiﬁcant taxonomic characters jus-
tifying Papilio subdivisions. For example, a granular fe-
male secretion attached to the egg surface, which is well
known in Troidini, is also reported from a number of
species of Papilio (Igarashi, 1984). In the context of our
phylogeny there is no consistent pattern for this charac-
ter among species groups, although it had been hypoth-
esized previously as a synapomorphy of Heraclides +
Chilasa (Igarashi, 1984). This character instead appears
to be either symplesiomorphic or derived in two sepa-
rate lineages (Chilasa and Heraclides)inPapilio.
Extensive metallic iridescent scales have evolved inde-
pendently in three different lineages of Papilio: Achillides,
known as gloss-papilios; the P. nireus group from African
Princeps; and Papilio troilus, a mimic of Battus philenor.
This trait appears convergent with iridescent scales in
some Troidini (e.g., Trogonoptera).
Species with curved minute hairs on the body of the
last instar larvae form a monophyletic group, as was
suggested by Igarashi (1984), with two exceptions: the
P. demoleus group and P. nobilis. More species of Princeps
should be studied to clarify the evolution of this trait.
Some morphological traits of Papilio larvae (such as
the morphology of prolegs) are tightly related to their
behavior. The number of crochet rows in abdominal pro-
legs is related to the ability of the larva to attach to the
host plant (Hagen, 1999). Larvae with clasping prolegs,
with only a single row of crochets, can cling more ﬁrmly
to thin twigs, whereas larvae with gripping prolegs and
two rows of crochets usually produce a silk pad on the
leaf surface and hold on to this pad. Twig-clasping pro-
legs seem to be plesiomorphic in Papilio. Although data
on proleg morphology are incomplete, there are several
examples of apparently independent gains or losses of
the two-row gripping prolegs and the ability to spin a
silk pad (Fig. 4).
These examples demonstrate that both adult and im-
mature morphological characters are highly labile both
within and among species. Experimental hybridization
of swallowtails indicates that in some cases wing and
body coloration in Papilio is controlled by a single gene or
set of tightly linked loci in a supergene, providing pheno-
typic plasticity (Clarke and Sheppard, 1960, 1971; Fisher,
1977; Nijhout, 1991; Loeliger and Karrer, 2000; ffrench-
Constant and Koch, 2003). Additional genetic data per-
taining to the extent to which subsets of the genome
maintain their genetic integrity would be very helpful
for understanding the contribution of genetic architec-
ture to speciation (Sperling, 2003).
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
6. Comparison of Ae’s (1979) differentiation index versus
COI-COII uncorrected percentage nucleotide sequence divergence for
Papilio species pairs from the same species groups, different species
groups, and different subgenera.
Taxonomic Implications: Is Papilio a Single Genus?
Papilio remains a taxonomic enigma. Few if any Pa-
pilio subdivisions can be delineated precisely using adult
or larval color pattern or other morphological charac-
ters (Munroe, 1961; Hancock, 1983; Igarashi, 1984; Miller,
1987). No consistent conclusions on taxonomic rank and
species relationships can be derived from the comparison
with hybrid incompatibility index (Ae, 1979). Although
distantly related taxa with up to 15% nucleotide sequence
divergence usually have high incompatibility scores and
rare cases of successful matings between species from
different subgenera may produce sterile hybrids (gen-
erally in the heterogametic sex; Sperling, 1990, 2003),
crosses between species within the same species group
may either completely fail or result in viable and fer-
tile hybrids (Fig. 6). Although there is much overlap be-
tween Ae’s differentiation index and groupings based
on comparisons within species groups, between species
groups, and between subgenera, there is much better
correspondence between mtDNA divergence and these
taxonomic groupings. An uncorrected mtDNA sequence
divergence of about 5% seems to provide a clear bound-
ary between comparisons within versus between species
An alternative method for evaluating the taxonomic
rank of Papilio subdivisions would be to compare their se-
quence divergence with known intergeneric differences
in other Lepidoptera. The close relatives of Papilio, the
Troidini, include three genera of birdwing butterﬂies,
Trogonoptera, Ornithoptera, and Troides, that apparently
have the same evolutionary ages as major Papilio sub-
divisions, around 40 MY (Morinaka et al., 1999, 2000),
and exhibit comparable nucleotide sequence divergence.
Average values for intra- and intergeneric divergences
among accepted Graphiini and Troidini genera (Parsons,
1996a, 1996b; Morinaka et al., 1999, 2000) and putative
genera of Papilio are presented in Figure 7, based on the
taxa included in Figure 1. Average divergences between
FIGURE 7. Average values of uncorrected percentage sequence di-
vergence in two mitochondrial genes and one nuclear gene compared
at different taxonomic levels within the Papilionidae. Three putative
genera in Papilio represent Heraclides, Pterourus + Chilasa + Papilio alex-
anor, and Princeps + Eleppone + Princeps sensu stricto. Twelve putative
genera constitute all major lineages from Figure 1 as A, B, C, D, E, F,
G, H + I, J, K, L, and M. Solid bars indicate 95% conﬁdence interval;
dashed bars show ranges of values.
putative genera within Papilio are comparable to inter-
generic distances in Troidini and Graphiini, whereas di-
vergences within the putative genera are substantially
lower. Average divergences in EF-1α at different taxo-
nomic levels within Papilioninae are comparable to pair-
wise divergences in Noctuidae (primarily Heliothinae)
that fall within the range of 10.1–13.5% for intertribal
and 6.5–9.0% for intergeneric distances (Mitchell et al.,
1997). However, although these data can be considered
to support the elevation of various Papilio subdivisions
to generic level, we will not endorse speciﬁc changes
based on sequence divergences until a more thorough
sampling of species can be included.
