Dataset characterising dominant bacterial phylotypes across animal manure-enriched composting microcosms for crude oil waste sludge bioremediation Authors: Onyedikachi Ubani, Veronica M. Ngole-Jeme Published: 2026 ABSTRACT The dataset provides a complete record of microbial, functional, physicochemical, and contaminant dynamics from controlled co-composting microcosms designed to remediate petroleum refinery sludge using targeted animal manure amendments. Five treatments—cow, pig, horse, and poultry manures, as well as an unamended control—were monitored over 300 days. The study incorporated amplicon-based 16S rRNA gene sequencing, functional gene inference, culture-based validation, bulk chemistry, and chromatographic analyses. Illumina MiSeq profiling of the V1–V3 regions identified 359 bacterial genera (raw OTU-level assignments prior to quality and abundance filtering) across 17 phyla, with taxonomic inventories structured from phylum to genus. Alpha and beta diversity measures demonstrated treatment-dependent community assembly, with the highest richness and diversity observed in cow-manure microcosms—pig and poultry amendments selectively enriched hydrocarbon-degrading taxa, including Pseudomonas. Functional predictions generated using PICRUSt2 (NSTI = 0.02–0.16) indicated enrichment of pathways involved in xenobiotic degradation, aromatic compound metabolism, and benzoate catabolism. These predictions were supported by culture-based evidence, including redox indicator screening and detection of the cbzE gene, which encodes catechol 2,3-dioxygenase, a key enzyme in chlorobenzoate/chlorocatechol degradation pathways and functionally analogous to the widely recognised xylE gene in aromatic hydrocarbon-degrading microorganisms. Collectively, these determinations provide complementary genomic and phenotypic evidence for the biodegradation potential of the microbial community and its capacity to transform aromatic and other environmentally relevant xenobiotic compounds. Additional datasets document total organic carbon, nitrogen, and phosphorus profiles of feedstocks and sludge. At the same time, Soxhlet extraction GC–MS measurements quantify polycyclic aromatic hydrocarbon (PAH) attenuation, with up to 99.9% removal achieved for multiple compounds in pig, horse, and poultry-amended systems. Through the combination of taxonomic profiling, PICRUSt2-based functional inference, chemical transformation analyses, and degradation kinetic measurements, this dataset affords a comprehensive characterisation of microbial community structure, predicted metabolic potential, and biodegradation performance associated with crude oil sludge co-composting at the sampled time point. Specifications TableSubjectBioremediation, Biotechnology, Earth and Environmental Science, MicrobiologySpecific subject areaAmplicon-based microbial ecology and petroleum bioremediation focusing on manure-driven community assembly, PAH catabolism, and co-composting microcosm dynamics.Type of dataTable, Image, Chart, Graph, Figure, Analysed, Filtered, Processed.Data collectionData were collected through an integrated workflow combining molecular, microbiological, and chromatographic approaches. Five co-composting microcosms of crude oil sludge co-amended with cow, swine/pig, horse, and poultry manures plus an unamended control were incubated over 300 days. Bacterial communities were profiled using Illumina MiSeq sequencing (V1–V3 hypervariable regions, 16S rRNA gene); high-quality reads were imported into the Mothur pipeline (v1.40.0) [18] for OTU clustering and were subsequently processed using QIIME2, with taxonomic classification performed against the SILVA reference database. Feedstock physicochemistry was quantified via LECO TruMac CNS analysis for TOC, nitrogen, and phosphorus. PAHs were extracted by automated Soxhlet with dichloromethane and quantified by GC–MS. Culture-dependent validation was performed using PCR-based detection of the cbzE gene and 2,6-DCPIP redox assays. Functional pathway inference utilised PICRUSt2 with NSTI quality filtering.Data source locationInstitution: Department of Environmental Sciences, University of South Africa (UNISA), Corner of Christiaan de Wet Road and Pioneer Avenue, Florida, Roodepoort, South Africa. Experimental workflows, encompassing microcosm construction, molecular assays, amplicon sequencing, bioinformatic processing, and physicochemical characterisation, were executed within affiliated analytical and research facilities supporting microbiological culturing, 16S rRNA gene profiling, and chromatographic quantification.City/Town/Region: Crude oil sludge feedstock was sourced from a petroleum refinery in Durban, KwaZulu-Natal Province. Animal manure amendments(cow, swine, horse, and poultry) were obtained from the University of Pretoria Experimental Farm, Pretoria, Gauteng Province. Reference garden soil was collected from Tembisa, north of Kempton Park, East Rand, Gauteng Province. All microcosm incubation, compositional profiling, and PAH quantification were performed under controlled laboratory conditions at UNISA, Roodepoort, Gauteng Province.Country: South Africa.Latitude and longitude: S 26°1′0″, E 28°14′0″ (Tembisa soil sampling site); S 29°51′0″, E 31°2′0″ (Durban petroleum refinery, crude oil sludge origin); S 25°45′0″, E 28°16′0″ (University of Pretoria Experimental Farm, manure source).The dataset reflects a geographically distributed yet experimentally unified microbial resource anchored in South Africa's petroleum-industrial and agricultural landscapes. By integrating site-specific biological feedstocks, indigenous crude oil sludge microbiota, manure-associated bacterial consortia, and contextualised soil amendments, with advanced amplicon sequencing, functional gene profiling, and chromatographic analytical platforms, this resource strengthens the environmental applicability and reproducibility of co-composting bioremediation frameworks across comparable hydrocarbon-impacted matrices within similar climatic and geological contexts.Data accessibilityRepository name: NCBI Sequence Read Archive (SRA) under the associated BioProjectData identification number: PRJNA794053.Data associated with this study have been deposited at the NCBI database(https://www.ncbi.nlm.nih.gov/Traces/study/?acc=SRP353438): sequence archive (SRA) as BioProject ID PRJNA794053.Instructions for accessing these data: Direct Access.Direct URL to data: https://www.ncbi.nlm.nih.gov/Traces/study/?acc=SRP353438Repository name: www.Figshare.comDOI -10.6084/m9.figshare.3253117510.6084/m9.figshare.32531175Related research articleData were extracted from the first author's doctoral thesis, referenced in this manuscript as *Ubani, O. (2021). Development of an active bacterial formulation for degradation of complex crude oil wastes (Doctoral dissertation, University of South Africa (South Africa) no. February, pp. 1–447. SubjectBioremediation, Biotechnology, Earth and Environmental Science, MicrobiologySpecific subject areaAmplicon-based microbial ecology and petroleum bioremediation focusing on manure-driven community assembly, PAH catabolism, and co-composting microcosm dynamics.Type of dataTable, Image, Chart, Graph, Figure, Analysed, Filtered, Processed.Data collectionData were collected through an integrated workflow combining molecular, microbiological, and chromatographic approaches. Five co-composting microcosms of crude oil sludge co-amended with cow, swine/pig, horse, and poultry manures plus an unamended control were incubated over 300 days. Bacterial communities were profiled using Illumina MiSeq sequencing (V1–V3 hypervariable regions, 16S rRNA gene); high-quality reads were imported into the Mothur pipeline (v1.40.0) [18] for OTU clustering and were subsequently processed using QIIME2, with taxonomic classification performed against the SILVA reference database. Feedstock physicochemistry was quantified via LECO TruMac CNS analysis for TOC, nitrogen, and phosphorus. PAHs were extracted by automated Soxhlet with dichloromethane and quantified by GC–MS. Culture-dependent validation was performed using PCR-based detection of the cbzE gene and 2,6-DCPIP redox assays. Functional pathway inference utilised PICRUSt2 with NSTI quality filtering.Data source locationInstitution: Department of Environmental Sciences, University of South Africa (UNISA), Corner of Christiaan de Wet Road and Pioneer Avenue, Florida, Roodepoort, South Africa. Experimental workflows, encompassing microcosm construction, molecular assays, amplicon sequencing, bioinformatic processing, and physicochemical characterisation, were executed within affiliated analytical and research facilities supporting microbiological culturing, 16S rRNA gene profiling, and chromatographic quantification.City/Town/Region: Crude oil sludge feedstock was sourced from a petroleum refinery in Durban, KwaZulu-Natal Province. Animal manure amendments(cow, swine, horse, and poultry) were obtained from the University of Pretoria Experimental Farm, Pretoria, Gauteng Province. Reference garden soil was collected from Tembisa, north of Kempton Park, East Rand, Gauteng Province. All microcosm incubation, compositional profiling, and PAH quantification were performed under controlled laboratory conditions at UNISA, Roodepoort, Gauteng Province.Country: South Africa.Latitude and longitude: S 26°1′0″, E 28°14′0″ (Tembisa soil sampling site); S 29°51′0″, E 31°2′0″ (Durban petroleum refinery, crude oil sludge origin); S 25°45′0″, E 28°16′0″ (University of Pretoria Experimental Farm, manure source).The dataset reflects a geographically distributed yet experimentally unified microbial resource anchored in South Africa's petroleum-industrial and agricultural landscapes. By integrating site-specific biological feedstocks, indigenous crude oil sludge microbiota, manure-associated bacterial consortia, and contextualised soil amendments, with advanced amplicon sequencing, functional gene profiling, and chromatographic analytical platforms, this resource strengthens the environmental applicability and reproducibility of co-composting bioremediation frameworks across comparable hydrocarbon-impacted matrices within similar climatic and geological contexts.Data accessibilityRepository name: NCBI Sequence Read Archive (SRA) under the associated BioProjectData identification number: PRJNA794053.Data associated with this study have been deposited at the NCBI database(https://www.ncbi.nlm.nih.gov/Traces/study/?acc=SRP353438): sequence archive (SRA) as BioProject ID PRJNA794053.Instructions for accessing these data: Direct Access.Direct URL to data: https://www.ncbi.nlm.nih.gov/Traces/study/?acc=SRP353438Repository name: www.Figshare.comDOI -10.6084/m9.figshare.3253117510.6084/m9.figshare.32531175Related research articleData were extracted from the first author's doctoral thesis, referenced in this manuscript as *Ubani, O. (2021). Development of an active bacterial formulation for degradation of complex crude oil wastes (Doctoral dissertation, University of South Africa (South Africa) no. February, pp. 1–447. Value of the Data •This dataset represents an integrated, multi-layered resource that combines amplicon-based community profiling, physicochemical characterisation, functional gene detection, and long-term PAH attenuation monitoring within a unified co-composting framework. It includes taxonomic inventories spanning 17 phyla and 359 bacterial genera (raw OTU-level assignments before quality and abundance filtering), as well as alpha and beta diversity indices, PICRUSt2-predicted KEGG pathway abundances, and culture-validated catechol 2,3-dioxygenase (cbzE) gene records. In contrast to datasets restricted to a single timepoint, this resource documents longitudinal microbial community trajectories throughout a 300-day incubation period across five manure-amendment treatments, thereby preserving the temporal and componential relationships between microbial assembly and hydrocarbon attenuation.