Present and Future Salmonid Cytogenetics

Salmonids are extremely important economically and scientifically; therefore, dynamic developments in their research have occurred and will continue occurring in the future. At the same time, their complex phylogeny and taxonomy are challenging for traditional approaches in research. Here, we first provide discoveries regarding the hitherto completely unknown cytogenetic characteristics of the Anatolian endemic flathead trout, Salmo platycephalus, and summarize the presently known, albeit highly complicated, situation in the genus Salmo. Secondly, by outlining future directions of salmonid cytogenomics, we have produced a prototypical virtual karyotype of Salmo trutta, the closest relative of S. platycephalus. This production is now possible thanks to the high-quality genome assembled to the chromosome level in S. trutta via soft-masking, including a direct labelling of repetitive sequences along the chromosome sequence. Repetitive sequences were crucial for traditional fish cytogenetics and hence should also be utilized in fish cytogenomics. As such virtual karyotypes become increasingly available in the very near future, it is necessary to integrate both present and future approaches to maximize their respective benefits. Finally, we show how the presumably repetitive sequences in salmonids can change the understanding of the overall relationship between genome size and G+C content, creating another outstanding question in salmonid cytogenomics waiting to be resolved.


Introduction
The taxonomic species diversity of peri-Mediterranean and Near East brown trout is still not well understood, and new species are expected to be discovered and/or resurrected from the westernmost tip of the trout distribution area in Morocco [1,2], across the Iberian peninsula [3], Italy, the Balkans (reviewed in Kottelat and Freyhof [4]) and Greece [5]. Similarly, new trout species were recently described in the territory of Turkey in the Mediterranean Sea, Black Sea and Persian Gulf river drainages [6][7][8][9][10][11], presently encompassing 12 species. However, several authors have already recognized the taxonomic diversity of this region's brown trout ( [9,12,13] and references therein). Among these for Invitrogen FITC-Streptavidin) and superimposed using Adobe Photoshop software, version CS5. Karyotypes from Giemsa-stained chromosomes were arranged in Ikaros (Metasystems) software. In the case of CMA 3 /DAPI staining, the CMA 3 signal was inverted into the red channel while the DAPI signal went into the green channel to enhance the contrast between these two signal types. At least 25 metaphases (of the highest possible quality) per individual and method were analysed, some of them sequentially. Chromosomes were classified according to Levan et al. [38], but modified as m = metacentric, st = subtelocentric and a = acrocentric, where st and a chromosomes were scored as uni-armed to calculate the NF value (Nombre Fondamental, number of chromosome arms sensu Matthey [39].

Cytogenomic Analyses
First, we reviewed current genomic resources (Ensembl and NCBI/genome) for fish and especially salmonid genome assemblies. Second, we applied the novel Python tool EVANGELIST (= EVAluatioN on GEnome LIST) based on the non-overlapping sliding window to visualize and quantify percentage of repeats and GC% in both repeats and non-repetitive DNA simultaneously, introduced by Matoulek et al. [24]. With this tool, we produced the prototypical virtual karyotype for a salmonid (S. trutta). Third, we extracted and manually curated data on genome size and GC% from currently available fish genomes assembled to the chromosome level (the best genome quality) using the NCBI online tool https://www.ncbi.nlm.nih.gov/genome/browse#!/overview/. Finally, we processed these data with R [40] and compared them with our previous results [29].

Karyotypes and Molecular Cytogenetic Traits
The diploid chromosome number was 2n = 80 and the karyotype was composed of 7 pairs of metacentric, 5 pairs of subtelocentric, 2 pairs of distinctly large acrocentric and 26 pairs of moderate sized acrocentric chromosomes, decreasing gradually in size (Figures 1 and 2a, b). The NF value equalled 96 ( Figure 1). DAPI/CMA 3 fluorescence showed CMA 3 -positive, i.e., highly GC-enriched, signals in p arms of the largest subtelocentric chromosome pair (Figure 2c). FISH with the 28S rDNA probe clearly visualized signals in the same position as CMA 3 , while 5S rDNA sites were located in pericentromeric regions of one middle-sized metacentric chromosome pair (Figure 2e). FISH with the telomeric probe labelled the terminal regions of all chromosomes and did not reveal any interstitial signals ( Figure 2d). Finally, Ag-NOR impregnation marked the same CMA 3 and FISH positive region, i.e., the p arms of the largest sub-telocentric chromosome pairs ( Figure 2f).
Genes 2020, 11, x FOR PEER REVIEW  4 of 18 software, version CS5. Karyotypes from Giemsa-stained chromosomes were arranged in Ikaros (Metasystems) software. In the case of CMA3/DAPI staining, the CMA3 signal was inverted into the red channel while the DAPI signal went into the green channel to enhance the contrast between these two signal types. At least 25 metaphases (of the highest possible quality) per individual and method were analysed, some of them sequentially. Chromosomes were classified according to Levan et al. [38], but modified as m = metacentric, st = subtelocentric and a = acrocentric, where st and a chromosomes were scored as uni-armed to calculate the NF value (Nombre Fondamental, number of chromosome arms sensu Matthey [39].

