4. Discussion
Deep-sea biodiversity assessments are one of the major gaps in marine ecology and evolution and are strongly needed for international policy and management (Costa, Fanelli, Marini, Danovaro, & Aguzzi, 2020). Metagenomics (MTG) allows high-resolution inventories relying on strong phylogenetic reconstructions but is still too expensive and time consuming, as well as biased toward prokaryotes that represent most biomass in sediment samples (Danovaro, Snelgrove, & Tyler, 2014). Although metabarcoding (MTB) is relatively fast and less expensive, it suffers biases and restricts resolution to the species level.
In this study, we show that capture by hybridization (CBH) can offer a good balance between MTG and MTB, allowing the detection of a broader spectrum of phyla than MTB, and the reconstruction of full-length barcodes of up to 2000 bp allows the potential for robust phylogenetic reconstruction to improve the resolution of taxonomic assignments of poorly studied taxa (Wilson, Brandon-Mong, Gan, & Sing, 2018).
The approach applied here, initially developed for prokaryotes by Gasc and Peyret (2018), confirms that CBH can be adapted to describe bacterial communities in the marine realm. Furthermore, the first assay of transfer of this methodology to deep-sea metazoan communities proved successful, with five more phyla (upon 16) detected with CBH than MTB. This is in line with other recent CBH studies targeting more restrictive sets of metazoan phyla from bulk DNA, their conservative ethanol or eDNA samples (here freshwater and sediment; Gauthier et al., 2020; Shokralla et al., 2016; Wilcox et al., 2018). All these studies also demonstrated the ability of CBH to overcome biases due to PCR steps in MTB and enhance detection rates, leading not only to a more reliable representation of the species present in the samples but also of their relative biomasses, highlighting possibilities for quantitative analyses. Here, CBH results confirmed the improvement of biodiversity inventories allowed by direct taxonomic identification using Kraken2, which is based on exact k-mer matches and clustering of detected taxa to the lowest common ancestor (Wood & Salzberg, 2014). The sequence reads included different fragments within the 18S gene and were not merged. One must thus keep in mind this may contribute to a larger amount of taxa detected with CBH than with MTB, where sequences are merged pairwise, filtered, clustered and are all from identical positions, yet this clearly does not explain the broadest spectra of diversity with 5 (>30%) more phyla detected with CBH than MTB.
The second aim of this study was the reconstruction of full barcodes to detect a broader spectrum of diversity and to enable robust phylogenetic reconstruction with long sequences. This has been proven feasible and underlines that a much greater sequencing depth than the one used in this first assay will be required. An example of successful reconstruction of long fragments and increased resolution is the case of Porifera taxa in sample 1A. While MTB detected only 3 taxa, 20 identifications assigned to distinct taxa were recorded with CBH short, resulting in 29 long sequences with CBH long, assigned to 12 clearly identified species. The different hypervariable regions of 18S used for MTB vary in their resolution for the different phyla and nucleotide positions (Machida & Knowlton, 2012); using a full barcode thus clearly improves the number and quality of detected taxa, as exemplified in sample 1A for Porifera . The full 18S should thus be targeted to obtain the best out of the two objectives of CBH: i) increase the breadth of phyla detected and improve their taxonomic identification and ii) enable several robust phylogenetic and phylogeographic approaches to describe and unravel the biodiversity of deep-sea communities. In addition, focusing on longer fragments will limit sources of “contamination” and focus on contemporary communities by limiting the contribution of extracellular DNA from marine sediments (Torti, Lever, & Jørgensen, 2015).