A DNA barcode library for the water mites of Montenegro

Abstract Water mites (Acari, Hydrachnidia) are a significant component of freshwater ecosystems inhabiting a wide range of aquatic habitats. This study provides a first comprehensive DNA barcode library for the water mites of Montenegro. DNA barcodes were analysed from 233 specimens of water mites morphologically assigned to 86 species from 28 genera and 15 families. In the course of the study, four species, i.e. Lebertiareticulata (Koenike, 1919), Atractidesinflatipalpis K.Viets, 1950, A.latipes (Szalay, 1935) and Parabrachypodamontii (Maglio, 1924) were molecularly confirmed as new for Montenegro and three species, i.e. Protziaoctopora Lundblad, 1954, Pionalaminata (Thor, 1901) and Unionicolaypsilophora (Bonz, 1783) are new for the Balkan Peninsula. Results are analysed using the Barcode Index Number system (BIN) and the Refined Single Linkage (RESL) of BOLD. The BIN assigned sequences to 98 clusters, while the RESL reveal 103 operational taxonomic units (OTUs). Unique BINs were revealed for 72 species (83.7%), whereas twelve species (14%) were characterised by two BINs and two species (2.3%) with three BINs. Amongst the studied taxa, 14 species were found with a high intraspecific sequence divergences (˃ 2.2%), emphasising the need for additional comprehensive morphological and molecular analysis of these species.


Introduction
Hydrachnidia, also known as water mites, is a most diverse and abundant group of arachnids in freshwater habitats (Davids et al. 2007). With nearly 7,500 species grouped into 550 genera (Smit 2020), they inhabit a wide range of aquatic habitats, including lotic, lentic, interstitial and temporary waters. Water mites have a complex life cycle that includes two pupa-like resting stages, i.e. protonymph and tritonymph and three active stages: larva is almost always parasitic, deutonymphs and adults that are predators of minute invertebrates (Davids et al. 2007). Some recent studies have shown that water mites can be good indicators of ecosystem health, especially of groundwater-dependent ecosystems (Pešić et al. 2019b). However, their time-consuming taxonomic identification has been identified as a major constraint for more significant involvement in rapid assessment programmes (Weigand et al. 2019).
Traditional morphology often underestimates the true diversity of water mites and, in recent years, it has been successfully replaced by an integrative approach that combines both morphological characteristics and molecular data (Martin et al. 2010, Pešić et al. 2017, Fisher et al. 2017, Pešić et al. 2019a, Pešić et al. 2020d, Pešić and Smit 2020. This process has been enhanced by the formation of the comprehensive DNA barcode reference libraries, such as the BOLD System (https://www.boldsystems.org/) and GenBank (https://www.ncbi.nlm.nih.gov/). DNA barcodes have been proposed and successfully adopted for water mites as an efficient method for detecting previously overlooked and/or misidentified species (Martin et al. 2010, Pešić et al. 2017. The significant increase in the number of studies using DNA barcodes in recent years, especially in some regions, has laid the foundations for building a comprehensive library of DNA barcodes at the national and/or regional level (e.g. Blattner et al. 2019).
