Overlooked Species Diversity and Distribution in the Antarctic Mite Genus Stereotydeus

: In the harsh Antarctic terrestrial ecosystems, invertebrates are currently conﬁned to sparse and restricted ice free areas, where they have survived on multi-million-year timescales in refugia. The limited dispersal abilities of these invertebrate species, their speciﬁc habitat requirements, and the presence of geographical barriers can drastically reduce gene ﬂow between populations, resulting in high genetic differentiation. On continental Antarctica, mites are one of the most diverse invertebrate groups. Recently, two new species of the free living prostigmatid mite genus Stereotydeus Berlese, 1901 were discovered, bringing the number of Antarctic and sub-Antarctic species of this genus up to 15, of which 7 occur along the coast of Victoria Land and in the Transantarctic Mountains. To examine the biodiversity of Stereotydeus spp., the present study combines phylogenetic, morphological and population genetic data of specimens collected from nine localities in Victoria Land. Genetically distinct intraspeciﬁc groups are spatially isolated in northern Victoria Land, while, for other species, the genetic haplogroups more often occur sympatrically in southern Victoria Land. We provide a new distribution map for the Stereotydeus species of Victoria Land, which will assist future decisions in matters of the protection and conservation of the unique Antarctic terrestrial fauna.


Introduction
Due to Antarctica's isolation and extreme environmental conditions, the continent's terrestrial biota has limited species level diversity and many higher taxonomic groups are completely missing or very poorly represented [1,2]. As a result of the climatic factors and the typically low availability of organic nutrients in soils, lichens and mosses are the only macroscopic flora present on the continent [1,[3][4][5][6]. Similarly, the Antarctic terrestrial fauna consists of a small number of microarthropod species (mites and springtails) as well as other microscopic invertebrates (nematodes, tardigrades and rotifers), making the continental region amongst the simplest ecosystems on Earth [2,7].
The challenging environmental conditions, isolation and the patchy distribution of ice free areas have been recognized as the main factors affecting and defining populations of the Antarctic terrestrial invertebrate fauna, both physiologically and genetically [8]. As a consequence, in order to survive the harsh Antarctic conditions, these terrestrial animals have evolved impressive biochemical and physiological adaptations, to tolerate prolonged periods of freezing and dry conditions, amongst other severe stresses [9][10][11][12]. Behavioral strategies also play a role. For instance, continental Antarctic springtails (Collembola) and At the same time, given the recent discovery of the two new Stereotydeus species in Northern Victoria Land in a study that also reviewed the morphological characters relevant to the identification of Antarctic Stereotydeus species [48], the question of a possible overlap between these new taxa with the species already known from the area (S. belli, S. punctatus and S. delicatus) and with S. mollis from Southern Victoria Land has to be addressed. In addition to that, the current lack of genetic knowledge of a species morphologically described more than fifty years ago needs addressing, not only for the systematic understanding of the genus, but also to contribute to the future development and implementation of sustainable conservation planning in Antarctica. Although Antarctica is often assumed to be a pristine continent, it is increasingly clear that Antarctic ecosystems and biodiversity are facing the same threats as in the rest of the world, particularly from climate change, pollution, biological invasions and an increase in direct human impacts and activities [51][52][53]. In this context, the poor existing knowledge of species diversity and their dispersal ability are considered limiting factors to their effective management and conservation [31,54,55].
In the current study, we investigated, using a combined taxonomic approach, the distribution, phylogenetic relationships and the population genetics of the genus representatives of the Stereotydeus present in Victoria Land, with the support of morphological characteristics fundamental for species identification. In the Antarctic Conservation Biogeographic Regions system (ACBRs [52,56,57]), Victoria Land is divided into Northern and Southern Victoria Land. Nevertheless, the area between Mount Melbourne and the Drygalski Ice Tongue has been singled out for its unusual biogeographic connections and possible role in the promotion of the genetic differentiation of terrestrial taxa in numerous studies targeting Collembola [27,58,59]. As such, this region, named "Central" for convenience, has been separated from the northern ACBR in our analyses. Furthermore, we provide more than 150 new sequences for the mitochondrial barcode region cox1, and the nuclear 28S, of five different Stereotydeus species from Victoria Land.

Sample Collection
Stereotydeus specimens were collected from nine different localities in Victoria Land ( Figure 1; Table 1) during the 2017-2018 and 2018-2019 austral summer expeditions of the Italian National Antarctic Research Program (PNRA: PNRA16_00234), and were immediately preserved in >99.5% ethanol and stored at −80 • C. A total of 159 individuals were used for the molecular analyses. Of these, the whole body of 137 specimens was used for the genetic analyses (Table 1; see Section 2.2). The remaining 22 individuals (see Section 2.3) were used in the morphological investigation, with only 2-4 legs used for the DNA extraction. Table 1. Coordinates and altitudes of sampling localities and ID codes for the different populations sampled; the numbers of individuals (n.) extracted and used for the molecular analyses and the species found at each locality, are given.   Table 1 for locality abbreviations) and in published studies of S. mollis [33,34,36]

Molecular Dataset
Total genomic DNA was extracted from 137 whole individuals from the nine collection sites (Table 1) and the outgroup specimen, the winter grain mite Penthaleus major (Acari: Penthaleidae; Accession number cox1: MZ350753; Accession number 28S: MZ442288; Table S1 in the Supplementary Materials) using the Wizard ® SV genomic DNA Purification System (Promega, Madison, WI, USA) and eluting in 50 μl ddH2O.

