Enrichment hydrocarbon‑degrading bacterial communities from the southern Gulf of Mexico in long‑term stored sediments

TL;DR

This study explores how bacterial communities in Gulf of Mexico sediments break down petroleum hydrocarbons over time, revealing resilient microbes that thrive in cold, dark conditions.

Contribution

The study provides new insights into the long-term resilience and succession patterns of hydrocarbon-degrading bacterial communities in deep-sea sediments.

Findings

Bacterial communities in Gulf of Mexico sediments showed stable initial composition followed by significant shifts after 12 months of incubation.

Hydrocarbon-degrading genera like Colwellia, Alcanivorax, and Shewanella increased in abundance during incubation.

Chemical analysis showed significant depletion of alkanes and polycyclic aromatic hydrocarbons, indicating active biodegradation.

Abstract

The Gulf of Mexico is chronically exposed to petroleum hydrocarbons from natural seeps and anthropogenic activities, sustaining diverse microbial communities capable of hydrocarbon degradation. To investigate natural bacterial succession associated with long-term hydrocarbon degradation, sediment samples from shallow (< 1000 m) and deep (> 2500 m) sites in the southern Gulf of Mexico were incubated at 4 °C for up to 24 months. Temporal changes in bacterial community composition were analyzed by 16S ribosomal RNA gene sequencing, and residual hydrocarbons were quantified by gas chromatography-mass spectrometry. Initial communities differed significantly between shallow and deep sediments but remained stable during the first six months before shifting markedly after 12 months of incubation. Gammaproteobacteria, Alphaproteobacteria, and Bacteroidota increased in relative abundance, whereas Deltaproteobacteria declined. Genera such as Colwellia, Alcanivorax, Shewanella, and Neptunomonas, which include well-known hydrocarbon-degrading species, displayed dynamic, time-dependent enrichment patterns. Chemical analyses revealed substantial depletion of alkanes and polycyclic aromatic hydrocarbons, indicating sustained biodegradation activity. These results demonstrate that Gulf of Mexico sediment communities harbor metabolically resilient consortia capable of long-term hydrocarbon mineralization under low-temperature, dark conditions. The integrated microbial and geochemical data provide new insights into ecological succession and the persistence of hydrocarbon-degrading bacteria in deep-sea sediments, contributing to a better understanding of natural attenuation processes in petroleum-impacted marine environments.

The online version contains supplementary material available at 10.1007/s10532-026-10260-7.

Figures

Sampling map. The green marks indicate the sites where the samples were taken near the coast, at depths below 1000 m, while the red marks show the locations of the samples taken in the open sea, between 2500 and 3000 m deep. The red bathymetric line depicts the 1000-m depth contour. Black diamonds indicate areas of natural hydrocarbon seepage

OTUs generated at 97% of sequence identity profiles of sGoM sediment samples using the Bray–Curtis dissimilarity plotted in a Nonmetric Multidimensional Scaling (NMDS). A Clustering based on sampling depth depicting shallow in green markers and deep in red markers. B OTUs profile clustered by incubation time. For clarity, samples corresponding to 0, 4 and 6 months were grouped as “short-time storages”. Ellipses describe the clustering of the samples at 0.9 of confidence. ANOSIM and pairwise ADONIS analysis results are reported in each graph. Significant differences between groups are highlight

Taxonomic identification of bacteria. Each color represents a taxon at the A class level and the B genus level. Sample analysis was initiated at 0 months, and then evaluated at 4, 6, 12, and 20 months

Full text

Introduction

The Gulf of Mexico (GoM), located in the tropical and subtropical North Atlantic Ocean, is a marginal sea and oceanic basin defined by the coastlines of the United States, Mexico, and Cuba. It hosts a diverse mosaic of marine habitats of high ecological and economic importance. Yet, despite its biodiversity significance, the GoM also harbors one of the planet’s richest crude-oil reserves, supporting an extensive network of offshore platforms, tankers, and subsea pipelines dedicated to its extraction (Daneshgar et al. 2016). Persistent natural oil seeps, decades of intensive exploitation, and episodic catastrophic spills -most notably from the 2010 Macondo well, the largest marine oil spill on record- have left the GoM chronically saturated with crude oil (Atlas and Hazen 2011; MacDonald et al. 2015).

