Phylogenomics, reticulation, and biogeographical history of Elaeagnaceae

The angiosperm family Elaeagnaceae comprises three genera and ca. 100 species distributed mainly in Eurasia and North America. Little family-wide phylogenetic and biogeographic research on Elaeagnaceae has been conducted, limiting the application and preservation of natural genetic resources. Here, we reconstructed a strongly supported phylogenetic framework of Elaeagnaceae to better understand inter- and intrageneric relationships, as well as the origin and biogeographical history of the family. For this purpose, we used both nuclear and plastid sequences from Hyb-Seq and genome skimming approaches to reconstruct a well-supported phylogeny and, along with current distributional data, infer historical biogeographical processes. Our phylogenetic analyses of both nuclear and plastid data strongly support the monophyly of Elaeagnaceae and each of the three genera. Elaeagnus was resolved as sister to the well-supported clade of Hippophae and Shepherdia. The intrageneric relationships of Elaeagnus and Hippophae were also well resolved. High levels of nuclear gene tree conflict and cytonuclear discordance were detected within Elaeagnus, and our analyses suggest putative ancient and recent hybridization. We inferred that Elaeagnaceae originated ca. 90.48 Ma (95% CI = 89.91–91.05 Ma), and long-distance dispersal likely played a major role in shaping its intercontinentally disjunct distribution. This work presents the most comprehensive phylogenetic framework for Elaeagnaceae to date, offers new insights into previously unresolved relationships in Elaeagnus, and provides a foundation for further studies on classification, evolution, biogeography, and conservation of Elaeagnaceae.

Methods
Data Collection:

Leaf material was gathered from multiple herbaria and the field. Total DNA extraction was performed via a modified CTAB method.

Sequencing and Processing:

Hybridization enrichment sequencing (Hyb-seq) was used to capture 100 low-copy nuclear genes. Raw sequenced reads underwent cleaning and filtering processes, which included trimming Illumina adapter sequence artifacts, discarding low-quality reads, and trimming low-quality read ends using TRIMMOMATIC v0.32. Assembly of processed nuclear reads was executed using HybPiper v1.2. As a result, 83 loci were kept for further analysis.

Genome skimming sequencing data were assembled into high-quality plastome sequence using GetOrganelle, then the plastomes were annotated with the software PGA, coding sequences of the plastome were extracted using the python script “get_annotated_regions_from_gb.py” from https://github.com/Kinggerm/PersonalUtilities.

Sequence Alignment and Cleaning:

The sequences of 83 nuclear genes were aligned using MAFFT, following which the original alignments were cleaned to reduce errors, removing gap-heavy and ambiguously aligned sites. To reduce errors in our alignments (i.e., gap-heavy and ambiguously aligned sites), the original alignment of each gene was cleaned using the pipeline of KewHybSeqWorkshop.