Changes in soil microbial communities at Jinsha earthen site are associated with earthen site degradation


 Background: Jinsha earthen site in Chengdu, China, plays an important role in understanding the ancient culture and history of Shu civilization. The site is undergoing soil degradation due to physical, chemical and biological factors, while very little is known about the influence of biological factors on earthen sites. To investigate the biological factor, we analyzed microbial communities and physicochemical properties from samples with no obvious, mild, moderate and severe degradation, referred to as S1, S2, S3 and S4 sample groups, respectively.Results: Amplicon sequencing targeting the 16S rRNA gene and ITS for bacteria and fungi, respectively, revealed high bacterial and relatively low fungal diversity; the bacterial OTUs were assigned into 36 phyla and 617 genera and the fungal OTUs into 5 phyla and 205 genera. The relative abundances of Bacteroidetes, Proteobacteria and Firmicutes were higher and that of Actinobacteria lower with higher degree of degradation. In the genus level, the relative abundances of Bacteroides and Ralstonia were higher and that of Rubrobacter lower with higher degree of degradation. The distribution of the fungal genera in the four sample groups seemed more random than that of bacteria; however, the relative abundance of the yeast genus Candida was highest in the severely degraded sample group. For both bacteria and fungi, the differences in community composition were associated with differences in EC, moisture, pH, and the concentrations of NH 4 + , K + , Mg 2+ , Ca 2+ and SO 4 2- .Conclusion: Taken together, the microbial communities in soil with different degree of degradation were distinctly different at Jinsha earthen site, and degradation was accompanied with bigger changes in the bacterial than in the fungal community.

environmental impacts such as erosion due to severe winds and heavy rainfall, earthquakes, fluctuation in temperature and humidity, and the growth and metabolism of microbes [3,4]. The problems in earthen sites include weathering, cracks, alkali-bleaching, collapses and shedding [4][5][6].
The soil of earthen sites is different from farmland or forest soil and characterized by low mechanical properties and overturning resistance [7]. Over time, the cohesive forces between soil particles have weakened or disappeared, and the distance between particles has increased, which has changed the surface of earthen sites. Erosion in earthen sites is mediated by physical, chemical and biological processes that cause degradation and result in the loss of cultural information [8][9][10]. The loss will be permanent, causing immeasurable damage to the study of history and culture. Most research has concentrated on degradation of earthen sites due to the physicochemical environmental factors [1,2,[11][12][13]. Soil microorganisms play an important role in weathering, soil formation and development, and probably cause serious damage to earthen sites [14]. However, information on microbial community structures in earthen sites is scarce. In particular, very little is known about the relationship between these communities and the degradation of earthen sites.
In this study, we used high-throughput sequencing to analyze bacterial and fungal communities in soil with different degree of degradation at Jinsha earthen site. The objective was to (1) characterize the communities, (2) compare overall community structure and the relative abundances of individual taxa in soil with different degree of degradation, and (3) determine the correlation of microbial communities with environmental factors. We hypothesized that community composition would vary depending on degradation degree.

Results
The soil physicochemical properties Soil characteristics varied with the degradation of earthen site (Table 1). Soil moisture was higher in moderately and severely degraded S3 and S4 samples than in not or slightly degraded S1 and S2 samples (p > 0.05), and pH ranged from 7.32 in S4 to 7.48 in S1 to S4 (p > 0.05). EC value and the concentrations of Mg 2+ , Ca 2+ and SO 4 2− were higher in the more degraded samples (P < 0.05), whereas the concentrations of NH 4 + and K + were lower (p > 0.05). In the SEM-EDS spot analysis, the relative proportions of C, S, O and Mg elements were higher in S3 and S4 samples than in S1 and S2 samples, and those of Al, Si and K were lower (p > 0.05) (Additional File 1: Table S1).  Figure S2).
The α diversity was estimated using Chao1 and Shannon indices, in order to determine the richness and diversity of the microbial community in soil with different degree of degradation from Jinsha earthen site. Chao1 index reflect the microbial community richness and Shannon index was used to characterize the microbial diversity. For bacteria, richness was higher in S4, than in the other sample groups, and diversity was higher in S3 and S4 than in S1 and S2 (p < 0.05) ( Table 2). For fungi, richness was higher in S1 than in S4 (p < 0.05) ( Table 2), and diversities were on the same level in all the sample groups.
Altogether 330 bacterial and 200 fungal OTUs were detected in all the four sample groups ( Figure 1).
The highest number of unique bacterial OTUs was detected in S4 and the lowest in S2. The highest number of unique fungal OTUs was detected in S4 ( Figure 1). Both the bacterial and fungal communities in not degraded and severely degraded sample groups were clearly separated in the principal component analysis (PCA) ( Figure 2).

