Diploid numbers and morphology of chromosome pairs
Evolution of chromosome numbers
Male diploid numbers of pholcids range from 9 (three Micropholcus species) [25, this study] to 33 (Artema spp.) [this study]. The diploid number of the Micropholcus species is the lowest one found in araneomorph spiders with monocentric chromosomes so far [25]. Particular pholcid subfamilies differ by their range of numbers of chromosomal pairs (NCPs). Reported data suggest a low diversity of NCPs in modisimines (7–8 pairs) [25,26,27, this study], ninetines (12–13 pairs) [this study], and smeringopines (11–13 pairs) [9, 28,29,30,31,32, this study]. The other two subfamilies show a much higher range of NCPs: artemines from 6 to 15 pairs [30, 32, 33, this study] and pholcines from 4 to 11 pairs [25, this study]. A high range of NCPs in artemines could reflect the possible non-monophyly of this group (see Fig. 15). In pholcines, there are also reports on 12 chromosome pairs in two species, but these data are dubious. The number of chromosome pairs in the species analysed by Sharma and Parida [34] can not be verified from the published information. In the species studied by Wang et al. [35], the presented data do not allow to determine unequivocally the NCPs and SCS.
Remarkably, the karyotype of an early-diverging pholcid, Artema (2n♂ = 33, X1X2Y, chromosomes biarmed) is very similar to karyotypes of the haplogynes Filistata insidiatrix (Filistatidae; 2n♂ = 33, X1X2Y) [9], Paculla sp. (Pacullidae; 2n♂ = 33, X1X2Y) [14], and Hypochilus pococki (Hypochilidae; 2n♂ = 29, X1X2Y) [9]. Whereas Artema, Filistata, and Paculla exhibit the same diploid number and biarmed morphology of chromosome pairs, the karyotype of Hypochilus is slightly different, showing lower NCPs and a higher ratio of monoarmed chromosome pairs. The karyotype of Hypochilus can be derived from the pattern found in Artema, Filistata, and Paculla by two chromosome fusions and several pericentric inversions. This pattern of karyotypes suggests that the ancestral pholcid karyotype was close to those found in Artema, Filistata and Paculla (2n♂ = 33, X1X2Y, chromosomes biarmed).
Recent studies on spider phylogenomics suggest two primary lineages of haplogynes, one consisting of filistatids and hypochilids, and another one comprising synspermiate haplogynes, including pholcids and pacullids [10, 36, 37] (Fig. 1). Considering the placement of Artema, Filistata, Paculla, and Hypochilus in this phylogeny, the karyotype structure found in the three first genera (2n♂ = 33, X1X2Y, biarmed chromosomes) could be also the ancestral haplogyne karyotype. However, this conclusion conflicts with the hypothesis supposing that the ancestral karyotype of haplogynes was very close to that of entelegynes (20 chromosome pairs + X1X20) [9]. This hypothesis is supported by the occurrence of very similar karyotypes in the haplogyne family Drymusidae (17 biarmed pairs + X1X2Y) and the family Austrochilidae (18 biarmed pairs + XY) [9], which is an early-diverging lineage of a clade formed by protoentelegynes and entelegynes [37]. The observed karyotype pattern of austrochilids and drymusids could reflect an early separation of drymusids within haplogynes, specifically before derivation of the clade including filistatids, hypochilids, pacullids, and pholcids. To determine the correct phylogenetic position of drymusids, more detailed phylogenomic analysis of early-diverging araneomorph clades is needed.
A fundamental trend of spider karyotype evolution is the reduction of chromosome numbers, which took place independently in many clades [38]. Our analysis suggests the same pattern in pholcids. Available data indicate the reduction to 13 chromosome pairs in the last common ancestor of smeringopines and pholcines, and separately in ancestral ninetines, or alternatively in the last common ancestor of ninetines, smeringopines and pholcines (see “Phylogenetic implications”, p. 29). Furthermore, a comparison of cytogenetic and molecular data suggests a further reduction to 11 pairs in ancestral pholcines, to 8 pairs in the common ancestor of modisimines, and to 7 pairs in the common ancestor of the artemine genera Holocneminus, Physocyclus, and Wugigarra (Fig. 15). The longest one or two chromosome pairs of pholcids are often prominent, which could reflect the formation of these pairs by fusion. In Leptopholcus (2n♂ = 17, X1X2Y), the first three pairs are considerably longer than the remaining ones. This set could be derived by three fusions from a karyotype containing ten chromosome pairs. Reductions of chromosome numbers also took place during the evolution of pholcid genera. However, the frequency of these events on genus level is low (Additional file 1: Table S1).
Variation of NCPs has also been suggested on intraspecific level in pholcids, namely in Crossopriza lyoni [30]. This widespread synanthropic species has been karyotyped by several authors. Recent studies consistently report 11 biarmed chromosome pairs in this species; analysed populations came from Brazil, India, and Vietnam [30, 32,33,34, 39, this study]. This number of pairs is probably an ancestral feature of the clade formed by Crossopriza, Holocnemus caudatus, H. hispanicus, and Stygopholcus (Fig. 16). In contrast, initial cytogenetic studies on C. lyoni suggested a higher NCPs in Indian populations, namely 12 [29] or 13 [28]. While the first study does not contain information on chromosome morphology, the karyotype with 13 pairs was exclusively biarmed. This pattern could be derived from a set with 11 biarmed pairs by fissions of two pairs and subsequent inversions of four newly formed pairs. However, our results indicate a very low frequency of chromosome fissions during pholcid evolution and a low intraspecific variability of pholcid karyotypes. Therefore, differentiation of the karyotype of C. lyoni by multiple fissions is very unlikely. Diversity of NCPs in C. lyoni is most probably an artifact, possibly caused by species misidentifications. Reported data on NCPs also differ in Micropholcus fauroti. While Araujo et al. [26] reported eight metacentric pairs in a Brazilian population, we found only four metacentric pairs in another Brazilian population and African populations (Cape Verde, South Africa). Notably, we frequently found in our preparations of Micropholcus clusters formed by the fusion of chromosome plates. This artefact, along with the low number of observed plates, could have led Araujo et al. [26] to report double NCPs in this species. Four chromosome pairs were also reported in other karyotyped Micropholcus species [25].
