Skip to main content
Browse Subject Areas

Click through the PLOS taxonomy to find articles in your field.

For more information about PLOS Subject Areas, click here.

  • Loading metrics

Instances of erroneous DNA barcoding of metazoan invertebrates: Are universal cox1 gene primers too “universal”?

  • Monika Mioduchowska ,

    Roles Conceptualization, Data curation, Formal analysis, Funding acquisition, Investigation, Methodology, Visualization, Writing – original draft

    Affiliation Department of Genetics and Biosystematics, Faculty of Biology, University of Gdansk, Gdansk, Poland

  • Michał Jan Czyż,

    Roles Conceptualization, Data curation, Funding acquisition, Investigation, Methodology, Writing – original draft

    Affiliation Research Centre of Quarantine, Invasive and Genetically Modified Organisms, Institute of Plant Protection – National Research Institute, Poznań, Poland

  • Bartłomiej Gołdyn,

    Roles Conceptualization, Funding acquisition, Investigation, Methodology, Writing – original draft

    Affiliation Department of General Zoology, Institute of Environmental Biology, Faculty of Biology, Adam Mickiewicz University, Poznań, Poland

  • Jarosław Kur,

    Roles Conceptualization, Investigation, Writing – original draft

    Affiliation Institute of Nature Conservation, Polish Academy of Science, Kraków, Poland

  • Jerzy Sell

    Roles Conceptualization, Supervision, Visualization, Writing – review & editing

    Affiliation Department of Genetics and Biosystematics, Faculty of Biology, University of Gdansk, Gdansk, Poland


The cytochrome c oxidase subunit I (cox1) gene is the main mitochondrial molecular marker playing a pivotal role in phylogenetic research and is a crucial barcode sequence. Folmer’s “universal” primers designed to amplify this gene in metazoan invertebrates allowed quick and easy barcode and phylogenetic analysis. On the other hand, the increase in the number of studies on barcoding leads to more frequent publishing of incorrect sequences, due to amplification of non-target taxa, and insufficient analysis of the obtained sequences. Consequently, some sequences deposited in genetic databases are incorrectly described as obtained from invertebrates, while being in fact bacterial sequences. In our study, in which we used Folmer’s primers to amplify COI sequences of the crustacean fairy shrimp Branchipus schaefferi (Fischer 1834), we also obtained COI sequences of microbial contaminants from Aeromonas sp. However, when we searched the GenBank database for sequences closely matching these contaminations we found entries described as representatives of Gastrotricha and Mollusca. When these entries were compared with other sequences bearing the same names in the database, the genetic distance between the incorrect and correct sequences amplified from the same species was c.a. 65%. Although the responsibility for the correct molecular identification of species rests on researchers, the errors found in already published sequences data have not been re-evaluated so far. On the basis of the standard sampling technique we have estimated with 95% probability that the chances of finding incorrectly described metazoan sequences in the GenBank depend on the systematic group, and variety from less than 1% (Mollusca and Arthropoda) up to 6.9% (Gastrotricha). Consequently, the increasing popularity of DNA barcoding and metabarcoding analysis may lead to overestimation of species diversity. Finally, the study also discusses the sources of the problems with amplification of non-target sequences.


In 1994 Folmer et al. [1] described “universal” DNA primers: forward LCO1490 and reverse HCO2198 (commonly referred to as the “Folmer primers”), to amplify the most conservative protein-coding gene in metazoan invertebrates—the mitochondrial cytochrome c oxidase subunit I (cox1) gene fragment of the length approximately 700 bp. These primers were successfully used in amplifying the cox1 gene fragment for more than 80 invertebrate species, proving their universality across diverse metazoan taxa [2]. This in turn resulted in the cox1 gene, due to its robustness and reliability, being accepted as the standard taxon barcode for most animals (e.g., [3]). Moreover, a high rate of substitution in the third codon position of this molecular marker allowed to apply it for phylogeny both at the population and species level [4]. In consequence, Folmer primers appear to be irreplaceable in polymerase chain reaction (PCR) for phylogenetic studies of diverse invertebrate organisms, and they make phylogenetic analyses at the species (and also higher) taxonomic level less complicated. Moreover, “barcoding” COI sequences ( have also been widely used to determine unknown species’ identity, and the records are given the keyword BARCODE in the GenBank. Unknown COI sequences are matched with the sequences deposited in the GenBank, and identification of species is frequently carried out by referring the database for comparison [5]. More recently, the utility of these barcode primers have also been shown in application to metabarcoding analyses (e.g., [6]).

In the period between 1994 and 2016 the number of scientific articles referring to Folmer primers reached nearly 17000 in Google Scholar. Moreover, the citation count for the original Folmer et al. [1] paper in the databases at Web of Science crossed 7600 and this number is constantly increasing. In order to maintain the broad taxonomic utility of COI sequences, a high level of degeneracy in updated Folmer primers is necessary (e.g., [7]). Nonetheless, it has been already shown that in numerous analyses of COI sequences these primers mismatch many metazoans [810]. To overcome the limitations of the Folmer primers, alternative primers have been developed [11]. In 2013 the Barcode of Life Database (the BOLDSystems) revealed c.a. 418 different primers targeting COI sequences for various taxa. Besides this, some researchers also applied considerably reduced primer annealing temperatures to resolve these amplification problems [12].

Nevertheless, in the DNA barcode approach, species identification could be carried out without any a priori knowledge about the investigated samples. This approach might be used for DNA barcode recovery from mixed environmental probes or identification of problematic specimens with ambiguous morphology—especially when the material is damaged or only parts of the body are present. Thus, in those cases universal primers, e.g. Folmer primers, seem to remain a reasonable solution. However, they can also co-amplify the nuclear mitochondrial pseudogenes—numts or “COI-like” sequences [13, 14] and microorganisms [15]. Finally, it can lead to overestimation of species diversity, as well as global biodiversity patterns [14]. Siddall et al. [15] analysed homologous priming site sequences obtained from prokaryotic and eukaryotic whole genome data, and they indicated the presence of targeted Folmer primer sequences. What is more, they showed that several genomes of bacterial species exhibited more contiguously matched nucleotides for the reverse primer than most animals.

The phenomenon of contaminant amplification of non-target taxa is quite common and mismatched sequences are still published. Using the universal Folmer primers to amplify the cox1 gene of crustacean fairy shrimp Branchipus schaefferi (Fischer 1834) we also obtained bacterial sequences which were described as metazoan in GenBank. This finding encouraged us to examine more closely the cases of mismatched and mislabelled sequences stored in the available on-line databases. In this case we focused on (i) analysis of the bacterial sequences (obtained with the universal Folmer primers) deposited in the GenBank as invertebrate sequences and their influence on DNA barcoding (and metabarcoding) of metazoan species, (ii) estimating the proportion of incorrectly flagged sequences deposited in the GenBank using the random sampling procedure. In addition, we discuss the problems with the amplification of non-target sequences and the possibilities of preventing incorrectly flagged sequences in further analyses.

