Spatio‐temporal patterns of eukaryotic biodiversity in shallow hard‐bottom communities from the West Antarctic Peninsula revealed by DNA metabarcoding

We studied molecular eukaryotic biodiversity patterns in shallow hard‐bottom Antarctic benthic communities using community DNA metabarcoding. Polar ecosystems are extremely exposed to climate change, and benthic macroinvertebrate communities have demonstrated rapid response to a range of natural and anthropogenic pressures. However, these rich and diverse ecosystems are poorly studied, revealing how little is known about the biodiversity of the Antarctic benthos associated with hard‐bottom habitats.


| INTRODUC TI ON
The human perception of cold waters as an unpleasant extreme environment is not shared by polar organisms. Despite sub-zero water temperatures, the biodiversity and biomass of polar organisms in Antarctica are high, thanks to their adaptation and evolution over 300 million years of isolation (Peck, 2018). However, Antarctica is a fragile environment under threat from increased human activities (Pertierra et al., 2021;Tin et al., 2009). The importance of conserving global biodiversity has come to the public's attention in recent decades due to unprecedented rates of species extinctions driven by human activities (Duarte et al., 2020;Luypaert et al., 2020). The West Antarctic Peninsula (WAP) has experienced warming at significantly greater rates than the rest of the Antarctic continent and possibly greater than any other region on Earth over the past 50 years (Turner et al., 2005;van Wessem et al., 2015). These changes have driven habitat and community shifts with significant impacts on biodiversity and ecosystem functioning. Although large-scale ecological conditions (i.e. low and relatively stable temperatures, seasonality of primary production and low terrigenous input) are similar around most parts of the continent (Bullivant, 1959;Clarke & Leakey, 1996), the benthic biodiversity in the Antarctic region is remarkably patchy (Almond et al., 2021) and depends on complex interactions between physical and biological factors that are not easily defined (Smale & Barnes, 2008).
In Antarctica, marine habitats shallower than 100 m occupy an estimated total area of approximately 25,000 km 2 (Clark et al., 2015).
Antarctic benthos has been categorized as a relatively homogenous biological unit (Downey et al., 2012;Smale, 2008). The benthos is the richest element of the food web in terms of numbers of macrospecies, dominated by suspension feeders in the shallows and deposit feeders in deeper waters (Griffiths, 2010), although their roles and interactions are poorly known. Over 4100 benthic species have been reported from the Southern Ocean (SO), with polychaetes, gastropods and amphipods being the taxa with highest number of described species (Clarke & Johnston, 2003). Gutt et al. (2004) estimated 11,000-17,000 macrozoobenthic species alone using statistical techniques to extrapolate species diversity from a sampled area in the Weddell Sea, pointing out that even this large figure may be an underestimate of true diversity. Other taxa with high species richness include bryozoans and sponges (Arntz et al., 1994). Bivalve molluscs and isopods show lower species richness in the SO than in equivalent areas of shelf elsewhere, while some groups of decapod crustaceans are completely absent, and pycnogonids, echinoderms and many suspension feeders are rich and diverse (Clarke, 1990;Clarke & Johnston, 2003). All marine phyla are present in Antarctic waters (de Broyer et al., 2014). Recent estimates suggest that between 33% and 91% of all marine species around the world have never been named (Appeltans et al., 2012;Mora et al., 2011), and taxonomic knowledge gaps have limited our ability to investigate patterns of diversity beyond a few indicator groups (Tittensor et al., 2010).
Nearshore benthic communities of Antarctica have demonstrated a rapid response to a range of natural and anthropogenic pressures and therefore, their assessment can be used as an indicator of ecosystem health (Magni, 2003). Unfortunately, visual morphology, widely used for species identification in coastal and marine communities, is cumbersome and entails limitations (timeconsuming, expensive, requires extensive taxonomic expertise; Aylagas et al., 2016;Wood et al., 2013). Despite hard-bottoms usually support higher abundance and diversity than soft-bottoms, very few studies have consistently evaluated the benthic diversity of hard-bottom habitats at different spatial scales along the WAP. The real extent of biodiversity and its temporal and spatial patterns remain unknown for the majority of the WAP, despite being one of the most studied areas in Antarctica. Hard-substrate habitats in shallow coastal Antarctic waters tend to display a gradient of benthic communities related to ice cover (Clark et al., 2013(Clark et al., , 2015. Furthermore, iceberg scouring causes a significant reduction in benthic biomass and biodiversity at small spatial scales (Conlan & Kvitek, 2005). Shallow areas of heavily disturbed sites are characterized by assemblages of low diversity and biomass able to be rapidly re-colonized after impacts (Gutt & Piepenburg, 2003;Peck et al., 1999;Smale, 2007;Teixidó et al., 2004). Intermediate frequencies of ice disturbance are thought to enhance diversity by preventing species domination and creating a patchwork of habitat and communities in various stages of recovery (Brown et al., 2004;Conlan & Kvitek, 2005;Smale et al., 2007). Thus, evaluations of biodiversity in benthic fauna are of critical importance for understanding ecosystem functioning, sustainability and resilience.
Documenting the diversity of marine life is challenging because many species are cryptic, small and/or rare, and belong to poorly known groups (Leray & Knowlton, 2015). Complex hard substrates, particularly in Antarctica, provide huge challenges for consistent sampling, because the communities are largely inaccessible to exhaustive qualitative or quantitative biodiversity assessments (de Broyer et al., 2014), and also because of the abundance of small epibionts, and the massive, colonial or modular morphology of many communities are probably a consequence of a combination of several biotic and abiotic factors (i.e. ice disturbance, food supply and competition).

