We performed a molecular and cytogenetic analysis on different Mantellinae species and revised the available chromosomal data on this group to provide an updated assessment of its karyological diversity and evolution. Using a fragment of the mitochondrial 16S rRNA, we performed a molecular taxonomic identification of the samples that were used for cytogenetic analyses. A comparative cytogenetic analysis, with Giemsa’s staining, Ag-NOR staining and sequential C-banding + Giemsa + CMA + DAPI was performed on eight species: Gephyromantis sp. Ca19, G. striatus (Vences, Glaw, Andreone, Jesu et Schimmenti, 2002), Mantidactylus (Chonomantis) sp. Ca11, M. (Brygoomantis) alutus (Peracca, 1893), M. (Hylobatrachus) cowanii (Boulenger, 1882), Spinomantis prope aglavei “North” (Methuen et Hewitt, 1913), S. phantasticus (Glaw et Vences, 1997) and S. sp. Ca3. Gephyromantis striatus, M. (Brygoomantis) alutus and Spinomantis prope aglavei “North” have a karyotype of 2n = 24 chromosomes while the other species show 2n = 26 chromosomes. Among the analysed species we detected differences in the number and position of telocentric elements, location of NOR loci (alternatively on the 6^th, 7^th or 10^th pair) and in the distribution of heterochromatin, which shows species-specific patterns. Merging our data with those previously available, we propose a karyotype of 2n = 26 with all biarmed elements and loci of NORs on the 6^th chromosome pair as the ancestral state in the whole family Mantellidae. From this putative ancestral condition, a reduction of chromosome number through similar tandem fusions (from 2n = 26 to 2n = 24) occurred independently in Mantidactylus Boulenger, 1895 (subgenus Brygoomantis Dubois, 1992), Spinomantis Dubois, 1992 and Gephyromantis Methuen, 1920. Similarly, a relocation of NORs, from the putative primitive configuration on the 6^th chromosome, occurred independently in Gephyromantis, Blommersia Dubois, 1992, Guibemantis Dubois, 1992, Mantella Boulenger, 1882 and Spinomantis. Chromosome inversions of primitive biarmed elements likely generated a variable number of telocentric elements in Mantella nigricans Guibé, 1978 and a different number of taxa of Gephyromantis (subgenera Duboimantis Glaw et Vences, 2006 and Laurentomantis Dubois, 1980) and Mantidactylus (subgenera Brygoomantis, Chonomantis Glaw et Vences, 1994, Hylobatrachus Laurent, 1943 and Ochthomantis Glaw et Vences, 1994).

Amphibia, chromosome evolution, karyotype, Madagascar, NORs

Madagascar is one of the richest biodiversity hotspots and an ideal region to study evolutionary dynamics (Myers et al. 2000; Ganzhorn et al. 2001; Vences et al. 2009). The native amphibians of Madagascar belong to four distinct anuran families: Hyperoliidae, Mantellidae, Microhylidae and Ptychadenidae (Glaw and Vences 2007). Among them, the family Mantellidae includes ca 230 described species (AmphibiaWeb 2021; Frost et al. 2021), representing the most species-rich amphibian group of the island.

Mantellidae are characterized by an extraordinary ecological and morphological diversity (Glaw and Vences 2007; Wollenberg et al. 2011; AmphibiaWeb 2021) and are subdivided into three subfamilies: Laliostominae with an overall low species diversity, including the genera Laliostoma Glaw, Vences et Böhme, 1998 (1 species) and Aglyptodactylus Boulenger, 1919 (6 species); Boophinae, a species-rich clade of about 80 described species of tree frogs, all belonging to the genus Boophis Tschudi, 1838, and Mantellinae, which is by far the most species-rich group including nine genera and more than 140 described species (Glaw and Vences 2007; AmphibiaWeb 2021).

The last three decades have seen the flourishing of the use of molecular techniques, with numerous taxonomic and systematic studies that clarified the relationships among the major groups within this subfamily (Glaw et al. 1998; Vences et al. 1998; Richards et al. 2000; Glaw and Vences 2006; Wollenberg et al. 2011; Kaffenberger et al. 2012). Similarly, these tools have been used in the identification of candidate species (Vieites et al. 2009; Perl et al. 2014) and have later contributed to the formal description of many of them (e.g. Andreone et al. 2003; Crottini et al. 2011a; Cocca et al. 2020; AmphibiaWeb 2021).

However, in contrast to the fast-growing amount of molecular data on Mantellidae, the available chromosomal data remain limited, leaving the karyological diversification of the family mostly unexplored. In particular, available cytogenetic data on the subfamily Mantellinae, obtained using different methods, come from a handful of studies (Morescalchi 1967; Blommers-Schlösser 1978; Pintak et al. 1998; Odierna et al. 2001; Andreone et al. 2003), together providing the description of the karyotype of ca. 40 species. These studies revealed the occurrence of a conserved chromosome number in most genera (2n = 26), but a marked difference in chromosome morphology, location of NORs and heterochromatin distribution (see Odierna et al. 2001 and Andreone et al. 2003). Differences in chromosome number (2n = 24) were also identified, with five species of the subgenus Brygoomantis all sharing this state, thus suggesting that the state of 2n = 24 is a derived feature of the group (Blommers-Schlösser 1978).

