The importance of being integrative: a remarkable case of synonymy in the genus Viridiscus (Heterotardigrada: Echiniscidae)

There are two predominant sources of taxonomically useful morphological variability in the diverse tardigrade family Echiniscidae: the internal structure and surface sculpture of the cuticular plates covering the dorsum (sculpturing) and the arrangement and morphology of the trunk appendages (chaetotaxy). However, since the appendages often exhibit intraspecific variation (they can be reduced or can develop asymmetrically), sculpturing has been considered more stable at the species level and descriptions of new echiniscid species based solely on morphology are still being published. Here, we present a case study in which a detailed analysis of the morphology and multiple genetic markers of several species of the genus Viridiscus shows that cuticular sculpture may also exhibit considerable intraspecific variation and lead to false taxonomic conclusions. In a population collected from the eastern Nearctic, in the type locality of the recently described species V. miraviridis, individuals with transitional morphotypes between those reported for V. viridissimus and V. miraviridis were found. Importantly, all morphotypes within the viridissimus–miraviridis spectrum were grouped in a single monospecific clade according to rapidly evolving markers (ITS-1, ITS-2 and COI). Given the morphological and genetic evidence, we establish V. miraviridis as a junior synonym of V. viridissimus. This study explicitly demonstrates that a lack of DNA data associated with morphological descriptions of new taxa jeopardizes the efforts to unclutter tardigrade systematics. Additionally, V. perviridis and V. viridissimus are reported from Lâm Đồng Province in southern Vietnam, which considerably broadens their known geographic ranges. Supplementary Information The online version contains supplementary material available at 10.1186/s40851-021-00181-z.

Viridiscus is an unappendaged (sensu [15]) genus of echiniscids that displays green to almost black body coloration and composite sculpturing, comprising a dense endocuticular sponge layer and flat epicuticular granules [15][16][17][18]. In the redescription of the type species Viridiscus viridis [19], Pilato et al. [18] highlighted morphological differences in dorsal sculpturing between the known representatives of Viridiscus. Last year, a new Nearctic species, Viridiscus miraviridis, was described based on its extraordinarily developed epicuticular layer forming sclerotized ridges, a character previously unknown in the genus [20]. However, attempts to obtain DNA barcode data were unsuccessful, so the original description was based solely on morphological characters. Thus, to amend the description and pinpoint the phylogenetic position of V. miraviridis, we sequenced five genetic markers, including four nuclear (18S rRNA, 28S rRNA, ITS-1 and ITS-2) and one mitochondrial (COI) marker, of topotypic specimens of the species and of Viridiscus viridissimus [21] and Viridiscus aff. viridianus [17], which we also found in a moss sample from Tennessee. For comparative purposes, we sequenced additional Viridiscus specimens from Madeira and Vietnam.
Unexpectedly, we found that despite clear morphological differences between V. viridissimus and V. miraviridis, individuals representing both morphotypes shared the same DNA barcodes, and their conspecificity was confirmed by three species delineation methods (ABGD, [22]; bPTP, [23]; and ASAP, [24]). This discovery gave us an opportunity to discuss the currently used taxonomic criteria and note potential problems induced by describing limnoterrestrial tardigrade species without associated genetic barcodes.

Specimens and morphology
Populations of Viridiscus were obtained from a total of seven moss samples collected in three locales: All specimens extracted from the samples (using standard methods described in [25]) were subsequently divided into groups used for light microscopy analyses and Sanger sequencing (Table 1). Some specimens were mounted in a small drop of Hoyer's medium on permanent slides and examined by phase-contrast microscopy (PCM) under an Olympus BX53 light microscope with an Olympus DP74 digital camera at Jagiellonian University. Syntypes of Viridiscus perviridis [16], paratypes of Viridiscus viridianus [17], and American specimens Genotyping DNA was extracted from individual animals following a Chelex® 100 resin (Bio-Rad) extraction method [26], with modifications according to [27]. Vouchers (specifically hologenophores) [28] were obtained when possible. Five DNA fragments were sequenced: the 18S rRNA small ribosomal subunit, the 28S rRNA large ribosomal Table 1 List of species and populations used in analyses. Types of analyses: PCMimaging and morphometry in PCM; DNA -DNA sequencing. The number in each analysis indicates how many specimens were utilized for a given method: ♀sexually mature females, ♂sexually mature males, jjuveniles, llarvae, vvouchers (please note that in some cases, the same specimens were used for both DNA and LCM analyses) Population also utilized in [15] b The first records of both species from Southeast Asia and the Indomalayan realm subunit, the internal transcribed spacers ITS-1 and ITS-2, and the cytochrome oxidase subunit I (COI). All fragments were amplified and sequenced according to the protocols described in [27]; the primers and original references for the specific PCR programs are listed in Supplementary Material 1. GenBank accession numbers for all specimens are provided in Table 2. We were not able to obtain COI barcodes for V. perviridis (see SM1). All ITS and COI sequences were aligned with sequences of Echiniscus succineus as an outgroup using the ClustalW Multiple Alignment tool [29] implemented in BioEdit [30]. The aligned fragments were edited in BioEdit, with Table 2 GenBank identifiers for sequenced Viridiscus specimens analyzed in the present study (new sequences are indicated in bold) gaps left intact in the case of ITS sequences. The alignments are provided as Supplementary Materials 2, 3 and 4. The 18S rRNA and 28S rRNA sequences were not used in developing primary species hypotheses, as they are too conservative [31] and thus not suitable for molecular species discrimination. Nevertheless, since these markers can be used in phylogenetic studies, they are also provided here.

