The lineages to be tested are those proposed by Caraballo and Rossi (2017) (see above), plus a group of populations referred to as the Manantiales lineage, which formed a separate SSR cluster linked to C. roigi (Fernández et al. 2012) (Table 1). Caraballo and Rossi (2017) classified Manantiales as a member of C. perrensi based on partial chromosomal data and mtDNA affinities; however, SSR-based clustering contradicted this assignment. The sublineages previously identified as Iberá I and Iberá II were treated as a single lineage in the present study, as their differentiation was based solely on a single chromosomal rearrangement.

To gain more insight into karyotypic differentiation in the Corrientes group, we compared the 2n/FN formulas of all lineages and summarized the minimum number of structural rearrangements among the three recognized species (using the karyotype of their type localities) and Iberá (using the more widespread karyotype, 2n = 42, FN = 76), inferred from G-banding patterns (Ortells 1995).

A total of 42 mitochondrial haplotypes from Corrientes, consisting of partial cytochrome b (cyt b), cytochrome oxidase I (COI), and control region (CR) sequences obtained from Caraballo et al. (2012, 2016; File S1), were used to infer phylogenetic relationships and estimate divergence times (GenBank accession numbers JX275502–JX275655 and KT818638–KT818684). The alignment comprised a total of 2178 bp, including 693 bp from COI, 1079 bp from cyt b, and 406 bp from the control region. Genes were concatenated from single individuals (there are no chimeric sequences). A Bayesian phylogenetic analysis was performed using BEAST v2.7.7 (Bouckaert et al. 2019), allowing for different substitution rates for each locus, estimated with MrModeltest (Nylander 2004), all of them being HKY + I + G. Two haplotypes from the sister group Ctenomys pearsoni (from Médanos, Entre Ríos, Argentina) were included as outgroups. A Yule tree prior and a relaxed lognormal clock were used for 5 × 10⁷ Markov Chain Monte Carlo (MCMC) generations in two independent chains, sampling every 5 × 10³ generations, with a burn-in of 10%. Convergence of the MCMC runs was confirmed using Tracer v1.7.2 (Rambaut et al. 2018). Different clock rate priors for each marker were used based on the literature, with values of 0.02 and 0.0295 substitutions/site/million years for cyt b and CR, respectively (Caraballo and Rossi 2018; Carnovale et al. 2021). In the case of COI, as there are no reported rates for caviomorphs, we used the same value as cyt b, based on a broad mammalian comparative study, which found that average between-species COI divergence is often comparable to that of cyt b (Tobe et al. 2010). Trees were visualized in FigTree v1.4.4 (Rambaut 2014).

A Poisson Tree Processes (PTP) model was applied to infer putative species boundaries on a given phylogenetic input tree, using the software bPTP (Zhang et al. 2013). The topology obtained in the BEAST analysis was used as input, running the program for 1 x 10⁵MCMC generations, with a burn-in of 10%.

The nucleotide alignment used in the phylogenetic analysis was utilized to infer a haplotype network. A median-joining network was obtained using the software PopArt (Leigh and Bryant 2015), for each marker separately, and for the concatenated dataset.

To compare the genetic distances between members of the Corrientes group and the rest of the genus, we generated a major p-distance table with all available Ctenomys cyt b sequences (Table SS1). A total of 823 partial or complete cyt b sequences were downloaded from GenBank. This dataset was curated by updating species names under the current taxonomy and retaining only sequences from which we could corroborate species status. Short sequences (e.g., fossil sequences) that produced no overlap with other sequences and pseudogenes were removed. The final alignment comprises 361 nucleotide sequences belonging to 49 Ctenomys species, plus 46 sequences that correspond to the Corrientes group. Corrientes group sequences were grouped according to each putative species/lineage: Ctenomys perrensi, C. dorbignyi, C. roigi, Iberá, Santa Rosa, Sarandicito, and Manantiales.