Materials and methods

DNA extraction, PCR amplification, and DNA sequencing

Samples of muscular tissues from 25 specimens of Branchipus schaefferi (Crustacea: Anostraca) collected in 2014 from a temporary water body located in a military training ground in western Poland (Poznań-Biedrusko; ca. 52°29'N, 16°51'E) were immediately isolated from the fresh thorax after the sampling. The total DNA was extracted from the samples in accordance with the protocol of the Biotrace Genomic Extraction GPB Mini Kit (GenoPlast) (see also [16]).

Until now, only the 18S rRNA gene has been used in genetic analyses of several specimens of B. schaefferi [16, 17]. Thus, molecular identification of species was conducted with these barcode sequences in accordance with the procedure described by Mioduchowska et al. [16].

The fragment of mtDNA spanning the cox1 gene was amplified with the universal primers HCO2198 and LCO1490 [1]. The polymerase chain reactions (PCRs) were performed in a 20 μL volume containing 0.8× JumpStart Taq ReadyMix (1 U of JumpStart Taq DNA polymerase, 4 mM Tris–HCl, 20 mM KCl, 0.6 mM MgCl2, 0.08 mM of dNTP; Sigma-Aldrich, Germany), 0.4 μM of forward and reverse Folmer’s primers, and about 100 ng of DNA. The cox1 gene fragments were amplified under the following conditions: the initial denaturation at 94°C for 5 min was followed by 30 cycles of 94°C for 30 s, 44°C for 1 min (gradient PCR amplification ranging from 42 up to 52°C was applied to determine the optimal annealing temperatures), and 72°C for 1 min and at the end at 72°C for 5 min. The reactions were conducted in a Biometra TProfessional thermocycler. The products of amplification were separated by 1% agarose gel electrophoresis in a 1x SB buffer and then visualized with ethidium bromide in UV light. The PCR products were purified with alkaline phosphatase and exonuclease I (Thermo Scientific, USA) treatment in accordance with the protocols and sequenced with the BigDye terminator cycle sequencing method.

The obtained sequences have been deposited in the GenBank under accession numbers given in Table 1.

Table 1. The list of correctly and incorrectly flagged sequences of eukaryotic and microbial cox1 gene fragment applied in our study.

DNA sequence analysis and phylogenetic tree reconstruction

The obtained sequences were aligned manually using BioEdit 5.0.9 [18] and consensus sequences were created. The alignments were prepared in ClustalX 1.81 [19] with the default settings for the COI marker. The BLAST application (Basic Local Alignment Search Tool; [20]) was used for the homology search and to browse the sequences deposited in the NCBI database. To check the indels and internal stop codons, the sequences of the cox1 gene fragments were translated into amino acid sequences with the EMBOSS-TRANSEQ application [21].

The phylogenetic analyses of the obtained sequences, which closely matched the query sequences downloaded from the GenBank, were carried out with the MEGA software package (version 6.06; [22]). Some of these sequences were found to be described in the GenBank as belonging to taxa that are phylogenetically distant from Crustacea, mislabelled as marine gastropod Patelloida striata (accession: AB161589.1; [23]) and the marine gastrotrich Tetranchyroderma sp. 3 [24]. Thus, we performed another search for other sequences with similar descriptions (or closely related) taxa, and we found entries for Patelloida striata (accession: AB238524.1; [25]) and Tetranchyroderma pachysomum [24]. We included them together with the closely matched query sequences in the phylogenetic reconstruction tree. The genetic distances were calculated with the MEGA software package (version 6.06; [22]) as p-distance values of Nei & Kumar [26]. The evolutionary model was established with jModelTest 2.1.4 [27] with the assumptions of the Akaike Information Criterion (AIC) [28]. The General Time Reversible (GTR) model with the gamma-distributed rate parameter was chosen as the best-fit and the Maximum Likelihood (ML) method was also applied. For the assessment of the node support 1000 bootstrap iterations were performed.

In the current paper we use the term “correct sequences” for sequences obtained from target species and “incorrect sequences” for those acquired from non-target species.

Testing the probability of detecting incorrectly flagged sequences deposited in the GenBank

We checked the NCBI nucleotide database for all COI sequences of Gastrotricha, Mollusca, and Arthropoda to estimate the percentage of incorrectly identified sequences. These analyses were performed for taxonomic groups in which sequence errors have been already observed. Since there is no obvious consensus on the metadata definition and gene vocabulary in the NCBI database (e.g. some of the deposited sequences had only “cytochrome c oxidase subunit I” in definition part of metadata, other entries had CO1 acronym instead of COI), we used following query with proper conditions to obtain as many as possible sequences related only to cytochrome c oxidase subunit I: where for <taxon name> we substituted: Gastrotricha, Mollusca or Arthropoda. According to the binomial distribution, the probability of detecting at least one misidentified record in a sample is given by: where P is the probability of detection, p is the proportion of misidentified records in a population, and n is the sample size (e.g., [29]). Thus, we put forward a hypothesis that at least 1% of sequences deposited for each investigated taxa are misidentified. To calculate the sample size the above equation needs to be transformed into:

Using 0.95 as the probability of detection and 0.01 as the ratio of misidentified sequences in the population, the calculated sample size equals 300. In other words, there is a 95% chance of finding at least 1 misidentified sequence in a sample of size 300, assuming that the true ratio of misidentified sequences in the whole population is 0.01. Taking these calculations into account, we decided to randomly sample 300 sequences for each taxon. In the case of Gastrotricha the total number of IDs was 363, and hence we decided to use all of them in the analyses to calculate the parameter for this population. Then, we downloaded all sequences related to the sampled IDs. This procedure was performed in R 3.4.1 [30] with package rentrez 1.1.0 [31]. For the obtained results we calculated 95% confidence intervals using the exact method with R package binom 1.1–1 [32]. The comparison with the GenBank records was carried out with BLAST to identify the sequences that were homologous to the obtained cox1 gene fragment.

Additionally, a stochastic calculation was performed to estimate the 5th, 50th, and 95th percentile of the potential number of misidentified sequences in the database (absolute values). It was calculated by multiplying 10000 random probabilities from the beta distribution Beta (α, β) by the number of all obtained sequences for Arthropods and Molluscs separately. The shapes of the beta distribution were calculated as follows: α = k +1 and β = n − k + 1, where k is the number of positive results and n is the sample size. We have also performed a simple stochastic simulation of the number of erroneous sequences that might be sampled from the GenBank database for Arthropods and Molluscs with 3 different sample sizes: 300, 1000 and 10000. To perform this calculations we ran 10000 iterations over binomial distributions B (n, p), where n is the sample size and p is the probability from the previous Beta (α, β) iterations. We have presented the 5th, 50th and 95th percentile for all results. The above calculations were performed in R 3.4.1 [30].


