Phylogeny and biogeography of Triatominae(Hemiptera: Reduviidae): molecular evidence of a

New World origin of the Asiatic clade

V�aaclav Hyp�ssa,a,b,* David F. Tietz,a,b Jan Zrzav�yy,a,c Ryan O.M. Rego,a,b Cleber Galvao,d

and Jos�ee Jurbergd

a Faculty of Biological Sciences, University of South Bohemia, �CCesk�ee Bud�eejovice, Czech Republicb Institute of Parasitology, Academy of Sciences, �CCesk�ee Bud�eejovice, Czech Republicc Institute of Entomology, Academy of Sciences, �CCesk�ee Bud�eejovice, Czech Republic

d Laborat�oorio Nacional e Internacional de Refereencia em Taxonomia de Triatom�ııneos, Departamento de Entomologia, Instituto Oswaldo Cruz,

Fiocruz, Av. Brasil 4365, 21045-900 Rio de Janeiro, Brazil

Received 24 July 2001; received in revised form 12 December 2001

Abstract

The most representative sample of molecular data, especially 16S and 12S rDNAs, is used to study the phylogeny and evo-

lution of 57 species of three tribes, Rhodniini, Linshcosteini, and Triatomini, of the subfamily Triatominae. For the first time both

New World and Old World species are brought together in a single phylogenetic analysis. Maximum-parsimony and distance

estimation place both the Asiatic representatives, Linshcosteus and Triatoma rubrofasciata, as sister groups. The Linshcosteus–T.

rubrofasciata clade nests firmly within Triatomini, in most analyses branching as a basalmost lineage, thus supporting a mon-

ophyletic origin of Triatominae. A paraphyly of ‘‘Triatoma’’ with respect to Linshcosteus, Dipetalogaster, Eratyrus, and Pan-

strongylus and the paraphyly of ‘‘Rhodnius’’ with respect to Psammolestes is observed in most of the analyses. Reinterpretation of

triatomine biogeography points to the origin of Triatominae in northern areas of South America, in Central America, or in the

southern region of North America. A few taxonomic changes are proposed: (1) reinclusion of Linshcosteus in Triatomini, (2)

inclusion of Psammolestes in Rhodnius, (3) elevation of the ‘‘T. flavida complex’’ to the full genus Nesotriatoma (including

N. flavida, N. bruneri, and N. obscura), (4) inclusion of the ‘‘T. spinolai complex’’ in Mepraia (including M. spinolai, M. gajardoi,

M. eratyrusiformis, and M. breyeri), and (5) inclusion of ‘‘T.’’ dimidiata in Meccus (M. dimidiatus). � 2002 Elsevier Science

(USA). All rights reserved.

Keywords: Triatominae; Phylogeny; Biogeography; 16S rRNA

1. Introduction

Although generally recognized as a monophyleticsubfamily of the Reduviidae, adapted to a blood-feedingstrategy (Usinger, 1944; Lent and Wygodzinsky, 1979;Clayton, 1990; Schuh and Slater, 1995), the subfamilyTriatominae poses several serious phylogenetic questionswith respect to its origin and geographical distribution.The current classification of Triatominae rests mainly on

an extensive review by Lent and Wygodzinsky (1979),followed by several considerable modifications by Lentet al. (1994) (elevation ofMepraia to full genus), Jurbergand Galvao (1997) (elevation of Hermanlentia), Car-cavallo et al. (1998) (description of Torrealbaia), andCarcavallo et al. (2000) (elevation ofMeccus, separationof Linshcosteus as a tribe). Six tribes with 18 genera(Triatomini: Triatoma,Meccus,Dipetalogaster,Mepraia,Eratyrus, Panstrongylus, Hermanlentia, Paratriatoma;Linshcosteini: Linshcosteus; Rhodniini: Rhodnius,Psammolestes; Cavernicolini: Torrealbaia, Cavernicola;Bolboderini: Bolbodera, Belminus, Parabelminus, Mi-crotriatoma; Alberproseniini: Alberprosenia) composed

Molecular Phylogenetics and Evolution 23 (2002) 447–457

MOLECULARPHYLOGENETICSANDEVOLUTION

www.academicpress.com

*Corresponding author. Fax: +420-38-5300388.

E-mail address: vacatko@paru.cas.cz (V. Hypsa).

1055-7903/02/$ - see front matter � 2002 Elsevier Science (USA). All rights reserved.

PII: S1055 -7903 (02 )00023-4

of 133 species, the great majority of them found in theNew World, are now recognized. While several speciesrepresenting 2 genera also inhabit continental India (5spp. of Linshcosteus) and south and southeast Asia toNew Guinea and northern Australia (8 spp. of theTriatoma rubrofasciata complex), no autochthonousendemic species are found in Africa. Trying to solve thisbiogeographical puzzle, Schofield (1988, 2000) proposeda concept of polyphyletic origin of the triatomines. Ac-cording to his view, the Asiatic fauna is composed of atleast two independent lineages. The first Asiatic lineageconsists of several species of Triatoma, which hadevolved from the originally New World species T. ru-brofasciata after its introduction into the OldWorld. Theother lineage is represented by the genus Linshcosteus, asupposedly autochthonous Asiatic lineage of blood-feeding reduviids. Although frequently discussed, thequestion of Triatominae origin and evolution had neverbeen seriously addressed using a rigorous cladistic ap-proach until a recent analysis combining morphological,ecological, geographical, behavioral, and molecularcharacters across all Triatominae (Tietz, 2000). The re-sults of this study nested both Linshcosteus and the T.rubrofasciata complex within the monophyletic Triato-mini and placed an origin of Triatominae in the areabetween southwest North America and central Mexico.Recently, several phylogenetic analyses using moleculardata have been published (Garcia and Powell, 1998;Stothard et al., 1998; Lyman et al., 1999; Garc�ııa, 1999;Monteiro et al., 2000; Marcilla et al., 2001). Althoughbased on limited taxon sampling, and hence unable tosolve the key questions of triatomine phylogeny, theseanalyses indicated that the morphological studies mayoverestimate the phylogenetic significance of conspic-uous morphological autapomorphies. In such cases,phylogenetic affiliations of highly derived and mor-phologically aberrant species or lineages may not beproperly recognized, resulting in taxonomic isolation ofsuch taxa and their classification at a high taxonomicalrank. This phenomenon is well documented by theobvious paraphyly of the two largest genera, i.e.,Rhodnius and Triatoma, with respect to Psammolestesand other genera of the Triatomini, respectively(Lyman et al., 1999; Monteiro et al., 2000; Marcillaet al., 2001).In this paper, we use the most representative sample

