Experimental Study on the Anti-Erosion of the Exterior Walls of Ancient Rammed-Earth Houses in Yangjiatang Village, Lishui

Abstract

Yangjiatang Village traces its origins to the late Ming and early Qing dynasties. It has evolved over more than 400 years of history. There are 78 rammed-earth buildings left, making it one of the most complete and largest rammed-earth building complexes in East China. This study investigated the traditional rammed-earth houses in Yangjiatang Village, Songyang County, Zhejiang Province. By combining field investigation, microscopic characterization, and experimental simulation, we systematically revealed the erosion resistance of rammed earth in a subtropical humid climate was systematically revealed. Using a combination of advanced techniques including drone aerial photography, X-ray diffraction (XRD), microbial community analysis, scanning electron microscopy (SEM), and soil leaching simulations, we systematically revealed the anti-erosion mechanisms of rammed-earth surfaces in Yangjiatang Village. The study found that (1) rammed-earth walls are primarily composed of Quartz, Mullite, lepidocrocite, and Nontronite, with quartz and lepidocrocite being the dominant minerals across all orientations. (2) Regulating the community structure of specific functional microorganisms enhanced the erosion resistance of rammed-earth buildings. (3) The surface degradation of rammed-earth walls is mainly caused by four factors: structural cracks, surface erosion, biological erosion and roof damage. These factors work together to cause surface cracking and peeling (depth up to 3–5 cm). (4) This study indicates that the microbial communities in rammed-earth building walls show significant differences in various orientations. Microorganisms play a dual role in the preservation and deterioration of rammed-earth buildings: they can slow down weathering by forming protective biofilms or accelerating erosion through acid production.

1. Introduction

1.1. Research Background

As one of the oldest forms of architecture in China, rammed-earth architecture can be traced back to the Neolithic Age and is a critical, a vital material carrier of Chinese civilization [1,2,3]. Rammed-earth architecture ranks among the oldest building forms in China, with origins traceable to the Neolithic Age. It represents a crucial material carrier of Chinese civilization [4,5,6]. It is a low-carbon and energy-saving building material. Before the emergence of modern building materials such as cement, polymers, and reinforced concrete, rammed earth was a traditional building material widely used in many parts of the world [7,8]. This form of architecture, which uses raw earth as the primary raw material, is made by layering and compacting materials such as clay, sand, and lime and has significant ecological advantages and regional adaptability. The production process of rammed-earth architecture embodies the wisdom of the ancient working people. Its typical process includes soil selection → ingredients (adding organic adhesives such as glutinous rice paste and brown sugar) → formwork installation → layered compaction (each layer is about 15 cm) → maintenance → demolding and other links (Figure 1). This building type offers advantages such as low carbon emissions, environmental protection, low cost, and thermal insulation, while also embodying rich regional culture and historical memory.

This study selected Yangjiatang Village in Songyang County, Lishui City, Zhejiang Province, as the research object. The village dates back to the late Ming and early Qing dynasties, with a history of more than 400 years. There are 78 rammed-earth buildings, which is one of the most complete and largest rammed-earth building complexes in East China [9,10]. Yangjiatang Village was designated as a Chinese Traditional Village in 2014, as part of the third batch. Its rammed-earth buildings have the following significant features: (1) the wall thickness is 50–80 cm, forming a unique microclimate regulation system; (2) local yellow soil is used, and oyster shell powder and other materials are added to improve durability; and (3) the building layout is based on the terrain, reflecting the traditional construction wisdom of “harmony between man and nature”. Driven by the subtropical monsoon climate, this region receives an average annual precipitation of 1837 mm and experiences a high acid rain frequency of 42.3%. These climatic factors collectively contribute to the severe erosion observed in the rammed-earth buildings. According to our survey, 37% of the village’s rammed-earth buildings have suffered structural damage in the past ten years, 12 of which are rated as dangerous. This situation poses a serious safety threat to older people and our adults living alone who still live in these buildings (accounting for 68% of the current population) and also puts precious architectural the cultural heritage at risk of irreversible damage.

1.2. Literature Review

The conservation of rammed-earth heritage is now a pressing priority, as accelerated deterioration driven by climate change and environmental factors threatens these structures globally. In Yangjiatang Village, Lishui, ancient rammed-earth houses are particularly vulnerable to rainfall, wind, and biological colonization, reflecting challenges documented across global heritage sites [11].

In recent years, scholars have conducted in-depth discussions on the deterioration mechanism of rammed-earth buildings. In humid subtropical regions, research has established that the primary mechanisms leading to soil disintegration are dry-wet cycles and capillary water migration. These processes are significantly exacerbated by the colonization of biological communities such as mosses and lichens, which accelerate surface softening and structural degradation. On this basis, some studies attempt to reveal the coupling effect of climatic conditions and biological interactions through laboratory simulations and on-site observations, thereby providing a scientific basis for protective measures [12,13].

At the forefront of protective technology research are methods for material modification and surface treatment aimed at enhancing durability. Traditional methods mostly employ lime or silicate stabilizers to enhance the rain erosion resistance of walls. The introduction of novel nanomaterials into rammed-earth conservation has drawn attention owing to their excellent hydrophobicity and minimal alteration to appearance. However, long-term compatibility, reversibility, and potential impact on the original materials remain the focus of academic debate [14].

International research provides a valuable comparative perspective. For instance, in Europe, sacrificial mud coatings are often used as replaceable “sacrificial layers” to withstand wind and rain and maintain the authenticity of the material. In Australia, strategies to mitigate water erosion often prioritize passive design solutions, including extended roof overhangs and enhanced drainage systems. In Latin America, community participation and simple protection measures have proven to be equally effective in arid or semi-arid environments. These experiences indicate that integrating local traditional knowledge with modern scientific methods may be the key approach to achieving sustainable protection [15,16].

