Biosurfactant-Producing Bacteria Isolated from a Microbial Consortium Previously Subjected to Adaptive Laboratory Evolution in Oily Sludge

Abstract

Microbial bioprospecting in contaminated environments is a promising strategy for identifying biosurfactant-producing bacteria; however, translating environmentally adapted strains into predictable cultivation processes remains challenging. In this study, a microbial consortium subjected to long-term evolutionary laboratory adaptation in oily sludge was investigated to evaluate strain-specific phenotypic responses related to biosurfactant production. Phylogenetic analysis based on 16S rDNA sequencing identified three taxonomically distant isolates: Faucicola sp. strain BS5C, Pseudomonas sp. strain BS16B, and Enterobacter sp. BS14MR. Biosurfactant production was evaluated using a sequential Design of Experiments (DOE) approach, including fractional factorial and central composite rotatable designs, with the emulsification index (E24) used as a semi-quantitative response variable. Initial screening revealed a statistically significant negative effect (p < 0.10) of high dextrose concentrations for all isolates. Strain-specific differences in model adequacy were observed, with a statistically adequate quadratic model obtained for Pseudomonas sp. BS16B (R² = 0.8658, p = 0.0225), whereas the other isolates showed significant lack of fit (p < 0.05). ATR-FTIR analysis revealed spectral profiles consistent with lipopeptide-like compounds. Overall, these results indicate that isolates derived from the same long-term adapted system may differ substantially in process predictability, suggesting that productivity-based screening alone may be insufficient for selecting robust strains.

1. Introduction

Replacing petroleum-derived synthetic surfactants with more sustainable functional equivalents is extremely relevant for modern biotechnology. Therefore, it is safe to say that microbial biosurfactants represent the most promising biomolecules for this purpose [1]. Biosurfactants are amphiphilic compounds that exhibit high biocompatibility, stability, and functional versatility, unlike synthetic surfactants, which can present environmental toxicity and low biodegradability [2]. Both their sustainability and multiple functional capabilities make biosurfactants considered efficient and promising in the surfactant market, with a growing trend toward replacing synthetic surfactants [3]. From an ecological and biotechnological perspective, biosurfactant-producing microorganisms are commonly found in oily sludge, as these hydrocarbon-rich environments exert strong selective pressure favoring microorganisms that develop adaptive strategies to overcome the low bioavailability of hydrocarbons, thereby enabling their utilization as carbon sources. [4].

Even with this potential, large-scale biosurfactant production still faces many challenges associated with high production costs [5]. Low yields from submerged fermentation and the high costs associated with downstream processes make biosurfactants less competitive compared to synthetic equivalents [1,6,7]. Nevertheless, the environmental and health-related costs associated with synthetic surfactant use are not easily quantifiable; however, their broad impacts have driven the search for more sustainable alternatives, such as microbial surfactants [8]. Therefore, to better understand factors influencing biosurfactant production, the optimization of culture media and cultivation conditions has gained increasing attention [9]. Thus, the statistical methodology of Design of Experiments (DOE) is a valuable tool for systematically identifying critical variables and evaluating biological responses in bioprocesses [10,11]. However, despite advances in culture optimization, the stability and predictability of biosurfactant production processes, particularly those involving environmentally derived strains, remain poorly understood and insufficiently addressed in the literature.

The need to use robust strains in industrial production directly confronts the challenge of the innate complexity of biological systems [12]. While conventional microbial strains are already well characterized and adapted for stable and predictable perfomance in bioprocesses [13], microbial isolates from complex environments or cultures subjected to long-term stress conditions may reflect distinct evolutionary trajectories [14]. In this context, prolonged exposure to stressful selective conditions, including during adaptive laboratory evolution processes, reinforces the need to specifically assess strain stability under bioprocess-relevant conditions, as such pressures may favor the emergence of phenotypic diversity as a survival strategy [15]. According to the literature, this may manifest as both increased stress tolerance and undesirable phenotypic instability, even under controlled cultivation [16]. Therefore, it is necessary to evaluate not only the potential for biosurfactant production by new strains, but also the stability of their behavior under bioprocess-relevant conditions, an aspect that remains comparatively less explored from a phenotypic and process-oriented perspective.

