Small is beautiful: the first phylogenetic analysis of Bryodelphax Thulin, 1928 (Heterotardigrada, Echiniscidae)

The phyletic relationships both between and within many of tardigrade genera have been barely studied and they remain obscure. Amongst them is the cosmopolitan Bryodelphax , one of the smallest in terms of body size echiniscid genera. The analysis of new -ly-found populations and species from the Mediterranean region and from South-East Asia gave us an opportunity to present the first phylogeny of this genus, which showed that phenotypic traits used in classical Bryodelphax taxonomy do not correlate with their phyletic relationships. In contrast, geographic distribution of the analysed species suggests their limited dispersal abilities and seems to be a reliable predictor of phylogenetic affinities within the genus. Moreover, we describe three new species of the genus. Bryodelphax australasiaticus sp. nov. , by having the ventral plate configuration VII:4-4-2-4-2-2-1, is a new member of the weglarskae group with a wide geographic range extending from the Malay Peninsula through the Malay Archipelago to Australia. Bryodelphax decoratus sp. nov. from Central Sulawesi (Celebes) also belongs to the weglarskae group (poorly visible ventral plates VII:4-2-2-4-2-2-1) and is closely related to the recently described Bryodelphax arenosus Gąsiorek, 2018, but is differentiated from the latter by well-developed epicuticular granules on the dorsum. Finally, a new dioecious species, Bryodelphax nigripunctatus sp. nov. , is described from Mallorca and, by the reduced ventral armature (II/III:2-2-(1)), it resembles Bryodelphax maculatus Gąsiorek et al., 2017. The latter species, known so far only from northern Africa, is recorded from Europe for the first time. A taxonomic key to the genus members is also presented.

The aim of this study was to elucidate the phylogeny of Bryodelphax in relation to morphological traits used in its taxonomy, with application of the integrative approach, i.e. DNA barcoding and both qualitative and quantitative morphology, based on three new species that are described and illustrated herein. Our analyses reveal no congruence between the topology of the phylogenetic tree and the traditional taxonomic divisions of the genus (based on the presence of ventral armature), the reproductive mode or the development of dark, contrasting epicuticular granules on the dorsal plates. On the other hand, we show that phylogeny is tightly correlated with geography. In addition, an amended and updated key to the genus Bryodelphax is provided.

Materials and methods
Sample collection and processing, comparative material Specimens of the genus Bryodelphax were extracted from various moss and lichen samples collected in numerous European and Asian locales (details in Table 1). The animals were divided into three groups used in different analyses: (I) qualitative and quantitative morphology investigated in phase contrast microscopy (PCM) and Nomarski differential interference contrast microscopy (NCM), collectively termed as light contrast microscopy (LCM); (II) qualitative morphology in scanning electron microscopy (SEM); and (III) DNA sequencing (details in Table 2). For morphological comparisons, the type series of B. aaseae Kristensen et al., 2010, B. amphoterus (Durante Pasa & Maucci, 1975, B. asiaticus Kaczmarek & Michalczyk, 2004, B. brevidentatus Kaczmarek et al., formed or twisted and their orientations were suitable. Body length was measured from the anterior to the posterior end of the body, excluding the hind legs. The sp ratio is the ratio of the length of a given structure to the length of the scapular plate expressed as a percentage (Dastych 1999). The bs ratio is the proportion