Phylogeny of the Drosophila obscura group as inferred from one- and two-dimensional protein electrophoresis

The phylogenetic relationships of 15 species of the obscuru group of Dro.sophih were analysed by use of one- and two-dimensional electrophoresis. Genetic distances based on two-dimensional data are five times smaller than those based on native proteins. From the data, it is proposed that the species radiation of the obscuru group happened in two evolut~onary bursts, the first one giving rise to at least four palearctic proto-lineages (bifasciata, obscura (including D. subsilvestris), subobscura, and microlabis) and one or two proto-nearctic lineages (aflnis, pseudoobscura), and the second, more recent bust giving rise to the current speciation within lineages.


Introduction
The phylogenetic relationships among species of the obscura group of Drosophila have been intensively studied using traits of different complexity. On the basis of morphological characters, the group has been divided into two subgroups, obscura and afinis, both including nearctic as well as palearctic species (Sturtevant 194% Buzzati-Traverso and Scossiroli 1955). The use of isozymes for &acing the evolution of the group has given rise to some discrepancies with respect to this traditional phylogeny. In a summary of their pioneering work, Lakovaara et al. (1976) divided the group into three distinct evolutionary branches, one lineage comprising the Eurasian obscura subgroup species, a second the American afirzis subgroup species, and a third the American obscura subgroup species. Nevertheless, two species, D. alpina and D. helvet~ca, belonging morpho~ogically to the obscura and the affink subgroup, respectively, have not proven to be appreciably related to the other species. This group subdivision was supported by later enzymatic studies (Pinsker and Buruga 1982;h u k a s et al. 1984). A fourth subgroup comprising the recently discovered new African species (Tsacas et al. 1985) has been added and termed the microlabis subgroup (Chriou et al. 1988). However, other authors, also using enzymatic traits, have obtained different phylogenies, mainly with respect to the obscura subgroup. Some palearctic species of this subgroup appear to be at least as distantly related to the other species of the obscura subgroup as to the nearctic species of the pseudoobscura subgroup (Marinkovi6 et al. 1977; Cabrera et al. 1983). Several new phylogenies of the group, obtained using n~itochondrial-DNA-sequence (mtDNA) data have recently been proposed. The results can be summarized as follows: the &a&tional division of the group into two subgroups ( a f i i s and obscura) does not correspond to the true phylogeny of the group (Barrio et al. 1992); D. afinzs and D. pseudoobscura subgroups seem to be monophyletic groups with closer affinities to one another than to the palearctic D. obscura subgroup species (Beckenbach et a1. 1993). The palearctic obscura subgroup seems to be a heterogeneous assemblage that could further be subdivided into several independent complexes (Gonzhlez et al. 1990; Barrio et al. 1994).
With the aim of reconciling the different phylogenies proposed, new molecular studies with representative species of all the subgroups hitherto described have been started. In this study, the phylogenetic relationships of 15 species were studied using one-and two-dimensional protein electrophoresis, a technique, that, by the inclusion of proteins with slower rates of evolutionary changes than the traditional allozymes, appears to be more useful for inferring the phylogenies of distantly related taxa (Spicer 1988).

Species
A total of 15 different Drosophila obscuru group species were used in this study. Their geographic origin and sources are listed in Table 1. Sample preparation For two-dimensional electrophoresis (2DE), 20 mg adult male flies of each species, approximately 20 individuals each, were homogenized in TE buffer (10 mM TrisHCl, 1 mM EDTA, pH 8) with 8M urea, 1% Nonidet P-40 detergent, 1% mercaptoethanol, and 5% Pharmacia ampholytes (2-D Pharmalyte). For one-dimensional native electrophoresis (IDNE), samples were simply homogenized in TJ3 buffer. For one-dimensional sodium-dodecyl-sulfate (SDS) denatured electrophoresis (IDDE), samples were homogenized in TE buffer with 2.5% SDS and 5% mercaptoethanol, and heated at 100OC for 5 min. In all cases, homogenates were centrifuged at 300 000 x g for 15 min in a Beckman L7-55 ultracentrifuge and the Supernatants kept at -70Â° until use. On gels, lpl of each sample, at a concentration of 40 pg of wet fly weighvpl, was applied.

Eiectrophoresis
Electrophoresis were carried out using the PhastSystem equipment from Pharmacia LKB Biotechnology. Protocols and programmed conditions were, with minor modifications, those described in the PhastSystem owner's manual. For ZDE, first-dimension isoelectric focusing was performed using PhastGel IEF 3-9 and the PhastGel sample applicator 811. After electrophoresis, gels were stained with Coom&sie blue to locate the sample lanes, which were cut and used in the second dimension with PhastGel gradient 8-25, and PhastGel SDS buffer strips. For lDNE, PhastGel gradient 8-25 and PhastGel native buffer strips were used. For lDDE, PhastGcl gradient 8-25 was also used but with PhastGel SDS buffer strips. ACOSTA, PINTO, HERNANDEZ, GONZALEZ, CAE~REM and LARRUGA Detection Gels were first Coomassie-blue stained and then silver stained following the protocol of the PhastGel-silver-kit instmction manual.

