Natural Hydrocarbon-Contaminated Springs as a Reservoir of Microorganisms Useful for Bioremediation: Isolation and Multilevel Analysis of Hydrocarbonoclastic Bacteria from the Agri Valley (Southern Italy)

Abstract

This research aimed to characterise hydrocarbonoclastic bacteria isolated from naturally hydrocarbon-contaminated springs and the surrounding soils in the Agri Valley (Southern Italy) and to assess the effectiveness of bioaugmentation using a four-strain microbial consortium for removing hydrocarbons from artificially diesel-contaminated lake waters in mesocosm experiments. Four novel bacterial strains were selected for the experimentation: Gordonia amicalis S2S5, Rhodococcus erythropolis S2W2, Acinetobacter tibetensis S2S8, and Acinetobacter puyangensis S1W1. The four isolates can use diesel oil as their sole carbon source, and some exhibited a relatively high emulsifying capacity and ability to adhere to hydrocarbons. Furthermore, genome analyses revealed the presence of genes associated with the degradation, detoxification, and transport of various contaminants. Mesocosm experiments demonstrated that the bioaugmentation enhanced the capacities of the native lake microbial communities to remove hydrocarbons, although drastic changes in their composition (analysed through Next-Generation Sequencing—NGS) were observed. Taken together, these results suggest that naturally contaminated environments can serve as a valuable reservoir of microorganisms with significant biotechnological potential, particularly in the field of bioremediation. However, a complete understanding of the ability of the isolated bacterial strains to efficiently degrade contaminants requires further research to fully assess their capabilities and limitations across different settings.

1. Introduction

Environmental hydrocarbon contamination is a critical global challenge, with far-reaching implications for ecosystem sustainability [1]. Hydrocarbons, which are ubiquitous and can have long-term toxic effects on living organisms, pose a significant concern for environmental health and the future of Earth’s ecosystems [2,3,4]. These chemical compounds, primarily derived from crude oil and its derivatives, serve dual roles as both essential resources and environmental hazards [5].

Crude oil and its by-products are strategic commodities that drive modern society’s socio-economic and industrial advancements, representing key energy sources for transportation, agriculture, and manufacturing [6,7]. As a result, due to the crucial role of crude oil in shaping various aspects of modern life, global demand for this resource continues to grow steadily [8].

The International Energy Agency (IEA) highlighted this persistent world oil demand in its November 2024 Oil Market Report (OMR), projecting an increase by 920 thousand barrels per day (kb/d) in 2024, reaching a total consumption of 102.8 million barrels per day (mb/d), and nearly 1 million barrels per day (mb/d) in 2025, leading to a total consumption of 103.8 mb/d. These figures demonstrate the enduring reliance on hydrocarbons as a fundamental component of global energy systems [9].

Despite their utility, the extensive exploitation of hydrocarbons has led to significant environmental consequences [10,11,12].

One of the most pressing concerns is oil spills, which can occur due to both natural processes and human activities [13,14]. The environmental impacts of such contamination are diverse, affecting soil, water, air quality, and biodiversity [15]. Consequently, hydrocarbon pollution represents a significant ecological burden, requiring comprehensive, interdisciplinary approaches to mitigate its impact [16].

Addressing these challenges necessitates the development of advanced strategies focused on balancing hydrocarbon utilisation and environmental protection [17,18].

Among the various remediation strategies, physico-chemical approaches such as chemical oxidation and soil washing are costly, invasive, and often associated with undesirable side effects, including the generation of toxic by-products. Biological approaches, in contrast, offer a more environmentally friendly and cost-effective alternative. These strategies leverage the unique capabilities of microorganisms and plant species to remove or degrade persistent and hazardous pollutants [19], presenting a sustainable solution [15,20,21].

Microorganisms play a significant role in maintaining ecosystem stability and can be employed to mitigate oil pollution. For instance, bacteria can adapt to oil-contaminated environments, utilising hydrocarbons as a source of carbon and energy for growth [22].

Specific enzymatic biodegradation mechanisms, in the framework of aerobic or anaerobic metabolism, facilitate their transformation into less harmful by-products or mineralisation and conversion into CO₂, CH₄, and H₂O [23].

Bioremediation strategies may be classified into two principal categories: biostimulation, which enhances environmental conditions (e.g., pH and nutrient availability) to optimise microbial activity [18,24,25], and bioaugmentation, which involves the introduction of specific microbial strains or consortia to improve remediation effectiveness [21,26,27]. Nevertheless, the large-scale application of biological approaches faces several challenges. These include the following: (1) the environmental variability, which may limit the effectiveness of introduced microorganisms [15,21]; (2) the toxicity of contaminants, which may inhibit microbial growth [28]; and (3) the complexity of contaminants, which often requires cooperative interactions between different microbial species, highlighting the need for a deeper understanding of microbial community composition and dynamics [21,28].

Moreover, integrating advanced technologies such as genetic engineering and meta-genomic analysis could significantly enhance the efficacy and sustainability of bioremediation strategies, despite the challenges involved.

The primary aims of this study were to isolate and characterise hydrocarbonoclastic bacteria from naturally hydrocarbon-contaminated springs located in the municipality of Tramutola (Agri Valley, Southern Italy), and to preliminarily assess their potential for remediating polluted environments. This was accomplished through a combination of traditional microbiological techniques, genome analyses, and mesocosm experiments. Despite the challenges posed by oil pollution, indigenous bacteria that have evolved in petroleum-enriched ecosystems—capable of metabolising or degrading hydrocarbons for growth and reproduction—may, in fact, represent promising candidates for the remediation of oil-contaminated sites and the enhancement of purification efficiency [29].

2. Materials and Methods

2.1. Study Area and Sample Collection

A sampling campaign was conducted in June 2022 at two natural hydrocarbon springs, designated as S1 and S2 (S1: 40.3228 N, 15.7598 E; S2: 40.3227 N, 15.7585 E), located within the municipality of Tramutola (Basilicata region, Southern Italy) (Figure 1), to isolate hydrocarbonoclastic bacteria and characterise the native microbial communities. The springs are situated in the Agri Valley, a prominent northwest–southeast oriented intramountain basin that extends approximately 30 km in length and has an average width of 12 km [29]. This valley is geologically significant, as it hosts Europe’s largest onshore oil and gas field, with complex geological structures [30].

Water and soil samples were collected from S1 and S2 springs and surrounding areas for microbial community analysis and isolation of hydrocarbon-degrading bacterial strains. Water samples (2 L) were collected from each spring in sterile 1 L bottles, while 20 g of soil was sampled using sterile spoons and stored in 50 mL tubes. To maintain sample integrity, they were immediately refrigerated and transported to the laboratory.