of molecular data published so far to study the phy-logeny and evolution of 57 species of Triatominae rep-resenting the tribes Rhodniini, Linshcosteini, andTriatomini. For the first time, both New World and OldWorld species are brought together in a single phylo-genetic analysis, thus allowing us to address the questionof Triatominae monophyly. Based on the phylogeneticanalyses, the origin, evolution, and biogeography oftriatomines are reinterpreted and several revisions of thegeneric level classification are proposed.

2. Materials and methods

2.1. DNA extraction, PCR amplification, and sequencing

Total DNA was extracted and purified from ho-mogenized thoracic muscles using the Dneasy Tissue kit(Qiagen). Primer pairs mt32–mt34 and mt35–mt36 (in-sect mtDNA set; Biotechnology Laboratory, Universityof British Columbia) were used to amplify approx. 500-and 400-bp-long fragments of 16S rDNA and 12SrDNA, respectively. PCR products were ligated eitherinto the pGem-T Easy vector (Promega) or the pCR-TOPO vector (Invitrogen) and sequenced using T7 andSP6 as forward and reverse primers, respectively, in anABI PRISM sequencer (Perkin–Elmer Model 310). Thespecies analyzed and sequences’ accession numbers arelisted in Table 1.

2.2. Alignments and phylogenetic analysis

Since the complete 16S rDNA matrix included 62species and covered a large taxonomic span, the matriceswere likely to be affected by misalignment due to anumber of homoplasies, particularly within the variableregions. To address this problem, four matrices differingin taxa sampling were designed and analysed. (1) The‘‘16S’’ matrix included partial 16S rDNA sequences of57 Triatominae species representing the tribes Rhodni-ini, Linshcosteini, and Triatomini; Fulgora laternaria,Halobates matsumarai, Pseudovelia tibialis, and 2 non-triatomine reduviid species, Arilus cristatus and Redu-vius personatus, were used as outgroups. (2) The‘‘Rhodnius’’ matrix included partial 16S rDNA of 12Rhodnius and 2 Psammolestes species; 4 species repre-senting the Triatomini and 2 nontriatomine reduviidspecies were used as outgroups. (3) The ‘‘SA’’ matrixincluded 16S rDNA of all South American species ofTriatoma, Panstrongylus, and Mepraia; 3 Asiatic andNorth American species (T. rubrofasciata, T. dimidiata,and T. rubida) were used as outgroups. (4) The ‘‘com-bined’’ matrix included all species for which at least twoof five genes (Table 1) were available. Since the ITS-2genes of Rhodniini and Triatomini show a very low levelof homology and since only three ITS-2 sequences areavailable for Rhodniini, the Rhodniini ITS-2s were notincluded in this matrix, to avoid an increase in numberof homoplastic characters.The matrices were aligned either by the MALIGN

program (Wheeler and Gladstein, 1994; Rhodniini) orby ClustalX (16S and SA). In ClustalX, each matrix wasaligned under nine different combinations of parameters(Ts/Tv ratios 1:1, 1:2, and 1:3; gap opening/gap exten-sion penalties 5/3, 8/5, and 12/8). In MALIGN, a heu-ristic algorithm ‘‘build, score 4’’ was performed with 10random rearrangements of sequence input order fol-lowed by branch-swapping procedures ‘‘treeswap’’ and

448 V. Hyp�ssa et al. / Molecular Phylogenetics and Evolution 23 (2002) 447–457

‘‘alignswap.’’ Substitution cost to (internal) gap costratios were 6:3, 4:3, 3:3, 3:4, and 3:6; leading and trailinggap costs were set to 20; the Ts/Tv ratios were set to 1:1,1:2, and 1:3. In the combined matrix, only one arbi-trarily selected alignment was used for each noncodinggene, to avoid an exponential increase with number ofmatrices. Multiple alignments were analyzed usingPAUP* 4.0b4 (Swofford, 1998). Maximum-parsimony(MP) analyses were performed by the TBR algorithmwith 10 randomizations of sequence order. Each matrixwas analyzed under the 1:1, 1:2, and 1:3 Ts/Tv ratios.Four different models (uncorrected distance (p); Ha-segawa et al., 1985 (HKY85); Tamura and Nei, 1993(TamNei); and log-determinant (LogDet)) were used indistance estimation. In all analyses, gaps were treated asmissing data.

3. Results

3.1. 16S, Rhodnius, and SA matrices

The length of the 16S rDNA sequences varies from505 to 510 bp; after aligning with 24 additional 16SrDNAs retrieved from GenBank (Table 1), the resultingalignment is 533–557 positions long, depending on gappenalties and Ts/Tv ratio setting. Accordingly, thenumber of variable and parsimony-informative posi-tions varies from 302 to 315 and from 213 to 225, re-spectively. The topologies obtained by MP analysisvaried in dependence on parameters and produced onlypoorly resolved consensus trees (Fig. 1). All analysesrecognized Rhodniini and Triatomini–Linshcosteini asmonophyletic sister clades. Within Rhodniini, bothPsammolestes species clustered together with the Rhod-nius prolixus complex. The R. pallescens complex was abasalmost lineage of Rhodniini in the majority of thetrees. MP analysis of the Rhodnius matrix provided re-sults fully compatible with those inferred from the 16Smatrix (Rhodniini in Fig. 4). The three complexes (R.pictipes, R. pallescens, and R. prolixus complexes) wererecognized as monophyletic clades by all analyses. Ofthree possible topologies, (R. pallescens (R. pictipes+R.prolixus)) was supported by a majority of the trees.Psammolestes, R. neivai, and R. domesticus clusteredwithin the R. prolixus complex.In Triatomini, a large polytomy formed a basal por-

