Fungal Diversity in Intertidal Mudflats and Abandoned Solar Salterns as a Source for Biological Resources

Intertidal zones are unique environments that are known to be ecological hot spots. In this study, sediments were collected from mudflats and decommissioned salterns on three islands in the Yellow Sea of South Korea. The diversity analysis targeted both isolates and unculturable fungi via Illumina sequencing, and the natural recovery of the abandoned salterns was assessed. The phylogeny and bioactivities of the fungal isolates were investigated. The community analysis showed that the abandoned saltern in Yongyudo has not recovered to a mudflat, while the other salterns have almost recovered. The results suggested that a period of more than 35 years may be required to return abandoned salterns to mudflats via natural restoration. Gigasporales sp. and Umbelopsis sp. were selected as the indicators of mudflats. Among the 53 isolates, 18 appeared to be candidate novel species, and 28 exhibited bioactivity. Phoma sp., Cladosporium sphaerospermum, Penicillium sp. and Pseudeurotium bakeri, and Aspergillus urmiensis showed antioxidant, tyrosinase inhibition, antifungal, and quorum-sensing inhibition activities, respectively, which has not been reported previously. This study provides reliable fungal diversity information for mudflats and abandoned salterns and shows that they are highly valuable for bioprospecting not only for novel microorganisms but also for novel bioactive compounds.


Introduction
As interest in marine living resources worldwide has increased, research on marine fungi has progressed considerably over the past two decades, including the discovery of new species and novel natural compounds [1]. It is still insufficient compared to the research on terrestrial fungi, but the study of marine fungi has been extended to intertidal zones such as mangrove forests and coastal wetlands [2,3]. In addition, there are other unique intertidal environments such as mudflats and abandoned salterns.
There are different types of tidal flats, and mudflats are differentiated from sandy tidal flats. Mudflats are a kind of coastal salt marsh made of clay deposited by waves and rivers and are a unique environment as they are exposed to the atmosphere twice a day depending on the tide. In the Yellow

Fungal Diversity and Recovery Assessment
The fungal communities and diversity in these intertidal environments were investigated by high-throughput sequencing of the sediment samples. Through the community analysis, the natural recovery of the decommissioned salterns was also assessed.
At the phylum level of the fungal community composition, Ascomycota and Basidiomycota, the two major phyla of the kingdom Fungi, were the dominant phyla ( Figure 2A). Meanwhile, the community composition of the abandoned saltern in Yongyudo was clearly different. It was highly dominated by Entorrhizomycota, while Chytridiomycota and Mortierellomycota dominated the other abandoned salterns and mudflats. Considering that the phylum Entorrhizomycota is composed of plant pathogenic fungi, the difference may be due to the introduction of their host plants at the initial stage of ecological succession after the saltern was abandoned. Meanwhile, a large number of fungi could not be identified, indicating a lack of DNA-based phylogenetic information on fungi in these intertidal environments. As the community compositions of the other salterns and mudflats were similar to each other except for that of the saltern in Yongyudo, a nonmetric multidimensional scaling (NMDS) plot was constructed using the database of fungal operational taxonomical units (OTUs) and their abundance to determine the distance between the fungal communities. According to the NMDS results, it appeared that the distance between the abandoned saltern and the mudflats in Yongyudo was much longer than the distance between the two environments on the other islands ( Figure 2B). This indicates that the abandoned salterns in Yongyudo that were abandoned less than a year ago were the least recovered to mudflat. Since the salterns of Yubudo and Gopado were abandoned 20 and 35 years ago, respectively, it was expected that natural recovery would have occurred. This was supported by the short distance between the communities of the two salterns and those of nearby mudflats on the NMDS plot. Thus, we clustered the samples by the expected recovery