Evolution of chromosome pairs
Similarly to the majority of other haplogynes with monocentric chromosomes [9], karyotypes of most pholcids are predominated by metacentric chromosomes. A different pattern has been reported only in two Physocyclus species [30, this study] and Pholcus manueli [35]. The karyotype of the latter species is suggested to be composed exclusively of monoarmed chromosomes. However, this idea is based only on the pattern of constitutive heterochromatin, i.e. a marker that could also be localised in regions other than the centromere. Therefore, this hypothesis should be tested by determination of the centromere position during mitotic metaphase or metaphase II. During metaphase II, chromatids of spider chromosomes are separated, except for the centromere.
At first glance, chromosome pairs of pholcids seem to be conservative elements, due to their predominantly metacentric morphology. However, comparison of closely related species revealed a dynamic nature of these pairs. Closely related taxa having the same number of pairs often differ by the morphology of one or several pairs, whose metacentric morphology is changed to submetacentric or even monoarmed (Additional file 1: Table S1). This pattern suggests the operation of pericentric inversions or some translocation variants. In some cases, the size of the pair is retained even after the change of its morphology to non-metacentric (e.g., acrocentric pair of Artema nephilit), which points to the operation of pericentric inversions. An enormous increase of the first pair (Crossopriza lyoni, Holocnemus caudatus) or considerable reduction of the last pair (some Psilochorus species, Stygopholcus, Wugigarra [this study], Mesabolivar spinulosus [25]), which are not accompanied by changes of NCPs, probably reflect an origin of these pairs by non-reciprocal or unequal reciprocal translocations. In the last three taxa, the last pair exhibits a subtelocentric morphology, which probably reflects translocation of most of an arm of an original biarmed pair to another chromosome. In Aetana, a considerable reduction of the last pair is accompanied by a change in morphology to acrocentric. This pattern could reflect centric fission of a biarmed chromosome pair, followed by integration of one product into another chromosome.
Diploid numbers of pholcids could be decreased by centric fusions of monoarmed pairs. Beside this, the number of pairs could be reduced by nested fusions. In contrast to centric fusions, this process involves biarmed chromosomes.
In conclusion, our data suggest the frequent involvement of fusions, inversions, and translocations of chromosome pairs in the karyotype evolution of pholcids. These kinds of rearrangements can be very effective in the formation of interspecific reproductive barriers [40, 41], and may thus have played a role in speciation processes in pholcids.
Sex chromosomes
X1X2Y system
Ancestral composition and origin of the X1X2Y system
Pholcids exhibit a considerable diversity of sex chromosomes. We found six SCS in these spiders, namely X0, X1X20, X1X2X30, XY, X1X2Y, and X1X2X3X4Y. The X1X2Y system has been reported in seven haplogyne families, namely Sicariidae [9, 42, 43], Filistatidae [9, 14,15,16], Drymusidae, Hypochilidae, Pholcidae [9], Pacullidae [14], and Plectreuridae [15]. Specific morphology and the meiotic pairing of chromosomes X1, X2, and Y suggest that the X1X2Y system arose once in haplogynes [9]. Together with molecular and paleontological data, this points to an ancient origin of the haplogyne X1X2Y system. Plectreurids have been found in Jurassic strata [44]. A recent molecular phylogeny suggests the origin of spiders possessing an X1X2Y system during the early Mesozoic [37]. The phylogenetic distribution of the X1X2Y system using recent phylogenomic trees [10, 36, 37] suggests that this sex chromosome determination is ancestral for haplogynes, including pholcids. It could have arisen even earlier, namely before the separation of entelegyne and haplogyne spiders [16] or before the separation of mygalomorphs and araneomorphs in ancient opisthothele spiders. If so, the supposed ancestral sex chromosome system X1X20 of mygalomorphs [6] and entelegynes [9] arose from the X1X2Y system by the loss of the Y chromosome. The ancestral X1X2Y system probably consisted of two large metacentric X chromosomes of similar size and a metacentric Y microchromosome [9]. Species with this pattern have been found in almost all X1X2Y families [9, 16, 18, this study]. The X1X2Y system was supposed to arise by rearrangements between autosomes and sex chromosomes [43]. In this case, it would be an ancient neo-sex chromosome system. According to our hypothesis, chromosomes X1, X2, and Y are derived from CSCP. It was suggested that multiple X chromosomes of spiders arose by nondisjunctions of the X chromosome of the CSCP [7]. The Y chromosome could originate in a similar way, namely by a nondisjunction of the Y chromosome of the CSCP and subsequent degeneration of the newly formed element. Nondisjunctions could be a major mechanism of formation of multiple X chromosomes in spiders [7, 18, 45]. This unusual origin of sex chromosomes is supported by the inactivation of multiple X chromosomes during meiosis of spider females. This unique behaviour probably evolved to avoid the negative effects of duplicated X chromosomes on female meiosis [5, 7].
According to our data, chromosomes of the X1X2Y system are dynamic elements. Although they exhibit a conservative pairing during male meiosis, they underwent frequent rearrangements during pholcid evolution. Remarkably, chromosomes X1, X2, and Y differ by their pattern of morphological evolution and evolutionary plasticity. See Additional file 28: Appendix S1 for evolution of particular sex chromosomes of the X1X2Y system.
