Phylogenetic Relationships in Drosophila: A Conflict between Molecular and Morphological Data 1

Richard H. Thomas* and John A. Hunt?

*Department of Zoology, The Natural History Museum; and TDepartment of Genetics, University of Hawaii

Grimaldi recent cladistic classification of genera in the family Drosophilidae, based s on adult morphological characters, is evaluated for those taxa for which alcohol dehydrogenase DNA sequences are also available. These data allow us to look at relationships of the Drosophila subgenera Sophophora and Drosophila with the Hawaiian Drosophila (Idiomyia) and with Scaptomyza, when Scaptodrosophila is used as an outgroup. The molecular data give strong support for the broad relationships hypothesized by Throckmorton, who contended that the subgenus Sophophora is a sister group to the remaining ingroup taxa. The Drosophila subgenus Engiscaptomyza, which Throckmorton regarded as intermediate between the Idiomyia (Hawaiian Drosophila) and Scaptomyza, is shown to be much more closely allied with Scaptomyza, in agreement with Grimaldi results from adult mors phology. The Hawaiian taxa, both Idiomyia and Scaptomyza, are firmly located as a sister group to the subgenus Drosophila, and these in turn all form a sister group to the subgenus Sophophora. Grimaldi classification of these taxa is quite s different and places the Hawaiian Drosophila (Idiomyia) as sister group to the remaining ingroup taxa (Scaptomyza and the subgenera Sophophora and Drosophila). Our results show that Grimaldi new classification of these taxa results s in paraphyletic groups, just as does the traditional classification under Throckmorton interpretation of relationships. Additional data are required to produce s a robust classification of this huge paraphyletic genus. Introduction

Evolutionary biologists have long been fascinated by the diversity of Drosophila ( Sturtevant 192 1; DeSalle and Gximaldi 199 1) , and, increasingly, molecular geneticists and developmental biologists are using this diversity as a tool to understand evolutionary processes at subcellular and cellular levels. Interspecific comparisons of Drosophila are being used in studies on the evolution of gene structure and function (e.g., see Sullivan et al. 1990) and on the distribution and evolution of transposable elements (Daniels et al. 1990; Maruyama and Hart1 199 1; Brezinsky et al. 1992). An accurate phylogeny of these flies is therefore of wider interest than simply understanding the relationships of a very interesting group of organisms. This paper discusses the phylogeny of clades of drosophilid flies for which Grimaldi ( 1990) has recently used the following nomenclature: subgenera Drosophila, Sophophora, and Engiscaptomyza of the genus Drosophila, and the genera Idiomyia and

1. Key words: alcohol dehydrogenase,Drosophila, Hawaiian Drosophila, Scaptomyza, Scaptodrosophila, molecular systematics. Address for correspondence and reprints: Richard H. Thomas, Department of Zoology, The Natural History Museum, Cromwell Road, London, SW7 5BD, United Kingdom.
Mol. Biol. Evol. 10(2):362-374. 1993. 0 1993 by The University of Chicago. All rights reserved.

Downloaded from http://mbe.oxfordjournals.org/ at Universidade Federal De Minas Gerias on June 11, 2012

0737-4038/93/1002-0008$02.00 362

Drosophila Phylogeny

363

Scaptomyza. As we will show, this classification results in a paraphyletic concept of the genus Drosophila, but Grimaldi s ranks and taxa will nevertheless be used here for their immediate convenience. Of these clades, D. (Drosophila) and D. (Sophophora) are essentially continental, Idiomyia (traditionally known as Hawaiian Drosophila) and D. (Engiscaptomyza) are endemic to the Hawaiian archipelago, and Scaptomyza are found primarily in Hawaii, with a few species scattered about the rest of the world. Grimaldi ( 1990) has produced a major revision of the genera in the family Drosophilidae, on the basis of an analysis of 2 17 adult morphological characters from 120 representative species. He undertook this study in an effort to construct a phylogenetic classification of the family and to evaluate the phylogenetic hypotheses of Throckmorton, which are the result of internal morphological studies (Throckmorton 1962, 1966, 1975) and which have formed the basis for most discussions on evolution in Drosophila and its relatives. However, nonmorphological evidence conflicts with Grimaldi s findings. In our analysis of the DNA sequences of alcohol dehydrogenase (ADH; gene Adh) (E.C. 1.1.1.1) genes (Thomas and Hunt 199 1)) we showed that the Hawaiian Drosophila form a monophyletic group that, for the purposes of the present paper, may be less confusingly referred to by the name resurrected by Grimaldi ( 1990)) Idiomyia. We tried to place these flies in context by comparisons with the available sequence data from mainland Drosophila. Although we had no sequence from a known outgroup with which to root our tree, the use of distance measures and a midpoint rooting technique supported the relationships hypothesized by Throckmorton ( 1966, 1975) (fig. la) and are also in general agreement with both an analysis of immunological distances from a larval hemolymph protein (Beverley and Wilson 1982, 1984, 1985) and an analysis of 905 bp of mitochondrial DNA (DeSalle 1992a). As a result of Throckmorton s work, the traditional classification of the Drosophilidae (Wheeler 198 1) has long been recognized to contain paraphyletic groups. Grimaldi ( 1990) results from his analysis of the adult morphology of these taxa are very s different (fig. lb), and he uses them to propose new generic concepts and a new classification of drosophilid genera, which he claims reflect phylogenetic relationships. Because these two views are almost mirror images of one another, we think it is important to reassess the conflict, by reporting new DNA sequence data. Consequently, we report here the first Adh sequence from a member of the drosophilid genus Scaptomyza, a group that attains its greatest diversity in the Hawaiian Islands, and analyze it along with other available Adh sequence data. Recently another Adh DNA sequence has become available, for Scaptodrosophila lebanonensis, a drosophilid (Alabat and Gonzalez-Duarte 1990). All lines of evidencemorphological (Throckmorton 1962, 1975; Grimaldi 1990) and molecular (Beverley and Wilson 1982, 1984; DeSalle 1992b)-indicate that Scaptodrosophila (Grimaldi 1990) is well outside a monophyletic genus Drosophila, despite its traditional classification as a subgenus of Drosophila (Wheeler 198 1) . This allows us to use Scaptodrosophila Zebanonensis as an outgroup in our previous analysis (Thomas and Hunt 199 1) and therefore allows us to examine sister-group relationships among D. (Drosophila) , D. (Sophophora) , D. ( Engiscaptomyza ) , Idiomyia, and Scaptomyza. Material and Methods Table 1 lists the species whose Adh sequences are analyzed here, their sources, and their data-base accession numbers. The Adh gene was isolated from a genomic

