Mite specimens were retrieved from 24 soil samples collected between 2018 and 2020 in different localities of laurel forests, pine forests, heathlands and crops of the islands of Tenerife and Fuerteventura (Suppl. material 1). At each site, superficial and deep soil samples were taken. The superficial sample included the litter and the first centimetres of the soil and the deep sample included 25 litres of soil from a hole about 30-40 cm in depth. Every sample was processed following the flotation-Berlese-flotation protocol (FBF) that allows for the ‘clean’ extraction of arthropod mesofauna (Arribas et al. 2016). Briefly, the FBF protocol is based on soil flotation in water, which allows the extraction of the organic matter and soil mesofauna from raw soil samples. Subsequently, the organic portion is placed in a modified Berlese apparatus to capture specimens alive and preserve them in absolute ethanol at -20ºC. The last step of the FBF protocol is an additional flotation of the ethanol-preserved arthropods, resulting in ‘clean’ bulk specimen samples. From the 24 cleaned samples of bulk mesofauna, which were examined under a Motic SMZ-171 stereoscope, 239 mites were selected, with representatives from the orders Mesostigmata, Sarcoptiformes and Trombidiformes and maximising the morphological variability of the subset. Each specimen was assigned a unique identifier number (voucher number) and henceforth is referred to as ‘vouchers’.

Non-destructive DNA extractions were performed for each voucher. For that purpose, the exoskeleton of each specimen was punctured with an entomological pin and kept individually in 1.5 ml vials. Pins were sterilised under a Bunsen burner to avoid contaminations between samples. A volume of 110 µl of digestion buffer (pK ratio 1/10) was added to each voucher and digestion was done overnight at 60°C. The supernatant (DNA lysate) was transferred to the corresponding well within a deep-well plate and the (DNA-extracted) vouchers were maintained in the vials with ethanol for further morphological study (see above). Subsequent steps of the DNA extractions used the MAg-bind Blood & Tissue DNA extraction Kit (Omega Bio-tek) in the KingFisher robotic system (Thermo Fisher Scientific inc.). The default protocol was followed, but DNA lysate and reagent volumes were cut in half. The resulting 100 µl of genomic DNA extraction were split into a ‘stock plate’ directly frozen at -20ºC and a ‘working plate’ that was used to quantify DNA concentration using absorbance values within an Infinite M Nano (Tecan Trading AG).

PCR amplification was done for the 5’ end COI gene (standard barcode region for Metazoa; Hebert et al. (2003b)) using degenerate Folmer barcode primers (Fol-degen-for: ‘TCNACNAAYCAYAARRAYATYGG’; Fol-degen-rev: ‘TANACYTCNGGRTGNCCRAARAAYCA’; Folmer et al. (1994), Yu et al. (2012)). For PCR reaction, 5 µl of extracted DNA was used with 15 µl of PCR mix, which consisted in: 9.72 µl molecular-grade water, 2 µl 10x NH₄ buffer, 1.2 µl MgCl₂, 0.4 µl dNTPs, 0.4 µl of BSA, 0.6 µl 10 µM Fol-degen-for and 0.6 µl 10 µM Fol-degen-rev primers and 0.08 µl Taq polymerase (BIOTAQ™ DNA Polymerase, Bioline) per sample. PCR conditions were: 10 min at 95°C in 10 min, followed by 44 cycles of 30 s at 95°C, 30 s at 48°C and 3 min at 72°C; 10 min at 72°C and holding at 10°C. After viewing PCR products in agarose gel 1%, the positive PCR products were purified. The cleaning mix was prepared with 0.005 µl of exonuclease (exo1), 0.050 µl of rapid alkaline phosphatase (rAP) and 1.945 µl of double distilled water. A volume of 2 µl per 5 µl of DNA sample was used to run a 30 min protocol in the thermal cycler (30 min at 37°C, 5 min at 95°C and holding at 12°C). After cleaning, 5 µl of Fol-degen-for 5 µM primer were added to each sample. The purified PCR product for each voucher was Sanger-sequenced with ABI technology in Macrogen, Spain.

