The species included in the present study are listed in Table 1. Most of the individuals originated from aquarium trade from different breeders. Further, various collectors or ornamental-fish importers donated several samples for DNA analysis. Species were identified following Kullander and Nijssen (1989), Kullander and Prada-Pedreros (1993) and Staeck and Schindler (2004), and part of the analyzed fish was deposited in ICCU (Ichthyological Collection of Charles University, Prague). See Table 1 and Table 2.

Chromosomes were obtained by non-destructive isolation procedure from caudal fin regenerates as developed by Völker et al. (2006) and modified by Kalous et al. (2010). This method is based on regeneration of the caudal fin tissue after cutting a small part (2–3mm) from its margin. After five to seven days the regenerated tissue was cut and incubated in the solution with colchicine for two hours at room temperature. A few drops of fixative (methanol, acetic acid 3:1) were added to the tissue after this incubation and it was placed for 30min at 4°C. The tissue was washed twice in fixative, always staying for 30min at 4°C after the wash. Next, the tissue was placed into a drop of 20% acetic acid and gently mashed through a fine sieve. The suspension was dropped on a slide on a hot fig (45°C). After 20 seconds the drop of suspension was sucked up from the slide and dropped to a different place in the slide. Metaphase chromosomes were stained in 4% Giemsa solution in phosphate buffer (pH=7). Generally 5–50 metaphases per individual were evaluated. Chromosomes were classified according to Levan et al. (1964), to be consistent with most of the recent studies on cichlid fishes (Marescalchi 2004, Fedlberg et al. 2003, Poletto et al. 2010) and arranged to karyotypes by using ADOBE PHOTOSHOP, version CS7. The CDD fluorescent banding (Chromomycin A₃/methyl green/DAPI) was performed following Mayr et al. (1985) and Sola et al. (1992).

DNA was extracted from the ethanol-preserved samples by the commercially available kits (QiaGen), and four target genes (cyt b, 16S rRNA, S7 first intron, RAG1) were amplified by PCR using primers according to Musilová et al. (2009). Sequences of the PCR products were obtained by commercial sequence-service company (Macrogen, South Korea, Netherlands). Sequences were aligned in BIO EDIT (Hall 1999) software and genes were concatenated for the bayesian analysis in MRBAYES 3.2. (Ronquist et al. 2012). Analysis parameters were: number of generations = 10, 000, 000, number of chains = 4, number of runs = 2, model set for every gene separately (and unlinked) based on the jModeltest (Posada 2008) results. Three additional species (Bujurquina vittata, Aequidens metae and Laetacara thayeri) from the same taxonomic tribus Cichlasomatini as Nannacara + Ivanacara were analyzed as well, and one species of the different tribus Geophagini (Geophagus brasiliensis) was determined as an outgroup for the phylogenetic analysis. Sequences were uploaded to GenBank (Table 1).

Results are summarized in Fig. 1 and Table 3. Examined individuals of the species of genera Nannacara, Ivanacara and Cleithracara showed the diploid chromosome number 2n = 44 to 50 chromosomes. All three species of the genus Nannacara possessed 44 chromosomes and karyotype composed of 18 metacentric (m)-submetacentric (sm)+26 subtelocentric (st)-acrocentric (a) or 16m-sm+28st-a chromosomes, while Ivanacara adoketa had 2n = 48 and karyotype of 16m-sm+32st-a chromosomes, and Cleihtracara maronii had 2n = 50 composed of 14sm+36st-a chromosomes. Karyotypes of all studied species are shown in Fig. 1.

In the karyotypes of four studied species, namely Cleithracara maronii, Ivanacara adoketa, Nannacara anomala, and Nannacara taenia, the CMA₃-positive signals were found on one chromosome pair, although probably not homologous in different species. In Cleithracara maronii the CMA₃-positive signals were located on terminal parts of the largest m-sm chromosome pair, whereas in Ivanacara adoketa and Nannacara taenia the CMA₃ signals were located a chromosome pair from st-a group, terminal parts in Nannacara taenia and around the centromere in Ivanacara adoketa. In Nannacara anomala the CMA₃ signals were found on the terminal parts of a chromosome pair from m-sm group, but not on the largest pair. Contrarily, in the karyotype of Nannacara aureocephalus, the CMA₃ signals were located on three m-sm chromosome pairs including the largest chromosome pair in the centromeric region. See Table 3 for more detail about the karyotype formulas and CMA₃ phenotypes and Fig. 1 for representative metaphases and results of different staining steps.