tion of the MP tree without any clear rooting tendency.More consistent results were obtained when the 16Smatrix was analyzed under four distance estimationmodels (p, HKY85, TamNei, and LogDet; Fig. 2). Bothsemistrict and majority-rule consensus trees of the dis-tance analyses placed the Linshcosteus–T. rubrofasciataclade as the sister group of all New World species.Within the New World Triatomini, the South Americanspecies of Triatoma, with the exceptions of T. vitticeps

and T. tibiamaculata, formed a robust monophyleticclade (SA Triatoma clade) present in all analyses. Aninternal topology of this clade was compatible with thatobtained by MP and contained several well-supportedspecies complexes. MP analysis of the SA matrix pro-duced a tree compatible with the results of 16S analysis,but provided better resolution. While the position of thePanstrongylus species remained uncertain, T. eratyrusi-formis+Mepraia spinolai constituted a sister lineage ofthe SA Triatoma clade. Within the SA Triatoma clade,three robust clades designated the T. infestans, T. sor-dida, and T. circummaculata complexes, were recognizedby all analyses (Fig. 4).

3.2. Combined matrix

MP analysis of conservative, unequivocally alignedsites produced 44 trees. The strict consensus was wellresolved, retaining three large monophyletic cladescorresponding to Rhodniini, North American Triato-mini, and South American Triatomini (including T.vitticeps and Panstrongylus megistus; Fig. 3). If theequivocally aligned sites were also included in theanalysis, the only change of the topology was a switch ofP. megistus to the base of the North American branch,supported by a low Bremer index (Fig. 3).The preferred tree discussed below (Fig. 4) was con-

structed as follows. Inner topologies of Rhodniini andSA Triatoma clade are based on Rhodnius and SA ma-trices, respectively. The basal position of the Linshcos-teus–T. rubrofasciata clade corresponds to one of threepossible rootings found by MP and is, moreover, sup-ported by the majority-rule consensus of distance ana-lyses of the 16S matrix. A basal split into twomonophyletic branches represented by North Americanand South American species is adopted from the com-bined matrix and is compatible with the majority-ruleconsensus trees of both MP and distance analyses of 16Smatrix. The position of the Panstrongylus clade withinthe South American branch is preferred due to a betterfit with the biogeography and the distribution of sym-biotic bacteria (see below). The monophyly of Pan-strongylus is based on results obtained aftercombination of the 16S matrix with morphological datafrom Tietz (2000).

4. Discussion

4.1. Origin of the Triatominae: monophyly versus poly-phyly

A consensus of all analyses performed in this study isconsistent with the scenario held by most authors whoconsider the Triatominae a monophyletic clade of re-duviids adapted to a blood-feeding strategy (Usinger,

V. Hyp�ssa et al. / Molecular Phylogenetics and Evolution 23 (2002) 447–457 449

Table 1

List of species and accession numbers of genes analyzed in this study

Tribe Group Subgroup Complex-LW Complex-C Species 16S rDNA Cyt. b COI ITS2 12S rDNA

Fulgora laternaria AF158057�

Halobates matsumarai AB026607�

Pseudovelia tibialis AB026613�

Arilus cristatus AF045712� AF045729�

Reduvius personatus AY035436 AF394517

Triatomini Dipetalogaster maximus AY035442 AF045728� AJ286887� AF394524

Eratyrus mucronatus AY035450

Linshcosteus sp. AF394595

T. protracta T. protracta Triatoma protracta AY035444 AF045727�

Triatoma nitida AF045702� AF045723�

T.lecticularia N Triatoma lecticularia AY035443

N Triatoma sanguisuga AF045696� AF045725�

T. rubrofasciata T. rubrofasciata N Triatoma rubrofasciata AY035468

N Triatoma rubida AY035445 AF394523

T. infestans T. phyllosoma Meccus Meccus pallidipennis AF045697� AF045724� AJ286882� AF394522

Meccus mazzottii AY035446 AJ286885�

Meccus picturatus AY035447

T. dimidiata Triatoma dimidiata AY035448 AF045726� AJ286880� AF301594�

T. spinolai Mepraia Mepraia spinolai AY035467

T. breyeri Triatoma eratyrusiformis AY035466

T. flavida T. flavida Triatoma flavida AY035451

Triatoma bruneri AF394594

T. infestans T. infestans Triatoma infestans AF021198� AF045721� AF021199� AJ289876� AF021197�

Triatoma platensis AF021201� AF021202� AF021200�

Triatoma delpontei AF028745�

Triatoma melanosoma AY035462

Triatoma rubrovaria AF021203� AF021204� AF021207�

T. maculata Triatoma maculata AY035465

Triatoma pseudomaculata AY035461

T. sordida Triatoma sordida AF021209� AF045730� AF021210� AJ293589� AF021208�

Triatoma garciabezi AY035455

Triatoma patagonica AY035464

Triatoma guasayana AF021192� AF021193� AF021196�

T. oliveirai Triatoma klugi AY035463

Triatoma jurbergi AY035456

Triatoma matogrossensis AY035454 AF394521

N Triatoma guazu AY035457

Triatoma williami AY035458

N Triatoma tibiamaculata AY035453

N Triatoma costalimai AY035459

N Triatoma arthurneivai AY035460

N Triatoma brasiliensis AF021183� AF021184� AJ293591� AF021187�

N Triatoma vitticeps AF021218� AF021219� AF021217�

T. circummacu-

lata

T. circummaculata Triatoma circummaculata AF021188� AF021190�

P. lignarius Panstrongylus herreri AY035452

N Panstrongylus megistus AJ243336� AF045722� AF021179� AJ286886� AF0211178�

N Panstrongylus geniculatus AF394593

N Panstrongylus lutzi AY035449

450

V.Hyp �ssa

etal./Molecu

larPhylogenetics

andEvolution23(2002)447–457

1944; Lent and Wygodzinsky, 1979; Clayton, 1990;Schuh and Slater, 1995). A dissenting view consideringTriatominae a polyphyletic assemblage of several redu-viid lineages has been proposed by Schofield (1988,2000), based predominantly on ecological and biogeo-graphical arguments. An independent origin of blood-feeding strategy, at least in Rhodniini and Triatomini,was recently reported to receive support from severalmolecular and physiological studies. The presence ofdifferent antihemostatic components in Rhodniini andR

