Edited by: Télesphore Sime-Ngando, CNRS, Centre National de la Recherche, France
Reviewed by: Anne Marie Delort, Blaise Pascal University, France; S. Venkata Mohan, Indian Institute of Chemical Technology (CSIR), India
*Correspondence: Satoshi Hiraoka
Wataru Iwasaki
This article was submitted to Aquatic Microbiology, a section of the journal Frontiers in Microbiology
This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) or licensor are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
The presence of microbes in the atmosphere and their transport over long distances across the Earth's surface was recently shown. Precipitation is likely a major path by which aerial microbes fall to the ground surface, affecting its microbial ecosystems and introducing pathogenic microbes. Understanding microbial communities in precipitation is of multidisciplinary interest from the perspectives of microbial ecology and public health; however, community-wide and seasonal analyses have not been conducted. Here, we carried out 16S rRNA amplicon sequencing of 30 precipitation samples that were aseptically collected over 1 year in the Greater Tokyo Area, Japan. The precipitation microbial communities were dominated by Proteobacteria, Firmicutes, Bacteroidetes, and Actinobacteria and were overall consistent with those previously reported in atmospheric aerosols and cloud water. Seasonal variations in composition were observed; specifically, Proteobacteria abundance significantly decreased from summer to winter. Notably, estimated ordinary habitats of precipitation microbes were dominated by animal-associated, soil-related, and marine-related environments, and reasonably consistent with estimated air mass backward trajectories. To our knowledge, this is the first amplicon-sequencing study investigating precipitation microbial communities involving sampling over the duration of a year.
Microbes are present and move around nearly everywhere in the Earth. Aerial microbes have received considerable attention within this context because the atmosphere not only is an unusual habitat for microbes but also likely represents a path by which microbes move exceptionally long distances (Kellogg and Griffin,
Precipitation, i.e., rainfall and snowfall, would bring aerial microbes in the troposphere to the ground surface. Quantitative polymerase chain reaction (PCR) has detected pathogenic bacterial sequences in precipitation samples (Kaushik et al.,
Here, we conducted 16S ribosomal RNA (rRNA) amplicon-sequencing analysis of 30 precipitation samples that were aseptically collected over 1 year in the Greater Tokyo Area, Japan. Microbial community analysis revealed seasonal variations in their composition. Notably, the estimated original habitats of precipitation microbes showed reasonable consistency with estimated air mass backward trajectories. Our results support a precipitation-mediated microbial cycle model in which soil, oceanic, and animal-associated microbes are spread in the atmosphere, transported for long distances, and deposited via precipitation.
Precipitation samples were collected at two sites in the Greater Tokyo Area, Japan: Kashiwa (35°54′00″N, 139°55′59″E, 50 m above sea level) and Hongo (35°42′55”N, 139°45'56″E, 30 m above sea level) (Figure
A map of the sampling sites (Kashiwa and Hongo, yellow) and meteorological observatories (Abiko and Tokyo, blue) (left panel), with photos of the sampling sites (right panel). At the Kashiwa site, a US-330 automatic precipitation sampler (Ogasawara Keiki, Tokyo, Japan) was installed. At the Hongo site, precipitation samples were manually collected.
We collected 25 and 5 precipitation samples containing sufficient amounts of microbial DNA at the Kashiwa and Hongo sites, respectively. The sampling dates spanned more than 1 year from May 2014 to October 2015, encompassing the rainy and typhoon seasons in Japan (Table
Sequencing statistics and meteorological characteristics of each precipitation sample.