Mitochondrial (COI-COII) plus nuclear (EF-1α) DNA
sequence data have provided considerable new resolu-
tion for the phylogeny of Papilio, a large and important
genus for research in ecology, genetics, and evolution-
ary and conservation biology. This phylogeny provides
a number of insights into the evolutionary history of the
212 SYSTEMATIC BIOLOGY VOL.
However, estimation of divergence times remains a
difﬁcult task that involves substantial uncertainties due
to a lack of reliable fossil records or well-dated vicari-
ance events. In this study, we relied on several calibra-
tion points that were applied in different combinations.
The inferred estimate for the divergence of Papilio from
the Troidini (up to 100 MYA, when all other calibration
points were enforced with no constraint age at the tree
root) and the age of subdivisions within Papilio (around
30–50 MYA) ﬁt well with the results of one other in-
sect molecular clock calibration (Gaunt and Miles, 2002).
These estimates agree with dating of the diversiﬁcation
of the major group of host plants in swallowtails, the Ru-
taceae (Magall´on et al., 1999). With few exceptions, the
dates conform with likely vicariant dispersal patterns in
the family (see Scriber et al., 1995) and are consistent with
known fossil records (see Emmel et al., 1992).
Papilio is relatively old and is a highly complex and
diverse taxonomic unit. Many characteristics of Papilio
have a polyphyletic origin, with multiple gains and
losses of particular morphological or ecological traits.
Although no deﬁnitive morphological synapomorphies
for generic subdivisions are broadly recognized, the evo-
lutionary age and degree of genetic divergence between
monophyletic groups within Papilio would support ele-
vation of at least ﬁve taxa, Heraclides, Pterourus, Chilasa,
Papilio, and Eleppone, to generic rank. A new genus name
would also be required for P. alexanor. Further studies
are required to clarify the composition of Princeps (sensu
Hancock, 1983), which in contrast to traditional taxon-
omy is paraphyletic and seems to include at least three
divergent groups. However, we refrain from offering for-
mal taxonomic conclusions until evolutionary patterns
are clariﬁed more consistently across Papilio.
We are grateful to S. A. Ae, A. V. Z. Brower, K. S. Brown, Jr., A.
Cieslak, C. A. Clarke, S. J. Collins, C. N. Duckett, P. P. Feeny, D. Goh, Y. F.
Hsu, R. V. Kelson, J. Y. Miller, J. Okura, N. D. Penny, R. D. Reed, A. A. So-
lis, B. J. Walsh, and J. D. Weintraub, who helped to provide specimens.
Collections in China were supported by the State Key Laboratory for
Biocontrol, Zhongshan University, and could not have been completed
without the gracious assistance of our colleagues from South China
Agricultural University (Guangzhou, People’s Republic of China). We
thank M. Dear, M. Kuo, and G. Taylor for laboratory support and S.
Morinaka for pointing out a sequence correction required for Atropha-
neura alcinous. We thank Chris Simon, Ted Schultz, and two anonymous
reviewers whose comments helped to improve this manuscript. The
study was supported by grants from NSERC Canada and California
AES to F.A.H.S.
Ackery, P. R., C. R. Smith, and R. I. Vane-Wright. 1995. Carcasson’s
African butterﬂies—An annotated catalogue of the Papilionoidea
and Hesperioidea of the Afrotropical Region. CSIRO, Canberra,
Australia and NMH, London, UK.
Ae, S. A. 1960. A study of hybrids betweeen Papilio xuthus and the
P. polyxenes-machaon group. J. Lepid. Soc. 14:5–18.
Ae, S. A. 1962. A study of interspeciﬁc hybrids in black swallowtails in
Japan. J. Lepid. Soc. 15:175–190.
Ae, S. A. 1979. The phylogeny of some Papilio based on interspeciﬁc
hybridization data. Syst. Entomol. 4:1–16.
Ae, S. A. 1990. A study of the Papilio bianor group mainly based on hy-
bridization. Tyˆo to Ga [Trans. Lepid. Soc. Jpn.] 41:13–19. (In Japanese.)
Ae, S. A. 1995. Ecological and evolutionary aspects of hybridization
in some Papilio butterﬂies. Pages 229–236 in Swallowtail butterﬂies:
Their ecology and evolutionary biology (J. M. Scriber, Y. Tsubaki, and
R. C. Lederhouse, eds.). Scientiﬁc Publishers, Gainesville, Florida.
Alvarez, W. 1972. Rotation of the Corsica–Sardinia microplate. Nat.
Phys. Sci. 235:103–105.
Andersen, N. M., L. Cheng, J. Damgaard, and F. A. H. Sperling.
2000. Mitochondrial DNA sequence variation and phylogeography
of oceanic insects (Hemiptera: Gerridae: Halobates spp.). Mar. Biol.