Molecular Dataset
Total genomic DNA was extracted from 137 whole individuals from the nine collection sites (Table 1) and the outgroup specimen, the winter grain mite Penthaleus major (Acari: Penthaleidae; Accession number cox1: MZ350753; Accession number 28S: MZ442288; Table S1 in the Supplementary Materials) using the Wizard ® SV genomic DNA Purification System (Promega, Madison, WI, USA) and eluting in 50 µL ddH 2 O.
Region II of mtDNA cytochrome c oxidase subunit I (cox1) was amplified using the mite specific primers COI-2F (5 -TTYGAYCCIDYIGGRGGAGGAGATCC-3 ) and COI-2R (5 -GGRTARTCWGARTAWCGNCGWGGTAT-3 ) [61]. A preliminary amplification of the 28S gene was performed on a restricted pool of five Stereotydeus individuals from each of six localities (CHA, CCI, CJO, CIC, INE and SNU) and including all the species, with the primer pair D1a (5 -CCCSCGTAAYTTAAGCATAT-3 ) and D5b1 (5 -ACACACTCCTTAGCGGA-3 ) [62]. A new specific primer pair (Ste-28S-F (5 -GGACGTGAAACCGCTTGTA-3 ) and Ste-28S-R (5 -TCTGACGATCGATTTGCAC-3 )) was designed in conserved regions (750 bp) and used to amplify all the remaining Stereotydeus specimens and the outgroup. PCRs were performed in 25 µL reaction volume containing: 2.5 µL of genomic DNA, 0.5 mM of each primer, 0.2 mM of each deoxynucleotide triphosphates (dNTPs), 2.5 mM of MgCl 2 , 5 µL of Green GoTaq Flexi buffer and 0.625 U of GoTaq Flexi DNA Polymerase (Promega, Madison, WI, USA). The amplifications were performed in a GeneAmp ® PCR System 2700 (Applied Biosystems, Foster City, CA, USA) thermal cycler. The initial denaturation step was set at 95 • C for 5 min, followed by 35 cycles at 95 • C for 1 min, 45 • C (for cox1) or 50 • C (for the 28S) for 1 min, and 72 • C for 90 s, and a final extension step at 72 • C for 7 min. PCR products were then purified using the kit Wizard ® SV Gel and PCR Clean-up (Promega, Madison, WI, USA) and sequenced on both strands (with the same primers used for PCRs) with a DNA Analyzer ABI 3730, at the core facility of the Bio-Fab Research Lab (Rome, Italy). The sequences were assembled and manually corrected using the MacVector™ software (MacVector, Inc., USA; version 16.0.8- [63]).
In addition to the new samples extracted for this study, all 56 publicly available cox1 sequences for the genus Stereotydeus were downloaded from GenBank (Table S2 in Supplementary Materials) and included in the analyses. These included 50 of S. mollis, 2 of S. shoupi, 1 of S. belli, 1 of S. villosus and 2 of Stereotydeus sp. together with a second outgroup, another eupodid mite Eriorhynchus sp. (Acari: Eriorhynchidae; see Table S2).
The two haplotypes, DQ305366 (S2- [34]) and DQ305388 (B- [33]), were excluded from this analysis because they are homonyms of DQ305362 and DQ305389, respectively. An error in naming them may have occurred when deposited in GenBank, therefore, following the analyses of Demetras et al. [36], the latter two were used in our analyses. Although we included all the remaining deposited haplotypes, some incongruences are noted in three other sequences: (i) for DQ305362 (S2- [34]), coordinates are missing because the precise sampling site in Wright Valley is not clear (W3 and/or W5); (ii) for DQ305382 (S20-V11 from Victoria Valley [34]), coordinates were not included in the original article [34]; (iii) DQ305367 (S6- [34]) was used in Demetras et al. [36] but is missing in the original article of McGaughran et al. [34], therefore, the coordinates are not shown (see Table S2). For the specimens from Demetras et al. [36], only the generic location of southern Dry Valleys (i.e., Garwood, Marshall and Miers Valleys, Shangri La and vicinity, according to Collins et al. [64]) was given, but not the exact coordinates, so they are not shown in this study.

Combined Morphological Analysis
In parallel to this study, morphological analyses have been performed on numerous specimens (between 20-50 for each sampled species, data not published). The morphological comparisons clearly defined the boundaries between all the Stereotydeus species occurring on Victoria Land, as recently published in Brunetti et al. [48], where not only the new species of S. ineffabilis and S. nunatakis are described, but also all the characters so far used to describe and distinguish the Antarctic Stereotydeus species are reviewed (see [48] Tables A1-A7), and the keys to identification are provided. Unfortunately, the lack of specimens of S. mollis, S. shoupi and S. villosus from accessible localities prevented us from improving the original descriptions with the new characters studied in these species and, therefore, were not available for combined morphological analyses.
In addition, after a quick molecular screening, we decided to deeply investigate the morphological aspects of S. delicatus and S. ineffabilis in relation to their genetic differentiation. We focused our attention on Campo Icaro, Inexpressible Island, Prior Island and Starr Nunatak, due to the presence at those localities of the new species described (S. ineffabilis and S. nunatakis). We also questioned the exact correspondence of previously published sequences to specific Stereotydeus taxa. In this respect, the combination of morphological and molecular analyses performed on the same specimens, collected in the central and southern sites of our sampling area, and the recent taxonomic description of new species of the genus (i.e., S. ineffabilis), challenged the attribution of some haplotypes to S. mollis.
Due to the small size of the specimens and, consequently, of the characteristics useful for an accurate taxonomic determination, 22 adult individuals (13 S. ineffabilis from four localities and 9 S. delicatus from Campo Icaro; Table 2) were selected for the joint morphological/molecular investigation and also used in all the molecular analyses. Only adult specimens were considered in the morphological comparison because, at the nymphal stages, most of the characteristics useful for the positive identification of Stereotydeus species are not yet developed (e.g., small size, sex structures not developed, division of the femora absent or incomplete, reduced number of aggenital and genital setae, and reduced number of rhagidial organs; see [48]).
From each specimen, 2-4 legs were removed (to perform the genetic analyses) while the remainder of the body was incubated on a slide with few drops of lactic acid (20%) at 37-45 • C for 30 min to clear the samples, which were then observed under a Leica DM RBE microscope for morphological analysis. The morphological characters considered for identification of S. delicatus and S. ineffabilis were: (a) the length (µm); (b) the division of the femora (presence/absence); (c) the position of the anal pore; (d) the number of aggenital and (e) the number of genital setae; (f) the length of the 4th segment of the pedipalp compared to the 3rd segment; (g) the shape of the epirostrum; and (h) the disposition of the rhagidial organs on tarsi I and II.

