Intense genomic reorganization in the genus Oecomys (Rodentia, Sigmodontinae): comparison between DNA barcoding and mapping of repetitive elements in three species of the Brazilian Amazon

Abstract Oecomys Thomas, 1906 is one of the most diverse and widely distributed genera within the tribe Oryzomyini. At least sixteen species in this genus have been described to date, but it is believed this genus contains undescribed species. Morphological, molecular and cytogenetic study has revealed an uncertain taxonomic status for several Oecomys species, suggesting the presence of a complex of species. The present work had the goal of contributing to the genetic characterization of the genus Oecomys in the Brazilian Amazon. Thirty specimens were collected from four locations in the Brazilian Amazon and three nominal species recognized: Oecomys auyantepui (Tate, 1939), Oecomys bicolor (Tomes, 1860) and Oecomys rutilus (Anthony, 1921). COI sequence analysis grouped Oecomys auyantepui, Oecomys bicolor and Oecomys rutilus specimens into one, three and two clades, respectively, which is consistent with their geographic distribution. Cytogenetic data for Oecomys auyantepui revealed the sympatric occurrence of two different diploid numbers, 2n=64/NFa=110 and 2n=66/NFa=114, suggesting polymorphism while Oecomys bicolor exhibited 2n=80/NFa=142 and Oecomys rutilus 2n=54/NFa=90. The distribution of constitutive heterochromatin followed a species-specific pattern. Interspecific variation was evident in the chromosomal location and number of 18S rDNA loci. However, not all loci showed signs of activity. All three species displayed a similar pattern for 5S rDNA, with only one pair carrying this locus. Interstitial telomeric sites were found only in Oecomys auyantepui. The data presented in this work reinforce intra- and interspecific variations observed in the diploid number of Oecomys species and indicate that chromosomal rearrangements have led to the appearance of different diploid numbers and karyotypic formulas.


Introduction
The order Rodentia is divided into nine taxonomic families in Brazil. The family Cricetidae contains the most members, among which the subfamily Sigmodontinae includes 86 genera and 395 species (sensu Reig 1980) according to Prado and Percequillo (2013). Oryzomyini is the most diverse tribe of the Sigmondontinae, and the genus Oecomys Thomas, 1906 is one of the most diverse of the tribe Oryzomyini (Prado and Percequillo 2013). However, its morphological and karyological distinction and generic status were only recognized relatively recently (Andrades-Miranda et al. 2001, Carleton and Musser 1984, Gardner and Patton 1976, Reig 1984, 1986 as cited in Musser and Carleton 2005). Similarity among species and the limited understanding of morphological variations in Oecomys (including interspecific, intraspecific, geographic, and specimen age-inherent variations) have rendered species identification difficult.
Currently, 16 species are recognized within this genus Carleton 2005, Carleton et al. 2009), but only nine species have been studied for karyotypes, showing 11 different diploid numbers, varying between 54 and 86 chromosomes (Table 1). In Brazil 12 species were registered and 9 of which can be found in Amazon biome; O. auyantepui Tate, 1939, O. bicolor (Tomes, 1860, O. concolor (Wagner, 1845), O. paricola (Thomas, 1904), O. rex Thomas, 1910, O. roberti (Thomas, 1904, O. rutilus Anthony, 1921, O. superans Thomas, 1911 andO. trinitatis (J. A. Allen &Chapman, 1893) (Bonvicino et al. 2008;Flores 2010). Variations in fundamental number have also been reported in species with the same diploid number, which is an indicator of chromosomal rearrangements within the group (Rosa et al. 2012). However, morphological and morphometric analysis in conjunction with molecular and cytogenetic approaches revealed uncertainty in the delimitation and distribution of Oecomys species, suggesting the presence of a complex of species (Patton and Sherwood 1983, Emmons and Feer 1997, Patton et al. 2000, Musser and Carleton 2005, Carleton et al. 2009, Flores 2010, Rosa et al. 2012.
Hence, in the present study, we used classic and molecular cytogenetics approaches in order to enable the genetic characterization of three species of the genus Oecomys from the Brazilian Amazon. Further, we used DNA barcoding to evaluate the intraand interspecific distances, and infer the utility in species identification by combining this dataset with sequences deposited in GenBank.

Samples
Thirty specimens were collected from five locations in the Brazilian Amazon (Fig. 1 It must be noted that the collections took place outside of conservation units and that these species are not threatened with extinction. Samples were collected from the hematopoietic organ of each individual following euthanasia to obtain chromosome preparations and muscle tissue for DNA extraction.

