Relationship between the Microbiome and Indoor Temperature/Humidity in a Traditional Japanese House with a Thatched Roof in Kyoto, Japan

In our living environment, there are various microorganisms that are thought to affect human health. It is expected that excessive microbial suppression can have a negative effect on human health and that the appropriate control of the microbiome is beneficial to health. To understand how the physical environment, such as temperature and relative humidity, or housing itself affects the microbiome in a rural house, we measured temperature and humidity and collected microbial samples in a traditional Japanese house with a thatched roof. The relative humidity of outdoor air was over 60% most of the day throughout the year. Indoor and outdoor air temperature and humidity were closer to each other in summer than in winter. The DNA concentration of indoor surfaces correlated with the relative humidity, especially with the lowest annual relative humidity. In the thatched roof, outside surface relative humidity often reached 100%, and the occurrence of condensation can affect the DNA concentrations. A high percentage of archaea were detected in the house, which is not a common characteristic in houses. In addition, the microbial community was similar outdoors and indoors or in each room. These characteristics reflect the occupants’ behaviour, including opening the windows and partitions in summer. In the future, it will be necessary to conduct continuous surveys in various houses, including traditional and modern houses, in Japan.


Introduction
In modern life, we spend approximately 87% of our time indoors [1], and the impact of the living environment on human health is considered to be significant. Risk management for human health in the field of architecture is mainly based on air quality control, such as temperature/humidity control and dilution of carbon dioxide and VOCs by ventilation, and there has been little investigation from the viewpoint of microbiology in public health. Some studies have been conducted on methods of preventing mould growth and material degradation of exterior walls. Abe et al. evaluated the rate of mould growth in a given location based on temperature and relative humidity data [2]. Another study

Temperature/Humidity Measurements
Temperature/humidity measurements were taken continuously in the house beginning on 25 April 2018. prior to this study. The measurement points of temperature/humidity are shown in Figure 2. Outdoor air temperature/humidity was measured under

Temperature/Humidity Measurements
Temperature/humidity measurements were taken continuously in the house beginning on 25 April 2018. prior to this study. The measurement points of temperature/humidity are shown in Figure 2. Outdoor air temperature/humidity was measured under the southeast eave using an outdoor temperature/humidity sensor (HOBO Pro V2 U23-002A, Onset Computer Corporation, Bourne, MA, USA) and a weather station (HOBO Weather Station Kits, Onset Computer Corporation, Bourne, MA, USA) located nearby. Indoor air temperature/humidity was measured using indoor temperature/humidity sensors (HOBO UX100-011A, Onset Computer Corporation, Bourne, MA, USA) in the earth floor space, the dining room, the Japanese-style room, the second-floor storage room, and the space under the ceiling. The temperature/humidity inside the thatched roof was measured using outdoor temperature/humidity sensors (HOBO Pro V2 U23-002A, Onset Computer Corporation, Bourne, MA, USA) at the south eave, the west eave, and the south side of the thatched roof ( Figure 2). The surface temperature of the thatched roof was measured at the west eave using thermocouples (HOBO UX100-014 M, Onset Computer Corporation, Bourne, MA, USA). The measurement intervals for all locations were 30 min. The relative humidity of the thatched roof surface was approximated by a calculated value using the temperature of the roof surface and the water vapour pressure calculated by the temperature and the relative humidity of the outdoor air.
(c) (d) Figure 1. Photographs of (a) the subject house, (b) the earth floor space, (c) the Japanese-style room and dining room, and (d) underside of the thatched roof.

Temperature/Humidity Measurements
Temperature/humidity measurements were taken continuously in the house beginning on 25 April 2018. prior to this study. The measurement points of temperature/humidity are shown in Figure 2. Outdoor air temperature/humidity was measured under the southeast eave using an outdoor temperature/humidity sensor (HOBO Pro V2 U23-002A, Onset Computer Corporation, Bourne, MA, USA) and a weather station (HOBO Weather Station Kits, Onset Computer Corporation, Bourne, MA, USA) located nearby. Indoor air temperature/humidity was measured using indoor temperature/humidity sensors (HOBO UX100-011A, Onset Computer Corporation, Bourne, MA, USA) in the earth floor space, the dining room, the Japanese-style room, the second-floor storage room, and the space under the ceiling. The temperature/humidity inside the thatched roof was measured using outdoor temperature/humidity sensors (HOBO Pro V2 U23-002A, Onset Computer Corporation, Bourne, MA, USA) at the south eave, the west eave, and the south side of the thatched roof ( Figure 2). The surface temperature of the thatched roof was measured at the west eave using thermocouples (HOBO UX100-014 M, Onset Computer Corporation, Bourne, MA, USA). The measurement intervals for all locations were 30 min. The relative humidity of the thatched roof surface was approximated by a calculated value using the temperature of the roof surface and the water vapour pressure calculated by the temperature and the relative humidity of the outdoor air. The temperature/humidity measurement points (red and orange dots) and the microbial sampling locations (blue, yellow, and green dots) were plotted on the plan. Coriolis (yellow plots) means sampling locations using a biological air sampler (Coriolis Micro, Bertin Technologies, France). ▲ indicates the main entrance, and △ indicates the back entrances.

