An agroecological structure model of compost–soil–plant interactions for sustainable organic farming

The study addresses how thermophile-fermented compost affects crop productivity, quality, and the soil–plant nitrogen cycle in the context of sustainable agriculture. With global food insecurity and environmental constraints linked to synthetic nitrogen and phosphorus inputs, organic approaches that recycle nutrients are needed. Compost efficacy is variable due to heterogeneous raw materials and fermentation conditions and its interaction with native soil microbiota. Prior work suggests thermophilic Bacillaceae can stabilize composting and act as plant growth-promoting bacteria (PGPB). This study uses carrots as a model to examine compost–soil–plant interactions via integrated multi-omics and to test the hypothesis that compost fermented with thermophilic Bacillaceae improves carrot growth and quality by reshaping plant metabolism and soil microbial communities, particularly involving Paenibacillus-mediated nitrogen processes.

Background literature emphasizes environmental limits of excessive mineral fertilizers and the importance of organic nitrogen for crop yield. Reviews establish Bacillus spp. as PGPB for disease suppression and growth promotion. The authors’ prior studies reported thermophilic Bacillaceae-based compost stabilized at >70 °C with marine animal resources, exhibiting antifungal activity, denitrification that reduces plant nitrate accumulation, and benefits across animal models. Compost-derived thermophiles efficiently produce L-lactate under anaerobic conditions. Collectively, these works imply that Bacillaceae-rich composts can modulate microbial communities and host metabolism, motivating investigation into their roles in agroecosystems and nitrogen cycling.

Field design: Two adjacent 1.8 m² plots (~30 cm apart) were established. Carrot (Takii Seeds Phytorich Series, Kyo Kurenai) seeds were sown in Aug 2016; thinning in Oct; harvests in Nov 2016 and Feb 2017. One plot received thermophile-fermented compost (with thermophilic Bacillaceae); the other served as control. Measurements: Plant growth indices (height, root length/diameter, stem+leaf and root fresh weights), color (RGB pixel analysis via image processing in R and Python), and sensory taste (panel in Nov n=4; Feb n=8) were evaluated. Antioxidant activity (DPPH assay) was measured in leaves and roots (n=5). Metabolomics: Targeted analyses assessed carotenoids (α-carotene, β-carotene, lycopene) and broader metabolite profiles in leaves and roots; soil metabolomics employed an instrument appropriate to soil matrix. Microbiome: 16S rRNA V1–V2 sequencing profiled soil bacterial communities; α- and β-diversity assessed (UniFrac PCoA), and taxonomic shifts analyzed at phylum and genus levels. Statistics and modeling: Normality (Shapiro–Wilk) and variance homogeneity (F test) guided parametric (unpaired or Welch t-tests) vs nonparametric (Wilcoxon signed-rank) tests; significance at p<0.05, trends 0.05≤p<0.20. Association analyses identified factors linked to compost. Structural equation modeling (SEM) built multiple regression models for antioxidant activity and metabolite groups; model fit indices (e.g., χ² p, CFI, TLI, RMSEA, SRMR, GFI, AGFI) compared across models. Causal mediation analysis (CMA) and BayesLINGAM were used to probe causal structures among grouped variables. Isolation and functional genomics: Compost-derived colonies were screened for nitrogen fixation genes (nif) leading to isolation of Paenibacillus macerans HMSSN-036 and Paenibacillus sp. HMSSN-139. Whole-genome sequencing and phylogenetics were performed; genomes inspected for PGPB-related functions (nitrogen fixation, auxin biosynthesis/transport, phosphate solubilization, siderophore systems). In vitro assays tested auxin production, siderophore activity, and phosphate solubilization. Soil-function assays: Lab soil microcosms assessed nitrogen fixation using 15N tracer with Arabidopsis thaliana and quantified N2O emission using a setup with fungal stimulation (potato dextrose addition) and ITS-based fungal community check. Field soil chemistry was analyzed post-harvest (e.g., total N, EC, phosphorus availability/activation [PAC], iron). Software: R (v4.0.5), Prism (v9.1.2), Microsoft Office (v16.66.1).

