Hebert PDN, Penton EH, Burns JM, Janzen DH, Hallwachs W. Ten species in one: DNA barcoding reveals cryptic species in the neotropical skipper butterfly Astraptes fulgerator. Proc Natl Acad Sci USA 101: 14812-14817

Abstract

Astraptes fulgerator, first described in 1775, is a common and widely distributed neotropical skipper butterfly (Lepidoptera: Hesperiidae). We combine 25 years of natural history observations in northwestern Costa Rica with morphological study and DNA barcoding of museum specimens to show that A. fulgerator is a complex of at least 10 species in this region. Largely sympatric, these taxa have mostly different caterpillar food plants, mostly distinctive caterpillars, and somewhat different ecosystem preferences but only subtly differing adults with no genitalic divergence. Our results add to the evidence that cryptic species are prevalent in tropical regions, a critical issue in efforts to document global species richness. They also illustrate the value of DNA barcoding, especially when coupled with traditional taxonomic tools, in disclosing hidden diversity.

Full text

Ten species in one: DNA barcoding reveals cryptic
species in the neotropical skipper butterfly
Astraptes fulgerator
Paul D. N. Hebert*
†
, Erin H. Penton*, John M. Burns
‡
, Daniel H. Janzen
§
, and Winnie Hallwachs
§
*Department of Zoology, University of Guelph, Guelph, ON, Canada N1G 2W1; ‡Department of Entomology, National Museum of Natural History,
Smithsonian Institution, Washington, DC 20560-0127; and §Department of Biology, University of Pennsylvania, Philadelphia, PA 19104
Contributed by Daniel H. Janzen, August 20, 2004
Astraptes fulgerator, first described in 1775, is a common and
widely distributed neotropical skipper butterfly (Lepidoptera: Hes-
periidae). We combine 25 years of natural history observations in
northwestern Costa Rica with morphological study and DNA bar-
coding of museum specimens to show that A. fulgerator is a
complex of at least 10 species in this region. Largely sympatric,
these taxa have mostly different caterpillar food plants, mostly
distinctive caterpillars, and somewhat different ecosystem prefer-
ences but only subtly differing adults with no genitalic divergence.
Our results add to the evidence that cryptic species are prevalent
in tropical regions, a critical issue in efforts to document global
species richness. They also illustrate the value of DNA barcoding,
especially when coupled with traditional taxonomic tools, in dis-
closing hidden diversity.
We are driven to find and describe our planet’s unrecog-
nized biodiversity because it is disappearing before our
eyes. Yet some of this uncharacterized biodiversity has been
staring us in the face, almost from the taxonomic start. Consider
the neotropical skipper butterf ly Astraptes fulgerator (Hesperi-
idae) (Fig. 1) described in 1775 (1). This butterfly has long been
regarded as a single species that is common, variable, and
wide-ranging: from the far southern United States to northern
Argentina, from the near desert to deep rain forest, from
lowlands to middle elevations, and from urban gardens to
pristine habitats. However, this view blocks perception of its real
complexity.
The rearing of ⬎2,500 wild-caught caterpillars of A. fulgerator
during 25 years of biodiversity inventory in the dry forest, rain
forest, and cloud forest of the Area de Conservacio´n Guanacaste
(ACG) in northwestern Costa Rica (http:兾兾janzen.sas.
upenn.edu; refs. 2–5) revealed a range of dicotyledonous food
plants far too broad for one species of pyrgine hesperiid (as
demonstrated by some 31,000 other ACG pyrgine rearing
records (http:兾兾janzen.sas.upenn.edu). Moreover, divergent
color patterns of the caterpillars (Fig. 2) segregated in accord
with food plants. Although dissections of 67 male and female
genitalia disclosed none of the morphological differentiation
that often distinguishes cryptic species of skippers (see, for
example, refs. 6–9), close study of adults, sorted by their
caterpillar food plant, showed subtle differences in color, pat-
tern, size, and wing shape. Synthesis of information on food plant
use, caterpillar color pattern, and adult external phenotypes
indicated that A. fulgerator from the ACG was a complex of at
least six or seven species. However, it seemed that several more
years of linking caterpillar and adult characteristics with food
plants would be needed to fully delimit species.