140521K | 140521(01:00)–140521(18:00) | – | 36 | 16.56 | 994.31 | 3.16 | 50 | 8,340 | 246 | 44 | 2.62 |
140630K_50 | 140628(01:00)–140630(05:00) | Rainy season | 22 | 21.98 | 999.23 | 1.74 | 50 | 8,622 | 1,092 | 27 | 1.63 |
140630K_100 | 140628(01:00)–140630(05:00) | Rainy season | 22 | 21.98 | 999.23 | 1.74 | 100 | 7,444 | 1,287 | 39 | 1.26 |
140630K_200 | 140628(01:00)–140630(05:00) | Rainy season | 22 | 21.98 | 999.23 | 1.74 | 200 | 7,462 | 1,118 | 44 | 1.48 |
140810K_50 | 140810(00:00)–141810(23:00) | Typhoon | 31.5 | 25.46 | 998.46 | 3.89 | 50 | 7,621 | 275 | 76 | 3.70 |
140810K_100 | 140810(00:00)–141810(23:00) | Typhoon | 31.5 | 25.46 | 998.46 | 3.89 | 100 | 8,441 | 108 | 53 | 4.41 |
140810K_200 | 140810(00:00)–141810(23:00) | Typhoon | 31.5 | 25.46 | 998.46 | 3.89 | 200 | 6,664 | 371 | 129 | 3.82 |
140926K | 140925(02:00)–140926(04:00) | – | 6.5 | 21.48 | 1,001.08 | 2.14 | 200 | 1,941 | 18 | 15 | 2.66 |
141014K | 141013(13:00)–141014(07:00) | Typhoon | 32.5 | 19.58 | 992.99 | 4.47 | 200 | 1,641 | 317 | 157 | 4.76 |
141023K | 141021(05:00)–141023(18:00) | – | 31.5 | 14.89 | 1,010.85 | 2.03 | 200 | 1,354 | 120 | 54 | 3.66 |
150107K | 150106(16:00)–150106(18:00) | – | 4 | 12.35 | 992.30 | 5.10 | 200 | 3,410 | 37 | 22 | 2.93 |
150116K | 150115(11:00)–150116(00:00) | – | 40.5 | 4.97 | 1,005.15 | 3.05 | 1,000 | 2,494 | 72 | 55 | 3.86 |
150202K | 150130(05:00)–150130(19:00) | Snow | 12.5 | 1.11 | 1,015.18 | 2.17 | 400 | 9,705 | 1,256 | 194 | 4.69 |
150409K | 150407(03:00)–150408(17:00) | – | 20.5 | 6.39 | 1,017.32 | 2.13 | 200 | 8,337 | 473 | 111 | 4.15 |
150412K | 150410(17:00)–150411(14:00) | – | 16 | 8.99 | 1,018.06 | 1.60 | 200 | 6,805 | 771 | 125 | 4.16 |
150414K | 150413(11:00)–150414(17:00) | – | 36.5 | 10.13 | 1,013.38 | 1.91 | 200 | 5,818 | 655 | 116 | 3.98 |
150513K | 150512(21:00)–150513(01:00) | Typhoon | 23 | 20.00 | 997.43 | 5.60 | 200 | 2,223 | 47 | 25 | 3.92 |
150604K | 150603(08:00)–150603(13:00) | Rainy season | 13 | 20.46 | 998.26 | 1.44 | 200 | 4,663 | 8 | 7 | 2.88 |
150628K | 150626(19:00)–150627(12:00) | Rainy season | 13.5 | 20.92 | 997.42 | 1.24 | 150 | 9,634 | 284 | 46 | 3.41 |
150711K | 150708(15:00)–150709(20:00) | Rainy season | 17.5 | 19.27 | 1,013.22 | 1.48 | 200 | 9,557 | 31 | 20 | 1.91 |
150718K | 150716(04:00)–150717(13:00) | Typhoon | 16.5 | 25.99 | 1,004.62 | 3.26 | 200 | 5,189 | 5 | 4 | 4.03 |
150816K | 150814(05:00)–150814(22:00) | – | 43 | 25.11 | 1,000.38 | 1.91 | 200 | 6,864 | 269 | 68 | 3.05 |
150827K | 150826(00:00)–150826(17:00) | Typhoon | 27 | 20.10 | 1,005.27 | 1.92 | 200 | 7,993 | 1,041 | 226 | 3.39 |
150926K | 150924(19:00)–150926(06:00) | – | 21 | 17.63 | 1,005.28 | 1.90 | 200 | 2,988 | 6 | 4 | 2.75 |
151014K | 151011(01:00)–151011(11:00) | – | 8 | 16.94 | 1,008.92 | 1.00 | 200 | 6,357 | 129 | 30 | 1.33 |
150414H | 150413(07:00)–150414(12:00) | – | 39.5 | 10.03 | 1,014.99 | 3.15 | 200 | 7,215 | 882 | 125 | 3.76 |
