The Influence of Kerosene on Microbiomes of Diverse Soils

Abstract

One of the most important challenges for soil science is to determine the limits for the sustainable functioning of contaminated ecosystems. The response of soil microbiomes to kerosene pollution is still poorly understood. Here, we model the impact of kerosene leakage on the composition of the topsoil microbiome in pot and field experiments with different loads of added kerosene (loads up to 100 g/kg; retention time up to 360 days). At four time points we measured kerosene concentration and sequenced variable regions of 16S ribosomal RNA in the microbial communities. Mainly alkaline Dystric Arenosols with low content of available phosphorus and soil organic matter had an increased fraction of Actinobacteriota, Firmicutes, Nitrospirota, Planctomycetota, and, to a lesser extent, Acidobacteriota and Verrucomicobacteriota. In contrast, in highly acidic Fibric Histosols, rich in soil organic matter and available phosphorus, the fraction of Acidobacteriota was higher, while the fraction of Actinobacteriota was lower. Albic Luvisols occupied an intermediate position in terms of both physicochemical properties and microbiome composition. The microbiomes of different soils show similar response to equal kerosene loads. In highly contaminated soils, the proportion of anaerobic bacteria-metabolizing hydrocarbons increased, whereas the proportion of aerobic bacteria decreased. During the field experiment, the soil microbiome recovered much faster than in the pot experiments, possibly due to migration of microorganisms from the polluted area. The microbial community of Fibric Histosols recovered in 6 months after kerosene had been loaded, while microbiomes of Dystric Arenosols and Albic Luvisols did not restore even after a year.

Keywords: bearing capacity; biodegradation; controlled study; ecological indicators; gasoline; jet-fuel; soil metagenome; soil pollution; total petroleum hydrocarbons; xenobiotic compounds.

Figure 1

Alpha-diversity estimated by the Shannon index at the ASV level. The numbers above the histograms represent the initial kerosene loads, in g/kg. 16S rRNA regions are shown by the dot (V3V4) and triangle (V4V5).

Figure 2

Differentiation of the studied soils in the pot and field experiments: (A) principal component analysis (PCA) of the soil chemical properties; (B) principal coordinate analysis (PCoA) of the microbiome composition based on weighted UNIFRAC as a metric; (C) the relative abundance of the top 12 most frequent phyla among all uncontaminated samples in all experiments (all time points). The soil is shown by the color: (dark) red, Albic Luvisols; green, Dystric Arenosols; blue, Fibric Histosols. The microbial composition was estimated using the V3V4 region of 16S rRNA gene.

Figure 3

The relative abundance of the top 10 most frequent bacterial phyla (A); the top 40 most frequent bacterial families in the studied soils (B). Families are sorted and colored according to phyla. The numbers above the histograms represent the initial kerosene loads, in g/kg. The bacterial composition was assessed using the V3V4 region of 16S rRNA; the results for the V4V5 region are provided in Supplementary Figures S12 and S13. The families whose relative abundance was increased after kerosene pollution in all, or almost all, soils are shown in red in the legend. The dominant families whose relative abundance increased after kerosene pollution in a specific soil type and experiment setup are shown in bold.

Figure 4

Temporal changes in the concentration of kerosine (g/kg) during the pot and field experiments (0, 1, 5, 10, 25, and 100 are the initial concentrations of kerosene in g/kg). The red line indicates the lowest detectable level.

Figure 5

Principal coordinate analysis (PCoA) plots for soils with different initial kerosene loads. The load in g/kg is shown by the dot color, and the day after kerosene treatment is shown by the dot shape (see legend). The beta-diversity is estimated by the weighted UNIFRAC metric (the V3V4 16S rRNA region). The names of the most abundant bacterial families mark the main shifts in the soil microbial composition. The families whose relative abundance increased after kerosene pollution in all, or almost all, soils are in red. Dominant families whose relative abundance increased after kerosene pollution in a specific soil type and experimental setup are in boldface. The PCoA for the V4V5 region and for Bray–Curtis dissimilarity are in Supplementary Figures S8–S11.

Figure 6

PCA plot based on the relative abundance of metabolic pathways in samples, as predicted by Picrust2 on the V3V4 data. All soils were plotted together in one plane and then separated to four subplots for better readability. Separate PCA plots for individual soils are shown in Supplementary Figure S15.

Figure 7

Heatmap with the relative abundance of bacterial families with a significant difference between highly contaminated (kerosene load = 25 or 100, day > 3) and control (kerosene load = 0, day > 3) samples of Albic Luvisols, the pot experiment. In the left panel, the abundance of each bacteria family is scaled (Z-score) across all samples to highlight the relative changes between samples. In the right panel, the medians of-centered log-ratio (clr) transformed abundances of bacterial families in highly contaminated and control samples are shown. The V3V4 data are shown, a similar heatmap for the V4V5 region is shown in Supplementary Figure S19.

Figure 8

Heatmap with the relative abundance of the top 50 most abundant bacterial families with significant differences between highly contaminated (kerosene load = 25 or 100, day > 3) and control (kerosene load = 0, day > 3) samples of Dystric Arenosols, the pot experiment. Notation as in Figure 7. The V3V4 data are shown. All bacterial families with significant differences between highly contaminated and control samples according to V3V4 data are shown in Supplementary Figure S20, a similar heatmap for the V4V5 region is shown in Supplementary Figure S21.

Figure 9

Heatmap with the relative abundance of bacterial families with the significant difference between the highly contaminated (kerosene load = 25 or 100, day > 3) and control (kerosene load = 0, day > 3) samples of Albic Luvisols, the field experiment. Notation as in Figure 7. The V3V4 data are shown, a similar heatmap for the V4V5 region is shown in Supplementary Figure S22.

Figure 10

Heatmap with the relative abundance of bacterial families with significant differences between highly contaminated (kerosene load = 25 or 100, day > 3) and control (kerosene load = 0, day > 3) samples of Fibric Histosols, the field experiment. Notation as in Figure 7. The V3V4 data are shown, a similar heatmap for the V4V5 region is shown in Supplementary Figure S23.