Previous studies have extensively investigated biosurfactant production by environmental and engineered microorganisms, often focusing on optimizing cultivation conditions to maximize productivity under controlled settings [7,9,10,17]. However, comparatively little attention has been given to how distinct bacterial isolates originating from the same long-term adapted system respond phenotypically to process-oriented cultivation and evaluation strategies. Thus, the present study aims to evaluate biosurfactant production phenotypes in three bacterial isolates derived from a long-term oily sludge–adapted consortium, focusing on strain-specific responses under controlled cultivation conditions and their implications at the process level. We hypothesize that bacterial isolates derived from the same long-term oil sludge-adapted consortium may belong to distinct microbial genera and display comparable biosurfactant-producing phenotypes, while differing in phenotypic stability and process robustness under controlled cultivation conditions.

2. Materials and Methods

Strains derived from long-term adaptation to oil sludge represent biologically complex systems shaped by prolonged selective pressure, which may result in heterogeneous and strain-specific phenotypic responses. Therefore, an exploratory and process-oriented methodological approach was adopted to systematically assess cultivation responses and process predictability.

2.1. Origin of Microorganisms

In this study, three bacterial isolates from the culture collection of the Microbial Biotechnology Laboratory (LABMI/UESC) were used, namely BS5C, BS16B, and BS14MR. These were obtained from microbial isolation of a consortium subjected to evolutionary laboratory adaptation (ALE) for 18 years in mineral minimum liquid medium (MM) [0.1% (w/v) K₂HPO₄, 0.1% (w/v) NH₄NO₃, 0.05% (w/v) MgSO₄, 0.001% (v/v) saturated FeSO₄ solution, 0.001% (v/v) saturated CaCl₂ solution] using 1% (v/v) crude petroleum as the sole carbon source. In this study, the term adaptive laboratory evolution (ALE) refers to the long-term maintenance of the consortium under defined laboratory conditions with sustained selective pressure for hydrocarbon utilization and biosurfactant production, without implying demonstrated genomic changes or speciation events. The cultures were incubated at 28 ± 2 °C under constant orbital shaking of 150 rpm in a shaker incubator model SL-222 (Solab Científica, Piracicaba, Brazil). This consortium was originally constituted from contaminated soil samples from landfarming units of the Mataripe Refinery (formerly Landulpho Alves Refinery—Petrobras), located in São Francisco do Conde, Bahia [18] and subsequently maintained under controlled laboratory conditions (28 ± 2 °C, 150 rpm) under continuous hydrocarbon selective pressure. The initial taxonomic diversity of the consortium was not determined at the time of establishment, and community composition was not systematically monitored throughout the enrichment period. The isolates characterized in the present study represent cultivable members of the consortium at the time of isolation. The reactivation of the three isolates was performed in rich medium (RM) [0.5% (w/v) meat extract, 1% (w/v) peptone, 0.5% (w/v) NaCl] with incubation at 28 °C and 150 rpm for 24 h.

2.2. Molecular Identification and Phylogenetic Analysis

2.2.1. Extraction, Quantification and Assessment of Genomic DNA Purity

The strains were grown in RM medium as described previously, and the purity of the cultures was verified by Gram staining upon observation of the slides under a Motic B1 Advanced Series optical microscope (Motic, Xiamen, China) with a 100× objective under oil immersion. Genomic DNA was extracted using the EasyPure^® Bacteria Genomic DNA Kit (TransGen Biotech, Beijing, China) according to the manufacturer’s instructions. The samples were quantified, and the purity was assessed by spectrophotometry (Nanodrop ND-1000, Thermo Fisher Scientific, Waltham, MA, USA). Samples with absorbance ratios A260/A280 between 1.8 and 2.0 and A260/A230 between 2.0 and 2.2 were considered pure. Finally, the genomic DNA samples were stored at −20 °C at LABMI/UESC and identified with the same codes as the source isolates (BS5C, BS16B, and BS14MR).

