Molecular Systematics of the Phoxinin Genus Pteronotropis (Otophysi: Cypriniformes)

The genus Pteronotropis is widely distributed along the gulf slope of eastern North America from Louisiana to Florida and rivers in South Carolina along the Atlantic slope. Pteronotropis have very distinctive, flamboyant coloration. The habitats most frequently associated with these species include heavily vegetated backwater bayous to small sluggish or flowing tannin-stained streams. Although Pteronotropis is recognized as a valid genus, no phylogenetic analysis of all the species has corroborated its monophyly. In recent years, four additional species have been either described or elevated from synonymy: P. merlini, P. grandipinnis, P. stonei, and P. metallicus, with the wide-ranging P. hypselopterus complex. To examine relationships within this genus and test its monophyly, phylogenetic analyses were conducted using two nuclear genes, recombination activating gene 1, RAG1, and the first intron of S7 ribosomal protein gene in both maximum parsimony and Bayesian analyses. In no analysis was Pteronotropis, as currently recognized, recovered as monophyletic without the inclusion of the currently recognized Notropis harperi, herein referred to as Pteronotropis. Two major clades are supported: one inclusive of P. hubbsi, P. welaka, and P. harperi and the second inclusive of P. signipinnis, P. grandipinnis, P. hypselopterus plus P. merlini sister to P. euryzonus, and P. metallicus plus P. stonei.


Introduction
The subfamily (or family) Leuciscinae includes all cyprinid species in North America, except Notemigonus, and species across Eurasia. Many of the species of this North American fauna have been examined in different phylogenetic studies at varying degrees of universality using both morphological and molecular data. Initial morphological studies by Mayden [1] and Coburn and Cavender [2] revealed exciting new relationships and a reclassification of the North American fauna. These studies were followed with several molecular analyses of different major lineages, genera, and species groups that supported many, but not all, of the monophyletic groups previously identified in one or both of the above studies [3][4][5][6][7]. However, not all proposed genera have been examined for species relationships using molecular markers.
One such genus in North America with an increasing and intriguing diversity, biology, and geographic distribution, as well as complex taxonomic history, is Pteronotropis. This genus contains one of North America's most colorful shiners. Pteronotropis hubbsi and P. welaka are relatively slender-body species that, in breeding males, possess enlarged dorsal fins, whereas the remaining species, P. euryzonus, P. hypselopterus, P. merlini, P. grandipinnis, P. stonei, P. metallicus, and P. signipinnis, are more deep-bodied and lack enlarged dorsal fins in breeding males. The most frequently associated habitat of these species across their ranges includes deep, backwater bayous, small sluggish tannin-stained streams, and flowing tannin-stained streams, all with ample aquatic vegetation. However, despite several studies on shiners and relatives to date, Pteronotropis has received essentially no recent attention as to their relationships and has been proposed to be an unnatural grouping. Herein, we provide the first examination of phylogenetic relationships of all species in the genus (formerly subgenus of Notropis [1]) and a test of the monophyly of this purported lineage. Two nuclear genes are used in this analysis because of their previously demonstrated genetic distances and resulting ability to resolve nodes deeper than at the crown of trees. These genes have been used successfully for resolution of more basal lineages of North American cyprinids by several recent papers [3][4][5][6][7][8][9][10].
Resulting phylogenetic inferences of species of this group and their eventual placement relative to other North American cyprinids are critical as they largely facilitate more process-level questions as to the evolution of the biology of the species and other lineages to better understand the processes of anagenesis and speciation. While multiple papers listed above have made groundbreaking strides in providing a phylogenetic framework where one previously did not exist for North American cyprinids, Hollingsworth et al. [10] provide an excellent evaluation of a subset of the fauna and a novel hypothesis as to habitat shifts for clades with differing rates of speciation. Given that no study has examined all of the species of Pteronotropis, we provide a review of the history of the genus and molecular phylogenetic analyses of the species using two nuclear genes that result in identical species relationships based on mitochondrial genes in Mayden and Allen [11].