The examined specimens were classified as Branchipus schaefferi on the basis of morphology as well as 18S rRNA sequences (GenBank accession number: KU645889). The Folmer primers used in the analysis amplified not only cox1 of B. schaefferi but also of bacterial origin. There were no differences in the size and quality of the obtained PCR products. Despite the high quantity and quality of the PCR products visualized on agarose gel, the sequencing results were poor due to the presence of mixed amplicons. Only 7 out of 25 analysed samples were successfully sequenced, while the other sequence chromatograms showed the effect of heteroplasmy [33]. Finally, we obtained two COI sequences (one haplotype) of B. schaefferi and five non-crustacean sequences (three haplotypes). These were further analysed as to the expected matches and nearest neighbours (BLAST searches), and the analysis has revealed non-target amplification of microbial contaminants from Aeromonas sp. No stop codons and indels have been observed for these seven sequences, and the translation was successfully carried out with -2nd reading frame. We have also managed to overcome the problems with the co-amplification by designing species-specific primers for weak sequences derived from mixed products. In this way we have obtained only target species—the COI sequences of B. schaefferi–and we have acquired non-target species with another pair of species-specific primers—the COI of Aeromonas sp. (Mioduchowska et al., in preparation).

We also reconstructed the phylogenetic relationships between our sequences and the nearest neighbours deposited in the GenBank. The Maximum Likelihood analyses (Fig 1) placed the obtained sequences among general clades of Metazoa (Arthropoda above all) and bacteria (all of them were Proteobacteria). Performed bootstrap analysis shows high support values: 100% support for a Eukaryota/Prokaryota bipartition, 98% support for Metazoa genomic COI, and 100% support for bacteria.

Fig 1. Maximum Likelihood tree showing evolutionary relatedness of eukaryotic and prokaryotic COI barcoding sequences.

The bootstrap values are shown at the nodes. The used colors of asterisk indicate both correctly and incorrectly described sequences for the same species. In the case of the sequences obtained in our study for B. schaefferi the sequences were correctly labeled and put into the GenBank. More information about the applied sequences is provided in Table 1.

Surprisingly, two sequences of the nearest neighbours classified in the analysis as bacterial clades had labels of clearly metazoan species in the GeneBank database. These sequences were described as gastropod Patelloida striata ([23]; AB161589.1, BOLD: ACY9706) and gastrotrichan Tetrachyroderma sp. ([24]; JF432035.1). However, other sequences assigned to these species are also present in the GeneBank, and these in turn match the Metazoa clade, and they have been placed on the phylogenetic tree in positions fairly fitting to their phylogenetic relationships. These were one sequence of P. striata ([25]; AB238524.1, BOLD: AAI4206) and one sequence of Tetrachyroderma pachysomum ([24]; JF432029.1) (Fig 1).

The average genetic distance (p-distance) between the COI sequences of Metazoa species and the analogous COI fragment derived from whole-genome sequencing of bacteria was 65%. The mean p-distance values between the sequences in the clades ranged between 2% and 67%–the values for the Metazoan clade and between 2% and 27%–values for the bacteria clade. Furthermore, the genetic distance between the incorrect and correct sequences of the same species was as follows: Branchipus schaefferi– 62%, Patelloida striata– 64%, and Tetrachyroderma sp.– 68%.

According to our query, the number of all Mollusca and Arthropoda COI sequences deposited in the GenBank was 109502 and 2008162 respectively. In the case of Gastrotricha we analysed all 363 sequences deposited in the GenBank. In our random sample of sequences, we have observed errors in Gastrotricha and Mollusca. However, these errors not always were related to the description of the metazoan sequence as a bacterium, but some mistakes have been observed also among the metazoan species. Among the sequences of Mollusca we identified one incorrectly flagged sequence, which had been classified as a metazoan sequence (GenBank accession number: JN100021.1; deposed in 2011 by Valenzuela and co-authors; there is no information indicating exactly which primers were used). In the case of Gastrotricha two sequences were of bacterial origin (GenBank accession numbers: JF432035.1, JF432024.1; [24]; Folmer primers were used), one was of fungal origin (GenBank accession number: KP878838.1; deposited in 2015 by Atherton; Uro3HF and Uro3HR primers were used), and 22 sequences came from other metazoan species (nonetheless, some of the incorrectly flagged sequences could be numts, see supplementary data S1 Table; [24, 3439]).

On the basis of our analyses we can conclude with 95% probability that less than 1% of the sequences among Arthropoda and 1% of Mollusca sequences are erroneously classified. When expressed in terms of 95% confidence intervals, the ratio of misidentified sequences in Arthropoda is close to 0, with lower and upper boundary 0 and 0.012 respectively, while in Mollusca the ratio is 0.003 with lower and upper boundary > 0.001 and 0.018 respectively. In other words, approximately 4601 (5th percentile: 348, 95th: 19890) of the erroneous sequences are deposited in the Arthropoda group and 615 (5th percentile: 132 and 95th: 1673) of the erroneous sequences are deposited for Mollusca. The results of the simulation show that when we sampled 300, 1000 or 10000 sequences deposited in the GenBank for Arthropoda, we found 1 (0, 4), 2 (0, 11) or 24 (1, 101) erroneous sequences, respectively. For Molluscs the values were 1 (0, 6), 5 (0, 17), 56 (11, 154)–for sample size 300, 1000 and 10000, respectively (5th and 95th percentile values in brackets). There are also 25 incorrect sequences deposited in the GenBank for Gastrotricha, roughly 0.6% of the sequences belong to bacteria, 0.3% to fungi, and surprisingly, 6.06% of them belong to other metazoan species and probably to numts. The incorrect sequences represent approximately 6.9% of all deposited sequences in this group of organisms.

The sequences presented in Table 1 that were described as correct (originating from the target species), as well as those incorrect (originating from the non-target species) were amplified with the Folmer primers. Similarly, almost all amplifications of the incorrect Gastrotricha that are included in S1 Table were carried out with these universal primers. However, there is no information about which primers were used to amplify the cox1 gene fragment of the incorrectly flagged four Urodasys poculostylis sequences (Atherton 2015, unpublished; S1 Table). Finally, newly designed internal primers, combined with the modified Folmer primers, were used for one cox1 gene sequence of Mesodasys sp. ([36]; S1 Table).


