Phylogenomics and classification of Notropis and related shiners (Cypriniformes: Leuciscidae) and the utility of exon capture on lower taxonomic groups

Introduction

Among North American cyprinoids, the shiners and related minnows have been among the most taxonomically complex groups of fishes. The group is currently placed in the Leuciscidae, subfamily Pogonichthyinae after the elevation of subfamilies to family rank in cyprinoids (Schönhuth et al., 2018; Tan & Armbruster, 2018). Ichthyologists tasked with assembling species into meaningful genera initially described a dizzying array of genera and subgenera of minnows. Species were moved between various categories (genera, subgenera, tribes) based primarily on phenetic similarity; many leuciscid genera remained relatively stable but one, Notropis (and many taxa formally placed in this genus), has continued to be difficult. A search of the Catalog of Fishes (Fricke, Eschmeyer & van der Laan, 2022) for ‘Notropis’ results in 325 species, representing a considerable bulk of nominal species of North American freshwater fishes.

Starting with Mayden (1989) comprehensive morphological phylogenetic analysis of Cyprinella and other North American taxa, a large monophyletic group was recognized as the Open Posterior Myodome clade (OPM). This clade includes most of the eastern North American leuciscids including Notropis and related genera. Mayden (1989) concluded that Notropis is an artificial group due to convergence of morphological characters and classification by phenetic similarity, and enacted nomenclatural changes elevating some subgenera in Notropis, such as Cyprinella, Luxilus, Lythrurus, and Pteronotropis, recognized other genera and reallocated species. He moved some species from Notropis to Hybopsis (including N. boucardi, N. calientis, N. dorsalis, N. longirostris, N. sabinae, N. alborus, N. bifrenatus), and differentiated six species of Notropis (i.e. N. topeka, N. mekistocolas, N. atrocaudalis, N. stramineus, N. chihuahua and N. procne) from the large clade that included all of the genera previously included in Notropis. A few now recognized genera were then not recognized (i.e. Codoma was within Cyprinella; Opsopoedus was within Notropis). Still, Notropis remained as a “taxonomic repository for small, silvery fishes of unknown relationship” (Gidmark & Simons, 2014: 379; see also Mayden et al., 2006) with approximately 91 currently recognized species loosely organized into subgenera (Jordan, 1885). Primarily because of the large number of taxa, coupled with conserved morphologies, few studies have attempted to tackle the remaining species allocated to the genus or other orphaned taxa of unknown taxonomic placement (Mayden et al., 2006; Hollingsworth et al., 2013; Schönhuth et al., 2018). However, even when necessary taxonomic decisions for species within the genus Notropis were made (sensu Mayden et al., 2006; Gidmark & Simons, 2014), they have not been widely accepted, and traditional taxonomic groups have been preferred by some for nomenclatural stability until a stronger consensus is reached on proposed nomenclatorial changes (Fricke, Eschmeyer & van der Laan, 2022).

Phylogenetically, most previous studies on shiners and relatives have focused on resolving relationships within purported monophyletic subgenera (for example Snelson, 1972; Buth, 1979; Raley, Wood & McEachran, 2001; Cashner, Piller & Bart, 2011) with varied results, and without investigation into relationships among the subgenera or to genera that have been segregated from Notropis. Mayden et al. (2006) attempted a comprehensive study of the so-called Notropin clade (the clade name does not refer to a taxonomic rank) to resolve relationships using cytb (mitochondrial marker) and revealed a nonmonophyletic Notropis as reported in prior studies either based on morphological or molecular data (Mayden, 1989; Simons, Berendzen & Mayden, 2003). The former study, in addition to corroborating monophyly of genera synonymized with Notropis as Cyprinella, Erycymba, Hybognathus, Hybopsis, Lythrurus, also identified a more restricted Notropis by recognizing the additional genera Alburnops, Aztecula, Graodus, Hudsonius, Miniellus, and Yuriria. These authors could not resolve Hydrophlox, Luxilus and Pteronotropis as monophyletic, and identified a ‘Notropis’ longirostris clade that included seven species of Notropis. Despite all these monophyletic groups having available names, their compositions were not always as initially proposed, and relocation of some species were made by Mayden (1989), Coburn & Cavender (1992), and Mayden et al. (2006). However, following this alternative phylogenetic classification of Notropin shiners, many species remained relegated to ‘Notropis’ (in single quotes) because of their uncertain placement due to weak support in analyses, and relationships among the genera listed above remained unclear. While these additional genera from within a nonmonophyletic Notropis were recognized and elevated, most subsequent studies reverted back to a larger encompassing Notropis, with perhaps recognition of some of these genera as subgenera or species groups (Bird & Hernandez, 2007; Rüber et al., 2007; Zhang et al., 2008; Chen & Mayden, 2009; Fang et al., 2009; Gaubert, Denys & Oberdorff, 2009; Scott et al., 2009; Bufalino & Mayden, 2010; Houston, Shiozawa & Riddle, 2010; Cashner, Piller & Bart, 2011; Wang et al., 2012; Hollingsworth et al., 2013; Imoto et al., 2013; Fricke, Eschmeyer & van der Laan, 2022).