Taxonomic History. Pteronotropis currently includes nine species in rivers and streams distributed along the gulf slope from Louisiana to Florida and along the Atlantic slope as far north as South Carolina. One species, P. hubbsi, currently occurs only in southern Arkansas and northern Louisiana but was likely to be more widely distributed in lowland habitats; the conservation status of this species is of concern, as it has not been found in some locations (including southern Illinois) for several decades.
In a study focusing on 566 morphological traits of a large number of cyprinids, Mayden [1] elevated Pteronotropis to generic level and included P. welaka, P. signipinnis, P. hypselopterus, and P. euryzonus within the genus but left P. hubbsi in Notropis. Recently, Suttkus and Mettee [12], with no characters, phylogenetic analysis, or substantive phylogenetic argument, maintained that Pteronotropis was a subgenus within the genus Notropis (as classified before Mayden's [1] analysis) and that this subgenus contained only P. euryzonus and the P. hypselopterus complex (P. hypselopterus, P. grandipinnis, P. stonei, P. metallicus, and P. merlini).
The phylogenetic relationships of Pteronotropis have been somewhat enigmatic over the years. Species share derived and distinctive color patterns that include bright red-orange to yellow striped dorsal, caudal, and anal fins and a broad dark lateral band extending from the head to the caudal peduncle. The genus was divided into two groups based on morphological and molecular characters [1,4]. Pteronotropis signipinnis was described by Bailey and Suttkus [13] and was considered a member of the genus Notropis (subgenus Pteronotropis by Fowler [14]), along with P. hypselopterus. Pteronotropis euryzonus [15] was later added to this subgenus and was considered a close relative to P. hypselopterus; however, neither of the above two studies included P. hubbsi or P. welaka and they were conducted in a prephylogenetic era. Pteronotropis hubbsi was described by Bailey and Robison [16] and was thought to be closely related to P. welaka; at that time, neither species was allocated to the subgenus Pteronotropis. In a study utilizing twenty-one allozyme loci, Dimmick [17] examined nine species (mostly Pteronotropis). This allozyme analysis revealed Pteronotropis as nonmonophyletic, with P. hubbsi and P. welaka as distantly related and N. signipinnis and N. hypselopterus as sister species. Consequently, Dimmick [17] argued that all of the morphological characters of Bailey and Robison [16], thought to indicate a close relationship between P. hubbsi and P. welaka, were the result of convergent evolution.
In the first sequence analysis of this group, Simons et al. [18] used mitochondrial cytochrome b gene and failed to corroborate Pteronotropis as a monophyletic group. With both parsimony and likelihood analyses, P. euryzonus was sister to P. hypselopterus and an unrelated clade included P. signipinnis sister to P. hubbsi plus P. welaka. Later, in a subsample of Pteronotropis species, Simons et al. [4], using two mitochondrial genes (12S, 16S), and Bufalino and Mayden [5,6], using two nuclear loci (RAG1, S7), found Pteronotropis as monophyletic but, again, only with the inclusion of "Notropis" harperi; however, neither of these analyses included all species of the genus. Other early molecular data and analyses also failed to resolve the phylogenetic relationships of the above species that were generally phenetically similar. Most recently a study by Hollingsworth et al. [10], using one mtDNA gene and nDNA genes, corroborated the monophyly of a subsample of species of Pteronotropis that also included N. harperi.
While there have been several efforts testing the monophyly of Pteronotropis, its composition, and at resolving the phylogenetic relationships of species since its elevation to genus, no single study has included all of the species in the genus and appropriate outgroups based on earlier studies and some did not include the morphologically similar Notropis harperi. With the elevation of species from synonymy with P. hypselopterus and the description of a new species [12], the complexity involved in testing the monophyly of the genus and species relationships have become even more biologically interesting. While Suttkus and Mettee [12] did provide dialogue invoking phylogenetic terminology as to species relationships, their study contained no phylogenetic analyses, no discussions of character homology, or any morphological or molecular synapomorphies. To date, no investigation has been completed for this group inclusive of all of the purported species of Pteronotropis. Thus, the objectives of the current study are twofold: (1) testing the monophyly of the genus and (2) examining relationships of all of the purported species of the genus using two nuclear genes.

