Data analyses
Before analysis, data was subjected to Shapiro-Wilk test for checking
its normal distribution. Two-way analysis of variance (ANOVA) followed
by Least Significant Difference (LSD) post hoc tests with p-value
adjusted using ”BH” method were applied to analyze the difference in
microbial communities and soil physicochemical properties between the
plots and among sites. Data was also subjected to principal component
analysis (PCA) using the FactoMineR
(https://CRAN.R-project.org/package=
FactoMineR) and factoextra
(https://CRAN.R-project.org/package=
factoextra) packages, to determine whether the overall physicochemical
properties of the invaded plots were different from those of uninvaded
plots across different sites along the altitudinal gradient.
We also used the community ecology R package‘vegan_2.5.6’
(https://github.com/vegandevs/vegan) for plotting the rarefaction
curves for assessing the sequencing depth of individual samples both in
bacteria and fungi at species level. We measured the α-diversity of
bacterial and fungal communities within the samples. Both richness (i.e.
number of different species present in a sample) and evenness (i.e. how
common in number different species in a sample are) measures were
computed for the invaded and uninvaded plots across all the sampling
sites along an altitudinal gradient using QIIME v1.9.1. Total observed
OTUs and Chao-1 estimator were employed to measure the observed and real
richness (i.e. observed plus less frequent species) of the samples. For
measurement of species diversity in each sample, we used Shannon index
and Simpson index – the two commonly employed diversity indices in
community ecology.
We determined the β-diversity (the differences of bacterial and fungal
diversity between the samples) using non-metric multidimensional scaling
(NMDS) plots based on Bray-Curtis dissimilarity index in vegan_2.5.6.
This index measures the relatedness of the species composition of soil
microbial communities across different sites and is a distance-based
method that maximizes rank dependent correlation between the original
distances and the distances between samples from multidimensional space
into new reduced 2D ordination space. Analysis of similarities (ANOSIM),
a non-parametric statistical test, was carried out to determine the
variation in the community composition.
Similarly, the most significant soil nutrient parameters governing the
composition of microbial communities at different sites were determined
by Canonical Correspondence Analysis (CCA). Forward selection method of
explanatory variables was carried out to find out the model that
includes only those parameters which contributed significantly to the
overall microbial diversity at different sites using ‘vegan_2.5.6’.
Chart.Correlation() function from the “PerformanceAnalytics” package
was used to estimate Spearman’s rho (ρ) statistic, which
is a rank-based measure of association between environmental variables
(https://github.com/braverock/PerformanceAnalytics). For data
visualization, all plots were created using the package ggplot 2_3.3.2
(https://ggplot2.tidyverse.org) in R software (R Core Team 2020).