Seasonal shift in airborne microbial communities

Preparing to load PDF file. please wait...

0 of 0
100%
Seasonal shift in airborne microbial communities

Transcript Of Seasonal shift in airborne microbial communities

Seasonal shift in airborne microbial communities
Romie Tignat-Perrier, Aurélien Dommergue, Alban Thollot, Olivier Magand, Pierre Amato, Muriel Joly, Karine Sellegri, Timothy M. Vogel, Catherine Larose
To cite this version:
Romie Tignat-Perrier, Aurélien Dommergue, Alban Thollot, Olivier Magand, Pierre Amato, et al.. Seasonal shift in airborne microbial communities. Science of the Total Environment, Elsevier, 2020, 716, pp.137129. ￿10.1016/j.scitotenv.2020.137129￿. ￿hal-02481717￿

HAL Id: hal-02481717 https://hal.archives-ouvertes.fr/hal-02481717
Submitted on 6 Nov 2020

HAL is a multi-disciplinary open access archive for the deposit and dissemination of scientific research documents, whether they are published or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.

L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.

1

Seasonal shift in airborne microbial communities linked to land use

2 3 Romie Tignat-Perrier1,2*, Aurélien Dommergue1, Alban Thollot1, Olivier Magand1, Pierre 4 Amato3, Muriel Joly3, Karine Sellegri3, Timothy M. Vogel2, Catherine Larose2

5 1Institut des Géosciences de l’Environnement, Université Grenoble Alpes, CNRS, IRD, 6 Grenoble INP, Grenoble, France

7 2Environmental Microbial Genomics, CNRS UMR 5005 Laboratoire Ampère, École Centrale 8 de Lyon, Université de Lyon, Écully, France

9 3Institut de Chimie de Clermont-Ferrand, CNRS UMR 6096 Université Clermont Auvergne10 Sigma, Clermont-Ferrand, France

11 *[email protected] 12 13 Highlights 14 -Airborne microbial communities showed a seasonal shift at the puy de Dôme elevated site 15 -Dominant microbial taxa showed different trends throughout the year 16 -Summer results in higher concentrations of plant-associated microbes in the air 17 -Winter results in higher concentrations of soil and dead material-associated microbes 18 -Seasonal changes in the underlying ecosystems likely drive microbial seasonal shift 19 20 21 22 23 24 25 Abstract
1

26 Microorganisms are ubiquitous in the atmosphere. Global investigations on the geographical 27 and temporal distribution of airborne microbial communities are critical for identifying the 28 sources and the factors shaping airborne communities. At mid-latitude sites, a seasonal shift in 29 both the concentration and diversity of airborne microbial communities has been 30 systematically observed in the planetary boundary layer. While the factors suspected of 31 affecting this seasonal change were hypothesized (e.g., changes in the surface conditions, 32 meteorological parameters and global air circulation), our understanding on how these factors 33 influence the temporal variation of airborne microbial communities, especially at the 34 microbial taxon level, remains limited. Here, we investigated the distribution of both airborne 35 bacterial and fungal communities on a weekly basis over more than one year at the mid36 latitude and continental site of puy de Dôme (France; +1465 m altitude above sea level). The 37 seasonal shift in microbial community structure was likely correlated to the seasonal changes 38 in the characteristics of puy de Dôme’s landscape (croplands and natural vegetation). The 39 airborne microbial taxa that were the most affected by seasonal changes trended differently 40 throughout the seasons in relation with their trophic mode. In addition, the windy and variable 41 local meteorological conditions found at puy de Dôme were likely responsible for the 42 intraseasonal variability observed in the composition of airborne microbial communities. 43 44 Keywords: atmospheric microorganisms, bioaerosols, planetary boundary layer, amplicon 45 sequencing, biosphere-atmosphere interactions 46 47 Introduction 48 Thousands to millions of diverse microbial cells per cubic meter of air are transported among 49 aerosols with their diversity shown to depend on geographic location1 and time of year2. 50 These microorganisms might be active, since some airborne microbial isolates were shown in
2