While this query was proceeding, it became apparent that
DNA sequencing of a standard gene region or ‘‘DNA barcoding’’
(10) might speed a solution. DNA barcoding can be helpful in
species diagnosis because sequence divergences are ordinarily
much lower among individuals of a species than between closely
related species (11–13). For example, congeneric species of
moths show an average sequence divergence of 6.5% in the
mitochondrial gene cytochrome coxidase I (COI), whereas
divergences among conspecific individuals average only 0.25%
(11). Similar values were obtained in birds, with intraspecific
divergences at COI averaging 0.27%, whereas congener diver-
gences averaged 7.93% (14).
In this study, the addition of DNA barcodes to data on food
plants, ecological distributions, caterpillar color patterns, and
adult facies indicates that A. fulgerator consists of 10 largely
sympatric species in the ACG. This result raises the prospect
that, over its huge neotropical range, A. fulgerator may comprise
many more hidden species. Imagine the biodiversity implications
of this result for other wide-ranging, common, and ‘‘somewhat
variable’’ species of neotropical animals and plants.
Materials and Methods
Field Biology. The ACG is a 110,000-hectare mosaic of many ages
of succession and old growth tropical dry forest, rain forest, and
cloud forest, as well as their various intergrades under conser-
vation in northwestern Costa Rica (www.acguanacaste.ac.cr and
http:兾兾janzen.sas.upenn.edu). Since 1978, tens of thousands of
caterpillars have been reared annually from thousands of species
of plants (e.g., refs. 2–5 and 15). Through 2003, these rearings
included 2,592 caterpillars of A. fulgerator. Each caterpillar was
reared individually, and its rearing data were collated under a
unique voucher code (e.g., 93-SRNP-3774), which is accessible
on the project web site (http:兾兾janzen.sas.upenn.edu).
Approximately half of these A. fulgerator caterpillars produced
adults; 968 were frozen on the day of eclosion, thawed within 2
months, pinned, spread, oven-dried, and stored at ambient
temperatures. These specimens were collected under multiple
research and export permits issued to D.H.J. by the Ministerio
del Ambiente y Energı´a of Costa Rica, and they have been
deposited in the National Museum of Natural History, where
they remain stored at ambient temperatures.
We attempted to assign DNA barcodes to 484 of these adults.
Where possible, those chosen included at least 20 individuals
reared from each species of food plant, extremes and interme-
diates of adult and caterpillar color variation, and representa-
tives from the three major ACG terrestrial ecosystems (dry
forest, cloud forest, and rain forest) and their intergrades. All 30
available individuals from wild-caught pupae were barcoded
even though their food plants are unknown (caterpillars of A.
fulgerator often pupate off their food plant). One leg was plucked
from each individual, placed in a dr y Eppendorf tube, and sent
to the University of Guelph for DNA analysis. Sampled adults
Freely available online through the PNAS open access option.
Abbreviations: ACG, Area de Conservacio´ n Guanacaste; COI, cytochrome coxidase I; NJ,
neighbor-joining.
Data deposition: The sequences reported in this paper have been deposited in the GenBank
database (accession nos. AY666597–AY667060, AY7224411, and AY7224412).
†To whom correspondence should be addressed. E-mail: phebert@uoguelph.ca.
© 2004 by The National Academy of Sciences of the USA
14812–14817
兩
PNAS
兩
October 12, 2004
兩
vol. 101
兩
no. 41 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0406166101

received yellow labels stating ‘‘Legs away兾for DNA.’’ Digital
photographs of the upper and lower side of each adult, together
with its collection details, are available on the Barcodes of Life
(BoLD) web site (www.barcodinglife.com) and on the inventory
web site, which also has images of hundreds of the adults and
their caterpillars.
Genetic Analysis. Total DNA was extracted from each of the 484
dry legs by using the GenElute Mammalian Genomic DNA
Miniprep kit (Sigma Genosys) according to the manufacturer’s
specifications, and the resultant DNA was eluted in 30
␮
lof
double-distilled H
2
0. Analysis ordinarily examined sequence
diversity in a specific 648-bp fragment of the mitochondrial COI
gene (the COI 5⬘region). This sequence was amplified by using
the following primer pair designed for Lepidoptera: LEP-F1,
5⬘-ATTCAACCAATCATAAAGATAT-3⬘; and LEP-R1, 5⬘-
TAAACTTCTGGATGTCCAAAAA-3⬘. When PCR amplifi-
cation with these primers failed to generate a product, the
LEP-F1 primer was combined with another reverse primer
(5⬘-CTTATATTATTTATTCGTGGGAAAGC-3⬘) to generate
a 350-bp product. This combination was necessary in ⬇5% of the
extractions; it was necessary more frequently with specimens
10–20 years old than with newer material.
