A Phylogenetic Test of Classical Species Groups in Argia (Odonata: Coenagrionidae)

A PHYLOGENETIC TEST OF CLASSICAL SPECIES GROUPS IN ARGIA(ODONATA: COENAGRIONIDAE)

RYAN M. CAESAR1,2

AND JOHN W. WENZEL1

1Department of Entomology, Ohio State University, Columbus, OH

Abstract.—We present the first cladistic analysis of Argia species, focusing on those occurring in North America

north of Mexico. Our analysis is based on mitochondrial 16S rDNA and morphological characters of both sexes of

adults and immatures. We reexamine classical work on Argia taxonomy and phylogeny. Our results agree

considerably with previous hypotheses based morphology in an absence of phylogenetic analysis, and thus our work

represents and independent test of these previous hypotheses. Argia is recovered as monophyletic. The clade

composed of A. funcki plus A. lugens is basal among the species studied here. The species A. fumipennis, including the

three subspecies, appears to be a paraphyletic assemblage, and thus may warrant being considered separate species as

originally described. The feasibility of producing a thorough phylogenetic analysis of the entire genus using multiple

sources of data is discussed.

Key words: Argia, Coenagrionidae, Odonata, damselfly, phylogeny.

INTRODUCTION

Rambur described the genus Argia (Odonata:

Zygoptera: Coenagrionidae: Argiinae) in 1842.

Argia is extremely speciose in comparison to other

odonate genera, with 118 valid described species

(Garrison, 1994; Garrison and von Ellenrieder,

2007). The genus occurs throughout the New

World, with its greatest diversity occurring in the

neotropical region. There are approximately thirty

six species that can be found in North America

north of Mexico, with the highest diversity in the

southwestern United States; twenty three species

are known to occur in southeastern Arizona

alone. The remaining species occur in subtropical,

tropical and temperate regions of Meso- and

South America. Populations of common Argia

species can be quite large, and, like all odonates,

they are voracious predators in all life stages. As

such, they represent vital components of the

trophic webs of aquatic ecosystems.

Most species of Argia prefer low to mid order

streams (Westfall and May, 1996), unlike the

remainder of coenagrionids that tend to occur in

lentic systems (Dunkle, 1990). Several species are

rare, endemic, or threatened, and may be of

conservation concern. For example, in Ohio, A.

bipunctulata (Hagen) is listed as an endangered

species, due to its restriction in the state mainly to

the 182 hectare Cedar Bog Preserve and a few

other fen habitats (Moody, 2002), although the

species is fairly common in adjacent states. The

recently described species A. sabino Garrison is

known from very few localities in Arizona, most

of which are continually threatened by forest fire,

and thus it is considered a species of concern by

biologists in that state (D. Turner, pers. comm.)

Some of the more common species have been

thoroughly studied from a behavioral and ecolog-

ical perspective (Borror, 1934; Bick and Bick,

1965, 1971, 1982; Robinson et al., 1983; Conrad,

1992; Conrad and Pritchard, 1988, 1990). Little is

known about the biology of most of the tropical

species, and there has never been a phylogenetic

study of the genus.

The current taxonomy of Argia species has been

established in the absence of an explicit phyloge-

netic hypothesis based on modern comparative

methods. In his work on Central American

Odonata, Calvert treated forty eight species of

Argia (Calvert, 1901), eighteen of which occur in

the United States (Garrison, 1994). Kennedy

included some Argia species in his ‘‘phylogeny’’

of Zygoptera (Kennedy, 1920a), and he described

several species (Kennedy, 1918, 1919). Leonora

Gloyd did considerable taxonomic work on Argia

throughout her life (Gloyd, 1958, 1968a, 1968b),

although she died before much of her work was

completed (Garrison, 1994). The larvae of some

Mexican species were treated by Novelo-Gutierrez

(1992). Garrison (1994) provided a thorough

2 E-mail address for correspondence: caesar.6@

osu.edu

Entomologica Americana 115(2):97–108, 2009

synopsis of the species of Argia occurring north of

Mexico, including taxonomic keys for adults

(these keys are reproduced in Westfall and May

1996), and several informal species groups were

outlined. Forster (2001) provides updated taxo-

nomic keys for some of the more common Central

American species of Argia. Garrison is currently

continuing the work of Leonora Gloyd on

revising the tropical species (pers. comm.). Many

undescribed species are thought to exist, and new

species descriptions continue to be published

(Daigle, 1991, 1995; Garrison, 1994, 1996; Garri-

son and von Ellenrieder, 2007). Here we provide

the first modern phylogenetic hypothesis for Argia

species based on multiple data sources and using

two phylogenetic optimality criteria. These pre-

liminary phylogenetic analyses allows us to test

existing taxonomic hypotheses, contribute an

improved understanding of species relationships,

and provide a foundation for resolving the

phylogeny of the genus.

MATERIALS AND METHODS

Taxon sampling

Our matrix includes thirty eight of the 118

described Argia species, including nearly all of

those found north of Mexico as well as several

from Mesoamerica. A. bipunctulata, A. barretti

Calvert and A. carlcooki Daigle are the only

species currently known to occur north of Mexico

that are missing from our matrix. Collection

records for specimens used in this study are listed

in Table 1, along with GenBank accession num-

bers. Where possible, we include in this study

adult specimens that were recently collected in the

field using a hand held aerial net and deposited

directly into 95% ethyl alcohol for preservation.

