The Drosophila Serido Speciation Puzzle: Putting New Pieces Together

The Drosophila serido speciation puzzle: putting new pieces together

Alfredo Ruiz1, Alessandra M. Cansian2, Gustavo C. S. Kuhn2, Maurilio A. R. Alves2 &Fabio M. Sene3 1Departament de Genètica i de Microbiologia, Facultat de Ciències, Universitat Autònoma de Barcelona, 08193 Bellaterra (Barcelona), Spain (Phone: 93-581-2729; Fax: 93-581-2387; E-mail: [email protected]); 2Departamento de Biologia, Faculdade de Filosoﬁa, Ciencias e Letras de Ribeirão Preto, Universidade de São Paulo, Ribeirão Preto, SP, 14049-900 Brazil; 3Departamento de Genética, Faculdade de Medicina de Ribeirão Preto, Universidade de São Paulo, Ribeirão Preto, SP, 14049-900, Brazil

Accepted 3 August 2000

Key words: cactophilic Drosophila, host plant speciﬁcity, inversions, karyotype, speciation

Abstract The D. serido superspecies is a complex mosaic of populations distributed over a vast part of South America and showing various degrees of genetical divergence. We have analyzed its chromosomal constitution in 16 new localities of southeastern and southern Brazil. Both the metaphase and salivary gland chromosomes show a sharp split of these populations in two groups. Four populations, fixed for inversion 2e8 and showing the type I karyotype, represent the southwestern limit of D. serido type B, which inhabits the Cerrado in central-western Brazil. The remaining populations are homozygous for 2x7, an inversion also fixed in the Caatinga populations of northeastern Brazil. However, their karyotype, in those populations analyzed, belong to a different type (V) from that of the Caatinga populations. Populations in this second group are polymorphic for five inversions on chromosome 2 plus another on chromosome 5 and show considerable interpopulation differentiation. The breakpoints of chromosome 2 inversions are described and the inversion loops of several heterokaryotypes are presented. Biogeographical information suggests that there are clear ecological differences between the two groups of populations as well as among the populations within the second group. The possible role of host plants in promoting the genetic divergence among the D. serido populations is discussed.