150513H | 150512(20:00)–150513(06:00) | Typhoon | 58.5 | 20.25 | 997.43 | 7.23 | 200 | 3,198 | 59 | 38 | 4.83 |
150627H | 150626(15:00)–150627(10:00) | Rainy season | 16 | 21.33 | 998.27 | 2.22 | 150 | 6,450 | 667 | 108 | 1.24 |
150710H | 150708(10:00)–150710(00:00) | Rainy season | 22 | 20.20 | 1,013.22 | 2.33 | 200 | 9,280 | 159 | 48 | 3.00 |
151014H | 151011(02:00)–151011(10:00) | – | 15 | 18.01 | 1,008.80 | 1.90 | 200 | 6,342 | 286 | 67 | 3.70 |
Microbial DNA on the Sterivex filters was retrieved using a ChargeSwitch Forensic DNA Purification Kit (Invitrogen) according to the supplier's protocol with one exception: the filters were directly suspended in the extraction solution from the kit during the cell lysis process. The V5-V6 region of the prokaryotic 16S rRNA gene was amplified using a standard PCR protocol with TaKaRa Ex Taq (TaKaRa) and the following high-performance liquid chromatography-purified primers: 784F (5′- RGGATTAGATACCC -3′) and 1064R (5′- CGACRRCCATGCANCACCT -3′) (Wang and Qian,
For raw sequence data from both precipitation and negative control samples, sequence regions at both ends that contained low-quality bases (quality score < 20) were trimmed using DynamicTrim (Cox et al.,
Amplicon-sequencing data of aerosol and cloud water samples were downloaded from NCBI SRA database (the accession numbers are shown in Supplementary Table
The data on the amount of precipitation, temperature, wind speed, and atmospheric pressure were retrieved from the website of the Japan Meteorological Agency (
The amplicon sequence data were deposited in the DDBJ/ENA/GenBank database under BioSample IDs
A total of 64,100 high-quality sequences 231 ± 45 bp in length were generated from 30 precipitation and eight negative control samples. The precipitation samples included typhoon rain, rainy season rain, and snow. After removing sequences exhibiting >97% similarity to the negative control samples, 12,089 “effective” sequences comprising 1,297 OTUs remained. To make our analyses based on reads that were not likely from contamination as much as possible, we took a conservative and strict filtering approach, whose extent of read number reduction was similar to that in a previous study (Cho and Jang,
Hierarchical cluster analysis of OTU composition in the precipitation samples indicated samples collected during the same precipitation event with different volumes (50, 100, and 200 mL) that were highly similar to each other (Figure
Hierarchical clustering of precipitation samples based on OTU composition. The distance matrix was calculated based on the Bray-Curtis dissimilarity, and clusters were calculated using Ward's method. Open symbols indicate samples that were collected during the same precipitation event with different volumes. Closed symbols indicate samples that were collected on the same day at different sites (Kashiwa and Hongo).