Meanwhile, the development of monitoring and diagnostic technologies has also significantly enhanced the level of research. Remote sensing, 3D laser scanning, and digital photogrammetry provide high-precision data support for the deterioration monitoring of rammed-earth buildings, making it possible to track and quantitatively analyze the erosion process dynamically. Such technologies provide a new methodological basis for the systematic protection of large-scale sites [17,18].

In conclusion, significant progress has been made in existing research regarding the analysis of deterioration mechanisms, the development of material modification strategies, and the advancement of technical conservation methods. However, in-depth exploration is still needed in the aspects of long-term effect evaluation, regional adaptability, and cultural inheritance. The sustainable protection of Yangjiatang Village hinges on three critical efforts: advancing eco-material applications, achieving a harmonious integration of traditional and modern techniques, and implementing monitoring-based data management. These efforts are fundamental to guaranteeing both the physical security of structures and the preservation of cultural identity.

1.3. Problem Statement and Objectives

Therefore, the restoration and protection of rammed-earth buildings have multiple values: from a cultural perspective, it is a living inheritance of traditional construction skills; from a technical standpoint, the ecological wisdom it contains is of enlightenment significance to the development of modern green buildings; and from a social perspective, the protection work improves the living conditions of villagers and promotes the growth of rural tourism. A professor from the School of Architecture at Tsinghua University emphasized, “The rammed-earth buildings of Yangjiatang Village are textbooks of living history. Their preservation is tantamount to safeguarding the cultural DNA of the Chinese nation”.

The purpose of the study is to deeply analyze the erosion resistance of rammed-earth buildings in East China. Using a multi-technique approach, including advanced drone aerial photography, XRD, microbial community analysis, SEM, and soil leaching simulations, this study systematically elucidates how microbial communities and mineral phase changes influence the durability of rammed earth. The methods selected for this study—XRD, SEM, microbial diversity analysis, and soil leaching simulations—were chosen to comprehensively characterize the material properties and biological factors influencing erosion resistance. These methods are well-established for mineral identification, microstructural visualization, and community profiling. While XRF could provide elemental composition, it was not prioritized as XRD already identified the key mineral phases. Future research could incorporate mechanical strength tests, porosity analysis, and specific surface area measurements to correlate microstructure with macroscopic durability further. At the same time, based on the innovative idea of integrating traditional craftsmanship with modern technology, a set of low-intervention restoration technology systems suitable for the characteristics of mountain villages is planned. This study is not only committed to solving the key scientific problems currently faced in the protection of rammed-earth buildings but also focuses on exploring the potential protection value of natural materials, providing innovative solutions that are both scientific and practical for the sustainable protection of traditional rammed-earth buildings, and establishing a replicable and popularizable technical paradigm for the protection and restoration of similar cultural heritage.

2. Materials and Methods

2.1. Study Area: Yangjiatang Village in Lishui City

Yangjiatang Village is in Sandu Township, Songyang County, Lishui City, Zhejiang Province (Figure 2). It is located between 119°10′ and 119°42′ east longitude and 28°14′ and 28°36′ north latitude. It is surrounded by mountains at an altitude of 400–600 m. Songyang County is the area with the most significant number of ancient villages and the best preservation in East China [19,20]. Songyang County boasts over 100 traditional villages with well-preserved structures and diverse styles, 78 of which are designated as national-level conventional villages. These settlements—including alpine, stepped, waterside, terraced, flat-valley, and Hakka types—are scattered across its mountains. Among them, Yangjiatang Village, established around 1630 AD during the late Ming and early Qing dynasties, has a history of nearly 400 years (Figure 3). It is the most representative traditional rammed-earth building settlement in southwestern Zhejiang. In 1655, the 12th year of the Shunzhi reign of the Qing Dynasty, the Song family settled in this area. They developed it rapidly into a high-mountain village where their clan became dominant. Despite this demographic shift, the original name, Yangjiatang Village, has been retained to the present day. The town has more than 300 residents, with the Song family being the most prominent surname in the area. They are descendants of Song Lian, the first civil official of the Ming Dynasty (1368–1644). The village is terraced along the mountain slope, and its overall layout remains well-preserved. are 78 rammed-earth buildings from the Ming and Qing Dynasties (1368–1840) to the Republic of China (1912–1949), with a building density of 85%. Zhejiang Province has identified 11 of these buildings as historical and cultural. The most representative buildings, such as “Song’s Ancestral Hall” and “Yang’s Hall” (Figure 4), have been listed as cultural relics protection units in Zhejiang Province.

The region is characterized by a typical subtropical monsoon climate, featuring an average annual temperature of 17.5 °C, substantial precipitation reaching 1837 mm per year, consistently high relative humidity maintained above 80% throughout the year, and relatively limited sunshine hours totaling only 1750 yearly (Figure 5). Although this warm and humid climate is conducive to agricultural production, it also poses a severe challenge to the protection of rammed-earth buildings. Data indicate that 37% of rammed-earth buildings have been damaged by rainwater erosion in the past decade, 12 of which were subsequently designated as dangerous.

The rammed-earth buildings of Yangjiatang have a special status in the residential system of southern Zhejiang, and their construction techniques integrate the characteristics of She and Han cultures. A specialized “three-in-one soil” formulation is employed in the building’s construction. This formulation leverages a base of local yellow soil (≈70%), which is enhanced with fine sand (20%), slaked lime (8%), and oyster shell powder (2%) to improve its properties. Tests confirm that this specific composition achieves a compressive strength of 2.5 MPa and a permeability coefficient lower than 10⁻⁶ cm/s, underscoring its superior durability. The thickness of the building wall is generally between 50 and 80 cm, which not only has good thermal insulation performance but also forms a unique microclimate regulation system.

However, the continuous high humidity environment and frequent precipitation have caused serious threats to rammed-earth buildings. On-site investigations show that the south-facing wall is most severely eroded by acid rain, with a surface peeling depth of 3–5 cm; the north-facing fence is in a humid environment for a long time, and microbial erosion leads to a loose structure; while the east–west wall is mainly affected by temperature difference stress, and the crack width generally exceeds 1 cm (Figure 6).

