Soil texture and environmental conditions influence the biogeochemical responses of soils to drought and flooding

Projected changes in precipitation timing and distribution under climate change are expected to increase the frequency of droughts and extreme inundation events. Such moisture extremes drive key soil biogeochemical processes, yet predictive understanding remains limited. Antecedent moisture conditions exert strong control on soil CO2 flux responses, including the well-known rewetting-induced respiration pulse (Birch effect), which arises from multiple physicochemical and biochemical destabilization mechanisms of soil organic carbon. Moisture shifts also alter soil solution ionic strength—up to nine orders of magnitude higher in dry versus saturated soils—affecting selective desorption and solubilization of organic molecules from minerals and protected microsites. Microbial processes are likewise influenced: drought imposes osmotic stress in dry pores, while saturation enhances diffusion and connectivity through the pore network. This study aims to develop a generalized understanding of how soil water extremes (drought and flood) influence soil carbon cycling across diverse soils through microbial, chemical, and physical traits. The authors hypothesized: (i) simulated drought would increase the proportion of complex aromatic C species and expression of osmoprotectants; and (ii) extended saturation would increase expression of motility factors (e.g., flagella and gliding motility).

Prior studies show that historical and antecedent moisture strongly shape soil respiration responses to current moisture and that rewetting typically induces a CO2 pulse (Birch effect). Mechanisms include destabilization of mineral-associated and physically protected organic matter via changes in ionic strength, sorption-desorption dynamics, and microbial cell lysis. Soil pore connectivity regulates access of microbes to substrates and the diffusion of solutes, with wetter conditions increasing hydrologic connectivity relative to field-moist, partially disconnected pore networks. Microbial communities exhibit stress responses to drying (e.g., osmolyte production) and adapt to local hydrologic regimes, leading to site-specific behavior under moisture extremes. Together, these findings suggest that both soil physicochemical properties (texture, mineralogy, pore architecture) and environmental history shape biogeochemical responses to drought and flooding.

Study sites and soils: Intact 5 cm diameter, 15 cm height soil cores were collected from three U.S. sites representing contrasting climates and pedology: (1) Alaska (Caribou-Poker Creeks Research Watershed; Gilmore silt loam; black spruce–moss vegetation); (2) Florida (Disney Wilderness Preserve; sandy Immokalee Series; pine flatwoods); (3) Washington (Secret River tidal freshwater floodplain; Ocosta silty clay loam; Sitka spruce). Cores were shipped on ice to PNNL within 36 hours. Experimental design: For each site, cores were randomly assigned to three 30-day treatments at 21 °C in the dark: field moist (maintained at in-situ water content), drought (air-dried to constant weight, then incubated dry), and flood (saturated from below and incubated saturated). Additional controls included (i) baseline (immediately deconstructed upon arrival) and (ii) time-zero saturation (saturated then immediately deconstructed). Target gravimetric moistures (%, w/w) during incubation: Alaska 37 (field), <0.01 (drought), 114 (flood); Florida 9, <0.01, 36; Washington 194, 62, 197. Replication: n=5 cores per site-treatment combination. Gas fluxes: CO2 and CH4 were measured every 3 days during the 30-day incubations and hourly during a post-incubation 24 h saturation (to capture rewetting pulses) using a Picarro G2301 analyzer. Fluxes were computed from headspace concentration change using chamber volume, soil mass, pressure, temperature, and the gas constant. Pore-water extraction and chemistry: After the 30-day incubation and 24 h saturation, pore water was extracted using Tempe Pressure Cells at −1.5 kPa (coarse pores >200 µm) and −50 kPa (fine pores 6–20 µm) corresponding to pore throat diameters via the Kelvin equation. Samples were filtered (0.22 µm) and analyzed for DOC concentration by combustion catalytic oxidation (Shimadzu TOC-5000A). Ultrabroad molecular characterization was conducted by electrospray ionization FT-ICR-MS (21 T; mass accuracy <1 ppm; 200<m/z<1200), following PPL solid-phase extraction. Molecular formulas (C, H, O, N, S, P) were assigned (S/N>7; error <1 ppm) and classified into Van Krevelen-based compound classes (lipid-, protein-, amino sugar-, carbohydrate-, lignin-, tannin-, condensed hydrocarbon-like, etc.). Diversity was assessed by Shannon index on detected molecular formulas. Soil carbon pools: Water-soluble organic carbon (WSOC) was extracted from air-dried soil (1:5 soil:water, 2 h shaking at 4 °C), filtered (0.22 µm), quantified by TOC analyzer, and characterized by FT-ICR-MS. Total C and TOC were measured on a VarioEL Cube elemental analyzer. Microbial genomics and transcriptomics: DNA and RNA were extracted (Qiagen PowerSoil kits). DNA (shotgun metagenomes) sequenced on Illumina HiSeq 1500. RNA quality checked (Bioanalyzer RIN≥8), rRNA depleted (Ribo-Zero), cDNA libraries prepared (ScriptSeq v2), and sequenced on HiSeq 2500 (2×250 bp). Reads were cleaned (bbduk), merged (bbmerge), annotated to TIGRFAMs (MaxRebo). Ribosomal genes were removed, abundances normalized; rare genes (<0.001% in all samples) were excluded. Pore structure and hydrology: X-ray micro-CT (voxel 28 µm) assessed pore size distribution and connectivity; image analysis performed in ImageJ. Water retention curves were generated using HYPROP and fitted with the van Genuchten model to quantify water content vs. matric potential. Statistics: FT-ICR-MS compound class relative abundances analyzed by MANOVA; PCA used for visualization (ggbiplot). Treatment×pore size interactions were considered, analyzing 1.5 and 50 kPa classes separately. Metagenome and metatranscriptome community structure tested by PERMANOVA (vegan adonis; Euclidean distances). Discriminatory genes identified via LEfSe (log score >2). Differential expression under drought or flood vs. baseline was tested with DESeq2; normalized abundances visualized in heatmaps (cividis colormap). Data management: Analyses performed in R 3.6.0 with dplyr and ggplot2; code and data archived in public repositories.