Phylogenetic Analyses
For both mitochondrial and nuclear markers, 159 sequences were obtained and the datasets were separately aligned using the online tool Clustal Omega [65]; and manually corrected and trimmed (147 bp and 54 bp were trimmed for the cox1 and 28S respectively) using the MacVector™ software (MacVector, Inc., Cary, NC, USA; version 16.0.8- [63]). The resulting cox1 dataset was then aligned with the two outgroups, while the 28S dataset was aligned only with the P. major outgroup, due to the lack of the ribosomal DNA sequence in Genbank for Eriorhynchus sp. The outgroups were selected from mite families related to ingroups in order to reduce the phylogenetic distance with the Antarctic Stereotydeus spp. In detail, the species P. major (from a closely related family to that of ingroups) was selected as outgroup both for combined and single locus analyses. In addition, the cox1 sequence of Eryorinchus sp. was also included as outgroup because it has been widely used in previous studies on Antarctic Stereotydeus spp. The cox1 dataset was concatenated to the 28S alignment to generate a multilocus dataset through FaBox [66], with the online tool Fasta alignment joiner (available at https:// users-birc.au.dk/palle/php/fabox/alignment_joiner.php; accessed on 18 September 2020).
The multilocus alignment was then run on the Gblocks server 0.91b ( [67]; available at http://molevol.cmima.csic.es/castresana/Gblocks_server.html; accessed on 18 September 2020) under strict settings and the hypervariable regions of the 28S alignment were discarded. After the run, 1034 positions, out of the 1171 of the initial dataset (88%), were kept. Ultimately, the four single-and the multilocus alignments used for the phylogenetic and population genetics analyses were: (i) cox1 with outgroup; (ii) cox1 all haplotypes; (iii) combined cox1-28S; and (iv) combined cox1-28S with associated morphological information ( Table 3). Table 3. List of the datasets (single and multilocus), number of new sequences obtained and used in each dataset (n.), markers, reference sequences (Ref.) and outgroups used for the analyses and models of nucleotide evolution that best fitted, divided according to the partition applied and to the respective tree search optimization criteria.

n.
Single To identify the haplotypes and their frequencies within populations, all the alignments were run with the online software DNA-Collapser [66]. The sequences of the resulting haplotypes were used to calculate the genetic distances between the haplotypes using the software R 3.6.1 [68] with the "ape 5.3" package [69]. The best evolutionary models were selected before the tree search (Table 3), partitioning the datasets with the software PartitionFinder 2.1.1 [70] based on Akaike's information criterion (AIC) and a greedy strategy: 1st, 2nd and 3rd codon positions for the cox1 protein-encoding gene and one single partition were considered for the 28S (Table 3). Bayesian analysis was performed with MrBayes 3.2.7 software [71], applying four chains (three hot and one cold) for 10 6 generations, with a sampling frequency of one tree every 1000 iterations and with 25% of the tree topologies discarded (burn in step) from the final result. For better visualization, the resulting phylogenetic trees were then zoomed and expanded and the node labels (posterior probabilities) were added with FigTree 1.4.4 software [72]. The new Stereotydeus mitochondrial and nuclear sequences were deposited in GenBank (cox1 Accession numbers: MZ350724-MZ350752; 28S Accession numbers: MZ442270-MZ442287; Table S1 in Supplementary Materials).

Population Structure Analyses
The population genetics study was performed using the cox1 dataset without the outgroups applied for the phylogenetic analysis. S. mollis sequences were not included in the analysis. This was due to: (i) the incongruences found in the Genbank sequences (see Section 2.2. and Table S2 in Supplementary Materials), (ii) the fact that no morphological investigations were performed on these individuals, and (iii) because new S. mollis specimens were not available for a morphological analysis during this study. Haplotype frequencies were obtained using the online tool DNA collapser [66]. The network clade analysis was performed on TCS 1.21 [73] using a connection limit of 98% and visualized with the online tool tcsBU ( [74]; available at https://cibio.up.pt/software/tcsBU/; accessed on 28 November 2020) to estimate the haplotype networks for each species. To investigate the genetic characteristics of populations and to test for the presence of population structure, Arlequin version 3.11 [75] was used for each species separately. The haplotype (h) and nucleotide (π) diversity indices [76], as well as the mean number of pairwise differences (θ) and segregating sites (θS), were computed at the population level. Analysis of molecular variance (AMOVA; [77]) was used to measure the extent to which genetic variance could be assigned to the hierarchical structure of population organization (testing them with the structure according to the populations: "Cape Hallett", "Crater Cirque", "Cape Jones" and "Kay Island" for S. belli; "Campo Icaro", "Vegetation Island" and "Inexpressible Island" for S. delicatus; "Campo Icaro", "Inexpressible Island", "Prior Island" and "Starr Nunatak" for S. ineffabilis and "Prior Island" and "Starr Nunatak" for S. nunatakis), with the statistical significance of variance components tested with 16,000 permutations. Pairwise differences between haplotypes (Φ ST values) were calculated using simple distances and these were used to look for significant relationships between population genetic distance (Φ ST ).