Fungal Diversity and Recovery Assessment
The fungal communities and diversity in these intertidal environments were investigated by high-throughput sequencing of the sediment samples. Through the community analysis, the natural recovery of the decommissioned salterns was also assessed.
At the phylum level of the fungal community composition, Ascomycota and Basidiomycota, the two major phyla of the kingdom Fungi, were the dominant phyla ( Figure 2A). Meanwhile, the community composition of the abandoned saltern in Yongyudo was clearly different. It was highly dominated by Entorrhizomycota, while Chytridiomycota and Mortierellomycota dominated the other abandoned salterns and mudflats. Considering that the phylum Entorrhizomycota is composed of plant pathogenic fungi, the difference may be due to the introduction of their host plants at the initial stage of ecological succession after the saltern was abandoned. Meanwhile, a large number of fungi could not be identified, indicating a lack of DNA-based phylogenetic information on fungi in these intertidal environments. As the community compositions of the other salterns and mudflats were similar to each other except for that of the saltern in Yongyudo, a nonmetric multidimensional scaling (NMDS) plot was constructed using the database of fungal operational taxonomical units (OTUs) and their abundance to determine the distance between the fungal communities. According to the NMDS results, it appeared that the distance between the abandoned saltern and the mudflats in Yongyudo was much longer than the distance between the two environments on the other islands ( Figure 2B). This indicates that the abandoned salterns in Yongyudo that were abandoned less than a year ago were the least recovered to mudflat. Since the salterns of Yubudo and Gopado were abandoned 20 and 35 years ago, respectively, it was expected that natural recovery would have occurred. This was supported by the short distance between the communities of the two salterns and those of nearby mudflats on the NMDS plot. Thus, we clustered the samples by the expected recovery status, consisting of recovered saltern (RS), nonrecovered saltern (NS), and mudflat. In fact, the NMDS plot showed that the fungal communities of the two environment types in Yubudo and Gopado were almost similar but were not completely clustered, especially in Gopado. This indicated that the fungal communities of the abandoned salterns were not fully recovered even after 35 years. Similarly, Bernhard and his colleague reported that salt marshes that were impounded and subsequently restored took more than 30 years to recover and become similar to the neighboring tidal flats [22]. They suggested that the regular inflow of tidal water increases the stability of the ecosystem. The abandoned salterns require a great deal of time to return to their original state due to the disturbance of human activity blocking the tidal water for a long time, resulting in a significant decrease in stability of microbial communities. Thus, with natural restoration alone, an artificially isolated environment requires an extremely long time to construct a microbial ecosystem similar to that of the adjacent environment, and it may require more than 35 years in the case of the abandoned salterns. status, consisting of recovered saltern (RS), nonrecovered saltern (NS), and mudflat. In fact, the NMDS plot showed that the fungal communities of the two environment types in Yubudo and Gopado were almost similar but were not completely clustered, especially in Gopado. This indicated that the fungal communities of the abandoned salterns were not fully recovered even after 35 years. Similarly, Bernhard and his colleague reported that salt marshes that were impounded and subsequently restored took more than 30 years to recover and become similar to the neighboring tidal flats [22]. They suggested that the regular inflow of tidal water increases the stability of the ecosystem. The abandoned salterns require a great deal of time to return to their original state due to the disturbance of human activity blocking the tidal water for a long time, resulting in a significant decrease in stability of microbial communities. Thus, with natural restoration alone, an artificially isolated environment requires an extremely long time to construct a microbial ecosystem similar to that of the adjacent environment, and it may require more than 35 years in the case of the abandoned salterns.  To verify the differences between the fungal communities, Shannon-Wiener and Gini-Simpson indices of each community were calculated. These two indices have been commonly used to calculate α-diversity, and the value of Shannon-Wiener and Gini-Simpson indices are more affected by species richness and evenness, respectively. The α-diversity indices of the NS were significantly lower in both the indices compared to those of the other environments, as expected (Table 1). This indicates that artificially blocking tidal inflow and conducting solar salt production reduce the species richness and evenness of the microbial community in the saltern, and the reduced diversity does not recover in a short period of time. We tested the significance of the difference by the clustered group using the Kruskal-Wallis rank sum test. There was no significant difference between the α-diversity indices of RSs and mudflats, while those of the NS were significantly lower than those of the others. The abandoned saltern in Yongyudo was an environment continuously interrupted by humans until recently. Thus, the ecosystem of the NS could be easily dominated by a small number of species that could survive or adapt under this harsh condition, resulting in significantly lower microbial diversity. In fact, the diversity indices of RSs and mudflats were high when compared to those of fungal communities in other environments, such as livestock manure, compost, and soil [23][24][25][26]. This implies that diverse fungi exist in these unique environments, thus making them valuable for bioprospecting.

