Introduction

The last decade witnessed an accelerated development of whole genome sequencing technologies and the concomitant development of the bioinformatics and analytical tools required for genome assembly and annotation as well as for whole genomes evolutionary comparisons both within species and across phylogenies. In many taxa, these comparative analyses unveiled an unprecedented level of structural variation, including duplications, transpositions and inversions1,2,3,4,5. In the pre-genomic era, this kind of variation had only been extensively studied in a few genera as its detection did not only require using laborious cytogenetic techniques but it also required that the size of the chromosomes in the species under study was large enough to detect the potential structural differences. In Diptera, the banding pattern of polytene chromosomes provides detailed information on each chromosome organization. Given that structural changes alter this organization, the nature and number of the structural changes that segregate within populations and of those that become fixed in different species can be cytologically inferred through banding pattern comparison. In both the Drosophila and Anopheles genera, the cytological detection of chromosomal inversions segregating in some species propelled their study across space and time (e.g., as summarized in Krimbas and Powell6 for Drosophila). Moreover, comparison of the detailed cytogenetic maps obtained for genera such as Drosophila and Anopheles allowed the detection of fixed inversions between related species and the subsequent generation of cytological phylogenies7.

Comparison of 12 genomes across the Drosophila phylogeny confirmed the cytological observation that paracentric inversions were major contributors to the genus chromosomal evolution2,3. It also confirmed, and extended, the cytological observation that some regions had been multiply disrupted by fixed inversions, and that the X chromosome is the fastest evolving chromosome in the genus3. Moreover, the molecular identification of inversion breakpoints through genome comparison at short time scales [i.e., either between closely related species such as Drosophila melanogaster, D. simulans and D. yakuba8, or within species9] allowed the detailed characterization of inversion breakpoints, which revealed that inversions could originate by the staggered breaks mechanism in addition to the cut-and-paste and ectopic recombination mechanisms.

Both the cytological and genome-based approaches to identify inversions and to finely localize their breakpoints through banding pattern and genome comparison, respectively, have limitations. The most obvious limitation of the cytological approach in species with polytene chromosomes is that imposed by the size of the structural change, as this approach precludes the identification of small inversions. The time elapsed since the divergence of the species under study would constitute a second limitation to identify fixed inversions as chromosomal changes accumulate through time. This accumulation implies that the banding pattern comparison might render the cytological identification of homologous fragments uncertain or even impossible when distantly related species are compared. This uncertainty or impossibility would preclude the identification of the multiple structural changes that are fixed at the long time scale. Even at a shorter time scale, this limitation might differentially affect those chromosomes with a higher rate of chromosomal evolution by paracentric inversions, as it is the case of the X chromosome in Drosophila3. Concerning the genome-based approach, the quality of the genomes to be compared would be the major limitation. Indeed, even if the number of species with sequenced genomes is steadily increasing, only a few of the newly generated genome assemblies are composed of super-scaffolds where a relatively reduced number of scaffolds that account for most of each chromosome length are ordered and oriented. These assemblies are especially suitable for the detailed molecular characterization of fixed inversions breakpoints through genome comparison. In contrast, the draft assemblies of the remaining species genomes would be inadequate to address questions on chromosomal evolution, as they would preclude the fine molecular identification of fixed inversions breakpoints.

The subobscura cluster of Drosophila is composed of three species: D. subobscura, D. madeirensis and D. guanche. The former species is widely distributed whereas the other two species are island endemics that originated in Madeira and the Canary Islands, respectively, upon their independent colonization by D. subobscura. The karyotype of these species consists of five rod-like acrocentric chromosomes and one dot-like chromosome named A (X), J, U, E, O and dot that correspond to Muller A, D, B, C, E and F elements, respectively. These species structural variation has been extensively studied at the cytological level (as summarized in10). Unlike the two island endemic species that are monomorphic at the chromosomal level11,12,13, D. subobscura exhibits a rich inversion polymorphism that presents adaptive latitudinal clines in both its original palearctic distribution area and in the newly colonized areas in the west coast of both American subcontinents14. The breakpoints of several of the D. subobscura polymorphic inversions have been molecularly identified and characterized through chromosome walking, which has revealed that the staggered breaks mechanism that generates duplications in the derived arrangement is the prevalent mechanism originating inversions in this species15,16,17,18,19,20, like it also is in D. melanogaster9,21.

