Natural chemical control of marine associated microbial communities by sessile Antarctic invertebrates

Organisms living in the sea are exposed to fouling by other organisms. Many benthic marine invertebrates, including sponges and bryozoans, contain natural products with antimicrobial properties, since microbes usually constitute the first stages of fouling. Extracts from 4 Antarctic sponges (Myxilla (Myxilla) mollis, Mycale tylotornota, Rossella nuda, and Anoxycalyx (Scolymastra) joubini) and 2 bryozoan species (Cornucopina pectogemma and Nematoflustra flagellata) were tested separately for antifouling properties in field experiments. The different crude extracts from these invertebrates were incorporated into a substratum gel at natural concentrations for an ecological approach. Treatments were tested by submerging plates covered by these substratum gels under water in situ during 1 lunar cycle (28 d) at Deception Island (South Shetland Islands, Antarctica). Remarkably, the butanolic extracts of M. tylotornota and C. pectogemma showed complete growth inhibition of microscopic eukaryotic organisms, one of the succession stages involved in biofouling. Our results suggest that different chemical strategies may exist to avoid fouling, although the role of chemical defenses is often species-specific. Thus, the high specificity of the microbial community attached to the coated plates seems to be modulated by the chemical cues of the crude extracts of the invertebrates tested.


INTRODUCTION
Submerged substrates are potentially exposed to a myriad of fouling organisms, and thus, competition is very intense and includes many chemically mediated interactions (Buss 1990, Steinberg & De Nys 2002. The colonization of a substratum (biofouling) in the sea is a highly dynamic and complex process involving (1) adsorption on the new surface of dissolved organic molecules, forming a conditioning film, (2) colonization by bacteria with specific cell−surface, cell− cell, and interpopulation interactions shaping the structure, composition, and functions of surface-associated microbial communities (Dang & Lovell 2016), (3) colonization by microscopic, unicellular eukar yotes, such as diatoms, fungi, and heterotrophic eukaryotes, and (4) settlement and growth of multicellular eukaryonts, such as invertebrate larvae and algal spores (Wahl 1997, Maki & Mitchell 2002, Lema et al. 2019. Several physicochemical properties (e.g. surface hydrophobicity, wetability, and/or surface molecular topography) may determine the adhesion of different bacterial species and microbial community assembly in the biofilms (Wiencek & Fletcher 1997). Bacterial adhesion on submerged surfaces is a highly complex process, not only controlled by surface properties of the substrate, but also by surface properties of the bacterium itself (Harder & Yee 2009).
Being sessile organisms, sponges and bryozoans rely mainly on bioactive compounds for defense against predators, competition for space, and overgrowth by fouling (Proksch et al. 2002). Moreover, their associated microorganisms (symbionts) may also be involved in chemical protection against fouling (Dobretsov et al. 2005, Ortlepp et al. 2008. Most invertebrates may remain relatively free from macrofouling while they often present some degree of microfouling (Richmond & Seed 1991, Dobretsov et al. 2006.
From an ecological perspective, there are only a few reports which have analyzed the inhibition activity of organism extracts under field conditions. Only some of these studies were able to measure fouling settlement (Henrikson & Pawlik 1995, 1998, Angulo-Preckler et al. 2015, Dobretsov & Rittschof 2020. For Antarctic marine benthic organisms, very few studies have evaluated invertebrate antifouling defenses (Slattery et al. 1995, Angulo-Preckler et al. 2015, Patiño Cano et al. 2018). As mentioned above, potential colonization and overgrowth could be a potent selective pressure on marine benthic organisms, favoring the development of chemical defenses against fouling. Thus, it is ecologically relevant to perform in situ experiments to establish the activity of the organisms' extracts against environmental micro orga nisms under real environmental conditions. A field assay method for testing the antifouling activity of crude organic extracts of marine organisms was developed by Henrikson & Pawlik (1995). Accordingly, the main advantages of this methodology are (1) the extract is incorporated into the medium simulating natural situations, where the products are located within the organism, (2) antifouling substances are liberated slowly, as presumably occurs in living organisms, and (3) the physical characteristics of the settlement surface remain unchanged (Henrikson & Pawlik 1995).
This study aims to evaluate the potential antimicrofouling activity of sessile marine Antarctic invertebrates by comparing their crude extracts under real conditions. We hypothesize that these benthic invertebrates use chemical defense to regulate surfaceassociated microorganism colonization as an efficient way to control fouling pressure. Therefore, we used extracts (lipophilic and less hydrophobic extracts) from these invertebrates incorporated into artificial substrata that were submerged under in situ conditions. The biofilms (surface-associated communities de veloping on these substrata) were analyzed in terms of species composition and relative abundances (bacteria and eukaryotes) and used to infer the antifouling activity of each invertebrate species tested.