Conversion of the X1X2Y system into other sex chromosome systems
In some haplogynes, including pholcids, the X1X2Y system has been transformed into other SCS [9, 15, this study]. According to [9], the X1X2Y system underwent conversion to the X0 system, namely by X chromosome fusion (formation of the XY system) and subsequent loss of the Y chromosome. Our results suggest that patterns of conversion of the X1X2Y to the X0 system are more diversified. In some pholcids, loss of the Y chromosome preceded fusion of the X chromosomes, which resulted in the formation of the X1X20 and, subsequently, of the X0 system. We suppose that this scenario could be more frequent during the evolution of the haplogyne X1X2Y system, due to specific features of its Y chromosome, which could facilitate degeneration and extinction of this element (tiny size, absence of recombinations between the X and Y chromosome).
XY system
In Wugigarra the X1X2Y system was transformed into the XY system by X chromosome fusion [this study]. The XY system of another haplogyne, Diguetia (Diguetidae), is suggested to arise by the same process [9]. The sex chromosomes of Wugigarra and Diguetia retained their metacentric morphology, except for the acrocentric X chromosome of D. canities, which probably originated from metacentric X by pericentric inversion [9]. The Y chromosomes of these haplogynes are small elements. While sex chromosomes of Wugigarra retain their achiasmatic pairing by both ends, only one X chromosome end is involved in achiasmatic pairing in Diguetia [9]. Notably, the XY system has also been suggested in other pholcids, namely Smeringopus ndumo (reported as S. pallidus) [9] and Holocnemus pluchei [31]. Our data showed that these reports are erroneous, based on revision of original preparations (S. ndumo) and laboratory stock studied by previous authors (H. pluchei). An XY system was also discovered in several other spiders [6, 9, 46, 47]. Its origin is, however, different. It is suggested to have arisen from the X0 determination by a fusion of the X chromosome and an autosome(s), resulting in the formation of neo-X and neo-Y chromosomes. Another possible origin of the system might be by a fusion of the X chromosome and X chromosome of the CSCP.
Multiple X chromosome systems
X1X20 system
Although the X1X20 system is the most frequent sex chromosome determination in entelegyne araneomorphs [18, 45], it is rare in haplogyne araneomorphs [9, 15, this study]. Provided that the X1X2Y system is ancestral for araneomorphs [16], the X1X20 system has originated in the same way in both entelegynes and haplogynes, namely from the X1X2Y system by the loss of the Y chromosome. The X1X2Y system of the pholcids Aetana and Artema, including a minute Y chromosome [this study], could represent an evolutionary transition between the X1X2Y and X1X20 systems. In haplogynes, the X1X20 system has originated several times, namely in filistatids [15], plectreurids [9], and pholcids [this study].
In pholcids, we found the X1X20 system in two smeringopine genera, Hoplopholcus and Smeringopus. Mapping of the distribution of the X1X20 system in smeringopines suggests an origin of this system in their ancestor. Remarkably, we never found the X1X20 system in pholcids in which it was reported previously (see database of Araujo et al. [24]). Instead, these species exhibit the X1X2Y system, which had probably been mistaken for the X1X20 system, due to the small size of the Y chromosome. The X1X2Y system was revealed in pholcids less than two decades ago [9].
Remarkably, the X1 and X2 chromosomes of the X1X20 system show a similar evolution to the X chromosomes of the X1X2Y system in pholcids (see Additional file 28: Appendix S1 for evolution of chromosomes of the X1X2Y system). While the X1 is a conservative element retaining large size and metacentric morphology, X2 underwent reduction and frequent changes of morphology. The size of the X1 chromosome is more variable (7.9–15.1% of TCL) than that of the X2 element (5.1–7.4% of TCL), which could reflect insertions of fragments derived from CPs into the X1 chromosome. In each species, the X2 chromosome is reduced in comparison with the X1 chromosome (including representatives with an ancestral state, i.e. metacentric morphology of the X2 chromosome). Therefore, the X2 chromosome was probably already reduced in an ancestor of smeringopines (having a X1X20 or X1X2Y system). The original metacentric X2 chromosome has been transformed several times to a monoarmed one, namely in the ancestor of Hoplopholcus and in several Smeringopus species. The morphology of the X2 chromosome has most probably changed by a pericentric inversion or translocation.
X1X2X30 system
A sex chromosome system comprising three X chromosomes is relatively frequent in entelegyne [48] and mygalomorph spiders [6]. Among haplogynes, Smeringopus pallidus is the first reported species with this sex chromosome determination [this study]. It was obviously derived from the X1X20 system found in the other Smeringopus species. However, the origin of the third X chromosome of S. pallidus is unresolved. The behaviour of the three X chromosomes of this species during male meiosis (pairing, pattern of condensation and heteropycnosis) is the same as in X1X20 species, which contradicts the possible origin of the extra X chromosome from autosomes. Another possibility is nondisjunction of an X chromosome, which is supposed to be a mechanism of X1X2X30 formation in entelegynes [7, 45, 49]. This hypothesis is supported by the same meiotic behaviour of sex chromosomes in X1X20 and X1X2X30 species. The sex chromosomes of S. pallidus (two large metacentric X chromosomes of similar size + one small metacentric X chromosome) can be derived from the ancestral pattern of the X1X20 system in Smeringopus (two metacentric X chromosomes, X2 considerably smaller than X1) by nondisjunction of the X1 chromosome. Formation of an additional large X chromosome would substantially increase the sum of the relative lengths of the X chromosomes. However, the values do not differ substantially between S. pallidus (16.9% of TCL) and the karyotypically most similar S. ndumo (18% of TCL), and fall within the range of values found in other Smeringopus species (14.2–19.8% of TCL). Therefore, the X3 chromosome of S. pallidus more likely arose by a fission of the ancestral metacentric X1 chromosome into two acrocentric chromosomes, followed by their pericentric inversion. This hypothesis is supported (1) by a reduction of the X1 chromosome of S. pallidus in comparison with other Smeringopus species, and (2) by the sum of X1 and X3 sizes in S. pallidus, which is similar to the X1 size in the karyotypically most similar S. ndumo. To resolve the origin of the X1X2X30 system in S. pallidus, sequencing of its X chromosomes needs to be carried out.