- Pore-water DOC concentration responses: Alaska showed strong drought-induced DOC increases across pore classes relative to time-zero controls (coarse pores: drought 111.8±41.45 vs. time zero 26.3±2.78 mg L−1; fine pores: drought 186.9±62.65 vs. time zero 29.14±4.41 mg L−1; Dunnett’s test, P<0.05). Florida exhibited no statistically significant DOC change among treatments despite numerical declines (P>0.10). Washington showed elevated DOC across all incubations compared to time zero, suggesting a strong sampling/incubation effect; DOC did not differ significantly among treatments (ANOVA F=2.58, P=0.10). - Molecular diversity (Shannon index) generally declined during incubation across sites and pore classes, except for drought-incubated Alaska soils, which maintained higher diversity relative to time zero, consistent with destabilization or newly available molecules. - FT-ICR-MS composition: Initial DOC composition differed strongly by site (MANOVA F=14.71, P<0.001): Alaska dominated by lignin-like; Florida by amino sugar and carbohydrate-like; Washington by protein- and lipid-like. After drought, Alaska gained new peaks in high-oxygen aliphatic carbohydrate- and amino sugar-like regions (higher NOSC), indicating more oxidized, microbially favorable compounds. Florida lost simple aliphatic peaks and gained low-O polyphenolic and aliphatic peaks across all treatments, indicating enrichment in more complex, less energetically favorable compounds, consistent with a sampling/incubation effect overriding moisture treatment. Washington clustered tightly across treatments in PCA, indicating minimal qualitative changes in DOC composition despite DOC concentration increases. - Microbial functional potential vs. activity: Metagenomes separated by site (PERMANOVA F=18.22, P=0.001) but not by treatment (F=0.95, P=0.444), indicating stable functional potential over 30 days. Metatranscriptomes separated by site (F=5.20, P=0.007) and treatment (F=3.97, P=0.006), indicating treatment-responsive gene expression. - Site-specific baseline expression: Washington showed greater representation of anaerobic processes (reductases, nitrate respiration), consistent with frequent inundation; Florida showed osmoprotectant and stress response genes (trehalose and ppGpp synthetases), consistent with periodic wet/dry and sandy soils. - Drought and flood gene expression: Drought increased expression of multiple sporulation genes (spoVID, spoIIAB, spoIIID, spoIVA, spoIIAA, spoVT, spoOA, spoVD) and osmoprotectant synthesis genes (treC, betB), particularly in Alaska and Washington. Flood also increased expression of sporulation-associated genes (spoIIID, spoVT, spoIVA, spoOA, sigK, sigE), suggesting sporulation as a general stress response to hydrologic extremes and/or sampling disturbance. - Motility: In Alaska, drought decreased expression of gliding motility genes while increasing flagellar motility gene expression, implying a shift toward swimming apparatus under low-water conditions. Washington maintained high expression of gliding motility genes even under “drought,” likely due to high residual moisture (~62% w/w). Florida showed low gliding motility expression at baseline, reflecting low water content. - Gas fluxes: Washington drought cores exhibited a pronounced CO2 flush upon rewetting despite minimal pore-scale chemical changes, indicating transient macropore-driven responses; micropores retained ~60% water during drought. - Physical controls: Washington soils exhibited greater pore and hydrologic connectivity, buffering moisture stress; Alaska had moderate clay (~15%) likely stabilizing SOC with potential for desorption under drought; Florida’s sandy texture facilitated access to simple molecules that were preferentially consumed, enriching more complex, aromatic DOC classes across treatments. - Overall, drought effects were stronger and more consistent on respiration, pore-water C, and microbial expression than flood, but responses were not uniform across sites and were shaped by texture, pore architecture, and environmental history.