Downloaded from http://mbe.oxfordjournals.org/ at Universidade Federal De Minas Gerias on June 11, 2012

FIG. 1.-Alternative hypotheses relationships among drosophilid taxa for which both molecular and morphological data are available. a, Throckmorton s ( 1966, 1975) hypotheses of relationships, based principally on internal morphology. b, Grimaldi ( 1990) cladistic classification based on adult morphological s characters.

of

Drosophila Phylogeny

365

Table 1 Species Included in the Analysis of Adh Sequences, according to the Arrangement of Wheeler (1981) Catalog of the Family Drosophilidae s
GenBank Accession No.

Species Genus Scaptomyza Hardy Subgenus Tantalia Malloch albovittata (Malloch) ........... Genus Drosophila Fallen Subgenus Scaptodrosophila Duda lebanonensis Wheeler .......... Subgenus Engiscaptomyza Kaneshiro crassifemur Grimshaw .......... Subgenus Drosophila (Hawaiian Drosophila or Idiomyia) heteroneura (Perkins) ......... afinidisjuncta Hardy ......... adiastola Hardy ............. mimica Hardy .............. nigra Grimshaw ............. (repleta group) mulleri Sturtevant ........... Subgenus Sophophora Sturtevant melanogaster Meigen ........... pseudoobscura Frolova .........
Subheadings in parentheses are commonly

Source of Sequence

Present paper

M80925

Alabat and Gonzalez-Duarte Thomas and Hunt 199 1

1990

X54814 M60790

Rowan and Hunt 199 1 Rowan and Dickinson 1988 Thomas and Hunt 199 1 Thomas and Hunt 199 1 Thomas and Hunt 199 1 Fischer and Maniatis Benyajati et al. 198 1 Schaeffer and Aquadro 1985

M3678 1 M37262 M6079 1 M60792 M60793 X03048 Ml 1290 YOO602

1987

used but are not recognized in Wheelers ( 198 1) catalog.

library from Scaptomyza albovittata DNA, extracted according to a method described elsewhere (Thomas and Hunt 199 1) and produced by partial Mb01 digestion and ligation into bacteriophage lambda EMBL 4 digested with Sal1 and BamHI. The positive plaques containing the Adh sequences were detected by hybridization to a plasmid probe containing part of the Adh gene from Drosophila heteroneura, and the bacteriophage were purified by replating and reisolation. The Adh-containing region was subcloned into the chimeric plasmid pZfl8U, and the sequence of the region was determined by dideoxy sequencing of both the plasmid DNA and sequential deletions derived from exonuclease III digestion according to methods described elsewhere (Thomas and Hunt 199 1) . Additional sequences were obtained by PCR amplification using the primers 5 -TTAGATGCCCGAGTCCCAGTGCT-3 and 5 -AGAACATCATCTTTGTCGCTGGTCT-3 and subsequent cloning by blunt-ended ligation into the SmaI site of plasmid pZfl8U for sequencing. Figure 2 shows the sequence of the three coding exons and of the two internal introns of the Adh gene of S. albovittata, along with its inferred amino acid sequence. The introns and flanking regions will be discussed elsewhere, in the context of the other Adh sequences. Our analyses utilize only the protein-coding exons of the Adh locus. The Adh loci from all of the species analyzed have equivalent placement of introns interrupting the coding sequences, and the coding regions are all of the same length, with the exception of D. melanogaster, which has an additional two codons just 3 of the initiation codon. These two codons are not included in the analyses and pose no difficulties to the alignment of the sequences. For each of 11 species we use sequences of 762 bp

