The PCR-based analysis of homologous genes has become one of the most powerful approaches for species detection and identification, particularly with the recent availability of Next Generation Sequencing platforms (NGS) making it possible to identify species composition from a broad range of environmental samples. Identifying species from these samples relies on the ability to match sequences with reference barcodes for taxonomic identification. Unfortunately, most studies of environmental samples have targeted ribosomal markers, despite the fact that the mitochondrial Cytochrome c Oxidase subunit I gene (COI) is by far the most widely available sequence region in public reference libraries. This is largely because the available versatile (“universal”) COI primers target the 658 barcoding region, whose size is considered too large for many NGS applications. Moreover, traditional barcoding primers are known to be poorly conserved across some taxonomic groups. We first design a new PCR primer within the highly variable mitochondrial COI region, the “mlCOIintF” primer. We then show that this newly designed forward primer combined with the “jgHCO2198” reverse primer to target a 313 bp fragment performs well across metazoan diversity, with higher success rates than versatile primer sets traditionally used for DNA barcoding (i.e. LCO1490/HCO2198). Finally, we demonstrate how the shorter COI fragment coupled with an efficient bioinformatics pipeline can be used to characterize species diversity from environmental samples by pyrosequencing. We examine the gut contents of three species of planktivorous and benthivorous coral reef fish (family: Apogonidae and Holocentridae). After the removal of dubious COI sequences, we obtained a total of 334 prey Operational Taxonomic Units (OTUs) belonging to 14 phyla from 16 fish guts. Of these, 52.5% matched a reference barcode (>98% sequence similarity) and an additional 32% could be assigned to a higher taxonomic level using Bayesian assignment. The molecular analysis of gut contents targeting the 313 COI fragment using the newly designed mlCOIintF primer in combination with the jgHCO2198 primer offers enormous promise for metazoan metabarcoding studies. We believe that this primer set will be a valuable asset for a range of applications from large-scale biodiversity assessments to food web studies.

Analysis of environmental DNA is increasingly used to characterize ecological communities, but the effectiveness of this approach depends on the accuracy of taxonomic reference databases. The MIDORI databases, first released in 2017, were built to improve accuracy for mitochondrial metazoan (animal) sequences. MIDORI has now been significantly improved and renamed MIDORI2 (available at http://www.reference‐midori.info). Like MIDORI, MIDORI2 is built from GenBank and contains curated sequences of thirteen protein‐coding and two ribosomal RNA mitochondrial genes. Coverage has been substantially expanded to cover all eukaryotes, including fungi, green algae and land plants, other multicellular algal groups, and diverse protist lineages. MIDORI2 also now includes not only species with full binomials, but also taxa referred to by genus with species left unspecified (“sp.”). Another new feature is the updating of the databases approximately every two months with version numbers corresponding to each new GenBank release. Additional potentially erroneously annotated sequences have also been removed. Finally, the ability to export data files to BLAST+ has been added to the original ability to export preformatted data to five taxonomic assignment programs, and databases of amino acid sequences are also made available for protein‐coding genes. As a technical validation, we conducted a preliminary comparison of the performance of MIDORI2 with five taxonomic assignment programs. Results suggest that BLAST+ top hits performed better for assigning CO1 sequences than alignment‐free methods based on compositional features. Comparing MIDORI2 with two other commonly used curated databases of mitochondrial sequences, CO‐ARBitrator and BOLD, we show that MIDORI2 includes sequences from a broader range of metazoan and non‐metazoan taxa. Overall, in many contexts, MIDORI2 offers clear advantages: a higher diversity of taxa than other databases, a variety of user‐friendly features, and regular updates. MIDORI2 is particularly well‐suited for environmental DNA studies that target mitochondrial genes with broad primers.

Traditional methods of characterizing biodiversity are increasingly being supplemented and replaced by approaches based on DNA sequencing alone. These approaches commonly involve extraction and high-throughput sequencing of bulk samples from biologically complex communities or samples of environmental DNA (eDNA). In such cases, vouchers for individual organisms are rarely obtained, often unidentifiable, or unavailable. Thus, identifying these sequences typically relies on comparisons with sequences from genetic databases, particularly GenBank. While concerns have been raised about biases and inaccuracies in laboratory and analytical methods, comparatively little attention has been paid to the taxonomic reliability of GenBank itself. Here we analyze the metazoan mitochondrial sequences of GenBank using a combination of distance-based clustering and phylogenetic analysis. Because of their comparatively rapid evolutionary rates and consequent high taxonomic resolution, mitochondrial sequences represent an invaluable resource for the detection of the many small and often undescribed organisms that represent the bulk of animal diversity. We show that metazoan identifications in GenBank are surprisingly accurate, even at low taxonomic levels (likely <1% error rate at the genus level). This stands in contrast to previously voiced concerns based on limited analyses of particular groups and the fact that individual researchers currently submit annotated sequences to GenBank without significant external taxonomic validation. Our encouraging results suggest that the rapid uptake of DNA-based approaches is supported by a bioinformatic infrastructure capable of assessing both the losses to biodiversity caused by global change and the effectiveness of conservation efforts aimed at slowing or reversing these losses.