K E Y W O R D S
benthic invertebrates, bulk DNA, community DNA, community ecology, eukaryotic, metabarcoding, Southern Ocean species. Furthermore, an important constrain is that monitoring is typically focussed on macrofauna (>1 mm size), which is only a part of a community that also comprises microeukaryotes and other inconspicuous taxa. Surveys focussing only on macrofauna may underestimate the sensitivity to changes in the whole community (Lanzén et al., 2016;Leray & Knowlton, 2015). Thus, meiofauna assessments are needed to generate a more complete and mechanistic understanding of marine benthic ecosystems (Bourlat et al., 2013).
Meiofauna (the microscopic taxa generally between 45-500 μm) are important members of the benthic ecosystems, playing a critical role in carbon transfer and nutrient cycling (Fonseca et al., 2017;Schratzberger & Ingels, 2018).
Another limitation to improve our understanding of Antarctic marine benthic ecosystems is the ability to generate comparable biological time-series data. Time series are essential tools for studying changes within ecosystems and could be particularly useful for understanding the effects of anthropogenic impacts and global change, but the remoteness of Antarctic habitats represents a big challenge to sample such long-time series (Fonseca et al., 2022;Schratzberger & Ingels, 2018).
Molecular biodiversity assessment methods are promising not only for the fundamental understanding of diversity but also for biodiversity monitoring in the context of global change (Bourlat et al., 2013). DNA barcoding is a technique that uses a short gene sequence, from a standardized region of the genome as a diagnostic 'biomarker' for species (Hebert et al., 2003). Natural variability of the mitochondrial cytochrome c oxidase I gene (COI) enables the taxonomic resolution of metazoan taxa at the species level and allows to resolve cryptic species complexes (Leray & Knowlton, 2016;Wangensteen et al., 2018). DNA metabarcoding has emerged as a powerful tool for quantifying biodiversity using genetic sequences extracted from an environmental or bulk sample (i.e. water, sediment, community; Taberlet et al., 2012).
Community DNA from bulk samples can be defined as the organismal DNA extracted from whole individuals that were presumably alive (or recently dead) at the time of sampling (Rodriguez-Ezpeleta et al., 2021). One of the main advantages of community-DNA studies is that they may uncover larval, small (meiofaunal) or rare taxa that may be missed by traditional surveys, being a powerful tool for biodiversity monitoring. Bulk samples often contain many different taxa that vary several orders of magnitude in biomass. (Rodriguez-Ezpeleta et al., 2021). Few specimens of high biomass will dominate the dataset, potentially leading to smaller specimens remaining undetected. Sorting taxa by size and pooling them proportionately according to their abundance leads to a more equal amplification of taxa compared with the processing of complete samples without sorting (Elbrecht et al., 2017;Wangensteen & Turon, 2017). Thus, metabarcoding provides a cost-effective, ecosystem-wide method for the assessment of biodiversity.
Molecular methods are particularly powerful when combined with standardized sampling, allowing for direct comparisons across space and through time. The resulting barcode sequence reads can be subsequently matched with sequences assigned to taxon names accessed from databases such as NCBI GenBank (Sayers et al., 2020) and the Barcode of Life Data System (BOLD; Ratnasingham & Hebert, 2007). Although the use of extra-organismal DNA (the DNA released from cell lysis; Rodriguez-Ezpeleta et al., 2021) is increasingly applied to water samples, there is not robust evidence that the entire macroinvertebrate community can be detected using exclusively extra-organismal DNA (Antich, Palacin, Cebrian, et al., 2021;Antich, Palacín, Wangensteen, et al., 2021;Rey et al., 2020). Hence, metabarcoding of benthic communities is needed for providing valuable information on unknown benthic biodiversity (Fonseca et al., 2010;Leray & Knowlton, 2015), and it can therefore be used to assess Antarctic species richness. Comparisons of benthic community composition between different Antarctic regions are difficult due to the fact that researchers have used different collection methods, depths, sieve sizes and temporal ranges. Furthermore, Antarctic biodiversity has not so far been comprehensively explored at the molecular diversity level, with the exception of some specific groups (Grant et al., 2011). Many marine shelf, deep water and polar areas are notoriously undersampled.
Different DNA metabarcoding approaches have been recently applied to assess biodiversity from the SO. These studies have mainly focussed on microbial communities (Flaviani et al., 2018;Luria et al., 2014), fungal diversity (Ogaki et al., 2021), sponge microbiomes (Castro-Fernández et al., 2023;Sacristán-Soriano et al., 2020), metazoans (Clarke et al., 2021;Vause et al., 2019) and benthic meiofauna (Brannock et al., 2018;Fonseca et al., 2017Fonseca et al., , 2022. However, studies analysing community DNA extracted from hard-bottom communities are lacking, which would provide a powerful tool to complement existing approaches, and a timely opportunity to gain insight into alpha and beta-diversity patterns of Antarctic hard-bottom communities, by providing the needed baselines to assess the resilience of these ecosystems in the context of global warming.
The goal of this study is to provide a global description of the shallow benthic communities associated with hard-bottoms along the WAP using high-throughput sequencing of community DNA, to provide a baseline for future biomonitoring studies in hard-bottom ecosystems. Additionally, we evaluated the temporal patterns of two nearby communities during three consecutive years in Deception Island.