Additional specimens were donated by other

collectors utilizing various collection and preser-

vation techniques, or were borrowed from the

International Odonata Research Institute (IORI;

Gainesville, Florida). Additionally, we did not

attempt to collect molecular data from some

species for which few or single specimens were

available.

Species were identified on the basis of published

keys based on morphological characters (Garri-

son, 1994; Westfall and May, 1996; Abbott, 2005.)

A. sp. nov. is an undescribed species collected in

northern Mexico; it has been known for several

years but remains undescribed. The sister group of

Argia is not known; we include six outgroup taxa

representing other damselfly lineages. 16S sequenc-

es for these outgroup species were taken from

GenBank (Table 1.) Specimens used in this study

are deposited as vouchers at either the IORI or the

Charles Triplehorn Insect Collection at the Ohio

State University (OSUC) in Columbus, Ohio.

Character analysis

We coded ten morphological characters from

imagos of both sexes as well as larvae for 35 Argia

species. Characters of the head, thorax and

abdomen, including secondary sexual characters

of both sexes, are represented. Characters were

initially chosen based on information in published

dichotomous keys. These characters are unambig-

uous in their interpretation and do not seem to

vary within species. The morphological matrix is

presented in Table 2.

Character 1. Thorax and head with metallic

copper-red coloration (imagos): absent (0), pre-

sent (1). This condition is also associated with

copper-red to red eyes for specimens in vivo

(imagos.) Species with this coloration are often

referred to as being associated together in the

‘‘metallics’’ group. While pigmentation is gener-

ally not a very useful character for damselflies, as

it is often variable within species, this condition is

largely a result of structural coloration and is

invariant within species.

Character 2. Mesepisternal tubercles (females):

absent (0), reduced (1), prominent (2). See Fig. 1.

Character 3. Posterior lobes of mesostigmal

plates (females): absent (0), broad and flange-like

(1), elongate and finger-like (2). See Fig. 1.

Character 4. Mesothoracic pits (females): absent

(0), shallow (1), deep (2). This character has been

discussed in very limited context in the literature,

but it has been suggested that it might be

informative for phylogenetics (Gloyd, 1958). Here

we code it for the first time and show that it is useful.

Character 5. Hairs lining mesothoracic pits

(females): absent (0), sparse (1), dense (2).

Character 6. Mesepisternal pits costate (fe-

males): absent (0), present (1).

Character 7. Pronotal pits (females): absent (0),

shallow (1), deep (2). This character is not utilized

in keys or discussed much in the literature, but

seems to be variable enough within Argia to be

useful. Indeed, this region corresponds to the

placement of the dorso-posterior potion of the

98 ENTOMOLOGICA AMERICANA Vol. 115(2)

male paraprocts during tandem linkage and

copulation (see Fig. 1.)

Character 8. Shape of cerci (males): entire (0),

bifid (1), trifid (2). Cerci are a secondary sexual

character, utilized as part of the clasping process

during copulation. The cerci contact the female

mesostigmal plates and mesepisternal tubercules

when linked in copula and may be part of the

Table 1. Taxonomic information, collection records and GenBank accession numbers for Argia and outgroup

species used in phylogenetic analyses.