Sample collection and processing
Sessile Antarctic invertebrate fauna from 2 common phyla (Porifera and Bryozoa) were selected to evaluate their potential antifouling activity. Four common species of sponges, 2 demosponges (Myxilla (Myxilla) mollis Ridley & Dendy, 1886, Mycale tylo tor nota Koltun, 1964), and 2 hexactinellids (Ros sella nuda Topsent, 1901, Anoxycalyx (Scolymastra) joubini (Topsent, 1916)), together with 2 abundant bryo zoan species (Cornucopina pectogemma (Goldstein, 1882), and Nematoflustra flagellata (Waters, 1904)), were selected for the experiment (Table 1). The sponges were collected in the Eastern Weddell Sea during the ANT/XXI-2 cruise of R/V 'Polarstern' (Alfred-Wegener-Institut), during the austral summer of 2003/2004, through bottom and Agassiz trawls. Bryozoans were collected by SCUBA diving in the vicinity of Livingston Island (South Shetland Islands) during the austral summer of 2012. A portion of each sample was conserved for further taxonomical identification at the University of Barcelona (UB). The re maining material was frozen at −20°C until it was needed for the experiments.
Each organism was disaggregated into small pieces and ground with a mortar and a pestle homogenizing it in acetone to collect the crude extracts. Then, crude extracts were fractionated by polarity, separating the most polar compounds by extraction in butanol (BuOH) from the less polar lipophilic compounds by extraction in diethyl ether (Et 2 O). The extraction procedure has been extensively described in previous works of our team (e.g. Avila et al. 2000). Natural concentrations of each extract compounds (Et 2 O or BuOH) were calculated as the total dry weight (DW T ) of each sample (DW T = dry weight of the solid residue + dry weight of the aqueous residue + dry weight of the Et 2 O extract + dry weight of the BuOH ex tract; see Table 1). Fractionating by polarity is important to determine which type of compounds are responsible for any activity.

Experimental design
Gels were prepared by dissolving 1.57 g of Phyta -gel™ (Sigma Chemical) per 100 ml distilled water and stirring for 10 s. After heating until boiling, the gel was allowed to cool down before an aliquot of extract dissolved in 3 ml solvent (ether or methanol) was added and shaken to obtain the treatments. Each treatment combined 1 species and 1 type of extract. The amount of tissue extracted was equivalent to the amount of gel prepared, so that each experimental dish would have a natural concentration of metabolites, reflecting that in the extracted organism (Angulo-Preckler et al. 2015).
Three replicates of each extract treatment were prepared, as well as 3 gel controls for both extracts that contained only 3 ml of diethyl ether (control for Et 2 O extracts) or methanol (control for BuOH extracts), respectively. Extracts were diluted in Phyta -gel™ and poured into Petri dishes, and the gel was then allowed to completely solidify. The solvents were fully evaporated before the assays were performed. The Petri dishes were placed on 3 acrylic plates and covered with a metallic grid to prevent removal by predators while under deployment (Angulo-Preckler et al. 2015). The plates with their substratum gels were placed in Whalers Bay (62.99°S, 60.56°W), Deception Island (Antarctica) at about 20 m depth and were maintained underwater for a full lunar cycle in January 2013. Two small buoys were attached to each plate to avoid burial of the structures by sedimentation and to keep the plates perpendicular to the water flow (see Fig. S1 in the Supplement at www. int-res. com/ articles/ suppl/ a085 p197_ supp. pdf). Once removed from the water, the coatings of the plates with their attached microbial communities (Phytagel™ discs) were immediately frozen for further genetic analysis. The seawater temperature ranged from 0 to 2.0°C during January 2013.

