Utility of the Mitochondrial Cytochrome Oxidase II Gene for

gfplticsmlellntoifl

Molecular Phylogenetics and EvolutionVol. 16, No. 2, August, pp. 286–295, 2000doi:10.1006/mpev.2000.0807, available online at http://www.idealibrary.com on

Resolving Relationships among Black Flies (Diptera: Simuliidae)K. P. Pruess,* B. J. Adams,† ,‡ T. J. Parsons,† ,§ X. Zhu,* and T. O. Powers†

*Department of Entomology, University of Nebraska, Lincoln, Nebraska 68583-0816; †Department of Plant Pathology,University of Nebraska, Lincoln, Nebraska 68583; ‡Department of Nematology, University of Florida,

Gainesville, Florida 32601; and §Armed Forces DNA Identification Laboratory, Rockville, Maryland 20850

Received August 31, 1999, revised May 3, 2000

Chironomoidea with Chironomidae (midges) and Cera-

dT

lSstc1nrHc

smsgitu(

The complete mitochondrial cytochrome oxidase IIgene was sequenced from 17 black flies, representing13 putative species, and used to infer phylogeneticrelationships. A midge (Paratanytarsus sp.) and threemosquitoes (Aedes aegypti, Anopheles quadrimacula-tus, and Culex quinquefasciatus) were used as out-

roup taxa. All outgroup taxa were highly divergentrom black flies. Phylogenetic trees based on weightedarsimony (a priori and a posteriori), maximum like-

ihood, and neighbor-joining (log-determinant dis-ances) differed topologically, with deeper nodes be-ng the least well-supported. All analyses supportedurrent classification into species groups but relation-hips among those groups were poorly resolved. Theajority of phylogenetic signal came from closely re-

ated sister taxa. The CO-II gene may be useful forxploring relationships at or below the subgenericevel, but is of questionable value at higher taxonomicevels. The weighting method employed gave phyloge-etic results similar to those reported by other au-hors for other insect CO-II data sets. A best estimatef phylogenetic relationships based on the CO-II genes presented and discussed in relation to current blacky classification. © 2000 Academic Press

Key Words: black fly; Simuliidae; mitochondrialDNA; cytochrome oxidase subunit II; phylogeny; char-acter weighting.

INTRODUCTION

Black flies (Diptera: Simuliidae) are a large group ofmedically and economically important Diptera in thesuborder Nematocera which have speciated with littlemorphological change. A paucity of diagnostic charac-ters for all life stages complicates species identificationand therefore control efforts often target black flies ingeneral, rather than the few pest species.

Previous workers agree that Simuliidae is a mono-phyletic group. Wood and Borkent (1989) and Ooster-broek (1995) place Simuliidae in the superfamily

1055-7903/00 $35.00Copyright © 2000 by Academic PressAll rights of reproduction in any form reserved.

286

topogonidae (biting midges) as sister groups. Thesetaxa are included, with Culicidae (mosquitoes) andother families, in the infraorder Culicomorpha. Molec-ular data, based on 28s rDNA gene sequences(Pawlowski et al., 1996), did not support these relation-ships, suggesting that Thaumaleidae (solitary midges)and Simuliidae are sister groups and that Chironomi-dae may be a sister group to all other Nematocera.Using a different 28s rDNA sequence, Friedrich andTautz (1997) found the Culicomorpha to form astrongly supported monophyletic clade but the rela-tionships between families were poorly resolved.

The fossil record for Simuliidae is meager. The ear-liest known fossil, a pupa from middle Jurassic (170mybp) is considered indistinguishable from modernProsimulium, and larvae from the lower Cretaceous(120 mybp) of Australia are indistinguishable frommodern Simulium (Crosskey, 1990). Quite likely, the

ivergence of modern genera was complete in theriassic.Two subfamilies are commonly recognized in Simu-

iidae, with all but four species placed in the subfamilyimuliinae. Simuliinae is divided into two tribes, Pro-imuliini and Simuliini, but with disagreement as toribal placement of some genera. The genus Simuliumontains over 1000 species (Crosskey and Howard,997). The current classification of black flies recog-izes “species groups” within subgenera without infer-ing relationships among those groups (Crosskey andoward, 1997). Currie (1988) provides an alternative

lassification at the generic level.Black fly taxonomy is further complicated by sibling

pecies complexes whose component species cannot beorphologically differentiated in all life stages. Many

ibling species are best characterized by larval salivaryland chromosome polymorphisms. Sibling species arenferred from lack of heterozygotes in areas of sympa-ry (Rothfels, 1979). Larvae of these sibling species aresually isolated temporally, spatially, or ecologicallyAdler, 1987). Within species groups, a number of cy-

e1a

CCC3

TABLE 1

CSSSSSSSSSSSS

PAA

287BLACK FLY PHYLOGENETICS

tophylogenies based on chromosomal inversions exist(e.g., Rothfels, 1979), but these are unrooted.

The cytochrome oxidase subunit II (CO-II) gene isone of the best-known mitochondrial genes. Cooper etal. (1991) and Saraste (1990) review the structure andfunction of the polypeptide subunits. CO-II containsboth highly conserved and variable regions. The se-quence of this gene is potentially useful for phyloge-netic analyses over a wide taxonomic range and hasbeen explored for that purpose. However, Liu andBeckenbach (1992) were unable to resolve the evolu-tion of insect orders and concluded that divergencemay have occurred too rapidly for resolution usingmitochondrial genes. This sequence has been generallyuseful for reconstructing phylogenies among moreclosely related groups (Willis et al., 1992; Beckenbacht al., 1993; Brown et al., 1994; Sperling and Hickey,994; Emerson and Wallis, 1995; Spicer, 1995; Smithnd Bush, 1997; Foley et al., 1998). With the exception

of the Collembola (Frati et al., 1997), the utility of thissequence has not been well explored at the generic orlower levels for insects.