51 laboratory studies to sustain metabolic activity using organic acids found in the atmosphere3–5. 52 Understanding the global distribution of airborne microbial communities is critical for 53 determining how airborne microorganisms might influence atmospheric chemistry4, 54 meteorological processes such as cloud and precipitation formation6, as well as human and 55 crop health7. Recent large geographical and spatial investigations highlighted the major 56 contribution of the local landscapes and local environmental factors in the observed 57 distribution of airborne microbial communities in the planetary boundary layer1,8. The 58 composition of airborne microbial communities is closely related to the nature of the 59 surrounding landscapes (ocean, agricultural soil, forest etc.) from which local meteorology 60 (especially wind direction and speed) controls microbial cell emission rates1. Studies on 61 airborne microbial communities at mid-latitude sites (aerosol-, cloud water- and precipitation62 associated microorganisms) reported seasonal changes in both microbial biomass and 63 biodiversity2,9–13. The seasonal variability was associated to changes in surface conditions10,11, 64 meteorological conditions2,9 and/or changes in the global air circulation9,13. Yet, our 65 understanding on how these potential factors impact airborne microbial community 66 composition, and more specifically the microbial taxa individually, remains limited. Here, we 67 investigated the distribution of airborne microbial communities and specific microbial taxa at 68 the mid-latitude and continental site of puy de Dôme (France; +1465 m altitude above sea 69 level). We monitored the diversity and abundance of bacterial and fungal communities in the 70 troposphere on a weekly basis for more than a year (June 2016 to August 2017). These 71 microbial community metrics were evaluated in relation to the local meteorology and 72 particulate matter chemical composition, and puy de Dôme local landscape was evaluated 73 based on the MODIS satellite images. While a number of studies focusing on microorganisms 74 in clouds has been carried out at puy de Dôme3,4,14–17, no investigation was conducted in the 75 dry troposphere and at such a high temporal resolution. Puy de Dôme is surrounded mainly by
3

76 croplands and vegetation (i.e. > 80% of the surrounding landscapes in a perimeter of 50 km) 77 whose surface characteristics change drastically over the four different seasons (summer, 78 autumn, winter and spring). These seasonal changes in landscape were related to the temporal 79 variability of airborne microbial community composition at the microbial taxon level. 80 81 Material and Methods 82 Sites and Sampling 83 A size selective high volume air sampler installed at the puy de Dôme (PDD) meteorological 84 station terrace was used to collect particulate matter on quartz fiber filters every week from 85 June 2016 to August 2017 (Table S1). The sampler was equipped with a PM10 size-selective 86 inlet in order to collect particulate matter smaller than 10 µm (PM10) and sampling was done 87 as presented in Dommergue et al. (2019)18. Overall, the dataset was composed of fifty-three 88 samples with an average normalized collected volume of 9100 m3 (Table S1). Quartz fiber 89 filters were heated to 500°C for 8 hours to remove traces of organic carbon including DNA. 90 All the material including the filter holders, aluminium foils and plastic bags in which the 91 filters were transported were sterilized using UV radiation as detailed in Dommergue et al., 92 (2019)18. A series of field and transportation blank filters were carried out to monitor the 93 quality of the sampling protocol as presented in Dommergue et al. (2019)18. PDD is a mid94 altitude (+ 1465 m) site surrounded by croplands, an urban area (Clermont-Ferrand) and 95 forests within a 50 km perimeter (Fig. 1). Monthly NASA satellite images of puy de Dôme 96 surrounding surfaces (https://wvs.earthdata.nasa.gov/) are shown in Fig. S1. The Atlantic 97 coast and Mediterranean Sea are at around 320 km and 240 km from PDD, respectively. 98 99 100
4