Taxon Authority Sampling locality Collector

GenBank

accession

number

Calopteryx aequibilis Say, 1839 unknown GenBank AF170961

C. maculata (Beauvois, 1805) unknown GenBank AF170960

Hetaerina americana (Fabricius, 1798) unknown GenBank AF170951

Neoneura esthera Williamson, 1917 unknown GenBank AF170948

Ceriagrion nipponicum Asahina, 1967 unknown GenBank AB127067

Argia agrioides Calvert, 1895 USA: Oregon R. Caesar FJ592218

A. alberta Kennedy, 1918 USA: New Mexico J. Abbott FJ592211

A. anceps Garrison, 1996 Mexico: Hidalgo K. Tennessen FJ592233

A. apicalis (Say, 1839) USA: Ohio R. Caesar FJ592212

A. cuprea (Hagen, 1861) USA: Texas T. Gallucci FJ592227

A. emma Kennedy, 1915 USA: California C. Barrett FJ592228

A. extranea (Hagen, 1861) Mexico: Sonora R. Behrstock FJ592231

A. fumipennis atra (Burmeister, 1839) USA: Florida K. Holt FJ592230

A. fumipennis violacea (Burmeister, 1839) USA: Ohio R. Caesar FJ592232

A. funcki (Selys, 1854) Mexico: Sonora D. Paulson FJ592197

A. garrisoni Daigle, 1991 Mexico: Tamaulipas R. Behrstock FJ592213

A. harknessi Calvert, 1899 Mexico: Sonora D. Paulson FJ592199

A. hinei Kennedy, 1918 USA: Texas J. Abbott FJ592207

A. immunda (Hagen, 1861) USA: Texas A. Cognato FJ592214

A. lacrimans (Hagen, 1861) USA: Arizona R. Behrstock FJ592216

A. leonorae Garrison, 1994 USA: Texas R. Behrstock FJ592226

A. lugens (Hagen, 1861) USA: Arizona R. Caesar FJ592215

A. moesta (Hagen, 1861) USA: Texas A. Cognato FJ592229

A. munda Calvert, 1902 USA: Arizona R. Caesar FJ592223

A. nahuana Calvert, 1902 USA: Texas T. Gallucci FJ592225

A. sp. nov. Not applicable Mexico: Sonora D. Paulson FJ592198

A. oculata Hagen in Selys, 1865 Mexico: Tamaulipas R. Behrstock FJ592221

A. oenea Hagen in Selys, 1865 Mexico: San Luis Potosi R. Behrstock FJ592217

A. pallens Calvert, 1902 USA: Arizona W. Mauffray FJ592224

A. pima Garrison, 1994 USA: Arizona J. Daigle FJ592208

A. plana Calvert, 1902 USA: Texas J. Abbott FJ592196

A. pulla Hagen in Selys, 1865 Nicaragua: Jinotega J. Abbott FJ592222

A. rhoadsi Calvert, 1902 Mexico: San Luis Potosi R. Behrstock FJ592206

A. sabino Garrison, 1994 USA: Arizona J. Daigle FJ592202

A. sedula (Hagen, 1861) USA: Oklahoma H. Song FJ592209

A. tarascana Calvert, 1902 USA: Arizona R. Caesar FJ592200

A. tezpi Calvert, 1902 Honduras: Francisco

Morazan

S. Dunkle FJ592220

A. tibialis (Rambur, 1842) USA: Ohio R. Caesar FJ592203

A. tonto Calvert, 1902 USA: Arizona R. Caesar FJ592204

A. translata Hagen in Selys, 1865 USA: Texas A. Cognato FJ592210

A. ulmeca Calvert, 1907 Mexico: Tamaulipas R. Behrstock FJ592205

A. vivida Hagen in Selys, 1865 USA: California J. Abbott FJ592201

A. westfalli Garrison, 1996 Mexico: Tamaulipas R. Behrstock FJ592219

2009 ARGIA PHYLOGENY 99

mechanism by which females recognize and

evaluate males.

Character 9. Shape of paraprocts (males): entire

(0), bifid (1), trifid (2).

Character 10. Lateral gills with marginal fringe

of stout setae (larvae): present (0), absent (1).

Genomic DNA was extracted from specimens

that were either freshly collected and preserved in

95% ethyl alcohol, or older museum specimens

that were acetone-dried (up to 20 years old). A

modification of the animal tissue protocol for

Quiagen DNeasy extraction kits (Quiagen, Valen-

cia, CA, USA; Caesar et al., 2005) was used to

extract DNA from leg and thoracic muscle tissue

that was isolated from specimens using sterile

techniques. Dried specimens from which DNA

was extracted were labeled as DNA vouchers.

DNA template vouchers are stored at 280uC in

the Wenzel Laboratory at the Museum of

Biological Diversity, Ohio State University

(MBD.)

16S ribosomal DNA (0.6 kb) from the mito-

chondria was amplified by the polymerase chain

reaction (PCR) run at 35 cycles with annealing

temperature of 50uC. We used the primers LR-J-

12887 (59 CCGGTCTGAACTCAGATCACGT

39) and LR-N-13398 (59 CGCCTGTTTAA-

CAAAAACAT 39) (Simon et al., 1994), known

to be useful in several odonate studies (e.g., Misof

et al., 2000). Reactions were carried out in 25 ml

volumes, with 12.5 ml of a TAQ master mix

(Quiagen, Inc. Valencia, CA, USA), 3.5 ml of

H2O, 2.0 ml of each primer (0.5 pmol/ml), and

5.0 ml genomic DNA template. A layer of mineral

oil was applied to each sample, and reactions

Table 2. Morphological character matrix used for

parsimony analysis of Argia. See Methods for coding.

Taxon Characters (1–10)

Het.ameri ??????????A.plana1 0111202?11A.extrane 0101?01111A.vivida1 0211100011A.emma1 0222111111A.munda1 00???1?010A.anceps2 0110000000A.westfal 011120000?A.sabino1 011110211?A.lacrima 0?10001110A.tonto1 0010001110A.pima1 001??0?11?A.f.atra1 0121201211A.f.viola 0121201211A.pallens 0010001211A.funcki1 00212?110?A.lugens1 0122102100A.moesta1 0222201100A.hinei1 0020002211A.agrioid 0120001011A.alberta 0000000010A.leonora 000??0?01?A.nahuana 0110001011A.pulla1 0000001121A.rhoadsi 0022101001A.sedula1 0010102011A.tarasca 0221101110A.apicali 0100001210A.tezpi 0221101010A.tibiali 0000000110A.immunda 0001001111A.transla 0222102010A.ulmeca1 0101002110A.n.sp.1 0??????11?A.garriso 002000111?A.harknes 0?????????A.oculata 001100201?A.cuprea1 1221102100A.oenea1 1021102110Cal.aequa ??????????Cal.macul ??????????Ceri.nipp ??????????Neo.esthe ??????????