It is touching that there are still 23 households and 38 older adults and older people in the village who stick to these old houses passed down from their ancestors. Grandpa Yang, 78 years old this year, said, “This house was built by my great-grandfather. It is warm in winter and cool in summer, and it is comfortable to live in”. This persistence not only reflects the continuity of cultural heritage but also reflects the villagers’ profound attachment to their traditional lifestyle. In recent years, the Songyang County Government has attached great importance to the protection of Yangjiatang Village and has invested more than 12 million yuan in special funds to implement projects such as “Save Old Houses” and “Protection and Utilization of Traditional Villages”. In 2020, the village was included in the Important Architectural Heritage Protection List of Zhejiang Province, and in 2022, it was awarded the “China Traditional Village Digital Museum” construction pilot. These measures not only protect precious architectural heritage but also inject new vitality into rural revitalization.

2.2. Field Survey and Sample Collection

Yangjiatang Village is in a typical mountainous terrain in southwestern Zhejiang. The village is terraced along the mountain slope, with buildings of varying heights blending seamlessly into the natural environment. Sampling was conducted during clear weather conditions with temperatures ranging from 15 °C to 22 °C and relative humidity between 70% and 85%. These parameters were recorded to ensure consistency and to correlate with subsequent laboratory simulations. To fully understand the status of rammed-earth buildings, we conducted systematic aerial photography surveys using a DJI MINI 4 PRO drone (Shenzhen Dajiang Innovation Technology Co., Ltd., Shenzhen, China). To adapt to the cloudy and foggy conditions of Yangjiatang Village, the camera has been set to aperture f/2.8, shutter speed 1/100–1/500 s, and ISO 100-800. The flight altitude was maintained at 11–15 m to capture both building overviews and wall details. The drone’s flight altitude is strictly controlled within the range of 11–15 m, which can not only capture the overall shape of the building but also clearly identify the detailed features of the wall. For the high-resolution documentation of severely damaged rammed-earth walls, the image acquisition distance was reduced to 3–5 m. This close-range approach, supplemented by the camera’s digital zoom function, enabled the clear capture of millimeter-scale crack details, which is a critical step for subsequent quantitative analysis.

The ground survey adopted a multi-device collaborative working mode: using the DJI POCKET 3 (Shenzhen Dajiang Innovation Technology Co., Ltd., Shenzhen, China) primary camera for 4K high-definition shooting and iPhone Smartphone (Foxconn, Shenzhen, China), Huawei Smartphone (Huawei Technologies Co., Ltd., Shenzhen, China), OPPO Smartphone (Guangdong Oppo Mobile Telecommunications Corp., Ltd., Dongguan, China), and other mobile devices for auxiliary shooting to ensure the integrity and comparability of the image data. Using this ground-to-earth integrated survey method, we created a digital image library featuring high-definition images of rammed-earth buildings (Figure 7).

During the field investigation, we systematically recorded various typical damage modes: (1) Structural cracks: mainly distributed at the corners of the wall, with the maximum crack width reaching 12 mm and extending for more than 2 m, showing obvious shear crack characteristics; (2) Surface erosion: The peeling depth of the south-facing wall is generally 3–5 cm, exposing the internal rammed-earth layer; (3) Biological erosion: The moss coverage rate of the north-facing wall is as high as 65%, and the black mold area accounts for 40% of the wall area; (4) Roof damage: About 28% of the buildings have tile slippage, causing rainwater to erode the top of the wall directly. Researchers have shown great interest in the formation mechanisms of these diseases, particularly regarding the differences in damage characteristics of walls facing different directions, which suggests a complex environmental mechanism.

Sampling was conducted with prior consent from homeowners. A scientific sampling plan was designed: (1) Samples were collected from four orientations (E, S, W, N). (2) Sampling was performed 1.5 m above ground, avoiding openings. (3) Using a stainless-steel knife, the surface weathering layer (2 mm) was removed before collecting a 10 × 20 × 10–20 mm cross-section sample (Figure 8).

The sample processing process strictly complies with relevant specifications: (1) Solid samples are wrapped in acid-free paper and placed in polyethylene specimen boxes; (2) powder samples are placed in 15 mL centrifuge tubes. All containers have been marked with unique codes and record metadata, including the location of the sampled house, the direction of the sampled wall, and the sampling time (Figure 9). To ensure the freshness of the samples, the time from collection to laboratory analysis is strictly controlled within 7 days. We use anti-seismic packaging and temperature recorders during transportation to prevent any changes in the sample properties. Through field investigations, we have obtained a wealth of physical samples and image data. This foundational research provides critical support for the informed conservation of traditional rammed-earth buildings and establishes a robust framework for future investigations into the long-term performance of these materials in humid subtropical environments.

2.3. Analysis Methods and Process

2.3.1. X-Ray Diffraction (XRD) Analysis

Mineral composition was determined by X-ray diffraction (XRD) using a Rigaku SmartLab diffractometer (Rigaku Holdings Corporation, Tokyo, Japan) with Cu Kα radiation (λ = 1.5406 Å), operated at 40 kV and 40 mA. Data were collected over a 2θ range of 5° to 70° with a step size of 0.02° and a counting time of 2 s per step. Quantitative phase analysis was performed by using the Rietveld refinement method. The XRD analysis was performed by using a Rigaku SmartLab high-resolution diffractometer equipped with a Cu Kα X-ray source (wavelength λ = 1.5406 Å) operating at 35 kV and 25 mA. The measurements were conducted in a 2θ range from 5° to 70° with a step size of 0.02° and a scanning speed of 2° per minute. The diffractograms were collected in continuous mode, and phase identification was carried out by comparing the obtained patterns with the International Centre for Diffraction Data (ICDD) PDF-4+ database. Quantitative analysis of mineral phases was performed using the Rietveld refinement method with the software package provided by the instrument manufacturer. By analyzing the XRD diffractograms of the samples, the mineral composition and stability of the surface materials can be understood, which can provide an essential basis for interpreting the durability texture of traditional building materials.