Microbial Sampling
On 6 September 2020, indoor and outdoor air samples were collected in phosphatebuffered saline (PBS) liquid by a biological air sampler (Coriolis Micro, Bertin Technologies, Montigny-le-Bretonneux, France) with airflow set at 300 L/min for 10 min. The aerosol-containing liquid from the sampler was filtered on-site through a 20-mL syringe (Terumo Corporation, Tokyo, Japan) and 0.2-µm filter cartridge (Sterivex, Millipore, Burlington, MA, USA). We swabbed the surfaces of the south and west sides of the exterior walls, the earth floor space roof, the earth floor, the dining room wall, the Japanese-style room wall and column, the second-floor storage wall, and the bathroom wall ( Figure 2) using a cotton swab (Fine Check, ASONE, Osaka, Japan). The size of the sampling area was 10 cm × 10 cm for each location. The thatch was pulled from six locations on the roof, as shown in Figure 2 (see also Supplemental Figure S1).

Numerical Analysis of Surface Temperature/Humidity
The temperature/humidity of wall surfaces and floor surfaces are different from the temperature/humidity of air. To evaluate close relationships between the temperature/humidity and microbial communities on surfaces, it is necessary to measure the surface temperature/humidity of the point where microbial samples were collected. Although the air temperature/humidity was measured, the temperature/humidity of surfaces where samples were collected was not measured directly in the subject house. We used the simultaneous heat and moisture transfer equation [31], which is a method commonly used in the field of architecture, to predict the temperature/humidity inside walls. We modelled the ground and mud walls of the earth floor and the Japanese-style room wall, respectively. We calculated the surface temperature/humidity using one-dimensional numerical analysis. The theory and methods of calculating surface temperature/humidity are explained in Appendix A. The temperature/humidity measurement points (red and orange dots) and the microbial sampling locations (blue, yellow, and green dots) were plotted on the plan. Coriolis (yellow plots) means sampling locations using a biological air sampler (Coriolis Micro, Bertin Technologies, France). indicates the main entrance, and indicates the back entrances.

Microbial Sampling
On 6 September 2020, indoor and outdoor air samples were collected in phosphatebuffered saline (PBS) liquid by a biological air sampler (Coriolis Micro, Bertin Technologies, Montigny-le-Bretonneux, France) with airflow set at 300 L/min for 10 min. The aerosolcontaining liquid from the sampler was filtered on-site through a 20-mL syringe (Terumo Corporation, Tokyo, Japan) and 0.2-µm filter cartridge (Sterivex, Millipore, Burlington, MA, USA). We swabbed the surfaces of the south and west sides of the exterior walls, the earth floor space roof, the earth floor, the dining room wall, the Japanese-style room wall and column, the second-floor storage wall, and the bathroom wall ( Figure 2) using a cotton swab (Fine Check, ASONE, Osaka, Japan). The size of the sampling area was 10 cm × 10 cm for each location. The thatch was pulled from six locations on the roof, as shown in Figure 2 (see also Supplementary Figure S1).

Numerical Analysis of Surface Temperature/Humidity
The temperature/humidity of wall surfaces and floor surfaces are different from the temperature/humidity of air. To evaluate close relationships between the temperature/humidity and microbial communities on surfaces, it is necessary to measure the surface temperature/humidity of the point where microbial samples were collected. Although the air temperature/humidity was measured, the temperature/humidity of surfaces where samples were collected was not measured directly in the subject house. We used the simultaneous heat and moisture transfer equation [31], which is a method commonly used in the field of architecture, to predict the temperature/humidity inside walls. We modelled the ground and mud walls of the earth floor and the Japanese-style room wall, respectively. We calculated the surface temperature/humidity using one-dimensional numerical analysis. The theory and methods of calculating surface temperature/humidity are explained in Appendix A.