hodniini

R.pallescens

Rhodniuspallescens

AF045706�

AF045720�

Rhodniusecuadorensis

AF028746�

AF045715�

AF045711�

Rhodniuscolombiensis

AY035438

R.pictipes

Rhodniuspictipes

AF045709�

AF045713�

AF028748�

Rhodniusstali

AY035437

R.prolixus

Rhodniusprolixus

AF045707�

AF045718�

AF394519

AF028747�

Rhodniusnasutus

AF028749�

AF394520

Rhodniusneglectus

AF045704�

AF045716�

Rhodniusrobustus

AF045705�

AF045717�

AF394518

Rhodniusdomesticus

AY035440

NRhodniusbrethesi

AF045710�

AF045714�

NRhodniusneivai

AY035441

Psammolestescoreodes

AF045708�

AF045719�

Psammolestestertius

AY035439

Note:Theinformaltaxonomicgroupings(‘‘groups,’’‘‘subgroups,’’and‘‘complexes’’)wereacceptedfromLentandWygodzinsky(1979)(‘‘complex-W’’)andCarcavalloetal.(2000)(‘‘complex-C’’).N,taxonnotassigned

toanyofthecomplexes.

*SequencesretrievedfromGenBank.

Fig. 1. Majority-rule consensus obtained from strict consensus trees of

all MP analyses of the 16S dataset. The numbers at nodes are boot-

strap supports obtained by 1000 replications of MP analysis of one,

arbitrarily selected 16S alignment under Tv:Ts¼ 1:1 assumption

(TL¼ 1310, CI¼ 0.37, RI¼ 0.61, RC¼ 0.22).

V. Hyp�ssa et al. / Molecular Phylogenetics and Evolution 23 (2002) 447–457 451

Triatomini (Ribeiro et al., 1998) and the sequence di-versity observed between the two tribes are consideredthe most significant arguments in favor of the Tri-atominae polyphyly and an independent origin of he-matophagy (for a review see Carcavallo et al., 2000). Inhis review on biosystematics of Triatominae, Schofield(1988) argues that not only Rhodniini, but even someother genera classified within Triatomini (such as Di-petalogater, Panstrongylus, and Eratyrus), may representunrelated groups that originated from different reduvi-ids. This scenario consistently produces an apparentgeographical disjunction of triatomines with the occur-rence of Linshcosteus and the T. rubrofasciata complexin Asia and the absence of any sylvatic triatomines inAfrica (Schofield, 1988, 2000; Gorla et al., 1997). Thisview was not confirmed nor supported by any of severalmolecular analyses published in the last few years. Onthe contrary, the paraphyly of Rhodnius with respect to

Psammolestes was unequivocally demonstrated in twopapers (Lyman et al., 1999; Monteiro et al., 2000). Thisfinding provides a good example of how rapid mor-phological and ecological changes result in a loss of the‘‘diagnostic’’ characters that are expected to be sharedby members of a given monophyletic group. We supposethat similar processes may have occurred within thetribe Triatomini, thus giving rise to several morpho-logically and bionomically distinct forms of triatomines.Indeed, the paraphyletic character of the genus Tria-toma with respect to at least some other Triatominigenera, particularly to Panstrongylus and Dipetalogas-ter, has become obvious (Marcilla et al., 2001).Although there is no unequivocal proof for Tri-

atominae monophyly at present, the following mor-phological characters pointed out by Lent andWygodzinsky (1979) should be seriously discussed aspossible autapomorphies of the Triatominae: (1) he-matophagous feeding habit, (2) elongate and nearlystraight labium with a flexible membranous connectionbetween segments 3 and 4, allowing upwardly pointeddistal rostral segment when the rostrum is in feedingposition, and (3) loss of dorsal abdominal scent glandsin nymphs. All these characters are primary homologiessensu de Pinna (1985) (¼ putative homology statementsprior to tree construction) and should be tested byconfrontation with other, e.g., molecular characters,

Fig. 2. Majority-rule and strict consensus (boldface lines) of distance

estimation analyses of the 16S dataset.

Fig. 3. Strict consensus of MP analyses of the combined dataset. The

line width is proportional to Bremer (decay) indices shown at indi-

vidual nodes (numbers above lines are Bremer indices for complete

alignment; numbers below lines are Bremer indices for conservative-

regions-only alignment). Two alternative positions of P. megistus are

shown by dashed lines.

452 V. Hyp�ssa et al. / Molecular Phylogenetics and Evolution 23 (2002) 447–457

because the test of congruence of shared derived char-acters (i.e., establishment of the secondary homology) isthe ultimate arbiter of the hypotheses on homology. Tothe best of our knowledge, no single putative apomor-phy shared by some triatomines and some nontriato-mines has been published so far.In addition to the putative morphological synapo-

morphies discussed above, the monophyly of Triatomi-nae is also supported by the observations derived fromthe present molecular study. The genera Panstrongylus,Dipetalogaster, and Eratyrus and the Asiatic cladecluster within the genus Triatoma and their sequences donot display significantly higher distances than those

observed among Triatoma species. Consequently, atleast the hypothesis of separate origins of individualgenera within the Triatomini is falsified. The transpacificdistribution of the Triatominae (see Schuh and Slater,1995, p. 40) is not exceptional in the Hemiptera—Col-obathristidae, Thaumastocoridae, and the reduviidsubfamilies of Physoderinae, Peiratinae, and Vesciinaeshare this type of distribution. In some hemipterans(smaller clades within Miridae and Rhyparochromidae),the sister groups of the transpacific clades inhabit Africa(Schuh and Slater, 1995). This seems to support thevicariant origin of the transpacific patterns (that areabundant in the flowering plants, e.g., Chloranthaceae).