Specimens and DNA Extraction/Amplification and Alignment.
Museum catalogue numbers for vouchers in this study include UAIC (University of Alabama Ichthyological Collection) and SLUM (Saint Louis University Museum). Specimens examined in this study were either frozen at Saint Louis University, preserved in 95% ethanol, or captured alive and transported to Saint Louis University (Table 1). Outgroup taxa included species from the genera Cyprinella, Lythrurus, and Notropis. Species of Cyprinella were included  seconds. This was followed by an annealing temperature of 55 ∘ C for 1 minute, an initial extension of 72 ∘ C for 90 seconds, and a final extension of 72 ∘ C for 5 minutes. Conditions for S7 were identical except the annealing temperature was set at 59 ∘ C. For the S7 intron, products that failed to amplify using the universal primers were reamplified using nested PCR reactions with the same conditions except for specific annealing temperatures as specified by the chemistry for the internal primers. Taxa failing to amplify with internal primers were cloned using the pGEM-T Easy Vector System kit (PROMEGA, Madison, WI) as outlined in Lang and Mayden [9]. PCR products were purified using QIAGEN gel extraction kits (QIAGEN, Valencia, CA). Sequencing was performed using a BigDye labeled dideoxy sequencing kit (BigDye) and visualized on an ABI 377 automated sequencer (Auburn University Molecular Genetics Instrumentation Facility, Auburn, AL) or an ABI 3700 (Macrogen Sequencing Facility, Seoul, South Korea). Both the heavy and light strands were sequenced for all samples and the sequences were aligned with Clustal X [21] with reference to the accompanying electropherograms. Some individuals contained heterozygote peaks in the RAG1 data and these heterozygote base pair positions were coded using standard degeneracy codes.

Phylogenetic
Analyses. An incongruence-length difference analysis (ILD [22]) was performed with 1000 replicates to test for incongruence between the RAG1 and S7 data sets. Maximum parsimony (MP) analyses (MPA) were conducted in PAUP * 4.0b10 [23]. All analyses consisted of a heuristic search model with 1000 random addition sequence replicates and TBR. Support for the parsimony analyses was generated using bootstrap analysis (BS) with 1000 bootstrap pseudoreplicates [18]. Bayesian analyses (BA) were conducted in Mr. Bayes 3.0b4 [24]. S7 intron all gaps were treated as missing data. The model of sequence evolution was determined using Modeltest v3.04 [25] with single partitions for each marker; the best-fit model for S7 was HKY + G and that for Rag1 was TrN + I + G. BA included four heated Markov chains using the default temperature setting. Log-likelihood scores were plotted against generation time to establish burn-in; trees prior to stationarity were discarded. Post-burn-in trees were used to develop the 50% majority rule consensus tree. Posterior probabilities (PP) were used as an indication of nodal support in BA.

Results and Discussion
As the ILD test was nonsignificant for heterogeneity between RAG1 and S7, the gene sequences were analyzed both individually and as a concatenated data set. MP analysis of the aligned 1001 bp of S7 (aligned sequence lengths ranged from 839 to 919 bp) yielded 245 bp parsimony informative sites (12.9%). Analyses of these data resulted in 90 equally parsimonious trees (Figure 1; length = 697, CI = 0.803, and RI = 0.875). The more conservative RAG1 sequences included 1521 bp with 151 bp sites (9.9%) being parsimony  informative. MP analyses of RAG1 resulted in 46,668 equally parsimonious trees (Figure 1; length = 371 steps, CI = 0.658, and RI = 0.866). Individual BA analyses for each gene resulted in some variations in sister-group relationships but all were consistent and supported the monophyly of Pteronotropis (Figure 2). Both MPA and BA of the combined S7 + RAG1 data recovered identical topologies ( Figure 3).