| Sample collection
Samples were collected along the WAP and South Shetland Islands

| Sample processing
Benthic samples were filtered through a 63 μm mesh to discard the seawater and immediately kept at −20°C until processed in the laboratory. Then, the samples were separated into three different size fractions (A: >10 mm; B: 1-10 mm; C: 63 μm-1 mm) using a stainlesssteel mesh sieve column. Each fraction was homogenized with a blender and stored in ethanol at −20°C until DNA extraction. All the equipment was carefully bleached between samples. Our sample dataset thus consisted of 99 benthic samples (7 communities × 3 replicates × 3 fractions and 2 communities × 3 years × 3 replicates × 3 fractions).

| DNA extraction, PCR amplification and library preparation
All procedures were performed in a laminar flow cabinet sterilized with UV light between samples. DNA from benthic samples was extracted using 10 g of homogenized material using the DNeasy PowerMax Soil Kit (Qiagen). Extraction of negative controls was included for each extraction event. We amplified the partial COI attached to the primers to improve sequence diversity for Illumina processing. PCR blanks were run by amplifying the PCR mixture without any DNA template. Also, negative controls were added by processing sand samples that were charred in a furnace (400°C for 24 h) and then sieved and processed as the benthic samples.
PCR condition for COI amplification followed (Wangensteen  et al., 2018) and the success of PCR amplification was checked through gel electrophoresis. Then, DNA was purified and concentrated using MinElute PCR Purification Kit (Qiagen) and measured with a Qubit 3.0 fluorometer (Thermo Fisher Scientific).
Amplification products were pooled to build two Illumina libraries using Nextflex PCR-free library preparation kit (BIOO Scientific).
Both libraries were sequenced together in an Illumina MiSeq V3 run using 2×250 bp paired-end sequencing.