Hollingsworth et al. (2013) expanded upon the cytb study by adding the RAG1 (nuclear) molecular marker for a phylogenetic reconstruction to test for a correlation between a shift from benthic to pelagic lifestyles and increased diversification rates. Unsurprisingly, this analysis also resulted in a relatively poorly resolved overall phylogeny with moderate support for non-monophyly of Notropis, but again illustrated the importance of understanding relationships to better inform our understanding of ecological and evolutionary processes.

To promote further study into this group, Gidmark & Simons (2014) amassed much of the knowledge reported for the shiners (distributions, histories, ecologies, etc.) and proposed using the designations made by Mayden et al. (2006) with the understanding that the relationships among them still remain unclear, despite support for the Shiner Clade as a whole (Simons, Berendzen & Mayden, 2003; Mayden et al., 2006; Schönhuth et al., 2008). Recently, a classification of the Holarctic family Leuciscidae based on nuclear and mitochondrial genes was proposed, where the Shiner Clade was a well-supported group within the Pogonichthyinae (Schönhuth et al., 2018). Despite the increase in both taxon and character sampling, Notropis was not resolved as monophyletic, as species of this genus were found in different parts of the Shiner Clade, and some of the genera formerly included in Notropis were not supported as monophyletic (Alburnops, Notropis s.s., Hudsonius), or their composition differed slightly (i.e. Miniellus, Hydrophlox) from that previously proposed (Coburn, 1982; Mayden et al., 2006; Cashner, Piller & Bart, 2011).

Recent advances in sequencing technologies have provided the opportunity to re-examine the shiner clade using phylogenomic markers. Most phylogenomic-scale studies thus far have focused on higher taxonomic levels (Lemmon, Emme & Lemmon, 2012; Bond et al., 2014; Eytan et al., 2015; Prum et al., 2015; Hamilton et al., 2016), but decreases in costs and the establishment of universal loci specific to fishes (Betancur-R et al., 2013; Arcila et al., 2017; Hughes et al., 2018, 2021) have helped overcome the hurdles associated with applying a phylogenomic approach to the shiner clade. In this study, we employ the probes developed by Arcila et al. (2017) in an attempt to tackle the systematic problems of Notropis and related genera, and follow the taxonomy discussed in Gidmark & Simons (2014). We also examined two specimens of three species from different geographic locations to test the utility of the markers on a smaller geographic scale. Previous research (Rincon-Sandoval, Betancur-R & Maldonado-Ocampo, 2019) had shown good utility at population-level scales, but wih the shiners, we wanted to determine if the markers could lead to a stong phylogenetic hypothesis for a group with very rapid divergence (Hollingsworth et al., 2013; Burress et al., 2017).

Materials and Methods

Results

The final alignment yielded 1,004 loci, 286,455 base pairs, and only 0.42% missing data. Of those sites, 32,349 (11.29%) were phylogenetically informative. The range of locus size was 196–1,748 bp, with an average bp length of 285 (Fig. 1). In the resulting concatenated ML tree, 78% of nodes are 100% bootstrap supported with only six nodes collapsing below the 70% bootstrap threshold (Fig. 2). Species tree analysis produced highly congruent results, particularly at the genus level (Fig. 3). At deeper nodes there is less support in the species tree for the placement of a few clades (i.e. Hudsonius hudsonius + H. altipinnis; ‘Notropis’ atrocaudalis + ‘N.’ bifrenatus + ‘N.’ heterolepis), resulting in remaining uncertainty as to the relationships among the genera. Nevertheless, with our focus on resolving within-genera relationships, both concatenation and species tree approaches resolve the same patterns with strong support.