2.2. Isolation and Genome Sequencing of Hydrocarbon-Oxidising Bacterial Strains

To isolate hydrocarbon-oxidising bacterial strains, a slightly modified version of the method described previously by Rizzo et al. [29] was followed. Aliquots of water samples (10 mL) and soil samples (1 g) were inoculated in 40 mL Bushnell–Haas medium (MgSO₄ 0.2 g/L; CaCl₂ 0.02 g/L; KH₂PO₄ 1.0 g/L; K₂HPO₄ 1.0 g/L; NH₄NO₃ 1.0 g/L; FeCl₃ 0.05 g/L), supplemented with diesel oil at a concentration of 2% v/v as the sole carbon source. The diesel oil was purchased from an Eni Spa petrol station, an Italian multinational corporation operating in the oil and gas sector. Cultures were incubated at 28 °C for 7 days with agitation (100 rpm). The enrichment step was repeated for four cycles. From the first enrichment cycle, 100 μL aliquots of the cultures were spread on Bushnell–Haas (BH) agar medium supplemented with diesel oil (2% v/v) as a carbon source. The colonies grown on the plates were repeatedly streaked over the entire surface of fresh BH agar medium supplemented with diesel oil and finally onto LB agar plates to ensure that pure cultures were obtained.

The isolated hydrocarbonoclastic bacterial strains were initially characterised phenotypically through the analysis of colony morphology, microscopic observations, Gram staining, and catalase production tests. Four strains (named S1W1, S2W2, S2S5, and S2S8), which exhibited distinct phenotypic traits and faster growth in BH medium supplemented with diesel oil (optical density measurements at a wavelength of 600 nm), were selected for further investigation and subjected to 16S rDNA partial sequencing.

The obtained 16S rDNA sequences were then compared with those stored in the GenBank database at the NCBI (National Center for Biotechnology Information) by using the BLAST v.2.13.0 (Basic Local Alignment Search Tool) program.

To obtain a comprehensive genetic characterisation, the genome of the four selected isolated bacterial strains was sequenced using the MiSeq Software v4.0 (Illumina, San Diego, CA, USA) at GenProbio srl, Parma, Italy. Individual genomic libraries were obtained using the Nextera XT preparation kit and loaded into a MiSeq flow cell (MS-102-3003, Illumina) with 500-cycle sequencing and paired-end reads of 250 bp. The MiSeq sequencer was used for sequencing according to the manufacturer’s instructions. The raw reads in FASTQ format obtained from the genome sequencing phase were filtered and assembled using the MEGAnnotator2 pipeline [31]. After quality control, the assembled reads were assigned to taxonomic categories using the taxonomic assignment module integrated in MEGAnnotator2 [31]. Subsequently, the obtained contigs were grouped according to the assigned taxonomy. Accounts smaller than 5000 bp were discarded. The whole-genome sequences decoded in this study have also been deposited in the NCBI Sequence Read Archive under the accession number PRJNA1223154.

2.3. Comparative Genome Analysis of the Selected Hydrocarbon-Oxidising Bacterial Strains

The assembled genomes were then annotated by the MEGAnnotator2 pipeline [31] using the Prokka software v1.14.6 [32] to identify and annotate the gene content of bacterial strains obtained from NCBI, as well as microorganisms isolated from the natural hydrocarbon springs and the surrounding soils. This process also enables the prediction of protein-coding sequences from the genomic data.

Protein sequences shared among all genomes of the same species were selected to reconstruct the core genome of each species under analysis. Protein domains with known functions for each of these sequences were evaluated using the InterPro database. In addition, shared protein families among phylogenetically close strains were selected to reconstruct a common central genome. Specifically, the PanX pipeline (pan-genomic analysis and exploration) [33] was employed to analyse the core genes of species closely related to those isolates in this study. The resulting phylogenetic relationships were visualised using iTOL v6.8.1 (interactive Tree Of Life) [34], an online tool for annotating, managing, and graphically representing phylogenetic trees.

2.4. Prediction of Genes Involved in Toxic Compound Degradation

To explore the arsenal of genes coding for enzymes involved in processes leading to the degradation of toxic compounds, heavy metals, PCBs, and possible transporters of these substances, these structures were searched through a literature search. Gene sequences encoding these enzymes were identified from the MetaCyc database [35] and downloaded from NCBI. Their presence in the genomes selected for this study was evaluated through DIAMOND [36]. The BLASTp function is based on an optimised local database focused on the analysis of coding sequences for enzymes involved in the degradation of toxic compounds. For DIAMOND hits, these options were used: --evalue 1 × 10⁻²⁵, --subject-cover 80, --query-cover 80, --sensitive, and --max-target-seqs 1. The Artemis program was used to evaluate the gene surrounds of the sequences identified through the degradation database, while the genetic similarities of the sequences were checked using the Clustalx 2.1 software [37].

2.5. Emulsification Properties and Microbial Adhesion to Hydrocarbons Assays

The experimental protocol described by Rani et al. [38] was followed for the determination of the emulsifying capacity (CE%) and emulsion index (E%). The isolated bacterial strains were grown in both BH and LB medium containing 2% v/v diesel oil and incubated at 28 °C with agitation (100 rpm). After three incubation days, the cultures were centrifuged at 7500× g for 5 min at 15 °C. The cell-free supernatants (2 mL) were injected into glass test tubes, 200 μL of diesel fuel was added, and tubes were vigorously vortexed for 30 s and left for 10 min. Samples showing an emulsion were selected, and once the emulsion had stabilised, more diesel oil (200 μL) was added; the procedure was repeated until a distinct and clear fraction of non-emulsified diesel oil was observed on the emulsion surface.

The emulsifying capacity (EC%) was calculated using the following formula (Equation (1)):

$$EC={\displaystyle \frac{V_{totaloilused}}{V_{initialacqueousfase}}}\times 100$$

(1)

The emulsion index (EI%) was calculated after the emulsifying capacity test; the change in the emulsion layer was measured after 24 and 48 h (EI₂₄ and EI₄₈, respectively) using the following formula (Equation (2)):

$$EI={\displaystyle \frac{Heightoftheemulsionlayer}{Heightofthetotalmixture}}\times 100$$

(2)

Hydrophobicity was measured following the method of Choundhary et al. [39] with some modifications.

An overnight culture of each isolate (50 mL) (incubated at 28 °C with agitation at 100 rpm) was harvested by centrifugation at 3500 rpm at 25 °C for 10 min. The supernatants were discarded, and the cell pellets were washed twice in PUM buffer and suspended in the same buffer (K₂HPO₄ 22.2 g/L; KH₂PO₄ 7.26 g/L; urea 1.8 g/L; MgSO₄ × 7H₂O 0.2 g/L).

The initial absorbance (A₀) of the bacterial suspensions was adjusted to 0.5 at 600 nm. The bacterial suspensions (5 mL) were transferred into tubes containing diesel oil (200 µL) and incubated at 28 °C for 10 min; they were then vortexed for 2 min. The suspensions were kept undisturbed for 1 h for phase separation.