between the maximal body width and the body length in dorsoventrally orientated specimens (Gąsiorek et al. 2018a). Morphometric data were handled using the Echiniscoidea ver. 1.2 template available from the Tardigrada Register, www.tardigrada.net (Michalczyk and Kaczmarek 2013). Ventral plate configuration is given according to Kaczmarek et al. (2012). Genotyping and phylogenetics DNA was extracted from individual animals (all examined under a 400× magnification PCM prior to DNA extraction) following a Chelex 100 resin (Bio-Rad) extraction method (Casquet et al. 2012;Stec et al. 2015). Three DNA fragments were sequenced: the small ribosome subunit 18S rRNA (primers 18S_Tar_Ff1 and 18S_Tar_Rr2 from Stec et al. 2017and Gąsiorek et al. 2017b, PCR programme from Zeller 2010, the large ribosome subunit 28S rRNA (primers 28S_Eutar_F and 28SR0990 from Gąsiorek et al. 2018b andMironov et al. 2012 and the PCR programme from Mironov et al. 2012) and the internal transcribed spacer ITS-1 (primers ITS1_Echi_F and ITS1_Echi_R from Gąsiorek et al. 2019a, PCR programme from Wełnicz et al. 2011). Some of the less conservative markers, such as ITS-2 and COI, are often difficult to amplify for Bryodelphax. All fragments were amplified and sequenced according to the protocols described in Stec et al. (2015). Sequences of 18S rRNA and 28S rRNA were aligned using the default settings of MAFFT version 7 (Katoh et al. 2002;Katoh and Toh 2008), with Echiniscus lineatus Echiniscus testudo (Doyère, 1840) used as the outgroup. The obtained alignments were edited and checked manually in BioEdit v7.2.6.1 (Hall 1999) and then trimmed to 967 bp (18S rRNA) and 754 bp (28S rRNA), respectively. The aligned sequences were concatenated using SequenceMatrix (Vaidya et al. 2011). PartitionFinder version 2.1.1 (Lanfear et al. 2016) with applied Bayesian Information Criterion (BIC) and greedy algorithm (Lanfear et al. 2012) were used to test for the best scheme of partitioning and substitution models for posterior phylogenetic analysis. The analysis was performed solely for MrBayes purposes. The preferred evolution model was GTR+G for both markers (Nei and Kumar 2000), which was finally chosen for further analyses.
ModelFinder (Kalyaanamoorthy et al. 2017) under the Akaike Information Criterion (AIC) and corrected AIC (AICc) was used to find the best substitution models for two predefined partitions (Chernomor et al. 2016). The programme indicated the following models: TVMe+G4 (18S rRNA) and TIM3e+G4 (28S rRNA). Maximum-likelihood (ML) topologies were constructed using IQ-TREE (Nguyen et al. 2015;Trifinopoulos et al. 2016). Strength of support for internal nodes of ML construction was measured using 1000 ultrafast bootstrap replicates (Hoang et al. 2018). Bootstrap (BS) support values ≥ 85% on the final tree were regarded as well supported and those > 70% as moderately supported. Bayesian inference (BI) marginal posterior probabilities were calculated using MrBayes v3.2 (Ronquist and Huelsenbeck 2003). Random starting trees were used and the analysis was run for ten million generations, sampling the Markov chain every 1000 generations. An average standard deviation of split frequencies of < 0.01 was used as a guide to ensure the two independent analyses had converged. The programme Tracer v1.7 (Rambaut et al. 2018) was then used to ensure Markov chains had reached stationarity and to determine the correct 'burn-in' for the analysis which was the first 10% of generations. The Effective Sample Size values were greater than 200 and consensus tree was obtained after summarising the resulting topologies and discarding the burn-in. In the BI consensus tree, clades recovered with posterior probability (PP) between 0.95 and 1 were considered well supported, those with PP between 0.90 and 0.94 were considered moderately supported and those with lower PP were considered unsupported. All final consensus trees were viewed and visualised by FigTree v.1.4.3 (http://tree.bio.ed.ac.uk/software/ figtree). MEGA7.0.26 (Kumar et al. 2016) was used for calculation of uncorrected pairwise distances (Srivathsan and Meier 2012
