In order to establish the most up-to-date checklist of macroinvertebrates species recognised and described occurring in freshwater aquatic habitats in metropolitan France, three checklists were complied. The first one comes from PERLA, an interactive tool managed by a French regional environmental agency (DREAL Auvergne-Rhône-Alpes) accessible at http://www.perla.developpement-durable.gouv.fr/index.php. PERLA serves as a national comprehensive checklist for water managers and encompasses larvae, nymphs and adults across various taxonomic groups, including insects, molluscs, crustaceans and worms, amongst others, found in rivers and aquatic ecosystems. In the manuscript, we will refer to this checklist as the ‘French aquatic ecosystems checklist’. The second checklist is coming from macroinvertebrate surveys conducted in four Alpine lakes (Lake Geneva, Lake Annecy, Lake Bourget and Lake Tignes) from 2015 to 2022. In the manuscript, we will refer to this checklist as the ‘French Alpine lakes checklist’. The third checklist we used was from a French NGO, Opie-benthos (https://www.opie-benthos.fr/opie/monde-des-insectes.html), studying freshwater insect’s taxonomy and diversity. Currently, Opie-benthos gathers, curates and maintains certainly the most up-to date freshwater insect species list for the French metropolitan territory. In the manuscript, we will refer to this checklist as the ‘French aquatic insect’s species list’. These three checklists were merged into a single one. When the taxonomic resolution of taxa was limited to a level above species (genus, family, class, phylum), we consulted the taxonomic referential of the National Inventory of Natural Heritage (INPN - Inventaire national du patrimoine naturel) (TAXREF v.17.0 2024, available at https://inpn.mnhn.fr/telechargement/referentielEspece/taxref/17.0/menu) to select species from those ranks. The phylum Acanthocephala and Rotifera, the classes Copepoda and Ostracoda (Arthropoda, Crustacea) and the families Succineidae (Arthropoda, Insecta, Diptera) and Hydrachnidae (Arthropoda, Arachnida, Trombidiformes), initially listed in French aquatic ecosystems checklist, were omitted from our search from TAXREF v.17.0 2024 as they are not considered as freshwater macroinvertebrates according Tachet et al. (2000). Once those macroinvertebrates species were retrieved for each taxonomic level without species identification, several filters from TAXREF were applied to select only French metropolitan freshwater species from this taxonomic referential. First, we selected only the species occurring in the following habitats: freshwater, marine and freshwater, brackish water and continental (land and freshwater). Then, from this updated list, only species occurring in metropolitan France were kept. Finally, species characterised by one those eight statuses over 15 statuses in total were selected (present (native or undetermined), endemic, sub-endemic, cryptogenic, introduced, invasive introduced, non-established introduced (including cultivated/domestic) and occasional). The checklists described above and the summarised species list resulting from these compilations, inclusive of all species from the mentioned inventories, is referred to as the “French metropolitan freshwater macroinvertebrates’ species list” and all together are available at https://doi.org/10.57745/LMOXEW in the ‘Checklists composition data’ file for future reference.

