Microbiota in Waterlogged Archaeological Wood: Use of Next-Generation Sequencing to Evaluate the Risk of Biodegradation

: Waterlogged archaeological wood (WAW) is considered a precious material, ﬁrst-hand account of past civilizations. Like any organic material, it is subjected to biodegradative action of microorganisms whose activity could be particularly fast and dangerous during the phases of excavation, storage and restoration. The present work aimed to characterize the microorganisms present in WAW during these tricky periods to evaluate the biological risk it is exposed to. The bacterial and fungal communities inhabiting woods coming from two archaeological sites (Pisa and Naples) were investigated through Next-Generation Sequencing (NGS). High-throughput sequencing of extracted DNA fragments was performed using the reversible terminator-based sequencing chemistry with the Illumina MiSeq platform. The analyses revealed that the two archaeological sites showed distinct richness and biodiversity, as expected. In all the WAWs, the bacterial community harbored mainly Proteobacteria, whereas Bacteroidetes was well represented only in Naples communities and taxa belonging to the phyla Chloroﬂexi only in the Pisa site. Concerning the fungal community, the two sites were dominated by di ﬀ erent phyla: Ascomycota for Naples samples and Basidiomycota for Pisa. Interestingly, most of the identiﬁed bacterial and fungal taxa have cellulolytic or ligninolytic ability. These results provide new and useful background information concerning the composition of WAW microbiota and the threat it represents for this precious material.


Introduction
Archaeological wood is defined as "wood, used by an extinct human culture, that may or may not have been modified for or by use, and that was discarded by intent or accident into a specific Wood remains from Naples were recovered in 2015 and were stored in water at 4 • C in the dark. They pertain the shipwreck named F and are dated back to the end of the II century AD [32]. The samples from Pisa were excavated in 2000 and were stored in water at room temperature. The fragments were sampled from different wooden remains pertaining the urban harbor of Etruscan and Roman Pisa (VII century BC to V century AD) [33][34][35]. The samples from both sites had never been treated with a biocide. When the analyses were performed, the storage water appeared more or less turbid, depending on the samples, and no biological colonization was perceivable by naked eye in the water or on the wood surface.
In 2008, the wooden remains from Pisa had been characterized and the wood level of degradation had been determined in the frame of a thesis project of the ICR restoration school [36]. In order to evaluate the present state of wood preservation with a quantitative method, to compare it to results obtained in the 2008 situation, and to establish a correlation between the wood degradation and the microbiota composition, analyses for the physical characterization of wood were performed on the Pisa samples selected for the present work. In particular, the maximum water content (MWC) and basic density (D bd ) were measured, and the lost wood substance (LWS) was calculated according to the Italian standard and to the most followed protocols [37][38][39][40][41].
The decrease of physical properties accounts for biodegradation processes which occur both during the lying in the waterlogged sites and after the recovery, and the two contributions cannot be distinguished. As no data concerning the state of preservation of Naples remains before the beginning of the present work were available, a certain correlation between the degradation of wood occurred during the storage phase and the present composition of the microbiota could not have been established. Therefore, no physical analyses on these samples are reported here.
The protocols for the DNA extraction were developed basing on the extraction kits Maxwell®RSC Plant DNA Kit (AS1490) (Promega Corporation, Madison, WI, USA) and Maxwell®RSC PureFood GMO and Authentication Kit (AS1600) (Promega Corporation, Madison, WI, USA). In order to compare the extraction efficiency on the same microbial community, both protocols were applied to the same wood sample ( Table 2). The only exception is represented by samples PL1 and PL2, for which the wood material was not sufficient to perform a double extraction.
DNA was extracted from 100 mg of wood frozen in liquid nitrogen and homogenized using mortar and pestle. In both protocols, every sample was incubated 5 minutes at 95 • C in a lysis buffer (A509C) (Promega Corporation, Madison, WI, USA) modified with 3% of PVPP. Then, proteinase K (MC5005) (Promega Corporation, Madison, WI, USA) was added in the buffer and the samples were incubated 25 minutes at 65 • C. Finally, after a 13,000-rpm centrifugation, the samples were processed by the two different kits and total genomic microbial DNA was extracted with Maxwell®RSC Instrument (Promega Corporation, Madison, WI, USA) The concentration and purity of the extracted DNA were evaluated using the Nanodrop microvolume sample retention system (Thermo Fisher Scientific, Waltham, MA, USA).