Pairwise genetic distances were calculated with MEGA 11 (Tamura et al. 2021). We used the uncorrected p-distance since this is used in broad comparisons of Ctenomys (Parada et al. 2011), and this mitochondrial gene displayed a relatively low genetic variability. Sequences were grouped according to each species and three types of genetic distances were computed: intraspecific distances, interspecific distances, and distances among the Corrientes group as a third test category. Within Corrientes, only putative interspecific distances were considered, since the aim of this analysis was to test if these distances were comparable to intra- or interspecific distances in other species. A histogram displaying distance frequencies was generated with ggplot2 (Wickham 2016). The same procedure was applied using only sequences from the torquatus group, to which the Corrientes group lineages belong, to have a direct comparison with more closely related species, with a smaller bias towards high distance values found between less-related species in the Ctenomys phylogeny. To assess the presence of a barcode gap between intraspecific and interspecific genetic distances, we used the bgd function from the spider R package (Brown et al. 2012). This method evaluates the overlap between the distributions of pairwise distances within and among species to statistically test for the existence of a barcode gap.

The analysis of diploid (2n) and fundamental numbers (FN) of the different lineages of the Corrientes group shows at least four breaks that can be interpreted as complex rearrangements that involve changes in the FN (Fig. 3A). Lineages such as C. dorbignyi, C. perrensi, Manantiales, and Sarandicito can be grouped in an FN 84 group, probably differing only in Robertsonian translocations. In contrast, C. roigi appears as a more reorganized karyotype (FN 80), different from the FN 84 group, but also from the Iberá karyotypes (FN 76–78). Finally, Santa Rosa appears as a different karyomorph, probably more related to the FN 84 group, but differing in an arm addition or a pericentric inversion. To determine which of these changes actually occurred and to quantify them, chromosome banding is required to identify regions of intra- and interchromosomal homology. Based on G-banded karyotypes from the type localities of the three nominal species of the Corrientes group, as well as the more widespread karyotype of the Iberá lineage (2n = 42, FN = 76), we classified and quantified the types of rearrangements involved between each pair of karyotypes: Robertsonian (Rb) translocations, pericentric inversions, and chromosomal arms lacking defined homology or absent. These latter categories (arm acquisition and loss) are interpreted as arising from multiple microrearrangements that alter linear DNA sequences and disrupt chromosome homologies, thus indicating major divergence. However, it is important to note that all three pairwise comparisons involving C. dorbignyi showed a high proportion of non-homologous chromosomal arms, likely due to the small size of many of its chromosomes, which hampers G-band-based homology assessment. When this potential artifact is disregarded, the most frequent rearrangements are Rb translocations which alter the 2n, and inversions which modify the FN. In some cases, rearrangements are superimposed; for example, a pericentric inversion may occur in a chromosomal arm involved in an Rb translocation. Additionally, karyotypic formulas that suggested simple Rb changes were shown by G-banding to involve other important chromosome differences.

When comparing the G-banded karyotypes of the nominal species and the Iberá lineage (Fig. 3B, C), C. dorbignyi and Iberá are the most differentiated as expected given their extreme karyotypes, differing by 20 chromosomal arms without defined homology, 12 Rb translocations (one with monobrachial homology), and six pericentric inversions. The chromosome complement most similar to Iberá—based on the lowest number of inferred pairwise chromosomal rearrangements—is that of C. roigi. Nevertheless, these karyomorphs are still separated by at least one chromosomal arm without homology, four Rb translocations (two with monobrachial homologies), and one pericentric inversion involving an Rb chromosome. In contrast, the Iberá lineage and C. perrensi differ by four chromosomal arms without defined homology, four Rb translocations (two with monobrachial homologies), and one pericentric inversion affecting an Rb chromosome. Finally, the differences between the nominal species, C. perrensi and C. roigi, include three chromosomal arms without homology, four Rb translocations (two with monobrachial homologies), and one pericentric inversion.

The phylogenetic analysis corroborates the relationships previously reported for the Corrientes group (Caraballo et al. 2012; 2016) (Fig. 4). Briefly, C. roigi and C. dorbignyi are monophyletic, whereas C. perrensi comprises two distinct haplogroups, one closely related to Sarandicito, and the other exhibiting paraphyly relative to haplotypes from Santa Rosa and a new lineage proposed herein, based on SSR clustering called Manantiales (see Table 1). The Iberá populations are monophyletic except for one locality (Curuzú Laurel), while the Manantiales lineage is split into two different clades, one related to C. roigi and the other to Santa Rosa. Notably, Loma Alta contains haplotypes from both clades. The use of three mitochondrial markers provides a more precise age estimate for the Corrientes group, ranging between 330,000 and 530,000 years before present (median: 420,000 ybp).