•The dataset is suitable for reuse by researchers in microbial ecology, environmental biotechnology, and bioinformatics for comparative modelling of organic-amendment-driven community assembly, benchmarking PAH biodegradation kinetics under co-composting regimes, and validating functional inference tools against culture-confirmed genetic markers. Treatment-resolved alpha diversity indices, beta diversity ordinations, and compound-specific removal metrics deliver structured reference baselines for evaluating hydrocarbon-degrading consortia. Furthermore, the dataset supports metabolic pathway reconstruction, ecological null-model testing, and calibration of predictive microbiome frameworks for organic-amendment and petroleum-contaminated matrix conditions.•Integration of manure feedstock physicochemistry, total organic carbon, nitrogen, and phosphorus profiles with shifts in community composition and substrate transformation rates establishes this dataset as a reproducible reference for nutrient-mediated biostimulation strategies. Quantitative GC–MS-derived PAH residual measurements document compound-class-dependent attenuation trajectories spanning high, medium, and low-molecular-weight aromatic fractions, providing a functional perspective on amendment-specific hydrocarbon catabolism. The combined chemical and ecological data extend analytical detail beyond datasets that address microbial diversity or chemical endpoints in isolation, offering co-resolved biological and physicochemical descriptors relevant to co-composting system optimisation.•The dataset allows benchmarking of petroleum bioremediation approaches by providing manure-treatment-specific PAH removal metrics, with attenuation rates reaching up to 99.9%, molecular-weight-dependent transformation patterns, and statistically validated diversity performance descriptors. Documented community richness trade-offs include a Chao1 value of 422.60 and a Shannon index of 3.726 in cow manure microcosms, as well as selective hydrocarbonoclastic enrichment profiles in pig and poultry treatments. The multi-treatment structure and validated performance indicators allow systematic cross-amendment evaluation of petroleum-derived contaminant profiles, without restricting analysis to a single variable or endpoint.•The dataset offers a transferable methodological model for compost-based bioremediation technology development, encompassing standardised microcosm construction protocols, manure amendment ratios, sequencing pipeline configurations, and PAH extraction and quantification procedures. Complete records of inoculum sources, incubation parameters, and sampling intervals are provided, together with bioinformatic processing steps from QIIME2 quality filtering and SILVA taxonomic assignment to PICRUSt2 NSTI-controlled functional prediction, ensuring full reproducibility across experimental and real-world contexts. This well-structured methodology supports adaptation to different organic amendments, petroleum waste matrices, and climatic conditions in both laboratory and field-oriented remediation workflows.•The dataset is context-specific, focusing on petroleum refinery sludge from a single South African industrial source, four animal manure types, and controlled laboratory microcosm conditions. These factors may limit direct transferability to other sludge compositions, climatic regimes, or field-scale operating environments. Certain functional predictions are computationally inferred rather than metagenomically validated, and large-scale pilot or field confirmation is beyond the dataset's scope. Nevertheless, the dataset is still valuable for comparative taxonomic analysis, bioinformatic tool validation, amendment optimisation, and co-composting system design in analogous hydrocarbon-impacted matrices.•All processed data products are hosted on Figshare. •This dataset represents an integrated, multi-layered resource that combines amplicon-based community profiling, physicochemical characterisation, functional gene detection, and long-term PAH attenuation monitoring within a unified co-composting framework. It includes taxonomic inventories spanning 17 phyla and 359 bacterial genera (raw OTU-level assignments before quality and abundance filtering), as well as alpha and beta diversity indices, PICRUSt2-predicted KEGG pathway abundances, and culture-validated catechol 2,3-dioxygenase (cbzE) gene records. In contrast to datasets restricted to a single timepoint, this resource documents longitudinal microbial community trajectories throughout a 300-day incubation period across five manure-amendment treatments, thereby preserving the temporal and componential relationships between microbial assembly and hydrocarbon attenuation. This dataset represents an integrated, multi-layered resource that combines amplicon-based community profiling, physicochemical characterisation, functional gene detection, and long-term PAH attenuation monitoring within a unified co-composting framework. It includes taxonomic inventories spanning 17 phyla and 359 bacterial genera (raw OTU-level assignments before quality and abundance filtering), as well as alpha and beta diversity indices, PICRUSt2-predicted KEGG pathway abundances, and culture-validated catechol 2,3-dioxygenase (cbzE) gene records. In contrast to datasets restricted to a single timepoint, this resource documents longitudinal microbial community trajectories throughout a 300-day incubation period across five manure-amendment treatments, thereby preserving the temporal and componential relationships between microbial assembly and hydrocarbon attenuation. •The dataset is suitable for reuse by researchers in microbial ecology, environmental biotechnology, and bioinformatics for comparative modelling of organic-amendment-driven community assembly, benchmarking PAH biodegradation kinetics under co-composting regimes, and validating functional inference tools against culture-confirmed genetic markers. Treatment-resolved alpha diversity indices, beta diversity ordinations, and compound-specific removal metrics deliver structured reference baselines for evaluating hydrocarbon-degrading consortia. Furthermore, the dataset supports metabolic pathway reconstruction, ecological null-model testing, and calibration of predictive microbiome frameworks for organic-amendment and petroleum-contaminated matrix conditions. The dataset is suitable for reuse by researchers in microbial ecology, environmental biotechnology, and bioinformatics for comparative modelling of organic-amendment-driven community assembly, benchmarking PAH biodegradation kinetics under co-composting regimes, and validating functional inference tools against culture-confirmed genetic markers. Treatment-resolved alpha diversity indices, beta diversity ordinations, and compound-specific removal metrics deliver structured reference baselines for evaluating hydrocarbon-degrading consortia. Furthermore, the dataset supports metabolic pathway reconstruction, ecological null-model testing, and calibration of predictive microbiome frameworks for organic-amendment and petroleum-contaminated matrix conditions. •Integration of manure feedstock physicochemistry, total organic carbon, nitrogen, and phosphorus profiles with shifts in community composition and substrate transformation rates establishes this dataset as a reproducible reference for nutrient-mediated biostimulation strategies. Quantitative GC–MS-derived PAH residual measurements document compound-class-dependent attenuation trajectories spanning high, medium, and low-molecular-weight aromatic fractions, providing a functional perspective on amendment-specific hydrocarbon catabolism. The combined chemical and ecological data extend analytical detail beyond datasets that address microbial diversity or chemical endpoints in isolation, offering co-resolved biological and physicochemical descriptors relevant to co-composting system optimisation. Integration of manure feedstock physicochemistry, total organic carbon, nitrogen, and phosphorus profiles with shifts in community composition and substrate transformation rates establishes this dataset as a reproducible reference for nutrient-mediated biostimulation strategies. Quantitative GC–MS-derived PAH residual measurements document compound-class-dependent attenuation trajectories spanning high, medium, and low-molecular-weight aromatic fractions, providing a functional perspective on amendment-specific hydrocarbon catabolism. The combined chemical and ecological data extend analytical detail beyond datasets that address microbial diversity or chemical endpoints in isolation, offering co-resolved biological and physicochemical descriptors relevant to co-composting system optimisation. •The dataset allows benchmarking of petroleum bioremediation approaches by providing manure-treatment-specific PAH removal metrics, with attenuation rates reaching up to 99.9%, molecular-weight-dependent transformation patterns, and statistically validated diversity performance descriptors. Documented community richness trade-offs include a Chao1 value of 422.60 and a Shannon index of 3.726 in cow manure microcosms, as well as selective hydrocarbonoclastic enrichment profiles in pig and poultry treatments. The multi-treatment structure and validated performance indicators allow systematic cross-amendment evaluation of petroleum-derived contaminant profiles, without restricting analysis to a single variable or endpoint. The dataset allows benchmarking of petroleum bioremediation approaches by providing manure-treatment-specific PAH removal metrics, with attenuation rates reaching up to 99.9%, molecular-weight-dependent transformation patterns, and statistically validated diversity performance descriptors. Documented community richness trade-offs include a Chao1 