The metazoan DNA barcoding with the Folmer primers enables quick and easy species identification, detection and discovery. However, some authors (e.g., [5153]) have showed that the Folmer primers, one of the most frequently used research tools for COI amplification, are not so “universal” for some taxa, though at the same time they seem too “universal” for non-target bacterial genomes (e.g., [15]). Moreover, these primers are poorly conserved across some metazoan representatives, e.g. gastropods [8], nematodes [9, 54], and echinoderms [10].

The universality of the Folmer DNA primers was discussed by Sharma and Kobayashi [55], who showed some weaknesses of this method. First of all, the limitations in the universality are related to the original design procedure based on three sense and six anti-sense strands and comparison of the cox1 gene conserved regions across the 15 taxa. During the analysis of 130,843 variations in the primer region of COI sequences from the kingdom Animalia they observed that the reverse primer was truly conserved, but the forward primer contained only four conserved regions and approximately 50% conserved regions for vertebrates, fungi, and ascidians. The problems associated with one or both Folmer primers (especially the high nucleotide variability in the forward primer) sometimes cause a relatively low amplification rate reported for certain taxonomic groups. This has led to designing species-specific primers (e.g., [56]).

The correct species identification based on DNA barcoding might also be hindered due to numt co-amplification. Moulton et al. [57] showed that even increased primer specificity was not sufficient to eliminate completely numt co-amplification, which was contradictory to the suggestions of Song et al. [14]. Moreover, Moulton et al. [57] showed that many numts do not have indels and stop codons, which is a serious problem for DNA barcoding. In addition, these findings make it more difficult to distinguish numt sequences from mitochondrial orthologues. Furthermore, Siddall et al. [15] compared homologous priming site sequences obtained from complete prokaryotic genome data and mitochondrial eukaryotic genome data, and they showed that the first 10 nucleotide positions of the forward Folmer primer demonstrated variability across a wide range of eukaryotes and prokaryotes. The reverse Folmer primer showed variation only in the third codon positions and all eukaryotic nucleotide variants were also present among the prokaryotes. Furthermore, some gamma-proteobacteria exhibited more contiguously matched nucleotides for the reverse Folmer primer than from the metazoan taxa.

We have also observed amplification and sequencing of non-target bacterial templates in our analysis. We found that the bacterial sequences amplified from Branchipus schaefferi were in 88% identical (query coverage was 100% and E-value was 0.0) to Aeromonas sp. sequences. The source of the bacterial DNA in our samples could derived from the gut contents or the surface of the thorax tissues (see also [7]). Nonetheless, this remains a speculation and more detailed research, such as metagenomic analysis is required. We were very careful during the sampling of B. schaefferi tissues and we applied laboratory procedures to minimize anomalous results and to exclude bacterial contamination or mixed templates (e.g., laboratory errors). However, anaerobic Aeromonas occur quite frequently and autochthonously in aquatic environments [58]. Besides this, Aeromonas sp. is known to be a fish pathogen [59] and commonly occur in the Medicinal leech Hirudo medicinalis in symbiosis [60, 61] or in nematodes for which it is toxic [62].

It is intriguing that bacterial COI sequences were found also in other metazoa species, and they have been erroneously deposited in a public database like the GenBank, e.g. as Patelloida striata [23] and Tetranchyroderma sp.3 [24]. Interestingly, the same authors described both correctly [24, 25] and incorrectly flagged sequences [23, 24] but they were not aware of the fact that those sequences had been erroneously labelled. Unfortunately, we cannot say anything about the potential sources of the bacterial DNA in these samples. However, in the case of marine gastrotrich Tetranchyroderma sp., as well as the marine limpet mollusc Patelloida striata, it is not surprising that bacterial sequences from these species match to the marine gamma-proteobacteria clade. The prokaryotic COI sequences were in 83% identical (query coverage was 100% and E-value was 1e-158) to those of Endozoicomonas montipore—aerobic bacterium isolated from the encrusting pore coral Montipora aequituberculata collected from seawater [47, 63]. Species of Endozoicomonas were frequently found also in various marine invertebrates, e.g., the sea slug [64], sponge [65], and the comb pen shell [66]. Due to the fact that Endozoicomonas have positive association with healthy individuals (especially in corals), these bacteria are considered as an essential affiliate of the host holobiont [47].

In most cases not only the primer sequences but also the annealing temperatures typically used for DNA barcoding are crucial for the successful amplification of the target taxon. Given the fact that the Folmer primers are such a poor match to many metazoan species, associating surface bacteria as well as gut bacteria, could be also preferentially amplified under low annealing temperatures [15]. Consequently, too low annealing temperatures are the main reason of contaminant bacterial COI sequence amplification. In the case of our B. schaefferi sequencing we applied an annealing profile of 44°C that preferentially amplified Aeromonas species. As a result, the highest annealing temperature used in the analysis generated weak or no PCR products. Many barcoding sequences deposited in the GenBank have in fact incorrect metazoan labels, as they are products of bacterial contaminants (e.g., [15]). In the case of the correct COI sequences of Patelloida striata and Tetranchyroderma pachysomum the amplifications were carried out with an annealing profile of 45–50°C and 46°C, respectively [24, 25]. Notwithstanding, the incorrectly flagged COI sequences of these metazoan species were amplified with an annealing profile as follows: 54°C for Patelloida striata and 46°C for Tetranchyroderma sp. [23, 24]. The obtained results are consistent with the observations of Siddall et al. [15], who claim that stringent annealing temperatures should not be expected to remediate the problem of bacterial contaminant amplification of COI sequences, and poor primer complementarity to metazoan genomes necessitates liberal amplification regimes.

According to Kress & Erickson [67] there are several traits that molecular markers should have to serve as efficient DNA barcode sequences, i.e., they should correctly discriminate the examined species and they should show a range of taxonomic diversity and the presence of universal primer pairs for PCR amplification. Moreover, the relative rate of evolution of mitochondria, as well as nuclear genes, which vary in different organisms, should be also considered [68]. In addition to COI barcode sequences, some other potential barcoding markers exist for various organisms, e.g., the 18S rRNA gene (e.g., [69]). Flanking regions of rRNA gene sequences can be used for universal primer amplification due to their high conservation. Moreover, the 18S rRNA gene is widely regarded as the primary candidate for reconstruction of metazoan relationships [70]. Recently, it has also been shown that this conservative molecular marker is a suitable tool for molecular identification of B. schaefferi [16].

