Brassica Napus Phyllosphere Bacterial Composition Changes with Growth Stage

Plant Soil https://doi.org/10.1007/s11104-021-04965-2

REGULAR ARTICLE

Brassica napus phyllosphere bacterial composition changes with growth stage

Jennifer K. Bell & Bobbi Helgason & Steven D. Siciliano

Received: 18 December 2020 /Accepted: 14 April 2021 # The Author(s) 2021

Abstract Keywords Phyllosphere . Brassica napus . BBCH . Aims Phyllosphere bacteria play critical roles in plant Microbiome growth promotion, disease suppression and global nutrient cycling but remain understudied. Methods In this project, we examined the bacterial com- Introduction munity on the phyllosphere of eight diverse lines of Brassica napus for ten weeks in Saskatoon, Saskatche- The phyllosphere is one of the largest microbial habitats wan Canada. Results The bacterial community was shaped largely by globally yet remains understudied (Vorholt 2012). The plant growth stage with distinct communities present phyllosphere can encompass all the aerial parts of the before and after flowering. Bacterial diversity before plant (Vorholt 2012), but most commonly is referred to flowering had 111 core members with high functional as the surface and interior of the leaves (Redford et al. potential, with the peak of diversity being reached dur- 2010;Vacheretal.2016). Phyllosphere microorganisms ing flowering. After flowering, bacterial diversity provide benefits to the plant, such as increased nutrient dropped quickly and sharply to 16 members of the core cycling, disease suppression, and growth promotion community, suggesting that the plant did not support the among many others. Fürnkranz et al. (2008) found that same functional potential anymore. B. napus line had some species of tropical plants had higher rates of little effect on the larger community, but appeared to beneficial nitrogen fixation by diazotrophic have more of an effect on the rare bacteria. phyllosphere bacteria. Vogel et al. (2016) and Jarvis Conclusions Our work suggests that the dominant bacte- et al. (2015) both found that the pre-existing rial community is driven by plant growth stage, whereas phyllosphere bacterial communities combined with differences in plant line seemed to affect rare bacteria. The plant genetics, impacted the reaction to and the extent role of these rare bacteria in plant health remains of disease on plants that had been inoculated with unresolved. known pathogens. Phyllosphere microbial communities also play important roles in promoting plant growth (Batool et al. 2016), and the diversity and abundance of these communities can impact the levels of insect Responsible Editor: Tim S. George. herbivory (Humphrey et al. 2014). Finally, phyllosphere bacteria are likely an important driver in the global * : : J. K. Bell ( ) B. Helgason S. D. Siciliano methane cycle and other global nutrient cycling process- Department of Soil Science, College of Agriculture and Bioresources, University of Saskatchewan, 51 Campus Drive, es (Iguchi et al. 2015). Saskatoon, Saskatchewan S7N 5A8, Canada Despite the importance of phyllosphere microbial e-mail: [email protected] communities for overall plant and ecosystem Plant Soil functioning, they remain poorly described, particularly these communities has also been largely ignored. Many in agricultural systems, which account for 37% (1385 of the studies on the canola microbiome have both mil. ha) of the global land use in 2017 (FAOSTAT limited sampling dates and canola lines. In order to 2017). With the global population predicted to reach 9 successfully breed B. napus to have a more robust billion by 2050, agricultural intensification will increase microbiome we must first gain an understanding of (Glibert et al.2010). Producing more food on decreasing how the microbiome develops and changes throughout arable land and in more erratic climate conditions poses the growing season for the whole plant, not just in the a unique and difficult problem for crop breeders and rhizosphere. In this study, we examined the bacterial farmers. Breeding crops to have a more robust phyllosphere microbiome of eight different lines of microbiome may be a sustainable way to improve crop B. napus in Saskatchewan over a ten-week period to yield without additional inputs or an increase in culti- determine the composition of the phyllosphere and what vated land (Ryan et al. 2009). However, before crops factors are important for the structuring of bacterial can be successfully bred to optimize microbiome con- communities on leaf surfaces. We hypothesized that tributions, we must first understand the diversity and the phyllosphere bacterial community composition dynamics of the microbiome of crop plants. would vary with developmental stage and B. napus line. Canola (Brassica napus) is produced for its high- quality oil, and increasingly, to be used as high-quality animal feed and for the production of biofuels. Canada Methods is the leading producer of canola, with an estimated 3 million tons produced in 2018 (AFFC 2019). The im- Eight lines of B. napus (Mason et al. 2017;Clarkeetal. portance and large-scale production of canola makes it a 2016) were seeded on May 28 or 29, 2017 in the field at good target for microbiome studies. Canola also has the Agriculture and Agri-Food Canada (AAFC) re- large nitrogen requirements. Using microbiome manip- search farm outside of Saskatoon, Saskatchewan, Can- ulation to increase nitrogen use efficiency could help ada (52.1718° N, 106.5052° W) which lies in the dark make canola a more sustainable crop. Additionally, the brown soil zone (Chernozem). The plots were seeded use of genetically modified crops, including canola, is using a 6 row, double disc drill with 1284 B. napus widely banned in Europe. Breeding canola with im- seeds applied per plot (6.1 m long by 1.8 m wide). A proved microbiomes could help lessen the disease pres- granular blend of 20–8–0-20 fertilizer was applied on sure placed on canola or rapeseed crops that cannot be May 23 at the rate of 139 kg ha−1. These lines of modified genetically to ameliorate some of this pressure. Brassica napus are part of the AAFC canola breeding Work on the microbiome of canola and oilseed crops program and thus were not commercial canola varieties, has focused primarily on the roots and rhizosphere but rather different lines created by nested associating (Copeland et al. 2015; Cordero et al. 2020; mapping, referred to as NAM lines (Clarke et al. 2016). Gopalakrishnan et al. 2015; Glaeser et al. 2020), or on They ranged in seed origin and color, fiber content, cultured microbial isolates (Romero et al. 2019)or erucic acid content and seed glucosinolate levels (SI pathogens from Brassica species, and not on the Table 1) meaning that some of them were not technical- microbiome as a whole (Wassermann et al. 2017). In ly canola which requires a low erucic acid content, but some cases, the interaction of canola root- and all lines were varying varieties of B. napus. Differences rhizosphere-associated microbiomes with other crops in these traits will result in differences in oil quality and was considered (Hilton et al. 2018; Schlatter et al. are related to phenological development. Bazghaleh 2019). To date, only one study has examined the canola et al. (2020) have described the experimental design phyllosphere microbiome (Copeland et al. 2015)ob- extensively, but briefly, the site consisted of three repli- serving differences compared to common bean and cate blocks with each B. napus line seeded randomly soybean. They further observed that the canola within each block, but not repeated within a block. All phyllosphere microbiome was affected by rainfall and lines were planted on May 28, 2017. The site received became increasingly different from the rhizosphere 127.9 mm of precipitation throughout the growing sea- community as the growing season progressed. son with a mean air temperature of 16.4 °C. Several leaf Not only does the composition of the phyllosphere samples were collected from each of the eight lines in remain understudied but what impacts the assembly of each block every week for ten weeks beginning on Plant Soil