As in previous studies involving species of Pteronotropis, nuclear sequence variation, neither individual nor combined [5,6], resolved Pteronotropis as a monophyletic group if Notropis harperi is excluded from the genus. Constraining Pteronotropis to be monophyletic in the S7 + RAG1 data set without N. harperi resulted in a significantly worse tree (1246 steps). In both BA and MPA, Notropis harperi is resolved as sister to P. welaka within the ingroup, a sister-group relationship with strong PP and BS support (Figures 1 and  2). Pteronotropis hubbsi is resolved as sister to this clade, also with strong PP and BS support. All three of these taxa (P. hubbsi (P. welaka + N. harperi)) are resolved as monophyletic and sister to the remaining species traditionally referred to as Pteronotropis (PP 95, bootstrap 75; Figure 2). The strong support for the monophyly of the (P. hubbsi (P. welaka + N. harperi)) clade (Figures 1 and 2) is logical as the three species are phenetically and ecologically similar. They possess aspects of similar body coloration in life when not in breeding condition and have similar habitat associations [5,26,27]. They are found in deep pools with ample aquatic vegetation and in areas where P. welaka and N. harperi are sympatric they are often taken syntopically in a sample (pers. obs.). The authors are unaware of any studies corroborating nest association in N. harperi, as observed in P. welaka and P. hubbsi [28][29][30]. In light of the relationships presented here and in Bufalino and Mayden [4,5] and Hollingsworth et al. [10], studies of N. harperi may reveal ecological and behavioral synapomorphies.
In all analyses, P. signipinnis is resolved as sister to a clade of remaining species of Pteronotropis (Figures 1 and 2). In analyses of S7 and S7 + Rag1 data sets, the latter clade formed two clades: one inclusive of P. hypselopterus, P. grandipinnis, and P. merlini and the other inclusive of P. euryzonus, P. stonei, and P. metallicus. Resolution of the former clade was not complete in either Rag1 or S7 analyses, but both are fully consistent with the phylogeny recovered with the Rag1 + S7 data set. These relationships are in contrast to those hypothesized by Simons   P. welaka + P. hubbsi. However, this latter study did not include all of the then or currently known species of Pteronotropis.

Conclusions
Given the consistent sister-group relationship between formerly recognized Notropis harperi and Pteronotropis welaka, the former species is herein referred to as Pteronotropis.
Nuclear genes RAG1 and S7 support the long-standing question/hypothesis regarding the monophyly of Pteronotropis and provide new insight into the phylogenetic placement of Pteronotropis harperi and the basal-most relationships between the species groups (P. hubbsi, P. welaka, and P. harperi) relative to the remaining species of Pteronotropis. These relationships are also consistent with those presented by Bailey and Suttkus [13] using mitochondrial gene ND2. In recent years, the general trend in phylogenetics has been to place greater emphasis on the use of nuclear genes, largely because of issues associated with hybridization, intergradation, lineage sorting, and disagreement between gene and species trees [13]. While these nuclear genes have shown a greater ability to resolve relationships at supraspecific levels for this group with greater consistency and stronger branch support, the results presented herein illustrate the benefit in using nuclear genes. However, it is also true that mitochondrial genes have been extremely useful in phylogenetic resolutions [26,27], and like nuclear genes they also vary in their degree of anagenesis and abilities to resolve trees at different levels of universality. While these and other nuclear genes used in the above-cited papers for Cypriniformes clearly display a reduced phylogenetic signal and are more limited in phylogenetic resolution for relationships of populations and species, they are essential for resolution of deeper nodes. This is to be expected as rates of mutation of many nuclear genes (especially protein coding) are generally not as high as that typically found in most mitochondrial genes.