X0 system
Phylogenetic distribution and origins of the pholcid X0 system
The X0 system is common in pholcids. It has been reported in some artemines, pholcines, smeringopines, and all modisimines karyotyped so far [24, this study]. Character mapping suggests at least five origins of the X0 system in pholcids, namely in the (1) artemines with low diploid numbers (see "Phylogenetic implications", p. 29), (2) ancestor of modisimines (Additional file 29: Fig. S23), (3) common ancestor of smeringopines Holocnemus, Crossopriza, and Stygopholcus (Additional file 30: Fig. S24), (4) pholcine Belisana, and (5) common ancestor of the pholcines Cantikus and Micropholcus (Additional file 31: Fig. S25). The X0 system was also reported in Pholcus manueli [35]. However, chromosome plates obtained in P. manueli (male mitoses) do not allow us to determine the SCS of this species unequivocally. Similar to P. phalangioides (X1X2Y) [9], the karyotype of P. manueli is composed of 25 chromosomes and contains an odd peculiar chromosome formed exclusively by constitutive heterochromatin [35]. The odd heterochromatic element of P. phalangioides is a Y chromosome [9]. Therefore, P. manueli exhibits with all probability the X1X2Y system.
The X0 system is common in spiders. Phylogenetic distribution of the X0 system in spiders suggests multiple origins of this system [6, 18, 45]. X0 systems of entelegynes and mygalomorphs arose by chromosome fusions from multiple X chromosome systems [6, 18, 45]. In haplogynes, the X0 system has been found in nine families (see database [24]). The X0 system of haplogynes with monocentric chromosomes arose from XY or X1X20 systems. The X0 system of smeringopine pholcids arose from the X1X20 system by X chromosome fusion [this study]. In contrast, the X0 system of a clade formed by Chisosa, Holocneminus, Physocyclus, and Wugigarra probably arose from the XY system (Additional file 29: Fig. S23). Formation of the X0 system in the other pholcids is unresolved. Our discovery of both X0 [this study] and ancestral X1X2Y systems [J. Král and O. Košulič, unpublished] in the same genus, the pholcine Belisana, indicate a relatively fast formation of the X0 system during the evolution of this clade. An increase in the size of the Y chromosome during the evolution of pholcines (see Additional file 28: Appendix S1 for evolution of the Y chromosome) did not prevent this element from being subjected to reduction and loss in the common ancestor of Cantikus and Micropholcus (X0).
Evolution of the X0 system in pholcids
Despite its multiple origins, the single X chromosome of pholcids exhibits an extremely conservative morphology. Except for Modisimus it is always mediocentric (Additional file 1: Table S1), which indicates a strong selection pressure to keep this feature. We suppose that this morphology could be essential to ensure the self-association and regular segregation of the X chromosome univalent during male meiosis. In mice, the meiotic stability of the sex chromosome univalents is promoted by their self-association [50].
In contrast to morphology, the size of the X chromosome varies considerably, namely from 8.89 (Belisana sabah) to 21.21% of TCL (Chisosa diluta) (Additional file 1: Table S1). A considerable diversity of X chromosome sizes occurs even on a genus level (Crossopriza, Holocnemus, Psilochorus). The increase in size of the X chromosome could be a consequence of an integration of material from CPs, as suggested, for example, by the large difference of X chromosome size between two Holocnemus species (H. caudatus and H. hispanicus).
X1X2X3X4Y system
Comparison of data on the male karyotype and meiosis suggests that the X1X2X3X4Y system occurs in Kambiwa. This pholcid exhibits one chromosome pair less than the other studied ninetine, Pholcophora, which displays a X1X2Y system [this study]. This pattern indicates the origin of the X1X2X3X4Y system by rearrangements between chromosomes of the X1X2Y system and a chromosome pair. The Y microchromosome was not involved in the rearrangements. Sex chromosome systems arising by rearrangements between chromosome pair(s) and sex chromosomes have also been reported in some other araneomorphs [7, 9, 47, 51,52,53] and in some mygalomorphs [6, 46]. These events are always apomorphies of low-level taxa (species or species groups), which indicates their relatively recent origin.
Sex chromosomes and pholcid speciation
In summary, our data suggest frequent and diverse structural changes of sex chromosomes during pholcid evolution. These events include inversions and translocations of sex chromosomes, integration of fragments or even whole chromosomes into the SCS, and loss of the Y chromosome. Since structural changes of sex chromosomes are often very potent in formation of interspecific reproduction barriers [54,55,56,57], they have probably been involved in the speciation process in pholcids. In keeping with this view, closely related pholcids often differ in sex chromosome morphology [this study].
Particular sex chromosomes of pholcid males do not recombine, which suggests a high degree of differentiation of these chromosomes. This degree of sex chromosome differentiation is associated with markedly stronger reproductive isolation [58]. Although sex chromosome rearrangements could be less detrimental in meiosis of spider males due to achiasmatic sex chromosome pairing, in female heterozygotes they can lead to a collapse of homologous sex chromosome pairing.
Nucleolus organizer regions
NOR bearing pairs
The pattern of NORs has only been determined in a low number of spiders so far. Usually, these structures were detected by silver staining, a technique that visualizes NORs that were active in the preceding interphase only [59]. Therefore, it is impossible to reconstruct the evolution of NORs in particular families or even higher taxa of spiders. Karyotypes of most spiders contain, however, a low number of these structures. The common ancestral pattern of opisthothele spiders comprised probably one or two chromosome pairs bearing a terminal NOR locus [6].