value of 422.60 and a Shannon index of 3.726 in cow manure microcosms, as well as selective hydrocarbonoclastic enrichment profiles in pig and poultry treatments. The multi-treatment structure and validated performance indicators allow systematic cross-amendment evaluation of petroleum-derived contaminant profiles, without restricting analysis to a single variable or endpoint. •The dataset offers a transferable methodological model for compost-based bioremediation technology development, encompassing standardised microcosm construction protocols, manure amendment ratios, sequencing pipeline configurations, and PAH extraction and quantification procedures. Complete records of inoculum sources, incubation parameters, and sampling intervals are provided, together with bioinformatic processing steps from QIIME2 quality filtering and SILVA taxonomic assignment to PICRUSt2 NSTI-controlled functional prediction, ensuring full reproducibility across experimental and real-world contexts. This well-structured methodology supports adaptation to different organic amendments, petroleum waste matrices, and climatic conditions in both laboratory and field-oriented remediation workflows. The dataset offers a transferable methodological model for compost-based bioremediation technology development, encompassing standardised microcosm construction protocols, manure amendment ratios, sequencing pipeline configurations, and PAH extraction and quantification procedures. Complete records of inoculum sources, incubation parameters, and sampling intervals are provided, together with bioinformatic processing steps from QIIME2 quality filtering and SILVA taxonomic assignment to PICRUSt2 NSTI-controlled functional prediction, ensuring full reproducibility across experimental and real-world contexts. This well-structured methodology supports adaptation to different organic amendments, petroleum waste matrices, and climatic conditions in both laboratory and field-oriented remediation workflows. •The dataset is context-specific, focusing on petroleum refinery sludge from a single South African industrial source, four animal manure types, and controlled laboratory microcosm conditions. These factors may limit direct transferability to other sludge compositions, climatic regimes, or field-scale operating environments. Certain functional predictions are computationally inferred rather than metagenomically validated, and large-scale pilot or field confirmation is beyond the dataset's scope. Nevertheless, the dataset is still valuable for comparative taxonomic analysis, bioinformatic tool validation, amendment optimisation, and co-composting system design in analogous hydrocarbon-impacted matrices. The dataset is context-specific, focusing on petroleum refinery sludge from a single South African industrial source, four animal manure types, and controlled laboratory microcosm conditions. These factors may limit direct transferability to other sludge compositions, climatic regimes, or field-scale operating environments. Certain functional predictions are computationally inferred rather than metagenomically validated, and large-scale pilot or field confirmation is beyond the dataset's scope. Nevertheless, the dataset is still valuable for comparative taxonomic analysis, bioinformatic tool validation, amendment optimisation, and co-composting system design in analogous hydrocarbon-impacted matrices. •All processed data products are hosted on Figshare. All processed data products are hosted on Figshare. Background The persistence of crude oil refinery waste sludge is a critical environmental liability due to its recalcitrant composition of high-molecular-weight polycyclic aromatic hydrocarbons (PAHs), asphaltenes, heavy metals, and hydrophobic organic fractions that resist abiotic degradation and compromise resident biota through mutagenic and carcinogenic mechanisms [[1], [2], [3], [4], [5]]. Conventional physicochemical interventions, including incineration, chemical oxidation, and stabilisation, are cost-prohibitive, generate secondary pollutants, and remain ecologically unsustainable, particularly in resource-constrained settings [[6], [7]]. This dataset was generated to address these constraints within a structured framework grounded in microbial community ecology and compost-based bioremediation. It adopts the established theoretical premise that organic amendments [8], particularly animal manures, drive nutrient-mediated biostimulation, selectively enrich functionally specialised hydrocarbonoclastic taxa, and modulate bacterial community assembly trajectories within petroleum-contaminated matrices [[9], [10], [11], [12]]. Guided by this framework, co-composting microcosms amended with cow, swine, horse, and poultry manures were constructed and maintained over a 300-day incubation period to generate longitudinal datasets on adapted community and substrate transformation under controlled conditions. The methodological design integrates culture-independent amplicon sequencing of the 16S rRNA gene (Illumina MiSeq; V1–V3 hypervariable regions) with culture-dependent functional screening approaches to resolve both taxonomic and functional dimensions of bacterial community assembly [13,14]. Bioinformatic processing through QIIME2 with SILVA-based taxonomic assignment, UCHIME chimera filtering, and mothur-supported OTU clustering provides a reproducible analytical framework for characterising community structure across treatments [[15], [16], [17], [18], [19], [20]]. Functional inference via PICRUSt2 extends the taxonomic inventory toward metabolic pathway prediction, while PCR-based detection of the catechol 2,3-dioxygenase (cbzE) gene, 2,6-DCPIP redox screening, and GC–MS-based PAH quantification following automated Soxhlet extraction provide culture-validated functional and chemical layers that anchor bioinformatic predictions to empirically confirmed biodegradation activity [1,[21], [22], [23], [24]]. Physicochemical characterisation of manure feedstocks, encompassing total organic carbon, nitrogen, and phosphorus, was conducted using LECO TruMac CNS analysis to document the nutrient substrates driving manure-specific community selection [[25], [26], [27], [28], [29], [30]]. This data article complements the associated research manuscript by providing structured, transparent access to the underlying amplicon sequencing datasets, alpha and beta diversity matrices, functional prediction outputs, culture-validated genetic records, and compound-specific PAH attenuation profiles across all five treatments. It enables independent verification of the methodological workflow and supports reuse in comparative community ecology modelling, co-composting protocol optimisation, functional inference tool calibration, and multi-scale bioremediation framework development across microbiology, environmental biotechnology, and petrogenic waste management research contexts [22,[31], [32], [33], [34], [35], [36], [37]]. Data Description The dataset is deposited in the NCBI Sequence Read Archive and GenBank under BioProject accession PRJNA794053, comprising raw paired-end Illumina MiSeq amplicon FASTQ files for five co-composting treatments: poultry manure (PM), horse manure (HM), swine/pig manure (SPM), cow manure (COW), and an unamended control (CT, crude oil sludge with soil and wood chips, no manure addition). Sequencing targeted the V1–V3 hypervariable region of the 16S rRNA gene using primers 27F/518R on a MiSeq platform with 300 bp paired-end reads and v3 chemistry. Operational taxonomic units were defined at 97% sequence similarity; high-quality reads were imported into the Mothur pipeline (v1.40.0) [18] for OTU clustering and were subsequently processed using QIIME2, with taxonomic classification performed against the SILVA reference database released 128 and 132. Table 1 reports fourteen sequencing and alpha diversity variables per treatment: total reads, observed OTUs, Chao1, Shannon entropy, Simpson diversity, dominance, evenness, Pielou's equitability, Brillouin index, Menhinick and Margalef richness, Fisher's α, Berger–Parker dominance, and Good's coverage computed as single treatment-level values after rarefaction-aware normalisation.Table 1Summary of sequencing output and ecological diversity indices for bacterial communities in crude oil sludge co-compost microcosms amended with different animal manures.Table 1 dummy alt textDiversity MetricAnimal Manure-Amended Co-Compost TreatmentsHMaSPMbCOWcPMdCTeTotal reads26,88425,97215,37747,97540,129Observed OTUs212120422309105Chao1 richness212.00120.40422.60309.00105.00Shannon (H′)2.8141.6983.7264.0653.388Simpson (1–D)0.8790.5830.9040.9540.933Dominance (D)0.1210.4170.0960.0460.067Evenness (eH/S)0.0790.0460.0980.1890.282Equitability (J)0.5250.3550.6160.7090.728Brillouin2.7981.6883.6754.0483.380Menhinick1.2930.7453.4031.4110.524Margalef20.69011.71043.67028.5809.811Fisher α31.39016.27080.21044.20013.080Berger–Parker0.2380.6310.2780.1440.189Goods coverage (%)99.798.599.298.699.0Index: Shannon (H′) = Shannon–Weaver entropy index quantifying community diversity; Simpson (1–D) = probability that two randomly drawn individuals belong to different OTUs; Dominance (D) = complement of Simpson (1–D), reflecting community unevenness; Evenness (eH/S) = exponential Shannon divided by richness; Equitability (J) = Pielou’s evenness index (H′/ln S); Brillouin = diversity index appropriate for fully censused collections; Menhinick = richness estimator (S/√N); Margalef = richness estimator ((S−1)/ln N); Fisher α = log-series alpha sensitive to rare taxa; Berger–Parker = proportional abundance of the single most dominant taxon; Chao1 = non-parametric richness estimator accounting for unseen species. OTU = operational taxonomic unit (97% similarity); PAH = polycyclic aromatic hydrocarbon.^aHM = horse manure-amended compost treatment;^bSPM = swine/pig manure-amended compost treatment;^cCOW = cow/dairy manure-amended compost treatment;^dPM = poultry/layer manure-amended compost treatment;^eCT = unamended control (crude oil sludge + soil + wood chips, no manure addition). Table 1Summary of sequencing output and ecological diversity indices for bacterial communities in crude oil sludge co-compost microcosms amended with different animal manures.Table 1 dummy alt textDiversity MetricAnimal Manure-Amended Co-Compost TreatmentsHMaSPMbCOWcPMdCTeTotal reads26,88425,97215,37747,97540,129Observed OTUs212120422309105Chao1 richness212.00120.40422.60309.00105.00Shannon (H′)2.8141.6983.7264.0653.388Simpson (1–D)0.8790.5830.9040.9540.933Dominance (D)0.1210.4170.0960.0460.067Evenness (eH/S)0.0790.0460.0980.1890.282Equitability (J)0.5250.3550.6160.7090.728Brillouin2.7981.6883.6754.0483.380Menhinick1.2930.7453.4031.4110.524Margalef20.69011.71043.67028.5809.811Fisher α31.39016.27080.21044.20013.080Berger–Parker0.2380.6310.2780.1440.189Goods coverage (%)99.798.599.298.699.0Index: Shannon (H′) = Shannon–Weaver entropy index quantifying community diversity; Simpson (1–D) = probability that two randomly drawn individuals belong to different OTUs; Dominance (D) = complement of Simpson (1–D), reflecting community unevenness; Evenness (eH/S) = exponential Shannon divided by richness; Equitability (J) = Pielou’s evenness index (H′/ln S); Brillouin = diversity index appropriate for fully censused collections; Menhinick = richness estimator (S/√N); Margalef = richness estimator ((S−1)/ln N); Fisher α = log-series alpha sensitive to rare taxa; Berger–Parker = proportional abundance of the single most dominant taxon; Chao1 = non-parametric richness estimator accounting for unseen species. OTU = operational taxonomic unit (97% similarity); PAH = polycyclic aromatic hydrocarbon.