Adults. Body pink, pearly opalescent; eyes absent or not visible after mounting in Hoyer's medium. Primary and secondary clavae small and conical. Cirri interni and externi with poorly-developed cirrophores. Cirri A of typical length for Bryodelphax, i.e. reaching around 25% of the total body length. All dorsal plates with barely-discern-ible intra-cuticular pillars (better visible under a 1000× magnification), the centro-posterior portion of the caudal (terminal) plate has evident, largest pillars ( Fig. 2A). Dark epicuticular granules absent (Figs 3; 4A, B), but lateral margins of all dorsal plates and internal margins of facets constituting the scapular plate distinctly thicker and, consequently, darker ( Fig. 2A). Pores large and easily detectable ( Venter with seven rows of faint, greyish plates (VII:4-4-2-4-2-2-1), of which two plates of the first, subcephalic row are located in a more ventrolateral position (Figs 2B,3C,D,10). Under SEM, only the central subcephalic and genital plates are visible as true cuticular thickenings, whereas other plates are visible only as darker areas on the cuticular surface (Fig. 3C, D). Leg papillae undetectable under LCM (Fig. 2), but papillae IV visible under SEM (Fig. 3B, C). Both pulvini and pedal plates present, the former developed as thin rectangular stripes in the proximal leg portions (Figs 2A, 3C) and the latter -as large swellings in the central leg portions (Figs 2A, 3C). Pedal plate IV toothless, but with a distinct dark margin ( Fig. 2A). External claws spurless, but internal ones with minute spurs positioned close to the claw bases (  Larvae. Body 80 μm long in a single found two-clawed specimen. Dorsal and ventral plates developed similarly to adults. Scapular plate 12.7 μm long. Claws 4.0-4.4 μm long.
Eggs. Up to one egg in exuvia was found.
Remarks. Two ventrolateral plates were not drawn in Claxton (2004), which is an unpublished PhD dissertation, thus the species described therein are not valid. However, having ascertained that these structures exist in specimens from Australia, the compared populations from both continents appeared identical in terms of morphology.  (Fig. 5), but their number varies considerably between both specimens and different elements of the armour, with the largest numbers present on the antero-central portion of the scapular plate and median plate 2 (23-48 pores/100 μm 2 , x ̅ = 32 and 23-47 pores/100 μm 2 , x ̅ = 33, respectively; N = 12) and lower numbers on the central portions of the caudal plate and paired segmental plate II (1-38 pores/100 μm 2 , x ̅ = 19 and 10-27 pores/100 μm 2 , x ̅ = 17, respectively; N = 12). Scapular plate with lighter rectangles (pseudofacets) between three or four transverse rows of merged dark epicuticular gran-  Supplementary lateral platelets present and detectable at lateral-most margins of the segmental plates (Fig. 5B). Venter with extremely weakly delineated plates (VII:4-2-2-4-2-2-1), only slightly darker than the surrounding ventral cuticle and without clear, sclerotised margins. Dark epicuticular granules and intra-cuticular pillars absent. Leg papillae undetectable under LCM. Both pulvini and pedal plates absent (Fig. 5B). Dentate collar IV absent. External claws spurless, but internal ones with minute spurs barely divergent from the claw branches (Fig. 5A, insert).

Juveniles. Not found.
Larvae. Not found.
