Soil minerals affect taxon-specific bacterial growth

Soil microbial activity drives major terrestrial carbon fluxes but arises from highly diverse communities structured by the physically and chemically complex soil environment, including its mineral matrix. Different mineral assemblages influence microbial community development via surface properties (charge, surface area), nutrient content, weathering stage, pore and particle size distributions, aggregation, and rates of processing of soil organic carbon (SOC). Clay minerals have reactive surfaces that affect nutrient accessibility, moisture, and SOC accumulation and persistence; bacteria predominantly inhabit microaggregates and the clay fraction, often attached to mineral surfaces. While minerals can protect microbes from predation and stress, they can also limit mobility and access to organic matter (OM). Fresh OM can be stabilized by occlusion, adsorption, and metal complexation, leaving mineral-associated bacteria on surfaces with little accessible OM, potentially relying on fresh OM delivered by diffusion. Low-molecular-weight substrates are more readily consumed than polymers requiring extracellular enzyme activity. Prior work shows phylogenetically clustered taxa respond to simple substrates, whereas fewer, phylogenetically dispersed taxa respond to complex polymers. Thus, bacterial responses likely depend on both mineral composition (especially clays and short-range order, SRO, minerals) and the chemical composition of fresh carbon inputs. SRO minerals (e.g., allophane, imogolite, ferrihydrite) strongly influence SOC dynamics by protecting new OM and controlling SOC content and residence time. Observing in situ taxon-specific phenotypes like growth has been challenging. This study applies quantitative stable isotope probing (qSIP) with 18O-water to estimate taxon-specific bacterial growth in situ across three soils with different parent materials and SRO contents but similar vegetation and climate, and tests how carbon additions (root exudates or litter) and taxonomy modulate mineral effects on growth.

Design and sites: Three A-horizon (0–11 cm) soils were collected in June 2012 from the California Sierra Nevada with similar mixed conifer vegetation, climate, slope, and clay content, differing primarily in parent material: andesite (highest SRO), basalt (intermediate SRO), and granite (low SRO; dominated by hydroxy-interlayered vermiculite and kaolinite). SRO content (sum of oxalate-extractable Fe and allophane) correlated positively with SOC. Soil pH ranged from 5.8 (andesite) and 6.0 (granite) to 6.5 (basalt). At each site, five field replicates collected 15 m apart were sieved (2 mm) and composited to one field sample per site. Soils were stored at 4°C to reduce activity.

Incubations: In July 2013, 1 g dry-equivalent soil aliquots were placed in 15 mL tubes, rewetted to 60% field capacity for 5 days after an initial 24 h air-dry step, then amended with 200 μL water: either 97 atom% H2 18O (labeled) or natural-abundance deionized water (unlabeled) in parallel. Treatments: 3 soils × 3 carbon additions (water-only; root exudates; ground Pinus ponderosa litter) × 2 isotope levels × 4 replicates = 144 microcosms. Root exudate mix (350 μg C g−1 soil at 80 mg C mL−1) included carbohydrates (fructose, glucose, sucrose, lactate; 4 parts), organic acids (succinate, malate, citrate; 2 parts), and amino acids (serine, cysteine, alanine; 1 part). Litter addition was dried, ground pine needles at the same C rate. An additional 40 g soil incubation (run separately) quantified CO2-C respiration across soils and C additions. During the 7-day incubation, tubes were re-aerated on days 2 and 5; median O2 declined slightly (21% to 20.3%). Samples were harvested at day 7 and stored at −40°C.

DNA extraction and density fractionation: DNA was extracted from ~0.5 g dry soil using the FastDNA Spin Kit for Soil. For isopycnic centrifugation, 800–1000 ng DNA was added to CsCl gradients (final density ~1.9 g mL−1) with gradient buffer (Tris, KCl, EDTA) in OptiSeal tubes, spun at 127,000 × g for 96 h (Beckman TLN-100 rotor). Gradients were fractionated into ~14 fractions (150–200 μL), density measured by refractometer, DNA purified by isopropanol precipitation, resuspended in TE, and quantified by Qubit.

qPCR: 16S rRNA gene copies were quantified in all fractions using primers F515/R806 with Phusion polymerase chemistry. Standard curves (soil-derived standards) had efficiencies >90% and R2 >0.995. Thermal profile: 95°C 1 min; 40 cycles of 95°C 30 s, 64.5°C 30 s, 72°C 1 min.