Finally, the aqueous phase was carefully removed using a Pasteur pipette, and the final absorbance (A₁) was recorded at 600 nm using a spectrophotometer. The percentage of hydrophobicity was calculated using the following formula (Equation (3)):

$$H=1-({\displaystyle \frac{A_0-A_1}{A_0}})\times 100$$

(3)

2.6. Water and Soil DNA Extraction and 16S rDNA Gene Sequencing

Spring water and soil samples collected at springs S1 and S2 were used for microbial community analyses. Water samples (1 L) were filtered through sterile mixed cellulose ester membranes (S-Pak^TM Membrane Filters, 47 mm in diameter, 0.22 μm pore size, Millipore Corporation, Billerica, MA, USA) within 24 h of collection.

DNA extractions from filters and soils were performed using the DNeasy PowerSoil Pro Kit (QIAGEN, Hilden, Germany), following the manufacturer’s instructions.

To evaluate DNA integrity and quantity, 2 μL of extracted DNA was visualised on a 1% agarose gel (w/v). Gel electrophoresis was conducted in 0.5× TAE buffer at 100 V for 40 min, using a 1 kb DNA ladder as a molecular weight marker (Thermo Fisher Scientific, Waltham, MA, USA).

The microbial community profiles in the samples were generated using Next-Generation Sequencing (NGS) technologies at the BMR Genomics Srl Laboratory, according to Caprari et al. [40,41]. Partial 16S rDNA gene sequences were obtained from the extracted DNA through polymerase chain reaction (PCR) using the primer pair Pro341F and Pro805R, targeting both the bacterial and archaeal V3–V4 hypervariable regions [42] and evaluated in 1.5% agarose gel. The PCR reaction was conducted as follows: 1 min of denaturation at 94 °C, 25 cycles of denaturation for 30 s at 94 °C, annealing for 30 s at 55 °C and extension for 45 s at 68 °C, and a final phase of 7 min at 68 °C.

The amplicons were purified with Thermolabile Esonuclease I (NEB), diluted 1:2, and amplified with short cycle with Nextera XT Index. The amplicons were normalised with SequalPrep (Thermo Fisher Scientific, Waltham, MA, USA) and multiplexed. The pool was purified with Agencourt XP 1X magnetic beads, loaded onto MiSeq and sequenced with V3 Chemistry—300 paired-end strategy.

In silico analysis was performed using QIIME2 tools version 2021.4. [43]. The obtained reads were cleaned of primers by using the Cutadapt software (v. 2021.4) and then processed with the denoised paired plugin of the DADA2 software v1.26 [44]. Briefly, sequences have been trimmed at the 3′ end and filtered by quality and length. Then, they were de-replicated and merged to obtain unique sequences, and chimeras were eliminated. The Amplicon Sequence Variants (ASVs) with a frequency <0.01% were not considered. All reads from bacterial communities were classified to the lowest possible taxonomic rank using a reference dataset from the SILVA database (v. 138) [45]. Alpha-diversity was calculated by using the Shannon index [46].

The 16S rDNA gene sequences generated in this study have been deposited in the NCBI Sequence Read Archive under the accession number PRJNA1219130.

2.7. Bacterial Consortium and Mesocosm Experiment Setup

A bacterial consortium was assembled by combining the four isolated hydrocarbon-oxidising bacterial strains, following the methodologies described by Yu et al. and Chen et al. [47,48], with modifications, in order to assess the capabilities of the analysed microorganisms to efficiently remove hydrocarbons from freshwater collected from Castel San Vincenzo Lake (Molise region, Southern Italy; Figure 2) and artificially contaminated with diesel oil, and the impact of the bioaugmentation process on the native lake microbial communities. Each strain was individually cultured in LB medium (50 mL) for 16 h; cells were harvested by centrifugation at 7000× g for 10 min, washed, and resuspended in sterilised lake water. Each bacterial suspension was adjusted to an optical density (OD₆₀₀) of 1.0. Subsequently, equal volumes of the suspensions were mixed to constitute the hydrocarbon-degrading consortium (10 mL), which has been used in the mesocosm experiments.

Water samples (15 L) were collected from Castel San Vincenzo Lake in sterile bottles and used to determine the main physical-chemical features (temperature, pH, conductivity, total hydrocarbon concentration, and major ions) and the mesocosm setup.

In detail, pH and electrical conductivity was measured using multiparameter instruments (Hanna Instruments Edge, Woonsocket, RI, USA). The determination of inorganic anions and cations was carried out through ion chromatography (ICS-1000 Thermo Fisher Scientific, Waltham, MA, USA). For the total hydrocarbon determination, the sample was acidified (HCl 1:1, pH ≤ 2) and extracted with a mixture of n-hexane (80%) and methyl-tert-butyl ether (20%). The extract, obtained via liquid–liquid extraction in a separatory funnel, was passed through a column packed with silica gel to remove any polar substances. The eluate was collected in a container and evaporated, and the residue was weighed gravimetrically (IRSA-APAT, 2004; [49]).

Mesocosms (1 L) were prepared in 2 L conical flasks (Figure 3), according to the following experimental plan: (1) a “bioaugmentation” mesocosm comprising lake water contaminated with diesel oil (2% v/v) and supplemented with the constructed bacterial consortium (2% v/v), (2) a “natural attenuation” mesocosm containing only lake water and diesel oil, and (3) a “negative control” mesocosm consisting of sterilised lake water contaminated with diesel oil to account for hydrocarbon leaks due only to abiotic factors. Each mesocosm was prepared in duplicate and incubated under static conditions at 28 °C for 28 days.

After the incubation period, molecular analyses were conducted to characterise microbial communities, following the protocol described in Section 2.6.

The biodegradation efficiency of diesel oil was calculated using the following formula (Equation (4)):

$$E={\displaystyle \frac{C_0-C_t}{C_0}}\times 100$$

(4)

where E represents the biodegradation efficiency (%), C₀ is the initial diesel oil concentration (mg/L), and C_t is the concentration at the final time (mg/L) [47].

3. Results

3.1. Isolation, Characterisation, and Genome Sequencing of Hydrocarbonoclastic Bacteria, Along with Microbial Community Analysis of Spring Waters and Surrounding Soils

After four enrichment cycles in Bushnell–Haas (BH) medium supplemented with diesel oil as the sole carbon source, four pure cultures of distinct hydrocarbon-oxidising bacterial strains exhibiting different phenotypic traits were selected and considered for genome sequencing and further investigation. The strains were named S1W1, S2W2, S2S5, and S2S8. Strains S1W1 and S2W2 were isolated from S1 and S2 spring water, respectively, whereas strains S2S5 and S2S8 were isolated from the surrounding soil of the S2 spring.

Strain S1W1 was a Gram-negative bacterium with a coccobacillary morphology, forming translucent colonies on the LB agar plates. Strain S2W2 was a Gram-positive coccobacillus, producing creamy white colonies on LB medium. Strain S2S5 was Gram-positive with rod-shaped cells, producing pink colonies on LB agar plates, while strain S2S8 was a Gram-negative short rod-shaped bacterium, producing light yellow colonies. All the isolates were catalase-positive. The partial sequencing of the 16S rDNA gene, conducted for preliminary taxonomic assignment, revealed that the strains S2S8 and S1W1 belong to the genus Acinetobacter, and S2W2 and S2S5 belong to the genera Rhodococcus and Gordonia, respectively.