The coalescent-based delimitation analysis using bPTP indicates that the posterior credible interval for the number of species in the dataset ranges from 13 to 33, with an average estimate across all MCMC samples of ~23 species (Fig. S1).

The networks obtained from each locus analyzed separately (File S2) are broadly congruent with the network based on the concatenated dataset (Fig. 5). As expected, the least resolved network corresponds to CR, not only because it is a non-coding region but also because it is the shortest fragment (406 bp) and, therefore, provides less information and lower resolving power. Because these loci are linked, their combined use maximizes the available information; accordingly, the most appropriate approach is to include all loci in a single network inference. The inferred mtDNA haplotype network is coherent with the Bayesian phylogenetic tree (Fig. 5). One remarkable aspect is that Sarandicito is distantly related to its sister group, one of the two clades of C. perrensi. It differs in 30 substitutions from the closest haplotype of C. perrensi, as well as from that of C. dorbignyi, and 32 from the closest haplotype of the Iberá populations (and 37 from the diverging haplotype found in Curuzú Laurel). In contrast, the two haplotypes found in Santa Rosa differ by only 3 and 4 substitutions from the closest haplotypes of the Manantiales lineage and C. perrensi, respectively. It is interesting to note that C. roigi is separated by 7–8 substitutions from the closest haplotypes of the Manantiales lineage. These haplotypes correspond to the localities Manantiales, Loma Alta, and Pago Alegre, and differ in at least 12 substitutions from the closest C. perrensi haplotype. The presence of two non-related haplotypes in the Manantiales lineage will be further discussed.

The network also confirms the presence of two mtDNA lineages within Iberá (Fig. 5), both of which are significantly differentiated from other lineages of the Corrientes group. One of these lineages includes all Iberá localities except Curuzú Laurel. This main group differs by at least 14 substitutions from any other lineage. The branch leading to the haplotypes found in Curuzú Laurel has 14 mutational steps, indicating that it is highly divergent from any other lineage.

The genetic distance values obtained when comparing individuals from different populations and lineages within the Corrientes group fall within a range that overlaps with both intra- and interspecific distances at the genus level (Fig. 6A). This is also confirmed when analyzing only sequences from the torquatus species group, to which the Corrientes group belongs (Fig. 6B). In both datasets, the barcode genetic gap test revealed that there are no significant differences between intra- and interspecific distances. Taken together, these results suggest that most of the variation in Corrientes probably pertains to the infraspecific level, but there might also be some degree of species-level differentiation.

The intralineage distances are quite low, indicating the genetic closeness of their members, ranging from 0 to 0.87% (Table S3). As expected, the highest values are found in lineages that do not form monophyletic groups in the phylogenetic tree (Fig. 4), such as C. perrensi, Iberá, and, to a lesser extent, Manantiales, which are composed of unrelated haplotypes. It is interesting to note that intralineage distances become more similar to each other when divided by the number of localities, suggesting a potential sampling bias influencing these values (Table S3).

The interlineage distances are more disparate (Table S4). The genetic distances between members of the Corrientes group and its sister group, C. pearsoni, are very similar, ranging between 2.0–2.2%. However, there are important differences between lineages within the Corrientes group. The most differentiated lineage is Sarandicito, with distances of 1.4–1.6% compared to any other lineage. In most of the remaining cases, distance values are around 1% except for Manantiales, which shows 0.5% divergence from both Santa Rosa and C. roigi.

Almost the entire sample of studied specimens shares the same skull anatomy, without clear variations in the character states among the craniodental traits defined by De Santi et al. (2020, 2021). We did not find major differences between our scoring of the skull characters and that of De Santi et al. (2021) for C. dorbignyi, C. perrensi, and the samples from Iberá. However, some minor but constant morphological differences were found between C. roigi and the other lineages, including more proodont upper incisors, shorter rostra, shorter premaxillae that end at the same level of the nasals, and larger paraoccipital processes (Fig. 7).