Cytogenomic Analyses
First, we reviewed current genomic resources (Ensembl and NCBI/genome) for fish and especially salmonid genome assemblies. Second, we applied the novel Python tool EVANGELIST (= EVAluatioN on GEnome LIST) based on the non-overlapping sliding window to visualize and quantify percentage of repeats and GC% in both repeats and non-repetitive DNA simultaneously, introduced by Matoulek et al. [24]. With this tool, we produced the prototypical virtual karyotype for a salmonid (S. trutta). Third, we extracted and manually curated data on genome size and GC% from currently available fish genomes assembled to the chromosome level (the best genome quality) using the NCBI online tool https://www.ncbi.nlm.nih.gov/genome/browse#!/overview/. Finally, we processed these data with R [40] and compared them with our previous results [29].

Karyotypes and Molecular Cytogenetic Traits
The diploid chromosome number was 2n = 80 and the karyotype was composed of 7 pairs of metacentric, 5 pairs of subtelocentric, 2 pairs of distinctly large acrocentric and 26 pairs of moderate sized acrocentric chromosomes, decreasing gradually in size (Figures 1 and 2a, b). The NF value equalled 96 ( Figure 1). DAPI/CMA3 fluorescence showed CMA3-positive, i.e., highly GC-enriched, signals in p arms of the largest subtelocentric chromosome pair (Figure 2c). FISH with the 28S rDNA probe clearly visualized signals in the same position as CMA3, while 5S rDNA sites were located in pericentromeric regions of one middle-sized metacentric chromosome pair (Figure 2e). FISH with the telomeric probe labelled the terminal regions of all chromosomes and did not reveal any interstitial signals ( Figure 2d). Finally, Ag-NOR impregnation marked the same CMA3 and FISH positive region, i.e., the p arms of the largest sub-telocentric chromosome pairs ( Figure 2f).

Virtual Karyotype and Cytogenomics in Salmonids
Thanks to the close relatedness between the S. platycephalus analysed here and the species S. trutta [41], which is among the best cytogenetically analysed salmonid fishes [18], it is relevant to compare their karyotypes. Utilizing the EVANGELIST Python tool, we produced the first virtual karyotype for the latter species ( Figure 3). The virtual karyotype of S. trutta was confronted with its cytogenetics-based congeneric karyotypes. Both virtual karyotypes of the genus Salmo show a homogenization in GC% along chromosomes and repetitive as well as non-repetitive fractions reaching 50-60%. The comparison with actual karyotypes enabled an assignment of just the three largest chromosomes (Figure 3) at this stage and shows the need to improve the virtual analysis by including an option to visualise the two nuclear ribosomal fractions in the next step. Visualization of 5S rDNA can already be performed in Ensembl, showing that chromosome No. 1 bears the majority of 5S rDNA sequences ( Figure 3, labelled with a blue arrowhead). Moreover, one more 5S rDNA site was identified on chromosome No. 20 ( Figure 3, orange arrowhead). This site is probably below the detection range of FISH; however, it shows the potential of virtual karyotyping to visualize DNA sequences otherwise hidden for FISH.

