Paper Obscuras Barrio

7/25/2019 Paper Obscuras Barrio

1/15

Evolutionof theDro sop hila o bscur aSpeciesGroup

InferredfromtheG p d h andSodGenesELADIO BARRIO1 AND FRANCISCO J. AYALA2

Department of Ecology and Evolutionary Biology, University of California, Irvine, California 92717

Received February 28 ,1 99 6; revised July 10 ,1 99 6

The D r osoph i l a obscu r a group consists of severaldozen Nearctic and Palearctic species, with a phylog-

eny that remains largely unresolved in spiteof numer-ousmorphological, cytogenetic, and molecular investi-gations. We havepartially sequenced two genes, G p d h

(about 1000 bp) and S od(about 700 bp) in 1213speciesand 1 subspecies in order to settle the issues. Difficul-ties in resolvingthephylogeny emanatefromthe rapid

sequence of successive radiations. Nevertheless, thefollowing conclusions are warranted: (1) The Palearc-tic speciesincludetwo monophyletic subgroups,su bob -

scu r a (which may be the most ancient clade of thewhole group) and obscu r a ,plus two other species withunresolved phylogenetic positions, D . b i f a s c i a t a and

D . su bsi l vest r i s;(2) the Nearctic species form a mono-phyletic group consisting of two monophyletic sister

clades, the a ffi n i s subgroup and the p seu d oob scu r asubgroup. The Palearctic radiation may have resultedfromadaptation to expanding temperateforests in theOld World. A second radiation occurred during the

colonization of thedeciduous forests of theNew Worldby the descendants of a single lineage that soon splitnto the a ffi n i s and p seu doobscu r a subgroups. r 1997

Academic Press

INTRODUCTION

Th e D r osophi la obscur a group consists of 35 de-scribed species and has been the subject of extensivenvest iga tions concerning genetics, ecology, et hology,

cytology, a nd evolution. Yet the phylogeny an d ta x-onomy of the group remain controversial. Based onmorphological traits, Sturtevant (1942) distinguishedwo subgroups: obscura w i t h Pa le a rc t ic a n d N e a rc t ic

species, a nd af f inis, w i t h m os t ly N e a rct ic s p eciesThrockmorton, 1975). Theobscur agroup a lso includes

a set of African species, mostly discovered after 1985

a nd classified in the m i c r ol a b i s subgroup (Tsacas et al.,1985; Cariou et al., 1988; B rehm a nd Krimbas, 1990,1992, 1993; Brehm et al., 1991; Bachmann et al., 1992;R u t t k a y et al . , 1992), which will not be further dis-

cussed in this pa per.Extensive allozyme investigations support the splitof Sturtevants obscur asubgroup into tw o subgroups, an ew obscura subgroup encompassing the P alearcticspecies, and the pseudoobscura subgroup encompass-ing the Nearctic species (Lakovaara et al., 1972, 1976;Ma rin kovi cet al.,1978; La kovaa ra a nd K era nen, 1980;L a k o v a a ra a n d S a u ra , 1 9 8 2 ; s e e a ls o Ca b re ra et al .,1983; Cariou et al., 1988). These studies further sug-gest that (1) the Nearctic pseudoobscura subgroup isphylogenetica lly closer to the (mostly) Nearctic a f fi n i s subgroup tha n to the P alearcticobscur asubgroup; an d

(2) th eobscur a subgroup may not be monophyletic, buteither polyphyletic or pa ra phylet ic. One or both of th esetw o inferences a re also supported by mitochondrialDNArestriction analysis (Latorreet al.,1988; G onza lezet al., 1990; Barrio et al .,1992) an d sequencing (Rut t-ka y et al., 1992; B eckenba ch et al., 1993; Barrio et al.,1994), DNA-DNA hybridization (Goddard et al., 1990)a nd cytogenetic compar ison of chromosoma l elements(Brehm et al., 1991; B rehm a nd Krimba s, 1990, 1992,1993). Barrioet al., (1994) have thus proposed that theobscura subgroup be split into two, one retaining thena me and most species, and the subobscur a subgroup,

consisting of th e widely distr ibuted D. subobscura a ndtwo island endemics, D . guanche(Ca n a ry Is la n d s ) a n dD . m ad eir ensis(Madeira).

B a r r i o et al. (1994) ha ve averred t ha t t he rema iningphylogenetic uncerta inties concerning the obscuragroupma y be a tt ributa ble, in pa rt, t o th e specia l evolutionar ydynamics of the D r osophi lamitochondrial genome, butit ma y also be tha t successive phylogenetic radia t ionsoccurred very rapidly in the group and thus are difficultt o d i s cer n . I n t h e pr es en t p a per w e t es t t h e r a p idphyletic radiat ion hypothesis and seek to resolve thephylogenetic uncertaint ies, by sequencing segments

from tw o nuclear genes, G p d h a n d Sod. Gpdhencodest h e a-glycerophosphate dehydrogena se, which ha s akey role in insect fl ight, r egenera ting NADH for cellu-

1 Current addr ess: Depar tment de G enetica, Un iversitat de Valen-cia, C/Dr. Moliner, 50. 46100-Va len cia, Spa in.

2 To w hom c or r espond enc e an d r epr int r eq uests should be a d -

dressed. Fa x: 714-824-2474. E -ma il: fja ya [email protected].

M O L E C U L A R P H Y L O G E N E TI C S A N D E V O L U T I O N

Vol. 7, No. 1, February, pp. 7993, 1997

ARTICLE NO. FY960375

791055 -7903/97 $25.00Copyright r1997 by Academic P ressAll rights of reproduction in a ny form r eserved.


2/15

ar respiration through the a-glycerophosphate cycle inhe thoracic flight muscle (Sacktor, 1975). Sodencodesh e Cu , Zn s u perox id e d ism u t a s e , a h igh ly s pe cifi c

enzyme that protects aerobic cells against the toxicityof free oxygen radicals (Fridovich, 1986).

MATERIALSANDMETHODS

Drosophila Species

We use st ra ins from 13 species a nd 1 subspeciesb elon g in g t o t h e D r osophi la obscur a group, mostlyobtained from the National D rosophi la Species StockCenter a t B owling G reen, OH (stock reference numberswill be given for these); the sources of the other strainsa re given below.

Seven species belong to the D . obscur a subgroup, asollows. D . subobscur a subgroup, thr ee species: D .

subobscura from Helsinki, Finland (strain H271 fromValencia, Spain); D. m adeir ensis,endemic in th e Islan dof M a d eira , P ort u g a l ; a n d D. guanche, endemic in theCa na ry Islands, Spain from Tenerife. D. obscura sub-group, four species: D . a m b i gu a (14011-0091.1), fromPort Coquit lam, Brit ish Columbia, Canada; D. obscurarom G irona, S pain; D. subsil vestri sof unkn own origin;

a nd D . bi fasciata(14012-0181.0) from Lake Akan-Ko,J apa n. D . guan che, D. mad eir ensis, D. obscur a, a nd D .subsilvestris were kindly supplied by Dr. Mara Mon-clus, U niversit y of B a rcelona . Thr ee add itiona l speciesbelong to the D . a f f i n i s subgroup: D . a f f i n i s (14012-

0 1 4 1 . 0 ) fro m Cry s t a l L a k e , H a s t in g s , N E ; D. azteca14012-0171.6) from Ma th er, CA; D . tol teca (14012-0201.0) from Coroico, B olivia. Fin a lly, th ree species an don e s u bs pe cies a re m em b ers of t h e pseudoobscurasubgroup: D . m i r a n d a (14011-0101.1) from M a t her, C A;D. p. pseud oobscur a(14011-0121.43) also from Ma th er,C A;D . p. bogotan a(14011-0121.69)f rom B ogota , C olom-b i a ; a n d D . p e r s i m i l i s (14011-0111.2) from Mt. Sa nJ acint o, CA.

D. melanogaster Gpdh(Bewley et al., 1989) and SodKwiatowski et al., 1989) sequences are used as out-

groups. Additional Sod a nd G p d h sequences, used in

some ana lyses, ar e from K wia towski et al.(1994 a nd inpress, respectively).

D N A E xtr act ion, A m pl ificat ion, C loning, and

Sequencing

Th e D rosophi la G pdh gene consists of 8 exons thatencode 3 isoenzymes that arise by differential expres-sion of exons 6 t o 8. The reg ion sequ enced in t he presen tstu dy comprises a bout 1000 bp of exons 3 to 6 (Fig. 1),corresponding to posit ions 3020 t o 4241 in the D .melanogaster sequence (Bewley et al., 1989). The Sodgene consists in D rosophi laof 2 exons a nd one intron.

The region sequenced in this s tu dy comprises about 700bp that include most of the coding region (except the 58a n d 38 ends) and the intron (Fig. 2), positions 438 to

1504 in the D. melanogaster sequence (Kwiatowski etal . ,1989).