The data obtained from the genome sequencing of the four hydrocarbonoclastic bacteria were processed through the MEGAnnotator2 pipeline [31] to reconstruct their genomic sequences. A taxonomic assessment was performed using MEGAnnotator2, which involved a thorough analysis of the complete 16S rDNA gene and whole-genome average nucleotide identity (ANI). This analysis assigned strain S2S5 to Gordonia amicalis (ANI score of 98%) and strain S2W2 to Rhodococcus erythropolis (ANI score of 99%) (

Table 1. Comprehensive genomic and taxonomic characterisation of hydrocarbon-degrading bacterial strains S1W1, S2W2, S2S5, and S2S8. The table details sequencing metrics, genome assembly parameters, completeness, contamination levels, average coverage, and key genomic features, including genome length and gene counts.

	S2S5	S2W2	S2S8	S1W1
Sequencing output:	1,172,742	785,928	1,115,851	1,002,674
High quality reads:	1,132,596	756,097	1,085,988	970,118
Filtered reads:	1,132,594	756,087	1,085,987	970,117
Contigs generated at:	K127	K127	K127	K127
16S rRNA gene identity:	Gordonia amicalis JCM 11271 100.000	Pseudomonas viridiflava 100.000	Bacillus mycoides 98.813	Acinetobacter soli 98.046
ANI screening:	Gordonia amicalis 98.4177	Rhodococcus erythropolis 98.6463	Acinetobacter kanungonis 85.9859	Acinetobacter populi 91.6293
Genome completeness:	99.76	99.94	99.73	100.00
Genome contamination:	0.3	0.09	0.96	0
Average coverage:	110,7166	58,0149	152,2452	120,9935
Number of contigs:	110	32	54	83
Genome length:	5,030,312	6,366,085	3,445,643	3,858,766
Number of genes:	4629	5885	3269	3625
Number of rRNA genes:	5	6	5	6
Number of tRNA genes:	49	54	67	59
Certain taxonomical level:	Gordonia amicalis	Rhodococcus erythropolis	Acinetobacter sp.	Acinetobacter sp.

). Conversely, for strains S2S8 and S1W1, their classification within the genus Acinetobacter was established solely through ANI analysis (Table 1). Additionally, a quality control assessment was conducted on the newly assembled genomes using CheckM, revealing a completeness level of 99.8% for G. amicalis strain S2S5, 99.9% for R. erythropolis strain S2W2, and 99.7% and 100% for Acinetobacter strains S2S8 and S1W1, respectively. The genome lengths of these strains were comparable to other strains of the same species available in the NCBI repository. Gene prediction analyses indicated that the genomes of strains S2S5, S2W2, S2S8, and S1W1, encompass 4629 genes (including 5 rRNA and 49 tRNA genes), 5885 genes (including 6 rRNA and 54 tRNA genes), 3269 genes (including 5 rRNA and 67 tRNA genes), and 3625 genes (including six rRNA and 59 tRNA genes), respectively. Detailed genomic features for these strains are summarised in Table 1. Based on these findings, the reconstructed genomes of strains S2S5 and S2W2 met all quality criteria necessary for subsequent large-scale comparative genomic analyses.

The NGS analysis provided detailed information about the composition of microbial communities in spring waters and soil samples.

The rarefaction analysis, a measure used to estimate alpha-diversity in the samples and assess whether sequencing captured microbial diversity, indicated that the soil and water samples from spring S2 exhibited relatively higher biodiversity than those from spring S1 (Supplementary Figure S1). The Shannon index values for S2-water and S2-soil samples were 6.56 and 7.11, respectively, whereas those for S1-water and S1-soil samples were 4.30 and 5.84, respectively.

As shown in Figure 4a, the main three bacterial phyla found in S1 spring water were Campilobacterota (46.95%), Bacteroidota (21.12%), and Proteobacteria (18.16%), collectively representing 86.23% of the prokaryotic community. In contrast, the microbial community of the S1 spring’s surrounding soil was dominated by Proteobacteria (80.6%), followed by Firmicutes (5.62%), and Actinobacteriota (4.35%). High relative abundance values for the phylum Proteobacteria were also observed in the water and soil samples from S2 (65.82% and 41.23%, respectively). In S2 spring water, Proteobacteria were followed by Bacteroidota and Firmicutes (12.46% and 6.55%, respectively), while in the surrounding soil, Actinobacteriota (21.42%) and Bacteroidota (10.40%) were the second and third dominant phyla. At a lower taxonomic level (Figure 4b), the bacterial genera Sulfurovum (44.72%), Chlorobium (20.02%), and Thiovirga (10.01%) were dominant in S1 spring water, whereas Malikia (12.30%), C39 (6.26%), and Novosphingobium (5.35%) were the most abundant genera in S2 spring water. In contrast, the top three genera representing microbial communities of soils were Acidosoma (21.26%), KCM-B-112 (10.72%), and Pseudoxanthomonas (7.40%) for samples collected near the S1 spring, and KCM-B-112 (13.25%), Mycobacterium (8.63%), and a genus within the Thermaceae family (5.71%) for samples collected near the S2 spring.

Microorganisms belonging to the domain of Archaea were found in relatively low abundances, ranging from 0.61% in soil collected near the S1 spring to 3.56% in the S2 spring water. The phylum Nanoarchaeota was the most represented in both spring water samples, while Thermoplasmatota was the most abundant archaeal phylum in the soils.

3.2. Comparative Genome Analysis of the Four Isolated Hydrocarbonoclastic Bacterial Strains

To achieve a comprehensive overview of the genetic and evolutionary differences within strains of the Gordonia amicalis and Rhodococcus erythropolis species, all the publicly available genomes were retrieved from the NCBI repository. In detail, 12 strains belonging to G. amicalis and 51 strains of R. erythropolis were recovered (Supplementary Tables S1 and S2). Subsequently, all available genomes belonging to G. amicalis, including the newly isolated S2S5 strain, were subjected to ANI analysis to obtain information on the genetic variability of the species based on whole genome comparisons. The same investigation was performed for the 52 genomes of the R. erythropolis species (including the S2W2 strain isolated from naturally hydrocarbon-contaminated spring water). The obtained ANI results highlighted the presence of two main genomic clusters of G. amicalis strains, as shown by hierarchical clustering analysis (HCA) in Figure 5.

In particular, the first cluster (cluster 1-A) encompassed six strains, including strain S2S5, and the second cluster included seven strains (cluster 1-B) (average ANI score of 98.8). Interestingly, the G. amicalis S2S5 strain had a higher nucleotide identity with strains isolated from similar environments (hydrocarbon-contaminated or uncontaminated soil) rather than strains isolated from other sources (water or oil). This result suggests a potential genetic correlation among strains inhabiting similar habitats, hinting at possible specific adaptations of these microbes to these environments.