Distribution of microbial community in sample groups
The bacterial OTUs were assigned into 36 phyla and 617 genera. Actinobacteria, Bacteroidetes, Proteobacteria and Firmicutes were the most abundant phyla ( Figure 3a). The relative abundances of Actinobacteria were highest in the sample groups S1 and S2, and those of Proteobacteria and Firmicutes in S3 and S4 (p < 0.05) (Additional File 1: Table S2).
On genus level, the relative abundances of Rubrobacter were highest in all sample groups except S4 where that of Bacteroides was highest ( Figure 3b). Compared with the S1 and S2, the relative abundances of Bacteroides and Corynebacterium were higher in S3 and S4 (p>0.05). The results showed that the bacterial community compositions in sample groups with different degree of degradation were significantly different (Additional File 1: Table S3).
The fungal communities were assigned into 5 phyla and 205 genera. In all sample groups, Ascomycota was the most abundant phylum and Basidiomycota the second most abundant with relative abundances ranging from 95.9% to 98.8% and 1.1% to 2.7%, respectively (Figure 4a, Additional File 1: Table S4). Glomeromycota and Chytridiomycota were detected only in sample S4. At the genus level, Toxicocladosporium, Cladosporium and Alternaria were the most abundant genera in sample groups S1 and S3, Fusarium in S2 and Candida in S4 (p < 0.05) (Figure 4b, Additional File 1: Table S5).
Liner discriminant analysis (LDA) coupled with effect size (LEfSE) was used to identify differentially abundant taxa. A total of 46 bacterial taxa were differentially abundant among the four sample groups. Three taxa were significantly more abundant in S1 than in the other three sample groups, six taxa in S2, sixteen taxa in S3, and 21 taxa in S4 (Additional File 1: Figure S3a). A total of 30 fungal taxa were differentially abundant. Fifteen taxa were significantly more abundant in S1 than in the other three sample groups, ten taxa in S3, and five taxa in S4 (Additional File 1: Figure S3b).

The correlation between the microbial community and environmental factors
The relationship between community compositions and environmental factors was analyzed using redundancy analysis (RDA). For bacteria, the RDA axes 1 and 2 accounted for 24.81% and 14.77%, respectively, of the total variation ( Figure 5a); for fungi, 30.01% and 14.96%, respectively ( Figure 5b).
For both bacteria and fungi, the differences in community composition were associated with differences in EC, moisture, pH, and the concentrations of NH 4 + , K + , Mg 2+ , Ca 2+ and SO 4 2-.

Discussion
In this study, we analyzed bacterial and fungal community composition in soil with different degree of degradation in Jinsha earthen site, Chengdu, China, using high-throughput sequencing approach. For bacteria, the relative abundance of Actinobacteria was four times lower in the moderately and severely degraded sample groups S3 and S4 than in the not or mildly degraded sample groups S1 and S2; those of Proteobacteria and Firmicutes were highest in S3 and S4. However, it should be noted that changes in absolute abundances cannot be concluded from the relative abundance data [15].
Actinobacteria have been frequently found as a dominant group in subterranean micro-niches, including cultural relics in caves and wall paintings in catacombs [16][17][18][19][20]. Previous work demonstrated that some Actinobacteria are potentially harmful to the preservation of cultural relics [21][22][23]. Proteobacteria are commonly the most abundant bacteria in soil [24,25], and Firmicutes that are tolerant to extreme temperatures and low humidity are often found in extreme environments [20]. These bacteria may play an important role in the microecological balance of earthen sites.
Rubrobacter was the most abundant genus in S1, S2 and S3. Rubrobacter was considered connected with the biodegradation of cultural relics and the rosy discolouration of masonry and lime wall paintings in historical buildings in Austria and Germany [26]. Resistance to desiccation might be a selective advantage for Rubrobacter growth and efflorescence on walls might be due to Rubrobacter strains [27]. Therefore, Rubrobacter may play a crucial role at the early stage of degradation of earthen sites. However, in the moderately and severely degraded sample groups S3 and S4, the relative abundances of Rubrobacter were remarkably lower than in S1 and S2, and those of Bacteroides, Ralstonia, Bacillus and Acinetobacter were higher. Bacteroides have the capability to produce acid [28], and the acid can dissolute minerals and further damage the structure of soil.
Moreover, Bacillus and Acinetobacter in other cultural heritage sites have been found to possess efficient degradation ability [20,29]. Therefore, the higher relative abundances of Bacillus and Acinetobacter may potentially accelerate the rate of degradation at Jinsha earthen site.
The distribution of the fungal genera in the four sample groups seemed more random than that of bacteria. The communities in S1 and S3 were similar, indicating that the fungal composition varied only little with the degradation degree of earthen site. The filamentous fungi Cladosporium, Fusarium and Toxicocladosporium were the most abundant genera in sample groups S1, S2 and S3, respectively. These genera are widely distributed in wall paintings in caves, catacombs and churches and have been isolated from severely decayed areas of stone artwork [18,30,31]. Cladosporium, Fusarium and Toxicocladosporium have been reported to produce extracellular enzymes and abundant mycelia that contribute to the mineral dissolution and mechanical destruction of soil structure [18,30,31]. The yeast Candida was most abundant in the severely degraded sample group S4. Candida species have the ability to secrete extracellular metabolites and to acidify soil [32].
Potentially, the accumulation of Candida plays an important role in the degradation of Jinsha site.
Since information on Candida species in cultural relics is scarce, further work is needed to understand the role of this fungal genus.
The diversity and distribution of microorganisms in soil are affected by environmental variables. The microbial community structure can rapidly change in response to altered environmental conditions [33,34]. In our study, the differences in microbial communities were associated with differences in moisture, pH, EC and concentrations of soluble salt ions. Moisture has been found a major factor in affecting microbial communities and their activities [35,36]. The higher moisture in the moderately and severely degraded sample groups S3 and S4 was associated with higher microbial diversity. Soil pH is another major factor connected with soil microbial community structure. In our study, the pH was higher in the moderately and severely degraded sample groups S3 and S4 than in the not or mildly degraded sample groups S1 and S2. Bacterial communities have been found more sensitive to changes in pH than fungal communities [37]. This could partly explain the more pronounced difference in bacterial communities than in fungal communities along the difference in degradation degree.
EC is apparently associated with soil salinity [38]. As in an earlier study [39], the relative abundance of Bacteroidetes was found to correlate positively with EC. Soluble salts are considered to cause damage on earthen sites [5,11]. We found that the differences in microbial communities were associated with differences in soluble salt concentrations. The bacterial and fungal communities in samples with no obvious degradation correlated positively with K + concentration, and those in  [8]. Plausibly, the weathering of these components lead to higher concentrations of Mg 2+ and Ca 2+ in the moderately and severely degraded sample groups S3 and S4.
The nutrients released by weathering could support the growth and metabolism of microorganisms.
As the microbially caused damage for cultural heritage relics is the result of microbial growth and the corresponding metabolism [41], the higher microbial diversity and possible more active metabolism may have potential harm to the preservation of Jinsha earthen site.