DNA Extraction and Sequencing
All samples were obtained using aseptic techniques and appropriate negative controls. A swab and a filter were directly placed in a bead tube of a DNeasy PowerBiofilm Kit (QIAGEN, Germantown, MD, USA) under a laminar flow cabinet, and DNA was extracted according to the manufacturer's protocol with some modifications, as follows [32]. Instead of using glass beads in a PowerBiofilm bead tube, 400 µL of sterilized ϕ0.5 mm zirconia beads (TORAY, Tokyo, Japan) and two grains of ϕ5 mm zirconia beads (TORAY) were used for homogenization. The samples were beaten with a Multi-bead shocker ® (Yasui Kikai Corporation, Osaka, Japan) at 2700 rpm for 10 min. The DNA was eluted in 100 µL of elution buffer and then purified and concentrated with a Dr. GenTLE precipitation carrier (Takara BIO, Tokyo, Japan). The concentration and purity of the DNA were measured with a DS-11FX + Spectro/Fluorometer (DeNovix, Wilmington, NC, USA).
The extracted DNA was amplified according to the Earth Microbe project [33]. The V4 region of the 16S rRNA gene and V9 region of the 18S rRNA gene were used as target regions for sequencing on the Illumina MiniSeq platform. The detailed library preparation procedures are described in [34]. PCR amplification reactions contained 2.5 µL each of 1 µM primers, 12.5 µL of 2× MightyAmp PCR Buffer v. 3 (TaKaRa Clontech, Mountain View, CA, USA), 0.5 µL of MightyAmp DNA Polymerase v.3 (1.25 U/µL), 5 µL of PCR grade water, and 2.5 µL of DNA templates. Amplification was performed under the following conditions: for the 16S rRNA gene, 98 • C for 2 min, followed by 35 cycles of 95 • C for 30 s, 50 • C for 1 min, and 68 • C for 1 min, for the 18S rRNA gene, 98 • C for 2 min, followed by 35 cycles of 95 • C for 30 s, 65 • C for 30 s, and 68 • C for 30 s. A reaction containing no template served as the negative control and confirmed the absence of nonspecific amplification. After amplification, PCR products were examined in a 2% w/v agarose-TAE gel, stained with Safelook Load-Green (Wako, Osaka, Japan), visualized on Printgraph CMOS I (ATTO, Tokyo, Japan), and cleaned using a Pronex ® Size-Selective Purification system (Promega, WI, USA). Primers with different barcodes (short artificial DNA sequences) were used for different samples to identify each sample. For index PCR amplification reactions, 12.5 µL of 2× KAPA HiFi HotStart ReadyMix (Kapa Biosystems, Woburn, MA, USA), 2.5 µL of each forward and reverse index primer (1 µM), and 7.5 µL of purified PCR product DNA were used. Amplification was performed under the following conditions: 95 • C for 3 min, followed by 8 cycles of 95 • C for 30 s, 55 • C for 30 s and 72 • C for 30 s, and a final elongation of 72 • C for 5 min. The libraries were verified by the fragment analyser TapeStation 4000 series (Agilent Technologies, Palo Alto, CA, USA), cleaned using a Pronex ® Size-Selective Purification system (Promega), and quantified with a DS-11FX + Spectro/Fluorometer (DeNovix). Based on these DNA concentrations, samples were pooled at equimolar concentrations into one tube and diluted to 1 nM. The pooled amplicons were denatured and diluted to 30 pM, and 30% PhiX DNA was added according to the manufacturer's recommendations. The mixed library at 0.8 pM was sequenced with Illumina MiniSeq using the 150-cycle Mini Reagent Kit (Illumina).

Data Analysis and Availability
A total of 6872,122 for 18S rRNA gene and 3,150,556 for 16S rRNA gene of raw sequences were obtained from MiniSeq Illumina sequencing. The raw sequenced reads were filtered for low-quality reads and adapter regions using Trimmomatic v. 0.39 [35] with the following settings: TrimmomaticSE -threads 2; LEADING 20; TRAILING 25; SLIDINGWINDOW 4:15; ILLUMINACLIP 2:30:10; MINILEN 120. After filtering and the removal of the primer sequences, the 18S and 16S rRNA gene sequences were trimmed and quality filtered using the DADA2 package [36] in the program R [37] with a procedure adopted from the DADA2 pipeline version 1.14.1 with the following settings: maxLen 150; miniLen 100; maxEE 2; truncQ 2; rm.phix TRUE; compress TRUE; verbose TRUE; multithread TRUE. The taxonomy was assigned using the silva_nr_v132_train_set.fa.gz database. For 16S rRNA gene analysis, chloroplast and mitochondrial OTUs were manually removed. Samples with more than 1000 reads were used for further analysis, and microbial data analysis was conducted in R using the packages phyloseq v1.36.0 [38] and vegan v2.5.7; ggplot2 v3.3.4 was used for visualization [39].

Outdoor Air and Indoor Air Temperature/Humidity
Temperature/humidity measurements have been conducted since 25 April 2018. In this study we used temperature/humidity data form 2019, when the data was available throughout the year, since there was no significant difference by year and the data represents the temperature/humidity conditions of the subject house (Supplementary Table S1). The daily mean of the outdoor temperature around the subject house rose from below 5 • C in winter to nearly 30 • C in summer (Figure 3a). The temperature gradually rose in spring and decreased in autumn ( Figure 3a). The relative humidity was high throughout the year, with few days when the daily mean value was below 60% (Figure 3b). In the summer, the indoor air temperature fluctuated with the outdoor air temperature, and the daily range was narrower than that of the outdoor air. The air temperature of the space under the ceiling did not rise as high as the outdoor air ( Figure 4a). This was due to the high thermal insulation performance and high thermal capacity of the thatched roof. In a modern house with a steel roof, the attic space air temperature can rise higher than the outdoor temperature [40]. The relative humidity in the same season was higher at night outdoors and in the earth floor space, but the difference among indoor locations was small ( Figure 4c).