Results
Using the cox1 haplotypes of the 50 S. mollis specimens already available on GenBank as templates, 495 bp of a uniform and unambiguous alignment from 159 sequenced individuals were used for all genetic analyses. For 28S, 1034 positions of the 159 sequenced individuals, together with the outgroup P. major, were used for phylogenetic analyses.
For each Stereotydeus species, between 2-14 cox1 and 1-9 28S haplotypes were found (Table 4) while, for each locality, between 1-11 cox1 and 1-4 28S haplotypes were found. Most 28S haplotypes were unique at the species level, with the only exception being RX1 from CIC, shared by both S. delicatus and S. ineffabilis. In addition, for the combined set of cox1 and 28S, from 3-16 and from 2-9 haplotypes were found for the Stereotydeus species and the localities, respectively. The number of Stereotydeus species identified per site ranged from 1-2 (Table 5).

Haplotype Network Analyses
The total number of nucleotide substitutions (absolute changes) ranged from 1 (S. nunatakis in SNU) to 117(S. ineffabilis in CIC) within all the populations of the five different taxa. Four subnetworks were found for S. belli, with two single haplotypes not connected with any other haplotype: MB1 and MB7, from CHA and KAY, respectively. Within the species and within the clusters, the number of nucleotide substitutions ranged from a minimum of nine, recorded in CJO, to a maximum of 21, in KAY (mean 7.50 ± 25.48) (Figure 2). For S. punctatus, one single network was observed where all haplotypes were connected with each other within an upper range of seven nucleotide changes (Figure 2). Three clusters were found for S. delicatus, with two single haplotypes not connected with any other haplotype: MD5 (VEG and CIC) and MD6 (VEG and INE). The number of nucleotide substitutions ranged from 42, in VEG, to 46, in CIC, in this species and within the populations (mean 29.33 ± 9.95) (Figure 2). Six networks were found for S. ineffabilis, with three single haplotypes not connected with any other haplotype: MI11 (SNU and PRI), MI12 (CIC and INE) and MI13 (CIC). These haplotypes are also placed together in a different position in the phylogenetic trees, with the respect to the other conspecific haplotypes (see Section 3.2). The differences within both species and populations ranged from 79, in SNU, to 117, in CIC (mean 89 ± 18.67) (Figure 2). For S. nunatakis, only two haplotypes were observed, differing by a single substitution (mean 0.50 ± 0.71) (Figure 2).

Haplotype Network Analyses
The total number of nucleotide substitutions (absolute changes) ranged from 1 (S. nunatakis in SNU) to 117 (S. ineffabilis in CIC) within all the populations of the five different taxa. Four subnetworks were found for S. belli, with two single haplotypes not connected with any other haplotype: MB1 and MB7, from CHA and KAY, respectively. Within the species and within the clusters, the number of nucleotide substitutions ranged from a minimum of nine, recorded in CJO, to a maximum of 21, in KAY (mean 7.50 ± 25.48) (Figure 2). For S. punctatus, one single network was observed where all haplotypes were connected with each other within an upper range of seven nucleotide changes (Figure 2). Three clusters were found for S.