G e n om ic D N A w a s p rep a r ed f r om 10 t o 20 fl i esa ccording to th e method of Ka w a saki (1990). B oth gener eg ion s w e r e a m p li fi e d b y P C R u si ng h i gh -fi d el it yconditions (Kwiatowski et al., 1991). Primers for PCRamplification of the G pdhr egion (Fig. 1) were designedb y com p a rin g a v a i la b le s eq u e nce s from ot h e r D r o - sophi la species (Tomina ga et a l . , 19 95 ): G N L , 58-C C C G AC C TG G TTG AG G C TAG C C AAG AATG C -38;G N R , 58-ACATATG C TC AG G G TG C TAG C G TATG C A-38(both contain a N h eI s ite). P rimers for th e Sod region(Fig. 2) are described elsewhere (Kwiatowski et al . ,1994): N, 58-C C TC TAG AAATG G TG G TTAAAG C TG T-NTG C G T-38(cont a ining a X baI site); an d C, 58-C TTG C -TG AG CTC G TG TC C AC C C TTG CC C AG ATC ATC -38(containing a SacI sit e).

PCR p ro d u c t s w e re c le a n e d w it h t h e W iz a rd PCR

FIG. 1. S t r u ct u r e o f t h e G p d h g en e r eg io n a n a l y z ed i n t h epresent study. The black boxes represent exons 3 to 6. The smallar r ows r epr esent pr imer s used f or P CR amplific ation a nd seq uenc-

ing. The amplified segment c or r espond s t o the br ac keted r egionbetween pr imer s M L and N R. Long ar r ows ind ic ate the extent anddirection of sequencing.

FIG. 2. St ructure of the Cu,Zn Sod gene region. The black boxesrepresent t he tw o exons, and th e small arr ows indicate the position ofthe primers used for PCR amplification and sequencing. The ampli-fi ed segment corresponds to the bra cketed region between primers Nan d C. The long arr ows indicate t he extent a nd direction of sequenc-

ing.

80 B ARRI O AND AYALA


3/15

P reps DNA purifi cat ion kit (Promega) a nd cloned intoplasmids. The G p d h region w a s cloned in pU C19 usinghe compatibility betw een the N h eI a n d X baI restric-ion sites present in the P CR primers a nd t he cloning

vector. Endonuclease X baI was included in the ligationreaction, thus increasing approximately 10 t imes the

cloning efficiency. TheS odregion w a s cloned either int opUC 19 by sta nda rd procedures or into the pCRI I vectorrom the TA-cloning kit (Invit rogen) a ccording to the

manufacturers protocol.One clone of each gene wa s sequenced by the S a nger s

dideoxynucleotide sequencing method for denat ureddouble-stra nded plasmid D NA by using t he Sequena sev. 2.0 DNAsequencing kit (United States BiochemicalsAmersh a m). Compressions and a mbiguit ies w ere solvedby multiple sequencing of both stra nds. When singlenucleotide substitu tions a ppea red in only one species,his could be due to mistakes introduced during the

P CR reaction by the Taqpolymera se; ad ditional clonesrom different PCR reactions were then sequenced to

check the presumptive substitutions. Five oligonucleo-ides w ere used t o sequence the G p d h gene r egion (Fig.

1 ) : L 4 BN , 58-C C ATG YG C TG TC TTG ATG G G -38; L 4 E ,58-G ATC TG ATC ACG ACG TG TTAC-38; L5E, 58-C G CG T-C TG AGG C G TTTG T-38; R5B, 58-CTCAGAGACGCGGC-G G TTACG G CC AC-38; an d R4M, 58-ACAGCCGCCTTG-GTGTTGTCGCC-38. Four oligonucleotides were usedor sequencing th e Sodgene r egion (Fig. 2): I N, 58-G A-

CATGCAKCCGTTRGTGTTG-38; I R , 58-GACAACACCA-AYG G CTG CATG TC-38; M L , 58-TG G AATTGATG AATAT-

TG C-38; and MR 5

8-G AGC TG CG CACTG TTATTS G AC-3

8.

n a ddition, we used the st a nda rd M13/pUC sequenc-ng oligonucleotides.

The Genbank accession numbers for the G pdh a n dSod sequences a re a s follow s (the t wo numbers a re forG pdha nd Sod, respectively): a f fi n i s (U47874, U47879);a m b i g u a (U47880, U47868);azteca (U47875, U47866);bifasciata(U47883, U47869);guanche(U47878, U47889);madeirensis (U 47890, U 47887); m i r a n d a (U47882,U47870); obscura (U47881, U47892); persimilis(U47886,U47873); p. bogotan a (U47891, U47872); p. pseudoob-scura(U 47885, U 47871);subobscura(U 47877, U 47888);

subsilvestris (U47884, G p d h only); tolteca (U47876,U47867).

Phylogenetic Inference

Sequences were a ligned using t he C LU STAL V pro-g ra m (H ig g in s a n d S h a rp , 1 988). F or p h y log en et ica na lysis, we used th e neighbor-joining (NJ ) (Sa itou an dNei, 1987), ma ximum-par simony (MP ) (Fitch, 1971),a nd ma ximum-likelihood (ML) (Felsenstein, 1981)m et h od s . N J t r ees a n d t h ei r s t a t i st i ca l t e st s w e rep erform ed u s in g t h e com pu t e r p rog ra m M E G A 1. 0

Kumar et al . , 1993). The statistical confidence of a

p a rt icu la r clu st e r o f s e q u en ces in t h e N J t re es w a sevalua ted by the bootstra p test (1000 pseudorepli-cates). B ootst ra p va lues were corrected, wh enever pos-

sible, by using th e complete-a nd-par tia l bootstr a p (CP B )technique (Li and Zharkikh, 1995). The MP trees andtheir bootstrap tests (1000 pseudoreplicates), and theML trees, were obta ined using the progra ms DNAPARS ,SE QB OOT, an d D NAML, respectively, implemented inthe P HYLIP packa ge 3.5 for Windows (Felsenstein,

1993). Alternative trees were compared by the maxi-mum par simony t est proposed by Templeton (1983) a nddeveloped by Felsenstein (1985), and by the maximumlikelihood test of Kishino and Hasegawa (1989). Theset e s t s w e re p e rfo rm e d w it h t h e p ro g ra m s DN A PA R Sa nd D NAML, respectively, from the P HYLI P package.

RESULTS

Phyl ogenetic An al yses of Gp dh Sequences

The par tia l sequences of th eG p d h gene corr espond t oa r egion th a t ra nges in size from 952 bp, inD . aff ini s,t o

1010 bp, inD . subsil vestr is,distr ibuted a s follows (Fig.1): 168 bp from exon 3, 66104 bp from in tr on II I; 373bp from exon 4, 5474 bp from intron IV, 154 bp fromexon 5, 6481 bp from intron V, and 64 bp from exon 6.

Due t o the difficulty of un ambiguously a ligning t heG p d h intron sequences from the obscura species withthose from the outgroup, we examine separately thecoding sequences and the intron sequences. D. m el ano-gaster is only included in the alignment of the codingsequences.

The 15 G pdh coding sequences yield 759 nucleotidepositions aligned for all species. Nucleotide polymor-

phisms occur a t 152 (20.0%) positions, of w hich 78(10.3%) a re phylogenetica lly informa tive. Nine va ria blesites correspond to first codon positions (4 are informa-tive), 142 to thir d codon positions (74 informa tive), an donly one to a second position. Only fi ve sites (1 fi rst, 1second, a nd 3 th ird codon positions) show nonsynony-mous substitutions, corresponding to replacements infour amino acid sites (two nonsynonymous substitu-tions occur a t t he fi rst a nd second position of the sa mea mino a cid site). There a re no a mino a cid indels. Thereis not much bia s in G1C content: the overall nucleotidefreq uencies a re 0.250 A, 0.255 T, 0.230 C, a nd 0.265 G ,

a lthough t here is some excess of G1C content at thirdcodon positions (60.8%). G iven t he a bsence of subst a n-tive composit ion bias , we est ima te nucleotide diver-gence according to Kimuras (1980) tw o-para metermodel (Ta ble 1, a bove dia gona l). The phy logenetic t reederived by t he neighbor-joining m ethod is shown in Fig.3A. Th e fi r s t b r a n ch in g of t h i s N J t r ee s h ow s t w omonophyletic subgroups, a f f i n i s a nd obscura, consis-tent with the classification proposed by Sturtevant in1942. The obscura subgroup splits into two monophy-letic complexes, one corresponding to the Nearctic andth e other to the P a learctic species. The next bra nching

splits the subobscura t riplet from the rest of the OldWorld obscur aspecies. These bra nchings a re supportedby high bootstra p values.