We obtained the complete sequence of the mitochon-drial CO-II gene for 17 black flies. Our primary pur-pose was to explore the utility of this gene for resolvingrelationships among black flies at different taxonomiclevels (tribe to sibling species). A phylogeny based onparsimony analysis, using empirically derived weights,is presented and discussed in relation to current blackfly classification.

Collection Data for B

Species

Heledon dicentum Dyar & ShannonProsimulium onychodactylum Dyar & ShannonCnephia sp. A (putative C. dacotensis Dyar & Shannon)C. sp. B [putative C. pecuarum (Riley)]

. sp. C (putative C. pecuarum)imulium (Eusimulium) pilosum (Knowlton & Rowe) (5 S. aureum. (Eusimulium) donovani Vargas G1 (5 S. aureum “G”). (Eusimulium) donovani G2. (Psilozia) vittatum Zetterstedt sibling IIIL-1. (Psilozia) vittatum “Iceland”. (Psilozia) encisoi Vargas & Diaz Najera. (Psilopelmia) bivittatum Malloch A. (Psilopelmia) bivittatum B. (Simulium) piperi Dyar & Shannon (hunteri group). (Simulium) jacumbae Dyar & Shannon (hunteri group). (Simulium) luggeri Malloch (jenningsi group). (Simulium) confusum Moulton & Adler (jenningsi group)

Out

aratanytarsus sp. (Chironomidae)edes aegypti (L.) (Culicidae)nopheles quadrimaculata Say (Culicidae)

Culex quinquefasciatus Say (Culicidae)

MATERIALS AND METHODS

Black Fly Collection and DNA Extraction

Larval black flies were either preserved in ethanolupon collection or returned to the laboratory alive andpreserved individually in TE buffer at 270°C. Themidge was obtained as an adult. With two exceptions(to be discussed), all specimens were obtained fromlocalities where, according to collecting and rearingdata, other closely related species did not occur. Col-lection data are given in Table 1. DNA was obtainedusing proteinase K digestion, phenol/chloroform ex-traction, and ethanol precipitation as described byWilkerson et al. (1993).

PCR and Sequencing

Using the notation of Simon et al. (1994), primersused to amplify the complete CO-II gene wereCI-J-2797, 59-CCTCGACGTTATTCAGATTACC-39;C2-J-3123, 59-GGACTACAAGATAGAGCCTC-39;C2-J-3407, 59-CATCAATGATACTGAAGTTATGA-39;

2-J-3684, 59-GGTCAATGTTCAGAAATTTGTGG-39;2-N-3380, 59-TCAATATCATTGATGACCAAT-39;2-N-3686, 59-CAATTGGTATAAAACTATGATTTGC-9; A8-N-3926, 59-AATAAAGATAATCAACTAATAGC-

39; and A6-N-4126, 59-AGTCCTAAGAATGTTCTTA-ATCA-39.

When possible, the entire CO-II gene was amplifiedusing primers C1-J-2797 and A6-N-4126 or A8-N-3926

ck Flies Sequenced

Source

California, Napa Co., 22 March 1990Colorado, Jackson Co., 5 Aug. 1989New Hampshire, Strafford Co., 4 May 1990Texas, Rains Co., 4 Jan. 1991Texas, Rains Co., 4 Jan. 1991

”) Nebraska, Lancaster Co., 24 April 1989California, Mono Co., 25 June 1991California, Mono Co., 25 June 1991Nebraska, Lancaster Co., 25 March 1988Iceland, R. Laxa, 31 Aug. 1990New Mexico, Dona Ana Co., 2 June 1992Nebraska, Loup Co., 12 Oct. 1990Wyoming, Crook Co., 21 May 1994Nebraska, Scottsbluff Co., 11 July 1989Nebraska, Lancaster Co., 30 Oct. 1992Nebraska, Loup Co., 12 July 1990Nebraska, Saline Co., 6 July 1990

ups

Pennsylvania, Chester Co., Jan. 1993Ho et al. (1995) (GenBank L34412)Mitchell et al. (1993) (GenBank L04272)Ho et al. (1995) (GenBank L34351)

la

“B

gro

with other internal primers used for sequencing. In a few

Sgepsfcrseq2

itive results between congruence and accuracy and

288 PRUESS ET AL.

cases, other overlapping combinations were employed.Two sequencing methods were used. Most se-

quence data were obtained by direct sequencing ofdouble-stranded PCR products. Primers were end-labeled with T4 polynucleotide kinase (Promega,Madison, WI) and g-32P and used with Promega’sfmol sequencing system according to manufacturer’srecommendations. Alternatively, products werecloned before sequencing. Amplified products werepurified with Gene Clean II (Bio 101, La Jolla, CA),end-repaired with T4 DNA Polymerase (Promega),ethanol precipitated, and ligated into Smacut pBlue-

cript SK1 vector (Stratagene, La Jolla, CA). Li-ated DNAs were transformed into Stratagene Esch-richia coli XL1-Blue supercompetent cells andlated on selective medium with blue/white colorelection. Individual white colonies were screenedor proper size insert using the initial PCR amplifi-ation conditions except the number of cycles waseduced to 23. Colonies with appropriately sized in-erts were grown, and single-strand DNA was recov-red using VCS M13 Helper Phage (Stratagene). Se-uences were obtained using Sequenase kit version.0 (U.S. Biochemical, Cleveland, OH) with 35S-label-

ing.To the extent possible, both strands were sequenced.

Sequence of the opposite strand usually resolved prob-lems due to compressions in direct sequence. Multipleclones (3 or 4) were sequenced for each taxon to deter-mine a consensus; a few instances of presumed Taqpolymerase error were observed in single clones. Se-quences were derived from a single individual of mostspecies. In two cases, however, additional individualswere sequenced to help resolve questions about thespecies involved. We include all sequences that differ.Sequences reported in this paper have been submittedto the GenBank DNA sequence bank and are availableunder Accession Nos. M76433 and AF08357–AF08373.

