This is the first study designed to include all described species of Ictalurus (except for I. meeki) within a molecular phylogenetic analysis, considering not only the currently recognized and synonymized species (2), but also individuals from well-differentiated populations, regarded as undescribed taxa in several published accounts (2,18), and from several previously unstudied populations. This could allow an improved estimation of the species composition of this genus. Our multilocus and species delimitation approach resolved the phylogenetic relationships among Ictalurus species, and revealed inconsistencies with respect to the previous taxonomic scheme. Our results also revealed the existence of genetic lineages that could represent undescribed species, and corroborated previous synonymies between species proposed on morphological grounds.
Agreement and conflicts between individual genes, concatenated analysis, and species trees: recovering the evolutionary relationships
Differences among genes were mainly found in the relationships of terminal nodes, which can be attributed to the higher content of variable and informative sites for cytb compared to the other mitochondrial genes, coxI and atpase8/6 (25-27), as shown by the number of informative characters and substitution rates. Previous studies found that coxI is more conserved than cytb, in which more than half of all the amino acid sites were invariable across 250 fish species (28). Moreover, for the nuclear gene RAG1, the lack of resolution in the more derived groups, as was the case with the pricei and lupus complexes, could be associated with incomplete lineage sorting from this nuclear gene. As evidenced by previous studies, the mitochondrial locus showed a high mutation rate, which is related to its smaller effective population size (29), and is also corroborated by the observed substitution rates.
Although the most informative cytb gene placed I. dugesii as the sister group of the pricei and lupus complexes, the concatenated tree, the coalescent species tree, and the coxI, atpase8/6 and RAG1 genes placed I. dugesii as closely related to the pricei complex. Nevertheless, the phylogenetic conflict among trees derived from individual genes was present in the low support shown in the species tree. This latter relationship is consistent biogeographically, since the western limit of the distribution range of I. dugesii (the Chapala-Ameca River) is adjacent to the southern distribution range of the pricei complex, which comprises the river basins of the Sierra Madre Occidental, including the Santiago River (Fig. 1). Similar biogeographic patterns have been reported in studies of other co-distributed fish species, including the Goodeids (30), Cyprinids (31) and Catostomids (32). All of these freshwater fish groups feature species that inhabit the central Mexican highlands, the sister species of which are distributed in river drainages of the Sierra Madre Occidental. Previous empirical analyses show that discordances between concatenated and coalescent-based analyses tend to occur when the branches of concatenated analyses are short and weakly supported as a consequence of signal conflict between individual gene trees (33). Based on this, we regard the species tree as a consistent phylogenetic hypothesis with which to depict the relationships and evolution of the genus Ictalurus (Fig. 4).
Evolutionary relationships and their taxonomic implications
According to the BPP analyses, the second scenario for a small ancestral population and low divergences (IGθ = 3, 0.002, IG τ0 = 3, 0.004) and the fourth scenario for a small ancestral population and deep divergences (IGθ = 3, 0.002, IG τ0 = 3, 0.4) in the BPP A10 and A11 analyses support almost all of the splits and assignments in the species delimitation, with the exception of Ictalurus furcatus and I. meridionalis, and the differentiated populations I. aff. lupus Cuatro Cienegas and I. aff. lupus Conchos, which were either supported by only one scenario, or unsupported in either of the analyses (Tables 4 and 5). The results highlight the particular case of I. balsanus, which was supported by only one scenario in the A11 analysis, as discussed below. Previous studies, in which different prior parameters were tested, and with values that varied considerably, found unstable support patterns, i.e., higher support for one prior combination and lower support for another (34-36). Indeed, BPP analyses have found to be sensitive and potentially misleading, especially regarding the age of the root (τ0) that, with increasingvalues (i.e., 0.04 or 0.4), produced spuriously high posterior probabilities possibly because high values of τ push coalescent events closer to the tips (36). However, this is not the case in the present study. Despite the increase in τ0 (0.004 to 0.4), under the second and fourth scenarios, the supports were unaffected, thus eliminating any misleading results and/or artifacts of the coalescent events. Moreover, there are cases in which θ prior increases considerably (by at least two orders of magnitude), as occurred in the present case (i.e., 0.002 – 0.2) and speciation probabilities tend to decrease (37). This is seemingly associated with the use of an extremely high θ, which is not suitable in all cases (34). Thus, the high values for θ used in scenarios 1 and 3 (0.2) resulted in an unsuitable set of priors for Ictalurus. This indicates that scenarios 2 and 4, which supported 13 of the 15 proposed groups as potentially well-delimited taxa, are more suitable for discarding a misleading result.