Species
Voucher polymerase chain reaction (PCR) using the universal primers described by Ivanova et al. (2007). The PCR products were purified with the ExoSap® kit (GE Healthcare) and sequenced using the method described by Sanger et al. (1977) on an ABI 3130XL automatic sequencer. The resulting sequences were submitted to the NCBI database under the following accession numbers: KT258600-KT258632. Sequences were manually aligned using BioEdit v7.2.2 software (Hall 2001) and compared with sequences deposited in GenBank using BLASTn (Basic Local Alignment Search Tool). A Bayesian phylogenetic analysis was conducted with MrBayes 3.2 (Ronquist and Huelsenbeck 2003). For this analysis, Markov Chain Monte-Carlo sampling was conducted every 20,000 th generation until the standard deviation of split frequencies was <0.01. A burn-in period equal to 25% of the total generations was required to summarize the parameter values and trees. Parameter values were assessed based on 95% credibility levels to ensure that the analysis had run for a sufficient number of generations. A genetic distance matrix was constructed using the MEGA 6 program (Tamura et al. 2013) and was obtained according to the Kimura 2 parameter (K2p) model. For Bayesian analysis, 53 Oecomys COI sequences available in GenBank were included (Appendix 1). One specimen of Euryoryzomys macconnelli was used as an outgroup.
18S rDNA loci were visualized on chromosome pairs 10 and 14 of both karyomorphs, while the single 5S rDNA loci was located on pair 5 of karyomorph "a" and pair 7 of karyomorph "b" (Figs 2e, 3e). Both karyomorphs presented interstitial telomeric sequences (ITSs) in the centromeric region of the X chromosome (Figs 2f and 3f).

Mitochondrial DNA identification
A total of 86 Oecomys mitochondrial COI gene sequences were compared: 33 originating from the present work and 53 deposited in GenBank (Appendix 1). NJ, Bayesian and ML tree retrieved the same topology and showed differences mainly in relation to branch support values. The similarity index was greater than 98%, which allowed molecular identification of the species. The phylogenetic trees ( Figure 6) grouped O. rutilus into two clades, one comprising individuals from Brazil, Suriname and Guyana (I), while the other consisted of one individual from Ecuador (J). The genetic distance between clades I and J was 7.33%, whereas the genetic distance within clade I was 1.62%. The individuals of O. auyantepui were grouped into a single clade (H), comprising individuals from Brazil, Guyana and Suriname, with an intraspecific genetic distance of 1.41%. One individual from Ecuador, whose species was not defined in GenBank, belonged to a distinct lineage (clade E). Two other specimens without species level-definition were grouped with Oecomys concolor (branch F), with a genetic distance of 0.79%. One other individual (clade G), also identified as O. concolor in GenBank, exhibited a distinct lineage, showing a large genetic distance (12.88%) from branch F. All O. roberti specimens were grouped together (clade D), with a genetic distance of 0.39%. Oecomys bicolor formed three clades (A, B and C) with large genetic distances: individuals from the Guyanas and Suriname were grouped together with high support, forming a moderately supported clade (C) with an individual from the Central Amazon (INPA 6775). Oecomys bicolor and Oecomys sp. from Ecuador and the Negro river (INPA 6770) formed a group with moderate-to-high support (clade B). One individual from the Purus River (clade C) showed a highly supported association with clade B, with a genetic distance of 7.12%. The genetic distance between clades A and B was 8.4%, and that between A and C was 9.89%. Oecomys rex also formed two clades (K and L), with a large genetic distance between them (11.92%).