81EVOLUTION OF G p d h AND Sod I N Drosophila


4/15

Sy nonymous substit utions, however, ca n become sa tu-rated, which will introduce bias in a phylogeny whenmost substitut ions a re synonymous (presumably owingo strong purifying selection a ga inst a mino acid repla ce-

ments), a s is the case in the obscur agroup. One escaperom this bias is to rely exclusively on conservative

changes (Swofford and Olsen, 1990; Hillis et al., 1993),ha t is, in our case, tra nsversions, which accumulat e a t

s low e r r a t e s t h a n t r a n s it i on s (B r o w n et a l . , 1982;Moritz et a l . , 1987). (Nonsynonymous substitut ions

a l s o a c cu m ul a t e a t a s low e r r a t e t h a n s y non y m ou ss u bs t i t u t ion s , b u t on ly s ev en , d ist r ib ut e d ov er fi v esites, occur in the G p d h d a t a s et , w h ich m a k es t h e mnsufficient for the present purposes.) The nucleotide

distances based on tra nsversions, a ccording to K imu-ra s (1980) tw o-para meter model, ar e presented inTa ble 1, below t he dia gona l. The NJ tr ee derived fr omhese va lues is shown in Fig. 3B . This t ree reverses th e

sequence of the fi rst t hree bra nchings: fi rst t he subob-scura tr iplet splits from t he rest (CP B va lue of 90%),h e n t h e a f f i n i s s u bg r ou p, a n d fi n a l ly t h e N ea r ct i c

pseudoobscura s u b grou p s plit s f rom t h e re ma in d er

Pa le a rc t ic obscura. H o w e ve r, t h e t h re e s u bg rou p s,affini s, obscur a, a nd pseudoobscura, appear practicallya s a polyt omy (CP B betw een 40 an d 77%).

Differences in the tra nsit ion/tra nsversion ra t io a remport a nt for inferring phylogenetic trees by t he ma xi-

m u m l ik el i h ood m e t h od . Th e a v e r a g e t r a n s i t i on /r a n s v er s ion r a t i o f or t h e obscura species is 2.12,

ra nging from 0.5, betw een D. obscuraa n d D . subsi l ves-t r i s , to 5.0, between D . a f f i n i s a n d D. tolteca. I f w econsider very closely relat ed ta xa (beca use they betterre fl ect t h e t ru e ra t io; B row n et al., 1982; Hillis et al.,1993), the range is about the same: a rat io of 0.7 for

subobscuraa n d m adeir ensis,2.0 for pseudoobscuraa n dpersim i l is, a nd, a s noted, 5.0 for aztecaa n d tolteca. Wehave, therefore, obtained ML trees using a variety of

tr a nsition/tr a nsversion ra tios, including 2.1, t he a ver-a g e o b s e rv e d in o u r d a t a s e t . T h e t re e b a s e d o n t h ea vera ge ra t io (Fig. 4A) shows unlikely rela t ionships,which also appear in other ML trees based on rat ioss m a lle r t h a n 5 . W h e n ra t io s $5 . 0 a re u s e d , t h e M Ltr ees exhibit t he topology depicted in Fig. 4B . As in F ig.3B, thesubobscur atr iplet diverges fi rst; but in contr a stw i t h t h a t t r ee, t h e N ea r ct i c s u b gr ou ps a f f i n i s a n dpseudoobscuraappear a s s ister taxa .

Maximum parsimony a na lysis (da ta not shown) yieldsimpla usible t opologies, s imilar to the ML trees ob-ta ined with 2.1 or other low tra nsit ion/tra nsversionratios. Thus, the MP tree shows melanogaster w i t h int h e a f f i n i s s u b g ro u p , a s a s is t e r t a x o n o f D. tolteca,although with low bootstrap values, suggesting exces-sive h omoplas y.

Th e a n a l y s es b a s ed on on ly t h e i n t ron (F ig . 5)sequences (which do not include the outgroup D . m ela- nogaster) s h ow t h e t w o P a le a rct ic s u b grou ps (subob-scuraa nd obscur a) as a pair of s ister clades and the tw oNearctic subgroups (a f f i n i s a n d pseudoobscura) a s a n -

ot h e r p a ir. Th es e re la t ion s h ip s a re fa ir ly s im ila r t oth ose in the ML tr ee (Fig. 4B), and in t he tra nsversion-only NJ tree (Fig. 3B), if we ignore the low-bootstraprelat ive positions of the obscur aa n d a f fi n i s subgroups.

From the different analyses based on the two sets ofa lignments, w e conclude tha t the 14 obscura t a x a a r egrouped into four significantly monophyletic clusters:t h e subobscur a, obscur a, aff ini s, a n d pseudoobscuras u bg rou ps . We m a y n ot e t h a t D . b i f a sci a t a a nd D .subsi lvestr is t y pi ca l l y cl us t er w i t h t h e a m b i g u a obscura pair, but this is not well supported in mostcases by bootstra p values.

We ha ve car ried out t wo a ddit ional s ta t is t ical testsseeking to resolve phylogenetic relationships left un-sett led by the previous an a lyses (Fig. 6). These tests a re

TABLE 1

Pairwise Nucleotide Divergences for All Substitutions(Upper Triangle), or Only Transversions (LowerTriangle) between G p d h Coding Sequences Estimated According to Kimuras (1980) Two-Parameter Model

S pecies 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1. melanogaster . 16 82 . 17 05 . 168 5 . 1 662 . 1613 . 1659 . 1662 . 1675 . 1604 . 1503 . 1661 . 1677 . 16 94 . 16 962. subobscura . 0412 . 0066 . 0120 . 0365 . 0323 . 0296 . 0379 . 0666 . 0665 . 0814 . 0550 . 0593 . 0578 . 0550

3. madeirensis . 0369 . 0040 . 0133 . 0379 . 0337 . 0323 . 0393 . 0694 . 0679 . 0829 . 0564 . 0607 . 0592 . 05644. guanche . 0383 . 0026 . 0040 . 0324 . 0282 . 0269 . 0310 . 0636 . 0622 . 0800 . 0507 . 0550 . 0536 . 05075. a m b i g u a . 0426 . 0147 . 0161 . 0120 . 0093 . 0160 . 0200 . 0520 . 0506 . 0696 . 0338 . 0380 . 0366 . 03386. obscura . 0397 . 0147 . 0161 . 0120 . 0026 . 0120 . 0187 . 0505 . 0491 . 0681 . 0352 . 0394 . 0380 . 03527. subsi lvestris . 0455 . 0147 . 0161 . 0120 . 0080 . 0080 . 0146 . 0548 . 0534 . 0738 . 0366 . 0380 . 0366 . 03668. bi fasciata . 0426 . 0174 . 0188 . 0147 . 0107 . 0107 . 0080 . 0592 . 0591 . 0798 . 0394 . 0436 . 0422 . 0394

9. a f f i n i s . 0469 . 0188 . 0229 . 0188 . 0202 . 0202 . 0229 . 0202 . 0133 . 0325 . 0492 . 0535 . 0520 . 04920. azteca . 0484 . 0202 . 0215 . 0174 . 0188 . 0188 . 0215 . 0215 . 0040 . 0242 . 0492 . 0535 . 0521 . 04921. tolteca . 0469 . 0215 . 0229 . 0188 . 0174 . 0174 . 0229 . 0229 . 0053 . 0040 . 0623 . 0696 . 0681 . 06522. m i r a n d a . 0440 . 0161 . 0174 . 0134 . 0093 . 0093 . 0120 . 0120 . 0161 . 0147 . 0161 . 0093 . 0079 . 00533. p. pseud oobscur a . 0455 . 0174 . 0188 . 0147 . 0107 . 0107 . 0107 . 0134 . 0174 . 0161 . 0174 . 0040 . 0013 . 0040

4. p. bogotana . 0455 . 0174 . 0188 . 0147 . 0107 . 0107 . 0107 . 0134 . 0174 . 0161 . 0174 . 0040 . 0000 . 00265. p e r s i m i l i s . 0440 . 0161 . 0174 . 0134 . 0093 . 0093 . 0120 . 01 20 . 016 1 . 014 7 . 0161 . 0026 . 0013 . 0013



5/15

h e ma ximum pa rsim ony (MP ) of Templeton (1983), a sm od ifi e d b y F e ls en s t ein (1993 ), a n d t h e m a x imu mikelihood (ML) of Kish ino a nd H a sega w a (1989), which

compar e phylogenies a s w holes w ith one another. One

qu estion is th e position ofD. subsil vestr is(s in Fi g. 6),w h i ch i n ou r a n a l y s es a p pea r s a s s oci a t ed w i t h D .bifasciata (b in Fig. 6), but ha s been previously

(Lakovaara et al.,1972; G onza lez et al., 1990; B a rrioetal . ,1994) clustered w ith th e couplet ofD . a m b i g u a a n dD. obscura (ao in Fig. 6). The other question is thebra nching sequence or topologica l confi gura tion of the

main clusters: a f fi n i s (A in Fig . 6, consis ti ng ofaf f ini s,azteca, a n d tolteca), subobscura (S, with subobscura,madeirensis, a n d guanche) , a n d pseudoobscura (P,

FIG. 3. Neighbor-joining trees ba sed on Gpdh coding sequences using all subst itut ions (A) or only tra nsversion (B ). B ra nch lengths a reproportional to the scale given in substitutions per nucleotide. Percentage bootstrap values and complete-and-partial bootstrap values (inparentheses) based on 1000 pseudoreplicates are given on the nodes.