June 20, 2017 until August 22, 2017. Plants on the plastid peptide nucleic acid blocker (pPNA) and 2 ul edges of the block were avoided as these leaves were (5 μM stock) mitochondrial peptide nucleic acid blocker visibly dusty and thus would likely have a different (mPNA) (PNA Bio, California, USA) and 10.3 μlnu- microbiome than most leaves on the interior of the rows. clease free water, and 2 μl of the standardized template Leaves that were clearly diseased, had extensive insect DNA. PNAs were included to block the amplification of damage or were senescing as the plant matured were host DNA, plant mitochondria and chloroplasts, which also avoided, as these leaves did not represent the ma- are a common contaminant from plant tissues (Ray and jority of the leaves present and would again bias the Nordén 2000, Von Wintzingerode et al. 2000). The microbiome results. During flowering, uppermost PCR conditions were 95 °C for 5 min as an initial leaves were avoided as these leaves had a thick layer denaturization, followed by 95 °C for 30 s., 78 °C for of petals on them and differentiating the leaf 10 s., 54 °C for 45 s., 72 °C for 60 s. for 35 cycles, and a microbiome from the flower microbiome would have final elongation of 72 °C for 7 min. Negative controls been impossible. However, some petal contamination and PCR duplicates were included. PCR product was was impossible to avoid, but was bypassed when feasi- purified to eliminate primers and impurities using 1:1 ble. Finally, during the later stages of seed development ratio of Nucleomag NGS clean-up and size select (D- and ripening, leaves that were mostly senesced were mark Biosciences, Scarborough, Ontario). Randomly avoided as dead tissue would cause large changes in selected duplicates were included during DNA extrac- the microbiome. Leaf samples, consisting of only one or tion, amplification, and sequencing stages adding in 56 two leaves were placed into sterile whirl-pak bags duplicates, bringing the total sample size to 326. After (Nasco, Wisconsin, USA) and placed onto ice. Leaf purification, samples were indexed following the samples from the same NAM line, but from different Illumina protocol (Illumina, San Diego, California), pu- blocks were not combined and plants were not destruc- rified again to remove excess index primers, quantified tively sampled as only a single leaf or two were sam- and standardized to 4 nM, and pooled. Pooled libraries pled. One sample from each NAM line in each block were then sequenced using the Illumina MiSeq platform was collected (24 samples) as well as one randomly using V3 chemistry. selected duplicate sample within block (line was not A total of 326 samples were sequenced resulting in considered here), leading to the collection of 27 samples 10,839,325 reads with an average of 18,186 reads per collected weekly, or 270 samples over the entire grow- sample. Sequences were imported into QIIME2 v ing season. Samples were then returned to the lab and 2019.1 (Bolyen et al. 2018) and primers were removed stored at −80 °C until further processing. using cutadapt (Martin 2011). Reads were then proc- Frozen, brittle leaves were crumbled manually in the essed into amplicon sequence variants (ASVs) and chi- whirl-pak and a 0.05 g subsample was taken and ex- meras were removed using Deblur (Amir et al. 2017) tracted using Qiagen PowerPlant extraction kit (Hilden, resulting in 1968 ASVs. ASVs were then classified Germany) following manufacturer instructions. After using a 515/806 trained Greengenes database classifier extraction, DNA was tested for quantity and quality (Desantis et al. 2006). Host mitochondria and chloro- following the standard Qubit protocol (Thermo Fisher plasts were removed after classification. Host DNA Scientific, Waltham Massachusetts). Template DNA ranged from 6% to 100% of the read in each sample was standardized to 4 ng μl−1 prior to amplification. with an average of 32% across samples. The abundance Bacterial diversity was assessed by amplifying the V4 and taxonomy artifacts produced in QIIME2 were region of the bacterial 16S rRNA using the primer set exported to BIOM format (McDonald et al. 2012)for 515F with Illumina adapters (5’ TCGTCGGC processing in R v. 3.5.3 (R Core Team 2018). Where AGCGTCAGATGTGTATAAGAGACA available, species level classification was used although GGTGYCAGCMGCCGCGGTAA 3′) and the 806R (5′ the authors recognize that there are issues (Johnson et al. - GTCTCGTGGGCTCGGAGATGTGTATA 2019), this is standard practice at the time of writing. If a AGAGACAG GGA CTA CCG GGG TAT CT - 3′) species level classification was not available, the ASV (Walters et al. 2015). The PCR reaction mix consisted of was marked as Unclassified. This was also done for 7 μl Invitrogen Platinum SuperFi PCR master mix higher levels of taxonomy. (Thermo Fisher Scientific, Waltham, Massachusetts), To select the core microbiome (Shade and Handelsman 0.1 μl of each primer (10 μMstock),3μl(5μMstock) 2012; Lundberg et al. 2013; Vandenkoornhuyse et al. Plant Soil

2015; Shade and Stopnisek 2019), each ASV was con- index places more emphasis on species evenness (Kim verted from number of reads to presence or absence. No et al. 2017). Unlike the previous two indices, the ACE singletons or doubletons formed part of the core, with all index takes into consideration the abundance of the members being both prevalent (present every week during species present and for this study most closely reflected sampling) and abundant (> 0.05% of all reads). The ASVs the observed diversity. Shannon-Weaver index, ACE that were present across all lines and weeks were selected (Chao and Lee 1992), Simpson index and species rich- as the Core and represents 15 ASVs, which fall into four ness (Jost 2007) estimates were calculated and showed different classifications (Table 1). Upon this conversion, it similar trends, with peak diversity occurring across all was noted that there appeared to be a shift in community lines during week 4 of sampling and declining steadily composition after flowering. The alpha diversity indices with plant maturity (Table 2). However, both the for the weeks before and after flowering were similar Shannon-Weaver and the Simpson indices showed an despite there being a shift in community composition. To initially very high level of diversity, but this is likely an investigate if there was a shift in composition, that was not over estimation due to a high abundance of rare ASVs being reflected accurately in the diverisity indices, the present in the first few weeks. community was subset into (i) ASVs present in all weeks The BBCH-scale is a scale used to uniformly identify up to and including flowering (Flowering) and (ii) those and quantify the phenological stages of plant develop- present in all weeks after flowering (Pod), using the same ment, with scales developed for species specific devel- prescence and absence matrix that was used to select the opment (Lancashire et al. 1991). Despite differences in core that was present across all ten weeks. In order for an crop species as well as varietal differences, all plants ASV to be selected for either the Flowering or Pod com- have the same growth stages of leaf development, for- munities, the ASV had to be present consistently every mation of side shoots, stem elongation, harvestable veg- week either before or after flowering. There were 111 etative plant part development, inflorescence emer- ASVs present before and during flowering (Flowering) gence, flowering, seed development, ripening and se- and 16 present post flowering (Pod), a small porportion nescence. All B. napus lines were assigned BBCH of the original 1968 ASVs. This separation into Flowering weekly using the Canola Council of Canada BBCH and Pod allowed for the conservation of the shift in com- guide (Canola Council of Canada 2020a, b) and were munity composition that would not have been captured by averaged. This was done because despite identical plant- examining the entirety of the sampling period together, nor ing times, the eight B. napus lines did exhibit some by looking at diversity metrics alone. The Flowering com- differences in plant development. Sampling week 1 munity has distinct ASVs from that of the Pod community and 2 took place during the leaf development stage for as many of the ASVs present during and before the most B. napus lines sampled, with bolting happening flowering stage disappeared once the plants matured into during week 3 (Fig. 1). Peak flowering was reached for seed and pod development. However, for analysis both of most lines during sampling week 4 (Flowering) with these communites do inlcude the original 15 Core ASVs, seed development occurring in the following two weeks as this small subset of ASVs were present all ten weeks. (Pod). The last four weeks of sampling were character- This resulted in Flowering consisting of 126 ASVs and ized by ripening of the B. napus seed pods (Pod). Podhaving31ASVs.UnweightedUniFracdistancema- Permutational analysis of variance (PERMANOVA) trices were performed on the entire data set of 1968 ASVs was performed using the adonis function in the vegan (Total) and well as the combined Core, Flowering and Pod package in R (Oksanen et al. 2017). The weighted and communities (Expanded Core). unweighted UniFrac (Lozupone and Knight 2005)dis- Alpha diversity was calculated using the package tances were calculated using the phyloseq package vegan in R (Oksanen et al. 2017). All diversity measures (McMurdie and Holmes 2013), for the entire communi- calculate the variety of organisms in a community based ty (Total), the Expanded Core (Core, Flower and Pod on combinations of species richness and evenness (Kim communities) as well as the communities present before et al. 2017). The Shannon-Weaver index, Abundance- and during flowering (Flower) and the communities based Coverage Estimate (ACE) and the Simpson Index present after flowering (Pod). UniFrac distances were were chosen because they represent a mix of approaches used instead of Bray-Curtis or other similar methods, to to diversity. The Shannon-Weaver index places more maintain the phylogenetic signals as there was a shift in emphasis on species richness whereas the Simpson community composition from the community present Plant Soil