^aHM = horse manure-amended compost treatment;^bSPM = swine/pig manure-amended compost treatment;^cCOW = cow/dairy manure-amended compost treatment;^dPM = poultry/layer manure-amended compost treatment;^eCT = unamended control (crude oil sludge + soil + wood chips, no manure addition). Summary of sequencing output and ecological diversity indices for bacterial communities in crude oil sludge co-compost microcosms amended with different animal manures. Index: Shannon (H′) = Shannon–Weaver entropy index quantifying community diversity; Simpson (1–D) = probability that two randomly drawn individuals belong to different OTUs; Dominance (D) = complement of Simpson (1–D), reflecting community unevenness; Evenness (eH/S) = exponential Shannon divided by richness; Equitability (J) = Pielou’s evenness index (H′/ln S); Brillouin = diversity index appropriate for fully censused collections; Menhinick = richness estimator (S/√N); Margalef = richness estimator ((S−1)/ln N); Fisher α = log-series alpha sensitive to rare taxa; Berger–Parker = proportional abundance of the single most dominant taxon; Chao1 = non-parametric richness estimator accounting for unseen species. OTU = operational taxonomic unit (97% similarity); PAH = polycyclic aromatic hydrocarbon. HM = horse manure-amended compost treatment; SPM = swine/pig manure-amended compost treatment; COW = cow/dairy manure-amended compost treatment; PM = poultry/layer manure-amended compost treatment; CT = unamended control (crude oil sludge + soil + wood chips, no manure addition). Fig. 1 renders the thirteen diversity indices from Table 1 as treatment-grouped bar panels, allowing simultaneous visual comparison of richness, evenness, dominance, and rare-taxon sensitivity across treatments. Fig. 2 plots rarefaction curves for all five treatments (observed OTUs against sequencing depth at 97% similarity), with endpoint markers and per-curve Good's coverage annotations defined in the caption.Fig. 1Alpha diversity indices across the co-compost treatments.Fig 1 dummy alt textFig. 2Rarefaction curves indicating the observed number of operational taxonomic units (OTUs) within the 16S rRNA gene sequences of crude oil sludge samples composted using different animal manures. OTUs are defined at the 3% genetic distance threshold (97% sequence similarity). Treatments: PM = poultry manure; HM = horse manure; COW = cow/dairy manure; SM = swine/pig manure; CT = unamended control. Endpoint markers (●, ■, ▲, ◆, ▼) indicate final observed OTU counts at total sequencing depth. Curve plateauing confirms sufficient sampling coverage across all treatments (Good's coverage > 98%).Fig 2 dummy alt text Alpha diversity indices across the co-compost treatments. Rarefaction curves indicating the observed number of operational taxonomic units (OTUs) within the 16S rRNA gene sequences of crude oil sludge samples composted using different animal manures. OTUs are defined at the 3% genetic distance threshold (97% sequence similarity). Treatments: PM = poultry manure; HM = horse manure; COW = cow/dairy manure; SM = swine/pig manure; CT = unamended control. Endpoint markers (●, ■, ▲, ◆, ▼) indicate final observed OTU counts at total sequencing depth. Curve plateauing confirms sufficient sampling coverage across all treatments (Good's coverage > 98%). Fig. 3 presents community structure at the phylum and class level across three panels: (a) stacked relative-abundance bar charts for the top 17 phyla by treatment; (b) class-level resolution of Proteobacteria into α-, β-, and γ-lineages per treatment; and (c) order-level resolution of Firmicutes into Bacillales and Clostridiales proportions per treatment. Fig. 4 visualises cross-treatment phylum connectivity as a Circos chord diagram, with treatment arcs (right semicircle, equal arc length) linked to phylum arcs (left semicircle, arc length proportional to cumulative relative abundance) via ribbons scaled to phylum abundance per treatment.Fig. 3Comparative Taxonomic Distribution of Bacterial Communities Across Co-Compost Treatments with poultry (PM), horse (HM), cow (CM), and swine (SM) manures and control samples (CT): Multi-Level Profiling Reveals Manure-Specific Community Architectures. Diversity at the phylum level (a) and Proteobacteria into α-, β-, and γ-class level lineages is shown in (b). The Firmicutes dominant orders in the phylum (c).Fig 3 dummy alt textFig. 4Circos chord diagram illustrating the phylogenetic connectivity between the crude oil sludge co-compost treatments and the bacterial phyla detected across the combined 16S rRNA gene amplicon dataset. The right semicircle displays the composting treatments (PM, Poultry Manure; HM, Horse Manure; SPM, Swine/Pig Manure; CM, Cow Manure; CT, Control), each allocated equal arc length (20.0%) reflecting the balanced experimental design. The left semicircle displays bacterial phyla, with arc length proportional to cumulative relative abundance across all treatments: Proteobacteria (74.3%), Firmicutes (11.6%), Actinobacteria (11.0%), Bacteroidetes (2.1%), Acidobacteria (0.4%), TM6/Dependentiae (0.3%), and other minor phyla (0.3%). Chord ribbons connect each treatment to the phyla detected in that microcosm, with ribbon width scaled proportionally to relative abundance (%). The asymmetric distribution of Proteobacteria ribbons is widest for HM and PM and narrowest for CT, reflecting the phylum's differential relative abundance across treatments.Fig 4 dummy alt text Comparative Taxonomic Distribution of Bacterial Communities Across Co-Compost Treatments with poultry (PM), horse (HM), cow (CM), and swine (SM) manures and control samples (CT): Multi-Level Profiling Reveals Manure-Specific Community Architectures. Diversity at the phylum level (a) and Proteobacteria into α-, β-, and γ-class level lineages is shown in (b). The Firmicutes dominant orders in the phylum (c). Circos chord diagram illustrating the phylogenetic connectivity between the crude oil sludge co-compost treatments and the bacterial phyla detected across the combined 16S rRNA gene amplicon dataset. The right semicircle displays the composting treatments (PM, Poultry Manure; HM, Horse Manure; SPM, Swine/Pig Manure; CM, Cow Manure; CT, Control), each allocated equal arc length (20.0%) reflecting the balanced experimental design. The left semicircle displays bacterial phyla, with arc length proportional to cumulative relative abundance across all treatments: Proteobacteria (74.3%), Firmicutes (11.6%), Actinobacteria (11.0%), Bacteroidetes (2.1%), Acidobacteria (0.4%), TM6/Dependentiae (0.3%), and other minor phyla (0.3%). Chord ribbons connect each treatment to the phyla detected in that microcosm, with ribbon width scaled proportionally to relative abundance (%). The asymmetric distribution of Proteobacteria ribbons is widest for HM and PM and narrowest for CT, reflecting the phylum's differential relative abundance across treatments. Genus-level structure and core microbiome overlap appear in Fig. 5. Panel (a) is a heatmap of log₁₀-transformed relative abundance for the top 40 genera by read count, colour-scaled from low to high abundance. Panel (b) is a five-treatment Venn diagram of shared and treatment-unique genera; panel (c) restricts this comparison to the four manure-amended treatments, resolving a shared manure-core set alongside pairwise exclusive overlaps. Exact genus counts and overlap totals are reported directly on each panel.Fig. 5Relative taxonomic distribution of genera and common microbiome diversity in crude oil sludge co-compost treatments. (a) Heatmap of normalised abundance at the genus level for bacteria across five treatment metagenomes: PM (poultry), HM (horse), SPM (swine/pig), CM (cow), and CT (control). Colour scale represents log₁₀-transformed relative abundance (%), with red indicating higher and blue indicating lower abundance. Values >3% are annotated directly on cells. The top 40 genera ranked by total read count are displayed. (b) Venn diagram showing shared and unique genera across all five treatments. A total of 316 genera were detected in the curated Venn-diagram/heatmap, with 13 shared across all treatments and 71 unique to the control. (c) Venn diagram showing shared and unique genera within the four manure-amended treatments only. A core of 14 genera was shared across all manure treatments, with pairwise overlaps indicating treatment-specific recruitment patterns.Fig 5 dummy alt text Relative taxonomic distribution of genera and common microbiome diversity in crude oil sludge co-compost treatments. (a) Heatmap of normalised abundance at the genus level for bacteria across five treatment metagenomes: PM (poultry), HM (horse), SPM (swine/pig), CM (cow), and CT (control). Colour scale represents log₁₀-transformed relative abundance (%), with red indicating higher and blue indicating lower abundance. Values >3% are annotated directly on cells. The top 40 genera ranked by total read count are displayed. (b) Venn diagram showing shared and unique genera across all five treatments. A total of 316 genera were detected in the curated Venn-diagram/heatmap, with 13 shared across all treatments and 71 unique to the control. (c) Venn diagram showing shared and unique genera within the four manure-amended treatments only. A core of 14 genera was shared across all manure treatments, with pairwise overlaps indicating treatment-specific recruitment patterns. Beta diversity is represented across three complementary ordinations. Fig. 6 presents the full pairwise Bray–Curtis dissimilarity matrix as a 5×5 heatmap, with the diagonal masked and the dissimilarity scale and treatment codes defined in the caption. Fig. 7 shows a non-metric multidimensional scaling ordination based on Bray–Curtis dissimilarity, with each treatment plotted as a single point and cluster groupings labelled directly on the plot. Fig. 8 shows a principal component analysis biplot of standardised OTU relative abundances, with variance explained by each axis annotated in the caption and treatments labelled by point.Fig. 6Bray–Curtis dissimilarity matrix of bacterial communities across five treatments. Pairwise dissimilarity values range from 0.589 (Horse Manure vs Poultry Manure) to 0.982 (Cow Manure vs Horse Manure/Swine/Pig Manure). Colour intensity reflects dissimilarity magnitude; diagonal values (self-comparisons) are masked. CM, Cow Manure; CT, Control Treatment; HM, Horse Manure; PM, Poultry Manure; SM, Swine/Pig Manure.Fig 6 dummy alt textFig. 