Plant diversity drives positive microbial associations in the rhizosphere enhancing carbon use efficiency in agricultural soils

The study addresses how plant diversity in agricultural systems affects soil carbon cycling via interactions in the rhizosphere microbial community. While biodiversity–ecosystem functioning research has established that plant diversity enhances productivity, stability, and resilience, belowground mechanisms remain less understood. Soils represent the largest terrestrial carbon reservoir, and microbial decomposition controls the fate of plant-derived carbon. The authors posit that plant diversity can modify rhizosphere microbial community structure and interactions (e.g., complementarity, facilitation, cross-feeding), thereby shifting the balance between microbial growth and respiration and increasing microbial community carbon use efficiency (CUE). They test the overarching hypothesis that undersown plant diversity influences microbial associations in the rhizosphere of a main crop (barley), with consequences for nutrient cycling and soil C dynamics. Specific hypotheses: (1) plant diversity positively influences microbial community structure and associations in the rhizosphere; (2) soil microbial community structure and positive network connectivity influence microbial community CUE; and (3) higher plant diversity increases plant biomass, contributing to soil C inputs. The work seeks to inform management strategies for enhancing soil carbon retention in agroecosystems.

The paper synthesizes evidence showing that biodiversity enhances ecosystem functioning (productivity, stability, resilience) and that many mechanisms operate belowground. Prior studies indicate plant diversity can increase soil microbial activity and carbon storage, and that microbial community composition shapes soil organic matter (SOM) chemistry and thermal stability, potentially leading to longer SOM residence times. Microbial necromass significantly contributes to SOM formation. The rhizosphere is highlighted as a hotspot of plant–microbe interactions where root exudates can drive microbial assembly and function. Work on microbial CUE suggests community composition, biotic interactions, and environmental conditions modulate growth versus respiration allocation. Network-based analyses have linked microbial network connectivity with ecosystem processes, though inferring ecological interactions from co-occurrence presents challenges. The study builds on these insights by empirically connecting plant diversity, microbial association networks, and community CUE in an agricultural field context.

Field site and experimental design: Soils were collected in August 2020 from the TwinWin agricultural experiment in southern Finland (60°13′N, 25°07′E), established in 2019 to test ecological intensification via undersowing diverse species with barley (Hordeum vulgare L. var. Harbiney) fertilized at 80 kg N ha−1 yr−1. Barley was sown in 12 cm rows; undersown species were intercropped in 10 cm rows one week later. Diversity treatments included barley monoculture and barley plus 1, 2, 4, or 8 undersown species. Undersown species spanned functional traits (rooting depth, N-fixation) and included Trifolium repens, T. hybridum, Festuca rubra, Medicago sativa, Lupinus multiflorus, Phleum pratense, Restuca andarehalia, Trifolium dubium, Lolium perenne. Plots (40 m × 10 m) were randomized across four blocks; replication: D1 (three plots per undersown species; total 24 plots), D2 (10 plots), D4 (6 plots), D8 (4 plots), plus 4 bare fallows (total 60 plots). For this study, rhizosphere sampling targeted barley monoculture, D1, D4, and D8. Sampling and biomass: Barley rhizosphere soil was collected (n intended = 24 samples per treatment; total n = 168). Spring and summer aboveground biomass were measured annually in 50 × 50 cm quadrats, separating barley from undersown species + weeds; samples dried 48 h at 60 °C and weighed. Barley yield was measured from a central 2 m strip at harvest. Soil processing: Rhizosphere soil adhering to barley roots was collected, sieved (2 mm). Subsamples were dried at 65 °C to determine moisture; others used for microbiological assays and SOM analysis. Microbial CUE (18O-water method): Community CUE was estimated using the H2 18O labeling approach to infer microbial growth from 18O incorporation into DNA, alongside CO2 respiration measurements after 24 h incubations at 22 °C in sealed vessels. For each rhizosphere sample (168), two technical replicate incubations received H2 18O and one control received natural abundance water, yielding 840 DNA extractions in total; technical extracts were pooled prior to quantification. DNA δ18O was measured via pyrolysis-IRMS (relative to VSMOW). Analytical precision for enriched samples was 2.6% (QC n = 9). CUE was calculated following established protocols (Spohn et al.). SOM quantity and quality: Rock-Eval 6 thermal pyrolysis-oxidation characterized SOM. Dried, ground soil (50–70 mg) underwent pyrolysis (220–650 °C) then oxidation (220–850 °C). Total organic carbon (TOC, %) and thermal stability indices (e.g., R-index) were derived. Microbial community profiling: Bacterial and fungal communities were profiled by amplicon sequencing (Novogene). Reads were processed in QIIME2 to 97% OTUs; taxonomy assigned with Greengenes (bacteria) and UNITE (fungi). Data were rarefied to 47,000 reads/sample (bacteria) and 19,000 reads/sample (fungi). Resulting datasets comprised 7,891 bacterial OTUs and 2,437 fungal OTUs. Community structure was analyzed using UNIFRAC NMDS; treatment effects tested with PERMANOVA (Adonis, vegan). Network analysis: Co-occurrence networks were constructed for bacteria and fungi at each diversity level, based on OTUs present in at least 10% of samples and focusing on more ubiquitous OTUs (≥5%). Positive and negative association networks used Spearman correlations with stringent thresholds (rho > 0.6; FDR-corrected P < 0.05). Network topology metrics (degree centrality, eigen centrality) were computed (igraph). To isolate diversity effects, network arithmetic subtracted intersecting correlations between successive diversity treatments. Network parameters were converted to sample-specific values weighted by OTU relative abundances and incorporated into structural equation models. Statistical analyses: Linear mixed models (location in field: block/plot as random effects) assessed relationships of plant diversity with soil C, thermal stability, biomass, growth, respiration, and CUE (transformations applied as needed; outliers flagged at 1.5× IQR). One-way ANOVA with Tukey HSD compared treatments. R version 3.6.3 with packages including vegan, nlme, agricolae, igraph, ggplot2, piecewiseSEM was used. Structural equation modeling (SEM): Piecewise SEM evaluated direct and indirect pathways linking plant diversity, stand plant biomass (cumulative spring/summer biomass 2019–2021), soil properties (composite of pH, C/N, exchangeable cations, CEC), soil carbon quantity, bacterial community composition (NMDS1), positive network connectivity (positive eigen centrality), mass-specific respiration (Respiration/MBC), mass-specific growth (Growth/MBC), and CUE. Model fit assessed by Fisher’s C; standardized path coefficients reported.