All the sequences for those species listed in the French metropolitan freshwater macroinvertebrates’ species list, whatever the gene, were downloaded from October 2023 to May 2024, from the Barcode of Life Data System v.4 -Bold- repository (https://boldsystems.org/index.php/databases). We opted to utilise BOLD as the reference library instead of GenBank due to concerns regarding the latter’s unverified submission process, which frequently results in misannotated sequences (Kozlov et al. 2016; Locatelli et al. 2020; Steinegger and Salzberg 2020). The absence of sequences for a given species may indicate either that the search in BOLD returned “Unmatched terms” or that, while the species was found, no sequences were publicly available, rendering them private sequences (regardless of the reason for their absence, they were all defined as “not available sequences” for this analysis). Before concluding that a species from the French aquatic ecosystems’ checklist and French Alpine lakes checklist lacked sequences in BOLD, synonyms were searched using INPN. The details of taxa resulting from this taxonomic harmonisation are available in Suppl. material 1: table S1. From this file containing all the sequences retrieved (whatever the gene) for the queried species of the French metropolitan freshwater macroinvertebrates’ species list, the COI-5P sequences belonging to the COI gene were selected. All the remaining sequences were aligned with the fwhF2/fwhR2N primer pair using MAFFT version 7 (Auto) (https://mafft.cbrc.jp/alignment/server/), by genus or species group (i.e. all sequences assigned to the same genus or species were aligned together). Then, sequences characterised by gaps, insufficiencies in length relative to the COI barcode produced by the fwhF2/fwhR2N primer pair (shorter than 205 bp) or identical (same genetic sequence for one species) from the sequences file, were removed using Jalview (https://www.jalview.org/). Due to the possibility of identical sequences being shared amongst different species, certain species may have been excluded during this stage, as a result of sequence overlap with other taxa. To address this issue, the list of species obtained from this step was cross-referenced with the initial species list following the alignment process. This comparison identified any missing species, whose sequences were then re-aligned individually to ensure their accurate representation. This final reference library containing only genetic sequences capable of being aligned by the fwhF2/fwhR2N primer pair is named ‘Aligned DNA library’ hereafter. The reference barcode libraries (with all genes, COI-5P fragments only and the shorter COI-5P fragment matching fwhF2/fwhR2N primers) and a summary table of the number of sequences and Barcode Index Numbers (BINs) as potential proxy of cryptic species are available at https://doi.org/10.57745/LMOXEW for future reference.

We firstly analysed the taxonomic composition of the French metropolitan freshwater macroinvertebrates’ species list. Secondly, to highlight any gaps in the availability of sequences in our list of species, the taxonomic coverage of barcodes at various taxonomic levels was assessed. To estimate the importance of cryptic diversity for the different taxonomic groups, we used the number of Barcode Index Numbers (BINs) per species. We also reported the number of BINs lumping sequences attributed to different species. To account for the unbalanced availability of the DNA sequences (and sampling effort) amongst species, we also evaluated the correlation between the number of BINs and sequences. Thirdly, we highlighted the taxonomic composition of species with COI-5P sequences that did not align with fwhF2/fwhR2N primers to reveal which species and taxonomic groups could be misrepresented in metabarcoding studies using that primer pair. Then, to explore the species haplotype diversity, the number of different haplotypes available relative to the number of species by phylum was examined. The categorisation of unique barcodes ranging from limited (< 5 haplotypes) to moderate (5–25 haplotypes) to good (> 25 haplotypes), was adopted from Trebitz et al. (2015) and the species with the highest number of haplotypes exceeding 100 were identified. Finally, in order to demonstrate the ability of the primers to discriminate each species by a unique barcode, groups of taxa sharing the same haplotype were identified.

The French metropolitan freshwater macroinvertebrates’ species list is composed of 10 phyla, 16 classes, 50 orders, 222 families, 670 genera and 2841 species (Table 1).

Fig. 1 provides insights into the taxonomic composition, highlighting that the highest diversity is observed within the phylum Arthropoda, followed by Mollusca and Nematoda, with 2469, 195 and 99 species, respectively. On the other hand, taxa such as Entoprocta and Nemerta are represented by only one species (Fig. 1, Table 1).

Figure 1.

Taxonomic composition of the French metropolitan freshwater macroinvertebrates’ species list. Percentage of the number of species according to phylum (a), class within the arthropod phylum (b) and order within the insect class (c). The number above each bar represents the total number of species affiliated to each taxonomic rank.