At present, DNA-based community analyses have offered some alternatives to the traditional methods of species identification, and metabarcoding is one of the most promising techniques to understand taxon diversity. This method enables species detection from a bulk of samples, e.g., environmental samples, gut or faecal contents [36], and it might be a good solution to the problems with proper species identification in the case of cryptic species (e.g., [71]) or poor species-specificity of cyst morphology and biometry, which is used when no active specimens are present (e.g., [72]). Metabarcoding is based on PCR amplification followed by deep sequencing of homologous gene regions. The resulting barcode sequences are then compared to homologous sequences deposited in databases (e.g. the GenBank, BOLDSystems, or created by research groups with special reference libraries) for multiple species identification in complex samples. However, due to the fact that many sequences have been incorrectly annotated in the reference libraries, this method is also prone to generate errors in species identification. Erroneous data that may otherwise potentially lead to misidentification of species should be removed following the first essential step (after obtaining NGS data), where sequence errors can be removed during quality filtering. Nevertheless, species misidentification can also occur during the next step, when the clustering method is applied into operational taxonomic units (OTUs), if only the obtained sequences match the incorrectly described references deposited in the DNA barcodes library (which, as we showed in this study, can be one of the problems with the use of the GenBank database) [73].

It seems that using BLAST search engine to compare homologous sequences deposited in the GenBank is sufficient for correct species identification at the molecular level. Nevertheless, there is a need to develop new algorithms and procedures to find the incorrectly labelled sequences in a more efficient way. For example, comparison of p-distance values calculated between new sequences and the match entries in the GenBank database should be helpful to filter out non-target bacterial genomes contaminants. In our study the values of the p-distance between the correct and incorrect sequences was ca. 65%, and the mean range of such values for the species should be ca. 2% (see also supplementary data S1 Table).

Although the ratio of the metazoan incorrectly flagged sequences annotated in the GenBank depends on the systematic group and varies between less than 1% and 6.96%, the probability of duplicating these errors in barcoding and metabarcoding analyses is high. Due to the small size of some metazoans, e.g. Gastrotricha, and the presence of microsporidian parasites [74] and unknown type of interaction between these organisms and protists [75], we expected that more incorrectly flagged sequences would be observed within these taxonomic groups. The results of the present study confirm our preliminary assumptions. Nevertheless, although we have not found bacterial sequences in Mollusca and Arthropoda, it has been shown that some mistakes are occurring also in these groups (this study, [15]). The ratio of detecting incorrectly described sequences of Gastrotricha originating from other metazoan taxa or probably numts was very high (6.06%). Thus, this is a serious drawback and therefore molecular taxonomy of these groups should be verified by specialists in the field with more attention.

The main problem which arises when researchers obtain a good quality sequence is that it can mislead to believe that it comes from the target species. Moreover, verification of the correctly flagged sequences with the BLAST can be difficult in the case of a new species, for which there are no reference sequences in the available databases. One should also remember that there is only one sequence of the cox1 deposited in NCBI databases for most of the Gastrotricha species described in S1 Table (10 out of 17 species), which is another difficulty that one can encounter in verification of taxonomic identification. We have identified the incorrectly described Gastrotrich sequences that come from other metazoan species. We consider these sequences to be true mitochondrial sequences due to the translation capacity and lack of gaps. Some of these sequences were highly homologous to the orthologous mtDNA sequence of Archaphanostoma ylvae and Palaemon floridanus with 99–100% query coverage in the BLAST search results, and the E value from 2e-167 to 0.0, and values of p-distance ranging between 0.000 and 0.220 (S1 Table, No 16–22).

The analysis has also revealed that some incorrectly flagged Gastrotricha sequences with a significant match to non-target metazoan species were marked as “cytochrome c oxidase subunit I like (COI) gene” (S1 Table, No 4, 5, 7–9); some sequences contained internal stop codons, disrupting the open reading frame of the amino acid sequence (S1 Table, No 5, 7, 9), and there were also sequences with gaps (S1 Table, No 1–15). Thus, it seems that the these sequences probably come from nuclear genes (numts). Similarly, Todalo et al. [24] described another c oxidase subunit I-like gene of Gastrotricha (accession: JF432020, JF432021, JF432022, JF432025, JF432034). These sequences do not conform to the general invertebrate mitochondrial sequence pattern, as well as the genetic code of other organisms deposited in the available databases; the sequences with gaps and without translation to their corresponding amino acid sequence. The authors also suggest that the ostensible COI sequences could be numts or one or more modifications of the mitochondrial genetic code within Gastrotricha. Although numts are highly divergent from true mitochondrial sequences, these sequences were significantly similar to other Gastrotricha sequences found in the NCBI database. Hence, it might seem that they come from the target species, and this approach is also a major problem.

The lack of a unified coding system for the already identified sequences makes finding the relevant data difficult. During the procedure of finding the proper query statement for the NCBI database we managed to find many sequences related to one gene (cox1) with a few different formats of description (e.g. "COI" vs. "CO1", "cytochrome c oxidase" vs. "cytochrome oxidase"). This lack of uniformity and consistency results in difficulties in formulating the right query statement which would give the desired results. Therefore, we postulate that there is still an urgent need of unifying the description format of the sequences deposited in the GenBank.

There is a number of databases which are useful in the Linnaean system of animal taxonomy (e.g.; The Barcode of Life Data System ( is currently the most popular integrated bioinformatic platform, which supports the analytical pathway ranging from specimen collection to tightly validated barcode library. It contains almost 6,000,000 barcodes, providing a direct link between the specimen data, digital images, taxonomy and assigning sequences to the BINs (Barcode Index Numbers) (e.g., [76, 77]). Furthermore, the BOLD data analysis tools enable screening the submitted sequences for some common contaminants and endosymbionts, such as Wolbachia, and flagging them as contaminants. However, it is quite striking that incorrect sequences were also found in the BOLD database. For example, there are both correct [25] and incorrect sequences [23] for Patelloida striata in this database. This raises a question whether it is possible to reduce considerably the number of such mistakes in databases even with strict measures and procedures. Nevertheless, the barcoding method should not be impugned, but a more careful approach is needed in those initiatives. There is no doubt that this will be the most valuable approach in identification of species in the future.

We are also determined to continue the research into the identification of incorrectly described sequences. A good solution to the problems associated with errors in species identification could be an on-line application, which would be based on integrative taxonomy (a combination of both morphology and DNA data). A preliminary concept of integrative taxonomy as a combination of both morphology and DNA data in an on-line application has been already proposed by Kur et al. [78], and similar research is in progress in our laboratories. A similar approach in developing a standardized on-line data repository has been adopted in the case of tardigrade taxonomy (, [79]). The digitization of morphological keys will be a novelty in our application when compared to the BOLD system (which despite numerous advantages is not based on the integrated taxonomy, but it is only an integrated bioinformatic tool, compare e.g., [80]). The application will allow species identification at the morphological level on the basis of a diagnosis of a number of morphological features. Moreover, the integration of comparative morphological and molecular taxonomy will also allow precise species identification and delimitation. We are convinced that the conceptual multi-platform web application, based on a combination of molecular and morphological data, will be available for both specialists in this field of research and broad public.