As with the earlier phylogenetic hypothesis based on morphology (2), a dichotomy between the furcatus and punctatus groups was found in our analyses. Interestingly, our study revealed several differences in species relationships when compared with the most recent phylogenetic hypothesis involving catfishes of the family Ictaluridae, in which the molecular data and morphological characters of extant and fossil species were used (11).
The furcatus group
The results of the species delimitation analyses do not support the split in the A10 analysis of Ictalurus meridionalis and I. furcatus, and the assignments in the A11 analysis for either of these taxa were highly supported only by the second BPP scenario, and in the case of I. meridionalis only moderately supported by the third scenario. It is therefore not possible to validate these two taxa and based on the principle of priority, I. meridionalis is considered a synonym of I. furcatus, corroborating its status as a synonymized species as previously stated by Lundberg (2). This is consistent with the values of genetic divergence of 1.5% found with cytb between I. meridionalis (Hondo River and Peten samples) and I. furcatus from across a wide geographic range (Mississippi to Coatzacoalcos samples). This divergence value lies below the minimum between-species level within Ictaluridae (1.8%-3.6%) (16, 22-24), and these results indicate that the morphological differentiation reported (38) could be the result of geographical variation within the widespread I. furcatus. However, limited sampling size at their southern limit in the present study prevents detailed analysis of the genetic variation along its geographic range. A future phylogeographic study including optimal sampling along the distribution range of the species is therefore necessary to determine the genetic pattern of the two divergent groups.
Unexpectedly, assignment of the species Ictalurus balsanus, which is one of those that diverged early within the genus, was only supported by the second scenario in the A11 analysis. However, the deep divergence level of this taxon strongly validates it as a recognized species. Therefore, based on the results presented herein, we consider the furcatus group to be formed by the species I. balsanus and I. furcatus, as originally was established (2), pending a phylogeographic study with a wider geographic range for I. furcatus.
Our results do not entirely support the previous phylogenetic hypothesis (13), which included molecular data and morphological characters as a total evidence analysis. In this study (13), Ictalurus balsanus was found as the early divergent species, and as the sister species of the clade containing all other living and fossil species within the genus. This discrepancy might have been the result of one of the following: the possible influence of the different phylogenetic signal from the genes used (16s, 12s, and RAG2, which were not used in the present study), or the methodological implications associated with the considerable amount of missing data, particularly the lack of complete molecular data for most taxa in the analysis by Arce-H et al. (13). For instance, in I. balsanus, only two of the five genes were included in that analysis. Moreover, it is known that a large number of missing data, together with high rates of change, can lead to long-branch attraction (39). This occurs more frequently using Maximum Parsimony than probabilistic methods (41) such as those used in Arce-H et al. (13).
The punctatus group
The punctatus group included the five remaining recognized species of Ictalurus (I. pricei, I. lupus, I. dugesii, I. mexicanus, and I. punctatus) and five well-differentiated populations (I. aff. dugesii Santiago, Ictalurus sp. Mezquital, I. aff. pricei Culiacan/San Lorenzo, I. aff. lupus Conchos + I. aff. lupus Cuatro Cienegas, and Ictalurus sp. Nazas), four of which were previously recognized as undescribed species (2, 18, 20), and one new well-differentiated population revealed in the present study (I. aff. lupus Soto la Marina). All 11 of these groups were supported by BPP analyses (Tables 4 and 5).