Fig. 1. Photo of the pro- and mesothorax of a

female of Argia emma, lateral oblique angle. Arrows

indicate some of the characters coded in the

morphological matrix.


occurred on a PTC-100 Peltier thermal cycler (MJ

Research, Ramsey, Minnesota, USA). Amplified

DNA was separated in an agarose gel via

electrophoresis and visualized under UV light

for verification of presence of amplicon and to

check for contamination. PCR products were

purified using QIAquick PCR purification kit

(Quiagen, Valencia, CA, USA), and purified

DNA was sequenced at the Plant Microbe

Genomics Facility (Ohio State University, Co-

lumbus, OH) or at Cogenics (Houston, TX).

Editing and assembly of raw sequences was

performed using Sequencher 4.1 (Gene Codes

Corp., Ann Arbor, Michigan, USA). We se-

quenced PCR products in both directions, and

and differences or ambiguities between strands

were resolved by eye after examining the elecro-

pherograms of each sequence. Edited sequences

were initially aligned by eye in Sequencher 4.1,

then submitted to CLUSTALW (Thompson et al.,

1994) for further alignment using a 1:1 gap

opening: extension cost. After editing and align-

ment, we utilized 551 base pairs (bp) as our

molecular data matrix. All new sequences gener-

ated in this study were deposited in GenBank

(Table 1).

Phylogenetic analysis

Calopteryx was designated as outgroup for all

phylogenetic analyses. Parsimony analysis was

performed on the molecular data only, and in

combination with the morphological data. We

used the parsimony ratchet implemented in

NONA/WinClada (Nixon, 2000). By default, this

implements TBR branch swapping. Gaps in the

molecular matrix were treated as missing data.

Bremer support values were generated in NONA

using the command ‘‘.hold 15,000; bsupport 5;’’

to estimate relative clade support. This command

generates as many as 15,000 trees that are up to

five steps longer than the most parsimonious

trees.

Concordance among trees produced by differ-

ent methods of phylogenetic reconstructions is

considered by some to be a good test of

phylogenetic accuracy (Huelsenbeck, 1997). Thus,

we also performed Maximum Likelihood (ML)

analysis on the molecular data. We did this using

the program GARLI, version 0.96 (Zwickl, 2006),

which uses a genetic algorithm to rapidly search

for the nucleotide substitution model parameters,

branch lengths, and topology that maximize the

log likelihood score. Default parameters were

used for 100 search replicates.

RESULTS

In some cases, DNA extraction and/or PCR

amplification of the target sequence failed for

older museum specimens that were otherwise

available for study. Thus, our analyzed matrices

do not include all available species. The ClustalW-

aligned molecular matrix contains 126 informative

sites and yields 38 most parsimonious trees (489

steps, CI 5 0.54, RI 5 0.61) in the MP analysis.

The strict consensus (499 steps) of these is shown

in Fig. 2. A monophyletic Argia is recovered with

good support, but a basal polytomy with several

weakly supported clades follows. The ML analysis

of the molecular data produced two trees that

differ only in the resolution of the clade contain-

ing the three species A. apicalis (Say), A. tarascana

Calvert, and A. tezpi Calvert. The ML tree with

the best likelihood score is shown in Fig. 3.

Analysis of the morphological matrix (Table 2)

alone yields trees (not shown) that are poorly

resolved and not informative. Combining the

morphological and molecular characters, we

obtained six most parsimonious trees (573 steps,

CI 5 0.49, RI 5 0.58). The consensus (578 steps)

is shown in Fig. 4. These morphological data in

the combined analysis improved resolution in

comparison with the molecular data alone, with

few changes in topology. Again Argia is recovered

as monophyletic, although the Bremer support

value improves from three to five. The basal

polytomy resulting from the molecular only data

is resolved, and we recover the clade of A. funcki

(Selys) + A. lugens (Hagen) as basal to the rest of

the genus. In the combined tree, A. moesta

(Hagen) is more basal, closer to A. translata

Hagen; A. munda Calvert falls out of the clade to

which it belonged, sister to the component

including A. pima Garrison and A. rhoadsi

Calvert. Some taxa are derived more apically in

the combined tree compared to the molecular tree,

such as A. oculata Hagen, which is no longer sister

to A. oenea Hagen in the basal paraphyly at the

ingroup node of Fig. 2. The polytomy in the

middle of the molecular tree (the component

including A. alberta Kennedy, A. nahuana Calvert,

A. leonorae Garrison, to A. anceps Garrison) has

better resolution, and the A. pima-A. westfalli

Garrison component is now sister to the remain-


Fig. 2. Strict consensus of 38 equally parsimonious trees resulting from analysis of 16S data alone. Clades with

dots are recovered in both ML and parsimony analyses. Numbers below branches are Bremer support values.


der, with the A. emma Kennedy-A. extranea

(Hagen) group coming out next.