2.3.2. Detection of Diversity Composition of Microbial Communities

This study used high-throughput sequencing technology to systematically study the microbial communities of rammed-earth building walls in Yangjiatang Village. The 16S rRNA gene V3-V4 region of samples from walls in four directions (East, South, West, and North) was sequenced using the Illumina NovaSeq platform, and an average of 78,452 valid sequences (length from 400 to 440 bp) were obtained. Data analysis was performed using QIIME2 (version 2019.4), and representative ASV sequences were obtained after DADA2 denoising. Species annotation was based on the SILVA 138 database, using the Naive Bayes classifier with a confidence threshold of 0.7. The study focused on analyzing microbial alpha diversity (Shannon, Chao1 index), beta diversity (PCoA, UPGMA clustering), and functional prediction (PICRUSt2).

2.3.3. Scanning Electron Microscope (SEM) Analysis

In this study, we performed high-resolution microscopic morphology analysis on rammed-earth samples using a German ZEISS GeminiSEM 360 field emission scanning electron microscope (Carl Zeiss AG, Oberkochen, Germany). The device is equipped with a secondary electron detector (SE) and a backscattered electron detector (BSE), which can achieve an ultra-high resolution of 1.2 nm at an accelerating voltage of 10 kV. The sample was pretreated by coating a 5 nm thick gold film with an Quorum Q150T ES ion sputtering coater (Quorum Technologies Ltd., Laughton, UK) to improve conductivity. During the analysis, the working distance was set to 5 mm, the beam intensity was 100 pA, and the magnification was adjusted from 200× to 20,000×. The particle morphology, pore structure, and interface characteristics of the samples facing the wall in different directions were focused on. Finally, the acquired images were subjected to quantitative analysis in ImageJ 1.54p bundled with 64-bit Java 8 to measure key parameters: particle size distribution (n > 500), porosity (calculated by threshold segmentation), and surface roughness (determined using a grayscale gradient algorithm).

3. Results

3.1. XRD Analysis Results

The XRD diffractograms revealed distinct mineralogical compositions across the four orientations (Figure 10 and Figure 11). Four main minerals are clearly marked in Figure 10 and Figure 11: Quartz (SiO₂), Mullite (Al₆Si₂O₁₃), Lepidocrocite (γ-FeOOH), and Nontronite (Na₀.₃Fe₂(Si,Al)₄O₁₀(OH)₂·nH₂O).

Quartz is a very stable mineral with high hardness, durability, heat resistance, and water resistance, which can improve the overall stability of the material. As can be seen from Figure 10, the Quartz diffraction peaks are apparent in all samples, indicating it is not only a significant component but also the dominant phase in the material. Among them, the proportion in the north wall and the east wall is higher than that in the other two walls (the north wall accounts for 26% and the east wall accounts for 21%). It is noteworthy that while the diffraction peak intensity of quartz appears particularly prominently in the east sample diffractogram (Figure 11), its quantified percentage (21%) is more comparable to other orientations when considering the complete mineral composition (Figure 10). This observation can be attributed to factors influencing peak intensity beyond mere abundance, such as the crystallinity and preferred orientation of quartz crystals within the sample. Therefore, quantitative analysis provides a more reliable measure of the actual mineral content, confirming that quartz is a dominant and critical stabilizing phase across all wall orientations. From the diffraction peaks of the north wall and the east wall, the peak intensity of the east wall is higher, indicating that Quartz is easier to crystallize.

Mullite is a high-temperature phase mineral with excellent thermal and chemical stability. It can be used for thermal insulation of building exterior walls, interior walls, and partition walls, reducing the overall heat loss and energy consumption of buildings. From the figure, Mullite is concentrated on the south wall because it is exposed to more sunlight. Rammed-earth buildings are prone to shrinkage and cracking when drying. The micro-expansion characteristics of Mullite can partially offset the shrinkage force, improving overall stability. Mullite acts as a humidity buffer due to its microporous network, which actively manages moisture by absorbing and slowly releasing it. Synergizing with the natural breathability of rammed-earth construction, this process aids in stabilizing the humidity levels specifically on the south-facing wall, thereby positively influencing the microclimate. Conversely, iron hydroxide is an environmentally sensitive mineral prone to phase changes, potentially causing material weathering and chromatic alterations. The significantly higher content detected on the south wall (24%) versus the north wall (14%) underscores the role of orientation-dependent factors, such as solar exposure and wet-dry cycles, in driving deterioration. The south-facing wall experiences significant diurnal temperature fluctuations, which promote water vapor condensation. This moisture, in turn, accelerates the oxidation of iron compounds within the material. Consequently, protective strategies for the south wall should prioritize enhancing waterproofing capabilities to mitigate the disruptive expansion and contraction cycles associated with iron hydroxide. Conversely, the north-facing wall is characterized by lower temperatures, higher humidity, and reduced sunlight exposure. These conditions suppress the oxidation of Fe²⁺, allowing iron hydroxide to persist in a colloidal state.