Comparative cytogenetics, especially when linked to phylogenetic inference, offers the possibility to identify plesio- and apomorphic states, and recognizes different evolutionary lineages (see e.g. Mezzasalma et al. 2013, 2014, 2017a). However, both the limited taxon sampling and the outdated taxonomy used in most previous works limited the possibility to draw robust comparisons and consistent hypotheses on the evolution of chromosomal diversification in the subfamily.

In this study we performed a comparative cytogenetic analysis on eight mantellid species belonging to the genera Gephyromantis Methuen, 1920, Mantidactylus Boulenger, 1895 (subgenera Chonomantis, Brygoomantis and Hylobatrachus) and Spinomantis Dubois, 1992, using a combination of standard coloration and banding methods. We coupled cytogenetic analyses with the molecular taxonomic identification of the samples and synthesized previously available information on this subfamily to produce an overview of their chromosomal diversity. This, enable us to propose a hypothesis on the chromosome diversification in mantelline frogs.

We here provide new karyological data on eight frog species belonging to the subfamily Mantellinae and discuss the available chromosome data on this subfamily to provide a first comprehensive assessment of its karyological diversity.

Available data on representatives of the other two Mantellidae subfamilies (Boophininae and Laliostominae) highlight the occurrence of a conserved karyotype structure in terms of chromosome number and morphology. In particular, the first karyological studies by Blommers-Schlössers (1978) on 12 species of Boophis (Boophininae) and on Aglyptodactylus madagascariensis (Duméril, 1853) (Laliostominae) revealed a conserved karyotype of 2n = 26 with all biarmed chromosomes.

Following studies by Aprea et al. (1998, 2004) expanded the knowledge on the karyological uniformity to the position of NORs loci, invariably on the sixth chromosome pair both in Boophis and A. madagascariensis, but evidenced different patterns of heterochromatin composition and distribution. Similar karyological characters were described also in different species of the genus Mantella (belonging to the subfamily Mantellinae), all showing a karyotype of 2n = 26 with all biarmed chromosomes (Blommers-Schlössers 1978; Odierna et al. 2001). A karyotype of 2n = 26 with all biarmed elements should thus be considered the primitive condition in the whole family Mantellidae, as it is highly conserved in all subfamilies, genera and most subgenera (see Blommers-Schlösser 1978; Aprea et al. 1998, 2004; Odierna et al. 2001, see Table 2). Nevertheless, species of other genera of the subfamily Mantellinae show a wider karyological variability, both concerning chromosomes number, morphology, localizations of NORs loci and heterochromatin composition and distribution (Blommers-Schlösser 1978; Odierna et al. 2001; present study) (see also Table 2).

Concerning the variability of the chromosome number, a 2n = 26 karyotype is still the most common chromosomal configuration, but karyotypes with a reduced chromosome complement (2n = 24) have been documented in 9 species of three different genera (6 species of Mantidactylus (subgenus Brygoomantis), 2 Spinomantis and Gephyromantis striatus) (See Fig. 1 and Table 2). Furthermore, while the 2n = 26 configuration occurs in all three subfamilies of the family Mantellidae (Mantellinae, Boophinae and Laliostominae) (e.g. Aprea et al. 1998, 2004; present study), karyotypes with 2n = 24 seem to occur in just a few phylogenetically lineages (genus Gephyromantis, Mantidactylus and Spinomantis), where the 2n = 26 configuration is also present (Blommers-Schlösser 1978; present study). In turn, the subfamily Boophinae, with all the species showing a 2n = 26 karyotype (Aprea et al. 1998, 2004), has been depicted as a basal group in the Mantellidae radiation (see e.g. Wollenberg et al. 2011). These evidences suggest that a reduction of the chromosome number from 2n = 26 to 2n = 24 occurred repeatedly and independently in different lineages of the subfamily Mantellinae, probably involving chromosome inversions and a fusion (translocation) between two elements of the smallest pairs (6–13), giving rise to an additional large (6^th) chromosome pair in several species (e.g. G. striatus, M. (Brygoomantis) alutus, and S. prope aglavei “North”) (Fig. 3; Table 3). Interestingly, a similar reduction of the chromosome number driven by tandem fusions (from 2n = 26 to 24) has been documented also in the family Ranidae (Miura et al. 1995).

Other than tandem fusions, chromosome inversions of primitive biarmed elements also had a significant role in the morphological chromosome diversity observed in mantelline frogs. These mechanisms generated a variable number of telocentric elements in different evolutionary lineages (see Figs 1, 3 and Table 2).

Figure 3.

Hypothesized general model of chromosome reduction in Mantellinae from n = 13 (2n = 26) to n = 12 (2n = 24) by means of chromosome fusions. Red dots highlight the NOR bearing chromosome.

