Studies of the species barrier between Drosophila subobscura and D. madeirensis V: the importance of sex-linked inversion in preserving species identity

Authors


Mahnaz Khadem, Centro of Life Sciences, University of Madeira, Campus Penteada, Funchal, 9000, Portugal.
Tel.: 351 291 705265; fax: 351 291 705266; e-mail: khadem@uma.pt

Abstract

The X chromosome is known to exert a disproportionately large effect on characters related to post-zygotic reproductive isolation. There is also growing evidence about the important role of the chromosomal regions with reduced recombination (such as inversions) in maintaining the identity of closely related species. Using molecular markers, we examine the effect of different regions of the X chromosome on determination of hybrid traits (viability, testes size, sperm motility and morphological anomalies) in hybrid males between Drosophila madeirensis and Drosophila subobscura. The preponderant effect of a region localized inside the A2 inversion in the X chromosome in all hybrid traits is identified. Other marked regions exert a weaker influence or only influence some of the hybrid trait. Our results confirm the crucial role of sex-linked chromosomal inversion in preserving the identity of species with incomplete reproductive isolation. The specific genomic make-up of parental lines used to perform crosses has a great effect on hybrid fitness.

Introduction

Mechanisms that limit gene flow between populations and reduce hybrid fitness are essential to the formation of new species. As originally proposed by the Dobzhansky–Muller model, incompatible epistatic interactions between different genes of the two species are responsible for the reduction in hybrid fitness (Dobzhansky, 1937; Muller, 1940). The subsequent studies on speciation have established two rules known as Haldane’s rule and Large X effect. Haldane’s rule proposes a reduction in the hybrid’s fitness in the heterogametic compared to the homogametic sex (Haldane, 1922), and it is considered to be generally obeyed, regardless of sexing systems, although the genetic causes of this rule are still under debate (see for review, Marsely & Presgraves, 2007). The Large X effect suggests a strong influence of the X chromosome on hybrid sterility and inviability (Charlesworth et al., 1987; Coyne & Orr, 1989; Coyne, 1992). In the speciation process, the special and determinant role of the X chromosome is explicit in these two rules.

Some studies have revealed genes and genomic regions responsible for conferring sterility and inviability in hybrids and also that factors related to these intrinsic mechanisms are disproportionally accumulated on the X chromosome (Noor, 1995; Orr & Irving, 2001). Hybrid sterility seems to evolve faster than hybrid inviability and moreover the concentration of these factors is higher in the regions of low recombination (such as centromeric and inverted regions). Inverted regions play an important role in maintaining species identity, especially in sympatric species, where gene flow may occur in colinear regions but not in inverted regions. Factors involved in hybrid sterility and inviability are mainly accumulated in the inverted chromosomal regions between the two sympatric species Drosophila pseudoobscura and Drosophila persimilis, where extensive gene flow has been detected in the colinear regions (Noor et al., 2007). In this species pair, genomic sequence data have identified regions of introgression outside the inversions on the autosomes but not on the X chromosome (Kulathinal et al., 2009), probably because of the higher density of factors related to reproductive isolation on the X chromosome than on the autosomes (Marsely & Presgraves, 2007). Between the two allopatric species of Drosophila pseudoobscura bogotana and D. persimilis, the factors conferring reproductive isolations to hybrids are found both in inverted and colinear regions (Noor et al., 2001; Machado & Hey, 2003; Brown et al., 2004). As a general trend, good association between genomic regions without gene flow and factors linked with hybrid sterility and inviability have been found (Ting et al., 2000; Machado et al., 2002). Therefore, genomes of two incipient species can be considered as a mosaic of regions with nil or very limited, gene flow and introgressed regions that can be identified by the number of fixed and shared polymorphisms between them (Wang et al., 1997).

In the present work, we employ five molecular markers located on the X chromosome with known cytological positions to examine the effect of different X-linked regions in causing abnormal hybrid traits (viability, small testes size, sperm immobility, abnormal head shape and presence of extra sex combs) in crosses between the two closely related species of Drosophila subobscura and Drosophila madeirensis. The genetic barrier between the two species is incomplete, and they can be crossed under laboratory conditions in both directions, producing mainly fertile females and sterile males (Khadem & Krimbas, 1993).

Drosophila madeirensis is an endemic species to Madeira Island that only inhabits the laurel forest, whereas D. subobscura is a Palaearctic species also found in three of the Macaronesian Archipelagos (Canary, Madeira and Azores) and has recently invaded North and South America (for review see Krimbas, 1993). In spite of different population sizes, the two species have similar levels of nucleotide variability (Khadem et al., 2001; Nóbrega, 2007). At the chromosomal level, D. subobscura is highly polymorphic for different arrangements whereas D. madeirensis seems to be monomorphic.