All PCR mixes had a total volume of 50
␮
l and contained 2.5
mM MgCl
2
, 5 pmol of each primer, 20
␮
M dNTPs, 10 mM
Tris䡠HCl (pH 8.3), 50 mM KCl, 10–50 ng (1–5
␮
l) of genomic
DNA, and 1 unit of TaqDNA polymerase. The thermocycling
profile consisted of one cycle of 1 min at 94°C, six cycles of 1 min
at 94°C, 1 min and 30 sec at 45°C, and 1 min and 15 sec at 72°C,
followed by 36 cycles of 1 min at 94°C, 1 min and 30 sec at
51°C, and 1 min and 15 sec at 72°C, with a final step of 5 min at
72°C. PCR products were electrophoresed in 1.0% TBE agarose
gels, stained with ethidium bromide, and visualized under UV
light. Two microliters (20–50 ng) of the PCR products from these
reactions were cycle sequenced without further cleanup by using
the LEP-F1 primer, the ABI Prism TaqFS dye terminator kit
(Applied Biosystems), and BIG DYE (version 3.1). Sequencing
reactions had a total volume of 10
␮
l and included 10 pmol of
each primer. The sequencing amplification protocol consisted of
one cycle of 1 min at 96°C, followed by 30 cycles of 10 sec at 96°C,
5 sec at 55°C, and 4 min at 60°C. Sequences were analyzed on an
ABI 377 sequencer (Applied Biosystems) and were aligned
subsequently by eye in BIOEDIT (16). Sequence divergences
among individuals were quantified by using the Kimura-2-
Parameter distance model (17) and graphically displayed in a
neighbor-joining (NJ) tree (18). All sequences obtained in this
study and the original chromatograms are available in a com-
pleted project file (Astraptes fulgerator Complex) on the BoLD
web site. The sequences have also been deposited in GenBank
(accession nos. AY666597–AY667060, AY7224411, and
AY7224412).
Results
COI Divergences. A full-length PCR product was recovered from
465 of the 484 individuals (96%), and a 350-bp product was
recovered from 14 of the 19 remaining specimens. The COI
sequences were easily aligned, as no insertions or deletions were
detected. However, 13 sequences showed heterozygosity (as
evidenced by dual peaks of similar height in the electrophero-
grams) at 16–28 nucleotide sites, suggesting either heteroplasmy
or coamplification of a nuclear pseudogene with its mitochon-
drial counterpart. A second DNA extraction from these 13
individuals, followed by sequence analysis, confirmed their
heterozygosity. We discuss these individuals below, but we
excluded them from our initial analyses. The 137 different COI
sequences among the remaining individuals displayed consider-
able divergence (Appendix 1, which is published as supporting
information on the PNAS web site), with Kimura-2-Parameter
distances among individuals averaging 2.76% (range, 0.0 –
7.95%).
The 10 Taxa. Mapping caterpillar兾adult morphology and food
plants onto the NJ tree of COI divergences reveals 10 haplotype
clusters that covary with morphological and ecological traits
(Fig. 3), suggesting the presence of 10 species. Sequence diver-
gences for the 45 pairwise NJ comparisons among these 10 taxa
average 2.97% and range from 0.32% to 6.58% (Appendix 2,
which is published as supporting information on the PNAS web
site). To aid discussion, we code each species by key biological
attributes: 7 of the 10 taxa are coded according to their primary
food plants (TRIGO, CELT, LONCHO, LOH AMP, HIHAMP,
BYTTNER, and INGCUP), and the other three taxa are coded
by their main food plants plus a color character of the adult
(SENNOV, YESENN, and FABOV). Two small COI groups of
three (MYST) and four (NUMT) individuals are treated sepa-
rately for reasons that are justified later. In the remainder of this
section, we briefly describe key features (ecological, ethological,
and morphological) for each of the 10 presumptive taxa.