Data are available from NCBI under accession number PRJNA641785 with SRA datasets: ITS SUB7670266, 16S SUB7669523.

Bioinformatic Analyses
The marker data were analyzed using qiime2 [42] (https://qiime2.org), according to the standard pipelines [43]. Briefly, quality trimming and OTU-picking was done using DADA2 [44], representative sequences were aligned using mafft [45], uninformative positions were masked and a phylogenetic tree was built with fasttree [46]. Alpha diversity values and beta diversity (i.e., UniFrac distance) were calculated on rarefied samples. Assessment of significant variation of alpha diversity between categories was determined using the Kruskal-Wallis test. Beta diversity significance (among categories) test was calculated with PERMANOVA and Mantel test, respectively. Taxonomic assignment was given to representative sequences using the most updated version of the SILVA database (release 132) [47], or (for fungal data) the UNITE database [48]. The feature classifier was trained using the qiime2 classify-sklearn plugin on the database; the same plugin also classifies the reads in the real dataset. Graphics about taxonomic composition and multivariate analyses were done using Calypso [49]. Table 3 reports the values of DNA concentration and purity for the two protocols. For all double extracted samples, the DNA concentration obtained with Protocol 1 is higher compared to that yielded by Protocol 2. The purity of the extracted DNA was evaluated in term of the ratio of absorbance at 260 and 280 nm. The high impurity of some of the samples treated with Protocol 2 may probably indicate that the kit's reagents acted also on wood extractives. This hypothesis is confirmed by the dark color of some of the DNA suspensions. The extracts obtained with both protocols were used for the preparation of the libraries to check if there were differences in the library amplification efficiency. All samples were able to generate libraries with a concentration of 4 nM and were loaded onto the flow cell. The raw data (Table S1) showed that all libraries yielded an adequate number of reads.

Results and Discussion
The Principal Component Analysis (PCA) calculated on the ecological matrix for 16S at the level of order, as expected, well separates N woods from the P ones ( Figure 1), regardless of the extraction method.
Appl. Sci. 2020, 10, x 5 of 21 probably indicate that the kit's reagents acted also on wood extractives. This hypothesis is confirmed by the dark color of some of the DNA suspensions. The extracts obtained with both protocols were used for the preparation of the libraries to check if there were differences in the library amplification efficiency. All samples were able to generate libraries with a concentration of 4 nM and were loaded onto the flow cell. The raw data (Table S1) showed that all libraries yielded an adequate number of reads.
The Principal Component Analysis (PCA) calculated on the ecological matrix for 16S at the level of order, as expected, well separates N woods from the P ones ( Figure 1), regardless of the extraction method. The samples from Naples (red dots) and Pisa (blue square) show distinct clusters on the first axis that accounts for the 54% of the total variance. The samples from Naples (red dots) and Pisa (blue square) show distinct clusters on the first axis that accounts for the 54% of the total variance.

Bacterial Community
The Beta diversity reported in Figure 2a shows the generic distribution of microbial community considering the difference in the bacterial communities present in Naples and Pisa samples. The graphic highlights that the samples cluster in two well-defined groups, each representing an archaeological site. Alpha diversity, based on observed OTUs and on Shannon index, indicates the richness and the biodiversity of each sample community. Figure 2b shows the boxplot of Shannon index in the two sites. The median Shannon values for the two communities are similar but the index varies in a wider range for Naples samples.
Appl. Sci. 2020, 10, x 6 of 21 The Beta diversity reported in Figure 2a shows the generic distribution of microbial community considering the difference in the bacterial communities present in Naples and Pisa samples. The graphic highlights that the samples cluster in two well-defined groups, each representing an archaeological site. Alpha diversity, based on observed OTUs and on Shannon index, indicates the richness and the biodiversity of each sample community. Figure 2b shows the boxplot of Shannon index in the two sites. The median Shannon values for the two communities are similar but the index varies in a wider range for Naples samples. Sequencing results showed that the bacterial communities of almost all the analyzed samples were dominated by taxa belonging to the phyla Proteobacteria, Acidobacteria, and Planctomycetes ( Figure 3a). It is interesting to observe that the community composition obtained with the two extraction methods for the same wood sample is always very similar. This suggests that the results obtained are not biased by an extraction limit and that it can be considered as representative of the actual microbial community.
