Limits...
Using a Control to Better Understand Phyllosphere Microbiota

View Article: PubMed Central - PubMed

ABSTRACT

An important data gap in our understanding of the phyllosphere surrounds the origin of the many microbes described as phyllosphere communities. Most sampling in phyllosphere research has focused on the collection of microbiota without the use of a control, so the opportunity to determine which taxa are actually driven by the biology and physiology of plants as opposed to introduced by environmental forces has yet to be fully realized. To address this data gap, we used plastic plants as inanimate controls adjacent to live tomato plants (phyllosphere) in the field with the hope of distinguishing between bacterial microbiota that may be endemic to plants as opposed to introduced by environmental forces. Using 16S rRNA gene amplicons to study bacterial membership at four time points, we found that the vast majority of all species-level operational taxonomic units were shared at all time-points. Very few taxa were unique to phyllosphere samples. A higher taxonomic diversity was consistently observed in the control samples. The high level of shared taxonomy suggests that environmental forces likely play a very important role in the introduction of microbes to plant surfaces. The observation that very few taxa were unique to the plants compared to the number that were unique to controls was surprising and further suggests that a subset of environmentally introduced taxa thrive on plants. This finding has important implications for improving our approach to the description of core phytobiomes as well as potentially helping us better understand how foodborne pathogens may become associated with plant surfaces.

No MeSH data available.


NDMS of 16S Communities from Store, Control and Phyllosphere Environments at 4 time-points.3a Bray-Curtis ordination colored by environment (control (plastic), phyllosphere, and store bacterial communities). A very clear separation of store microbiota (green) is evident, however no significant separation between communities associated with control (red) and phyllosphere (purple) is evident. 3b Bray Curtis ordination colored by time-point. Time-point 0 (T0) and Time-point 1 (T1) appear to separate from each other and also from time-points 2 (T2) and 3 (T3), however T2 and T3 are more similar. For both 1a and 1b, the ellipse defines the upper 95th percentile limit of the assumed distribution.
© Copyright Policy
Related In: Results  -  Collection

License
getmorefigures.php?uid=PMC5036865&req=5

pone.0163482.g003: NDMS of 16S Communities from Store, Control and Phyllosphere Environments at 4 time-points.3a Bray-Curtis ordination colored by environment (control (plastic), phyllosphere, and store bacterial communities). A very clear separation of store microbiota (green) is evident, however no significant separation between communities associated with control (red) and phyllosphere (purple) is evident. 3b Bray Curtis ordination colored by time-point. Time-point 0 (T0) and Time-point 1 (T1) appear to separate from each other and also from time-points 2 (T2) and 3 (T3), however T2 and T3 are more similar. For both 1a and 1b, the ellipse defines the upper 95th percentile limit of the assumed distribution.

Mentions: Little separation by treatment (control and phyllosphere) was evident using nonmetric Multi Dimensional Scaling (nMDS) ordinations to look at Bray Curtis dissimilarity of bacterial communities (Fig 3). Microbiota that was washed off plastic plants pre-surface sterilization however, was clearly different from microbiota associated with plastic and live plants from the field environment. The washing step was performed to ensure that microbiota from the store environment where plants were purchased was not erroneously described as part of the environmentally driven consortia. Fig 3A shows nMDS ordination of 16S libraries for all environments: store, control (plastic plants) and phyllo (phyllosphere, live tomato plants). Fig 3B shows an nMDS ordination of the same data separated by time point. To further test the global hypothesis of independence for covariate variables (store, phyllo, enviro, and time-points: T0, T1, T2, T3) and associated response variables, a conditional inference regression tree was modeled onto the data. Nodes were generated by performing a binary split on all covariates (where the hypothesis could not be rejected) using a p value set at 0.1 as a cutoff (Fig 4) [31]. For ordination component MDS2, the time at which the samples were collected determines their position in the ordination (Fig 4A). However, for component MDS1, whether or not the plant was real (phyllo) or plastic (control) does not play a significant role in its location on the ordination chart. Nonparametric significance tests comparing the MDS1 component for time point 0 (p < 0.001) and time-points 1, 2, and 3 (p < 0.001) support this conclusion. The distances between control and phyllosphere samples from the same time-point were significantly closer to each other than they were to other time-points of same treatment (P = 2e-16; Mann-Whitney), suggesting that temporal changes are more influential drivers of community composition than treatment alone (Fig 4B). Additionally, we performed an analysis of multivariate homogeneity of group dispersions (using the permutest in the Vegan package). Significant differences in dispersion were identified between T0 compared to T1 and T1 compared to T2 (P<0.007 for each comparison). We did not identify significant differences in dispersions among time-points within each environment (control and phyllosphere), most likely due to limited group sizes (S2 Fig).