6/15

w it h p. pseudoobscura, p. bogotana, a n d p er s i m i l i s ), asw ell as th e four species just m entioned.There are 30 possible unrooted trees relating these

groupings. Figure 6 displays t he six confi gura tions tha ta re sta tist ica lly superior t o the other 24; but th ese six,however, are not s ignificantly different from one an-other. This result remains the same whether we useonly the coding sequences, only the introns, or bothcombined (wh ich can be a ccomplished in t he case of MPowing to the assumption of chara cter independence).Th e s ix t r ees a l l s h a r e t h e a s s oci a t i on of t h e t w oNearctic subgroups, a f f i n i s a nd pseudoobscura, bu t

eave unresolved w hether D . subsil vestr isclusters w ithD. bifasciataor with t he obscur aam bi guacouplet (seeFig . 7).

The number of possible trees is 210 if we add D .

melanogaster as an outgroup to the 30 unrooted trees,

since the root can be in any of the seven branches ofeach tree. The ML test yields 48 trees superior to therest, but n ot signifi cant ly different from one another. Ofthese 48 t rees, 42 correspond to the six topologies

FIG. 4. Ma ximum-likelihood t rees derived from G pdh codingequences. The t ra nsition/tra nsversion ra te is 2.1 in A, but 5 in B .

The scale is for substitutions per site.

FIG.5. Neighbor-joining tree based on Gpdh in tron sequences. Bra nch lengths a re proportional to the scale given in substitut ions per nucleotide.Parent bootstrap values and the complete-and-partial bootstrap values (in parentheses)based on 1000 pseudoreplicates are given on the nodes.

FIG. 6. Six Gpdh topologies that are statistically superior to theother possible 24 t opologies, a ccording t o ma ximum parsim ony (Temple-ton 1983) or maximum likelihood (Kishino a nd Ha segawa 1989). Thelineages represented are the subgroups: aff inis(A), pseud oobscur a (P),

a nd subobscura (S) and the species D. ambigua (a), D. obscura (o), D .subsilvestris(s), andD. bifasciata(b). These six trees a re not signifi cant lydifferent, according to either test applied to the coding, the intron, or thetotal (intron 1 coding) sequences. The ML test based on t he codingsequences including D. melanogaster a s outgroup yields a number ofsuperior but indistinguishable trees (of 210 possible), namely those inwhich the position of the root is indicated by a black dot. Also used as

outgroup sequences areD. nebulosa(n), D. paulistorum (p), D. simulans

(s), D . t eissieri (t), and D. w i l l istoni (w). Trees tha t sh a re a root positionindicat ed by a sma ll line and letters are stat istically equivalent. Ast ar isused when all the outgroups root equivalent trees at the same position.



7/15

shown in Fig. 6, where the position of the root, deter-mined by th e outgr oupD . m el an ogaster,is indica ted bya black dot. The MP test yields 34 superior equiva lentrees, 28 of which correspond to the six t opologies

s h ow n in F ig . 6 . We re pea t e d t h e M P t e st u s in g a sou t g rou p ea c h of fi v e d iff er en t s peci es of t h e S o-phophora subgenus (the subgenus to which the obscur a

species belong), namely D. nebul osa, D. paul istoru m, D.sim ul ans, D. teissieri ,a n dD . w i l l i st on i (sequences fromKw ia t o w s k i et al., in press). Most of the stat is t icallyequivalent tr ees a ga in correspond to t he six t opologiesdepicted in Fig. 6.

Compositional bias may influence phylogenetic topol-ogy (Saccone et al., 1993). In our sequences composi-ional bias appea rs primarily in the third-codon posi-ions, which is where most of the substitutions occur.

We have used Saccone et al.s (1990, Eq. 21, p. 576)x2

est for stationarity that ascertains whether nucleo-ide frequencies a t equivalent codon positions a re con-

stant in the extant species and their ancestors. The x2va lues a re show n in Ta ble 2. Va lues a bove 1.5 (shown inbold) a re thought t o violat e the sta tionar ity conditions.

On the whole, the stationarity condition is satisfied fort h e obscura group species. The only exception is D .tolteca, w h ich m a y a c cou n t for t h e p os it ion of t h isspecies (or the a f fi n i s subgroup) as the fi rst branchingclade in some trees, a possible consequence of conver-gence with outgroup species.

Phylogenetic Analyses of Sod SequencesSod sequences were obta ined from 12 D . obscur a

species an d 1 subspecies, the sa me for w hich the G p d h region was sequenced except for D. subsilvestris. Th esequences in clude 42 bp (of 66) of t he coding r egion ofthe first exon, the intron (320397 bp), and 300 bp (of396) of th e second exons coding r egion (Fig. 2). As in th ecase of G p d h ,intr on divergence betw een D. melanogas-t er (725 bp) and the obscura species is much too higha nd t he a lignment becomes uncert a in. Thus, a s before,we separately consider coding sequences and intronsequences. The outgroup D. melanogaster is only in-

cluded in the coding sequence alignment.The alignment of the 14 Sod coding sequences con-

sists of 342 nucleotide sites, of w hich 91 (26.6%) a re

FIG.7. S ix G p d h unrooted topologies tha t a re not significan tly different a ccording to the MP a nd ML tests, obta ined using total sequencesby th e maximum -likelihood meth od. Bra nch lengths a re proportional to t he scale given in subst itut ions per nucleotide.



8/15

va ria ble a nd 45 (13.7%) ar e informa tive. S ixteen va ri-a ble sites ar e fi rst codon positions (6 a re informat ive),13 a re second positions (4 informa tive), and 62 a re th irdpositions (35 informa tive). Twent y-six sites (13 fi rsta nd 13 second codon positions) show nonsynonymoussubstitutions, corresponding to replacements in 20a mino acid sites. There a re no a mino acid indels.

In contra st to Gpdh , Sodsequ ences ha ve an excess ofG1C (overa ll content 80%), wh ich is part icular ly la rgeor third codon positions (where the adenine is particu-arly scarce; i ts frequency is 28.4, 31.5, and 5.5% at

fi rst, second, a nd thir d position, respectively). B eca useof this composition bias, we h a ve estimat ed nucleotidedivergence not only according to Kimuras (1980) two-pa ra meter model (Ta ble 3, above the dia gona l), but a lsoa ccording to Ta jima a nd Nei (1984), Ta mur a (1992),

a n d Ta m u ra a n d N ei (199 3). We h a v e u s ed t h e seva rious dista nces for constr ucting phy logenetic trees bythe NJ method, but the result ing topologies were notdifferent from those based on Kimuras method andth us will not be display ed.

F igu re 8A re p res en t s t h e N J t re es d erived fromlea s t -s q u a re s e st im a t e s of b ra n ch len g t h s b a s ed onKimuras distances. The D. obscura group species areclustered in four divergent lineages: the couplet D .am bigu aD. obscur a, t he D . b i f a sci a t a l in ea g e , t h emonophyletic subobscura subgroup, a nd a monophy-let ic clu st e r form ed b y t h e t w o N ea rc t ic s u b grou psa f f i n i s a n d pseudoobscura,each of wh ich forms , in tur n,a monophyletic cluster. The branching sequence ofth ese four lineages is, however, uncert a in a ccording tot h e CP B t e s t .