Those samples corresponding to C. dorbignyi, C. perrensi, and Iberá have medium-sized to large, robust skulls, not flattened in lateral profile; nasals with nearly straight and posteriorly convergent outer borders; broad and quadrate interorbital regions, with moderate to well-developed postorbital processes; the width across the auditory meatus slightly narrower than the zygomatic breadth; proportionally large and wide incisive foramina; medium-sized and pyriform auditory bullae; and mesopterygoid fossae extending to the level of M2 (Fig. 7).

Descriptive statistics are provided in Table S5. The first two axes of the PCA explained 84.5% of the total variance (PC1 = 78.8%, PC2 = 5.7%, Fig. 8; Table 2). The four lineages studied here exhibited a moderate to large overlap in both components. A minor separation can be detected along PC1, being males positioned towards positive values, indicating a tendency for a larger size. In turn, females were positioned toward negative values, depicting a smaller overall size.

The DFA analysis reproduced the results shown by the PCA, although depicting C. roigi as the most distinctive nominal form and a moderate to large degree of overlap among C. dorbignyi, C. perrensi, and the samples from Iberá (Fig. 8; Table 2). The distinction of C. roigi occurred both along the first axis (37.6% of the total variance) as well as the second (23.9%), while the other three lineages slightly differed along the first axis.

The MANOVA showed an overall significant intergroup variation (λ = 0.0076, df = 119, 336, P < 0.001). However, posterior pairwise comparisons using Bonferroni-corrected P values showed that all groups do not differ significantly from each other. This pattern is consistent with limited morphological differentiation among recently diverged lineages and is expected for taxa recognized at the subspecific level.

Unlike many other species groups of Ctenomys, the Corrientes group is characterized by remarkable chromosomal diversity, with many distinct karyotypes comparable to the total number of sampled localities. This pattern suggests that chromosomal evolution is an ongoing process in this group and that the rate of fixation is rapid, as—except for some Robertsonian rearrangements—most localities are monomorphic. The lack of allozymic (also corroborated with genetic studies, including this study) and morphological differentiation in the presence of the extensive karyotypic diversification found in these lineages made previous authors claim the Corrientes group to be a case of chromosomal speciation (Ortells et al. 1990; Ortells and Barrantes 1994; Ortells 1995). However, the type and number of chromosomal changes required to cause reproductive isolation remain a matter of debate, both in Ctenomys and across mammals and other taxa (Rieseberg 2001).

Chromosomal differences have been historically claimed to be barriers to gene flow, given that heterozygotes for extensively rearranged chromosome complements would produce unbalanced gametes, leading to varying degrees of hybrid sterility, or promote divergence between heteromorphic homologues via recombination suppression, which may result in negative heterotic effects (King 1995; Dobigny et al. 2005; Faria and Navarro 2010). However, some rearrangements, such as Robertsonian (Rb) translocations, which alter the 2n but not the FN, do not necessarily constitute reproductive barriers, particularly if they are in low numbers (Lanzone et al. 2002; 2007; Dobigny et al. 2005). This previously led to a first delimitation of the Corrientes group, whereby populations characterized by different FN were considered members of independent lineages (Caraballo and Rossi 2017). This is because FN-altering rearrangements, such as inversions and tandem fusions, can produce unbalanced gametes in hybrids (King 1995; Dobigny et al. 2017); another possible source of FN changes is whole-arm heterochromatin additions and deletions, but these types of modifications were detected in low frequency within the group (Argüelles et al. 2001). On the contrary, individuals with the same FN were considered as potentially interfertile (Caraballo and Rossi 2017). However, heterozygotes for multiple Rb and, to a lesser extent, tandem fusions in natural rodent populations (Kartavtseva et al. 2021), along with viable hybrids between divergent populations (Maputla et al. 2011), show that extensively rearranged karyotypes do not necessarily act as strong barriers to gene flow. Thus, although it may be indicative of evolutionary independence, there may be some degree of gene flow between karyotypically divergent populations.

