Microbial community dispersal in sourdough

Understanding how microbes disperse in ecosystems is critical to understand the dynamics and evolution of microbial communities. However, microbial dispersal is difficult to study because of uncertainty about the vectors that may contribute to their migration. This applies to both microbial communities in natural and human-associated environments. Here, we studied microbial dispersal among French sourdoughs and flours used to make bread. Sourdough is a naturally fermented mixture of flour and water. It hosts a community of bacteria and yeasts whose origins are only partially known. We analyzed whether flour is a carrier of sourdough yeast and bacteria and studied whether microbial migration occurs between sourdoughs. The microbial community of a collection of 46 sourdough samples, as well as that of the flour from which each was made, was studied by 16S rDNA and ITS1 metabarcoding. No sourdough yeast species were detected in the flours. Sourdough lactic acid bacteria (LAB) were found in only five flour samples, and they did not have the same amplicon sequence variant (ASV) as found in the corresponding sourdough. The species shared between the sourdough and flour samples are commonly found on plants and are not known to be alive in sourdough. Thus, the flour microorganisms did not appear to grow in the sourdough microbial community. Dispersal between sourdoughs was also studied. Sourdoughs shared no yeast ASV, except in few cases where groups of three to five bakers shared some. These results suggest that there is little migration between sourdoughs, except in a few situations where bakers may exchange sourdough or be vectors of yeast dispersal themselves.


185
Although sourdoughs had a higher microbial density than flour, their microbial communities 186 were less diverse than those in flour. Alpha diversity indexes calculated on the number of bacterial 187 and fungal species were significantly lower in sourdough than in flour in terms of both richness 188 (Wilcoxon-Mann-Witney test, bacteria W = 1725.5, P < 0.001, fungi W = 1555.5, P < 0001) 189 and evenness (Wilcoxon-Mann-Witney test, bacteria W = 1929, P < 0.001, fungi W = 1467, 190 P < 0001 ; Figure 1). This difference was greater for bacteria than for fungi, with averages of 191 four and 11 species for bacteria in sourdough and flour, respectively, and 10 and 13 species for 192 fungi in sourdough and flour, respectively.

195
The microbiota compositions of sourdough and flour were characterized by different families.

196
The bacteria in the sourdoughs were almost entirely composed of Lactobacillaceae, while flour 197 contained mainly Erwiniaceae and Pseudomonadaceae. In sourdough, all samples but three contai-198 ned Fructilactobacillus sanfranciscensis as the dominant bacterial species ; the others contained 199 Companilactobacillus paralimentarius. Less frequently, the presence of Levilactobacillus brevis, 200 Latilactobacillus sp. and Lactilactobacillus sp. was found. In flour, Erwiniaceae, Pantoea agglo-201 merans, an unidentified Pantoae sp., and Pseudomonadaceae were generally detected. Among

202
Pseudomonas sp., some were P. graminis, P. rhizospherae or P. donghuensis. As for fungi, Sac-  Any overlaps between the sourdough and flour communities were analyzed using the Weighted 212 Bray-Curtis distance calculated on the basis of species diversity. The Weighted Bray-Curtis was 213 used to build two PCoAs, one for the bacterial community and the other for the fungal community.

214
PCoA axis 1 and 2 explained 79.1% and 8.5% of variance for bacteria, and 28.5% and 13.6% communities. Flour and sourdough dissimilarity matrices were not correlated (Mantel test, z = 219 836, p = 0.667 for bacteria and z = 854, p = 0.13 for fungi). Close microbial communities among 220 flours did not lead to close microbial communities among sourdoughs. 221 We analyzed bread-making practices in order to determine whether they might be related 222 to microbial communities in sourdough and flour. Two groups of bread-making practices could 223 be distinguished ( Figure S1). Farmer-baker practices (cluster 1) were more frequently associated 224 with the use of non-commercial yeast, ancient wheat landraces, small production runs and leng-225 thy fermentation while artisanal practices (cluster 2) were generally characterized by larger scale 226 production, short fermentation, and the use of commercial yeast and modern wheat varieties.

