, Tween 20 for 2h30 under constant agitation 143 (140 rpm) at room temperature. After this seed soak that enables the recovery of epiphytic 144 and endophytic microorganisms

, following the manufacturer's procedure. Seed 147 suspensions in PBS are classically performed in seed pathology (see International Seed 148 Testing Association, DNeasy PowerSoil HTP 96 Kit (Qiagen)

. Barret, which target a portion of gyrB and 155 the fungal ITS1 region, respectively. The cycling conditions for ITS1F/ITS2, were an initial 156 denaturation at 94°C for 3 min, followed by 35 cycles of amplification at 94°C (30 s), vol.50, p.45, 2009.

, 68°C (90 s), and a final elongation step at 68°C for 10 min. The cycling conditions for 158 gyrB were: initial denaturation at 94°C for 3 min, 35 cycles of amplification at 94°C

, A second PCR amplification was 161 performed to incorporate Illumina adapters and barcodes. PCR cycling conditions were 162 identical for the two molecular markers: a first denaturation step at 94°C (1 min), 55°C (45 s) and 68°C (90 sec), and final elongation step at 68°C for 10 min. Amplicons were 160 purified with magnetic beads (Sera-Mag TM , Merck)

, Changes in microbial assemblage composition were assessed with Bray-Curtis (BC) index 188 and weighted UniFrac (wUF) distance (Lozupone & Knight, 2005.

A. , , 2001.

L. , Common ASV 195 was recorded as present if detected in all subsamples (n=5) of one seed lot. One seed lot 196 corresponded to seeds collected each year from each plant genotype. Visualization of 197 genotype-specific ASVs and shared ASVs between plant genotypes was assessed for each 198 year with the R package UpSetR, A presence/absence matrix of common ASVs was constructed as follows, 2014.

. Barret, Distribution of bacterial and fungal ASVs in other Brassicaceae seeds was assessed by 204 investigating their prevalence in 479 seed samples associated to Brassica oleracea, Brassica 205 rapa and Raphanus sativus, 2015.

, Overall 4,451 bacterial and 359 fungal ASVs were detected within gyrB and ITS1 datasets

. Callahan, this descriptor was 213 therefore used for estimating the taxonomic structure of the seed microbiota of B. napus, ASV provided a finer resolution than OTU, p.214, 2017.

, According to coverage estimates, the number of gyrB sequences was not sufficient for 215 reaching the plateau after rarefaction, while saturation was achieved for fungal reads

. Hence, Although 218 bacterial and fungal richness were significantly different between the two harvesting years 219 (P < 0.01), seed samples collected in Y2 contained on average fewer bacterial and more 220 fungal ASVs than Y1 seed sample (Fig. 1A, D). The increase of fungal richness in Y2 was 221 associated with a significant decrease in fungal diversity (Fig. 1E). While bacterial diversity 222 did not differ between years (Fig. 1B), bacterial phylogenetic diversity (Faith's PD), Chao1 index) was assessed for bacterial assemblages, while 217 observed richness (number of ASVs) was measured for fungal assemblages

, Similarities in composition of bacterial and fungal assemblages were estimated with 226 weighted UniFrac (wUF) distance and Bray-Curtis (BC) index, respectively. Ordination of wUF 227 and BC revealed a significant clustering of seed-associated bacterial and fungal assemblages 228 according to the harvesting year (Fig. 2, Table 1). The relative influence of the harvesting

, However, according to average 281 germination rate (n=100), germination of seed samples harvested in Y1 was significantly (P < were also observed between genotypes within specific year. For instance, seeds 284 from Astrid germinated on average earlier (P <0.01) than the other genotypes in Y1 but not 285 in Y2 (Fig. 6), There were no significant differences between years and genotypes for seed area, radicle 280 elongation and imbibition speed (Supplementary Fig. S6)

, Moreover, 3 bacterial ASVs and 5 fungal ASVs were specifically associated to Tenor in Y2 290

. Figures, Tables and Supporting information captions

, Community profiling of 630 the seed microbiota was performed on five seed subsamples per genotype. (A) Estimated 631 richness (Chao1 index), (B) diversity (Inverse Simpson's index) and (C) phylogenetic diversity 632 (Faith's phylogenetic diversity index) were monitored with gyrB ASVs for bacteria. (D) 633 Observed richness (number of detected ASVs) and (E) diversity (Inverse Simpson's index) UniFrac distance (A) and Bray-Curtis index for bacterial and fungal assemblages (B), 639 respectively, Diversity of seed-associated microbial assemblages of B. napus

, UniFrac distance and Bray-Curtis index. Samples are coloured according to their genotypes

, Taxonomic composition of B. napus seed microbiota. Relative abundance of the most 644 abundant bacterial (A) and fungal (B) order within B. napus seed samples collected from nine 645 genotypes across two successive generations

. Bartoli, ASVs was performed with an in-house gyrB database, vol.1, p.647, 2010.

L. , Plant genotype 653 associated with each particular intersection is highlighted with a dot. Intersection associated 654 with all genotypes is displayed in red. The histogram located on the left of each UpSetR 655 representation, Generalist and specific bacterial ASVs. Prevalence of bacterial ASVs across plant 651 genotypes were investigated with UpSetR, 2014.

L. , Plant genotype(s) 661 associated with each particular intersection is (are) highlighted with a dot. Intersection 662 associated with all genotypes is displayed in red. The histogram located on the left of each phenotyping platform, Generalist and specific fungal ASVs. Prevalence of fungal ASVs across plant genotypes 659 were investigated with UpSetR, vol.669, 2004.

, Each genetic group is represented with a colour. A cultivar belongs to the group of 677 which the colour is majority, WOSR genetic groups. Cultivars were genotyped using 628 SNP (Clarke 675, vol.676, 2009.

, A) Rainfall level in mm (RR-DC) and 681 mean humidity in % (UM-MD). (B) Mean calculated temperature (TMC), minimal (TN) and 682 maximal (TX) temperatures in °C. All data are reported for 10 days (decade) during the years 683 2015-2016 (Y1) and 2016-2017 (Y2) from records at Le Rheu station 35240002 -La Grande-684 Verrière, S2 Climatic data for the cropping periods Y1 and Y2, p.685

. Fig and . S3, Similarities in seed microbial composition were estimated 688 with weighted UniFrac distance (A, B) and Bray-Curtis index (C, D) for bacterial and fungal 689 assemblages, respectively. Principal coordinate analysis (PCoA) was used for ordination of 690 weighted UniFrac distance and Bray-Curtis index. Samples are colored according to their 691 genotypes

. Fig, S4 Number of bacterial (A) and fungal (B) ASVs shared between years or specific of one 694 year, for each genotype

. Fig, S5 Relationships between B. napus genetic distances and similarities of microbial

, Michener genetic distance and unweighted UniFrac distance (bacteria; A, B) or Jaccard index 700 (fungi; C, D) for Y1 (A, C) and Y2

. Fig, S6 Different phenotyping parameters calculated for seeds of each year and genotype

, A) Seed surface at initial time; (B) Seed imbibition 8h after initiation of imbibition

, Radicle elongation 8h after initiation of germination; (D) Time at which a half of the seeds

, Quality: 00 or ++ is for absence or 708 presence of Glucosinolate-Erucic acid, WOSR groups are defined from the genetic diversity 709 study of 116 WOSR accessions with 628 SNPs (Fig. S1)

, Table S2 Cropping conditions for the B. napus seed production

, Table S3 Genotype-specific bacterial and fungal ASVs for Y1

, Table S4 Genotype-specific bacterial and fungal ASVs for Y2