Table 1 The Core microbiome present across all B. napus lines formation ability are all adaptations giving these core bacteria throughout the ten-week sampling period. The relative abundance advantages to colonizing the leaf surface. Expression of these traits is expressed as a percentage of the total bacterial abundance, the are denoted with a plus sign if the bacteria has the capability and a core relative abundance, for all lines for the entire sampling period. negative sign if the bacteria do not express this capability. These While this table represents 15 ASVs in the Core, these ASVs bacterial capabilities are supported by the literature but were not (amplicon sequence variants) classified into only four different implicitly tested in this study species. Plant pathogenesis, pigment production, and biofilm

Order Genus Species Relative Relative Abundance Pathogenic Pigmented Biofilm Abundance Core Forming

Pseudomonadales Pseudomonas Viridiflava 12.9 31.9 + 1 – + 2 Enterobacteriaceae Pantoea Unclassified 10.6 12.5 ± 3 + 3 + 3 Oxalobacteraceae Massilia Unclassified 5 13.1 ––– Xanthomonadales Stenotrophomonas Retroflexus 2.9 5.6 ––+ 4

1 Hu et al. 1998, 2 Bartoli et al. 2014, 3 Walterson and Stavrinides 2015, 4 Ren et al. 2015 pre-flowering and post-flowering. Other distance matri- investigate whether NAM line influenced the commu- ces that do not contain phylogenetic information would nity when it was not outweighed by the importance of not accurately capture this shift. growth stage. The residuals of this model were extract- Distance-based redundancy analysis (dbRDA) ed, and an additional dbRDAs were performed (Legendre and Andersson 1999) was performed using constraining by NAM line. By doing this we accounted the capscale function in the vegan package in R for the effect of growth stage without the influence of (Oksanen et al. 2017). The weighted and unweighted NAM line, then after this variation was accounted for, UniFrac matrices for both the Total and the Expanded we could isolate the additional variation caused by Core communities were constrained by NAM line and NAM line alone without the large signal of growth stage BBCH growth stage, as well as the interaction, and these overshadowing the effect of NAM line. constraints were then analyzed using the anova function (base R). Given that block was found to not be significant for any of the communities in the PERMANOVA Results results it was not included in the dbRDAs. To isolate the effect of B. napus NAM line more The leaf bacterial community was dominated by a small thoroughly, a dbRDA was performed constraining each subset of the total ASVs present. The leaf microbial UniFrac matrix by growth stage only. This was done to community (Fig. 2)of1968ASVswasprimarily

Table 2 Average diversity metrics of all eight lines of B. napus the Abundance-based Coverage Estimate. Larger numbers repre- for each week. Observed diversity is simply the average number of sent higher diversity ASVs (amplicon sequence variant) per sample present and ACE is

Stage Week/BBCH Observed ACE Shannon Inverse Simpson

Leaf Development 1/12–14 52.7 36.6 3.4 34.2 Leaf Development 2/14–19 68.5 84.7 2.5 7.7 Bolting 3/34–55 76.1 91.4 2.5 7.5 Flowering 4/59–65 77.4 94.4 2.7 11.1 Seed Development 5/61–72 48.6 63.9 2.2 7.2 Seed Development 6/70–79 45.1 55.2 2.2 6.4 Ripening 7/77–79 41.3 49.8 2.2 6.6 Ripening 8/77–79 44.4 53.3 2.5 8.9 Ripening 9/80–81 39.2 46.3 2.1 5.1 Ripening 10/80–85 37.5 44.4 2.3 7.4 Plant Soil

Fig. 1 Abundance-based Coverage Estimate (ACE) average of bars representing the standard error of the estimate. The BBCH the bacterial community on the leaf surface against the average range represents the range of BBCH observed across the eight BBCH of all 8 lines of B. napus across the ten sampling weeks. different lines for that sampling week. Change in BBCH stage is Each point is the average ACE for that week for the community indicated by changes in background color with the name of the microbiome (1968 ASVs (amplicon sequence variants) with error stage indicated at the top of the colored bar

Proteobacteria (79%), followed by Bacteroidetes (7%), reaching a peak during the flowering period. The ini- Acidobacteria (6%) and Firmicutes (6%). The tially high diversity found during week one was likely Proteobacteria were primarily Gammaproteobacteria due to the plants being recently emerged from the soil (74%), followed by Betaproteobacteria (20%) and and thus still being colonized by primarily by soil bac- Alphaproteobacteria (4.7%). However, despite 1968 teria. After flowering, all the diversity indices declined ASVs being present, the majority of the community steadily to levels similar to, or lower than those observed was dominated by a few species. The Expanded Core during the leaf development stages. There was a sharp microbiome of 136 ASVs were primarily Pseudomonas decrease in diversity between weeks six and seven (Fig. virdiflava (Table 1) (12.9% of the Total, 31.9% Core), 2 and Table 2) which could have been caused by low followed by Pantoea sp. (10.6% Total, 12.5% Core), precipitation. Diversity increased again during weeks Massilia sp. (5% Total, 13.1% Core) and seven to nine, only to decrease again at week ten. The Stenotrophomonas retroflexus (2.9% Total, 5.6% Core) overall decline in diversity after flowering could have (Fig. 2). However, it should be noted that these identi- been caused by the gradual leaf senescence that occurs fications are tentative due to the difficulty of assigning with B. napus during seed production. taxonomy using only one region of the bacterial PERMANOVA results indicated that stage was the genome. primary factor shaping the community structure. The Bacterial diversity was highest during week four, or PERMANOVA of the Total weighted UniFrac dis- the flowering period for both the observed diversity (the tances (Table 3) showed that plant development stage number of ASVs present) and abundance-based cover- (BBCH) significantly affected the community composi- age estimator (ACE) (Table 2 and Fig. 1). The highest tion (p = 0.026, R2 = 0.029) and that NAM line was diversity for the Shannon index and inverse Simpson nearly significant (p = 0.06, R2 = 0.021) (Table 3). was observed during the first sampling week (Table 2), There were no significant terms for the PERMANOVA however there was a sharp decline in both of these of the Total Unweighted UniFrac Distances (Table 3). indices during the second sampling week with both For the Expanded Core, stage was significant in the Plant Soil

Fig. 2 Relative abundance of genera that composed the Core as separate. For this graphic, the ASVs present in the Expanded Core well as the communities present during the Flowering and Pod have been grouped into genus instead of showing all 136 individ- stages (Expanded Core). Many of the ASVs (amplicon sequence ual ASVs in the Expanded Core. Colors are grouped as genus and variants) in the Expanded Core were classified into the same genus the weekly range of BBCH observed is present on the x axis or even species, but due to the nature of ASVs, they remained

Weighted Unifrac distances (p =0.018,R2 =0.026).In (p =0.048;Table3 and Fig. 3b). There was a marginally the unweighted Flower community, there was only a significant NAM line by growth stage interaction for the marginally significant interaction between NAM line, unweighted Total community (p =0.058;Table 3)and stage and block (Table 3). Stage was significant for both for the weighted, Expanded Core community (p = the Weighted (p =0.001,R2 = 0.039) and Unweighted 0.030; Table 3), likely caused by slight differences in (p = 0.001, R2 = 0.052) UniFrac distances of the Pod plant development times between lines. In the dbRDA community. Additionally, NAM line was significant models of the Total community, BBCH stage and NAM (p =0.045,R2 = 0.048) for the weighted UniFrac dis- line explained 28% of the variance in the weighted tances in the Pod community (Table 3). Unifrac analysis (Fig. 3a) and 24% of the variance in Given the significant, but low, explanatory values the unweighted UniFrac analysis (Fig. 3b). The slight given by the PERMANOVAs, db-RDAs were done to decrease in explanatory power between the weighted further isolate the effects driving community composi- and unweighted analysis could have been caused by tion while maintaining the phylogenetic relationships the bacterial community being largely dominated by a provided by the UniFrac distance matrices. Similar to small number of ASVs, which would not have been the PERMANOVAs, the db-RDAs indicated that reflected in the unweighted distance matrix. When only growth stage was the dominate factor shaping the bac- the Expanded Core was analysed using dbRDA, this terial community. BBCH stage significantly (p <0.001) increased to 41% (SI Fig. 1a) and 32% (SI Fig. 1b)for affected Total bacterial community structure for both the weighted and unweighted community analyses weighted and unweighted UniFrac dbRDA analyses (Table 3). This increase in explanatory power is likely (Fig. 3). NAM line was significant only for the un- caused by the reduction in the number of ASVs included weighted UniFrac analysis of the Total community in the analysis (1968 ASVs in the Total to 136 ASVs in Plant Soil