227
Sourdough from farmer-bakers frequently contained K. bulderi as the dominant yeast species.

228
However, analysis of the association between sourdough and flour microbial community dissimila-229 rity and the geographical distances between bread-making practices did not reveal any correlation

259
We then studied the occurrence of ASVs in the most abundant bacteria, F. sanfranciscensis, 260 and found they were present in all sourdoughs. By contrast, the ASVs of the dominant sourdough 261 yeast species (S. cerevisiae, K. bulderi and K. humilis) were generally specific to a single sourdough 262 ( Figure 6). However, some ASVs were found in several sourdoughs. Sourdoughs from bakers B12, 263 B15, B26 and B63 shared one K. bulderi ASV. Sourdoughs from bakers B04, B17, B31, B56 and 264 B58 shared one to three K. humilis ASVs. Sourdoughs from bakers B29, B55 and B74 shared one 265 to two S. cerevisiae ASVs and those from bakers B01, B16, B17 and B32 shared another ASV.

266
Bakers who shared a yeast ASV generally belonged to the same bakery practices cluster. The 267 group of three bakers who shared one or two S. cerevisiae belonged to cluster 2 (corresponding 268 to artisanal bakery practices) while three of the four bakers who shared another S. cerevisiae ASV 269 shared at least one K. humilis ASV belonged to cluster 2, but the two others, who belonged to 271 cluster 1, shared more ASVs than with the three others ( Figure 5). An evaluation of the association 272 between sourdough fungal community dissimilarity and geographical distances did not reveal any 273 significant correlation (mantel z = 363535.1, P = 0.547). The only link that could be made from 274 the data on sourdough exchanges concerned farmer-baker B15, who shared a K. bulderi ASV with 275 farmer-baker B12, and started his sourdough using B12 sourdough. to a low number of reads. The reads might also not have passed the quality filtering or merging 289 steps in the bioinformatics analysis, particularly if the ITS region was too long. This is a limitation 290 of the dada2 software, where reads that are too long to be merged are lost. However, this does 291 not concern the ITS database, as in this case the ASV would have been found but not assigned 292 to a species.     The exchange or gifting of sourdoughs between bakers can lead to yeast dispersal, as was 364 found between bakers B15 and B12 who were regularly in contact and exchanged their sourdoughs.

365
However, this practice is not common, as bakers prefer to develop their own sourdough when they 366 lose one. Finding the same sourdough yeast species in dough from several farmer-bakers could 367 be explained by the development of networks of bakers who meet to share their knowledge and 368 skills, and yeast dispersal may be promoted through handshakes. Student bakers traveling between 369 different bakeries may also be a vector for dispersal. Bakers belonging to the same bakery practices 370 cluster (artisanal or farmer-baker) tended to share more ASVs than with bakers from the other 371 cluster (see Figure 6). However, the number of bakers sharing yeast ASVs was quite low : four, 372 six, and eight bakers shared at least one K. bulderi, K. humilis and S. cerevisiae ASV respectively, 373 so we were not able to perform a robust statistical analysis.   B01  B02  B04  B07  B09  B11  B12  B13  B14  B15  B16  B17  B18  B23  B25  B26  B29  B31  B32  B42  B45  B52  B53  B54  B55  B56  B57  B58  B59  B60  B61  B62  B63  B64  B65  B66  B67  B68  B69  B70  B71  B72  B73  B74 Figure 5 -Number of shared species (on the right) and ASV (on the left) between sourdoughs and the flour used to make them. Results for bacteria are shown at the top and for fungi at the bottom.  Figure 6 -Sourdoughs sharing K. bulderi, K. humilis and S. cerevisiae ASVs. Top, the heatmaps show the number of shared ASV between sourdoughs, each tile being colored according to the number of shared ASV. In the diagonal, the number of ASVs of the considered species in each sourdough are displayed, and the tiles are underlined according to the cluster of bread-making practices (1 = farmer-baker and 2 = artisan-baker). At the bottom, the maps of France show the locations of each baker. Bakers are represented by a point when the species considered was not detected in their sourdough, and in the other case the pie charts show the composition of their sourdoughs. ASVs that are shared between at least two different sourdoughs are colored and their identifiers displayed in the legend, while the ASVs of species considered to be specific to one sourdough are represented in black (SourdSpe in the legend), while ASVs from other species are in grey.