Amongst the insects, Coleoptera dominate, followed by Diptera and Trichoptera, with 684, 630 and 527 species, respectively. The predominant phyla with the highest number of species in our checklist matches with those identified in a study by Specchia et al. (2020), who conducted a gap analysis of DNA barcodes available in international repositories using the aquatic macroinvertebrate species checklist of a south-eastern Apulia region in Italy. Furthermore, our findings align with the prevailing understanding that Arthropoda and thus insects, represent the most diverse group of animals, exerting dominance in freshwater ecosystems inventories (Yeates and Wiegmann 1999; Grosberg et al. 2012; Dijkstra et al. 2014; Choudhary and Ahi 2015). Moreover, Diptera emerge as the most species-rich group utilised in biomonitoring across Europe, with chironomids recognised for their prevalence and diversity in freshwater habitats (Pinder 1986). Alongside Diptera, Coleoptera (beetles) stand out for their exceptionally high species numbers across various ecoregions and countries in Europe and serving as the most abundant group of aquatic insects (Jäch and Balke 2008; Short 2018). Following these orders, Trichoptera (caddisflies), Plecoptera (stoneflies) and Ephemeroptera (mayflies), collectively known as EPT, emerge as the next three orders in terms of species richness from our checklist. These organisms spend their immature stages in freshwater and are widely employed as biological indicators for freshwater quality assessment (Hering et al. 2004; Sweeney et al. 2011) and ecological investigations, demonstrating robust responses to pollution or climate change (Álvarez-Troncoso et al. 2015). Additionally, Nematoda emerges as a highly diverse group, ranking third in species richness amongst all phyla listed in our checklist. This outcome may be attributed to the interest in this phylum in ecological assessments, as nematodes have been used for this purpose for a long time (e.g. Bongers (1990); Moreno et al. (2011)). From all the sequences retrieved, based on the species of the French metropolitan freshwater macroinvertebrates’ species list, 85.4% belonged to COI-5P gene (Suppl. material 1: fig. S1). This result was expected as BOLD is the main repository for COI sequences (Ratnasingham and Hebert 2007). Overall, 56% of the 2841 species listed in the French metropolitan freshwater macroinvertebrates’ species list possessed at least one COI-5P genetic sequence publicly available in BOLD database (Fig. 2).

Figure 2.

Barcoding coverage of the French metropolitan freshwater macroinvertebrates’ species list. The barcoding coverage gives for each phylum, for each class within Arthropoda and for each order within Insecta the total percentage of species with available COI-5P public sequence in BOLD. The number above each bar represents the total number of species listed in the Freshwater macroinvertebrate checklist.