Crude oil (petroleum) is a viscous and complex liquid composed mainly of hydrocarbons. Roughly two-thirds of it consists of paraffins (alkanes), naphthenes (cycloalkanes), and related derivatives, while the remaining fraction is largely made up of polycyclic aromatic hydrocarbons (PAHs) and other aromatic compounds. In addition to hydrocarbons, crude oil contains a range of non-hydrocarbon trace constituents. These include organic compounds bearing heteroatoms of sulfur, oxygen, and nitrogen, as well as trace amounts of diverse heavy metals, such as vanadium and nickel (Loring and Rantala 1992; Tirado et al. 2023; Tissot and Welte 1984).

Hydrocarbon solubility strongly influences microbial bioavailability. When oil contacts seawater, low-molecular-weight (LMW) hydrocarbons, including LMW PAHs, aromatic BTEX compounds (benzene, toluene, ethylbenzene, and xylenes), phenols, and other heterocyclic compounds readily dissolve into and disperse throughout the water column (Ghosal et al. 2016; Prince et al. 2013). The water-soluble fraction (WSF) is relevant because it directly impacts aquatic organisms and is a major contributor to the toxicity of crude oil spills (Neff and Anderson 1980; Murawski et al. 2021).

Persistent hydrocarbon contamination in the GoM has driven the selection of native microbial communities equipped with a broad range of catabolic enzymes -such as alkane monooxygenases and PAH dioxygenase-, that channel hydrocarbons into central metabolic pathways, enabling their use as primary sources of carbon and energy (Rodríguez-Salazar et al. 2021). Indeed, following the capping of the Macondo well, oil concentrations in the water column dropped significantly within just three weeks (Atlas and Hazen 2011). While this rapid attenuation has been attributed to several physicochemical factors -such as the relatively light composition of the Macondo crude, its high-pressure release at ~ 1500 m depth, and its efficient dissolution and dispersion-, microbial biodegradation is believed to have played an equally important role (Atlas and Hazen 2011).

In contrast to the components of the WSF, poorly soluble petroleum compounds -i.e. highly branched alkanes, high-molecular-weight polycyclic aromatic hydrocarbons (PAHs), and heavy metals- tend to accumulate in benthic sediments. In the GoM, bacterial communities in the water column are compositionally distinct from those inhabiting benthic sediments (Godoy-Lozano et al. 2018). Sediments, richer in organic substrates, harbor denser and more metabolically diverse microbial consortia, some of which are capable of utilizing recalcitrant hydrocarbons as potential carbon sources (petroleum hydrocarbon-degrading bacteria -PHDB-). However, the degradation of these compounds proceeds over substantially longer time scales compared to the biodegradation of the WSF (Ghosal et al. 2016; Ma et al. 2021). The capacity of bacterial communities inhabiting marine sediments to degrade petroleum has been well documented in several regions (Ma et al. 2021; Hoshino et al. 2020; Domingues et al. 2020), and has also been demonstrated in the GoM (Bacosa et al. 2018; Kimes et al. 2013; Liu and Liu 2013; Orcutt et al. 2017; Ramírez et al. 2020; Sánchez-Soto Jiménez et al. 2018). Previous studies by our group have shown that PHDB from genera such as Haliea, Reinekea, Colwellia, and Pseudomonas are prevalent in sediments from the southern Gulf of Mexico (sGoM) (Godoy-Lozano et al. 2018; Rodríguez-Salazar et al. 2021; Raggi et al. 2020). However, the patterns of bacterial succession during hydrocarbon degradation remain poorly understood.

Here, we investigated natural bacterial succession over time (0, 4, 6, 12, 20, and 24 months) in sediment samples collected from shallow and deep waters of the sGoM. Samples were maintained at 4 °C in the dark, and high-throughput DNA sequencing was employed to track temporal variations in community composition. In parallel, we assessed changes in the concentrations of petroleum derivatives at the beginning and end of the experiment. By integrating microbial and geochemical analyses, this work provides new insights into the processes underlying bacterial succession and the natural mineralization of hydrocarbons in marine sediments, contributing to the development of more effective bioremediation strategies for petroleum pollution in the sGoM.