The yellow-ringed caterpillars of TRIGO eat the two species
of Trigonia (Trigoniaceae) in the ACG, whereas those of CELT
eat only Celtis iguanaea (Celtidaceae兾Ulmaceae). These food
plants are ignored by the remainder of the A. fulgerator complex
(and by other ACG hesperiids, as well). Conversely, TRIGO and
CELT do not use the food plants of the other eight members of
the complex. The lone record of TRIGO eating Licania arborea
(Chrysobalanaceae) is real but exceptional (the other ⱖ750
caterpillar records from this species of plant are of other species
of Lepidoptera). CELT and TRIGO are sympatric in the ACG
lowland rain forest (up to ⬇400 m), but only TRIGO extends
into the dry forest.
LOHAMP and LONCHO, which have similar yellow-disk-
marked caterpillars, are likewise faithful to their food plants,
although with some instructional exceptions. LOH AMP normally
eats Hampea appendiculata (Malvaceae), but 1 of 47 barcoded
individuals ate Lonchocarpus oliganthus (Fabaceae), and 2 individ-
ual ate Styrax argenteus (Styracaceae), a plant that was not other-
wise fed on by any member of the A. fulgerator complex. LONCHO
regularly eats L. oliganthus or L. costaricensis, but 9 of 41 barcoded
individuals used Senna (Fabaceae), and 2 used H. appendiculata.
Hence, LOHAMP and LONCHO can survive on each other’s
principal food plant, whereas LONCHO also rarely eats the pri-
mary food plants (Senna) of three other A. fulgerator species.
LONCHO is sympatric with five species in the complex at the lower
Fig. 1. Newly eclosed female A. fulgerator (species LOHAMP, voucher code
02-SRNP-9770) from the ACG.
Hebert et al. PNAS
兩
October 12, 2004
兩
vol. 101
兩
no. 41
兩
14813
ECOLOGY

margin of the ACG cloud forest but does not follow the many
species of Lonchocarpus into the rain forest or dry forest lowlands.
By contrast, LOHAMP follows H. appendiculata throughout its
highland to lowland range and so coexists w ith eight other members
of the A. fulgerator complex. The blue on the upperside of the wings
of fresh reared adults is perceptibly deeper and darker in LONCHO
than it is in LOHAMP.
DNA barcoding revealed HIHAMP in an unexpected manner.
The original HIHAMP group included just three adults reared
from caterpillars but 11 adults from wild-caught pupae. Each of
these pupae was found 1–2 m above the ground, under a different
tall adult H. appendiculata. One caterpillar was in mature foliage of
an adult Hampea crown, whereas both records ‘‘from’’ Capparis
frondosa (Capparidaceae) were likely prepupal caterpillars de-
scended from the Hampea overhead. A directed search in 2004
located three more HIHAMP caterpillars in the crowns of adult
Hampea, two of which survived to produce adults. Both had
sequences identical with those of the other HIHA MP. The cater-
pillars of HIHAMP are ringed (whereas those of LOHAMP bear
yellow discs), and they feed on mature foliage in the Hampea crown,
whereas those of LOHAMP feed on low, young foliage (usually on
saplings). HIHAMP appears to be a middle-elevation cloud-forest
species that coexists with five other members of the A. fulgerator
complex, whereas the partly sympatric LOH AMP ranges down into
the rain forest lowlands. However, HIHAMP is ecologically and
microgeographically parapatric with FABOV, which is closest to it
in the NJ tree, and feeds on a very different plant family than
FABOV.
The ringed caterpillars of BYTTNER resemble those of
several other A. fulgerator species. Their apparent monophagy on
Byttneria catalpaefolia (Sterculiaceae) is still tentative because
just four adults of this dry forest species were available for
analysis. Despite their rarity, these specimens were obtained
over a 15-year period, providing evidence for a persistent linkage
between this food plant and a particular COI lineage.
INGCUP is striking because it eats multiple species in two
Fig. 2. Last-instar caterpillars of 10 species in the A. fulgerator complex from the ACG. Interim names reflect the primary larval food plant and, in some cases,
a color character of the adult.
14814
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0406166101 Hebert et al.

genera, Inga (Fabaceae) and Cupania (Sapindaceae), in different
plant orders, in all three major ACG ecosystems. Only 3 of 66
barcoded INGCUP used other A. fulgerator food plants. Al-
though their color patterns on both food plant genera vary from
complete rings to dorsally broken rings to lateral patches that
range from intense yellow to deep orange, the barcode group is
extremely cohesive, with 95% having identical sequences.