There are some exceptions with Nemerta, which have a barcode coverage of 100% and the Entoprocta that have 0% coverage (both phyla with only one species referenced). Mollusca (195 species) and Nematoda (99 species), which are the second and third most diverse phyla in terms of species in the French metropolitan freshwater macroinvertebrates’ species list, have the worst barcode coverage of all phyla (41% and 5.1%, respectively). Within Arthropoda, most classes have a barcode coverage above 60%, with the highest being in the Malacostraca (65 species, 73.8%) and the lowest in the Branchiura (3 species, 66.7%). Finally, within insects, Hymenoptera (1 specie), Lepidoptera (5 species), Megaloptera (3 species), Odonata (89 species) and Hemiptera (83 species) have the highest barcode coverage (> 80%). The three most species diverse insects’ orders (Coleoptera (684 species), Diptera (630 species) and Trichoptera (527 species)) have a barcode coverage of 73.4%, 55.9% and 67.9%, respectively (Fig. 2). Compared to previous gap analyses carried out in other countries, our analysis in freshwater ecosystems in France revealed a slightly lower barcoding coverage for freshwater macroinvertebrates (56%). For instance, investigations in specific regions such as the Apulia Region of southeast Italy reported DNA barcode availability for 58% of listed aquatic Macroinvertebrate species (Specchia et al. 2020), while a study in Atlantic Iberia documented coverage for 63% of macroinvertebrates (Leite et al. 2020). Similarly, a comprehensive assessment of 4502 freshwater invertebrate species used in ecological quality assessments indicated that 60% possessed one or more barcodes (Weigand et al. 2019). Our findings of the lowest barcode coverage at the phylum level align with previous studies, which also reported very low barcode coverage for freshwater Platyhelminthes, with only three species having sequences deposited in examined databases (two in our study). The limited barcode coverage observed for this phylum may be attributed to challenges associated with amplification using standard COI primers. For instance, a study conducted across 11 rivers reported the absence of taxa detection from this phylum using three pairs of standard COI primers (Poyntz-Wright et al. 2024). A similar trend was observed for the phylum Nematoda, which also exhibits low barcode coverage and was not successfully amplified in the previously mentioned study. Furthermore, studies have demonstrated that the COI gene is an unsuitable molecular marker for free-living marine nematodes due to extensive haplotype hypervariation and frequent mitochondrial genome recombination (Da Silva et al. 2010; Hyman et al. 2011).The low taxonomic coverage for Mollusca in our study can be attributed to the high number of DNA barcodes deposited in GenBank rather than in BOLD (Weigand et al. 2019 and references therein). Insects exhibited the highest number of available COI sequences, aligning with previous studies highlighting the over-representation of Arthropoda, particularly insects (Meglécz 2023). The barcode coverage for Insecta was 69%, aligning with other findings such as Weigand et al. (2019), who reported that 66% of monitored insect species were barcoded and Trebitz et al. (2015), who found approximately 70% representation in BOLD for Great Lakes fauna. Within Insecta, Diptera had the lowest coverage on BOLD at 55.9%, while Odonata and Hemiptera were the best covered, with over 80% of species barcoded in each group, similar to Weigand et al. (2019). The large number of DNA barcodes accessible within public databases often reflects the intensity of dedicated studies and associated barcoding projects. Certain taxonomic groups, such as Arthropoda, receive disproportionate attention, resulting in a heightened focus and increased deposition of sequences within genetic databases (Briski et al. 2011, 2016; Ardura 2019). Consequently, the absence of comprehensive reference databases may lead to false-negative outcomes (Klymus et al. 2017), while inaccuracies within these databases can engender false-positive identifications, as evidenced by instances of misreporting species presence (Port et al. 2016). Furthermore, the false assignment of sequences to closely-related species may ensue when references for the true species are absent (Schenekar et al. 2020; Couton et al. 2022). Resolving this issue necessitates the sequencing of new specimens of target taxa and their subsequent integration into reference libraries. Currently, some projects such as Biodiversity Genomics Europe (BGE) aim at producing such data, for example, DNA barcodes from 45,000 specimens of 15,000 species and metabarcoding data from environmental samples in Europe have been sequenced. In addition, some taxonomic studies released barcodes (e.g. Vuataz et al. (2024)) and ecological and evolutionary studies delivered large amounts of DNA sequences for certain European taxa (e.g. Saclier et al. (2024)).

The 1811 species, with a BIN and at least one COI-5P sequence, encompassed 3348 unique BINs, which might indicate a high rate of cryptic diversity. In one hand, this diversity is probably underestimated due to 1689 BOLD identifiers, belonging to 410 species (for which 31 are not in the 1811 species), which have no associated BINs, representing 2.4% of the total identifiers recovered (69083). This can be explained by the fact that the sequences associated with these identifiers did not meet the conditions required to be evaluated for inclusion in the BINs according to the algorithm used to delimit this cryptic diversity (Ratnasingham and Hebert 2013). On the other hand, the cryptic species estimated with the BINs could also be overestimated for the French metropolitan territory as the sequences retrieved originate from many countries (Suppl. material 1: fig. S2). Indeed, for taxa with broad geographic distributions and possibly high genetic diversity, spatial heterogeneity in the sampling may increase the splitting of divergent haplotypes attributed to a single species into multiple BINs. Furthermore, we cannot totally rule out the possibility that a BIN might also originate from contamination errors, meaning that the sequences generated and uploaded belong to bacteria or endosymbionts, for example, rather than to the intended freshwater species (Meyer and Paulay 2005; Pilgrim et al. 2021). Interestingly, 267 BINS are associated with 410 distinct species (Suppl. material 1: table S2), which highlights that molecular techniques tend to group morphologically distinct taxa together. This may be due to the inability of the COI to differentiate between these taxa (non-discriminatory barcode) and/or to morphological over-division, based on morphological criteria that are not in agreement with the COI.