DNA extraction and PCR amplification
In the home laboratory, the Phytagel™ discs were thawed (3 replicates of each extract treatment) and subsequently swabbed with sterile cotton buds for the collection of surface bacteria following our previously described protocol (Angulo-Preckler et al. 2015). The genomic DNA from each individual cotton bud was extracted using the MO BIO PowerWater DNA Isolation Kit (MO BIO Laboratories). Extraction procedures were identical for all samples. DNA concentrations were determined using a Nanodrop 2000p (Thermo Scientific™).
The enzyme Taq DNA polymerase and AccuPrime specific primers were used to amplify the 16S (Bacteria/Archaea) and 18S rRNA (eukaryotes) genes ( at 46°C for 30 s, and extension at 72°C for 35 s, preceded by 5 min denaturation at 94°C and followed by 7 min extension at 72°C. PCR products were purified using the QIA quick PCR purification kit (Qiagen), ob taining pure DNA. The quality of DNA was assessed by 1% agarose gels. To control for false-positive PCR signals, 1 l of MilliQ water was frozen, thawed, and subjected to the same DNA extraction procedure. The concentration of the samples was adjusted to between 10 and 25 ng μl −1 . PCRs were carried out using the thermo cyclers GeneAmp PCR System 9700 (Applied Biosystems) and Ptc 200 Peltier Thermal Cycler (MJ Research).

Cloning, sequencing and phylogenetic analysis
PCR-amplified DNA fragments were cloned using the TOPO ® TA Cloning ® Kit (Invitrogen). The ligation product was introduced into competent cells of E. coli (Strain Machi1-T1) for transformation by heat shock. The cells were inoculated on LB agar with ampicillin (100 μg ml −1 ) and X-Gal (50 μg ml −1 ) and plates were incubated at 37°C for 24 h. White colonies were selected, inoculated on plates containing TB (terrific broth) and ampicillin and incubated at 37°C for 18 h. The pellets obtained were used to extract the plasmid DNA with the fluid handling ep -Motion 5075 Vac robot (Eppendorf AG). Samples were then sequenced using a 48 capillary sequencer ABI 3730 XL (Applied Biosystems). Read-lengths of up to approximately 1000 bp were achieved. A total of 900 bacterial and 900 eukaryotic clones were sequenced. Sequences were analyzed with UCHIME, to identify and remove chimeric reads, and classified to eliminate those that could be considered contaminants (Edgar et al. 2011). Operational taxonomic units (OTUs) were identified using BLAST at the NCBI database (http://ncbi.nlm. nih. gov/ BLAST). Representative sequences were aligned using Clustal X 2.0 (Larkin et al. 2007).
Sequences obtained in this study were deposited in the National Center for Biotechnology Information (NCBI) sequence database under the accession numbers KX214587−KX214606 for Bacteria, and KX232671−KX232675 for eukaryotes. No sequences were obtained for Archaea.
Phylogenetic trees were obtained using MEGA version X (Kumar et al. 2018) with parsimony, neighborjoining, and maximum likelihood analyses. In all cases, general tree topology and clusters were stable, and reliability of the tree topologies was confirmed by bootstrap analysis using 1000 replicate alignments. Analytic Rarefaction 1.3 software (https:// strata. uga. edu/ software) was used to calculate rarefaction curves ( Fig. S2 in the Supplement). It re vealed that rarefaction curves reached saturation at 3% sequence divergence, indicating that the samples contained almost all the diversity at this genetic distance.