The insular population of D. subobscura in Madeira Island also has a high level of genetic variability at molecular level although its gene arrangements are reduced compared to continental populations. The insular population of D. subobscura in Madeira Island also has a high level of genetic variability at molecular level although its gene arrangements are reduced compared to continental populations. Regarding the X chromosome, they segregate for two inversions, A2 and AST. In 1972, their frequencies was estimated as 90% and 10% respectively (Provosti, 1972). In 2002, the frequency of A2 was estimated as 92% (C. Nóbrega, personal communication). In Europe and north of the Africa, AST inversion shows a north–south cline; its frequency decreases from north to south whereas A2 inversion shows a south–north cline. This inversion is fixed in Canary Islands and is highly frequent (more than 80% in north of Africa). In southern Europe, the frequency of A2 inversion varies from 50% to 70% (for review see Krimbas, 1993). Studies of the long-term changes in the chromosomal inversion polymorphism of D. subobscura from South-western Europe showed that the frequencies of inversions typical of southern latitudes (nonstandard gene arrangements) increased whereas the ones typical of northern latitudes (standard arrangements) decreased during the last three decades (Soléet al., 2002). The X chromosome of D. madeirensis, compared to D. subobscura X chromosome with standard arrangement, has two specific arrangements, Am1 and inversion A16BCD.

The estimated divergence time between D. subobscura and D. madeirensis is 0.6–1.0 My (Ramos-Onsins et al., 1998), and between the two inversions AST and A2 it is about 133 000 years, the former being ancestral (Nóbrega et al., 2008). It has been hypothesized that Madeira Island was colonized twice by ancestral populations of D. subobscura. The first wave of colonization resulted in speciation of D. madeirensis and the second one originated the extant population of D. subobscura in the island (Khadem et al., 2001). Therefore, assuming that the A2 arrangement originated in mainland populations, the second colonization event took place during the last 130 000 years.

The results of the present work show that (i) all hybrid traits are strongly influenced by factor(s) near rpII/sxl region of the X chromosome, (ii) Drosophila madeirensis origin in this region is strongly associated with the appearance of all abnormal traits in hybrid males; (iii) the localization of rpII/sxl region within A2 inversion indicates its possible role in speciation process between D. subobscura and D. madeirensis, (iv) other genomic regions of the X chromosome weakly influence traits directly related to the reduction in fitness and (v) the strains used to perform crosses have a great influence in the expression of the hybrid traits.

Materials and methods

Strains and crosses

A single inbred strain of D. madeirensis (mad 12), originally collected in Ribeiro Frio, Madeira Island in 2001, and the isofemale line of D. subobscura (ch cu strain) were used. The D. subobscura X (=A) chromosome has a standard gene arrangement. Comparing to this species, D. madeirensis has two fixed inversions: Am1, a large inversion near the centromeric end and A16BCD, a small inversion at the tip (Krimbas & Loukas, 1984; Papaceit & Prevosti, 1989).

Drosophila madeirensis virgin females were crossed to D. subobscura males. F1 females were backcrossed to males of D. subobscura and B1 males were used to study hybrid traits.

Hybrid traits

Hybrid viability was measured by the number of adult flies. Hybrid male sterility/fertility was accessed by (i) testes size; (ii) sperm motility. Testes size was categorized in two classes (small and normal). Small testes size indicates that the male possessed atrophied, empty testes having dark red colour. These males are invariably sterile. Normal testes can vary slightly in size but the general aspect is normal (compared to parental species). Normal testes also present light orange colour, typical of parental species. The size of testes was determined by removing them from 7-day-old B1 males in Ringer’s solution. B1 males with normal testes size can be sterile or fertile.

The same preparations were used to determine sperm motility. It was diagnosed by the presence of motile sperms, established by microscopic examination of slightly crushed testes. Although the percentage of motile sperm varied among males, they were considered fertile when having at least one motile sperm. Males without any sperm or with few sperm or lots of sperm but without motility were classified as sterile (Zeng & Singh, 1993). In the present work, the term ‘fertile’ is employed as synonymous of sperm motility.

The hybrids can present two morphological anomalies (head shape and extra sex combs).

Head anomaly can be described as the deformation of the head part, including swollen antennae, eyes, mouth part and bending aristae. Males may be severely or moderately affected by these anomalies (Khadem & Krimbas, 1991b). Extra sex combs (esc) is the presence of the sex combs, with fewer teeth, in the second and some times third pair of legs in addition to the sex combs of the first pairs of legs (Khadem & Krimbas, 1991b; Papaceit et al., 1991).