Phylogenetic Analysis

Published sequences for three mosquitoes, Aedesaegypti (Ho et al., 1995), Anopheles quadrimaculata(Mitchell et al., 1993), and Culex quinquefasciatus(Ho et al., 1995), were included as outgroup taxa. Wealso sequenced a chironomid, Paratanytarsus sp.,presumed to be more closely related to black flies.DNA sequences were aligned by eye. We also ana-lyzed the data using Paratanytarsus as the sole out-group taxon. Phylogenetic signal was estimated us-ing relative apparent synapomorphy analysis (RASA2.3) (Lyons-Weiler, 1999; Lyons-Weiler et al., 1996).To identify the effect of taxon sampling on signalcontent, we did separate analyses on pruned sets oftaxa (i.e., outgroup taxa, closely related sister taxaremoved). Cunningham (1997) showed that six-pa-rameter weighted parsimony can give consistent pos-

in the present study weights were derived a prioribased on a modification of the six-parameter model ofRodriquez et al. (1990) (Appendix). This model as-sumes that base frequencies of all sequences, includ-ing the ancestral sequence, are at equilibrium. Allpossible pairwise comparisons of nucleotide use weremade by codon position. We evaluated weights basedon nucleotide use at only variable sites and also at allsites (Waddell and Steel, 1997). Outgroup taxa wereincluded in computing weights except for the last sixcodons where alignment was questionable. Theweights used were proportional to the ratio of expect-ed/observed values based on nucleotide frequency.Integer values were entered into step matrices foranalysis with PAUP*4.0.0d64 for PPC and UNIX(provided by David L. Swofford). Starting trees forheuristic searches using branchswapping were ob-tained by stepwise addition. Branchswapping op-tions consisted of random addition sequence using 30replications with tree bisection–reconnection (TBR).

A heuristic search was also used to recover a maxi-mum likelihood (ML) tree. Search options included get-ting the starting tree by stepwise addition, using the“as-is” algorithm of sequence addition, and TBRbranchswapping. The substitution model for the max-imum likelihood analysis assumed unequal transition-to-transversion rates. The ratio of transitions to trans-versions and nucleotide base frequencies wereestimated from the data set. The analysis also assumedthe HKY two-parameter model for unequal base fre-quencies (Hasegawa et al., 1985). Variable sites wereassumed to reflect a gamma distribution with six ratecategories, the shape parameter being estimated fromthe data set. Additionally, a neighbor-joining tree wasproduced using log-determinant distances.

Competing topologies were evaluated using parsi-mony and maximum likelihood criteria according toKishino and Hasegawa (1989) and Templeton’s non-parametric test of parsimony (Templeton, 1983) usingthe Treescores option in PAUP*. To study the effect ofthe outgroup taxa on tree structure, we removed all ofthe outgroup taxa from the analysis with the exceptionof Paratanytarsus and duplicated the tree-buildinganalyses. Because the relationships among the out-group taxa could alter tree lengths and likelihoods ofthe overall tree, and because some trees were gener-ated in the absence of these taxa, all outgroup taxa(except Paratanytarsus) were removed for tree compar-isons. For parsimony trees, nodal support was esti-mated by bootstrapping (500 replicates), and Bremersupport indices (Bremer, 1988, 1994; Kallersjo et al.,1992) were calculated using AutoDecay version 2.9.7(Eriksson, 1997). Trees were visualized using Tree-View version 1.5 (Page, 1998).

McnSn1

WeightsTABLE 2

289BLACK FLY PHYLOGENETICS

RESULTS AND DISCUSSION

CO-II Gene

The CO-II gene in all black flies had 229 codons,including an ATG initiation codon, and a completeputative TAA termination. The A 1 T bias at all codonpositions (Table 2) was typical of that in other se-quenced insects. Within black flies, there was aminoacid variation in 27 codons (Fig. 1) of which 12 werephylogenetically informative. Outgroups varied at 73amino acids.

From a functional standpoint, the amino acid substi-tutions observed in black flies are not surprising. Al-most all amino acid substitutions noted in black fliesalso occur in other insects (Liu and Beckenbach, 1992).In some cases closely related species, with little nucle-otide divergence, had amino acid replacements. Resi-due 42 in S. confusum is a conservative change (L for

). Residue 49 (T for K) in S. jacumbae might beonsidered a nonconservative replacement but wouldot greatly affect the hydropathic profile of this region.uch amino acid differences, even within a species, areot unusual (Willis et al., 1992; Sperling and Hickey,994).

Nucleotide Use by Codon Position in Black Flies

Position

Nucleotide

A C G T

1 1321 681 831 10602 1006 790 536 15613 1767 369 118 1639

FIG. 1. Variable amino acids in CO-I

Although weighting of nucleotide characters is com-mon, there seems little agreement as to how suchweights should be derived and applied. Because wewere interested in relationships at different taxonomiclevels, we felt it desirable to retain third position nu-cleotides which, potentially, could have been the onlypositions differing between closely related species. It isalso becoming evident that third positions, eventhough highly homoplastic, may contain most of thephylogenetic structure in a data set (e.g., Bjolund,1999). Although our weighting scheme resulted in rel-atively large weights for some types of changes,weights were similar to those derived by other methodsin that the third position has lowest weight with tran-sitions less than transversions. This pattern also holdsat the first position, but C-T changes (primarily amongleucine codons) receive a lower weight than any thirdposition transversion. Relative weights assigned by po-sition (ca. 4:14:1) are similar to those computed for acytochrome b data set (4:17:1) by Arnason and Gull-berg (1994) and further explored by Milinkovitch et al.(1996). Our approach, similar to that of Williams andFitch (1990), further subdivided the position weightsinto specific nucleotide changes. This resulted in adeparture from weights assigned by other workers forthe second position, with transitions and transversionsbeing similarly weighted. A second position C-G trans-version in codon 174 was the only second codon posi-tion nucleotide which entered into parsimony analysisof the ingroup, the few other second position differ-ences between black flies all being autapomorphies.Our weighting scheme often violated the triangle in-equality. This would also be the case when applied to

black flies and Nematocera outgroups.