Ictalurus punctatus: The channel catfish presents a native distribution stretching from the north in the Great Lakes, southern Canada, Hudson Bay (Red River drainage), and Missouri-Mississippi River basins, extending southwards to the Panuco River on the Atlantic slope of Mexico (18, 20, 41). In the present study, we included samples from along the entire distribution range of this species, between Canada and central-eastern Mexico, and all samples identified as Ictalurus punctatus (excluding hybrids) were recovered as a monophyletic assemblage in the concatenated and species tree analyses. In addition, low genetic differentiation between geographic samples was detected with all genes, indicating that all of these populations should be treated as a single genetic group or species. Interestingly, was uncovered a similar pattern when dealing with a specific parasite of channel catfishes (42). The trematode Phyllodistomum lacustri,which occurs in the urinary bladder of its hosts, was genetically identical in channel catfishes sampled in Canada, USA, and northern Mexico (42). However, uncovered a cryptic species complex of P. lacustri in populations of catfishes occurring in several river basins of Mexico, where Ictalurus has experienced a great diversification process (42).
All of the specimens of channel catfish sampled were morphologically identified or genotyped (with mitochondrial genes) as I. punctatus but, in four cases, specimens from the Santa Maria River in the Panuco River basin, and Pantepec River in the Tuxpan River basin, were assigned to the I. mexicanus clade with the nuclear locus, and a specimen from Sabinas River within the Bravo River basin was assigned as part of the lupus complex with the nuclear locus. Due to the high genetic divergence values (7.5% based on cytb in the former two, and 7.8-8.7% in the latter) between I. punctatus and the I. mexicanus - lupus complex, as well as the phylogenetic position with both mtDNA and nDNA of most samples from the I. mexicanus - lupus complex, the possibility of an incomplete lineage sorting can be discarded. Such findings therefore suggest the occurrence of hybridization events between I. punctatus and specimens of the I. mexicanus - lupus complex. Previous studies have found hybridization events between I. punctatus and other species, such as I. pricei in the Yaqui River (43-44) and I. lupus in the Bravo River (45). Interestingly, the populations of the trematode Phyllodistomum lacustri from the Pantepec River in Veracruz (identified by the authors as I. punctatus) differed slightly from those of P. lacustri from northern areas when analyzed through the mitochondrial gene coxI (42).
Ictalurus australis and Ictalurus mexicanus:Both species were described by Meek (1904), the type-locality of Ictalurus australis is the Forlon River, a tributary of the Guayalejo River sub-basin of the Panuco River Basin, and extends its distribution along the Panuco, Tuxpan, Cazones, Tecolutla and Nautla Rivers (47), reaching the Blanco River within the Papaloapan River basin (48). The type-locality of I. mexicanus is the Gallinas River, and it is considered endemic to this tributary in the Upper Panuco River (18, 49-50). The taxonomic status of I. australis has been questioned and the species has even been considered a junior synonym of I. punctatus (2, 51-52). However, several studies still consider I. australis as a valid taxon (20, 53-54); however, according to Miller et al. (18), I. mexicanus is restricted to the Gallinas River and the populations in the Panuco basin are considered a closely related but undetermined taxon.
We included specimens from the type-locality for both species and, according to the low genetic divergence shown between them (<0.5% with the mitochondrial cytb gene) and the lack of reciprocal monophyly within Ictalurus mexicanus/australis with the cytb and coxI genes, we consider that the samples of I. mexicanus from Gallinas River up- and downstream of the Tamul waterfall, those distributed along the Verde River, and those from the Guayalejo River (according to RAG1 gene) do not represent two distinct lineages, suggesting that I. mexicanus, I. australis and/or Ictalurus sp. sensu Miller et al. (18) are part of the same taxonomic unit. Thus, based on the principle of priority, and pending the inclusion of more samples across the distribution range of this taxon and integrative taxonomic analyses, we considered recognizing the population from the Gallinas River up- and downstream of the Tamul waterfall, as well as the populations distributed in the Verde and Guayalejo Rivers within the Panuco basin, as I. mexicanus. The lack of I. australis samples from the southern basins such as the Tuxpan, Cazones, Tecolutla, Papaloapan, and Nautla Rivers (47-48), indicates the need to include samples from the entire potential distribution range in a phylogeographic analysis in order to explore the relationship with, and patterns of differentiation from, I. mexicanus of the Panuco basin.