Phylogenetic reconstruction based on the DNA sequences of up to four genes shows monophyly of the genus Nannacara (three species used in this study) and its sister relationship with the genus Ivanacara (one species present in our study). The monotypic genus Cleithracara (Cleithracara maronii) represents then basal lineage to the rest of Nannacara + Ivanacara (Fig. 2). The observed karyotype characteristics, i.e. the diploid chromosome number, the karyotype and the phenotype, were mapped on the phylogenetic tree and allowed reconstruction of the scenario of genome/karyotype evolution in the studied cichlids as well as to reconstruct as well as of the most likely hypothetical karyotype of an ancestor of the whole group. An ancestral karyotype of 2n = 48 was hypothesized as (16m-sm + 32 st-a) and was estimated as a basal stage for the clade by the most parsimonious reconstruction based on our material. The ancestor also had most likely only one pair of CMA₃ sites (Fig. 2).

Two of the five species presented within this study have been previously studied in Thompson (1979), Marescalchi (2004) and reviewed in Feldberg et al. (2003). The karyotype of Nannacara anomala corresponds in both the chromosomal number (2n=44) and the karyotype (18m-sm+26st-a) to the results of Thompson (1979). The karyotype of Cleithracara maronii corresponds with various previous studies in chromosomal number (2n = 50; Marescalchi 2004, see Feldberg et al. 2003), but slightly differs in the karyotype description: while in our study we recognized seven pairs of sub-metacentric chromosomes (14m-sm+36st-a), Marescalchi (2004) found only six pairs of those. However, inspecting the study of Marescalchi (2004), we found one additional pair of sub-metacentric chromosomes in their original karyotype data as well, so it is fully comparable with our results.

In the clade of Neotropical cichlids, three trends in karyotype differentiation can be distinguished (Feldberg et al. 2003). First trend - also called Karyotype “A” by Thompson (1979) – is characterized by maintaining the ancestral karyotype of 2n=48 with mostly subtelocentric-acrocentric elements (karyotype of 48st-a, although not exclusively) and evolved mostly by the pericentric inversions (during which the centromere is shifted from the central position of chromosome). Second evolutionary trend is similar to the previous one and additionally suppose the chromosomal breakage/fission events (Feldberg et al. 2003), leading to the increasing diploid chromosome number usually to the 2n=50 or 52, extremely up to 2n=60). This karyotype is dominated by uniarmed chromosomes. The third evolutionary trend - also called Karyotype “B” in Thompson (1979) – is represented by the opposite evolutionary scenario - mostly centric fusions played role in evolution from the ancestral karyotype, which lead to reduction of diploid chromosome number accompanied by increasing number of metacentric and submetacentric chromosomes (Thompson 1979, Poletto et al. 2010). This trend of chromosome number reduction seems to be parallel to some other fish groups like it was uncovered in killifishes (Cyprinodontiformes, Nothobranchiidae) Völker et al. (2008).