Comparison of the cytological maps of D. subobscura and D. guanche had revealed that the D. guanche chromosomes differ from the standard arrangement of all chromosomes of D. subobscura by 13 inversions11,13,22 —six on the A chromosome and seven in the four large acrocentric chromosomes. Ten of these inversions became fixed since the D. subobscura-D. guanche split whereas the other three originated and still segregate in D. subobscura10. Although a high-quality assembly of the D. guanche genome has been recently obtained23, this is not yet the case for D. subobscura, which precludes the identification of inversions breakpoints in both species through genome comparison. In D. subobscura, over 500 sequence-based markers have been cytologically mapped, with a small subset also mapped in D. madeirensis and D. guanche23,24. Given the extended collinearity previously detected between D. subobscura and D. guanche in most autosomal regions and in parts of the A chromosome, we have combined the information provided by the newly generated D. guanche genome and by previously mapped markers with known sequence to molecularly identify the breakpoints of the ten inversions —four in autosomes J, E and O and six on the A chromosome (Fig. 1)— that became fixed since the D. subobscura-D. guanche split11,13. The identification and molecular characterization of these inversions breakpoints will not only allow us to contrast the available, and sometimes differing, cytological information on these inversions but it might provide information on their mechanism of origin. It will additionally allow us to identify in which lineage each inversion occurred and became fixed, and therefore to infer a molecularly based phylogeny of the paracentric inversions underlying the chromosomal evolution of the subobscura cluster.

Figure 1
figure 1

Drosophila guanche and D. subobscura chromosomes affected by fixed inversions. The localization of inversion breakpoints in D. subobscura chromosomes J, E, O and A (Muller elements in parentheses) through their flanking markers is given on a schematic representation of the Kunze-Mühl and Müller33 map (not at scale). Localization of each marker sequence in the D. guanche genome sequence is represented above on a horizontal line (in Mb units). Fragments delimited by differently colored lines include inversion breakpoints, with the D. guanche fragments spanning both breakpoints of a given inversion labeled with the same low-case letter (e.g., a, proximal breakpoint; a*, distal breakpoint). The O chromosome of D. guanche differs from the Ost arrangement of D. subobscura by two overlapping inversions, the Of inversion and the D. subobscura polymorphic inversion O3. Although both inversions are shown, only markers flanking the Of inversion —fixed since the two species divergence— are given. For the four overlapping inversions of the A chromosome (#), fragments labeled a, b, c and d correspond to the breakpoints of inversions Af1, Af2, Af3 and Af4, respectively. Markers delimiting inversion breakpoints are labeled 1 to 39.

Full size image

Results

Identification of fixed inversions breakpoints

The relevant cytological and molecular information used to identify the breakpoints of the four autosomal and six A chromosome inversions fixed since the divergence of D. subobscura and D. guanche is summarized in Fig. 1. These inversions have been named Jf, Ef1, Ef2, Of and Af1 to Af6, according to the chromosome affected and numbered in each chromosome according to their centromere-proximal position. Our strategy was based on (i) the cytological localization of 335 markers with known sequence that had been previously mapped on the J, E, O and A chromosomes of D. subobscura, as summarized in Puerma et al.23, and (ii) the localization of these markers in the recently assembled D. guanche genome sequence23. Comparison of these datasets revealed discontinuities in markers order —two for each of the independent inversions Jf, Ef1, Ef2, Of and Af5 and overlapping inversions Af1 to Af4, and only one for inversion Af6, as it is a terminal inversion (Fig. 1). These discontinuities allowed us to identify those markers flanking the breakpoints of each inversion (Fig. 1) and therefore to delimit a large region spanning each breakpoint in D. guanche, with the only exception of the distal breakpoint of inversion Af6.