The pricei complex: northwestern pacific group:This monophyletic group is represented by one recognized species, Ictalurus pricei, and three well-differentiated populations with a divergence of 1.8 - 1.9% for the cytb locus. This value exceeds the minimum divergence found between well-recognized species within Ictaluridae (16, 22-24). These four divergent groups were corroborated by the BPP species delimitation test as independent taxonomic units. The northernmost taxon corresponds to I. pricei sensu stricto from the Yaqui and Fuerte rivers. We did not include samples from other basins such as the Sonora, Mayo, and Casas Grandes Rivers (18, 44), and these populations therefore remain to be analyzed in order to determine whether or not they can be nested with I. pricei from the Yaqui and Fuerte rivers.
Another differentiated group was represented by the specimens of Ictalurus aff. pricei from Culiacan/San Lorenzo rivers, confirming previous morphological findings that indicated this population as a potential undescribed taxon (21). Similarly, the specimens of Ictalurus sp. from Mezquital River were corroborated as representing a potential undescribed taxon in our analyses, as previously recognized by other authors (2,19-20). Finally, the differentiated population of I. aff. dugesii from the Santiago River, which corresponds to the southernmost independent taxonomic unit of the pricei complex, was also corroborated as a potential undescribed taxon, originally referred to by Miller et al. (18) as an undescribed species related to I. dugesii. The results of the study by Rosas-Valdez et al. (42) on recognition of a cryptic species complex of the trematode Phyllodistomum lacustri may lend further support. The populations of the parasite sampled in catfishes of the Mezquital River basin (identified by the authors as Ictalurus sp), and those from the Lerma River basin at San Juanico Dam (identified as I. dugesii), but also including introduced specimens of I. punctatus, were retrieved in the molecular phylogenetic analyses of two genes (28s and coxI) as cryptic species, each representing a separate species.
Ictalurus dugesii:The Ictalurus samples from the Lerma-Chapala system, including specimens identified as I. ochoterenai from Chapala Lake and the populations from the Ameca and Armeria River basins, were nested within the I. dugesii clade. With regard to I. ochoterenai and I. dugesii, we found a lack of reciprocal monophyly, as well as genetic divergence (<0.5%) that was below the minimum limit recorded for Ictaluridae using cytb (1.8-3.6%) (16, 22-24). Based on the above, the present study confirms I. ochoterenai as a synonym of I. dugesii, as was proposed by other authors (55, 2).
An undescribed form of I. dugesii was proposed by Miller et al. (18), corresponding to the populations of I. aff. dugesii Armeria. In this case, even though we found reciprocal monophyly with cytb, this group was nested within I. dugesii when data were analyzed using the other two mitochondrial genes. It also presented low genetic divergence, which suggests a relatively recent isolation event within the Armeria River basin. Thus, the distribution of I. dugesii is confirmed for three disjunct river basins in central Mexico, i.e., the Lerma-Chapala system and the Ameca and Armeria Rivers. This is in accordance with other co-distributed fish species in the region, such as the catostomid Moxostoma austrinum Bean, 1880 (56) and several species of goodeids (57-58).