Discussion
The identification of Rodentia species is often difficult using morphological criteria alone (Granjon et al. 2002, Lecompte et al. 2005, Ben Faleh et al. 2010. Such difficulties are evident in this order mainly because of the existence of cryptic species (Granjon et al. 2002, Musser and Carleton 2005, Lecompte et al. 2005 and new species are continually described (Helgen 2005, Musser et al. 2005. Species identification via molecular methods, such as molecular barcoding using a short genetic marker (Hebert et al. 2003), has been proposed to overcome some of the weaknesses of the traditional approach, which will aid non-taxonomists by fulfilling the urgent requirement for rapid and accurate species identification tools (Teletchea 2010). This approach is potentially useful in the study of rodents (Borisenko et al. 2008, Tamrin and Abdullah 2011, Barbosa 2013. In the present work, employing COI sequences as a tool for species identification was shown to be satisfactory, as the obtained distance patterns provided sufficient information for the identification of specimens whose taxonomic identification at the species level is not straightforward. Most of the species were recovered as monophyletic groups. The available chromosomal data for Oecomys species consist mostly of descriptions of diploid and fundamental numbers, which restricts comparisons with the data obtained in the present work (Table 1). However, high karyotypic diversity can be observed, with countless chromosomal rearrangements between Oecomys species being responsible for this diversity. Neither of the two O. auyantepui karyomorphs reported in this work had been previously described in the literature. Both individuals (INPA 6751,INPA 6754) showing these two karyomorphs were captured on the same bank (right) of the Jatapú river, approximately 1 km from each other. The karyomorphs only exhibited one ITS, located on X chromosome. ITSs have been observed in other rodents as well (Castiglia et al. 2007, Rovatsos et al. 2011, Suárez-Villota et al. 2013). Short telomeric sequences (TTAGGG) n have been primarily classified as components of satellite DNA (Adegoke et al. 1993). These sequences may be located in subtelomeric and interstitial chromosome positions (Garrido-Ramos et al. 1998) and are subjected to amplification (Arnason et al. 1998, Castiglia et al. 2006. They may also appear during the double-stranded DNA nick repair process (Nergadze et al. 2004(Nergadze et al. , 2007. However, the most commonly accepted scenario is that ITSs signal recent chromosomal rearrangements, such as the transposition of functional telomeric sequences to an interstitial position (Dobigny et al. 2003, Zhdanova et al. 2005, or chromosome fusion events, with the latter being the main source of ITSs in many organisms (Lee et al. 1993, Slijepcevic 1998). Nevertheless, it was not possible to determine the occurrence of either an increase in the diploid number from 2n=64 as a result of a fission event or a decrease from 2n=66 due to a fusion event.
Establishing the evolutionary direction of chromosomal rearrangements is not always possible because most of the available painting data for the Sigmodontinae group are incomplete, and it is not possible to draw definitive conclusions regarding the composition of a putative Sigmodontinae ancestral karyotype (Romanenko et al. 2012). The same is true for Oecomys, where it cannot be determined whether the diploid number has increased or decreased because the in situ hybridization method used in this study likely does not detect very short (< 1 kb) stretches of (TTAGGG)n sequences. Thus, even if chromosome fusions that would result in a decrease in diploid number have occurred, the fused chromosomes will not always possess an ITS, which may have been lost prior to the fusion or been subjected to molecular erosion (Mandrioli et al. 1999).
Both Oecomys auyantepui karyomorphs exhibit similar, predominantly centromeric, subtle heterochromatic blocks. Their NORs are also similar, with three different labeled sites being observed on the same chromosome pairs. The largest chromosomes of both karyomorphs are homologous -those carrying 5S rDNA loci in particular -sharing the same chromosomal region (subtelocentric chromosomes), position (long arm, proximal) and number of labeled sites, as inferred based on the increased resolution provided by G-banding. Thus, much like the NOR-carrying pairs, these chromosomes were not involved in chromosomal alteration processes leading to the occurrence of two different diploid numbers in O. auyantepui. Mitochondrial DNA analysis grouped all O. auyantepui specimens onto a single branch (Fig. 6) with a high support value and low intraspecific genetic distance (1.41%), indicating that the occurrence of these two karyomorphs may be due to chromosomal polymorphism and not to the existence of two differentiated evolutionary units, as the intraspecific genetic distance is consistent with available data for other Sigmodontinae and the family Cricetidae in general (Smith and Patton 1993, Patton 1999,Ventura 2009).
Current phylogenetic and karyotypic data suggest the existence of a complex of O. bicolor species (Smith and Patton 1999, Flores 2010, Andrade and Bonvincino 2003. Four different diploid numbers have previously been characterized in the Brazilian Amazon, varying from 54 to 86 chromosomes, with 2n=80 being the most common (Gardner and Patton 1976, Patton et al. 2000, Andrades-Miranda et al. 2000, Andrades-Miranda et al. 2001, Lira 2012. Comparison of the karyotypic patterns of O. bicolor captured along the Jatapú and Purus rivers revealed a similar chromosomal organizational pattern for individuals with 2n=80 chromosomes. However, the karyotypic pattern of individuals collected on the banks of the Jari river diverges, with a diploid number 2n=82 and FN=116 (Lira 2012). The NORs described in the present work (7 labeled sites) occurred in larger numbers than what had been previously described for the species (1 to 4 labeled sited) (Andrades-Miranda 2001, Lira 2012). These NORs do not refer to the labeling of acidic heterochromatic regions, as fluorescent in situ hybridization using 18S rDNA probes revealed the existence of twelve chromosome pairs carrying these sequences. A larger number of sites compared with the number identified through silver