The analysis of the karyotypic diversity of the Corrientes group suggests at least four (possibly five) chromosomal lineages that differ in several chromosome pairs. The most widespread FN = 84, which includes C. dorbignyi and C. perrensi, as well as Sarandicito and Manantiales, comprises a complex of Rb forms. At first glance, it may appear confusing that in the G-banding analysis of C. perrensi and C. dorbignyi, the total number of chromosomal changes was higher than that of C. perrensi and C. roigi or Iberá (Fig. 2), which have more rearranged karyotypes as revealed by the inspection of 2n, chromosomal morphology, and FN, as well as by DAPI-banding (this study; Caraballo et al. 2015; Buschiazzo et al. 2018). However, all comparisons involving C. dorbignyi include a large number of arms where homology could not be determined. Additionally, it is noteworthy that C. dorbignyi and C. perrensi have no computed pericentric inversions, nor chromosomal arms additions, and instead the identified rearrangements would be exclusively Rb fusions/fissions (again, with lower reproductive isolation outcomes). The remaining differences correspond to a lack of homology between chromosomes, which is in opposition to the DAPI-banding comparison, and could be attributable to artifactual causes that prevent accurate matching of bands, especially considering the small sized chromosomes of C. dorbignyi. Another interesting aspect is the relative similarity between the karyotypes of C. roigi and the most frequent form found in Iberá (2n = 42, FN = 76) as revealed by conventional staining, G-bands, and DAPI-bands. These are the karyotypes with the lowest FN of the Corrientes group, and although a nearly complete arm homology could be established between them, there are two pericentric inversions, one of which took place in an Rb chromosome, which would represent a strong reproductive barrier. In fact, the inspection of DAPI-banding patterns requires that more rearrangements should have taken place between these two karyomorphs to explain their chromosomal differences (Buschiazzo et al. 2018). The latter chromosomal lineage would be that of Santa Rosa, which depicts high 2n and FN = 86.

Monophyletic groups are the cornerstone of genetic approaches for species identification. However, mitochondrial DNA can sometimes lead to inaccurate phylogenetic reconstruction due to several limitations. Since mtDNA is maternally inherited and does not undergo recombination, it does not provide a complete picture of a population’s genetic history (Ballard and Whitlock 2004). Moreover, mtDNA is susceptible to selective sweeps and strong genetic drift, which can reduce genetic diversity and mask true evolutionary connections (Christie and Beekman 2017). Additionally, introgression can introduce foreign mtDNA into a population, further complicating phylogenetic analyses (Bágeľová Poláková et al. 2021). The rapid evolution of mtDNA compared to nuclear DNA can also result in homoplasies—similar traits arising independently rather than from a common ancestor—leading to incorrect assumptions about species relatedness (Galtier et al. 2006). These factors can distort phylogenetic trees and mislead our understanding of evolutionary histories. In this sense, the species delimitation inference performed in this study is based on a single mitochondrial DNA phylogeny; therefore, it reflects only maternal lineages and may be influenced by the distorting factors mentioned above. To avoid potential oversplitting or undersplitting, these results should be evaluated in combination with additional sources of data, such as nuclear markers, morphology, ecology, and geographic distribution, to provide a more robust framework for species delimitation.

On the other hand, microsatellites have been widely used to estimate gene flow or isolation between populations (Mirol et al. 2010; Fernández et al. 2012; Kim and Sappington 2013). However, incorrect cluster inference of panmictic populations using SSR markers can be caused by several factors (Putman and Carbone 2014). Insufficient marker polymorphism and a small number of markers might fail to capture genetic diversity adequately. A small sample size may not represent the population’s genetic diversity accurately, leading to biased allele frequency estimates and incorrect clustering. Hidden population structure or substructure can also interfere with the assumption of random mating in panmictic populations. Additionally, genotyping errors, such as allele dropout or stutter bands, can introduce inaccuracies in the data, resulting in erroneous cluster assignments.

Below, we present the delineation of the different independently evolving lineages found in the Corrientes group. As we will discuss, there is some level of discordance among lines of evidence, specifically between mtDNA phylogeny and SSR clusters. However, these incongruences are expected in groups with short divergence times, such as the Corrientes group, where processes such as hybridization followed by introgression or incomplete lineage sorting are more prone to occur (Avise 1994).

