Unraveling historical introgression and resolving phylogenetic discord within Catostomus (Osteichthys: Catostomidae)

Abstract

Background: Porous species boundaries can be a source of conflicting hypotheses, particularly when coupled with variable data and/or methodological approaches. Their impacts can often be magnified when non-model organisms with complex histories of reticulation are investigated. One such example is the genus Catostomus (Osteichthys, Catostomidae), a freshwater fish clade with conflicting morphological and mitochondrial phylogenies. The former is hypothesized as reflecting the presence of admixed genotypes within morphologically distinct lineages, whereas the latter is interpreted as the presence of distinct morphologies that emerged multiple times through convergent evolution. We tested these hypotheses using multiple methods, to including multispecies coalescent and concatenated approaches. Patterson's D-statistic was applied to resolve potential discord, examine introgression, and test the putative hybrid origin of two species. We also applied naïve binning to explore potential effects of concatenation.

Results: We employed 14,007 loci generated from ddRAD sequencing of 184 individuals to derive the first highly supported nuclear phylogeny for Catostomus. Our phylogenomic analyses largely agreed with a morphological interpretation,with the exception of the placement of Xyrauchen texanus, which differs from both morphological and mitochondrial phylogenies. Additionally, our evaluation of the putative hybrid species C. columbianus revealed a lack introgression and instead matched the mitochondrial phylogeny. Furthermore, D-statistic tests clarified all discrepancies based solely on mitochondrial data, with agreement among topologies derived from concatenation and multispecies coalescent approaches. Extensive historic introgression was detected across six species-pairs. Potential endemism in the Virgin and Little Colorado Rivers was also apparent, and the former genus Pantosteus was derived as monophyletic, save for C. columbianus.

Conclusions: Complex reticulated histories detected herein support the hypothesis that introgression was responsible for conflicts that occurred within the mitochondrial phylogeny, and explains discrepancies found between it and previous morphological phylogenies. Additionally, the hybrid origin of C. columbianus was refuted, but with the caveat that more fine-grain sampling is still needed. Our diverse phylogenomic approaches provided largely concordant results, with naïve binning useful in exploring the single conflict. Considerable diversity was found within Catostomus across southwestern North America, with two drainages [Virgin River (UT) and Little Colorado River (AZ)] reflecting unique composition.

Keywords: Catostomus; Hybridization; Introgression; Patterson’s D-statistic; Phylogenetic incongruence; ddRAD.

Fig. 1

Map of sampling locations colored by species. Map split into two panels with the left panel containing members of the former Pantosteus and the second containing all other Catostomus and Xyrauchen samples

Fig. 2

Phylogeny of Catostomus with branch lengths derived via RAXML. Letters at nodes correspond to columns in Fig. 3 and present support values for all analyses. Nodes are collapsed according to species and level of support. Those representing operational taxonomic units (OTUs) are discussed. Dotted lines represent significant introgression events per D-statistic tests. Numbers in parentheses represent individuals at each collapsed node

Fig. 3

Nodal support values for all phylogenetic methods. Numbers to right of Astral = number of loci binned for each run where s2 = filtered data. Column headers = nodes in Fig. 2. Numbers below column headers = bootstrap support. Blue boxes with no values = 100% bootstrap support (1.0 posterior probability). Cell color: Blue = higher support, red = lower support, white(−) = no support, with cell colors varying from blue to red. Trees that supported individual binned loci for each Astral run are presented as colored cells at lower right corner of table

Fig. 4

Alternative phylogenetic hypotheses for taxa in ‘discobolus’ group, as derived by (a) concatenated SNP approaches (RAXML, MRBAYES), and (b) multispecies coalescent approach (SVDQUARTETS). Letters at nodes correspond with columns in Fig. 3 that contain support values for all analyzes

Fig. 5

Depiction of the bias on concatenation caused by introgression. Top left phylogeny (black) represents the proposed species phylogeny from morphological [35] and mitochondrial [54] data, with a red dotted line representing significant introgression detected by D-statistic tests. Resulting topologies of non-introgressed (blue) and introgressed (red) loci are shown on top. Below represents the binned loci (solid bars) and corresponding mutations (arrows above loci) that are colored according to the topology supported. Introgressed loci carry more mutations supporting the introgressed topology (red arrows) due to the long divergence between C. platyrhynchus and C. discobolus / C. virescens. As binning increases, every binned locus that contains both introgressed and non-introgressed loci will reflect the introgressed topology, resulting in more binned loci supporting the introgressed topology as binning increases. Phylogenies to the left of the loci represent the topology supported by ASTRAL for each level of binning with colors corresponding to above

Fig. 6

Map of Colorado River Basin and Bonneville Basin