Studied regions

The cytological positions of the five gene markers are given in Fig. 1. These markers are achaete gene (ac) on segment 2B (A. Munté, personal communication); ribosomal protein II (rpII) on segment 10A; sex lethal (sxl) on segment 10B; 6- phosphogluconate dehydrogenase (6pgd) on segment 13A (Segarra & Aguadé, 1992) and swallow (sw) on segment 16C (C. Segarra, personal communication). In X chromosome with standard gene arrangement, the distances between ac- - -rpII/sxl- - - pgd- - -sw are 65.21, 32.61, 32.21 cM, respectively. rpII and sxl are closely linked and are located within the inverted region of A2.

Figure 1.

 The positions of the genes studied on the A (X) chromosome of Drosophila subobscura, schemes of D. subobscura and Drosophila madeirensis with their respective inversions (photo map was kindly provided by Prof. Krimbas).

Although A2 is the most frequent gene arrangement in Madeira Island, it does not directly influence the present data as the ch cu strain used in this study has standard inversion.

In D. madeirensis, the ac gene region is near the break point of the Am1 inversion, considering its position on standard X chromosome (2B). The distance between this gene and rpII/sxl region is calculated approximately as 11.04 cM. The sw gene region is positioned in the middle of inversion 16BCD; therefore, this inversion does not influence the distance between this region and other gene regions used and the distances are approximately as those in D. subobscura.

DNA extraction, PCR amplification and analysis

Genomic DNA was extracted from B1 males previously scored, according to protocol no 48 in Ashburner (1989). Amplification primers and PCR conditions are given in Table 1.

Table 1.   PCR primers and amplification conditions.
MarkersPrimersPCR conditions
acForward 5′CTGAGCAAAGTCTCGAC3′30× (94 °C, 30 s; 48 °C, 30 s 68 °C, 40 s)
Reverse 5′CGTTGTGTGGTTGCGACTAGC3′
rpIIForward 5′CAGGTGAAGCGAGTACAG3′40× (95 °C, 30 s; 41 °C, 30 s; 63 °C, 45 s)
Reverse 5′CCGTTATTCCCCACAAAT3′
sxlForward 5′AACAACAAACCCTAAACA3′40× (92 °C, 45 s; 43 °C, 30 s; 65 °C, 50 s)
Reverse 5′GATTGAGGTTCGGGTTGA3′
pgdForward 5′CACTGGCGGTTCCAAAATAAG3′40× (92 °C, 45 s; 52 °C, 30 s; 73 °C, 30 s)
Reverse 5′CATCATCGGCCACCTATACAC3′
swForward 5′AAGTACATAGCGGACATC340× (93 °C, 30 s; 48 °C, 30 s; 63 °C, 30 s)
Reverse 5′GCAGATAGTCGCTGTAG3`

Fixed indels were used to assign each PCR product to the corresponding parental species. For rpII, sxl and pgd, published primers or sequences were employed (Martín-Campos, 1998; Llopart, 1999; Nóbrega, 2007; respectively) and for ac and sw, new primers were designed using available D. pseudoobscura sequences. The indels were detected after aligning each gene sequence, and appropriate primers were designed to amplify short fragments that include the indels.

For the ac gene, the primers amplified a fragment of about 500 bp. In this gene, there are two deletions in D. madeirensis compared to D. subobscura: one of three nucleotides (nt; GCA) and another of 6 nt (GCAGCC).

In the rpII gene region, primers amplified a fragment of about 250 nucleotides (in the first intron of the gene). D. subobscura has a deletion of 8 nt (TGCAAAAA) compared to D. madeirensis.

In the sxl gene, a fragment of 1000 bp was amplified; D. madeirensis has a deletion of about 40 nt compared to D. subobscura.

In the 6pgd genes, primers amplified a fragment of about 250 bp, with a deletion of 7 nt (CATCCGG) in D. subobscura.

Finally, in the sw gene, a fragment of about 330 nts was amplified. In the second exon of this gene, the D. subobscura strain used has three tandem repetitions of CCACAG motif whereas D. madeirensis has two repetitions. Therefore, the two species can be distinguished by a 6 nt indel.

Except for sxl amplification, the PCR products were separated in gel electrophoresis with T7C5 polyacrylamide and stained according to Budowel et al., 1991. Agarose gels were used for screening sxl amplifications. The origin of the genomic region (D. madeirensis or D. subobscura) in B1 males for each of the marker regions was determined comparing the relative mobility of the PCR product on the gels.