Outdoor Air and Indoor Air Temperature/Humidity
Temperature/humidity measurements have been conducted since 25 April 2018. In this study we used temperature/humidity data form 2019, when the data was available throughout the year, since there was no significant difference by year and the data represents the temperature/humidity conditions of the subject house (Supplemental Table S1). The daily mean of the outdoor temperature around the subject house rose from below 5 °C in winter to nearly 30 °C in summer ( Figure 3a). The temperature gradually rose in spring and decreased in autumn ( Figure 3a). The relative humidity was high throughout the year, with few days when the daily mean value was below 60% (Figure 3b). In the summer, the indoor air temperature fluctuated with the outdoor air temperature, and the daily range was narrower than that of the outdoor air. The air temperature of the space under the ceiling did not rise as high as the outdoor air ( Figure 4a). This was due to the high thermal insulation performance and high thermal capacity of the thatched roof. In a modern house with a steel roof, the attic space air temperature can rise higher than the outdoor temperature [40]. The relative humidity in the same season was higher at night outdoors and in the earth floor space, but the difference among indoor locations was small ( Figure 4c).
In the winter, the difference in temperature and relative humidity among locations was larger than in summer. In addition, there was a wider range of fluctuations and a shorter cycle in the indoor air temperature in winter than in summer (Figure 4b,d). In spring and autumn, the difference in indoor and outdoor air temperature/humidity was intermediate between that in summer and winter.
These seasonal variations are supposed to reflect room temperature control methods. In this house, occupants opened the windows and partitions without air conditioning in summer, so indoor air temperature and humidity were similar to those of the outdoor air. In winter, they used a wood-burning stove and closed windows. Thus, the temperature and humidity of the indoor air were different between the room with the stove and the room without the stove, and these differed greatly from the outdoor air. In the winter, the difference in temperature and relative humidity among locations was larger than in summer. In addition, there was a wider range of fluctuations and a shorter cycle in the indoor air temperature in winter than in summer (Figure 4b,d). In spring and autumn, the difference in indoor and outdoor air temperature/humidity was intermediate between that in summer and winter.
These seasonal variations are supposed to reflect room temperature control methods. In this house, occupants opened the windows and partitions without air conditioning in summer, so indoor air temperature and humidity were similar to those of the outdoor air. In winter, they used a wood-burning stove and closed windows. Thus, the temperature and humidity of the indoor air were different between the room with the stove and the room without the stove, and these differed greatly from the outdoor air.  The mean annual temperature was low outdoors and in the earth floor space and high indoors. The annual highest temperature was above 30 °C in all locations, and the annual range was more than 25 °C (Table 1a). This indicates that there was a wide annual range in temperature everywhere in the house. For the relative humidity, the mean value was high in the outdoor and the earth floor space, and the highest value exceeded 85% in all locations during the summer. The annual range outdoors was wide. In the storage room, the annual range of relative humidity was narrowest, although it was nearly 60% (Table 1b). This means that the annual range of relative humidity was also wide everywhere. The temperature and humidity measurement results show that the earth floor space is a buffer space between the outdoor and indoor environments.
Furthermore, the eaves provide shading and suppress a rise of indoor air temperature [41]. Japanese modern houses have large openings in each direction and have few eaves for design purposes, while in traditional Japanese houses the deep eaves of the  The mean annual temperature was low outdoors and in the earth floor space and high indoors. The annual highest temperature was above 30 • C in all locations, and the annual range was more than 25 • C (Table 1a). This indicates that there was a wide annual range in temperature everywhere in the house. For the relative humidity, the mean value was high in the outdoor and the earth floor space, and the highest value exceeded 85% in all locations during the summer. The annual range outdoors was wide. In the storage room, the annual range of relative humidity was narrowest, although it was nearly 60% (Table 1b). This means that the annual range of relative humidity was also wide everywhere. The temperature and humidity measurement results show that the earth floor space is a buffer space between the outdoor and indoor environments.
Furthermore, the eaves provide shading and suppress a rise of indoor air temperature [41]. Japanese modern houses have large openings in each direction and have few eaves for design purposes, while in traditional Japanese houses the deep eaves of the thatched roof provide shading [40]. In addition to the high thermal insulation and high thermal capacity of the thatched roof, the shading effect of the roof moderates the indoor air temperature rise in summer in the subject house. Table 1. Annual average value, annual highest value, annual lowest value, and annual range of (a) temperature ( • C) and (b) relative humidity (%) in 2019 at each location. All the values in these tables are rounded off to one decimal place. (a)

Annual Mean
Annual

Result of Numerical Analysis of Surface Temperature/Humidity
The swabbed surface temperature/humidity was not measured. Therefore, to understand how the surface temperature/humidity fluctuations are influenced by air, we calculated the corresponding values using the simultaneous heat and moisture transfer equation. The temperature of the earth floor surface changed following the temperature of the earth floor space air, but the surface temperature did not rise as high as that of the air. The relative humidity of the earth floor surface fluctuated little and was approximately the highest daily value of the relative humidity of the air (Figure 5a).
For the Japanese-style room wall, the daily range of surface temperatures was approximately 2 • C narrower than that of the air, and the surface relative humidity fluctuated a few hours behind that of the air. (Figure 5b).

