Ecological succession and stochastic variation in the assembly of Arabidopsis thaliana phyllosphere communities

Abstract

Bacteria living on the aerial parts of plants (the phyllosphere) are globally abundant and ecologically significant communities and can have significant effects on their plant hosts. Despite their importance, little is known about the ecological processes that drive phyllosphere dynamics. Here, we describe the development of phyllosphere bacterial communities over time on the model plant Arabidopsis thaliana in a controlled greenhouse environment. We used a large number of replicate plants to identify repeatable dynamics in phyllosphere community assembly and reconstructed assembly history by measuring the composition of the airborne community immigrating to plant leaves. We used more than 260,000 sequences from the v5v6 hypervariable region of the 16S rRNA gene to characterize bacterial community structure on 32 plant and 21 air samples over 73 days. We observed strong, reproducible successional dynamics: phyllosphere communities initially mirrored airborne communities and subsequently converged to a distinct community composition. While the presence or absence of particular taxa in the phyllosphere was conserved across replicates, suggesting strong selection for community composition, the relative abundance of these taxa was highly variable and related to the spatial association of individual plants. Our results suggest that stochastic events in early colonization, coupled with dispersal limitation, generated alternate trajectories of bacterial community assembly within the context of deterministic selection for community membership.

Importance: Commensal bacteria associated with plants help protect their hosts against infection and promote growth. Bacteria associated with plant leaves (the "phyllosphere") are highly abundant and diverse communities, but we have very limited information about their ecology. Here, we describe the formation of phyllosphere communities on the plant model organism Arabidopsis thaliana. We grew a large number of plants in a greenhouse and measured bacterial diversity in the phyllosphere throughout the Arabidopsis life cycle. We also measured the diversity of airborne microbes landing on leaves. Our findings show that plants develop distinctive phyllosphere bacterial communities drawn from low-abundance air populations, suggesting the plant environment is favorable for particular organisms and not others. However, we also found that the relative abundances of bacteria in the phyllosphere are determined primarily by the physical proximity of individual plants. This suggests that a mixture of selective and random forces shapes phyllosphere communities.

FIG 1

Experimental design. Plants were placed in four growth trays inside two custom-built shade boxes side by side on a greenhouse bench. Arabidopsis Col-0 plants were placed in trays 1 and 2 (gray). Trays 3 and 4 (crosses) contained Arabidopsis gl-1 plants (data not shown). Trays were rotated at every watering and sampling event by each tray being moved one place to the right. Note that trays 1 and 2 remained in separate shade boxes until day 55 of the experiment. Glass slides coated with CellTak (small black rectangles) were placed in the center of each tray and sampled/replaced simultaneously with sampling of plants. Flats within each tray were rotated back to front and front to back as shown every time trays were rotated.

FIG 2

Mature plant-associated microbial communities have a distinct membership from air communities and show a clear trajectory over time. (A) Principal coordinate plots using the membership-based Jaccard index to measure beta diversity, based on OTUs clustered at 97% similarity. Each dot represents a single community (blue, air; green, plant). Dot size is scaled by sampling day (day 19, small; day 73, large). Arrow indicates the trajectory from early to mature phyllosphere communities. (B) Mean Jaccard distance between replicate plants on each sampling day shows increasing similarity with time. (C) Abundance trajectories over time of the day 60+ air biomarker OTU 1. Blue indicates relative abundance (percentage of total read count) in air, and green indicates relative abundance in plants. Error bars show the standard deviation across triplicates (plants) and duplicates (air). Abundance trajectories over time of day 60+ plant biomarker OTU 4, a day 55 biomarker (D), and OTU 10, a day 50 biomarker (E). Coloring is identical to that described for panel C.

FIG 3

Spatially associated plants share similar taxon abundances. (A) Principal coordinates plot using the abundance-based Morisita-Horn index to measure beta diversity, based on OTUs clustered at 97% similarity. Blue, air; green, tray 1 plants; orange, tray 2 plants. Scaling is the same as described for Fig. 1. Arrows indicate trajectories from newly colonized plants to tray-specific communities. (B) Abundance trajectory of the dominant tray 1 biomarker Pseudomonas; error bars indicate the standard deviation of replicate plants. Green bars indicate relative abundance of the OTU in tray 1, orange bars in tray 2, and blue bars in air. (C, D) Abundance profiles for major tray marker taxa identified by LEfSe at successive time points. (C) Abundance of the tray 1 biomarker Rhodococcus. (D) Abundance of the tray 2 biomarker Methylobacterium. (E) Abundance of the dominant tray 2 biomarker Acinetobacter. Note that three plants were sampled at every time point, chosen randomly; possible configurations for sampling between trays 1 and 2 were (2,1), (1,2), (3,0), or (0,3).

FIG 4

Sub-OTU diversity reveals ecological processes operating at fine taxonomic scale. Within the genus Acinetobacter, sequences belonging to the dominant OTU 0 (top panel, 20% of the data set) could be decomposed into 4 major oligotypes (bottom 4 panels). Each oligotype shows a distinctive distribution indicating fine taxonomic selection within this dominant taxon; oligotype 1 in tray 2 plants, oligotype 2 in air, oligotype 3 in tray 1 plants, and oligotype 4 in both tray 1 and tray 2 plants.