Introduction to occur in Oceanic Islands, e.g. the Hawaiian Ar- chipelago, scenarios prone to founder effects are also In the genus Drosophila, allopatric speciation is al- possible in continental settings (Giddings, Kaneshiro most certainly the rule, although there is evidence that & Anderson, 1989). sympatry may enhance premating isolation (Coyne & The second question concerns the role that natural Orr, 1997; Powell, 1997). Two unresolved questions selection plays in the speciation process. Under the related to allopatric divergence remain. One is the size founder effect model, genetic divergence is achieved of the divergent populations. In many cases, species at least in part without the concourse of, or even seem to arise by slow genetic divergence and sub- against the effect of, natural selection, which oth- sequent reproductive isolation of geographically sep- erwise plays the preeminent role under the adaptive arated and differentially adapted races and subspecies divergence model. In the latter case, however, the (Dobzhansky, 1970). By contrast, divergence may be simple acceptance of the operation of natural selection relatively faster when a new population is established leaves us with an incomplete account of the speci- by a few original founders carrying only a small frac- ation process. Identification of the selective agents tion of the genetic variation of the parental population involved is necessary, these selective agents may be (Mayr, 1963; Carson, 1975). While this is more likely abiotic factors (such as temperature, light or humidity) 218 or biotic factors (such as predators, competitors or observation, Ruiz and Wasserman (1993) proposed to parasites). Although we may ultimately conclude that include them tentatively within D. koepferae. Tidon- each speciation event is unique in this regard, we sug- Sklorz and Sene (1995a) placed them in a dendogram gest that some ecogeographical rules of speciation in closely related to D. koepferae. Populations of D. Drosophila will eventually emerge. serido from the Cerrado in central-western Brazil are In addition to the traditional advantages of Droso- homozygous for inversion 2e8 (Tosi & Sene, 1989), phila for genetical studies, cactophilic species of the an inversion which is also fixed in its close relative D. D. repleta group (Barker & Starmer, 1982; Barker, borborema (Wasserman & Richardson, 1987). Finally, Starmer & MacIntyre, 1990) are particularly suitable the populations inhabiting the rocky fields of the Ca- to the study of speciation. One of the more intricate deia do Espinhaço, States of Bahia and Minas Gerais, and intriguing situations is found in the neotropical are also fixed for 2e8 but show a consistent morpho- Drosophila serido superspecies which belong to the logical and genetic differentiation from the rest of D. buzzatii complex of the mulleri subgroup (Wasserman, serido populations and have been described as a separ- 1992; Ruiz & Wasserman, 1993). D. serido was de- ate species under the name D. seriema (Tidon-Sklorz scribed in 1977 from flies collected in Milagres, State & Sene, 1995b; Kuhn et al., 1996). In summary, the of Bahia (Brazil) (Vilela & Sene, 1977). At that time D. serido ‘superspecies’ is split at present into two its known geographical range was restricted to north- species, D. koepferae and D. seriema, plus a num- eastern and central Brazil. Subsequent work expanded ber of insufficiently known populations which are still its range, eastwards to the Bolivian Andes and south- designated with the specific name D. serido. wards to Argentina, and revealed a complex mosaic of We report here the chromosomal constitution of D. geographically dispersed populations with various de- serido populations from 16 new localities in southeast- grees of genetical divergence (Sene, Pereira & Vilela, ern and southern Brazil, a previously unstudied area. 1982, 1988; Ruiz, Fontdevila & Wasserman, 1982; We provide ecological information on this region in- Fontdevila et al., 1988). Up to six different meta- cluding the cactus species used as host plants by D. phase karyotypes (I–VI) were found (Baimai, Sene serido, describe three new polymorphic inversions and & Pereira, 1983) and up to five distinguisable ae- reinterpret some previous cytological results. Jointly, deagus types (A–E) were described (Silva & Sene, these genetical and ecological data considerably cla- 1991). These contrasting patterns of differentiation rify the biogeography of D. serido and shed light on made with various genetic markers made the polytypic the factors promoting genetic differentiation within D. serido a ‘speciation puzzle’. this polytypic taxon. Although not causally related with the origin of reproductive isolation (Zouros, 1982; Powell, 1997), chromosomal inversions usually provide considerable Materials and methods information on patterns of geographical differentiation and gene flow as well as the origin of particular Drosophila serido adults were obtained from 16 local- species (Carson, 1987; Ruiz, Heed & Wasserman, ities (Figure 1) in the States of Minas Gerais and São 1990). D. serido was found to be highly polymorphic Paulo in southeastern Brazil and the States of Parana, for inversions, chiefly on the second chromosome Santa Catarina and Rio Grande do Sul in southern (Ruiz, Fontdevila & Wasserman, 1982; Wasserman & Brazil. All localities belong to a general ecological Richardson, 1987; Tosi & Sene, 1989). From the dis- region classified as forest although clear ecological tribution of D. serido inversions, the following pattern differences among them do exist. In most cases, flies has emerged: populations from the Caatinga in north- were collected with banana-orange baits and a nylon eastern Brazil are fixed for inversion 2x7 (Wasserman net. Whenever possible, rotting cactus stems (rots) & Richardson, 1987; Tosi & Sene, 1989). Those from were located in the field, wrapped in newspaper, and northwestern Argentina and Bolivia are fixed for in- returned to the laboratory where they were placed in version 2j9 (Ruiz, Fontdevila & Wasserman, 1982; closed glass containers. Nearly 2000 flies emerged Tosi & Sene, 1989) and have been described as a from 39 rots in the first two weeks were collected with separate species under the name of D. koepferae (Font- an insect aspirator and classified to species (Table 1). devila et al., 1988). Likewise, the populations from the Most laboratory stocks were founded with single wild eastern Chaco in Argentina are presumably fixed for females but some were established using several fe- inversion 2j9 (Tosi & Sene, 1989), and based on this males. The specific identity of all D. serido stocks was 219

Figure 1. Geographical map of central-southern Brazil showing the position of the 16 localities where Drosophila serido was collected. Key (collection codes are given in parenthesis): 1. Peixoto, MG (H27); 2. Furnas, MG (H24/H26); 3. Morro do Forno, Altinópolis, SP (H6); 4. Analândia, SP (H32); 5. Sertãozinho, SP (H34); 6. Campinas, SP (H48); 7. Rio Ligeiro, PR (D92/D93); 8. Santiago, RS (H47); 9. Guaritas, RS (H44); 10. Osório, RS (H59); 11. Tramandaí, RS (H58); 12. Arroio Teixeira (H41/H61); 13. Capão da Canoa, RS (H42/H62); 14. Araranguá, SC (H63); 15. Florianópolis, SC (H64); 16. Guaratuba Beach, SP (H49). Major morphoclimatic domains are indicated. Other places shown on the map: Porto Alegre, Puerto Tyrol and Rio de Janeiro.

Table 1. Individuals of Drosophila species emerged from 39 cactus rots collected in four localities of southern Brazil

Locality Cactus species N D. serido D. buzzatii D. meridionalis D. mercatorum D. hydei Other species Total

8 Cereus hildmannianus 2 260 6 6 272 9 Opuntia monacantha 23 574 862 15 34 37 5 1527 Cereus hildmannianus 110101 21 12 Opuntia monacantha 12 8 59 6 73 13 Cereus hildmannianus 19 22 4 35