Table 3 Permutational analysis of variance (PERMANOVA) distances and unweighted UniFrac distances were analyzed for Tables. B. napus (NAM) line, growth stage, block and all the the Total community, the Expanded Core community and the interactions were used as covariates. Both weighted UniFrac communities present before and after flowering

Data Variable Component D.F. Sum of Squares Mean Squares F. Model R2 p value

Total Weighted Unifrac Stage Fixed 4 0.500 0.125 2.333 0.029 0.026* NAM Fixed 7 0.353 0.044 0.824 0.021 0.0667 Block Random 2 0.150 0.075 1.403 0.009 0.227 Stage x NAM 29 1.706 0.059 1.098 0.100 0.295 Stage x Block 8 0.270 0.034 0.631 0.016 0.889 NAM x Block 15 0.934 0.062 1.162 0.055 0.25 NAM x Stage x Block 56 2.207 0.039 0.736 0.129 0.968 Unweighted Unifrac Stage Fixed 4 1.132 0.283 1.133 0.014 0.237 NAM Fixed 7 1.584 0.198 0.792 0.020 0.938 Block Random 2 0.462 0.231 0.924 0.006 0.524 Stage x NAM 29 6.725 0.232 0.928 0.083 0.797 Stage x Block 8 2.105 0.263 1.054 0.026 0.327 NAM x Block 15 3.504 0.234 0.935 0.043 0.685 NAM x Stage x Block 56 13.966 0.249 0.998 0.173 0.505 Expanded Core Weighted Stage Fixed 4 1.294 0.324 1.980 0.026 0.018* NAM Fixed 7 1.201 0.172 1.050 0.025 0.362 Block Random 2 0.367 0.183 1.121 0.008 0.335 Stage x NAM 29 4.162 0.149 0.910 0.085 0.738 Stage x Block 8 1.377 0.172 1.053 0.028 0.36 NAM x Block 15 2.544 0.182 1.112 0.052 0.274 NAM x Stage x Block 56 8.992 0.167 1.019 0.184 0.442 Unweighted Stage Fixed 4 1.172 0.293 1.298 0.018 0.125 NAM Fixed 7 1.467 0.210 0.929 0.022 0.595 Block Random 2 0.402 0.201 0.891 0.006 0.558 Stage x NAM 29 6.405 0.229 1.014 0.096 0.408 Stage x Block 8 1.400 0.175 0.776 0.021 0.877 NAM x Block 15 3.204 0.229 1.014 0.048 0.455 NAM x Stage x Block 56 12.902 0.239 1.059 0.193 0.229 Flower Weighted Stage Fixed 2 0.107 0.053 0.647 0.011 0.647 NAM Fixed 7 0.719 0.103 1.246 0.073 0.253 Block Random 2 0.101 0.053 0.610 0.010 0.649 Stage x NAM 14 0.931 0.665 0.806 0.094 0.764 Stage x Block 4 0.423 0.106 1.283 0.043 0.247 NAM x Block 14 0.824 0.059 0.714 0.083 0.873 NAM x Stage x Block 28 2.178 0.778 0.944 0.220 0.587 Unweighted Stage Fixed 2 0.558 0.279 0.996 0.015 0.446 NAM Fixed 7 1.944 0.278 0.992 0.053 0.487 Block Random 2 0.558 0.279 0.996 0.015 0.435 Stage x NAM 14 3.905 0.279 0.996 0.107 0.526 Plant Soil

Table 3 (continued)

Data Variable Component D.F. Sum of Squares Mean Squares F. Model R2 p value

Stage x Block 4 1.288 0.322 1.150 0.035 0.160 NAM x Block 14 3.678 0.263 0.939 0.101 0.726 NAM x Stage x Block 28 8.794 0.314 1.122 0.242 0.06 Pod Weighted Stage Fixed 1 0.211 0.211 7.654 0.039 0.001* NAM Fixed 7 0.188 0.027 0.973 0.034 0.477 Block Random 2 0.107 0.054 1.947 0.020 0.084 Stage x NAM 7 0.224 0.032 1.164 0.041 0.282 Stage x Block 2 0.055 0.275 0.999 0.010 0.376 NAM x Block 14 0.358 0.026 0.930 0.066 0.556 NAM x Stage x Block 12 0.345 0.029 1.044 0.063 0.388 Unweighted Stage Fixed 1 1.870 1.875 10.660 0.052 0.001* NAM Fixed 7 1.724 0.246 1.401 0.048 0.045* Block Random 2 0.366 0.183 1.041 0.101 0.369 Stage x NAM 7 1.477 0.211 1.199 0.041 0.165 Stage x Block 2 0.411 0.205 1.168 0.011 0.284 NAM x Block 14 2.370 0.169 0.963 0.066 0.572 NAM x Stage x Block 12 2.604 0.217 1.234 0.072 0.099

the Expanded Core). Despite there being a large number growth stage alone explained 28% of the variation ob- of rare ASVs, they were not abundant thus reducing the served for the weighted UniFrac matrix (Fig. 3a)and noise for the analysis. For the Flowering community, 19%fortheunweightedmatrix(Fig.3b). NAM line stage was significant for both the weighted (p =0.002) explained 3% for the weighted and unweighted Expand- and unweighted (p = 0.001) UniFrac matrices, but nei- ed Core residuals and was not significant. The leaf is an ther NAM line nor the interaction of NAM line and inhospitable environment which can change drastically stage were significant for either (Table 3). For the Pod with plant growth stage, thus the small differences in Development community, stage was significant for both NAM line were likely insignificant to shaping the bac- the weighted (p = 0.003) and the unweighted (p = 0.001) terial community when compared to the much larger UniFrac matrices, but again, neither NAM line nor the force of plant growth stage. interaction was significant (Table 3). For the Total community dbRDA, BBCH stage alone explained 19% of the variation observed for the weight- Discussion ed UniFrac matrix (Fig. 3a) and 12% of the variation for the unweighted UniFrac matrix (Fig. 3b). NAM line was B. napus growth stage is the dominant factor shaping the observed to have slight differences in development time bacterial community present on the leaves, throughout and visual appearance of the leaves (observations from the growing season with a distinct community present the field) however these differences had little effect on before and after flowering. The distinct community is the bacterial communities present on the leaves. After evidenced by the Flowering community with 111 ASVs isolating the residuals of each of the above models, present every week before flowering, with most of these NAM line explained only 2% of the variation observed disappearing after flowering, leaving only 16 ASVs and was not significant for the weighted UniFrac dis- present in the Pod community. Additionally, significant tance residuals. Similarly, for the unweighted UniFrac differences in both PERMANOVAs and dbRDAs were distance residuals, NAM line explained 3% (p =0.1)of found when comparing the weighted UniFrac to the the variation. For the Expanded Core community, unweighted UniFrac distances, with BBCH stage being Plant Soil

Fig. 3 Distance-based redundancy analysis (dbRDA) of the weighted (28% of the variation) UniFrac distances (a) and unweighted (24% of the variation) UniFrac distances (b) of the Total bacterial community, across the entire 10-week sampling period, constrained by BBCH (p = 0.001) stage x B. napus line (NAM) (not significant) (biplots). Points are colored based on B. napus growth stage and correspond with the colors in Fig. 1

significant for both the total community and the Ex- bacterial community despite differences in NAM line as panded Core. Differences between weighted UniFrac, the Expanded Core remained abundant and constant and which uses phylogenetic relationships as well as abun- only the composition of the rare ASVs seemed to vary. dance, and unweighted UniFrac, which does not use Furthermore, NAM line significantly impacted the com- abundance, indicate that both community structure, munity structure only a limited number of times and evenness change throughout the sampling period (Table 3). It is unclear if this is a B. napus specific (Lozupone et al. 2007). If both community structure and finding or perhaps a phyllosphere specific finding. Crop evenness were not significantly impacted by BBCH variety has been shown to impact rhizosphere commu- growth stage, there would be no significant differences nities (Edwards et al. 2015, Coleman-Derr et al. 2016, between the weighted and unweighted UniFrac Dombrowski et al. 2017). Additionally, Bokulich et al. distances. (2014) found that microbial communities in grape must The dominance of growth stage appears to outweigh varied with grape cultivar. However, crop line has var- the importance of different B. napus lines on shaping the iable impacts on phyllosphere communities. Singh et al. Plant Soil