Of the Balkan countries, Montenegro is one of the best studied from the taxonomic and faunistic point of view . Water mite research began in 1903 when the Czech zoologist Karl Thon published the first list of 13 species (Thon 1903). For more than one century, a large number of papers on the Montenegrin water mites have been published (Musselius 1912, Viets 1936, Pešić 2001, Pešić 2002b, Pešić 2002d, Pešić 2002a, Pešić 2002c, Pešić 2003a, Pešić 2003c, Di Sabatino et al. 2003, Pešić 2003b, Pešić 2004b, Pešić 2004a, Smit and Pešić 2004, Baker et al. 2008, Pešić et al. 2010, Pešić et al. 2012, Pešić et al. 2017, Bańkowska et al. 2016, Pešić et al. 2019a, Pešić et al. 2019c, Pešić et al. 2020a, Pešić et al. 2020b, Pešić et al. 2020c, Pešić et al. 2020d, Zawal et al. 2020, Pešić and Smit 2020, Pešić et al. 2021b Currently, 201 species of water mites have been reported for Montenegro , Pešić et al. 2019c, Pešić et al. 2020c, Pešić et al. 2020d, Pešić and Smit 2020. This number makes up about 50% of the species known from the Balkans, which is estimated at about 400 species . This is still a small number for the area of such hydrogeological characteristics and the turbulent geological history as the Balkans. Therefore, there is no reason not to believe that the expected number of water mites in the Balkans is at least at the level of Central Europe which is home to approximately 745 species . The aim of the study is to develop and evaluate the first library of barcodes for water mites from Montenegro, targeting a COI fragment of ~ 658 bp. Taking advantage of publicly available DNA barcode reference libraries, such as the BOLD and the use of the universal Barcode Index Number (BIN), allows us to assess the molecular diversity of water mite species inhabiting the territory of Montenegro, as well as to explore their distribution patterns in Europe. Moreover, this approach will allow us to also identify problematic species groups both for traditional taxonomy and for DNA barcoding.

Material and methods
Water mites were collected by hand netting, sorted live in the field and immediately preserved in 96% ethanol (EtOH) for the molecular analysis. Water mites were collected from 54 sampling sites in Montenegro ( Fig. 1) during several sampling campaigns from 2018-2020. Photos from each studied specimen were taken before molecular work started. The photographs were made using a camera on a Samsung Galaxy smartphone.
Molecular analysis were conducted in the Canadian Centre for DNA Barcoding (Guelph, Ontario, Canada; (CCDB; http://ccdb.ca/) and in the Department of Invertebrate Zoology and Hydrobiology (DIZH), University of Łódź, Poland. For the methods used for cytochrome c oxidase subunit I (COI) gene amplification in DIZH, see Pešić et al. (2017). In CCDB, the specimens were sequenced for the barcode region of COI using standard invertebrate DNA extraction (Ivanova et al. 2007), amplification (Ivanova and Grainger 2007a) and sequencing protocols (Ivanova and Grainger 2007b). The DNA extracts were archived in −80°C freezers at the Centre for Biodiversity Genomics (CBG; biodiversitygenomics.net) and the specimen vouchers were stored in 95% EtOH and returned to the first author for morphological examination. Some of these vouchers were dissected as described elsewhere (Davids et al. 2007) and slide-mounted in Faure's medium, while the rest were transferred to Koenike's medium and stored in the collection of the first author at the Department of Biology in Podgorica.

DNA barcode analysis
In CCDB, the chromatograms were assembled into consensus sequences for each specimen and uploaded to BOLD. The taxonomic account, voucher specimen ID, collecting locality and voucher depositor were incorporated into the system for further analysis. Water mite sequences, obtained during this study, were grouped in the "MNHYD" (DNA barcode reference library of Montenegrin water mites) dataset. Detailed voucher information, taxonomic classifications, photos, DNA barcode sequences, primer pairs used and trace files (including their quality) were uploaded to the dataset "MNHYD" on the Barcode of Life Data Systems (BOLD; www.boldsystems.org).
The translation of the COI sequences into amino acids did not contain any stop codon positions and blasting the sequences confirmed the absence of contaminations. In cases of the four Unionicola ypsylophora mites, we amplified Anodonta exulcerata DNA instead of water mite DNA. These specimens were excluded from further analysis.