| Bioinformatic analyses
The initial bioinformatic steps used the OBItools package (Boyer et al., 2016). Illuminapairedend was used to align paired-end reads and keep only those with >40 alignment quality scores. Reads were demultiplexed and primer sequences were removed using ngsfilter. Those with mismatched primer tags at any end were discarded. Obigrep and obiuniq were used to perform a length filter (retaining only those between 310 and 317 bp) and dereplicate sequences. Uchime-denovo algorithm from VSEARCH v2.7.1 (Rognes et al., 2016), was used to remove chimeric amplicons. Sequences were then clustered into molecular operational taxonomic units (MOTUs) with SWARM v2.1.7 using a distance value of d = 13, which is the optimal value for this marker as explained in Antich, Palacin, Cebrian, et al. (2021) and Antich, Palacín, Wangensteen, et al. (2021). Singletons (MOTUs with just one read) were removed after this step to minimize data loss (Atienza et al., 2020). Taxonomic assignment was performed using ecotag (Boyer et al., 2016) and a custom reference database containing sequences from the EMBL and from the Barcode of Life Database (BOLD). This reference database is publicly available from https://github.com/uit-metab arcod ing/DUFA. Assignment of some metazoan sequences was further improved by querying the BOLD database. Sequences with a species name assigned and with an identity match >95% in BOLD were kept, whereas matches below this threshold, even if assigned to the species level by ecotag, were downgraded to genus level. The final refining steps consisted of deleting any MOTU for which reads in blank or negative controls represented more than 10% of total reads for that MOTU across all samples (Wangensteen & Turon, 2017). A minimum relative abundance filter was also applied, removing, for a given PCR replicate, the MOTUs that represented <0.005% of total reads of that replicate. We also removed MOTUs that had a combined total of <6 reads across all samples after the previous steps. All MOTUs assigned to Prokaryotes or classified as evident nontarget, contamination taxa (e.g. Insecta, Mammalia, Arachnida) were removed.

| Data analyses
All analyses were performed in R version 4.0.5 (https://www.r-proje ct.org/), and graphic visualizations were done with the ggplot2 package (Wickham et al., 2016). Rarefaction, subsampling and biodiversity estimates were calculated using the R package vegan (Oksanen TA B L E 1 Coordinates of each sampling location.

Lon (W)
Bahia  (Chen, 2018), to represent the MOTUs overlap between fractions. The three fractions of each biological sample were analysed separately as different replicates. To account for differences in total number of reads, samples were rarefied to the lowest number of sequences (7000 reads) before calculating MOTU richness. Total reads were then transformed to relative abundance. Alpha diversities were compared using different metrics: species richness, Shannon and Simpson diversity indices, and tested for significant differences (one-way analysis of variance (ANOVA) on sqrt-transformed data for the factor station). Potential dissimilarities between regions (beta diversity) in terms of MOTU composition were estimated using the Jaccard dissimilarity index to minimize the weight given to 'absence' values and implemented using vegdist function. Also, the mantel function was used to assess the correlation (Pearson correlation coefficient) between community dissimilarities and geographical distance. The Mantel approach is appropriate for testing the variation in beta diversity among groups of sites (Legendre et al., 2005).

| RE SULTS
We metabarcoded a total of 99 samples. After quality filtering, demultiplexing, dereplicating and chimaera removal, we had a total F value = 20.28; p < .0001).
Our dataset revealed a total of 10 different kingdom-level lineages (Table S1). Most MOTUs (1328)    Cormoran Town had the lowest diversity (Figure 4).