There was a significant correlation (R² = 0.6, p < 0.01) between the number of unique COI sequences per species and the number of BINs generated (Fig. 3).

Figure 3.

Number of BINs in relation to the number of sequences per species.

However, some groups like Annelida or Megaloptera have a high cryptic diversity, but a low genetic diversity and, on the contrary, Platyhelminthes or Lepidoptera have a high genetic diversity, but a low cryptic diversity (Table 2). Furthermore, uncovering the number of cryptic species is tricky as the molecular delineation of species, based on a mitochondrial barcode or a small portion of it, is far from being ideal (Rubinoff et al. 2006). There are many other algorithms to evaluate this cryptic diversity with different molecular species delimitation methods (Poisson Tree Process (Kapli et al. 2017), General Mixed Yule Coalescent (Fujisawa and Barraclough 2013), Assemble Species by Automatic Partitioning (Puillandre et al. 2021), for which distinct results can be obtained for the same pool of species (Eme et al. 2018). In addition, this mitochondrial gene can also reveal a singular evolution from the nuclear genome, the so called mitonuclear discordance, calling into question the use of a single barcode for species delineation (Després 2019 and references within).

The Fig. 4 represents the taxonomic composition of the 24 species from the French metropolitan freshwater macroinvertebrates’ species list that cannot be aligned with the primer pair fwhF2/fwhR2N.

Figure 4.

Number of species associated with COI-5P genetic sequences that cannot be aligned with fwh2 primer pair at the phylum rank (a), within the Arthropoda phylum (b) and within the Insecta and Malacostraca class (c, c’).

Arthropoda emerged as the predominant taxonomic groups facing alignment difficulties at the phylum level. Within insects (Arthropoda), Coleoptera and Diptera had the species with the most alignment issues. Specifically, within Malacostraca (Arthropoda), the gammarid (Gammaridae) and crayfish (Astacidea) families had the highest number of species with alignment issues (Fig. 4). Several reasons could explain these results. Firstly, although COI-5P sequences were available, some were found to lie outside the primer pairs’ intended region during the alignment process, either entirely or partially. In the latter case, where the sequence in question was shorter than the desired amplified barcode, it was subsequently discarded. Another possible explanation is the poor sequence annotation, such as mislabelling (e.g. COI-3P instead of COI5-P) or misidentification of specimens, leading to incorrect species assignments. Misidentification of voucher specimens has been highlighted as a major factor contributing to erroneous records, as morphological identifications of closely-related species can be challenging. This issue has been noted by Leite et al. (2020), Pentinsaari et al. (2020) and Paz and Rinkevich (2021), emphasising its impact on subsequent species identifications using databases like BOLD. This was evident in cases when some species of insects failed to align with the primer pair, while others belonging to the same genus did. Despite BOLD being a curated database with verification procedures during sequence deposition and a reported error rate of less than 1% for Metazoan sequences at the genus level (Leray et al. 2019), problematic records may still exist, as evidenced in various studies on marine macroinvertebrates (Radulovici et al. 2021) where up to 39% of sequences were considered ambiguous. For the Decapoda species’ (Malacostraca, Arthropoda), sequences that failed to align with the primers pair fwhF2/fwhR2N, comprising eight crayfish species and one crab species, their absence in the mock community during primer design could account for this discrepancy (Elbrecht and Leese 2015; Elbrecht et al. 2017). Although Gammarids (Amphipoda, Arthropoda) were included in primer design, the failure of some species’ sequences to align with the primers could be due to their vast diversity within aquatic environments (Horton et al. 2023). Molecular studies on Amphipoda have revealed extensive species diversity and the presence of cryptic species complexes (Jażdżewska and Mamos 2019), suggesting a high genetic variability within species (Lefébure et al. 2006) that may not be compatible with the primer pair. Furthermore, Vamos et al. (2017) indicated that their analysis of the efficiency of the primers pair displayed higher penalty scores for certain taxa of Turbellaria, Mollusca, Trichoptera and Isopoda, indicating potential under-representation due to primer bias. This aligns with findings from other studies suggesting preferential detection of taxonomic groups by different markers and primers (Leduc et al. 2019). Consequently, the incorporation of multiple genetic regions in DNA metabarcoding to ensure the broadest possible taxonomic detection may prove to be a good solution (Duarte et al. 2021). This could be particularly important to monitor the noble crayfish, Astacus astacus and the amphipod Gammarus roeselii which cannot be detected with fwhF2/fwhR2N primers pair (Fig. 4), whereas they are the most frequently monitored species of the malacostracans in Europe (Weigand et al. 2019). Analysis of the number of haplotypes relative to the number of species available within different phyla (Fig. 5) provides important information about the potential to detect taxa in natural samples.