Determination of Indicator Taxa
We tried to determine the major differences among the fungal communities. Various analytical methods were used to find indicator taxa that are responsible for the differences, including linear discriminant analysis effect size (LEfSe), indicator species analysis (ISA), RandomForest, and mvabund.
First, we visualized the result of the Kruskal-Wallis rank sum test with a LEfSe cladogram ( Figure 3 and Figure S1). It was obvious that the abundance of Ascomycota differs significantly between NS and the other environmental groups, tending to be lower in NS ( Figure S2). RS and mudflat both tended to be abundant in Chytridiomycota and Basidiomycota, but many of the taxa overlapped each other. This finding supports that there is no significant difference between RSs and mudflat, and the RSs were in the process of recovery to mudflats.
For the next step, we investigated the taxa specific to each group via ISA. ISA results in an "IndVal score" calculated by multiplying the relative abundance by the relative frequency of a taxon in each group. Microbial community data are known to be zero-rich and to have many rare species. In this case, OTUs with low mean abundance tend to have low mean variance; this can lead to statistical misinterpretations if some rare species are highly specific to a certain environment. Therefore, the results with low frequency have low fidelity, and taxa with frequencies less than 5 were screened out. As a result, several lineages of taxa were selected to be specific to each environment ( Table 2). Red, green, and blue nodes/shades indicate taxa that are significantly higher in relative abundance. The diameter of each node is proportional to the taxon's abundance. NS and RS mean non-recovered saltern and recovered saltern, respectively.
For the next step, we investigated the taxa specific to each group via ISA. ISA results in an "IndVal score" calculated by multiplying the relative abundance by the relative frequency of a taxon in each group. Microbial community data are known to be zero-rich and to have many rare species. In this case, OTUs with low mean abundance tend to have low mean variance; this can lead to statistical misinterpretations if some rare species are highly specific to a certain environment. Therefore, the results with low frequency have low fidelity, and taxa with frequencies less than 5 were screened out. As a result, several lineages of taxa were selected to be specific to each environment ( Table 2).  . Linear discriminant analysis effect size (LEfSe) cladogram of fungal composition in abandoned salterns and mudflats (logarithmic linear discriminant analysis (LDA) score > 2, p < 0.05). Red, green, and blue nodes/shades indicate taxa that are significantly higher in relative abundance. The diameter of each node is proportional to the taxon's abundance. NS and RS mean non-recovered saltern and recovered saltern, respectively. RandomForest analysis was carried out to determine whether there is a hierarchy-based relationship in the fungal community that can distinguish the sample groups. The results are shown as the mean decrease in accuracy and Gini impurity, which indicates the importance of the taxa in the generated decision tree. At the class to species level, two lineages of taxa, Gigasporales sp. and Umbelopsis sp., were repeatedly ranked at the top ( Figure 4). RandomForest analysis was carried out to determine whether there is a hierarchy-based relationship in the fungal community that can distinguish the sample groups. The results are shown as the mean decrease in accuracy and Gini impurity, which indicates the importance of the taxa in the generated decision tree. At the class to species level, two lineages of taxa, Gigasporales sp. and Umbelopsis sp., were repeatedly ranked at the top ( Figure 4). Finally, mvabund was used to determine which taxon abundances differed significantly by group. Recently, it was proved that model-based analysis of multivariate abundance data combined with negative binomial regression is excellent for detecting multivariate effects that otherwise would lead to statistical misinterpretation and is powerful when selecting indicator species using multivariate abundance data [27]. Several lineages of taxa, including the order Gigasporales and the family Umbelopsidaceae, appeared to differ significantly among the groups (Table 3).  Finally, mvabund was used to determine which taxon abundances differed significantly by group. Recently, it was proved that model-based analysis of multivariate abundance data combined with negative binomial regression is excellent for detecting multivariate effects that otherwise would lead to statistical misinterpretation and is powerful when selecting indicator species using multivariate abundance data [27]. Several lineages of taxa, including the order Gigasporales and the family Umbelopsidaceae, appeared to differ significantly among the groups (Table 3).
Various analyses were conducted to determine indicator taxa, but the results of these analyses were not identical. Therefore, the indicator species supported by multiple analyses were selected. Gigasporales sp. was supported by every analysis except for ISA, and Umbelopsis sp. was selected in every analysis except for LEfSe. Thus, these two fungal lineages can be regarded as definite indicator taxa for mudflats. In fact, the order Gigasporales belongs to the Glomeromycetes (Glomeromycota), the arbuscular mycorrhizal fungi. It was speculated that the abundance of Gigasporales sp. increases as it forms a symbiotic network with the indigenous halophytes of mudflats and as the abandoned saltern is restored to a mudflat [28]. Meanwhile, Entorrhizomycetes was selected by both ISA and mvabund as well as by a simple visual assessment of the fungal community composition ( Figure 2A). However, it was difficult to determine the indicator taxa for the NS because only one sampling site, the abandoned saltern in Yongyudo, represented the NS environment. In other words, this plant pathogenic fungus may be dominant there because of the existence of its unique host plants. Therefore, it can be regarded as an indicator taxon of the NS in Yongyudo but not of all the NSs in the Yellow Sea.