DISCUSSION

All but three clades from the MP consensus

from the combined matrix are found in the ML

trees, as indicated by the dots on the shared nodes

of the MP trees (Figs. 2, 4), and the ML trees

(Fig. 3), demonstrating that the solution to our

16S data is generally stable, despite optimality

criterion. The main difference is that the MP tree

places A. funcki and A. lugens rather basally, and

the A. rhoadsi and A. sedula (Hagen) group rather

apically, whereas the ML tree places these lineages

in the middle of the tree. This rather modest

difference has the undesirable effect of compro-

mising the monophyly of most groups along the

spine of the tree even if the great majority of

fundamental relationships remain the same.

Monophyly of Argia is strongly supported by

both data partitions and in all phylogenetic

analyses conducted for this study (Figs. 2, 3, and

4). This is not surprising, and there are additional

morphological features that are autapomorphic

for Argia relative to other Coenagrionidae that

support monophyly of Argia. These include the

long length of tibial spines (Westfall and May,

1996) and absence of an angulate frons (Carle et

al., 2008). O’Grady and May also recovered Argia

as monophyletic in their morphological analysis

of Coenagrionidae (2003), although only three

species were included. We think that this result

will be stable to the addition of data in future

analyses.

The solutions to the molecular-only (Fig. 2) or

combined (Fig. 4) matrices appear very different,

and a strict consensus of these trees provides little

resolution. Strict consensus trees are usually used

to illustrate groups that are monophyletic for all

solutions, but it is not very useful when there is a

great degree of agreement for certain networks of

taxa among otherwise competing solutions. Prob-

lematic taxa and rooting issues that interrupt

networks are often not disputed among very

different solutions. While we are always interested

in recovering monophyletic of lineages, especially

for taxonomic work, analysis of ‘‘species groups’’

should also focus on stable networks of terminals,

even if their placement is ambiguous. We offer

Fig. 5 to demonstrate that the solutions of Figs. 2

and 4 are easily interpretable to be closely related,

despite a poorly resolved consensus. Our discus-

sion is based on three agreement subtrees, each of

which is common to all solutions.

We concentrate on the clade that is sister to A.

oculata, where the appearance of discord is most

evident. Excluded from discussion are A. munda

and the pair A. agrioides Calvert and A. hinei

Kennedy, because they are not informative for

demonstrating agreement among trees. The first

comparison is the sedula-pulla-rhoadsi component

Fig. 3. Best maximum likelihood tree generated by analysis of the 16S data using GARLI.


(Fig. 5A). This component can be placed either

basally near A. oculata (Fig. 2), or apically in the

clade of interest (Fig. 4). In both cases the root of

this component is between A. sedula and the pair

A. pulla Hagen in Selys plus A. rhoadsi. Next we

discuss the component that includes the clades

emma-extranea, and pima-westfalli (Fig. 5B). This

combination of ten species is topologically iden-

tical in both Figs. 2 and 4, and also rooted in the

same place. The difference is that in Fig. 2 the

component is apical in the tree and monophyletic,

whereas in Fig. 4 it is deeper in the tree and

Fig. 4. Strict consensus of six equally parsimonious trees generated from combined 16S and morphology data.

Clades with dots are recovered in ML analyses of 16S alone. Numbers below branches are Bremer support values.


paraphyletic (but with components adjacent,

hence identical in topology to the monophyletic

interpretation).

Fig. 5C summarizes the placement of these

components by plotting the alternative arrange-

ments simultaneously on the topology common to

both solutions. The alternative placements in

Fig. 5C produce either Fig. 2 or Fig. 4. This

demonstration, requiring cropping only three

species from a total of 27, illustrates that the

molecular and combined matrices differ in modest

ways with respect to the connection of species to

each other, even if these differences have undesir-

able consequences for a strict consensus. Keeping

in mind that the total number of trees possible

from this set of 24 terminals is approximately 5 3

1026, it is evident that the matrices are concordant,

or congruent, except in the alternatives shown in

Fig. 5C. The fact that the entirety of the

component of Fig. 5B is identical in both trees is

also encouraging. We conclude from this explo-

ration of agreement subtrees that our analyses

identify the same close networks of species, even if

there is ambiguity regarding the placement of

Fig. 5. Alternate arrangement of three components account for a majority of the disagreement between the

molecular-only and combined analyses. Figure 5A shows the sedula-pulla-rhoadsi component, Fig. 5B shows the

pima-extranea component, and Figure 5C shows the alternate placements of 5A and 5B on the background of the

component that is rooted near oculata.


those networks. The main lesson is that morpho-

logical data provide adequate phylogenetic signal

to move the sedula-pulla-rhoadsi group apically,

away from A. oculata Hagen in Selys, and place

the component of Fig. 5b to a more basal portion

of the tree, even in combination with molecular

data. These findings demonstrate the importance

of morphological data to complement molecular

data in Argia, and warrant further investigation of

potentially informative morphological characters.

The combined parsimony tree (Fig. 4) in many

ways accords with traditional groupings of species

based on morphology alone. Having a phylogeny

with which to compare to these previous non-

phylogenetic hypotheses of relationship allows us

to provide a test, and a revised hypothesis.

Because our phylogeny is based on different

characters and analytical methods, it represents

an independent test of the classical morphology-

based species groups of Calvert, Kennedy, Gloyd,

Garrison and others.