Eggs. Not found. DNA sequences. Two 18S rRNA haplotypes (MT333469-70) and two 28S rRNA haplotypes (MT333462-3) and single ITS-1 haplotype (MT333478).  -5, 10, 13, 17 -20, 25, 27-29, 32-34, 41, 45, 47-48, 50-51) deposited in the Department of Zoology, Comenius University        Tables 5, 6). Cirri interni always shorter than cirri externi and both with poorly-developed cirrophores. Cirri A reach around 1/5-1/4 of the total body length (Tables 5, 6). Unappendaged. Cuticular sculpture consists of large epicuticular granules, true pores and intra-cuticular pillars (Fig. 10). In PCM, these structures appear, respectively, as conspicuous large dark spots, smaller bright spots and fine dark and dense punctuation (pseudogranulation). Epicuticular granules of irregular shape and size (up to ca. 1.6 μm) are merged together (as visible in SEM, Figs 8-10) and arranged in rows along the margins of all plates, although they are least visible or absent in posterior median plate 1 and posterior median plate 2. Moreover, in the cervical (neck) plate, the row or a double row of granules is also present along its transverse axis. Granules in rows appear as dark spots in PCM ( Fig. 7A-C). Similar rows of granules cover also folds which create the faceting of the scapular and caudal plate: median and two lateral longitudinal folds (at the level of cirrophores A) together with 3-4 posterior transverse folds on the scapular plate and two longitudinal folds dividing the caudal plate into three parallel portions (Figs 7A, B, 8, 9A, C). Finally, short longitudinal rows of granules divide both paired segmental plates and posterior portions of anterior m1-2 plates into left and right portions (Figs 7A, B,  8, 9B). Scattered granules also irregularly cover the surface of the cephalic, paired, caudal, anterior m1-2 and anterior parts of scapular plates (Figs 7A, B, 8, 9, 10A). Round, focusable pores (0.3-0.4 μm in diameter) are unequally distributed on dorsal plates in spaces between the scattered granules, on pedal plates IV and between transverse rows of granules in the scapular plate, but they are absent on the rows of granules (Figs 7C, 8, 9A-C, 10B). The density of pores varies between the sexes and elements of armour, with the largest pores present on the segmental plate II, anterior m2 and the scapular plate (21-40 pores/100 μm 2 , x ̅ = 29, 14-28 pores/100 μm 2 , x ̅ = 22 and 18-31pores/100 μm 2 , x ̅ = 26, respectively, N = 15 in females and 7-35 pores/100 μm 2 , x ̅ = 29, 0-34 pores/100 μm 2 , x ̅ = 25 and 11-33 pores/100 μm 2 , x ̅ = 25, respectively; N = 15 in males) and lowest density on caudal plate (16-27 pores/100 μm 2 , x ̅ = 22 in females and 7-28 pores/100 μm 2 , x ̅ = 20 in males; N = 15). On the scapular plate (in the area delimited with lateral longitudinal rows of granules), lines of pores tend to alternate with transverse rows of granules (Figs 7A-C, 8A, 9A, 10A, B). Regularly distributed round intra-cuticular pillars (0.1-0.2 μm in diameter) reinforce the entire cuticle (also under the granules), but they are well-visible only in the cephalic, scapular, both paired segmental, caudal, anterior median 1 and anterior median 2 plates (Fig. 10B), as well as on pedal plates IV. On the remaining cuticle, they are weakly (venter) or scarcely (legs) detectable. Cephalic plate with an anterior chalice-like incision. Each segmental and median plate consists of the anterior and the posterior portion separated each from other with a transverse bright poreless stripe in PCM. Therefore, paired segmental plates are subdivided into the narrow-er anterior (ca. 1/3-2/5 of the plate length) and the wider posterior portions, trapezoidal anterior median plate 1 is subdivided just behind its anterior margin, pentagonal anterior m2 (the largest amongst the median plates) is divided at approximately equally-long portions, triangular anterior portion of median plate 3 is ca. two times as long (along median body axis) as the posterior one with rounded posterior margin (dividing transverse line of anterior median plates 2-3 correspond with posterior corners of paired segmental plates). Pentagonal posterior median plates 1-2 subdivided at portions of about same lengths. On each body side, the first two pairs of supplementary lateral platelets are connected with anterior and posterior median plates 1-2 and the third pair is connected with the posterior portion of m3 and with the anterior edge of caudal plate. Anterior platelets of each pair have very distinctly-thickened lateral (lower) margins (Figs 7A,B,8,9).