DNA Concentrations and Relative Humidity of Indoor Surfaces
The relationship between the DNA concentration and the relative humidity of the swabbed surface was investigated.
For the surface of the Japanese-style room wall, the analytical values of the mud walls were used. The swabbed surface of the column in the Japanese-style room was adjacent to the surface of the wall, and the difference in relative humidity was considered to be small. Thus, the analytical values of mud walls were used for the surface of the column. For other indoor wall surfaces, the relative humidity should be that of the air of each room. This is because, from the results of the analysis using the simultaneous heat and moisture transfer equation, the relative humidity of the surface of indoor walls can be approximated by the relative humidity of the air with which the surface is in contact. For the earth floor surface, the analytical values of the ground were used.

DNA Concentrations and Relative Humidity of Indoor Surfaces
The relationship between the DNA concentration and the relative humidity of the swabbed surface was investigated.
For the surface of the Japanese-style room wall, the analytical values of the mud walls were used. The swabbed surface of the column in the Japanese-style room was adjacent to the surface of the wall, and the difference in relative humidity was considered to be small. Thus, the analytical values of mud walls were used for the surface of the column. For other indoor wall surfaces, the relative humidity should be that of the air of each room. This is because, from the results of the analysis using the simultaneous heat and moisture transfer equation, the relative humidity of the surface of indoor walls can be approximated by the relative humidity of the air with which the surface is in contact. For the earth floor surface, the analytical values of the ground were used.
No correlation between DNA concentration and the highest annual relative humidity was observed (Figure 6b, R 2 = 0.3537). While positive correlations with the DNA concentrations were observed for the annual mean value of relative humidity and the annual lowest value of relative humidity, with correlation coefficients of 0.6574 and 0.9994, respectively (Figure 6a,c). The annual range in relative humidity was negatively correlated with the DNA concentration ( Figure 6d, R 2 = 0.9996). The temperature and humidity of the surface may affect the numbers of microorganisms on indoor wall and floor surfaces. For the relationship between relative humidity and microorganisms, Danmiller et al. reported that the growth of microorganisms in house dust is accelerated in a high-humidity environment [30].
These results showed that there was a correlation between the DNA concentration on indoor wall and floor surfaces and the annual relative humidity, and the lowest relative humidity had a particularly large effect. No correlation between DNA concentration and the highest annual relative humidity was observed (Figure 6b, R 2 = 0.3537). While positive correlations with the DNA concentrations were observed for the annual mean value of relative humidity and the annual lowest value of relative humidity, with correlation coefficients of 0.6574 and 0.9994, respectively (Figure 6a,c). The annual range in relative humidity was negatively correlated with the DNA concentration (Figure 6d, R 2 = 0.9996). The temperature and humidity of the surface may affect the numbers of microorganisms on indoor wall and floor surfaces. For the relationship between relative humidity and microorganisms, Danmiller et al. reported that the growth of microorganisms in house dust is accelerated in a high-humidity environment [30].
These results showed that there was a correlation between the DNA concentration on indoor wall and floor surfaces and the annual relative humidity, and the lowest relative humidity had a particularly large effect.
The surface material is soil for the earth floor, mud wall for the dining room wall and the Japanese-style room wall, wood for the Japanese-style room column and the storage wall, and thatch for the earth floor roof surface. In the subject house, the DNA concentrations of microorganisms on surfaces varied more by sampling location than by surface material. Comparing the wood column and mud wall in the Japanese-style room, the DNA concentration was higher on the column surface. It has been reported that microbial biomass can reflect seasonal variations in soil moisture and temperature [42], and in damp or water-damaged building materials, wood is a material which is likely to support fungal growth, following plaster and concrete [29]. This indicates that the surface materials and those moisture conditions can affect the DNA concentration of microorganisms.

Thatched Roof Temperature/Humidity
Variations in temperature/humidity inside and on the surface of the thatched roof were significantly different. The temperature at the surface of the thatched roof can rise to much higher than that of the outdoor air, reaching up to 60 • C in summer (Figure 7a). At night, the surface temperature can drop below the outdoor air temperature (Figure 7b). This was due to the effect of nocturnal radiation. The temperature dropped significantly, especially in winter. However, on some days in winter, the roof surface temperature did not drop below approximately 0 • C, although the outdoor air temperature dropped below 0 • C (Figure 7b). When this occurred, the roof surface may have frozen due to snow covering the roof. The temperature inside the roof had a narrower daily range than the outside roof surface and the outdoor air and showed a narrower range at deeper locations inside the roof (Figure 7a,b). The surface material is soil for the earth floor, mud wall for the dining room wall and the Japanese-style room wall, wood for the Japanese-style room column and the storage wall, and thatch for the earth floor roof surface. In the subject house, the DNA concentrations of microorganisms on surfaces varied more by sampling location than by surface material. Comparing the wood column and mud wall in the Japanese-style room, the DNA concentration was higher on the column surface. It has been reported that microbial biomass can reflect seasonal variations in soil moisture and temperature [42], and in damp or water-damaged building materials, wood is a material which is likely to support fungal growth, following plaster and concrete [29]. This indicates that the surface materials and those moisture conditions can affect the DNA concentration of microorganisms.