366

Thomas and Hunt 60 20 119 31 178 139 238 59 298 F 79

ATGTGTATCGCTGGCAAGAATATCATCTTTGTCGCTGGTCTCGGTGGCATTGGACTGGAC M C I A G K N I I F V A G L G G I G ACCAATCGCGAGATTGTCAAGAGCGGTCCCAAG T NREIVK S G P

GTATATTTTAACTCACGAATACTTTA K AACCTAGTGATCCTTGATCGCAT N L V I L D R

AGTAATGTTTATGTTAAAGTGCGTTTCTCTGTTTAG

TGAGAATCCGGCTGCCATTGCCGAGCTGCAGGCGATTAATCCCAATGTGACCGTATCCTT E N PAAIAELQAI NPNVTVSF TTATCCCTACGATGTTACAGTAGCGTTGGCTGAAAGTGTCAAGCTGCTGAAGACAATCTT Y PYDVTVALAE SVKLLKTI CGCTAAGCTGAAGACAGTTGATTTGCTCGTCAATGGAGCTGGCATTCTGGACGATCATCA A K LKTVDL LVNGAGI L D D GATTGAGCGGACCATTGCAGTGAATTTCACTGGTACTGTGAATACCACAACTGCTATCAT I E R T I AVNFTGTVNTTTAI GGAGTTCTGGGATAAGCGCAAGGGTGGACCAGGTGGCGTTATTGCCAACATTTGCTCAGT I c E FWDKRKGGPGGV I A N GACAGGATTCAATTCCATTTACCAGGTGCCTGTATACTCTGCCTCAAAGGCGGCTGCCTT F N S I Y Q V PVYSASKAAAL T G GAGCTTTACCAGCTCCCTTGCG S FTSSLA GTAAGTTCATTAACAGATAATTCTATTCCATTCAACA

358 Q99 418 M 119 478 v139 538 159

Downloaded from http://mbe.oxfordjournals.org/ at Universidade Federal De Minas Gerias on June 11, 2012

597 166 656 175 716 195 776 215 836 235

GATGATAGTAATGCCTTTGTTTCTAACACAG

AGATTGGCACCTATAACTGGCGTAACGG R L A P I T G V T

TCTATTCTATTAATCCTGGCATTACTGACACCACATTGGTGCACAAATTCAATTCCTGGC v Y s I N P G I TDTTLVHKFNSW TCGATGTTGAGCCACGTGTGGCCGAGAAGCTCCTTGCCTTTCCCACACAGACATCTTTGG LDVEPRVAEKLLAFPTQTSL CATGTGCCAAGAACTTTGTCAGGGCCATTGAAGCCAATAAGAATGGCGCCATCTGGAAGC A C A K N F V R A I EANKNGAIWK TGGATCTTAACCGTCTGGATGAAATTGAATGGACCAAGCACTGGGACTCGGGCATC L D L N R L D E I E W T KHWDSGI

892 254

FIG. 2.-DNA sequence of the Scaptomyza ( Tantalia) albovittata Adh gene. Shown are the three coding exons and the two internal introns. The exons align precisely with those of the other species. For the aligned sequences of the remaining species, with the exception of Scaptodrosophila lebanonensis, see the work of Thomas and Hunt ( 199 1).

each. All sequences in this analysis have been deposited (table 1) and are also available from the authors.

into the sequence

data bases

Primary analyses were carried out by PAUP, version 3.0k (Swofford 1990), for parsimony and linear invariants, and by PHYLIP, version 3.3 1 (Felsenstein 199 I), for maximum-likelihood and distance methods. The details of their use will be given below. Base compositions, format conversions, statistical tests of equality of base compositions, and generation of random sequences were done with a program written by one of the authors (R.H.T.). Additional sequence manipulations were done with the sequence editor ESEE (Cabot and Beckenbach 1989 ) .