For each inversion, the large region spanning a breakpoint in D. guanche was initially compared to draft2 of the D. subobscura genome (Barcelona Subobscura Initiative [BSI]) in order to more narrowly delimit the breakpoint region. For breakpoints of the four autosomal inversions, this comparison allowed the identification of a rather short D. guanche fragment that spanned the breakpoint. For the A chromosome inversions, this approach was generally unsuccessful, which led us to initiate chromosomal walks from both ends of the large regions that in D. guanche spanned the breakpoints. For both autosomal and A chromosome breakpoint regions, probes were designed and amplified on the D. guanche genome and they were subsequently in situ hybridized on polytene chromosomes of D. subobscura until one of them gave two signals approximately at the locations expected according to either Moltó et al.11 or Brehm and Krimbas13, indicating that the probe spanned the breakpoint. In the case of autosomal inversions, one or a few probes were required for breakpoint identification whereas a higher number of probes was generally needed to narrow down and eventually cross each A chromosome inversion breakpoint. The final probes spanning the breakpoints of the four autosomal inversions and the A chromosome inversions Af2 to Af5 were additionally in situ hybridized on polytene chromosomes of D. guanche where they gave one signal at or rather close to the expected band (Supplementary Figs S1 to S8; Table 1). The adequate combination of primers designed on D. guanche was used to amplify the fragments that in D. subobscura spanned each inversion breakpoint and only in a few cases, new primers had to be designed on draft2 of the D. subobscura genome (BSI). When these fragments were hybridized on polytene chromosomes of both species, they generally gave a single signal on D. subobscura chromosomes and two signals on D. guanche chromosomes at the same bands that fragments spanning the D. guanche breakpoints had (Supplementary Figs S1 to S8). The D. subobscura fragments spanning the breakpoints were subsequently sequenced, annotated and compared to the D. guanche genome, which allowed us to delimit and characterize each breakpoint (see below).

Table 1 Proposed cytological boundaries of inversions fixed between D. subobscura and D. guanche.
Full size table

The breakpoints of inversions Af1 and Af6 could not be identified but they were narrowed down to rather short regions. Indeed, breakpoints of inversion Af1 were narrowed down to ~44-kb (flanked by genes GA14783 and GA13678) and ~33-kb long (flanked by genes GA17070 and GA15499) regions of the assembled A chromosome of D. guanche, and the proximal breakpoint of inversion Af6 to an ~20-kb long region (flanked by genes GA22805 and GA24354). Our failure to design new probes on D. guanche in each of the three intervening regions was due either to the discontinuity between two neighboring scaffolds in the D. guanche assembled A chromosome (proximal Af1 breakpoint) or because of the abundance of transposable elements and other repetitive sequences (two other breakpoints).

For the A chromosome inversions, results of the in situ hybridizations on D. subobscura polytene chromosomes revealed some discrepancies not only concerning the number of breakpoints but also their cytological localization relative to those previously proposed11,13 (Table 1). This led us to propose a modified version of the cytological map of the D. guanche A chromosome (Table 1 and Fig. 2). The major difference between the here proposed map and previous proposals stems from the more limited resolution of the classic cytological approach. Indeed, as opposed to the eight breakpoints here identified for the four overlapping inversions in segment I of the A chromosome, those studies could only identify seven breakpoints, which implied reuse of one of the breakpoints (Table 1).

Figure 2
figure 2

Newly proposed cytological map of the D. guanche A chromosome. Colored boxes represent conserved blocks between D. guanche and D. subobscura based on the localization of breakpoints in the recently assembled D. guanche genome (Puerma et al.23). Sections delimiting each conserved block, given in their upper part, were inferred from the localization in the Kunze-Mühl and Müller33 map of D. subobscura of the fragments spanning the six fixed inversions breakpoints in D. guanche (Puerma et al.23). Differently colored lines stemming from a given breakpoint in the D. guanche map indicate each of its flanking regions and their localization in the Kunze-Mühl and Müller33 map of D. subobscura, as revealed by in situ hybridization.

Full size image

Characterization of fixed inversions breakpoints

Figures 3 and 4 show the functional annotation of the breakpoint regions of the four autosomal inversions and the A chromosome inversions Af2 to Af5 here studied, respectively. It is worth noting that none of the breakpoints disrupt genic regions even though according to the D. pseudoobscura annotation, the proximal D. guanche breakpoints of inversions Jf and Ef1 would disrupt the 5′ UTRs of genes GA19578 and GA15509, respectively.

Figure 3
figure 3

Functional annotation of autosomal inversion breakpoints. Schematic representation of the sequenced and annotated breakpoint regions corresponding to the four autosomal inversions fixed since the D. subobscura-D. guanche split. D. guanche, above; D. subobscura, below. Dashed lines between chromosomal arrangements indicate the limits and orientation of homologous regions. Arrowed bars represent annotated coding regions whereas rhombuses represent annotated transposable elements and other repetitive sequences. Thick dark red lines above or below a particular breakpoint region indicate its collinearity relative to D. pseudoobscura.

Full size image
Figure 4
figure 4

Functional annotation of A chromosome inversion breakpoints. Schematic representation of the annotated breakpoint regions corresponding to X chromosome inversions Af2 to Af5 that originated and became fixed since the D. subobscura-D. guanche split. D. guanche, above; D. subobscura, below. Dashed lines between chromosomal arrangements indicate the limits and orientation of homologous regions. Arrowed bars represent annotated coding regions whereas rhombuses represent annotated transposable elements and other repetitive sequences. Thick dark red lines above a particular breakpoint region indicate its collinearity relative to D. pseudoobscura.?, missing information.