In our study, we analysed the evolution of spider NORs on a family level for the first time. Nucleolus organizer regions have been visualized only in three pholcids so far, in each case by silver staining [30, 32]. To reconstruct NOR evolution in pholcids, we analysed 30 pholcid species, including one species studied earlier by silver staining. To detect NORs, we used fluorescence in situ hybridization (FISH), which visualizes inactive NORs as well. According to our results, pholcids exhibit a considerable diversity of NOR numbers (from one to nine loci); congeneric species often differ in the number of NOR loci. Almost all detected NOR loci are homozygous for the presence of NOR, which suggests a high stability in the number of NOR loci at species level. Findings of loci heterozygous for the presence of NOR are rare in pholcids (Nipisa, Pholcus pagbilao, Smeringopus sp.) [this study]. In the case of Artema atlanta, intraspecific variability of NORs is doubtful. While the studied karyotype from South Africa contained one NOR locus [this study], two NOR loci were reported in an Indian population probed by silver staining [32]. Considering the low quality of the signals detected, the information of the latter authors should be verified by FISH.
Character mapping suggests that ancestral pholcids had a single biarmed CP bearing a terminal NOR (Additional file 29: Fig. S23). Whereas the number of NORs underwent multiple changes during pholcid evolution, the pattern of their location has remained conservative. Each NOR-bearing chromosome pair of pholcids includes a single NOR, except for exceptional pairs having NORs at both ends. These pairs were only found in some pholcines [this study]. NORs of pholcids are terminal, except for a pericentric NOR of Physocyclus globosus [30], which might originate from a paracentric inversion. The terminal location of pholcid NORs suggests that these structures have mostly spread by ectopic recombination. This mechanism could be promoted by hybridization of individuals belonging to populations differing in the number of NOR loci, which is indicated by the finding of heterozygotes for the presence of NOR in Nipisa, P. pagbilao, and, Smeringopus sp.
Our data indicate an increase in the number of NOR bearing CPs in some pholcid clades. In the ancestor of ninetines and the common ancestor of smeringopines and pholcines (Additional file 29: Fig. S23, Additional file 30: Fig. S24) or common ancestor of ninetines, smeringopines, and pholcines (see "Phylogenetic implications", p. 29), the number of NOR-bearing pairs increased to two. During subsequent evolution, NORs spread to three chromosome pairs in Kambiwa (Ninetinae) and in the ancestor of pholcines. Karyotypes of Smeringopus sp. and some pholcines contain four NOR-bearing pairs. In Belisana, there are even five pairs bearing NORs. On the other hand, character mapping suggests that the number of NOR-bearing pairs decreased several times in smeringopines (Stygopholcus, Holocnemus hispanicus, Hoplopholcus) and pholcines (Aetana, Muruta, Quamtana, common ancestor of Cantikus and Micropholcus) (Additional file 30: Fig. S24, Additional file 31: Fig. S25). Remarkably, these clades often exhibit reduced NCPs in comparison with their relatives; NORs retain a terminal position.
Sex chromosome-linked NORs
In contrast to other opisthothele spiders, haplogynes often exhibit sex chromosome-linked NORs [9]. These NORs are also common in pholcids. They have been found in all pholcid subfamilies [30, this study]. Sex chromosome-linked NORs of pholcids and other haplogynes are always restricted to the X chromosome(s) [9, 30, this study] except for Nipisa [this study]. Character mapping suggests at least five origins of these NORs in pholcids, namely in (1) Physocyclus (Arteminae, X0), (2) Kambiwa (Ninetinae, X1X2X3X4Y), (3) Psilochorus (Modisiminae, X0) (Additional file 29: Fig. S23), (4) the common ancestor of Crossopriza lyoni and Holocnemus hispanicus (Smeringopinae, X0) (Additional file 30: Fig. S24), and (5) during early evolution of pholcines (X1X2Y) (Additional file 31: Fig. S25). In modisimines, NOR data are available for Psilochorus only. Therefore, their sex chromosome-linked NORs could have arisen earlier than during the evolution of this genus. Sex-chromosome linked NORs of pholcids always have a terminal location, which suggests an origin of these structures by ectopic recombination.
Most pholcids with the X0 system and sex-chromosome-linked NORs display two terminal NORs (e.g., Physocyclus, Psilochorus) [this study]. The second NOR was probably formed by ectopic recombination between the arms of the X chromosome, which was facilitated by their meiotic association. Although the Physocyclus species analysed in our paper exhibit two sex chromosome-linked NORs, a study on P. globosus did not reveal these structures [30]. Since these authors used silver staining and gonial mitoses to detect NORs, it is possible that sex-chromosome linked NORs of this species are inactivated in the germline cells. Alternatively, P. globosus may belong to a Physocyclus lineage that lacks sex chromosome-linked NORs.
We revealed the most complex evolution of sex chromosome-linked NORs in the X1X2Y system of pholcines. According to our data, sex chromosome-linked NORs already appeared during the early evolution of pholcines, no later than before the separation of the clade containing Aetana. Paleontological [60] and molecular data [61] indicate relative antiquity of the pholcine lineage with sex chromosome-linked NORs (further SCL-NOR clade). According to the latter dataset, this clade evolved at the latest during the Jurassic. The ancestral pattern of sex chromosome-linked NORs was probably formed by a single terminal locus on the X1 chromosome, which is retained in Aetana. In another early-diverging pholcine, Belisana, the X1X2Y system has been transformed to X0. During this transformation, the terminally positioned sex chromomosome-linked NOR originating from the X1 chromosome was probably retained. The second NOR of Belisana was probably formed by ectopic recombination between the arms of the single X chromosome.
The subsequent evolution of NORs in the pholcine X1X2Y system probably involved gradual spreading of the NORs by ectopic recombination to the other X1 end, and after that to one or both ends of the X2 chromosome. Remarkably, the X chromosome ends are attached in male meiosis to ensure sex chromosome pairing, which could promote spreading of NORs among sex chromosomes by ectopic recombination. We suppose that spreading of NORs by ectopic recombination could also be facilitated by the association of X chromosome bivalents during female meiosis of spiders [5]. Besides pholcines, multiple SCS containing NORs have been found only in the ninetine Kambiwa. In contrast to pholcines, the end of the X chromosome containing NOR is not involved in chromosome pairing in this spider (see “Sex chromosome behaviour”, p. 26).