Early on, reared adults whose caterpillars were mainly found on
the same two species of Senna (Fabaceae) were sorted by J.M.B.
into the following two groups by ventral body color: SENNOV with
an orange venter and YESENN with a yellow one. It was later
recognized that their caterpillars are, respectively, fully ringed and
laterally disked. Moreover, SENNOV is a dry forest species using
chiefly Senna hayesiana, whereas YESENN is a rain forest species
using chiefly Senna papillosa. Both species also occur in cloud forest
clearings and in the intergrade between dry and rain forests where,
again, their two main food plants are common, and an individual
plant may have both species on it. Although both eat a smattering
of other legumes, SENNOV appears to be the more polyphagous
because it occasionally eats Karwinskia calderoni (Rhamnaceae).
Like adult ventral coloration and caterpillar patterns, barcodes
clearly separate these two species.
The COI sequences also point to yet another species
(FABOV), whose adults were initially grouped by J.M.B. with
SENNOV because of their orange venters and their caterpillar
food plants. However, when FABOV was revealed, it was seen
that, on average, its ventral orange is slightly paler than that of
SENNOV. Although predominantly Senna eaters (21 of 33
cases), the ringed caterpillars of FABOV feed on at least seven
species in six other genera of Fabaceae. This taxon occurs in the
lowland ACG, mostly in dry forest and in the intergrade between
dry forest and rain forest (once recorded in deep rain forest), but
it is not common anywhere.
Individuals with ‘‘Heterozygous’’ COI Sequences. Adults with two
different COI sequences came from caterpillars collected on
Celtis (1), Cupania (1), Hampea (1), Inga (3), Lonchocarpus (1),
and Senna (6). Because of the strong associations between
caterpillar food plants and COI sequences in other individuals,
it was possible to ascertain the likely genotypic characteristics of
these ‘‘heterozygotes.’’ By assuming that one of the COI se-
quences in each individual matched the typical sequence for
other A. fulgerator found feeding on its food plant, the second
sequence could be determined by subtraction. For example, the
three heterozygous individuals collected on Inga were assumed
to possess the standard sequence found in individuals of ING-
CUP and a second sequence that differed from it at all het-
erozygous positions. This analytical approach revealed that all 13
COI heterozygotes possessed a second sequence with high
similarity. This congruence, despite recovery from varied mem-
bers of the A. fulgerator complex, supports a pseudogene origin
because nuclear pseudogenes of mitochondrial origin (also
known as Numts; see ref. 19) regularly show sequence conser-
vation among lineages with divergence in the corresponding
mitochondrial sequence (20). These presumptive pseudogene
sequences are closely similar or identical to the NUMT se-
quence, suggesting that the four individuals in this group rep-
resent cases in which only this pseudogene amplified.
Fig. 3. NJ tree based on Kimura-2-Parameter distances for COI DNA se-
quences from 466 individuals of the A. fulgerator complex from the ACG.
Numbers in parentheses indicate the total sample size for each interim taxon,
rectangles caricature caterpillar color patterns, and black backgrounds indi-
cate groups of ⱖ10 conspecifics with identical sequences.
Hebert et al. PNAS
兩
October 12, 2004
兩
vol. 101
兩
no. 41
兩
14815
ECOLOGY

Discussion
Despite the infancy of DNA barcoding protocols, our study
demonstrates that dry museum specimens up to 23 years old can
be sequenced with considerable success (⬇98%; but often only
350-bp sequences were obtained from older specimens). Al-
though preservation methods can damage DNA (21, 22), earlier
studies have recovered PCR products from insect specimens
over 1 century old (23, 24). Moreover, augmented PCR protocols
with an initial DNA repair step promise advances in DNA
recovery, suggesting that a comprehensive barcode library could
be assembled through the sequencing of museum specimens.
Numts pose a potential interpretational hazard for any PCR-
based survey of mitochondrial DNA diversity (25), and 2.8% of
our COI sequences showed probable coamplification of a Numt
with its mitochondrial counterpart. However, the taxonomic
impact of these coamplifications was small; all such individuals
were identified as belonging to the A. fulgerator complex, and
most individuals could be assigned to one of its 10-component
taxa when the pseudogene sequence was determined. We em-
phasize, as well, that when sequencing is done with fresh
specimens, the use of RT-PCR provides strong protection
against Numt amplification (26), suggesting the use of this
approach in taxa with COI pseudogenes.
Our sequencing results support the prior conclusion that A.
fulgerator is a species complex. Its levels of COI diversity are
much higher than those typical of single-species populations.