Full size image

Comparison of the fragments spanning the D. guanche and D. subobscura breakpoints of the four autosomal and A chromosome Af2 and Af5 fixed inversions with the D. pseudoobscura genome allowed us to establish which of the two arrangements of each inversion exhibited the ancestral state. Indeed, the D. guanche fragments spanning the proximal breakpoints of inversions Jf, Of and Af2, and the distal breakpoints of inversions Ef2, Af3, Af4 and Af5 are collinear or partially collinear with D. pseudoobscura (Figs 3 and 4). The presence of gene Cenp-C in both the D. guanche and D. subobscura breakpoints of inversion Of reflects an intra-chromosomal transposition that occurred in these species ancestor. In contrast, the presence of the overlapping genes tapas and GA32843 in the distal D. subobscura breakpoint of inversion Af5 would reflect an inter-chromosomal transposition that occurred in D. subobscura after the D. guanche split. Concerning inversion Ef1, the D. subobscura arrangement is the ancestral arrangement for the affected region as revealed by the collinearity relative to D. pseudoobscura exhibited by its distal breakpoint region (Fig. 3).

In inversions Ef2, Of and Af3, the presence in both D. subobscura breakpoints of a duplicated fragment that was present in only one of their D. guanche breakpoints (Figs 3 and 4) is consistent with the three inversions having originated by the staggered-breaks mechanism from the D. guanche ancestral arrangement. Concerning the remaining inversions, there is no indication that they might have originated through ectopic recombination given the present absence of repeat motifs in inverted orientation at the two breakpoints of any of them in both D. guanche and D. subobscura (Figs 3 and 4).

Discussion

In the subobscura cluster of Drosophila, inversions fixed differentially between any of its three species pairs were identified either through banding pattern comparison between species11,25,26,27 or through the observation of polytene chromosomes in hybrids between D. madeirensis and either D. subobscura or D. guanche12,13,22. Concerning the more distantly related species —D. subobscura and D. guanche—, the number of inversions inferred through banding pattern comparison was the same in the different studies for both their autosomes and A chromosome11,13. There were, however, some discrepancies concerning the breakpoint assignment of some of the multiple overlapping inversions that had become sequentially fixed in segment I of the A chromosome since the D. subobscura-D. guanche split, and therefore also in the putative order in which these inversions occurred11,13. Moreover, comparison of the banding pattern of the three species of the subobscura cluster and those of six other species of the obscura group did not allow the establishment of how the different members of the subobscura cluster are related to the other six species, except for the J and E chromosomes, for which the subobscura cluster ancestor would have the arrangement presently found in D. guanche27.

Concerning the three species of the subobscura cluster, polytene chromosomes in hybrids between D. madeirensis and either D. subobscura or D. guanche12,13,22 revealed that at the cytological level (i) D. madeirensis only differed from the standard arrangement of the D. subobscura chromosomes by two A chromosome inversions and three autosomal inversions12,22, (ii) for the A chromosome, the differences exhibited by D. madeirensis relative to D. subobscura are shared by D. guanche13, and (iii) the most centromere-proximal A chromosome inversion present in D. madeirensis differed from the D. subobscura polymorphic A1 inversion12.