The reference library for the molecular identification of water mites sequenced in this study was analysed using the Barcode Index Number system (BIN) (Felsenstein 1985). The distribution of BINs was performed by the Barcode of Life Data System v.4 (accessed 15 November 2021). The two-phase BIN analysis system in the first phase applies a first threshold of 2.2% (that allows a rough differentiation between intraspecific and interspecific distances), followed by refinements through Markov clustering into the final BINs (Ratnasingham and Hebert 2013). BOLD ID and accession numbers for all specimens included in final dataset are given in Table 1. Sampling sites from Montenegro. The green colour represents the Danube Basin (Black Sea) and the yellow colour represents the Adriatic Basin.  (Kumar et al. 2018). The Neighbour-Joining (NJ) tree (edited in MEGA7, Kumar et al. 2016), based on K2P distances and pairwise deletion of missing data, was used to visualise similarity. The support for tree branches was calculated by the non-parametric bootstrap method (Felsenstein 1985) with 1000 replicates and shown next to the branches.

Results
DNA barcodes of 233 specimens morphologically assigned to 86 species from 28 genera and 15 families of water mites from Montenegro were newly generated for this study. The specimens were collected through the "DNA-Eco" (DNA barcode reference library as a tool for sustainable management of freshwater ecosystems in the highly threatened Lake Skadar Basin) project. The current study develops the first COI barcode reference library of water mites for Montenegro with the focus on Skadar/Shkodra Lake catchment area.
Fragment lengths of the analysed DNA barcode fragments ranged from 201 to 658 (mean: 636.2) base pairs, including no stop codons, insertions or deletions. The DNA barcode region was characterised by a high AT-content: the mean sequence compositions were A = 30.82 ± 0.1252%, C = 20.39 ± 0.1222%, G = 14.91 ± 0.0709% and T = 33.88 ± 0.1253%. The obtained results are similar to those found in other arthropod studies (e.g. Raupach et al. 2015).
The families Hygrobatidae Koch, 1842 and Lebertiidae Thor, 1900 are represented by the highest number of sequences (53 and 44, respectively). The opposite, the three families Arrenuridae Thor, 1900, Teutoniidae Koenike, 1910and Limnesiidae Thor, 1900 are represented each with two sequences and the two familes Athienemanniidae K. Viets, 1922 andWettinidae Cook, 1956 by the lowest number of sequences (each with one sequence). The most common genus was Lebertia Neuman, 1880, for which 44 barcode sequences (11 species) were generated, followed by Atractides Koch, 1837 (35 barcodes; 13 species), Torrenticola Piersig, 1896 and Sperchon Kramer, 1877 (29 and 26 barcodes, 10 and 8 species, respectively). Six genera were represented by a single specimen. The highest number of barcodes per species was reached for Atractides pennatus (K. Viets, 1922), Sperchon violaceus Walter, 1944 and Torrenticola meridionalis Di Sabatino and Cicolani, 1990 (each with 10 barcodes), followed by Lebertia inaequalis (Koch, 1837) and L. variolata Gerecke, 2009 (each with 8 barcodes) and Sperchon thienemanni Koenike, 1907 (6 barcodes). On the other hand, most species are represented by less than 5 DNA barcodes. Thirty-three species are represented by a single DNA barcode not allowing us to estimate the intraspecific distances. BOLD ID and accession numbers for all specimens included in final dataset are given in Table 1.
The mean intrageneric K2P distance was 20.2 ± 0.0% (range 6.09-42.37%). The mean intraspecific nucleotide K2P distances were 2.43 ± 0.01% (ranging from 0% to 24.16%). The summary statistics showing significant changes of average K2P distances within the different taxonomic levels are given in Table 2.  The BIN and RESL (OTU) analyses assigned sequences to 98 BINs and 103 OTUs, respectively. Fifty BINs (159 records) were concordant (51%) and 48 BINs were represented by a single sequence (49%). At the time of publication of the dataset, fifty-five (56.1%) of these BINs (with 102 sequences) included sequences only from Montenegro, while the remaining BINs included sequences also from other countries.