Nontronite is a layered silicate mineral with specific water absorption and expansion properties. Its presence will cause the material to change in volume when the humidity changes, thus affecting the stability of the structure. The east wall is affected by the monsoon climate, and the humidity changes significantly. Rainwater infiltration causes Nontronite to expand, which in turn affects the stability of the wall. Currently, its proportion constitutes only 9%, resulting in a negligible impact on the wall. Furthermore, the high feldspar content (46%) in the east wall effectively restricts these potential expansion spaces. The composition of the north sample is 14% iron hydroxide, 26% quartz, and 60% andalusite. The substantial quartz content in the northern sample contributes to the material’s enhanced hardness, durability, and cold resistance. The colloidal state of iron hydroxide can fill the gaps of andalusite and improve impermeability. The composition of the west sample is 10% iron hydroxide, 18% quartz, and 41% kaolinite. In the west sample, the Quartz content is significantly higher than that of others, with extremely high hardness and durability and excellent overall stability. However, in the diffraction peak diagram, it can be seen that the peak intensity of the quartz on the west wall is not high, which may be caused by wind erosion. It is necessary to strengthen the weathering protection of the wall surface. At the same time, the west wall is more humid than the east wall. Kaolinite has strong water absorption and a higher proportion. The composition of the east sample is Quartz accounting for 21%, Nontronite accounting for 9%, Feldspar accounting for 46%, kaolinite accounting for 21%, and iron hydroxide accounting for 4%. Quartz and feldspar can form a skeleton support body with good overall stability. The composition of the south sample is Quartz accounting for 14%, dolomite accounting for 9%, calcite accounting for 9%, Mullite accounting for 44%, and iron hydroxide accounting for 24%. Quartz is still the main mineral, which ensures the hardness and durability of the material. In addition, due to the water absorption and expansion of nontronite, the volume of the material may change when the humidity changes. Calcite and dolomite are carbonate minerals that can neutralize acidic rainwater, thereby improving the erosion resistance of the wall surface.

3.2. Analysis of the Microbial Diversity Detection Data

We analyzed the microbial communities from four wall orientations using 16S rRNA gene sequencing. The data was processed with QIIME2 software to identify bacterial types and their abundances. As shown in

Table 1. Analysis of DADA2 sequence denoising results.

Sample lD	Input	Filtered	Denoised	Merged	Non-Chimeric	Non-Singleton
North	60,951	51,354	51,220	50,881	50,630	50,629
South	76,418	63,719	60,721	47,966	44,304	44,099
West	82,270	68,601	68,543	68,104	67,048	67,048
East	67,285	56,416	55,846	52,831	49,873	49,823

, the original input sizes were 60,951, 76,418, 82,270, and 67,285 sequences. After filtering, these counts decreased to 51,354, 63,719, 68,601, and 56,416, respectively, indicating that preliminary quality control removed approximately 15%–17% of low-quality sequences. Following denoising, the sequence counts slightly varied to 51,220, 60,721, 68,543, and 55,846, respectively. During sequence merge, north retained 50,881 sequences, south significantly reduced to 47,966, west slightly decreased to 68,104, and east dropped to 52,831. After removing chimeras and singletons, the final valid sequence counts obtained for each sample were 50,629, 44,099, 67,048, and 49,823, respectively. These counts will be used for subsequent taxonomic analysis and diversity assessment. Notably, the west sample retained the highest proportion of its sequences (81.5%); in contrast, the south sample suffered the greatest attrition, with a final count representing only 57.7% of the original. This disparity may reflect variations in DNA quality and sequence complexity among different samples, as well as differences in their specific microbial community structures.

The research findings revealed significant differences in microbial community composition richness among the four samples. The south sample demonstrated the highest diversity across all taxonomic levels, with 23 phyla (Phylum), 54 classes (Class), 108 orders (Order), 152 families (Family), 216 genera (Genus), and 55 species (Species) detected. Consequently, the south sample possesses an exceptionally rich and complex microbial community structure, as evidenced by its total of 608 taxonomic units, a count notably higher than that of the other three samples. The east sample exhibited moderate taxonomic diversity, identifying 13 phyla, 25 classes, 46 orders, 61 families, 77 genera, and 40 species, totaling 262 taxonomic units. Notably, the east sample showed a relatively high number of species-level taxonomic units (40), second only to the south sample, suggesting a higher abundance of species-identifiable microbial groups. The north sample displayed relatively low taxonomic diversity, detecting eight phyla, 15 classes, 31 orders, 35 families, 41 genera, and seven species, totaling 137 taxonomic units. The west sample exhibited the lowest taxonomic diversity, identifying seven phyla, 10 classes, 25 orders, 26 families, 31 genera, and six species, totaling 105 taxonomic units. The bar chart visually demonstrates the distribution patterns of microbial taxa across taxonomic levels among the four samples. Sample south exhibits the highest overall abundance, with a significantly greater number of taxa across all taxonomic ranks. This dominance is particularly evident at the order, family, and genus levels, as indicated by the prominent blue, orange, and purple sections in the figure. Samples east and north possess relatively low taxonomic diversity. The west sample, however, is characterized by the least abundant and least diverse microbial community. These findings clearly reveal distinct differences in microbial community composition: The south demonstrates the richest biodiversity, whereas the west shows the lowest diversity (Figure 12).

This study looked at the tiny living things on the walls of rammed-earth buildings in four directions (east, south, west, and north) in Yangjiatang Village, showing a significant link between the types of microbes present and how well the buildings resist erosion.

As shown in Figure 13, this bar chart displays the relative abundance distribution of microbial communities at the genus level across four distinct samples (North, South, West, and East). The results reveal significant differences in microbial community composition among the sample points. In the North sample, microbial diversity is relatively low, dominated by “Others” taxa with nearly 100% relative abundance, indicating a relatively homogeneous microbial community in this region. The South sample exhibits richer microbial diversity, featuring multiple genera such as Sphingomonas, Pseudonocardia, and Rubrobacter, collectively constituting approximately 40% of the relative abundance, while “Others” taxa account for about 60%. In the West sample, Actinomycetospora was the dominant genus, accounting for approximately 70% of the relative abundance. It was followed by an orange-marked genus (possibly Rubrobacter or another species), which constituted about 25%. The East sample demonstrates the most diverse microbial composition, including Arthrobacter, Pseudonocardia, Rubrobacter, Bacillus, and other genera. These dominant bacterial genera account for the vast majority (80%) of the microbial community, while the “Others” category represents a minor fraction (20%). Overall, microbial diversity gradually increased from the North to the East sample. Concurrently, the community structure transitioned from a simpler to a more complex composition. In contrast to the higher diversity of the East sample, the West sample was marked by a clear predominance of specific bacterial taxa.

