Metabarcoding by capture using a single COI probe (MCSP) to identify and quantify fish species in ichthyoplankton swarms

Abstract

The ability to determine the composition and relative frequencies of fish species in large ichthyoplankton swarms could have extremely important ecological applications However, this task is currently hampered by methodological limitations. We proposed a new method for Amazonian species based on hybridization capture of the COI gene DNA from a distant species (Danio rerio), absent from our study area (the Amazon basin). The COI sequence of this species is approximately equidistant from all COI of Amazonian species available. By using this sequence as probe we successfully facilitated the simultaneous identification of fish larvae belonging to the order Siluriformes and to the Characiformes represented in our ichthyoplankton samples. Species relative frequencies, estimated by the number of reads, showed almost perfect correlations with true frequencies estimated by a Sanger approach, allowing the development of a quantitative approach. We also proposed a further improvement to a previous protocol, which enables lowering the sequencing effort by 40 times. This new Metabarcoding by Capture using a Single Probe (MCSP) methodology could have important implications for ecology, fisheries management and conservation in fish biodiversity hotspots worldwide. Our approach could easily be extended to other plant and animal taxa.

Fig 1. Density distribution of genetic distance.

A. Density distribution of genetic distances (Juke Cantor 1969) between the COI barcode sequences of 444 Amazonian fish species and the single-probe (red) and siluriform probes (black line). The dashed red bar line corresponds to the most distant species in our experiment with a genetic distance of 0.26. B. Density distribution of genetic distance for the 164 siluriform species only.

Fig 2. Rarefaction curves.

The curves represent the number of species identified among the 36 species (Characiformes and Siluriformes) in two mocks as a function of the number of COI barcode assigned. With the siluriform probes (red line), at most 25 species are identified with 2,000 COI barcodes (asymptote). With the single-probe (blue line), 34 species are already identified with 2,000 COI barcodes and all 36 species are identified with 5,000 COI barcodes (asymptote).

Fig 3. Correlation of frequencies.

Correlations between Sanger frequencies of species (actual frequencies) and frequencies estimated from NGS libraries enriched with siluriform probes (blue diamond) or single-probe (red square) in mock (A) and bulk samples (B).

Fig 4. Influence of probe proximity on bias in the frequency estimation.

For the siluriform probes, we found a significant correlation between the bias in the frequency estimation and the distance to the probes (red). We did not find such bias with the single-probes (blue).