(2019) found that grape cultivar impacted the microbial (Smalla et al. 2001,Wagneretal.2014, 2016, Copeland community present in the phyllosphere. Conversely, et al. 2015,Hiltonetal.2018) so it is not surprising that Johnston-Monje et al. (2016) found that maize genotype growth stage changes were found in B. napus leaves. had no effect on microbial communities in the Additionally, it has been well documented that plants phyllosphere. Most importantly, Copeland et al. (2015) undergo large physiological and biochemical changes did not find that B. napus line impacted the microbial during flowering and fruit production (Nitsch 1965, communities in the phyllosphere. The lack of impact of Mohan Ram and Rao 1984, Shu et al. 2010). B. napus line on bacterial community structure could be due to leaves have a thick, waxy cuticle which thickens with thick, waxy cuticle present on B. napus leaves. The thick age. This thickening of the cuticle would make access to cuticle likely makes the B. napus leaf an inhospitable plant derived carbon more limited as the plant ages, environment for bacteria with low availability of plant which could be the cause of the shifts observed during derived carbon (Vorholt 2012; Vacher et al. 2016). the growing season. Furthermore, during flowering, the More work is needed to characterize the phyllosphere petals are quickly shed, many of which land on the leaf. and impact of line on B. napus as well as other crops. This additional and large source of labile carbon could Copeland et al. (2015) found that the canola be one of the factors accounting for the large increase in phyllosphere was dominated by a few high abundant bacterial diversity and abundance seen around the time genera with many rare ASVs, with decreasing diversity of flowering. After flowering, as the seeds begin to as the growing season progressed. However, they exam- develop and ripen, many of the different lines started ined only sampling date, but not plant growth stage, so to show signs of leaf senescence and death, in some while the trends are similar, it is difficult to extrapolate cases shedding their leaves entirely. This slow leaf further. The dbRDAs, as well as the PERMANOVA senescence coupled with the thickened waxy cuticle results suggest that perhaps B. napus line plays a larger likely drives the rapid decrease in bacterial diversity role in structuring the rare ASVs present on the leaves, but seen after flowering. While both the PERMANOVA not the Expanded Core microbiome which was comprised results and the db-RDAs have relatively small R2,their of relatively abundant bacteria. This is especially true for cumulative effect explains quite a bit of the variation. It the Pod community which showed a significant effect of is well established that environmental variables or un- NAM line on the unweighted UniFrac distances, but not in measured variables explain most of the variation seen in the weighted UniFrac distances. Since the unweighted microbial communities and it is not uncommon to have UniFrac distances do not incorporate abundance, simply low R2 (Redford et al. 2010,Wagneretal.2014, presence or absence of an ASV, this indicates that there Coleman-Derr et al. 2016,Leffetal.2017,), therefore might be a larger influence of line on the rarer ASVs, these results are valid despite the relatively low explan- which are not given as much weight when using the atory power of each variable. abundance weighted UniFrac distance matrix. However, The most prevalent non-pathogenic ASVs in the there appears to be no consistent patterns in what rare phyllosphere appear to be well suited for life in this ASVs are present throughout the sampling season, unlike difficult environment. Stenotrophomonas retroflexus the Expanded Core community, that was generally highly forms biofilms, which could aid in the colonization of abundant, relative to other taxa. This lack of both consis- the leaf surface and protect the bacteria from desiccation tency and abundance suggests that perhaps these rare and UV stress (Ren et al. 2015, Lui et al. 2017). ASVs are transitory inhabitants of the leaves that may be Oxalobacteraceae massilia exists in the rhizosphere, selected for by more nuanced biochemical or morpholog- root surface and seed coat of cucumber plants, so it’s ical characteristics that are line specific, such as cuticle not unexpected to find it on the leaf surface as well thickness or unmeasured secondary metabolites. In other (Ofek et al. 2012). Additionally, this species varies with words, the composition of the rare ASVs is likely deter- plant development stage in cucumber plants so given mined by the stochastic effects of random replication, that we see large growth stage patterns in B. napus death and immigration, whereas the Expanded Core bac- leaves, we might expect to see a similar pattern here terial community is likely selected primarily by plant (Ofek et al. 2012). Enterobacteriaceae Pantoea is a growth stage, making it a more deterministic process. functionally diverse genera with plant growth promot- Changes in the plant bacterial microbiome with plant ing abilities, nitrogen fixing capabilities, can produce growth stage are well documented for rhizosphere soils antibiotics and might be an opportunistic pathogen Plant Soil