The molecular characterization of the breakpoints of eight of the ten inversions that became fixed since the divergence of D. subobscura and D. guanche and their comparison with the D. pseudoobscura genome has allowed us to detect collinearity and therefore to infer which was the chromosomal arrangement for each of these inversions that was present in the ancestor of the subobscura cluster. Our results would support that the D. subobscura population that colonized the Canary Islands archipelago had the same chromosomal arrangement than D. guanche presently has in seven of the eight regions affected by the fixed inversions with breakpoints here characterized, being the region affected by inversion Ef1 the only exception (Fig. 5). Moreover, previous cytological studies had revealed that D. guanche and D. madeirensis share the same arrangement at the regions affected by inversions Af1 and Af622. It can be therefore concluded that i) only inversion Ef1 would have occurred and become fixed upon the colonization of the Canary Islands by D. subobscura (i.e., in D. guanche), and ii) the chromosomal arrangement at the remaining nine regions of D. guanche would be a relict of the arrangements present at the ancestor of the subobscura cluster (Fig. 5). It can be, moreover, inferred that seven of these inversions —autosomal inversions Jf, Ef2 and Of as well as A chromosome inversions Af2 to Af5— became fixed in D. subobscura after this species colonized the Canary Islands but prior to the colonization of the Madeira island (i.e., between 1.8 and 0.6 million years ago according to previous estimates of the corresponding species divergence times28) whereas A chromosome inversions Af1 and Af6 would have become fixed thereafter in D. subobscura. Concerning the E chromosome, the present inference stands in contrast with the ancestral character of the Ef1+f2 arrangement inferred from the banding pattern comparison of D. guanche and other six species of the obscura group27, which might be due to the limited resolution of banding pattern comparison when distantly related species are compared. The much higher number of inversions fixed in the broadly distributed D. subobscura than in the island endemic D. guanche would be at odds with expectations if these inversions had not affected, or slightly affected, the fitness of their bearers. It should be however considered that the continental species possibly experienced more environmental challenges than the island species, and that the adaptive character of at least some of the inversions that emerged in D. subobscura might have contributed to cope with these challenges, and consequently, led them to fixation. The putative selective advantage of any of these inversions could be due to the structural change itself29,30,31 or to the particular variant of one or more of the genes included in the region affected by the structural mutation.

Figure 5
figure 5

Distribution across the subobscura cluster phylogeny of the ten fixed inversions since the D. subobscura-D. guanche split. The ancestral arrangement of chromosomes J, E, O and A (X), as inferred in the present study, is given in the cluster ancestor. Inversions that became fixed in these chromosomes are presented in the different branches of the phylogeny. The arrangement for each chromosome upon the fixation of the 10 inversions is given for each lineage, with numbers in parentheses referring to presently polymorphic arrangements as a result of inversions that originated thereafter in the four affected chromosomes. *Inversion O3 is currently polymorphic in D. subobscura and always associated with inversion O4.

Full size image

Concerning inversion Af1 and the D. subobscura A1 polymorphic inversion, the breakpoints of neither inversion could be identified20 (present work) but they were narrowed down to differently sized regions of the D. guanche assembled chromosome (see Results section). The localization of the proximal and distal Af1 breakpoint regions in D. guanche differs by at least 880 and 790 kb, respectively, from that inferred through BLAST search of the corresponding regions of the D. subobscura A1 inversion20, which corroborates the cytological results obtained in D. subobscura-D. madeirensis hybrids12,22.

Upon establishing the ancestral character of the D. guanche arrangement for segment I of the A chromosome, we aimed at inferring the order in which its four overlapping inversions occurred and became fixed in the D. subobscura lineage. Our identification of both breakpoints of each of these four inversions in D. guanche has resulted in a cytological map with increased resolution, which has allowed us to accomplish our goal. Figure 6 represents the two possible orders based on our results in which the four inversions might have accumulated and led from the A chromosome arrangement present in the subobscura cluster ancestor —now only present in D. guanche (Fig. 2)— to the Ast segment I arrangement of D. subobscura. Given that segment I of D. madeirensis only differs from the D. subobscura Ast arrangement by the Af1 inversion, it can be inferred that this inversion that occurred upon the D. madeirensis-D. subobscura split was the last one to occur13. This allows us to unambiguously establish that the order in which the four inversions occurred is the following: Af3, Af4, Af2 and Af1 (Fig. 6-I).

Figure 6
figure 6

Sequential order of occurrence and fixation in the D. subobscura lineage of the four A chromosome overlapping inversions. Schematic representation of the order established in the present study (I) and its discarded alternative (II) for the sequential occurrence and fixation of inversions Af1 to Af4 of the A chromosome in the D. subobscura lineage after its split from D. guanche (see text). Horizontal bars with differently colored boxes represent the different chromosomal arrangements, with the ancestral order (now only present in D. guanche) in the upmost part of the figure and the D. subobscura Ast order in its lowest part. Boxes colored as in Fig. 2 reflect conserved blocks relative to D. subobscura. Pairs of crossed lines between arrangements represent the regions affected by each of the four inversions. Arrows connecting the different chromosomal arrangements represent the sequential accumulation of inversions from the ancestral D. guanche arrangement (see text).