Results

Hybrid male viability

In backcross generation, the number of females was significantly higher than the number of males (489 females vs. 298 males, χ2 = 46.209, 1 d.f., P = 0.001), which confirms the earlier study (Khadem & Krimbas, 1993). Using five molecular markers, only 25 of all possible 32 combinations were recovered. The combinations that were not recovered at all, or appeared with a very low frequency (between 0.47% and 1.42%), were the results of crossover events between rpII and sxl that are rare because of the physical proximity of these two markers (approximately 7 cM); hence, their low frequency could not be only interpreted as an evidence of their inviability. Indeed, considering these two genes as a single genomic region, all 16 possible combinations were recovered. The frequency of each combination depends on the genetic distances between the markers. To avoid biased interpretation of data because of nonuniform distances between markers, only complementary combinations (classes) are compared (Table 2). The results show that under specific conditions (Y chromosome and the majority of the autosomes from D. subobscura origin), B1 hybrid males have the highest viability when carrying most of the D. subobscura X chromosome. In contrast, viability is severely reduced when the X chromosome is from D. madeirensis (class 1 in Table 2). Significant differences are also observed in complementary classes 2 and 4. In class 4, hybrid males carrying ac and rpII/sxl regions from D. subobscura and pgd and sw from D. madeirensis have higher viability compared to the males with the complementary genetic makeup. In class 2, males having D. subobscura ac, sxl/rpII and pgd origin and sw from D. madeirensis have higher viability than their complementary counterpart.

Table 2.   Backcross male viability is tested by comparing complementary classes. The chi-squares test the number of males in each two complementary classes and the influence of each marker (s = Drosophila subobscura; m = Drosophila madeirensis).
MarkersacrpII/sxlpgdswNoχ2
Classes
 assss3617.81***
mmmm8
 bsssm174.16*
mmms7
 cmsss293.13
smmm17
 dssmm167.12**
mmss4
 esmss71.31
msmm12
 fssms82.4
mmsm2
 gsmsm70.07
msms8
 hsmms81.64
mssm14
Markersχ2 (1 d.f.)
  1. No = sample size, *P < 0.05, **P < 0.01, ***P < 0.001.

ac5.12*
rpII/sxl32.00**,*
pgd5.12*
sw0.98

The results hint to the presence of three regions on the sex chromosome that influence the viability of the hybrid males. The strongest effect is found in the region near rpII and sxl markers. These markers are located in the second segment of the X chromosome, within the A2 inversion in D. subobscura, and are absent in D. madeirensis as well as in the ch cu strain of D. subobscura used in this study. A weaker effect is detected in regions marked by ac (segment I, within inversion Am1) and pgd (in segment II, outside inversion A2) of the X chromosome. It should be noted that the insignificant results can be caused by small sample size in some of the pair wise comparisons rather than lack of linkage between the gene regions and viability. Chi-square test performances for each marker individually also showed the strong effect of rpII/sxl region and a weaker influence of the ac, pgd regions (Table 2). In this latter test, the sample size is sufficiently big for reliable conclusions. However, it is possible that because of small sample size, some interactions between the loci in complementary classes remained undetected.

The two regions marked by the ac and pgd may either be containing real factors with weak effects or are reflecting the influence of strong factors of regions located in their vicinity. Indeed, the existence of a factor within Am1 inversion strongly influencing the viability of hybrid males has already been reported (Khadem & Krimbas, 1997).

The genetics of hybrid male sterility

Testis size

In B1 generation, 47% (139/293) of the males had small testes size and were invariably sterile because small testes do not contain spermatozoids. The percentage of the males with small testes size is much higher compared to the earlier study, where 29% (40 of the 137) of B1 males presented this anomaly (Khadem & Krimbas, 1991a). This difference could be attributed to different Drosophila strains used in the two studies.

Table 3 shows that more than 50% of the hybrid males have normal testes size when they carry rpII/sxl region from D. subobscura (genotypes 1–6) and < 18% of the hybrids have normal testes size when this region is from D. madeirensis (genotypes 7–11). The normal size is not fully restored in the males having all the regions from D. subobscura nor is it completely unrepresented in males with all regions from D. madeirensis. Other regions on the X chromosome and/or autosomes that influence this trait are yet to be detected.

Table 3.   Percentage of the B1 males with normal testes size and their respective genotypes. Total number is given in the last column (s = Drosophila subobscura, m = Drosophila madeirensis).
MarkersacrpIIsxlpgdsw% Normal testesTotal number
Genotypes
 1sssss7536
 2mssss6529
 3ssssm6816
 4sssms758
 5sssmm5016
 6mssmm5814
 7smmss148
 8smmsm148
 9smmmm1217
 10mmmms147
 11mmmmm128

Hybrid males carrying all the markers from D. subobscura background have significantly more normal testis than those with D. madeirensis background (Table 3, comparing genotypes 1 and 11, χ2 = 11.049, 1 d.f., P = 0.0009).