(Walterson and Stavrinides 2015; Coutinho and Venter et al. 2017). A similar phenomenon was observed by 2009). Many species of Pantoea are also pigmented, Copeland et al. (2015) where the Shannon diversity which also likely increases their ability to withstand UV index was very high initially and the community com- stress, making them well suited for life in the position was similar to the soil, but as the growing phyllosphere (Andrews and Harris 2000). Furthermore, season progressed, the diversity index decreased, and Pantoea spp. live in other extreme environments such as the leaves developed communities that were distinct oil sands (Mitter et al. 2017) and seeds (Feng et al. 2006) from the soil. indicating that they can live in difficult environments Increasing bacterial diversity between the initial sam- such as the phyllosphere. However, given the uncertain- pling week and the peak at flowering could have been ty of assigning taxonomy based on 16S rRNA sequenc- caused by a number of factors. It is well established that ing alone, a follow-up, culture dependent study to de- while phyllosphere communities are separate from rhi- termine the identity and functionality of these ASVs zosphere communities they do have considerable over- would be needed to confirm these potential functions lap in community composition (Wagner et al. 2014; in the B. napus phyllosphere. Copeland et al. 2015;Wagneretal.2016). This increase The most common ASV found was Pseudomonas in diversity leading up to flowering could stem from the virdiflava. This is a pathogenic species that causes foliar deposition of soil dust from nearby farming activities. blights, root rots, and stem necrosis (Marina et al. 2008; Vokou et al. (2012) found air samples and the Sarris et al. 2012). Furthermore, this species can cause phyllosphere of nine different species shared a similar disease in a wide variety of cultivated plants, including composition, suggesting that aerial deposition is a likely cantaloupe, celery and tomatoes. Typical pathology for source of bacterial diversity on plants. Aerial deposition infection caused by Pseudomonas virdiflava is the of soil dust would also explain the overlap in wilting and yellowing of leaves, which was seen in the community composition between the rhizosphere and B. napus leaves, but the plants did not have high disease phyllosphere seen in other studies as the rhizosphere levels, and this could have been normal leaf senescence. draws its community from the surrounding bulk soil Culture dependent methods would need to be employed which would also be the source of the dust. to verify if the observed wilting and yellowing leaves Additionally, this increase in diversity could be caused observed in the field was indeed caused by Pseudomo- by visiting pollinators or insect herbivory. Humphrey nas virdiflava as its presence alone does not indicate the et al. (2014) found that insect herbivory was positively cause of disease. correlated with bacterial diversity in the phyllosphere. The initially very high estimates of diversity during The most likely explanation for the increasing diversity the first sampling week (Table 1) was likely caused by is a combination of aerial soil deposition, insect drivers the plants being relatively small during the first sam- and the plants selecting for a beneficial community. pling week as they were still in the leaf development Pre-flowering, the bacterial community on the leaves stage, with two to six leaves. These plants had only had greater diversity members (Fig. 2, SI Table 2), recently emerged and may have been predominantly compared to the Pod community. These functionally colonized by bacteria originating from the soil. The rich bacteria included members of the large number of unique or low abundance ASVs found Bradyrhizobiaceae which are well documented for their during the first sampling week quickly disappeared in widespread symbiosis with plants and their ability to fix subsequent weeks as the leave matured (Fig. 2) and may nitrogen (Marcondes de Souza et al. 2014; have been selected against by the harsh conditions on Gopalakrishnan et al. 2015). There were other the leaf surface. Both the Shannon-Weaver index and nitrogen-associated bacteria present pre-flowering as the Simpson index include estimates of observed diver- well, such as Nitrosovibrio tenuis which oxidized am- sity as well evenness but not abundance into the esti- monium (Harms et al. 1976)andStreptomyces mirabilis mates, which likely lead to the over estimation of initial which produces the enzyme nitroreductase and has diversity as rare species are given equal weight (Kim shown antifungal and antibacterial activity (Bordoloi et al. 2017). We did not see the same initial trend in the et al. 2001;Yangetal.2012). Other functionally useful ACE metric likely because the ACE metric uses both bacteria present on the leaves pre-flowering that disap- evenness and abundance of rare ASVs to estimate the pear after flowering included Burkholderia bryophila likelihood of a particular ASV to be present or not (Kim which has been shown to have anti-fungal activity, Plant Soil carbon monoxide reduction capabilities and plant with rarer ASVs changing rapidly and seemingly with- growth promoting properties (Vandamme et al. 2007; out pattern. Initially, there was very high diversity, but Weber and King 2012) which would be useful to the this was likely caused by colonization by soil bacteria B. napus plant as many of its major pathogens are fungal that quickly selected against due to the difficult nature of (Canola Council of Canada 2020a, b ). life in the phyllosphere. During the development stages, Methylobacterium were also present on the leaves prior the bacterial community retained some of this diversity to flowering. This species can use both methane and which included a higher potential for functional diver- methanol as a carbon source, both of which are sity. The bacterial community reached peak diversity byproducts of plant metabolism (Hanson and Hanson during flowering and rapidly became much less diverse 1996;Nisbetetal.2009;Dorokhovetal.2018). Fur- during the pod development and ripening stage, thermore, Methylobacterium has the ability to form consisting primarily of ASVs from a small number of biofilms, making it well adapted for life in the genera. These swift changes in both diversity, abundant phyllosphere (Hanson and Hanson 1996). and rare ASV composition suggests that while the plant This bacterial diversity largely disappears after has some control over the community composition, flowering (Fig. 2,SITable2), with the majority of the likely based on growth stage, there is also a probable core species found post flowering simply being addi- large stochastic component to the community assembly tional AVSs of the core bacteria present throughout the processes. Community assembly process on the entire growing season and a large dip in diversity seen phyllosphere warrant further investigation as research between weeks six and seven. This decrease in diversity into harnessing the microbiome for plant health inten- (Fig. 2 and Table 2) could have been caused by weeks sifies. This work shows that the bacterial community five and six having relatively low precipitation and present will change drastically throughout the growing hotter temperatures (unpublished site data). The recov- season, and that multiple sampling points are necessary ery of some diversity in weeks eight and nine was likely to get an accurate look at the microbiome and how it caused by these weeks experiencing much cooler and interacts with the plant as it develops. wetter conditions, allowing the bacteria present the moisture needed to rebound. The phyllosphere is a harsh environment that can fluctuate rapidly in temperature, Supplementary Information The online version contains supplementary material available at https://doi.org/10.1007/s11104- moisture and ultraviolet radiation levels, all of which 021-04965-2. will have an impact on the bacterial communities living on the surface of these leaves (Kinkel 1997; Vorholt 2012;Vacheretal.2016). However, this diversity was Acknowledgements We would like to acknowledge Kira not maintained during week ten as most of the leaves on Blomquist for assistance in field collections and Alix Schebel the plants were undergoing senescence at this point. and Zayda Morales for laboratory assistance. We acknowledge that this work took place on Treaty 6 Territory, homeland of the Despite this one anomalous large dip in diversity, the Métis. We gratefully acknowledge Dr. Sally Vail and the team of overall trend of decreasing diversity stands with one field technical staff at Saskatoon who managed the field trial. exception. Firmicute Exiguobacterium was not found in the pre-flowering core community and was abundant Code availability https://github.com/jbell364/Phyllosphere in the Pod community. Exiguobacterium is not widely Funding This work is supported by a grant from the Plant studied creating large gaps in knowledge about this Phenotyping and Imaging Research Centre (P2IRC). P2IRC is a genus, however it has been shown to grow in difficult digital agriculture research center funded by the Canada First environments (Vishnivetskaya et al. 2009) which could Research Excellence Fund (CFREF) from the Natural Sciences be the plant preparing for transfer to the seed (Shade and Engineering Research Council (NSERC), managed by the Global Institute for Food Security (GIFS), and located at the et al. 2017) or the community adapting to the quickly University of Saskatchewan (U of S). senescing leaves. Plant growth stage, but not B. napus line, was the Data availability All raw sequence files can be found at the dominant factor in shaping the bacterial phyllosphere National Center for Biotechnology Information (NCBI) under across all eight lines of B. napus sampled. There was a Bioproject PRJNA 635907. Supplementary Information The online version contains supple- small, but abundant Core community present through- mentary material available at https://doi.org/10.1007/s11104-021- out the growing season, regardless of plant growth stage 04965-2. Plant Soil

Declarations Canola Council of Canada, (2020b)-05-25. https://www. canolacouncil.org/canola-encyclopedia/crop- Conflict of interest The authors declare no conflicts of interest development/growth-stages/ Chao A, Lee S-M (1992) Estimating the number of classes via SampleCoverage. J Am Stat Assoc 87(417):210–217 Open Access This article is licensed under a Creative Commons Clarke WE, Higgins EE, Plieske J, Wieseke R, Sidebottom C, Attribution 4.0 International License, which permits use, sharing, Khedikar Y et al (2016) A high - density SNP genotyping adaptation, distribution and reproduction in any medium or format, array for Brassica napus and its ancestral diploid species as long as you give appropriate credit to the original author(s) and based on optimised selection of single - locus markers in the source, provide a link to the Creative Commons licence, and the allotetraploid genome. Theor Appl Genet 129(10):1887– indicate if changes were made. The images or other third party 1899 material in this article are included in the article's Creative Com- Coleman-Derr D, Desgarennes D, Fonseca-Garcia C, Gross S, mons licence, unless indicated otherwise in a credit line to the Clingenpeel S, Woyke T, North G, Visel A, Partida- material. If material is not included in the article's Creative Com- Martinez LP, Tringe SG (2016) Plant compartment and bio- mons licence and your intended use is not permitted by statutory geography affect microbiome composition in cultivated and regulation or exceeds the permitted use, you will need to obtain native Agave species. New Phytol 209(2):798–811 permission directly from the copyright holder. To view a copy of Copeland JK, Yuan L, Layeghifard M, Wang PW, Guttman DS this licence, visit http://creativecommons.org/licenses/by/4.0/. (2015) Seasonal community succession of the phyllosphere microbiome. Mol Plant-Microbe Interact 28:274–285 Cordero J, de Freitas JR, Germida JJ (2020) Bacterial microbiome associated with the rhizosphere and root interior of crops in References Saskatchewan, Canada. Can J Microbiol 66:71–85 Coutinho TA, Venter SN (2009) Pantoea ananatis: an unconven- tional plant pathogen. Mol Plant Pathol 10(3):325–335 Amir A, Daniel M, Navas-Molina J, Kopylova E, Morton J, Xu Desantis TZ, Hugenholtz P, Larsen N, Rojas M, Brodie EL, Keller ZZ, Eric K, Thompson L, Hyde E, Gonzalez A, Knight R K, Huber T, Dalevi D, Hu P, Andersen GL (2006) (2017) Deblur rapidly resolves single-nucleotide community Greengenes, a Chimera-Checked 16S rRNA Gene Database – sequence patterns. Am Soc Microbiol 2:1 7 and Workbench Compatible with ARB. Appl Environ Andrews JH, Harris RF (2000) The ecology and biogeography of Microbiol 72(7):5069–5072 microorganisms on plant surfaces. Annu Rev Phytopathol Dombrowski N, Schlaeppi K, Agler MT, Hacquard S, Kemen E, 38:145–180 Garrido-Oter R, Wunder J, Coupland G, Schulze-Lefert P Bartoli C, Berge O, Monteil CL, Guilbaud C, Balestra GM, (2017) Root microbiota dynamics of perennial Arabis alpina Varvaro L, Jones C, Dangl JL, Baltrus DA, Sands DC, are dependent on soil residence time but independent of Morris CE (2014) The Pseudomonas viridiflava phylogroups flowering time. ISME J 11(1):43–55 in the P.syringae species complex are characterized by ge- Dorokhov YL, Sheshukova EV, Komarova TV (2018) Methanol netic variability and phenotypic plasticity of pathogenicity- in plant life. Front Plant Sci 871:1–6 – related traits. Environ Microbiol 16:2301 2315 Edwards J, Johnson C, Santos-Medellín C, Lurie E, Podishetty Batool F, Rehman Y, Hasnain S (2016) Phylloplane associated NK, Bhatnagar S, Eisen JA, Sundaresan V (2015) Structure, plant bacteria of commercially superior wheat varieties ex- variation, and assembly of the root-associated microbiomes hibit superior plant growth promoting abilities. Front Life Sci of rice. Proc Natl Acad Sci 112(8):E911–E920 9:313–322 FAOSTAT (2017) FAOSTAT, Food and Agriculture Bazghaleh N, Mamet SD, Bell JK, Moreira ZM, Taye ZM, Organization of the United Nations. http://www.fao. Williams S, Arcand M, Lamb EG, Shirtliffe S, Vail S, org/faostat Siciliano SD, Helgason B (2020) An intensive multilocation Feng Y, Shen D, Song W (2006) Rice endophyte Pantoea temporal dataset of fungal communities in the root and rhi- agglomerans YS19 promotes host plant growth and affects zosphere of Brassica napus. Data Brief 30:105467 allocations of host photosynthates. J Appl Microbiol 100: Bokulich NA, Thorngate JH, Richardson PM, Mills DA (2014) 938–945 Microbial biogeography of wine grapes is conditioned by Fürnkranz M, Wanek W, Richter A, Abell G, Rasche F, Sessitsch cultivar, vintage, and climate. Proc Natl Acad Sci U S A A (2008) Nitrogen fixation by phyllosphere bacteria associ- 111:E139–E148 ated with higher plants and their colonizing epiphytes of a Bolyen E et al. (2018) QIIME 2 : Reproducible , interactive , tropical lowland rainforest of Costa Rica. ISME J 2:561–570 scalable , and extensible microbiome data science. PeerJ. doi: Glaeser SP, Gabur I, Haghighi H, Bartz JO, Kämpfer P, Snowdon https://doi.org/10.7287/peerj.preprints.27295 R, Obermeier C (2020) Endophytic bacterial communities of Bordoloi GN, Kumari B, Guha A, Bordoloi M, Yadav RNS, Roy oilseed rape associate with genotype-specific resistance MK, Bora TC (2001) Isolation and structure elucidation of a against Verticillium longisporum. FEMS Microbiol Ecol new antifungal and antibacterial antibiotic produced by 96:1–15 Streptomyces sp. 201. Bioscience. Biotechnol Biochem 65: Gopalakrishnan S, Sathya A, Vijayabharathi R, Varshney RK, 1856–1858 Gowda CLL, Krishnamurthy L (2015) Plant growth promot- Canola Council of Canada. (2020a)-05-05. https://www. ing rhizobia: challenges and opportunities. 3 Biotech 5:355– canolacouncil.org/canola-encyclopedia/diseases/ 377 Plant Soil