2.2.2. Amplification of the 16S rDNA Gene by Polymerase Chain Reaction (PCR)

Partial amplification of the 16S rDNA region was performed with the universal primers 27F (5′-AGAGTTTGATCCTGGCTCAG-3) and 1525R (5′-AAGGAGGTGATCCAGCC-3), purchased from LGC Biosearch Technologies (Kidlington, UK) as described by Weisburg et al. [19]. The reactions were performed with the Platinum™ Taq DNA Polymerase kit (Invitrogen, Carlsbad, CA, USA), containing: 1 U of Platinum™ Taq DNA Polymerase, 0.2 µM of each primer, 0.2 mM dNTP, 1.5 mM MgCl₂, 1× buffer, and molecular grade water (HyClone, Cytiva, Marlborough, MA, USA), totaling 25 µL per reaction.

The amplification reaction was performed by polymerase chain reaction (PCR) in an Eppendorf Nexus MasterCycler Thermal Cycler (Eppendorf, Hamburg, Germany). The PCR protocol consisted of: initial denaturation at 94 °C for 5 min; 14 cycles of 94 °C for 30 s, 65 °C for 40 s, 72 °C for 1 min; followed by 20 cycles of 94 °C for 30 s, 50 °C for 40 s, 72 °C for 1 min; with a final extension at 72 °C for 7 min. The amplified products were stored at −20 °C until the purification step.

2.2.3. Electrophoresis, Purification and Preparation for Sequencing

The PCR fragments were analyzed by 0.8% agarose gel electrophoresis in 1× TAE buffer (Invitrogen) and revealed with SYBR™ Safe DNA Gel Stain (Invitrogen) by a Kodak Edas 290 Ultraviolet Transilluminator (Eastman Kodak Company, Rochester, NY, USA) UV photodocumentation system. Purification was performed by precipitation with 96% (v/v) ethanol according to the protocol adapted from Dorado Pérez [20]. The purified products were oven-dried (37 °C), resuspended in 6 µL of ultrapure water (Sigma Aldrich, Saint Louis, MO, USA) and quantified in Nanodrop ND-1000 to prepare sequencing samples (150 ng/5 µL).

2.2.4. Sequencing and Assembly of Sequences

Sequencing was performed by ACTGene Análises Moleculares Ltd.a. (Nova Alvorada, Brazil) using the Sanger method on a 3500 Genetic Analyzer automated sequencer (Applied Biosystems, Carlsbad, CA, USA) with 50 cm capillaries, POP7 polymer, and the BigDye Terminator v3.1 Cycle Sequencing Kit (Applied Biosystems). Labeling reactions used 2.5 pmol of specific primer in a final volume of 10 µL and followed a standard protocol (25 cycles, 96 °C for 10 s, 55 °C for 5 s, 60 °C for 4 min). Products were purified with 75% (v/v) isopropanol and diluted in Hi-Di™ Formamide (Applied Biosystems) before electroinjection. The generated electropherograms (.ab1) were converted into FASTA files using Sequence Analysis Software v6 (Applied Biosystems).

2.2.5. Phylogenetic Analysis

The electropherograms were visually inspected, and the forward and reverse sequences were used to generate a consensus sequence for each isolate using UGENE v52.1 [21] and CAP3 [22] software. A similarity search was performed with the BLASTn tool version 2.15.0. [23], submitting the NCBI reference database (RefSeq RNA) for preliminary identification. For robust phylogenetic analysis, a custom dataset was constructed for each isolate, including: (i) the three type sequences with the highest BLASTn identity; (ii) sequences from other species of the same genus; (iii) sequences from phylogenetically close genera; (iv) sequences from species of the same genus with biosurfactant production documented in the literature; and (v) an appropriate outgroup sequence.

For each dataset, multiple alignment was performed with MAFFT v7.505 [24]. Phylogenetic inference was conducted using the Maximum Likelihood (ML) method in IQ-TREE 2 v2.0.7 software [25]. The best nucleotide substitution model for each dataset was determined by ModelFinder based on the Bayesian Information Criterion (BIC), with the TIM3 + F + R2 models selected for the Pseudomonas sp. strain BS16B and Faucicola sp. strain BS5C datasets, and the TN + F + G4 model for the Enterobacter sp. strain BS14MR isolate. The statistical robustness of the tree topology was assessed with 1000 bootstrap replicates. Trees were rooted with Escherichia coli for isolates of the order Pseudomonadales (Pseudomonas and Faucicola) and with Pseudomonas aeruginosa for isolates of the order Enterobacterales (Enterobacter). The final trees were visualized and edited on the interactive platform iTOL (Interactive Tree Of Life) [26].