As shown in Figure 14, the composition of microbial communities (OTU/ASV) across the four sampling sites (North, South, West, and East) exhibited pronounced regional specificity, with limited overlap among them. A network analysis revealed that the South sample demonstrated the highest microbial diversity, containing 1296 unique OTUs accounting for 69.6% of total detected OTUs. The East sample followed closely, with 387 unique OTUs accounting for 20.78% of the total. In contrast, the West and North samples contained significantly fewer unique OTUs, at 74 (3.97%) and 71 (3.81%), respectively. Shared OTUs between any two sites were strikingly limited. The highest overlap was observed between the South and East samples (21 OTUs, 1.13%), while all other pairwise comparisons, such as North–South (5 OTUs, 0.27%) and South-West (2 OTUs, 0.11%), showed even lower numbers. Notably, only one core microbiome OTU (0.05%) was shared across all sites, highlighting significant regional variations in microbial composition.

As shown in Figure 15, this bar chart illustrates the relative abundance distribution of microbial communities at the phylum and genus levels across different sampling points and their combinations. At the phylum level (left panel), Actinobacteria (blue-green) and Proteobacteria (yellow) are dominant phyla in most samples. Actinobacteria was the dominant phylum across all samples, but its abundance varied: it was overwhelmingly dominant in the North sample (~95%), while the predominant phylum remained at a significant level (~75%) in the West and East samples. Notably, Proteobacteria show a significant increase in relative abundance in the South samples and their combinations with other samples. Firmicutes (light purple) also exhibit high abundance in East and its combinations. Significantly, Proteobacteria dominate the shared microbial communities between the North and the South, indicating that these two sampling sites share predominantly Microbomonas bacteria. At the genus level (right panel), differences become more pronounced. Rubrobacter (blue-green) constitutes a significant proportion in the North samples, whereas Actinoplanes (yellow) and Sphingomonas (pink) demonstrate higher abundance in the South samples. In the west sample, Methylobacterium (purple) and Actinomycetospora (pink) dominate. The East sample’s defining feature is its predominance of bifidobacteria (red). In shared microbial communities, Bacillus and Actinoplanes are the primary shared components between the North and South. Rubrobacter predominates in the microbiota shared between the North and West samples, while Actinomycetospora and Blastococcus constitute the core community shared by the South and East samples. Notably, Bacillus constitutes the main shared microbial component among the North, South, and West, whereas Delftia dominates the shared microbiota between the North, South, and East.

As shown in Figure 16, the heat map illustrates the relative abundance distribution and clustering relationships of microbial communities at the genus level across four sampling sites (North, South, West, and East). The color gradient from blue (low abundance) to red (high abundance) reflects relative abundance variations in genera across different samples. The heat map analysis reveals significant differences in microbial community structures among the four sites, each exhibiting characteristic dominant bacterial genera.

In the South sample, high abundance (red) is demonstrated by over 30 genera, including Brevundimonas, Micromonospora, Truepera, Bryobacter, Methylobacterium-Methylorubrum, Chloroplast, and Sphingomonas, indicating the highest microbial diversity at this site. The West sample, however, shows dominance by Actinomycetospora, Geodermatophilus, and Rubrobacter as key bacterial genera. The characteristic bacterial genera in the East samples primarily include Paenibacillus, Promicromonospora, Alkalibacterium, Gracilibacillus, Virgibacillus, Oceanobacillus, Marmoricola, Halobacillus, Nocardiopsis, Arthrobacter, Streptomyces, Bacillus, Nocardioides, Pseudonocardia, and Blastococcus. In the North samples, only Saccharopolyspora showed relatively high abundance, while other genera were generally low in abundance. The phylogenetic tree reveals that microbial genera form several major clades: one is predominantly enriched in the South samples, while another is primarily associated with the East samples, with a minor clade (e.g., Actinomycetospora, Geodermatophilus, Rubrobacter) showing high abundance in the West samples. This clustering pattern reflects the selective distribution characteristics of microbial communities under different environmental conditions.

From the perspective of the cluster tree structure, the microbial genera can be divided into several main clusters: one cluster is mainly enriched in the south samples; another is mainly enriched in the east samples; A small number of bacterial genera, such as Actinomycetospora, Geodermatophilus, and Rubrobacter, also showed high abundance in the west samples. This clustering pattern reflects the selective distribution characteristics of microbial communities under different environmental conditions.

In summary, the south-facing wall exhibited the most diverse microbial community, which may contribute to its distinct erosion patterns compared to other orientations.

3.3. SEM Analysis Results

3.3.1. Comparative Analysis of SEM Images

Furthermore, the researchers conducted SEM imaging analysis on samples from the east, south, west, and north, which revealed notable differences in their micromorphology, particle distribution, and surface characteristics (Figure 17).

(1)

The particles in the eastern samples are regular polygons, large in size, with significant size differences, relatively concentrated distribution, and a relatively smooth overall surface.

(2)

In contrast, the particles in the south are more irregular in shape, mostly polygonal in structure, unevenly distributed, and not obviously aggregated. Its surface is rough and accompanied by cracks, indicating that this orientation may have been affected by greater stress during its formation.

(3)

The particle morphology of the western samples is between the eastern and southern ones, mainly round and polygonal, with a large number of total particles; the particle size is relatively consistent, the particle size difference is slight, the distribution is relatively concentrated, and the surface is relatively flat, showing good particle uniformity. The particles in the northern samples are characterized by a fragmented or flaky morphology, a broad size distribution, and an uneven spatial arrangement.

(4)

The particle morphology of the northern samples is relatively dispersed, and the particles are mainly distributed on the right side. The particle size is relatively large. There are microorganisms in the north of samples. The organisms may aggregate small particles into larger clusters through secretions or produce fine fragments by decomposition. This microbial activity could lead to the observed heterogeneous particle size distribution, which may result from uneven microbial corrosion or deposition.