Materials and methods

Sampling and microbial enrichment

In August 2016, twelve sediment samples were collected from the sGoM using a box corer onboard the experimental ship Justo Sierra (http://www.buques.unam.mx/justo-sierra/). As depicted in Fig. 1, there were seven sampling sites (S01, S03, S04, S05, S09, S14, and S27) located at depths below 1000 m near the coast, and five sampling sites (S40, S33, S54, S55, and S59) taken from open sea locations between 2500 and 3200 m depth. Sediment was collected from the upper 10 cm of the seafloor. After collection, each sample was sub-sampled into sterile 50 mL Erlenmeyer flasks and stored in darkness at 4 °C until processing. An additional sediment sample was frozen and kept as a time-zero control.Fig. 1. Sampling map. The green marks indicate the sites where the samples were taken near the coast, at depths below 1000 m, while the red marks show the locations of the samples taken in the open sea, between 2500 and 3000 m deep. The red bathymetric line depicts the 1000-m depth contour. Black diamonds indicate areas of natural hydrocarbon seepage

DNA extraction and microbial amplicon library preparation for illumina MiSeq sequencing

Total DNA was extracted from 0.5 g of sediment sample after 0, 4, 6, 12, and 20 months of incubation. The PowerSoil® DNA Isolation Kit (MO BIO Laboratories, Inc., West Carlsbad, CA) was employed according to the manufacturer’s instructions. The variable regions V3-V4 of the 16S rRNA gene were amplified using the primers S-D-Bact-0341-b-S-17 and S-D-Bact-0785-a-A-21 (Klindworth et al. 2013). The 16S Ribosomal RNA Gene Amplicons for the Illumina MiSeq System were prepared following the instructions provided by Illumina (https://support.3illumina.com/documents/documentation/chemistry_documentation/16s/16s-metagenomic-library-prep-guide-15044223-b.pdf). The libraries were assessed using a Bioanalyzer 2100 System (Agilent, Inc.) and quantified with Qubit Fluorometer (Thermo Fisher Scientific, Waltham, MA, USA) before sequencing. The Illumina libraries were normalized to a concentration of 4 nM for sequencing on the Illumina MiSeq platform using a 600-cycle kit configured to generate 300 bp paired-end reads.

Sequence data processing and OTUs taxonomic assignment

The bcl2fastQ v2.20 software was used for Illumina FastQ base calling, and only reads with an average Phred Q >  = 20 quality score were retained for further analysis. The paired-end reads were overlapped, and the original amplicon region (between 400 and 490 bp) was reconstructed using Flash v1.2.7 software (Magoč and Salzberg 2011) with default parameters. Non-overlapping sequences were discarded. High-quality sequences and their extended amplicons were dereplicated, and chimeric sequences were removed with VSEARCH v2.13.4 (Rognes et al. 2016). The non-chimeric sequences were then clustered at 97% sequence identity using the ‘cluster_fast’ function. Clusters consisting of only one sequence per sample (singletons) were discarded. The filtered OTUs (Operational Taxonomic Units) at a 97% similarity level were normalized using the ‘cumNormMat’ function from the MetagenomeSeq R library (McMurdie and Holmes 2013).

Diversity indices were calculated using diverse methods: Good’s coverage, using the formula reported by Gilbert et al. 2010; Chao1 and Shannon alpha indexes by Phyloseq v1.19.1 R library (McMurdie and Holmes 2013); and beta diversity estimated by Bray–Curtis dissimilarity through Vegan v2.4–6 R library (Dixon 2003). We performed an ANOVA with Tukey’s multiple comparisons test for statistical comparisons across the different sampling events with Prism 6 (Magoč and Salzberg 2011). In parallel, the taxonomic annotation of the extended amplicon was performed before clustering using Parallel-meta v2.4.1 (Su et al. 2014) and the Metaxa2 database v2.1.1 (Bengtsson-Palme et al. 2015). The annotation tables were aggregated at genus rank and normalized by relative abundance in R (Bengtsson-Palme et al. 2015). Beta diversity differences between groups were visualized in a Non-metric Multidimensional Scaling plot (NMDS), generated using the Vegan v2.7-1-6 (Package ‘vegan’ 2025) R library. Comparison between groups was performed by the anosim function of the Vegan R package using the Bray–Curtis dissimilarity as a distance measure. Additionally, differentially abundant bacterial genera were identified using genus-level count matrices separated by sediment type (shallow and deep) and compared across incubation time points. Analyses were performed in R using the metagenomeSeq package (Paulson et al. 2013). Count data and associated sample metadata were integrated into MRexperiment objects, and low-abundance genera were filtered by requiring their presence in at least 80% of samples within each comparison group. Data were normalized using cumulative sum scaling, and differential abundance was assessed using the zero-inflated Gaussian model (fitZig). Genera with a p-value < 0.05 were considered differentially abundant.

