Seed tuber imprinting shapes the next-generation potato microbiome

Background Potato seed tubers are colonized and inhabited by soil-borne microbes, that can affect the performance of the emerging daughter plant in the next season. In this study, we investigated the intergenerational inheritance of microbiota from seed tubers to next-season daughter plants under field condition by amplicon sequencing of bacterial and fungal microbiota associated with tubers and roots, and tracked the microbial transmission from different seed tuber compartments to sprouts. Results We observed that field of production and potato genotype significantly (P < 0.01) affected the composition of the seed tuber microbiome and that these differences persisted during winter storage of the seed tubers. Remarkably, when seed tubers from different production fields were planted in a single trial field, the microbiomes of daughter tubers and roots of the emerging plants could still be distinguished (P < 0.01) according to the production field of the seed tuber. Surprisingly, we found little vertical inheritance of field-unique microbes from the seed tuber to the daughter tubers and roots, constituting less than 0.2% of their respective microbial communities. However, under controlled conditions, around 98% of the sprout microbiome was found to originate from the seed tuber and had retained their field-specific patterns. Conclusions The field of production shapes the microbiome of seed tubers, emerging potato plants and even the microbiome of newly formed daughter tubers. Different compartments of seed tubers harbor distinct microbiomes. Both bacteria and fungi on seed tubers have the potential of being vertically transmitted to the sprouts, and the sprout subsequently promotes proliferation of a select number of microbes from the seed tuber. Recognizing the role of plant microbiomes in plant health, the initial microbiome of seed tubers specifically or planting materials in general is an overlooked trait. Elucidating the relative importance of the initial microbiome and the mechanisms by which the origin of planting materials affect microbiome assembly will pave the way for the development of microbiome-based predictive models that may predict the quality of seed tuber lots, ultimately facilitating microbiome-improved potato cultivation. Supplementary Information The online version contains supplementary material available at 10.1186/s40793-024-00553-w.

The microbial community associated with a plant, referred to as the plant microbiome, can 22 significantly influence plant performance. The complex plant microbiome includes microbes that are 23 plant pathogens but also plant beneficial microbes that support plant growth by mobilizing scarce 24 nutrients from the soil or protect the plant against pathogens 1, 2, 3 . The plant microbiome 25 significantly expands the genomic potential of its host and is often referred to as the host's "second 26 genome" 1, 2, 4, 5, 6, 7 . 27 28 Potato is the 3rd most important crop for human consumption, with an annual global harvest of 29 approximately 375 million tons. Additionally, it is a key crop that is essential for global food security 30 and a source of raw materials for industry (www.fao.org, 2017). Potato is a space-efficient crop, 31 yielding five times more consumable weight per hectare than rice and wheat. As global demand for 32 potato increases, the UN-FAO identified it as a crop with great potential to become a game changer 33 for global food security 8,9 . 34 35 Potatoes are commonly propagated vegetatively by transplanting seed tubers from one field to the 36 next 10 . As potato tubers develop underground, they closely interact with the dense and diverse 37 microbial communities in soil 11 . Studies demonstrated that the potato tuber microbiome can have a 38 profound impact on plant health and productivity 12,13 . Potato is sensitive to a wide range of plant 39 pathogens 14, 15 , but it also hosts beneficial microbes that can promote plant growth 12, 13, 16, 17, 18 . 40 41 A batch of seed potatoes is of high vitality if it manifests in a large canopy and exhibits homogeneous 42 growth in the early stages of its development. Seed tubers of the same potato genotype that were 43 produced in different production fields can display significant differences in their vitality, resulting in 44 differences in growth of the emerging potato plants 19,20,21 . This may be caused by local 45 environmental factors in the fields of production that confer changes in tuber physiology, but also 46 the seed tuber microbiome likely impacts the vitality of the outgrowing potato crop. Many potato 47 pathogens can be seed tuber-borne 15,22 . Field experiments in which seed tubers were treated with 48 beneficial bacteria show that the applied microbes colonize the roots of plants that develop from the 49 treated tubers 23,24 . Such findings suggest that seed tubers can be an important inoculum source of 50 microbes for the potato plants that emerge from them and that potato plants may inherit at least 51 part of their microbiome from the seed tuber. However, there is limited information available on 52 tuber-borne transmission of microbes from one potato generation to the next. To gain insight into 53 intergenerational inheritance of the potato microbiome, we investigated whether the field of 54 production of potato seed tubers has an impact on the microbiomes of tubers and roots of plants 55 emerging from these seed tubers when planted together in a single trial field. 56

57