More importantly, there is clear covariation of the COI se-
quences with morphological, ethological, and ecological traits.
Although studies of classical characters indicated six or seven
species, the addition of COI data raised the count to 10. This
increase reflects the way in which the COI data added meaning
to isolated records on unusual food plants (e.g., Byttneria), to
lineages with behavioral and caterpillar color pattern divergence
(e.g., LOHAMP vs. HIHAMP), and to taxa with nearly indis-
tinguishable adult facies (e.g., SENNOV vs. FABOV). Whereas
our study reveals the power of DNA barcoding in helping to
resolve complex taxonomic situations, it also indicates the im-
perative of large sample sizes and supplemental morphology and
natural history. We emphasize that barcodes differ from the
standard traits used for species discrimination in the following
important way: they can be obtained in a mechanized manner.
Hence, they can be used without much background knowledge,
both for routine identifications and for the detection of hidden
species (13, 27).
Despite the variation in caterpillar color pattern and food
plants, adults in the A. fulgerator complex show little phenotypic
diversity. Their similarity probably ref lects not only recent
common ancestry but also stabilizing selection arising from
membership in a massive mimicry ring. At least 35 species in
three subfamilies of hesperiids from the ACG range from
general to exact mimics of A. fulgerator (see images of adults at
http:兾兾janzen.sas.upenn.edu). In the neotropics as a whole,
many more skipper species swell this mimetic assemblage. As
expected from their morphological similarity, members of the A.
fulgerator complex show less sequence divergence than most
other congeneric species pairs (10, 13). Only two taxa (CELT
and TRIGO) have minimum COI divergences (3.4% and 5.4%)
from all other members of the complex that exceed the sequence
threshold (3%) typically encountered between congeneric spe-
cies pairs recognized by morphological approaches (14). Diver-
gences among the remaining taxa are lower, but all exceed 1.1%,
which is well above usual intraspecific values, except among
members of the triad FABOV, HIHAMP, and INGCUP, which
show ⬍0.5% divergence. These three species possess distinct
COI sequence arrays (Fig. 3), feed on very different food plants,
and show subtle differences in both caterpillar and adult facies.
The MYST lineage apparently represents an exception to this
pattern of mitochondrial divergence as it includes one individual
from Inga and two from Senna, suggesting that it is shared by
INGCUP and FABOV. As such, this likely represents a case in
which lineage sorting is incomplete, a result that might have been
expected given the low genetic divergence (and presumed recent
origin) of these groups.
Past work has provided conf licting perspectives on the likely
efficacy of mtDNA markers in delineating species boundaries.
Some studies, including extensive analyses of GenBank data,
have indicated that even closely related species ordinarily show
marked mitochondrial divergence (10, 12). However, others
suggest that mtDNA markers will often encounter problems in
species resolution (28–30). For example, a review of case studies
(31) concluded that nearly one-fourth of all animal species fail
the test of mitochondrial monophyly. The A. fulgerator complex
represents a case in which mitochondrial markers might have
been expected to fail because its component species are both
extremely similar and sympatric, providing opportunities for
hybridization. However, our detection of reciprocal monophyly
for COI variants among its members means that shared ancestral
polymorphisms have been lost, either as a consequence of
stochastic lineage pruning or selective sweeps. Moreover, the
lack of shared haplotypes indicates either strict reproductive
isolation or ongoing selection against mitochondrial exchange
between members of the complex. Female Lepidoptera are both
the heterogametic sex and the primary agents of food plant
selection, which are factors that can also act to ensure rapid
divergence in mitochondrial markers (32, 33).
Should the 10 species of A. fulgerator identified in this study be
formally described despite their morphological similarity? Yes.
Although their recognition was facilitated by DNA barcoding,
the combination of their genetic distinctiveness and their cova-
rying caterpillar color patterns, food plant usage, and adult
morphology demonstrates that they are reproductively isolated
populations. The fact that these populations are largely sympa-
tric argues even more strongly for traditional binomials.