Statistical analysis
The mean total number of clones (bacteria and eukaryotes) was determined from the 3 replicates of each treatment. The data were 4 root-transformed to achieve normality and homoscedasticity (Shapiro & Wilk 1965). A 2-way factorial ANOVA block design (treatment [T] as a fixed effect factor and plate [P] as a random effect factor) was run to test for global significant differences between treatments (control and extracts were analyzed separately by polarity). No significant interactions for T × P were found and thus the data were reanalyzed running a 1-way ANOVA for each extract (Et 2 O and BuOH). Dunnett's post hoc tests were performed to determine which treatments showed differences from the controls.
Data from genetic analysis were used to build a matrix composed for clones grouped by phylotypes (OTUs). A dissimilarity matrix between samples was calculated using the Bray-Curtis distance after relative abundance data had been 4 square root transformed. A 3-way permutational multivariate analysis of variance (PERMANOVA; Anderson 2005) was used to test for any significant differences within and be tween factors. The PERMANOVA (unrestricted permutation of raw data method, using Monte Carlo test for testing pairwise differences between treatments) was run on a Bray-Curtis dissimilarity matrix with the PRIMER 6 + PERMANOVA software package (Plymouth Marine Laboratory). Effect sizes were calculated using the partial omega squared index (ω 2 p, Olejnik & Algina 2003). A non-metric multi dimensional scaling (NMDS) was also used to represent the results (Kruskal 1964). Furthermore, those data were used to calculate biodiversity indices for each treatment (number of taxa, OTU richness S, relative abundance N, and Shannon diversity index H'). The microbial community was evaluated alto gether, as well as bacteria and eukaryotes separately (see Table 2).

RESULTS
The number of Bacteria and Eukarya clones retrieved from the plates after 28 d of in situ incubation are shown for the different treatments (species extracted and control), prepared with their lipophilic diethyl ether and less hydrophobic butanol ex tracts (Fig. 1). On our plates, the number of unique OTUs was high. Furthermore, no diatoms were de tected on our plates using 18S cloning and sequencing.

Number of clones
Differences in the number of clones between experimental treatments and controls, determined by 1-way ANOVA with treatment as a fixed factor (Fig. 1), indicated that 3 out of 4 tests performed showed significant differences, i.e. bacteria on Et 2 O extracts (p = 0.00005), eukaryotes on Et 2 O extracts (p = 0.00005), and eukaryotes on BuOH extracts; (p = 0.0153), while bacteria on BuOH extracts (p = 0.9825) showed no differences. Differences be tween the numbers of clones in the controls were also observed.
The BuOH extracts of the bryozoan Cornucopina pectogemma and the sponge Mycale tylotornota completely inhibited the growth of eukaryotic clones. On the other hand, 2 extracts significantly increased the number of clones settled on the gels with respect to the pertinent control treatments. Both extracts belong to hexactinellid sponges, but while the Et 2 O extract of Rosella nuda showed an increase in the abundance of bacterial clones, the BuOH extract of Anoxycalyx (Scolymastra) joubini showed an increase in the settlement of eukaryotic clones. In contrast with the antifouling activity showed by several extracts, the Et 2 O extract of C. pectogemma and the BuOH extract of Nematoflustra flagellata and Myxilla mollis favored the settlement of eukaryotic clones. The data showed significant differences in the Et 2 O extracts with an opposite trend: a large decrease in the abundance of eukaryotic clones together with the highest abundance of bacterial clones.
As many as 22 OTUs were found in only 1 treatment (in 1, 2, or 3 replicates but restricted to the same treatment). The highest richness and diversity of microorganisms was found for both extracts of the sponge Mycale tylotornota, while the lowest diversity and richness were found on both extracts of the bryozoan Cornucopina pectogemma (see Table 2). Surprisingly, the highest values of bacterial diversity was found on BuOH extracts of M. tylotornota, which were also found to have the lowest richness and diversity for eukaryotic clones (Fig. 2). In general, a wide variability was observed within the different ex tracts tested, in both bacteria and eukaryotic communities. The differences in the microbial community composition showed contrasting patterns depending on the extract polarity. PERMANOVA ana lysis showed significant differences in the Species (Sp) factor and in the interaction between Species and Extract (Sp×Fr). Both showed the largest effect size (see Table 3), while the factors Plate and Extract showed no significant differences (p > 0.05). PERMANOVA results show that differences among treatments were the largest component of variability, with effect sizes (ω 2 p) of 0.78 and 0.75 for the Et 2 O and BuOH extracts, respectively (Table 4). The global microbial community (bacteria and eukaryotes) was completely different in all Et 2 O extracts, while only Myxilla mollis was significantly different in the BuOH extracts (Table 5).