Aubert, J., L. Legal, H. Descimon, and F. Michel. 1999. Molecular phy-
logeny of swallowtail butterﬂies of the tribe Papilionini (Papilion-
idae, Lepidoptera). Mol. Phylogenet. Evol. 12:156–167.
Avise, J. C. 1994. Molecular markers, natural history and evolution.
Chapman and Hall, New York.
Baker, R. H., and R. DeSalle. 1997. Multiple sources of molecular char-
acteristics and the phylogeny of Hawaiian drosophilids. Syst. Biol.
Barker, F. K., and F. M. Lutzoni. 2002. The utility of the incongruence
length difference test. Syst. Biol. 51:625–637.
Beckenbach, A. T., Y.-W. Wei, and H. Liu. 1993. Relationships in the
Drosophila obscura species group inferred from mitochondrial cy-
tochrome oxidase II sequences. Mol. Biol. Evol. 10:619–634.
Beerli, P., H. Hotz, and T. Uzzell. 1996. Geologically dated sea barriers
calibrate a protein clock for Aegean water frogs. Evolution 50:1676–
Beutelspacher, C. R. 1975. Una especie nueva de Papilio L. (Papilion-
idae). Rev. Soc. Mex. Lepid. 1:3–6.
Bossart, J. L., and J. M. Scriber. 1995. Genetic variation in oviposition
preference in tiger swallowtail butterﬂies: Interspeciﬁc, interpop-
ulation and interindividual comparisons. Pages 183–194 in Swal-
lowtail butterﬂies: Their ecology and evolutionary biology (J. M.
Scriber, Y. Tsubaki, and R. C. Lederhouse, eds.). Scientiﬁc Publishers,
Bowen, G. J., W. C. Clyde, P. L. Koch, S. Ting, J. Alroy, T. Tsubamoto,
Y. Wang, and Y. Wang. 2002. Mammalian dispersal at the Pale-
ocene/Eocene boundary. Science 295:2062–2065.
Braby, M. F. 2000. Butterﬂies of Australia, their identiﬁcation, biology
and distribution. CSIRO, Collingwood, Australia.
Bremer, K. 1988. The limits of amino acid sequence data in angiosperm
phylogenetic reconstruction. Evolution 42:795–803.
Britten, R. J. 1986. Rates of DNA sequence evolution differ between
taxonomic groups. Science 231:1393–1398.
Brower, A. V. Z. 1994. Phylogeny of Heliconius butterﬂies inferred from
mitochondrial DNA sequences (Lepidoptera: Nymphalidae). Mol.
Phylogenet. Evol. 3:159–174.
Caterino, M. S., R. D. Reed, M. M. Kuo, and F. A. H. Sperling. 2001.
A partitioned likelihood analysis of swallowtail butterﬂy phylogeny
(Lepidoptera: Papilionidae). Syst. Biol. 50:106–127.
Caterino, M. S., and F. A. H. Sperling. 1999. Papilio phylogeny based on
mitochondrial cytochrome oxidase I and II genes. Mol. Phylogenet.
Chapco, W., G. Litzenberger, and W. R. Kuperus. 2001. A molecular
biogeographic analysis of the relationship between North American
melanoploid grasshoppers and their Eurasian and South American
relatives. Mol. Phylogenet. Evol. 18:460–466.
Cho, S., A. Mitchell, J. C. Regier, C. Mitter, R. W. Poole, T. P. Friedlander,
and S. Zhao. 1995. A highly conserved gene for low level phyloge-
netics: Elongation factor-1 alpha recovers morphology-based tree for
heliothine moths. Mol. Biol. Evol. 12:650–656.
Clarke, C. A. 1952. Hand-pairings of Papilio machaon in February. En-
tomol. Rec. 64:98–100.
Clarke, C. A., C. G. C. Dickson, and P. M. Sheppard. 1963. Larval color
pattern in Papilio demodocus. Evolution 17:130–137.
Clarke, C. A., and P. M. Sheppard. 1955. The breeding in captivity of
the hybrid, Papilio rutulus female × Papilio glaucus male. Lepid. News
Clarke, C. A., and P. M. Sheppard. 1956. A further report on the genetics
of the machaon group of swallowtail butterﬂies. Evolution 10:66–73.
Clarke, C. A., and P. M. Sheppard. 1957. The breeding in captivity of the
hybrid Papilio glaucus × Papilio eurymedon. Lepid. News 11:201–205.
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
Clarke, C. A., and P. M. Sheppard. 1960. The genetics of Papilio dardanus.
Clarke, C. A., and P. M. Sheppard. 1971. Further studies on the genetics
of the mimetic butterﬂy Papilio memnon L. Philos. Trans. R. Soc. Lond.
Clary, D. O., and D. R. Wolstenholme. 1985. The mitochondrial DNA
molecule of Drosophila yakuba: Nucleotide sequence, gene organiza-
tion and genetic code. J. Mol. Evol. 22:252–271.
Collins, N. M., and M. G. Morris. 1985. Threatened swallowtail but-
terﬂies of the world. The IUCN Red Data Book. IUCN, Cambridge,
Cook, P. J. 1990. Australia: Evolution of a continent. BMR Palaeo-
geographic Group, Australian Government Publishing Service,
Crespi, B. J., D. A. Carmean, L. A. Mound, M. Worobey, and D.