Although the responsibility for the right molecular identification of species rests upon researchers, there is no doubt that many sequences deposited in the GenBank should be verified more thoroughly. This fact should be seriously taken into consideration especially by the researchers who use universal primers, e.g. Folmer primers, which have a high specificity to bacterial genome. As a consequence, some of the accounts of new species described only at the molecular level remain questionable. Moreover, this diagnostic method of species identification can also lead to overestimation of species number, which in turn can result in numerous mistakes. Therefore, there is an urgent need to review and correct the available data to prevent similar errors in further research. The present study clearly shows that the best solution to eliminate incorrect sequence flagging is to perform more thorough comparative analyses of homologous sequences deposited in the available databases, both the NCBI and BOLD.

Supporting information

S1 Table. The list of incorrectly flagged cox1 gene fragment sequences of Gastrotricha originated from another metazoan species.



We are grateful to Małgorzata Kolicka from Adam Mickiewicz University, in Poznań (Poland) for her support and clarification of doubts about the sources of incorrect sequences in Gastrotricha.


  1. 1. Folmer O, Black M, Hoeh W, Lutz R, Vrijenhoek R. DNA primers for amplification of mitochondrial c oxidase subunit I from diverse metazoan invertebrates. Mol. Mar. Biol. Biotechnol. 1994;3(5): 294–299. pmid:7881515
  2. 2. Simon C, Frati F, Beckenbach A, Crespi B, Liu H, Flook P. Evolution, weighting, and phylogenetic utility of mitochondrial gene sequences and a compilation of conserved polymerase chain reaction primers. Ann. Entomol. Soc. Am. 1994;87: 651–701.
  3. 3. Boehme P, Amendt J, Zehner R. The use of COI barcodes for molecular identification of forensically important fly species in Germany. Parasitol. Res. 2012;110: 2325–2332. pmid:22186975
  4. 4. Frati F, Simon C, Sullivan J, Swofford DL. Evolution of the mitochondrial cytochrome oxidase II gene in Collembola. J. Mol. Evol. 1997;44: 145–158. pmid:9069175
  5. 5. Buhay JE. ‘COI-like’ sequences are becoming problematic in molecular systematic and DNA barcoding studies. J. Crust. Biol. 2009;29: 96–110.
  6. 6. Elbrecht V, Leese F. Validation and development of freshwater invertebrate metabarcoding COI primers for Environmental Impact Assessment. PeerJ Preprints 2016.
  7. 7. Geller J, Meyer C, Parker M, Hawk H. Redesign of PCR primers for mitochondrial cytochrome c oxidase subunit I for marine invertebrates and application in all-taxa biotic surveys. Mol. Ecol. Resour. 2013;13: 851–861. pmid:23848937
  8. 8. Meyer CP. Molecular systematics of cowries (Gastropoda: Cypraeidae) and diversification patterns in the tropics. Biol. J. Linn. Soc. 2003;79: 401–459.
  9. 9. Derycke S, Vanaverbeke J, Rigaux A, Backeljau T, Moens T. Exploring the use of Cytochrome Oxidase c Subunit 1 (COI) for DNA barcoding of freeliving marine nematodes. PLoS One 2010;5: e13716. pmid:21060838
  10. 10. Hoareau TB, Boissin E. Design of phylum-specific hybrid primers for DNA barcoding: addressing the need for efficient COI amplification in the Echinodermata. Mol. Ecol. Res. 2010;10: 960–967.
  11. 11. Lobo J, Costa PM, Teixeira MAL, Ferreira MSG, Costa MH, Costa FO. Enhanced primers for amplification of DNA barcodes from a broad range of marine metazoans. BMC Ecology 2013;13: 34. pmid:24020880
  12. 12. De Waard JR, Ivanova NV, Hajibabaei M, Hebert PDN. Assembling DNA barcodes. Analytical protocols. Methods Mol. Biol. 2008;410: 275–294. pmid:18642605
  13. 13. Williams ST, Knowlton N. Mitochondrial Pseudogenes Are Pervasive and Often Insidious in the Snapping Shrimp Genus Alpheus. Mol. Biol. Evol. 2001;18: 1484–1493. pmid:11470839
  14. 14. Song H, Buhay JE, Whiting MF, Crandall KA. Many species in one: DNA barcoding overestimates the number of species when nuclear mitochondrial pseudogenes are coamplified. Proceedings of the National Academy of Sciences USA 2008;105: 13486–13491.
  15. 15. Siddall ME, Fontanella FM, Watson SC, Kvist S, Erseus C. Barcoding bamboozled by bacteria: convergence to Metazoan mitochondrial primer targets by marine microbes, Syst. Biol. 2009;58: 445–451. pmid:20525598
  16. 16. Mioduchowska M, Gołdyn B, Czyż MJ, Namiotko T, Namiotko L, Kur J et al. Notes on genetic uniformity in the fairy shrimp Branchipus schaefferi Fischer, 1834 (Branchiopoda, Anostraca) from Poland. N. West. J. Zool. 2018;14(1): 127–129.
  17. 17. Weekers PH, Murugan G, Vanfleteren JR, Belk D, Dumont HJ. Phylogenetic analysis of anostracans (Branchiopoda: Anostraca) inferred from nuclear 18S ribosomal DNA (18S rDNA) sequences. Mol. Phylogenet. Evol. 2002;25: 535–44. pmid:12450757
  18. 18. Hall TA. BioEdit: a user friendly biological sequence alignment editor and analysis program for Windows 95/98/NT. Nucleic Acids Symp. Ser. 1999;41: 95–98.
  19. 19. Thompson JD, Gibson TJ, Plewniak F, Jeanmougin F, Higgins DG. The CLUSTAL_X windows interface: flexible strategies for multiple sequence alignment aided by quality analysis tools. Nucleic Acids Res. 1997;25: 4876–4882. pmid:9396791
  20. 20. Altshul SF, Gish W, Miller W, Myers EW, Lipman DJ. Basic local alignment search tool. J. Mol. Biol. 1990;215: 403–410. pmid:2231712
  21. 21. Rice P, Longden I, Bleasby A. EMBOSS: the European molecular biology open software suite. Trends Genet. 2000;16: 276–277. pmid:10827456
  22. 22. Tamura K, Stecher G, Peterson D, Filipski A, Kumar S. MEGA6: Molecular evolutionary genetics analysis version 6.0. Mol. Biol. Evol. 2013;30: 2725–2729. pmid:24132122