Sequences were edited (trimming and primer removal) on Geneious Prime version 2020.0.3 (www.geneious.com). Edited sequences were deposited in BOLD Systems (Ratnasingham and Hebert 2007), together with images, chromatograms and complementary information of the specimens (sampling site, date, taxonomy, store and institution information). Edited sequences were aligned using MAFFT 6.240 (Katoh et al. 2002) with the FFT-NS-i-x2 method and an unweighted pair group method with arithmetic mean (UPGMA) tree from distances corrected under a HKY distance model was generated. In addition, Maximum Likelihood (ML) phylogenetic analyses were run on the IQ-TREE web server at http://iqtree.cibiv.univie.ac.at (Trifinopoulos et al. 2016) using the best fitting substitution model for each codon partition as estimated with ModelFinder (Kalyaanamoorthy et al. 2017). Nodal support was obtained by 1,000 ultrafast boot-strap (UFBoot) replicates (Minh et al. 2013) (Suppl. material 2).

Five different methods were applied for molecular species delimitation of the vouchers. First, the barcode index number (BIN) was implemented in the BOLD system. This approach provides an effective method for species delineation as each sequence cluster is assigned a unique alphanumeric (BIN URI, see Table 1), which reflects the patterning of intra- and interspecific divergences found in the overall BOLD database (Ratnasingham and Hebert 2007, but see Meier et al. (2022)). Second, the Assemble Species by Automatic Partitioning analysis (ASAP) was implemented. ASAP is a species delimitation method based on pairwise genetic distances ranked using a unique scoring system and consists of merging sequences into groups by an ascending hierarchical clustering until all sequences merge in a single group (Puillandre et al. 2020; https://bioinfo.mnhn.fr/abi/public/asap/). Analysis was implemented over the edited sequences using a K80 Kimura model and the lower ASAP scores as species delimiter parameter (Table 1). Third, we applied the Bayesian implementation of the Poisson tree processes model (PTP, Zhang et al. (2013); http://species.h-its.org/ptp/) by using the phylogenetic tree previously generated on IQ-tree. PTP works with a species delimitation hypothesis, based on the number of mutations (branch lengths) and evaluates species delimitation hypothesis using Maximum-Likelihood algorithms (mlPTP). Finally, as additional criteria for defining molecular entities, we also implemented 3% and 8% genetic distance thresholds using the distance-based UPGMA tree previously generated. The 3% threshold has been widely used to define OTUs in molecular studies (e.g. Hebert et al. (2003a), Magoga et al. (2021)), whereas the 8% threshold has been recently proposed as a conservative threshold in beetles (Salces-Castellano et al. 2021).

To evaluate the previously unrecorded diversity in the Canary Islands that our dataset contains, we performed searches of the different species identified in our data within BIOTA. For each species within our dataset, we annotated the previously existing records and their status as endemic, native or potentially introduced taxa in the archipelago context.

To identify potential cryptic diversity within our dataset, we compare the morphological assignment provided to specimens with the results provided by the different species delimitation methods implemented. The number of morphological species that were split, merged or maintained was estimated for each delimitation method and the concordance amongst the specimens grouping resulting from each approach was evaluated.

To quantify the overall unrecorded mite diversity within our dataset, we used the Barcode of Life Data (BOLD) as the major source of barcode reference sequences available. BOLD contains 17,789,385 specimen records, of which 13,911,307 have barcode sequences (accessed at 02/08/2023). Of these, a total of 219,759 records are from mites, of which 181,682 have barcode sequences. Records with barcode sequence and identification to species level represent a total of 4,350 mite species, included in orders Holothyrida (1), Ixodida (334), Mesostigmata (992), Sarcoptiformes (769) and Trombidiformes (2,254). We have compared our sequences with those available on the platform to check the consistency of morphological identifications, detect potential cryptic diversity at the global scale and to investigate to what extent the diversity within our dataset is already represented in the BOLD repository.

First, we performed BOLD Identification System (IDS) (default setting parameters) for each sequence against the overall BOLD system (21/04/2023), including public and private barcode records. We extracted the taxonomic identification, based on morphology (species or genus level) and the similarity percentage to the best match using the BOLD Identification System (IDS) results for each sequence. Using this information: i) we estimated the overall similarity with BOLD sequences for each of our species and ii) we checked the coherence between the taxonomic identifications from BOLD and the morphological identifications of our specimens. In the cases where species-level identification agreed between both datasets, the overall similarity, the monophyly of the Canary Islands sequences and the geographical origin of BOLD sequences were evaluated to identify potential cryptic diversity within those mite species. Finally, as an additional indicator of how the diversity in our dataset is already reported within BOLD, the number of BINs that were only composed by our sequences (i.e. not including sequences already present on the repository) was recorded.