Chemical control
Our results suggest that the Antarctic benthic invertebrates tested here could chemically control their associated microbial communities. Microbial settlement depends on the composition of the bacterial biofilms, and the production or absence of certain proteolytic enzymes (Qian et al. 2007, Almeida & Vasconcelos 2015. The high specificity of the microbial community attached to our coated plates seems to be modulated by the chemical cues of the extracts, with species identity being much more important   Table 3, Fig. 3). Thus, the colonization of invertebrate surfaces may be induced and/or inhibited by the natural products present in the surface tissues or excreted by them. Our 18S rRNA gene sequencing, however, failed to detect diatoms, de spite their common abundance in Antarctic waters. In a previous study, Toupoint et al. (2012) also failed to detect diatoms. This may be an artefact due to the primers used; diatoms are not commonly re trieved from environmental clone libraries, ex cept when they occur in high abundance (Potvin & Lovejoy 2009, Briand et al. 2018). Alternatively, diatoms may really not have been present after 28 d, which could be too short a period for their settlement and growth in the biofilms. Al though 1 mo has been considered to be long enough to achieve a relative stability and maturation in multispecies biofilms (Dang et al. 2008), this may take longer in Antarctic waters. It is important to recognize, however, that the metho do logy used here only reflects the number of different 16S rRNA and 18S rRNA genes retrieved from a sample, which may not reflect the numbers of different organisms originally in the sample. Any assay evaluating this has constraints when trying to reflect natural conditions (Angulo-Preckler et al. 2015). Furthermore, it is also necessary that the putative active chemicals are present in large enough concentrations to have a significant biological ef fect, in order to demonstrate that naturally produced chemicals mediate in a given biological interaction. Al though both solvents were completely evap- orated, they al ways leave a residue, which may affect micro organisms or may modify the gels in the plates.

Selective antimicrobial activity
We have proved here that selective antimicrobial activity with differential bacterial and eukaryotic attachment occurs, even when all the microorganisms in our experiment came from the same water column, with the same environmental factors and physicochemical properties of the initial surface (maturation biofilm). Not all the extracts showed antifouling activity, but all of them resulted in different microbial communities, with a high similarity within replicates. Microbial inhibition must, therefore, be a more selective process than just a reduction in surface biofilms. Specifically, when a bacterium grows on top of another, this may either provide positive settlement cues for innocuous larvae or negative cues for potential competitors (Walls et al. 1993).
In general terms, the more polar extracts in our experiments favored a higher number of clones per OTU than the non-polar extracts, which showed a higher antimicrobial activity. This trend can be clearly observed in the Colwellia clade, which was always more abundant in the more polar extract for most species tested. Although a total inhibition of the microbial community was never found, a shift in the number of clones per OTU and/or composition of its communities always occurred. This highlights the importance of the diversity rather than the abundance of microorganisms in the formation of biofilms. Moreover, a large variability in the effect of the extracts tested on both bacteria and eukaryotic communities was detected, including the total inhibition of eukaryotic clones by the extracts from the bryo zoan Cornucopina pectogemma and the sponge My cale tylotornota. The interference in eukaryotic communities is key for avoiding macrofouling adhesion (Almeida & Vasconcelos 2015  are usually quite species-specific in their ecological roles, even in sponges and bryozoans from polar waters (Avila et al. 2008, Angulo-Preckler et al. 2015, Núñez-Pons & Avila 2015, Figuerola et al. 2017, suggesting that different chemical strategies may exist to deal with repellence, allelopathy, and fouling. Further studies should chemically analyze the compounds directly responsible for these activities.