The study’s hypotheses were partially supported. Drought increased expression of sporulation and osmoprotectant genes (e.g., treC, betB) in Alaska and Washington, consistent with microbial responses to osmotic stress. Contrary to expectations, drought did not generally increase complex aromatic compounds; instead, Alaska shifted toward more oxidized, energetically favorable molecules, suggesting desorption of mineral-associated OM and/or microbial transformation of complex substrates. Flooding had weaker and less consistent impacts on DOC chemistry and gene expression, with Washington showing minimal response due to already saturated field conditions. The observed site-specificity underscores that soil texture, pore size distribution, and hydrologic connectivity govern substrate accessibility and chemical responses, while microbial communities reflect adaptation to historical moisture regimes, driving distinctive transcriptional responses. The strong sampling/incubation effect on Washington DOC concentrations and Florida DOC composition highlights the sensitivity of certain soils to physical disturbance, potentially overshadowing moisture treatment effects. Discrepancies between pore-scale chemistry and core-scale gas fluxes (e.g., Washington’s CO2 flush despite little DOC compositional change) indicate that macropore processes and transient physicochemical mechanisms can dominate respiration dynamics during rewetting. Overall, results emphasize that predicting biogeochemical responses to hydrologic extremes requires integrating pore-scale physics, organo-mineral interactions, and microbial ecology tailored to each soil’s environmental history.

Drought induced stronger and more pervasive changes than flooding in respiration, pore-water carbon, and microbial gene expression, but responses were highly site-specific. The first hypothesis was partially supported: drought increased sporulation and osmoprotectant expression (Alaska, Washington), but did not increase complex aromatic compound abundance; instead Alaska’s pore-water shifted toward more oxidized molecules. The second hypothesis was partially supported: flood did not significantly increase motility gene expression, but drought decreased gliding motility in Alaska while increasing flagellar motility. The study proposes that soil texture and associated pore size distribution and water spatial distribution primarily drive chemical responses, whereas environmental conditions and disturbance history primarily drive microbial responses to new hydrologic regimes. Because “drought” and “flood” impose different absolute water contents across soils, their stressfulness is soil-dependent (e.g., Washington’s field-moist ≈ saturated; Florida’s field-moist already very dry). Future work should expand sample sizes and sites, incorporate mineralogy and litter quality, quantify microbial fixed vs. planktonic fractions, and link pore-scale chemistry and hydrology to ecosystem-scale fluxes to improve predictions of soil C–water dynamics under intensifying hydrologic extremes.

- Small sample size per site-treatment (n=5) and three sites limit generalizability. - Operational definitions of drought (air-dry to constant weight) and flood (saturation) translate to very different absolute moisture states across soils, affecting stress intensity and comparability. - Potential sampling and incubation disturbance effects were evident (notably in Washington DOC and Florida DOC composition), possibly overshadowing treatment effects. - Florida metatranscriptome had fewer sequencing replicates due to low RNA yields, reducing power to detect expression changes. - The study did not explicitly control for or analyze all potentially influential variables (e.g., detailed mineralogy, litter quality inputs), and incubations lacked ongoing litter inputs, likely reducing SOM molecular diversity over time. - Laboratory conditions (constant temperature, dark, static cores) may not capture field heterogeneity, redox dynamics, and plant–microbe interactions.