Proteobacteria accounted for more than 50% of the total reads in all the Naples samples, reaching 96% for N88, and more than 30% for Pisa woods. The phylum Bacteroidetes was well represented in Naples communities, in some cases reaching relative frequencies of 10−20%, while it accounted for less than 5% in Pisa samples. The opposite results were registered for the phylum Chloroflexi.
At class level ( Figure S1), the communities of N samples were mainly composed of Gammaproteobacteria, Alphaproteobacteria, Planctomycetacia, and Bacteroidia with a neat prevalence of the first taxon. Instead, Alphaproteobacteria dominated in P samples and together with Gammaproteobacteria, Planctomycetacia, and Acidobacteriia accounted for more than 60% of the whole communities. Sequencing results showed that the bacterial communities of almost all the analyzed samples were dominated by taxa belonging to the phyla Proteobacteria, Acidobacteria, and Planctomycetes ( Figure 3a). It is interesting to observe that the community composition obtained with the two extraction methods for the same wood sample is always very similar. This suggests that the results obtained are not biased by an extraction limit and that it can be considered as representative of the actual microbial community.
Proteobacteria accounted for more than 50% of the total reads in all the Naples samples, reaching 96% for N88, and more than 30% for Pisa woods. The phylum Bacteroidetes was well represented in Naples communities, in some cases reaching relative frequencies of 10−20%, while it accounted for less than 5% in Pisa samples. The opposite results were registered for the phylum Chloroflexi.
The communities of Naples samples were mainly composed of microorganisms belonging to the families Pseudomonadaceae (g. Pseudomonas), Burkholderiaceae (g. Janthinobacterium), Methylophilaceae, Pirellulacecae, and Xanthobacteraceae. Instead, the genera of the families Xanthobacteraceae (g. Pseudolabrys), Caulobacteraceae, Solibacteraceae (g. Bryobacter), and Hyphomicrobiaceae (g. Hyphomicrobium) were enriched in P samples ( Figures S2 and 3b). Even if the role of bacteria in the biodegradation of WAW is well-known and several studies investigated the degradation patterns produced by these microorganisms [8,9,11,50,51], very little is known about the bacterial genera involved in this phenomenon. Some of the genera identified during the present work (e.g., Pseudomonas, Janthinobacterium, Flavobacterium, Brevundimonas, Sphingomonas, and Spirochaeta) are known as members of the microbial community present in waterlogged wood and as active degraders of cellulose and/or lignin [22][23][24][25]29,31]. Pseudomonas sp. characterized the community of all the samples from Naples, reaching more than 40% of the relative frequency in N88, while the taxon was present only in four of the analyzed Pisa samples, always with frequencies lower At class level ( Figure S1), the communities of N samples were mainly composed of Gammaproteobacteria, Alphaproteobacteria, Planctomycetacia, and Bacteroidia with a neat prevalence of the first taxon. Instead, Alphaproteobacteria dominated in P samples and together with Gammaproteobacteria, Planctomycetacia, and Acidobacteriia accounted for more than 60% of the whole communities.
Even if the role of bacteria in the biodegradation of WAW is well-known and several studies investigated the degradation patterns produced by these microorganisms [8,9,11,50,51], very little is known about the bacterial genera involved in this phenomenon. Some of the genera identified during the present work (e.g., Pseudomonas, Janthinobacterium, Flavobacterium, Brevundimonas, Sphingomonas, and Spirochaeta) are known as members of the microbial community present in waterlogged wood and as active degraders of cellulose and/or lignin [22][23][24][25]29,31]. Pseudomonas sp. characterized the community of all the samples from Naples, reaching more than 40% of the relative frequency in N88, while the taxon was present only in four of the analyzed Pisa samples, always with frequencies lower than 0.5%. Microorganisms belonging to the genus Pseudomonas are strict aerobes or facultative anaerobes. Some species are involved in denitrification or are implicated in sulfur and iron metabolism [23]. A study carried out on chips of Eucalyptus grandis x Eucalyptus urophylla, Populus canadensis, and Larix olgensis demonstrated that the strain Pseudomonas sp. PKE117 is able to produce a wood weight loss ranging from 8% to 27% in 60 days. The characterization of the degraded wood showed that the lignin structure was degraded more than the cellulose [52]. Other studies demonstrated that several species belonging to the genus Pseudomonas are able to degrade lignin and lignin model compounds, to oxidize carbohydrates and to degrade cellulose via different metabolic pathways [53][54][55][56][57][58][59].