I in

twrT

TABLE 3

290 PRUESS ET AL.

other large data sets such as that of Liu and Becken-bach (1992) and we used weights as computed, but notethat adjusting them to satisfy the triangle inequalitydid not alter tree topology. Cunningham (1997) foundthat six-parameter parsimony gave consistent positiveresults between congruence and accuracy.

Phylogeny

The data set contained significant phylogenetic sig-nal. Even though the outgroup taxa were phylogeneti-cally distant from the ingroup taxa, their contributionto unweighted signal content was appropriate to re-solving relationships among the ingroup taxa (Lyons-Weiler et al., 1998). However, uncontroversial relation-ships among closely related sister taxa was the onlysignificant source of phylogenetic signal (Table 3).Each method of treebuilding produced a single, optimaltree (Figs. 2 and 3). Maximum likelihood and un-weighted parsimony trees displayed disparate topolo-gies (Fig. 3), but the maximum likelihood tree wascongruent with the weighted maximum parsimony tree(Fig. 2) when the highly divergent outgroup taxa wereremoved from the maximum likelihood analysis. Wecautiously propose Fig. 2 as the best estimate of phy-logenetic relationships among the sampled taxa basedon the CO-II gene, with reservations as to the correctordering of nodes below those of closely related sistertaxa. The best estimate of phylogenetic relationshipsimplies a monophyletic Simulium, but the alternativehypotheses depict a paraphyletic Simulium.

Both Simuliidae and Culicidae were strongly sup-ported as monophyletic families. CO-II sequences tendto support the finding of Pawlowski et al. (1996) thatChironomidae may be a sister group to other Nemato-cera. Pruess et al. (1992) found that mosquito, Dro-sophila, and black fly were about equally divergent intRNA genes.

Nucleotide use at variable vs invariant sites differedsignificantly (x2, P , 0.01) at all codon positions. Al-hough weights differed somewhat, both methods ofeighting (nucleotide use at all vs only variable sites)

esulted in identical single most parsimonious trees.he resultant parsimony phylogeny, with four non-

RASA (Relative Apparent Synapomorp

Taxon status Obs

Unrooted, all taxaRootedRooted, prunedUnrooted, outgroup taxa removedRooted with Paratanytarsus, outgroup taxa removedUnrooted, outgroup taxa removed, prunedRooted with Paratanytarsus, pruned

Note. Values of P , 0.05 reject the hypothesis that the data set

black fly outgroup taxa (Fig. 2), fully agreed with cur-rent classification of black flies into species groups, butprovided poorer resolution of relationships among spe-cies groups. Nodes at deeper levels were more weaklysupported but resulted in phylogenies consistent withcurrent classification at the subgeneric or lower levels.Molecular data suggest, as does morphology, the fossilrecord, and cytological evidence, an early divergence ofcurrently recognized taxa at the subgeneric or deeperlevels. Subsequent to this early radiation, there ap-pears to have been little morphological differentiation.

Analysis) Tests of Phylogenetic Signal

ed slope Null slope tRASA Significance

466 5.535 20.2318 NS183 5.003 7.7149 P , 0.0005371 2.794 20.946 NS974 3.455 1.768 P , 0.025514 4.888 7.685 P , 0.0005481 2.238 25.338 P , 0.0005153 2.647 21.351 NS

s not contain a significant phylogenetic signal.

FIG. 2. Hypothesis of phylogenetic relationships by weightedparsimony (all taxa) and maximum likelihood (ingroup taxa andParatanytarsus only). Percentage bootstrap support (500 replicates)is given to the left of each node; Bremer support values appearinternal to their respective node and are preceded by the letter “d.”

hy

erv

5.8.2.3.7.1.2.

doe

chodactylum was as distant from H. dicentum as

PuwSbuai

CwtCwtoSu1ucCtfBtttCswdpTotm

i1wln“duc“S

291BLACK FLY PHYLOGENETICS

Prosimulium. Prosimuliini are generally acceptedto be the more primitive black flies but there is nogeneral agreement as to which genus, Heledon orProsimulium, should be considered most plesiomorphic(Rothfels, 1979). The two Prosimuliini species exam-ined here did not form a monophyletic clade. P. ony-

FIG. 3. Hypotheses of phylogenetic relationships by method ofreconstruction: (A) maximum likelihood (all taxa); (B) neighbor-join-ing tree based on log-determinant distances; (C) unweighted parsi-mony, reweighted parsimony.

H. dicentum was from all other black flies. P. onycho-dactylum is a species complex (Henderson, 1986), andour sequence was probably derived from cytospecies10.

Cnephia. Tribal placement of Cnephia is question-able. Crosskey and Howard (1997) place Cnephia in

rosimuliini. Sohn et al. (1975) and Teshima (1972),sing DNA:DNA hybridization, found that Cnephiaas as similar to some Simulium as species included inimulium were to each other. Both Currie (1988),ased on morphological analysis, and Moulton (2000),sing molecular data, place Cnephia in Simuliini. Ournalysis of the CO-II gene also provided support fornclusion of Cnephia basally in the Simuliini clade.

Our results pose an interesting taxonomic problem.nephia sp. A and sp. B, which differ by 7% un-eighted sequence divergence, were initially assumed

o be C. pecuarum, a serious livestock pest in Texas.. ornithophilia in this area may occur sympatricallyith C. pecuarum (Procunier, 1982), and thus C. orni-

hophilia may have been sampled. We were able tobtain a partial sequence from C. ornithophilia fromouth Carolina (where neither C. dacotensis nor C. pec-arum is known to occur). In 195 bp sequenced (codons11–175), we found C. ornithophilia to differ by 4.6%nweighted sequence divergence from putative C. da-otensis from New Hampshire but by 7.2–12.3% from. pecuarum. The situation was further complicated by

wo partial sequences we obtained from C. dacotensisrom Nebraska where no other Cnephia are known.oth were different, with each being similar to one of

he putative C. pecuarum sequences from Texas. Athis time we cannot definitely ascribe the correct nameo either sequence. The New Hampshire putative. dacotensis did not match either of the two partialequences obtained from Nebraska specimens. Thus,e have an anomalous situation in which we have fouristinct haplotypes for three putative species, with twoutative species each having two shared haplotypes.hese results point out how little is known about levelsf genetic divergence among sibling species, as well ashe great range of genetic diversity represented byorphologically indistinguishable black flies.