Total 39 851 872 112 35 37 21 1928

N = number of rots. 220

Table 2. Metaphase chromosomes of the 24 stocks of Dro- is also occasionally present at the southernmost areas sophila serido analyzed in this study (e.g locality 9) and can be used by D. serido as well Locality (number of stocks) Metaphase chromosome (Table 1). Five localities are sand dunes on the Atlantic XY6 Litoral of Santa Catarina (14) or Rio Grande do Sul (10–13). Here C. hildmannianus as well as O. mon- 2 (6), 3 (1), 4 (4) t st d acantha were present. Our rearing records (Table 1) 5 (1), 7 (5), 8 (1), 9 (2), t t t show that both Cereus and Opuntia are used by D. 12 (2), 13 (1), 15 (1) serido as host plants. The remaining two localities are t = telocentric; st = acrocentric; d = dot chromosome. peculiar. Guaratuba (locality 16) is very unusual for cactophilic Drosophila. A population of D. serido was found in a narrow strip of rocky shore separating the confirmed by visual inspection of the male genitalia Forest (Mata Atlantica) from the Ocean where a single (see Discussion). non-epifitic cactus, Cereus pernanbucensis, could be We analyzed the metaphase chromosomes of 24 seen. D. serido has previously been shown to emerge stocks derived from 10 localities (Table 2). Metaphase from this species of Cereus (Bizzo & Sene, 1982; plates were prepared (seven larvae per stock) from Pereira, Vilela & Sene, 1983). Florianópolis (locality larval brain ganglia (Baimai, Sene & Pereira, 1983). 15) shows a similar setting but differs by the presence In addition, the salivary gland chromosomes of 90 of C. hildmannianus and O. monacantha. laboratory stocks from 16 localities (Table 3) were studied. Third instar larvae were dissected in ethanol– Metaphase chromosomes of D. serido populations acetic acid (3 : 1) and the salivary gland chromosomes stained with a drop of acetic–lactic orcein (2%). Eight Two different karyotypes were found in the 10 pop- to 10 larvae were dissected per stock. Finally, we ulations analyzed (Table 2). One was found in three analyzed the salivary gland chromosomes of hybrid localities (2–4) from southeastern Brazil and consisted larvae (8–10 per cross) produced by two interspecific of four pairs of telocentric autosomes, one pair of combinations (female parental is given first): D. serido small-dot chromosomes, a telocentric X and an ac- (stock H24S3 from Furnas, MG) × D. seriema (stock rocentric Y. This karyotype was described as Type I D71C1 from Morro do Chapéu, BA); and D. koepferae by Baimai, Sene and Pereira (1983). The other one (stock B26D2 from Famatina, Argentina) × D. serido differed from the first by having a larger microchromo- (stock H34G8 from Sertãozinho, SP). some and a longer Y chromosome and corresponds to Type V in Baimai, Sene and Pereira (1983). These au- thors found the latter karyotype to be restricted to two Results localities only. Here we found a much wider distribution of this karyotype including seven more localities Biogeography and host plants (5, 7–9, 12, 13, 15) from southern Brazil and the Atlantic Litoral. All localities studied here (Figure 1) share the presence of one or more cactus species which provide the Salivary gland chromosomes of D. serido populations feeding and breeding substrate for D. serido. Four localities (1–4) are rocky fields on the top of isol- The chromosomes of all D. serido populations ana- ated hills (morros) of the States of Minas Gerais and lyzed here differ from the standard sequence of São Paulo. Here, a single cactus species, Pilosocereus the buzzatii cluster, Xabc;2abmnz7;3b;4;5 (Ruiz & machrisii (Zappi & Taylor, 1993), is present and in- Wasserman, 1993) by a fixed paracentric inversion on tensive collections at one site (3) have shown that it the second chromosome. However, two different in- is used as host plant by D. serido (Moraes, 1997). versions, 2e8 and 2x7 (Figure 2), were found fixed Five localities (5–9) are found in or near inland valleys in different populations. Four populations from south- of the Paraná River watershed (States of São Paulo, eastern Brazil (1–4) were homozygous for inversion Paraná and Rio Grande do Sul). The dominant cactus 2e8 whereas 12 populations from southern Brazil and in these places (sometimes the only one present) is the Atlantic Litoral (5–16) were fixed for inversion 2x7 Cereus hildmannianus, which is often associated with (Table 3). Thus, the ancestral arrangement of chromo- the relictual vegetation of the Chaco, and is also used some 2, 2abmnz7, which would have been called the by D. serido (Table 1). However, Opuntia monacantha D. serido standard if present, was not found. 221

Table 3. Chromosome arrangements found in the 90 stocks of Drosophila serido analyzed in this study

Locality Number Chromosome arrangement of stocks Xabc 2abmnz7 3b 4 5

11 +2e8 +++ 26 +2e8 +++ 31 +2e8 +++ 45 +2e8 +++ 54 +2x7/2x7y8 +++ 61 +2x7/2x7y8/2x7z8 +++ 71 +2x7/2x7y8 +++ 85 +2x7/2x7y8/2x7z8 +++/5e 95 +2x7/2x7y8/2x7z8 +++/5e 10 21 + 2x7/2x7y9/2x7z8 +++ 11 14 + 2x7/2x7y9/2x7z8 +++ 12 14 + 2x7/2x7y9/2x7z8 +++ 13 4 + 2x7/2x7y9/2x7z8 +++ 14 1 + 2x7/2x7y9 +++ 15 6 + 2x7y9x8w8 +++ 16 1 + 2x7y9x8w8 +++

Figure 2. Cytological map of chromosome 2 of D. serido.(a)2abmnz7, ancestral arrangement of the buzzatii cluster showing the breakpoints of inversions 2e8 and 2x7;(b)2abmnz7e8, arrangement found in the populations of central Brazil; (c) 2abmnz7x7, arrangement present in the populations of northeastern and southern Brazil, showing the breakpoints of the polymorphic inversions 2z8, 2y8 and 2y9;(d)2abmnz7 x7y9, arrangement present in the populations of southeastern Brazil, showing the breakpoints of inversions 2x8 and 2w8. The maps shown were constructed by cut-and-paste of the original D. repleta chromosomes drawn by Wharton (1942). 222

Figure 3. Micrographs of some inversion heterokaryotypes observed in the populations of D. serido.(a)2x7z8/x7, showing the single loop of the 2z8 inversion; (b) 2x7y8/x7, showing the single loop of inversion 2y8;(c)2x7y9/x7, showing the single loop of inversion 2y9;(d)2x7z8/x7y9, showing the double loop formed by the two overlapping inversions 2z8 and 2y9. 223

The break points of the five second chromosome inversions are shown in Figure 2c and d. In contrast to Tosi and Sene (1989), inversions 2z8 and 2y8 arose independently on a 2x7 chromosome and their break points are D3d-G1g and E1g-F6a, respectively (Fig- ure 2c). The two inversions overlap each other and also overlap inversion 2x7. Thus, there is no recom- bination between the three inversions and only three gene arrangements, 2x7, 2x7z8 and 2x7y8 are possible. Figure 3a and b shows the single inversion loops observed in the heterokaryotypes 2x7z8/x7 and 2x7y8/x7. The double loop corresponding to the 2x7y8/x7z8 heterokaryotype can be seen in Figure 3 of Tosi and Sene (1989). Inversions 2z8 and 2y8 were present in the populations from the inland valleys of the Parana river watershed (5–9). Inversion 2y9 also arose on a 2x7 chromosome and has break points C4f-E1d (Figure 2c). It overlaps and is mutually exclusive with inversions 2z8 and 2y8. The single inversion loop shown by heterokaryotype 2x7y9/x7 and the double inversion loop exhibited by heterokaryotype 2x7y9/x7z8 are depicted in Figure 3c and d. These are two of the six karyotypes that were observed in the populations of the Atlantic Litoral (10–14) where the three arrangements 2x7, 2x7y9 and 2x7z8 coexist (Table 3). The remaining two inversions, 2x8 and 2w8,aroseon a2x7y9 chromosome; 2w8, whose break points are E4g-D3d, is included within 2x8 with break points B2a-C6e (Figure 2d). These two inversions were ob- Figure 4. Updated chromosomal phylogeny of the buzzatii cluster. served only in homozygous state in the populations of Fixed inversions are shown in boxes, polymorphic inversions in el- Florianópolis (15) and Guaratuba (16). lipses. Data from Ruiz and Wasserman (1993), Kuhn et al. (1996) and this work. Salivary gland chromosomes of the interspecific hybrids × While no chromosomal variation was found D. serido (H24 Furnas, MG) D. seriema between or within the former four populations, the lat- All chromosomes were homosequential and showed ter, where inversion 2x7 is fixed, exhibited polymorph- very good pairing except in the proximal ends. This ism as well as a striking interpopulation differentiation is confirmation that the inversion fixed in the cent- (Table 3). Five inversions on chromosome 2, 2z8, 2y8, ral Brazil populations of D. serido is indeed 2e8,the 2y9, 2x8 and 2w8, plus one on chromosome 5, 5e, same inversion fixed in D. seriema and D. borborema are segregating in these populations. Inversions 2z8, (Wasserman & Richardson, 1987; Tosi & Sene, 1989; 2y8 and 5e, are seemingly identical to inversions ‘2d’, Kuhn et al., 1996). ‘2e’ and ‘5a’ observed by Tosi and Sene (1989) in two populations (Puerto Tirol and Resistencia) from north- D. koepferae × D. serido (H34 Sertãozinho, SP) eastern Argentina. We have changed their designation Chromosomes 3, 4 and 5 were homosequential and to accommodate it to the inversion nomenclature of showed good general pairing except in the proximal general use in the repleta group (Wasserman, 1992). ends. Chromosome X seems also homosequential but The other three inversions, 2y9, 2x8 and 2w8, have not showed asynapsis in the proximal one fifth of this been found previously and are described here for the chromosome. Chromosome 2 has a small distal loop first time. (corresponding to inversion 2l9 present in D. koep- 224 ferae) and a complex multiple inversion loop in the (8–9) and the two localities in the western Chaco proximal region (corresponding to heterokaryotype of Argentina previously analyzed (Baimai, Sene & 2j9m9/2x7y8). This confirms that the inversion fixed in Pereira, 1983; Tosi & Sene, 1989) show beyond doubt the D. serido populations from southern Brazil is 2x7 that they belong to the same population group. We and not 2j9, the inversion fixed in D. koepferae (Ruiz have shown that these populations are fixed for in- & Wasserman, 1993). version 2x7 and not for inversion 2j9 as reported by Tosi and Sene (1989) which seemingly misinterpreted it. Therefore, these populations do not share any in- Discussion version with D. koepferae and do not belong to this species, as suggested tentatively by Ruiz and Wasser- The chromosomal constitutions of D. serido in south- man (1993). Likewise, their location in the phylogen- eastern and southern Brazil show a clear cut split etic tree close to D. koepferae (Tidon-Sklorz & Sene, between two groups of populations. Four populations 1995a) should be revised accordingly. (1–4; see Figure 1) are fixed for inversion 2e8 and The populations of D. serido in southern Brazil three of them (2–4) show the karyotype denoted as contain abundant inversion polymorphism (Figure 4). Type I by Baimai, Sene and Pereira (1983). These are Five inversions on chromosome 2 and another one on characteristics of the Cerrado populations in central- chromosome 5 have been found segregating in these western Brazil (Figure 4). In addition, a morphometric populations. In addition, there are marked interpop- analysis of the male genitalia in these four populations ulation differences. Those populations of the inland (Monteiro & Sene, unpublished) has revealed a similar valleys of the Paraná watershed (5–9) are polymorphic shape of aedeagus as that of populations in central- for inversions 2y8, 2z8 and 5e. On the other hand, the western Brazil, the so called aedeagus type B (Silva five southernmost populations of the Atlantic Litoral & Sene, 1991). Thus both genetical markers and mor- (10–14) are polymorphic for inversions 2z8 and 2y9. phological traits fit perfectly and delineate a group of Finally, the two northernmost coastal populations (15– populations, referred to as D. serido type B, with a 16) are apparently fixed for 2y9, 2x8 and 2w8.This wide geographical distribution, from Pernambuco and chromosomal differences agree well with the biogeo- Bahia in northeastern Brazil to Mato Grosso do Sul graphy of the region. The populations of the inland in western Brazil and Paraguay, and to Minas Gerais valleys and those in the Atlantic Litoral are ecologic- and Sao Paulo in southeastern Brazil (Baimai, Sene ally quite different and remain separated by the Serra & Pereira, 1983; Tosi & Sene, 1989; Silva & Sene, do Mar, a mountain range that runs parallel to the At- 1991). Apparently, the four populations studied here lantic coast and ends near Porto Alegre, Rio Grande represent the southwestern limit of the range of D. do Sul (Figure 1). Neither there is a continuous habitat serido type B. Thus it is perhaps not surprising (Brus- for D. serido along the coast. Whereas the sand dunes sard, 1985) to find that these peripheral populations provide a more or less continuous cactus habitat in the are monomorphic whereas most other populations of southermost part of Rio Grande do Sul, this habitat this type are polymorphic for inversion ‘2a’ (Tosi & becomes fragmented into isolated patches northwards. Sene, 1989). Our data seem to be consistent with the morpho- The second group comprises the remaining 12 pop- metric analysis of the male genitalia (Silva & Sene, ulations (5–16; see Figure 1). They are homozygous 1991; Monteiro & Sene, unpublished). Populations for 2x7, an inversion fixed also in the Caatinga popu- 8 and 9 (Figure 1) show a distinct aedeagus shape, lations of northeastern Brazil, but show a karyotype designated as aedeagus type D by Silva and Sene (Type V) which is different from that of the Caat- (1991). Populations with this aedeagus type exhibit a inga populations (Type I; Baimai, Sene & Pereira, wide geographical range (Silva & Sene, 1991) which 1983). Furthermore, the inversion polymorphisms are include our localities 5–7 and the two localities in also quite different in these two regions (Figure 4). the western Chaco of Argentina, Puerto Tirol and Although linked by inversion 2x7, the historical rela- Resistencia, previously analyzed by Baimai, Sene tionships between the distant D. serido populations of and Pereira (1983) and Tosi and Sene (1989). On northeastern and southern Brazil are still unclear (see the other hand, populations 15 and 16 seem to have below). an aedeagus more similar to type A, which is that The similar karyotype and inversion polymorph- found in the northeastern Brazilian populations of ism observed in our two westernmost populations D. serido (Silva & Sene, 1991; Monteiro & Sene, 225 unpublished). Finally, populations 10–14 show an and Sinaloa. The chief host plant used by D. mojaven- intermediate distribution between aedeagus types A sis in each area is different, Ferocactus acanthodes, and D. Both morphometric and inversion data sug- Stenocereus gummosus and S. thurberi, respectively gest a genetic break between localities 14 and 15 (Ruiz, Heed & Wasserman, 1990; Etges et al., 1999). in the coast of the Santa Catarina State (Figure 1). A similar situation is found in D. buzzatii. Most pop- This interesting situation deserves further scrutiny as ulations in Argentina and worldwide feed and breed does the apparent morphological relationship between on Opuntia cladodes and are polymorphic for the 2j populations 15–16, and those farther north in the inversion. Populations in the Santa Maria Valley of Caatinga. Tucuman (Argentina) are unusual because they use Can the genetic differentiation among populations mostly the rotting stems of the giant columnar cactus and/or population groups be attributed to any identifi- Trichocereus tershekii (Hasson, Naveira & Fontdev- able ecological factors? The most evident ecological ila, 1992). These populations are also chromosomally variable for cactophilic Drosophila is the use of dif- peculiar: they have inversion 2j almost fixed and a ferent host plants. Female D. serido and other buzzatii high frequency of inversion 4s, present at very low cluster species oviposit on rotting cactus stems, usu- frequency in other Argentinian populations (Hasson ally fallen on the ground. The larvae feed on the de- et al., 1995). It appears thus that host plants may exert caying tissues and associated microorganisms (mostly an effect direct or indirect on the genetical constitution yeasts) until reaching the pupal stage. The adults feed of the resident Drosophila. on the juice exuding from cactus rots and mate also on Host plants may act as a selective factor in several the rots or on the vegetation nearby. Therefore, their ways. First, cactus species may differ in their chemical whole life-cycle is associated with the host plant and composition (e.g. lipids and alkaloids) and associated no natural populations can be sustained in the absence microbial communities (yeasts and bacteria) (Starmer of cacti. Most cactophilic Drosophila can use several et al., 1982; Fogleman & Heed, 1989). Accordingly, different cactus species, i.e. they are oligophagic, but they may vary in their ability to sustain larval growth show some degree of specificity (Pereira, Vilela & (Ruiz & Heed, 1988; Starmer & Aberdeen, 1990). Sene, 1983; Heed & Mangan, 1986; Ruiz & Heed, Second, rots of each cactus species have a character- 1988; Benado & Montero, 1988; Escalante & Benado, istic size and density (Heed & Mangan, 1986; Ruiz 1990; Hasson, Naveira & Fontdevila, 1992). & Heed, 1988; Etges et al., 1999). The availabil- Host plant differences are correlated with the in- ity of these ephemeral substrates determine a distinct version polymorphism variation observed in D. serido. selection regime on the resident Drosophila popula- The four populations of D. serido type B living on tions. Finally, the fauna, both arthropods and non- the top of isolated hills (1–4) and fixed for 2e8 utilize arthropods, associated with different cactus species Pilosocereus machrisii, which is a relatively small- may also vary. This means that a Drosophila species sized cactus with cylindrical stems 1–2 m long and may encounter diverse competitors and predators in 6–8 cm in diameter. On the other hand, the chief host each host plant (Benado & Montero, 1988; Escalante plants of the D. serido populations fixed for inversion & Benado, 1990). A plausible model for the effect 2x7 belong to the genus Cereus. C. hildmannianus of host plant on the inversion polymorphism may be (mandacarú-like) is a large columnar cactus with many proposed. Cactus with large (and usually sparse) rots branches 15–20 cm in diameter and up to 3 m tall. It is probably select for large adult body size and slow de- the primary host plant for D. serido populations 5–14, veloping larvae, whereas cactus with small (and more although Opuntia monacantha is also used occasion- abundant) rots will select small adult body size and ally. Finally, C. pernambucensis is the host plant used fast developing larvae (Heed & Mangan, 1986; Et- by populations 15–16 and very likely also those in the ges & Heed, 1987; Etges, 1993). To the extent that Caatinga (Bizzo & Sene, 1982; Pereira, Vilela & Sene, chromosomal inversions have an effect on body size 1983). and developmental time, as in D. serido close relative A correlation between inversion polymorphism D. buzzatii (Ruiz et al., 1991; Betrán, Santos & Ruiz, and host plant use has also been observed in other 1998), they will be selected by the host plant, and geo- cactophilic Drosophila. D. mojavensis, for instance, graphical variation in cactus composition will translate is divided into three chromosomally divergent races into among-population chromosomal variation. This which inhabit the Mojave desert in southern Califor- model might account for the peculiar chromosomal nia, the Peninsula of Baja California, Mainland Sonora constitution of the D. buzzatii in the Santa Marta 226