Data analysis
Pattems of bands or spots on the gels were compared between species by presence-absence criteria. Similarity was analysed using the Dice , where a is the number of matches (bands or spots shared by the two species compared) and u is the total number of mismatches (spots present in only one of these species). Following Spicer (1988), the logarithmic transformation of the similarity coefficient (-In SD) was used as a measure of distance. Phylogenetic relationships were analysed assuming constant (UP-GMA, Sneath and Sokall973) and variable (NJ. Saitou and Nei 1987) evolutionary rates, using thi MEGA 1.0< pr&gram (Kumar et a;.

Results
A mean of 125 k 8.3 protein spots were reliably scored in the 2DE gels, 37.8 4.1 bands was the mean for lDNE gels, and 41.6 1.0 bands was the mean for the l D D E gels. An example of a rutinary two-dimensional silver-stained gel is presented in Figure 1. Allelic differences among species are difficult to identify in two-dimensional electrophoresis (Avise 19831, therefore, spots were simply scored as either present or absent. Figure 2 shows located examples of the type of interspecific variation detected in 2DE gel comparisons. Matrices of genetic distances were calculated for 2DE (Table  21, lDNE (Table 3) and lDDE (Table 4)    with 2DE, subobscura-madeirensis was the closest pair: with 1DNE. the closest pair was guanche-madeirensis; and with lDDE, the order of the triad was not resolved at all (Figs 3,4. 5). Discrepancies in the location of the branching points of the main clusters seem to indicate that all are very distinct and well-differentiated lineages. It seems that the radiation of the obscuru group species had two important bursts, well separated in time. The older one gave rise to several branches. that, taking in account the species studied here. are actually represented by the bifasciata. pseudoob~cura. affit~is. microlabis, obscura (including s-~ibsi1vestri.s). and subob.scura lineages. The second radiation gave rise to the extant species within these main lineages.

Discussion
The accumulation of new data and the addition of newly discovered species have changed the classical phylogenetic view of the D. obscuru group, giving rise to a more congruent picture. It seems that the subdivision of the group into two subgroups, obscura and afinis, does not hold up anymore. All studies based on molecular data, beginning with the isozyme analysis by Lakoovara et al. (1972), as well as subsequent studics at the DNA level (Latorre et al. 1988;Goddard et al. 1990: Beckenbach et al. 19931, have demonstrated that the nearctic pseudoobscuru lineage of the obscuru subgroup is more rclated to the nearctic affinis subgroup than to the palearctic branch of its own subgroup. All clustering in this stud!. based on protein data. are also in accordance with this supposition. Chromosome homologies also support this assumption (Sturtevant and Novitski 1941;Lakoovara and Saura 1982  Abbreviations as in Table 1 subgroup than to the other palearctic species of the subgroup. The same situation, also using isozyme data, was later found for the palearctic species D. subobscuru (Marinkovie et al. 19781, D. ambigua (Cabrera et al. 1983), and D. ob.scura (Cariou et al. 1988). These findings led to the formulation of a new phylogenetic hypothesis for the species of the obscuru group: a common ancestor gave rise to various lines in the palearctic region, one of which led to the American species and the others to the different European clusters (Cabrera et al. 1983). This polyphyletic origin of the palearctic D. obscura subgroup representatives has been repeatedly confirmed by numerous studies of DNA data (Latorre et al. 1988;Gonzilez et al. 1990;Goddard et al. 1990;Barrio et al. 1992;Marfany and Gonzilez-Duarte 199% Ruttkay et al. 1992;Beckenbach et al. 1993;Barrio et al. 1994 proposed for these African species (Cariou et al. 1988). The results of this study, based on protein data, are in total agreement with both the polyphyletic origin of the palearctic lineages of the obscwa subgroup and the independent subgroup of the African species (Figs 3,4,5). However9 none of the data presented here clarify the chronology of the speciation events leading to the major lineages proposed. In this respect, these data only support the incorporation of D. dsiZvestris within the obscura lineage, as Lakoovara et al. (1972) already showed, and as the majority of studies carried out at DNA level have confirmed (e.g. Gonz6lez et al. 1990;Barrio et al. 1994). The phylogeny obtained with 1DDE data (Fig. 51, bringing the African species closer to the common ancestor of the group, though in accordance with one of the phylogenies proposed by Ruttkay et al. (19921, was not confirmed by either 2DE (Fig. 3) or 1DNE (Fig. 4) data, whose topologies are more in agreement with those proposed by Cariou et al. (1988) and Brehm et al. (1991). Finally, D.