Effect of potato genotype, production field, and storage on the tuber microbiome 58 In the autumn of 2018, seed tubers of two potato varieties, Colomba (hereafter Variety A) and 59 Innovator (hereafter Variety B), were harvested from 3 fields of production for Variety A and 3 60 different fields for Variety B ( Fig. 1a; Fig. S1a-b). To investigate the influence of plant genotype and 61 field of production on the tuber-associated microbiome, we isolated microbial DNA from 4 replicate 62 samples per field, each replicate containing peels of 6 tubers. Subsequently, we sequenced 16S rRNA 63 gene and ITS amplicons to profile the bacterial and fungal communities, respectively. Principal 64 coordinates analysis (PCoA) and permutational multivariate analysis of variance (PERMANOVA), 65 revealed that both the bacterial and the fungal microbiome on the tuber is determined primarily by 66 the field in which the potato was produced (Fig. 1b-d; Fig. S2a-c; Table S1; Table S2). The production 67 field significantly (P = 0.001) affected the tuber peel microbiome and accounted for up to 64% of the 68 variation in the bacterial community (R 2 = 0.64, Table S1) and 55% of the variation in the fungal 69 community (R 2 = 0.55, Table S2). In addition, the potato variety significantly (P = 0.001) affected 70 tuber microbiome composition, explaining 18% (R 2 = 0.18) and 17% (R 2 = 0.17) of the variation in 71 bacterial and fungal community composition, respectively ( Fig. 1b-d; Fig. S2a-c; Table S1; Table S2). 72 It is common agricultural practice to store seed tubers over the winter prior to planting in spring. To 73 study the effects of cold storage on tuber microbiomes, the above-mentioned seed tubers had 74 remained in cold storage at 4℃ in the dark for 7 months (Fig. 1a). These so-called post-storage seed 75 tubers were then processed in the same manner as the seed tuber samples, after which the bacterial 76 and fungal microbial communities were profiled by amplicon sequencing. Although there were 77 significant changes in the composition of the bacterial (P = 0.001) and fungal (P = 0.019) microbiome 78 before and after storage of the tubers ( Fig. S3a and c, Table S3), post-storage seed tubers clustered 79 closely with those of the pre-storage seed tubers from the same field of production ( Fig. S3b and d). 80 Notably, tubers from different fields of production maintained their distinct microbial community 81 patterns even after 7 months of cold storage (  Table S1; Table S2). On post-82 storage seed tubers, the production field accounted for up to 57% of the variation in the bacterial 83 community (R 2 = 0.57, Table S1) and 46% of the variation in the fungal community (R 2 = 0.46, Table  84 S2). 85 86 87 Fig. 1 Bacterial community composition of seed tuber and post-storage seed tuber samples. a Graphic representation of the sampling strategy. Seed tubers of two potato varieties (A and B) were harvested from 3 fields of production per variety and sampled for microbiome analysis before and after a 7-month cold storage period. Principle component analysis (PCoA) of 16S amplicon sequencing data representing bacterial communities on b) seed tubers of Variety A and B, c) seed tubers of Variety A only, or d) seed tubers of Variety B only, e) poststorage seed tubers of Variety A and B, f) post-storage seed tubers of Variety A only, and g) post-storage seed tubers of Variety B only. Each symbol represents the bacterial community of one replicate potato peel sample. Each sample consists of a pool of potato peels collected from 6 seed tubers. For each variety, 4 replicate of seed tuber samples and 6 replicate of post-storage seed tuber samples were collected from each of the 3 fields of production. Green symbols represent Variety A and orange symbols represent Variety B. Different shapes within a same color represent distinct production fields. The P from PERMANOVA is shown in each PCoA plot. Each ellipse represents a 68% confidence region and depicts the spread of data points within each group.

Seed tubers, roots of emerging plants, and daughter tubers harbor distinct microbiomes 88
Seed tubers of Variety A and B of the above-mentioned 6 production fields were subsequently 89 planted in a single trial field near Veenklooster, the Netherlands, in the spring of 2019 (Fig. S1a). The  90  emerging plants from these seed tubers were cultivated for three months after which roots and  91 daughter tubers were harvested (Fig. 2a). The microbiome composition of these potato samples was 92 analyzed by sequencing both 16S rRNA gene and ITS amplicons. Using PCoA, we observed that the 93 bacterial community composition of both roots and tubers harvested in 2019 from the trial field 94 clearly separated (P = 0.001) from the seed tuber samples harvested from the production fields in 95 2018 ( Fig. 2b; Table S4). In addition, the bacterial communities found on the roots are distinct from 96 those on daughter tubers, indicating that these two belowground potato organs harbor distinct 97 bacterial microbiomes within one field (P = 0.001, Fig. 2b; Table S4). A similar separation was 98 observed for the fungal communities on seed tubers, roots, and daughter tubers (Fig. S4a, Table S4). 99 We then focused on shared bacterial amplicon sequence variants (ASVs) between the microbiomes 100 of seed tuber, daughter tuber, and root samples (Fig. 2c). A total of 3986, 9205, and 11622 unique 101 bacterial ASVs were detected in seed tuber, daughter tuber, and root samples, respectively. 