Diversity of Culturable Fungi
The community analysis implied that these intertidal environments are valuable for bioprospecting. Thus, the culturable fungi were isolated from the sediments to evaluate their diversity and to compare the results of metagenome analysis.
A total of 53 fungal strains was isolated from the sediments of the intertidal mudflats and the abandoned salterns (Table 4). It was obvious that the number of fungal isolates was much higher in Yongyudo than in the other two regions. It was suspected that this could be due to the sediment properties, such as the texture and grain size, since there was no significant difference in the environmental factors among sampling locations ( Figure S1). The genera Talaromyces and Trichoderma were mostly isolated from the mudflat samples, and most Penicillium spp. were found in the NS. Meanwhile, the diversity and the number of fungal isolates in the NS were much higher than those of the other salterns, which was completely opposite to the result of the metagenome analysis, where the diversity of the NS was the lowest ( Table 1, Table 4 and Table S1). In addition, all of the isolates except for Phanerochaete chrysosporium KUC10791 belong to Ascomycota. These results support that the actual diversity of environmental microorganisms must be investigated by eDNA-based metabarcoding approaches, unless all possible culture conditions and isolation methods are employed [29].
All 53 strains were grouped into 43 groups by morphological analysis and ITS sequences. The best-fit model of ITS sequences is a general time reversible (GTR) + proportion of invariable sites (I) + gamma distribution (G) by MrModeltest and contains 121 taxa and 844 nucleotide characters. In the phylogenetic analysis, 53 strains were classified into 2 phyla, 5 classes, 11 orders, 20 families, 22 genera, and 43 species, based on current taxonomic concepts. The dominant genera of the ITS tree were Penicillium (number of strains = 9), followed by Talaromyces (number of strains = 7), Aspergillus (number of strains = 6), Trichoderma (number of strains = 6), and Cladosporium (number of strains = 2) ( Figure 5). Talaromyces and Trichoderma were the most complex clades and had low resolution. Thus, to precisely perform phylogenetic analysis of Trichoderma and Talaromyces, EF1-α for Trichoderma species and benA for Talaromyces species were amplified, and phylogenetic trees were constructed ( Figure S3 and S4). The best-fit model of EF1-α for Trichoderma is Hasegawa-Kishino-Yano (HKY) + I + G and contains 41 taxa and 794 nucleotide characters. Multiple sequence alignments of two loci (ITS & benA) were analyzed for Talaromyces. Both loci were assigned GTR + I + G as the best fit model, and concatenated datasets contained 64 taxa and 1,155 nucleotide characters (ITS, nchar = 598; benA, nchar = 557). Through the EF1-α phylogenetic analysis of Trichoderma, KUC21406, KUC21404, KUC21401, KUC21411, and KUC21394 are confidently classified as T. afroharzianum and T. harzianum in the EF1-α tree. Talaromyces sp. 2 KUC21413 was closely related to T. viridulus in the ITS tree. In the Mar. Drugs 2019, 17, 601 9 of 26 concatenated tree (ITS & benA), however, the position changed to be near T. galapagensis with low posterior probability. Therefore, Talaromyces sp. 2 KUC21413 was assigned as a novel species candidate. Talaromyces sp. 1 KUC21276 appeared to be closely related to T. angelicus (99.81% sequence similarity in ITS; 93.22% in benA). Talaromyces sp. 3 KUC21415 and Talaromyces sp. 3 KUC21421 appeared to be closely related to T. helices (99.81% sequence similarity in ITS; 96.7% in benA). Talaromyces sp. 4 KUC21408 appeared to be closely related to T. boninensis (97.97% sequence similarity in ITS; 93.72% in benA). Thus, they are suggested to be candidate novel species.   The dominant taxa of the intertidal sediments were Penicillium species (number of strains = 8) and Talaromyces species (number of strains = 6). A number of fungal isolates belonged to Eurotiomycetes, followed by Sordariomycetes. A total of 18 candidate novel fungal species was isolated from this study. This result demonstrates that many novel fungal candidates remain unexploited in abandoned salterns or intertidal mudflat sediments.