Most of the morphologically-identified species show an intraspecific variation of less than 2%. However, the 14 taxa listed in Table 3 showed a maximum interspecific divergence larger than 2%, resulting in these species in BOLD being spread over more than one BIN. Two species, Lebertia glabra Thor, 1897 and L. inaequalis appeared each with 3 BINs and twelve species, i.e. Lebertia maculosa Koenike, 1902, L. porosa Thor, 1900 (Soar, 1900), each with 2 BINs (Table 3). In total, unique BINs were revealed for 72 species (83.7%), two BINs for 12 species (14.0%) and three BINs for two species (2.3%).
The NJ analyses, based on K2P distances, revealed non-overlapping clusters with bootstrap support values > 95% for 50 species (58%) with more than one analysed specimen indicating a high congruence between BINs affiliation and morphological species identification. Moreover, specimens showing high intraspecific distances are also clearly separated into different clades. A more detailed topology of all analysed specimens is presented in the supporting information (Suppl. material 2).
Moreover, species identification, based on molecular data conducted during this project, extended the list of Montenegrin water mites by description of several species new for science, i.e. Atractides anae Pešić, 2020, Hygrobates lacrima Pešić, 2020, H. limnocrenicus Pešić, 2020, H mediterraneus Pešić, 2020and Mideopsis milankovici Pešić and Smit, 2020(Pešić et al. 2020a, Pešić et al. 2020c, Pešić et al. 2020d, Pešić and Smit 2020. All of these studies highlighted the importance of an integrated approach that combines the morphology-based taxonomy and DNA barcodes. Our study confirmed efficiency of DNA barcoding as a tool for the identification of water mites. In particular, 72 of the 86 morphologically-identified species exactly matched the BINs defined from BOLD. This result coincides with high identification efficiency rates through the BOLD Best Close Match analysis. Nevertheless, our data revealed also 14 species listed in Table 3 that showed high intraspecific distances (> 2.2%) suggesting possible cryptic and/or pseudocryptic diversification. Most of these possible cryptic and/or pseudocryptic species, as seen in Table 3, appear to be hidden within common species.
Three species, i.e. Lebertia maculosa, Monatractides madritensis and Torrenticola laskai appeared each with 2 BINs in our dataset. The intraspecific maximum distances between BINs within each of these species were below 3% (Suppl. material 1). On the other hand, the intraspecific maximum distances between BINs within each of the other eleven species in the dataset were greater than 5% (Suppl. material 1).  Stur (2017) showed that 18 specimens of L. porosa from Norway comprise 7 BINs with a mean intraspecific p-distance of 11.7% and maximum up to 18.5%. In our dataset, specimens, morphologically assigned to Lebertia porosa, were presented with two clusters. Based on the available records from BOLD, the first cluster (BIN:ACS0974) appeared to be well represented in the BOLD database with 133 records from different parts of Europe; the second cluster (BIN:AED4662) contained specimens only from Montenegro. In our study, specimens of the latter BIN were collected in large limnocrenic springs, such as Mareza and Vitoja, while specimens from the first cluster (BIN:ACS0974) were sampled in the lower reaches. The intraspecific K2P distance between these two L. porosa clusters in our dataset was estimated at 5.5% (Suppl. material 1).
Sperchon brevirostris, a species inhabiting low-and middle order streams in the study area (Pešić et al. 2010, was represented in our material by two clusters. Based on the available records from BOLD, the first cluster (BIN:ACP6107) includes specimens from Norway, Germany and one specimen from Montenegro, while the second cluster (BOLD:AED3857) contained three specimens from Montenegro and North Macedonia. The K2P distance between these two clusters was 8.1% (Suppl. material 1). Similarly, S. clupeifer, a species frequently reported from Western Palaearctic (Di , appeared with two clusters in our dataset. The first cluster (BIN:ACS1100) is well represented in BOLD and includes specimens from different part of Europe, while the second cluster (BIN:AEE4061) contained a single specimen from Montenegro. The intraspecific K2P distance between these two clusters in our dataset was estimated at 8.3% (Suppl. material 1).