Venter with transverse rows of weakly-developed plates unevenly sculptured with epicuticular granules similar to those on the dorsal plates, but a bit smaller. There are three rows of ventral plates in females (the plate formula III:2-2-1) and two rows in males (II:2-2) (Fig. 7D). The outer surface of legs with a narrow well-visible proximal pulvinus and a wide weak distal pedal plate placed in the central part of the leg. A single row of small epicuticular granules (similar to those on ventral plates) on the distal edge of pulvini in legs I-III (rarely, the second row also on their proximal     edge). Pedal plates I-III sculptured usually with three (sometimes with more) transverse rows of epicuticular granules, which can be either shortened or connected at their ends (Figs 7A-C, 8B). The pedal plate IV sculptured with distinct intra-cuticular pillars and scattered pores and distally hemmed with dentate collar. The collar with sharp teeth, always longer than the width of their bases and with the distance between teeth similar to their basal widths, although some pairs of teeth can be merged (Fig. 11). Papilla or spine on legs I-III absent, papilla on legs IV well developed. Claws slender, claws IV always slightly longer than claws I-III. External claws smooth, internal ones with a small spur pointing downwards and placed very close to the claw bases (Figs 7A-C, 7D, insert).
Juveniles. In appearance as adults, but smaller (111-112 μm) and with ventral plates just marked with rows of granules. Selected measurements of a shorter specimen: cephalic papilla 3.1 μm, scapular plate 14.9 μm long, claws I-III 5.1-5.6 and claws IV 6.5 μm long.
Larvae. 83-85 μm long. Dorsal plates (especially median ones) mostly with poorly-delineated margins, supplementary lateral platelets absent. Epicuticular granules less numerous than in adults, concentrated mainly on posterior margins of the cephalic, scapular, both paired and caudal plates. Cuticular pores less numerous than in adults, but intracuticular pillars, stripes of granules on the outer surface of legs, papilla on legs IV and dentate collar IV well developed. Ventral plates not visible in laterally orientated larvae. Claws with spurs formed as in adults. Some measurements of shorter specimen: cephalic papilla 2.5 μm, claws II-III 4.6-5.3 and claws IV 6.4 μm long.
Eggs. Not found.

Phylogeny and evolution of traits in Bryodelphax
Inter-generic tardigrade relationships are constantly being unravelled (Bertolani et al. 2014;Fujimoto et al. 2016;Gąsiorek et al. 2019a, b). Recently, Guil et al. (2019) proposed a new classification of Echiniscidae, with Bryodelphax included within Echiniscinae and having its own tribe Bryodelphaxini. Not only is such a proposal unjustified morphologically, as Bryodelphax is more similar to the Pseudechiniscus-like genera than to the Echiniscus lineage (Gąsiorek et al. 2018a), but, importantly, the current phylogenetic evidence is also not conclusive (different positions of the genus on echiniscid phylogenetic trees in Guil et al. 2019). In fact, the trait used to delimit putative Bryodelphaxini from Echiniscini, i.e. the presence of peribuccal cirrophores, is biased and unreliable -Bryodelphax has weakly outlined cirrophores due to the miniaturised body, but, essentially, the anatomy of cephalic cirri within both dubious tribes is identical. Therefore, the systematic distinction between Bryodelphaxini and Echiniscini is controversial and their status should be further verified. In terms of morphology, the genus should be currently recognised as a separate lineage of Echiniscidae, with the unsolved, long-standing problem of Bryochoerus (Kristensen 1987;Lisi et al. 2017;Gąsiorek 2018).
Regarding the phyletic relationships within the genus Bryodelphax, some intriguing conclusions can be drawn from the mapping of various phenotypic traits onto the phylogenetic tree (Fig. 1). Firstly, the division of the genus, based on the presence (weglarskae group) or absence (parvulus group) of ventral armature, has only practical significance for taxonomic purposes (see below), as the members of both groups are phylogenetically intermingled. This suggests that this trait is not conservative and its appearance should be regarded as convergent. Ventral plates are strongly sclerotised and evident in B. amphoterus, formerly affiliated within the parvulus group, which corroborates the supposition by Gąsiorek et al. (2017a) that these structures have been previously overlooked. Peculiarly, the reduction of ventral armature to plesiomorphic subcephalic and genital plate rows is known in Bryodelphax only in three Mediterranean species (B. amphoterus, B. maculatus and B. nigripunctatus sp. nov.).