Diversification in the A. fulgerator complex is clearly linked to
food plants, suggesting the importance of a detailed analysis of
shifting food plant use. The complex likely derives from a species
that fed on Fabaceae because most other species of Astraptes and
many of those in 19 allied genera (34) feed largely on plants in
this family (2). Based on standard calibrations for rates of
mitochondrial evolution (e.g., refs. 35 and 36), TRIGO sepa-
rated from the basal fulgerator clade ⬇4 million years ago,
probably onto Trigonia, whereas CELT separated ⬇2 million
years ago, likely onto Celtis. Subsequent speciation has involved
less radical food plant shifts, but some of these events probably
occurred within the last half-million years (e.g., FA BOV, HI-
HAMP, and INGCUP). This diversification may be linked to
regional variation in the species composition of plant commu-
nities and to the inclusion of novel species in the food plant
repertoire of some taxa. For example, only SENNOV eats K.
calderoni, a species that is chemically and morphologically
divergent from its standard food plants. If isolated in a setting
lacking its usual food plants, SENNOV might rapidly evolve into
aKarwinskia specialist.
Although members of the A. fulgerator complex are young in
evolutionary terms, their ranges have surely expanded well
beyond their areas of origin. As a result, we doubt that any of the
lineages revealed in this study are endemic to the ACG or to
Costa Rica. Rather, we expect that most of their distributions
span many degrees of latitude, extending wherever food plants
and ecological conditions permit (37). Given the high diversity
of the South American hesperiid fauna, and the fact that A.
fulgerator ranges from the southern United States to northern
Argentina, a comprehensive survey might uncover many more
species.
14816
兩
www.pnas.org兾cgi兾doi兾10.1073兾pnas.0406166101 Hebert et al.

This study has altered our view of a ‘‘species’’ that has been
known to science for more than 2 centuries. Its transformation
from a single, common, variable, and wide-ranging taxon to a
complex of ⱖ10 food plant specialists with differing ecological
attributes reveals a layer of biological complexity that needs
exploration. How often are widespread ‘‘species,’’ such as A.
fulgerator, really an amalgam of specialized, reproductively
isolated lineages? The answer to this question is crucial to
refining estimates of species diversity in the animal kingdom
because the level of food plant specialization in tropical
arthropods is a critical modulator of these values (e.g., refs. 38
and 39). Although few DNA-based studies have examined this
issue, our results contribute to an emerging pattern. The
number of species in a neotropical cerambycid beetle genus
doubled after an analysis of only 48 individuals from a small
geographic region (40). Similar evidence for cryptic species
was found in varied tropical pseudoscorpion lineages (41).
Collectively, these results may motivate a broad-ranging eval-
uation of the incidence of cryptic species in the tropics, an
effort that could be expedited by use of an efficient screening
procedure like DNA barcoding.
We thank A. Holliss for assistance with sequencing, I. Smith for
illustrations, S. Ratnasingham for aid with data analysis, the 17 para-
taxonomists in the ACG for collecting, rearing, and databasing the
caterpillars, D. Harvey and E. Klafter for dissecting genitalia, and
M. Stoeckle, S. Miller, N. Pierce, and C. Meyer for comments on the
manuscript. This research was supported by grants from the Natural
Sciences and Engineering Research Council (Canada) and the Canada
Research Chairs program (to P.D.N.H.) and by National Science Foun-
dation Grants 8307887, 8610149, 9024700, 9306296, 9400829, 9705072,
and 0072730 (to D.H.J.).
1. Walch, J. E. I. (1775) Naturforscher 7, 113–116, plate 1.
2. Janzen, D. H. (2003) in Arthropods of Tropical Forests. Spatio-temporal Dy-
namics and Resource Use in the Canopy, eds. Basset, Y., Novotny, V., Miller,
S. E. & Kitching, R. L. (Cambridge Univ. Press, Cambridge, U.K.), pp.
369–379.
3. Gauld, I. D. & Janzen, D. H. (2004) Zoo.J. Linn. Soc. 141, 297–351.
4. Janzen, D. H. (2004) in Biodiversity Conservation in Costa Rica, eds. Frankie,
G. W., Mata, A. & Vinson, S. B. (Univ. of California Press, Berkeley), pp.
80–96.
5. Burns, J. M. & Janzen, D. H. (2001) J. Lepid. Soc. 55, 15–43.
6. Burns, J. M. (1974) Psyche 81, 51–69.
7. Burns, J. M. (1984) Smithsonian Contrib. Zool. 405, 1–38.
8. Burns, J. M. (1994) J. Lepid. Soc. 48, 273–337.
9. Burns, J. M. (1996) J. Lepid. Soc. 50, 173–216.
10. Hebert, P. D. N., Cyw inska, A., Ball, S. L. & deWaard, J. R. (2003) Proc. R.
Soc. London Ser. B 270, 313–321.