Hanson RS, Hanson TE (1996) Methanotrophic bacteria. different insights into factors that structure microbial com- Microbiol Rev 60:439–471 munities. Appl Environ Microbiol 73(5):1576–1585 Harms H, Koops HP, Wehrmann H (1976) An ammonia-oxidizing Lundberg DS, Lebeis SL, Paredes SH, Yourstone S, Gehring J, bacterium. Nitrosovibrio tenuis nov gen nov sp Archives of Malfatti S, Tremblay J, Engelbrektson A, Kunin V, del Rio Microbiology 108:105–111 TG, Edgar RC, Eickhorst T, Ley RE, Hugenholtz P, Tringe Hilton S, Bennett AJ, Chandler D, Mills P, Bending GD (2018) SG, Dangl JL (2013) Defining the core Arabidopsis thaliana Preceding crop and seasonal effects influence fungal, bacte- root microbiome. Nature 488(7409):86–90 rial and nematode diversity in wheat and oilseed rape rhizo- Marcondes de Souza JA, Carareto Alves LM, de Mello Varani A, sphere and soil. Appl Soil Ecol 126:34–46 de Macedo Lemos EG (2014) The Family Bradyrhizobiaceae Hu FP, Young JM, Fletcher MJ (1998) Preliminary description of BT - The Prokaryotes: Alphaproteobacteria and biocidal (syringomycin) activity in fluorescent plant patho- Betaproteobacteria (E. Rosenberg, E. F. DeLong, S. Lory, genic Pseudomonas species. J Appl Microbiol 85:365–371 E. Stackebrandt, & F. Thompson (eds.); pp. 135–154). Humphrey PT, Nguyen TT, Villalobos MM, Whiteman NK Springer, Berlin Heidelberg (2014) Diversity and abundance of phyllosphere bacteria Marina M, Maiale SJ, Rossi FR, Romero MF, Rivas EI, Gárriz A, are linked to insect herbivory. Mol Ecol 23:1497–1515 Ruiz OA, Pieckenstain FL (2008) Apoplastic polyamine Iguchi H, Yurimoto H, Sakai Y (2015) Interactions of oxidation plays different roles in local responses of tobacco Methylotrophs with Plants and Other Heterotrophic Bacteria. to infection by the necrotrophic fungus Sclerotinia Microorganisms 3(2):137–151 sclerotiorum and the biotrophic bacterium Pseudomonas – Jarvis KG, White JR, Grim CJ, Ewing L, Ottesen AR, Beaubrun viridiflava. Plant Physiol 147(4):2164 2178 JJ-G, Pettengill JB, Brown E, Hanes DE (2015) Cilantro Martin M (2011) Cutadapt removes adapter sequences from high- – microbiome before and after nonselective pre-enrichment throughput sequencing reads. EMBnet.journal 17:10 12 for Salmonella using 16S rRNA and metagenomic sequenc- Mason AS, Higgins EE, Snowdon RJ, Batley J, Stein A, Werner ing. BMC Microbiol 15:160 C, Parkin IAP (2017) A user guide to the Brassica 60K Illumina Infinium ™ SNP genotyping array. Theor Appl Johnson JS, Spakowicz DJ, Hong BY, Petersen LM, Demkowicz – P, Chen L, Leopold SR, Hanson BM, Agresta HO, Gerstein Genet 130(4):621 633 M, Sodergren E, Weinstock GM (2019) Evaluation of 16S McDonald D, Clemente JC, Kuczynski J, Rideout JR, Stombaugh rRNA gene sequencing for species and strain-level J, Wendel D et al (2012) The biological observation matrix microbiome analysis. Nat Commun 10(1):1–11 (BIOM) format or: how I learned to stop worrying and love the ome-ome. GigaScience 464(1):1–6 Johnston-Monje D, Lundberg DS, Lazarovits G, Reis VM, McMurdie PJ, Holmes S (2013) Phyloseq: an R package for Raizada MN (2016) Bacterial populations in juvenile maize reproducible interactive analysis and graphics of microbiome rhizospheres originate from both seed and soil. Plant Soil census data. PLoS One 8:e61217 405(1–2):337–355 Mitter EK, de Freitas JR, Germida JJ (2017) Bacterial root Jost L (2007) Partitioning diversity into independent alpha and microbiome of plants growing in oil sands reclamation beta components. Ecology 88:2427–2439 covers. Front Microbiol 8:1–14 Kim BR, Shin J, Guevarra RB, Lee JH, Kim DW, Seol KH, Lee Mohan Ram HY, Rao IVR (1984) Physiology of flower bud JH, Kim HB, Isaacson RE (2017) Deciphering diversity growth and opening. Proc: Plant Sci 93(3):253–274 indices for a better understanding of microbial communities. – Nisbet RER, Fisher R, Nimmo RH, Bendall DS, Crill PM, J Microbiol Biotechnol 27:2089 2093 Gallego-Sala AV, Hornibrook ERC, López-Juez E, Lowry Kinkel LL (1997) Microbial population dynamics on leaves. Annu D, Nisbet PBR, Shuckburgh EF, Sriskantharajah S, Howe – Rev Phytopathol 35:327 347 CJ, Nisbet EG (2009) Emission of methane from plants. Proc Lancashire PD, Bleiholder H, Van Den Boom T, Langelüddeke P, RSocBBiolSci276:1347–1354 Stauss R, Weber E, Witzenberger A (1991) A uniform dec- Nitsch JP (1965) Physiology of flower and fruit development BT - imal code for growth stages of crops and weeds. Ann Appl Differenzierung und Entwicklung / Differentiation and – Biol 119:561 601 Development (A. Lang (ed.); pp. 1537–1647). Springer, Leff JW, Lynch RC, Kane NC, Fierer N (2017) Plant domestica- Berlin Heidelberg tion and the assembly of bacterial and fungal communities Ofek M, Hadar Y, Minz D (2012) Ecology of root colonizing associated with strains of the common sunflower, Helianthus Massilia (Oxalobacteraceae). PLoS One 7:e40117 annuus. New Phytol 214(1):412–423 Oksanen, Jari, Pierre Legendre, Bob O’Hara, M Henry H Stevens, Legendre P, Andersson MJ (1999) Distance-based redundancy Maintainer Jari Oksanen, and MASS Suggests. (2007). “The analysis: testing multispecies responses in multifactorial eco- Vegan Package.” Community Ecology Package 10: 631–37 logical experiments. Ecol Monogr 69:1–24 R Core Team (2018) R: A language and environment for statistical Liu W, Russel J, Røder HL, Madsen JS, Burmølle M, Sørensen SJ computing. R Foundation for Statistical Computing, Vienna, (2017) Low-abundant species facilitates specific spatial or- Austria URL https://www.R-project.org/ ganization that promotes multispecies biofilm formation. Ray A, Nordén B (2000) Peptide nucleic acid (PNA): its medical Environ Microbiol 19(7):2893–2905 and biotechnical applications and promise for the future. Lozupone C, Knight R (2005) UniFrac: a new phylogenetic meth- FASEB J 14:1041–1060 od for comparing microbial communities. Appl Environ Redford AJ, Bowers RM, Knight R, Linhart Y, Fierer N (2010) Microbiol 71:8228–8235 The ecology of the phyllosphere: geographic and phyloge- Lozupone CA, Hamady M, Kelley ST, Knight R (2007) netic variability in the distribution of bacteria on tree leaves. Quantitative and qualitative B diversity measures lead to Environ Microbiol 12(11):2885–2893 Plant Soil