Morris. 1998. Phylogenetics of social behavior in Australian gall-
forming thrips: Evidence from mitochondrial DNA sequence, adult
morphology and behavior, and gall morphology. Mol. Phylogenet.
Crozier, R. H., Y. C. Crozier, and A. G. Mackinlay. 1989. The CO-I and
CO-II region of honeybee mitochondrial DNA: Evidence for vari-
ation in insect mitochondrial evolutionary rates. Mol. Biol. Evol.
Darlu, P., and G. Lecointre. 2002. When does the incogruence length
difference test fail? Mol. Biol. Evol. 19:432–437.
DeSalle, R., T. Freedman, E. M. Prager, and A. C. Wilson. 1987. Tempo
and mode of sequence evolution in mitochondrial DNA of Hawaiian
Drosophila. J. Mol. Evol. 26:157–164.
Dietz, R. S., and J. C. Holden. 1970. Reconstruction of Pangeae: Breakup
and dispersion of continents, Permian to present. J. Geophys. Res.
Donnelly, T. W. 1988. Geologic constraints on Caribbean biogeography.
Pages 15–37 in Zoogeography of Caribbean insects (J. K. Liebherr,
ed.). Cornell Univ. Press, Ithaca, New York.
Drake, V. A., and A. G. Gatehouse. 1995. Insect migration: Tracking re-
sources through space and time. Cambridge Univ. Press, Cambridge,
Durdon, C. J., and H. Rose. 1978. Butterﬂies from the middle Eocene:
The earliest occurrence of fossil Papilionidae. Pearce-Sellards Ser.
Tex. Mem. Mus. 29:1–25.
Emmel, T. C., M. C. Minno, and B. A. Brummond. 1992. Florissant
butterﬂies. A guide to the fossil and present-day species of central
Colorado. Stanford Univ. Press, Polo Alto, California.
Eriksson, T. 1998. AutoDecay, version 4.0. Program distributed by the
author. Department of Botany, Stockholm Univ. Stockholm.
Farris, J. C., M. K¨alersj¨o, A. G. Kluge, and C. Bult. 1994. Testing the
signiﬁcance of incongruence. Cladistics 10:315–319.
Felsenstein, J. 1978. Cases in which parsimony or compatibility meth-
ods will be positively misleading. Syst. Zool. 27:401–410.
Felsenstein, J. 1981. A likelihood approach to character weighting and
what it tells us about parsimony and compatibility. Biol. J. Linn. Soc.
Felsenstein, J. 1985. Conﬁdence limits on phylogenies: An approach
using the bootstrap. Evolution. 39:783–791.
Felsenstein, J. 1988. Phylogenies from molecular sequences: Inference
and reliability. Annu. Rev. Genet. 22:521–565.
Felsenstein, J. 1989. PHYLIP—Phylogeny inference package (ver-
sion 3.2). Cladistics 5:164–166.
ffrench-Constant, R., and P. B. Koch. 2003. Mimicry and melanism in
swallowtail butterﬂies: Toward a molecular understanding. Pages
259–280 in Butterﬂies: Ecology and evolution taking ﬂight (C. L.
Boggs, W. B. Watt, and P. R. Ehrlich, eds.). Univ. Chicago Press,
Fisher, M. S. 1977. The taxonomy and identity of Papilio nitra W. H.
Edwards in Colorado (Papilionidae). Bull. Allyn Mus. 47:1–7.
Gaunt, M. W., and M. A. Miles. 2002. An insect molecular clock dates
the origin of the insects and accords with palaeontological and bio-
geographic landmarks. Mol. Biol. Evol. 19:748–761.
Gielly, L., and P. Taberlet. 1994. The use of chloroplast DNA to re-
solve plant phylogenies: Noncoding versus rbcL sequences. Mol.
Biol. Evol. 11:769–777.
Gillespie, J. H. 1986. Rates of molecular evolution. Annu. Rev. Syst.
Graham, S. W., J. R. Kohn, B. R. Morton, J. E. Echenwalder, and S. C. H.
Barret. 1998. Phylogenetic congruence and discordance among one
morphological and three molecular data sets from Pontederiaceae.
Syst. Biol. 47:545–567.
Hagen, R. H. 1999. Prolegs of Papilionini (Lepidoptera: Papilionidae):
Alternative solutions to the problem of attachment. Univ. Kans. Nat.
Hist. Mus. Spec. Publ. 24:237–251.
Hagen, R. H., and J. M. Scriber. 1991. Systematics of the Papilio glaucus
and P. troilus species groups (Lepidoptera: Papiliondae): Inferences
from allozymes. Ann. Entomol. Soc. Am. 84:380–395.
Hall, R. 1996. Reconstructing Cenozoic SE Asia. Pages 153–184 in Tec-
tonic evolution of Southeast Asia (R. Hall and D. Blundell, eds.)
Special Publication 106. Geological Society of London, London.
Hall, R. 2001. Cenozoic reconstructions of SE Asia and the SW Paciﬁc:
Changing patterns of land and sea. Pages 35–56 in Faunal and ﬂoral
migrations and evolution in SE Asia–Australia (I. Metcalfe, J. M. B.
Smith, M. Morwood, and I. D. Davidson, eds.). Swets and Zeitlinger,
Hancock, D. L. 1979. The systematic position of Papilio anactus Macleay
(Lepidoptera: Papilionidae). Aust. Entomol. Mag. 6:49–53.