TABLE 2

Pairwise Results of theStationarityx2-Test of Saccone e t a l .(1990) Measuring Similaritybetween Compositional Patterns at Third-Codon Positionsin G p d h

S pecies 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

1. melanogaster2. subobscura 12.1573. madeirensis 12.942 0.0164. guanche 10.590 0.140 0.1865. a m b i g u a 8.745 0. 468 0. 538 0 .1756. obscura 9.405 0. 277 0. 333 0 .132 0. 0377. subsi lvestris 8.720 0. 319 0. 412 0 .117 0. 03 7 0. 0278. bi fasciata 8.704 0. 586 0. 665 0 .175 0. 046 0. 144 0. 10 29. a f f i n i s 6.258 1. 031 1. 238 0. 537 0. 304 0. 443 0. 265 0. 2460. azteca 5.878 1. 114 1. 352 0. 656 0. 429 0. 549 0. 343 0. 390 0. 0201. tolteca 2.908 3.232 3.684 2.690 2.139 2.302 1.911 2.177 0.974 0.7402. m i r a n d a 7.423 0. 555 0. 723 0. 271 0. 180 0. 211 0. 088 0. 196 0. 082 0. 098 1.2503. p. pseud oobscur a 8.439 0. 533 0. 654 0. 148 0. 148 0. 231 0. 122 0. 065 0. 154 0. 246 1.845 0.0994. p. bogotana 8.022 0. 572 0. 711 0. 188 0. 157 0. 239 0. 116 0. 086 0. 102 0. 174 1.633 0.060 0.0065. p e r s i m i l i s 8.451 0. 482 0. 615 0. 143 0. 190 0. 249 0. 126 0. 116 0. 157 0. 226 1.754 0.07 6 0 .0 09 0 .0 09

N ote.Values a bove 1.5 (bold num bers) do not sa tisfy t he sta tionar y condition.

TABLE 3

Pairwise Nucleotide Divergences between S od CodingSequences for All Substitutions(Upper Triangle)

or Only Transversions (Lower Triangle) Estimated Accordingto Kimuras (1980) Two-Parameter ModelS pecies 1 2 3 4 5 6 7 8 9 10 11 12 13 14

1. melanogaster . 1948 . 2145 . 1908 . 1677 . 1751 . 1905 . 2292 . 2295 . 2373 . 2101 . 2142 . 2057 . 20972. subobscura .0722 .0208 .0208 .0610 .0642 .0644 .0968 .1003 .1104 .0609 .0641 .0673 .06413. madeirensis .0790 .0118 .0360 .0706 .0738 .0741 .1000 .1035 .1136 .0705 .0737 .0770 .07374. guanche .0722 .0059 .0179 .0609 .0641 .0643 .0900 .0934 .1034 .0737 .0770 .0802 .0770

5. a m b i g u a .0689 .0270 .0332 .0332 .0148 .0457 .0772 .0806 .0904 .0611 .0676 .0644 .05806. obscura .0722 .0301 .0364 .0364 .0089 .0487 .0738 .0772 .0870 .0643 .0708 .0676 .06117. bi fasciata .0756 .0209 .0270 .0270 .0059 .0089 .0807 .0842 .0941 .0645 .0645 .0614 .06148. a f f i n i s .1000 .0491 .0556 .0556 .0332 .0364 .0270 .0088 .0208 .0577 .0641 .0610 .05469. azteca .0965 .0459 .0524 .0524 .0301 .0332 .0240 .0029 .0148 .0610 .0674 .0643 .05780. tolteca .1036 .0524 .0589 .0589 .0364 .0395 .0301 .0089 .0118 .0705 .0771 .0739 .0674

1. m i r a n d a .0825 .0332 .0395 .0395 .0240 .0270 .0179 .0332 .0301 .0364 .0059 .0088 .0029

2. p. pseud oobscur a .0825 .0332 .0395 .0395 .0240 .0270 .0179 .0332 .0301 .0364 .0000 .0088 .00883. p. bogotana .0859 .0364 .0427 .0427 .0209 .0240 .0148 .0301 .0270 .0332 .0029 .0029 .00594. p e r s i m i l i s . 0859 . 0364 . 0427 . 0427 . 0209 . 0240 . 0148 . 0301 . 0270 . 0332 . 0029 . 0029 . 0000



9/15

To overcome t he potentia l problem of sat ura tion w eha ve relied on only tr a nsversions (Ta ble 3, below thediagona l). The NJ tr ee derived from t he tra nsversiondivergences is shown in Fig. 8B. This tr ee a ga in shows

h e t w o N ea rc t ic s u b g rou ps a s m on op h y let ic s is t erclades. It differs from th e previous (Fig. 8A) tr ee in t ha tt shows the subobscur a subgroup as the first clade to

diverge, the obscur asubgroup as a para phyletic taxonw it h t w o l in e a g e s (am bi gu aobscur a a nd bifasciata),a nd t he monophyletic Nea rctic subgroups as a monophy-letic cla de, but w ithout high CP B va lues.

In the case of protein-encoding genes, i t is oftenuseful to dea l separ a tely with syn onymous and nonsyn-on y m ou s s u bs t i t u t ion s , b eca u s e t h e la t t e r a re m ore

FIG. 8. Neighbor-joining t rees based on Sodcoding sequences, using all substitutions (A) or only transversions (B). Branch lengths areproportional to the scale given in substitutions per nucleotide. Percentage bootstrap values and complete-and-partial bootstrap values (in

parentheses) on the nodes are based on 1000 pseudoreplicates.



10/15

ikely t o be selectively constr a ined a nd thu s t he phylo-genetic s ignal is less likely to be erased by superim-posed or back substitutions. There are 26 polymorphicnonsynonymous sites in th e S oddat a set , and a total of45 nonsynonymous substitutions in those sites, whichm ig h t w a r r a n t t h ei r s epa r a t e u se f or p hy log en et i c

purposes (only 5 nonsynonymous polymorphic sitesappear in th e G p d h da ta set , with a total of 7 substitu-ions, which are obviously insufficient). Figure 9 shows

the NJ tr ees, obta ined by the unw eight ed method of Neiand Gojobori (1986), based on the nonsynonymous (9A)an d synonymous (9B ) dista nces r eported in Table 4.B oth NJ tr ees yield impla usible topologies. The synony-mous topology shown in Fig. 9B is very similar to theone in Fig. 8A, ba sed on w hole dista nces. The nonsyn-

onymous t ree (Fig. 9A) has low bootst ra p va lues ow ingt o t h e re du ce d n u m be r of s u bs t i t u t ion s , b u t s h ow sguanchea n d subobscur a(D. subobscurasubgroups) a s

FIG. 9. Neighbor-joining Sod trees based on nonsynonymous substitutions (A) or synonymous (B) substitutions. The scales for branchengths a re substitut ions per nonsynonymous (A) or synonymous (B ) position. Percenta ge complete bootst ra p values on the nodes ar e based on

000 pseudoreplicat es.



11/15

he fi rst an d second species to diverge from the otherobscuraspecies.

The search for MP tr ees genera ted 6 equa lly par simo-nious trees (results not shown), w hich ha ve similaropologies. I n a ll six t rees,D . obscur a a nd D . a m b i g u a

ar e the fi rst to diverge, but they form a monophyleticcla de in only t wo t rees. A second ra dia tion gives rise tohree lineages corresponding to D . bi fasciata, t he D .

subobscura subgroup, an d t he monophyletic cluster ofhe t w o Nea rctic subgroups.

We have performed the ML analysis on 1st-plus-2ndand 3rd codon positions separately. The 1st-plus-2ndanalysis yields a polytomous tree (Fig. 10A), showingguanchea n d subobscur a as the first diverging species,

u s t a s in F ig . 9 B. T h e M L t re e b a s e d o n 3 rd c o d o npositions (Fig. 10B) shows the unlikely relationshipsa lso ma nifest in th e synonymous NJ tr ee (Fig. 10B ).

Phylogenetic analyses of the intron sequences im-

p rov e a c cu ra c y, b u t t h e la ck of a re lia b le o u t grou psequence makes moot the posit ion of the root in thetrees. Phylogenies obtained by different methods allresult in similar topologies (Fig. 11). The two monophy-letic Nearctic subgroups, a f f i n i s a n d pseudoobscura,form in tur n a consistently monophyletic cluster in t heMP (100%bootstr a p replicat ions) a nd NJ (100%) trees.The species pair am bi gua obscur a a n d t h e D. subob-scurasubgroup are monophyletic clades (98100%boot-strap). D . bi fasciata c lu s t e rs w it h t h e D. subobscurasubgroup with low r elia bility on the basis of the intr onseq uences (CP B , 47%).

Alternat ive trees representing all the branching or-ders a mong subgroups a nd lineages can be comparedby the tests of Templeton (1983) a nd of Kishino a ndHasegawa (1989). We tested different topologies of theset of 6 clades ma nifested by G p d h , t h a t is , t h e af f ini s,pseudoobscura, a n d subobscur asubgr oups, the coupletam bigu aobscur a, D. bifa scia ta, and the outgroup D .melanogaster. There a re 15 possible unr ooted t rees a nd105 rooted trees. Of the 15 unrooted trees, 3 (Fig. 12)

a re superior to the others but not signifi cant ly differenta mong them, a ccording t o either t he MP or ML t est fort he Sod intron alignment, and also for t he Sod codingregion excluding the outgroup. The three trees canjoint ly be r epresented a s t he polyt omous unrooted t reedepicted in Fig. 13.