In general, the east and west samples exhibit more regular particle morphology, a more concentrated size distribution, and smoother surfaces, suggesting a more uniform and stable growth environment during their synthesis or processing. The south and north samples exhibit more complex morphology, a wider particle size distribution, and rougher surfaces with cracks, which may be attributed to factors such as crystal orientation, local temperature gradients, or stress concentration. Among them, the particle size in the north is more dispersed due to the influence of microorganisms. These phenomena indicate that the microstructure of the sample in different directions is affected by the synthesis or processing conditions, thus showing obvious anisotropic characteristics.

3.3.2. Particle Size Distribution Analysis

The researchers conducted a statistical analysis of the particle size of the four oriented samples of this rammed-earth wall, and the results are as follows (

Table 2. Particle size statistics in Yangjiatang Village.

Wall Orientation		East Wall	South Wall	West Wall	North Wall
Total particles		109	75	187	130
Particle size information	Average particle size	342.636 nm	361.141 nm	284.309 nm	266.304 nm
	Particle size standard deviation	217.108 nm	220.090 nm	159.866 nm	145.531 nm
	P10	243.434 nm	214.087 nm	218.898 nm	203.984 nm
	P50	755.318 nm	608.584 nm	671.127 nm	596.557 nm
	P90	1267.203 nm	1003.082 nm	1123.356 nm	989.13 nm

Source: statistics by the author.

):

Through statistics, it was found that there were apparent differences in the morphology and particle size of samples at different wall orientations, indicating that the samples may have been affected by different external environments (such as temperature, reaction rate, or stirring rate) during the past rammed-earth synthesis or processing, forming anisotropic structural characteristics (Figure 18). In terms of the total number of particles, the west wall has the most significant number of rammed-earth particles, reaching 187, while the south wall has the fewest, only 75. A larger number of particles may mean that the wall has higher standards in terms of pressure resistance and stability, or more rammed-earth materials were used during construction. The average particle size of the south wall is the largest, at 361.141 nm, and the average particle size of the north wall is the smallest, at 266.304 nm. The difference in average particle size may be associated with localized environmental factors, such as variations in wind force and solar radiation across the different orientations. Larger average particles may make the wall stronger but may also affect the fineness and esthetics of the wall. Using rammed earth with different average particle sizes in different directions may be to balance the physical properties and appearance of the wall. In addition, the standard deviation reflects the degree of dispersion of particle size. The standard deviation of the south wall is the largest, indicating that the particle size of the rammed earth on the south wall varies greatly. The heterogeneity in particle sizes may be attributed to the use of source materials with inherent size variations or to a relatively ad hoc construction methodology. The north wall exhibits the smallest standard deviation, suggesting that its rammed-earth particles have a highly uniform size and may have been subject to more stringent control during construction. P10 means that 10% of the particles are smaller than this particle size. The P10 particle size of the north wall is the smallest, indicating that the north wall has relatively more particles with smaller particle sizes. This may affect the porosity and air permeability of the wall. Finer particles can reduce wall porosity and compromise air permeability, but they may enhance waterproofing performance. P50 is the median, and the P50 particle size of the east wall is the largest. The particle size distribution of the east wall indicates a prevalence of larger mid-range particles, suggesting an overall coarser composition that may contribute to greater wall solidity. The P90 value (the threshold above which only 10% of particles lie) is largest for the east wall, reflecting a higher content of large-sized particles. This particle size distribution likely benefits load-bearing capacity and stability, albeit at the potential cost of greater construction difficulty. These data will help researchers select materials for subsequent repairs and provide a basis for restoration. Figure 19, Figure 20, Figure 21 and Figure 22 show the particle size distribution of rammed-earth particles from the four wall orientations.

3.4. Spray Water Results

In this experiment, we used a TD model electronic scale produced by Yuyao Jinnuo Tianping Instrument Co., Ltd. (Ningbo, China), and an electric sprayer produced by Delixi Electric (Yueqing, China). Original sample E (unrepaired) on the east side, initial reaction: after encountering water, it showed “rapid penetration” characteristics. Water penetrated the sample within 10 s, and a wet halo quickly appeared on the surface, indicating that it has high porosity and a loose structure. After 6 h, particles began to fall off at the edge, accompanied by small cracks extending to the center, forming an “edge collapse-center suspension” trend. Subsequently, the loss of primary structural support led to an inward collapse of the central section, resulting in a honeycombed fracture pattern. After 12 h, it completely disintegrated into a muddy state, with no block residue, and was determined to be a precarious structure (Figure 23a–c).

During the initial water reaction test, the unrepaired south sample (S) exhibited “surface runoff” behavior: water quickly slid down its slope with minimal infiltration, indicating a dense surface layer with low permeability. After 8 h, scouring erosion caused a “cutting-like” collapse at the water flow edges, forming an approximately 2 mm-deep annular groove, while the central 50% of the area remained intact. Following 12 h of testing, the sample’s outer region collapsed, causing stress redistribution and the layered sliding of large soil fragments; the final residual volume was about 35% of the initial volume. Despite this significant material loss, its erosion resistance was greater than that of samples E and N, owing to its dense surface and internal structural support (Figure 23d–f).

West repair sample W (after repair). The repair material (presumably a binder or curing agent) filled the pores and formed a mesh structure, significantly improving the overall bonding strength. The repaired sample demonstrated an approximately 60% reduction in water penetration rate compared to samples E and N, along with the formation of a short-term water film on its surface. After 9 h, small “scale-like” pieces fell off the edge (single piece diameter < 5 mm), and the damage did not extend to the center, showing the characteristics of “local loss-main body stability”. After 12 h, the width of the edge fracture zone was about 1–2 cm, and the central area maintained a high degree of integrity. The erosion resistance was significantly improved compared with the unrepaired sample, verifying the effectiveness of the repair process (Figure 23g–i).

