The genus Microniphargus (Crustacea, Amphipoda): evidence for three lineages distributed across northwestern Europe and transfer from Niphargidae to Pseudoniphargidae

Microniphargus leruthi Schellenberg, 1934 (Amphipoda: Niphargidae) was first described based on samples collected in Belgium and placed in a monotypic genus within the family Niphargidae. However, some details of its morphology as well as recent phylogenetic studies suggest that Microniphargus may be more closely related to Pseudoniphargus (Amphipoda: Pseudoniphargidae) than to Niphargus. Moreover, M. leruthi ranges over 1,469 km from Ireland to Germany, which is striking since only a few niphargids have confirmed ranges in excess of 200 km. To find out the phylogenetic position of M. leruthi and check whether it may be a complex of cryptic species, we collected material from Ireland, England and Belgium then sequenced fragments of the mitochondrial cytochrome c oxidase subunit 1 gene as well as of the nuclear 28S ribosomal gene. Phylogenetic analyses of both markers confirm that Microniphargus is closer to Pseudoniphargus than to Niphargus, leading us to reallocate Microniphargus to Pseudoniphargidae. We also identify three congruent mito-nuclear lineages present respectively in Ireland, in both Belgium and England, and in England only (with the latter found in sympatry at one location), suggesting that M. leruthi is a complex of at least three species with a putative centre of origin in England.


Introduction
Microniphargus leruthi Schellenberg, 1934 (family Niphargidae) was first described from Engihoul Cave in Wallonia (Belgium) and placed into a new, monotypic genus considered as closely related valachicus Dobreanu & Manolache, 1933 (> 3,200 km), two epigean species with enhanced dispersal via surface water (Copilaş-CioCianu et al. 2017). The wide range of M. leruthi could therefore be due to the presence of undetected species boundaries.
To resolve these uncertainties, we conducted a molecular study on M. leruthi collected in Ireland, England and Belgium using both 28S (nuclear) and COI (mitochondrial) markers. Our aims were (i) to confirm the phylogenetic position of Microniphargus relative to the genera Niphargus and Pseudoniphargus and (ii) to test for the possible existence of cryptic lineages within M. leruthi.

Sampling and sequencing
Although we carried out intensive and targeted sampling for M. leruthi, especially in caves around the type locality, we were only able to collect it at a single location on the European continent: the Grotte de Comblain (Wallonia, Belgium), which is 20 km away from the type locality. We also collected M. leruthi from one site in Ireland (Polldubh, Clare) and two sites in England (Sweetwater Pot, South Devon and Swildon's Hole, Somerset; Fig. 1). All the material (Table 1) was determined morphologically by one of us (L.K.). Specimens were collected by sweeping a long-handle net fitted with a 250 µm mesh collecting bag along the bottom and sides of cave pools, making sure to disturb the substrate to suspend both  sediment and specimens into the water column. The collected specimens were immediately preserved in 96% ethanol and kept at -20°C until DNA was isolated.
Due to the small size of M. leruthi, we used one entire specimen for each DNA isolation. DNA was extracted following the standard protocol of the NucleoSpin® Tissue Kit (Macherey-Nagel) except that we performed two elution steps, the first one with 60 µL and the second with 40 µL (instead of a single elution step with 100 µL) to achieve a higher concentration of DNA. The resulting DNA isolates are stored at -20°C in the collections of the Evolutionary Biology & Ecology research unit at the Université libre de Bruxelles (ULB).
We also sequenced Verovnik's fragment of the nuclear 28S ribosomal gene. The primers Niph15 and Niph16 (see Table 2) were used for amplification (verovnIK et al. 2005). The PCR mix for 28S contained 2μL of DNA template (variable concentration), 1μl of each primer (10pmol/μL), 0.2μL of REDTaq DNA Polymerase (Sigma-Aldrich), 5µL REDTaq reaction buffer and 15.8µL ultrapure water. PCR cycling conditions for 28S were an initial 3min denaturation step at 95°C; followed by 56 cycles of 30s denaturation at 94°C, 60 s annealing at 45°C, and 90s extension at 72°C.
The amplification success of each PCR reaction was verified using agarose gel electrophoresis, then PCR products were sequenced at Genoscreen (Lille, France). For COI the primer used for sequencing were the same as for PCR amplification, whereas for 28S we used the primers Niph20 and Niph21 (FLot et al. 2010b) as well as one or both of two new internal primers located slightly inward of the primers used for initial amplification (Niph15i and Niph16i; see Table 2).
The resulting chromatograms were assembled and cleaned using Sequencher version 4.1.4 (Gene Codes, USA). Whenever double peaks were observed in both the forward and reverse chromatograms of an  TABLE 2 Primers used in the present study.