The influence of D. madeirensis origin on an otherwise D. subobscura X chromosome is assessed for ac (comparing genotypes 1 and 2), for sw (comparing genotypes 1 and 3), pgd (comparing genotypes 1–4) and for rpII/sxl (comparing genotypes 1 and 7). A significant effect is only observed in the last case (χ2 = 9.510, 1 d.f., P = 0.0020).

The effect of ac and sw regions of D. subobscura origin, on an otherwise D. madeirensis X chromosome, do not have any influence in determination of testes size, whereas rpII/sxl from D. subobscura (comparing genotypes 6 and 11) increase the number of males with normal testis size, but not significantly (χ2 = 3.16, 1 d.f., P > 0.05).

According to the present results, only D. madeirensis factor in rpII/sxl region strongly reduces the hybrid testes size.

Male fertility/sperm mobility

Considering sperm motility, only 26 of 215 (12%) B1 males were fertile. In the earlier study, it was reported that 36 of 137 (26%) hybrid males were fertile (Khadem & Krimbas 1991a). This discrepancy is attributed to the different Drosophila strains used in each study.

Because of the low number of fertile males, the influence of each marker is assessed separately. When fertility is analysed as a binary character, data are highly significant for the rpII/sxl region (Table 4). Hybrid males are invariably sterile when carrying D. madeirensis background in this region whereas fertility is partially restored (18%) in males with D. subobscura origin. At a lower level, the region marked by sw shows a significant effect.

Table 4.   Number of fertile and sterile B1 males depending on the origin of parental species (mad = Drosophila madeirensis; sub = Drosophila subobscura).
MarkersSpecies originFertileSterileChi-square test
  1. Significance levels are determined by chi-square test.

  2. *P < 0.05 and **P < 0.001.

acmad12770.73 ns
sub14116
rpIImad07013.091**
sub26119
sxlmad07014.439**
sub26117
pgdmad8901.134 ns
sub1698
swmad6945.518*
sub1887

Considering the effect of the regions marked by ac, pgd and sw, on males having rpII/sxl of D. subobscura origin, Table 5 shows that depending on genotype combinations, male fertility varies between 6% and 43%. In general, it appears that D. madeirensis origin in sw region decreases male fertility, although not significantly.

Table 5.   Percentage of fertile hybrid males for each genotype (s = Drosophila subobscura, m = Drosophila madeirensis).
MarkersacrpII/sxlpgdsw% FertileTotal number
Genotypes
 1ssss2236
 2sssm616
 3ssms437
 4msss1828
 5ssmm1019
 6mssm1714
 7msms336
 8msmm714

Only the significant influence of rpII/sxl region in determining hybrid fertility is confirmed. Data from a larger number of fertile B1 males are needed to draw a decisive conclusion.

Morphological anomalies

Abnormal head shape

About 32% (96 of 298) of the backcross male hybrids had anomalies; males may be severely or moderately affected by these anomalies. The percentages of hybrids with head anomalies in this study is much higher than the one reported (17.8%) in an earlier study (Khadem & Krimbas, 1991b). The genetic basis of abnormal head shape was analysed, and the results are shown in Table 6. The percentages of males with this anomaly vary from 17% to 36%, when the markers are all or predominantly from D. subobscura origin or at least have the rpII/sxl region from this species (genotypes 1–6). A higher proportion of hybrid males (71–86%) have abnormal head shapes when carry predominantly regions from D. madeirensis origin, or at least have rpII/sxl region from this species (genotypes 7–11). Drosophila subobscura origin in this region in otherwise D. madeirensis background significantly reduces the percentage of males with this anomaly (comparing genotypes 6 and 11; χ2 = 6.806, 1 d.f., P = 0.009***) whereas rpII/sxl region from D. madeirensis origin in D. subobscura background significantly increases the head anomaly in hybrid males (comparing genotypes 1–7; χ2 = 5.847, 1 d.f., P = 0.0156**).

Table 6.   Percentage of B1 males with head anomalies and extra sex combs and their genotypes (s = Drosophila subobscura, m = Drosophila madeirensis).
MarkersacrpIIsxlpgdswabnHescTotal
  1. abnH, Abnormal head shape; esc, extra sex combs.

Genotypes
 1sssss362.836
 2mssss31329
 3ssssm25016
 4sssms2508
 5sssmm25016
 6mssmm17012
 7smmss86717
 8smmsm86867
 9smmmm826417
 10mmmms71437
 11mmmmm75508

Extra sex combs

In this study, 47 of 297 (15.8%) B1 males examined have extra sex combs. This value is slightly higher than the one (13.5%) reported earlier (Khadem & Krimbas, 1991b; Papaceit et al., 1991). The expression of this character also appears to be dependent on the strains of the species employed.