nitrate staining, which is a common occurrence and is observed in other groups (Lira 2012). This disparity stems from the fact that the latter technique labels proteins associated with the nucleolar structure and not ribosomal DNA regions, thus identifying only NORs that had been active in the preceding interphase (Miller et al. 1976). Thus, the difference in silver-stained sites between different populations may stem from the activity of ribosomal RNA genes. Because rDNA sequence hybridization had not been performed in individuals from the analyzed populations in previous studies, this hypothesis cannot be verified. In contrast, the heterochromatin distribution pattern is similar, with centromeric blocks extending to the short arms of the majority of metacentric and submetacentric chromosomes and both sex chromosomes. The diploid number determined for O. rutilus (2n=54, first described in Lira (2012) did not vary, regardless of the collection site, and no variations in karyotypic structure were observed in the present work. However, the three specimens described previously (Lira, 2012) exhibited differences in their autosomal fundamental number (82, 84 and 86). Such variation may be related to karyotype interpretation, given that it depends on the quality of chromosome preparations, DNA compaction patterns, size and number of chromosomes and errors in the measurement of chromosomal arms. The C-banding pattern observed in O. rutilus consisted of very subtle labeling on the majority of chromosomes but was consistent with the expected locations previously described for other Oecomys species and the tribe Oryzomyini (Yonenaga-Yassuda et al. 1987, Svartman and Almeida 1992, Silva and Yonenaga-Yassuda 1998, Aniskin and Volobouev 1999, Volobouev and Aniskin 2000, Andrades-Miranda et al. 2002, Bonvicino et al. 2005, Lira 2012).
Based on the amplitude of the genus distribution, Langguth et al. (2005) suggested O. catherinae (2n=60) as the ancestral taxon; the same finding was reported by Weksler (2006), based on phylogenetic analysis of the IRBP gene and morphological data for O. bicolor, O. catherinae, O. concolor, O. mamorae and O. trinitatis. Although the current phylogenetic analysis based on COI sequences was limited to a single marker and did not consider several of the taxa analyzed by Flores (2010), it showed similar results with high support values, such as monophyly of the genus Oecomys, which was also observed in molecular studies using other markers Patton 1999, Weksler 2006). However, considering O. rex as a sister group of O. catherinae, which would classify both species as ancestral taxa (Flores 2010), the basal diploid number would be approximately 60/62 chromosomes. Therefore, it must be noted that molecular analyses did not detect an increasing or decreasing tendency in the diploid number between the branches, suggesting a complex karyotypic structure, as shown by the different diploid numbers obtained for the same morphological species. Moreover, the phylogenetic analysis placed all individuals in a single group.
In the present work, NORs were found to be preferentially located in the terminal regions of chromosomes, and their number increased with the diploid number; this pattern is also present in other members of the family Cricetidae (Lira 2012, Romanova et al. 2006, Ventura 2009, Fagundes et al. 1997. These data agree with FISH results obtained using the 18S ribosomal DNA probe, confirming the presence of two labeled pairs for O. auyantepui and O. rutilus. Labeling of four NORs was observed in Oecomys rutilus, whereas three were detected in O. auyantepui. Lira (2012) described four labeled sites in O. rex, again suggesting that it may constitute a basal taxon. In Oecomys bicolor, five chromosome pairs exhibited labeling, though not all displayed labeling on both homologous chromosomes. The multiple 18S rDNA sites observed in O. bicolor likely derive from duplication and dispersion. Di Meo et al. (1993) reported that the difference in the NOR distribution in correlated species is ascribed to rearrangements that have accumulated since the divergence of the common ancestor, mainly via inversions and Robertsonian translocations. Grozdanov et al. (2003) and Britton-Davidian et al. (2012) stated that NOR diversity among rodents is an indica-tor of high intrachromosomal transposition rates in the absence of visible rearrangements, suggesting, once again, that this character represents a derived state for this taxon. Despite this fact, the interstitial position of 5S rDNA is related to sequence protection, thereby avoiding possible crossing-over or transposition events, which are more frequent in terminal regions (Martins and Galetti Jr., 1999). This scenario is made evident by comparing the degree of conservation in the position and location of this sequence compared with 45S rRNA. Ventura et al. (2012) described a similar situation in Akodontini, which shows conservation of 5S rDNA chromosomal sites, despite large chromosomal variability within the group.
Oecomys species have undergone intense chromosomal alteration processes, as confirmed by the observed karyotypic patterns, indicating high local diversity and an ample distribution for the taxa under study. However, the limited taxonomic sample available, in terms of both Oecomys individuals and molecular data renders the determination of which evolutionary processes have led to the variability in karyotype morphology more difficult. Furthermore, the current data reinforce the necessity for integrative taxonomy, where genetic tools should be used in conjunction with morphological analysis to delimit Oecomys taxa.

Conclusions
The intra-and interspecific variations observed in the diploid number of Oecomys species indicate that chromosomal rearrangements such as fusions/fissions, translocations and duplications have led to the appearance of different diploid numbers and karyotypic formulas. However, telomere sequence hybridization was not found to be a good indicator of autosomal chromosome rearrangements in the Oecomys species under study, as no autosomal ITSs could be observed. O. bicolor, which is considered to be a derived taxon of the genus (Flores 2010), exhibits the highest diploid number, possibly arising from chromosomal fission events that occurred during its evolutionary history.