Figure 5.

Taxonomic composition at the class rank for the species from the aligned DNA library. Each box at the phylum level represents the distribution of COI-5P haplotypes relative to the number of species.

In our reference library, for most phyla, the majority of species have less than five haplotypes, except cnidarians, which have the majority of their species with 5–25 haplotypes. This pattern of low haplotypes availability could be due to a lack of samples; therefore, too few taxa are sequenced to detect genetic variability. Only three species presented more than 100 unique barcodes: one trichopteran (Agraylea multipunctata), one isopod (Asellus aquaticus) and one gasteropod (Physella acuta). These findings are consistent with the observations of Trebitz et al. (2015), suggesting a noteworthy exception to the prevailing low barcoding rate for invertebrates, with some species being exceptionally well genetically represented. This phenomenon may be attributed to the scientific significance and ecological relevance of these species, such as being acknowledged as a reliable indicator. Indeed, Asellus aquaticus is a common species monitored in European countries (Weigand et al. 2019), known as a bioindicator for metal pollutants detection (e.g. O’Callaghan et al. (2019)). Physella acuta is also intensively studied as it is an invasive aquatic Gasteropoda with worldwide distribution (Banha et al. 2014; Vinarski 2017). The analysis on the ability of the fwhF2/fwhR2N primers to discriminate each species by a unique sequence demonstrated that 57 identical sequences were shared by 116 distinct species (Suppl. material 1: table S3). Amongst those sequences, 10 were shared by two or three different genera and 47 were shared by two or three different species. These results could be explained by the fact that some taxa could have been mislabelled or that, for this length of barcode (205 bp), no genetic variability is found between two related taxa; therefore, this barcode is not suitable to decipher those species. Several authors showed the necessity to have multiple sequences for each species to cover correctly their haplotypic diversity (Leite et al. 2020; Keck et al. 2023). This imperative arises from the recognition that the absence of intraspecific variants can pose significant challenges. In instances where a single sequence is available for a species exhibiting high genetic variability, the accurate identification of all haplotypes may be compromised. Moreover, inadequate representation of closely-related species within reference libraries can lead to the erroneous assignment of multiple species to a single taxonomic entity (e.g. Jackman et al. (2021)), potentially resulting in erroneous assessments of species diversity.

The authors have declared that no competing interests exist.

No ethical statement was reported.

This study was made possible thanks to a funding of the Carnot Eau & Environnement institute, a funding of the INRAE department AQUA and a funding of the pôle INRAE-OFB ECLA (ECosystèmes LAcustres).

Conceptualization: PG, FR. Data curation: PG. Formal analysis: DE, PG. Investigation: DE, PG. Methodology: PG, DE. Supervision: ID, FR. Validation: ID, DE, FR. Visualization: ID, DE, FR. Writing - original draft: PG. Writing - review and editing: DE, ID, FR.

Paula Gauvin https://orcid.org/0000-0002-9268-1856

David Eme https://orcid.org/0000-0001-8790-0412

Frédéric Rimet https://orcid.org/0000-0002-5514-869X

All of the data that support the findings of this study are available in the main text or Supplementary Information.