Phylogeny
The sequences of the ITS fragments were concatenated to generate a matrix of 1064 bp in SequenceMatrix [32]. Using PartitionFinder version 2.1.1 [33] with the application of the Bayesian information criterion (BIC) and a greedy algorithm [34], the best substitution model and partitioning scheme were chosen for posterior phylogenetic analysis. As the best-fit partitioning scheme, Parti-tionFinder suggested the retention of two partitions (I: ITS-1, II: ITS-2), and the best fit model was TVM + G for both partitions; in the case of the COI matrix (611 bp), the best model was TIM + G. Bayesian inference (BI) marginal posterior probabilities were calculated using MrBayes v.3.2 [35]. 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 that the two independent analyses had converged. Tracer v1.6 [36] was then used to ensure that Markov chains had reached stationarity and to determine the correct burn-in for the analysis (i.e., the first 10% of generations). The effective sample size values were greater than 200, and the consensus tree was obtained after summarizing the resulting topologies and discarding the burn-in. All final consensus trees were viewed and visualized by using  [24] to obtain three independent marker-based primary species hypotheses using uncorrected pairwise distances. The partitions with the lowest ASAP scores and p values < 0.05 were chosen as the best-fit hypotheses. In tandem, we applied another phenetic method of species delineation based on genetic distances (automatic barcode gap discovery (ABGD, [22]), with the default options) to the three alignments. Finally, Bayesian Poisson tree processes (bPTP, [23]) were applied to the Bayesian phylogenetic trees of the three markers. In all cases, we discarded the outgroup to protect against eventual biases caused by the distant relationship between the outgroup and ingroup taxa. The calculations were conducted with 100,000 MCMC generations, thinning the set to 100, with 10% burn-in, and with searches for maximum likelihood and Bayesian solutions.

Morphology (Figs. 1-2)
Except for samples PT.042 and US.077, all of the other samples analyzed in the present study contained mixed Viridiscus morphotypes ( Table 1). As in the original description of V. viridianus, the dorsal sculpture of V. aff. viridianus individuals from the USA (samples US.077, US.078, and US.081) was composed of densely packed epicuticular granules (Fig. 1a), whereas specimens of V. perviridis from Portugal (Madeira; sample PT.042) and Vietnam (samples VN.027 and VN.028) showed a similar phenotype but with a better developed endocuticular sponge layer (Fig. 1b), which is in agreement with the original description of V. perviridis. In both taxa, there was very little intraspecific morphological variation in the dorsal sculpturing. However, 8/21 (38%) of the analyzed specimens that otherwise fit the description of V. viridianus exhibited extremely long cirri A (50-100% of the body length), which are characteristic of V. perviridis (according to [17], the cirri A of V. viridianus reach a maximum length of only 20% of body length). Given these phenotypic discrepancies and the lack of available topotypic DNA sequences of V. viridianus s.s., we classified our specimens as Viridiscus aff. viridianus.
The American sample US.078 was also inhabited by individuals of the V. viridissimus (Fig. 2a) and the V. miraviridis (Figs. 2d) morphotypes as well as by tardigrades with two intermediate morphotypes (Fig. 2b-c). In other words, we found four morphotypes of dorsal sculpturing, constituting a viridissimus-miraviridis spectrum, where the first morphotype was attributable to V. viridissimus, the fourth morphotype was identifiable as V. miraviridis, and the two intermediate morphotypes were not classifiable as any known species. In brief, along this spectrum, the area covered with epicuticular granules increases, and adjacent round pores fuse into irregularly shaped pores (see Fig. 2 for PCM photomicrographs and a detailed description of the four morphotypes). Finally, the American (US.080 and US.081) and Vietnamese (VN.027 and VN.028) samples contained the V. viridissimus morphotype (Fig. 2a).
Integration of phenotype and genotype data (Figs. 1, 2 and 3) We attribute the oversplitting of lineages into putative candidate species by ABGD and bPTP to the weaker performance of the two methods in comparison to ASAP [24]. Given that all sequenced individuals representing the viridissimus-miraviridis spectrum formed a single well supported but internally poorly differentiated clade (Figs. 2-3), we conclude that V. miraviridis is a junior synonym of V. viridissimus, representing a rare morphotype of the senior species.