7Non-metric multidimensional scaling (NMDS) ordination of bacterial communities based on Bray–Curtis dissimilarity. Stress = 0.2334. Each point represents a single co-composting treatment. Spatial separation indicates compositional dissimilarity, with Cow Manure maximally separated from all other treatments.Fig 7 dummy alt textFig. 8Principal component analysis (PCA) of standardised OTU relative abundances. PC1 (33.6% variance explained) separates Cow Manure from Control along the primary compositional gradient; PC2 (28.2%) differentiates manure-amended treatments. Collectively, the first two components explain 61.8% of the variance in the community.Fig 8 dummy alt text Bray–Curtis dissimilarity matrix of bacterial communities across five treatments. Pairwise dissimilarity values range from 0.589 (Horse Manure vs Poultry Manure) to 0.982 (Cow Manure vs Horse Manure/Swine/Pig Manure). Colour intensity reflects dissimilarity magnitude; diagonal values (self-comparisons) are masked. CM, Cow Manure; CT, Control Treatment; HM, Horse Manure; PM, Poultry Manure; SM, Swine/Pig Manure. Non-metric multidimensional scaling (NMDS) ordination of bacterial communities based on Bray–Curtis dissimilarity. Stress = 0.2334. Each point represents a single co-composting treatment. Spatial separation indicates compositional dissimilarity, with Cow Manure maximally separated from all other treatments. Principal component analysis (PCA) of standardised OTU relative abundances. PC1 (33.6% variance explained) separates Cow Manure from Control along the primary compositional gradient; PC2 (28.2%) differentiates manure-amended treatments. Collectively, the first two components explain 61.8% of the variance in the community. Table 2, Table 3 report the corresponding multivariate significance tests. Table 2 gives PERMANOVA results (Bray–Curtis dissimilarity, 9,999 permutations) for the overall treatment effect in standard ANOVA format, alongside the global ANOSIM and PERMDISP statistics. Table 3 gives pairwise PERMANOVA results for all ten treatment comparisons, ranked by dissimilarity, with pseudo-F, R², p-value, and FDR-adjusted q-value reported per pair.Table 2PERMANOVA results for bacterial community composition across co-compost treatmentsTable 2 dummy alt textSourcedfSum of SquaresMean SquaresPseudo-FR²p-valueTreatment43.6360.90924.7510.9080.0001***Residual100.3670.037—0.092—Total144.003——1.000—PERMANOVA was conducted using Bray–Curtis dissimilarity with 9,999 unrestricted permutations. Significance: ***p < 0.001.Table 3Pairwise PERMANOVA and ANOSIM results.Table 3 dummy alt textComparisonBray–CurtisPseudo-FR²p-valueFDR q-valuePM vs HM0.37919.670.8310.1020.104PM vs SPM0.67214.870.7880.1010.104SPM vs CM0.68710.840.7300.1040.104HM vs SPM0.78620.820.8390.0910.104CT vs SPM0.79118.740.8240.1040.104PM vs CM0.80623.020.8520.1030.104HM vs CM0.81226.330.8680.1030.104CT vs CM0.93627.410.8730.0990.104PM vs CT0.95061.770.9390.1020.104HM vs CT0.97580.740.9530.0960.104Global ANOSIM: R = 0.959, p = 0.0001. PERMDISP: F₄,₁₀ = 154.56, p < 0.001.Note: Pairwise p-values approach but exceed 0.05 due to limited permutation space with n = 3 per group (only 10 unique permutations possible); however, the global PERMANOVA and ANOSIM are highly significant (p = 0.0001), and the high pseudo-F values confirm large effect sizes. Table 2PERMANOVA results for bacterial community composition across co-compost treatmentsTable 2 dummy alt textSourcedfSum of SquaresMean SquaresPseudo-FR²p-valueTreatment43.6360.90924.7510.9080.0001***Residual100.3670.037—0.092—Total144.003——1.000—PERMANOVA was conducted using Bray–Curtis dissimilarity with 9,999 unrestricted permutations. Significance: ***p < 0.001. PERMANOVA results for bacterial community composition across co-compost treatments PERMANOVA was conducted using Bray–Curtis dissimilarity with 9,999 unrestricted permutations. Significance: ***p < 0.001. Table 3Pairwise PERMANOVA and ANOSIM results.Table 3 dummy alt textComparisonBray–CurtisPseudo-FR²p-valueFDR q-valuePM vs HM0.37919.670.8310.1020.104PM vs SPM0.67214.870.7880.1010.104SPM vs CM0.68710.840.7300.1040.104HM vs SPM0.78620.820.8390.0910.104CT vs SPM0.79118.740.8240.1040.104PM vs CM0.80623.020.8520.1030.104HM vs CM0.81226.330.8680.1030.104CT vs CM0.93627.410.8730.0990.104PM vs CT0.95061.770.9390.1020.104HM vs CT0.97580.740.9530.0960.104Global ANOSIM: R = 0.959, p = 0.0001. PERMDISP: F₄,₁₀ = 154.56, p < 0.001.Note: Pairwise p-values approach but exceed 0.05 due to limited permutation space with n = 3 per group (only 10 unique permutations possible); however, the global PERMANOVA and ANOSIM are highly significant (p = 0.0001), and the high pseudo-F values confirm large effect sizes. Pairwise PERMANOVA and ANOSIM results. Global ANOSIM: R = 0.959, p = 0.0001. PERMDISP: F₄,₁₀ = 154.56, p < 0.001. Note: Pairwise p-values approach but exceed 0.05 due to limited permutation space with n = 3 per group (only 10 unique permutations possible); however, the global PERMANOVA and ANOSIM are highly significant (p = 0.0001), and the high pseudo-F values confirm large effect sizes. Fig. 9 links community composition to hydrocarbon-removal function via a canonical correspondence analysis triplot, relating 16S rRNA-derived ordination axes to reduction efficiency for 18 individual PAH compounds. Compound vectors are shown as biplot arrows and treatments as points in the same ordination space; the compounds annotated span two- to six-ring naphthalene-, pyrene-, and benzo-series aromatics.Fig. 9Canonical Correspondence Analysis (CCA) triplot relating bacterial community composition to PAH compound-specific reduction. Biplot arrows indicate the direction and strength of association between individual PAH compounds and community ordination axes. Treatments in proximity to arrow termini exhibit community compositions most strongly associated with degradation of those specific PAHs.Fig 9 dummy alt text Canonical Correspondence Analysis (CCA) triplot relating bacterial community composition to PAH compound-specific reduction. Biplot arrows indicate the direction and strength of association between individual PAH compounds and community ordination axes. Treatments in proximity to arrow termini exhibit community compositions most strongly associated with degradation of those specific PAHs. Differential abundance is reported in Fig. 10, Fig. 11. Fig. 10 presents a LEfSe-style ranking of the top 30 treatment-specific biomarker genera by LDA effect size, coloured by treatment of enrichment, with the LDA > 2 statistical thresholds marked. Fig. 11 presents DESeq2-style volcano plots for each manure treatment against the control, plotting log₂ fold change against effect magnitude per genus, with significantly enriched and depleted genera distinguished by colour and counted per panel.Fig. 10LEfSe-style analysis of differentially abundant genera across treatments. Horizontal bars represent LDA effect sizes for the top 30 genera; bar colour indicates the treatment in which each genus is most enriched. The dashed vertical line marks the significance threshold (LDA > 2).Fig 10 dummy alt textFig. 11DESeq2-style volcano plots of genus-level differential abundance for each manure treatment versus Control. Each point represents a genus; the x-axis shows log₂ (fold change) and the y-axis shows effect magnitude. Red points indicate significantly enriched genera (log₂FC > 1.5); blue points indicate significantly depleted genera (log₂FC < −1.5). Panel annotations indicate the number of enriched and depleted genera per comparison.Fig 11 dummy alt text LEfSe-style analysis of differentially abundant genera across treatments. Horizontal bars represent LDA effect sizes for the top 30 genera; bar colour indicates the treatment in which each genus is most enriched. The dashed vertical line marks the significance threshold (LDA > 2). DESeq2-style volcano plots of genus-level differential abundance for each manure treatment versus Control. Each point represents a genus; the x-axis shows log₂ (fold change) and the y-axis shows effect magnitude. Red points indicate significantly enriched genera (log₂FC > 1.5); blue points indicate significantly depleted genera (log₂FC < −1.5). Panel annotations indicate the number of enriched and depleted genera per comparison. Predicted functional potential is presented in Fig. 12, Fig. 13. Fig. 12 is a PICRUSt2-predicted MetaCyc pathway heatmap restricted to aromatic-degradation, xenobiotic-metabolism, and ring-cleavage pathways selected from the full predicted pathway set, with per-treatment relative proportions colour-scaled and NSTI values (0.02–0.16) reported as the prediction-reliability metric. Fig. 13 is a companion heatmap of the top 25 hydrocarbon-degradation-associated KEGG Orthologs selected from the full predicted KO set, annotated with enzyme identifiers spanning monooxygenase, dioxygenase, and transport-protein functions relevant to aromatic ring cleavage.Fig. 12PICRUSt2-predicted hydrocarbon degradation pathway heatmap. Of 412 MetaCyc pathways predicted, 123 are associated with aromatic compound degradation, xenobiotic metabolism, or ring-cleavage reactions. Values represent normalised relative proportions across treatments. Cow Manure shows elevated predictions for HMW aromatic catabolism routes.Fig 12 dummy alt textFig. 