Virtual Karyotype and Cytogenomics in Salmonids
Thanks to the close relatedness between the S. platycephalus analysed here and the species S. trutta [41], which is among the best cytogenetically analysed salmonid fishes [18], it is relevant to compare their karyotypes. Utilizing the EVANGELIST Python tool, we produced the first virtual karyotype for the latter species ( Figure 3). The virtual karyotype of S. trutta was confronted with its cytogenetics-based congeneric karyotypes. Both virtual karyotypes of the genus Salmo show a homogenization in GC% along chromosomes and repetitive as well as non-repetitive fractions reaching 50-60%. The comparison with actual karyotypes enabled an assignment of just the three largest chromosomes ( Figure 3) at this stage and shows the need to improve the virtual analysis by including an option to visualise the two nuclear ribosomal fractions in the next step. Visualization of 5S rDNA can already be performed in Ensembl, showing that chromosome No. 1 bears the majority of 5S rDNA sequences ( Figure 3, labelled with a blue arrowhead). Moreover, one more 5S rDNA site was identified on chromosome No. 20 ( Figure 3, orange arrowhead). This site is probably below the detection range of FISH; however, it shows the potential of virtual karyotyping to visualize DNA sequences otherwise hidden for FISH.  Virtual karyotype of Salmo trutta shows the haploid set of size-sorted chromosomes. The colour scale represents the proportion of repetitive (green) and non-repetitive (red) sequences. The y axis of each chromosome represents the scale of GC%. The karyotype based on cytogenetics in S. trutta enables us to roughly identify only the first three chromosomes according to their size-the largest acrocentric, the largest sub-telocentric and probably the largest metacentric chromosome. According to Ensembl, the 5S rDNA bearing chromosomes are chromosome No. 1 (the main site visualized also by FISH, blue arrow), i.e., the fourth largest chromosome, and chromosome No. 20 (orange arrow), which has a single 5S rDNA sequence.
Finally, we have taken advantage of the increasingly available data on genomic features (GC% and genome size) among teleost fishes with a special focus on salmonids. Including more species than in the previous analysis [29] confirmed earlier results that genome size negatively correlates with the genomic GC% in fish excluding salmonids. Moreover, the results revealed an inverse relationship between these two measures in salmonids in comparison with other teleosts (Figure 4).
Genes 2020, 11, x FOR PEER REVIEW 7 of 18 Finally, we have taken advantage of the increasingly available data on genomic features (GC% and genome size) among teleost fishes with a special focus on salmonids. Including more species than in the previous analysis [29] confirmed earlier results that genome size negatively correlates with the genomic GC% in fish excluding salmonids. Moreover, the results revealed an inverse relationship between these two measures in salmonids in comparison with other teleosts (Figure 4).

Discussion
The salmonid genus Salmo (Linnaeus, 1758) represents freshwater anadromous fishes that are originally widely distributed from the North Atlantic Basin, i.e., Northeastern North America and Europe (including European Arctic) to the upper parts of Amu-Darya R. in Central Asia [42]. The genus contains two sister lineages-a primarily anadromous Atlantic salmon, S. salar, and primarily freshwater resident fishes collectively known as brown trout (Salmo spp.), although both lineages include numerous anadromous and freshwater populations. However, these salmonids have been introduced and/or stocked outside their native range virtually around the world, mainly as objects of recreational fishery. As a result, several countries report an adverse ecological impact after their introduction [43][44][45][46][47]. Although brown trout are of limited interest in production aquaculture [48] (except the commercially important Atlantic salmon), brown trout have been and still are objects of intense investigations in various types of studies [12]. Similarly, chromosomes of different species and forms of brown trout were already extensively studied by numerous authors by the end of the 19th century (see review of Gas [49]). To compare results of our cytogenetic analysis of flathead trout, we summarized all available chromosome data in brown trout (Table 1). Summing up these studies, we excluded those without reliable locality data, without descriptions of cytogenetic methods used, without the number of individuals used and/or simply reports from other non-referenced sources. In some cases, we reinterpreted specific status of material examined under the name S. trutta (i.e., S. oxianus, S. cenerinus, S. farioides, S. lourosensis, S. peristericus) since the published locality data clearly pointed to species different from true S. trutta. We also included older data based on analyses of anaphase chromosomes from embryo squashes for species/forms not analysed afterwards (i.e., S. carpio, S. letnica, S. labrax, S. caspius) but with a sufficient chromosome quality to reliably infer 2n and karyotypes. On the other hand, we are aware that all of these summarized studies had significant flaws. First, none of these studies clearly claimed that examined fishes were deposited in any collection to enable later taxonomic identification of the material analysed [50]. Second, the results of some cytogenetic studies, especially in peri-Mediterranean and central European populations, could certainly have been affected by the stocking of non-autochthonous individuals and their subsequent genetic admixture (e.g., Kohout et al. [51], Leitwein et al. [52]). Nevertheless, our review clearly shows that cytogenetic and/or cytotaxonomic characteristics of flathead trout are nearly or even invariably the same as in other species/forms of brown trout. To explain this conclusion, we further examine in detail the data regarding 2n, karyotype composition and other chromosomal characteristics.