On the other hand, the ANI analysis performed on the 52 genomes of the R. erythropolis species (including the S2W2 strain isolated from naturally contaminated hydrocarbon spring water) allowed the identification of a larger cluster encompassing 38 strains, in which the newly isolated strain S2W2 is included (cluster 2-A), as well as a smaller cluster comprising 14 strains (cluster 2-B) (Figure 6). Furthermore, the ANI analysis again revealed in the whole a higher percentage of nucleotide identity among strains isolated from similar environments, such as contaminated or uncontaminated soils. However, it is noteworthy mentioning that some strains of R. erythropolis isolated from contaminated soils (such as R. erythropolis IEGM 1415, KB1, and NRRL B-16352) showed a lower percentage of nucleotide identity compared with those isolated from other environments. This discrepancy may be ascribed to the unspecified chemical contaminant in their isolated source, which may exert different ecological pressures on the evolution of the bacterial strains.

For strains S2S8 and S1W1, whose membership in the bacterial genus Acinetobacter had already been determined, ANI analysis was performed by including 91 genomes (Supplementary Table S3), representative of each of the known species of the genus Acinetobacter, to understand the species membership of strains S2S8 and S1W1, which were isolated from naturally hydrocarbon-contaminated soil and spring water, respectively (Figure 7). Specifically, strain S2S8 showed a higher percentage of nucleotide identity with strain Y-23 of Acinetobacter tibetensis (ANI score of 98%). In comparison, strain S1W1 showed a higher percentage of nucleotide identity (ANI score of 96.7%) with strain ANC 4466 of Acinetobacter puyangensis. Consequently, it can be inferred that strain S2S8 belongs to Acinetobacter tibetensis, while strain S1W1 belongs to Acinetobacter puyangensis.

3.3. Assessment of the Unique Genetic Repertoire and Prediction of Genes Involved in Toxic Compound Degradation

The G. amicalis and R. erythropolis genomes were used to perform a comparative genomics analysis to predict the pangenome and core genome of these species, providing a comprehensive view of the genetic peculiarities of the S2S5 and S2W2 strains isolated in this research [50] (Supplementary Figure S2, Tables S4 and S5).

In contrast, no comparative genomics analysis was conducted on the genomes of A. tibetensis (including the S2S8 strain) and A. puyangensis (including the S1W1 strain) due to the low number of strains available for these species, which makes comparative genomics investigations unsuitable.

The analysis of the pangenome of G. amicalis revealed the presence of 2621 gene sequences encompassing the core genome, in addition to 2700 dispensable genes, which were observed only in a subset of the 13 G. amicalis strains. Furthermore, a total of 3645 strain-specific Truly Unique Genes (TUGs) were identified. Of these, 142 were found exclusively in the genome of S2S5, representing the unique genetic features possessed by this newly isolated bacterial strain (Supplementary Figure S2a and Table S4). Thus, the PFAM and InterPro databases were employed to screen the TUGs of strain S2S5 for protein domains with known functions that potentially play a key role in the bacterium’s adaptability, colonisation, and survival mechanisms. Moreover, a BLASTp homology analysis was performed to support the assignment of a putative annotation to each of these genes.

In detail, the S2S5 strain of G. amicalis showed the presence of several genes related to DNA recombination, bacteriophage infection, and heavy metal transport. These genes could be critical for the bacterium’s adaptation to the environment. Specifically, a unique gene in G. amicalis S2S5 was predicted to code for an enzyme encompassing a “Heavy metal-associated domain, HMA” (InterPro accession IPR006121), which has been found in bacterial transporter genes involved in resistance to toxic metals. Consequently, it can be suggested that this gene sequence may be involved in the mechanisms by which strain S2S5 can handle resistance to heavy metals.

The analysis was subsequently extended to assess the existence of unique genes common to all G. amicalis strains belonging to the same phylogenetic clade (cluster 1-A) as our S2S5 isolate, leading to the identification of thirteen distinctive genes unique to this clade (Supplementary Tables S6 and S7). Amongst the predicted protein functions, the identification of the pyridoxamine 5′-phosphate oxidase (pdxH) gene is intriguing, as it is involved in the pyridoxal 5′-phosphate (PLP) biosynthesis pathway. Several studies have shown that PLP is essential for enzymes that degrade toxic compounds, including train-3-chloro-L-aspartate [51], and that increased PLP biomass is found in environments heavily contaminated with polycyclic aromatic hydrocarbons [52]. These findings support the hypothesis that the presence of the PLP-producing enzyme pdxH in the G. amicalis S2S5 strain could play an important role in bioremediation processes.

In parallel, the same analyses were performed on R. erythropolis, leading to the identification of 2384 genes constituting the core genome, 10,692 dispensable genes, and 12,409 TUGs (Supplementary Figure S2b and Table S5). Specifically, 25 TUGs with known protein domains were found in strain S2W2. Among them, a gene encoding a sulfatase was identified, suggesting that strain S2W2 may be capable of metabolising sulphates, providing an adaptive advantage in specific habitats. Additionally, a TadE-like domain typical of extracellular structures involved in adhesion to specific substrates, was identified, although the nature of these substrates cannot be definitively predicted. On the other hand, genome analysis of phylogenetically related strains of R. erythropolis, clustered with strain S2W2 (cluster 2-A), did not reveal any gene sequences that are characteristic and exclusive to this clade.

To better understand the evolutionary relationships among the bacterial strains included in this study, a phylogenetic analysis was conducted by aligning the core genes previously obtained through pangenomic analysis from the G. amicalis, R. erythropolis, and Acinetobacter reference strains. For the G. amicalis species, the strain S2S5 was phylogenetically closely related to strains isolated from contaminated soils. This correlation between isolation sources and evolutionary distance suggests the possible coevolution of these strains, driven by their adaptation to contaminated soil, thus explaining the conserved genomic features related to contaminant degradation and detoxification. Interestingly, R. erythropolis strain S2W2, despite being included in a macro-cluster with other strains isolated from different habitats (soil, contaminated soil, and water), was more closely related to a strain found in contaminated soils, even though it was isolated from spring waters.

To determine whether the S2S5 strain of G. amicalis, the S2W2 strain of R. erythropolis, the S2S8 strain of A. tibetensis, and the strain S1W1 of A. puyangensis could be used efficiently for bioremediation, a degradative database was created. This database included homologues of genes coding for subunits that constitute enzymes involved in the degradation processes of hydrocarbons, their derivatives, and other toxic compounds that may be present in soils. In addition, gene sequences encoding enzymes involved in the degradation and transport of PCBs and heavy metals were incorporated, as their presence could be crucial for successfully utilising a bacterial strain in bioremediation.