Species that are currently placed in Notropis are not resolved as monophyletic (Figs. 2 and 3). Notropis jemezanus, N. amabilis, N. micropteryx, N. rubellus, and N. amoenus form a clade with the type species, N. atherinoides, and a second clade included N. buchanani, N. wickliffi, N. volucellus, and N. spectrunculus. Species designated as ‘Notropis’ by Mayden et al. (2006) are found throughout the tree. The genera Hudsonius, Luxilus, and Alburnops were not monophyletic. Cyprinella forms a monophyletic group that excludes C. callistia, which forms a trichotomy with Opsopoeodus + Pimephales and the clade containing the remaining members of Cyprinella + Codoma. The results of the AU test were all significant (Cyprinella constrained, p = 3e−06; Hudsonius constrained, p = 6e−08; Luxilus constrained, p = 2e−18), indicating that all of the constrained topologies can be rejected as alternative tree hypotheses. For comparison, a summary of the four-gene phylogeny of Schönhuth et al. (2018) for the shiner clade is presented including the taxonomic changes proposed herein (Fig. 4). A list of all species, genera, and proposed taxonomic changes discussed below are given in Table S2.

Discussion

Alternative taxonomic and systematic classifications have been proposed for species included within the controversial genus Notropis and various purported relatives in several studies based on different characters (Swift, 1970; Coburn, 1982; Mayden, 1989; Bielawski & Gold, 2001; Mayden et al., 2006; Gidmark & Simons, 2014; Schönhuth et al., 2018). However, while several genera have been elevated from synonymy with Notropis and are monophyletic groups, consensus regarding number and composition of some of the different clades and groups of the genus Notropis has remained elusive, and classifications were considered provisional until a more comprehensive study including sufficient taxon and character sampling can produce a well supported analysis of relationships (Cashner, Piller & Bart, 2011; Gidmark & Simons, 2014; Schönhuth et al., 2018). This study provides the results on which to make some, but not all of the remaining taxonomic considerations (Table S2). In general, we made taxonomic changes if this study and that of Schönhuth et al. (2018) concur and if there was high support values in Figs. 2 and 3 (>93% bootstrap and/or a local posterior probability of 1 in the ASTRAL-II tree). Taxonomic discussion roughly follows Fig. 2. Boostrap support (BS, Fig. 2) and local posterior probability (PP, Fig. 3) are indicated in the discussion of clades.

All phylogenomic analyses (based on concatenation and species tree approaches) resolved the same patterns with strong support. The 42 species analyzed previously included in Notropis were not resolved as monophyletic; however, analyses resolved these species in several well-supported clades, and nine other well-recognized genera, together providing an alternative classification for this controversial group of shiners. This revisionary classification for Notropis supported the recognition of the five genera previously synonymized with Notropis (Alburnops, Hudsonius, Hydrophlox, Miniellus and Paranotropis) and re-allocated some species within these groups with other genera within the Shiner Clade that differed from prior studies (Mayden, 1989; Mayden et al., 2006; Gidmark & Simons, 2014). This study also highlights some interesting relationships with other closely related genera within the Shiner Clade.

Conclusions

This study provides an important first step in using phylogenomics to resolve the relationships and taxonomy of the problematic leuciscid minnows. By employing a publicly available probe set (Arcila et al., 2017), future research can include more specimens that were not sampled in this study and easily be combined with our dataset. Our phylogenies help in understanding why this group has been difficult to resolve. A phylogenomic approach provides far more characters that can break the polytomies at the base of the shiner clade that are the likely result of rapid divergence. Not only has the group been described as morphologically conserved (Gidmark & Simons, 2014), thus hampering morphological interpretations of relationships, but we would argue that the same is true genetically. We find over 88% similarity (or uninformativeness) in a dataset comprised of over 288,000 base pairs. However, we find strong agreement between this study and the four gene phylogeny provided by Schönhuth et al. (2018), indicating that dense taxonomic sampling, as done in the latter study, is also a key in resolving closely related taxa. Problems with elucidating shiner relationships have been exacerbated by studies focusing only on subsets of the shiner clade due to sampling or cost restrictions. We demonstrate the utility of the exon capture method of Arcila et al. (2017) to elucidate relationships of rapidly evolving clades, and demonstrate that the markers may be of use at the population level as well. This is important as studies utilizing the Arcila et al. (2017) markers have the potential of resolving deep and shallow relationships within a single analaysis (Hughes et al., 2021). With the continuing decrease in cost of phylogenomic methods, the demonstrable utility of the Arcila et al. (2017) markers at many phylogenetic levels, and the soon to be large number of fish taxa sampled using the Arcila et al. (2017) markers, we would encourage researchers to add to this dataset. Numerous issues remain in the taxonomy and systematics of North American leuciscids, and we will continue to add species to the analysis. This study continues the trend at subtending the shiner clade into genera, but several important clades still need to be resolved and described.

Supplemental Information