102 Whereas 86% ((6830+1050)/9205) of the bacterial ASVs on the daughter tubers were shared with 103 those on roots of the potato plants in the same trial field, only 13% ((1050+156)/9205) and 12% 104 ((1050+393)/11622) of the ASVs on the daughter tubers and roots, respectively, were also detected 105 on the seed tuber (Fig. 2c). Analysis of the fungal microbial communities showed similar results with 106 84% ((758+182)/1117) of the fungal ASVs from daughter tubers shared with those on roots, while 107 only 18% ((182+22)/1117) and 16% ((182+37)/1405) of the ASVs detected on the daughter tubers 108 and roots, respectively, were also detected on the seed tubers (Fig. S4b). This suggests that the 109 majority of microbes on potato daughter tubers and roots are not inherited from the seed tubers 110 but originate from the trial field. 111 The most abundant bacterial phyla in the microbiomes of all tuber and root samples were the 112 Proteobacteria, Actinobacteria, Bacteroidetes and Firmicutes. Whereas Bacteroidetes were relatively 113 abundant in samples from plants in the trial field, Firmicutes had relatively low abundance in 114 samples from this field, especially on the daughter tubers. On those daughter tuber samples 115 Bacteroidetes were relatively more abundant, whereas Actinobacteria, Firmicutes, and 116 Planctomycetes had higher relative abundance on the roots of the potato plants in the same field 117 (Fig. 2d). 118 119 120

Origin of seed tubers affects the root and tuber microbiomes of emerging plants 121
Within the trial field, the bacterial and fungal microbial communities of both potato roots and 122 daughter tubers were significantly (P = 0.001) affected by potato genotype (Fig. 3a- Fig. S2g and j, 123 Table S1, Table S2). The effect size of potato genotype was larger for the tuber samples (R 2 = 0.08) 124 than for the root samples (R 2 = 0.03). Interestingly, also the field of production of the seed tubers 125 had a significant effect on microbiome composition of daughter tubers (P = 0.001, R 2 = 0.07) and 126 Fig. 2 Analysis of bacterial communities on seed tubers from different production fields and their roots and daughter tubers the Veenklooster trial field. a Graphic representation of the experimental design. Seed tubers from 3 different production fields of each variety (n = 2) were planted together in a single trial field in Veenklooster (Fig. S1). For seed tubers from each production field, 4 replicate plots were randomly distributed across this trial field. Roots and daughter tubers from the emerging plants were harvested for microbiome analysis. b PCoA of potato-associated bacterial communities of seed tubers, daughter tubers, and roots. Square symbols represent Variety A and triangle symbols represent Variety B. Colors represent different sample types. Each ellipse represents a 68% confidence region and depicts the spread of data points within each group. c UpSet plot showing the number of bacterial ASVs that are shared between or are unique for seed tubers, daughter tubers and roots of both varieties combined. d Stacked bar chart of the taxonomic composition of bacterial communities of different sample types aggregated at the phylum level. Each stacked column represents an independent sample (n = 216). Different colors within a column represent different phyla. Only the top 10 most-abundant phyla were colored individually, all the rest are colored in gray and listed as "Other phyla". Samples are clustered by sample type and production field, which is shown by the colored bar on top of the stacked bar chart. roots (P = 0.001, R 2 = 0.08) of the plants emerging from these seed tubers ( Fig. 3c- Table  127 S1, Table S2). Thus, the impact of the production field stretches across a generation and influences 128 microbiome assembly on the roots and tubers of the daughter plants emerging from the seed tubers 129 in the subsequent growing season. 130 131 132

Inheritance of field-unique ASVs in daughter tubers and roots 133
We hypothesized that the intergenerational influence of the seed tuber production field on the 134 microbiome of roots and daughter tubers is the result of vertical, seed tuber-mediated transmission 135 of field-unique microbes from one generation of potatoes to the next. To be able to track the 136 vertical inheritance of field-unique microbes from seed tubers to the emerging plants, we focused 137 on Variety A seed tubers from Field 1 and identified bacterial and fungal ASVs that were uniquely 138 detected in seed tuber samples from Field 1. We observed that 50.6% of the bacterial ASVs on seed 139 tubers from Field 1 were not detected on seed tubers from Field 2 and 3 and defined these 952 ASVs 140 as Field-1-unique on seed tubers (Fig. S5a). With the same definition, we identified 1451 bacterial 141 ASVs (29.7% of total daughter tuber ASVs) as Field-1-unique on daughter tubers that originate from 142 Field-1 seed tubers and 1244 bacterial ASVs (20.2% of total root ASVs) as Field-1-unique on roots 143 that originate from Field-1 seed tubers ( Fig. S5c-d). An additional 54, 132 and 137 fungal ASVs were 144 defined as Field-1-unique on seed tubers, daughter tubers, and roots, respectively ( Fig. S5e-h). 145 Each symbol represents the bacterial community of one replicate potato peel sample. Each daughter tuber sample consisted of a pool of potato peels collected from 6 daughter tubers of one plant. Each root sample is a subset of the whole root of the same plant from which the daughter tubers were sampled. For each variety, 4 replicate samples were collected from each of the 4 randomly distributed replicate plots. Green symbols represent Variety A and orange symbols represent Variety B. Different shapes within a same color represent different production fields. The P from PERMANOVA is shown in each PCoA plot. Each ellipse represents a 68% confidence region and depicts the spread of data points within each group.