Drosophila Phylogeny

367

Results Tree Topology A bootstrapped (Felsenstein 1985 ) maximum-parsimony analysis of the full data set is given in figure 3. Details of the analysis are given in the figure caption. This tree has been rooted by assuming that Scaptodrosophila is the outgroup of the rest. Maximum parsimony applied to unordered characters results in the same tree length regardless of the position of the root (Swofford 1990). Designation of a taxon as an outgroup does not affect the reconstruction of ancestral character states. The first point is the close relationship between Scaptomyza albovittata and D. (Engiscaptomyza) crassijkmur. Throckmorton regarded Engiscaptomyza as intermediate between Scaptomyza and Drosophila, and this led him to conclude that Scaptomyza could have originated in Hawaii and then colonized the rest of the world. Grimaldi s ( 1990) results show the same close relationship of these flies as do the Adh data. We will return to this problem in the discussion. Because these taxa form a clade, we will refer to them collectively as Scaptomyza/ Engiscaptomyza. More important for the main argument of this paper is that the parsimony analysis strongly and significantly supports the placement of Scaptodrosophila on the internal branch of the unrooted tree joining the sophophoran species, D. melanogaster and D. pseudoobscura, and the remaining ingroup taxa. An additional 40 substitutions on the most parsimonious tree of 668 steps (fig. 3) are required to move Scaptodrosophila to the internal branch joining Scaptomyza/ Engiscaptomyza and Idiomyia, as hypothesized by Grimaldi ( 1990). Differences in base composition among sequences can bias methods for constructing trees from DNA sequence data (Gillespie 1986). Pair-wise x2 tests of the equality of base composition between pairs of species showed that these taxa do not differ among themselves in first- and second-codon-position base composition (P > 0.05) but do in some cases differ significantly in third-base-position usage (Thomas and Hunt 199 1) (table 2). Scaptomyza albovittata is similar to the other Hawaiian species in its third-base composition (P > 0.05 ) but differs from the D. (Sophophora) and D. (Drosophila) species (P -C0.05). Scaptodrosophila lebanonensis is significantly different only from D. (Sophophora) melanogaster in its third-base usage (P < 0.05 ). To eliminate both any potential bias due to base-composition differences at thirdposition sites and any noise arising from multiple substitutions at these predominantly synonymous sites, we repeated the bootstrapped maximum-parsimony analysis with only first- and second-position sites (total number 508), which are almost wholly nonsynonymous (and thus presumably more subject to selection). The tree topology for the subgenera and genera is the same as for the full data set. Only within the Idiomyia (Hawaiian Drosophila) does the topology differ, and this is not strongly supported, as would be expected for these closely related species analyzed under a more slowly evolving subset of the data. Again the position of Scaptodrosophila on the internal branch joining the sophophorans and the remaining taxa is supported, with the sophophorans again grouping together in 99% of replicates and with the remainder of the ingroup species grouping together in 9 1% of the replicates. An additional 16 substitutions on the most parsimonious tree of 22 1 steps are required to shift Scaptodrosophila to the position hypothesized by Grimaldi ( 1990). Though Scaptodrosophila is an outgroup to the rest of the taxa in these analyses, we ask whether its Adh sequence is so diverged from the ingroup taxa as to behave as a random outgroup (Wheeler 1990)) thus producing an artifactual rooting for the ingroup. On the basis of the average base composition of the variable sites (323) of

mbe.oxfordjournals.org/ at Universidade Federal De Minas Gerias on June 11, 2012

I

99

lebanonensis

(Scaptodrosophila)
54

melanogaster

(Sophophora)
56

pseudoobscura

(Sophophora)
51

mulleri

(Drosophila)
crassifemur

(Engiscaptomyza)
30

albovittata

(Scaptomyza)

45

40

35

30

25

20

15

10

Time (Myr)
FIG. 3.-Bootstrapped maximum-parsimony tree based on analysis of the full 762-bp A&-sequence data set. The tree was produced with PAUP (Swofford 1990) using a branch-and-bound search with furthest addition sequence. All minimal length trees were saved (MULPARS), and zero-length branches were collapsed. Numbers on branches are the inferred number of substitutions along the branch (with the number of synapomorphies in parentheses). The numbers on the nodes are the percentage of replicates in which the taxa to the right of the node occur together. This is a bootstrap 50%-majority-rule consensus tree based on 100 replicates. The tree is shown with Scaptodrosophila lebanonensisas an outgroup to root the tree. The X indicates the branch where Grimaldi ( 1990) analysis hypothesizes that the outgroup should root the s tree. There are 598 character states distributed over 240 informative characters. The tree is 668 steps long and is the single most parsimonious tree. The next most parsimonious tree is 12 steps longer. The time scale was drawn by assuming a per-lineage synonymous rate of 0.0 15 substitutions/nucleotide/Myr (Rowan and Hunt 199 1) and is the same as that presented by Thomas and Hunt ( 199 1)) with the addition of S. lebanonensisand Scaptomyzaalbovittata.

Drosophila Phylogeny

369

Table 2 Base Composition at Third-Base of Drosophilid Species

Positions in ADH-coding Regions

No. OF BASES
SPECIES

G 63 80 64 74 78 77 70

C 86 91 77 110 116 130 96

T 78 62 82 47 52 36 73

A 27 21 31 23 8 11 15

D. adiastola . . . . D. (Engiscaptomyza) crassifemur Scaptomyza albovittata . . D. (Drosophila) mulleri . . . D. (Sophophora) pseudoobscura . D. (Sophophora) melanogaster . . . . Scaptodrosophila lebanonensis . . .