Discussion
The salmonid genus Salmo (Linnaeus, 1758) represents freshwater anadromous fishes that are originally widely distributed from the North Atlantic Basin, i.e., Northeastern North America and Europe (including European Arctic) to the upper parts of Amu-Darya R. in Central Asia [42]. The genus contains two sister lineages-a primarily anadromous Atlantic salmon, S. salar, and primarily freshwater resident fishes collectively known as brown trout (Salmo spp.), although both lineages include numerous anadromous and freshwater populations. However, these salmonids have been introduced and/or stocked outside their native range virtually around the world, mainly as objects of recreational fishery. As a result, several countries report an adverse ecological impact after their introduction [43][44][45][46][47]. Although brown trout are of limited interest in production aquaculture [48] (except the commercially important Atlantic salmon), brown trout have been and still are objects of intense investigations in various types of studies [12]. Similarly, chromosomes of different species and forms of brown trout were already extensively studied by numerous authors by the end of the 19th century (see review of Gas [49]). To compare results of our cytogenetic analysis of flathead trout, we summarized all available chromosome data in brown trout (Table 1). Summing up these studies, we excluded those without reliable locality data, without descriptions of cytogenetic methods used, without the number of individuals used and/or simply reports from other non-referenced sources. In some cases, we reinterpreted specific status of material examined under the name S. trutta (i.e., S. oxianus, S. cenerinus, S. farioides, S. lourosensis, S. peristericus) since the published locality data clearly pointed to species different from true S. trutta. We also included older data based on analyses of anaphase chromosomes from embryo squashes for species/forms not analysed afterwards (i.e., S. carpio, S. letnica, S. labrax, S. caspius) but with a sufficient chromosome quality to reliably infer 2n and karyotypes. On the other hand, we are aware that all of these summarized studies had significant flaws. First, none of these studies clearly claimed that examined fishes were deposited in any collection to enable later taxonomic identification of the material analysed [50]. Second, the results of some cytogenetic studies, especially in peri-Mediterranean and central European populations, could certainly have been affected by the stocking of non-autochthonous individuals and their subsequent genetic admixture (e.g., Kohout et al. [51], Leitwein et al. [52]). Nevertheless, our review clearly shows that cytogenetic and/or cytotaxonomic characteristics of flathead trout are nearly or even invariably the same as in other species/forms of brown trout. To explain this conclusion, we further examine in detail the data regarding 2n, karyotype composition and other chromosomal characteristics. Notes: 1: Giemsa-stained chromosomes only; 2: Robertsonian polymorphism detected; 3: Replication banding pattern discovered cytotype variants in some chromosomes; 4: Ag-, CMA 3 -and/or C-banding, cytotype polymorphisms; 5: Ag-, CMA 3 -and rDNA ISH and/or FISH; *-Material was analysed under the name S. trutta but evidently out of the known autochthonous range of S. trutta s. str., thus likely representing another species of the genus, the species name was determined based on locality data in a given study according to geographical distribution of trout taxa in Kottelat and Freyhof [4]; †-counts reported both trout taxa without distinguishing between them. Studies with incomplete information (without data reflecting karyotype composition, geographic origin, number of examined individuals or methodically problematic studies; all mostly reviewed in Gas [49]) were excluded from this review; symbol "-" in chromosome counts represents observed range, symbol "/"in chromosome counts shows that both categories (m and sm or st and a) were determined.
The karyotype of flathead trout undoubtedly belongs to category A sensu Phillips and Ráb [18], i.e., salmonids with 2n =~80 and chromosome arm number NF =~100. The 2n = 80 found in flathead trout has been reported in a majority of studies (e.g., [16,37,48]). Differences from this value are mostly caused by centric fusions of acrocentric chromosomes and/or fissions of metacentric chromosomes as reported in nearly all studies so far. Some reports of different 2n were caused by (i) lower quality of metaphases examined and/or (ii) low number of analysed individuals (e.g., Kaidanova [62], Karakousis et al. [56]). On the other hand, some studies pointed definitively to different 2n such as 2n = 82 in S. obtusirostris [60], 2n = 84 in S. trutta "alabalach" [72], 2n = 82 in some forms of S. ischchan [71] and 2n = 78 to 82 in some taxonomically unidentified Transcaucasian trout [68]. Regardless, such variation in 2n, frequently documented in other lineages of salmonids with A type karyotype, could be explained despite minor chromosome rearrangements such as pericentromeric inversions that can convert acrocentric chromosomes into sub-telocentric ones [18].
The 2n, karyotypes, and hence NF of the examined species/forms of brown trout are remarkably similar (Table 1). Nevertheless, differences caused by a chromosome classification bias among individual reports exist. Most authors categorise uni-armed and bi-armed chromosomes according to Levan [38] but NF was originally designed to quantify the centric translocations or fissions of the Robertsonian type [39] only. However, some authors scored sub-telocentric chromosomes as bi-armed. Differences in the NF reported for the same form/species thus usually resulted from a difference in the scoring rather than from any real variation. In other words, most of the studies provide the number of metacentric and submetacentric chromosomes together, while a minority of them distinguish these categories, as was done in our study. Another problem in comparing reports on karyotype structures in brown trout is that most of the summarized studies analysed Giemsa-stained chromosomes, published karyotypes and/or metaphase plates of lower quality to infer karyotype structure in more details. The studies using conventional and/or molecular cytogenetic protocols [55,[64][65][66][74][75][76][77][78][79][80] revealed very similar or even identical karyotypes as we found in the flathead trout for this study. We can therefore conclude that the karyotype of brown trout typically consist of seven pairs of metacentric, five to six pairs of visibly sub-telocentric chromosomes and all remaining are acrocentric elements of gradually decreasing size. The brown trout's karyotype also contains several distinct chromosome markers-the first two pairs of acrocentric chromosomes distinctly larger from other acrocentric ones and the largest sub-telocentric pair, which is also the largest one in the complement. The short (p) arm of this marker chromosome pair bears the major rDNA sites, as revealed by FISH with 28S rDNA probe, corresponding to positive Ag-and CMA 3 -stainings [55,[64][65][66][75][76][77][78]84]. Intraspecific variation in the locations and sizes of the chromosomal nucleolar organizer regions (NORs), i.e., major rDNA sites, have been frequently documented [85] but available data for this marker consistently document the same karyotype location across brown trout diversity including flathead trout. However, as in other cases, some intraspecific variability has been observed [64][65][66]. In our study, we observed the variability in the size of the NOR-bearing p arm of this marker chromosome corresponding to the 28S rDNA signal, similar to Caputo et al. [28] in S. marmoratus. The intraspecific variability of the 5S rDNA cytotaxonomic marker is quite well known [85][86][87]. Surprisingly, the location of 5S rDNA genes in the genus Salmo was examined in two studies only. Pendás et al. [84] found multichromosomal sites of these genes in brown trout from northwestern Spain, while Caputo et al. [55] observed these sites in telomeres of one middle-sized metacentric pair only. Our study also detected this gene cluster in the pericentromeric region of one middle-sized metacentric pair only. Whether this 5S rDNA bearing chromosome pair is homologous remains to be demonstrated by a cross-species painting protocol (e.g., Ráb et al. [88]). We can therefore conclude that the 2n, and structure as well as number and position of NORs, i.e., the active 28S rDNA sites, of the endemic flathead trout karyotype entirely correspond to those found in other brown trout taxa