The different molecular species delimitation methods implemented resulted in a range from 60 to 85 molecular species delimited. The barcode index number (BIN) implemented in the BOLD System showed that our 168 voucher sequences were grouped in 71 BINs. Sixty-five of those BINs were newly generated by BOLD and assigned to sequences contributed in this study, while 11 sequences were grouped into six pre-existing BINs (access numbers in Table 1). ASAP analysis grouped our sequences in 60 molecular species according to the lower ASAP score, whereas PTP analysis delimited 71 molecular species for the mlPTP approach. Finally, the 3% genetic distance threshold resulted in 70 groups and the 8% genetic distance threshold in 64 (Figs 3, 4).

Figure 3.  

Distance-based UPGMA tree obtained using HKY corrected distances. Coloured horizontal blocks over the tree represent specimen clusters corresponding to morphological species. Vertical bars represent, from left to right: (i) morphological species, (ii) BINs classification in BOLD, (iii) species delimitation with ASAP, (iv) species delimitation with mlPTP, (v) 3% similarity clusters and (vi) 8% similarity clusters. At the bottom, each method's total number of species is presented. The X-axis represents genetic distance; with dotted lines corresponding 3% and 8% divergence thresholds (first half).

Figure 4.  

Distance-based UPGMA tree obtained using HKY corrected distances. Coloured horizontal blocks over the tree represent specimen clusters corresponding to morphological species. Vertical bars represent, from left to right: (i) morphological species, (ii) BINs classification in BOLD, (iii) species delimitation with ASAP, (iv) species delimitation with mlPTP, (v) 3% similarity clusters and (vi) 8% similarity clusters. At the bottom, each method's total number of species is presented. The X-axis represents genetic distance; with dotted lines corresponding 3% and 8% divergence thresholds (second half).

The concordance amongst the results of the different molecular delimitations implemented was relatively high. BINs, mPTP and 3% approaches resulted in a higher (but mostly concordant) number of delimited entities (Figs 3, 4). In all cases, molecular species delimitations exceeded the 58 entities morphologically delimited. In general, morphologically identified species and molecular species groupings were coincident, but at least eleven morphologically delimited species consistently included multiple molecular entities in at least one delimitation method (Phthiracarus cf. ligneus, Mesotritia cf. grandjeani, Gustavia longirostris, Damaeus recasensi, Galumna alata, Amerus cuspidatus, Eupelops acromios, Zygoribatula propinqua, Campachipteria petiti, and Acrogalumna longipluma) (Figs 3, 4).

Of the 45 morphological entities identified as known species in our dataset, 31 species (68%) were already registered as present in the archipelago, whereas the remaining 14 (31%) represented new species and genera records at the archipelago level (Table 1, Fig. 5a).

Figure 5.  

Evaluating unrecorded diversity within the Canary Islands and global context. a Proportion of species in our dataset that are already registered in the BIOTA database; b Category of origin (i.e. native non-endemic, introduced or endemic) for species recorded in BIOTA as reported within the database; c Similarity values of best matches of obtained sequences representing each species against BOLD Systems; d Species-level identification coherence between the specimens in our dataset and BOLD best matches. When there is no coherence, we specify if the BOLD best match was identified at species or higher taxonomic level (genus or family level).

Of the 31 species in our dataset that are present in the BIOTA, six species are endemic to the Islands (19%) and the rest are considered non-endemic native species (24 species, 77%) or introduced species (one species, 3%) (Table 1, Fig. 5b).

BOLD Identification System (IDS) results for the 168 sequences in our dataset resulted in a range of similarity percentages, with the best matches ranging from 74.48% to 99.5%. Amongst the 45 morphological entities identified as known species, 38 (84%) reported a similarity below 92%, four (9%) from 92% to 97% and three (7%) above 97% (Table 1, Fig. 5c). Regarding the taxonomic identity of the BOLD best matches for our sequences, the best match has no species-level identification for 26 (58%) entities. For the remaining cases where best BOLD matches have species-level identifications, three (6%) are concordant with the morphological identifications of our specimens and 16 (35%) showed no concordant species-level identifications (Table 1, Fig. 5d).