The nucleotide sequences of the isolates were deposited in GenBank under accession numbers PX219823, PX219872, and PX219883.

2.3. Assessment of Cultivation Conditions Influencing Biosurfactant Production

2.3.1. Design of Experiments (DOE) for Exploratory Evaluation of Biosurfactant Production

The influence of cultivation parameters such as pH, temperature (°C), dextrose concentration (% w/v), and inoculum concentration (% v/v) on biosurfactant production was evaluated using the Design of Experiments (DOE) approach as an exploratory tool to assess strain-specific responses. In the first step, a Fractional Factorial Design (FFD)$2^{4-1}$ matrix was performed for initial screening of the factors, followed by a Central Composite Rotatable Design (CCRD) $2^3$ matrix to further explore the experimental space and evaluate response patterns. The actual and coded values of each factor for each matrix are shown in

Table 1. Experimental ranges and levels of independent variables for the 2⁴⁻¹ Fractional Factorial Design (FFD) and the 2³ Central Composite Rotational Design (CCRD) to evaluate the production of biosurfactants by microorganisms BS16B, BS5C and BS14MR.

	Coded Variable	Dextrose—D [% (w/v)]	ph	Temperature—T (°C)	Inoculum *—I [% (v/v)]
FF	−1	1.0	5.0	20.0	1.0
	0	5.5	7.0	28.5	3.0
	1	10.0	9.0	37.0	5.0
CCRD	−1.681	0.000	8.50	35.0	-
	−1	0.203	9.21	37.6	-
	0	0.500	10.25	41.5	-
	+1	0.797	11.29	45.4	-
	+1.681	1.000	12.00	48.0	-

* The inoculum factor was fixed for the CCRD as I = 5% (v/v).

; to perform the CCRD, the inoculum concentration was set at 5% (v/v), as determined in the analysis of the FFD matrix results. All experiments were performed in triplicate for each isolate, with the emulsifying activity index ($E_{24}$) serving as a semi-quantitative response variable for comparative evaluation.

The statistical analysis of the results obtained was performed with the Protimiza Experimental Design^® software version 1.0. in accordance with Rodrigues and Iemma [11]. A confidence level of 90% was used for factor and coefficient screening to detect trends in this exploratory study, while ANOVA was performed at a 95% confidence level.

The cultures were carried out in a 250 mL Erlenmeyer flask containing 25 mL of culture medium [0.5% (w/v) yeast extract, 0.1% (w/v) (NH₄)₂SO₄, 0.06% (w/v) MgSO₄, 0.3% (w/v) KH₂PO₄, 0.6% (w/v) K₂HPO₄, 0.27% (w/v) NaCl] with addition of dextrose and pH adjustment with 1 M HCl and 1 M NaOH solutions in a benchtop pH meter model PHS3BW (Bel Equipamentos Analíticos, São Paulo, Brazil) according to the values proposed in Table 1; the incubation occurred in a Refrigerated Incubator with Orbital Shaking (Thoth Equipamentos, São Paulo, Brazil) under orbital shaking at 150 rpm and temperature control. The inoculum was standardized by spectrophotometry at 600 nm at an optical density (OD) of 0.135.

Statistical analyses were conducted with an exploratory objective to identify trends and influential factors rather than to generate predictive industrial-scale models, as commonly applied in DOE studies involving complex biological systems.

