Multivariate analyses of soil abiotic parameters, soil fungal, bacterial and plant communities
We tested the effects of conditioning time, conditioning plant community, and forb:grass ratio including all interactions on soil abiotic composition (soil nutrients and including soil pH) with a permutational analysis of variance (permanova; 999 permutations) using Euclidian distances. Furthermore, the effects on soil fungal (ITS2), soil bacterial (16S), and plant community composition were assessed with permanova (999 permutations) using Bray-Curtis dissimilarity. Fungal and bacterial data were square root-transformed and standardized using Wisconsin double standardization prior to calculating Bray-Curtis dissimilarity. Plot number was included in the models as a random effect to indicate that the one- and two-year conditioned subplots belong to the same plot. All multivariate analyses were performed in R, using the ‘vegan ’ package (R Core Team 2018; Oksanen et al. 2018) and community composition was visualized using ordination based on non-metric multidimensional scaling, using the ‘ggplot2 ’ package (Wickham 2016).
To assess whether responding plant communities responded to conditioning time and forb:grass ratio, and whether particular responding plant species drove these responses, we performed (restricted) redundancy analyses with either forb:grass ratio (categorical), or conditioning time (categorical) as explanatory variables. These analyses were performed and visualized for each of the six conditioning communities separately. Redundancy analyses and visualizations were performed in Canoco 5.03 (Microcomputer Power, Ithaca NY, USA).