| Taxonomic composition
Differences between stations were observed not only in the lowest taxonomic resolution but also in the phylum level within macrobenthic organisms (Figure 7a). Nonetheless, 100 of all MOTUs were present in all localities, forming a benthic Antarctic core community.
Five phyla encompassed more than 80% of the total reads (Annelida, Porifera, Echinodermata, Arthropoda and Nemertea), forming different macrobenthic settlements at each locality ( Figure 8). Due to the low species-level resolution, the calculation of the indicator value scores (IndVal) used to gather information about specific F I G U R E 3 (Top) Donut charts by stations representing the regional composition by size fraction (a; outer rings, b; medium rings and c; inner rings) of the average relative read abundance by phylum and (Bottom) lollipop charts showing the mean MOTU richness by fraction and station. Some minor phyla were grouped (Alveolata: apicomplexan and bigyra. Fungi: ascomycota, basidiomycota, blastocladiomycota and mucoromycota. Hacrobia; cryptophyta and haptophyta. Other Metazoa; acanthocephala, braquiopoda, cephalorhyncha, chaetognatha, platyhelminthes, rotifera and xenacoelomorpha).  Figure S3).

| Bahia Falsa (Livingston Island)
This location is a pristine habitat dominated by the proximity of the

| Punta Polaca (Livingston Island)
Locality with two research stations in the vicinity and very exposed to the waves with no apparent influence of ice. Several sponges form the main structure of this community (Haliclona sp., Dendrilla

| Cormoran Town (Deception Island)
Inside this active Antarctic volcano, we found the less diverse habitat of our study. The entrance of the island and the east part of Whaler

| Temporal patterns
Despite the close proximity between the two stations on Fildes values (see Table S3). In addition, a slight but important decrease in the evenness was observed along the years in both stations ( Figure S5). Moreover, no differences were found related to the number of reads between these two stations. The community composition showed significant differences between stations (PERMANOVA p < .0009) and years (PERMANOVA p < .0009).
Although the two communities were grouped separately in the nmMDS (Figure 5b), there was a tendency to converge among the years, probably due to the drastic annual decrease of the MOTUs richness in Fildes.

| Beta diversity
Overall, there were changes in benthic community compositions be-

| DISCUSS ION
The logistic difficulties of accessing the remote Antarctic ecosystems call for cost-effective and comprehensive tools for the evaluation of community diversity facing a changing climate. Despite the relatively easy way in which eDNA water sampling can be performed, water eDNA has been proved to be a poor surrogate for benthic structure and composition (Antich, Palacin, Cebrian, et al., 2021;Antich, Palacín, Wangensteen, et al., 2021;Cowart et al., 2018) and direct sampling methods are required for monitoring these complex communities via metabarcoding techniques. The sampling of benthic hard-bottom habitats requires direct access to the environments and involves more effort than sampling sediments or water, which can be accessed remotely.
An important challenge for molecular methods is the presence of large taxonomic gaps in global reference sequence databases.
Kvist ( habitats is yet to be taxonomically described in regard to the identification and development of associated barcodes. Only around 5% of our MOTUs had a full taxonomical match at species level (>99% similarity) against public databases. Despite limitations due to taxonomic assignment at the species level, this study shows interesting insights into the extend of biodiversity in Antarctic shallow benthos. Similar levels of diversity were found when compared to other marine studies carried out in temperate regions using the same methodology (Antich, Palacin, Cebrian, et al., 2021;Antich, Palacín, Wangensteen, et al., 2021). Additionally, our study pinpoints the importance of MOTU diversity and the relative read abundance of metazoans in samples from the Antarctic benthos. These findings are fundamental to identifying key species in the benthic communities, understanding trophic relationships and evaluating ecosystem dynamics.
Metabarcoding of benthic communities, using a broad range of eukaryotic marker (COI), retrieved substantial differences in community structure among localities at different latitudes and years. Our study provides a framework for the assessment of biodiversity associated with shallow rocky bottoms along the WAP and South Shetland Islands The amplification efficiency of the COI gene varies among groups and species, and this might bias the quantitative results (Descôteaux et al., 2021;Hajibabaei et al., 2011). However, even though correlations between read abundance and species biomass are not perfect, the relative read abundance from community-DNA metabarcoding datasets has shown to be significantly associated with the relative biomass of the species, which adds support to the development of assessment tools based on semiquantitative approaches for this kind of metabarcoding data (Elbrecht & Leese, 2015;Ershova et al., 2021).
Our benthic samples are community or bulk DNA (Rodriguez-Ezpeleta et al., 2021). This type of DNA is typical of significant quantity and high quality. In addition, the mesh size used here (63 μm), guarantees that most prokaryotes and the smallest fraction of the microeukaryotes were washed out, along with viruses, cell debris and extracellular DNA, which explains the high number of reads assigned to eukaryotes. When using quantitative metrics based on relative read abundances, replicates of each ecological location always clustered together and thus the combined replicate samples accurately reflected alpha and beta diversities from the WAP. Benthic community composition can be extremely variable even within small spatial scales, but all localities maintained a similar core community. Local patchiness and structure within these communities are probably a consequence of a combination of several biotic and abiotic factors (i.e. ice disturbance, food supply, competition, seabed topography and life-cycle strategies; Clarke et al., 2021;Fonseca et al., 2022).
Understanding spatial-temporal changes in ecological communities of coastal ecosystems is crucial in the context of climate change.