Amplicon sequencing: Fractions with densities 1.66–1.74 g mL−1 were sequenced (7–13 fractions per tube). The 16S rRNA V3–V4 region was amplified (triplicate reactions; Phusion HF buffer; DMSO), followed by a barcoded tailing reaction (12-nt Golay barcodes, Illumina adapters). Libraries were bead-purified, quantified (qPCR), pooled, and sequenced on Illumina MiSeq v2 chemistry (2×150; 30% PhiX). Data are in NCBI SRA PRJNA701328.

Bioinformatics and qSIP calculations: Reads were demultiplexed in QIIME 2 (2021.4) and denoised with DADA2 to amplicon sequence variants (ASVs; 100% identity). Taxonomy was assigned using a sklearn classifier against SILVA 138. Samples with <3000 reads and ASVs present in <3 samples were removed, yielding 11,320 ASVs across 13,699,617 reads (>99% retention). Excess atom fraction 18O (EAF18O), a proxy for taxon-specific growth, was calculated from shifts in weighted average DNA density between unlabeled and labeled gradients based on isotope substitution models. ASVs were retained if present in ≥3 fractions per sample and in ≥2 replicates per treatment, leaving 3,476 bacterial ASVs (archaea negligible). A tube effect on density was corrected per prior methods. EAF18O was computed in R 4.1.1 using data.table and publicly available qSIP scripts.

Statistical analyses: Community composition and growth structure were analyzed by PerMANOVA (adonis, 1000 permutations) on Euclidean distances of ASV relative abundances and EAF18O, visualized via NMDS. Variance partitioning (restricted maximum likelihood with nested random effects) was performed on 310 ASVs shared across all soils and belonging to sufficiently represented lineages (≥2 orders per phylum; families with ≥3 members): model EAF ~ 1 + 1|Phylum/Class/Order/Family/Genus/ASV/soil/substrate. Mean relative growth per community was computed as mean EAF18O across shared ASVs; two-way ANOVA with Tukey’s HSD tested soil and substrate effects (α=0.05). Family-level growth responses to substrates were assessed via bootstrap differences of means (1000 iterations; 95% CI). ASVs were classified as SRO-stimulated, -unaffected, or -suppressed by the significance and sign of the slope of EAF18O vs. soil SRO proportion (α=0.05).

• Soil respiration: Despite higher SOC, andesite and basalt (higher SRO minerals) had lower 7-day respiration than granite (low SRO). Under water-only conditions, granite respired ~4× more CO2-C than andesite and basalt. Fresh C increased CO2-C in all soils, with exudates > litter.
• Community composition: ASV richness: andesite 2319, basalt 1596, granite 1206 (across C treatments). 484 ASVs (13.9%) were shared across all soils/treatments and comprised >57% of andesite and basalt communities and >80% of granite under water-only. Soils differed significantly in composition (PerMANOVA F2,28=39.02, R2=0.71, p<0.01); substrate addition had no significant effect (F2,28=1.65, R2=0.03, p=0.13). Dominant phyla: Actinobacteria and Proteobacteria.
• Mean taxon-specific growth (EAF18O): Soil type, not substrate, primarily determined growth. Granite had ~0.1 higher EAF18O on average than andesite and basalt (p<0.01; Tukey); andesite and basalt did not differ. Substrate effects were generally weak, except exudate decreased growth in basalt vs water-only (p<0.05). For shared taxa, soil explained more variation in EAF18O than substrate (PerMANOVA: Soil F2,29=7.90, R2=0.34, p<0.01; Substrate F2,29=1.69, R2=0.04, p=0.09).
• Across soils, most taxa grew fastest in granite. Depending on treatment, 6–25× more shared ASVs had higher growth in granite vs andesite/basalt (95% CI difference of means). However, due to lower richness, total numbers of growing ASVs were 40% and 25% fewer in granite than andesite and basalt, respectively, under water-only.
• Variance partitioning of growth (EAF18O): Soil type explained 53.5% of the explained variance; taxonomy 37.7% (phylum 6.3%, family-to-ASV 31.4%); substrate 8.8%.
• Family-level substrate responses: In andesite, 48/70 families were stimulated by exudates vs water-only—approximately twice as many as by litter—and about half of families grew more under exudates than litter. In granite, 29 families were stimulated by exudates and 28 by litter. In basalt, fresh C suppressed growth in many families (22 suppressed by exudates; 15 by litter). Xanthomonadaceae showed high growth but no substrate effect; Sphingobacteraceae was stimulated by exudates across soils; Pseudomonadaceae increased under both litter and exudates regardless of soil. Streptomycetaceae responses depended on soil (suppressed by fresh C in basalt, stimulated in granite, unchanged in andesite). Acidobacteraceae (Subgroup 1) increased with exudates in andesite and basalt but decreased with litter; in granite it showed high growth without treatment change (EAF18O > 0.31).
• Effect of SRO minerals on growth: Among 484 shared ASVs, most were unaffected or suppressed by SRO presence. Water-only: 3 ASVs (<1%; Actinobacteria: Atopobium, Rothia, Actinomyces) were SRO-stimulated; 35.3% were SRO-suppressed. Exudates: 5 ASVs stimulated; 29.7% suppressed. Litter: 9 ASVs stimulated; 35.5% suppressed. SRO-suppressed taxa accounted for 26–48% of community 16S copies per soil, whereas SRO-stimulated taxa comprised <0.5%.