Thirdly, species exhibiting different modes of reproduction are scattered on the tree. Two of the three known dioecious Bryodelphax spp., B. instabilis and B. nigripunctatus sp. nov., are not directly related (Fig. 1). This pattern, that is parthenogenetic and dioecious taxa mixed on the tree, is consistent with recent data for Paramacrobiotus and Milnesium (Guidetti et al. 2019;Morek and Michalczyk 2020). However, most of the other echiniscid genera are more consistent in terms of the mode of reproduction (Kristensen 1987).
Last but not least, in contrast to phenotypic traits, geographic distribution of the analysed species suggests their limited dispersal abilities and seems to be a reliable predictor of phylogenetic affinities within the genus. This intriguing pattern has been recently shown in the genus Milnesium Doyère, 1840 by Morek and Michalczyk (2020). Considering the remote phyletic relationship and contrasting body sizes in the two groups (Milnesium comprises largest tardigrades), these results suggest that tardigrade species, in general, may have much more restricted geographic distributions than the "Everything is everywhere" hypothesis predicts (Beijerinck 1913).
It ought to be noted that the lengths of the tree branches in the case of the Oriental clade are considerably shorter than those for the Western Palaearctic clade (Fig. 1), whereas the taxa of both lineages are well-separated from each other (with the exception of B. arenosus and Bryodelphax sp. nov. from Seram, which requires more data to solve its phyletic relationship with other congeners). Deeper nodes in the Western Palaearctic clade may result from: (a) longer divergence time needed for cladogenesis in this region (e.g. Ricklefs 2004), (b) higher extinction rate in the tropics (e.g. Jablonski et al. 2006) or (c) from under-sampling of lineages in the Palaearctic (e.g. Chown and Gaston 2000). If the last possibility is excluded, then the observed pattern may mean that speciation could be more rapid in tropical tardigrades, as postulated by the tropical cradle biodiversity hypothesis, which assumes that young evolutionary lineages are prevalent in the tropics (Stebbins 1974;Stenseth 1984;Jablonski et al. 2006;Moreau and Bell 2013). A similar scenario was recently demonstrated for oribatid mites (Pachl et al. 2017), but a much greater sampling effort is required for Bryodelphax spp. in order to test this hypothesis.
Taxonomic key to the genus Bryodelphax Since the last key by Fontoura et al. (2008), the number of described Bryodelphax species almost doubled (from 15 spp. considered valid in 2008 to 25 spp. gathered in the present contribution). Moreover, the 2008 key has several inadequacies: (I) three members of the weglarskae group (B. iohannis, B. sinensis and B. weglarskae) had an under-estimated number of ventral plates (nine instead of ten, six instead of seven and eight instead of nine, respectively); (II) B. asiaticus was delimited from B. parvulus by the absence of supplementary lateral platelets, but these structures are present in both species (Gąsiorek 2018). Our analysis of B. amphoterus paratypes revealed the presence of reduced, but evident ventral armature (formula II:2-2), thus the species belongs to the weglarskae group (Fig. 12). All these facts inclined us to present a new key allowing for the delimitation of females of the genus members at the adult life stage. The distinction between members of the weglarskae group (12 spp.) is rather straightforward, but the identification of species within the parvulus group (13 spp.) may pose a problem for beginner taxonomists. Consequently, we advise the greatest caution when identifying its members. Importantly, B. lijiangensis Yang, 2002 is designated as nomen dubium due to the insufficient description and the general habitus not conforming to the characteristics of the genus (trunk cirri in all lateral positions suggest its affinity to Echiniscus) and hence it is omitted from our key.  (15) (Fontoura, 1982) 19 (13)