The close relationship between A. funcki and A.

lugens, and their sister relationship to the remain-

ing Argia, reflects the original placement of these

species in the genus Hyponeura by Selys (1865) on

the basis of veinational characters. Hyponeura,

along with Diargia Calvert (which included the

species bicellulata Calvert), was synonomized with

Argia after it was determined that venational

characters are unreliable for species distinction

(Gloyd, 1968a). Kennedy (1920b) divided about

55 species of Argia into five subgenera on the

basis of the morphology of the penes; indeed,

omitting those taxa that do not fall into the

geographic range of our study, we recover many

of Kennedy’s groups in our analyses, both

molecular and combined: Cyanargia includes A.

lacrimans (Hagen) and A. tonto Calvert; Heliargia

is composed of A. vivida Hagen, A. plana Calvert,

(plus A. immunda, which we pull out of this clade);

Chalcargia, including A. translata, A. harknessi, A.

cuprea, A. oenea, A. ulmeca, A. occulata, (A.

garrisoni and A. sp. nov. were not available to

Kennedy). He also placed here A. tezpi and A.

sedula although our analyses do not support this.

Kennedy defined his species groups solely on

morphology of the intromittent penes, so it is

satisfying that we find similar patterns based on

molecular and morphological characters not used

by Kennedy.

Additionally, taxonomists have proposed close

relationships among (A. vivida, A. plana, A.

extranea), (A. fumipennis and A. pallens), (A.

westfalli and A. anceps), and (A. tonto and A.

lacrimans) (Gloyd, 1958; Garrison, 1994, 1996;

Westfall and May, 1996), which our analyses

support. A. plana was originally described as a

variety of A. vivida by Calvert (1901), and Gloyd

later elevated it to species status on the basis of

clasper morphology (1958). Our results validate

Gloyd, as we recover A. vivida as sister to A. plana

plus A. extranea. The close relationship between

the latter was suggested on the basis of coloration

and morphology by Garrison (1994).

In a thorough study of some Argia larvae,

Novelo-Gutierrez separated species into groups

on the basis of a single character: the degree of

convexity of the ligula (1992). He placed species

into three groups: those with very prominent

ligulae (including A. emma, harknessi, insipida,

moesta, oenea, tezpi, translata and ulmeca),

moderately prominent ligulae (A. munda, taras-

cana, and tonto), and slightly prominent ligulae

(A. fumipennis, lacrimans, nahuana, plana, pulla,

rhoadsi, and sedula). Our analysis suggests that

this character is not homologous (Fig. 4), or at

least is insufficient to accurately infer phylogenetic

relationship. Larvae of Argia species remain

poorly studied. Our limited inclusion of larval

characters indicates their potential value as

sources of phylogenetic characters, but further

study of larval morphology is needed. Of the 118

species of Argia currently recognized, larvae are

known from relatively few species.

Argia fumipennis is composed of the three

subspecies atra, fumipennis, and violacea. These

were described as separate species largely on the

basis of wing color; A. f. violacea has the typical

clear wing color, A. f. fumipennis can have smoky-

hyaline wing color in parts of its range, while A. f.

atra has dark brown wing pigmentation. These

species were unified on the basis of several

morphological characters (Gloyd, 1968b). Closely

related to A. fumipennis, on the basis of general

size, appearance, and clasper morphology, is A.

pallens (Westfall and May, 1996), a species

restricted in the U.S. to southeastern Arizona.

Our analyses do not include A. f. fumipennis, but

based on the molecular data we place A. pallens

within the A. fumipennis group, sister to A. f.

violacea (Figs. 2, 3). The morphological data pull

A. pallens out as sister to A. fumipennis (Fig. 4) in

a monophyletic clade. Both results are only

weakly supported, so further investigation of


these relationships is warranted, as taxonomic

changes may be justified.

A somewhat surprising result is that we fail to

recover in our analyses a monophyletic assem-

blage of the ‘‘metallic’’ species- those that have the

cupreous coloration in the head and thorax (A.

cuprea, A. oenea, A. orichalcea). We only code one

morphological character related to the metallic

condition in this study. Further analysis of this

group is needed, and more detailed character

analysis may provide additional synapomorphy

for these species that have traditionally been

considered close relatives.

A plethora of additional characters, including

several morphological features unique to Argia,

are available for further study. For example,

males have a unique pair of pad-like structures

called ‘‘tori,’’ located posterodorsally on the tenth

abdominal segment between the cerci; the mor-

phology of these, and of the cerci and paraprocts,

are critical for species identification. These char-

acters are used in published taxonomic keys

(Garrison, 1994; Westfall and May, 1996; Forster,

2001), yet have not been thoroughly explored as

phylogenetic characters. Species of Argia also

differ dramatically in morphology of the penes

(Kennedy, 1920a), and while Kennedy used some

of these structures in his ‘‘phylogeny’’ of Zygop-

tera, his work predated by many years the formal

development of phylogenetic methodology. A

careful reexamination of these structures in Argia

is warranted. Additional mitochondrial genes, as

well as both ribosomal and protein-coding genes

such as 12S, 16S, cytochrome oxidases I and II,

are already known to be informative for odonate

phylogenetics (Chippendale et al., 1999; Misof et

al., 2000; Turgeon et al., 2005; Bybee et al., 2008).