Hancock, D. L. 1983. Classiﬁcation of the Papilionidae (Lepidoptera):
A phylogenetic approach. Smithersia 2:1–48.
Haq, B. U., J. Hardenbol, and P. R. Vail. 1988. Mesozoic and Cenozoic
chronostratigraphy and eustatic cycles. Pages 71–108 in Sea-level
changes: An integrated approach (C. K. Wilgus, B. S. Hastings, H.
Posamentier, J. V. Wagoner, C. A. Ross, and C. G. S. C. Kendall, eds.).
Special Publication 42. Society of Economic Paleontologists and Min-
eralogists, Tulsa, Oklahoma.
Harrison, R. G., and S. M. Bogdanowicz. 1995. Mitochondrial DNA
phylogeny of North American ﬁeld crickets: Perspectives on the
evolution of life cycles, song, and habitat associations. J. Evol. Biol.
Hasegawa, A. M., and T. Yano. 1984. Phylogeny and classiﬁcation of
Hominoidea as inferred from DNA-sequence data. Proc. Jpn. Acad.
Ser. B. 60:389–392.
Ha ¨user, C. L. 1993. Critical comments on the phylogenetic relationships
within the family Papilionidae. Nota Lepid. 16:34–43.
Ha ¨user, C. L., R. de Jong, G. Lamas, R. K. Robbins, C. Smith,
and R. I. Vane-Wright. 2002. Papilionidae—Revised GloBIS/GART
species checklist, 2nd draft. Available at http://www.insects-
Hay, W. W., R. M. DeConto, C. N. Wold, K. M. Wilson, S. Voigt,
M. Schulz, A. Wold-Russby, W.-C. Dullo, A. B. Ronov, A. N.
Balukhovsky, and E. S ¨oding. 1999. Alternative global Cretaceous
paleogeography. Pages 1–47 in Evolution of the Cretaceous ocean-
climate system (E. Barrera and C. C. Johnson, eds.). Special Paper
332. Geological Society of America, Boulder, Colorado.
Hebert, P. D. N., E. A. Remigio, J. K. Colbourne, D. J. Taylor, and C. C.
Wilson. 2002. Accelerated molecular evolution in halophilic crus-
taceans. Evolution 56:909–926.
Hillis, D. M. 1996. Inferring complex phylogenies. Nature 383:130–
Holloway, J. D., and E. S. Nielsen. 1999. Biogeography of the Lepi-
doptera. Pages 423–462 in Handbook of zoology, Part 35. Lepidoptera
(N. P. Kristensen, ed.). De Gruyter, New York.
Hsu, K. J. 1972. When the Mediterranean dried up. Sci. Am. 227:27–36.
Huelsenbeck, J. P. 1995. The performance of phylogenetic methods in
simulation. Syst. Biol. 44:17–48.
Huelsenbeck, J. P., and J. P. Bollback. 2001. Empirical and hierarchical
Bayesian estimation of ancestral states. Syst. Biol. 50:351–366.
Huelsenbeck, J. P., and K. A. Crandall. 1997. Phylogeny estimation and
hypothesis testing using maximum likelihood. Annu. Rev. Ecol. Syst.
Huelsenbeck, J. P., and F. Ronquist. 2001. MrBayes: Bayesian inference
of phylogeny. Bioinformatics 17:754–755.
Huelsenbeck, J. P., F. Ronquist, R. Nielsen, and J. P. Bollback. 2001.
Bayesian inference of phylogeny and its impact on evolutionary bi-
ology. Science 294:2310–2314.
Igarashi, S. 1984. The classiﬁcation of the Papilionidae mainly based on
the morphology of their immature stages. Tyˆo to Ga [Trans. Lepid.
Soc. Jpn.] 34:41–96.
Jacobs, L. L., and D. Pilbeam. 1980. Of mice and men—Fossil-based
divergence dates and molecular clocks. J. Hum. Evol. 9:551–555.
214 SYSTEMATIC BIOLOGY VOL.
Kamiie, K., H. Taira, H. Ooura, A. Kakuta, S. Matsumoto, S. Ejiri, and
T. Katsumata. 1993. Nucleotide sequence of the cDNA encoding silk
gland elongation factor 1 alpha. Nucleic Acids Res. 21:742.
Kimura, M. 1983. The neutral theory of molecular evolution. Cam-
bridge Univ. Press, Cambridge, U.K.
Klompen, J. S. H., W. C. Black, J. E. Keirans, and D. E. Norris. 2000. Sys-
tematics and biogeography of hard ticks, a total evidence approach.
Kurt´en, B. 1971. The age of mammals. Weidenfeld and Nicolson,
Lanave, C., G. Preparata, C. Saccone, and G. Serio. 1984. A new method
for calculating evolutionary substitution rates. J. Mol. Evol. 20:86–93.
Larget, B., and D. L. Simon. 1999. Markov chain Monte Carlo algo-
rithms for the Bayesian analysis of phylogenetic trees. Mol. Biol.
Lewis, P. O. 2001. A likelihood approach to estimating phylogeny from
discrete morphological character data. Syst. Biol. 50:913–925.
Li, W.-H. 1997. Molecular evolution. Sinauer, Sunderland,
Li, Z. X., and C. M. Powell. 2001. An outline of the palaeogeographic
evolution of the Australasian region since the beginning of the Neo-
proterozoic. Earth-Sci. Rev. 53:237–277.