Full size image

Concerning the eight fixed inversions with breakpoints here molecularly characterized, our results are consistent with three of them having originated by the staggered-breaks mechanism whereas they are not informative on the mechanism involved in the origin of the other five (Figs 3 and 4). Our results would seem to be in contrast with those previously obtained for nine polymorphic inversions of D. subobscura15,16,17,18,19,20 where the staggered-breaks mechanism was considered the most common (proposed in eight out of nine inversions with breakpoints molecularly characterized). It should be however noted that the point and structural mutations that accumulate through time at the breakpoint regions tend to erode the signals left by the different originating mechanisms. This erosion would render polymorphic inversions more adequate than fixed inversions to establish the role played by the different originating mechanisms given the differential time scale of fixed and polymorphic inversions occurrence. Finally, our analysis of the D. guanche and D. subobscura breakpoint regions revealed two transpositions. This observation together with the loss of collinearity relative to D. pseudoobscura detected at one of the two breakpoint regions of each ancestral arrangement indicates that additional structural changes have occurred in the vicinity of inversion breakpoints, as previously observed at the extended breakpoint regions of some D. subobscura polymorphic inversions16,17,18.

In summary, we have identified and molecularly characterized the breakpoints of eight of the ten chromosomal inversions —four autosomal and six sex-linked— that became fixed since the D. subobscura-D. guanche split. The breakpoints of the other two fixed inversions could only be narrowly delimited. Based on our molecular and cytological results, we propose a modified version of the cytological map of the D. guanche A chromosome. The molecular information here obtained together with the previous cytological results have allowed us to establish that in nine of the ten regions affected by fixed inversions, D. guanche can be considered a relict exhibiting the chromosomal arrangements present at the ancestor of the subobscura cluster. This information has resulted in one of the first chromosomal phylogenies based on the comparison of inversion breakpoint sequences, and constitutes therefore an important contribution to advance our knowledge on chromosomal evolution. Finally, we have been able to unambiguously establish the order in which the four A chromosome overlapping inversions occurred and became fixed.

Materials and Methods

One Drosophila guanche (GI_16) isogenic strain and four D. subobscura (ch cu, OF28, OF40 and FO31c) isogenic strains were used in the present study. The D. guanche GI_16 strain is homokaryotypic for all chromosomes32 whereas the D. subobscura strains are either homokaryotypic for the five acrocentric chromosomes (ch cu for Ast, Jst, Ust, Est and O3+4; and OF28 for Ast, J1, U1+2+8, Est and Ost) or for a subset of them (OF40 for Ast, J1, Est and O3+4+8; and FO31c for Jst, Ust, Est and O3+4+2). Isogenic lines were obtained by at least 12 generations of brother-sister sib-mating16,19.

The sequences of 335 molecular markers with known cytological information in D. subobscura were used to identify their D. guanche homologs through BLAST search. Comparison of these datasets allowed us to delimit a large region spanning each inversion breakpoint in D. guanche. In order to identify each inversion breakpoint, the sequence of each D. guanche large region was thereafter either i) compared to draft2 of the D. subobscura genome (BSI) to further delimit the breakpoint region, or ii) used to initiate chromosomal walks. Oligonucleotides to amplify additional probes were designed on the D. guanche genome23 and in a few cases on draft2 of the D. subobscura genome (BSI). Probes were amplified by PCR using TaKaRa DNA polymerase (Takara Bio Inc) and genomic DNA from either the D. guanche GI_16 or the D. subobscura ch cu strains [using the Puregen Cell kit B (Qiagen)]. Upon their Biotin-16-dUTP (Roche) labeling, they were in situ hybridized on polytene chromosomes of one of the four D. subobscura strains (ch cu, OF28, OF40 and FO31c) whereas those crossing each breakpoint were also hybridized on the D. guanche GI_16 strain. Hybridization signals were subsequently located on the D. subobscura cytological map33. All steps of the in situ hybridization procedure were performed as described in Montgomery et al.34 with minor modifications. Digital images at a 400 magnification were obtained using a Leica DFC290 camera mounted on a phase contrast Axioskop 2 Zeiss microscope.

Fragments spanning the breakpoints in D. subobscura were PCR amplified using DNA from the corresponding strain using TaKaRa DNA polymerase (Takara Bio Inc) and oligonucleotides anchored at each breakpoint flanking regions. The amplified fragments were sequenced using primer walking whenever necessary. Amplicons were purified with MultiScreen PCR plates (Millipore) prior to their sequencing with the ABI PRISM version 3.2 cycle sequencing kit. Sequencing products were separated on an ABI PRISM 3730 sequencer. Sequences were assembled using the DNASTAR package (Burland 2000). The D. subobscura sequences newly obtained have been deposited in the European Nucleotide Archive (ENA) under project number PRJEB27938.