The study demonstrates that soil mineral assemblage—particularly short-range order (SRO) minerals—exerts strong control over taxon-specific bacterial growth in situ, beyond effects on community composition. Bacteria shared across soils generally attained their highest growth in the granite soil with minimal SRO phases, likely due to weaker organo-mineral interactions and greater OM availability, consistent with higher CO2 production. Conversely, high SRO content slowed growth despite higher SOC, indicating that SOC quantity alone does not predict bacterial growth; mineral protection of OM reduces accessibility, supporting current paradigms of SOC persistence. High SRO soils exhibited greater bacterial richness, potentially due to more microsites and niche diversity associated with high specific surface area, even as growth was suppressed by strong sorption. Substrate additions had comparatively minor effects on mean community growth relative to soil type, and responses varied by soil. Exudates did not uniformly enhance growth over litter, possibly because over 7 days litter leached soluble DOM that was rapidly assimilated, and because mineral composition more strongly governs substrate accessibility. In basalt, fresh C suppressed growth for many bacterial families even as total respiration increased, suggesting potential competitive interactions (e.g., fungal stimulation) and/or strong sorption limiting bacterial access in a low-biomass community. Taxonomy explained about one-third of growth variation, driven largely by family-level and finer classifications, indicating evolutionary constraints on growth traits, while parent material and mineralogy explained the majority, underscoring environment-by-lineage interactions. Overall, bacterial–mineral–substrate interactions shape SOC processing and may influence soil C persistence and loss through taxon-specific growth dynamics.

Mineral composition, especially the abundance of short-range order (SRO) minerals, suppresses the relative growth of many bacterial taxa in soil. Soil type (parent material/mineral assemblage) explains substantially more variation in taxon-specific growth than fresh carbon additions, with taxonomy contributing at family-level and below. Most shared taxa grew fastest in the low-SRO granite soil, despite its lower SOC, highlighting the importance of organo-mineral accessibility rather than SOC quantity alone. Family-level responses to exudates vs litter were soil-dependent, and in SRO-rich or intermediate soils, fresh C did not consistently stimulate bacterial growth. These findings emphasize that mineral–microbe–substrate interactions are key determinants of SOC processing and persistence. Future work should employ longer-term experiments, controlled mineral gradients or synthetic soils for mechanistic inference, inclusion of fungal growth dynamics (e.g., fungal qSIP), and broader soil comparisons to generalize taxon-specific growth responses across mineral contexts.

• Site compositing and design: Five field replicates per site were composited to a single sample, so within-site variability was not assessed; the design preserves inter-soil comparisons only.
• Natural-soil lithosequence approach: While ecologically realistic, this approach provides weaker mechanistic inference than experiments with synthetic soils or defined mineral gradients.
• Short incubation duration (7 days): May favor detection of rapid responses to soluble substrates (e.g., leached DOM from litter) and may not capture longer-term decomposition or community shifts.
• Potential unmeasured biotic interactions: Suppressed bacterial growth with C additions in basalt may reflect fungal competition, which was not directly measured; archaeal contributions were negligible in these datasets and not analyzed.
• Technical considerations: One treatment (granite–litter) had only two replicates after filtering; tube-specific density effects were corrected but may introduce uncertainty. Results reflect conditions (e.g., moisture, aeration) of microcosm incubations.