Phylogenetic Analyses
• cox1 with outgroups For this single locus analysis, a total of 165 Stereotydeus sequences and two outgroups (Eriorhynchus sp. and P. major) were used. Before the addition of the outgroups, two unrooted analyses were also performed (Table S3 and Figure S1 in Supplementary Materials). One monophyletic group was formed by the haplotypes of S. belli (MB1-10) and includes 29 specimens from Northern Victoria Land (CHA, CJO and two from CCI), all those of the KAY population (Central Victoria Land) and also the single sequence of S. belli (specimen from Cape Hallett) (Figures 2 and 3). Another monophyletic group included all 12 S. punctatus sequences (MP1-4) belonging to the CCI population. One paraphyletic group included the S. delicatus specimens (MD1-6) and the S. ineffabilis specimens (MI1-10, 14), with individuals from Southern Victoria Land (CIC, VEG, INE for S. delicatus and CIC, INE, PRI and SNU for S. ineffabilis) (Figures 2 and 3). Three haplotypes of S. ineffabilis were not included in the latter group, but they were clustered together, although with low statistical support (Figures 2 and 3). These three sequences, together with the branch that carries the two S. nunatakis haplotypes, did not cluster with the remaining ingroup, due to the insertion of three sequences of other species, although with medium statistical support (pp = 0.74 and 0.87) (Figure 3).   Table 5). Accession numbers for reference sequences and Eriorynchus sp. [33,34,36] and P. major outgroups are also shown.  Table 5). Accession numbers for reference sequences and Eriorynchus sp. [33,34,36] and P. major outgroups are also shown.
• cox1 all haplotypes Fifty previously published S. mollis reference sequences were included for this analysis (Table S2). Despite the S. ineffabilis haplotypes being spread throughout the entire phylogenetic tree and not all nodes being statistically well supported, the two monophyletic groups of S. belli and S. punctatus were still distinct from the remaining species with good support at nodes (pp = 0.95 and 1, respectively). S. delicatus was, once again, recovered as a paraphyletic group: one cluster of four haplotypes (MD1-4) and two separated branches (MD5 and MD6), although with low support at nodes. The cluster of two haplotypes for S. nunatakis, together with MI11 and two S. mollis haplotypes (Sm49 and Sm50), was collapsed with the other three sequences at the base of the main cluster. Six (MI1, 4,6,9,11,12) out of the fourteen S. ineffabilis haplotypes were identical to previously published sequences (L, K, J, Sm44, R, O, respectively) originally assigned to S. mollis before the identification and description of S. ineffabilis as a new species [48] (Table 7). After the morphological identification of the specimens related to these haplotypes (Table 7), these sequences are now considered as S. ineffabilis. In addition, when sequences initially assigned to S. mollis clustered together with the S. ineffabilis haplotypes and were statistically well supported (pp > 0.85), we tentatively considered them as belonging to S. ineffabilis (e.g., Sm50, P; Figure 4). Table 7. S. ineffabilis specimens used for the haplotypic and morphological analyses (Slide) with cox1 haplotypes (cox1) identical to previously published sequences of S. mollis (haplo.). Sampling localities with their ID codes where the specimens were found and accession numbers (Acc. num.) of the S. mollis haplotypes are also provided. • Combined cox1-28S Following the phylogenetic analyses of the combined dataset of cox1 and 28S sequences (1034 bp), four phylogroups were detected: three monophyletic groups (S. belli, S. punctatus and S. nunatakis, although with variable support, pp = 0.55-1) and one paraphyletic clade (statistically low support, pp = 0.66, including S. ineffabilis and S. delicatus as mutually para/polyphyletic groups). The combination of the two datasets generated 14 unique haplotypes for S. belli from northern Victoria Land (CHA, CJO and CCI) and central Victoria Land (KAY), 4 unique haplotypes for S. punctatus from northern Victoria Land (CCI), 3 unique haplotypes for S. nunatakis from southern Victoria Land (PRI and SNU), 11 unique haplotypes for S. delicatus from southern Victoria Land (CIC, VEG and INE) and 16 unique haplotypes for S. ineffabilis from southern Victoria Land (CIC, INE, PRI and SNU) ( Figure 5).

•
Combined cox1-28S with morphology In order to further clarify the paraphyletic relationships, a table of some morphological characteristics was linked to the combined cox1-28S tree, restricted to S. ineffabilis and S. delicatus sequences. All the nodes clustering the deepest branches together were statistically well supported, with the exception of that separating the Si 4 haplotype from the main cluster (pp = 0.64) (Figure 6).   Table 5.  Table 5. In order to further clarify the paraphyletic relationships, a table of some morphological characteristics was linked to the combined cox1-28S tree, restricted to S. ineffabilis and S. delicatus sequences. All the nodes clustering the deepest branches together were statistically well supported, with the exception of that separating the Si 4 haplotype from the main cluster (pp = 0.64) (Figure 6). Figure 6. Phylogenetic tree of the combined mitochondrial and nuclear sequences of S. delicatus (orange labels) and S. ineffabilis (green labels) and table with codes of the morphological characters (see Table 8). Posterior probabilities are shown at the nodes of the phylogenetic tree. For the specimen ID (black, bold), see Table 2. Figure 6. Phylogenetic tree of the combined mitochondrial and nuclear sequences of S. delicatus (orange labels) and S. ineffabilis (green labels) and table with codes of the morphological characters (see Table 8). Posterior probabilities are shown at the nodes of the phylogenetic tree. For the specimen ID (black, bold), see Table 2.
The S. delicatus specimens had a mean length of 451.83 µm (± 27.39 µm), ranging from CI1 (389.99 µm) to CI7 (481.55 µm). The femora were divided in CI7, CI9, CI11 (Sd 5) and CI14 (Sd 11), undivided in CI1 (Sd 4) and CI5 (Sd 3) and with partial division in CI10, CI12 and CI13 (Sd 5). The position of the anal pore was always apical. In all the specimens observed with haplotype Sd 5 and CI14 (Sd 11), there were four pairs of aggenital setae, while CI1 and CI5 had five pairs. Six pairs of genital setae were present in all the specimens, with the exception of CI7, which had seven pairs. The length of the 4th segment of the pedipalp was longer than the 3rd in all the specimens with haplotypes Sd 5 and CI14 (Sd 11), while in CI1 and CI5 the two segments were comparable in length. The trilobe shaped epirostrum was weakly developed in CI1 and CI5, while, in the remaining specimens, it was evident and strongly developed. The three rhagidial organs on tarsi I and II showed an axis of symmetry in all specimens, with the exception, again, of CI1 and CI5. Table 8. Morphological characters considered for the identification of S. delicatus and S. ineffabilis. Every row of a character is represented by a number (1)(2)(3)(4)(5) used to link them to the combined morphology phylogenetic tree (Figure 6). A. Length (µm); B. Femora; C. Position of the anal pore; D. Aggenital setae; E. Genital setae; F. Length of the 4th segment of pedipalp compared to the 3rd; G. Epirostrum; H. Disposition of the rhagidial organs on tarsi I and II. The S. ineffabilis specimens had a mean length of 427.62 µm (± 18.61 µm), ranging from P1 (386.62 µm) to CI3 (460.44 µm). The femora were undivided except for individuals I3 (Si 3) and I4 (Si 6), where the division was only partial. The anal pore was always ventral in the terminal portion (see [48] Figures 1b and 5b). The number of aggenital setae was variable: two specimens (I2, I5) had four pairs, five (I3, CI3, P1, P3 and P5) had five pairs, two (S2, P2) had six pairs, while four had an intermediate number (I1 and I4 had 9 setae; S1 and S5 had 11 setae). Six pairs of genital setae were present in all specimens with the exception of S2 (Si 4), which presented an asymmetry with 13 setae. The length of the two terminal segments of the pedipalps was comparable in all the specimens examined except in S2 (Si 4), where the 4th segment was longer than the 3rd. The trilobed shape of the epirostrum was weakly developed in all specimens. The three rhagidial organs on tarsi I and II showed an axis of symmetry only in P2 (Si 16). P5 legs I and II were missing, so it was not possible to determine the positions of the rhagidial organs (for the morphological features see [48], Figures 1-5).