We subsequently investigated whether the Field-1-unique ASVs were transmitted to the roots and 146 daughter tubers of the plants emerging from these Field-1 seed tubers. To our surprise, the results 147 did not support our original hypothesis, and instead, we found only a very small overlap between 148 Field-1-unique ASVs of seed tubers and daughter tubers and roots derived from these Field-1 seed 149 tubers ( Fig. 4a-b and d-e, Fig. S6a-b and d-e). Namely, only 24 bacterial and 3 fungal Field-1-unique 150 ASVs were shared between seed tubers and the emerging daughter tubers. Similarly, only 26 151 bacterial and 1 fungal Field-1-unique ASVs were shared between seed tubers and the roots of 152 emerging plants ( Fig. 4a-b, Fig. S6a-b). Moreover, these ASVs were lowly abundant in daughter tuber 153 (Bacteria: 0.1%, fungi: 0.3%) and root (Bacteria: 0.05%, fungi: 0.09%) microbial communities (Fig. 4g-154 h, Fig. S6g-h). Thus, although we can distinguish ASVs unique to the field of production of the seed 155 tuber on the next season daughter tubers and roots, the large majority of the field-unique ASVs in 156 the daughter generation cannot be immediately traced back to the seed tuber. 157 When looking into the entire microbial community on seed tubers instead of only the field-unique 158 ones, we found that 83% (1556/1882, Fig. 4d) and 78% (1472/1882, Fig. 4e) of the seed tuber 159 bacterial ASVs were lost during vertical transmission to daughter tubers and roots, respectively. 160 Furthermore, 77.2% of the daughter tuber ( Fig. 4g) and 74.5% (Fig. 4h) of the root bacterial 161 communities were acquired from the environment during the 3 months of growth in the trial field. 162 Around a quarter of daughter tuber (22.8%, Fig. 4g) and root (25.6%, Fig. 4h) microbial communities 163 were shared with those on the peel of the seed tuber. However, since these ASVs were not Field 1-164 unique, it cannot be verified to what extent they are inherited from the seed tuber or simply 165 common in different fields. Similar results were observed for the fungal communities on the 166 daughter tubers and roots from Field 1 (Fig. S6). These results indicate that even though the field-167 unique ASVs were rarely inherited cross generations, we did observe vertical inheritance for other 168 ASVs from seed tubers to daughter tubers and roots. However, the majority of the microbial 169 population in daughter tubers and roots were acquired from the environment where they were 170 formed. 171 To investigate whether cold storage would already lead to the depletion of the above defined field-172 unique seed tuber microbes pre-planting, we examined the occurrence of ASVs on the post-storage 173 seed tubers from Field 1. These post-storage seed tubers were stored under cold and dark condition 174 much longer than common practice, thus used as an extreme case to study the influence of storage 175 on field-unique seed tuber microbes. We found that 66% (1051/1593) of the total bacterial ASVs 176 detected on the post-storage seed tubers were also detected on the pre-storage seed tubers from 177 the same field ( Fig. 4c and f) and that these ASVs represent 91.8% of the bacterial community ( Fig.  178 4i). These results indicated that the large majority of the seed tuber bacterial community persists 179 during cold storage. Moreover, a large part of the field-unique ASVs were maintained over the 180 storage period ( Fig. 4f and i, "Unique-Unique"). Similar results were observed for fungal 181 communities on the seed tuber and post-storage seed tubers from Field 1 ( Fig. S6f and i). 182 Fig. 4 Comparison of bacterial ASVs on daughter tubers, roots, post-storage seed tubers and seed tubers of Variety A originating from Field 1. Venn diagrams showing the overlap between a) seed tubers and daughter tubers, b) seed tubers and roots, c) seed tubers and post-storage seed tubers of Field-1unique bacterial ASVs (in red) or all bacterial ASVs (in blue). Sankey diagram of bacterial ASVs transferred from seed tubers to d, g) daughter tubers and e, h) roots that emerged from the seed tubers; and f, i) post-storage seed tubers. "Shared" in blue represents ASVs detected on both sample types. "Unique-Unique" in red represents the overlap of Field-1-unique ASVs on both sample types. The "Unique-Unique" in red is included in the "Shared" in blue. "Lost" in white represents ASVs lost from the seed tuber during vertical transmission. "Acquired" in light grey represents ASVs not transmitted from seed tubers but acquired from the environment. In a-f), numbers in the bars indicate the number of ASVs in each category mentioned above. In g-i), numbers in the bars indicate the accumulative relative abundance of ASVs in each category mentioned above.

Tracking the microbial transmission from different seed tuber compartments to sprouts
185 Even though the microbiomes on daughter tubers and roots of next-season potato plants could be 186 distinguished based on the field of production of the seed tuber, we found little evidence for direct 187 vertical transmission of microbes from the peel of the seed tuber to the peel of the tubers or roots 188 on the daughter plants. This could mean that: 1) potato daughter plants do not inherit their 189 microbiome from the peel but other compartments of the seed tuber; or 2) vertical transmission is 190 apparent only during early stages of plant development after which transmitted microbes are 191 replaced by members from the trial field resident microbiome. To gain further insight into the 192 potential of vertical microbiome transmission from seed tubers to next-generation daughter plants, 193 we investigated the contribution of different seed tuber compartments (namely peel, eye, heel end, 194 flesh, and adhering soil, Fig. S1c) in shaping the potato sprout microbiome. We made use of material 195 from a parallel study in which we harvested tubers from 6 potato varieties produced in 25 distinct 196 fields of production (Variety A from 5, Variety B from 5, Festien (Variety C) from 3, Challenger 197 (Variety D) from 5, Sagitta (Variety E) from 5, and Seresta (Variety F) from 2 fields, respectively; Fig.  198 S1b). Samples from 50 seed tubers were pooled into a single sample per compartment per field and 199 thus a total of 1250 (50 x 25) tubers were sampled from these 25 fields. DNA was isolated and 200 bacterial and fungal microbiome composition was determined through 16S rRNA gene and ITS 201 amplicon sequencing. 