13Heatmap of the top 25 KEGG Orthologs (KOs) associated with hydrocarbon degradation, selected from 586 degradation-related KOs identified across 7,160 total predicted KOs. Values represent relative proportions per treatment. Key enzymes include nitronate monooxygenase, alkanesulfonate monooxygenase, catechol 1,2-dioxygenase, protocatechuate 3,4-dioxygenase, and benzoate/toluate 1,2-dioxygenase.Fig 13 dummy alt text PICRUSt2-predicted hydrocarbon degradation pathway heatmap. Of 412 MetaCyc pathways predicted, 123 are associated with aromatic compound degradation, xenobiotic metabolism, or ring-cleavage reactions. Values represent normalised relative proportions across treatments. Cow Manure shows elevated predictions for HMW aromatic catabolism routes. Heatmap of the top 25 KEGG Orthologs (KOs) associated with hydrocarbon degradation, selected from 586 degradation-related KOs identified across 7,160 total predicted KOs. Values represent relative proportions per treatment. Key enzymes include nitronate monooxygenase, alkanesulfonate monooxygenase, catechol 1,2-dioxygenase, protocatechuate 3,4-dioxygenase, and benzoate/toluate 1,2-dioxygenase. Fig. 14 relates genus-level relative abundance to function directly, plotting the five most abundant genera (Pseudomonas, Achromobacter, Rhodanobacter, Stutzerimonas, Comamonas) against mean PAH reduction across the 18 compounds, with points labelled by treatment.Fig. 14Scatter plots of the five most abundant genera (Pseudomonas, Achromobacter, Rhodanobacter, Stutzerimonas, Comamonas) versus mean PAH reduction (%) across treatments. Each point is labelled by treatment code. The Control consistently occupies the lower-left region (low abundance, low reduction), while manure-amended treatments show genus-specific positive associations with degradation efficiency.Fig 14 dummy alt text Scatter plots of the five most abundant genera (Pseudomonas, Achromobacter, Rhodanobacter, Stutzerimonas, Comamonas) versus mean PAH reduction (%) across treatments. Each point is labelled by treatment code. The Control consistently occupies the lower-left region (low abundance, low reduction), while manure-amended treatments show genus-specific positive associations with degradation efficiency. Table 4 reports the chemical endpoint underlying all functional analyses above: final residual concentration (mg kg⁻¹, mean ± SE, n = 3) and percentage reduction for 13 selected PAH compounds after 300 days of co-composting, across all five treatments. Quantification followed automated Soxhlet extraction with dichloromethane, GC–MS analysis using a Restek semivolatile standard mixture, duplicate instrumental injections, and triplicate treatment replication.Table 4The average final reduction of selected PAHs in poultry (PM), horse (HM), cow (CM), pig (SPM), and Control (CT) compost piles after 300 days. Values are the mean of three ± standard error for the compost piles.Table 4 dummy alt textSelected PAHs §Initial Conc. (mg.kg−1)Residual concentration (mg.kg−1)†PMHMCMSMCTNaphthalene (2)98±6.20.07±0.02 (99.9)0.04±0.01 (99.9)0.03±0.00 (99.9)0.06±0.01 (99.9)2.91±0.10 (97.0)Acenaphtylene (3)6.0±0.80.37±0.06 (93.8)0.24±0.06 (96.0)0.49±0.07 (91.9)0.23±0.09 (96.2)0.96±0.08 (84.0)Acenaphthene (3)9.2±1.20.25±0.05 (97.3)0.01±0.00 (99.9)0.07±0.01 (99.2)0.01±0.00 (99.9)1.21±0.31 (86.8)Fluorene (3)27±3.10.91±0.07 (96.7)0.01±0.00 (99.9)0.11±0.04 (99.6)0.01±0.00 (99.9)4.39±0.91 (84.0)Anthracene (3)42±4.60.19±0.00 (99.5)0.01±0.00 (99.9)0.25±0.07 (99.4)0.01±0.01 (99.9)1.11±0.07 (97.3)Phenanthrene (3)15±2.13.55±0.42 (76.1)0.09±0.03 (99.4)0.17±0.03 (98.9)0.06±0.02 (99.6)4.44±1.21 (70.2)Fluoranthene (4)2.4±0.60.26±0.08 (89.1)0.18±0.05 (92.4)0.47±0.06 (80.3)0.15±0.06 (93.7)0.94±0.05 (60.5)Pyrene (4)14±1.10.96±0.00 (93.2)0.77±0.09 (94.5)2.97±0.43 (78.9)0.29±0.03 (97.9)3.23±0.81 (77.0)Chrysene (4)55±4.21.46±0.16 (97.3)1.19±0.21 (97.8)2.31±0.32 (95.8)0.64±0.12 (98.8)10.0±2.13 (81.7)Benzo [a] anthracene (4)4.1±1.21.15±0.03 (71.8)1.05±0.01 (74.3)2.59±0.61 (36.5)0.82±0.10 (79.9)2.64±0.57 (35.3)Benzo [b] fluoranthene (5)24±2.70.97±0.01 (95.9)1.00±0.04 (95.8)1.15±0.08 (95.1)0.87±0.09 (96.3)1.99±0.08 (91.6)Benzo [k] fluoranthene (5)2.6±0.31.07±0.08 (59.4)0.98±0.03 (62.9)0.81±0.03 (69.3)1.27±0.31 (51.9)1.27±0.21 (51.9)Benzo[a]pyrene (5)10.0±1.81.72±0.32 (82.7)1.75±0.8 (82.4)2.51±0.58 (74.8)1.71±0.23 (82.8)2.74±0.89(72.5)Perylene (5)5.1±2.01.52±0.28 (69.9)1.81±0.45 (64.2)2.15±0.01 (57.4)1.94±0.18 (61.6)2.52±0.23 (50.1)Indenol [1,2,3-cd] pyrene (6)10±2.92.98±0.01 (70.6)1.50±0.47 (85.2)2.59±0.63 (74.4)1.35±0.41 (86.7)4.35±1.16 (57.1)Dibenzo [a,h] anthracene (5)12±1.96.97±0.45 (39.7)2.92±0.33 (74.8)5.22±0.92 (54.9)2.12±0.18 (81.7)6.97±1.56 (39.8)Benzo [ghi] perylene (6)9.4±1.15.63±0.79 (40.3)2.58±0.00 (72.6)5.11±0.73 (45.8)2.10±0.78 (77.7)5.63±1.07(40.3)Benzo [e] acephenanthrylene (5)3.9±0.10.93±0.13 (75.9)0.98±0.05 (74.6)1.09±0.03 (71.8)0.81±0.05 (79.0)1.93±0.33 (50.0)§The number of benzene rings in different PAHs is given in brackets.†Percent degradation of each PAH under different manure composting is provided in brackets Table 4The average final reduction of selected PAHs in poultry (PM), horse (HM), cow (CM), pig (SPM), and Control (CT) compost piles after 300 days. Values are the mean of three ± standard error for the compost piles.Table 4 dummy alt textSelected PAHs §Initial Conc. (mg.kg−1)Residual concentration (mg.kg−1)†PMHMCMSMCTNaphthalene (2)98±6.20.07±0.02 (99.9)0.04±0.01 (99.9)0.03±0.00 (99.9)0.06±0.01 (99.9)2.91±0.10 (97.0)Acenaphtylene (3)6.0±0.80.37±0.06 (93.8)0.24±0.06 (96.0)0.49±0.07 (91.9)0.23±0.09 (96.2)0.96±0.08 (84.0)Acenaphthene (3)9.2±1.20.25±0.05 (97.3)0.01±0.00 (99.9)0.07±0.01 (99.2)0.01±0.00 (99.9)1.21±0.31 (86.8)Fluorene (3)27±3.10.91±0.07 (96.7)0.01±0.00 (99.9)0.11±0.04 (99.6)0.01±0.00 (99.9)4.39±0.91 (84.0)Anthracene (3)42±4.60.19±0.00 (99.5)0.01±0.00 (99.9)0.25±0.07 (99.4)0.01±0.01 (99.9)1.11±0.07 (97.3)Phenanthrene (3)15±2.13.55±0.42 (76.1)0.09±0.03 (99.4)0.17±0.03 (98.9)0.06±0.02 (99.6)4.44±1.21 (70.2)Fluoranthene (4)2.4±0.60.26±0.08 (89.1)0.18±0.05 (92.4)0.47±0.06 (80.3)0.15±0.06 (93.7)0.94±0.05 (60.5)Pyrene (4)14±1.10.96±0.00 (93.2)0.77±0.09 (94.5)2.97±0.43 (78.9)0.29±0.03 (97.9)3.23±0.81 (77.0)Chrysene (4)55±4.21.46±0.16 (97.3)1.19±0.21 (97.8)2.31±0.32 (95.8)0.64±0.12 (98.8)10.0±2.13 (81.7)Benzo [a] anthracene (4)4.1±1.21.15±0.03 (71.8)1.05±0.01 (74.3)2.59±0.61 (36.5)0.82±0.10 (79.9)2.64±0.57 (35.3)Benzo [b] fluoranthene (5)24±2.70.97±0.01 (95.9)1.00±0.04 (95.8)1.15±0.08 (95.1)0.87±0.09 (96.3)1.99±0.08 (91.6)Benzo [k] fluoranthene (5)2.6±0.31.07±0.08 (59.4)0.98±0.03 (62.9)0.81±0.03 (69.3)1.27±0.31 (51.9)1.27±0.21 (51.9)Benzo[a]pyrene (5)10.0±1.81.72±0.32 (82.7)1.75±0.8 (82.4)2.51±0.58 (74.8)1.71±0.23 (82.8)2.74±0.89(72.5)Perylene (5)5.1±2.01.52±0.28 (69.9)1.81±0.45 (64.2)2.15±0.01 (57.4)1.94±0.18 (61.6)2.52±0.23 (50.1)Indenol [1,2,3-cd] pyrene (6)10±2.92.98±0.01 (70.6)1.50±0.47 (85.2)2.59±0.63 (74.4)1.35±0.41 (86.7)4.35±1.16 (57.1)Dibenzo [a,h] anthracene (5)12±1.96.97±0.45 (39.7)2.92±0.33 (74.8)5.22±0.92 (54.9)2.12±0.18 (81.7)6.97±1.56 (39.8)Benzo [ghi] perylene (6)9.4±1.15.63±0.79 (40.3)2.58±0.00 (72.6)5.11±0.73 (45.8)2.10±0.78 (77.7)5.63±1.07(40.3)Benzo [e] acephenanthrylene (5)3.9±0.10.93±0.13 (75.9)0.98±0.05 (74.6)1.09±0.03 (71.8)0.81±0.05 (79.0)1.93±0.33 (50.0)§The number of benzene rings in different PAHs is given in brackets.†Percent degradation of each PAH under different manure composting is provided in brackets The average final reduction of selected PAHs in poultry (PM), horse (HM), cow (CM), pig (SPM), and Control (CT) compost piles after 300 days. Values are the mean of three ± standard error for the compost piles. The number of benzene rings in different PAHs is given in brackets. Percent degradation of each PAH under different manure composting is provided in brackets Experimental Design, Materials and Methods Sample provenance and feedstock characterisation Crude oil sludge (COWS) was obtained directly from an active petroleum refinery in Durban, KwaZulu-Natal Province, South Africa, and stored at 4°C in sealed, pre-tared containers pending characterisation. Prior to microcosm construction, PAH content was quantified by automated Soxhlet extraction (EPA Method 3541) using dichloromethane (DCM) as the extraction solvent, followed by gas chromatography–mass spectrometry (GC–MS; Agilent 7890B GC coupled to a 5975C MSD) operating in selected ion monitoring (SIM) mode, calibrated against the Restek EPA Method 8270 semivolatile standard mixture (catalogue no. 8270-1) across a concentration range of 10–50 ppm from a 1,000-ppm standard stock [38,39]. Detected PAH compounds included naphthalene, acenaphthylene, acenaphthene, fluorene, anthracene, pyrene, benzo[a]anthracene, chrysene, benzo[b]fluoranthene, benzo[k]fluoranthene, benzo[a]pyrene, indeno(1,2,3-cd)pyrene, benzo[g,h,i]perylene, and perylene, spanning 2–6 fused aromatic rings. Four animal manures, cow, swine/pig, horse, and poultry, were obtained from the University of Pretoria Farm, Pretoria, Gauteng Province, South Africa, and immediately double-bagged in sealed polyethylene bags for transport to the laboratory. Each manure sample was analysed in triplicate for total organic carbon (TOC), total organic nitrogen (TON), and total organic phosphorus (TOP) using standard methods [11,25]. Reference garden soil (20 cm depth) was collected from a maize Farm in Tembisa, north of Kempton Park, East Rand, Gauteng Province (S 26°1′0″, E 28°14′0″), homogenised by passing through a 2-mm sieve, air-dried, and characterised for soil texture, pH, water-holding capacity (WHC%), and elemental composition. Multi-element soil characterisation was performed by ICP-OES (PerkinElmer Optima 5300 DV) following concentrated aqua regia acid digestion (HNO₃:HCl, 1:3, v/v); the 23-element Merck standard stock solution (catalogue no. 43843; 100 mg L⁻¹) was used to prepare ten calibration levels spanning 0.2–8 mg L⁻¹ (ppm), with working standards prepared in 2% (v/v) HNO₃ and all filtrates from digested samples quantified under matrix-matched conditions [24,25]. Co-composting microcosm construction and incubation Five co-composting treatments were assembled under controlled laboratory conditions: four manure-amended microcosms (cow, swine/pig, horse, and poultry) and one unamended control. Microcosm preparation followed a standardised gravimetric procedure. For each treatment, 300 g of crude oil waste sludge (COWS) was dispersed in 400 mL of tetrachloromethane (CCl₄, 99.55%) and homogenised with 1.0 kg of air-dried garden soil; the mixture was air-dried at room temperature until the solvent had fully evaporated, yielding ≈1.30 kg of oil-amended, air-dried soil. Air-dried wood chips were then incorporated at a soil-to-wood-chip ratio of 1:2 (w/v; 1.30 kg oil-amended soil per 2000 mL wood chips), producing a structured, air-dried matrix of 2.34 kg per microcosm. Each matrix was amended with 1.17 kg of air-dried cow, horse, pig, or poultry manure at a matrix-to-manure ratio of 2:1 (w/w), yielding a total microcosm mass of 3.51 kg. An unamended matrix (2.34 kg, no manure) served as the control. All masses are reported on an air-dried-weight basis, while a control setup without manure amendment was maintained for comparison. All reagents were of analytical grade and used without further purification. Each treatment was replicated three times (n = 3), producing 15 individual PVC trough microcosms (dimensions: 30 × 18 × 20 cm) fitted with perforated bases for passive aeration. Microcosms were incubated at room temperature (22 ± 2°C) under ambient atmospheric pressure (100 kPa) for 300 days under static aerobic conditions. Moisture levels were adjusted periodically with deionised water to maintain volumetric moisture content within the range 50–80%. Compost pH was monitored monthly using a calibrated bench pH meter (Crimson Micro pH 2000™, calibrated at pH 4.0, 7.0, and 10.0). Ash content was evaluated gravimetrically by combustion at 550°C for 4 h at both the start (Day 0) and end (Day 300) of the experiment. Microbial respiratory activity and compost maturation were assessed at each monitoring interval using a