Rhizosphere microbial community composition shifts diurnally and in response to natural variation in host clock phenotype

Abstract

Plant-associated microbial assemblages are known to shift at time scales aligned with plant phenology, as influenced by the changes in plant-derived nutrient concentrations and abiotic conditions observed over a growing season. But these same factors can change dramatically in a sub-24-hour period, and it is poorly understood how such diel cycling may influence plant-associated microbiomes. Plants respond to the change from day to night via mechanisms collectively referred to as the internal "clock," and clock phenotypes are associated with shifts in rhizosphere exudates and other changes that we hypothesize could affect rhizosphere microbes. The mustard Boechera stricta has wild populations that contain multiple clock phenotypes of either a 21- or a 24-hour cycle. We grew plants of both phenotypes (two genotypes per phenotype) in incubators that simulated natural diel cycling or that maintained constant light and temperature. Under both cycling and constant conditions, the extracted DNA concentration and the composition of rhizosphere microbial assemblages differed between time points, with daytime DNA concentrations often triple what were observed at night and microbial community composition differing by, for instance, up to 17%. While we found that plants of different genotypes were associated with variation in rhizosphere assemblages, we did not see an effect on soil conditioned by a particular host plant circadian phenotype on subsequent generations of plants. Our results suggest that rhizosphere microbiomes are dynamic at sub-24-hour periods, and those dynamics are shaped by diel cycling in host plant phenotype. IMPORTANCE We find that the rhizosphere microbiome shifts in composition and extractable DNA concentration in sub-24-hour periods as influenced by the plant host's internal clock. These results suggest that host plant clock phenotypes could be an important determinant of variation in rhizosphere microbiomes.

Keywords: diel cylcing; dynamics; microbiome; rhizosphere.

Fig 1

Concentration of extracted DNA as a function of sampling time. DNA concentration (ng/μL; 16s locus) was divided by soil mass (milligram) to estimate microbial biomass within samples and is shown on the y-axis (see Discussion for caveats to this approach). Sampling time is shown on the x-axis. Panels (a) and (b) show DNA concentration for genotype 6 in cycling (a) and free-running conditions (b). Likewise, panels (c) and (d) show DNA concentration for genotype 15 in cycling (c) and free-running (d) conditions. Note that time in (b) and (d) is subjective, that is, day and night reflect the times that light and dark periods would have been experienced by the experimental plants based on the preceding cycling conditions. Bars show 95% confidence intervals around estimates.

Fig 2

Differences in relative abundances of several ASVs that were associated with day and night time points. Relative abundances are shown on the y-axis as estimates of Dirichlet pi parameters (sampling group-wide proportions). Treatment, either cycling or free-running, and genotype are shown on the x-axis. Colors of points denote sampling time. DMM confirmed a highly certain difference in relative abundances for these taxa between night and day time points.

Fig 3

Plant biomass as a function of rhizosphere microbiome conditioning via overstory history. For this experiment, plants of genotypes 6 and 15 were grown in soil that had been previously used to grow plants of a particular genotype (shown on the x-axis). Since each genotype influences the rhizosphere microbiome uniquely, this experiment tested if the conditioning of a particular genotype generally led to increased performance in subsequent generations. Genotypes differed in that they had short (21-hour) or long (24-hour) period clock phenotypes (as labeled). We observed plants of genotype 20 were larger than genotypes 4 and 15, but we did not see any consistent effects associated with clock phenotype nor did we observe treatment by genotype interactions (P = 0.71).