When D. melanogaster is included as an outgroup,t h e re a re 39 t re es (of 105) s t a t is t ica l ly eq u iva le n ta ccording to Kishino and H a segaw a s ML test, and only10 accordin g t o Templetons MP test . Fift een ML t reesa nd a ll 10 MP t rees correspond to rooted t rees derivedfrom t he 3 topologies depicted in Fig. 12, by loca tin g t he

r oot on a n y b r a n ch (M L t es t ) or on t h e b r a n ch esconnecting the Palearctic lineages (MP test). The otherML trees show either t he am bi gua obscur apair or the

TABLE 4

J ukes-Cantor Corrected Proportion of Synonymous (Upper-Right Matrix) and Nonsynonymous(Lower-Left Matrix) Divergences between S od Coding SequencesEstimated

According to Nei and Gojoboris (1986) Unweighted Method

S pecies 1 2 3 4 5 6 7 8 9 10 11 12 13 14

1. melanogaster . 9448 . 9448 1. 029 . 8071 . 8621 . 8071 . 6184 . 5654 . 7786 . 8621 . 9023 . 8240 . 8621

2. a f f i n i s .0965 .0382 .0646 . 4015 . 4009 . 3403 . 2849 . 2856 . 2948 . 1819 . 2142 . 1979 . 16623. azteca .0965 . 0000 .0381 . 4230 . 4224 . 3602 . 3033 . 3041 . 3135 . 1979 . 2309 . 2142 . 18194. tolteca .0966 . 0077 . 0077 .4888 . 4881 . 4207 . 3597 . 3606 . 3708 . 2468 . 2820 . 2642 . 22985. subobscura .0729 .0233 .0233 .0233 .0515 .0787 .2672 .2678 .2859 .2658 .2838 .3022 .28386. madeirensis .0879 .0273 .0273 .0273 .0116 .1067 .3033 .3041 .3231 .3018 .3206 .3399 .3206

7. guanche .0687 .0272 .0272 .0273 .0038 .0155 .2495 .2502 .2678 .3210 .3403 .3602 .34038. obscura .0771 .0193 .0193 .0194 .0115 .0154 .0154 .0386 . 1840 .2668 .3033 .2849 . 24929. a m b i g u a .0770 .0233 .0233 .0233 .0077 .0115 .0115 .0077 .1845 .2675 .3041 .2856 .24990. bi fasciata .0728 .0252 .0252 .0253 .0077 .0115 .0115 .0115 .0077 .2765 .2765 .2586 .25861. m i r a n d a .0836 .0233 .0233 .0233 .0077 .0116 .0116 .0115 .0077 .0096 .0253 .0382 .01252. p. pseud oobscur a .0836 .0233 .0233 .0233 .0077 .0116 .0116 .0115 .0077 .0096 .0000 .0382 .0382

3. p. bogotana .0836 .0233 .0233 .0233 .0077 .0116 .0116 .0115 .0077 .0096 .0000 .0000 .02534. p e r s i m i l i s . 0836 . 0233 . 0233 . 0233 . 0077 . 0116 . 0116 . 0115 . 0077 . 0096 . 0000 . 0000 . 0000

FIG. 10. Maximum likelihood Sod trees derived from 1st-plus-2nd (A) or 3rd (B) codon positions . The t ra nsit ion/tr a nsvers ion ra tios 1.68 in A and 1.26 in B . The scales a re for subst itut ions per codon

position.



12/15

subobscura s u bg rou p a s t h e fi r s t cl a de t o d iv er geresults not shown). Similar results are obtained when

other outgroup sequences are used, namely, D . s i m u - lans, D . sal tans, or D . w i l l i st on i (sequence data fromKw ia t o w s k i et al . , 1994). Most rooted tr ees derivedrom the 3 un rooted t rees connecting th e sam e set of 4

clades (Nearctic species, subobscurasubgroup, ambiguaobscura

pair, an dD. bifascia ta

) ar e sta t is t ically equiva-ent, a ccording to Templetons MP test. The u nusua l

result showing D . a m b i g u a a s closer to th e out groups ins om e t re es m a y b e a n a r t i fa c t d u e t o con v erg en ce.However, this cannot be due to differences in the Sodcompositional constraints among the species of the D .obscuragroup, beca use Sod third codon positions sat-isfy the sta t iona rity condit ion a ccording to S accone etal .s test (result s not sh own).

DISCUSSION

The phylogenetic relat ionships of the D . obscur ag rou p s pe cies h a v e b een in t en s iv ely s t u d ie d u s in gmolecular an d other approaches. Restrict ion an alysis

FI G. 11. Neighbor-joining t rees based on Sod intron sequences. Branch lengths are proportional to the scale given in substitutions pernuc leotid e. P er centage bootstr ap values a nd the c omplete-and -par t ial bootstr ap values ( in par entheses) on t he n odes a r e ba sed on 1 00 0pseudoreplicates.

FI G. 12. Three S od topologies that are statistically best, accord-ng to ma ximum pa rsim ony (Templeton 1983) or maxim um likelihoodKishino and Hasegawa 1989). Species, subgroups, and other sym-

bols a re a s described in th e legend of Fig. 6, except th at s a l t a n s (1) isnow used a s outgr oup, but nebulosa, paulistorum, a n d t eissier i a r e

not.

FI G. 13. P olytomous tree tha t incorporates the three equiva lentunrooted topologies of Fig. 11. This unrooted tree is obtained usingt o t a l Sod seq uences by the maximum-likelihood method . B r anc hl en g t h s a r e p r op or t i on a l t o t h e s ca l e g iv en i n s u bs t i t u t io n s p e r

nucleotide.



13/15

a nd sequencing of mitochondria l DNA ha ve been usefulor deciphering phylogenetic relat ionships w ithin sub-

groups but not between subgroups. Barrio et al . (1994)h a v e s u gg es t ed t h a t t h e d if ficu lt ie s in s olvin g t h ephylogeny of the subgr oups could be due eith er to th eirra pid phyletic ra diat ion, or to th e distinctive evolution-

a ry d y n a m ics of t h e m it och on d ria l g en om e in D r o - sophi la.

We ha ve now a na lyzed tw o nuclear gene regions, S oda n d G p d h , both of which yield consistent phylogeneticnterpreta tions. The obscur agroup a ppea rs a s a mono-

p h y le t ic cla d e con s is t in g of s ev era l l in ea g e s: (1) aNearctic cluster with tw o monophyletic subgroups,pseudoobscura a n d af f inis; (2) t he D . subobscur a-rela ted species (subobscurasubgroup according t o Ba r-rio et al., 1994); (3) the lineage of the sister taxa D .obscura a n d D . a m b i g u a , a nd (4) the lineage (or lin-eages) of D . bi fasciata a n d D. subsilvestris. O u r a t -

empts to determine the branching order of these fouro r fi v e , i f w e s e p a r a t e bifasciata a nd subsilvestris)ineages fail , owing to short internode lengths which

result in levels of sequence divergence among t he ta xaha t a re not differentia ble (Figs. 7 an d 13; Ta bles 1 a nd

3).Th e s t a t is t ica l in st a b il it y of t h e ou t g rou ps in t h e

r ees b a sed on t h e Sod a n d G p d h coding regionsndicates that the phylogenies are affected by stochas-ic noise. The essential role of the enzymes encoded by

both genes subjects t hem t o int ense purifying selection.This is clearly observed in t he G pdhgene, where a lmost

a l l v a r ia b le s it e s corre sp on d t o s y n on y m ou s cod onpositions (96.7%), but a lso, in th e Sod gene (71.4%s y n on y m ou s s u bs t i t u t ion s ). S a t u ra t ion a n d p a ra l lela n d b a ck m u t a t i on s m a y h a v e r a n d om iz ed t h e s e -quences w ith respect t o the common phylogenetic his-ory t h a t t h e obscur aspecies shar e wit h t he outgroupswhich a re t he closest outgroups ava ilable, a nd mem-

bers of the sa meSophophor asubgenus; see Throckmor-on, 1978; Kwiatowski et al . , 1994). Ra ndomizat ion

cou ld a ls o h a v e clou d ed t h e re la t io ns h ip s b et w e ensubgroups and lineages, especially because they areproducts of rapid phyletic radiat ion. Clouding would

occur even t hough th e exa mina tion of the skewness ofhe tree-length distributions (Hillis and Huelsenbeck,

1992) for both gene regions (data not shown), and theresolution of the phylogenetic relationships betweens peci es w i t h in s u bg r ou ps , i nd ica t e t h a t t h er e i s astrong phylogenetic signal.

Ra pid phyletic ra diat ion and/or a sat ura t ion effectmay account for the apparent impossibility to resolveh e ea r l y p hy l og en y of t h e D . obscur a group. Thensta bility of th e outgroups a nd t he convergent compo-

sit iona l patt erns observed in some species ar gue inavor of th e ra ndomizat ion of the sequences over time.