202 Again we found that potato genotype significantly influenced the composition of bacterial and 203 fungal communities in the distinct seed tuber compartments (Fig. 5a, Fig. S7a, Table S1, Table S2). 204 Moreover, we found that each distinct tuber compartment harbored a bacterial community that is 205 significantly different (P < 0.001) from the communities in the other compartments ( Fig. 5b-c, Table  206 S5), with the exception of the pairwise comparisons between eye and peel (P = 0.143) and between 207 eye and heel end (P =0.061). The richness of the bacterial communities decreased from the outside 208 of the potato to the inside, with highest diversity in the potato-adhering soil and increasingly lower 209 diversity in respectively the potato peel, heel end, eye, and flesh compartments (Fig. S8a). At phylum 210 level, Bacteroidetes and Proteobacteria have a higher relative abundance in the heel end 211 compartments compared to the other 4 tuber compartments (Fig. 5c). 212 Similar to the bacterial communities, fungal communities found in distinct compartments were 213 significantly different from each other (P < 0.001, Fig. S7b, Table S6), except for the eye and peel 214 compartments which harbored nearly identical fungal communities (P = 0.83, Table S6). The highest 215 richness for fungal communities was observed in adhering soil samples; however, diversity did not 216 differ significantly between the other compartments ( Fig. S8b). At family level, Cladosporiaceae was 217 most abundant in the adhering soil, whereas Plectosphaerellaceae was relatively more abundant in 218 the heel ends (Fig. S7c). 219 220 221 The spout is the first daughter tissue to emerge from the seed potato, and thus the most likely tissue 222 for vertical transmission of microbiota. To investigate vertical transmission of microbes from the 223 seed tuber to the emerging plant, seed tubers of all 6 varieties and from 2 fields per variety sprouted 224 on Petri dishes for 7 days. Subsequently, we isolated microbial DNA of sprouts of 5 replicate tubers 225 per field and analyzed microbiome composition of the samples through 16S rRNA gene and ITS 226 amplicon sequencing. The bacterial community composition of sprouts was significantly (P < 0.001) 227 different from those of all five distinguished compartments of the seed tuber (Table S7). At phylum 228 level, the bacterial community of the sprout was dominated by Actinobacteria, which were detected 229 at a relative abundance of 72% of the total community, whereas Firmicutes (15%) and 230 Proteobacteria (11%; Fig. S9) were also abundantly detected on sprouts. Also on sprouts, our 231 analysis revealed a significant impact of plant genotype on microbial community composition (P = 232 0.001; Fig. 6a). Interestingly, 4 of the 6 varieties of sprouts emerging from seed tubers originating 233 from different production fields had distinct microbiomes (Fig. 6b-g). These results indicate that the 234 sprout-associated microbiome is influenced by plant genotype, but also by the field of production of 235 the seed tuber. 236 Fig. 6 Field of production of the seed tuber affects the sprout microbiome. PCoA of bacterial sprout microbiomes of a) all varieties together and b-e) each variety separately. Each color represents one variety. Open and closed symbols represent distinct seed tuber production fields. The P from PERMANOVA is shown in each PCoA plot. Each sprout sample is a pool of 3-4 sprouts from one single tuber. Each ellipse represents a 68% confidence region and depicts the spread of data points within each group.
We next compared the microbiomes of the sprouts to the distinct compartments on the seed tubers 237 that were analyzed above to identify the sources for the sprout microbiome. For bacteria, the 238 analysis revealed that 79% (177 of 223) of the ASVs detected in the sprout microbiome were also 239 detected in the microbiomes of at least one of the five seed tuber compartments (Fig. 7a). Thirty-240 one percent of these ASVs (70 of 223) were present in all compartments, but these 70 ASVs 241 represented on average 60% of the total abundance of the sprout microbiome. Concomitantly, the 242 46 sprout-unique ASVs only made up 1.2% of the total bacterial abundance on the sprout (Fig. 7a). 243 Thus, with 98.8% of the total bacterial abundance on the sprout, the seed tuber was the main source 244 of the sprout microbiome in this soil-free system. Nonetheless, the taxonomic composition of the 245 sprout microbiome was distinct from the compartments on the seed tuber (Fig. S9), indicating that 246 the sprout compartment favors proliferation of a distinct subset of microbes that originate from the 247 seed tubers. 