  23. 23. Nakano T, Ozawa T. Systematic revision of Patelloida pygmaea (Dunker, 1860) (Gastropoda: Lottiidae), with a description of a new species. J. Molluscan Stud. 2005;71: 357–370.
  24. 24. Todaro MA, Kanneby T, Dal Zotto M, Jondelius U. Phylogeny of Thaumastodermatidae (Gastrotricha: Macrodasyida) inferred from nuclear and mitochondrial sequence data. PLoS ONE 2011;6: e17892. pmid:21455302
  25. 25. Nakano T, Ozawa T. Worldwide phylogeography of limpets of the order Patellogastropoda: Molecular, morphological and palaeontological evidence. J. Molluscan Stud. 2007;73: 79–99.
  26. 26. Nei M, Kumar S. Molecular Evolution and Phylogenetics. Oxford University Press, New York 2000.
  27. 27. Darriba D, Taboada GL, Doallo R, Posada D. jModelTest 2: more models, new heuristics and paralel computing. Nat. Methods 2012;9: 772.
  28. 28. Posada D, Buckley TR. Model selection and model averaging in phylogenetics: advantages of Akaike information criterion and Bayesian approaches over likelihood ratio tests. Systematic Biology 2004;53: 793–808. pmid:15545256
  29. 29. IPPC. ISPM no. 31 Methodologies for sampling of consignments. IPPC, FAO, Rome; 2008.
  30. 30. R Core Team. R: A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL 2017;
  31. 31. Winter D. rentrez: Entrez in R. R package version 1.1.0. 2017;
  32. 32. Dorai-Raj S. binom: Binomial Confidence Intervals For Several Parameterizations. R package version 1.1–1. 2014;
  33. 33. Mioduchowska M, Kaczmarczyk A, Zając K, Zając T, Sell J. Gender-associated mitochondrial DNA heteroplasmy in somatic tissues of the endangered freshwater mussel Unio crassus (Bivalvia: Unionidae): implications for sex identification and phylogeographical studies. J. Exp. Zool. A Ecol. Genet. Physiol. 2016;325A: 610–625.
  34. 34. Kanneby T, Todaro MA, Jondelius U. A phylogenetic approach to species delimitation in freshwater Gastrotricha from Sweden. Hydrobiologia 2012;683: 185–202.
  35. 35. Kanneby T, Todaro MA, Jondelius U. Phylogeny of Chaetonotidae and other Paucitubulatina (Gastrotricha: Chaetonotida) and the colonization of aquatic ecosystems. Zool. Scr. 2013;42(1): 88–105.
  36. 36. Leray M, Yang JY, Meyer CP, Mills SC, Agudelo N, Ranwez V, et al. A new versatile primer set targeting a short fragment of the mitochondrial COI region for metabarcoding metazoan diversity: application for characterizing coral reef fish gut contents. Front. Zool. 2013;10(1): 34.
  37. 37. Kanneby T, Wicksten MK. First record of the enigmatic genus Redudasys Kisielewski, 1987 (Gastrotricha: Macrodasyida) from the Northern hemisphere. Zoosystema 2014;36(4): 723–733.
  38. 38. Bekkouche N, Worsaae K. Neuromuscular study of early branching Diuronotus aspetos (Paucitubulatina) yields insights into the evolution of organs systems in Gastrotricha. Zoological Lett. 2016;2: 21. pmid:27688902
  39. 39. Kieneke A, Martinez Arbizu PM, Fontaneto D. Spatially structured populations with a low level of cryptic diversity in European marine Gastrotricha. Mol. Ecol. 2012;21(5): 1239–1254. pmid:22257178
  40. 40. Hebert PD, Remigio EA, Colbourne JK, Taylor DJ, Wilson CC. Accelerated molecular evolution in halophilic crustaceans. Evolution 2002;56(5): 909–926. pmid:12093027
  41. 41. Hebert PD, Ratnasingham S, Zakharov EV, Telfer AC, Levesque-Beaudin V, Milton MA, et al. Counting animal species with DNA barcodes: Canadian insects. Philos. Trans. R. Soc. Lond., B, Biol. Sci. 2016;371(1702).
  42. 42. Gao X, Jian J, Li WJ, Yang YC, Shen XW, Sun ZR, et al. Genomic study of polyhydroxyalkanoates producing Aeromonas hydrophila 4AK4. Appl. Microbiol. Biotechnol. 2013;97(20): 9099–9109. pmid:24000047
  43. 43. Chai B, Wang H., Chen X. Draft genome sequence of high-melanin-yielding Aeromonas media strain WS. J. Bacteriol. 2012;194(23): 6693–6694. pmid:23144426
  44. 44. Liu L, Li N, Zhang D, Fu X, Shi C, Lin Q, et al. Complete Genome Sequence of the Highly Virulent Aeromonas schubertii Strain WL1483, Isolated from Diseased Snakehead Fish (Channa argus) in China. Genome Announc. 2016;4(1): e01567–15. pmid:26798095
  45. 45. Nolan M, Sikorski J, Davenport K, Lucas S, Del Rio TG, Tice H, et al. Complete genome sequence of Ferrimonas balearica type strain (PAT). Stand. Genomic Sci. 2010;3(2): 174–182.
  46. 46. Lin J, Kong L, Li Q. DNA barcoding of true limpets (Order Patellogastropoda) along coast of China: a case study. College of Fisheries, Ocean University of China, No. 5 Yushan Road, Shinan District, Qingdao, Shandong 266003, P.R. China. Mitochondrial DNA A DNA Mapp. Seq. Anal. 2016;27(4): 2310–4. pmid:25812056
  47. 47. Ding JY, Shiu JH, Chen WM, Chiang YR, Tang SL. Genomic Insight into the Host-Endosymbiont Relationship of Endozoicomonas montiporae CL-33(T) with its Coral Host. Front. Microbiol. 2016;7: 251. pmid:27014194
  48. 48. Morohoshi T, Kato T, Someya N, Ikeda T. Complete Genome Sequence of N-Acylhomoserine Lactone-Producing Pseudomonas sp. Strain StFLB209, Isolated from Potato Phyllosphere. Genome Announc. 2014;2(5).
  49. 49. Gardes A, Kaeppel E, Shehzad A, Seebah S, Teeling H, Yarza P, et al. Complete genome sequence of Marinobacter adhaerens type strain (HP15), a diatom-interacting marine microorganism. Stand. Genomic Sci. 2010;3(2): 97–107.
  50. 50. Jeong H, Yim JH, Lee C, Choi SH, Park YK, Yoon SH, et al. Genomic blueprint of Hahella chejuensis, a marine microbe producing an algicidal agent. Nucleic Acids Res. 2005;33(22): 7066–7073. pmid:16352867
  51. 51. Simison WB. Evolution and phylogeography of new world gastropod faunas. Berkeley: University of California; 2000.