NOTE.-Drosophila adiastola is included as a representative of the Hawaiian Drosophila (Idiomyia). The Hawaiian Drosophila species are all very similar in their base composition (Thomas and Hunt 1991). Frequencies at first- and second-base positions do not differ significantly among these species (Thomas and Hunt 199 1).

the ingroup taxa, a random sequence was generated, and the maximum-parsimony analyses were repeated. The random sequence joins the tree on the branch leading to D. (Sophophora) melanogaster and results in a tree 100 steps longer than that observed with Scaptodrosophila lebanonensis as the outgroup. The sequence similarity between the ingroup taxa and Scaptodrosophila is much higher than that with the random sequence, suggesting that it is indeed a good outgroup. The other methods of analysis reported below also support Scaptodrosophila as a good outgroup. By pruning our tree to a number of taxa for which we can examine all possible trees (i.e., <lo), we can consider in more detail trees near the most parsimonious. The Idiomyia taxa- afinidisjuncta, adiastola, and mimica-were removed from the analysis, to leave a more regularly branching tree with eight taxa. All trees within 5% of the length of the most parsimonious tree of 600 steps were retained. The next most parsimonious tree is 6 12 steps in length. To break the monophyly of the Hawaiian taxa requires an additional 16 substitutions on the most parsimonious tree. None of the 20 trees within 5% of the length of the most parsimonious tree has the outgroup Scaptodrosophila located, as hypothesized by Grimaldi ( 1990)) within the Hawaiian taxa ( Idiomyia and Scaptomyza / Engiscaptomyza) . Methods of analysis other than parsimony that are applied to these data all produce the same result. A maximum-likelihood analysis (DNAML; Felsenstein 199 1) of the full data set, where rates of change at the different codon positions are categorized as by Thomas and Hunt ( 199 1) , produces a topology identical to that obtained by maximum parsimony, and this topology is significantly better (P < 0.001) (i.e., has a significantly higher likelihood) than that proposed by Grimaldi ( 1990). The method of Kishino and Hasegawa ( 1989) was used to calculate the mean and variance of loglikelihood differences between the trees. Similarly, Kimura two-parameter-model (Kimura 1980) pair-wise distances (DNADIST ) were calculated on 100 bootstrap replicates of the complete data set (SEQBOOT), and neighbor-joining trees (Saitou and Nei 1987) were produced from each (NEIGHBOR), The majority-rule consensus tree (CONSENSE) was identical to the parsimony tree, and the nodes were supported at similarly high levels (data not shown). Lake ( 1987) method of linear invariants, s which uses a restricted subset of transversional differences, also supports the placement

370

Thomas

and Hunt

of Scaptodrosophila with the mainland Drosophila species (subgenera Drosophila and Sophophora), rather than with the Hawaiian flies (both Idiomyia and Scaptomyza/ Engiscaptornyza) (P < 0.0 1). Times of Divergence Pairwise distances were calculated for synonymous positions, by the method of Li et al. ( 1985 ) (table 3 ) . Rowan and Hunt ( 199 1) molecular-clock s calibration value of 0.0 15 substitutions/ nucleotide/Myr for synonymous positions was used to place a time scale on the divergences shown in figure 3. This value was estimated from pairwise divergences among four species of the planitibia subgroup of the Hawaiian picture-winged Drosophila, for which the times of origin can be assumed to be near the age of the islands on which they are endemic. Because of their similarity in base composition at synonymous sites, we do not expect the divergence times of the Hawaiian species to be much affected by this factor. It is more likely to be important in estimating divergence times between the Hawaiian species and the other taxa, though the direction and magnitude of the effect depends on substitution processes about which very little is known (Thomas and Hunt 199 1). Evidence from Shields et al. ( 1988) indicates that for Adh the distances would be underestimated if their correlation of codon usage with synonymous rates is correct. There are a number of likely evolutionary processes that would result in these distance measures being underestimates, with the degree of bias being correlated with the length of the lineage (Gillespie 1986; Shoemaker and Fitch 1989). Thus the divergence times for the more distant species are probably minimum estimates. Discussion

Our results provide a robust phylogeny relating D. (Sophophora), D. (Drosophila), Idiomyia (Hawaiian Drosophila), and Scaptomyza/ Engiscaptomyza. This phylogeny is also corroborated by two other independent molecular data sets (Beverley and Wilson 1984, 1985; DeSalle 1992a) and by Throckmorton s ( 1975) interpretation of internal morphology. Grimaldi s ( 1990) cladistic analysis of adult morphology yields the same Table 3 .

Downloaded from http://mbe.oxfordjournals.org/ at Universidade Federal De Minas Gerias on June 11, 2012

Pairwise Distances between ADH-coding Regions of 11 Species of Drosophilids, with Estimated Times of Divergence
albovittata heteroneura . . afinidisjuncta adiastola . . . . mimica . . nigra . . . . . . crassifemur . . albovittata . . . mulleri . . . . . pseudoobscura melanogaster 0.7 19 0.7 19 0.662 0.609 0.613 0.306 + 0.095 (24.0 + 0.096 (24.0 k 0.09 1 (22.1 + 0.082 (20.3 f 0.082 (20.4 St 0.049 (10.2 Mya) Mya) Mya) Mya) Mya) Mya) 1.401 1.373 1.253 1.231 1.187 1.160 1.202 1.098 0.882 0.832 lebanonensis + 0.212 (46.7 zk 0.204 (45.8 + 0.177 (41.8 f 0.180 (41.0 + 0.166 (39.6 k 0.159 (38.7 + 0.166 (40.1 + 0.152 (36.6 f 0.12 1 (29.4 rt 0.108 (27.7 Mya) Mya) Mya) Mya) Mya) Mya) Mya) Mya) Mya) Mya)

... . .. .... . . .... .. .. .... ....