Ren D, Madsen JS, Sørensen SJ, Burmølle M (2015) High prev- the plant holobiont. New Phytol 206(4):1196–1206. alence of biofilm synergy among bacterial soil isolates in https://doi.org/10.1111/nph.13312 cocultures indicates bacterial interspecific cooperation. Vishnivetskaya TA, Kathariou S, Tiedje JM (2009) The ISME J 9:81–89 Exiguobacterium genus: biodiversity and biogeography. Romero FM, Rossi FR, Gárriz A, Carrasco P, Ruíz OA (2019) A Extremophiles 13:541–555 bacterial endophyte from apoplast fluids protects canola Vogel C, Bodenhausen N, Gruissem W, Vorholt JA (2016) The plants from different phytopathogens via antibiosis and in- Arabidopsis leaf transcriptome reveals distinct but also over- duction of host resistance. Phytopathology 109(3):375–383 lapping responses to colonization by phyllosphere commen- Ryan PR, Dessaux Y, Thomashow LS, Weller DM (2009) sals and pathogen infection with impact on plant health. New Rhizosphere engineering and management for sustainable Phytol 212:192–207 agriculture. Plant Soil 321(1–2):363–383 Vokou D, Vareli K, Zarali E, Karamanoli K, Constantinidou HIA, Sarris PF, Trantas EA, Mpalantinaki E, Ververidis F, Goumas DE Monokrousos N, Halley JM, Sainis I (2012) Exploring bio- (2012) Pseudomonas viridiflava, a multi host plant pathogen diversity in the bacterial Community of the Mediterranean with significant genetic variation at the molecular level. PLoS Phyllosphere and its relationship with airborne Bacteria. One 7:e36090 Microb Ecol 64(3):714–724 Schlatter DC, Hansen JC, Schillinger WF, Sullivan TS, Paulitz TC Von Wintzingerode F, Landt O, Ehrlich A, Göbel UB (2000) (2019) Common and unique rhizosphere microbial commu- Peptide nucleic acid-mediated PCR clamping as a useful nities of wheat and canola in a semiarid Mediterranean envi- supplement in the determination of microbial diversity. ronment. Appl Soil Ecol 144:170–181 Appl Environ Microbiol 66:549–557 Shade A, Handelsman J (2012) Beyond the Venn diagram: the Vorholt J a (2012) Microbial life in the phyllosphere. Nat Rev hunt for a core microbiome. Environ Microbiol 14(1):4–12 Microbiol 10:828–840 Shade A, Stopnisek N (2019) Abundance-occupancy distributions Wagner MR, Lundberg DS, Coleman-Derr D, Tringe SG, Dangl to prioritize plant core microbiome membership. Curr Opin JL, Mitchell-Olds T (2014) Natural soil microbes alter Microbiol 49:50–58 flowering phenology and the intensity of selection on Shade A, Jacques MA, Barret M (2017) Ecological patterns of flowering time in a wild Arabidopsis relative. Ecol Lett 17: seed microbiome diversity, transmission, and assembly. Curr 717–726 Opin Microbiol 37:15–22 Wagner MR, Lundberg DS, del Rio TG, Tringe SG, Dangl JL, Shu Z, Shi Y, Qian H, Tao Y, Tang D (2010) Distinct respiration Mitchell-Olds T (2016) Host genotype and age shape the leaf and physiological changes during flower development and and root microbiomes of a wild perennial plant. Nat Commun senescence in two freesia cultivars. HortScience 45(7):1088– 7:1–15 1092 Walters W, Hyde ER, Berg-lyons D, Ackermann G, Humphrey G, Singh P, Santoni S, Weber A, This P, Péros JP (2019) Parada A, Gilbert JA, Jansson JK (2015) Improved bacterial Understanding the phyllosphere microbiome assemblage in 16S rRNA gene (V4 and V4-5) and fungal internal tran- grape species (Vitaceae) with amplicon sequence data struc- scribed spacer marker gene primers for microbial community tures. Sci Rep 9(1):1–11 surveys. mSystems 1:1–10 Smalla K, Wieland G, Buchner A, Zock A, Parzy J, Kaiser S, Walterson AM, Stavrinides J (2015) Pantoea: insights into a Roskot N, Heuer H, Berg G (2001) Bulk and rhizosphere soil highly versatile and diverse genus within the bacterial communities studied by denaturing gradient gel Enterobacteriaceae. FEMS Microbiol Rev 39:968–984 electrophoresis: plant-dependent enrichment and seasonal Wassermann B, Rybakova D, Müller C, Berg G (2017) shifts revealed. Appl Environ Microbiol 67:4742–4751 Harnessing the microbiomes of Brassica vegetables for Vacher C, Hampe A, Porté AJ, Sauer U, Compant S, Morris CE health issues. Sci Rep 7:1–12 (2016) The Phyllosphere: microbial jungle at the plant– Weber CF, King GM (2012) The phylogenetic distribution and climate Interface. Annu Rev Ecol Evol Syst 47:1–24 ecological role of carbon monoxide oxidation in the genus Vandamme P, Opelt K, Knöchel N, Berg C, Schönmann S, De Burkholderia. FEMS Microbiol Ecol 79(1):167–175 Brandt E, Eberl L, Falsen E, Berg G (2007) Burkholderia Yang J, Xie B, Bai J, Yang Q (2012) Purification and character- bryophila sp. nov. and Burkholderia megapolitana sp. nov., ization of a nitroreductase from the soil bacterium moss-associated species with antifungal and plant-growth- Streptomyces mirabilis. Process Biochem 47:720–724 promoting properties. Int J Syst Evol Microbiol 57:2228– 2235 Publisher’snoteSpringer Nature remains neutral with regard to Vandenkoornhuyse P, Quaiser A, Duhamel M, Le Van A, jurisdictional claims in published maps and institutional Dufresne A (2015) The importance of the microbiome of affiliations.