Whole-genome duplication and host genotype affect rhizosphere microbial communities in Arabidopsis thaliana

The composition of complex microbial communities found in association with plants is influenced in part by host genotype. Yet, the salient genetic architecture is often unknown. Genome duplication events are common in the evolutionary history of plants, influence many important diverse plant traits, and may affect associated microbial communities. Using genotypes with experimentally induced whole genome duplication (WGD), we tested the effect of WGD on rhizosphere bacterial communities in Arabidopsis thaliana. Specifically, we performed 16s rRNA amplicon sequencing to characterize differences between microbiomes associated with specific host genotypes (Columbia vs. Landsberg) and ploidy levels (diploid vs. tetraploid). We modeled abundances of individual bacterial taxa by utilizing a hierarchical Bayesian framework, based on the Dirichlet and multinomial distributions. We found that host genotype and host ploidy level affected rhizosphere community composition. Further, the microbiome of the tetraploid Columbia genotype differed from that of other host genotypes. We then tested to what extent microbiomes derived from a given host genotype or ploidy level affected plant performance by inoculating sterile seedlings of each genotype with microbial communities harvested from a prior generation, using a full factorial design. We found a negative effect of the tetraploid Columbia microbiome on growth of all four plant genotypes. The findings suggest that while both host genotype and ploidy affect microbial community assembly, bacterial communities found in association with only some host genotypes may affect growth of subsequent plant generations.

To test the effects of whole genome duplication and plant genotype on rhizosphere 1 0 1 bacterial community composition and plant host performance, we selected two A. thaliana 1 0 2 diploid genotypes: Columbia (Col-2x) and Landsberg erecta (Ler-2x) and their tetraploid generations removed from initial colchicine treatment, and therefore can be assumed to be 1 0 5 Page 6 of 30 Julian C. Bennett Ponsford mutationally stable (Hollister et al., 2012). In our first experiment, each genotype was planted 1 0 6 into sterilized potting mix and inoculated with a microbial community from the Catsburg region  In both experiments, seeds were surface sterilized using a solution of 15% bleach, 0.1% USA) and 2ml of liquid inoculate (described below). To ensure sterility, potting mix was 1 1 9 autoclaved on a wet cycle for 60 min at 121°C, allowed to rest for at least one hour and 1 2 0 autoclaved for another 60 min. No microbes could be cultured on tryptic soy agar media using 1 2 1 serial dilutions of autoclaved soil as inoculum. To create our inoculants for the first experiment, to the potting substrate, which minimized potential selection by the common soil matrix on  For the second experiment, we used plant-conditioned soil collected from each host 1 3 0 genotype to create inoculate. In all experiments, after ten days of growth, seedlings were thinned 1 3 1 to one plant per pot. All experiments were performed at the Williams Conservatory or Agricultural Experiment Station at the University of Wyoming. To measure the influence of host ploidy on rhizosphere bacterial community composition, 20 1 3 7 replicates of Col-2x, Col-4x, Ler-2x, and Ler-4x were planted in a fully randomized four block were allowed to grow until senescence to quantify additional phenotypes. To account for 1 4 2 variation in microbial communities due to greenhouse conditions and intrinsic variation in soil substrate that were potted simultaneously with our experiment (N = 7).  Rhizosphere samples were collected from the root surface as described in (Bulgarelli et particles, subsequently placing roots with closely adhering soil in a buffer of sterile PBS with 1 6 0 0.1% Silwet in a 15 mL tube, and vortexing for 10 min on maximum speed. Rhizosphere 1 6 1 samples were centrifuged at 3714 rcf, and a total soil mass of no more than 250 mg was 1 6 2 transferred from each tube into a sterile Qiagen PowerSoil (Qiagen, Valencia, CA) bead tube.  We used the R package dada2 (ver. 1.5.8) to filter and trim reads based on quality, 1 6 9 estimate the error rate using 1 000 000 reads, dereplicate reads, infer amplicon sequence variants 1 7 0 (ASV), merge paired end reads, remove chimeras, and assign unique sequences to taxa using the 1 7 1 Silva 16s database (ver. 128) (Quast et al., 2013;Callahan et al., 2016). Next, we ascertained the 1 7 2 abundance of operational taxa by quantifying unique sequences and amplicon sequence variants  in sequencing data or any finite number of taxa observations from an assemblage), readily allows 1 9 0 information-sharing among replicates within a category and obtains estimates of the relative 1 9 1 abundance of each microbial taxon, while propagating uncertainty in those estimates. We 1 9 2 estimated differences in the relative abundance of each microbial taxon between experimental Briefly, DMM estimates the multinomial parameters describing the relative abundances 1 9 5 of each taxon in a replicate, denoted as a vector of parameters (p). These multinomial parameters 1 9 6 were informed by a Dirichlet distribution, with parameters characterizing the expected 1 9 7 frequencies of each taxon (π* Θ ), where π is a vector describing the expected relative abundance 1 9 8 of each taxon in the sampling group and Θ is an intensity parameter that models among-replicate there was little evidence that the relative abundance of the focal taxon differed between treatment 2 0 5 groups. The DMM offers several advantages over existing analytical methodologies. First,  Material. For each of four chains, the sampler was run for 1500 steps as a burn-in period and was  For those taxa that differed credibly among our treatment groups, we used NCBI BLAST  To assess effects on plant performance in experiment 2, we used fixed effect, three-way 2 3 0 ANOVAs, where inoculum, genotype, and block were the explanatory variables. For purposes of 2 3 1 data representation, residuals were plotted after statistically accounting for the effect of genotype 2 3 2 and block. Finally, we used planned comparisons to contrast plant performance between plants 2 3 3 grown in the Col-2x versus Col-4x inoculums and Ler-2x versus Ler-4x inoculums. After quality filtering, sample inference and chimera removal in dada2 as well as  of Jaccard and Bray-Curtis dissimilarities (P < 0.01).