Microlabis
Kiturnensis bijiasciata appeared as an independent lineage that is not closely related to any other species, a result that is also in agreement with the majority of the phylogenies constructed where this species is included. It seems that, for the resolution of the hierarchy in this early radiation of the group, analyses of slower-evolving sequences and ditTerent sets of out-group species should be used. With regard to the phylogenetic relationships among related species, it is evident that representatives of the ajJ?nis9 pseudoobscura and subobscura subgroups are all monophyletic clusters. The topology of the triad pseudoobscurapersirnilis-miranda with the first two species closer to each other (Figs 3, 4, 5) is in agreement with all previous analyses of this subgroup (Lakoovara et al. 1972Goddard et a]. 1990Beckenbach et al. 1993;Barrio et al. 1992Barrio et al. , 1994. With respect to the species triad obscura+zmbigua-tristis, the data in this study favour the obscura-ambigua pair as being more related (Figs 3, 4, 5). This disagrees with the first studies based on electrophoretic data (Lakoovara et al. 1972(Lakoovara et al. , 1976Pinsker and Bumga 1981) that considered obscura-trktis to be the closest pair. It is, however, in accordance with the majority of the more recent phylogenies obtained from very different sources of data such as chromosomal histone-gene localization (Eelger and Pinsker 198' 7), specific satellitemDNA evolution (Bach-AOSTAY PINTO, HERNANDEZ, GONZ~LEZ, CABRERA and LARRUGA cytb sequences. The relative position of the three species in the subobscura cluster is more controversial. With the 2DE data, ihe species subobscura-madeirensis form a cluster (Fig. 3); with the 1DNE data, madeirensis-guanche is the most related pair (Fig. 4); and the 1DDE information fails to resolve the k p l o g y of the triad at a11 (Fig. 5). Phylogenies suggesting a c k m relationship of D. madeirensis with D. subobscura, as W e d here with the 2DE set of data, were also obtained by b a u g a and Pinsker (1984) and hukas et al. (1984) from k q m e data. No discrepancies exist when chrommmal homologies (Krimbas and h u k a s 1984; Molto et al. 1987;-i t and Prevosti 1989;B r e h andKrimbas 1990a, b, l!W2,1993;Brehm et al. 1991), and analysis of hybrid sterility and developmental incompatibilities (Khadem and Krimbas lWla, b, 1993;Papaceit et al. 1991) among these species are e. The general agreement is that D. sub&scura and D. d u e m i s are the closer species pair. From the comparison of polytene chromosome patterns (Brehm and Krimbas 1992Krimbas , 1993, D. panche appeared to be the most ancestral species of this tiad. The existence of a D. guanche specific-satellite DNA, not present in the genome of its sibling species D. subobscura and D. mdeirensis (Bachmann et al. 1989), has been explained by considering the continental species D. subobscura as the parental species from which D. guanche and D. madeirensis originated through founder effects and geographic isolation. Using restriction-site maps of mtDNAy Gonzilez et al. (1990) found, at a significant level, subobscura-madeirensis to be the closest pair of species. From four independent mtDNA sequences of these species, all but one analysis clustered madeirensis-subobscura (Bamo et al. 19941, and unpublished results (T. Acosta et al.) based on cytb sequences also favour the same pair relationship. Taking all this into account, the species D. madeirensis and D. subobscura should definitively be considered as being more related to each other than to D. gwnche.
There are some hypothesis and time estimates for the evolution in the obscura group that are worth contrasting with the data presented here. In biogeographic analyses, Throck- morton (1975) proposed that the obscura group founder separated at the time of the temperate forest disjunction in mid-Miocene period, and that the final developments of present-day patterns were added by speciation events occurring within local regions during the Pliocene and Pleistocene ages that have continued into recent times. The hypothesis of two main radiation bursts is in good ageement with that proposition. If the calibration values for two-dimensional electrophoresis data (Spier 1988: estimated at 118 million Drosophila obsczua group protein phylogeny years per unit of is applied to the 2DE data in this study, t h p a h d c proto-lineages of bifasciata, obscura, mbdmnm, nicdubk and one or two proto-nearctic lineages

( -a )
would have radiated about 9-10 m i B k ~UI S ago m the mid-Miocene period. Applying the ai&ah 6u the second and more recent burst, that gave ~~~s p e c i a t i o n within these lineages, we obtain values in af 2 4 million years, which are in accordance with @@cal events and fit well with time estimates given by a m&cm for these radiations (