The only cases where species-level identification agrees with the BOLD dataset are Acrogalumna longipluma, Hypochthonius luteus and Odontocepheus elongatus. In the case of Acrogalumna longipluma, the overall similarity of Canarian specimens with BOLD sequences was 96.45%. In the case of Hypochthonius luteus, the overall similarity of our Canarian specimens with the only one registered on BOLD was 96.34%. Finally, in the case of Odontocepheus elongatus, the overall similarity of Canarian specimens with BOLD sequences was 74.48%.

Our study initiates the Barcode Database of soil fauna from the Canary Islands (CISoilBiota) by developing a standardised workflow that combines specific soil sampling, Berlese extraction, sample sorting, COI barcoding and traditional taxonomic identification of barcoded specimens. The workflow has been applied to 239 mite specimens, of which we recovered 168 sequences. This represents a success of 70%, similar to success rates in other barcoding studies (e.g. deWaard (2019), Salces-Castellano et al. (2021), Suárez et al. (2022), Suárez et al. (2023), Caterino and Recuero (2023)). We did not detect any pattern amongst failures regarding taxonomical assignments or geographic distribution of soil samples. Causes of failures may be indicative of a low quantity of DNA retrieved, considering that we worked with minute mesofauna specimens or poor quality of DNA, as specimens were collected using Berlese apparatus with water on the collecting recipients. Each Berlese apparatus was revised and specimens were transferred to ethanol every two days to minimise the degradation of DNA. Still, we cannot discard DNA degradation as affecting PCR performance in some cases. Although extracted DNA quantification and further dilutions or reconcentration will help to obtain a higher success rate, we consider this 70% success rate as a good starting point for further development, considering inherent difficulties of DNA work with small-sized soil mesofauna.

One of these difficulties is the incompatibility of the procedures used for the morphological study of these minute organisms (requiring microscopic preparations where specimens are cleared and fixed with different chemical products) and DNA preservation. Here, we solve that by implementing a protocol of specimen imaging and non-destructive DNA extractions for mites that allow the morphological study of the specimens after DNA extraction. Our results demonstrated that non-destructive DNA extraction of soil mites is feasible without compromising the morphological integrity of specimens.

Another difficulty in implementing barcoding to soil mesofauna is associated with the reduced body size of specimens and the low DNA concentration retrieved. The DNA extraction and PCR protocol performed here appears adequate under these low DNA conditions. DNA extraction was implemented using a magnetic-bead approach in a robotic platform; this semi-automated approach is optimal for implementing arthropod barcoding because it facilitates the standardised processing of high numbers of specimens while maximising the quality of DNA extracts for long-term storage (Arribas et al. 2022). We reduced reagent volumes in half without an evident impact on DNA extraction performance. Further tests with more reduced volumes would be desirable to minimise costs associated with DNA extraction. Finally, the high phylogenetic diversity within soil mites could challenge the selection of the primer sets for PCR amplification of the barcode fragment. Our results, aligned with previous studies (Young et al. 2012, Arribas et al. 2016, Arribas et al. 2019), demonstrate an overall good performance of the Folmer degenerate barcode primers (Yu et al. 2012) for the broad diversity of soil mites.

We expect that barcoding effort over soil mites can be additionally improved by the application of High Throughput Sequencing (HTS) approaches, at the same time that costs are reduced (Srivathsan et al. 2018, Creedy et al. 2021, Srivathsan et al. 2021, Emerson et al. 2022). Although Sanger sequencing is an efficient and optimised technique, multiplexing approaches combined with the strength of HTS can be used to generate thousands of barcode sequences in a faster and cheaper way (Srivathsan et al. 2019, Srivathsan et al. 2023, Vasilita et al. 2023). The rigorous implementation of HTS barcoding methodologies on the remarkably abundant and hyperdiverse soil mesofauna of arthropods holds great promise in addressing the global lack of knowledge on soil biodiversity. Still, to maximise the utility of obtained sequences as barcode references, protocols as proposed here, are fundamental, as the link amongst obtained sequences, properly preserved voucher specimens and images is maintained. These procedures include costly and time-consuming steps for specimen sorting, imaging and puncturing before DNA extraction, in addition to requiring great taxonomic expertise. We acknowledge that morphological identification is the main bottle-neck for the whole approach and, in agreement with Srivathsan et al. (2021), we envision a system where HTS barcoding can be used to obtain barcode sequences from high numbers of soil mesofauna specimens that, after the application of DNA similarity clustering and molecular identification methods, can be subsampled for a detailed morphological study of a much-reduced number of representative specimens.