Considering the position of the loci of NORs, our results and available literature data (Aprea et al. 1998, 2004; Odierna et al. 2001; Andreone et al. 2003), show that NORs occurrence on the sixth chromosome pair can be considered a primitive state, as it is described for all analysed species belonging to the genus Boophis, A. madascariensis and most Gephyromantis, Mantidactylus (subgenus Brygoomantis), and Spinomantis. On the other hand, a derivate configuration of NOR loci seems to have emerged multiple times in distinct lineages. The different positions of NOR loci in mantelline frogs suggest that these elements were also differently involved in the hypothesized chromosome fusions from 2n = 26 to 2n = 24, providing further support to multiple, independent rearrangements leading to similar karyotype configurations. In fact, while in M. (Brygoomantis) alutus the sixth large chromosome pair likely derived from a fusion involving the primitive NOR bearing pair and another smaller pair, in G. striatus and S. prope aglavei “North” the pair 6 does not include NOR loci, which are found on the 7^th and 10^th chromosome pair, respectively (see Fig. 1). In other species of Gephyromantis, Blommersia, Guibemantis and Mantella the relocation of NORs involved different pairs (1^st, 2^nd, 8^th, 9^th, 10^th or 11^th) (Odierna et al. 2001; Andreone et al. 2003; this study). It should be noted that Ag-NOR staining only evidences active NORs, and the existence of different inactive sites in the karyotypes of the studied species cannot be excluded based only on this analysis. However, we found correspondence in NOR location using both Ag-NOR and C-banding + CMA (in Figs 1, 2), which also has the power to uncover rDNA clusters (Schmid 1982; Zaleśna et al. 2017).

Various mechanisms may be responsible for NOR relocation, such as cryptic structural rearrangements, minute insertions, reintegration of rDNA genes amplified during ovogonial auxocytosis or the activation of silent sites (Nardi et al. 1977; Schmid 1978; King 1980; Mahony and Robinson 1986; Schmid and Guttenbach 1988; Mezzasalma et al. 2018). These mechanisms may be independent to other rearrangements, despite the resulting change in the configuration of NORs is a significant indicator of lineage divergence at different taxonomic level (e.g. Pardo et al. 2001; Mezzasalma et al. 2015, 2018, 2021).

Sequential C-banding did not evidence the occurrence of any sex-specific, largely heterochromatic chromosomes (generally related to differentiated heterogametic sex chromosomes, a condition not yet documented in the family Mantellidae), B chromosomes, or interchromosomal rearrangements leading to heteromorphic autosome pairs (e.g. Mezzasalma et al. 2014, 2016, 2017b; Sidhom et al. 2020). Nevertheless, C-banding showed a heterogeneous heterochromatin distribution in Mantellidae (see also Aprea et al. 1998, 2004; Odierna et al. 2001; Andreone et al. 2003), highlighting the occurrence of species-specific banding patterns. For example, G. striatus and M. (Hylobatrachus) cowanii show different amount and location of C-banding positive heterochromatin in comparison with closely related species with the same chromosome number and similar morphology (e.g. G. sp. Ca19 and M. (Brygoomantis) alutus). Interspecific variations in heterochromatin are generally due to different levels of amplification of highly repetitive DNA (Charlesworth et al. 1994). These differences mostly occurred without modifications of the chromosome morphology in Mantellidae (see also Aprea et al. 1998, 2004; Odierna et al. 2001; Andreone et al. 2003), probably by means of symmetrical addition/deletion of heterochromatin. The occurrence of distinctive species-specific banding patterns may be useful in evolutionary cytogenetic and cytotaxonomic studies in the subfamily, but comprehensive comparative analyses would benefit from more banding data on species of different genera and subgenera.

Finally, we also highlight the importance of a preliminary molecular taxonomic identification of mantellid frogs for a consistent karyotype attribution, and that future cytogenetic studies should focus on Laliostoma Glaw et al., 1998, Wakea Glaw et Vences, 2006, Boehmantis Glaw et Vences, 2006 and Tsingymantis Glaw et al., 2006, as well as on different undersampled genera and subgenera.

We provide new chromosomal data on eight species belonging to the subfamily Mantellinae, advancing the knowledge on their karyotype diversity, and suggesting that a reduction in the chromosome number and the relocation of NORs loci occurred repeatedly and independently in different genera of this subfamily. We hypothesize a karyotype of 2n = 26 with all biarmed elements and loci of NORs on the 6^th chromosome pair as the ancestral state in the whole family Mantellidae and propose a model for the reduction of the chromosome number from 2n = 26 to 2n = 24 by means of tandem fusions.

We are grateful to Malagasy authorities for granting research and export permits. Numerous colleagues helped us in the field: Gennaro Aprea, Frank Glaw, Miguel Vences. Portuguese National Funds through FCT (Foundation for Science and Technology) support the 2020.00823.CEECIND/CP1601/CT0003 contract to AC and the ICETA 2020-37 contract to MM. This work was supported by National Funds through FCT under the PTDC/BIA-EVL/31254/2017 project.

Marcello Mezzasalma https://orcid.org/0000-0002-7246-9831

Franco Andreone https://orcid.org/0000-0001-9809-5818

Fabio Maria Guarino https://orcid.org/0000-0002-1511-7792

Angelica Crottini https://orcid.org/0000-0002-8505-3050