Species belonging to this genus are considered as cosmopolites and no specific relations have been reported with one or more wood species. In the present study, Pseudomonas sp. was identified in silver fir, elm, ash and holm oak samples. Landy et al. [23] reported Pseudomonas spp. from WAW pilings from different sites across Europe. The species were associated with spruce (Picea abies (L.) H.Karst.), fir, Scots pine (Pinus sylvestris L.), oak and poplar (Populus sp.). Palla et al. [25] identified Pseudomonas spp. from pine samples belonging to the wood recovered inside a rostrum in the site of Acqualadroni (Messina, Sicily). Wagner et al. [29] identified three Pseudomonas strains from Quercus robur/petraea remains recovered from different sites across Europe.
In WAW, the genus Janthinobacterium is reported in association with Pseudomonas, in samples coming from sites characterized by limy soil [23,29]. Some species of the genus are reported from lake sediments, as part of the bacterioplankton of maritime Antarctic lake and from Antarctic snow [60][61][62]. In the present study, it was identified only in N88 and N39, respectively silver fir and elm, in association with Pseudomonas. Ravindran and Yang [63] demonstrated that the strain Janthinobacterium sp. AR-129 has cellulolytic activity and is able to produce high thermal stable cellulase.
Flavobacterium was identified in all the Naples samples but not in Pisa. The genus is reported as part of the bacterial community of WAW piles coming from two archaeological sites from the Netherlands [23] and it has also been identified in the storage water of lacquerware from the Nanhai No. 1 shipwreck (China) [31]. Members of the genus Flavobacterium have a wide distribution, they mostly occur in aquatic ecosystems ranging in salinity from freshwater to seawater, but have also been isolated from soil and sediments [64]. The species F. akiainvivens was isolated from decaying wood of the Hawaiian shrub Wikstroemia oahuensis [65]. The known species belonging to this genus are able to degrade cellulose derivatives but not crystalline cellulose, some strains have proved to be able to lysate algae [64,66].
Some of the identified genera never reported before in WAW deserve a special mention. The genus Methylovirgula was identified only in two samples N88 (silver fir) and PF (ash). It is part of the bacterial community present in decaying wood of the species Fagus sylvatica (beech), Picea abies and Pinus sylvestris [72,73]. The species Methylovirgula ligni has been found on beechwood blocks attacked by white-rot fungi [74]. Methylovirgula bacteria are obligated methylotrophs, and can use methanol as the sole carbon source. Methanol is produced during the decomposition of woody materials and this could explain why these microorganisms are found in wood colonized by fungi and in advanced stages of decay [72,73].
The genus Bryobacter was identified in both Naples and Pisa samples. In the latter, it reached a relative frequency of 13%, while in N wood the frequency was always equal to or lower than 1%. The genus comprises acidotolerant, strictly aerobic, slow-growing chemoorganotrophic bacteria, which inhabit acidic wetlands and soils and are capable of hydrolyzing several heteropolysaccharides [75]. Currently, only the species Bryobacter aggregatus is described for this genus, it was isolated from boreal Sphagnum peat bogs. This species is able to hydrolyze several substrates, among which pectin and starch, but not cellulose [76].

Fungal Community
As for the 16S data, the Beta diversity index (Figure 4a) shows a difference between the fungal communities of Naples and Pisa samples. The median Shannon values, indicating Alpha diversity, are similar for the two communities (Figure 4b).

Fungal Community
As for the 16S data, the Beta diversity index (Figure 4a) shows a difference between the fungal communities of Naples and Pisa samples. The median Shannon values, indicating Alpha diversity, are similar for the two communities ( Figure 4b). The sequencing results showed that the fungal communities of the wood samples coming from Naples were dominated by the phylum Ascomycota, accounting on average for more than 50% and reaching in some cases more than 90% of the total reads (Figure 5a). In Pisa samples, a neat prevalence of the phylum Basidiomycota was registered. Sequences attributed to this phylum represented more The sequencing results showed that the fungal communities of the wood samples coming from Naples were dominated by the phylum Ascomycota, accounting on average for more than 50% and reaching in some cases more than 90% of the total reads (Figure 5a). In Pisa samples, a neat prevalence of the phylum Basidiomycota was registered. Sequences attributed to this phylum represented more than 65% of the identified taxa in almost all analyzed samples. It is interesting to note the presence of sequences belonging to the fungal phyla Chytridiomycota and Rozellomycota in some of the P samples and of the phylum Cercozoa (supergroup Rhizaria) in both N and P samples (reaching more than 67% in sample PO). Again, the extraction method does not seem to have modulated the composition of the fungal community (Figure 5a). Differently from what has been observed in the bacterial communities, most of fungal genera were exclusively identified in one or two of the analyzed samples (Figure 5b). Usually, the relative frequencies obtained for each taxon with the two extraction protocols do not coincide; therefore, in the results, both values are reported.