The molecular species delimitations showed broad consistency amongst them and with morphological species identifications in our dataset (Figs 3, 4). ASAP and 8% threshold showed a higher agreement in delimited entities between them and with the morphological identification, whereas BINs, mlPTP and 3% resulted in additional splits for some lineages, likely representing over-splitting as previously reported in other studies (Copilaş-Ciocianu et al. 2022, Ranasinghe et al. 2022). Still, the comparison of the more conservative ASAP and 8% threshold with the morphological identification allowed us to detect several cases of inconsistency and potential cryptic diversity. The taxonomic challenge posed by cryptic species (two or more morphologically similar species classified as a single species) has been recognised as a significant limitation in quantifying soil mesofauna (Emerson et al. 2011, Pfingstl et al. 2021, Szudarek-Trepto et al. 2021, Yin et al. 2022). Different studies, implementing DNA sequencing, have revealed that cryptic diversity could be massive in specific soil lineages (Cicconardi et al. 2010, Cicconardi et al. 2013). In this study, despite our reduced sampling (168 barcodes from mite specimens), we evidenced several examples of potential cryptic diversity within the mites of the Canary Islands, where molecular delimitations identify multiple divergent lineages within a single morphological species.

We have detected a series of cases where specimens, morphologically identified as a species, show intraspecific divergences higher than 3%. Part of these cases consists of monophyletic lineages where internal divergences are higher than 3%, but lower than 8%. Here we found the cases of: (i) Phthiracarus cf. ligneus with three lineages with divergences over 6%; (ii) Damaeus recasensi with two lineages (one with a single specimen) with divergences above 5%; (iii) Eupelops acromios with two lineages with divergences above 3%; and (iv) Gustavia longirostris with two lineages with divergences above 3%. In these four cases, the moderately high intraspecific divergences found are compatible with a single species, which is also suggested by molecular species delimitation methods, such as ASAP (Figs 3, 4). In fact, high intraspecific variation can be expected for soil mites if we consider the huge population sizes reported for some soil taxa (Petersen and Luxton 1982, Endlweber et al. 2006) and the expectations from the neutral theory of molecular evolution, with genetic diversity increasing with a larger effective population size and the decreasing effects of drift (Kimura 1979). Still, these divergences also suggest that additional attention should be placed on these lineages, as they likely represent native species within the Canaries, which is a highly fragmented landscape that can contribute to geographic isolation and diversification (Juan et al. 2000). Alternatively, this pattern may also reflect the human-mediated introduction of new populations for Canarian native species or even cases of multiple introduction of non-native species. More detailed studies will be required to distinguish amongst the different alternatives for each lineage.

We have also detected other cases where specimens classified as a single species are split into different lineages with divergences higher than 8% for the barcode fragment (Figs 3, 4). These cases include: (i) Campachipteria petiti, represented by three specimens with divergences between 7.1 and 11.1%; (ii) Galumna alata with two lineages diverging 14.3%; (iii) Amerus cuspidatus represented by only two specimens having a divergence of 9.3%; (iv) Acrogalumna longipluma with two lineages with divergences of 11.5%; (v) Zygoribatula propinqua represented by only two specimens having a divergence of 17.5%; (vi) Acrotritia ardua, where the two morphological subspecies show molecular divergences of 21.7% and (vii) the two morphologically highly similar specimens classified as Archiphthiracarus sp., which are only distantly related phylogenetically (26.9% divergence). All these cases suggest speciation with reduced morphological differentiation, although deep mitochondrial DNA divergence has been also shown to not indicate distinct species in some lineages (e.g. Leo et al. (2010)). A detailed morphological study and the sequencing of additional genes from the nuclear genome and specimens from a wider geographical range will be needed to clarify their taxonomic status.

Further implementation of the proposed barcoding workflow within the Canary Islands will contribute to elucidating the status of the reported cases and, presumably, to detect additional cases of cryptic diversity. An integrative approach, with parallel and interactive morphological and molecular work, will contribute to accelerating species inventory and discovery. For example, in our dataset and within the genus Xenillus, two already-described species are detected, with additional specimens showing morphological variation not matching any described species. Molecular analysis shows consistent results with morphology, suggesting the existence of four additional new species within the genus with divergences above 13.2% (Figs 3, 4).