Using a Control to Better Understand Phyllosphere Microbiota
NDMS of 16S Communities from Store, Control and Phyllosphere Environments at 4 time-points.3a Bray-Curtis ordination colored by environment (control (plastic), phyllosphere, and store bacterial communities). A very clear separation of store microbiota (green) is evident, however no significant separation between communities associated with control (red) and phyllosphere (purple) is evident. 3b Bray Curtis ordination colored by time-point. Time-point 0 (T0) and Time-point 1 (T1) appear to separate from each other and also from time-points 2 (T2) and 3 (T3), however T2 and T3 are more similar. For both 1a and 1b, the ellipse defines the upper 95th percentile limit of the assumed distribution.
© Copyright Policy
Related In: Results  -  Collection

License
Show All Figures
getmorefigures.php?uid=PMC5036865&req=5

pone.0163482.g003: NDMS of 16S Communities from Store, Control and Phyllosphere Environments at 4 time-points.3a Bray-Curtis ordination colored by environment (control (plastic), phyllosphere, and store bacterial communities). A very clear separation of store microbiota (green) is evident, however no significant separation between communities associated with control (red) and phyllosphere (purple) is evident. 3b Bray Curtis ordination colored by time-point. Time-point 0 (T0) and Time-point 1 (T1) appear to separate from each other and also from time-points 2 (T2) and 3 (T3), however T2 and T3 are more similar. For both 1a and 1b, the ellipse defines the upper 95th percentile limit of the assumed distribution.
Mentions: Little separation by treatment (control and phyllosphere) was evident using nonmetric Multi Dimensional Scaling (nMDS) ordinations to look at Bray Curtis dissimilarity of bacterial communities (Fig 3). Microbiota that was washed off plastic plants pre-surface sterilization however, was clearly different from microbiota associated with plastic and live plants from the field environment. The washing step was performed to ensure that microbiota from the store environment where plants were purchased was not erroneously described as part of the environmentally driven consortia. Fig 3A shows nMDS ordination of 16S libraries for all environments: store, control (plastic plants) and phyllo (phyllosphere, live tomato plants). Fig 3B shows an nMDS ordination of the same data separated by time point. To further test the global hypothesis of independence for covariate variables (store, phyllo, enviro, and time-points: T0, T1, T2, T3) and associated response variables, a conditional inference regression tree was modeled onto the data. Nodes were generated by performing a binary split on all covariates (where the hypothesis could not be rejected) using a p value set at 0.1 as a cutoff (Fig 4) [31]. For ordination component MDS2, the time at which the samples were collected determines their position in the ordination (Fig 4A). However, for component MDS1, whether or not the plant was real (phyllo) or plastic (control) does not play a significant role in its location on the ordination chart. Nonparametric significance tests comparing the MDS1 component for time point 0 (p < 0.001) and time-points 1, 2, and 3 (p < 0.001) support this conclusion. The distances between control and phyllosphere samples from the same time-point were significantly closer to each other than they were to other time-points of same treatment (P = 2e-16; Mann-Whitney), suggesting that temporal changes are more influential drivers of community composition than treatment alone (Fig 4B). Additionally, we performed an analysis of multivariate homogeneity of group dispersions (using the permutest in the Vegan package). Significant differences in dispersion were identified between T0 compared to T1 and T1 compared to T2 (P<0.007 for each comparison). We did not identify significant differences in dispersions among time-points within each environment (control and phyllosphere), most likely due to limited group sizes (S2 Fig).

View Article: PubMed Central - PubMed

ABSTRACT

An important data gap in our understanding of the phyllosphere surrounds the origin of the many microbes described as phyllosphere communities. Most sampling in phyllosphere research has focused on the collection of microbiota without the use of a control, so the opportunity to determine which taxa are actually driven by the biology and physiology of plants as opposed to introduced by environmental forces has yet to be fully realized. To address this data gap, we used plastic plants as inanimate controls adjacent to live tomato plants (phyllosphere) in the field with the hope of distinguishing between bacterial microbiota that may be endemic to plants as opposed to introduced by environmental forces. Using 16S rRNA gene amplicons to study bacterial membership at four time points, we found that the vast majority of all species-level operational taxonomic units were shared at all time-points. Very few taxa were unique to phyllosphere samples. A higher taxonomic diversity was consistently observed in the control samples. The high level of shared taxonomy suggests that environmental forces likely play a very important role in the introduction of microbes to plant surfaces. The observation that very few taxa were unique to the plants compared to the number that were unique to controls was surprising and further suggests that a subset of environmentally introduced taxa thrive on plants. This finding has important implications for improving our approach to the description of core phytobiomes as well as potentially helping us better understand how foodborne pathogens may become associated with plant surfaces.

No MeSH data available.