As demonstrated in some animals, sex chromosome-linked NORs can ensure achiasmatic pairing of sex chromosomes [62, 63]. The specific terminal location of sex chromosome-linked NORs in pholcines suggests involvement of these structures in achiasmatic pairing as well. Pairing ensured by NORs could be more stable, which may have promoted an increase in Y chromosome size during the evolution of pholcines. We suppose that the terminal NOR(s) of pholcids with the X0 system have a similar function as in the X1X2Y system, i.e. they could strengthen the association of the X chromosome arms during male meiosis.
Remarkably, our data suggest the loss or degeneration of sex chromosome-linked NORs in some pholcids, especially in the SCL-NOR clade. The most obvious cases are Pholcus pagbilao and the clade formed by Cantikus and Micropholcus. In contrast to Pholcus phalangioides, we did not detect the target 18S rDNA motif on the sex chromosomes of P. pagbilao. This sequence could be degenerated or lost during the evolution of NORs involved in pairing. A similar pattern was found in some Drosophila species, in which most rDNA of sex chromosome-linked NOR is lost, except for motifs involved in pairing [64]. Loss/degeneration of sex chromosome-linked NORs in the common ancestor of the pholcines Cantikus and Micropholcus was accompanied by a considerable decrease of NCPs and formation of the X0 system in this lineage.
Some authors propose that sex chromosome-linked NORs could be a synapomorphy of some haplogyne clades [9, 30]. Our data support another scenario, namely multiple invasions of NORs on sex chromosomes of haplogynes by ectopic recombinations. Although ectopic recombination is probably also a frequent mechanism of NOR dispersion in entelegynes [65], these spiders exhibit a much lower frequency of sex chromosome-linked NORs than haplogynes [5]. It was suggested that the dispersion of NORs on spider sex chromosomes by ectopic recombination is reduced by sex chromosome inactivation in meiosis of both spider sexes [65]. Since sex chromosomes of haplogynes show a similar pattern of meiotic condensation and heterochromatinisation as entelegynes [9, this study], it would be interesting to determine the mechanisms facilitating the transmission of NORs on haplogyne sex chromosomes. It could be the lower degree of X chromosome condensation during some periods of male prophase I [this study]. In keeping with this hypothesis, NORs have almost never been detected on the Y chromosome of the haplogyne X1X2Y system [9, this study], which could be a consequence of its considerable condensation in the germline. Another possibility of frequent spreading of NORs on sex chromosomes of haplogynes could be the presence of NOR on their CSCP. Chromosomes of the CSCP are associated with the other sex chromosomes during male meiosis [7], which could facilitate spreading of NORs on these sex chromosomes.
Chromosome behaviour in the male germline
Modifications of meiotic division
The prophase of the first meiotic division is modified in pholcid males. Following pachytene, nuclei enter the so-called diffuse stage. The male diffuse stage was also found in other haplogynes, in the protoentelegyne family Leptonetidae [9], and in some clades of early-diverging mygalomorphs [6] and entelegynes [66]. The diffuse stage evolved in other animal groups as well. In some pholcids, the diplotene is reduced: recondensed bivalents exhibit a diakinetic morphology after the diffuse stage [this study]. This pattern has also been found in some other haplogynes [9, 67].
Pholcid karyotypes are predominated by biarmed chromosomes, which generally form more chiasmata than monoarmed ones [68]. In spite of this, most pholcids show a very low chiasma frequency in male meiosis (Additional file 2: Table S2, Additional file 7: Table S3, Additional file 12: Table S4, Additional file 19: Table S5). This pattern is probably an ancestral pholcid feature. The frequency of chiasmata is increased particularly in some pholcids with a low diploid number, namely some artemines, modisimines, and pholcines [25, 26, this study]. Since a reduction of diploid number is accompanied by decrease of total chiasma number per karyotype, increase of chiasma frequency per bivalent could be a compensatory mechanism to avoid reduction of genetic variability in these species. Pholcids usually exhibit a predomination of intercalar and distal chiasmata. A different pattern has been found in Modisimus and Physocyclus, which show a relatively high proportion of pericentric chiasmata. Relocation of chiasmata in these spiders could be a consequence of inversions, which can change the position of chiasmata [69].
Polyploid cells
Germline cells of spider males form spermiocysts, whose cells are connected by cytoplasmic bridges and synchronized during their development [70]. As a result, plates belonging to the same mitotic or meiotic phase are frequently fused on chromosome preparations [26, 30]. Besides these artefacts resembling polyploid plates, preparations of pholcid testes contained many large endopolyploid nuclei, which could belong to a specific cell lineage. Occurrence of these nuclei in testes reflects a high metabolic activity of testicular tissues. Overcondensation of sex chromosomes in endopolyploid nuclei of some pholcids could reflect transcriptional repression of these chromosomes. Endopolyploid nuclei were also reported in the testes of some other opisthothele spiders [5, 6, 71]. In pholcids, they have only been reported in the silk and venom glands so far [72].
Sex chromosome behaviour
Spider sex chromosomes are distinguished by a specific behaviour in the male germline [5, 6, 18, 38, 45, 73], which is sometimes already initiated at spermatogonial mitosis [5]. In some pholcids, the sex chromosomes exhibit a specific condensation during this division. Chromosomes of the X1X2Y system are usually associated in spermatogonia, whereas X chromosomes are often arranged in parallel [this study]. Moreover, sex chromosomes of this system are preferentially located in the middle of spermatogonial metaphases [9, this study]. Concerning the other pholcid sex chromosome systems, we were not able to determine the relative position of their chromosomes in spermatogonia. Notably, multiple X chromosomes of some entelegynes also show specific condensation, and similar arrangement and location at spermatogonia, as found in the X1X2Y system [5].