Fungal identification at species level through ITS barcoding is not very reliable due to problems linked to the insufficient hypervariability and/or amplicon length (especially for species-rich ascomycete genera), insufficient annotations in public DNA repositories, and the unreliable sequences deposited in the reference databases [96][97][98]. However, in the discussion of the results some of the identified species ( Figure S5) will be mentioned due to the interest they have for the At class level ( Figure S3), the communities of N samples were mainly composed of Sordariomycetes (on average 55% of the total reads), Leotiomycetes (on average 39%), and Dacrymycetes (5% of the N88 community). Agaricomycetes dominated the P samples, accounting for more than 70% of the total reads on average. The class Dothideomycetes was identified in almost all analyzed samples and reached more than 5% of the relative frequency in most of them. The class Eurotiomycetes was present only in sample PF accounting for 14% (Protocol 2) and 19% (Protocol 1) of total reads.
Differently from what has been observed in the bacterial communities, most of fungal genera were exclusively identified in one or two of the analyzed samples (Figure 5b). Usually, the relative frequencies obtained for each taxon with the two extraction protocols do not coincide; therefore, in the results, both values are reported.
Fungal identification at species level through ITS barcoding is not very reliable due to problems linked to the insufficient hypervariability and/or amplicon length (especially for species-rich ascomycete genera), insufficient annotations in public DNA repositories, and the unreliable sequences deposited in the reference databases [96][97][98]. However, in the discussion of the results some of the identified species ( Figure S5) will be mentioned due to the interest they have for the present work.
The results obtained for Naples samples will be analyzed first. Pleurotheciella is the most abundant genus in all analyzed N samples except for N88 where it is absent. Species of the genus Pleurotheciella are usually isolated from freshwater habitats, the species P. rivularia (identified in the N samples) was collected on decaying wood submerged in freshwater [99,100]. The literature has not reported a possible ligninolytic or cellulolytic activity of this species.
The genus Podospora accounted for more than 66% of total reads of the fungal community of sample N4. Podospora species are saprophytic, predominantly reported as coprophilous [107][108][109][110]. Acidomelania (species A. panicicola, 5.9%-Protocol 1 and 7.6%-Protocol 2) and Cerinosterus (species C. luteoalbus, frequencies 5.3%-Protocol 1 and 5.2%-Protocol 2) were found only in sample N88. The first species is closely related to endophyte species and is usually isolated form the roots of plants living in acid and nutrient-poor environments [111]. The latter was isolated from decayed historic wood on Deception Island (South Shetlands, Antarctica) and from arctic driftwood attacked by soft-rot fungi [105,106].
The fungal communities of Pisa samples are more difficult to interpret. Most of the obtained sequences remained unidentified; for some samples, more than 60% of total reads are generically reported as Fungi. However, in four of the analyzed woods, more than 71% of sequences are reported as belonging to the family Serendipitaceae ( Figure S4), therefore these data suggest that the fungal communities of the Pisa samples are dominated by basidiomycetes. Several fungal species belonging to the phylum Basidiomycota are well known wood degraders. A large number of basidiomycetes have been typified as brown or white rot fungi, able to degrade cellulose and lignin [5,[130][131][132][133][134]. Among the few identified species, the most abundant basidiomycetes are Angulomyces argentinensis and Schizophyllum commune. The first was identified in samples PO (17.2%-Protocol 1 and 0.9%-Protocol 2), PF (0.4%-Protocol 1, not detected with Protocol 2), PL1 (relative frequency 0.2%), and PL2 (relative frequency 0.4%). It has been isolated from water, pollen, soil and cellulosic material in tropical ponds [135,136]. S. commune, found only in sample PO (1.6%-Protocol 1, not detected with Protocol 2), is one of the most widely distributed white-rot fungi. Several studies proved its ability to degrade cellulose and lignin through different enzymatic processes [137][138][139][140][141][142][143].