Fig. 4. A preferred phylogeny of the Triatominae (see Discussion). Newly proposed taxonomic combinations are boldfaced (for discussion on the

generic names used in the tree see Systematic implications). SA, SA Triatoma clade; 1, T. infestans complex; 2, T. circummaculata complex; 3, T.

sordida complex.

V. Hyp�ssa et al. / Molecular Phylogenetics and Evolution 23 (2002) 447–457 453

4.2. Phylogeny of Rhodniini

Recently, Schofield and Dujardin (1999) published anevolutionary scenario for the origin and evolution of theRhodniini based on two assumptions: (1) the tribe rep-resents a monophyletic group of reduviids which hasdeveloped a blood-feeding lifestyle independently of thetribe Tratomini, and (2) an ancestor of Rhodniini, rep-resented by an extant widespread species, R. pictipes,originated in arboreal habitats of the Amazon–Orinocorainforests and then dispersed northwest and south togive rise to the R. pallescens and R. prolixus complexes.In contrast to this view, the R. pictipes complex wasnever found as the most primitive branch in any of ouranalyses. Depending on parameters used, two compet-ing topologies were obtained, placing the R. pictipescomplex as sister group of either the R. pallescens or theR. prolixus complex. Both topologies have been recentlypublished as results of molecular analyses (Lyman et al.,1999; Monteiro et al., 2000). Although the availablemolecular data do not allow for a convincing discrimi-nation between these two topologies, the tree with R.pallescens complex as a basal branch was preferred by amajority of the analyses performed in this study. Thesefindings indicate that an occurrence of widespread spe-cies, such as R. pictipes, R. prolixus, and R. robustus, inAmazonia may reflect a secondary invasion of this areaby highly successful species rather than the origin of thewhole tribe. This scenario is well compatible with thegeneral concept of triatomine invasion into the centralparts of South America from the northern parts of thesubcontinent or even from the southern parts of NorthAmerica (Tietz, 2000; for further discussion see below),and with the supposedly recent origin of the Amazonfauna compared to other parts of South America(Nores, 1999, and references therein; see also Zrzav�yyand Nedv�eed, 1999).

4.3. Higher-level phylogeny of Triatomini and Linshco-steini: the rooting problem, biogeography, and ecology

The Asiatic species Linshcosteus sp. and T. rubrofas-ciata form a monophyletic clade, branching within theTriatominae. In MP analyses, the position of this cladewithin Triatominae is unstable and varies with the pa-rameters (Fig. 1). The Linshcosteus–T. rubrofasiata cladeis recognized as one of three possible roots of the Tri-atomini, and this position is supported by a majority ofdistance-based analyses (Fig. 2). In addition to theLinshcosteus–T. rubrofasiata clade, two alternative rootsare indicated by MP analyses, one between Meccus andthe rest of Triatomini and the other between the T.protracta group and the rest of Triatomini. The rootingby T. rubrofasciata–Linshcosteus (shown in Figs. 2 and4) is thus the only arrangement supported by both dis-tance analyses and some MP. It has also been favored

over the T. protracta-rooted topologies by ITS-2 anal-ysis where T. barberi (representing the T. protractacomplex) is the sister group of Dipetalogaster maximus(Marcilla et al., 2001). It is important to stress that re-gardless of the variable position of the T. rubrofasciata–Linshcosteus clade, the monophyly of this group and itsclear affinity to all other Triatomini strongly favor themonophyly of both Triatomini (including Linshcosteus):the extant Asiatic species seem likely to represent a lin-eage which dispersed to Asia at an early stage of theTriatomini radiation. The three possible rootings differin evolutionary interpretation of the relationships be-tween the North and the South American faunas of theTriatominae. Whereas the Linshcosteus–T. rubrofasci-ata-rooted trees tend to show the basal split situatedbetween the North and the South American groups,with an uncertain position of the ‘‘Panstrongylus clade’’(see below), the other two topologies make the NorthAmerican species paraphyletic with respect to the SouthAmerican clade.

4.4. T. phyllosoma complex–Meccus

Recently, a genus Meccus was revalidated for largeMexican species with several morphological synapo-morphies (Carcavallo et al., 2000). Lent and Wygod-zinsky (1979), although recognizing a close relationshipand unique features of these species, treated them asmembers of the genus Triatoma (as T. phyllosomacomplex, including T. phyllosoma, T. picturata, T.mazzotiii, T. longipennis, and T. pallidipennis). Theyfurther stated that ‘‘T. dimidiata . . . is superficiallysomewhat similar to phyllosoma group and agrees in thefeatures of the abdomen of the fifth instar nymph.’’ Incontrast to this view, Carcavallo et al. (2000) arguedthat T. dimidiata differs morphologically from otherspecies in Mexico and the United States and establisheda separate T. dimidiata complex with putative origin innorthern South America. The results that we obtainedby analyzing the 16S and combined matrices stronglysupport the former view and place T. dimidiata nearMeccus, usually as its sister group. Clustering of T. di-midiata together with the species of Meccus was previ-ously demonstrated in sequence analyses of themitochondrial large subunit rRNA and cytochrome b(Lyman et al., 1999) and of ITS-2 (Marcilla et al., 2001),although on a more restricted taxon sample. Anotherspecies clustering within, or as a sister group of,Meccusin a majority of analyses was Triatoma sanguisuga.However, while the basal position of this species withrespect to the T. dimidiata–Meccus clade is recognizedby most MP analyses, the distance estimates place T.sanguisuga deep into Meccus. As the result of MPanalysis better fits morphological arguments and sinceonly short fragments of the 16S rRNA gene are avail-able for T. sanguisuga and M. pallidipennis, we assume

454 V. Hyp�ssa et al. / Molecular Phylogenetics and Evolution 23 (2002) 447–457

that the incongruence between the MP and the distancemethod may be due to incorrect distance calculation.

4.5. Panstrongylus clade

In addition to four Panstrongylus species, this cladeunexpectedly includes three species of Triatoma, viz., T.flavida, T. bruneri, and T. tibiamaculata. Although neverformally suggested, a close relationships of T. flavidaand T. bruneri on the one hand and Panstrongylus(namely, P. lignarius complex) on the other hand issupported by the overall color pattern and by thepronotum with an unusually wide posterior lobe (Lentand Wygodzinsky, 1979). In this paper, P. herreri is theonly species representing the P. lignarius complex. Itshould be noted that in both distance estimation andMP, P. herreri is the closest relative of the T. flavidacomplex, either as a single-species branch or togetherwith P. megistus (Figs. 2 and 4). A reliable cladogram ofPanstrongylus is needed to decide whether these mor-phological similarities reflect phylogenetic relationship.Within the Panstrongylus clade, the relationships of in-dividual species are poorly resolved, without any cleartendency of Panstrongylus spp. to cluster as a mono-phyletic group. This observation is difficult to explain,but since the genus Panstrongylus is a morphologicallywell-defined taxon, we speculate that the failure ofPanstrongylus species to form a monophyletic cladecould reflect the short time between the origin of Pan-strongylus and its subsequent radiation, rather than thenonmonophyly of Panstrongylus. The position of thePanstrongylus clade within the Triatomini is sensitive toparameters used during the alignment process—it clus-ters as the basalmost offshoot of either the SouthAmerican or the North American group. A comparisonof all-sites and conservative-sites-only analyses shows alow support for either topology. Indirect evidence sup-porting close relationships between Panstrongylus andthe South American clade comes from 98% 16S rDNAsimilarity of Arsenophonus triatominarum, the Triatomainfestans’s intracellular symbiotic bacterium (Hyp�ssa andDale, 1997), with the Panstrongylus megistus’s symbiont,whereas no symbiotic bacteria have been detected inRhodniini and Dipetalogaster maximus (unpublishedresults). The symbiont distribution thus supportsmonophyly of Panstrongylus+ the South Americanclade, unless secondary losses of the symbionts in otherTriatomini are hypothesized. Based on analysis of ITS-2sequences, Marcilla et al. (2001) reported a basal di-chotomy between North American species on the onehand and Panstrongylus+South American species onthe other, with R. prolixus used as an outgroup. How-ever, the authors acknowledge a ‘‘substantial variationin the alignment of all triatominae ITS-2 sequences’’ andemphasize a great dissimilarity between the sequences ofRhodniini and Triatomini. Further, a comparison of

ITS-2 genes of all available triatomines shows clearlythat while there are diagnostic features characterizingboth North American species and species representingthe SA Triatoma clade, the P. megistus sequence differsconsiderably from both groups. Thus, using Rhodniuswith its highly divergent ITS-2 sequence as an outgroupmay be misleading and may cause an incorrect rootingof the triatomine tree. In our opinion, although an exactposition of the Panstrongylus clade remains to be veri-fied, an occurrence of several Panstrongylus species inCentral America and/or northern parts of SouthAmerica (P. chinai, P. geniculatus, P. humeralis, P.lignarius, and P. rufotuberculatus) and the Greater An-tillean distribution of the T. flavida group suggest thatan ‘‘intermediate’’ position of the Panstrongylus clademay well fit the basal dichotomy between North andSouth American species.

4.6. Mepraia and the T. spinolai complex

Lent and Wygodzinsky (1979) characterized the T.spinolai complex, viz. T. eratyrusiformis, T. breyeri, andT. spinolai as a taxonomically and geographically iso-lated group of species that inhabit semiarid areas insouthern South America. Based on several morpholog-ical characters, Lent et al. (1994) removed the last spe-cies from the genus Triatoma, revalidating the genusMepraia Mazza, Gajardo & Jorg, 1940 (later an addi-tional species, M. gajardoi, was described by Frias et al.(1998)). In our analysis, Mepraia spinolai and T. eraty-rusiformis form a stable and well-supported branchplaced within the South American clade, often as a mostbasal lineage of the SA Triatoma clade (MP analyses;Fig. 1) or as a sister group of the Panstrongylus clade(majority of distance analyses; Fig. 2).

4.7. The SA Triatoma clade

The SA Triatoma clade represents one of the mostrobust clades, present in all analyses performed. It en-compasses 20 species, recognized by Lent and Wygod-zinsky (1979) on morphological grounds as members ofthe T. infestans, and T. circummaculata complexes (Ta-ble 1). However, 2 species previously considered mem-bers of the T. infestans complex, T. tibiamaculata, and T.vitticeps, display an unstable position, always branchingoutside the SA Triatoma clade. T. tibiamaculata shows atendency to cluster in the vicinity of the Panstrongylusclade (see above) while T. vitticeps’s unstable position isreflected by a separate position of this species within alarge polytomy encompassing all North American spe-cies in addition to the monophyletic SA Triatoma clade(Fig. 1). Forcing T. vitticeps to form a monophyleticgroup together with the SA Triatoma clade produces atree only one step longer and places this species at thebase of the clade. The same position is obtained without

V. Hyp�ssa et al. / Molecular Phylogenetics and Evolution 23 (2002) 447–457 455

any constraint by analyzing a matrix combining 16SrDNA data with the other genes (Fig. 3). Moreover,calculation of Bremer indices for the combined matrixreveals T. vitticeps+SA Triatoma clade as one of themost robust clades in the tree. In this respect, it is per-tinent to note that while all species clustering within theSA Triatoma clade share the same karyotype (20+XY),two different patterns, 20+XXXY and 20+XXY, havebeen identified for T. vitticeps and T. tibiamaculata, re-spectively (Schreiber and Pellegrino, 1950; Schreiber etal., 1972; Panzera et al., 1996). Within the SA Triatomaclade, two species, T. brasiliensis and T. maculata, dis-play quite unstable positions while the rest of speciesform three well-supported groups, We adopt the well-established terms T. infestans, T. circummaculata, and T.sordida complexes to name these groups (Fig. 4), al-though they differ considerably in their species contentsfrom those suggested recently on morphologicalgrounds (Carcavallo et al., 2000).

4.8. Systematic implications

Within the Rhodniini, it is evident that Psammolestesspp. represent aberrant species of Rhodnius. Splitting thetraditional (i.e., paraphyletic) genus Rhodnius into sev-eral monophyletic genera seems inappropriate becauseall Rhodnius s. str. species are conspicuously similar toone another morphologically and because there are nogeneric-rank names proposed by earlier authors. Con-sequently, we propose here to include all the Psammo-lestes species into Rhodnius, as R. arthuri (Pinto, 1926)comb. n., R. coreodes (Bergroth, 1911) comb. n., and R.tertius (Lent and Jurberg, 1965) comb. n.The interrelationships among genera and species of

the Triatomini are shown as quite unstable in this study.However, the tribal separation of Linshcosteus proposedby Carcavallo et al. (2000) is not supported by thephylogenetic relationships because T. rubrofasciata, thetype species of Triatoma Laporte, 1832, is an immediatesister group of Linshcosteus.The paraphyletic nature of ‘‘Triatoma’’ is evident. In

all analyses, the ‘‘Triatoma’’ flavida–‘‘T.’’ bruneri cladegroups as a close relative to Panstrongylus. Genericname Nesotriatoma (Usinger, 1944), has been proposedfor ‘‘T.’’ flavida, a taxonomic decision followed here.The genus Nesotriatoma is a well-characterized GreaterAntillean clade including three species, N. flavida (Ne-iva, 1911) (type species), N. bruneri (Usinger, 1944), andN. obscura Maldonado and Farr, 1962.The case of ‘‘T.’’ tibiamaculata is more problematic:

this species tends to group near to the Panstrongylus–Nesotriatoma clade and might eventually be elevated tothe generic rank (Eutriatoma Pinto, 1926). However,some of the species that were sometimes (in a ratherchaotic manner) referred to as Eutriatoma spp. are in-cluded in the present study, and all except for Neso-

triatoma and ‘‘T.’’ tibiamaculata group within the SATriatoma clade. At present, we do not propose to clas-sify Eutriatoma as a full genus because of its uncertainspecies content.Monophyly of Meccus and its distinct phylogenetic

position within the Triatomini is supported by thisstudy. The results indicate that ‘‘T.’’ dimidiata should beincluded in this genus also, as Meccus dimidiatus(Latreille, 1811) comb. n. The phylogenetic position of‘‘T.’’ sanguisuga and its possible inclusion inMeccus hasto be reinvestigated when more molecular data areavailable.‘‘T.’’ eratyrusiformis groups with no exceptions as a

sister species of Mepraia spinolai, forming the aridsouthern South American T. spinolai complex (Lent andWygodzinsky, 1979) monophyletic. Because the positionof this species couple is quite uncertain (and basal incombined trees), we have decided to classify this clade asa full genus including four species, M. spinolai (Porter,1934),M. gajardoi Frias, Henry and Gonzalez, 1998,M.eratyrusiformis (Del Ponte, 1929) comb. n., and M.breyeri (Del Ponte, 1929) comb. n.The other ‘‘Triatoma’’ spp. should be left unclassified

at present. It is probable that at least three separategroups hidden under a common generic name are rep-resented in this study—(i) the probably paraphyletic ‘‘T’’.sanguisuga–protracta–lecticularia–nitida–rubida group,possibly related to (or even congeneric with) Dipetalog-aster; (ii) T. rubrofasciata (closely related to Linshcos-teus); and (iii) SA Triatoma clade, a monophyletic cladeincluding the rest of ‘‘Triatoma’’ species.

Acknowledgments

Randall T. Schuh (New York) and an anonymousreviewer helped to improve the manuscript. This workwas supported by Grants No. A6022801 (Grant Agencyof the Academy of Sciences of the Czech Republic) andNo. 96066 and MSM 123100003 (Ministry of Educa-tion, Czech Republic) and by Conselho Nacional deDesenvolvimento Cient�ııfico e Tecnol�oogico, CNPq andFundac�a o Nacional de Sa�uude, Brazil.

References

Carcavallo, R.U., Jurberg, J., Lent, H., 1998. Torrealbaia martinezi,

gen. nov. sp. n., da tribo Cavernicolini (Hemiptera, Reduviidae,

Triatominae): Uma abordagem filogen�eetica. Entomol. Vect. 5,

143–150.

Carcavallo, R.U., Jurberg, J., Lent, H., Noireau, F., Galvao, C., 2000.

Phylogeny of the Triatominae (Hemiptera: Reduviidae). Proposal

for taxonomic arrangements. Entomol. Vect. 7 (Suppl. 1), 1–99.

Clayton, R.A., 1990. A Phylogenetic Analysis of the Reduviidae

(Hemiptera: Heteroptera) with Redescriptions of the Subfamilies

and tribes. Ph.D. dissertation, George Washington University,

Washington, DC.

456 V. Hyp�ssa et al. / Molecular Phylogenetics and Evolution 23 (2002) 447–457

de Pinna, M.C.C., 1985. Concepts and tests of homology in the

cladistic paradigm. Cladistics 4, 367–394.

Frias, D.A., Henry, A.A., Gonzalez, C.R., 1998. Mepraia gajardoi: A

new species of Triatominae (Hemiptera: Reduviidae) from Chile

and its comparison withMepraia spinolai. Rev. Chil. Hist. Nat. 71,

177–188.

Garc�ııa, B.A., 1999.Molecular phylogenetic relationships among species

of the genus Triatoma. In: Carcavallo, R.U., Gal�ıındez-Gir�oon, I.,

Jurberg, J., Lent, H. (Eds.), Atlas of Chagas Disease Vectors in the

Americas, vol. III. Editora Fiocruz, Rio de Janeiro, pp. 971–980.

Garc�ııa, B.A., Powell, J.R., 1998. Phylogeny of species of Triatoma

(Hemiptera: Reduviidae) based on mitochondrial DNA sequences.