Screening of the strains selected for this study revealed that strain S2S5 of G. amicalis, along with other strains of the same species, contains genes encoding subunits that constitute enzymes involved in hydrocarbon degradation (Supplementary Figure S3a and Table S8).

Notably, strain S2S5 has multiple gene sequences coding for the alpha subunit of a hydroxylase involved in the degradation of benzene (an aromatic hydrocarbon). Since the genome of the G. amicalis S2S5 strain possesses only the gene sequence encoding the alpha subunit of this hydroxylase, we manually screened the genes surrounding the sequences coding for the alpha subunit. Intriguingly, we found the presence of gene sequences coding for proteins with unknown functions.

It was then evaluated whether the strains of G. amicalis selected for this study could degrade alkanes. In this regard, gene sequences encoding the three enzymatic subunits of a hydroxylase involved in alkane degradation (alkB system) and a gene encoding the enzyme belonging to the cytochrome CYP153 family, again involved in alkane oxidation, were identified. The G. amicalis strains selected for this study were also found to possess genes encoding components of enzymes involved in detoxification from heavy metals. Notably, strain S2S5 exhibited coding gene sequences related to the detoxification from mercury. Additionally, the genomes of G. amicalis strains contained gene sequences involved in the degradation of other toxic compounds that could be present in contaminated soils, such as tetrathionate, cyanide, acetaldehyde, ethylene glycol, and atrazine, as well as plastic compounds and their derivatives such as nylon, polylactic acid (PLA), and dibutylphthalate (DBP).

Performing the same analysis on R. erythropolis strains selected for this study revealed that, like G. amicalis strains, they encode genes for enzymatic subunits involved in the degradation of benzene, catechols, and 4-nitrobenzoate (Supplementary Figure S3b and Table S9). Additionally, fifty bacterial genomes, including the chromosome sequence of strain S2W2, contain gene sequences encoding one of the components of a reductase complex involved in the degradation process of benzoyl-CoA, an intermediate in the benzene degradation process. This was not detected in the G. amicalis genomes analysed (Supplementary Figure S3a). Furthermore, gene sequences encoding the three subunits (alkB, rubB, and alkG) of a hydroxylase involved in alkane degradation, as well as genes encoding an enzyme from the cytochrome CYP153 family involved in alkane oxidation, were also identified.

Moreover, we also observed that all the genomes of the R. erythropolis strains possess genes encoding a reductase involved in mercury detoxification, and the chromosome sequences of forty-six strains, including strain S2W2, possess genes encoding a cadmium transporter. In addition, genes encoding components of enzymes potentially involved in the degradation of other toxic compounds were also identified in the genomes of the R. erythropolis strains. Specifically, gene sequences encoding components of a reductase involved in the degradation of tetrathionate were identified (whereas in the G. amicalis strains, only the gene sequences coding for the catalytic alpha subunit and gamma subunit were detected). Additionally, gene sequences predicted to encode enzymes involved in the degradation of cyanuric acid, commonly used in the production of herbicides like atrazine [53], and plastic compounds and their derivatives, such as nylon, poly(3-hydroxybutyrate-co-3-hydroxyvalerate) (PHBV), PLA, DBP, polyethylene terephthalate (PET), polyethylene (PE), and polybutylene succinate (PBS), were identified.

The same screening was also performed on the genomes of the Acinetobacter tibetensis and Acinetobacter puyangensis strains selected for this study (Supplementary Figure S3c and Table S10).

This analysis showed that the genome of the A. tibetensis S2S8 strain possesses specific gene sequences predicted to encode components and subunits of enzymes involved in benzene degradation, such as the alpha subunit of a hydroxylase deputed to the degradation of this aromatic hydrocarbon. Moreover, the genome of strain S2S8 contains gene sequences predicted to code enzymes involved in the degradation of benzene derivatives, such as catechol and 4-nitrobenzoate, as well as two components of a reductase involved in the degradation of benzoyl-CoA.

The chromosome sequence of the A. tibetensis S2S8 strain also appears to possess genes predicted to encode enzymes involved in detoxification from heavy metals such as mercury and nickel.

Further screening revealed the presence of genes in the genome of strain S2S8 predicted to code enzymes involved in the degradation of additional toxic compounds, including the gene sequence for one of the subunits of a reductase involved in the degradation of tetrathionate, as well as genes for enzymes involved in the degradation of acetaldehyde, atrazine, ethylene glycol, and plastic compounds and their derivatives (PE, PLA, PHBV, and nylon).

In parallel, strain S1W1 of A. puyangensis has been shown to possess specific genomic sequences predicted to synthetise enzymes involved in the degradation of catechol and benzoyl-CoA. Additionally, the S1W1 strain appears capable of reducing the bioavailability of mercury and nickel in the environment, as well as degrading toxic compounds, such as tetrathionate, acetaldehyde, atrazine, ethylene glycol, and plastic compounds like PE, PHBV, and nylon.

3.4. Assessment of Emulsification Properties and Microbial Adhesion to Hydrocarbons

The assessment of the emulsification capacity (EC) of the four bacterial strains isolated in this research, conducted in both Bushnell–Haas (BH) and Luria–Bertani (LB) growth media, revealed the formation of emulsions in all cases. Notably, Acinetobacter puyangensis strain S1W1 and Gordonia amicalis strain S2S5 achieved EC values approaching 50% in the BH medium. In the LB medium, G. amicalis strain S2S5 exhibited an even higher EC value, reaching 80%. Emulsion stability (EI) in BH medium was observed exclusively for G. amicalis strain S2S5, although a slight decline was noted after 48 h. In the LB medium, all four bacterial strains maintained emulsion stability for 24 h (EI₂₄), with values exceeding 50%.

Interestingly, after 48 h in LB medium, high values of this parameter were still recorded for A. puyangensis strain S1W1 and G. amicalis strain S2S5. In addition to emulsification tests, the ability of the four bacterial isolates to adhere to hydrocarbons was also evaluated. The test of microbial adhesion to hydrocarbons was not performed on Acinetobacter puyangensis strain S1W1, as strong cell aggregation in the form of flocs was observed after the resuspension in the buffer solution used. Acinetobacter tibetensis strain S2S8 demonstrated low hydrophobicity, resulting in minimal adhesion to diesel oil. In contrast, Rhodococcus erythropolis strain S2W2 and Gordonia amicalis strain S2S5 exhibited high cellular hydrophobicity values of 90.62% and 76.00%, respectively, demonstrating a strong capacity to adhere to diesel oil. The results of the different tests are summarised in

Table 2. The table provides an indirect evaluation of the hydrocarbon-oxidising activities of bacterial strains isolated from hydrocarbon-contaminated environments. The assessment includes the following parameters measured in both BH and LB growth media: emulsification capacity (EC), emulsification index at 24 h (EI₂₄), emulsification index at 48 h (EI₄₈), and microbial adhesion to hydrocarbons (MATH).