Loeliger, E. A., and F. Karrer. 2000. Unusual demonstration of autoso-
mal dominant inheritance of the black coloration of one of America’s
broods of the hybrids Papilio polyxenes asterius with
Papilio machaon gorganus (Papilionidae). Nota Lepid. 23:40–49.
Machado, C. A., E. M. Jousselin, F. Kjellberg, S. G. Gompton, and E.
A. Herre. 2001. Phylogenetic relationships, historical biogeography
and character evolution of ﬁg-pollinating wasps. Proc. R. Soc. Lond.
Maddison, D. R., and W. P. Maddison. 2000. MacClade 4: Analy-
sis of phylogeny and character evolution. Sinauer, Sunderland,
Maddison, W. P., and D. R. Maddison. 2003. Mesquite: A mod-
ular system for evolutionary analysis, version 1.0. Available at
Magall´on, P. S., P. S. Herendeen, and P. R. Grane. 1999. Phylogenetic pat-
tern, diversity, and diversiﬁcation of eudicots. Ann. Mo. Bot. Gard.
Makita, H., T. Shinkawa, K. Kondo, L. Xing, and T. Nakazawa. 2003.
Phylogeny of the Graphium butterﬂies inferred from nuclear 28S
rDNA and mitochondrial ND5 gene sequences. Trans. Lepid. Soc.
Mantovani, B., M. Passamonti, and V. Scali. 2001. The mitochondrial
cytochrome oxidase II gene in Bacillus stick insects: Ancestry of
hybrids, androgenesis, and phylogenetic relationships. Mol. Phylo-
genet. Evol. 19:157–163.
Matsui, H., R. Tada, and T. Oba. 1998. Low-salinity isolation event in
the Japan Sea in response to eustatic sea-level drop during LGM:
Reconstruction based on salinity-balance model. Quat. Res. 37:221–
McKenzie, D. P., and J. C. Sclater. 1973. The evolution of the Indian
Ocean. Sci. Am. 228:63–72.
Miller, J. S. 1987. Phylogenetic studies in the Papilioninae (Lepidoptera:
Papilionidae). Bull. Am. Mus. Nat. Hist. 186:365–512.
Mitchell, A., S. Cho, J. C. Regier, C. Mitter, R. W. Poole, and M.
Matthews. 1997. Phylogenetic utility, nucleotide sequence variation
in a 1240-bp fragment of the elongation factor-1 alpha in Noctuioidea
(Insecta: Lepioptera): The limits of synonymous substitution. Mol.
Biol. Evol. 14:381–390.
Morinaka, S., T. Maeyama, K. Maekawa, D. Erniwati, S. N. Prijono, I. K.
Ginarsa, T. Nakazawa, and T. Hidaka. 1999. Molecular phylogeny of
birdwing butterﬂies based on the representatives in most genera of
the tribe Troidini (Lepidoptera: Papilionidae). Entomol. Sci. 2:347–
Morinaka, S., N. Minaka, M. Sekiguchi, D. Erniwati, S. N. Prijono, I. K.
Ginarsa, and T. Hidaka. 2000. Molecular phylogeny of birdwing but-
terﬂies of the tribe Troidini (Lepidoptera: Papilionidae)—Using all
species of the genus Ornithoptera. Biogeography 2:103–111.
Munroe, E. 1961. The classiﬁcation of the Papilionidae (Lepidoptera).
Can. Entomol. Suppl. 17:1–51.
Nijhout, H. F. 1991. The development and evolution of butterﬂy wing
patterns. Smithsonian Institution Press, Washington, D.C.
Noonan, G. R. 1988. Faunal relationship between eastern North Amer-
ica and Europe as shown by insects. Mem. Entomol. Soc. Can. 144:39–
Oshima, K. 1990. The history of straits around the Japanese Islands in
the late Quaternary. Quat. Res. 29:193–208.
Page, R. D. M., and E. C. Holmes. 1998. Molecular evolution. A phylo-
genetic approach. Blackwell Science, Oxford, U.K.
Page, R. D. M., P. L. M. Lee, S. A. Becher, R. Grifﬁths, and D. H. Clayton.
1998. A different tempo of mitochondrial DNA evolution in birds and
their parasitic lice. Mol. Phylogenet. Evol. 9:276–293.
Parsons, M. J. 1996a. Gondwanan evolution of the Troidine swallow-
tails (Lepidoptera: Papilionidae): Cladistic reappraisal using mainly
immature stage characters, with focus on the birdwings Ornithoptera
Boisduval. Bull. Kitakyushu Mus. Nat. Hist. 15:43–118.
Parsons, M. J. 1996b. A phylogenetic reappraisal of the birdwing genus
Ornithoptera (Lepidoptera: Papilionidae: Troidini) and a new theory
of its evolution in relation to Gondwanan vicarience biogeography.
J. Nat. Hist. 30:1707–1736.
Pawlowski, J., I. Bolivar, J. F. Fahrni, C. de Vargas, M. Gouy, and L.
Zaninetti. 1997. Extreme differences in rates of molecular evolution of
Foraminifera revealed by comparison of ribosomal DNA sequences
and the fossil record. Mol. Biol. Evol. 14:498–505.