101 102 Fig. 1. Geographical location and landscape of the sampling site. Map showing the 103 location of the puy de Dôme mountain in France and relative surfaces types surrounding the 104 site in a perimeter of 50 km based on the MODIS satellite images. Cropland and vegetation 105 areas comprise > 80% of the surrounding landscapes, while forest and urban areas (mainly 106 Clermont-Ferrand) comprise < 20% of the surrounding landscapes. 107 108 109 DNA extraction 110 We extracted DNA from 3 punches (diameter of one punch: 38 mm) from the quartz fiber 111 filters using the DNeasy PowerWater kit with some modifications as detailed in Dommergue 112 et al. (2019)18. During cell lysis, an one hour heating step at 65°C followed by a 10-min 113 vortex treatment at maximum speed and a centrifugation using a syringe to separate the filter 114 debris from the lysate were added to the DNeasy PowerWater DNA extraction protocol18. 115 DNA concentration was measured using the High Sensitive Qubit Fluorometric 116 Quantification (Thermo Fisher Scientific) then stored at -20°C. 117 118 Real-Time qPCR analyses
5

119 The bacterial cell concentration was approximated by the number of 16S rRNA gene copies

120 per cubic meter of air and the fungal cell concentration was approximated by the number of

121 18S rRNA gene copies per cubic meter of air. Primers and methodology are presented in 122 Tignat-Perrier et al. (2019)1.

123

124 MiSeq Illumina amplicon sequencing

125 16S rRNA gene sequencing: library preparation, reads quality filtering and taxonomic

126 annotation. The V3-V4 region of the 16S rRNA gene was amplified and libraries were 127 prepared as presented in Tignat-Perrier et al. (2019)1. The amplicons were sequenced by

128 paired-end MiSeq sequencing using the V3 Illumina technology with 2 x 250 cycles. Reads

129 were

filtered

based

on

quality

using

FASTX-Toolkit

130 (http://hannonlab.cshl.edu/fastx_toolkit/), assembled using PANDAseq19, and annotated using

131 RDP Classifier20 and the RDP 16srRNA database as detailed in Tignat-Perrier et al. (2019)1.

132 RDP classifier was used in part to avoid errors due to sequence clustering. The number of

133 sequences per sample and the percentage of sequences annotated at the genus level were

134 evaluated using a home-made R script. The sequences annotated as chloroplasts by RDP were

135 manually removed.

136 ITS rRNA gene sequencing: library preparation, reads quality filtering and taxonomic

137 annotation. The ITS2 region of the ITS was amplified libraries were prepared as presented in 138 Tignat-Perrier et al. (2019)1. The amplicons were sequenced by a paired-end MiSeq

139 sequencing using the technology V2 of Illumina with 2 x 250 cycles. Reads were filtered

140 based on quality using FASTX-Toolkit, assembled using PANDAseq, and annotated at the 141 species level21 using RDP Classifier and the RDP fungallsu database as detailed in Tignat142 Perrier et al. (2019)1. The number of sequences per sample and the percentage of sequences

143 annotated at the species level were evaluated using a home-made R script.

6

144 The number of reads per sample and per sequencing (V3-V4 regions of the 16S rRNA gene 145 and the ITS2 region of the ITS) is presented in the Table S1. Samples with less than 6000 146 reads were removed. Samples from the same season were pooled and rarefaction curves per 147 season were done (Fig. S2). 148 149 Estimation of the trophic mode of the fungal species 150 We used the FUNGuild software22 to assign the trophic mode of the fungal species (RDP 151 classifier based annotation). Fungal species annotated to a trophic mode with the level 152 confidence “Possible” were grouped in the “not classified” fungi. Then, we calculated the 153 percentage represented by each trophic mode per sample. Heatmaps were done using the R 154 package gplots. 155 156 Chemical analyses 157 The elemental carbon (EC), organic carbon (OC), sugar anhydrides and alcohols 158 (levoglucosan, mannosan, galactosan, inositol, glycerol, erythriol, xylitol, arabitol, sorbitol, 159 mannitol, trehalose, rhamnose, glucose, fructose and sucrose), soluble anions (MSA, SO42-, 160 NO3-, Cl-, Ox) and cations (Na+, NH4+, K+, Mg2+, Ca2+) concentrations were analyzed as 161 presented in Dommergue et al. (2019)18. 162 163 Meteorological data 164 Meteorological parameters such as wind speed and direction, temperature, relative humidity 165 and UV were collected (Vaisala instrument). For each sample, the backtrajectories of the air 166 masses were calculated over 3 days before the sampling using HYSPLIT23 (maximum height 167 above ground level: 1 km). 168
7