Microbial interaction analysis

The MetaMIS software was used with default parameters, to infer for each site and time points data, the possible interactions between the genera present in the community. MetaMIS (Shaw et al. 2016) uses a Lotka-Volterra model to infer positive and negative interactions between factors (genera) and for this work, we generated network graphs for only 0.01% of all possible interactions, which were evaluated by the software as follows:

Mutualism (+ / +): both species benefit from the interaction.

Competition (−/−): both species are harmed by the interaction.

Parasitism or predation ( ±): one species benefits while the other is harmed.

Commensalism (+ /0): one species benefits while the other is unaffected.

Amensalism (−/0): one species is harmed while the other is unaffected.

The input matrix for MetaMIS is included as Supplementary Material.

Hydrocarbon quantification

Hydrocarbon compounds were extracted from the 0- and 24-month sediment samples using a hexane: dichloromethane mixture in a ratio of 1:1, utilizing an ASE 300 accelerated solvent extractor. Following extraction, the samples were concentrated and purified using a chromatographic column packed with sulfite, silica, alumina, and granulated copper, as described by (Tolosa et al. 2004). The purified extracts were subsequently analyzed using a GC–MS system (Agilent Technologies 6890N/5973MS).

The estimation of organic matter was carried out using the Walkley–Black method. Specifically, 1 mL of diphenylamine was added to the sample, and the organic matter and carbon content were estimated based on back titration with a 0.5 N ferrous sulfate solution, following the procedure described by (Rich 1958).

Results

The bacterial community composition shifts during long-term sediment storage

Total DNA was extracted from all sediment samples after 0, 4, 6, 12, and 20 months of incubation at 4 °C. Sequencing yield was 67,000 reads in every sample, which is sufficient to capture a maximum number of Operational Taxonomic Units (OTUs) expected per sample, according to the Chao 1 index (Supplementary Material). Accordingly, the Good’s coverage index ranged from 0.895 to 0.997, indicating that only 0.3–10.5% of OTUs remained unsampled (Table S1).

OTUs profiles from shallow and deep regions are significantly different (P-val < 0.1) as shown in Fig. 2A. Analysis of the same profile over time showed that the taxonomic composition remained stable during the first six months of incubation at 4 °C (Fig. 2B). However, substantial shifts in OTU composition were observed after 12 and 20 months.Fig. 2OTUs generated at 97% of sequence identity profiles of sGoM sediment samples using the Bray–Curtis dissimilarity plotted in a Nonmetric Multidimensional Scaling (NMDS). A Clustering based on sampling depth depicting shallow in green markers and deep in red markers. B OTUs profile clustered by incubation time. For clarity, samples corresponding to 0, 4 and 6 months were grouped as “short-time storages”. Ellipses describe the clustering of the samples at 0.9 of confidence. ANOSIM and pairwise ADONIS analysis results are reported in each graph. Significant differences between groups are highlighted in bold

An independent analysis of bacterial genera with relative abundance > 0.1% confirmed these patterns. Microbial communities at 12 and 20 months displayed distinct profiles compared to those at short sampling times (0, 4, and 6 months). Nonetheless, a core set of 450 bacterial genera, previously reported for the sediments in the sGoM (Godoy-Lozano et al. 2018), persisted across all samples throughout the duration of this study (Supplementary Material).

Assessing microbial shifts in marine sediments

Clear changes in microbial diversity were observed in all sediments incubated at 4 °C. Gamma-, Delta-, and Alphaproteobacteria and their corresponding genera dominated the communities but varied in relative abundance over time (Fig. 3A and B). The abundance of Alpha- and Gamma-proteobacteria increased gradually, while that of Deltaproteobacteria decreased. Among Bacillota, Clostridia declined between 0 and 12 months but recovered by 20 months, whereas the Bacilli group and Actinobacteria, which remained stable from 0 to 6 months, peaked at 12 months before decreasing. In addition, members of Bacteroidota, such as Cytophagia, increase only after 20 months of incubation.Fig. 3. Taxonomic identification of bacteria. Each color represents a taxon at the A class level and the B genus level. Sample analysis was initiated at 0 months, and then evaluated at 4, 6, 12, and 20 months