All of the species within the studied evolutionary lineage have a higher proportion of sub-metacentric chromosomes in their karyotypes compared with the rest of cichlids (Poletto et al. 2010). Especially considering the fact that the ancestral cichlid karyotype has been postulated as 2n=48 and 48st-a, i.e. no sub-metacentric chromosomes are present (Poletto et al. 2010), the whole Nannacara – Ivanacara – Cleithracara clade seems to have evolutionary derived karyotype within cichlids. Based on Thompson’s (1979) classification, the whole lineage possess the karyotype type “B” characterized by higher proportion of the sub-metacentric chromosomes, although not all the species have the lower number of chromosomes then the ancestral stage, which is usually characteristic for the karyotype “B” as well (Thompson 1979). Interestingly, the chromosome rearrangements and formation of karyotype “B” occurred several times independently in cichlid evolution, as from 41 examined Neotropical cichlids, the karyotype “B” has been found in three unrelated lineages: in the species Bujurquina vittata (Heckel, 1840) (tribe Cichlasomatini), in the genus Apistogramma Regan, 1913 (tribe Geophagini) and in the genus Symphysodon Heckel, 1840 (tribe Heroini; sister tribe of Cichlasomatini; Thompson 1979). Strikingly, the most similar karyotype formula possessed by all the species of the genera Apistogramma (22-24m-sm+16-22st-a) and Dicrossus Steindachner, 1875 (12m-sm+34st-a), which also represent another two unrelated lineage of the dwarf cichlids (Thompson 1979, Feldberg et al. 2003), and then a few other species like Cichlasoma paranaense Kullander, 1983 (14-20m-sm+28-34st-a), Mesonauta festivus/insignis (Heckel, 1840) (12m-sm+36st-a), Crenicichla niederleinii (Holmberg, 1891) (14m-sm+34st-a) and Astronotus ocellatus (Agassiz, 1831) and Astronotus crassipinnis (Heckel, 1840) (12-18m-sm+30-36st-a, Feldberg et al. 2003). Note, that although the karyotype composed of mostly subtelocentric-acrocentric chromosomes is considered as ancestral for the cichlids, it is not generally ancestral trait for other fish groups. Therefore, the emergence of karyotype “B” (with more sub-metacentric chromosomes) probably represents secondary change back to the “common teleost karyotype” (Thompson 1979, Arai 2011).

The CMA₃ signals represent usually the GC-rich DNA segments of heterochromatic regions, often correlated with the location of active or inactive NORs, usually represented by the rDNA regions in genome (Schmid and Guttenbach 1988, Ráb et al. 1999, Poletto et al. 2010, but see Fontana et al. 2001, Gromicho et al. 2005 or Saitoh and Laemmli 1994). The number of CMA₃ signals found within this study corresponds to what has been previously observed in cichlids – i.e. the most common number of NORs in Neotropical cichlids is one pair, but in some species were found up to three pairs (Feldberg et al. 2003, Poletto et al. 2010). In the Nannacara – Ivanacara – Cleithracara clade, all species except for Nannacara aureocephalus possess only one pair of CMA₃ signals in their karyotype. Nannacara aureocephalus has three pairs of CMA₃ signals, which is usually interpreted as the result of inversion followed by the multiplication of the rDNA regions (Poletto et al. 2010). Further, one of the observed CMA₃ regions in this species is located in the centromeric region.

After Feldberg et al. (2003), one pair of NORs on the larger pair of chromosomes represents the most common NOR phenotype for the whole family Cichlidae. Further, Hsu et al. (1975) suggested that species with the single pair of NORs should be considered as more primitive that the karyotype with several NOR pairs hinting that the ancestral karyotypes possess less NORs than the evolutionary derived. Multiplication of NORs is usually caused by the chromosomal rearrangements, such as translocation or inversion but recently an increasing number of studies has shown the cases of rDNA multiplication caused by the activity of transposable elements.(Cioffi et al. 2010, Symonová et al. 2013, Schneider et al. 2013). As summarized in Feldberg et al. (2003), five out of 15 analysed species of the subfamily Cichlasomatinae (tribes Heroini + Cichlasomatini) possess multiple NOR pairs, i.e. Caquetaia spectabilis (Steindachner, 1875) (Feldberg et al. 2003), Cichlasoma paranaense Kullander, 1983 (Feldberg et al. 2003), Mesonauata insignis and Mesonauata festivus (Heckel, 1840) (Feldberg et al. 2003) and Symphysodon aequifasciatus Pellegrin, 1904 (Feldberg et al. 2003).