In addition, it has been demonstrated that nuclear

genes such as Histone 3, Elongation Factor 1-

alpha, 18S and 28S rDNA show adequate

variation for species-level analyses in odonates

(Ware et al., 2007; Bybee et al., 2008; Carle et al.,

2008). We continue to code, refine, and expand

upon our data set as we continue our work on the

systematics and evolution of Argia.

ACKNOWLEDGMENTS

We thank several collectors who kindly donatedspecimens that were used in this study, listed inTable 1. Bill Mauffray of the IORI made thecollection available on several occasions and provid-ing and generously loaned specimens for this study.For discussion of various aspects of this study, we

thank Joe Gillespie, Dennis Paulson, Mark McPeek,Ola Fincke, and members of the Wenzel Lab at OSU.The suggestions of two anonymous reviewers im-proved an earlier version of our manuscript. We owespecial gratitude to Rosser Garrison for sharing hisimmense expertise on Neotropical Odonata with us.

LITERATURE CITED

Abbott, J. C. 2005. Dragonflies and Damselflies of Texas

and the South-Central United States. Princeton

University Press, Princeton, New Jersey, USA.

Borror, D. J. 1934. Ecological studies of Argia moesta

Hagen (Odonata: Coenagrionidae) by means of

marking. Ohio Journal of Science 34: 97–108.

Bick, G. H. and J. C. Bick. 1965. Color variation and

significance of color in reproduction in the

damselfly, Argia apicalis (Say) (Zygoptera: Coena-

griidae) (sic). Canadian Entomologist 97(1): 32–41.

Bick, G. H. and J. C. Bick. 1971. Localization, behavior,

and spacing of unpaired males of the damselfly

Argia plana Calvert (Odonata: Coenagrionidae).

Proceedings of the Entomological Society of

Washington 73: 146–152.

Bick, J. C. and G. H. Bick. 1982. Behavior of adults of

dark-winged and clear-winged subspecies of

Argia fumipennis (Burmeister) (Zygoptera: Coa-

nagrionidae). Odonatologica 11(2): 99–107.

Bybee, S. M., T. H. Ogden, M. A. Branham and M. F.

Whiting. 2008. Molecules, morphology and

fossils: a comprehensive approach to odonate

phylogeny and the evolution of the odonate wing.

Cladistics 24(4): 477–514.

Caesar, R. M., N. Gillette and A. I. Cognato. 2005.

Population Genetic Structure of an Edaphic

Beetle (Ptiliidae) Among Late Successional Re-

serves within the Klamath-Siskiyou Ecoregion,

California. Annals of the Entomological Society

of America 98(6): 931–940.

Calvert, P. P. 1901. Odonata, in Biologia Centrali

Americana: Insecta Neuroptera. R. H. Porter

and Dulau and Co., London, UK.

Carle, F. L., K. M. Kjer and M. L. May. 2008.

Evolution of Odonata, with special reference to

Coenagrionoidea (Zygoptera). Arthropod Sys-

tematics & Phylogeny 66(1): 37–44.

Chippendale, P. T., D. Varshal, D. H. Whitmore and J. V.

Robinson. 1999. Phylogenetic relationships of

North American damselflies of the genus Ischnura

(Odonata: Zygoptera: Coenagrionidae) based on

sequences of three mitochondrial genes. Molecular

Phylogenetics and Evolution 11: 110–121.

Conrad, K. F. 1992. Relationships of larval phenology

and imaginal size to male pairing success in Argia

vivida Hagen (Zygoptera: Coenagrionidae). Odo-

natologica 21(3): 335–342.

Conrad, K. F. and G. Pritchard. 1988. The mating

behavior of Argia vivida Hagen: as an example of


female-control mating system (Zygoptera: Coe-

nagrionidae). Odonatologica 17: 179–185.

Conrad, K. F. and G. Pritchard. 1990. Pre-oviposition

mate guarding and mating behavior of Argia

vivda (Odonata: Coenagrionidae). Ecological

Entomology 15: 363–370.

Daigle, J. J. 1991. Argia garrisoni spec. nov. from

Mexico (Zygoptera: Coenagrionidae). Odonato-

logica 20(3): 337–342.

Daigle, J. J. 1995. Argia carlcooki spec. nov. from

Mexico (Zygoptera: Coenagrionidae). Odonato-

logica 24(4): 467–471.

Dunkle, S. W. 1990. Damselflies of Florida, Bermuda

and the Bahamas. Scientific Publishers,

Gainesville, Florida, USA.

Forster, S. 2001. The dragonflies of Central America

exclusive of Mexico and the West Indies. Gunnar

Rehfeldt, Wolfenbuttel, Germany.

Garrison, R. W. 1994. A synopsis of the genus Argia of the

United States with keys and descriptions of new

species, Argia sabino, A. leonorae, and A. pima

(Odonata: Coenagrionidae). Transactions of the

American Entomological Society 120(4): 287–368.