But the persistent inability to resolve the phylogenywith either nuclear or mitochondrial genes supportshe hy pothesis of rapid phyletic ra diat ion (Hoelzer a nd

Melnick, 1995). Very likely w e a re confronting bothphenomena, randomization of the phylogenetic signal

a nd short int ernode time lapses.Constrained positions and conservative changes are

thought to be particularly reliable in the phylogenetica na lysis of D NA sequences, beca use th eir phylogenetic

signal is less likely erased by par allel and back muta -t ions and saturat ion (Swofford and Olsen, 1990, pp.

497500; Hillis et al . , 1993; Miyamoto et al . , 1994).When we rely on conservative changes (transversions

in the NJ a nd MP a na lyses of both genes; or, forSod,NJnonsyn onymous subst itu tions or ML 1st -plus-2nd posi-

t i on s ), or a s s um e t h a t t r a n s it i on s a r e m u ch m or ef r eq u en t t h a n t r a n s v er s ion s i n t h e M L a n a l y s es ofG p d h , the phylogenetic reconstructions show the D .

subobscura s u bg r ou p a s t h e fi r s t t a x on t o d iv er g e.These trees are not statistically robust but both genes

yield the sa me result. We thus t enta tively propose tha tt he D. subobscurasubgroup may have been the fi rst t o

diverge from the rest . This hypothesis is consistentwith the proposal t hat the metapha se confi gurat ion of

five acrocentric chromosomes is the ancestral one inD rosophi la(Buz za ti-Tra verso a nd S cossiroli, 1955). D .

subobscur a, D. guan che, a nd D . m ad eir ensish a v e fi v eacrocentric chromosomes (Lakovaara and Saura, 1982),

whereas the prevailing condition in most obscura spe-cies is one m eta centr ic X chromosome plus thr ee a cro-

centric autosomes.We hypoth esize th a t t he obscur agroup species derive

from two main radiat ions. One resulted from adapta-t ion t o n ew h a b it a t s t h a t a p p ea re d d u rin g t h e e xp a n -

sion of t empera te forests in the Old World. This r a dia-tion gave rise to theD. subobscur asubgroup, several D .

obscura s u b g ro u p l in e a g e s , a n d t h e a n c e s t o r o f t h eNearctic species. Products of this first radiat ion mayh a v e a l s o b een t h e a n ces t or s of t h e D . m i c r ol a b i s

subgroup (Car iou et al., 1988) and perhaps of severalra re species of unclear ta xonomic posit ion, l ike D .

a l p i n a a nd D. helvetica (Lakovaara and Saura, 1982),a n d D. inexspectata (Tsa cas , 1988). A second ra dia t ion

took place during the colonization of deciduous forestsof t h e N ew Wor l d b y t h e d es cen d a n t s of a s in g lecolonizing lineage that gave origin to the two Nearcticsubgroups, a f fi n i s a n d pseudoobscura (G oddar d et al.,1990; B eckenba chet al .,1993).

In conclusion, our analysis of the G pdha nd Sodgenesequences supports tw o previous proposals : (1) thedivision of the para phyletic former D . obscur a sub-group in, a t lea st , t wo subgroups (obscura a n d subob-scura) ( B a r r i oet a l .,1994, bas ed on mit ochondria l DNAsequences); an d (2) the monophyletic origin of theNearctic species, including both the a f fi n i s a nd pseudo-

obscurasubgroups (based on DNADNA hybridiza tion,G o d d a rd et a l . , 1990 ; a n d m it och on d ria l COII s e-quences, Beckenbach et al.,1993).



14/15

ACKNOWLEDGMENTS

This w ork has been supported by a postdoctoral fellowship to E.B .

rom th e P rograma G eneral de F ormacion del P rofesorado, Ministe-

io de Educacion y C iencia, Spa in, a nd by the Na tional I nstit utes of

Health Gr a nt GM 4 23 97 t o F.J .A. We a r e gr atef ul t o D ouglas W.

Skarecky for technical advice and to Brigita Fody, Nelsson Becerra,

a n d J u l is s a Ag u il a r f or v a l u a b le h e lp . C o m pu t e r a n a l y se s w e r ec ar r ied out using the S er vei d e Bioinfor matic a (U niver si ty of

Valencia, S pain), supported by G ra nt P B93-0034. We a re indebted t o

P rofessor W.-H. L i for th e NJ BOOTK2 program .

REFERENCES

Bac hmann, L. , M uller , E. , Car iou, M . L. , and S per l ic h, D. (1 9 9 2 ).

Cloning and c har ac ter izat ion of K M 1 9 0 , a spec ific satel l i te DN A

f amily of D r o sop h i l a k i t u m en s i s a nd D . m i cr o l a b i s . G en e 120:

267269.

B a r r io , E . , L a t o r r e, A. , a n d M oy a , A. (1 994). P h y l og en y o f t h e

D r o sop h i l a ob scu r a species group deduced from mitochondria lDNAsequences. J . M o l . E v ol . 39: 478488.

Barrio, E., Latorre, A., Moya, A., and Ayala, F. J . (1992). Phylogenetic

reconstruction of the Drosophila obscura gr oup, on the basis of

mitochondria l D NA. M ol . B i ol . E v o l . 9: 621635.

B eckenba ch, A. T., Wei, Y. W., a nd L iu, H . (1993). Relat ionsh ips in t he

Dr osophila obscura species group, inferred from mitochondria l

cytochrome oxidase-II sequences. M ol . B i ol . E v o l . 10: 619634.

B ewley, G . C., C ook, J . L., Kusa kabe, S ., Mukai, T., Rigby, D. L., a nd

Chambers, G. K. (1989). Sequence, structure and evolution of the

gene coding for sn-glycerol-3-phosphate dehydrogenase in D r o -

sophi la melan ogaster. N ucleic Acid s Res. 17: 85538567.

Brehm, A., and Krimbas, C. B. (1990). The phylogeny of 9 species of

t h e Drosophila obscuragroup inferred by t he ba nding h omologies

of chromosomal regions. 2. Element E. H er e d i t a s 113: 157168.


t h e Dr osophil a obscuragroup inferred by t he ba nding h omologies

of chromosomal regions. 3. Element D. Genome35: 10751085.


t h e Dr osophil a obscuragroup inferred by t he ba nding h omologies

of the chromosoma l regions. 4. Element C. H er e d i t y 70: 214220.

B rehm, A., Krimba s, C. B ., Sourdis, J . , an d C ar iou, M. L. (1991). The

phylogeny of 9 species of the Dr osophil a obscuragroup inferred by

the band ing homologies of c hr omosomal r egions. 1 . Element B.

Genome34: 464471.

B rown, W. M., P ra ger, E . M., Wan g, A., an d Wilson, A. C . (1982).

M itoc hond r ial DN A seq uenc es of pr imates: tempo and mod e of

evolution. J . M o l . E v ol . 18: 225239.

B uzza ti-Tra verso, A., a nd S cossiroli, R. E. (1955). The obscur agroup

of the genus D rosophi la. A dv. Genet. 7: 4792.

Ca brer a , V. M., G onza lez, A. M., La rr uga , J . M., a nd G ullon, A. (1983).

Genetic distance and evolutionary relationships in the Drosophila

obscuragroup. E v o l u t i on 37: 675689.

Ca r iou, M . L. , La c haise, D . , Tsac as, L. , S our d is , J . , Kr imbas, C. B. ,

and Ashburner, M. (1988). New African species in the Drosophila

obscuraspecies group: genetic var iat ion, differentia tion a nd evolu-

tion. Heredity61: 7384.

Felsenstein, J . (1981). Evolutionar y trees from DNA sequences: a

ma ximum likelihood approach. J . M o l . E v ol . 17: 368376.

Felsenstein, J . (1 98 5). Confid enc e l imits on phy logenies with a

molecular clock.Syst. Zool. 34: 152161.Felsenst ein, J . (1993). PH YLIP : phylogeny in ference packag e, v. 3.5c.

Un iversity of Wash ington, Seat tle.

Fitch, W. M. (1971). Towa rd defin ing the course of evolution: mini-

mum change for a specified tree topology. Syst. Zool. 20: 406416.

Fridovich, I. (1986). Superoxide dismuta ses. A d v . E n z ym o l . 58:

6197.

God d ar d , K. , Ca c cone, A ., a nd P owell , J . R. (1 99 0). Evolutionar y

implicat ions of DNA divergence in th e Drosophila obscura group.

E v o l u t i o n 44: 16561670.

G onza lez, A. M., H erna ndez, M ., Volz, A., P esta no, J ., La rrug a , J . M.,

S p er l ich , D . , a n d C a b r e r a , V. M . (1990). M it o ch on d r ia l D N A

evolution in the obscura species subgroup of D r o s o p h i l a . J . M o l .

E v o l . 31: 122131.