| Spatial patterns
Although no clear trends in MOTUs richness or Shannon diversity index were observed with latitude, a correlation between the geographical distance and the dissimilarity of the community composition at each locality was detected. Therefore, as samples became physically more separated their community composition became more dissimilar. Furthermore, some differences in community composition were found between the WAP stations and the SSI stations and even between localities within areas. The contribution of each site to beta diversity was positively correlated with MOTU richness. This is especially remarkable for Rothera station, the most geographically distant station, which has comparatively lower values for both alpha and beta diversity compared with other localities. The lowest diversity values were found in Cormoran Town, at Deception Island. Surprisingly, Fildes Point, also in Deception Island, and separated only by a few tens of metres from Cormoran Town, was the station with the highest MOTU richness in our study. This is a striking example of the well-documented patchiness in benthic Antarctic communities (Gutt & Piepenburg, 2003;Smale, 2008) and may be related to the exposure to strong currents in this site, closer to the Neptune Bellows (bay entrance).
Antarctic Peninsula sites can be remarkably rich in species, as well as having high biomass and abundance values relative to those in the Arctic or even temperate regions (Arntz et al., 1997;Brey & Clarke, 1993, Fonseca et al., 2017, 2022 MOTUs were shared within all seven stations analysed here. This common core community in all these distant localities, separated by more than 500 km may be explained by the high number of Antarctic species with circumpolar distribution, the similarity of conditions in the sea around the continent and the circumantarctic current systems (Arntz et al., 1994), although the traditional view that many SO benthic invertebrate taxa have a circumpolar distribution has come recently under scrutiny (Harder et al., 2016;Hauquier et al., 2017;Janosik et al., 2011;Wilson et al., 2007). Previous studies in the re-  Gutt et al., 2017), the so-called 'Animal Forests' (AF; Rossi, 2013). The complexity of the AF depends on the ecological structuring organisms. Indeed, patches of sponges can be highly three-dimensionally complex and diverse, influencing hydrodynamics and supplying shelter and food to very diverse associated fauna (Rossi et al., 2017). A high geographical turnover of microbenthic assemblages within larger regions can generally be explained mainly by sea-ice conditions and proxies for food supply, such as current and pigments in the sediment (Gutt, 2007). All communities studied here present several MOTUs representing these habitat builders. Moreover, these species are represented with high relative read abundances in our dataset, probably due to the high relative biomass in the samples (sponges, bryozoans, macroalgae, annelids and echinoderms). We detected high diversity of some specific groups, mostly in accordance with general Antarctic diversity patterns, where pycnogonids, polychaetes, ascidians, amphipods and bryozoans are particularly species-rich around Antarctica (Clarke & Johnston, 2003).
The most diverse taxa in our samples were Maxillopods (mainly copepods) with 108 MOTUs, followed by diatoms (99 MOTUs The composition of these structuring organisms (the trees of the marine animal forests) conditioned the associated fauna living on them. Furthermore, due to the high patchiness of Antarctic benthic communities, especially in shallow waters, the randomization of the sampling and the small surface sampled (625 cm 2 ) might greatly condition the obtained results. Therefore, we can assume that the communities obtained here are only representative of the sampled area at a local scale (few metres) but not at the regional scale (hundreds of metres). There were significant differences between stations, just a few metres apart, and variation at the scale of <5 m (replicates) was also high. The distribution of species in Antarctic shallow waters is inevitably highly heterogeneous due to the physical pressure of ice disturbance . The high heterogeneity found inside Deception Island, a place highly protected from ice-scouring, gives more weight to other physical factors such as current flows or food availability, or even to biological factors (niche-driven).