The ocean constitutes one of the vastest and richest biomes on our planet. Most recent estimations, all based on indirect approaches, suggest that there are millions of marine eukaryotic species. Moreover, a large majority of these are small (less than 1 mm), cryptic and still unknown to science. However, this knowledge gap, caused by the lack of diagnostic morphological features in small organisms and the limited sampling of the global ocean, is currently being filled, thanks to new DNA-based approaches. The molecular technique of PCR amplification of homologous gene regions combined with high-throughput sequencing, routinely used to census unculturable prokaryotes, is now also being used to characterize whole communities of marine eukaryotes. Here, we review how this methodological advancement has helped to better quantify the magnitude and patterns of marine eukaryotic diversity, with an emphasis on taxonomic groups previously largely overlooked. We then discuss obstacles remaining to achieve a global understanding of marine eukaryotic diversity. In particular, we argue that 18S variable regions do not provide sufficient taxonomic resolution to census marine life, and suggest combining broad eukaryotic surveys targeting the 18S rRNA region with more taxon-focused analyses of hypervariable regions to improve our understanding of the diversity of species, the functional units of marine ecosystems. This article is part of the themed issue ‘From DNA barcodes to biomes’.

DNA metabarcoding, the PCR-based profiling of natural communities, is becoming the method of choice for biodiversity monitoring because it circumvents some of the limitations inherent to traditional ecological surveys. However, potential sources of bias that can affect the reproducibility of this method remain to be quantified. The interpretation of differences in patterns of sequence abundance and the ecological relevance of rare sequences remain particularly uncertain. Here we used one artificial mock community to explore the significance of abundance patterns and disentangle the effects of two potential biases on data reproducibility: indexed PCR primers and random sampling during Illumina MiSeq sequencing. We amplified a short fragment of the mitochondrial Cytochrome c Oxidase Subunit I (COI) for a single mock sample containing equimolar amounts of total genomic DNA from 34 marine invertebrates belonging to six phyla. We used seven indexed broad-range primers and sequenced the resulting library on two consecutive Illumina MiSeq runs. The total number of Operational Taxonomic Units (OTUs) was ∼4 times higher than expected based on the composition of the mock sample. Moreover, the total number of reads for the 34 components of the mock sample differed by up to three orders of magnitude. However, 79 out of 86 of the unexpected OTUs were represented by <10 sequences that did not appear consistently across replicates. Our data suggest that random sampling of rare OTUs (e.g., small associated fauna such as parasites) accounted for most of variation in OTU presence–absence, whereas biases associated with indexed PCRs accounted for a larger amount of variation in relative abundance patterns. These results suggest that random sampling during sequencing leads to the low reproducibility of rare OTUs. We suggest that the strategy for handling rare OTUs should depend on the objectives of the study. Systematic removal of rare OTUs may avoid inflating diversity based on commonβdescriptors but will exclude positive records of taxa that are functionally important. Our results further reinforce the need for technical replicates (parallel PCR and sequencing from the same sample) in metabarcoding experimental designs. Data reproducibility should be determined empirically as it will depend upon the sequencing depth, the type of sample, the sequence analysis pipeline, and the number of replicates. Moreover, estimating relative biomasses or abundances based on read counts remains elusive at the OTU level.

The significance of symbioses between eukaryotic hosts and microbes extends from the organismal to the ecosystem level and underpins the health of Earth’s most threatened marine ecosystems. Despite rapid growth in research on host-associated microbes, from individual microbial symbionts to host-associated consortia of significantly relevant taxa, little is known about their interactions with the vast majority of marine host species. We outline research priorities to strengthen our current knowledge of host–microbiome interactions and how they shape marine ecosystems. We argue that such advances in research will help predict responses of species, communities, and ecosystems to stressors driven by human activity and inform future management strategies.