Sperchonopsis verrucosa, a species often reported from the Holarctic Region (Gerecke et al. 2016), was represented in our study with two clusters. The first cluster (BIN:ACS9705) was more represented in BOLD and includes specimens from Norway, Italy and one specimen from Montenegro. The second cluster (BIN:AEK8297) includes two specimens from Montenegro and Romania. The intraspecific K2P distances between these two clusters was 11.2%, indicating the need for additional integrative analysis.
Torrenticola meridionalis, a species originally described from Italy, is widely distributed in Montenegro, inhabiting mainly low order streams . It is morphologically closely related to T. elliptica which remains distinguishable in the male sex only, based on the stouter genital field. In our COI tree (Suppl. material 2), T. elliptica appeared as a sister clade to the clade that includes two clusters morphologically assigned to T. meridionalis (BIN:AEI3402 and BOLD:AED7519, respectively). The intraspecific K2P distances between T. elliptica and T. meridionalis clusters ranged from 8.6-9.0%. On the other hand, the K2P distance between T. meridionalis clusters in our dataset was estimated at 6.6%.
Atractides nodipalpis, a rhitrobiontic species, is the most frequently reported species of the genus in Europe (Gerecke et al. 2016). In our dataset, sequences of the specimens, morphologically assigned to the latter species, appeared as two clusters. Interestingly, specimens of both clusters were recorded syntopically. The first cluster (BIN:ACR0209) in the BOLD database was represented with 41 specimens from Norway (country of the type locality), The Netherlands, Montenegro and Russia, but also from Greenland. The second cluster (BIN:AED3547) includes two specimens from Montenegro. The intraspecific K2P distance between these two clusters was 18.8%, indicating the need for a comprehensive revision of this species complex.
Hygrobates calliger, a rhitrobiontic species widely distributed in the Palaearctic (Di , was represented by two clusters in our dataset, each with two records in the BOLD database. The first cluster (BIN:AEF4261) includes specimens from Norway and Montenegro, while the second cluster (BIN:AEL5782) includes specimens from Germany and Montenegro. The intraspecific K2P distance between these two clusters was 20.9% (Suppl. material 1), suggesting the existence of possible hidden cryptic and/or pseudocryptic species.
The sequences of Atractides gibberipalpis, a rhitrobiontic species often reported from the Palaearctic (Pešić et al. 2021a), in our dataset were assigned to two different barcode clusters, each represented by a single specimen from Montenegro. The intraspecific K2P distance between these two clusters (BIN: BOLD:AEK7766 and BIN: BOLD:AEI3946, respectively) was estimated at 5% (Suppl. material 1).
Unionicola minor, a species widely distributed in Europe (Gerecke et al. 2016), was presented with two clusters in our dataset. Based on available data from BOLD, the first cluster (BIN:AAU0335) includes specimens from Norway and The Netherlands and one specimen from Lake Šasko in Montenegro. The second cluster (BIN:AAU0335) includes specimens only from Montenegro. The intraspecific K2P distances between these two clusters in our dataset was 23.8% (Suppl. material 1), suggesting the existence of cryptic (or pseudocryptic, see Pešić and Smit (2016) for a discussion about pseudocryptic speciation in water mites) species. Stålstedt et al. (2013) showed that the Swedish population of Unionicola minor consists of at least three cryptic species, emphasising the need for further research of the species in this complex.
Taxonomic studies of the above species were outside the scope of this paper. Further studies with material from a wider geographical area, were needed to clarify taxonomy and elucidate the delimitation of the species in the above complexes. This process should be accompanied by sufficient barcode coverage to allow the detection of phylogeographic patterns and/or even the existence of possible overlooked cryptic species. The build-up of DNA barcode library for water mites of Montenegro represents a long-term task, aimed at improving molecular identification and inclusion of this group in environmental assessment programmes and, on the other hand, to stimulate further biodiversity research of this limnofaunistic group in Montenegro and the Balkans.