248 We further analyzed whether specific compartments on the seed tuber contribute differentially to 249 the sprout microbiome. Of the 223 bacterial ASVs detected on sprouts, 148 ASVs (66%) were also 250 detected in adhering soil, 124 in heel end (56%), 128 in peel (57%), 109 in eye (49%) and 103 in flesh 251 compartments (46%; Fig. 7b). We subsequently identified the top 18 most-abundant bacterial ASVs 252 (ASVs with relative abundances over 1%) in the sprouts that made up 80% of the total bacterial 253 sprout community and were able to trace them back in at least 2 of the 5 tuber compartments, but 254 with significantly lower abundances comparing within the sprouts (Fig. 7c). For fungi, 8 ASVs out of 255 the 74 ASVs that were detected in sprout samples were not found in any of the tuber compartments, 256 and the 8 ASVs represented only 2% of the sprout fungal community (Fig. S10a). On the other hand, 257 46% (34 of 74) of the sprout ASVs were present in all compartments and represent on average up to 258 65% of the total abundance of the sprout fungal community (Fig. S10b). Furthermore, the top 16 259 most-abundant fungal ASVs (ASVs with relative abundances over 1%) in the sprout totaled 95% of 260 the fungal sprout community (Fig. S10c). The relative abundance of these 16 fungal ASVs in the 261 sprout did not differ significantly (ANOVA, Turkey, P > 0.05) between the distinct tuber 262 compartments (Fig. S10c). vegetatively propagated by transplantation of relatively large seed tubers that contain a complex 273 microbiome. Here we studied how the microbiome of seed potatoes is affected by the field of 274 Fig. 7 The sprout microbiome is derived from diverse seed tuber tissues. a UpSet plot shows shared and unique ASVs of each compartment of Variety A. Each row represents a sample type, and each column represents a set of ASVs, where filled-in black dots with an edge between the dots indicates that these ASVs are present in multiple sample types. The sets are ordered by the number of ASVs as indicated by the bar plot above each category. The total ASVs in each sample type is indicated by the rotated bar plot on the left. The inlay shows the abundance of ASVs (46) that are unique to sprouts and of sprout ASVs (70) that are shared with all tuber compartments. b Venn diagrams of ASVs shared between each tuber compartment and the sprout of Variety A. Color represents different compartment. c The distribution of the top 18 most-abundant sprout ASVs in all compartments of Variety A. Color represents the genus of the ASVs. The percentage under each figure shows the relative abundance of these top sprout ASVs in each compartment. Capital letters indicate significant difference (P < 0.05) in agglomerated abundance of the top sprout ASVs as determined by ANOVA with Tukey's post-hoc test. production and whether the seed tuber microbiome associated with production fields is transmitted 275 to the emerging potato plant in the next season. 276 First, we analyzed two important factors that likely determine potato tuber microbiome 277 composition. Soil has been reported to be the main source of microbes that colonize potato roots 278 and tubers 20, 28, 29 . In addition plant genotype is a factor that shapes plant-associated microbiomes 279 30 . Plant roots actively and dynamically secrete root exudates that can selectively promote or deter 280 specific microbes 31, 32 . Although up to 85% of the total dry matter produced by the potato plant can 281 accumulate in the tubers 33 , it is unclear whether the tuber actively exudes metabolites to interact 282 with the microbiome. In this light, it has been reported that the tuber surface is low in nutrients and 283 that the limited nutrients that are available to the microbiome are a result of cell decay or lesions 284 only 34 . Tubers might therefore control soil microbiota to a much smaller extent compared to roots. 285 In line with this, Buchholz, Antonielli 20 and Nahar, Floc'h 35 reported that the microbiome found on 286 potato tubers is largely independent from the potato genotype. Also, Weinert et al. show that tuber-287 associated bacteria were not strongly affected by the plant genotype although a few cultivar-288 dependent taxa were identified 36, 37 . 289 290 In our study, however, when growing different genotypes in the same field we observed that not 291 only root, but also the tuber-associated bacterial and fungal communities were significantly affected 292 by the potato genotype (Fig. 3). Moreover, we found that the influence of potato genotype is larger 293 on the tuber microbiome than on the potato root microbiome (Fig. 3, Table S1). This suggests that 294 potato plants do exert control on the tuber microbiota, just like they selectively shape their root 295 microbiomes. Nonetheless, up to half of the bacterial ASVs found on seed tubers harvested from one 296 field were not found on seed tubers from the same variety that originated from other production 297 fields (Fig. 1, Fig. S5). Field of production determined more than half of the bacterial variation of the 298 seed tubers (Fig. 1, Table S1, S2). These results indicated that field of production is dominating over 299 genotype and is playing an even more vital role in tuber-associated microbiome assembly than 300 potato genotype, confirming previous findings 20, 29, 35 . 301 302 Interestingly, we observed that both roots and daughter tubers in our trial field harbored 303 microbiomes that were distinguishable by the production field of their seed tuber. This implies that 304 there is intergenerational or vertical transmission of microbes from the seed tuber to the emerging 305 plant and subsequently to the newly emerging tubers, the latter most likely via the stolon. In this 306 light, Vannier et al. 38 reported that both bacteria and fungi of the clonal plant Glechoma hederacea 307 can be transmitted to daughter plants through the stolon. In potato, some bacteria may migrate via 308 the xylem or intracellular spaces to the above ground tissues of the potato plants as well as the 309 stolon 39 and subsequently into the emerging tubers 20 . These studies suggest that vertical 310 transmission of microbes from one potato generation to the next is possible. In our study, we 311 observe around a quarter of bacterial and up to half of fungal communities in the daughter tubers 312 and roots overlapped with the seed tuber microbiomes (Fig. 4, Fig. S6). However, when we looked at 313 ASVs that were uniquely found on roots and daughter tubers that originate from seed tubers from a 314 specific production field, we see that a very small part of these ASVs (< 0.5%) is