Biological Activities of Fungi from Intertidal Mudflats and Abandoned Salterns
To evaluate the potential value of the intertidal environments for bioprospecting, a variety of biological activities of the fungal extracts were investigated.

Tyrosinase Inhibitory Activity
A total of five fungal extracts (Cladosporium sphaerospermum KUC21388, Lulwoana sp. KUC21398, P. citrinum KUC21390, Trichoderma afroharzianum KUC21411, and Westerdykella capitulum KUC21407) showed tyrosinase inhibitory activity (Table 5). Tyrosinase is able to oxidize l-3,4-dihydroxyphenylalanine (l-DOPA) to dopaquinone, which is eventually converted to pheomelanin [48]. As melanin biosynthesis is responsible for darkening of the skin tone, researchers have tried to discover tyrosinase inhibitors. Tyrosinase inhibitors can suppress melanin formation in the skin, so they can be developed as whitening agents. In particular, kojic acid, the most widely used tyrosinase inhibitor, was discovered through the screening of 600 marine fungi [49].
Considering that all the activities were measured using crude extracts, the tyrosinase inhibitory activity of C. sphaerospermum KUC21388, P. citrinum KUC21390, and Trichoderma harzianum KUC21411 extracts were notable. In particular, the IC 50 value of C. sphaerospermum KUC21388 extract was even comparable to that of the positive control, kojic acid. This is the first report of tyrosinase inhibition activity in the genus Cladosporium, which is generally known to produce melanin.

Antifungal Activity
Antifungal experiments on Asteromyces cruciatus and Lindra thalassiae were conducted using the extracts of fungi isolated from intertidal mudflats and abandoned salterns. The two target fungi were selected considering that our fungal isolates were derived from the marine environment. A. cruciatus is a potentially harmful fungus to brown algae because it degrades alginate, a major constituent of brown algae. Alginate plays an important role in brown algae by forming the structure of the algal biomass and physically protecting algae from pathogens. L. thalassiae is a pathogenic fungus of sea plants and brown algae that causes raisin disease [53][54][55]. The results showed that the extracts of 17 fungal strains showed inhibitory effects against mycelial growth in A. cruciatus or L. thalassiae (Table 5).
All the extracts of the four Penicillium sp. 1 strains had potential antifungal activities, as they could inhibit the growth of A. cruciatus at a minimum concentration of 50 µg/mL and could inhibit the growth of L. thalassiae as well. Since they were identified as the same species from the same isolation source, the potential antifungal secondary metabolites are highly likely to be identical. There are a number of antifungal compounds reported from the genus Penicillium, for example, brefeldin A from P. brefeldianum; griseofulvin from P. griseofulvum; atpenins A4, A5, and B from P. atramentosum [56]; xanthocillin X and penicisteroid A from P. chrysogenum [56,57]; calbistrins from P. restrictum [58]; canadensolide from P. canadense [59]; macrocyclic polylactones from P. verruculosum [60]; patulin from P. carneum , citrinin from P. melinii (previously known as P. damascenum), palitantin and arthrographol from P. implicatum [61]; and compactin from P. brevicompactum [62]. There is no report of antifungal compounds or activity from the genus Pseudeurotium. In the case of the genus Talaromyces, some antifungal compounds have been reported, for example, talaroconvolutins from T. convolutus [63], talaron from T. flavus [64], 3-O-methylfunicone from T. pinophilus, wortmannin from T. wortmannii, macrophorin A from T. purpurogenus, and botryodiplodin from T. stipitatus [56]. In particular, Penicillium sp. 1 KUC21389 and Pseudeurotium bakeri KUC21422 exhibited significantly lower MIC values than the other strains. To the best of our knowledge, this is the first report of the antifungal activity of the genus Pseudeurotium. It is strongly speculated that these antifungal metabolites can affect other pathogenic fungi, so they could be developed as biocontrol agents.