Natural products from the studied species
The natural products from the species studied herein are mostly unknown. Only a taurine-conjugated anthranilic acid, glassponsine, has been found in the BuOH extract of the hexactinellid Anoxycalyx (Scolymastra) joubini (Carbone et al. 2014; http:// pubs. rsc.org/marinlit/). A moderate anti fungal activity for the crude extract of A. joubini has been re ported (Berne et al. 2016), but we only found a significant bioactivity in its BuOH extract. Actually, it seems to stimulate the attachment of eukaryotic organisms, but this could be due to the low abundances of organisms settled on the controls for the BuOH extracts, composed solely of the basidio mycota Cryptococcus sp. Some Rossella spp., in cluding R. nuda, display selective cytotoxicity against human tumor cell lines and the early development of sea ur chin embryos, with undescribed metabolites (Tabo ada & Avila 2010, Figuerola et al. 2013b, Turk et al. 2013, Berne et al. 2016. Moreover, anti-predatory activity in the extracts of R. nuda and R. fibulata Schulze & Kirkpatrick, 1910 has also been reported (McClintock et al. 1993, 2000, Núñez-Pons et al. 2012, Taboada et al. 2013, Núñez-Pons & Avila 2014. The observation that one extract prevents eukaryotic attachment while an other extract from the same species promotes bacterial attachment confirms that the composition of the microbial community within the biofilm is more important than the number of attached microorganisms. Interestingly, 2 species whose extracts significantly promoted an increase in the abundance of attached organisms belong to the hexactinellid sponges, and the different bioactivities of the ex tracts may perhaps be associated with speciesspecific life history traits. Mycale sponges present several cytotoxic compounds (e.g. pateamine, peloruside, mycalamide;Hood et al. 2001, Singh et al. 2011. M. tylotornota Kol tun, 1964, a barely studied, rare sponge only found 5 times before in the surroundings of the South Shetland Islands, represents here the first record for the Weddell Sea (see www.gbif.org/). M. tylotornota showed the highest antifouling activity, with very low levels of eukaryotic clones on the lipophilic ex tract, and a complete absence of eukaryotic clones on the more polar extract. This more polar extract, together with that of C. pectogemma, were the only ones able to inhibit the growth of the most abundant basidiomycota, Cryptococcus sp. Also, some Antarctic Myxilla species have shown antibacterial activity (Angulo-Preckler et al. 2018). Lipophilic extracts of M. (Myxilla) mollis inhibited growth of the green algae Pseudo kirchneriella subcapitata and were active against Staphylo coccus aureus (a human methicillin-resistant strain) (Berne et al. 2016). Sacristán-Soriano et al. (2017) also studied the potential antibacterial activity of a wide panel of Antarctic invertebrates in the laboratory. Four species (3 sponges and 1 bryozoan species) were also included in our study. In their study, no antibacterial activity was found for the sponges R. nuda and A. joubini, while the sponge M. mollis and the bryozoan Nema to flustra flagellata showed activity against Antarctic bacteria (Bacillus aquimaris and Paracoccus sp., respectively). Our data show that the lipo philic extract of M. mollis was active in decreasing the number of eukaryotic as well as bacterial clones.
Crude extracts of the bryozoans C. pectogemma and N. flagellata have also been shown to inhibit the QS indicator strains Chromobacterium violaceum CV026 and C. violaceum VIR07 in the laboratory (Figue rola et al. 2017). Antibacterial activity was de tected in the lipophilic extract, while the more polar one did not show any antimicrobial activity. In addition, both bryozoans display many other chemical defensive strategies (e.g. repellence against generalist macroinvertebrate predators; Figuerola et al. 2013aFiguerola et al. , 2017, suggesting their natural products are used for a wide array of ecological roles. Some cold-water bryozoan species possess inhibitors of QS-regulated gene expression found in diverse marine bacterial strains: e.g. Flustra foliacea harbors alkaloids with antimicrobial activity (Lippert & Iken 2003, Peters et al. 2003. In bryozoans, many alkaloids and polyketides have been found to be responsible for different ecological defensive activities, although only 1 Antarctic bryozoan species has been chemically studied so far (Lebar et al. 2007, Sharp et al. 2007). However, bryozoans show similar antifoulant activity to sponges, being a promising source of pharmacologically interesting compounds (Figuerola & Avila 2019).