Plant performance and quality: Compost amendment increased carrot growth metrics, with effects more pronounced by Feb 2017. In Nov: height 46.5±3.5 cm vs 43.8±4.4 (p<0.05); root length 19.1±1.8 cm vs 16.4±1.5 (p<0.05). In Feb: height 43.4±6.0 vs 40.5±3.0 (p<0.05); root length 21.4±3.0 vs 18.5±2.0 (p<0.01); root diameter 5.4±0.5 cm vs 4.9±0.5 (p<0.05); stem+leaf weight 29.0±13.0 g vs 18.0±7.0 g (p<0.05); root weight 291.0±97.0 g vs 201.0±67.0 g (p<0.01). Color: Increased red pixel rate in Nov (p=0.012) and increased blue pixel rate in Feb (p=0.031). Taste: Significant differences in flavor attributes (Nov: χ²=23.333, df=3, p=0.0086; Feb: χ²=11.667, df=3, p=3.44×10⁻⁵). Antioxidant activity and metabolites: Root DPPH increased ~40% (p=0.2612); leaf DPPH decreased ~40% (p=0.0014). Root carotenoids: α-carotene (p=0.0103) and lycopene (p=0.0146) increased; β-carotene trended higher (p=0.1188). Leaf metabolites: GABA increased (p=0.0072); methionine sulfoxide and methionine decreased (p=0.0007); flavonoid glycosides decreased: kaempferol-Gal-Rha (p=0.0023), cyanidin 3-O-rutinoside (p=0.0031), apigenin 7-O-neohesperidoside (p=0.0035), quercetin-Glc (p=0.0147). Other leaf changes: indole-3-carboxaldehyde increased (p=0.009); L-2-aminoadipate, malate, 1,3-dimethylurate, methylmalonate, phenyllactate decreased. Root metabolites: tryptophan (p=0.0177), phenylalanine (p=0.0952), tyrosine (p=0.0449), L-2-aminoadipate (p=0.0355) decreased; arginine increased (p=0.0315). Soil omics: Soil metabolomics indicated increased trend in S-methyl-L-cysteine and decreased nicotinamide. Microbiome: β-diversity shifted; Proteobacteria decreased significantly (p=0.0308); Firmicutes, Gemmatimonadetes, Verrucomicrobia trended higher; Planctomycetes, Acidobacteria, Spirochaetes trended lower. At genus level, Marmoricola increased (p=0.0031); Paenibacillus and Geobacillus trended higher (p≈0.067–0.069). SEM and causal analyses: Optimal SEM linked compost to leaf amino acids, flavonoids, and root carotenoids explaining DPPH variation; a leaf–root SEM including leaf flavonoids (apigenin 7-O-neohesperidoside, kaempferol-Gal-Rha, quercetin-Glc), root L-2-aminoadipate and phenylalanine had best fit; soil SEM included Paenibacillus with DL-2-aminoadipate, nicotinamide, and S-methyl-L-cysteine. CMA showed no single significant mediation, suggesting group-level interactions; BayesLINGAM highlighted compost as a frequent causal source among top interactions. Paenibacillus isolation and functions: Two compost-derived nitrogen-fixing strains were isolated (Paenibacillus macerans HMSSN-036; Paenibacillus sp. HMSSN-139) harboring nif genes (e.g., nifH, nifE/N/B/X/U), auxin production genes, phosphate transport/solubilization, and siderophore regulation (Fur); in vitro assays confirmed auxin, siderophore, and phosphate solubilization activities. Soil function assays: In 15N soil–plant microcosms, nitrogen fixation rates ~20% were observed in plant and soil; Arabidopsis growth increased nonsignificantly. Compost markedly reduced N2O emissions in fungal-stimulated soils compared to controls; dominant fungi were Gibberella (66.1%) and Fusarium (28.7%). Field soil chemistry post-harvest showed slightly increased total N, higher EC, increased PAC, and decreased iron, consistent with phosphate solubilization and siderophore effects. Overall, compost improved carrot yield and quality, modulated plant metabolomes, altered soil microbiota, and supported nitrogen cycling with reduced N2O in laboratory tests.

Findings support the hypothesis that thermophile-fermented compost enhances crop performance by restructuring plant and soil biochemical networks. The integrated omics and SEM indicate coordinated shifts in amino acids, flavonoids, and carotenoids under compost, aligning with improved root antioxidant capacity and color/taste traits. Paenibacillus emerges as a key soil mediator in the SEM, consistent with its genomic capacity for nitrogen fixation and PGPB traits (auxin, phosphate solubilization, siderophores). Soil trends—decreased nicotinamide and increased S-methyl-L-cysteine—suggest altered nutrient signaling and potential antipathogenic dynamics; siderophore-mediated iron acquisition may link to carotenoid biosynthesis via heme-dependent P450 enzymes. Although isolated Paenibacillus strains did not exactly match soil amplicon sequences, the observed increase of Paenibacillus in compost-treated soils and known quorum-sensing capacities imply genus-level functional contributions. Laboratory assays demonstrate compost can promote nitrogen fixation and suppress N2O generation in fungal-rich soils, indicating environmental co-benefits. Altogether, compost acts through a complex cascade influencing the nitrogen cycle and plant secondary metabolism, improving yield and quality while potentially reducing greenhouse gas emissions.

Using carrots as a model crop and a thermophile-fermented, Bacillaceae-rich compost, the study integrates multi-omics with structural equation modeling to reveal a systems-level compost–soil–plant interaction network. Compost increased carrot yield and quality, shifted plant metabolite profiles (amino acids, flavonoids, carotenoids), altered soil microbiota (including Paenibacillus), and modulated soil metabolites. Isolation and genomic/functional validation of nitrogen-fixing Paenibacillus strains corroborate SEM predictions and support mechanisms involving nitrogen fixation, phosphate solubilization, and siderophore-mediated iron dynamics. Laboratory soil assays further indicate compost can reduce N2O emissions. These results provide a framework for chemically independent, nitrogen-efficient organic farming. Future work should optimize composting conditions and formulations, verify field-level N2O mitigation, link specific Paenibacillus taxa to in situ functions, and test generality across crops and environments (including aquatic plant systems).

Field plots were limited (two areas, 1.8 m² each) and initial soil was not analyzed prior to cultivation (though plots were mixed by a skilled operator), potentially confounding baseline differences. Several observed effects had marginal p-values or were trends rather than statistically significant. Different analytical platforms for plant and soil metabolomics complicated direct comparison (e.g., L- vs DL-2-aminoadipate). CMA did not identify significant single mediations, indicating complex group interactions that are difficult to causally resolve. The isolated Paenibacillus genomes did not match soil 16S sequences exactly; genus-level inference may mask species-level specificity. N2O suppression and nitrogen fixation were demonstrated in laboratory microcosms, not measured at field scale; Arabidopsis growth promotion was not statistically significant. Some associations (e.g., nicotinamide and S-methyl-L-cysteine roles) remain mechanistic hypotheses requiring targeted validation.