Pellmyr, O., and J. Leebens-Mack. 1999. Forty million years of mutual-
ism: Evidence for Eocene origin of the yucca–yucca moth association.
Proc. Natl. Acad. Sci. USA 96:9178–9183.
Posada, D., and K. A. Crandall. 1998. Modeltest: Testing the model of
DNA substitution. Bioinformatics 14:817–818.
Rambault, A. 2002. Sequence-alignment editor, version 2.0. Available
Rambaut, A., and L. Bromham. 1998. Estimating divergence dates from
molecular sequences. Syst. Biol. 49:579–591.
Raven, P. H., and D. I. Axelrod. 1972. Plate tectonics and Australasian
paleobiogeography. Science 176:1379–1386.
Reed, R. D., and F. A. H. Sperling. 1999. Interaction of process partitions
in phylogenetic analysis: An example from the swallowtail butterﬂy
genus Papilio. Mol. Biol. Evol. 16:286–297.
Remington, C. L. 1960. Wide experimental crosses between Papilio
xuthus and other species. J. Lepid. Soc. 13:151–164.
Remsen, J., and R. DeSalle. 1998. Character congruence of multiple
data partitions and the origin of the Hawaiian Drosophilidae. Mol.
Phylogenet. Evol. 9:225–235.
Sampson, S. D., L. M. Witmer, C. A. Forster, D. W. Krause, P. M.
O’Connor, P. Dodson, and F. Ravoavy. 1998. Predatory dinosaur re-
mains from Madagascar: Implications for the Cretaceous biogeogra-
phy of Gondwana. Science. 280:1048–1051.
Sanderson, M. J. 1997. A nonparametric approach to estimating diver-
gence times in the absence of rate constancy. Mol. Biol. Evol. 14:1218–
Sanderson, M. J. 2002. Estimating absolute rates of molecular evolution
and divergence times: A penalized likelihood approach. Mol. Biol.
Sanderson, M. J., and J. Kim. 2000. Parametric phylogenetics? Syst. Biol.
Scott, J. A. 1986. The butterﬂies of North America: A natural history
and ﬁeld guide. Stanford Univ. Press, Palo Alto, California.
Scriber, J. M. 1995a. Hybridization studies with North American swal-
lowtails. Pages 269–282 in Swallowtail butterﬂies: Their ecology and
evolutionary biology (J. M. Scriber, Y. Tsubaki, and R. C. Lederhouse,
eds.). Scientiﬁc Publishers, Gainesville, Florida.
Scriber, J. M. 1995b. Overview of swallowtail butterﬂies: Taxonomic and
distributional latitude. Pages 3–8 in Swallowtail butterﬂies: Their
ecology and evolutionary biology (J. M. Scriber, Y. Tsubaki, and
R. C. Lederhouse, eds.). Scientiﬁc Publishers, Gainesville, Florida.
Scriber, J. M., R. C. Lederhouse, and R. V. Dowell. 1991. Hybridiza-
tion of the Brazilian Papilio (Pyrrhosticta) (section V) with the North
American Papilio (Pterourus) (section III). J. Res. Lepid. 29:21–32.
Scriber, J. M., Y. Tsubaki, and R. C. Lederhouse (eds.). 1995. Swallow-
tail butterﬂies: Their ecology and evolutionary biology. Scientiﬁc
Publishers, Gainesville, Florida.
Shimogori, Y. 1997. Achillides butterﬂies (Papilionidae), Volume 6. End-
less Science, Tokyo.
Simon, C., F. Frati, A. Beckenbach, B. J. Crespi, H. Liu, and P. Flook. 1994.
Evolution, weighting, and phylogenetic utility of mitochondrial gene
2004 ZAKHAROV ET AL.—PAPILIO PHYLOGENY
sequences and a compilation of conserved polymerase chain reaction
primers. Ann. Entomol. Soc. Am. 87:651–701.
Smith, A. B., B. Lafay, and R. Christen. 1992. Comparative variation of
morphological and molecular evolution through geologic time: 28S
ribosomal RNA versus morphology in echinoids. Philos. Trans. R.
Soc. Lond. B 338:365–382.
Smith, D. G., and R. G. Coss. 1984. Calibrating the molecular clock:
Estimates of ground squirrel divergence made using fossil and geo-
logical time markers. Mol. Biol. Evol. 1:249–259.
Soltis, P. S., D. E. Soltis, V. Savolainen, P. R. Crane, and T. G.
Barraclough. 2002. Rate heterogeneity among lineages of tracheo-
phytes: Integration of molecular and fossil data and evidence for
molecular living fossils. Proc. Natl. Acad. Sci. USA 99:4430–4435.
Sperling, F. A. H. 1987. Evolution of the Papilio machaon species group
in western Canada. Quaest. Entomol. 23:198–315.
Sperling, F. A. H. 1990. Interspeciﬁc hybrids in Papilio butterﬂies: Poor
taxonomy or interesting evolutionary problem?Can. J. Zool. 68:1790–
Sperling, F. A. H. 1991. Mirochondrial DNA phylogeny, speciation,
and host-plant coevolution of Papilio butterﬂies. Ph.D. Dissertation,
Cornell Univ. Ithaca, New York.
Sperling, F. A. H. 1993a. Mitochondrial DNA phylogeny of the Papilio
machaon species group (Lepidoptera: Papilionidae). Mem. Entomol.