Sediment Sampling and Fungal Isolation
Intertidal sediments 10 cm deep from the surface were sampled into sterile polyethylene jars using a sterile knife. The sediment samples were sieved using 2-mm sieves and stored at −80 • C for further use. To isolate fungi from the sediment samples, 1 gram of wet sediment was mixed with 10 mL of sterilized D.W. and vortexed. Preliminary tests with a variety of different intertidal sediment samples showed that the optimal concentration for isolating fungi from intertidal sediments was 5-10 times higher than those typically used when using forest sediments. The reason may be that the amount of organic matter in the sediment is relatively small, so that the amount of microbial biomass is small. The sediment suspension was serially diluted to a ratio of 1 g : 1000 mL, and 3 mL of the diluted suspension was added to 47 mL of the culture medium composed of 2% malt extract, 3.59% artificial sea salt (Instant Ocean, Aquarium Systems, Mentor, OH, USA), and 0.01% streptomycin sulfate (Sigma-Aldrich, St. Louis, MO, USA) for inhibiting the growth of bacteria. After that, 150 µL of the inoculated medium was distributed into each well of four sterilized 96-well plates. The plates were incubated at room temperature for 2-4 weeks until the fungal mycelia were visible to the naked eye. Each fungal colony was transferred to an individual 2% malt extract agar plate with 0.01% streptomycin sulfate. The plates were incubated at room temperature for a week and subcultured to isolate single strains. The isolated fungal strains were stored in 10% glycerol solution at 4 • C for further use and deposited in the Korea University Culture (KUC) collection.  Table 1.

Phylogenetic Analysis
All the obtained sequences were assembled, proofread and edited using MEGA v. 7.0 [120]. Edited sequences were aligned using MAFFT 7.130 [121] and ambiguously aligned positions manually modified using MacClade 4.08 [122]. Through MrModeltest 2.3 using Akaike information criterion criteria, the best-fit models were calculated for the Bayesian phylogenetic analysis [123]. The Bayesian phylogenetic analyses were performed for 10 million generations using MrBayes 3.2.1 [124]. Phylogenetic trees were sampled every 100 th generation. After sampling, the last 75% of the trees were selected. The constructed phylogenetic tree followed the 50% majority-rule and the tree reliability was presented by posterior probability.

DNA library Preparation and Amplicon Sequencing
Total DNA was extracted from 0.3 g of frozen sediment using a PowerSoil DNA isolation kit (MoBio, Carlsbad, CA, USA) following the manufacturer's protocol. DNA libraries were constructed using the Illumina MiSeq platform with the fungal ITS rDNA gene. An approximate 300-350 bp region of the ITS2 region was amplified with forward primer fITS7 [125] and reverse primer ITS4 [116]. Primer fITS7 contained a unique 12-nt barcode at the 5 end for MiSeq sequencing detection. Paired-end read sequences were generated by high-throughput sequencing technology.

Bioinformatics Analyses
The sequences obtained in this study were processed using QIIME v1.9.1 [126]. The primer, key, and barcode sequences were trimmed from both ends. Sequences composed of homopolymers (n > 6) and chimeras were removed, and the remaining sequences were analyzed. The sequences were then clustered as operational taxonomic units (OTUs) based on a ≥ 97% similarity threshold and the average linkage method using Vsearch [127]. The representative sequence that was most abundant from each OTU was taxonomically assigned using the UNITE [128] reference database.
Nonmetric multidimensional scaling (NMDS) analysis applying the Bray-Curtis similarity index was performed to plot the similarity of the fungal communities in a way such that distances could be represented in two dimensions, using the vegan package in R statistical software, version 3.5.3. Alpha diversity indices were calculated using QIIME, and differences among the sediments from the abandoned salterns and mudflats were compared. After taxonomic profiling, we also investigated which fungal taxa differed in relative abundance among the intertidal environments. The relative abundances of the fungal communities were compared using linear discriminant analysis effect size (LefSe) analysis [129] with a LDA score threshold of 2.0 at the phylum, class, order, and family levels. Indicator species analysis was performed using the labdsv package in R to determine the key fungal taxa that best represent the Yellow Sea intertidal environments. Indicator species values are based on how specific and widespread the taxa are within a particular group and are independent of the relative abundance of other fungi [130]. RandomForest analysis was performed to identify key predictors of the environments among the fungal taxa. The accuracy importance measure was calculated for each tree and averaged over the forest (10,000 trees). This analysis was performed using the randomForest package in R. To determine which fungal taxa differed among the environments, the multivariate statistical package mvabund [131] was used in R; this approach is reported to have much greater power than distance-based approaches. In the generalized linear model, the relative abundance of each fungal taxon was modeled on the negative binomial distribution.