Subgenus Eusimulium. S. aureum in North Amer-ca is a complex of perhaps 10 sibling species (Dunbar,959; Leonhardt, 1985), none of which is conspecificith the European nominal species. Of the Eusimu-

ium taxa we studied, we have confidence only in theame identification of S. pilosum (formerly S. aureumB”). The two specimens labeled S. donovani A and S.onovani B came from a collection in which all individ-als cytologically determined (Peter Adler, personalommunication) were S. donovani (formerly S. aureumG”). However, Adler (personal communication) found. pilosum at this site in a previous year. The close

similarity to S. pilosum from Nebraska prompted us to

rtcs

iNPfltwb

a1sIcnbtslTfsmhfppammwcbt

supported clade, but its relationship to other species

292 PRUESS ET AL.

sequence a second individual. Each differed from theother two by ,1% sequence divergence. This couldepresent within-species variation or, at the other ex-reme, three different species. Leonhardt (1985) indi-ated that “B” and “G” were sister species in chromo-omal phylogeny.

Subgenus Psilopelmia. The subgenus Psilopelmias a complex of about 50 named species, primarilyeotropical, but with representatives in the Westernalearctic. S. bivittatum B came from a stream withuctuating water levels and often carrying silt, in con-rast to A which was from a clean sandy-bottom streamith stable flow, typical of the habitat in which S.ivittatum occurs in Nebraska.

Subgenus Psilozia. Based on cytological (Rothfelsnd Featherston, 1981) and ecological (Adler and Kim,984) evidence, S. vittatum is a complex of two siblingpecies, designated IS-7 and IIIL-1. S. vittatum fromceland, lacking a diagnostic sex chromosome system,annot be reliably assigned to either currently recog-ized sibling. Our sequences from IIIL-1 from Ne-raska and a specimen from Iceland differed by lesshan 1%. Zhu et al. (1998) found that the two putativeibling species shared many restriction fragmentength polymorphisms in the mitochondrial genome.ang et al. (1996) failed to identify sibling-specific dif-

erences in either 12S or 16S mitochondrial ribosomalequences. Zhu et al. (1998) identified sequence poly-orphism in the ND4 gene, but the most commonaplotype was present in both siblings. S. encisoi dif-ered from S. vittatum by 9% with which it formed aoorly supported clade. Psilopelmia and Psilozia, bothresumed to be of Neotropical origin, were supporteds a clade. A single amino acid change (D in Psilopel-ia at 129 vs E in other black flies) suggests an apo-orphic character state in Psilopelmia. Had greatereight been given third position transversions which

hange amino acids, S. encisoi would more clearly haveeen placed in the Psilozia clade, a phylogenetic posi-ion supported by morphology.

Subgenus Simulium. The subgenus Simulium isdivided into 18 species groups. S. piperi and S. jacum-bae, members of the hunteri group, are morphologicallydistinct as larvae and pupae, but adults can be reliablyseparated only by genitalic characters. S. piperi has amore northern distribution. Where sympatric, bothtend to occur in small, spring-fed streams which differprimarily in water temperature, S. piperi occurring incooler streams. These two species, with Psilozia andPsilopelmia, form a clade, in agreement with theNearctic or Neotropical distribution of all included spe-cies.

The two species placed in the jenningsi speciesgroup, S. luggeri and S. confusum, form a strongly

groups is unclear. In their revision of the jenningsigroup, Moulton and Adler (1995) placed S. luggeri ba-sally with S. confusum treated as derived. Includedspecies in the jenningsi group are all eastern Nearcticin distribution. The sequence of a member of the east-ern Palearctic or western Nearctic S. malyschevi spe-cies group, considered by Moulton and Adler to be alogical sister group, might help resolve the relationshipof the jenningsi group to other black flies.

Our weighting method is readily computed from thedata, providing an a priori estimate of weights, andrequires little computer time, permitting bootstrap es-timates of support. Although lower level relationshipswere weakly supported, those relationships were bothlogical and better supported than obtained by un-weighted parsimony analyses and distance methods.Inclusion of more species, and better outgroup taxa,would be suggested to determine its full utility. TheParatanytarsus which we used is a highly derived par-thenogenetic species. At the time this study was con-ducted, a specimen of Thaumaleidae was not availablenor were we yet aware of the evidence for a sister grouprelationship with Simuliidae. Inclusion of additionaltaxa in Prosimuliini would perhaps aid in breaking uplong branches (Graybeal, 1998) and better resolve theplacement of Cnephia.

We applied this weighting method to other CO-IIdata sets. For insect orders (Liu and Beckenbach,1992), we also found strong bootstrap support formonophyly of cockroach and termite (97%) and Hyme-noptera (93%) but all other relationships remainedpoorly, and illogically, resolved. It is apparent that theCO-II gene is not informative at the insect order level,a condition that is unlikely to be solved by any methodof weighting or analysis. The low differential in posi-tion weights (2:3:1) and transversion:transitionweights (1.5:1, 1.5:1, 2:1) suggests that saturation wasbeing approached at all positions.

Results for Collembola were almost identical to thosereported by the authors (Frati et al., 1997). Differentialbetween position weights (3:5:1) and transversion:transition (1.5:1, 1:1, 3:1) suggested that although sat-uration was being approached, a strong phylogeneticsignal was retained.