- Plant diversity increased total organic carbon (TOC) in rhizosphere soils and was associated with greater plant productivity. A positive relationship between undersown plant diversity and spring biomass (undersown species + weeds) was observed. - Microbial physiology: Along the undersown diversity gradient, mass-specific growth increased significantly with plant diversity, whereas respiration did not change significantly. Consequently, community CUE increased with plant diversity in the barley rhizosphere. - Microbial networks: Bacterial association networks exhibited increased positive connectivity relative to negative connectivity with rising plant diversity. The ratio of degree centrality (positive to negative networks) increased along the diversity gradient, with significant differences among treatments (ANOVA Type III, P < 5 × 10−7; df = 16.3; n = 167; biological replicates per treatment ≈ 24–25). Fungal networks did not show significant responses to plant diversity. - Phylum-level links: Shifts in the relative abundances of bacterial phyla differentially related to network parameters and physiology. Acidobacteria, Gemmatimonadetes, and Pseudomonadota were more strongly linked to increases in positive network connectivity, whereas Actinobacteria, Chloroflexi, and Verrucomicrobia were negatively related to positive network parameters. Changes in phyla were distinctly related to respiration, growth, and CUE; increased Bacilli abundance was negatively associated with CUE. - SEM insights: Plant diversity positively influenced positive microbial network connectivity and bacterial community composition. Positive network connectivity had a positive effect on CUE. Soil properties influenced CUE indirectly via effects on microbial community composition and mass-specific growth. Soil carbon quantity negatively affected positive network connectivity and directly influenced mass-specific respiration; mass-specific growth was affected by soil properties. A substantial fraction of CUE variance remained unexplained, indicating additional unmeasured drivers. - Management relevance: Increasing plant diversity under barley (undersowing) can strengthen beneficial microbial associations in the rhizosphere, enhancing microbial community efficiency and contributing to higher soil carbon retention.

The findings support the hypothesis that plant diversity promotes beneficial (positive) microbial associations in the rhizosphere, which shift microbial physiology toward greater growth efficiency and higher CUE. Enhanced positive network connectivity likely reflects mechanisms such as cross-feeding and facilitation among bacterial taxa, increasing resource-use complementarity and reducing the relative cost of growth. The resulting higher microbial turnover and biomass formation can augment microbial-derived SOM, consistent with observed higher TOC in more diverse treatments and with prior literature linking diversity to increased soil carbon storage. The lack of response in fungal networks, contrasted with bacterial network sensitivity, suggests taxa-specific dynamics: bacteria respond rapidly to root exudates, whereas fungi may be more involved in litter decomposition or require longer-term signals to reconfigure networks. SEM results underscore that biotic interactions—captured by positive network connectivity—mediate the link from plant diversity to microbial efficiency, while soil properties modulate growth rather than directly driving CUE. Overall, the study provides empirical evidence that managing plant diversity in agroecosystems can steer rhizosphere microbial communities toward configurations that favor carbon retention, aligning soil functioning with climate mitigation goals.

This study demonstrates that undersown plant diversity in barley fields enhances positive microbial associations in the rhizosphere, increases microbial growth efficiency, and elevates community CUE, contributing to higher soil organic carbon. Bacterial networks, rather than fungal networks, were primarily responsible for these diversity-driven changes. Structural equation modeling revealed that plant diversity affects CUE both directly and indirectly through shifts in microbial community composition and increased positive network connectivity, with soil properties exerting indirect effects via microbial growth. These insights highlight plant diversity as a practical management lever to enhance soil carbon retention in agricultural systems. Future research should: (1) disentangle the effects of plant diversity from plant/root biomass and trait effects (e.g., rooting depth, N-fixation); (2) evaluate long-term dynamics, particularly fungal responses and SOM persistence; (3) integrate additional mechanistic drivers such as extracellular enzyme activities and dissolved organic carbon; and (4) explore crop breeding for variety mixtures to leverage complementarity, alongside socioeconomic frameworks that facilitate adoption of diversified agroecosystems.

- Causality and disentangling factors: The design does not fully separate plant diversity effects from plant/root biomass and trait influences; thus some observed microbial changes may be directly or indirectly driven by biomass inputs. - Temporal scope: The TwinWin experiment was recently established; fungal network responses and longer-term SOM dynamics may require more time to manifest. - Network inference constraints: Co-occurrence networks infer associations and are sensitive to compositional and environmental confounding; they do not prove direct interactions. - Unexplained variance: SEM left a substantial fraction of CUE variation unexplained, indicating missing variables such as extracellular enzyme activities or dissolved organic carbon. - Treatment coverage: Although diversity levels 0, 1, 4, and 8 were sampled with robust replication, intermediate levels and species identity effects across all combinations were not fully resolved. - Methodological notes: Some text indicates pseudo-replication for network construction and technical pooling in isotope-based CUE assays; one anomalous sample was excluded due to negative 18O-excess. These factors may influence precision but are mitigated by high replication and stringent QC.