1.062 + 0.144 (35.4 Mya) 1.059 + 0.141 (35.3 Mya) 1.259 + 0.181 (42.0 Mya)

NOTE.-Data are mean f standard error distances for synonymous changes and were calculated by the method of Li et al. (1985). Data in parentheses are times of divergence and are based on distances from synonymous differences, under an estimated rate of change of 0.0 15 substitutions/nucleotide/Myr (Rowan and Hunt 199 1). For further explanation, see text.

Drosophila Phylogeny

37 1

unrooted topology for the ingroup taxa; the difference lies in the rooting of these taxa. The Adh sequence data unambiguously place the outgroup taxon, Scaptodrosophila, between D. (Sophophora) and D. (Drosophila). A variety of analyses that make different assumptions about substitution processes all yield the same result-and with high degrees of confidence, where it is possible to estimate confidence intervals. Both relativerate tests ( Sarich and Wilson 1967) and a maximum-likelihood test (Felsenstein 199 1) of a molecular clock show very little or no heterogeneity of evolutionary rate over this tree (fig. 3). Rooting the ingroup taxa between Idiomyia (Hawaiian Drosophila) and Scaptomyza, as hypothesized by Grimaldi ( 1990)) would require us to postulate large rate differences for which we have no evidence. Two additional molecular data sets-larval hemolymph protein immunological distances (Beverley and Wilson 1984, 1985) and mitochondrial DNA sequences (DeSalle 1992a)-result in topologies that are consistent with that produced from Adh sequence data. Throckmorton s ( 1962, 1966, 1968, 1975) work is more difficult to assess, because, as Grimaldi ( 1990) points out, many of Throckmorton s data have not been published, and his methods for inferring phylogenies have not been made completely explicit. Our results are consistent with Throckmorton s 1975 view of relationships among these taxa. Examination of the trees in the publications of Throckmorton ( 1975) and Grimaldi ( 1990) shows the large numbers of drosophilid taxa that are not included in our analysis. It may well be that the similarity of our results to those of Throckmorton ( 1975) is an effect of our sampling of taxa. Late in the writing of the present paper another drosophilid Adh sequence appeared in the literature (Maruyama and Hart1 199 1), from Zaprionus tuberculatus, a member of what Throckmorton ( 1975) calls the immigrans-Hirtodrosophila radiation, which in his view also contains the Idiomyia and Scaptomyza. Both Throckmorton ( 1975) and Grimaldi ( 1990) would place Zaprionus on the internal branch between the Hawaiian taxa and the mainland taxa (fig. 1). Maximum-parsimony and neighbor-joining analyses of the Adh data, performed as above, consistently place Zaprionus on the internal branch between the subgenera Sophophora and Drosophila (fig. 3), though not with overwhelming bootstrap support. The base composition of Zaprionus is not significantly different (P > 0.01) from that of the Sophophora species. DeSalle ( 1992a) analysis of mitos chondrial sequence data places Zaprionus on the branch between Scaptodrosophila and Sophophora. Thus, in neither molecular analysis does Zaprionus group with the Hawaiian taxa, in opposition to the results from morphology. Times of divergence of lineages estimated from the Adh data (Rowan and Hunt 199 1; Thomas and Hunt 199 1) are consistent with available biogeographic and fossil data. Grimaldi ( 1987) has described a Scaptomyza species from Dominican amber that is thought to date from the early Miocene ( =24 Mya). Our estimation that the split between the Scaptomyza/Engiscaptomyza lineage and the lineage giving rise to Idiomyia (Hawaiian Drosophila) occurred ~27 Mya corresponds well with this fossil date. The estimation that the split between D. (Engiscaptomyza) crasszjemur and Scaptomyza albovittata occurred = ld Mya is intriguing because these Hawaiian taxa appear at opposite extremes of the adult morphology-based cladogram (Grimaldi 1990) for scaptomyzoid species (sensu Kaneshiro 1976). As a result of the similarity in estimated times for the apparent diversification of the Idiomyia and scaptomyzoid lineages, we tentatively prefer the explanation that would have these lineages each arriving separately on the Hawaiian archipelago some 10 Mya (Thomas and Hunt 199 1) . Alternatively, and cladistically more parsimoniously, our data could be said