2.3.2. Emulsification Activity Index (${\mathrm{E}}_{24}$) as a Semi-Quantitative Response Variable

Biosurfactant-related emulsifying activity was evaluated by determining the emulsifying activity index (${\mathrm{E}}_{24}$, %) [27]. For this purpose, equal volumes (4 mL) of cell-free supernatant and kerosene were homogenized by vortexing at maximum speed for 2 min. The emulsion height (h emulsion, cm) and the total height (which encompasses both the emulsion and aqueous phases) (h total, cm) were measured after 24 h and ${\mathrm{E}}_{24}$ was calculated according to Equation (1). SDS (sodium dodecyl sulfate) and distilled water were used as positive and negative controls, respectively, to ensure assay consistency. The ${\mathrm{E}}_{24}$ index was employed as a comparative and screening parameter rather than as a quantitative measure of biosurfactant yield.

$${\mathrm{E}}_{24}={\displaystyle \frac{h_{emulsion}}{h_{total}}}\times 100\mathrm{\%}$$

(1)

2.4. Preliminary Structural Characterization of Biosurfactants

2.4.1. Precipitation of Biosurfactants

The biosurfactants produced by the isolates were extracted using the cold acetone precipitation method [28]. For this purpose, each isolate was cultured in 200 mL of fermentation medium under orbital shaking (150 rpm) in a Refrigerated Orbital Shaking Incubator (Thoth Equipamentos, São Paulo, Brazil) for 72 h. After cultivation, the broths were centrifuged in a Refrigerated Digital Centrifuge TH.9700R (Thoth Equipamentos, São Paulo, Brazil) at 4400× g for 20 min to remove the cells. The supernatants were collected and subjected to precipitation by the addition of ice-cold pure acetone (Êxodo Científica, São Paulo, Brazil) at a ratio of 3:1 (v/v). The mixtures were kept at 4 °C for 15 h to allow the formation of precipitates. The supernatants were then discarded and the precipitates recovered by centrifugation under the same conditions. The precipitated biosurfactants were resuspended in sterile ultrapure water and concentrated in a Concentrator 5301 vacuum evaporation system (Eppendorf, Hamburg, Germany).

2.4.2. Preliminary Characterization of Biosurfactants by FTIR-ATR

To assess the main functional groups present in the produced biosurfactants, FTIR-ATR analysis was performed at the Fungal Biology Laboratory of the Center for Biotechnology and Genetics of the State University of Santa Cruz (CBG/UESC). The concentrated samples were applied directly to the equipment crystal and analyzed in transmittance mode, with spectra recorded in the range of 4000 to 650 cm⁻¹, a resolution of 4 cm⁻¹, and an average of 10 scans per sample. The analyses were performed on a Spectrum 400 FTIR spectrometer (PerkinElmer, Waltham, MA, USA), with spectral interpretation based on literature data for biosurfactant functional groups. FTIR-ATR was used as a preliminary analytical tool to assess major functional groups and infer the general class of biosurfactants produced by the isolates. While suitable for rapid chemical fingerprinting, this technique does not allow definitive molecular identification. Therefore, FTIR-ATR was applied to support class-level comparisons rather than full structural elucidation.

3. Results

3.1. Taxonomic Diversity and Phenotypic Convergence Within the Microbial Consortium of Origin

Phylogenetic analysis revealed taxonomic diversity among the biosurfactant producers studied, identifying isolates BS5C, BS14MR, and BS16B as members of three evolutionarily distant bacterial genera: Faucicola, Enterobacter, and Pseudomonas, respectively. The topology of the maximum likelihood trees corroborates the identifications and demonstrates, in all cases, clustering of the isolates with reference strains of their respective genera, including species that have been reported as biosurfactant producers in the literature (Figure 1).

Isolate BS5C was placed in a monophyletic clade with 92% support with species of the genus Faucicola (Figure 1A), presenting 97.38% 16S rDNA sequence identity with strain Faucicola osloensis A1920 (Supplementary Table S1) and was designated as Faucicola sp. strain BS5C (Accession number: PX219872). Isolate BS14MR grouped with strains of the genus Enterobacter with 62% support (Figure 1B), having greater sequence similarity with Enterobacter quasiroggenkampii (98.31% identity) (Supplementary Table S1) and was designated as Enterobacter sp. strain BS14MR (Accession number: PX219883). Isolate BS16B was inserted with 100% support into a clade with Pseudomonas aeruginosa species, which includes numerous reference strains and well-documented biosurfactant producers in the literature (Figure 1C), presenting 97.24% identity with strain DSM 50071 (Supplementary Table S1) and was designated as Pseudomonas sp. strain BS16B (Accession number: PX219823).