Thatched Roof Temperature/Humidity
Variations in temperature/humidity inside and on the surface of the thatched roof were significantly different. The temperature at the surface of the thatched roof can rise to much higher than that of the outdoor air, reaching up to 60 °C in summer (Figure 7a). At night, the surface temperature can drop below the outdoor air temperature (Figure 7b). This was due to the effect of nocturnal radiation. The temperature dropped significantly, especially in winter. However, on some days in winter, the roof surface temperature did not drop below approximately 0 °C, although the outdoor air temperature dropped below 0 °C (Figure 7b). When this occurred, the roof surface may have frozen due to snow cov- ering the roof. The temperature inside the roof had a narrower daily range than the outside roof surface and the outdoor air and showed a narrower range at deeper locations inside the roof (Figure 7a,b).
The relative humidity at the surface of the thatched roof had a wider daily range than that of the outdoor air (Figure 7c,d). At night, when the relative humidity of the outdoor air was high, that of the thatched roof surface was also high, reaching 100% on many days throughout the year, and more frequently in winter (Figure 7c,d). The reason for this is that the temperature of the thatched roof surface was lower than the outdoor temperature due to nocturnal radiation, where condensation occurs. The relative humidity inside the roof fluctuated less than that outside the roof surface and the outdoor air. At a depth of 250 mm inside the roof, the variation was different from that of the outdoor air or at a depth of 50 mm. This was due to the effects of solar radiation and rainfall on the outside roof surface on the heat and moisture transfer inside the roof (Figure 7c,d).
The relative humidity of the thatched roof surface reached 100% on more than half the days of the year, which was more than double that of the outdoor air. On the other hand, there were no days when the relative humidity inside the thatched roof reached 100% during this period (Table 2). The relative humidity at the surface of the thatched roof had a wider daily range than that of the outdoor air (Figure 7c,d). At night, when the relative humidity of the outdoor air was high, that of the thatched roof surface was also high, reaching 100% on many days throughout the year, and more frequently in winter (Figure 7c,d). The reason for this is that the temperature of the thatched roof surface was lower than the outdoor temperature due to nocturnal radiation, where condensation occurs. The relative humidity inside the roof fluctuated less than that outside the roof surface and the outdoor air. At a depth of 250 mm inside the roof, the variation was different from that of the outdoor air or at a depth of 50 mm. This was due to the effects of solar radiation and rainfall on the outside roof surface on the heat and moisture transfer inside the roof (Figure 7c,d).
The relative humidity of the thatched roof surface reached 100% on more than half the days of the year, which was more than double that of the outdoor air. On the other hand, there were no days when the relative humidity inside the thatched roof reached 100% during this period (Table 2). Table 2. The number of days and total time that the relative humidity reached 100% per year are shown for the outdoor air, thatched roof surface, and 250 mm inside the thatched roof. For the thatched roof surfaces and inside the roof, there were times when relative humidity data were not available. The number of days was calculated by counting the days when the relative humidity reached 100%, even temporarily. Since the measurement interval was 30 min, the total time (hours) was calculated as the number of data points at which the relative humidity reached 100% multiplied by 30 min.

DNA Concentration and Relative Humidity of the Thatched Roof
For microbial samples extracted from thatched roofs, the relationship between DNA concentration and relative humidity was different from the relationship found in indoor surface samples. There was no correlation between mean annual relative humidity and DNA concentration (Figure 8a, R 2 = 0.159). However, the higher the number of days or longer the total time that the relative humidity reached 100%, the higher the DNA concentration (Figure 8b, R 2 = 0.9986). Thus, the length of time that the relative humidity reaches 100% is supposed to have an effect on microorganisms on thatched roofs. As we will discuss later, the thatch of the roof surface, where condensation frequently occurred, deteriorated.

Outdoor Air
Roof Surface

DNA Concentration and Relative Humidity of the Thatched Roof
For microbial samples extracted from thatched roofs, the relationship between DNA concentration and relative humidity was different from the relationship found in indoor surface samples. There was no correlation between mean annual relative humidity and DNA concentration (Figure 8a, R 2 = 0.159). However, the higher the number of days or longer the total time that the relative humidity reached 100%, the higher the DNA concentration (Figure 8b, R 2 = 0.9986). Thus, the length of time that the relative humidity reaches 100% is supposed to have an effect on microorganisms on thatched roofs. As we will discuss later, the thatch of the roof surface, where condensation frequently occurred, deteriorated.
(a) (b) Figure 8. Correlation between the DNA concentrations of microorganisms and relative humidity on the west side of the thatched roof. The horizontal axis represents the (a) annual mean relative humidity and (b) total hours when the relative humidity reached 100%. The samples were collected from three locations (under side of the eaves, roof surface side, and inner side of roof), and the relative humidity of outdoor air, roof surface 1, and 250 mm inside the roof were adapted.