Code
Although character C seems the only listed character that sharply sorts out the two species, when few exceptions of specimens are not considered, the list of characters increases (see [48] for the keys and the synoptic Tables A1-7 of the Antarctic Stereotydeus species).

Population Structure Analyses
Haplotype diversity (h) for cox1 in S. belli ranged from 0 to 0.743 (mean 0.336). Within populations, CJO had the highest haplotype diversity and CHA and CCI the lowest. Nucleotide diversity (π) was low for all four populations, with the highest value being in the KAY population (0.010) ( Table 9). The values of mean nucleotide pairwise differences θ(π) and mean number of segregating sites θ(S) ranged from 0 to 5.200 (mean 1.194 ± 2.486) and from 0 to 7.423 (mean 2.521 ± 3.501), respectively. The KAY population had the highest values of both θ(π) and θ(S), while CHA and CCI had the lowest. For S. delicatus, h ranged from 0 to 0.553 (mean 0.384), with the highest values in CIC (0.553) and the lowest in INE. Measures of π showed a similar pattern to haplotype diversity, with the highest values found in CIC (0.030). The highest values of θ(π) and θ(S) were recorded in CIC (14.966) and in VEG, respectively, while the INE population had the lowest values for both parameters. In S. ineffabilis populations, h ranged from 0.380 to 0.801 (mean 0.647). Within the populations, CIC, again, had the highest haplotype diversity, while PRI had the lowest. Reflecting the h measures, π had the highest value in CIC (0.071), with the lowest recorded in INE (0.026). The values of θ(π) and θ(S) ranged from 13.121 to 35.375 (mean 21.460 ± 9.829) and from 22.317 to 29.579 (mean 26.202 ± 3.497), respectively. The CIC population had the highest values of both θ(π) and θ(S), while INE and PRI had the lowest. These parameters were also calculated for the two S. nunatakis populations. However, because, in PRI, only one haplotype was detected, all parameters for this population were 0, while in SNU the values were 0.250 for h and θ(π) and 0.001 and 0.386 for π and θ(S), respectively. Table 9. Population genetic parameters for cox1 in S. belli, S. delicatus, S. ineffabilis and S. nunatakis sampled across Victoria Land (Area). n, number of individuals; N H , number of haplotypes within the populations and their frequencies; h, haplotype diversity; π, nucleotide diversity; θ(π), mean number of pairwise differences; θ(S), mean number of segregating sites; haplotypes shared between populations are indicated in italics (see Table 5 for details).
When group structure was assigned to populations for each species, the AMOVA analysis revealed more variation among groups and within populations (for S. ineffabilis) than among populations within groups (Table 10). In particular, for S. belli and S. delicatus, the Φ CT values were similar (10.48068 and 9.51162, respectively). while for S. ineffabilis the value was only 2.94891. In contrast, Φ ST values were higher in S. ineffabilis (10.89525) than in S. belli and S. delicatus (1.25345 and 6.66210, respectively).

Discussion
This study provides over 150 new sequences representing all species of the mite genus Stereotydeus from Victoria Land. Combined with the morphological assessments that we provided, this information sheds light on an understudied taxon and provides a good starting point for further taxonomic studies of the species of the genus (Figure 7).  [48] (blue) and previous molecular data from [33,34,36] (grey). (coastlines from ADD Simple Basemap, NPI/Quantarctica 3 [60]).

North Victoria Land Taxa
Based on the analyses performed in this study, we found a latitudinal pattern in the distribution of Stereotydeus species in the Victoria Land coastal region. The presence of S. belli characterizes all populations from Cape Hallett (CHA) to Kay Island (KAY), while S. punctatus has, so far, been detected only in Crater Cirque  [48] (blue) and previous molecular data from [33,34,36] (grey). (coastlines from ADD Simple Basemap, NPI/Quantarctica 3 [60]).