3.2. Analysis of Cultivation Conditions Associated with Biosurfactant Production

The initial screening of the culture conditions [dextrose (% w/v), pH, temperature (ºC) and inoculum (% v/v)] was performed according to Table 1 for a Fractional Factorial Design (FFD)$2^{4-1}$ (

Table 2. Coded Fractional Factorial Design (FFD) 2⁴⁻¹ matrix for the factors: dextrose concentration (D, %, w/w), pH, temperature (T, °C) and inoculum concentration (I, %, v/v) and the response emulsification index ${(E}_{24}, \%)$ for biosurfactant production by the strains BS16B, BS14MR, and BS5C.

Assay	Coded Variables				Results ${\mathbf{E}}_{24}, \mathit{\%}$
	Dextrose—D [% (w/v)]	pH	Temperature—T (°C)	Inoculum—I [% (v/v)]	BS16B	BS14MR	BS5C
1	−1	−1	−1	−1	52.28	3.73	0.00
2	+1	+1	−1	−1	29.73	9.26	15.96
3	+1	−1	+1	+1	4.49	0.00	0.00
4	−1	+1	+1	+1	61.29	45.25	65.84
5	+1	−1	−1	−1	12.61	0.00	0.00
6	−1	+1	−1	−1	59.82	59.98	25.68
7	−1	−1	+1	+1	40.61	30.80	68.19
8	+1	+1	+1	+1	32.36	35.27	53.72
9	0	0	0	0	6.35	16.23	19.33
10	0	0	0	0	4.41	12.97	26.93
11	0	0	0	0	53.52	28.58	11.76

The factors relating values are presented in Table 1 in Materials and Methods section.

) with the emulsification index ($E_{24}$, %) (Figure S1) being the response variable evaluated for the three strains. The results obtained for $E_{24}$ (Table 2) presented a general range of values from 0.00 to 68.19% indicating marked variability in emulsifying activity, with mean values at the central points of: 21.42 ± 27.81% (BS16B), 19.34 ± 7.59% (BS5C) and 19.26 ± 8.23% (BS14MR). The Effects Analysis (Supplementary Table S2) revealed that the factors exerted distinct and strain-specific influences (positive and negative effects) on biosurfactant production. Dextrose concentration was the only variable with a statistically significant negative effect (p < 0.10) indicating a consistent trend toward higher E24 values at lower dextrose levels (D = 1.0% w/v) would be favorable conditions for $E_{24}$, within the experimental range evaluated. The curvature at the central points was not statistically significant (p > 0.10), suggesting an approximately linear response within the explored experimental space. Also, according to the results of the statistical analysis (Supplementary Table S2), higher pH values (9.0) were associated with higher $E_{24}$ values (statistically significant positive effect, p < 0.10) for isolates BS5C and BS14MR. As for temperature, values closer to 37 °C, in turn, would be more favorable for isolate BS5C (due to the statistically significant positive effect, p < 0.10), while the inoculum concentration close to I = 5% (v/v) proved to be the best for isolate BS14MR (statistically significant positive effect, p < 0.10).

Although the inoculum had a significant effect (p < 0.10) for isolate BS14MR (Supplementary Table S2), it did not have a significant effect for the other two strains. This result highlights the strain-dependent nature of the inoculum effect. Previous studies indicate that multiple inoculum-related variables, including age, cell density, and preparation method, can influence metabolic activity, fermentation kinetics, and product profiles [29,30]. Therefore, considering the multiple variables that can influence the inoculum and which were not the focus of this study, the inoculum concentration was fixed at 5% (v/v) to standardize subsequent experiments, as previously reported in biosurfactant studies [31]. Accordingly, new values were defined for the other three factors (D, pH and T) (Table 1). Considering the joint analysis of the three isolates and in order to further explore the cultivation conditions, the Central Composite Rotatable Design (CCRD) $2^3$ was performed. The results obtained (

Table 3. Central Composite Rotatable Design (CCRD) 2³ matrix coded and the resulting emulsification index ${(E}_{24}, \%)$ for biosurfactant production by the strains BS16B, BS14MR, and BS5C.