Cytotaxonomy and Diversity of Eurasian Trouts
The species S. trutta has long been considered a single but highly polymorphic species broadly distributed in the European ichthyo-geographic region (see Bañarescu [42]) forming three ecotypes-marine migratory, lacustrine and brook/riverine [4]. In line with this, several subspecies or even distinct species have been described but most of them are simply considered as interindividual and/or interpopulation variability. Though even nominal subgenera of the genus Salmo have been described, i.e., Acantholingua (for A. ohridanus), Salmothymus (for S. obtusirostris) and Platysalmo (for P. platycephalus), collectively called´archaic trout´, they are closely related to the S. trutta species complex at the molecular level (e.g., Sušnik et al. [89], Phillips et al. [90]). However, recent detailed investigations of brown trout life histories, biology, distribution and taxonomy suggest that the biological and hence taxonomic diversity of the Eurasian genus Salmo is considerably greater than the taxonomy that was accepted up until the 1990s would suggest [4,91], a situation similar for freshwater trout of the genus Oncorhynchus [92]. Recently, FishBase [93] lists 50 formally described Salmo species. However, many molecular phylogeneticists and phylogeographers question this biological species concept of taxonomic diversity of the genus Salmo by pointing to negligible and/or weak genetic differentiation among some of those populations/taxa (to cite from numerous ones e.g., [13,89,[94][95][96][97][98][99]). Yet, other colleagues detected significantly larger genetic differences (e.g., [100][101][102][103][104]). How can the cytotaxonomy of the genus Salmo contribute to this debate? Our results of the cytogenetic analysis of flathead trout compared with available cytogenetic data for other trout populations and/or taxa ( Table 1) clearly demonstrate that 2n, karyotype structures and other chromosomal markers, especially the position of major rDNA sites, are rather stable or even invariable across trout diversity as described for several lineages of salmonid fishes with A type karyotypes [18,105]. At first glance, this conclusion would support/conform to the view of molecular-based studies. However, the stability of 2n and similar and/or even identical chromosomal characteristics were observed, i.e., karyotype stasis is widely documented in a group of taxonomically different species and/or even lineages. Such uniform stasis has been discovered in groups as diverse as plants [106][107][108], amphibians [109,110] and birds [111]. Among teleost fishes, multiple groups display such apparent karyotype stasis persisting in significantly long stages of lineage divergences, e.g., pikes of the genus Esox [112,113], fishes of the family Leuciscidae ( [88,[114][115][116][117] and references therein), Gobionidae [34], Xenocyprinidae [118] and especially many percomorph groups [119][120][121][122][123][124]. The underlying evolutionary mechanisms for this mode of karyotype (non) differentiation have not been identified so far but they may be at least partially linked with the functional arrangement of chromatin within the interphase nucleus and the degree of tolerance to its change [125,126]. We therefore conclude that, from the cytotaxonomic point of view, apparent karyotype stasis found in trout of the genus Salmo does not challenge their existing and evident taxonomic diversity.