Our results for the beetle phylogeny of Emerson andWallis (1995), with tRNALEU excluded, supported theirphylogeny for nodes with .60% bootstrap support ineither tree, but arrangements differed considerably atless-supported nodes. We computed much greatertransversion:transition weights than used by the au-thors at both first (18:1) and third (10:1) codon posi-tions.

Our analysis of the mosquito data (Foley et al., 1998)gave somewhat different results. Because the authorsreported only relative improvement obtained byweighting, instead of bootstrap support, it is difficult to

basis of chromosomal, or apparent ecological, differ-

Fep

TABLE 4

Pd

AACCGT

LE

AACCG

293BLACK FLY PHYLOGENETICS

make a direct comparison of results. We obtained thesame sister taxa when bootstrap support by our anal-ysis was greater than 60%, but deeper relationshipswere poorly resolved with branching quite differentfrom that of the authors. Without knowing their actualbootstrap values, we cannot judge if the phylogeniesdiffered significantly, nor can we compare the com-puted weights used by us with the unstated weightsused by the authors. We computed 5:17:1 positionweights and transversion:transition weights within po-sition of 7:1, 1:1, and 4:1.

We conclude that the weighting method which wepropose for this type of data will provide results as goodas those obtained by any other method. This studysupports results of other authors that the CO-II gene islikely to be useful for exploring relationships at subge-neric or lower levels, but is of questionable value athigher taxonomic levels.

Of more practical value is the contribution of molec-ular data as a source of characters for identification ofthe taxonomically challenging Simuliidae. The pres-ence of sibling species is usually first suspected on the

Frequency of Nucleotide Use at Variable Sites andairwise Changes by Codon Position in the Mitochon-rial CO-II Gene for Black Flies and Outgroups

Nucleotide

Relative frequency of use at codon position

1 2 3

A 0.3363 0.2601 0.4536C 0.1738 0.2010 0.0919G 0.2143 0.1373 0.0311T 0.2756 0.4017 0.4233

Change Number of pairwise changes

A 7 C 209 208 1,1817 G 1044 102 1,6517 T 801 128 6,4127 G 11 266 977 T 2046 366 5,2907 T 397 118 302

otal 4508 1188 14,933

TAB

Weights Assigned to Nucleotide Changes Comp

Change

Arithmetic position weights

1 2

A 7 C 3.59 6.137 G 1.23 4.717 T 2.72 6.107 G 50.08 5.307 T 0.46 6.267 T 4.03 7.32

ences. Cytology, unfortunately, permits identificationonly of mature larvae collected and preserved underideal conditions and examined by persons with highlyspecialized skills. Molecular data, on the other hand,are applicable to all life stages and in black flies pro-vide a comparatively rich source of variation relative tohighly conserved morphology. Once discovered, molec-ular characters, ideally from independent loci, can thenbe used in combination with morphology and/or cytol-ogy to define species boundaries.

APPENDIX: COMPUTATION OF WEIGHTS

Weights were computed as expected/observedchanges based on a stochastic model assuming basefrequencies are at equilibrium at each codon position.Here we provide a simple algebraic solution illustratedwith computation of codon position weights and weightassigned A7 C changes at codon position 1. Althougheach weight can be computed from a single equation,we will use a two-step process in which positionweights are first computed, then multiplied by relativeweights for each type of change within position. Theexample below is based on nucleotide frequencies atvariable positions, but is equally applicable if one de-sires to include invariant sites.

The expected relative probability of change at eachcodon position, Pc, where fcn is frequency of nucleotiden at codon position c, is

Pc 51 2 ¥ n 5 1

4 f cn2

¥ c 5 13 ~1 2 ¥ n 5 1

4 f cn2 !

. (1)

rom Table 4, given 20,629 pairwise differences in thentire data set, expected frequency of changes at codonosition is 0.7182/2.0309 5 0.3536. Observed fre-

quency is 4508/20629 5 0.2185 and position weight,Wc, is expected/observed 5 0.3536/0.2185 5 1.62. Notethat if nucleotide frequency were the same at eachposition, then the expected changes at each positionunder a stochastic model would be 1/3.

5

ed from Total Data Base Including Outgroups

Integer position weights

3 1 2 3

0.75 25 42 50.15 8 32 10.59 19 42 40.65 343 36 40.19 3 43 10.89 28 50 6

ut

Within codon position, expected relative frequency of

qt2

t

A

A

B

B

B

B

B

Cooper, C. E., Nicholls, P., and Freedman, J. A. (1991). Cytochromec oxidase: Structure, function, and membrane topology of the

C

C

C

C

D

E

E

F

F

F

G

H

H

H

K

K

L

L

L

L

294 PRUESS ET AL.

each type of nucleotide change, ni 7 nj, is given:

E~fc:i7j! 52fci fcj

1 2 ¥ n 5 14 f cn

2(2)

Again for codon position 1, observed frequency f(c:i 7 j) ofA 7 C changes is 209/4508 5 0.0464, expected fre-uency is 2(0.2928)(0.1260)/0.7182 5 0.1027, and rela-ive within-position weight W(c:i 7 j) is 0.1027/0.0464 5.2142.Weight for A7 C changes within codon position 1 is

he product of the two weights:

Wc(A7 C) 5 WcW(c:A7 C) 5 2.2142(1.62) 5 3.5869.

For some phylogenetic analyses, only integer weightsare permitted. These are readily derived fromWc (i 7 j)/Wmin as reported in Table 5.

Upon request, we can provide a program written inMicrosoft GW-BASIC which permits rapid computa-tion of weights.

ACKNOWLEDGMENTS

Peter Adler, Ronnie Byford, Gisli Gislason, Bai Xiong, and JamesRobinson provided some of the black flies used in this study. RobertaWeber provided the Paratanytarsus. We thank David Swofford forpermission to use the test version of PAUP*. Linda Young and DavidTaylor provided constructive reviews and assistance in analysis.This study is published with approval of the Director as JournalSeries 11874, Agricultural Research Division, University of Ne-braska, and is supported in part by NIH Grant AI29456 to T.O.P.