The lupus complex: The distribution range of the headwater catfish, I. lupus comprises several rivers of the Gulf of Mexico slope, including the Colorado, Guadalupe, and Nueces drainages in Texas, the Bravo River drainage in the United States and Mexico, as well as the San Fernando, Soto la Marina and Conchos Rivers, and the endorheic Cuatro Cienegas basin in Coahuila, Mexico (45, 59). For the cytb analyses, I. lupus sensu stricto from the Pecos and Devil Rivers presented a relatively high genetic divergence with respect to the other populations (I. aff. lupus Conchos, I. aff. lupus Cuatro Cienegas, and I. aff. lupus Soto la Marina), ranging from 2 to 2.8%. Although I. aff. lupus Conchos and I. aff. lupus Cuatro Cienegas presented a high divergence value (1.9% in the cytb gene), the species delimitation test resolved only three groups as taxonomic independent units: I. lupus, I. aff. lupus Conchos + Cuatro Cienegas, and I. aff. lupus Soto la Marina (Tables 4 and 5).
The genetically differentiated I. aff. lupus Conchos seems to correspond to the previously reported divergent population of Ictalurus from Conchos River, which also occurs in sympatry with I. lupus in the lower Bravo River (2, 18-19, 51). Additionally, Ictalurus aff. lupus Soto la Marina was supported as a candidate species in the BPP analyses and presented the highest genetic divergence within the lupus complex, merging as a novel recognized independent evolutionary lineage and putative species. According to our findings, and pending an integrative taxonomic study with more samples as well as a detailed study of the morphoanatomy of this Ictalurus population, we support the existence of three independent taxonomic units within the lupus complex, I. lupus sensu stricto, Ictalurus aff. lupus from Conchos River and Cuatro Cienegas valley, and Ictalurus aff. lupus from Soto la Marina River.
Ictalurus sp. Nazas: Our study resolved the specimens of Ictalurus from the Nazas River as a highly divergent group within the punctatus complex. All of the analyses confirmed this population as a reciprocally monophyletic group, which was apparently the first to diverge within the central-western drainages, indicating this as a potential taxon in the BPP analyses. Our results support previous studies that considered the population of Ictalurus from the Nazas as an undescribed species, although these studies provided no details regarding the morphological characters that distinguished this from other congeners (2, 19, 20, 60) and, consequently, the description of the new species is pending. Other studies indicate that the Nazas River is an area where independent genetic lineages of fishes occur (61-65).
Divergence times and evolution of Ictalurus
The genus Ictalurus is confirmed as an ancient group of fishes; our dating analysis placed the earliest split between the punctatus and furcatus groups at ca. 33-23 Ma (Fig. 4), which is partially consistent with previous studies that estimate the age of the crown of Ictalurus at 38-20 Ma (66) and 37-30 Ma (13). The considerable species richness in the fossil record, together with the inferred divergence times, indicate that the genus experienced several extinction events during its evolutionary history, particularly in the case of early-diverging species. The most abundant fossil record for Ictalurus has been found in North America (13, 67-68), and includes the oldest fossil, Ictalurus rhaeas, discovered at the Cypress Hill Oligocenean Formation, located in southern Saskatchewan, Canada. Hydrographically, this formation is included in the Missouri-Mississippi drainage system (67). The region originated in the Late Eocene and is one of the most important in terms of explaining the evolution of the North American freshwater fish fauna, including ancient fish fauna elements such as gars (Lepisosteidae), bowfin (Amiidae), and mooneyes (Hiodontidae), as well as the most recent representatives of North American fish fauna, such as catostomids, salmonids, percids, cyprinids, and catfishes (69). The first lineages, derived during the evolution of the genus Ictalurus in the early Miocene, show high dispersal capacity and adaptation to the environmental conditions that allowed them to expand their distribution range along the Atlantic slope of North America, as reflected in the widespread distribution of I. furcatus and I. punctatus, from southern Canada through the US and Mexico southwards to Belize, including the dispersal of the ancestor of I. balsanus to the Balsas River on the Pacific slope of Mexico.