Ctenomys roigi is likely the most distinct lineage within the Corrientes group. This species is characterized by a unique karyotype (2n = 48, FN = 80; Caraballo et al. 2015), forms an exclusive SSR cluster (Mirol et al. 2010; Fernández et al. 2012), and its mtDNA variants are monophyletic (Caraballo and Rossi 2017; this study), which confirms its genetic distinctiveness and independent evolution.

Ctenomys perrensi consists of populations with varying diploid numbers (50–62), but a shared FN of 84. These populations are grouped into two or three SSR clusters (Mirol et al. 2010; Fernández et al. 2012) and two mtDNA groups. Interestingly, the SSR clusters and mtDNA phylogeny show different patterns. At the mtDNA level, the northern and southern localities fall into two unrelated groups (Fig. 4), but the SSR clusters interweave these localities (Fig. 2; Table 1). This suggests that the populations may form a continuum of chromosomal forms, with the extremes possibly connected by intermediate karyotypes, resembling the pattern observed in ring species (Coyne and Orr 2004). This is congruent with the high frequency of Rb heterozygotes in some of these populations (Giménez et al. 2002; Lanzone et al. 2002; 2007; Caraballo et al. 2015; Buschiazzo et al. 2018). Its inclusion in two distant clades suggests deep coalescence of the mitochondrial variants found in these populations.

The populations assigned to C. dorbignyi, corresponding to the originally described northern nucleus of this species (Contreras and Contreras 1984), contain an exclusive haplotype and form their own SSR cluster, confirming its independence from all other lineages. Its karyotype (2n = 70, FN = 84) is suggested to be the ancestral form of the Corrientes group, a characteristic also shared with C. pearsoni and the population from Sarandicito (Caraballo and Rossi 2017; Buschiazzo et al. 2022). Interestingly, a recent study employing a high-throughput genomic approach that includes C. dorbignyi and a topotype of C. perrensi reveals that their genomic distances fall within the range of intraspecific variation (Tomasco et al. 2024), thus corroborating the low level of divergence between these nominal taxa.

The individuals from Sarandicito share a similar karyotype with C. dorbignyi, yet they exhibit a distinct haplotype and possess a unique SSR cluster. Sarandicito stands out as the most divergent lineage based on mtDNA distances (Table S4), a pattern also evident in the haplotype network (Fig. 5). Its interlineage distances are closer to interspecific values than to any other comparisons in the Corrientes group. This combination of distinctiveness and substantial differentiation establishes Sarandicito as a well-delineated independent lineage, with some degree of affinity to C. perrensi, as revealed by the hierarchical SSR clustering and mtDNA phylogeny. Further studies using a genomic approach at the assemblage level will help to identify additional chromosome modifications which remain undetectable with classical cytogenetic banding techniques.

The Iberá lineage represents a distinct, independent evolutionary path within the Corrientes group. Its most notable characteristic is a quite derived karyotype with the lowest 2n and FN values observed in Corrientes (Fig. 3). Similar to C. perrensi, karyomorphs within the Iberá lineage exhibit slight differences in 2n, attributed to Rb translocations (Caraballo et al. 2015; Buschiazzo et al. 2018), the most common rearrangement among rodents (Patton and Sherwood 1983; Dobigny et al. 2005). The chromosome modification accounting for FN differences among Iberá lineage localities is proposed to be a tandem fusion involving the smallest chromosome pair (Caraballo et al. 2015). Although this type of rearrangement potentially promotes reproductive isolation, the chromosome abnormalities more frequently tolerated, producing viable individuals, are those involving the smallest elements (Labaroni et al. 2023). This is more in line with the evidence obtained from SSR and mtDNA that indicates that these populations are interconnected, functioning as a single evolutionary entity. Thus, there is no conclusive evidence to subdivide the Iberá lineage into two sublineages. Nevertheless, some discordances warrant biological interpretation. All Iberá lineage populations are monophyletic except for one locality: Curuzú Laurel. Remarkably, this population shares the same karyotype as the majority of Iberá populations (2n = 42, FN = 76) and belongs to exclusive Iberá SSR clusters (Mirol et al. 2010; Fernández et al. 2012). In the mtDNA-based phylogenetic analysis, Curuzú Laurel forms part of a primary polytomy encompassing all northern Correntinean populations. This pattern could arise from hybridization with a distantly related lineage or retention of an ancestral variant (deep coalescence). If hybridization led to the origin of this haplotype, there should be additional data supporting Curuzú Laurel’s clustering with another lineage of the Corrientes group. Our study employs a network approach, revealing that Curuzú Laurel’s haplotype corresponds to a distinct mitochondrial lineage, not closely related to any other lineage, likely representing an ancient haplotype potentially fixed through genetic drift.