372

Thomas and Hunt

to support a single introduction to Hawaii >27 Mya, on the assumption that habitable islands have existed continuously for that length of time, a situation that is by no means clear. There is evidence from bathymetry and projection (H. L. Carson, personal communication) that supports the status of Necker and Nihoa as high islands from -7 Mya, respectively. It will be necessary to determine Adh sequences -1OMyaand from mainland Scaptomyza and related taxa such as Grimshawomyia, to test Throckmorton ( 1975 ) conjecture that Scaptomyza evolved in Hawaii and spread from there s to the rest of the world. This work is in progress. In our opinion the sum of data from morphological and molecular sources is not yet adequate to produce a reliable classification of this complex group. Our data do show that Grimaldi ( 1990) classification of the taxa discussed in the present paper s is paraphyletic, just as is the traditional classification (Wheeler 198 1) under Throckmorton ( 1975 ) interpretation s of relationships. The available data provide a robust set of relationships among the taxa examined here, which may be of some comfort to nonsystematists. Additional Adh sequences are needed from a wide range of subgenusDrosophila species, from mainland Scaptomyza, and particularly from representatives of the mycophagous genera with which Grimaldi s ( 1990) analysis allies Zdiomyia (Hawaiian Drosophila). Sequence data from additional nuclear and mitochondrial loci will likewise be very useful. The interesting problem of determining the processes responsible for the disparity between the adult-morphology results and the molecular results awaits further work. If our view of these taxa is correct, then we have to explain the apparent reversals to primitive states for several characters of adult morphology in the Zdiomyia (Hawaiian Drosophila) and Scaptomyza/ Engiscaptomyza. These characters mostly involve the head, such as the eye setulae, vibrissae, lacinia, cibarium, and, mostly strikingly of all to our vertebrate eyes, the facial carina. Conversely, if we are wrong, we must explain, at the least, a substantial number of character reversals and homoplasies on the molecular level, in three independent data sets. The Drosophilidae provide us with the first group for which we may have the tools-molecular, genetic, and developmentalto answer such questions ( DeSalle and Grimaldi 199 1) . Acknowledgments

We are grateful to David Grimaldi for his extensive comments on an early version of the manuscript and for a prepublication copy of a paper. Rob DeSalle discussed his results with us and supplied copies of his papers prior to publication. Bill Heed and Henar Alonso-Pimentel made comments on the manuscript, and H. L. Carson provided helpful discussions. Discussions with Nick Arnold and Darrell Siebert of the BM( NH) helped alleviate our confusion over the comments of two anonymous reviewers of an earlier version and improved the presentation. They are all blameless for our conclusions. We gladly acknowledge the SERC Daresbury Laboratory SEQNET molecular biology computing facility for providing access to up-to-the-minute versions of sequence data bases. Part of this work was funded by National Science Foundation grant BSR 85-2 18 10 to J.A.H.
LITERATURE CITED

ALABAT, R., and R. GONZALEZ-DUARTE. 1990. Nucleotide sequence of the A& gene of Drosophila lebanonensis. Nucleic Acids Res. l&6706. BENYAJATI,C., A. PLACE, D. POWERS,and W. SOFER. 198 1. Alcohol dehydrogenase gene of

Downloaded from http://mbe.oxfordjournals.org/ at Universidade Federal De Minas Gerias on June 11, 2012