The large diversification observed in the punctatus group shows a first split with the cladogenetic event of Ictalurus mexicanus in certain tributaries of the Panuco River basin. Although no explanations have been proposed for the origin and isolation of this species in the Verde or Gallinas rivers (currently inhabited by I. mexicanus), this region has served as an area of speciation for different groups, with the occurrence of a significant number of endemic species of cyprinids (62), goodeids (57), cichlids (70-71) and poecilids (18). This high level of endemism seems to be related to the intense volcanism of the Miocene-Pliocene, which promoted a sudden subsidence of the graben structure in the basin and the formation of shallow lakes (72). Apparently, I. mexicanus followed the same speciation pattern as other fish species in the area.
The next diversification event was the cladogenesis of Ictalurus sp. Nazas, and the lupus complex distributed in the Northern basins of the Bravo and Soto la Marina Rivers, followed by the formation of the I. dugesii and the pricei complex with wide distribution along the Pacific slope drainages, from the Yaqui River in Sonora southwards to the Lerma River in Michoacan and Guanajuato. Several fish groups experienced the same speciation pattern, including the cyprinids and catostomids (32, 62). These speciation events associated with the Bravo, Nazas, and Northern pacific slope drainages have been explained by hypothetical connections along the extensive paleo-hydrological system in the Chihuahuan desert region, which dates back to the Oligocene and included the Conchos River, the main channel of the Bravo River, the Nazas River basin, and the headwaters of several western Pacific river drainages (i.e., the Yaqui and Mezquital Rivers) (63), and the isolation of which is estimated at ca. 5 Ma (19). The biogeographic role of the Chihuahuan desert paleo-hydrological system is closely associated with the tectonic activity of the Bravo River rift, together with the arid conditions prevalent since the Miocene (73). This paleo-hydrological system could have had two main roles in the evolution of the genus Ictalurus, which is similar to that reported for other fish groups (32, 61, 63-64, 74-76): 1) most of the diversification events occurred within or were promoted across the region, and 2) the region acted as a corridor for the punctatus group, allowing them to colonize the Pacific slope, where they subsequently diversified.
In general, derived from a temporary and geographically extensive evolutionary history, most of the speciation events within the genus Ictalurus occurred in the river basins of Mexico. This implies an increase in species richness and levels of endemism. The results of our study pose a huge challenge for research of the diversity and conservation of this representative and important group of North American freshwater fishes. The most recent evaluation of the IUCN red list for fish species in Mexico categorizes only I. pricei and I. mexicanus in the “at risk” category (77), mainly based on the premise of the high distribution range of most of the species. However, our research contradicts this notion of the wide distribution of the species and raises the need for a more detailed conservation evaluation of the independent evolutionary units or undescribed species found in the present study, in order to adequately protect the true diversity of the genus Ictalurus in Mexico. Furthermore, the discovery of hybridization events between I. punctatus and other endemic Ictalurus species highlights the necessity to review aquacultural policies that involve the uncontrolled introduction of non-native species into other basins, leading to dramatic effects on the ecosystems and the potential introduction of invasive parasite species. New policies should guarantee and fully incorporate the study of the possible environmental consequences of such introductions.
Finally, parasitological data is frequently used, along with the phylogenetic history of hosts, to explain concurrent patterns. In this context, the members of the punctatus group were typical hosts of the trematode Phyllodistomum lacustri, a parasiteof the urinary bladder of several species of ictalurids (42). Molecular phylogenetic analyses of specimens of P. lacustri from Ictalurus punctatus sampled from its natural distribution range across Canada, USA, and northern Mexico show a pattern congruent with that of the hosts (42). Interestingly, also was reported the presence of a cryptic species complex of P. lacustri that includes other species of Ictalurus, such as I. dugesii in the Lerma Santiago basin, and I. pricei (Ictalurus sp. Mezquital in this study)from the Mezquital basin (42). These river basins of western Mexico appear to be important areas for the diversification of I. dugesii and I. pricei and their parasites. We believe that future studies aimed at uncovering diversification patterns of freshwater fishes should, whenever possible, utilize other information sources in order to corroborate hypotheses. We also consider that the history of the host-parasite association could constitute a robust proxy with which to test such patterns of diversification.