| Temporal patterns
Compared with traditional surveys, eDNA has increased the detection sensitivity of organisms (Bohmann et al., 2014 Our results show that benthic assemblages changed between 2016 and 2018 (temporal beta diversity), and these changes were linked to species loss rather than to species replacement. Due to the high degree of patchiness in the Antarctic hard-bottom communities, we cannot assert that the biodiversity is lower at the Fildes station (Table S3) or that biodiversity is accumulating at the Cormoran Town station. Although it may be a sampling artefact, a convergence between the most diverse and the least diverse stations was observed throughout the years. We detected a significant turnover in Cormoran Town (MOTU replacement; Figure S6) but an overall nestedness (MOTU loss) for Deception Island. Due to the oceanographic conditions, we should expect a higher biodiversity of sessile suspension feeders (habitat builders providing architectural complexity) in Fildes, which would allow a rich and diversified community of smaller organisms to settle. At the same time, the high rates of sedimentation inside Deception Island (Baldwin & Smith, 2003), can be the reason behind the less diverse community in Cormoran Town. The main abiotic factors affecting the benthic communities are common for both stations (ice disturbance, depth, sediment type). The main difference between stations is the currents (food supplied, sedimentation) since Fildes Point is located just in the entrance channel while Cormoran Town is inside the protected Whaler Bay. The highest values of alpha diversity of Fildes Point confirm the presence of an important biodiversity hotspot, probably due to the absence of physical disturbance such as ice-scouring and anchor ice at the entrance of Deception Island, which allows the development of these benthic assemblages, despite the background of recent volcanic eruptions (Angulo-Preckler et al., 2018. Future studies should focus on verifying if this decreasing trend in biodiversity is maintained and try to unmask the possible factors behind this process.

| CON CLUS ION
Recent advances in molecular and sequencing methodologies enabled us to evaluate biodiversity levels from even the most remote habitats. Measuring species and biological diversity of the Antarctic shelf is notoriously difficult as a result of high community patchiness and complex hierarchical scales of spatial variation. Hard-bottom assemblages exhibit high spatial variability and heterogeneity, not related to depth, which is challenging to design large-scale sampling programmes to assess the state of the benthos in the Southern Ocean. There are many unknowns as to how communities will respond to climate change, and studies like this contribute to understanding the present spatial variability in Antarctic shallow hard-bottom communities and serve as a baseline for future comparisons. The identification of a remarkable core community shared within all the localities, regardless of the large distance between some stations, reinforces the traditional view of circumpolar distribution in several invertebrate taxa. As the accuracy of metabarcoding taxonomic assignment relies upon the completeness of public sequence databases, continued efforts in the improvement of these databases are essential to provide a holistic description of Antarctic marine ecosystems. An effort to increase the coverage of the sampled areas is necessary to achieve sufficient representability.

ACK N O WLE D G E M ENTS
We thank all members of the DISTANTCOM and BLUEBIO research projects for their support. We especially thank the crews of the scientific vessels BIO Hesperides and Sarmiento de Gamboa, as well as the staff of the Spanish Antarctic Bases for their logistic support during the diving operations. We also want to thank Dr. Xavier Turon and Dr. Adria Antich for their help in processing the samples.

CA-P was funded by the Ramon Areces Foundation. UiT the Arctic
University of Norway is thanked for financial support to KP. This is an AntICON (SCAR) contribution. The publication charges for this article have been funded by a grant from the publication fund of UiT The Arctic University of Norway.

CO N FLI C T O F I NTER E S T S TATEM ENT
The research reported here has been conducted in an ethical and responsible manner and complies with all relevant legislation. The authors declare there is no potential conflict of interest.

PEER R E V I E W
The peer review history for this article is available at: https:// www.webof scien ce.com/api/gatew ay/wos/peer-revie w/10.1111/ ddi.13703.

DATA AVA I L A B I L I T Y S TAT E M E N T
The raw sequencing data of this study have been deposited in the NCBI Short Read Archive (SRA). Project number: PRJNA914641.