Differential abundance analysis at the genus level (Fig. 4A) revealed the prevalence of similar genera across samples, albeit with significant variations in abundance. While most genera of non-hydrocarbon-degrading bacteria, such as Rhodovibrio, Geoalkalibacter, Desulfonatronum, Desulfovibrio, and Thioprofundum (Fig. 4A, names labelled in grey), maintained relatively constant abundance levels throughout the incubation period, obligate (Fig. 4A, in red) and facultative (Fig. 4A, in blue) hydrocarbon-degrading bacteria exhibited notable changes. For instance, Colwellia and Oleispira exhibited a pronounced increase in abundance at 4 months, whereas Shewanella, Marinobacter, Alcanivorax, and Neptunomonas showed a significant increase after 4 months, even in deep-water sediments, before decreasing by 12 months. These temporal patterns suggest that changes in nutrient availability drive the succession of petroleum hydrocarbon-degrading bacteria (PHDB) during long-term incubation.Fig. 4A Relative abundance of bacterial genera detected over time in shallow and deep marine sediment samples. Circle size indicates relative abundance. Genera highlighted in red correspond to obligate hydrocarbon-degrading bacteria, while those in blue represent facultative hydrocarbon degraders; remaining genera are considered non-degraders. Shallow (light gray) and deep (dark gray) sediment samples are distinguished by background shading at the bottom of each panel. Vertical black lines separate incubation time points (0, 4, 6, 12, 20 months). The cladogram on the left illustrates the phylogenetic relationships among genera, with shaded blocks indicating taxonomic class (Firmicutes, Alphaproteobacteria, Deltaproteobacteria, Gammaproteobacteria). B and C Pairwise differential abundance analysis of bacterial genera in shallow (B) and deep (C) sediments, expressed as log fold change (logFC) relative to the initial time point (0 months). Bars represent comparisons between 0 and 4 months (teal), 0–6 months (red), 0–12 months (blue), and 0–20 months (green). Only genera with statistically significant differential abundance are shown (padj < 0.05). Positive logFC values indicate enrichment relative to the initial community

In line with the temporal patterns observed in Fig. 4A, pairwise differential abundance analyses (Fig. 4B and C) indicate that statistically significant changes are mainly associated with hydrocarbon-degrading genera. In shallow sediments (Fig. 4B), the strongest differential signals are concentrated between 4 and 12 months, with obligate and facultative degraders such as Alcanivorax, Marinobacter, Pseudomonas, and Halomonas showing the highest logFC values. Comparisons involving later incubation time points show fewer significant changes, suggesting reduced community dynamic after the initial response. Similarly, in deep sediments (Fig. 4C), hydrocarbon-degrading genera exhibit marked changes primarily within the same 4–12 month window, whereas comparisons at 6 and 20 months reveal more limited variation. This consistency across sediment depth supports the presence of a defined temporal window in which the succession of PHDB is most pronounced.

Microbial interactions networks were inferred using MetaMIS software based on microbial community profiles and a consensus network was generated for each site across all time points (Supplementary Material), resulting in 1.1 M interactions per site on average. Due to the amount of interactions, only the top 0.01% were retained for visualization and interpretation. Focusing on the genera highlighted in Fig. 4, two taxa were consistently identified in all sites as main interaction hubs. We found that Colwellia presented the most interactions with other bacterial genera in 10 of the 12 sites (except for sites S01 and S27) acting as the main interaction hub, where their positive and negative interactions were a balance between mutualism, parasitism/predation and competition. The shift between these three interactions is variable depending on the site. In addition Thioprofundum a sulfur-oxidizing bacteria genus was present as a main interaction hub in 5 of the 12 sites with the second most interaction amount (S01, S04, S09, S27 and S40), with a positive and negative interactions balance with site dependent shifts between mutualism, parasitism/predation and competition.