J. Med. Entomol. 35, 232–238.

Gorla, D.E., Dujardin, J.P., Schofield, C.J., 1997. Biosystematics of

Old World Triatominae. Acta Trop. 63, 127–140.

Hasegawa, M., Kishino, H., Yano, T., 1985. Dating the human-ape

split by a molecular clock of mitochondrial DNA. J. Mol. Evol. 22,

160–174.

Hyp�ssa, V., Dale, C., 1997. In vitro culture and phylogenetic analysis of

‘‘Candidatus Arsenophonus triatominarum,’’ an intracellular bacte-

rium from the triatomine bug, Triatoma infestans. Int. J. Syst.

Bacteriol. 47, 1140–1144.

Jurberg, J., Galvao, C., 1997. Hermanlentia n. gen. da tribo Triato-

mini, com um rol de esp�eecies de Triatominae (Hemiptera, Redu-viidae). Mem. Inst. Oswaldo Cruz 92, 181–185.

Lent, H., Jurberg, J., Galvao, C., 1994. Revalidac�ao do ganero

Mepraia Mazza, Gajardo & Jorg, 1940 (Hemiptera, Reduviidae,

Triatominae). Mem. Inst. Oswaldo Cruz 89, 347–352.

Lent, H., Wygodzinsky, P., 1979. Revision of the Triatominae

(Hemiptera, Reduviidae) and their significance as vectors of

Chagas’ disease. Bull. Am. Mus. Nat. Hist. 163, 123–520.

Lyman, D.F., Monteiro, F.A., Escalante, A.A., Cordon-Rosales, C.,

Wesson, D.M., Dujardin, J.P., Beard, C.B., 1999. Mitochondrial

DNA sequence variation among Triatominae vectors of Chagas’

disease. Am. J. Trop. Med. Hyg. 60, 377–386.

Marcilla, A., Bargues, M.D., Ramsey, J.M., Magallon-Gastelum, E.,

Salazar-Schettino, P.M., Abad-Franch, F., Dujardin, J.P., Scho-

field, C.J., Mas-Coma, S., 2001. The ITS-2 of nuclear rDNA as a

molecular marker for population, species, and phylogenetic rela-

tionships in Triatominae (Hemiptera: Reduviidae), vectors of

Chagas Disease. Mol. Phylogenet. Evol. 18, 136–142.

Monteiro, F.A., Wesson, D.M., Dotson, E.M., Schofield, C.J., Beard,

C.B., 2000. Phylogeny and molecular taxonomy of the Rhodniini

derived from mitochondrial and nuclear DNA sequences. Am. J.

Trop. Med. Hyg. 62, 460–465.

Nores, M., 1999. An alternative hypothesis for the origin of Amazo-

nian bird diversity. J. Biogeogr. 26, 475–485.

Panzera, F., P�eerez, R., Hornos, S., Cestau, R., Panzera, Y., Delgado,V., Nicolini, P., 1996. Chromosome numbres in the Triatominae

(Hemiptera-Reduviidae): A review. Mem. Inst. Oswaldo Cruz 91,

515–518.

Ribeiro, J.M.C., Schneider, M., Isaias, T., Jurberg, J., Galvao, C.,

Guimaraes, J.A., 1998. Role of salivary antihemostatic components

in blood feeding by Triatomine bugs (Heteroptera). J. Med.

Entomol. 35, 599–610.

Schofield, C.J., 1988. Biosystematics of the Triatominae. In: Service,

M.W. (Ed.), Biosystematics of Haematophagous Insects. Claren-

don, Oxford, pp. 285–312.

Schofield, C.J., 2000. Biosystematics and evolution of the Triatominae.

Cad Sa�uude P�uublica Rio de Janeiro 16, 89–92.

Schofield, C.J., Dujardin, J.P., 1999. Theories on the evolution of

Rhodnius. Actual Biol. 21, 183–197.

Schreiber, G., Bogliolo, A.R., Coelho de Pinho, A., 1972. Cytogenetics

of Triatominae: Caryotype, DNA content, nuclear size and

heteropycnosis of autosomes. Rev. Brasil Biol. 32, 255–263.

Schreiber, G., Pellegrino, J., 1950. Eteropicnosi di autosomi come

possible meccanismo di speciazione (Ricerche citologishe su alcuni

Emitteri neotropic). Sci. Genet. 3, 215–226.

Schuh, R.T., Slater, J.A., 1995. True Bugs of the World (Hemiptera:

Heteroptera): Classification and Natural History. Cornell Univer-

sity Press, Ithaca, NY.

Stothard, J.R., Yamamoto, Y., Cherchi, A., Garcia, A.L., Valente,

S.A.S., Schofield, C.J., Miles, M.A., 1998. A preliminary survey of

mitochondrial sequence variation in Triatominae (Hemiptera:

Reduviidae) using Polymerase chain reaction-based single-strand

conformational polymorphism (SSCP): Analysis and direct se-

quencing. Bull. Entomol. Res. 88, 553–560.

Swofford, D.L., 1998. Phylogenetic Analysis Using Parsimony (*and

Other Methods), ver. 4. Sinauer, Suterland, MA.

Tamura, K., Nei, M., 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.

Tietz, D.F., 2000. Molecular and morphological analysis of phylogeny

of subfamily Triatominae (Heteroptera: Reduviidae). M.Sc. thesis,

Faculty of Biological Sciences, University of South Bohemia, Cesk�eeBudejovice (in Czech).

Usinger, R.L., 1944. The Triatominae of North and Central America

and the West Indies and their public health significance. Publ.

Health Bull. 288, 1–83.

Wheeler, W.C., Gladstein, D., 1994. MALIGN, ver. 1.99 (ftp.amn-

h.org).

Zrzav�yy, J., Nedv�eed, O., 1999. Evolution of mimicry in the New

World Dysdercus (Hemiptera: Pyrrhocoridae). J. Evol. Biol. 12,

956–969.

V. Hyp�ssa et al. / Molecular Phylogenetics and Evolution 23 (2002) 447–457 457