Cytogenomics in Salmonids
The cytogenomic approach represents a logical continuation of the traditional molecular cytogenetics, which was crucial for understanding fish genome evolution. Cytogenomics effectively integrates the huge body of evidence generated by karyological and cytogenetic research with the genomic approach based on currently extensive genome sequencing [127]. The sequencing effort of fish genomes is still accelerating and highly ambitious; hence, with about 32,000 fish species [128], fish cytogenomics has a good chance of fast becoming as equally crucial as molecular cytogenetics despite the small fraction of genomes that had been sequenced so far in comparison with the number of species already analysed cytogenetically. Already, virtual karyotyping has taken another step forward with the potential to visualize more details with better resolutions than through the use of microscopes for most small-sized fish chromosomes.
At this initial stage, our tool for virtual karyotyping utilizes masking of repeats in the DNA sequence via soft-masking, i.e., identified repetitive sequences become lower-case, whereas the remaining sequences retain their upper-case. It means that the quality of the input assembly and its soft-masking is crucial and cannot be influenced by the tool itself. This tool has been introduced in this special issue to outline the potential future of fish cytogenomics, and so far its functionality has been utilized to address general questions on GC% and repeats evolution not only in fish but also across vertebrates. This means that the tool has not yet been used systematically in cytogenetically analysed fish species and results of both approaches have not yet been compared.
The inverse relationship between the GC% and genome size had been initially ascribed to the extremely dynamic and often highly amplified ribosomal genes [105,[129][130][131] that represent the GC-richest genome fraction [132,133]. However, further molecular cytogenetic results based on FISH with rDNA probes in further salmonids (continuously summarized by the database by Sochorová et al. [134]) as well as the results obtained here do not support these initial assumptions. The genomic approach is less useful here, because the rDNA is mostly disregarded and/or even discarded in the genome assemblies.