Hydrocarbon profiles were analyzed in sediments collected from shallow and deep ocean waters at 0 and 24 months of incubation at 4 °C and statistically evaluated (Table 1). In general, the hydrocarbon degradation capacity of bacterial consortia did not differ between shallow- and deep-water sediments, with some exceptions; measured short-chain alkanes (C11–C13) and the polycyclic aromatic hydrocarbon 2,3,5-trimethylnaphthalene were not significantly degraded in shallow-water sediments, whereas the long-chain alkanes tetracosane (C24), hexacosane (C26), and nonatriacontane (C39) remained unaltered in deep-water sediments.Table 1. Hydrocarbon content determined by gas chromatography in shallow- and deep-water sediments from the southern Gulf of Mexico (sGoM)Hydrocarbon fractionShallow 0 mShallow 24 mΔ%Deep 0 mDeep 24 mΔ%Undecane (C-11)33.235.57.038.635.1− 9.0Tridecane (C-13)50.751.61.962.351.8− 16.9Tetracosane (C-24)96.889.4− 7.7112.2114.82.3Pentacosane (C-25)147.292.5− 37.1149.111.6− 92.2Hexacosane (C-26)97.187.4− 10.0101.6101.90.2Heptacosane (C-27)244.7152.4− 37.7193.8191.9− 1.0Nonatriacontane (C-39)130.135.8− 72.5000.0Naphtalene4.43.7− 16.12.52.3− 5.6Biphenyl55.554.8− 1.239.213.9− 64.62,3,5-trimethylnaphthalene4.34.31.64.54.1− 9.5Triphenylene2.30− 1000.70− 100Chrysene1.50− 1000.70− 100% MO2.42.2− 5.01.61.5− 7%CO1.41.3− 4.00.90.9− 7Values correspond to averages per zone (shallow or deep) and overall means, measured at the start (0 months, 0 m) and after 24 months (24 m) of incubation at 4 °C. Values represent means of all samples analyzed. Δ% denotes the relative percentage change between 0 and 24 months; negative values indicate degradation, while positive values indicate persistence or increase

Discussion

The metabolic capacity of bacterial communities inhabiting marine sediments of the sGoM to degrade petroleum has been extensively documented by several groups, including ours (Bacosa et al. 2018; Godoy-Lozano et al. 2018; Kimes et al. 2013; Liu and Liu 2013; Orcutt et al. 2017; Raggi et al. 2020; Ramírez et al. 2020; Rodríguez-Salazar et al. 2021; Sánchez-Soto Jiménez et al. 2018). In the present work, we expand on this knowledge by showing that hydrocarbon-degrading bacterial communities remain detectable but undergo marked shifts in community structure during long-term storage in a closed environment.

Although site depth initially influenced community structure, its effect diminished over time. Nonetheless, microbial communities from shallow and deep sediments in the sGoM remain structurally distinct (Fig. 2A), indicating that community composition is strongly associated with ocean depth (Godoy-Lozano et al. 2018), and they are clearly defining the hydrocarbon degradation processes differently. The observed differences in bacterial composition can be associated with temporal shifts over the course of the experiment in both shallow- and deep-water sediments (Fig. 2B, 3, and Supplementary Material). Gamma-, Delta-, and Alphaproteobacteria (Fig. 3) were the most abundant groups in both types of sediments; however, the trends -such as the increase in alpha- and gammaproteobacteria alongside the decline of deltaproteobacteria- suggest metabolic specialization as incubation progresses. Similar microbial succession patterns have been observed in natural marine environments following hydrocarbon perturbations (Vigneron et al. 2023; Bacosa et al. 2018; Kimes et al. 2013; Liu and Liu 2013; Orcutt et al. 2017; Ramírez et al. 2020; Sánchez-Soto Jiménez et al. 2018). Although these successions are often cryptic and shaped by complex microbial networks, it is remarkable that comparable community peaks reappear even in mesocosms. The reproducibility of these patterns indicates that hydrocarbonoclastic bacterial succession is ecologically robust and reinforces the importance of continued investigation of microbial succession through controlled experimental approaches. The analyses presented here aim to characterize temporal changes in bacterial community composition during long-term sediment storage. While these patterns provide valuable ecological context, the experimental design and data resolution do not support mechanistic inference, which will require targeted functional and genome-resolved approaches in future studies.

Bioremediation processes can occur under both aerobic and anaerobic conditions. However, in marine sediments, oxygen is rapidly depleted within a few millimeters below the surface due to intense microbial respiration (Revsbech et al. 1980). As a result, hydrocarbon degradation proceeds at relatively low rates, primarily through anaerobic respiration and fermentative metabolism. In this context, we observed distinct temporal patterns in the abundance of several bacterial genera during the incubation of sGoM sediments collected from shallow and deep waters (Fig. 4). Members of the genera Oleispira and Colwellia displayed peak abundance at 4 months, while Shewanella peaked at 6 months. Marinobacter, Alcanivorax, Pseudomonas, and Cycloclasticus reached their highest abundance between 6 and 12 months of incubation. These genera are recognized as key hydrocarbon degraders and, although typically present as a minor fraction of marine bacterial communities (Atlas and Hazen 2011), they are known to rapidly bloom following exposure to oil (Bacosa et al. 2018; Godoy-Lozano et al. 2018; Gutierrez et al. 2013; Kimes et al. 2013; Liu and Liu 2013; Orcutt et al. 2017; Rodríguez-Salazar et al. 2021; Sánchez-Soto Jiménez et al. 2018; Ramírez et al. 2020). These temporal dynamics likely reflect the sustained availability of hydrocarbons or their degradation byproducts as carbon sources in the sediments.