Preparation of Fungal Extracts
All the fungal isolates were precultured on 90-mm Petri dishes (10090, SPL Life Sciences Co., Pocheon, Republic of Korea) containing 20 mL of potato dextrose agar (PDA) at 25 • C for a week. After that, three agar plugs with actively-growing mycelia were inoculated to 150-mm Petri dishes (10150, SPL Life Sciences Co., Pocheon, Republic of Korea) containing 50 mL of PDA and incubated at 25 • C for 7 days in the dark. The cultures were extracted with 200 mL of MeOH and filtered with Whatman No.1 filter paper. The filtrates were dried at 35 • C using a rotary evaporator, and the dried extracts were dissolved in 15 mL of D.W. and 15 mL of EtOAc. After 6 hours, the EtOAc layers were transferred to 20-mL scintillation vials and dried using the rotary evaporator, and the dried extracts were stored at 4 • C until use. All the extracts were prepared in triplicate.

Biological Assays
The biological assays were performed with methods that were the same as or slightly modified from the methods used in our previous study [132]

Antifungal Assay
Asteromyces cruciatus SFC20161110-M19 was obtained from Marine Fungal Resource Bank (MFRB) at Seoul National University as a marine bioresource bank of Korea by the Ministry of Oceans and Fisheries, and Lindra thalassiae NBRC106646 was purchased from Biological Resource Center under National Institute of Technology and Evaluation. Antifungal activity was determined in a 96-well plate using the microtiter broth dilution method [134]. Twenty-five microliters of spore suspensions (4 × 10 5 conidia/mL) of the target fungi were added to each well containing 25 µL of 4X potato dextrose broth and 49 µL of D.W. The fungal extracts were added to a final concentration of 100 µg/mL. The inoculated plates were incubated at 25 • C for 3 days. The extracts with lower concentrations (50, 25, 12.5, 6.5 µg/mL) were tested to determine MIC, the minimum concentration of an antimicrobial agent that causes microbial death.

Quorum Sensing Inhibition Assay
QSI screening was performed based on the inhibition of violacein production by Chromobacterium violaceum CV026 strains under culture conditions supplemented with an exogenous QS molecule, 3-oxo-C6-HSL [135]. The C. violaceum CV026 cultured overnight was diluted with LB medium (5 g of yeast extract, 10 g of tryptone, and 5 g of NaCl in a liter of D.W.) to an OD 600 nm of 0.1, and 97 µL of the diluted culture was added to each well of a 96-well plate. One microliter of 3-oxo-C6-HSL (final concentration of 10 µM; Sigma-Aldrich, St. Louis, MO, USA) and 2 microliters of fungal extract (10 mg/mL DMSO) were added to each well. As a negative and positive control, two microliters of DMSO and two microliters of 100 nM (Z)-4-bromo-5-(bromomethylene)-2(5H)-furanone (Furanone C-30; Sigma-Aldrich, St. Louis, MO, USA) were used, respectively. After 16 h at 28 • C, 100 µL of DMSO was added to each well and vigorously shaken for an hour at room temperature to determine the production of violacein.
The selected extracts were further examined using a paper disc method. Fifteen microliters of the diluted culture of C. violaceum CV026 was spread on an LB agar plate, and a paper disc (diameter 8 mm; Advantec, Tokyo, Japan) impregnated with 40 µL of extract and 1 µL of 1 mM 3-oxo-C6-HSL was placed in the center of the plate. DMSO was used as a negative control, and piericidin A isolated from Streptomyces xanthocidicus KPP01532 was used as a positive control [135]. The plates were incubated overnight at 28 • C.

Statistical Analyses
The half maximal inhibitory concentration (IC 50 ) values were calculated by nonlinear regression analysis using SigmaPlot 12.0 (Systat Software Inc., San Jose, CA, USA).