Soc. Can. 165:233–242.
Sperling, F. A. H. 1993b. Mirochondrial DNA variation and Haldane’s
rule in the Papilio glaucus and Papilio troilus species group. Heredity
Sperling, F. A. H. 2003. Butterﬂy molecular systematics: From species
deﬁnitions to higher-level phylogenies. Pages 431–458 in Butterﬂies:
Ecology and evolution taking ﬂight (C. L. Boggs, W. B. Watt, and
P. R. Ehrlich, eds.). The Univ. Chicago Press, Chicago.
Sperling, F. A. H., and P. Feeny. 1995. Umbellifer and composite feeding
in Papilio: Phylogenetic frameworks and constraints on caterpillars.
Pages 299–306 in Swallowtail butterﬂies: Their ecology and evolu-
tionary biology (J. M. Scriber, Y. Tsubaki, and R. C. Lederhouse, eds.).
Scientiﬁc Publishers, Gainesville, Florida.
Sperling, F. A. H., and R. G. Harrison. 1994. Mitochondrial DNA vari-
ation within and between species of the Papilio machaon group of
swallowtail butterﬂies. Evolution 48:408–422.
Sperling, F. A. H., J. R. Spence, and N. M. Andersen. 1997. Mitochondrial
DNA, allozymes, morphology, and hybrid compatibility in Limno-
porus water striders (Heteroptera: Gerridae): Do they all track species
phylogenies? Ann. Entomol. Soc. Am. 90:401–415.
Su, Z.-H., O. Rominaga, M. Okamoto, and S. Osawa. 1998. Origin
and diversiﬁcation of hindwingless Damaster ground beetles within
the Japanese Islands as deduced from mitochondrial ND5 gene se-
quences (Coleoptera, Carabidae). Mol. Biol. Evol. 15:1026–1039.
Swofford, D. L. 1998. PAUP
: Phylogenetic analysis using parsi-
and other methods), version 4.0. Sinauer, Sunderland,
Tamura, K., and M. Nei. 1993. Estimation of the number of nucleotide
substitutions in the control region of mitochondrial DNA in humans
and chimpanzees. Mol. Biol. Evol. 10:512–526.
Thompson, J. D., T. J. Gibson, F. Plewniak, F. Jeanmougin, and D. G.
Higgins. 1997. The ClustalX Windows interface: Flexible strategies
for multiple sequence alignment aided by quality analysis tools. Nu-
cleic Acids Res. 25:4876–4882.
Thorne, J. K., H. Kishino, and I. S. Painter. 1998. Estimating the rate of
evolution of the rate of evolution. Mol. Biol. Evol. 15:1647–1657.
Tiffney, B. H. 1985. The Eocene North Atlantic bridge: Its importance in
Tertiary and modern phytogeography of the Northern Hemisphere.
J. Arnold Arbor. 66:243–273.
Tyler, H. A., K. S. Brown, and K. Wilson. 1994. Swallowtail butterﬂies of
the Americas: A study in biological dynamics, ecological diversity,
biosystematics, and conservation. Scientiﬁc Publishers, Gainesville,
Vane-Wright, R. I., D. C. Raheem, A. Cieslak, and A. P. Vogler. 1999.
Evolution of the mimetic African swallowtail butterﬂy Papilio dar-
danus: Molecular data conﬁrm relationships with P. phorcas and
P. constantinus. Biol. J. Linn. Soc. 66:215–229.
Vogler, A. P., and A. Welsh. 1997. Phylogeny of North American Ci-
cindela tiger beetles inferred from multiple mitochondrial DNA se-
quences. Mol. Phylogenet. Evol. 8:225–235.
Walker, J. J. 1882. A life history of Papilio paeon, Roger. Entomol. Month.
Wallace, A. R. 1864. On the phenomena of variation and geographical
distribution as illustrated by the Papilionidae of the Malayan region.
Trans. Linn. Soc. Lond. 25:1–71.
Williams, C. B. 1930. The migration of butterﬂies. Oliver and Boyd,
Yagi, T., T. Katoh, A. Chichvarkhin, T. Shinkawa, and K. Omoto. 2001.
Molecular phylogeny of butterﬂies Parnassius glacialis and P. stubben-
dorﬁi at various localities in East Asia. Genes Genet. Syst. 76:229–
Yagi, T., G. Sasaki, and H. Takebe. 1999. Phylogeny of Japanese papil-
ionid butterﬂies inferred from nucleotide sequences of the mitochon-
drial ND5 gene. J. Mol. Evol. 48:42–48.
Yang, Z. 1996. Among-site rate variation and its impact on phylogenetic
analysis. Trends Ecol. Evol. 11:367–372.
Zuckerkandl, E., and L. Pauling. 1962. Molecular disease, evolution,
and genetic heterogeneity. Pages 189–225 in Horizons in biochem-
istry (M. Kasha and B. Pullman, eds.). Academic Press, New York.
Zuckerkandl, E., and L. Pauling. 1965. Evolutionary divergence and
convergence. Pages 97–166 in Evolving genes and proteins (V. Bryson
and H. J. Vogel, eds.). Academic Press, New York.
First submitted 13 December 2002; reviews returned 2 April 2003;
ﬁnal acceptance 15 November 2003
Associate Editor: Ted Schultz