The original sample N (unrepaired) on the north side has special geological characteristics. Due to its high sand content and uneven pore distribution, water rapidly infiltrates along cracks during the initial stage. Within 30 s, this leads to the formation of 2–3 small gullies deeper than 3 mm, demonstrating a chain reaction characteristic of “pitting-gully erosion”. After 6 h, the structure showed significant degradation through block collapse (1–3 cm³ per block) and gully expansion (>1 cm width); ultimately, after 12 h, it was completely fragmented, retaining only 10% of its original mass (Figure 23j–l). Erosion resistance is second only to sample E, highlighting the erosion resistance of the loose sand structure. The overall erosion resistance of the samples can be ranked in descending order as follows: W > S > N > E (

Table 3. Experimental process and result record of spray rammed-earth test block changes.

Sample Name	Sample Initial Weight/g	Weight After 6 h of Showering/g	Weight After 12 h of Showering/g
East	102.49	68.7	35.6
South	343.06	96.3	58.4
West	340.2	86.8	43.2
North	41.15	24.5	10.5

Source: The author calculated based on experimental results.

).

4. Discussion

The experimental results demonstrate that the erosion resistance of rammed-earth walls is influenced by both mineral composition and microbial activity. These findings directly inform the “Save Old Houses” project by highlighting the need for orientation-specific repair strategies and the potential use of microbial regulators in conservation practice. The integration of mineralogical, microbiological, and microstructural data provides a comprehensive explanation for the varying erosion resistance observed in different wall orientations. The high quartz content and regular particle morphology contribute to the stability of the east and west walls (Figure 10, Figure 18, Figure 19 and Figure 21), while the complex microbial community and presence of expansive minerals like Nontronite correlate with the severe erosion patterns on the south and north walls (Figure 13, Figure 17, Figure 20 and Figure 22). The above experimental results show that the “Save Old Houses Action” project identified Songyang County as a pilot county in 2016. The “saving” focuses on cultural relics in traditional Chinese village. Still, it does not include national key cultural relics, Zhejiang provincial cultural relics, or privately owned cultural relics that are part of the general immovable cultural relics listed in the Third National Cultural Relics Survey. The funding for the restoration of old houses in Songyang is shared by three parties: the China Cultural Relics Protection Foundation subsidizes approximately 50%, the Songyang County Government contributes 20% to 30%, and the villagers cover the remaining cost. Let the villagers participate in part of the funds, and the results of repair and protection will be more long-lasting. According to the principle of combining protection and utilization, past repair projects have restored the previously collapsed and dilapidated old houses; these projects not only effectively protected ancient cultural relic buildings but also took into account future use by installing sanitary facilities, re-laying strong and weak electricity pipelines, rectifying the surrounding environment, and demolishing some later-built pig pens and dry toilets, which beautified the living environment and improved sanitary conditions. The guiding principle of maximizing the original appearance mandates the use of traditional craftsmanship and locally sourced materials like soil, wood, and stone in the restoration. The construction workers are mainly composed of the owners of the old house, villagers, and carpenters and stonemasons from neighboring villages. Compared with international conservation strategies, such as the use of sacrificial mud coatings in Europe or community-based maintenance in Latin America, the approach proposed in this study—integrating microbial regulation and local material science—offers a novel, low-intervention alternative. While European methods focus on replaceable protective layers, and Australian strategies emphasize drainage design, our findings highlight the potential of bio-based solutions tailored to subtropical humid climates. In the future, further research into the repair or improvement of rammed-earth walls still has great potential and development prospects. All of this will rely on existing experimental data and microstructural analysis, which will provide a reference for subsequent restoration projects.

5. Conclusions

Through sampling and test analysis of the rammed-earth wall materials of the Yangjiatang Village residence, we can determine its local anti-erosion performance: (1) The minerals on the wall surface are mainly composed of quartz, mullite, lepidocrocite, and nontronite. The minerals contained in all four directions are mainly quartz and lepidocrocite. (2) The particle morphology of the east and west samples is more regular, the particle size distribution is relatively concentrated, and the surface is smooth, indicating that the growth environment of these two directions is more uniform and stable during synthesis or processing. The south and north samples exhibit more complex morphology, a wider particle size distribution, and rougher surfaces with cracks, which may be attributed to factors such as crystal orientation, local temperature gradients, or stress concentration. (3) The overall corrosion resistance shows the order of sample W > sample S > sample N > sample E. (4) The surface degradation of rammed-earth walls is mainly caused by four factors: structural cracks, surface erosion, biological erosion, and roof damage. These factors together caused surface cracking and peeling (depth of 3–5 cm).

While this study provides a comprehensive analysis of the anti-erosion mechanisms of rammed-earth walls through multi-technique integration, it primarily focuses on short-term and micro-scale characteristics. The advantages lie in the detailed characterization of mineral composition, microbial community structure, and microstructural features, which offer immediate insights for restoration strategies. However, the limitations include the lack of long-term environmental monitoring and real-time data on microbial dynamics under varying climatic conditions. Future studies should incorporate continuous environmental sensing and periodic sampling to track changes over decades, thereby enhancing the predictive capacity of microbial regulation technologies. To improve the protection of traditional rammed-earth buildings, future efforts could integrate intelligent monitoring systems (e.g., IoT-based sensors for humidity, temperature, and crack propagation) with eco-friendly materials such as bio-based consolidants or geopolymer coatings. These systems would enable real-time damage assessment and early warning, while environmentally friendly materials would ensure compatibility with the original structure and reduce ecological impact. Such an approach would support sustainable conservation by combining traditional wisdom with modern technology. In the future, work can be integrated with intelligent monitoring systems to evaluate the long-term effectiveness of restoration technology and explore more eco-friendly materials, thereby advancing the sustainable conservation of traditional buildings.

While this study focuses on Yangjiatang Village, the methodological framework—combining multi-scale characterization techniques to decipher the complex interplay between material composition, microbiology, and environment—applies to the conservation of rammed-earth heritage in other humid subtropical regions worldwide. These insights into the dual role of microorganisms and the orientation-specific degradation mechanisms provide a scientific basis for developing targeted, sustainable conservation strategies beyond this specific case study, contributing to the global effort to preserve earthen architectural heritage.