The lineage of Manantiales also possesses an exclusive SSR cluster, which is hierarchically clustered with C. roigi (Table 1), indicating its independence from C. perrensi, previously assigned by Caraballo and Rossi (2017). Further, mtDNA network analysis partially supports the connection with C. roigi, where three of its localities—Loma Alta, Pago Alegre, and Manantiales—share haplotypes more closely related to this species than to any other lineage (Fig. 5). However, another haplotype in Loma Alta is noteworthy. It is shared with Mburucuyá, unrelated to Manantiales (locality), closely related to Santa Rosa, and secondarily to C. perrensi (Figs 4, 5). The presence of these two unrelated haplotypes in Loma Alta was confirmed by analyzing an additional 10 individuals, resulting in a 1:1 ratio, demonstrating this polymorphism. This haplotype was probably introduced through introgression from Santa Rosa or the nearby Saladas locality of C. perrensi (Figs 1, 5). Although karyotypic data for the Manantiales lineage are incomplete, available evidence suggests a differentiated chromosomal complement with an ancestral FN = 84 (Table 1).

The final independent lineage identified within the Corrientes group is Santa Rosa, which presents several inconsistencies. It has a unique FN of 86 and a polymorphic derived karyotype with a high chromosome number (2n = 65–66) (Giménez et al. 2002; Buschiazzo et al. 2018). At the mtDNA level, interlineage distances are comparable to other comparisons within the Corrientes group, but genetic distances with Loma Alta and Mburucuyá are notably low. Microsatellite analyses are conflicting: in one study, Santa Rosa clustered with members of C. perrensi (Mirol et al. 2010), while in another, it grouped with San Miguel, a member of the Iberá lineage (Fernández et al. 2012). This raises the question of whether Santa Rosa serves as a point of contact between these two lineages. Regardless, it cannot be unambiguously assigned to any previously identified lineage, suggesting it may represent a distinct, independent lineage.

The Corrientes group traces its origins to a period between 530,000 and 330,000 years ago, after diverging from its sister group, C. pearsoni, and expanding its range from south to north (Caraballo and Rossi 2018; our Supplementary Online Material SOM1). The ancestor of the Corrientes group exhibited a diploid number of 70 chromosomes and a fundamental number of 84, characteristics shared with certain forms found in both the Corrientes group and C. pearsoni (Buschiazzo et al. 2022; Caraballo 2023). During the diversification of the Corrientes group, two primary mtDNA lineages emerged. The southern lineage encompasses the austral populations of C. perrensi and Sarandicito, all of which retained the ancestral FN of 84. Sarandicito also retained the ancestral 2n = 70 and was isolated from the rest of the Corrientes group, with no subsequent gene flow. In contrast, the northern lineage underwent extensive chromosomal diversification. Although some populations maintained the ancestral FN of 84—with C. dorbignyi retaining the relictual 2n = 70—certain populations of the southern and northern mtDNA lineages, sharing karyotypic compatibility, have remained genetically connected, constituting C. perrensi. However, other populations remained isolated, giving rise to chromosomally divergent lineages, likely facilitated by habitat instability.

During the Pliocene, the Paraná River flowed into the Uruguay River, situated to the east of the Corrientes Province. More recently, during the Late Pleistocene and the Holocene—particularly between 18,000 and 6000 years ago—the Paraná River occupied multiple overlapping channels that initially flowed from northeast to southwest across Corrientes province, before shifting westward and then northward toward the Paraguay River basin (Popolizio 1977). This movement was followed by a dry period, succeeded by a humid one. Due to the gentle slope to the west of the Ituzaingó–La Paz fault, excess water from this last humid period accumulated in a large depression, giving rise to several shallow lakes, marshes, channels, and small rivers, ultimately forming the Iberá wetland around 3000 years ago (Iriondo 1991). Because the Correntinean populations of Ctenomys broadly predate these events, the current distribution likely reflects remnants of past extinctions and/or displacements associated with climatic and hydrological changes.