The genetic basis of extra sex comb anomaly was analysed, and the results are shown in Table 6. The percentages of males with extra sex combs vary from zero to three when the markers are all or predominantly from D. subobscura origin or at least have the rpII/sxl region from this species (genotypes 1–6). When the majority of the markers are from D. madeirensis origin, 43–86% of B1 males have extra sex combs genotypes (7–11). The only exception is detected when all regions are from this species and only pgd is from D. subobscura origin (not shown in Table 6), none of the hybrid males having this character, most probably because of the very low number of individuals (2) recovered from this genotype. A strong influence of the region marked by sxl/rpII is evident, comparing genotypes 1–7 (χ2 = 23.004, 1 d.f., P = 0.000***); 6–11 (χ2 = 6.806, 1 d.f., P = 0.009**) and also 6–7 (χ2 = 8.647 1 d.f., P = 0.0032**).

The origin of D. madeirensis in sxl/rpII region in otherwise D. subobscura background significantly increases, whereas D. subobscura origin in predominantly D. madeirensis background significantly decreases the percentage of the B1 males with extra sex combs.

Discussion

Hybrid traits

This study employs five molecular markers to further decipher the effect of different regions of the X chromosome on five hybrid abnormal traits. The unprecedented and main achievement of this study is detecting the strong role of the genomic regions encompassing A2 inversion on all hybrid traits. However, this work is quite distinct from the earlier report on the same subject (Khadem & Krimbas, 1991a, 1993, 1997). Firstly, the position of the markers is accurately determined from in situ hybridization data enabling calculation of the distances between them, although in the earlier studies they were grossly calculated from interspecific hybrid data. Secondly, the second segment of the X chromosome has been marked by four markers: rpII and sxl located within A2 inversion, pgd outside this inversion and sw within inversion A16BCD. In the earlier study, this segment was marked only by one marker (beadex) with unknown cytological localization. From interspecific hybrid data, the distances between beadex and diaphorase (dia), and pgd and sw are estimated as 38 and 36.7 cM, respectively. As dia and sw are occupying similar positions within the inverted region of A16BCD, it is assumed that beadex should be located near to pgd but outside the inversion A2.

The first segment of the X chromosome is marked by one marker (ac) near the breakpoint of Am1 inversions although two visible markers (antennae colour and vermillion eye colour, ve), both within inversion Am1, have been used earlier.

Three of these traits, inviability, small testes size, sperm immobility, are directly related to reduction in hybrid fitness; the other two (abnormal head shape and extra sex combs) are morphological anomalies. The former trait, depending on the extent of head malformation, can have a moderate or strong effect on hybrid fitness by reducing severely the longevity of the carriers. The latter trait (extra sex combs) does not seem to have any influence on hybrid fitness. However, these two anomalies appear to be linked (Khadem & Krimbas, 1991b).

Under the specific genetic make-up (Y chromosome and majority of autosomes from D. subobscura origin), the region marked by rpII/sxl (second segment of X chromosome) predominantly influence all hybrid traits. Backcross hybrid males carrying this region from D. madeirensis origin are significantly less viable, less fertile, have small testes size and demonstrate more morphological anomalies comparing to their siblings with D. subobscura origin of the same region (Fig. 2). Other regions of the X chromosome might have either a weak effect (ac and pgd in viability, sw in fertility) or no effect (testes size, hybrid anomalies). A preponderant effect of the first segment of the X chromosome in all hybrid traits, marked by ve and antennae colour, has been reported (Khadem & Krimbas, 1993). In the case of hybrid inviability, the influence of the same chromosomal region with different proximity to the viability factor is probably being detected. If so, then the factor located within inversion Am1 is closer to the markers used in the previous study (antennae colour and ve) than the ac used in this study. However, this same argument does not hold for other traits (small testes size, fertility and anomalies) that ac does not seem to have any influence. It is possible that there are two different factors in the first segment: one excreting strong and the other weak effect on viability, and only one factor (close to antennae colour and ve) in the case of other hybrid traits. But considering that these markers are located within inversion Am1, it is expected that they behave similarly, unless crossing-over can occur freely between antennae colour, ve and ac regions. Alternatively, these former markers (antennae colour, ve) were not localized correctly in the earlier study. It should be mentioned that no recombination was observed between antennae colour and ve, either because of proximity of these two loci or recombination suppression within inversion. This discrepancy can be resolved by using new markers in the first segment of the X chromosome.

Figure 2.

 The strong influence of the genomic region marked by rpII/sxl, shaded in grey, on hybrid traits (small testes size, head malformation and extra sex combs). The chromosomal inversions associated with the markers are also indicated.

In the second segment, the three regions marked by rpII/sxl, pgd and sw are located at convenient distances (32.7; 36.7 cM, respectively) allowing them to behave independently. There is a good agreement on the effect of sw and dia (located at A16BCD arrangement) on hybrid traits (influencing weakly fertility) between the present and published work. Factors influencing other hybrid traits are closely linked to sxl/rpII region, moderately linked to beadex region (Khadem & Krimbas, 1997) and only in the case of fertility are weakly linked to pgd region.

Furthermore, we should take into consideration the influence of autosomal factors and their possible interactions with X-linked factors on determination of reported hybrid traits. The presence of factors in major autosomes affecting the fertility of hybrid males between D. subobscura and D. madeirensis, but not interactions between different chromosomal regions, except one between E and O, was reported (Khadem & Krimbas, 1991a). Regarding head anomalies, an interaction between the X chromosome (exerting major effect) with the E and O chromosomes was detected. The extra sex combs anomaly was considered to be determined by at least four autosomal loci, one in each autosome (Papaceit et al., 1991), and other authors detected the strong influence of the X chromosome interacting only with the E chromosome (Khadem & Krimbas, 1991b). The former study did not directly detect the effect of the X chromosome because of the location of the single marker (diaphorase) used. This marker is localized near the tip of the X chromosome in or near inversion 16BCD, and the region that the present report shows not to exert any effect on determination of hybrid anomalies. A clearer picture of the genetics of hybrid traits in the two species employed in this study can only be achieved by employing many markers along the sex chromosomes as well as autosomes.

The role of the inversions

Chromosomal inversions play an important role in maintaining species identity by significantly reducing recombination in interspecific hybrids. Even in intraspecific crosses, gene flow between different chromosomal arrangements can be highly restricted, depending on the size and the particular location of the genes within the inverted regions (Muntéet al., 2005; Nóbrega, 2007).

In D. pseudoobscura subgroup, it has been demonstrated that the factors associated with reproductive isolation are localized in the inverted chromosomal regions in sympatric species, whereas in allopatric species these factors are localized both in inverted and noninverted regions. Extensive gene flow has only occurred in noninverted regions between the two sympatric species (D. pseudoobscura and D. persimilis) but not between the two allopatric species (D. ps. bogotana and D. persimilis) (Machado & Hey, 2003 and Chang & Noor, 2007).

Any possible interspecific gene flow between D. subobscura and D. madeirensis could have only occurred in Madeira Island, the only place where the two species coexist. In Madeira Island, A2 is the most frequent arrangement in segment II of the A chromosome of D. subobscura, and AST is also present but at a lower frequency (90% and 10%, respectively, Provosti, 1972; C. Nóbrega, personal communications). The present data strongly suggest the preponderant role of the region marked by rpII/sxl genes in all hybrid traits studied (viability, testes size, fertility and morphological anomalies). These two genes are closely linked and are located in segment II of the X chromosome within inversion A2. AST is considered to be the ancestral arrangement, showing a higher level of variation compared to A2. The estimated age of A2 inversion is about 133 000 years, whereas the estimated age between D. madeirensis and D. subobscura is about 630 000 years (Nóbrega, 2007). Therefore, the extant populations of D. subobscura in Madeira Island are representing the second waves of colonizer that entered the island, more recently, after the speciation event of D. madeirensis.

The prevalence of A2 inversion in Madeira Island and the preponderant effect of rpII/sxl region in the determination of all hybrid traits points to the importance of this inversion in maintaining the species integrity. This conclusion is based on two gene markers; we assume that same magnitude influence is excreted along A2 inversion.

Although recombination is reduced within the inverted segment in heterokaryotypes (Sturtevant, 1917; Roberts, 1976), genetic exchanges might occur by double crossing-over or gene conversion. Nóbrega et al. (2008) studied five regions along the A2 inversion; no evidence of genetic exchange was found between AST and A2 inversions. As the size of A2 inversion (41.3 cM) seems to be large enough for double crossovers to occur, the action of selection against the recombinant chromosome was postulated by the authors. Regarding the role of this inversion in determinations of hybrid traits, more loci are needed to be studied to enable a final conclusion to be made.

Inversion A16BCD, near the tip of X chromosome of D. madeirensis, has some minor effect on hybrid fertility but seems to be free of factors regarding other hybrid traits. This inversion is a short one near the tip of the chromosome where recombination is expected to be reduced.

In segment I of the X chromosome, a strong role of the markers located within inversion Am1 has already been reported in all hybrid traits but it is difficult to ascertain the role of this inversion as a whole rather than the role of a specific gene region within inversion Am1. Our data show that ac region, located near the breakpoint of inversion Am1, has some weak effect on hybrid inviability, but not the preponderant effect on all traits detected using other markers (Khadem & Krimbas, 1997).

This discrepancy cannot be explained by the location of the markers within this inversion. Am1 is considered a large inversion; it encompasses almost all segment I of the D. madeirensis X chromosome (Fig. 1). Genetic exchange could have occurred between different loci within this inversion. Although no recombinants were recovered between these two visible markers (antennae colour and ve) employed in the earlier study (Khadem & Krimbas, 1997), a stronger effect of a locus near the break point of the inversion would be expected as the result of strong crossover suppression compared to the loci more centrally located within the inversion. However, the results indicate the opposite effect.

To shed light on this discrepancy and to ascertain the exact role of inversion Am1 in the speciation process between D. madeirensis and D. subobscura, more markers with known cytological positions in this segment of the X chromosome need be employed.

The role of the strain

In this study, the percentages of sterile and abnormal males were higher compared to earlier reports (Khadem & Krimbas, 1991a, 1993, 1997). However, hybrid males consistently showed lower viability than the females with the same order of magnitude (1 : 2 respectively) in all similar crosses (female F1 from D. madeirensis mother crossed with male D. subobscura). Strain-dependent viability was reported when female D. subobscura was crossed to male D. madeirensis (Khadem & Krimbas, 1993) and also in crosses between Anopheles gambiae and A. arabiensis (Slotman et al., 2004). Hybrid female sterility in crosses between D. yakuba and D. santomea appeared only when some strains of drosophila were used (Coyne et al., 2004), but no intraspecific polymorphism was detected for alleles associated with sterility among strains of D. pseudoobscura or D. ps. bogotana (Brown et al., 2004).

In the species pairs, except for sex ratio trait, other differences were considered to be strain dependent. Different alleles segregating in different strains are responsible for the manifestation of hybrid traits. Moreover, at least some of alleles should be simultaneously determining different hybrid traits as the higher percentage of sterile males is directly correlated with the higher percentages of morphological anomalies. Different results were obtained when different strains of D. madeirensis were crossed to the same strain of D. subobscura, and the existence of an analogous system like hybrid male rescue (hmr) in the species of the Drosophila melanogaster group (Hutter & Ashburner, 1987; Hutter et al., 1990) was suggested in D. madeirensis (Khadem & Krimbas, 1991a, 1993, 1997). The new data show that D. subobscura is also polymorphic for alleles related to hybrid traits. When strains of this species from Madeira Island were used to perform crosses, about 44% (58 of 133) of B1 males were fertile, 14% (20 of 140) had abnormal head and only 2% (three of 143) displayed extra sex combs (M. Khadem, personal data). These numbers are quite different from the ones reported here or in the earlier studies. Indeed, the existence of fertile hybrid F1 was reported by Rego et al., 2007 although all previous studies as well as present data clearly indicated absolute hybrid male sterility in crosses between these two species (Krimbas & Loukas, 1984; Papaceit & Prevosti, 1991; Khadem & Krimbas, 1991a, 1997). This difference can be explained by the methodology employed in different studies. Whereas Rego et al. mass crossed populations of the two species from Madeira Island, other studies used isofemale or inbred lines of continental D. subobscura, kept in the laboratory for many generations. Genetic variability is reduced through the process of inbreeding. Reduction in genetic variability can explain the detection of fertile hybrid males in one but not in other studies. However, we found significantly different results (regarding the percentages of fertile males and morphological anomalies, Khadem unpublished data) using isofemale strains of D. subobscura from Madeira Island compared to continental ones. These differences can be explained by the existence of genetic differences between continental and islanders populations of D. subobscura. No genetic differentiation was detected between populations of D. subobscura from Spain and Madeira (Khadem et al., 2001; M. Khadem, unpublished data), so these results can be specific to the loci studied that might not reflect the other parts of the genome. For example, two populations can differ from each other in loci that confer to local adaptations while being undifferentiated in other loci. The same phenomenon was observed between the races of D. melanogaster, where sexual isolation is known to be determined by many autosomal loci but molecular data failed to show much differentiation between the races (Hollocher et al., 1997; Hasson et al., 1998; Tsaur et al., 1998).

Alternatively we can hypothesize that some gene flow has occurred between sympatric populations of the two species, making them more compatible with each other. However, no gene flow has been detected between sympatric populations of the two species (Khadem et al., 1997, Nóbrega, 2007).

Acknowledgments

We are grateful to Costas Krimbas, Agusti Munté, Thomas Dellinger for valuable comments and suggestions on the manuscript; Montse Papceit, Carmen Segarra and Montse Aguadé for in situ hybridization and providing sequencing facilities. A special thanks to John Micklethwaite for revising the manuscript and Fábio Reis for graphic help. This work was supported by Isoplexis, Germobanco of the University of Madeira.

Ancillary