169 Graphical and Statistical analyses 170 Environmental variables. For chemical species, air concentrations in ng per cubic meter of air 171 were used in the analyses. The chemical table was log10-transformed to approach a Gaussian 172 distribution (verified on a Q-Q plot and tested using the Shapiro-Wilk test), and a hierarchical 173 cluster analysis (average method) was done on the Euclidean distance matrix using the vegan 174 and ade4 R packages. Meteorological data were used to do the wind roses using the openair R 175 package24. Backtrajectories of the air masses over three days were plotted on maps using the 176 openair R package, and the relative surfaces of the landscapes (MODIS land surfaces) air 177 masses over flown were calculated. 178 Diversity statistics and Multivariate analyses. Before doing the multivariate analyses, the raw 179 abundances of the taxa (bacterial genera and fungal species) were transformed into relative 180 abundances to counter the heterogeneity in the number of sequences per sample. A 181 hierarchical cluster analysis (average metric) on the Bray-Curtis dissimilarity matrix was done 182 using the vegan and ade4 R packages25. We have defined the seasons as following: we 183 adjusted the beginning and ending administrative dates of the seasons (i.e. summer: 20th of 184 June to 22th of September; autumn: 23th of September to 21th of December; winter: 22th of 185 December to 19th of March; spring: 20th of March to 19th of June) based on the weekly mean 186 temperature (Fig. S3). France goes through a cycle of four strong seasons characterized by 187 distinctive weather: the summer is characterized by higher temperatures, a longer daylight 188 period and the period of fructification of fruiting plants; the autumn is the harvest period of 189 most annual crops and a preparation period for dormancy; winter is characterized by the 190 lowest temperatures, frequent precipitation (rain and snow) and high air humidity; and finally 191 the spring is characterized by milder weather, snow melting, budding and flowering. To 192 access the variability of microbial population structure within the seasons, we averaged the 193 degrees of dissimilarity obtained from the Bray-Curtis matrix for each pair of samples from
8

194 the same season, subtracted these values from 1 to get similarity values and divided the 195 similarity values by the standard deviation (as detailed in Tignat-Perrier et al., 20191). 196 Spearman correlations were calculated to test the correlation between microbial abundance 197 and richness and quantitative environmental factors using the Hmisc R package26. ANOVA 198 analyses were used to test the influence of qualitative factors such as season and year on both 199 bacterial and fungal abundance and richness using the vegan R package, followed by 200 TukeyHSD tests to identify which group revealed a significantly different mean. A distance201 based redundancy analysis (RDA – linear or non-linear correlation) was carried out to 202 evaluate the part of the variance between the samples explained by the seasons, chemistry 203 and/or meteorology, and an ANOVA was carried out to test each variable using the vegan and 204 ade4 R packages. Venn diagrams using the R package VennDiagram were done to access the 205 shared and unique bacterial genera and fungal species from each season after rarefaction on 206 the raw abundances (rarefaction at the minimum number of reads). A Mantel test between the 207 Bray-Curtis matrices based on the bacterial genera and fungal species was used to evaluate 208 similarities in the distribution of the samples. A Mantel test was done between the Bray-Curtis 209 matrix based on either the bacterial or the fungal diversity and the Euclidean distance matrix 210 based on the chemical variables or meteorology to evaluate the similarities in the distribution 211 of the samples. 212 213 214 Results 215 Temporal distribution of airborne microbial communities 216 Airborne microbial abundance. Airborne bacterial and fungal concentrations (estimated by 217 the number of 16S rRNA and 18S rRNA gene copies) were positively correlated (r=0.77, 218 pvalue=7.9×10-11) and varied between 1.8×103 and 2.1×107 cells per cubic meter of air and 3
9
CommunitiesPuyDômeSeasonsShift