also detected 315 uniquely on the seed tubers from that production field (Fig. 4, Fig. S6). We conclude that, based on 316 the tractable vertical transmission of field-unique microbes, intergenerational transmission of 317 microbiota is minimal and cannot explain the effects of field of production on microbiomes in the 318 subsequent crop. 319 To better understand the early events in transmission of specific microbiome members from the 320 seed tuber to plants emerging from these tubers, we analyzed the microbial composition of sprouts geminated in a soil-free system and compared it to the microbial communities of different 322 compartments of the seed tubers. Firstly, we observed that the tuber's adhering soil, peel, heel end, 323 eye and flesh constitute distinct compartments that have significantly different microbiomes (Fig. 5,  324 Fig . S7). Apparently the physical and chemical characteristics and activities in these distinct 325 microhabitats 40, 41 select for different microbes. Moreover, the bacterial richness decreased from 326 the surface of the tuber inwards (Fig. S8). Arguably this is a result of physical exclusion of microbes 327 by the barrier function of the distinct tuber tissues and increased selective pressure inside the tuber 328 by a combination of e.g., plant immunity and oxygen limitation 42 . 329 In order to focus on the transmission from seed tuber to its sprouts without the interference of the 330 soil, we subsequently analyzed the microbiomes of sprouts emerging from the seed tubers in a soil-331 free system. Our results showed that the early stage of microbial community assembly in the sprouts 332 are genotype related. Moreover, sprouts emerging from tubers of the same genotype but originating 333 from different production fields still show to some extent distinct microbial patterns (Fig. 6). These 334 results indicate that the influence of tuber genotype and the field of seed tuber production can 335 largely determine the early-stage microbial assembly on the potato sprouts. Moreover, the top 18 336 most abundant bacterial ASVs, comprising almost 80% of the total bacterial communities on the 337 sprouts, could be traced back to the seed tuber compartments that we analyzed (Fig. 7, Fig. S10). 338 However, these sprout-abundant ASVs microbiome comprised a significantly smaller part of the total 339 bacterial microbiome in the different seed tuber compartments. This suggests that the most 340 abundant ASVs on the sprouts originate from diverse compartments of the seed tuber, and their 341 proliferation was specifically stimulated by the sprout. 342 Together our data show that microbiome composition is intergenerationally affected by the field of 343 production of the seed tuber. The potato tuber and root microbiomes on the daughter plants were 344 comprised mostly of microbes derived from the soil environment in which the next-season potato 345 plants were cultivated. The composition of a potato tuber microbiome is typically influenced by a 346 combination of factors: the resident soil microbiome, potato genotype, and the specific physical, 347 chemical, and (micro)biological conditions under which the tubers develop. In this study we 348 demonstrate that the potato tuber microbiome is also affected by the field in which the seed tuber 349 was produced. However, although we show that vertical transmission of microbes can occur from 350 seed tuber to the emerging sprouts in a soil free system, most microbes that occur on the roots and 351 daughter tubers of field-grown potato cannot be traced back to the population of seed tubers from 352 which they emerged. We speculate that the abiotic and biotic environmental conditions in the fields 353 of production differentially imprinted the seed tubers, leading to so far unknown epigenetic and/or 354 metabolic changes in the seed tubers that in turn differentially altered interactions of the emerging 355 plant with the soil microbiome, resulting in distinguishable microbiome signatures on daughter 356 tubers and roots, depending on the field of production of the mother seed tuber. 357 In conclusion, we show that seed tuber imprinting by the field of production shapes the microbiome 358 of the emerging potato plant. As it is accepted that plant microbiomes contribute to plant nutrition 359 and health, the initial microbiome is a much-undervalued trait of seed tubers specifically, or planting 360 materials in general. Elucidating the relative importance of the initial microbiome and the 361 mechanisms by which the origin of planting materials affect microbiome assembly will pave the way 362 for the development of microbiome-based predictive models that may predict the quality of seed 363 tuber lots, ultimately facilitating microbiome-improved potato cultivation. 364

366
Potato varieties 367 In total, 5 potato varieties form the Royal HZPC Group and Averis Seeds B.V. were used in this study, 368 namely variety Colomba (Variety A), Innovator (Variety B), Festien (Variety C), Challenger (Variety D), 369 Sagitta (Variety E) and Seresta (Variety F). 370 Sampling of seed tubers and post-storage seed tubers 371 In the autumn of 2018, seed tubers of two potato varieties (labelled A and B in this study to protect 372 the commercial interests of the potato breeding companies that produced them) were harvested 373 from 3 fields of production for Variety A and 3 other fields for Variety B (Fig. S1a-b). These tubers 374 were shipped to a central location where they were subsequently stored in the dark at 4 °C. Seed 375 tubers were taken from cold storage and sampled in December 2018 as "seed tuber" and July 2nd, 376 2019, as "post-storage seed tuber". For seed tuber samples, peels were sampled from 24 seed 377 tubers per production field and the peels of 6 tubers were pooled into a composite replicate sample, 378 resulting in 4 replicated samples per variety per field. For post-storage seed tuber samples, peels 379 were sampled from 36 seed tubers per field and the peels of 6 tubers were pooled into a composite 380 replicate sample, resulting in 6 replicated samples per variety per field. In total, 144 seed tubers and 381 216 post-storage seed tubers were sampled and resulted in 24 seed tuber samples and 36 post-382 storage seed tuber samples. These samples were frozen in liquid N2, freeze-dried and stored in 50-383 mL falcon tubes at -20 °C prior to analysis. 384 385 Seed tubers of Variety A and B of the above-mentioned 6 production fields were subsequently 386 planted in a single trial field near Veenklooster ( Fig. S1a; GPS location: 53.30353, 6.02670), the 387