modified closed-jar CO₂ evolution method, in which enclosed gas-phase CO₂ was measured as a proxy for active microbial metabolism, expressed as µg CO₂-C g⁻¹ dry weight day⁻¹ [26,27]. PAH monitoring over the 300-day incubation PAH concentrations were monitored throughout the 300-day composting period using automated Soxhlet extraction with dichloromethane (DCM) as the solvent, following EPA Method 3541. Extracted samples were concentrated under a nitrogen stream to 1 mL, and 18 target PAH compounds spanning 2–6 aromatic rings were quantified by GC–MS (Agilent 7890B GC/5975C MSD) under the following chromatographic conditions: DB-5MS capillary column (30 m × 0.25 mm × 0.25 µm); injector temperature 280°C; oven temperature programme: 40°C (2 min hold) ramped at 10°C min⁻¹ to 310°C (final hold 10 min); carrier gas: helium at 1 mL min⁻¹; MS operated in SIM mode. Calibration was performed using the Restek semivolatile standard mixture (catalogue no. 8270-1). All chromatographic values were expressed as mean ± standard error (n = 3) per treatment per sampling interval. PAH removal efficiency per compound was calculated as [(C₀ − Ct)/C₀] × 100, where C₀ = initial concentration (mg kg⁻¹) and Ct = residual concentration at each time point (mg kg⁻¹) [39,40]. Culture-dependent enrichment and isolation of hydrocarbon-degrading bacteria To enrich autochthonous hydrocarbon-adapted bacteria from the co-composting matrix, 15 g of composite compost material was collected from each replicate microcosm at the end of the incubation period and inoculated into 100 mL of mineral salts medium (MSM) supplemented with 10 mL of crude oil sludge as the sole carbon and energy source, in 250 mL Erlenmeyer flasks. Cultures were incubated at 28°C under continuous orbital agitation (150 rpm) for 21 days to sustain selective pressure for hydrocarbon-adapted organisms, with sub-culturing every seven days onto fresh MSM–COWS plates. Following enrichment, cultures were serially diluted from 10⁻⁶ to 10⁻⁸ in sterile phosphate-buffered saline (PBS), and 100 µL aliquots were spread-plated onto mineral salt agar (MSA) in triplicate. Morphologically distinct colonies were picked after 72 h incubation at 28°C, streak-purified over three successive rounds on nutrient agar (Oxoid, UK), and archived as glycerol stocks (25%, v/v) at −80°C. Isolate viability was confirmed by re-streaking frozen stocks onto nutrient agar prior to each downstream assay. Metabolic activity on crude oil sludge substrates was assessed using the 2,6-dichlorophenolindophenol (2,6-DCPIP) redox indicator assay: individual isolates were inoculated into 50 mL MSM broth amended with crude oil sludge (1%, w/w) and 0.04% (w/v) 2,6-DCPIP in 250 mL Erlenmeyer flasks, incubated at 28°C, 150 rpm, with colour change recorded daily from Days 3 to 7 and scored semi-quantitatively as: − (no change; blue retained), + (minimal decolourisation), ++ (moderate decolourisation), or +++ (complete colour loss). Isolates expressing confirmed 2,6-DCPIP decolourisation were forwarded for further molecular and functional characterisation. DNA extraction from pure bacterial isolates Total genomic DNA was extracted from pure bacterial isolate cultures using the Quick g-DNA™ kit (Zymo Research, USA) according to the manufacturer's protocol. Isolates were cultured overnight in nutrient broth (28°C, 150 rpm), harvested by centrifugation (8,000 × g, 5 min), and cell pellets were processed for lysis and column-based nucleic acid purification. DNA integrity was confirmed by 1% (w/v) agarose gel electrophoresis in 1× TAE buffer, visualised under UV light after SYBR Safe staining (Supplementary File 1). Quantification was performed using a NanoDrop 2000 spectrophotometer (Thermo Scientific, USA); samples with A260/A280 purity ratios of 1.8–2.0 and concentrations of 20–150 ng µL⁻¹ were accepted for downstream amplification. Culture-independent DNA extraction from co-compost matrices For community-level 16S rRNA gene profiling, total community DNA was extracted from the co-composting matrix samples using the Faecal/Soil Total DNA™ extraction kit (Zymo Research Corporation, USA) following the manufacturer's protocol, optimised for recovery from complex petroleum-contaminated matrices with high humic content. For each sample, 15 g of thoroughly homogenised compost material was suspended in 50 mL sterile phosphate-buffered saline, incubated overnight at 4°C with gentle agitation to facilitate cell detachment, homogenised using a bead-beating step, and centrifuged at 12,000 rpm for 5 min at 4°C. The resulting supernatant was used as the DNA extraction input. DNA integrity and purity were assessed using NanoDrop 2000 spectrophotometry for purity ratio (A260/A280 = 1.8–2.0) and concentration (20–150 ng µL⁻¹). Total community genomic DNA extracted from both pure bacterial isolates and compost matrices was pooled per treatment for downstream amplicon library construction [38]. PCR amplification, library preparation, and illumina MiSeq sequencing 16S rRNA gene amplification was performed using a two-step nested PCR strategy to maximise taxonomic resolution across V1–V3 hypervariable regions. In the first step, universal primers 27F (5′-AGAGTTTGATCCTGGCTCAG-3′) and 1492R (5′-ACGGCTACCTTGTTACGACTT-3′) were used to amplify full-length 16S rRNA gene sequences. In the second step, targeted V1–V3 amplification was performed using 27F and 518R (5′-GTATTACCGCGGCTGCTGG-3′) to generate amplicons of approximately 500 bp, selected for their superior resolution of hydrocarbon-degrading taxa in complex environmental matrices. PCR reactions were performed in 25 µL total volumes containing 12.5 µL of 2× DreamTaq Green Master Mix (Thermo Fisher Scientific), 1 µL each of forward and reverse primers (10 µM), 1 µL of template DNA (∼50 ng µL⁻¹), and 9.5 µL nuclease-free water. Thermocycler conditions: initial denaturation at 95°C for 10 min; 32 cycles of 95°C for 30 s (denaturation), 55°C for 30 s (annealing), and 72°C for 1 min (extension); followed by a final extension at 72°C for 10 min. Amplicons were purified using the Zymo DNA Clean & Concentrator™ kit (Zymo Research, USA). Sequencing libraries were prepared by attaching Illumina-compatible dual-index adapter sequences, following standard Illumina 16S library preparation guidelines [41]. Paired-end sequencing (2 × 300 bp reads, v3 chemistry) was performed on the Illumina MiSeq platform (Illumina Inc., San Diego, CA, USA) at the Sequencing Unit, University of South Africa (UNISA), Florida Campus. Catechol 2,3-dioxygenase (cbzE) gene detection All isolates were screened for the catechol 2,3-dioxygenase functional gene (cbzE/C23O) using targeted PCR with primers CbzE-F (5′-GCACTGCACATCGACTTCAA-3′) and CbzE-R (5′-GTAGTAGTCGATGCCGATGA-3′), generating an amplicon of approximately 392 bp. Reaction volumes, master mix composition, and thermocycler conditions were identical to those described for 16S rRNA gene amplification, except that the annealing temperature was adjusted to 58°C. PCR products were resolved on 1.5% (w/v) agarose gels stained with SYBR Safe, and amplicons of the expected size were scored as positive (+); absence of a visible band was scored as negative (−). All cbzE PCR assays were performed in triplicate alongside a no-template negative control. Bioinformatic processing of amplicon sequencing data Raw paired-end sequencing reads were subjected to quality assessment and pre-processing using ngsShoRT software [17], which performed adaptive trimming of low-quality bases (quality threshold Q < 20), removal of ambiguous nucleotide calls, and filtering of PCR artefacts. Chimeric sequence detection and removal were carried out using the UCHIME algorithm [16,17], with default parameters and the SILVA reference database for chimera checking. High-quality reads were imported into the Mothur pipeline (v1.40.0) [18] for OTU clustering and were subsequently processed using QIIME2, with taxonomic classification performed against the SILVA reference database. Sequences were aligned against the SILVA 16S rRNA reference database (releases 128 and 132) [22], and clustering into OTUs at 97% pairwise sequence similarity was performed using the OptiClust algorithm. Taxonomic classification of representative OTU sequences was performed using the Naïve Bayesian classifier with an 80% bootstrap confidence threshold [23], with classification validated against the NCBI nucleotide database using BLAST+ [42]. All accepted quality-filtered sequences were deposited under NCBI BioProject accession PRJNA794053. Alpha and beta diversity analysis Alpha diversity indices, Chao1, Shannon (H′), Simpson (1–D), dominance (D), Pielou's equitability (J), Brillouin, Menhinick, Margalef, Fisher's α, Berger–Parker, and evenness (eH/S), were calculated from rarefied OTU-level count tables at a genetic distance of 0.03 within the Mothur pipeline (v1.40.0). Good's coverage estimator was computed per treatment to assess sequencing depth adequacy. Rarefaction curves were generated by plotting cumulative OTU count against sequencing depth from 1,000 to maximum reads per treatment. Beta diversity analyses were performed in R (v3.6.1) using the vegan and phyloseq packages [35,36]. Bray–Curtis dissimilarity matrices were computed from relative abundance OTU tables and visualised as a pairwise heatmap. Non-metric multidimensional scaling (NMDS) ordination and principal component analysis (PCA) of standardised OTU relative abundances were performed to resolve treatment-level compositional separations in reduced ordination space. Canonical Correspondence Analysis (CCA) was conducted to directly correlate microbial community composition with PAH compound-specific reduction efficiencies across the five treatments. Permutational multivariate analysis of variance (PERMANOVA; adonis2 function in vegan) was executed using Bray–Curtis dissimilarity with 9,999 unrestricted permutations to test for significant differences in community composition among treatments [36]. Pairwise PERMANOVA comparisons were corrected for multiple testing using the Benjamini–Hochberg false discovery rate (FDR) procedure. Analysis of similarities (ANOSIM) [45] was performed as a complementary rank-based statistical test. Homogeneity of multivariate group dispersions was assessed using the PERMDISP procedure [46], to verify PERMANOVA's assumption of equal within-group dispersion. Similarity percentage analysis (SIMPER) [45] was used to decompose between-treatment Bray–Curtis dissimilarity to identify genera contributing most to compositional divergence. The Mantel test (9,999 permutations) assessed the correlation between compositional dissimilarity matrices and PAH degradation profiles. Environmental vector fitting projected PAH degradation variables (LMW reduction, HMW reduction, mean PAH reduction, mean residual concentration) onto the NMDS ordination space. Differential abundance analysis: LEfSe