4 7
Ploidy level did not significantly affect whole rhizosphere community composition replicates. However, because we observed differential growth effects of the Col-4x vs. all other 2 5 2 rhizospheres, we performed a post hoc comparison of the rhizosphere microbiome of the Col-4x 2 5 3 genotype to that of other sampling groups including Col-2x. We found that Col-4x differed  diversity across all samples was 4.59 (+/-0.043, n = 29).

6 0
We utilized DMM to identify individual taxa with significantly different relative 2 6 1 abundance due to host genotype and host ploidy level. This model allowed us to quantify 2 6 2 differences in the abundance of individual taxa among treatments that might not be evident in 2 6 3 whole community dissimilarity analyses. Based on this approach we identified 25 taxa that were 2 6 4 significantly enriched in both Ler genotypes and 29 taxa that were more abundant in the Col 2 6 5 genotypes (Figure 2a). A taxon of genus Pedobacter was the most highly enriched in Ler 2 6 6 associated bacterial communities with an 8-fold increase. In Col bacterial communities, the most 2 6 7 enriched taxon was Pseudomonas at a 4.8-fold increase in abundance. DMM also identified 17 2 6 8 taxa that were significantly enriched in all diploid genotypes. Of these taxa, the most enriched 2 6 9 taxon was a member of the genus Bacillus, enriched 3.1-fold relative to tetraploid communities. Tetraploid communities' most enriched member was Mucilaginibacter, which was present at 4-2 7 1 fold greater abundance compared to diploid rhizosphere bacterial community abundance. 23 taxa 2 7 2 were significantly enriched across both tetraploid genotypes (Figure 2b). Our model identified enriched in all other communities. This bacterium appeared at 5.8-fold greater frequency in these 2 7 7 communities than Col-4x. Across all comparisons, members of the phylum Proteobacteria were 2 7 8 more commonly differentially abundant than any other phylum.  The soil microbiome had no effect on phenological characteristics or fruit number 2 8 6 (Supplemental Table 2). Whole genome duplication is estimated to have occurred across 30-70% of the 2 9 0 angiosperm phylogeny over its evolutionary history and is a common genetic feature of many 2 9 1 economically important plants (Alix et al., 2017). We found that host genotype and WGD 2 9 2 influenced numerous individual taxa within the rhizosphere microbiome. Despite this, 2 9 3 dissimilarity analyses identified no differences in overall community composition attributable to 2 9 4 WGD. Furthermore, host-genotype specific effects on rhizosphere microbiomes influenced plant 2 9 5 growth. In contrast to prior studies (Bai et al., 2015), our experiments also utilized a soil with a 2 9 6 history of Arabidopsis occurrence as inoculant. This should in principle allow genotypes to 2 9 7 assemble root microbiota from existing and potentially coevolved plant-proximal microbes. Previous studies examining the root microbiota found in association with common lab  al., 2009;Lundberg et al., 2012). These studies report a < 10% shift in composition between the 3 0 1 Ler and Col, which we corroborate here (Figure 1). As noted above, we did not find a 3 0 2 significant effect of ploidy on bacterial community composition based on dissimilarity analysis 3 0 3 but did identify individual taxa influenced by genome duplication (Figure 2a). This provides 3 0 4 support for the hypothesis that these specific taxa are directly associated with genome exudates, or other metabolites produced by the plant, in response to WGD could be partially 3 1 0 responsible for the association between ploidy and the relative abundance of microbial taxa that 3 1 1 we observed (Edger et al., 2015;Lebeis et al., 2015). We conclude, that while host ploidy level 3 1 2 affects the abundance of some taxa, it does not lead to broad-scale changes in the rhizosphere 3 1 3 microbiome (Figure 1).

1 4
We found that differences in the rhizosphere bacterial microbiome lead to differences in grown in soils inoculated with microbial communities from all other genotypes (Figure 3).

1 8
Although 16 ASVs were identified as significantly enriched in Col-4x root associated 3 1 9 communities, the nature of 16s rRNA amplicon data prevents us from obtaining the species level  Therefore, further study is required to elucidate the effects of these bacteria on Arabidopsis' ability to confer drought tolerance to Arabidopsis. Applied Soil Ecology 68, 1-9.   Supplemental Figure 1:a) Principal coordinate analysis of Jaccard dissimilarities (n=29).

3 8
Rhizosphere community composition does not differ significantly by host plant ploidy level (P =