North Victoria Land Taxa
Based on the analyses performed in this study, we found a latitudinal pattern in the distribution of Stereotydeus species in the Victoria Land coastal region. The presence of S. belli characterizes all populations from Cape Hallett (CHA) to Kay Island (KAY), while S. punctatus has, so far, been detected only in Crater Cirque (CCI). This is the first genetic study to be conducted on the latter species and, although only comprising a limited number of samples (12 individuals analyzed resulting in 4 cox1 and 1 28S unique haplotypes), the presence of genetic variability is already evident. In addition, this is an easy species to identify morphologically due to the peculiar dorsal position of the anal opening that does not occur in any other Stereotydeus species. Early records of this taxon were reported by Strandtmann [44] and Gressitt and Shoup [40], also from Cape Adare and Cape Hallett.
For S. belli it is possible that historical events, such as habitat fragmentation due to glacial events, divergence in isolation and subsequent range expansion [32][33][34][78][79][80], are responsible for the patchy distribution of these populations and their genetic isolation. All the S. belli populations were clearly distinct, with KAY and CCI clustered together and separated from both CHA and CJO. This fragmented and apparently disjointed distribution is consistent with reports from other terrestrial invertebrate species in north Victoria Land (e.g., [31,64]). The same studies have reported that invertebrate populations in the region of the Tucker Glacier are genetically more closely related to populations in central-south Victoria Land, compared to others in relatively closer geographical proximity. Recent studies of springtail species endemic to Victoria Land, including Cryptopygus cisantarcticus [64] and Friesea gretae and F. propria [31] (F. grisea in Collins et al. [64]), highlighted the important role of the Tucker Glacier as an insurmountable barrier leading to high levels of genetic divergence between populations from either side of the glacier, plausibly representing distinct species. Combining the inferences made in the current study with previous springtail studies specific to northern Victoria Land [31,64], a comparably important role may be played by Crater Cirque, where S. belli and S. punctatus occur in sympatry.

Central-South Victoria Land Taxa
This study presents the first record in the central part of Victoria Land of S. delicatus, which was originally discovered and described by Strandtmann [44] (although with only one individual from each location) from Cape Adare and Edisto Inlet; thus, our new data considerably expand the known distribution of this species southwards. Our sampling area is located in a part of Terra Nova Bay that is affected by powerful winds, while the Hells Gate moraine creates an abrupt interruption between Inexpressible Island (INE) and the peninsula of the Northern Foothills where Campo Icaro (CIC) is located. The distribution of the haplotypes found in this area suggests a possible role played by Vegetation Island (VEG) in acting as a bridge to connect CIC and INE. It is plausible that gene connectivity bypassed the inhospitable Hells Gate channel by using VEG as a midpoint between CIC and INE, although further intermediate steps may have been available at different points in the past.
Considering the phylogenetic relationship of S. delicatus with the other Stereotydeus taxa reviewed in this study, the link with the newly described species S. ineffabilis is immediately evident. Even though the species are morphologically distinct ( Figure 6, Table 8; see also [44,48] for species descriptions), individuals of the two species from the CIC locality share the same unique haplotype (RX1) for the nuclear marker 28S, although the combined analyses of the latter with the mitochondrial marker cox1 and the morphological characteristics provided a good resolution of the boundary between the two species. A possible explanation for these results is that these taxa have "recently" undergone a speciation process and, because of the different resolutions of the two markers, it is possible that the large ribosomal subunit may not yet have accumulated sufficient mutations to enable distinguishing between the two sympatric species. A "slow" nucleotide substitution rate in 28S is not unusual and has recently also been recorded in Friesea lineages from Victoria Land [31]. Specifically in relation to understanding the geographic distribution and genetic diversity of S. delicatus, it is now crucial to expand the sampling and study effort to include the north Victoria Land locations of Cape Adare and Edisto Inlet, where the species was first recorded and described by Strandtmann [44].
While S. delicatus shows a well defined pattern of distribution, that of S. ineffabilis appears to be more complex. As for S. delicatus, the presence of the Hells Gate moraine isolates the populations north of the Drygalski Ice Tongue but, observing the haplotype networks, it is possible that, in the past, the two areas were linked, with the populations starting to differentiate only when the connection was broken. It is notable that the two S. ineffabilis populations south of the Drygalski Ice Tongue show a genetic connection to the population north of the glacier, although also showing some differentiation. As the Drygalski Ice Tongue is considered the geographical barrier that sharply delimits the faunas of north and south Victoria Land, our data provide a first indication of geneflow between north and south Victoria Land, and the first record of a terrestrial microarthropod species shared between the two regions.
In comparing the genetic diversity present in north and south Victoria Land, this study included also Stereotydeus spp. cox1 haplotypes reported in previous studies [33,34,36] in the phylogenetic analyses performed. A striking outcome of these analyses is the strong link that emerged between S. mollis and S. ineffabilis sequences. The great genetic variability of the cox1 marker alone proved ineffective in drawing a clear distinction between the two taxa. In order to stabilize the phylogenetic signal of the mitochondrial marker, it will be crucial to include one or more nuclear markers in future studies, as well as combining genetic and morphological approaches. In the absence of nuclear DNA sequence data from the Stereotydeus specimens, several morphological characteristics (e.g., the smaller size of the adults, the asymmetry in the tarsal rhagidial organs, the position of the solenidia on the tibiae and the genua, the number of the aggenital setae; see [48] for more details) were useful in identifying boundaries between S. mollis and S. ineffabilis. A high level of genetic diversity of recent origin (see branching pattern on Figures 4 and 5) is generally interpreted as an indication of recent demographic expansion. However, the present distributions of the S. ineffabilis, S. delicatus and S. mollis phylogroups may best be interpreted as being the result of alternative and temporally disjunct colonization events and speciation processes that occurred several times and started from different glacial refugia over a time interval of more than 10 Myr.
Together with S. ineffabilis, S. nunatakis was also present in the Prior Island (PRI) and Starr Nunatak (SNU) sampling locations [48]. Although the number of samples for genetic and morphological analyses was low, some variability and divergence was apparent. Based on the combined mitochondrial and nuclear phylogenetic analysis and the computation of genetic distances, S. nunatakis appears to be more closely related to S. punctatus, from north Victoria Land, than to the other species from south Victoria Land, S. ineffabilis, S. mollis and S. shoupi.

