Biological and experimental factors that define the effectiveness of microbial inoculation on plant traits: a meta-analysis

Abstract

Bacterial and fungal microbiomes associated with plants can significantly affect the host's phenotype. Inoculating plants with one or multiple bacterial and fungal species can affect specific plant traits, which is exploited in attempts to increase plant performance and stress tolerance by microbiome engineering. Currently, we lack a comprehensive synthesis on the generality of these effects related to different biological (e.g. plant models, plant traits, and microbial taxa) and experimental factors. In a meta-analysis, we showed that the plant trait under consideration and the microbial taxa used to inoculate plants significantly influenced the strength of the effect size. In a methodological context, experiments under sterilized conditions and short-term periods resulted in larger positive effects on plant traits than those of unsterilized and long-term experiments. We recommend that future studies should not only consider (short-term) laboratory experiments with sterilized plants and single inoculants but also and more often (long-term) field or greenhouse experiments with naturally occurring microbial communities associated with the plants and inoculated consortia including both bacteria and fungi.

Keywords: bacteria; experimental setting; fungi; inoculation; microbes; plant traits.

Figure 1

The individual effect of biological factors on the strength of the effect size. The estimated overall effect size (lnRR) of microbial inoculation on the quantitative traits of the plant (A) and the effect of biological factors on the strength of lnRR (B–F). Solid dots with thick black bars represent the effect size and 95% CI. The thin bars are representative of the PI, indicating the heterogeneity among the effect sizes or expected values that future studies may find. The transparent, colorful circles show individual data points whose size is adjusted according to the precision of the study. The size of each colorful circle is proportional to its relative weighting in the overall model and the number of replicates (n), thus giving more weight to well-replicated studies. The k-value represents the number of observations (data points) in the model, while the value in parentheses shows the number of studies. If the CIs do not cross the vertical dotted lines (lnRR = 0), the effect size is significant at P < .05. If the effect size was statistically significant for each biological factor (at P < .05), we performed pairwise comparisons of subgroups within each biological factor using Tukey’s HSD post-hoc test. Only for those significant biological factors are shown; asterisks that are denoted to sublevels that are significant within each biological factor (if the CIs do not cross the vertical dotted lines). Letters represent significant pairwise differences between sub-levels based on post-hoc Tukey’s test. ^* For the flower, only the number of days required for flowering was included in this sub-category. Therefore, the significant negative effect size means that the plants inoculated with microbes flowered earlier than the non-inoculated control plants. See the main text for more explanation.

Figure 2

The individual effect of experimental factors on the strength of the effect size. Asterisks are denoted to sub-levels that are significant within each biological factor, while letters represent significant pairwise differences between sub-levels based on post-hoc Tukey’s test. See Fig. 1 for a general explanation of the orchid plot.

Figure 3

The effect of soil parameters on microbial inoculation outcomes. Spearman correlation test was performed to check for the relationship between soil parameters and lnRR. Soil properties included pH, organic matter content (%), soil water holding capacity (%), electrical conductivity (ds.m), total nitrogen (%), and total phosphate (%).

Figure 4

Future research directions in using microbes to improve plant traits. In term of microbial inoculation, we recommend that future studies should consider inculcating plants with a consortium containing both bacteria and fungi strains instead of a single strain. The short- and long-term effects (experimental periods) of such microbial consortiums on plant traits should be tested not only under sterilized and non-sterilized growth conditions but also under robust field experiments (experimental conditions) with naturally occurring microbial communities associated with the plants.