Garrison, R. W. 1996. A synopsis of the Argia fissa

group, with descriptions of two new species, A.

anceps sp. n. and A. westfalli sp. n. (Zygoptera:

Coenagrionidae). Odonatologica 25(1): 31–47.

Garrison, R. W. and N. von Ellenrieder. 2007. The true

Argia difficilis Selys, 1865, with the description of

Argia yungensis sp. nov. (Odonata: Coenagrioni-

dae). Transactions of the American Entomolog-

ical Society 133(1): 189–204.

Gloyd, L. K. 1958. The dragonfly fauna of the Big Bend

region of Trans-Pecos Texas. Occasional Papers

of the University of Michigan Museum of

Zoology 658.

Gloyd, L. K. 1968a. The union of Argia fumipennis

(Burmeister, 1839) with Argia violacea (Hagen,

1861), and the recognition of three subspecies

(Odonata). Occasional Papers of the University

of Michigan Museum of Zoology.

Gloyd, L. K. 1968b. The synonomy of Diargia and

Hyponeura with the genus Argia (Odonata:

Coenagrionidae: Argiinae). Michigan Entomolo-

gist 1(8): 271–274.

Huelsenbeck, J. P. 1997. Is the Felsenstein zone a fly

trap? Systematic Biology 46: 69–74.

Kennedy, C. H. 1918. New species of Odonata from the

southwestern United States. Part I. Three new

Argias. Canadian Entomologist 50(7): 256–261.

Kennedy, C. H. 1919. A new species of Argia (Odonata).

Canadian Entomologist 51: 17–18.

Kennedy, C. H. 1920a. The phylogeny of the zygopter-

ous dragonflies as based on the evidence of

the penes. Ohio Journal of Science 21: 19–32.

Kennedy, C. H. 1920b. Forty-two hitherto unrecognized

genera and subgenera of Zygoptera. The Ohio

Journal of Science 21: 83–88.

Misof, B., C. L. Anderson and H. Hadrys. 2000. A

Phylogeny of the Damselfly Genus Calopteryx

(Odonata) Using Mitochondrial 16S rDNA

Markers. Molecular Phylogenetics and Evolution

15(1): 5–14.

Moody, D. 2002. Coenagrionidae: Narrow-winged dam-

selflies. In Glotzhober, R. C. and D. McShaffrey

(eds.), The Dragonflies and Damselflies of Ohio.

Ohio Biological Survey, Columbus, Ohio, USA.

Nixon, K. C. 2000. Winclada version 0.99 (BETA).

Published by the author, Ithaca, New York

(available at www.cladistics.com).

Novelo-Gutierrez, R. 1992. Biosystematics of the larvae

of the genus Argia in Mexico (Zygoptera:

Coenagrionidae). Odonatologica 21: 39–71.

O’Grady, E. W. and M. L. May. December 2003. A

phylogenetic reassessment of the subfamilies of

Coenagrionidae (Odonata, Zygoptera). Journal

of Natural History 37: 2807–2834.

Rambur, M. P. 1842. Histoire Naturelle des Insectes

Neuropteres. Librarie Encyclopedique de Roirot,

Paris, France.

Robinson, J. V., J. E. Dickerson and D. R. Bible. 1983.

The demographics and habitat utilization of

adult Argia sedula (Hagen) as determined by

mark-recapture analysis (Zygoptera: Coenagrio-

nidae). Odonatologica 12(2): 167–172.

Simon, C., F. Frati, A. Beckenbach, B. Crespi, H. Liu and

P. Flook. 1994. Evolution, weighting, and phylo-

genetic utility of mitochondrial gene sequences and

a compilation of conserved polymerase chain

reaction primers. Annals of the Entomological

Society of America 87(6): 651–701.

Thompson, J. D., D. G. Higgins and T. J. Gibson. 1994.

CLUSTAL W: improving the sensitivity of

progressive multiple sequence alignment through

sequence weighting, positions-specific gap penal-

ties and weight matrix choice. Nucleic Acids

Research 22: 4673–4680.

Turgeon, J., R. Stoks, R. A. Thum, J. M. Brown and M.

A. McPeek. 2005. Simultaneous Quaternary

Radiations of Three Damselfly Clades across the

Holarctic. American Naturalist 165: E78–E107.

Ware, J. M., M. L. May and K. Kjer. 2007. Phylogeny

of the higher Libelluloidea (Anisoptera: Odo-

nata): An exploration of the most speciose

superfamily of dragonflies. Molecular Phyloge-

netics and Evolution 45: 289–310.

Westfall, M. J. and M. L. May. 1996. Damselflies of North

America. Scientific Publishers, Gainsville, FL.

Zwickl, D. J. 2006. Genetic algorithm approaches for the

phylogenetic analysis of large biological sequence

datasets under the maximum likelihood criterion.

Ph.D. dissertation, The University of Texas at

Austin.

Received 10 February 2009; accepted 22 November 2009


A Phylogenetic Test of Classical Species Groups in Argia (Odonata: Coenagrionidae)

Documents

Transcript of A Phylogenetic Test of Classical Species Groups in Argia (Odonata: Coenagrionidae)