  52. 52. Lohman DJ, Prawiradilaga DM, Meier R. Improved COI barcoding primers for Southeast Asian perching birds (Aves: Passeriformes). Mol. Ecol. Resour. 2009;9: 37–40. pmid:21564563
  53. 53. Schubart CD. Mitochondrial DNA and decapod phylogenies: the importance of pseudogenes and primer optimization. In: Martin JW, Crandall KA, Felder DL, Boca Raton FL, editors. Decapod Crustacean Phylogenetics. CRC Press, Taylor & Francis Group 2009;47–65.
  54. 54. Creer S, Fonseca VG, Porazinska DL, Giblin-Davis RM, Sung W, Power DM, et al. Ultrasequencing of the meiofaunal biosphere: practice, pitfalls and promises. Mol. Ecol. 2011;19: 4–20.
  55. 55. Sharma P, Kobayashi T. Are”universal” DNA primers really universal? J. Appl. Genet. 2014;55: 485–496. pmid:24839163
  56. 56. Gibson JF, Kelso S, Jackson MD, Kits JH, Miranda GFG, Skevington JH. Diptera-specific polymerase chain reaction amplification primers of use in molecular phylogenetic research. Ann. Entomol. Soc. Am. 2011;104(5): 976–997.
  57. 57. Moulton MJ, Song HJ, Whiting MF. Assessing the effects of primer specicity on eliminating numt coampli-cation in DNA barcoding: a case study from Orthoptera (Arthropoda: Insecta). Mol. Ecol. Resour. 2010;10: 615–627. pmid:21565066
  58. 58. Janda JM, Sharon LA. The Genus Aeromonas: Taxonomy, Pathogenicity, and Infection. CMR 2010;35–73.
  59. 59. Gudmundsdóttir BK. Infections by atypical strains of the bacterium Aeromonas salmonicida. Buvisindi, Icel. Agr. Sci. 1998;12: 61–72.
  60. 60. Graf J. Symbiosis of Aeromonas and Hirudo medicinalis, the medicinal leech. ASM News 2000;66: 147–153.
  61. 61. Tasiemski A, Massol F, Cuvillier-Hot V, Boidin-Wichlacz C, Roger E, Rodet F, et al. Reciprocal immune benefit based on complementary production of antibiotics by the leech Hirudo verbana and its gut symbiont Aeromonas veronii. Sci. Rep. 2015;5: 17498. pmid:26635240
  62. 62. Couillault C, Ewbank JJ. 2002. Diverse bacteria are pathogens of Caenorhabditis elegans. Infect. Immun. 70:4705–4707. pmid:12117988
  63. 63. Yang CS, Chen MH, Arun AB, Chen CA, Wang JT, Chen WM. Endozoicomonas montiporae sp. nov., isolated from the encrusting pore coral Montipora aequituberculata. Int. J. Syst. Evol. Microbiol. 2010;60: 1158–62. pmid:19666790
  64. 64. Kurahashi M, Yokota A. Endozoicomonas elysicola gen. nov., sp. nov., a γ-proteobacterium isolated from the sea slug Elysia ornata. Syst. Appl. Microbiol. 2007;30: 202–206. pmid:16904280
  65. 65. Nishijima M, Adachi K, Katsuta A, Shizuri Y, Yamasato K. Endozoicomonas numazuensis sp. nov., a gammaproteobacterium isolated from marine sponges, and emended description of the genus Endozoicomonas Kurahashi and Yokota 2007. Int. J. Syst. Evol. Microbiol. 2013;63: 709–714. pmid:22544802
  66. 66. Hyun DW, Shin NR, Kim MS, Oh SJ, Kim PS, Whon TW, et al. Endozoicomonas atrinae sp. nov., isolated from the intestine of a comb pen shell Atrina pectinate. Int. J. Syst. Evol. Microbiol. 2014;64(7): 2312–8.
  67. 67. Kress WJ, Erickson DL. DNA barcodes: genes, genomics, and bioinformatics. Proc. Natl. Acad. Sci. USA 2008;105: 2761–2762. pmid:18287050
  68. 68. Bruder K, Medlin LK. Molecular assessment of phylogenetic relationships in selected species/genera in the naviculoid diatoms (Bacillariophyta). I. The genus Placoneis. Nova Hedwigia 2007;85: 331–352.
  69. 69. Jahn R, Zetzsche H, Reinhardt R, Gemeinholze B. Diatoms and DNA barcoding: A pilot study on an environmental sample. Proceedings of the 1st Central European Diatom Meeting, Berlin 2007; 113.
  70. 70. Meyer A, Todt C, Mikkelsen NT, Lieb B. Fast evolving 18S rRNA sequences from Solenogastres (Mollusca) resist standard PCR amplification and give new insights into mollusk substitution rate heterogeneity. BMC Evol. Biol., 2010;10: 70 75. pmid:20214780
  71. 71. Pinceel T, Vanschoenwinkel B, Waterkeyn A, Vanhove MPM, Pinder A, Timms BV, et al. Fairy shrimps in distress: a molecular taxonomic review of the diverse fairy shrimp genus Branchinella (Anostraca: Thamnocephalidae) in Australia in the light of ongoing environmental change. Hydrobiologia 2013;700: 313–327.
  72. 72. Czyż MJ, Woliński P, Gołdyn B. Cyst morphology of large branchiopod crustaceans (Anostraca, Notostraca, Laevicaudata, Spinicaudata) in western Poland. Biological Letters 2016;53(1): 73–82.
  73. 73. Staats M, Arulandhu AJ, Gravendeel B, Holst-Jensen A, Scholtens I, Peelen T, et al. Advances in DNA metabarcoding for food and wildlife forensic species identification. Anal. Bioanal. Chem. 2016;1–16.
  74. 74. Manylov OG. First finding of a microsporidian parasite in the gastrotrich, Turbanella lutheri (Gastrotricha: Macrodasyida). Protistology 1999;1: 17–19.
  75. 75. Kisielewska G, Kolicka M, Zawierucha K. Prey or parasite? The first observations of live Euglenida in the intestine of Gastrotricha. Eur. J. Protistol. 2015;51(2): 138–41. pmid:25779973
  76. 76. Ratnasingham S, Hebert PDN. BOLD: The Barcode of Life Data System ( Mol. Ecol. Notes. 2007;7: 355–364. pmid:18784790
  77. 77. Ratnasingham S, Hebert PDN. A DNA-based registry for all animal species: the barcode index number (BIN) system. PLoS One. Public Library of Science 2013;8: e66213. pmid:23861743
  78. 78. Kur J, Mioduchowska M, Petković M. Trying to solve current issues with invertebrate taxonomy—the conceptual web-based application. World Scientific News 2016;57: 664–673.
  79. 79. Michalczyk Ł, Kaczmarek Ł. The Tardigrada Register: a comprehensive online data repository for tardigrade taxonomy. J. Limnol. 2013;72(s1): 175–181.
  80. 80. Yeates DK, Seago A, Nelson L, Cameron SL, Joseph L, Trueman JWH. Integrative taxonomy, or iterative taxonomy? Syst. Entomol. 2011;36: 209–217.