High-throughput methods for efficiently building massive phylogenies from natural history collections

Abstract

Premise: Large phylogenetic data sets have often been restricted to small numbers of loci from GenBank, and a vetted sampling-to-sequencing phylogenomic protocol scaling to thousands of species is not yet available. Here, we report a high-throughput collections-based approach that empowers researchers to explore more branches of the tree of life with numerous loci.

Methods: We developed an integrated Specimen-to-Laboratory Information Management System (SLIMS), connecting sampling and wet lab efforts with progress tracking at each stage. Using unique identifiers encoded in QR codes and a taxonomic database, a research team can sample herbarium specimens, efficiently record the sampling event, and capture specimen images. After sampling in herbaria, images are uploaded to a citizen science platform for metadata generation, and tissue samples are moved through a simple, high-throughput, plate-based herbarium DNA extraction and sequencing protocol.

Results: We applied this sampling-to-sequencing workflow to ~15,000 species, producing for the first time a data set with ~50% taxonomic representation of the "nitrogen-fixing clade" of angiosperms.

Discussion: The approach we present is appropriate at any taxonomic scale and is extensible to other collection types. The widespread use of large-scale sampling strategies repositions herbaria as accessible but largely untapped resources for broad taxonomic sampling with thousands of species.

Keywords: destructive sampling; herbaria; herbariomics; museomics; phylogenomics.

FIGURE 1

Visual representation of the workflow. Sample envelopes with QR code are placed in the center of the image to represent their centrality in tracking steps of the workflow. Arrows indicate relations to specimen images and transcription efforts, wet lab and sequencing efforts, and finally arrows repatriating results to the SLIMS database (bottom).

FIGURE 2

A screenshot of a sample selection report. The report is used to prioritize which samples to select for sequencing. There are three levels (family, genus, species) of reporting. The family and genus levels are collapsible by clicking on the buttons (“+” or “‐”) to the left. The meanings of the glyphs are given when the mouse hovers over them. Green marks indicate that the sample met the criteria for sequencing, and the black cross indicates that the sample was rejected for sampling because the yield was too low. See an interactive example at: https://rafelafrance.github.io/nitfix/assets/sample_selection.html.

FIGURE 3

Locus recovery statistics. (A) Kernel density plot of overall locus recovery; the dotted vertical line is the median (87.2%). (B) Boxplots of locus recovery for every family with at least 50 samples.