While some northern lineages remained disconnected for a sufficient period to develop evolutionary independence, secondary contact between certain lineages has taken place, as detected through SSR and mtDNA analyses. Examples of such secondary interactions include Santa Rosa and Iberá, Manantiales and Santa Rosa, and among distinct mtDNA lineages of C. perrensi (Figs 1, 5). Notably, all interlineage genetic exchanges took place in the same geographical region—a central part of the distribution of all known populations—while other demes outside this area remained isolated. This suggests this is a highly unstable zone where colonizations, extinctions, and reconnections likely occurred iteratively. Ctenomys dorbignyi and C. roigi likely remained isolated due not exclusively to ecological reasons, but most probably as a result of substantial karyotypic divergence from neighboring populations. Therefore, these populations may function as containment belts, limiting genetic exchange with other potentially interbreeding lineages.

If arriving at a universal species concept is difficult —even impossible —achieving universal agreement on infraspecific divisions is likely even more challenging. However, subspecies, races, and varieties are often considered real entities in nature, and recognized as biologically meaningful groups (Reydon and Kunz 2019). Given that infraspecific groups are, by definition, subdivisions of species, their status depends entirely on the status of their respective species.

In mammals, subspecies recognition is controversial: some authors deny the existence of this taxonomic category as a real entity and claim that it should not be used (Burbrink et al. 2022). Others, in a typological approach, highlight the importance of heritable phenotypic differentiation to determine subspecies (Patten 2015; Molinari 2023). However, the distinction between genetically and environmentally determined character variability (i.e., phenotypic plasticity) is usually difficult to establish. Other approaches focus on detecting incompletely separated lineages with semi-independent evolution (de Queiroz 2020; Reydon and Kunz 2021), which aligns more closely with the species concepts currently used (Reydon and Kunz 2021). The case of the Corrientes group is an example of chromosomal radiation (Kavalco and Pasa 2023). The closest population groups are differentiated by several chromosomal rearrangements, which are expected to generate some degree of hybrid sterility proportional to the cytogenetic divergence (King 1995). In addition, contact between them is restricted by their patchy distribution and the solitary, subterranean lifestyle characteristic of the genus Ctenomys (Reig et al. 1990). Among these populations, morphological differentiation is low, as expected due to the constraints imposed by the subterranean environment (Lessa 1990). The semi-independence of the lineages proposed in this work is supported by the integration of different genetic data (mitochondrial and nuclear). Accordingly, we propose a new taxonomic arrangement for the Corrientes group that is more compatible with the data obtained to date. However, due to the deep chromosomal divergence between populations and the low frequency of heterozygotes (involving only up to two pairs), other independent or semi-independent lineages may require additional taxonomic recognition (e.g., Sarandicito, Santa Rosa).

The integration of different lines of genetic evidence (i.e., SSR, mtDNA, cytogenetics, genomic data) plus qualitative and quantitative morphological analyses, supports the existence of at least four different taxa within populations currently referred to as the Corrientes group. These four groups are here regarded as two species, one of them with three subspecies. We therefore propose that C. perrensi and C. roigi are sufficiently distinctive to be considered full species, while C. dorbignyi plus an unnamed subspecies from Iberá, which is described below, are part of C. perrensi.

Additional support for this taxonomic arrangement comes from a recent study employing a high-throughput genomic approach, which found that the genomic distances between C. dorbignyi and a topotype of C. perrensi fall within the range typically observed for intraspecific comparisons (Tomasco et al. 2024). This indicated that our taxonomic decision to incorporate part of the diversity displayed by the Corrientes group at the subspecies level into the species category is more in line with the data obtained to date (mitochondrial and nuclear genomes, plus morphological data). Despite some populations showing exclusive haplotypes, unique chromosome complements, and/or forming their own SSR clusters, these genetic markers can evolve fast and be neutral enough not to establish species boundaries.

As was stated above, the lineage from Iberá has no available name, so we describe and diagnose it as follows: