A new perspective on studying burial environment before archaeological excavation: analyzing bacterial community distribution by high-throughput sequencing

Burial conditions play a crucial role in archaeological heritage preservation. Especially, the microorganisms were considered as the leading causes which incurred degradation and vanishment of historic materials. In this article, we analyzed bacterial diversity and community structure from M1 of Wangshanqiao using 16 S rRNA gene amplicon sequencing. The results indicated that microbial communities in burial conditions were diverse among four different samples. The samples from the robber hole varied most obviously in community structure both in Alpha and Beta diversity. In addition, the dominant phylum in different samples were Proteobacteria, Actinobacteria and Bacteroidetes, respectively. Moreover, the study implied that historical materials preservation conditions had connections with bacterial community distribution. At the genus level, Acinetobacter might possess high ability in degrading organic culture heritage in burial conditions, while Bacteroides were associated closely with favorable preservation conditions. This method contributes to fetch information which would never recover after excavation, and it will help to explore microbial degradation on precious organic culture heritage and further our understanding of archaeological burial environment. The study also indicates that robbery has a serious negative impact on burial remains.

. WS4 was water sample so some physicochemical properties weren't detected. Soil moisture changed notably from 28.57 to 45.32 because of precipitation and surface water influence on WS3. Soil pH value ranged from 6.98 and 7.37. It can be explained that the raw soil was close to neutralization but the precipitation was slightly alkaline. Furthermore, ancient people may artificially add lime to the caesious mud called Qinggaoni, which can be inferred by the high concentrations of calcium in WS3 and WS4. Soil salinity was lower in all samples compared with agricultural soil. WS2 revealed lowest carbon and nitrogen contents while WS3 displayed highest carbon contents. However, the results suggested that most of physicochemical properties varied slightly, while some abnormal data can be explained by special microenvironment created by man-made.
Microbial diversity and community structure. After quality filtering the raw reads, a total of 613299 effective sequences were recovered from all 12 samples (replicates contained). Each library contained 22567 to 112681 reads, while different phylogenetic OTUs (determined at ≥ 97% similarity) were in the range of 230-447, with WS3-3 having the highest and WS2-1 having the lowest.
The alpha diversity indices for all samples were shown in Table 2, of which the Chao1 and ACE estimators reveal the community richness and Shannon and Simpson index indicate the community diversity. The results showed that WS2 achieved significantly lowest index of Chao1, ACE, Shannon and Simpson, suggesting that WS2 was different from other samples in microbial richness and diversity. Both of them were lower in WS2, while the other three groups shared approximate microbial richness and evenness. Rarefaction curves of Chao1 tended to approach the saturation plateau, which indicated the results reasonable. The sequence of different samples were classified at different taxonomic levels using RDP classifier. There was no significant difference in kingdom level, where Bacteria occupied at least 97% population. However, the histogram ( Fig. 1) witnessed marked shift at the phylum and genus level, especially in sample group WS2. The composition of microbial communities included 6 different bacterial phyla (Proteobacteria, Actinobacteria, Bacteroidetes, Firmicutes, Verrucomicrobia, GAL15) with selecting the relative abundance above 1%. Microbial community exhibited little difference in diversities but large in abundances at Phylum level. Proteobacteria was the most abundant group in WS2, accounting for approximate 90% of total bacteria. While Actinobacteria was the dominant phyla in WS1 and WS4, followed by Proteobacteria. Larger percentages of Bacteroidetes (31.41%) and Actinobacteria (31.62%) were observed in WS3. When it comes to genus level, Acinetobacter comprised nearly 50% of the OTUs in WS2, but was found less than 0.1% relative abundance in microbial distribution of other samples. Bacteroides played a predominant role in microbial structure of WS3. Meanwhile, samples of WS1 and WS4 still had the most frequently genus Nocardioides in common, representing about 30% of the OTUs. Generally speaking, taxonomic distribution varied strikingly in WS2 compared with other samples, and indicated higher similarity in WS1 and WS4.
At the genus level, observed differences between samples were represented in Fig. 2. The number of shared OTUs among these samples was 211. WS2 and WS3 possessed more unique OTUs than WS1 and WS4. The largest overlap was detected between WS1 and WS3, followed by WS3 and WS4, while the smallest overlap was found between WS1 and WS2.
The top 50 abundant genera in all samples were selected as the representative genera shown in the heat map (Fig. 3). There was a significant variation of microbial community between WS2 and other samples. The other samples had close Figure 1. Microbial community at phylum (A) and genus (B) level. The OTUs whose proportion was less than 1% was assigned as other. Refer to Table 3    connection with each other so that some of repetitions were separated. The dominant genera in WS3-1, WS3-2 and WS1-3 were different from the rest, while W1-1 and W1-2 gathered firstly with WS4-1 and WS4-3.
Principal Co-ordinates Analysis which was shown in Fig. 4 is a universally used method in analyzing Beta diversity of microbial distribution. PCoA based on unweighted unifrac distance matrix and weighted unifrac distance matrix revealed clear distinctions between bacterial populations between different samples, respectively. Samples WS2 grouped together, whereas the other samples had no clear boundary.
Although the microbial communities were diverse obviously, physicochemical properties of different samples differed slightly. However, the surprising difference in WS2 had close relationship with the robbery situation, which may affect the results strongly.

Discussion
Bacteria inhabiting in the soil subsurface plays a vital role in organic matter decomposition, especially when soil depth is increasing 14 . In this work, soil and liquid bacteria community structure of the No. 1 Wangshanqiao Chu tomb were studied using 16 S rRNA gene amplicon sequencing, which broadens the understanding of burial conditions.
Contamination was effectively avoided in the samples by using standard coring methods, which guaranteed the reliability. The depths of samples included 4 m and 10 m, 11 m, which overstepped the general study range in many related research field like agriculture and environmental sciences 15,16 . Many factors including low temperature, less oxygen and nutrient availability restricted microbial activity in the deeper soil horizons. Previous studies have thought fewer microorganism can survive under these circumstances 17 . However, the overall microbial communities in burial conditions were unexpectedly diverse, harboring a wide variety of taxa, some of which was common in soil environment 18 . Since the bacterial taxa data is short in archaeological burial conditions, comparative study of community structures can be difficult. It is noteworthy that this burial environment is similar to that of sediment and permafrost. A large number of research were reported on these bacteria with unique function 19,20 , which provides information for our study.
The results indicated that bacterial communities in different samples were surprisingly diverse, which had significant implication of perceiving biodegradation underground. The samples from the robber hole (WS2) varied obviously in community structure both in Alpha and Beta diversity. The predominant phylum in them was Proteobacteria, which differed obviously from others. Proteobacteria which was observed frequently to take the lead in soils 18,21 , is of great importance to global carbon and nitrogen cycling 22 . As nutritional bacteria, it can utilize complex organic matter to metabolize. Moreover, the genus Acinetobacter accounted for nearly the half in the samples from the rob hole. This bacteria is ubiquitous microorganism in soil, water, and sewage 23 , and was referred to occupy nearly 0.001% of the total heterotrophic aerobic population of soil and water 24 . Much research has reported efficient degradation ability of Acinetobacter, widespreadly found not only in decomposing cellulose, hemicelluloses, lignin, but also in cleansing waste water and petroleum pollution [25][26][27] . For instance, an Acinetobacter sp. strain was isolated from the fluid steeping with the erosive bamboo slips that were unearthed from the archaeological site of Three Kingdoms in China 28 . Comamonadaceae and Pseudomonas which occupied in the third and fourth place in the whole population were also reported in studies on the microbial degradation of lignin 28,29 . Especially, Pseudomonas was implicated in the process of influencing the crystalline structure of cellulose microfibrils so that it was closely associated with the decay of wooden relics 30 . As a result, it can be inferred that the predominant bacteria detected in WS2 would play a vital role in degrading organic culture heritage in burial conditions.
Undisturbed caesious mud showed an abundant proportion of Bacteroidetes, which is widely distributed in the environment, within soil, sediments, and sea water. Previous studies suggested that the phylum Bacteroidetes makes up an average of 5% of soil bacterial communities 18 . What is remarkable is that Bacteroidetes is one of the phylum that was frequently associated with permafrost soils 31 . Moreover, Bacteroides which played a predominantly role at the genus level is an important anaerobic member of the gut microbiome 32 and seemed to be rare in soils. Firmicutes was another phylum abundant in undisturbed caesious mud. However, it is deduced that these spore-forming species might be well adapted to cold conditions since it was frequently discovered in permafrost-like habitats 33,34 . Furthermore, it must be stressed that Actinobacteria was the most abundant group in raw soil and liquid as well as in undisturbed caesious mud. Actinobacteria was previously found with a high abundance in permafrost 31 , which was thought to be caused by their maintenance of metabolic activity and DNA repair mechanisms at low temperatures 35 . Nocardioides took the lead in raw soil and liquid at the genus level. It is noteworthy that Streptomyces accounted for nearly 5% in raw soil while the proportion was few in caesious mud and liquid. Streptomyces was found to have efficient degradation ability of lignin by in vitro assays 36 .
There are several excavated historical tombs famous for their luxuriant organic archaeological remains such as fragile silk, lacquer in humid region of China. These tombs concentrated regularly on specific area like Jingzhou and Changsha, which attracted archaeologists to search for answer. Generally speaking, silk and other organic matter excavated from No. 1 Wangshanqiao Chu tomb weren't preserved well. But the preservation conditions varied with shifts on microenvironment. The straw mat near undisturbed caesious mud was better preserved. Silk fabrics found in caesious sludge soil degraded seriously, losing their mechanical strength and even turning into powder, while silk fabrics floating in outer coffin liquid were much better.
The study indicated that burial cultural heritage preservation conditions has connections with their bacterial community distribution. We can speculate that burial conditions of No. 1 Wangshanqiao Chu tomb was favourable for organic matter maintenance at the beginning with anaerobic environment according to the experimental results and previous archaeological discovery nearby. The bacterial tended to degrade slowly in low-oxygen environments and the dominant species may have less capability in degrading organic material like wood and silk. Nevertheless, when the balance was broken due to robbery, the community structure changed observably. Proteobacteria which possessed high degradation ability dominated in whole bacterial, with the degree of biodeterioration aggravating. Later, precipitation seeped into the tomb and was filled with the whole outer coffin. The water may be filtered by soil and then cleansing slowly during a long time, leaving microbial community observed nowadays similar to the raw soil and protecting some precious organic historical relics.
Our current knowledge on molecular microbial diversity in burial conditions is extremely poor. The soil microbial community has arguably the highest level of prokaryotic diversity of any environment 37 . Currently, less than 1% of this diversity can be cultivable by traditional techniques 38 . Fortunately, high-throughput sequencing technologies provide a new method to study the burial environment. We attempted to distinguish the bacteria community distribution in a case study applying 16 S rRNA gene amplicon sequencing and drew interesting conclusions successfully.
This method contributes to fetch information which would never recover after excavated. Firstly, the information will play an important role in research on degradation mechanism of precious culture heritage both organic matter and metals in the future, and thus explore potential cultural information of which the value may be underestimated nowadays. The ignored information would contribute to explain complex archaeological phenomenon. For instance, the results indicated that ancient people had realized how to utilize burial conditions to prevent decomposition. Secondly, this method evaluates likelihood of survival about organic matter like silk fabrics in terms of microbial activity. Some indicator may predict where burial microbial condition is likely to preserve treasures before excavation. Thus, it will lead to select appropriate conservation in archaeological site according to predicting objects condition before excavation, which will benefit conservation and cost reduction. Thirdly, the study indicates that robbery has a serious negative impact on burial remains, far more than taking away treasures. Further study can draw public attention towards preservation. Moreover, the results can assess if rob hole reaches the aim successfully before excavation. Lastly, the large organic artifacts tend to reburial in situ for protection nowadays in lack of economic resources for excavation and conservation, especially in Europe 39,40 . It is no doubt that data of burial microbial environment will provide reference for management.

Methods
Sample Site. The study took place at the No. 1 Wangshanqiao Chu tomb in Jingzhou, Hubei Province. The tomb was built about 2300 years ago in Warring States. Jingzhou witnessed a Golden Age of Chu Kingdom. Plenty of sites of Warring States were found in Jingzhou, and many large noble tombs were also buried here. No. 1 Wangshanqiao Chu tomb is one of the highest grade and largest scale Chu noble tombs that has ever been excavated. It locates in the lower Yangtze region. Figure 5 shows the appearance of the tomb and the arrangement is a typical representation of Chu tomb. However, the large tomb was seriously robbed around Qin-Han Dynasties which caused the turbid water to flow into and stuff the chamber, but a large amount of exquisite cultural relics still survived including bronze, lacquer, jade, textile 41 . Renowned historical tombs were discovered in the district nearby, including No. 1 Mashan Chu tomb which is called treasure-house filled with silk and No. 1 Wangshan Chu tomb which is well-known for Sword of Goujian. Unfortunately, some organic matter, for instance, the silk that was unearthed from No. 1 Wangshanqiao Chu tomb is influenced by robbery (Fig. 5). Although burial environment were disturbed in one spot, the previous burial conditions which are considered as favourable soil conditions for the preservation of archaeological remains and far away from robber hole have been well-retained before excavation. Hence this tomb is a superexcellent case for our study.
Undisturbed soil was sampled by standard coring methods while scene excavation was moving to the 5 meters above outer coffin. Outer Coffin liquid was sampled once the outer coffin board was slung (Table 3). Detailed plots were marked in Fig. 6. The number of samples is influenced by archaeological excavation limit. Soil and liquid for sequencing analysis were stored at − 20 °C, and soil and liquid for other physicochemical properties applications were stored at 4 °C.
Scientific RepoRts | 7:41691 | DOI: 10.1038/srep41691 Soil Chemical Analysis. Soil moisture was determined on fresh soil material after drying for 24 h at 105 °C.
Soil pH was estimated on 1:2 soil/deionized water extracts which was prepared by passing through a 100 mesh screen. Conductivity of 1:5 soil/water extracts was measured with conductivity meter to determine the soil salinity. Total carbon and total nitrogen contents were measured using an element analyzer (Elementar, Germany).  Data Analysis. Paired-end reads from the original DNA fragments were merged using FLASH 42 . Paired-end reads were assigned to each sample according to the unique barcode. The sequences with high quality (length > 200 bp, without ambiguous base 'N' , and average base quality score > 30) were screened for chimeraschecking using UCHIME algorithm 43 . Sequences were clustered into operational taxonomic units (OTUs) at 97% identity threshold using UPARSE-OUT ref algorithms 44 . Representative sequences were picked for each OTU. Taxonomy were assigned using the greengenes database 45 and Ribosomal Database Project classifier 46 . Representative sequences were aligned using PyNAST 47 . In case of the influences of sequence depth on community diversity, the OTU table was rarified to make all samples holding the same sequence number. All data analyses were performed using R 3.3.0 48 . Beta diversity metrics were calculated in Vegan 49 . Unweighted and weighted Unifrac distances were calculated in QIIME 50 . Principal Coordinate Analysis (PCoA) were performed using package Ape 51 . Some pictures were made using SPSS19.0 (IBM Corp, Armonk, NY, USA).