Despite the considerable diversity of SCS in pholcids, pairing of their sex chromosomes during male meiosis is conservative. Similar to other spiders [5], it is already established during premeiotic interphase [9, this study]. Pairing of chromosomes of the X1X2Y system is initiated by the parallel attachment of X chromosomes [this study]. The mode of chromosome pairing within the pholcid X1X2Y system is the same as in other haplogynes [9]. Biarmed chromosomes of this system pair without chiasmata, namely by the ends of both arms [9, this study]. This sex chromosome attachment is probably an ancestral mode of achiasmatic sex chromosome pairing in spiders [5]. It is retained in achiasmatic XY and multiple X chromosome systems of pholcids [this study]. Remarkably, pairing of the monoarmed X2 chromosome is modified in some pholcids. In Artema (X1X2Y) and smeringopines (X1X20), the monoarmed X2 shows standard pairing. However, attachment of the short arm is less stable in the latter group during late prophase and metaphase I. In some pholcines with monoarmed X2, only the long arm of the X2 takes part in pairing during this period [9, this study]. In contrast, there are pholcines with monoarmed X2, both ends of which participate in pairing. This pattern indicates that restriction of pairing to the long arm of X2 does not depend only on the morphology of this chromosome. Concerning the X0 system, arms of the sex chromosome are associated together in pholcids during late prophase and metaphase I [this study]. This pattern is also common in other spiders exhibiting the X0 system and biarmed sex chromosome [5, 9].
Our data suggest a possible pattern of sex chromosome pairing in the X1X2X3X4Y system of Kambiwa. As indicated above, two of its X chromosomes are probably biarmed and two other X are monoarmed. X chromosomes of Kambiwa probably pair by the ends of their arms during male meiosis. According to our model (Fig. 17), both arms of the biarmed X chromosomes are involved in pairing. One monoarmed X chromosome bears a NOR at the end, which is not involved in pairing. Since this NOR is placed in the short arm, only the long arm of this chromosome is involved in pairing. We hypothesize the same mode of pairing for the second monoarmed X chromosome. Our data indicate that the sex chromosome multivalent of Kambiwa contains two X chromosome pairs; each of them consists of one biarmed and one monoarmed chromosome. Pairing between these pairs is less stable than within them. We assume a similar location of the Y chromosome as in the achiasmatic sex chromosome systems of other haplogynes with more than two X [14], namely in the middle of the multivalent.
Meiotic segregation of the sex chromosomes is modified in pholcid males. Regardless of the type of SCS, it is usually delayed in anaphase I. The sex chromosome(s) of some pholcids show precocious or delayed division during metaphase II. In some pholcids, X chromosome segregation is also delayed during the second meiotic division [this study].
As in other spiders, the sex chromosomes of pholcids are usually placed at the periphery of the plate during premeiotic interphase, prophase and metaphase I. Notably, our data suggest relocation of sex chromosomes to the middle of the plate during late prophase I in pholcids with multiple X systems. The same behaviour of X chromosomes has evolved in males of avicularioid mygalomorphs [6]. Its function is unclear. After segregation during anaphase I, X chromosomes of pholcids retain their association until the end of meiosis, whereas the Y chromosome tends to be placed in the middle of the plate [this study].
As in other spiders, the course of sex chromosome condensation and pycnosis is complicated and species-specific in male meiosis of pholcids. The Y chromosome is often more condensed than the X chromosomes. In some pholcids, condensation of the X chromosomes is delayed during diplotene and diakinesis (e.g. Nipisa, Leptopholcus, most Pholcus species) [this study].
It should be underlined that chromosomes of the CSCP can also possess a specific behaviour in the male germline of spiders. In some mygalomorphs, they are associated and exhibit precocious chromatid separation in spermatogonial mitosis. Furthermore, they are heterochromatic in some male meiotic phases of some haplogynes and mygalomorphs [5, 6, 9]. Therefore, a large heterochromatic pair observed in male prophase I of several pholcids is probably a CSCP [this study].
Phylogenetic implications
Despite the relatively limited sample of studied species, our study emphasizes the potential of karyotype data as an independent source of information for phylogenetic reconstruction. Based on character mapping, many chromosome features were identified as apomorphies, which can be potentially used to reconstruct pholcid phylogeny. Most of these features concern the number of chromosome pairs, chromosome morphology, SCS, and NOR pattern. Character mapping also suggests, however, a high level of homoplasy and many characters that need to be mapped on terminal branches, especially those concerning chromosome morphology. In general, this suggests a limited use of certain karyotype data for the reconstruction of pholcid phylogeny. However, numerous clades established on the basis of morphological and/or molecular data are in fact supported by karyotype data:
-
At the level of subfamilies: e.g. Smeringopinae (2 characters) and Pholcinae (3 characters), sister relationship of Smeringopinae and Pholcinae (2 characters) (Figs. 16, 18).
-
At the level of genus-groups, e.g.:
-
Sister-group relationship between Quamtana and the Pholcus group of genera (three characters) (first proposed based on morphology [76], later supported by molecular data [20]) (Fig. 18).
-
Clade formed by all artemine genera except Artema (in our sample Chisosa, Wugigarra, Holocneminus, and Physocyclus) (1 character) (Fig. 15).
-
Separation of Stygopholcus and Hoplopholcus (Fig. 16) (only character 15 supports a close relationship between the two genera, as originally claimed e.g. by Brignoli [75,76,77]).
-
Sister group relationship between Cantikus (previously in Pholcus) and Micropholcus (three characters) (strongly supported by molecular data, [20]) (Fig. 18).
-
At the level of genera: e.g., Artema (three characters), Hoplopholcus (two characters) (Figs. 15, 16).
In some cases, the karyotype data suggest plausible alternative topologies that should be tested by molecular (ideally phylogenomic) approaches.