Primer Bases
Marker PCR Sequencing Reference individual, we considered this individual as polymorphic and called its two haplotypes (determined using the approach summarized in Fontaneto et al. 2015) "a" and "b" in downstream analyses.

Phylogenetic and species delimitation analyses
We compiled comprehensive sets of COI and 28S including all sequences available in GenBank to date, then curated them manually to remove duplicates. The resulting set of 1384 COI sequences was aligned manually, whereas for the 255 sequences of 28S (including two gammarids Gammarus fossarum and Gammarus pulex and two crangonyctids Crangonyx subterraneus and Synurella ambulans as outgroups) we used MAFFT 7's E-INS-i mode (Katoh et al. 2019

Results
For both COI and 28S, we successfully sequenced nine Microniphargus leruthi specimens. For COI, four individuals (one from Belgium and two from England) displayed one double peak each and were therefore represented by two sequences 'a' and 'b' (with a single base difference between them) in all downstream analyses. One M. leruthi individual (from Belgium) displayed a double peak in its 28S chromatograms and was therefore represented by two sequences (with a single base difference between them) in all downstream analyses, whereas all other individuals were homozygous for the 28S marker (Fig. 2). Figure 3 -COI maximum-likelihood phylogeny of Microniphargus and Pseudoniphargus (with two Niphargus and two crangonyctids as outgroups). The tree was turned into a haploweb by adding connections between haplotypes found co-occurring in the same individual.
The COI phylogeny supported a (Microniphargus + Pseudoniphargus) clade with 96% of ultrafast bootstrap replicates ( Fig. 3 and Fig. S1) and revealed Microniphargus to be composed of three main clades A (found only in Ireland), B (found both in Belgium and in England) and C (found only in England), with > 99% ultrafast bootstrap support for each of them. Clade B contained two subclades comprising respectively Belgian and English sequences, also with > 99% ultrafast bootstrap support. Clade B and C co-occurred at one sampling site (Fig. 1). The sister-clade relationship between Pseudoniphargus and Microniphargus was supported with 100% of ultrafast bootstrap replicates in the comprehensive 28S phylogeny, which supported also the monophyly of lineages A and C (with 100% and 96% bootstrap replicates, respectively) but not of B, which was paraphyletic using this marker ( Fig. 4 and Fig. S2).
The COI lineages A, B and C were separated by average p-distances of 0.073-0.081 between A and B, 0.072 between A and C, and 0.066 between B and C; whereas the p-distance between the two sublineages of B was 0.029. ABGD's initial partitioning of our comprehensive COI dataset supported a three-species hypothesis for Microniphargus leruthi, whereas the recursive partitioning favoured a fourspecies hypothesis separating the Belgian and English sub-lineage of lineage C. The KoT analysis of Figure 4 -28S maximum-likelihood phylogeny of Microniphargus, Niphargus and Pseudoniphargus (with two crangonyctids as outgroups). The tree was turned into a haploweb by adding connections between haplotypes found co-occurring in the same individual.  the Microniphargus COI sequences supported a four-species scenario as well, with a K/θ ratio of 19.3 between the two subclades within lineage B, itself separated by a gap with a K/θ of 14.0 from lineage C, and finally separated by a gap with a K/θ ratio of 16.8 from lineage A (Fig. 5). By contrast, the 28S haploweb revealed three fields for recombination (FFRs sensu DoyLe 1995, i.e., putative species following the criterion of mutual allelic exclusivity; FLot et al. 2010a), corresponding to clades A, B and C (Fig. 6).

Discussion
Key novel, high-quality sequences were acquired  . 2017) is 100% identical to our complete sequences from Ireland, but with the first 59 bp and last 156 bp lacking; whereas the three Pseudoniphargus sequences available till now were also highly incomplete. The high-quality 28S and COI sequences we obtained from representative individuals of C. subterraneus from Germany, Pseudoniphargus italicus from Sicily and P. spiniferus from Basses Pyrénées in France, as well as from each of the three lineages of M. leruthi identified in our study, will make it easier to include these species as outgroups in future studies of Niphargus, Pseudoniphargus and other related genera.