Importantly, pairwise differential abundance analyses (Fig. 4B–C) provide quantitative support for these successional patterns, showing that the most pronounced community shifts occur within a defined temporal window between 4 and 12 months. During this interval, differentially abundant genera are predominantly associated with hydrocarbon degradation, whereas most non-degrading taxa remain relatively stable throughout the incubation. Several taxa exhibit large effect sizes, with logFC values exceeding six-fold in some cases, including Roseobacter and Spongiispira, indicating abrupt rather than gradual changes in abundance. Together, these results suggest that long-term confinement primarily amplifies functional shifts within specialized hydrocarbon-degrading guilds rather than inducing broad restructuring of the entire bacterial community.

The magnitude of the observed logFC changes further suggests that certain taxa may transiently dominate key metabolic steps during hydrocarbon degradation, potentially reflecting shifts in substrate availability or the accumulation of intermediate compounds. For example, Alcanivorax spp., which are specialized in the degradation of aliphatic hydrocarbons such as alkanes, and Marinobacter spp., commonly associated with the degradation of aromatic hydrocarbons, may become differentially abundant at distinct stages of the incubation. However, given the exploratory nature of this study, these interpretations remain descriptive and warrant future genome-resolved and functional validation.

In addition, microbial interaction analysis results suggest that Colwellia genus consistently exhibits a high number of inferred interactions across sites, supporting its potential ecological relevance in the metabolism of hydrocarbons in cold marine environments such as the analyzed samples in this work. Its interactions were a balance between mutualism, parasitism/predation and competition, but the shift between these three interactions is variable depending on the site (Supplementary Material, Microbial interaction networks inferred by MetaMIS) suggesting that the co-occurrence of other bacteria and the hydrocarbons present in each site, are not the only variables that could explain these ecological interactions.

Notably, Kordiimonas was the only genus to peak at 20 months. Although not commonly considered a prominent oil-degrading bacterium, some strains such as Kordiimonas gwangyangensis have been reported to degrade high-molecular-mass polycyclic aromatic hydrocarbons, including benzo[a]pyrene and pyrene (Kwon et al. 2005), suggesting a potential role in the later stages of microbial succession.

Indeed, strains of Oleispira antarctica isolated from Antarctic coastal marine environments have been shown to preferentially degrade aliphatic hydrocarbons (Yakimov et al. 2007), a trait typical of marine hydrocarbonoclastic bacteria such as Alcanivorax and Marinobacter (Yakimov et al. 2007). However, unlike Oleispira, both Alcanivorax and Marinobacter are capable of degrading a broader spectrum of hydrocarbon species, similar to what has been reported for Pseudomonas and Shewanella (Al-Mailem et al. 2013; Gutierrez et al. 2013; Hara et al. 2003; Li et al. 2024; Neethu et al. 2019; Whyte et al. 1997). These successional patterns in hydrocarbonoclastic communities are reflected in the capacity of the bacterial consortia to degrade hydrocarbons (Table 1). The chemical analysis of sediments confirmed that not all hydrocarbon fractions were equally degraded after 24 months of incubation. Short-chain alkanes (C11–C13) remained largely unchanged in shallow sediments, whereas long-chain alkanes such as tetracosane (C24), hexacosane (C26), and nonatriacontane (C39) persisted in deep sediments. This pattern suggests a differential capacity of bacterial consortia to degrade hydrocarbons depending on chain length and solubility. The near-complete removal of aromatic hydrocarbons such as triphenylene and chrysene (Δ% = −100) indicates that certain taxa, including Cycloclasticus and Colwellia, may have contributed to their degradation, consistent with their known specialization in polycyclic aromatic hydrocarbons (PAH) metabolism (Gutierrez et al. 2013; Yakimov et al. 2007). By contrast, the limited degradation of alkanes with high molecular weight highlights the metabolic challenge posed by more recalcitrant compounds, which may require longer incubation times or syntrophic interactions for efficient breakdown (Head et al. 2006). Taken together, these results provide new insights into the dynamic succession, resilience, and functional potential of specialized hydrocarbon-degrading taxa in sediments from both shallow and deep waters of the sGoM and underscore the critical role of microbial community restructuring in sustaining long-term hydrocarbon degradation in marine sediments.

