A first complete phylogenomic hypothesis for diploid blueberries (Vaccinium section Cyanococcus)

Abstract

Premise: The true blueberries (Vaccinium sect. Cyanococcus; Ericaceae), endemic to North America, have been intensively studied for over a century. However, with species estimates ranging from nine to 24 and much confusion regarding species boundaries, this ecologically and economically valuable group remains inadequately understood at a basic evolutionary and taxonomic level. As a first step toward understanding the evolutionary history and taxonomy of this species complex, we present the first phylogenomic hypothesis of the known diploid blueberries.

Methods: We used flow cytometry to verify the ploidy of putative diploid taxa and a target-enrichment approach to obtain a genomic data set for phylogenetic analyses.

Results: Despite evidence of gene flow, we found that a primary phylogenetic signal is present. Monophyly for all morphospecies was recovered, with two notable exceptions: one sample of V. boreale was consistently nested in the V. myrtilloides clade and V. caesariense was nested in the V. fuscatum clade. One diploid taxon, Vaccinium pallidum, is implicated as having a homoploid hybrid origin.

Conclusions: This foundational study represents the first attempt to elucidate evolutionary relationships of the true blueberries of North America with a phylogenomic approach and sets the stage for multiple avenues of future study such as a taxonomic revision of the group, the verification of a homoploid hybrid taxon, and the study of polyploid lineages within the context of a diploid phylogeny.

Keywords: Ericaceae; HybSeq; Vaccinium; alleles; homoploid hybridization; phasing; phylogenetics; target enrichment.

Figure 1

Geographic distribution maps for diploid Cyanococcus morphospecies. Black symbols indicate populations included in our broad survey of ploidy and morphology. Yellow symbols indicate a subset of those samples sequenced and included in phylogenomic analyses.

Figure 2

Comparison of topologies recovered from concatenated and species‐tree analyses for the diploid Cyanococcus clade (highlighted in blue). Note the inconsistent placement of V. pallidum and V. elliottii populations between analyses and data sets. Sample numbers refer to the voucher table in Appendix S1. Values above branches indicate support (bootstrap or posterior probability). Values below branches indicate gene concordance factors (gCF) and site concordance factors (sCF). These are reported as gCF/sCF. Intraspecific (population‐level) support values are not shown. (A) Phylogenetic estimate from IQ‐TREE analysis of the concatenated IUPAC data set. (B) Species tree inferred from SVDquartets analysis of the concatenated IUPAC data set. (C) Species tree inferred from ASTRAL‐III analysis of the IUPAC data set. (D) Species tree inferred from ASTRAL‐III analysis of the allele data set.

Figure 3

Comparison of species trees inferred from IUPAC and allele data. In both instances, alleles and IUPAC sequences were assigned to species. Note the inconsistent placement of V. pallidum and V. elliottii between data sets. (A) Species tree inferred from ASTRAL‐III analysis of the IUPAC data set. (B) Species tree inferred from ASTRAL‐III analysis of the allele data set. Values on branches indicate local posterior probability support.

Figure 4

Evidence for the homoploid hybrid origin of Vaccinium pallidum. (A) Network inferred from the allele data set in which alleles were assigned to species. Values on hybrid edges are the estimated genomic contributions from each parent (gamma). (B) Posterior distribution of Bayesian species‐tree analysis. The lowest 5% of trees from the posterior distribution are depicted in yellow, showing alternative placement of V. pallidum sister to V. myrtilloides and V. boreale. (C) Network inferred from IUPAC data set with increased population sampling. Note that the hybrid event predates divergence of all sampled V. pallidum populations.