Higgins, D. G. , and S ha r p, P . M . (1 98 8). C LU S TAL: a pac kage f or

performing multiple sequence alignment on a microcomputer. Gene

73: 237244.

Hillis, D . M., Allar d, R. W., a nd Miy a moto, M. M. (1993). Ana lysis of

DN A seq uence d ata : Ph y logenetic inf er ence. M et h o d s E n z ym o l .

224: 456487.

Hil l is , D. M ., a nd Huelsenbeck, J . P . (1 99 2). S ignal , noise, a nd

r eliabi li ty in molecular phy logenetic ana ly sis . J . H e r ed i t y 83:

189195.

Hoelzer, G. A., a nd Melnick, D . J . (1995). P at terns of speciat ion a ndlimits to phylogenetic resolution. Trends Ecol. E vol. 9: 104107.

Hudson, R. R., B ailey, K., Ska recky, D., Kw iat owski, J . , an d Aya la, F.

J . (1994). Evidence for positive selection in the superoxide dismu-

t a s e (Sod) region of Drosophila melanogaster. Genetics136: 1329

1340.

Ka wa saki , E . S . (1 99 0). S a mple pr epar a tion f r om blood , c ells , a nd

other fluid s. I n P CR P rotocols: A G uide to Meth ods a nd Applica-

tions (M. A. I nnis, D. H. G elfand, J . J . S ninsky, an d T. J . White,

Ed s.), pp. 146152, Academ ic Pr ess, Sa n D iego, CA.

Kimura , M. (1980). Asimple method for estima ting evolutionary rat e

of base substi tution thr ough c ompar ative stud ies of nuc leotid e

sequences. J . M o l . E v ol . 16: 111120.

Kishino, H . , an d H asegaw a, M . (1 98 9). E valuat ion of the ma ximumlikelihood estimate on the evolutionary tree topologies from DNA

seq uenc e d ata , and the br anc hing or d er in Hominoid ea. J . M o l .

E v o l . 29: 170179.

K u m a r , S . , Ta m u r a , K . , a n d N ei , M . (1993). M E G A: m o le cu l a r

evolutionar y genetic s a naly sis , v . 1 .0 . The P ennsy lvania S ta te

U niver sity , U niver si ty P ar k, P ennsy lvania .

K w i a t o w s k i, J . , P a t e l, M ., a n d Ay a l a , F. J . (1 989). Drosophila

melanogasterC u,Zn superoxide dismuta se gene sequence. N u c l ei c

Acid s Res. 17: 1264.

Kwia towski , J . , S kar ec ky, D. , Ba i ley, K. , and Ay ala , F. J . (1 99 4).

Ph y logeny of Drosophila a n d r el a t e d g e n er a i n fe rr e d f r om t h e

nuc leotid e seq uenc e of the Cu,Zn Sod gene. J . M o l . E v o l . 38:443454.

Kwia towski , J . , S kar ecky, D . , H er nand ez, S . , Ph am, D. , Quijas , F. ,

and Ay ala , F. J . (1 99 1). High fi d eli ty of th e poly mer ase c hain

reaction. M ol . B i ol . E v o l . 8: 884887.

Kw iat owski, J . , Kra wczyk, M., J aw orski, M., Skar ecky, D., a nd Aya la,

F. J . (1996). Erratic evolution of glycerol-3-phosphate dehydroge-

nase in Drosophila, Chymomyza,a nd C er a t i t i s . J . M o l . E v ol . 43: inpress.

La kovaa ra , S., a nd K eran en, L. (1980). P hylogeny of the Drosophila

obscuragroup. Genetika12: 157172.

Lakovaara, S., and Saura, A. (1982). Evolution and speciation in the

Drosophila obscura group. I n The Genetics and Biology of D r o -

sophila (M. Ashburner, H . L. Ca rson, a nd J . N. Thompson, Eds.),

pp. 159, Academ ic P ress, N ew York.

La kovaa ra , S., Sa ura , A., an d Fa lk, C. T. (1972). Genetic dista nce andevolutionary r elationships in the Dr osophil a obscuragroup. E v o l u -

t ion26: 177184.



15/15

La torre, A., B arr io, E., Moya, A., a nd Aya la, F. J . (1988). Mit ochon-d r ial D N A evolution in t he Drosophila obscura group. M ol . B i ol .

E v o l . 5: 717728.

Li, W.-H., an d Zharkikh, A. (1995). S ta tistical tests of DNA phylog-

enies. Syst. Bi ol. 44: 4963.

Ma rinkovic, D., Aya la, F. J ., a nd Andjelkovic, M. (1978). G eneticpolymorphism a nd phy logeny of Dr osophil a subobscura. Evolu tion

32: 164173.Miyam oto, M. M., Allard, R. W., Adkins, R. M., J an ecek, L. L ., a nd

Honey cutt , R. L. (1 99 4). A c ongr uence test of r el iabi li ty usinglinked mitochondrial D NAsequ ences. Syst. Bi ol. 43: 236249.

Moritz, C., Dowling, T. E ., a nd B rown, W. M. (1987). E volution ofanimal mitochondrial DNA: Relevance for population biology andsy stematic s .Annu. Rev. Ecol. Syst. 18: 269292.

Nei, M., a nd Gojobori, T. (1986). S imple methods for estima ting thenumbers of synonymous and nonsynonymous nucleotide substitu-

tions. M ol . B i ol . E v o l . 3: 418426.

Ruttka y, H. , S olignac , M ., and S per lich, D. (1 99 2). N uclear andmitochond r ial r ibosomal RN A var ia bi li ty in the obscura group ofDrosophila. Genetica85: 131138.

S a c cone, C. , Lana ve, C. , and Pesole , G . (1 99 3). Time a nd biose-quences. J . M o l . E v o l . 37: 154159.

S a c con e , C . , L a n a v e, C . , P e s ol e, G . , a n d P r e p a r a t a , G . (19 90).

Influenc e of base c omposit ion on q uanti tat ive est imates of geneevolution. M ethods Enzymol. 183: 570583.

Sa cktor, B . (1975). B iochemist ry of insect fl ight .I nIn sect B iochemis-tr y and Func tion (D. J . Ca nd y and B. A. Kilby, Ed s.), pp. 3 8 8,

Wiley, New York.

Saitou, N., and Nei, M. (1987). The neighbor-joining method: a newmethod for reconstructing phylogenetic trees. M ol . B i o l . E v ol . 4:406425.

St urteva nt, A. H . (1942). The classifi cation of t he genus D r o sop h i l a

with the description of nine new species. Univ. Texas Publ. 4213:551.

Swofford, D. L., and Olsen, G. J . (1990). Phylogeny reconstruction. I n

M olec ular S y stemat ics (D. M . Hil l is and C. M or itz , Ed s.), pp.

411501, Sinauer, Sunderland, MA.

Tajima , F., a nd Nei, M. (1984). E stima tion of evolutionary dista nce

between nucleotide sequences. M ol . B i ol . E v o l . 1: 269285.Tam ura , K. (1992). E stima tion of th e number of nucleotide substitu-

t i o n s w h e n t h e r e a r e s t r o n g t r a n s i t i o n - t r a n s v e r s i o n a n d G 1C -

content biases. M ol . B i o l . E v o l . 9: 678687.

Ta m u r a , K . , a n d N ei , M . (19 93). E s t i m a t i on of t h e n u m b er o f

nuc leotid e substi tutions in the c ontr ol r egion of mitoc hond r ial

DN A in humans a nd c himpanzees. M o l . B i o l . E v ol . 10: 512526.

Templeton, A. R. (1983). P hylogenetic inference from restriction

end onuc lease c leava ge si te ma ps w ith par t ic ular r efer enc e to t he

evolution of huma ns an d apes. E v o l u t i on 37: 221224.

Throckmorton, L. H. (1975). The phylogeny, ecology, and geography of

D r o sop h i l a . I n Ha nd book of G enetics (R. C . King, Ed .), pp.

421469, Plenu m P ress, New York.

Tominaga , H. , S hiba, T., and N ar ise , S . (1 99 5). S tr uc tur e of D r o - sop h i l a v i r i l i s glycerol-3-phosphate dehydrogenase gene and com-par ison with t he Drosophila melanogaster gene. Biochim. Biophys.

Acta1131: 233238.

Tsa cas, L. (1988). U ne cinquieme espece a fricain e du groupe D r o - sophi la obscura (D i p t e r a : D r oso ph i l i d a e ). Ann. Soc. Ent. Fr. (N S )24: 331335.

Ts a c a s , L . , C a r i ou , M . L . , a n d L a c h a i se , D . (1985). L e g r ou p eDrosophila obscuraen Afrique de lEst , description de tr ois nouv-

elles especes (D i p t e r a : D r o sop h i l i d a e ). Ann. Soc. Ent. Fr. (N S) 21:413424.


Documents

Paper Obscuras Barrio