Non-hydrocarbon degraders (chemolithoheterotrophic or chemolithoautotrophic bacteria) remain at high abundance during the whole storage time, showing prevalence of sulfate-reducing bacteria Desulfovibrio spp. and Desulfonatronum spp. and their synergistic companions, the sulfur oxidizing Thioprofundum spp. This indicates a complex and probably active sulfur metabolism, as already observed in previous analysis of the sGoM (Raggi et al. 2020). Sulfur metabolism is active thanks to the non-limiting presence of SO₄ and the sulfur-containing hydrocarbons. Sulfur metabolism has long been associated with hydrocarbon degradation processes (Borgne and Ayala 2010; Zhao et al. 2025), even finding sulfur-oxidizing chemoautotrophs degrading hydrocarbons (Wang et al. 2020). Microbial interaction network analyses indicate that Thioprofundum genus ranked second in the number of inferred interactions (top 0.01%) in 5 of the 12 analyzed sites, although its presence and abundance is quite constant throughout all the sites. Despite having a variable shift between mutualism, parasitism/predation and competition, these interactions do not affect their presence or abundance in all sites, suggesting an ecologically versatile role and metabolism adapted to sulfur-oxidizing and hydrocarbon-rich environments. Comparison of metaMIS networks revealed contrasting topological patterns between shallow and deep-sea communities. Shallow networks were characterized by a single dominant central hub, whereas deep-sea networks exhibited a more distributed architecture with multiple hubs. These differences suggest distinct community organization and potentially different metabolic interaction structures between depth regimes, a pattern further supported by the differential degradation of hydrocarbon compounds discussed above. Environmental conditions associated with depth—such as low temperature, high hydrostatic pressure, and limited energy availability—are expected to further shape microbial selection and succession in situ.

This time-traced study reveals successional patterns and helps identify different microbial genera that emerge at various stages of hydrocarbon degradation, offering deeper insights into microbial adaptation and community dynamics over time. Although the “bottle effect” is often criticized for distorting natural microbial community dynamics by introducing artificial selection pressures (Hammes and Egli 2010), it can be strategically leveraged to enrich and isolate microorganisms with specialized metabolic functions. In particular, when hydrocarbons are present or added to confined systems, selective pressure favors the proliferation of hydrocarbon-degrading microorganisms, which can become highly adapted and efficient over time (Yakimov et al. 2007). Such enrichment techniques have been widely used to isolate potent biodegraders for environmental applications, especially in marine systems where natural hydrocarbon degraders may be present in low abundance but respond rapidly under selective conditions (Head et al. 2006). While the resulting community may only partially reflect in situ microbial diversity, the bottle environment serves as a useful tool for functional screening and the discovery of robust microbial strains for bioremediation.

Supplementary Information

Below is the link to the electronic supplementary material.Supplementary file1 (PNG 1300 KB)—Changes in bacterial diversity over time. The Venn diagrams displayed at the top of the figure depict the shared genera between shallow and deep-sea sediments incubated for 0, 4, 6, 12, and 20 months, in relation to the previously defined "core" bacterial profile by Godoy-Lozano et al. (2018) for marine sediments in the southern Gulf of Mexico labelled as SwGoM. A significant decrease in Shannon diversity index is observed after 12 months of incubation (* <0.05, ** < 0.01, *** < 0.001, **** < 0.0001).Microbial interaction networks inferred by MetaMIS. MetaMIS input. MetaMIS input consisted of genus-level relative abundance tables for all samples, including formatted tables for each station prepared for analysis in MetaMIS. MetaMIS output. MetaMIS output comprised three components: (i) consensus interaction network figures for each station, inferred from the time-series data using the Lotka–Volterra model implemented in MetaMIS (blue arrows indicate negative and red arrows positive interactions); (ii) figures summarizing the types of inferred interactions for each genus, which are direction-sensitive and represent relative abundance changes between interacting genera (mutualism, competition, parasitism/predation, commensalism, and amensalism); and (iii) a ranked list of genera based on their average relative abundance across samples, including genus IDs to facilitate interpretation of the interaction figures.Supplementary file2 (ZIP 439 KB)Supplementary file3 (ZIP 2234 KB)Supplementary file4 (XLSX 17 KB)—Summary of sequencing quality metrics and alpha diversity indices for all samples. Values include read statistics, richness (Observed, Chao1), diversity (Shannon), sampling coverage, sampling time, station, and group (shallow vs. deep).