Research Article | Ecological and Evolutionary Science
Invasive Plants Rapidly Reshape Soil Properties in a Grassland Ecosystem
Sean M.Gibbons, YlvaLekberg, Daniel L.Mummey, NaseerSangwan, Philip W.Ramsey, Jack A.Gilbert
Sean M. Gibbons
Graduate Program in Biophysical Sciences, University of Chicago, Chicago, Illinois, USABioscience Division, The Microbiome Center, Argonne National Laboratory, Argonne, Illinois, USAMPG Ranch, Missoula, Montana, USADepartment of Biological Engineering, Massachusetts Institute of Technology, Cambridge, Massachusetts, USA
Graduate Program in Biophysical Sciences, University of Chicago, Chicago, Illinois, USABioscience Division, The Microbiome Center, Argonne National Laboratory, Argonne, Illinois, USADepartment of Surgery, The Microbiome Center, University of Chicago, Chicago, Illinois, USAMarine Biological Laboratory, The Microbiome Center, Woods Hole, Massachusetts, USA
Principal-coordinate analysis (PCoA) (Hellinger distance metric) of plant community (A), soil chemistry (B), 16S rRNA gene (prokaryotic) (C), and ITS (fungal) (D) community structure colored by the aboveground community type (green for native, red for cheatgrass, blue for spotted knapweed, and yellow for leafy spurge) for the field plots. All sequenced samples from the field sites are plotted (including pseudoreplicates within sites). These data were collapsed by site prior to statistical analyses. PC1 and PC2, principal coordinates 1 and 2, respectively.
Changes in soil physicochemical variables relative to native plant communities across the invasion gradients, field plots, and experimental (exp) plots. P values were calculated using paired t tests (native versus invaded; significance threshold of P < 0.05). The uptick symbols denote a significant increase relative to controls, while the downticks indicate a significant decrease.
Heatmap showing Spearman’s rho values for significant correlations between weed abundances and bacterial classes (i.e., OTU data pooled at the class level; FDR-corrected P < 0.05). There were no bacterial classes that showed significant correlations with more than one weed species.
Differences in Verrucomicrobia genera across plant community types. The heatmap shows differences in the abundances of genera across plant communities (row-normalized z scores), where asterisks highlight significant differences (FDR-corrected P < 0.05). The plot to the right of the heatmap shows the rank-ordered abundance of each genus in the heatmap. The solid blue circles in the rank abundance plot are aligned with the genera identified in the heatmap.
Above- and belowground community structure along cheatgrass, spotted knapweed, and leafy spurge invasion transects. The bar plots at the top of the figure show the abundance of each invasive plant species along the linear invasion transects sampled in 2012. Scatter plots are PCoAs for prokaryotic (top row) and fungal (bottom row) communities for the three different gradients (left to right, see labels). PCoA points are colored by distance along the gradient: yellow denotes samples taken near the native side of the gradient, and black denotes samples taken toward the invaded side of the gradient.
Heatmaps of soil physicochemical metadata (row-normalized z scores) across invasion gradients. Samples were binned into four categories: (i) native, (ii) 25 to 50% invaded, (iii) 50 to 75% invaded, and (iv) invaded. Asterisks indicate variables that show a significant Pearson’s correlation with invader abundances across the gradient (P < 0.05).
Results from ANOVA showing significant differences in soil physicochemical variables in the experimental plots for year 1 and year 3. Letters (i.e., a and b) denote significant groupings based on Tukey's post hoc test.
Cartoon of the three different sampling designs presented in this paper. See the actual locations of the field sites, gradients, and experimental plots in Fig. S2. Extreme ends of the gradients (native and invaded) were used as replicate filed plots. Download FIG S1, TIF file, 0.2 MB.
Clustermap of Pearson’s correlation coefficients between measured soil chemical parameters (hierarchical clustering based on Euclidean distances). Most metals, including soil organic matter (OM), are weakly positively correlated with pH. However, Zn, Fe, Mn, and Cu are negatively correlated with pH. Download FIG S3, TIF file, 0.2 MB.
Jackknifed beta-diversity plots for bacterial (A) (weighted UniFrac) and fungal (B) (Hellinger distance) communities. (A) Bacterial communities were rarefied to 3,900 sequences per sample 10 times. (B) Fungal communities were rarefied to 2,000 sequences per sample 10 times. In both cases, opaque red spheres show the mean principal coordinate, and the translucent red spheroids represent the variance around the mean over 10 independent rarefactions. Overall, principal coordinates were robust to multiple subsamplings. Download FIG S4, TIF file, 1 MB.
PCoA plots showing Hellinger-transformed data for bacterial (16S), fungal (ITS), and plant communities. Vectors show correlation between significant soil chemical variables from Fig. 2 and the PC axes. Download FIG S5, TIF file, 0.4 MB.