DrosophilaPhylogeny 373 Drosophila melanoguster: relationship of intervening sequences to functional domains in the proteins. Proc. Natl. Acad. Sci. USA 78:27 17-272 1. BEVERLEY,S. M., and A. C. WILSON. 1982. Molecular evolution in Drosophila and higher Diptera. I. Micro-complement fixation studies of a larval hemolymph protein. J. Mol. Evol. l&25 l-264. 1984. Molecular evolution in Drosophila and the higher Diptera. II. A time scale for fly evolution. J. Mol. Evol. 21:1-13. -. 1985. Ancient origin for Hawaiian Drosophilinae inferred from protein comparisons. Proc. Natl. Acad. Sci. USA 82:4753-4757. BREZINSKY,L., T. D. HUMPHREYS, and J. A. HUNT. 1992. Evolution of the transposable element Uhu in five species of Hawaiian Drosophila. Genetica 86:2 l-35. CABOT, E. L., and A. T. BECKENBACH.1989. Simultaneous editing of multiple nucleic acid and protein sequences with ESEE. CABIOS 5:233-234. DANIEL& S. B., A. CHOVNICK, and I. A. BOUSSY. 1990. Distribution of hobo transposable elements in the genus Drosophila. Mol. Biol. Evol. 7:589-606. DESALLE,R. 1992a. The origin and possible time of divergence of the Hawaiian Drosophilidae: evidence from DNA sequences. Mol. Biol. Evol. 9:905-916. . 1992b.The phylogenetic relationships of flies in the genus Drosophila deduced from mtDNA sequences. Mol. Phylogenet. Evol. 1:3 l-40. DESALLE, R., and D. A. GRIMALDI. 199 1. Morphological and molecular systematics of the Drosophilidae. Annu. Rev. Ecol. Syst. 22:447-475. FELSENSTEIN,J. 1985. Confidence limits on phylogenies: an approach using the bootstrap. Evolution 39:783-79 1. . 199 1. PHYLIP, version 3.4. University of Washington, Seattle. FISCHER, J. A., and T. MANIATIS. 1985. Structure and transcription of the Drosophila mulleri alcohol dehydrogenase genes. Nucleic Acids Res. 13:6899-69 17. GILLESPIE,J. H. 1986. Rates of molecular evolution. Annu. Rev. Ecol. Syst. 17:637-665. GRIMALDI, D. A. 1987. Amber fossil Drosophilidae (Diptera), with particular reference to the Hispaniolan taxa. Am. Museum Novitates 2880: l-23. . 1990. A phylogenetic, revised classification of genera in the Drosophilidae ( Diptera) . Bull. Am. Museum Nat. Hist. 197: I- 139. KANESHIRO, K. Y. 1976. A revision of generic concepts in the biosystematics of Hawaiian Drosophilidae. Proc. Hawaiian Entomol. Sot. 22:255-278. KIMURA, M. 1980. A simple method for estimating evolutionary rate of base substitutions through comparative studies of nucleotide sequences. J. Mol. Evol. 16: 11 l-120. KISHINO, H., and M. HASEGAWA. 1989. Evaluation of the maximum likelihood estimate of the evolutionary tree topologies from DNA sequence data, and the branching order in Hominoidea. J. Mol. Evol. 29: 170-179. LAKE, J. A. 1987. A rate-independent technique for analysis of nucleic acid sequences: evolutionary parsimony. Mol. Biol. Evol. 4: 167-19 1. LI, W.-H., C.-I. Wu, and C.-C. LUO. 1985. A new method for estimating synonymous and nonsynonymous rates of nucleotide substitution considering the relative likelihood of nucleotide and codon changes. Mol. Biol. Evol. 2: 150-l 74. MARUYAMA,K., and D. L. HARTL . 199 1. Evidence for interspecific transfer of the transposable element mariner between Drosophila and Zaprionus. J. Mol. Evol. 33:5 14-524. ROWAN,R. G., and W. J. DICKINSON.1988: Nucleotide sequence of the genomic region encoding alcohol dehydrogenase in Drosophila ufinidisjuncta. J. Mol. Evol. 28:43-54. ROWAN, R. G., and J. A. HUNT. 199 1. Rates of DNA change and phylogeny from the DNA sequences of the alcohol dehydrogenase gene for five closely related species of Hawaiian Drosophila. Mol. Biol. Evol. 8:49-70. SAITOU,N., and M. NEI. 1987. The neighbor-joining method: a new method for reconstructing phylogenetic trees. Mol. Biol. Evol. 4:406-425.

374

Thomas and Hunt

SARICH, V. M., and A. C. WILSON. 1967. Rates of albumin evolution in primates. Proc. Natl. Acad. Sci. USA 58: 142- 148. SCHAEFFER,S. W., and C. F. AQUADRO. 1987. Nucleotide sequence of the Ad/z gene region of Drosophila pseudoobscura: evolutionary change and evidence for an ancient gene duplication. Genetics 117:6 l-73. SHIELDS, C., P. M. SHARP,D. G. HIGGINS, and F. WRIGHT. 1988. Silent sites in Drosophila D. genes are not neutral: evidence of selection among synonymous codons. Mol. Biol. Evol. 5:704-7 16. SHOEMAKER, S., and W. M. FITCH. 1989. Evidence from nuclear sequences that invariable J. sites should be considered when sequence divergence is calculated. Mol. Biol. Evol. 6:270289. STURTEVANT,A. H. 192 1. The North American species of Drosophila. Carnegie Institution, Washington, D.C. SULLIVAN,D. T., P. W. ATKINSON, and W. T. STARMER. 1990. Molecular evolution of the alcohol dehydrogenase genes in the genus Drosophila. Evol. Biol. 24: 107- 147. SWOFFORD,D. L. 1990. PAUP: phylogenetic analysis using parsimony, version 3.0. Illinois Natural History Survey, Champaign. THOMAS, R. H., and J. A. HUNT. 199 1. The molecular evolution of the alcohol dehydrogenase locus and the phylogeny of Hawaiian Drosophila. Mol. Biol. Evol. 8:687-702. THROCKMORTON,L. H. 1962. The problem of phylogeny in the genus Drosophila. Univ. Tex. Publ. 6205:207-343. 1966. The relationships of the endemic Hawaiian Drosophilidae. Univ. Tex. Publ. 6615:335-396. -. 1968. Concordance and discordance of taxonomic characters in Drosophila taxonomy. Syst. Zool. 17:355-387. . 1975. The phylogeny, ecology and geography of Drosophila. Pp. 421-469 in R. C. KING, ed. Handbook of genetics. Plenum, New York. WHEELER,M. R. 198 1. The Drosophilidae: a taxonomic overview. Pp. l-97 in M. ASHBURNER, H. L. CARSON, and J. N. THOMPSON,JR., eds. The genetics and biology of Drosophila. Academic Press, London. WHEELER, W. C. 1990. Nucleic acid sequence phylogeny and random outgroups. Cladistics 6:363-367.
Downloaded from http://mbe.oxfordjournals.org/ at Universidade Federal De Minas Gerias on June

MARTIN KREITMAN, reviewing Received Accepted April 20, 1992; revision September 29, 1992

editor received September 17, 1992