11. Moore, W. S. (1995) Evolution (Lawrence, Kans.) 49, 718–729.
12. Av ise, J. C. & Walker, D. (1999) Proc. Natl. Acad. Sci . USA 96, 992–995.
13. Hebert, P. D. N., Ratnasingham, S. & deWaard, J. R. (2003) Proc. R. Soc.
London Ser. B 270, S596–S599.
14. Hebert, P. D. N., Stoeckle, M. Y., Zemlak, T. S. & Francis, C. M. (2004) PLoS
Biol., in press.
15. Janzen, D. H., Walker, A. K., Whitfield, J. B., Delvare, G. & Gauld, I. D. (2003)
J. Hymenopt. Res. 12, 42–76.
16. Hall, T. (1999) Nucleic Acids Symp. Ser. 41, 95–98.
17. Kimura, M. (1980) J. Mol. Evol. 16, 111–120.
18. Saitou, N. & Nei, M. (1987) Mol. Biol. Evol. 4, 406 –425.
19. Lopez, J. V., Yukhi, N., Masuda, R., Modi, W. & O’Brien, S. J. (1994) J. Mol.
Evol. 39, 174–190.
20. Perna, N. T. & Kocher, T. D. (1996) Cur r. Biol. 6, 128 –129.
21. Dean, M. D. & Ballard, J. W. (2001) Entomol. Exp. Appl. 98, 279–285.
22. Dillon, N., Austin, A. D. & Bartowsky, E. (1996) Insect Mol. Biol. 5, 4–25.
23. Townson, H., Harbach, R. E. & Callan, T. A. (1999) Syst. Entomol. 24, 95–101.
24. Goldstein, P. Z. & DeSalle, R. (2003) Mol. Ecol. 12, 1993–1999.
25. Bensasson, D., Zhang, D., Hartl, D. L. & Hewitt, G. M. (2001) Trends Ecol.
Evol. 16, 314–321.
26. Collura, R. V., Auerbach, M. R. & Stewart, C.-B. (1996) Curr. Biol. 6,
1337–1339.
27. Janzen, D. H. (2004) Philos. Trans. R. Soc. London B 359, 731–732.
28. Lipscomb, D., Platnick, N. & Wheeler, Q. (2003) Trends Ecol. Evol. 18, 65–66.
29. Mallet, J. & Willmott, K. (2003) Trends Ecol. Evol. 18, 57–59.
30. Will, K. & Rubinoff, D. (2004) Cladistics 20, 47–55.
31. Funk, D. J. & Omland, K. E. (2003) Annu. Rev. Ecol. Evol. Syst. 34, 397–423.
32. Janz, N. (2003) in Butterflies: Ecology and Evolution Taking Flight, eds. Boggs,
C. L., Watt, W. B. & Ehrlich, P. R. (Univ. of Chicago Press, Chicago), pp.
229–240.
33. Sperling, F. (2003) in Butterflies: Ecology and Evolution Taking Flight, eds.
Boggs, C. L., Watt, W. B. & Ehrlich, P. R. (Univ. of Chicago Press, Chicago),
pp. 431–458.
34. Evans, W. H. (1952) A Catalogue of the American Hesperiidae Indicating the
Classification and Nomenclature Adopted in the British Museum (Natural
History). Part II. Pyrginae. Section I (British Museum, London), plates 10–25.
35. Fleischer, R. C., McIntosh, C. E. & Tarr, C. L. (1998) Mol . Ecol. 7, 533–542.
36. Knowlton, N. & Weigt, L. A. (1998) Proc. R. Soc. London Ser. B 265, 2257–2264.
37. Janzen, D. H. (1985) Oikos 45, 308 –310.
38. Odegaard, F. (2000) Biol. J. Linn. Soc. 71, 583–597.
39. Novotny, V., Basset, Y., Miller, S. E., Weiblen, G. D., Bremer, B., Cizek, L. &
Drozd, P. (2002) Nature 416, 841–844.
40. Berkov, A. (2002) Biol. J. Linn. Soc. 76, 195–209.
41. Wilcox, T. P., Hugg, L., Zeh, J. A. & Zeh, D. W. (1997) Mol. Phylogenet. Evol.
7, 208–216.
Hebert et al. PNAS
兩
October 12, 2004
兩
vol. 101
兩
no. 41
兩
14817
ECOLOGY