REFERENCES

Adler, P. H. (1987). Ecology of black fly sibling species. In “BlackFlies: Ecology, Population Management, and Annotated WorldList” (K. C. Kim and R. W. Merritt, Eds.), pp. 62–76. PennsylvaniaState Univ., University Park, PA.dler, P. H., and Kim, K. C. (1984). Ecological characterization of twosibling species, IIIL-1 and IS-7, in the Simulium vittatum complex(Diptera: Simuliidae). Can. J. Zool. 62: 1308–1315.

rnason, U., and Gullberg, A. (1994). Relationship of baleen whalesestablished by cytochrome b gene sequence comparison. Nature367: 726–728.eckenbach, A. T., Wei, Y. W., and Liu, H. (1993). Relationships inthe Drosophila obscura species group, inferred from mitochondrialcytochrome oxidase II sequences. Mol. Biol. Evol. 10: 619–634.

jolund, M. (1999). Are third positions really that bad? A test usingvertebrate cytochrome b. Cladistics 15: 191–197.remer, K. (1988). The limits of amino acid sequence data in angio-sperm phylogenetic reconstruction. Evolution 42: 795–803.

remer, K. (1994). Branch support and tree stability. Cladistics 10:295–304.rown, J. M., Pellmyr, O., Thompson, J. N., and Harrison, R. G.(1994). Phylogeny of Greya (Lepidoptera: Prodoxidae), based onnucleotide sequence variation in mitochondrial cytochrome oxidaeI and II: Congruence with morphological data. Mol. Biol. Evol. 11:128–141.

polypeptide subunits. Biochem. Cell. Biol. 69: 586–607.

rosskey, R. W. (1990). “The Natural History of Blackflies.” Wiley,Chichester.

rosskey, R. W., and Howard, T. M. (1997). “A New Taxonomic andGeographical Inventory of World Blackflies (Diptera: Simuliidae).”Natural History Museum, London.

unningham, C. W. (1997). Is congruence between data partitions areliable predictor of phylogenetic accuracy? Empirically testing aniterative procedure for choosing among phylogenetic methods.Syst. Biol. 46: 464–478.

urrie, D. C. (1988). Phylogeny of primitive Simuliidae (Insecta:Diptera: Culicomorpha). Ph.D. dissertation, Univ. Alberta, Edm-onton, Canada.

unbar, R. W. (1959). The salivary gland chromosomes of sevenforms of black flies included in Eusimulium aureum Fries. Can. J.Zool. 37: 495–525.

merson, B. C., and Wallis, G. P. (1995). Phylogenetic relationshipsof the Prodontria (Coleoptera: Scarabaeidae: subfamilyMelolonthinae), derived from sequence variation in the mitochon-drial cytochrome oxidase II gene. Mol. Phylogenet. Evol. 4: 433–447.

riksson, T. (1997). AutoDecay ver. 2.9.7 (Hypercard Stack distrib-uted by the author). Botaniska Institutionen, Stockholm Univ.,Stockholm.

oley, D. H., Bryan, J. H, Yeates, D, and Saul, A. (1998). Evolutionand systematics of Anopheles: Insights from a molecular phylogenyof Australian mosquitoes. Mol. Phylogenet. Evol. 9: 262–275.

rati, F., Simon, C., Sullivan, J., and Swoffford, D. L. (1997). Evolu-tion of the mitochondrial cytochrome oxidase II gene in Collem-bola. J. Mol. Evol. 44: 145–158.

riedrich, M., and Tautz, D. (1997). Evolution and phylogeny of theDiptera: A molecular phylogenetic analysis using 28s rDNA se-quences. Syst. Biol. 46: 674–698.

raybeal, A. (1998). Is it better to add taxa or characters to a difficultphylogenetic problem? Syst. Biol. 47: 9–17.

asegawa, M., Kishino, H., and Yano, T. (1985). Dating of the hu-man–ape splitting by a molecular clock of mitochondrial DNA. J.Mol. Evol. 22: 160–174.

enderson, C. A. P. (1986). A cytological study of the Prosimuliumonychodactylum complex (Diptera: Simuliidae). Can. J. Zool. 64:32–44.

o, C-M., Liu, Y-M., Wei, Y-H., and Hu, S-T. (1995). Gene for cyto-chrome c oxidase subunit II in the mitochondrial DNA of Culexquinquefasciatus and Aedes aegypti (Diptera: Culicidae). J. Med.Entomol. 32: 174–180.

allersjo, M., Farris, J. S., Kluge, A. G., and Bult, C. (1992). Skew-ness and permutation. Cladistics 8: 275–287.

ishino, H., and Hasegawa, M. (1989). Evaluation of the maximumlikelihood estimate of the evolutionary tree topologies from DNAsequence data, and the branching order of the Hominoidea. J. Mol.Evol. 29: 170–179.

eonhardt, K. G. (1985). A cytological study of species in the Eu-simulium aureum group. Can. J. Zool. 63: 2043–2061.

iu, H., and Beckenbach, A. T. (1992). Evolution of the mitochondrialcytochrome oxidase II gene among ten orders of insects. Mol.Phylogenet. Evol. 1: 41–52.

yons-Weiler, J. F. (1999). RASA 2.3 software and documentation forthe Mac. http://test1.bio.psu.edu/LW/rasatext.html.

yons-Weiler, J., Hoelzer, G. A., and Tausch, R. J. (1996). Relativeapparent synapomorphy analysis (RASA) I: The statistical mea-surement of phylogenetic signal. Mol. Biol. Evol. 13: 749–757.

Lyons-Weiler, J., Hoelzer, G. A., and Tausch, R. J. (1998). Optimaloutgroup analysis. Biol. J. Linnean Soc. 64: 493–511.