Sequence Results and Comparison of Microbial Communities
Sixteen and 13 samples were used for analysis of the 18S rRNA gene and 16S rRNA gene, respectively. Good's coverage values were greater than 99.5% for all the samples (Supplemental Tables S2 and S3). In terms of the richness and evenness within the samples, water displayed the highest values but did not differ significantly among the samples (p > 0.05) (Figure 9a). Because of the characteristics of traditional Japanese houses with thatched roofs and earth floorspace, we collected samples from various indoor and outdoor locations (mud walls, wooden walls, earth floor, thatched roof, water, and air) to

Sequence Results and Comparison of Microbial Communities
Sixteen and 13 samples were used for analysis of the 18S rRNA gene and 16S rRNA gene, respectively. Good's coverage values were greater than 99.5% for all the samples (Supplementary Tables S2 and S3). In terms of the richness and evenness within the samples, water displayed the highest values but did not differ significantly among the samples (p > 0.05) (Figure 9a). Because of the characteristics of traditional Japanese houses with thatched roofs and earth floorspace, we collected samples from various indoor and outdoor locations (mud walls, wooden walls, earth floor, thatched roof, water, and air) to understand the microbial community structure (Figure 2). Regarding community similarity (i.e., beta diversity and hierarchical clustering), samples were grouped by phenotype, but there was a high degree of diversity among samples. Interestingly, indoor and outdoor aerosol samples overlapped (MK-Cw_18S, MK_Cd_18S and MK_Co_18S). This is different from the results of previous studies conducted in modern houses [5,26]. In general, the microbial community structure inside and outside the house is very different, and most microbes inside the house are of human origin. Even doorknobs outside the house have a very different microbial community origin from those inside the house [43]. It is likely that there is sufficient microbial exchange between the outside and inside (Figures 9b and 10). Buildings affect the biogeography and the pattern of microbial spread indoors [44,45] via the building materials, surfaces, and products used [42,46,47], indoor environmental conditions (temperature, humidity, light, airflow, etc.) [30,48,49], indoor-outdoor connections, and associated microbes [19,50], etc. All of these factors affect the location of microorganisms in the built environment and their survival there. Although the built environment may seem inconvenient for microorganisms [51], they can survive indoors for months [52][53][54][55], and indoor environmental conditions can promote intermittent growth of bacteria and fungi [30]. In traditional Japanese houses with thatched roofs and earth floors, a character-istic temperature and humidity environment is formed (Figure 4). In addition, microbial diversity is presumably increased by the intervention of bringing in soil and mud walls wherein microorganisms may be fermenting plant matter, making this study distinct from many studies on microorganisms in modern buildings.
indoors [44,45] via the building materials, surfaces, and products used [42,46,47], indoor environmental conditions (temperature, humidity, light, airflow, etc.) [30,48,49], indooroutdoor connections, and associated microbes [19,50], etc. All of these factors affect the location of microorganisms in the built environment and their survival there. Although the built environment may seem inconvenient for microorganisms [51], they can survive indoors for months [52][53][54][55], and indoor environmental conditions can promote intermittent growth of bacteria and fungi [30]. In traditional Japanese houses with thatched roofs and earth floors, a characteristic temperature and humidity environment is formed (Figure 4). In addition, microbial diversity is presumably increased by the intervention of bringing in soil and mud walls wherein microorganisms may be fermenting plant matter, making this study distinct from many studies on microorganisms in modern buildings.  Table S2 for detailed explanation of each sample label.  Table S2 for detailed explanation of each sample label.

Taxonomic Composition and Biotic Interactions
For eukaryotes, Ascomycota, Phragmoplastophyta, and Basidiomycota dominated the samples. Arthropoda such as spiders, collembola, and flies were the dominant taxa in air samples. The high concentration of Arthropoda in the aerosol samples was probably due to the aspirated faeces and shell fragments. In the thatch group, most of the plants were grasses, especially MK_T_C and MK_T_D, indicating that the thatch genes from eukaryotes were not destroyed by UV. MK_T_C and MK_T_D were collected from the interior of the thatch and under the eaves, respectively. The appearance of the thatch differed depending on its location (Supplementary Figure S1), the temperature and humidity environment, and microbial features (Figures 4, 7 and 11). The surface thatch sample (MK_T_B) was brittle like soil. This was presumably the result of microbiological degradation, as Aspergillaceae was dominant in this sample.
What is unique microbiologically in this study is the high percentage of archaea detected. Note that archaea appear in the eukaryotic data as well as in the prokaryotic data, as the universal primer used in this study amplifies archaea [35]. In water samples, Nitrosotateaceae, an ammonia-oxidizing archaea, was dominant. In prokaryotic data, Candidatus_Nitrosotalea, an ammonia-oxidizing archaea, was also present ( Figure 12). Interestingly, Halococcus, an extremely halophilic archaea, was present on the earth floor. The water in the houses is well water and has not been treated. In addition, the earth floor was also dominated by archaea, with an unusually large percentage of archaea present for a house. It has been reported that the risk of asthma in children is reduced in farm home-like indoor environments, and the microbiota of farm dust is characterized by high levels of archaea [33]. This indicates that the microbiome in the traditional Japanese house could have a positive effect on human health.  Figure 10. Hierarchical clustering of the eukaryotic community based on distance measurements using the Bray-Curtis dissimilarity index and a clustering algorithm using the Ward distance. The samples from the wall and thatch could not be readily distinguished.