Hydrocarbon-Oxidising Bacterial Strains	EC % in BH	EI24 % in BH	EI48 % in BH	EC % in LB	EI24 % in LB	EI48 % in LB	MATH %
Acinetobacter puyangensis S1W1	50.00	-	-	30.00	66.67	66.67
Rhodococcus erythropolis S2W2	26.67	-	-	20.00	55.55	-	90.62
Gordonia amicalis S2S5	53.33	53.33	46.67	83.33	100.00	90.48	76.00
Acinetobacter tibetensis S2S8	13.33	-	-	40.00	93.33	20.00	11.11

.

3.5. Construction of a Microbial Consortium and Bioaugmentation of Artificially Hydrocarbon-Contaminated Lake Waters to Assess the Degradative Capacity of the Four Hydrocarbonoclastic Strains in a Different Environmental Context

To assess the potential of the four strains to metabolise hydrocarbons when inoculated into an environment different from that of their isolation, a consortium was established and used to conduct a bioaugmentation in lake waters collected from Castel San Vincenzo Lake, artificially contaminated with diesel, through mesocosm experiments. Before setting up the mesocosms, waters of the Castel San Vincenzo Lake were analysed to determine their main chemical-physical characteristics, assess whether hydrocarbons were present, and investigate the native microbial communities. The results of these investigations are reported in

Table 3. Physico-chemical parameters of Castel San Vincenzo Lake water. The table outlines key environmental factors, including pH, temperature, conductivity, total hydrocarbon concentration, and the concentrations of major anions (Cl⁻, SO₄²⁻) and cations (Na⁺, K⁺, Mg²⁺, Ca²⁺).

Parameters	Values
pH	7.89
Temperature (°C)	23.50
Conductivity (µS/cm)	246.30
Total hydrocarbons (mg/L)	<10.00
Cl− (mg/L)	2.06
SO42− (mg/L)	7.34
Na+ (mg/L)	1.69
K+ (mg/L)	0.68
Mg2+ (mg/L)	4.17
Ca2+ (mg/L)	49.75

and showed that the concentration of total hydrocarbons in the lake waters fell below the detection limit of the method used (10 mg/L).

As previously described, three different mesocosms were prepared: (1) a “bioaugmentation” mesocosm comprising lake water and supplemented with the constructed bacterial consortium; (2) a “natural attenuation” mesocosm containing only lake water; and (3) a “negative control” mesocosm consisting of sterilised lake water. All the mesocosms were artificially contaminated with diesel oil (concentration of 2%). After 28 days of incubation at 28 °C, the total hydrocarbon concentration in each mesocosm was determined, and microbial community composition was analysed in the “natural attenuation” and “bioaugmentation” experiments.

Chemical investigations revealed a loss of 11.84% of hydrocarbons in the control mesocosm, likely due to abiotic processes. In mesocosms augmented with the microbial consortium, a decrease of 30.95% of contaminants was detected. Furthermore, the “natural attenuation” mesocosm demonstrated a 17.85% loss of hydrocarbons, attributable to both abiotic processes and the contribution of native lake water microbial communities. These results suggested that, although the abiotic processes may partially account for the recorded loss of hydrocarbons, native microbial communities can play a key role in contaminant removal. Furthermore, the enrichment with the four hydrocarbonoclastic strains isolated from the Tramutola study site can enhance the degradation efficiency.

The results obtained from the molecular investigations showed, already at the phylum level, notable variations in mesocosm microbial communities compared to non-treated lake waters, likely induced by both the artificial diesel contamination and bioaugmentation, as well as by the experimental conditions applied (Figure 8a). The top three phyla retrieved from Castel San Vincenzo Lake were Proteobacteria, Cyanobacteria, and Bacteroidota, which accounted for 81.02% of the prokaryotic community. A drastic reduction in Cyanobacteria was observed in the “natural attenuation” and “bioaugmentation” mesocosms, as well as a decrease in Bacteroidota, accompanied by an increase in Proteobacteria, with relative abundance values of 81.91% and 66.49%, respectively.

Actinobacteriota represented the second most abundant phylum in the “bioaugmentation” mesocosm, with a relative abundance of 18.43%, higher than that detected in the lake waters and “natural attenuation” mesocosm. Verrucomicrobiota values ranged from 3.54% in the “natural attenuation” mesocosm to 6.19% in lake waters and 7.96% in the “bioaugmentation” experiment. No Archaea were detected.

At the taxonomic level of genus (Figure 8b), the main known taxa found in the lake waters were Cyanobium_PCC-6307 (24.19%), Limnobacter (6.97%), and Dinghuibacter (6.88%). On the other hand, Niveispirillum, Caulobacter, and Arcicella were found among the most represented genera in the “natural attenuation” mesocosm. In the “bioaugmentation” experiment, the genus Azospirillum predominated over the others with a relative abundance value of 42.37%, followed by Rhodococcus (9.21%), Gordonia (7.91%) (genera to which two of the strains constituting the microbial consortium belonged), and Comamonas (7.27%). Interestingly, the genus Acinetobacter (to which the strains S1W1 and S2S8 belonged) that was not detected in the lake water, was recovered with a very low relative abundance percentage (<1%) in the “bioaugmentation” mesocosm. This observation supported the hypothesis that, among the four strains used for the consortium construction, only S2W2 and S2S5 were able to thrive in Castel San Vincenzo Lake water and were responsible for the observed hydrocarbon loss.

The rarefaction analysis revealed a reduction in microbial diversity in the “bioaugmentation” mesocosm (Shannon index value 4.53) compared to lake water (Shannon index value 5.44) (Supplementary Figure S4). In contrast, a Shannon index value (5.48) close to that of lake water was observed for the “natural attenuation” experiment.

4. Discussion and Conclusions

The growing significance of global oil pollution and oil-based products has become a pressing issue in recent decades, driven mainly by rapid population growth, industrialisation, and increased human activities. These determinants have led to the build-up of toxic substances in various environmental systems, posing a significant challenge due to their harmful effects on ecosystems, human health, and economic stability [54]. Remediating hydrocarbon-contaminated sites is particularly difficult and often requires substantial financial resources and time [55].