Conclusions
The results of the fungal community analysis showed that only the abandoned saltern in Yongyudo was significantly different from the other abandoned salterns or mudflats. The fungal communities in the abandoned salterns in Yubudo and Gopado had been restored to a similar status as those in the mudflats, but that was not the case of the saltern in Yongyudo, which was abandoned less than a year ago. The results also implied that it takes more than 35 years to restore abandoned salterns to mudflats via natural restoration. The α-diversity of the fungal community in the NS was significantly lower than that of the RSs and mudflats. The diversity indices were high in general, which implies that these unique environments are valuable for bioprospecting.
Through various statistical analyses, the lineages of Gigasporales sp. and Umbelopsis sp. were selected as indicator taxa of mudflats in the Yellow Sea in South Korea. We demonstrate that the ecological recovery of the abandoned salterns can be assessed by using the two indicator taxa to indicate not only the recovery state but also the recovery direction, i.e., whether the recovery process is moving in the right direction. In this study, several analytical methods were applied to investigate indicator taxa. However, the results were not identical, even though we only used methods that use the absolute abundance data rather than processed data such as ecological distances that could lead to statistical misinterpretation [27]. Therefore, indicator taxa must be selected by carefully comparing the results of various types of analysis. We also proved that the fungal community is highly suitable for use in ecosystem assessment, as it has a sufficient level of identification with good resolution. The limitation was that the number of samples, especially for the NS and RS environments, was insufficient. A larger sample size is recommended for more accurate statistical analysis with high fidelity.
To evaluate the fungal diversity and compare it with the results from the metagenome analysis, the culturable fungi were isolated from the sediment samples. A total of 53 fungal strains was isolated, and the number of fungal isolates differed by sampling location rather than by environment type. Based on the phylogenetic analysis, Talaromyces was the most diverse genus, consisting of seven different species. The genera Talaromyces and Trichoderma were mostly isolated from the mudflat sediments, and most Penicillium spp. were found in the NS. The diversity and the number of isolates in the NS were higher than those in the other salterns, which is inconsistent with the results of metagenomics, and every isolate except P. chrysosporium belongs to the Ascomycota. This result showed that metagenomics must be applied when examining the microbial diversity within an actual environment, as has been suggested over the past decades [29]. Interestingly, a total of 18 fungal isolates was identified as candidate novel species.
To evaluate the potential value of the intertidal environments for bioprospecting, a variety of biological activities of the fungal extracts were investigated. Regardless of the actual diversity, the isolates from both intertidal environments exhibited a variety of biological activities. Over half of the culturable fungal isolates (29 strains) exhibited biological activities, and Penicillium spp. exhibited the most varied activities. The antioxidant compound of Phoma sp. KUC21426, the tyrosinase inhibitors of C. sphaerospermum KUC21388, the antifungal compounds of Penicillium sp. 1 KUC21389 and P. bakeri KUC21422, and the quorum-sensing inhibitor of A. urmiensis KUC21392 will be separated and identified in the near future. Since bioactive compounds with these activities have never been reported from these species, the likelihood of discovering novel compounds is high.
In this study, we presented the fungal community and diversity of the abandoned solar salterns and the intertidal mudflats on three different islands located in the Yellow Sea of South Korea, and we identified indicator taxa for the assessment of the restoration of abandoned salterns to mudflats. Additionally, we provided reliable DNA information for the 53 fungal isolates from the intertidal sediments and their exploitable biological activities. We demonstrated that these unique environments are highly valuable in bioprospecting not only for novel microorganisms but also for novel bioactive compounds. Restoring the reduced biodiversity of abandoned salterns is important in terms of environmental protection and conservation. However, many novel fungal species were found from the abandoned salterns, and interestingly, all fungal strains selected for high bioactivity were isolated from NS or RS. Therefore, active measures to recover abandoned salterns to mudflats should be carefully determined and planned, taking into account their value as a source of biological resources.
Supplementary Materials: The following are available online at http://www.mdpi.com/1660-3397/17/11/601/s1, Figure S1: Linear discriminant analysis (LDA) scores of the fungal taxa of which score > 2. NS and RS mean non-recovered saltern and recovered saltern, respectively, Figure S2: Relative abundance of Ascomycota in the three different intertidal environments. NS and RS mean non-recovered saltern and recovered saltern, respectively, Figure S3: The Bayesian analysis tree based on EF1-α for Trichoderma complex. Bayesian posterior probabilities (PP) at the nodes are presented if > 75. All of reference strains are type and ex-type strains. The fungi isolated from this study are in bold. The scale bar means the number of nucleotide substitutions per position, Figure