-
Position of Ninetinae. Previous hypotheses placed ninetines either as sister to all other pholcids [61, 78] or together with Artema as sister to all other pholcids [20]. Ancestral ninetines probably exhibited 13 chromosome pairs, as found in Pholcophora. The ancestor of another studied ninetine, Kambiwa (12 pairs), probably had the same NCPs as Pholcophora; one pair has most probably been incorporated into the SCS. Mapping of chromosome data on the molecular tree suggests two chromosome changes in the ancestors of Ninetinae (decrease of NCPs from 15 to 13 and increase of NOR number to two loci) and the same two changes in the common ancestor of Smeringopinae + Pholcinae (Figs. 15, 16). Our chromosome data thus suggest a sister-group relationship between Ninetinae and Smeringopinae + Pholcinae. Clearly, the position of ninetines continues to be unclear.
-
Phylogeny of artemines with low diploid numbers. The molecular phylogeny suggests the following topology: Chisosa (Physocyclus (Wugigarra + Holocneminus)) [21]. In this clade, the X0 system could either arise once, with subsequent reversion to the XY system (in Wugigarra), or it possibly evolved three times from the XY system, namely in Chisosa, Holocneminus, and Physocyclus (Fig. 15, Additional file 29: Fig. S23). The first hypothesis is improbable, since the X and Y chromosomes of Wugigarra exhibit the same mode of meiotic pairing as the chromosomes of the ancestral pholcid X1X2Y system. This mode of pairing would probably not be retained, if a reversion to the XY system had occurred. Although the second hypothesis (multiple convergent origin of the X0 system from the XY system) is supported by several independent origins of the X0 system in other pholcids, karyotype data lend some support to an alternative topology (Wugigarra (Chisosa (Holocneminus + Physocyclus))), which includes only a single origin of the X0 system from the XY system in the clade, namely at the base of the clade formed by the genera Chisosa, Holocneminus, and Physocyclus. This scenario contradicts a fairly strong node linking Wugigarra and Holocneminus (bootstrapping support 81) [21], but the alternative topology supported by karyotype data should be further tested too.
-
Position of Spermophora. This spider differs from most other pholcines by the absence of sex chromosome-linked NORs. According to the molecular phylogeny, Spermophora belongs to an early-diverging clade of pholcines, which also includes the genera Aetana and Belisana [21], both of which have sex chromosome-linked NORs (Fig. 18, Additional file 31: Fig. S25). In this phylogeny, the absence of sex chromosome-linked NORs in Spermophora reflects NOR loss. An alternative hypothesis, based on chromosomes, is that the absence of sex chromosome-linked NORs in Spermophora represents a symplesiomorphy of pholcids rather than a loss, suggesting a sister-group relationship between Spermophora and all other pholcines. This is a plausible scenario because the position of Spermophora (and also that of Aetana and Belisana) was not strongly resolved in the molecular phylogeny.
-
Phylogeny of the CHS clade (formed by Crossopriza, Holocnemus, and Stygopholcus). The molecular phylogeny suggests the topology H. hispanicus (Stygopholcus (H. caudatus + Crossopriza))) (Fig. 16). Karyotype data show a specific medium-sized submetacentric pair (feature 6-1) in H. caudatus and H. hispanicus, which evolved either once or twice. This uncertainty is related to the question of Holocnemus monophyly: H. caudatus and H. hispanicus might be sister taxa, sharing feature 6-1. Molecular data did not support such a relationship, but also did not strongly contradict it [20]. The two species are geographic neighbors and are morphologically similar, which lends further credibility to the karyotype data. The type species of Holocnemus, H. pluchei, is a morphologically isolated species and probably not closely related to H. caudatus and H. hispanicus (B.A. Huber, unpublished data). Chromosome data suggest it is probably sister taxon to the other taxa of the CHS clade (Fig. 16). Mapping of the chromosome information on a molecular cladogram suggests that eleven CPs is a synapomorphy of Stygopholcus, Holocnemus caudatus, H. hispanicus, and Crossopriza (Fig. 16). The molecular cladogram also suggests another synapomorphy for this clade, namely a sex chromosome-linked NOR. This structure is already present at the base of this clade in H. hispanicus, i.e. before the separation of Stygopholcus (Fig. 16). In contrast to H. hispanicus, however, Stygopholcus does not exhibit sex chrososome-linked NOR, indicating secondary loss of this marker. The chromosome data suggest a more plausible explanation of the pattern found in Stygopholcus. Similar to Spermophora, the absence of sex chromosome-linked NOR may represent a symplesiomorphy of pholcids rather than a loss, which suggests a sister-group relationship between Stygopholcus and a clade formed by H. caudatus, H. hispanicus, and Crossopriza.
-
Smeringopus phylogeny. Molecular data suggested the topology S. pallidus (S. cylindrogaster (S. atomarius (S. similis + S. peregrinus))) (Fig. 16) (accepting that S. peregrinus, which was not included in the analysis, and S. peregrinoides are closely related). Mapping of chromosome data on this cladogram suggests two independent origins of the specific shortest chromosome pair exhibiting an acrocentric morphology, namely in S. cylindrogaster and S. atomarius. This pair might arise by pericentric inversion from a short subtelocentric pair, which is a synapomorphy of S. similis and S. peregrinus. Chromosome data indicate that the shortest chromosome pair exhibiting an acrocentric morphology may be a synapomorphy of S. cylindrogaster and S. atomarius. This is in agreement with a morphological cladistic analysis [79], suggesting that the sister-group relationship between S. cylindrogaster and S. atomarius indicated by the karyotype data is a plausible alternative to the molecular hypothesis.
-
Position of Pholcus phalangioides. Mapping of chromosome information on the cladogram derived from molecular data suggests an independent decrease of NCPs from eleven to ten and a considerable decrease of X2 size in the two Pholcus clades, namely P. pagbilao and African Pholcus species (Fig. 18). Chromosomes suggest an alternative topology, namely a single origin of these features at the base of a clade formed by African and South Asian members of Pholcus. A closer relationship of African Pholcus with the Southeast Asian P. pagbilao than with P. phalangioides is plausible, as the latter was identified as a rougue taxon in [20], and is thus likely misplaced in Fig. 18.