Target Capture Sequencing Unravels Rubus Evolution

Abstract

Rubus (Rosaceae) comprises more than 500 species with additional commercially cultivated raspberries and blackberries. The most recent (> 100 years old) global taxonomic treatment of the genus defined 12 subgenera; two subgenera were subsequently described and some species were rearranged. Intra- and interspecific ploidy levels and hybridization make phylogenetic estimation of Rubus challenging. Our objectives were to estimate the phylogeny of 94 taxonomically and geographically diverse species and three cultivars using chloroplast DNA sequences and target capture of approximately 1,000 low copy nuclear genes; estimate divergence times between major Rubus clades; and examine the historical biogeography of species diversification. Target capture sequencing identified eight major groups within Rubus. Subgenus Orobatus and Subg. Anoplobatus were monophyletic, while other recognized subgenera were para- or polyphyletic. Multiple hybridization events likely occurred across the phylogeny at subgeneric levels, e.g., Subg. Rubus (blackberries) × Subg. Idaeobatus (raspberries) and Subg. Idaeobatus × Subg. Cylactis (Arctic berries) hybrids. The raspberry heritage within known cultivated blackberry hybrids was confirmed. The most recent common ancestor of the genus was most likely distributed in North America. Multiple distribution events occurred during the Miocene (about 20 Ma) from North America into Asia and Europe across the Bering land bridge and southward crossing the Panamanian Isthmus. Rubus species diversified greatly in Asia during the Miocene. Rubus taxonomy does not reflect phylogenetic relationships and subgeneric revision is warranted. The most recent common ancestor migrated from North America towards Asia, Europe, and Central and South America early in the Miocene then diversified. Ancestors of the genus Rubus may have migrated to Oceania by long distance bird dispersal. This phylogeny presents a roadmap for further Rubus systematics research. In conclusion, the target capture dataset provides high resolution between species though it also gave evidence of gene tree/species tree and cytonuclear discordance. Discordance may be due to hybridization or incomplete lineage sorting, rather than a lack of phylogenetic signal. This study illustrates the importance of using multiple phylogenetic methods when examining complex groups and the utility of software programs that estimate signal conflict within datasets.

Keywords: biogeography; caneberries; genetic resources; phylogenomics; plant migration; systematics; taxonomy.

Figure 1

Topological relationships between genus Rubus Groups 1–8 in phylogenetic analyses of exon or chloroplast sequences. Nodes with strong support (Bootstrap > 75 for SVDQuartets phylogenies; Posterior Probability > 0.95 for ASTRAL-II phylogenies) are marked with a star.

Figure 2

ASTRAL-II phylogeny estimated from exon sequence gene trees from all Rubus taxa. Posterior probability values (0–1) are shown to the right of each node. Branch lengths are in coalescent units and measure discordance in the underlying gene trees. Groups are labelled with colored bands. Taxa are labelled with their subgeneric classification.

Figure 3

Super network for all Rubus taxa estimated with SuperQ from exon gene trees estimated with RAxML. Colored shapes correspond to Groups 1–8. Top inset placement of R. allegheniensis and R. nepalensis due to limited sequence data for these samples (38).

Figure 4

Rubus ancestral range estimation using the DEC model for all taxa. Time scale is in millions of years. Pie charts represent relative probability of each area being the ancestral range. P, Pliocene; Q, Quaternary; N, North America (including Mexico and Guatemala); S, South America; A, Asia; E, Europe; O, Australia; Z, New Zealand. Combinations of letters indicate presence across multiple areas. Ancestral nodes for major groups are labelled numerically.

Figure 5

Maximum likelihood phylogeny estimated with RAxML for chloroplast sequences from all Rubus taxa. Bootstrap values (0-100) are shown to the right of each node. Branch lengths represent relative evolutionary change. Groups are labelled with colored bands. Taxa are labelled with their subgeneric classification sensu GRIN (2019).