Taxonomic Composition and Biotic Interactions
For eukaryotes, Ascomycota, Phragmoplastophyta, and Basidiomycota dominated the samples. Arthropoda such as spiders, collembola, and flies were the dominant taxa in air samples. The high concentration of Arthropoda in the aerosol samples was probably due to the aspirated faeces and shell fragments. In the thatch group, most of the plants were grasses, especially MK_T_C and MK_T_D, indicating that the thatch genes from eukaryotes were not destroyed by UV. MK_T_C and MK_T_D were collected from the interior of the thatch and under the eaves, respectively. The appearance of the thatch differed depending on its location (Supplemental Figure), the temperature and humidity environment, and microbial features (Figures 4, 7 and 11). The surface thatch sample (MK_T_B) was brittle like soil. This was presumably the result of microbiological degradation, as Aspergillaceae was dominant in this sample. What is unique microbiologically in this study is the high percentage of archaea detected. Note that archaea appear in the eukaryotic data as well as in the prokaryotic data, as the universal primer used in this study amplifies archaea [35]. In water samples, Nitrosotateaceae, an ammonia-oxidizing archaea, was dominant. In prokaryotic data, Candi-datus_Nitrosotalea, an ammonia-oxidizing archaea, was also present ( Figure 12). Interestingly, Halococcus, an extremely halophilic archaea, was present on the earth floor. The ingly, Halococcus, an extremely halophilic archaea, was present on the earth floor. The water in the houses is well water and has not been treated. In addition, the earth floor was also dominated by archaea, with an unusually large percentage of archaea present for a house. It has been reported that the risk of asthma in children is reduced in farm homelike indoor environments, and the microbiota of farm dust is characterized by high levels of archaea [33]. This indicates that the microbiome in the traditional Japanese house could have a positive effect on human health.

Conclusions
We measured the temperature and humidity outdoors, indoors, and inside the roof of a thatched-roof house with an earth floor surrounded by mountains and forests. The

Conclusions
We measured the temperature and humidity outdoors, indoors, and inside the roof of a thatched-roof house with an earth floor surrounded by mountains and forests. The house was ventilated by opening the windows in summer and heated by a wood burning stove in winter, so the indoor air temperature and relative humidity were closer to the outdoor air in summer than in winter. The high thermal insulation, the high thermal capacity and the shading effect of the thatched roof, which is a characteristic of Japanese traditional houses, moderated the indoor air temperature rise in summer. For relative humidity, the annual mean value was lower and the annual range was narrower indoors than outdoors. The DNA concentration of microorganisms was greater on the earth floor surface than on indoor surfaces. The DNA concentrations on the swabbed indoor surfaces were correlated with the mean annual relative humidity and the lowest annual relative humidity. In particular, the correlation of the DNA concentrations with the lowest annual relative humidity was high. This result shows that relative humidity has a significant effect on the DNA concentration of microorganisms, including bacteria, similar to what has been reported for fungi. This suggests that lowering the minimum relative humidity is effective in suppressing the growth of microorganisms. Microbial surveys were also conducted on the surface and inside of the thatched roof. The DNA concentration of microorganisms on the outside surface of the roof was the highest, while the concentrations on the inside were lower. On the surface of the thatched roof with a high DNA concentration, condensation occurred frequently and the thatch deteriorated. It is necessary to consider the effect of the occurrence of condensation on the durability of thatched roofs. The microbial community in the subject house differed from that in a typical modern house in that there was less difference between indoor and outdoor settings or between indoor settings at each location. In addition, a high percentage of archaea were detected in the house. The earth floor was dominated by archaea, and there was an unusually large percentage of archaea present for a house. This indicates that the microbiome in the traditional Japanese house could have a positive effect on human health.
In this study, the relationship between temperature/humidity and DNA concentration was considered. How temperature and humidity, and the surface material affect the microbiome in the building environment should be considered in future work. This survey was conducted in the summer of 2020, but given that temperature and humidity environments vary greatly from season to season, it is necessary to conduct multiple surveys to examine seasonal variations. Furthermore, it also needs to be investigated in various other houses, including traditional and modern houses in Japan.

Institutional Review Board Statement: Not applicable.
Informed Consent Statement: Not applicable.

Data Availability Statement:
The authors declare that all the data supporting the findings of this study are available within the paper (and its Supplementary files) and that raw data were presented where possible. The raw MiniSeq data reported in the paper (Figures 9-12, Supplementary Tables S2 and S3) have been uploaded to the DDBJ database under the accession number DRR313886-DRR313922.