and DESeq2 Treatment-specific bacterial biomarkers were identified using linear discriminant analysis effect size (LEfSe) [47]. The LEfSe workflow applied a Kruskal–Wallis test (α = 0.05) for between-treatment differences, followed by pairwise Wilcoxon tests for internal consistency, and LDA (log-transformed relative abundances) to estimate the effect size of each discriminating taxon. Only genera with LDA scores exceeding the significance threshold (LDA score > 2) were retained as biomarkers. Genus-level differential abundance between each manure treatment and the unamended control was further quantified using the DESeq2 framework with a negative binomial model, size factor normalisation by median-of-ratios, and shrinkage estimation of log₂ fold changes using the apeglm method. Genera with |log₂FC| > 1.5 and a Benjamini–Hochberg-adjusted p-value < 0.05 were classified as significantly enriched (log₂FC > 1.5; red) or depleted (log₂FC < −1.5; blue) relative to the control. Functional pathway prediction: PICRUSt2 and KEGG ortholog analysis Predictive functional profiling was performed using the PICRUSt2 pipeline (Phylogenetic Investigation of Communities by Reconstruction of Unobserved States, version 2) [22,23,45]. Placement of OTU representative sequences onto a reference phylogenetic tree was performed using EPA-ng (maximum parsimony criterion), followed by hidden-state prediction using the CASTOR algorithm. Functional gene content was predicted at the KEGG Orthology (KO) level, and MetaCyc pathway abundances were calculated by aggregating KO-level predictions. The Nearest Sequenced Taxon Index (NSTI) was computed per OTU as a quality metric for prediction accuracy, with per-treatment mean NSTI values retained in the dataset (range: 0.02–0.16). KOs were classified as hydrocarbon degradation-related based on KEGG PATHWAY and BRITE functional annotations associated with aromatic compound catabolism, xenobiotic biodegradation, and ring-cleavage reactions. MetaCyc pathway heatmaps and the top-25-KO heatmap were constructed in R using the ggplot2 package [39], with normalised relative proportions per treatment as the input variable. Taxonomic visualisation: Circos chord diagrams and genus-level heatmaps Phylum-level connectivity between co-composting treatments and bacterial taxa across the combined dataset was visualised using a Circos chord diagram generated in R using the circlize package, with ribbon width scaled proportionally to relative taxon abundance and outer arc length reflecting each taxon's cumulative contribution to total classified reads. Genus-level abundance heatmaps were produced from log₁₀-transformed relative abundance values using the heatmap.2 function in the gplots package within R (v3.6.1). Venn diagrams of shared and unique genera across treatments were generated using the VennDiagram package in R. Scatter plots correlating genus-level abundance with mean PAH reduction were generated in R with ggplot2 [39,40]. Statistical computing environment All data processing, statistical modelling, diversity calculations, ordination analyses, and visualisation were performed in R (v3.6.1 and v4.3.1) [40], using the following packages: vegan (v2.6) for diversity indices, beta diversity, PERMANOVA, ANOSIM, PERMDISP, and SIMPER [44]; phyloseq (v1.42) for microbiome data management and ordination [43]; ggplot2 (v3.4) for graphical output [39]; circlize for Circos diagram generation; DESeq2 for negative-binomial differential abundance modelling; and gplots/heatmap.2 for heatmap generation. Phylogenetic tree construction for taxonomic confirmation of culture-validated isolates was performed using MEGA6 (Molecular Evolutionary Genetics Analysis version 6.0) with the neighbour-joining method and 1,000 bootstrap replications [48]. All analysis scripts, raw OTU count tables, normalised relative abundance matrices, diversity index outputs, and PAH concentration files are archived within the associated data repository under NCBI BioProject accession PRJNA794053. Limitations Several data-specific constraints should be acknowledged. Amplicon sequencing was performed at a single terminal time point (Day 300) rather than at multiple intervals throughout the incubation period; therefore, temporal community dynamics between Days 0 and 300 were inferred from physicochemical trends and culture-based observations rather than directly resolved through longitudinal sequencing. Although each co-composting treatment was represented by three experimental replicates for physicochemical measurements and downstream statistical analyses, microbial community profiling was generated from a single pooled DNA sequencing library per treatment (n=3 samples pooled to a single sample). Accordingly, the sequencing dataset does not permit estimation of within-treatment microbial variation or formal statistical testing of sequencing-derived diversity metrics. In addition, DNA used for amplicon sequencing was pooled before library preparation, and the resulting taxonomic profiles were therefore interpreted as treatment-level community representations rather than independent replicate-level observations. Functional pathway profiles were generated using PICRUSt2 inference from 16S rRNA gene data rather than shotgun metagenomic sequencing. While NSTI values indicated high prediction reliability (0.02-0.16), the inferred pathways remain computational predictions and do not constitute direct evidence of functional gene abundance or expression. Consistent with this limitation, the reported biodegradation potential was supported by culture-dependent validation through PCR-based detection of the cbzE gene and 2,6-DCPIP redox screening, but not by metagenomic or metatranscriptomic analyses. The NMDS ordination yielded a stress value of 0.2334, slightly above the conventional threshold for robust two-dimensional representation, reflecting the pronounced compositional heterogeneity among the five treatment groups. Crude oil sludge was obtained from a single petroleum refinery in Durban, South Africa, and all animal manures were sourced from a single farm, which may limit the generalisability of the observed microbial community structures to substrates derived from other geographic locations or management systems. Furthermore, biosurfactant production data were not included in the final dataset and not inferred from the functional interpretations presented herein. Finally, no pilot-scale or field-scale validation was conducted, and caution should therefore be exercised when extrapolating these findings to operational bioremediation settings. Ethics statement Human subjects Not applicable. This dataset was generated exclusively from environmental microbiological and physicochemical laboratory experiments. No human participants were recruited, enrolled, or involved at any stage of sample collection, experimental processing, sequencing, or data analysis. No informed consent procedure, ethical committee approval, Declaration of Helsinki compliance, or protocol registration was required or sought. Animal experiments Not applicable. Although animal-derived manure materials, sourced from cow, swine/pig, horse, and poultry livestock maintained at the University of Pretoria Farm, Pretoria, South Africa, were incorporated as organic co-composting amendments, no live animals were handled, experimentally manipulated, sacrificed, or subjected to any invasive or non-invasive procedure at any stage of this research. Manure collection was performed non-invasively as agricultural residue under standard farm management practices and did not constitute an animal experiment under the U.K. Animals (Scientific Procedures) Act 1986, EU Directive 2010/63/EU, or the NIH Guide for the Care and Use of Laboratory Animals (NIH Publications No. 8023, revised 1978). ARRIVE guideline compliance was accordingly not required. Social media and personal data Not applicable. No data were collected from social media platforms, online repositories of personal information, or from any identifiable individuals at any stage of this research. The entire dataset consists of wholly experimental laboratory-generated data derived from environmental microorganisms, agricultural residues, and petroleum industrial waste materials. No participant anonymisation, platform data redistribution policy compliance, or informed consent for personal data use was required. General ethical compliance The authors confirm that they have read and fully comply with the ethical requirements for publication in Data in Brief. All experimental work was conducted at the Department of Environmental Sciences, University of South Africa (UNISA), Florida Campus, Roodepoort, Gauteng, South Africa, in strict accordance with institutional biosafety guidelines and standard microbiological containment practices under Biosafety Level 1–2 conditions. Crude oil sludge substrate was obtained from an active petroleum refinery facility in Durban, KwaZulu-Natal Province, South Africa, under a material transfer arrangement for non-commercial academic research purposes exclusively; no regulated clinical, human-derived, or vertebrate animal-derived biological material was incorporated at any experimental stage. Animal manures were obtained as unregulated agricultural by-products from the University of Pretoria Farm under standard inter-institutional research cooperation arrangements. Garden soil was collected from an agricultural plot in Tembisa, Gauteng Province (S 26°1′0″, E 28°14′0″), a non-protected, non-restricted land-use zone, without the requirement for environmental sampling permits under applicable South African legislation. Raw 16S rRNA gene amplicon sequencing data and associated community composition datasets have been deposited in the publicly accessible NCBI Sequence Read Archive and BioProject repository under accession number PRJNA794053, in full compliance with NCBI data deposition, access, and reuse policies. The funding body, the South African National Research Foundation (NRF; PhD Research Grant), imposed no restrictions on data access, open deposition, or publication timelines. All authors have reviewed and approved the final dataset and accompanying manuscript in its entirety and collectively confirm that all data described herein are original, unpublished beyond the associated doctoral thesis repository, and entirely free from fabrication, falsification, selective reporting, or undisclosed duplication. CRediT Author Statement Onyedikachi Ubani: Conceptualisation, study design, field sampling, laboratory processing, DNA extraction and library preparation, workflow development, data curation, formal analysis, visualisation, and writing – original draft preparation. Veronica M. Ngole-Jeme: Methodology refinement, supervision, project administration, resource provision, critical review and editing of the manuscript, and overall research oversight.