The phylogenetic reconstruction of the Nannacara – Ivanacara – Cleithracara clade (also called NIC clade in Musilová et al. 2008, 2009) corresponds to the results observed in the previous studies (Musilová et al. 2008, 2009). This suggests the basal position of the monotypic genus Cleithracara followed by the Ivanacara (one species) sister to the rest of fishes from the genus Nannacara (three species). Within Nannacara, the Nannacara taenia has basal position and Nannacara anomala + Nannacara aureocephalus represent the sister species. In this study, we did not include two species of the studied clade, i.e. Nannacara quadrispinae and Ivanacara bimaculata, which we failed to obtain either as live individuals for cytogenetics, or as samples for DNA analysis. Especially Ivanacara bimaculata would be crucial for confirmation of monophyly of the genus Ivanacara, since Ivanacara bimaculata was previously found as closely related to the fishes of the genus Nannacara then to Ivanacara adoketa based on morphological data set (Musilová et al. 2009).

Within Nannacara aureocephalus, more distinct forms are known; some of them were introduced into the aquarium trade under different names. So far no robust revision of Nannacara is available, and it is therefore difficult to make any taxonomic conclusion based on our data set. However, at least two different forms of Nannacara aureocephalus are spread among the aquarium hobbyist within Central Europe (Germany, Poland, Czech Republic, Slovakia) – one of them called “blue” and the other one called “brown” both included in our analyses. These forms are not of artificial origin, as usually F1 progeny of the wild caught individuals has been studied. Intuitively, the blue morph shows more light-blue coloration with iridescent elements both on the face and body, while the “brown” form doesn’t have the iridescent coloration and possess darker brown to dark-green coloration pattern. We have shown that those two morphs are genetically distinct; however, more detailed future work is necessary on this species/genus.

Cichlid karyotypes show some general common features - for example many species from African and Neotropical cichlids possess one pair of significantly larger chromosomes. Although the homology of the largest chromosome within the African lineage has been proved (Ferreira et al. 2010) as well as high synteny conservation of African cichlid genomes (Mazzuchelli et al. 2012), it is, however, not yet clear to what extent is the homology present across the whole family Cichlidae (Valente et al. 2009).

Although all the studied species from the Nannacara – Ivanacara – Cleithracara clade are characterized by the karyotype “B” (Thompson 1979), they underwent different evolutionary paths in past. The phylogenetic reconstruction of the karyotype evolution suggests the following scenario: from the ancestral karyotype, first the karyotype of the Cleithracara maronii (2n = 50; 14mt-sm + 36 st-a) evolved by fission event of one sub-metacentric chromosome pair, falling apart into two additional pairs of subtelocentric-acrocentric chromosomes. While the karyotype of Ivanacara adoketa remained unchanged compared with the ancestral one, in the lineage of Nannacara, two fusions occurred decreasing chromosomal number to 2n = 44. These fusions were followed by pericentric inversions, which again decreased the number of sub-metacentric chromosomes. At least one pericentric inversion happened in the base of all Nannacara, and additional pericentric inversion happened in the Nannacara taenia lineage. Finally, two inversion impacting CMA₃ regions happened in Nannacara aureocephalus leading to the multiplicaiton of these signals.

The proposed mechanisms of chromosomal rearrangements are described in cichlids as well as in other fish species. Usually the sub-metacentric chromosome arises during the (centric) fusion, when two acrocentric-telocentric chromosomes fuse (Thompson 1979). However, the number of sub-metacentric chromosomes in karyotype is not evolutionarily stable. The sub-metacentric chromosome changes back to the acrocentric-subtelocentric chromosome by inversion, which involves the centromere, i.e. the pericentric inversion (Feldberg et al. 2003, Poletto et al. 2010). Further, those pericentric inversions are considered as the main mechanism generally contributing to changes in chromosome arms size in various percomorph lineages (Galetti et al. 2000, Affonso 2005). In general, the taxon sampling within such comparative studies is however still too low to be able to make a strong conclusion about the general trends in cichlid karyotype evolution (Feldberg et al. 2003, Poletto et al. 2010).

To conclude, we aimed to provide a comparative study on a small scale of three genera combining molecular and cytogenetic approaches. Assuming that cytogenetic data provide additional information, which is undetectable by molecular genetics (Ráb et al. 2007), we expected a broad insight into the genome evolution of the studied group. In the dwarf cichlid genus Nannacara and its relatives (Ivanacara and Cleithracara), we reconstructed the phylogeny and we found substantial amount of karyotype characteristics, which we were able to interpret in the evolutionary context.