Valley since the favoured chromosome arrangement Etges, W.J. & W.B. Heed, 1987. Sensitivity to larval density in 2j increases body size which is adaptive to exploit a populations of Drosophila mojavensis: influences of host plant variation on components of fitness. Oecologia 71: 375–381. giant cactus such as T. tershekii. Likewise it might ex- Etges, W.J., 1993. Genetics of host–cactus response and life-history plain, at least in part, the inversion variation observed evolution among ancestral and derived populations of cactophilic in D. serido. Drosophila mojavensis. Evolution 47: 750–767. Etges, W.J., W.R. Johnson, G.A. Duncan, G. Huckins & W.B. Heed, 1999. Ecological genetics of cactophilic Drosophila:inversion polymorphism in Drosophila mojavensis and Drosophila Acknowledgements pachea, pp. 164–214 in Ecology of Sonoran Desert Plants and Plant Communities, edited by R.H. Robichaux. Univ. Arizona Press, Tucson. We thank S.G. Monteiro and F.C. Val for their help Fogleman, J.C. & W.B. Heed, 1989. Columnar cacti and desert during the field trips to collect Drosophila, R. Tidon- Drosophila: the chemistry of host plant specificity, pp. 1– Sklorz and D. Zappi for their aid in the identification 24 in Interactions Among Plants and Animals in the Western Deserts, edited by J.O. Schmidt. University of New Mexico of Drosophila and cactus species, respectively, and Press, Albuquerque. Marcos R. de Souza for preparing the map. We also Fontdevila, A., C. Pla, E. Hasson, M. Wasserman, A. Sanchez, H. thank Antonio Barbadilla, Bill Etges and an anonym- Naveira & A. Ruiz, 1988. Drosophila koepferae, a new member ous referee for helpful comments on a previous version of the Drosophila serido (Diptera, Drosophilidae) superspecies taxon. Ann. Entomol. Soc. America 81: 380–385. of the manuscript. Work supported by several grants Giddings, L.V., K.Y. Kaneshiro & W.W. Anderson, 1989. Genetics, from FAPESP, CNPq, FINEP and USP to F.M. Sene. Speciation and the Founder Principle. Oxford Univ. Press, NY. Partial funds were provided also by grant PB95-0607 Hasson, E., H. Naveira & A. Fontdevila, 1992. The breeding sites of from DGICYT (Spain) to Alfredo Ruiz. the Argentinian species of the Drosophila mulleri complex (sub- genus Drosophila, repleta group). Revista Chilena de Historia Natural 65: 319–326. Hasson, E., C. Rodriguez, J.J. Fanara, H. Naveira, O. Reig & A. Fontdevila, 1995. The evolutionary history of Drosophila buzzatii. XXVI. Macrogeographic patterns of inversion polymorphism References in New World populations. J. Evol. Biol. 8: 369–384. Heed, W.B. & R.L. Mangan, 1986. Community ecology of the Baimai, V., F.M. Sene & M.A.Q.R. Pereira, 1983. Heterochro- Sonoran Desert Drosophila, pp. 311–345 in The Genetics and matin and karyotypic differentiation of some neotropical cactus- Biology of Drosophila, Vol. 3e, edited by M. Ashburner, H.L. breeding species of the Drosophila repleta species group. Genet- Carson and J.N. Thompson, Jr. Academic Press, NY. ica 60: 81–92. Kuhn, G.C.S., A. Ruiz, M.A.R. Alves & F.M. Sene, 1996. The meta- Barker, J.S.F. & W.T. Starmer (eds), 1982. Ecological Genetics phase and polytene chromosomes of Drosophila seriema (repleta and Evolution. The Cactus–Yeast–Drosophila Model System, group; mulleri subgroup). Brazilian J. Genetics 19: 209–216. Academic Press, Sydney. Mayr, E., 1963. Animal Species and Evolution. Belknap, Cam- Barker, J.S.F., W.T. Starmer & R.J. MacIntyre, 1990. Ecological and bridge. Evolutionary Genetics of Drosophila. Plenum Press, NY. Moraes, E.M., 1997. Aspectos Ecológicos e Populacionais de Éspe- Benado, M. & C. Montero, 1988. Competition between the cacto- cies do Gênero Drosophila do Morro do Forno, Altinópolis, SP. philic species Drosophila starmeri and D. uniseta. Revista Monografía de Bacharelado. FFCLRP, USP. Chilena de Historia Natural 61: 187–190. Pereira, M.A.Q.R., C.R. Vilela & F.M. Sene, 1983. Notes on breed- Betrán, E., M. Santos & A. Ruiz, 1998. Antagonistic pleiotropic ing and feeding sites of some species of the repleta group of the effects of second chromosome inversions on body size and early genus Drosophila (Diptera, Drosophilidae). Ciência e Cultura life-history traits in Drosophila buzzatii. Evolution 52: 144–154. 35: 1313–1319. Bizzo, N.M.V. & F.M. Sene, 1982. Studies of the natural pop- Powell, J.R., 1997. Progress and Prospects in Evolutionary Biology: ulations of Drosophila from Peruíbe (SP), Brazil (Diptera, The Drosophila Model. Oxford Univ. Press, NY. Drosophilidae). Revista Brasileira de Biologia 42: 539–544. Ruiz, A., A. Fontdevila & M. Wasserman, 1982. The evolution- Brussard, P.F., 1984. Geographical patterns and environmental ary history of Drosophila buzzatii. III. Cytogenetic relationships gradients: the central-marginal model in Drosophila revisited. between two sibling species of the buzzatii cluster. Genetics 101: Ann. Rev. Ecol. and Systematics 16: 25–64. 503–518. Carson, H.L., 1987. Tracing ancestry with chromosomal sequences. Ruiz, A., W.B. Heed & M. Wasserman, 1990. Evolution of the Trends in Ecology and Evolution 2: 203–207. mojavensis cluster of cactophilic Drosophila with descriptions Carson, H.L., 1975. The genetics of speciation at the diploid level. of two new species. J. Heredity 81: 30–42. American Naturalist 109: 83–92. Ruiz, A. & W.B. Heed, 1988. Host-plant specificity in the cacto- Coyne, J.A. & H.A. Orr, 1997. ‘Patterns of Speciation in Droso- philic Drosophila mulleri species complex. J. Animal Ecol. 57: phila’ Revisited. Evolution 51: 295–303. 237–249. Dobzhansky, Th., 1970. Genetics of the Evolutionary Process. Ruiz, A., M. Santos, A, Barbadilla, J.E. Quezada-Díaz, E. Hasson & Columbia Univ. Press, NY. A. Fontdevila, 1991. Genetic variance for body size in a natural Escalante, A. & M. Benado, 1990. Predation on the cactophilic population of Drosophila buzzatii. Genetics 128: 739–750. fly Drosophila starmeri in the columnar cactus Pilosocereus Ruiz, A. & M. Wasserman, 1993. Evolutionary cytogenetics of the lanuginosus. Biotropica 22: 48–50. Drosophila buzzatii species complex. Heredity 70: 582–596. 227