Microbial communities attached to the coated plates
Symbiotic bacteria producing bioactive compounds have been obtained from a wide variety of marine orga nisms, such as sponges, corals, mollusks, crustaceans, bryozoans, and ascidians in different geo graphical areas (see Piel 2009 and references therein). However, few studies on symbiotic bacteria in Antarctic marine invertebrates have so far been done (Giudice et al. 2019, Sacristán-Soriano et al. 2020. Some microorganisms identified here are especially resistant to the antifouling effect produced by invertebrate extracts. Here, the most resistant bacterium was Colwellia. Colwellia contains proteins, such as the chaperones DnaK and DnaJ, that allow them to adapt to different environments (Yamauchi et al. 2004). In C. maris, the DnaK gene is essential for growth and viability under diverse environmental conditions (García-Descalzo et al. 2011). Also, some of the bacteria identified here (i.e. Colwellia, Pseudoaltero monas) are able to produce lipases (Urbanek et al. 2018), enzymes that hydrolyze ester bonds in lipids, i.e. enable the bacteria to feed on lipids. Lipases may neutralize or decrease the antifouling effect of the bioactive compounds from invertebrate extracts (Prabhawathi et al. 2014). An other possible mecha nism for antifouling inhibition could be marine invertebrates controlling the secondary metabolism and colonization behaviors of the microorganisms they host. As a result, the microorganisms modify their effect on their host. For ex ample, it has been reported that Pseudo altero monas inhibit antibiotic and pigment production by the corals they inhabit (Dobretsov et al. 2013). This has not been ob served in Antarctica so far.
Some members of the Roseobacter group, such as Loktanella and Roseobacter sp., have also been described as primary colonizers of eukaryotic hosts (Michael et al. 2016). This could explain the high number of close relatives within the Gamma-and Alphaproteobacteria, both well-known psychrophiles from Antarctica, in most treatments (Figs. S3 & S4 in the Supplement). Five different clades within Rhodobacteracea were found in almost all the treatments (except on Rosella nuda extracts), al though Roseo bacter was only found on M. mollis ex tracts, probably reflecting the shift from these early primary colonizers to the secondary structure community.
Eukaryotic microorganisms may also prevent the antifouling effect of marine invertebrates. Here, we identified mainly 2 types of fungi: Cryptococcus and Mrakia. The fungus Cryptococcus has a polysaccha-ride capsule, composed mainly of glucurunoxylomannan, and can form biofilms (Martinez & Casadevall 2007). In the laboratory, forming a biofilm makes them less susceptible to environmental stresses than their planktonic counterparts (Martinez & Casadevall 2007). The predominance of Cryptococcus in cold waters could derive from their ability to produce polysaccharides and utilize available nutrients in oligotrophic systems (Margesin & Miteva 2011).

CONCLUSION
We cannot conclude that the invertebrates tested here are directly responsible for the observed activity, because invertebrate-associated bacteria may also play a role, and it is indeed beneficial for the host to harbor epi biotic bacteria with antifouling properties. However, it is plausible to assume that the host chemically modulates the associated microbial community to gain benefits from such interactions, although the origin of the activity remains unclear. It should also be taken into account that, in most previous studies, 'bioactivities' were not estimated at natural concentrations or under ecological conditions. Although certain species can allocate biofouling defenses in certain specific tissues (Cronin & Hay 1996, Furrow et al. 2003, Angulo-Preckler et al. 2015, we assumed a homogeneous distribution throughout the organism for the purpose of the present the study. This in fact means that the antifouling capacity of the invertebrates may have been underestimated and could be more effective when the antifouling compounds occur concentrated in surface tissues. Even if the responsible metabolites have not yet been described, our results suggest that the bryozoan Cornucopina pectogemma and the sponge Mycale tylotornota are promising potential new sources for antifouling compounds, being able to disrupt colonization of a substrate by microscopic eukaryotes. Further studies should be devoted to fully developing the bioactive potential of these species. Furthermore, genetic analyses were supported by grants CTM2010-12134-E/ANT and CTM2011-16003-E, with E.G.L. as recipient of a PTA fellowship (PTA2016-12325-I). Also, we thank the anonymous reviewers and the editor for providing very helpful comments on the manuscript. This is an AntECO (SCAR) contribution.