Discussion
The complex of species previously known as the Echiniscus viridis group but currently classified in the recently erected genus Viridiscus has always drawn the attention of tardigrade taxonomists due to the persistence of the extraordinary green body pigmentation after mounting [15-18, 20, 38]. Despite the crucial revisions by Pilato et al. [17,18], not all of the described Viridiscus spp. are properly delineated. In fact, none of the species in the genus has been described or redescribed under the integrative taxonomy framework. For example, one of the key characteristics separating V. perviridis and V. viridianus is the length of cirri A, which greatly exceeds 50% of the body length in the former. However, we encountered single individuals of V. aff. viridianus exhibiting particularly long cirri A (50-100% of the body length) in the samples from Tennessee. Individuals with such long cirri may have prompted Maucci [39] to identify North American Viridiscus specimens as V. perviridis. Likewise, Nelson et al. [20] identified Tennessee specimens with long cirri as V. perviridis. However, given that the Tennessee specimens analyzed in this study with perviridis-like long cirri exhibited viridianus-like sculpturing, we identified them as V. aff. viridianus, together with similar specimens that have short cirri. Nevertheless, removing the uncertainty from the taxonomic identification of Tennessee V. aff. viridianus will require topotype DNA sequences of V. perviridis and V. viridianus. This illustrates the power and value of genetic data associated with type (or neotype/topotype) series and shows how problematic the lack of such data can be.
However, our study provides an even more explicit example demonstrating the importance of integrating classical methods (morphology and morphometry) and molecular tools (phylogeny and genetic delineation) for precise taxonomic inference. Although our observation of the morphological viridissimus-miraviridis spectrum itself was an indication that the validity of V. miraviridis  was questionable, it did not allow us to determine whether the spectrum represents a single species or two closely related and interbreeding species (males of both morphotypes were found, which could favor the latter hypothesis). Interestingly, all specimens from the Maucci collection originating from Tennessee [39] present a "classical" V. viridissimus morphotype (such as shown in Fig. 2a). Thus, only the use of variable genetic markers, such as ITS and COI, could ultimately verify the phylogenetic position and, thus, the taxonomic identity of the observed morphotypes. Studies addressing milnesiids, possibly one of the most speciose morphologically static limnoterrestrial tardigrade lineages [40], have already emphasized that basing further descriptions of new limnoterrestrial tardigrade species solely on a morphological analysis of a small number of specimens may be detrimental to tardigrade classification [41]. Although echiniscids are the richest in taxonomically informative traits among limnoterrestrial tardigrades [11], distinguishing between intra-and interspecific variability using phenotypes alone may be unreliable and misleading [42]. The necessity of an integrative approach has also been stressed in other studies conducted on Echiniscidae (e.g., [43,44]). Thus, the more we know about limnoterrestrial tardigrade diversity and evolution, the clearer it becomes that abandoning phenotype-based taxonomy and adopting an integrative approach is the only way to make real progress in describing and understanding tardigrade diversity, biogeography and evolution. Otherwise, we will likely face an unprecedented rate of taxonomic inflation (i.e., increases in the number of synonyms [45]), considering how much unknown tardigrade diversity likely exists and how few taxonomically useful phenotypic characters limnoterrestrial tardigrades exhibit (e.g., see recent papers addressing Pseudechiniscus diversity: [46][47][48]).
There is concern that DNA tools are not available to everyone and that they may limit the development of young taxonomists and 'citizen scientists', especially in developing countries. While it is true that genetic analysis entails additional costs, the price per sequenced barcode has been rapidly decreasing over the last two decades. More importantly, there are a number of laboratories around the world that are willing to provide genetic expertise through collaboration. The association of even a single variable marker, such as COI or ITS-2, with the morphological characterization of a new taxon greatly reduces the chance of taxonomic inflation without being costly in terms of effort or money. Moreover, integrative redescriptions, especially for the type species of genera and species complexes, seem more important than the description of 'regular' new species because poorly described type taxa often constitute a serious obstacle to elucidating the biodiversity of a given lineage (e.g., see [48] for heterotardigrades and [49] for eutardigrades). Abandoning classical taxonomy means that when it is not possible to obtain DNA data (e.g., when only old material or specimens preserved on slides are available), the description of some taxa will be postponed indefinitely until new material becomes available. However, in such cases, we need to consider whether it is more important to publish a description of a new species based solely on morphology and risk the further cluttering of limnoterrestrial tardigrade taxonomy or to wait and perform genetic analysis to advance scientific progress in the field.
Through molecular and comparative phylogenetic analyses and the integration of phenotypic and genetic data, taxonomy and systematics should gradually become more objective and more testable [50]. Fortunately, even though tardigrade species are still being described based solely on morphology, the integrative approach has become the "gold standard" since the first such study was published a decade ago [51], and the proportion of integrative works is constantly increasing [52]. Thus, hopefully by the end of this decade, journal editors and reviewers will become reluctant to accept descriptions of new limnoterrestrial taxa and, eventually, faunistic records without genetic evidence.

Conclusions
Neither morphology nor molecular methods should be used alone to delineate tardigrade species, as this leads to for the accumulation of taxonomic issues over many decades of work. We want to raise awareness that further describing species based solely on morphology will inevitably result in serious taxonomic inflation and unreliable biogeographic data.