Sene, F.M., M.A.Q.R. Pereira & C.R. Vilela, 1982. Evolution- Tidon-Sklorz, R. & F.M. Sene, 1995b. Drosophila seriema:a ary aspects of cactus breeding Drosophila in South America, new member of the Drosophila serido (Diptera, Drosophil- pp. 97–106 in Ecological Genetics and Evolution. The Cactus– idae) superspecies taxon. Ann. Entomol. Soc. America 88: Yeast–Drosophila Model System, edited by J.S.F. Barker and 139–142. W.T. Starmer. Academic Press, Sydney. Tosi, D. & F.M. Sene, 1989. Further studies on chromosomal Sene, F.M., M.A.Q.R. Pereira & C.R. Vilela, 1988. Contrasting pat- variability in the complex taxon Drosophila serido (Dip- terns of differentiation inferred from traditional genetic markers tera, Drosophilidae). Revista Brasileira de Genetica 12: 729– in the process of speciation. Paciﬁc Sci. 42: 81–88. 745. Silva, A.F.G. & F.M. Sene, 1991. Morphological geographic variab- Vilela, C.R. & F.M. Sene, 1977. Two new neotropical species of the ility in Drosophila serido (Diptera, Drosophilidae). Revta. Bras. repleta group of the genus Drosophila (Diptera, Drosophilidae). Ent. 35: 455–468. Papéis Avulsos de Zoologia 30: 295–299. Starmer, W.T. & V. Aberdeen, 1990. The nutritional importance of Wasserman, M., 1992. Cytological evolution of the Drosophila re- pure and mixed cultures of yeasts on the development of Dro- pleta species group, pp. 455–552 in Inversion Polymorphism in sophila mulleri larvae in Opuntia tissues and its relationship to Drosophila, edited by C.B. Krimbas and J.R. Powell. CRC Press, host plant shifts, pp. 145–160 in Ecological and Evolutionary Inc., Boca Raton, Florida. Genetics of Drosophila, edited by J.S.F. Barker, W.T. Starmer Wasserman, M. & R.H. Richardson, 1987. Evolution of Brazilian and R.J. MacIntyre. Plenum Press, NY. Drosophila mulleri complex species. J. Heredity 78: 282– Starmer, W.T., H.J. Phaff, M. Miranda, M. Miller & W.B. Heed, 286. 1982. The yeast ﬂora associated with the decaying stems of Wharton, L.P., 1942. Analysis of the repleta group of Drosophila. columnar cacti and Drosophila in North America. Evol. Biol. 14: Univ. Texas Publ. 4228: 23–52. 269–295. Zappi, D. & N.P. Taylor, 1993. Eastern Brazilian cacti. Piante Tidon-Sklorz, R. & F.M. Sene, 1995a. Evolution of the buzzatii Grasse 13 (4 suppl.): 65–78. cluster (Drosophila repleta species group) in the Northeastern Zouros, E., 1982. On the role of chromosomal inversions in South America. Evolución Biológica 8/9: 71–85. speciation. Evolution 36: 414–416.