Biological remediation techniques, which harness the metabolic abilities of microorganisms, have emerged as effective alternatives for addressing hydrocarbon pollution. However, the success of bacterial strains in degrading petroleum hydrocarbons is influenced by a range of interrelated elements across ecological, biochemical, and genetic variables [15]. Understanding these factors is crucial for optimising bioremediation strategies to mitigate the harmful effects of pollutants. This study aimed to isolate and characterise novel bacterial strains from naturally hydrocarbon-contaminated springs (S1 and S2) and the surrounding soils in the Agri Valley (Southern Italy). This site is known for its natural outcrops of hydrocarbons, which have been recognised since the late 19th century [29,56]. Specifically, the hydrocarbon springs S1 and S2 emerge along a fault zone that promotes fluid up-flow, facilitating the mixing of hydrocarbons and gas (mainly CH₄ and H₂S) [29] with the local groundwater, which is primarily fed by the nearby carbonate aquifer. Previous studies [57] have identified several different PAHs, including Naphthalene, Benzo(b)fluoranthene, Benzo(k)fluoranthene, Benzo(g,h,i)perylene, Indene(1,2,3-c,d)pyrene, Chrysene, Anthracene, Phenanthrene, Fluoranthene, benzo(j)fluoranthene, Dibenzo(a,h)anthracene, Pyrene, and Dibenzo(a,l)pyrene in S1 and S2 spring waters and the surrounding soils. Moreover, the presence of H₂S has been confirmed in spring S1. Continuous exposures to hydrocarbons and H₂S have shaped microbial communities, naturally selecting strains capable of biodegrading these pollutants or utilising hydrogen sulphide. NGS analyses provided evidence of this, revealing that in S1 spring water, microorganisms predominantly belonged to the genera Sulfurovum, Thiovirga, and Chlorobium—taxa that include strict chemolithoautotrophs sulphur-oxidising bacteria (SOB) [58,59], as well as microorganisms capable of using reduced sulphur compounds as electron donors for anoxygenic photosynthesis [60]. In the S2 spring and the surrounding soils of the S1 and S2 springs, microorganisms of the genera Malikia, C39, Novosphingobium, Acidisoma, KCM-B-112, Pseudoxanthomonas, and Mycobacterium, including species known for their ability to degrade contaminants/hydrocarbons, were found.

Following a traditional cultivation-based approach, which involved applying selective pressure through diesel oil supplementation, the attention has been focused on four different bacterial strains identified as Gordonia amicalis S2S5, Rhodococcus erythropolis S2W2, Acinetobacter tibetensis S2S8, and Acinetobacter puyangensis S1W1. Members of the genera Gordonia and Rhodococcus are widely distributed in nature across various environments and are known for their remarkable catabolic abilities, allowing them to utilise a broad range of hydrocarbons with diverse chemical structures [61,62]. For instance, Hao et al. [63] showed that Gordonia amicalis strain LH3 could degrade long-chain n-alkanes under both aerobic and anaerobic conditions, while Sowani et al. [64] demonstrated that Gordonia amicalis strain HS-11, a tropical soil isolate, degraded 92.85% of the provided diesel oil after 16 days of aerobic incubation. Additionally, Thi-Mo et al. [65] found that strains X5 and S67 of Rhodococcus erythropolis are capable of degrading hydrocarbons (hexadecane) even in their solid form, achieving 30–40% degradation at low temperatures (10 °C) within 18 days.

Similarly, the genus Acinetobacter includes both biotechnologically relevant species and nosocomial pathogens, with some members capable of degrading diesel oil components [66,67,68], n-alkanes [69,70,71,72], and aromatics [72,73,74].

A study by Dohare et al. [75] demonstrated that the Acinetobacter pittii strain ED1 degraded 68.61% of diesel oil within 7 days, increasing to 90% after 21 days, when grown at 30 °C, pH 7.0, and 1% diesel concentration.

All the strains isolated in this study could grow in BH medium with diesel oil as the sole carbon source, a finding further supported by genome analyses that revealed various genes predicted to be involved in hydrocarbon biodegradation. Additionally, the presence of several other genes encoding putative proteins responsible for heavy metal detoxification and the breakdown of other environmental pollutants suggested that these microorganisms have promising bioremediation potential, not only in environments contaminated by hydrocarbons but also in sites affected by multiple pollutants.

Furthermore, experimental trials conducted on the novel selected bacterial strains unequivocally demonstrated the ability of some of them to enhance the bioavailability of hydrocarbons. This capacity may be attributable to the production of biosurfactants, which are naturally derived surfactants produced by biological entities, particularly microorganisms. These biosurfactants could be utilised as a cost-effective and environmentally friendly method to enhance the bioremediation of oil components [76]. Additionally, the high cellular hydrophobicity exhibited by these strains suggests a significant capacity to adhere to diesel, further contributing to their bioremediation efficacy [77].

The mesocosm experiments—aimed at assessing the biodegradation efficiency of a microbial consortium encompassing the four newly isolated hydrocarbon-oxidising bacterial strains—were performed using water samples collected from Castel San Vincenzo Lake, artificially contaminated with diesel oil. The results of microbiological investigations revealed significant changes in the composition of native microbial communities, driven by the addition of contaminants and the bioaugmentation process. Initially, the native microbial communities were dominated by picocyanobacteria of the genus Cyanobium, small photoautotrophs inhabiting aquatic systems, which play a crucial role in global oxygen production and CO₂ fixation on the planet [78]. However, after 28 days of incubation at 28 °C, the sole artificial contamination with diesel oil reshaped the communities, with a shift towards genera such as Niveispirillum (including nitrogen-fixing bacteria) [79], Caulobacter, and Arcicella, which include species isolated from various sources. The alterations in the composition of the microbial communities were linked to an approximately 6% reduction in total hydrocarbons in the “natural attenuation” mesocosm (excluding the hydrocarbon loss due to abiotic processes, as measured in the control experiment), indicating the presence of hydrocarbon-oxidising strains in the lake microbiota. On the other side, bioaugmentation with the microbial consortium and the addition of diesel oil promoted the increase in microorganisms from the genera Gordonia and Rhodococcus, but particularly in the genus Azospirillum (mainly represented by Azospirillum picis), which includes species isolated from various niches [80], such as the roots of wild plants [81,82], cultured plants [83,84], aquatic environments [85,86,87], and contaminated areas [83,85,88,89,90]. Several studies have shown that Azospirillum strains can degrade xenobiotics and hydrocarbons and tolerate different heavy metals [80]. In our case, a reduction of approximately 19% of total hydrocarbons was observed in the “bioaugmentation” mesocosm (attributable to both the activities of the newly isolated strains and some components of the native lake microbial community), demonstrating that the addition of the microbial consortium enhanced degradation efficiency. Interestingly, NGS analyses revealed very low relative abundance values of representatives of the Acinetobacter genus, indicating that the experimental conditions used to carry out the experimentation may have inhibited the proliferation of microorganisms belonging to this taxon and their degradation activity. However, the exact factors underlying this observation would need to be explored further.

In conclusion, this study has clearly shown that the natural presence of hydrocarbon in the environment can serve as an invaluable reservoir of microbial biodiversity, providing a precious source of new microorganisms with unique biodegradative potentials that can be harnessed for biotechnological applications, such as bioremediation. However, the results also emphasised that, despite the favourable genetic traits and metabolic capabilities, the actual effectiveness of individual strains, as well as microbial consortia applied in environments different from their origin, requires significant efforts to fully understand the factors that influence their proliferation and can promote the contaminant removal process. Furthermore, a careful and detailed analysis of the impact of introducing allochthonous microorganisms on native microbial communities is critical, as these communities may undergo significant variations, and the potential effects on ecosystem equilibrium should be carefully considered.