Beyond the prevalence of cryptic diversity within the soil mites of the Canary Islands, our results point to the generality of this pattern globally. The analyses comparing our sequences with the BOLD database found three paradigmatic cases of potential cryptic diversity within worldwide distributed species. The first one is the case of Acrogalumna longipluma, with available barcode sequences from Canada, Germany, Finland, UK and the Canary Islands, forming five differentiated geographically coherent lineages with similarities below 97%. Of these, two lineages are exclusively found in the Canary Islands with divergences higher than 12% and not showing a sister taxa relationship (Suppl. materials 5, 6). This pattern suggests that, under the taxonomic name Acrogalumna longipluma, we currently enclose a complex of species, that could include two endemic species for the Canary Islands and that requires a detailed morphological revision. The second case is that of Hypochthonius luteus, where the Canarian specimens form a lineage sister to a specimen from Belgium, with an overall similarity of 96.34%, again pointing to potential cryptic diversity within this morphologically described species (Suppl. material 7). In the third case, that of Odontocepheus elongatus, available sequences from Norway, Finland and the Canary Islands form three differentiated lineages with similarities below 75%, one of these lineages exclusively composed of Canarian specimens, showing again cryptic diversity within a highly similar morphology (Suppl. material 8).

If a reference database is poorly populated for a specific group, the probability of inacurate taxonomic assignment is higher and placement to high taxonomic ranks is frequent (Somervuo et al. 2017). Our results illustrate the high magnitude of unrecorded diversity within soil mites and associated issues when comparing obtained barcodes with reference databases. Most of the taxa within our dataset were not previously represented in the BOLD database, with most generated barcode sequences showing low similarity values with the best available matches in BOLD. In addition, a high proportion of best matches corresponded to specimens without a species-level identification in BOLD (26 cases) or to different species to those here identified morphologically (16 cases, all with low similarity). The lack of morphological identifications to the species level can be understood given: (i) the great species richness, abundance and local-scale heterogeneity of mesofauna communities, (ii) the poor taxonomic background knowledge of many soil lineages and geographic areas and (iii) the general lack of taxonomic expertise on soil mites (André et al. 1994, Decaëns 2010, Cameron et al. 2018, Guerra et al. 2020, White et al. 2020). Consequently, species widely distributed, naturally or by mean of human introductions, are likely better represented in public repositories. In contrast, those species locally endemic with small distributional ranges are represented to a lesser extent and only from a few geographical regions (Porter and Hajibabaei 2018). This lack of representation and taxonomic resolution in reference databases limits their potential to provide a reliable taxonomic assignment to newly-generated barcodes, highlighting the imperative need for further barcode projects on soil fauna integrating taxonomic expertise.

Within the context of the Canary Islands, our results point to a massive under-representation of the diversity of soil mites in biodiversity databases. The BIOTA database contains records for 474 species and subspecies of mites, of which 425 are considered native species and 49 introduced species. Of those species classified as native, 110 species are considered endemic to the Canary Islands and 104 species endemic to the Macaronesian Region. Regarding our data, 14 of the 45 (31%) species, for which we obtained a species-level identification, represent the first record for the Canary Islands, all of them also providing the first record at the genus-level. All of these are species known from outside the Canaries and are now reported to the Canaries for the first time. These species may correspond to native non-endemic species or introduced species, according to their known distribution outside the Canaries, but given the absence of reference sequences for most of these species, we cannot discard that they represent additional cases of cryptic diversity. For example, in our dataset, we found four cases in which, although it is not the best match in BOLD, there has been a match with a sequence of the same species. Acrotritia ardua ardua and A. penicillata have a 78.86-79.57% similarity with several sequences named Rhysotritia ardua (junior synonym) from Canada, Poland and Norway (see Suppl. materials 9, 10). Phthiracarus cf. globosus has a 77.78-77.89% similarity with sequences from Finland (see Suppl. material 11). Xenillus tegeocranus has a 78.15-80.07% similarity with sequences from Canada, Finland, Norway and Slovakia (see Suppl. material 12). These worldwide matches suggest that species that have been taxonomically assigned to a single species may constitute cryptic lineages. To advance our knowledge of the magnitude of biodiversity and level of endemicity of the Canary Islands, we will need further efforts in generating barcode reference sequences, with reliable taxonomic identification, from inside the Canaries, but also from other regions. Barcoding specimens unambiguously associated with a particular species considering both its morphology and geographic origin is needed. These data will be key to distinguishing amongst native, endemic and introduced species within the Canaries, allowing us to generate reliable local inventories of soil fauna.