Conclusions
In this study, high throughput sequencing was applied to explore microbial community structures in soil with different degree of degradation from Jinsha earthen site. The dominant bacterial phyla and genera showed more variability than fungal phyla and genera. Furthermore, differences in bacteria and fungus community composition were associated with differences in soil physicochemical properties. Microbial diversity, soil moisture and the concentrations of Mg 2+ , Ca 2+ and SO 4 2− were higher and soil pH was lower with the increased degree of earthen site soil degradation. referred to as S1, S2, S3 and S4 sample groups, respectively (Figure 6b, Figure 6c). Samples were taken using minimally invasive sampling techniques and aseptic procedures, and transported on ice to the laboratory.

Physicochemical properties analyses
Due to the minimal intervention principle in sampling in an archaeological site, the sample quantities were not enough to meet the requirements of routine soil property analyses. We analyzed pH, moisture, electrical conductivity (EC) and soluble salt contents that are considered to play significant roles in the degradation of earthen sites [1,3,11,42]. The samples were air-dried, crushed and sieved through a ø 1 mm sieve. The determination of pH and EC were made on 1:1 slurry of air-dried soil and water [43]. Moisture was determined by oven drying method [44]. Soluble salt contents were measured using ion chromatography (IC): 2 g sample was suspended into 20 mL deionized water, the suspension was shaken for 30 minutes at 150 rpm, filtered first through medium pore sized filter paper and finally through 0.2 μm pore size syringe membrane filter [45]. Ions in the extracts were determined using a Dionex ICS-3000

16S rRNA and ITS amplicon sequencing
The DNA samples were sequenced at Novogene Bioinformatics Technology, Co., Ltd. (Beijing, China).

Bioinformatic and statistical analyses
Single-end reads were assigned to samples according to their unique barcodes and primers and barcodes were cut off. Low-quality sequences and reads with ambiguous nucleotides were removed using Cutadapt V1.9.1 [46]. Chimeric reads were filtered out using UCHIME v. 4.2.4.0 [47]. Sequences were assigned to operational taxonomic units (OTUs) at ≥ 97% similarity using UPARSE v7.0.1001 [48]. The representative sequences of the OTUs were assigned to taxa using Silva Database based on Mothur algorithm [49]. Chao1 and Shannon indices were calculated using the phyloseq package [50].
Venn diagrams were done at VennDiagramWeb [51]. Principal Component Analysis (PCA) was done using CANOCO 5 [52]. Differential taxa at phylum to species levels were identified using linear discriminant analysis coupled with effect size (LEfSe) [53]. To analyze the bacterial and fungal community distribution and their correlation with environmental factors, redundancy analysis (RDA) was carried out using CANOCO 5 [52]. Statistical differences among groups were analyzed by One-Way ANOVA with repeated measures followed by a post hoc least significance difference test (SPSS 17.0) [54]. Differences were taken statistically significant at p < 0.05.
The amplicon sequencing data were deposited into the NCBI Sequence Read Archive (SRA) under the accession numbers SRR9678166-SRR9678177 and SRR9678184-SRR9678195.