R

R

R

S

S

Smith, J. J., and Bush, G. L. (1997). Phylogeny of the genus Rhago-letis (Diptera: Tephritidae) inferred from DNA sequences of mito-

S

S

S

T

T

T

W

W

W

W

W

Z

295BLACK FLY PHYLOGENETICS

Milinkovitch, M. C., Leduc, R. G., Adachi, J., Farnir, F., Georges, M.,and Hasegawa, M. (1996). Effects of character weighting and spe-cies sampling on phylogeny reconstruction: A case study based onDNA sequence data in Cetaceans. Genetics 144: 1817–1833.

Mitchell, S. E., Cockburn, A. F., and Seawright, J. A. (1993). Themitochondrial genome of Anopheles quadrimaculatus species A:Complete nucleotide sequence and gene organization. Genome 36:1058–1073.

Moulton, J. K. (2000). Molecular sequence data resolves basal diver-gences within Simuliidae (Diptera). Syst. Entomol. 25: 95–113.

Moulton, J. K., and Adler, P. H. (1995). Revision of the Simuliumjenningsi species-group (Diptera: Simuliidae). Trans. Am. Ento-mol. Soc. 121: 1–57.

Oosterbroek, P. (1995). Phylogeny of the nematocerous families ofDiptera (Insecta). Zool. J. Linnaean Soc. 115: 267–311.

Page, R. D. M. (1998). TreeView ver. 1.5 (distributed by the author).Division of Environmental and Evolutionary Biology IBLS, Univ.Glasgow.

Pawlowski, J., Szadziewski, R., Kmieciak, D., Fahrni, J., and Bittar,G. (1996). Phylogeny of the infraorder Culicomorpha (Diptera:Nematocera) based on 28S RNA gene sequences. Syst. Entomol.21: 167–178.

Procunier, W. S. (1982). A cytological study of species in Cnephia s.str. (Diptera: Simuliidae). Can. J. Zool. 60: 2852–2865.

Pruess, K. P., Zhu, X., and Powers, T. O. (1992). Mitochondrialtransfer RNA genes in a black fly, Simulium vittatum (Diptera:Simuliidae), indicate long divergence from mosquito (Diptera: Cu-licidae) and fruit fly (Diptera: Drosophilidae). J. Med. Entomol. 29:644–651.odriquez, F., Oliver, J. L., Martin, A., and Medina, J. R. (1990). Thegeneral stochastic model of nucleotide substitution. J. Theor. Biol.142: 485–501.othfels, K. H. (1979). Cytotaxonomy of black flies (Simuliidae).Annu. Rev. Entomol. 24: 507–539.othfels, K., and Featherston, D. (1981). The population structure ofSimulium vittatum (Zett.): The IIIL-I and IS-7 sibling species.Can. J. Zool. 59: 1857–1883.

araste, M. (1990). Structural features of cytochrome oxidase. Q.Rev. Biophys. 23: 331–366.

imon, C., Frati, F., Beckenbach, A., Crespi, B., Liu, H., and Flook,P. (1994). Evolution, weighting, and phylogenetic utility of mito-chondrial gene sequences and a compilation of conserved polymer-ase chain reaction primers. Ann. Entomol. Soc. Am. 87: 651–701.

chondrial cytochrome oxidase II. Mol. Phylogenet. Evol. 7: 33–43.

ohn, U. I. K., Rothfels, K., and Straus, N. A. (1975). DNA–DNAhybridization studies in black flies. J. Mol. Evol. 5: 75–85.

perling, F. A. H., and Hickey, D. A. (1994). Mitochondrial DNAsequence variation in the spruce budworm species complex (Cho-ristoneura: Lepidoptera). Mol. Biol. Evol. 11: 656–665.

picer, G. S. (1995). Phylogenetic utility of mitochondrial cytochromeoxidase gene: Molecular evolution of the Drosophila buzzatii spe-cies complex. J. Mol. Evol. 41: 749–759.

ang, J., Pruess, K., and Unnasch, T. R. (1996). Genotyping NorthAmerican black flies by means of mitochondrial ribosomal RNAsequences. Can. J. Zool. 74: 39–46.

empleton, A. R. (1983). Convergent evolution and non-parametricinferences from restriction fragment and DNA sequence data. In“Statistical Analysis of DNA Sequence Data” (B. Weir, Ed.), pp.151–179. Dekker, New York.

eshima, I. (1972). DNA–DNA hybridization in blackflies (Diptera:Simuliidae). Can. J. Zool. 50: 931–940.

addell, P. J., and Steel, M. A. (1997). General time-reversibledistances with unequal rates across sites: Mixing G and inverseGaussian distributions with invariant sites. Mol. Phylogenet. Evol.8: 398–414.

ilkerson, R. C., Parsons, T. J., Albright, D. G., Klein, T. A., andBraun, M. J. (1993). Random amplified polymorphic DNA (RAPD)markers readily distinguish cryptic mosquito species (Diptera:Culicidae: Anopheles). Insect Mol. Biol. 1: 205–211.

illiams, P. L., and Fitch, W. M. (1990). Phylogeny determinationusing dynamically weighted parsimony method. Methods Enzy-mol. 183: 615–626.

illis, L. G., Winston, M. L., and Honda, B. M. (1992). Phylogeneticrelationships in the honey bee (Genus Apis) as determined by thesequence of the cytochrome oxidase II region of mitochondrialDNA. Mol. Phylogenet. Evol. 1: 169–178.

ood, D. M., and Borkent, A. (1989). Phylogeny and classification ofthe Nematocera. In “Manual of Nearctic Diptera” (J. F. McAlpine,Ed.), Vol. 3, pp. 1333–1370. Canadian Government PublishingCentre, Hull, Quebec.

hu, X., Pruess, K. P., and Powers, T. O. (1998). Mitochondrial DNApolymorphism in a black fly, Simulium vittatum (Diptera: Simu-liidae). Can. J. Zool. 76: 440–447.