Speciation in Action
The patterns of diversity observed today in many Antarctic species can be traced back to historic events, such as habitat fragmentation, divergence in isolation and subsequent range expansion, that influenced the distribution of species particularly at local scales [32][33][34]81]. The resulting patterns of genetic variation can be used to infer ecological factors (e.g., effective population size, dispersal capacity), as well as those affecting speciation processes. Allopatric speciation in populations that are geographically separated appears to be characteristic for populations of many terrestrial invertebrate species native to Victoria Land, and is considered the result of the different fragmentation and isolation events of ancestral and widespread lineages [19,20,27]. As for these other invertebrates, we suggest that, due to their limited dispersal abilities and the presence of physiological barriers such as low tolerance to desiccation and abiotic barriers, our resulting populations also started to differentiate independently. However, especially for the southern Victoria Land species of Stereotydeus, the scenario appears to be more complex, due to the presence of four geographically and genetically closely related species. In recent decades, the suggestion that speciation might also occur in populations that are not geographically isolated (i.e., sympatric speciation [82][83][84]) has become increasingly accepted. It is possible, for example, that, when limited resources are available to members of sympatric populations, interactions through both direct (i.e., interference) and indirect (i.e., exploitation) competition could lead groups of individuals, especially those belonging to populations of large size, to adopt different behaviors, select different habitats, establish temporal shifts of activity patterns or avoid mating or generating hybrids with low fitness. Thus, ecologically based barriers to gene flow evolve between populations resulting in an "ecological selection" [83,85,86]. This selection can occur under different geographic conditions [83], so it cannot be excluded that this process may also have contributed to the current patterns of variability and distribution of Stereotydeus species in Antarctica.
Although the biogeographical patterns of springtails and mites in coastal Victoria Land share some similarities [28,33,35], their intra-and interspecific genetic distances are not entirely comparable. Interspecific genetic distances calculated between species of Acari are generally greater than those observed in comparisons between Collembola (e.g., [33,34]), and it is not possible to exclude this being influenced by the different survival strategies and/or life histories of free living mites [12]. It is possible that all aspects of the life history strategy of Antarctic terrestrial invertebrates (e.g., including generation time, life cycles, physiology and metabolism), in combination with environmental conditions, could be major factors influencing evolutionary rates (nucleotide substitutions). However, it is also not clear, in general, how rates of evolution differ across species or, if they do, what factors drive these differences. The factors responsible for the high levels of divergence shown by mites have previously been suggested to include the smaller size of the animals, their shorter generation time and higher activity levels [37] and their greater recolonization/dispersal abilities [33] in comparison with springtail taxa. However, these hypotheses have not been explicitly tested. Prostigmatid mites lack an impermeable cuticle, and behavioral strategies, such as microhabitat selection, along with physiological acclimatization [12] are likely to play a fundamental role in the isolation of populations and their survival. As suggested by Demetras et al. [36], some behavioral differences may have a role in increasing genetic divergence, as has also been noted for some Antarctic springtail species [87]. Thus, through combining morphological, genetic and ecological studies of terrestrial fauna, we can better understand the evolutionary origins, dispersal history and current distribution of Antarctic invertebrates.
Due to the close phylogenetic relationships between the central and southern species (S. ineffabilis, S. delicatus and S. mollis), in the future it will be fundamental to carry out and implement new combined taxonomical studies and enlarge the number of specimens in the analyses. The inclusion of a more recent revision of the original materials used for the first description of S. delicatus and S. mollis will help to identify additional morphological characters, if any, necessary to distinguish these species with respect to S. ineffabilis. In fact, when the amount of divergence at inter-and intraspecific level is overlapping, morphology is important to identify species boundaries. In addition, the genetic differentiation of species of "recent" origin may be less variable with respect to more ancient ones. Thus, the combination of new morphological analyses and a deeper genetic screening through the incorporation of more nuclear markers and/or genome comparisons will be the starting point to better define some of the phylogenetic relationships of all the Victoria Land Stereotydeus species.
In summary, the contemporary distributions of species of Stereotydeus occurring in Victoria Land follow defined latitudinal patterns, including two major features. These are characterized by, first, a more genetically defined cluster in the north Victoria Land populations of S. belli and S. punctatus and, second, a more complex, in terms of species composition, cluster including populations in south Victoria Land.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/ 10.3390/d13100506/s1, Figure S1: Unrooted phylogenetic trees of Stereotydeus specimens with pos-terior probabilities shown at nodes, Table S1: Accession numbers of the cox1 and 28S sequences of Stereotydeus species and Penthaleus major deposited on GenBank and included in the analyses, Table S2: Accession numbers of the cox1 sequences of Stereotydeus species and one Eriorhynchidae mite downloaded from GenBank and included in the analyses, Table S3: List of the datasets, number of new sequences obtained and used in each dataset, markers, reference sequences and outgroups used for the analyses and models of nucleotide evolution that best fitted, divided according to the partition applied and to the respective tree search optimization criteria.