Both COI and 28S sequences of Microniphargus were found to contain double peaks
Out of the nine M. leruthi individuals whose COI marker was sequenced, four (three from Belgium and one from England) presented a double peak in their COI chromatograms, resulting in an intraindividual polymorphism level of 0.15% in these individuals. For the three Belgian specimens the double peak was an R = A or G transition in position 101, whereas for the English specimen it was an S = C or G transversion in position 189 (Fig. 4). These mutations were not synonymous but corresponded to N (asparagine) ↔ D (aspartate) and A (alanine) ↔ G (glycine) mutations in the translation amino acid sequences. Such mitochondrial double peaks are rare in niphargids: for instance, no double peak was observed in the COI chromatograms of the 67 Romanian specimens sequenced in FLot et al. (2014) nor reported for any of the hundreds of niphargids sequenced in eme et al. (2018). The presence of two distinct COI sequences in M. leruthi individuals may be the result of heteroplasmy, i.e., the presence of two distinct mitochondrial lineages in the cells of an organism, or of a recent numt, i.e., a nuclear pseudogene of a mitochondrial sequence following the transfer and integration of a copy of this sequence in a nuclear chromosome (DIercKxSenS et al. 2020). Determining which one of these two hypotheses is correct in the present case will require whole-genome sequencing, which is beyond the scope of the present study, but in any case, the very limited divergence between the COI sequences found cooccurring in some individuals (with a single double peak per individual) did not hinder downstream phylogenetic analyses.

The genus Microniphargus is more closely related to Pseudoniphargus than to Niphargus
Our COI and 28S phylogenetic trees confirm that all collected specimens assigned to the morphospecies Microniphargus leruthi form a monophyletic group that is clearly distinct from Niphargus and Pseudoniphargus, thereby confirming its status as a separate genus previously established on the sole basis of morphological characters (ScheLLenBerg 1934 As mentioned in the introduction, a similarity between the two genera can be found in the shape of the telson (which is widely incised and carries one spine on each lobe), and also partly the shape of gnathopod 1. This shape of telson as well as the protrusion on the carpus of gnathopod 1 are found also in Bogidiellidae (another family placed in recent phylogenetic trees not far away from the clade Niphargidae+Pseudoniphargidae: Copilaş-CioCianu et al. 2020) and may be simply symplesiomorphic, in which case the deeply incised, bilobated telson of Niphargus would represent an apomorphic character of this genus. However, the small size of Microniphargus, the reduced setation of mandibular palp and gnathopods, the lack of elongation of the third uropod in males, and the 1-articulated accessory flagellum of antennulae suggest a major role of paedomorphosis, making it difficult or impossible to correctly allocate this genus within current amphipod taxonomy and phylogeny based on morphological characters alone.

Microniphargus leruthi comprises at least three cryptic lineages
Our COI phylogeny, ABGD's initial partitioning of our comprehensive COI dataset and our haploweb analysis of 28S sequences of Microniphargus support the hypothesis that Microniphargus leruthi is composed of three distinct, putatively species-level lineages: clade A found in Ireland, clade B found both in England and in Belgium (with two COI sub-clades consistent with the geographic distance between these two locations), and clade C found so far only in England. By contrast, ABGD's recursive partitioning supports a four-species hypothesis, and so does the KoT analysis. However, the p-distances between the three main lineages A, B and C are all well above the 3% species-level threshold traditionally considered in barcoding studies (heBert et al. 2003), whereas the average p-distance between the two COI sub-clades of B falls below this symbolic threshold. These arguments, together with the fact that all individuals of lineage B (and only these individuals) display double peaks in their COI chromatograms, lead us to consider tentatively the two sub-clades of lineage B as conspecific and therefore to distinguish at present only three putative species-level lineages within M. leruthi.
Although lineage A (found only in Ireland to date) appears geographically separated from the other two, lineages B and C occur in sympatry in at least one location (Swildon's Hole in Somerset), bringing further support to the hypothesis that these two lineages are distinct species. The phylogenetic analysis based on COI could point to an origin of the genus Microniphargus in England with subsequent dispersals to Ireland and to Belgium; however, more samples and analyses will be required to test this hypothesis. The fact that lineage B still occurs on both sides of the English Channel is not overly surprising since the land connection between England and continental Europe was only severed about 8,000 years ago (WaLLer & Long 2003).
The hypothesis that the three Microniphargus leruthi lineages identified here represent distinct cryptic (or pseudo-cryptic) species will need to be tested further. Doing so will require further collecting and sequencing, as well as detailed morphological analyses using microscopy techniques appropriate for such small specimens. Figure S2 -Detailed version of the 28S maximum-likelihood phylogeny.

taBLe S1
List of all the sequences included in the COI phylogeny, including species names and GenBank accession numbers.