Among the ascomycetes identified from Pisa samples, the genus Cladophialophora, found in sample PF (19.5%-Protocol 1 and 14.3%-Protocol 2), deserves a mention. Species belonging to this genus have been reported as degrader of wooden artifacts from the Deception Island (South Shetlands, Antarctica) [105]. The species C. bantiana, identified in this study, is a thermotolerant multinucleated saprophytic black mold, isolated from decayed vegetation, wood and soil able to cause myelitis in humans [144,145].

Microbiota and Wood Decay
For both 16S and ITS2 sequences, it was observed that the communities of the samples treated according to the two extraction protocols were very similar, as it can be clearly observed in Figures 3 and 5. This suggests that the obtained results can be considered as representative of the actual communities present in the wood. The differences in the relative frequency registered for almost all identified taxa could be attributed to the extraction method or to the anisotropic distribution of the microorganisms inside the wood.
On the contrary, the neat differences observed in the composition of microbial communities in Naples and Pisa samples should be attributed to the lying site. In fact, it is well know that the heterogeneous environmental conditions of the soil (e.g., pH, climate, organic carbon availability) influence the microbial community [146,147], and soil represents the main source of colonizers for wood in ground contact [148,149]. This statement is obviously valid also for WAW that lies at a certain depth included in waterlogged soil for centuries and that, during excavation and recovery, comes into contact with the most superficial ground layers, the richest in term of microbial biomass and diversity.
By analyzing more in depth the differences among the bacterial communities of the two sites, it is evident that Naples samples were characterized by the highest number of potential biodeteriogen bacterial genera (Pseudomonas, Janthinobacterium, Flavobacterium, Sphingomonas, Methylovirgula, Bryobacter, Acidithiobacillus, Thiomonas, and Rhodanobacter) whereas only three of these taxa, Pseudomonas, Methylovirgula, and Bryobacter were present in Pisa woods. Regarding the fungal communities, 10 of the 13 potential biodeteriogen taxa identified were present only in Naples samples (Pleurotheciella, Cadophora, Podospora, Acidomelania, Cerinosterus, Reticulascus, Mollisia, Lecythophora, Penicillium, and Exophiala) while three were exclusive of Pisa (Angulomyces argentinensis, Schizophyllum commune, Cladophialophora). In this case, it worth remembering that lot of the sequences extracted from Pisa samples remained unidentified so the presence of other ligninolytic and/or cellulolytic species cannot be excluded. Table 4 reports the results of the physical analyses carried out to evaluate the level of degradation of Pisa woods in 2008 and 2018. Data clearly show a general increase in the degradation of wood. For holm oak and oak, an increase by 100−200% of MWC and by 4% of LWS values was registered while for elm and ash MWC almost doubled and LWS increased by ca 5−10%. Degradation occurring during the 10 years of storage could be attributed to the ongoing activity of biodeteriogens present inside the wood, particularly to the fungi (mostly basidiomycetes) which are better competitors with respect to bacteria. It is worth underlining that basidiomycetes proved to have a high degradative potential even in condition of complete imbibition, usually considered as protective for WAW against these biodeteriogens. Finally, it is interesting to note that the bacterial genus Methylovirgula was present only in sample PF. As discussed, these bacteria are obligated to use methanol as carbon source. As this molecule is a degradation product of woody materials, it is probable that it was particularly concentrated in PF which was characterized by an advanced state of decay.

Conclusions
As the discussed results demonstrate, much of the identified bacterial and fungal taxa have cellulolytic or ligninolytic abilities. The potential biodeteriogens were present in the samples from both archaeological sites and no correlation emerged between their presence and the wood species and/or the storage conditions. This means that all WAW recovered from the sites should be considered at risk of biological degradation during the storage pre-restoration phase. The results of physical analyses carried out on wood stored at room temperature showed that a neat increase of the level of degradation can be registered over a 10-year period. It can be supposed that, in the wood preserved at 4 • C, the microorganisms' activity is reduced but it cannot be completely excluded.
Hence, the results obtained in the present work should increase the awareness of conservators and restorers on the importance of adopting suitable practices for the prevention of biodeterioration. The supplement to the Official Gazette of the Italian Republic n. 244 (2001), also known as "Museum Standards", prescribes the use of a biocide during storage and restoration phases, but when it is not possible to attend this prescription, alternative strategies should be taken into account (e.g., change the storage water frequently, use of UV light to sterilize water, shorten storage and restoration time). Finally, the use of high-throughput sequencing analyses during these tricky periods could help in defining the complexity of the microbiota present in WAW highlighting the presence of possibly biodeteriogen taxa.