Among the 12,089 effective sequences, 11,994 (99.2%) were taxonomically assigned at the phylum level. Almost all sequences were assigned to 24 phyla in the domain Bacteria with the exception of 4 (0.03%) and 219 (1.7%) sequences assigned to Archaea and mitochondria, respectively. This strong bias toward bacterial sequences may reflect the actual composition but may also be attributable to amplification bias introduced by primer specificity. The top three and six most abundant bacterial phyla accounted for >80 and >95%, respectively, of the sequence pool of all precipitation samples (Figure
Relative abundances of sequences at the phylum
Several OTUs were assigned to genera that potentially contain INA bacteria, i.e.,
Taxonomic distribution exhibited seasonal variability (Figure
Correlation analysis between relative abundances of sequences at the order level and meteorological data. The color scheme represents Spearman's rank correlation coefficient.
To estimate the environments from which microbes in precipitation originated, we performed a microbial habitat index analysis using MetaMetaDB (Yang and Iwasaki,
Estimated ordinary habitats of precipitation microbes. Because the ordinary habitat for an individual 16S rRNA sequence cannot be conclusively determined, the microbial habitability index (MHI) was calculated to estimate the probability of an ordinary habitat (Yang and Iwasaki,
The estimated backward trajectories of air masses that led to the precipitation events at the Kashiwa and Hongo sites were classified as terrestrial, oceanic, and hybrid routes. The terrestrial route typically originated from the middle of the Eurasian continent and passed through the East China Sea, the Yellow Sea, and the Sea of Japan; the oceanic route typically originated from the Pacific Ocean and passed through the East China Sea or the Sea of Okhotsk; and the hybrid route comprised both the terrestrial and oceanic areas. Consistent with the typical pattern of the seasonal winds in Asia, the terrestrial and oceanic routes dominated in winter and summer, respectively (Figure
Soil, oceanic, and animal-associated microbes are spread in the atmosphere and transported for long distances (Morris et al.,
Microbes are present nearly everywhere in the Earth, even in precipitation from the sky. Precipitation is supposed to make microbes in the atmosphere finally fall down to the ground surface. In this study, we thoroughly observed microbial communities in precipitation samples that were collected over 1 year in the Grate Tokyo area, Japan. To our knowledge, this is the first amplicon-sequencing study investigating precipitation microbial communities involving sampling over the duration of a year. Most importantly, our results suggest seasonal variations in the microbial communities in precipitation, and their community structures were significantly associated with the estimated air mass trajectories. These results highlight importance of precipitation in long-range microbial immigration via the atmosphere, which may answer how tiny microbes can dynamically travel around the globe.
SH designed and performed the bioinformatics analyses and wrote the manuscript. MM designed the experiments and performed the sample collection, DNA extraction, DNA sequencing, and bioinformatics analyses. KF and AM designed the experiments and performed the sample collection, cell counting, DNA extraction, and DNA sequencing. WI conceived of and designed the study, wrote the manuscript, and supervised the project. All authors read and approved the final manuscript.
The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.
We thank Kazuhiro Kogure, Minoru Ijichi, and Suguru Nishijima for their helpful suggestions.
The Supplementary Material for this article can be found online at:
Rarefaction curves for each precipitation sample.
Nonmetric multidimensional scaling plot for OTU compositions. The distance matrix was calculated based on the Bray-Curtis dissimilarity. The stress value of the final configuration was 20.46%.
The estimated air mass backward trajectories 240 h prior to precipitation events.
Estimated ordinary habitats of precipitation microbes for three ecosystem groups. The abundance values in each ecosystem group are summation for habitats described below. Marine-related: “aquatic”, “marine”, “marine sediment”, “fish”, and “hot spring”; Animal-associated: “human”, “human gut”, “human lung”, “human nasal pharyngeal”, “bovine gut”, and “mouse gut”; and Soil-related: “hydrocarbon”, “rhizosphere”, “soil”, and “terrestrial.” The estimated route of the air mass before each precipitation event is indicated in the right column.
Estimated ordinary habitats of microbes in aerosol and cloud water samples. Estimated ordinary habitats demonstrating <5% abundance were summarized as “Others.”
Amplicon-sequencing data of aerosol and cloud water samples.