Sampling of daughter tubers and roots emerging from seed tubers
Netherlands. The chemical composition of this sandy field was analyzed by Normec Groen Agro 388 Control B.V. and found to contain 1630 mg N/kg, 34 mg P2O5/l, 108 mg K/kg, 216 mg MgO/kg, 9 mg 389 Na/kg, 3.4% organic matter and a sulfur supply capacity 7.2kg S/ha per year. The field pH was 5.1 390 and the cation exchange capacity was 57 mmol/kg. On April 16 th , 2019, 24 seed tubers were planted 391 in each of the 4 replicate plots which were randomly distributed across the field. On July 2 nd , 2019, 4 392 potato plants were collected from the centre of each plot, from which the root material of each 393 plant was sampled as a root sample, resulting in 4 root samples per plot. In detail, for each plant, the 394 loosely attached soil was shaken off the roots, then the roots were cut into 5 cm fragments by sterile 395 scissors and a random subset of the root fragments were stored in a 50-mL falcon tube. In the 396 meantime, the peel of 6 newly formed tubers of each plant were samples and pooled as a composite 397 daughter tuber sample, resulting in 4 daughter tuber samples per plot. For both tuber and root 398 samples, the soil tightly attached to the peel and root was retained. In total, 96 potato plants and 399 576 daughter tubers were sampled resulting in 96 root and 96 daughter tuber samples. These 400 samples were freeze-dried and stored in 50-mL falcon tubes at -20 °C prior to analysis. 401 402 To dissect the contribution of microbiomes of different seed tuber compartments, namely peel, 403 eyes, heel ends, flesh, and adhering soil (Fig. S1c), in shaping the sprout microbiome, we made use 404 of material from a parallel study in which we harvested tubers from 6 potato varieties produced in 405 25 distinct production fields (Variety A from 5, Variety B from 5, Variety C from 3, Variety D from 5, 406

Sampling of seed tuber compartments
Variety E from 5, and Variety F from 2 fields, respectively; Fig. S1). In detail, the adhering soil was 407 gently rubbed from the tuber surface and collected in 50-mL falcon tubes. Subsequently, 1 cm thick 408 cores were sampled from potato heel ends and eyes using a sterilized Ø 0.6-cm metal corer. Then, 409 peel was sampled from around the minor axes of a tuber using a sterilized peeler. Flesh was sampled 410 by halving a tuber using a sterile scalpel and sampling 1-cm core using a sterile Ø 0.6-cm metal corer 411 from the centre of the tuber. Samples from 50 seed tubers were pooled into a single sample per 412 compartment per field. In total, 1250 tubers were sampled to access the microbial composition of 413 different tuber compartments, resulting in 125 compartment samples. These samples were freeze-414 dried and stored in 50-mL falcon tubes at -20 °C prior to analysis. 415 416 To study early events in transmission of specific microbiome members from seed tubers to plants 417 emerging from these tubers, the sprout microbiome was characterized. Seed tubers of all 6 varieties 418

Sampling of sprouts
(Variety A-F) from 12 of the above mentioned 25 fields were employed to study the sprout 419 microbiome (Fig. S1a-b). Five replicate tubers collected from each production field were germinated 420 on sterile Petri dishes in dark conditions (20 ℃ and RH 68%). These 60 seed tubers were randomized 421 in 6 trays and the position of the trays were rotated every day. After 7 days, 3−4 sprouts were 422 removed from each tuber using a sterile scalpel and pooled as a composite sample. These 60 sprout 423 samples were freeze-dried and stored in 2-mL Eppendorf tubes at -20 °C prior to analysis. 424 Sample grinding 425 To grind the samples in high-throughput, four 5-mm sterile metal beads were added to freeze-dried 426 samples in 50-mL falcon tubes and placed in a custom-made box. The samples were ground for 9 427 min on maximum intensity in a SK550 1.1 heavy-duty paint shaker (Fast & Fluid, Sassenheim, the 428 Netherlands). Freeze-dried sprout samples were ground in 2-mL Eppendorf tubes with one 5-mm 429 sterile metal bead per tube with a Tissuelyzer at 30 Hz for 1 min. 430 DNA isolation, library preparation and sequencing 431 Genomic DNA was isolated from ±75 mg potato powder per sample using a Qiagen Powersoil KF kit. 432 The KingFisher™ Flex Purification System machine was used for high throughput DNA isolation. DNA 433 was quantified using a Qubit® Flex Fluorometer with the Qubit dsDNA BR Assay Kit (Invitrogen,  434 Waltham, MA, USA) and normalized to a concentration of 5 ng/µl. The resulting DNA samples were 435 then stored at -20 °C. 436 Bacterial 16S ribosomal RNA (rRNA) genes within the V3-V4 hypervariable regions were amplified 437 using 2.5 µL DNA template, 12.5 µL KAPA HiFi HotStart ReadyMix (Roche Sequencing Solutions, 438 Pleasanton, USA), 2 µM primers B341F ( 5'-439 TCGTCGGCAGCGTCAGATGTGTATAAGAGACAGCCTACGGGNGGCWGCAG-3') and B806R (5'-440 GTCTCGTGGGCTCGGAGATGTGTATAAGAGACAGGACTACHVGGGTATCTAATCC-3') 43 with Illumina 441 adapter sequences in combination with 2.5 µM blocking primers mPNA (5'-GGCAAGTGTTCTTCGGA-442 3') and pPNA (5'-GGCTCAACCCTGGACAG-3') in 25 µL reactions. Blocking primers were used to avoid 443 the amplification of mitochondrial (mPNA) or plastidial (pPNA) RNA from the plant host 44