Assay	Coded Variables			Results ${\mathit{E}}_{24}, \mathit{\%}$
	Dextrose—D [% (w/v)]	pH	Temperature—T (°C)	BS16B	BS14MR	BS5C
1	−1	−1	−1	28.32	47.74	57.88
2	1	−1	−1	16.75	38.29	56.05
3	−1	1	−1	50.31	13.93	42.75
4	1	1	−1	22.38	8.75	3.85
5	−1	−1	1	55.78	41.63	55.53
6	1	−1	1	28.76	58.21	59.24
7	−1	1	1	36.58	38.35	3.56
8	1	1	1	45.54	50.66	3.16
9	−1.681	0	0	60.25	55.99	29.13
10	1.681	0	0	49.64	5.10	3.14
11	0	−1.681	0	17.42	33.00	23.66
12	0	1.681	0	4.49	3.26	5.16
13	0	0	−1.681	43.43	34.73	44.15
14	0	0	1.681	59.65	38.48	9.76
15	0	0	0	47.29	43.71	37.98
16	0	0	0	51.56	38.78	33.31
17	0	0	0	52.52	37.07	40.31

The actual values are presented in Table 1 in materials and methods.

) showed a general range of 3.85 to 60.25% for the $E_{24}$ values and the values of the central points presented average values of: 50.45 ± 2.78% (BS16B), 37.20 ± 3.57% (BS5C) and 39.85 ± 3.45% (BS14MR).

The coefficient analysis for the quadratic models to be fitted to the points of each microbial isolate is presented in Supplementary Table S3; according to the results of this analysis, for the quadratic models of isolates BS5C and BS14MR, only the linear term for temperature (T) was statistically significant (p < 0.10), and the coefficient of determination ($R^2$) of these full quadratic models were, respectively, 0.7133 and 0.6672, respectively. For isolate BS16B, the quadratic terms for temperature ($T^2$) and the linear term for dextrose concentration (D) were statistically significant (p < 0.10), with the $R^2$ of the full quadratic model being 0.8658.

Non-significant coefficients were retained in the quadratic models to preserve the original experimental structure; the ANOVA of the three quadratic models indicated that the regression term was statistically significant (p < 0.05) only for the quadratic model of isolate BS16B, highlighting contrasting model behaviors among isolates. (Supplementary Table S4). The regression terms for isolates BS5C and BS14MR were not statistically significant according to ANOVA with 95% confidence, with high dispersion between predicted and experimental values observed for isolates BS5C and BS14MR (Figure 2A,B), consistent with the lack of fit detected for these models (p < 0.05). In contrast, Figure 2C shows that the quadratic model for isolate BS16B showed lower data dispersion and a non-significant lack of fit (p > 0.05 and Fcalc = 15.252) (Supplementary Table S4) supporting its adequacy to describe the E₂₄ response for this isolate (Equation (2)).

$$\begin{gathered} E_{24-BS16B}\left(\%\right)=50.644-5.521·\left(\mathrm{p}\mathrm{H}\right)+0.941·\left({\mathrm{p}\mathrm{H}}^2\right)+0.253·\left(\mathrm{T}\right)-14.612·\left({\mathrm{T}}^2\right)+5.578·\left(\mathrm{D}\right)-0.263· \\ \left({\mathrm{D}}^2\right)-2.453·\left(\mathrm{p}\mathrm{H}\right)·\left(\mathrm{T}\right)+2.680·\left(\mathrm{p}\mathrm{H}\right)·\left(\mathrm{D}\right)-3.755·\left(\mathrm{T}\right)·\left(\mathrm{D}\right) \end{gathered}$$

(2)

Using the quadratic model for isolate BS16B (Equation (2)), the response surface and contour plots generated and presented in Figure 3 were analyzed. These images indicate that, considering the conditions evaluated, no distinct maximum within the evaluated experimental space was observed; however, ranges associated with higher E₂₄ responses could be identified: pH between 8.5 and 9.5, temperature between 39 and 43 °C, and dextrose concentration of 0.5 and 1.0% (w/v). The theoretical stationary point (dy/dx = 0) calculated from the model corresponded to the following conditions: pH = 7.9, T = 38.6 °C, and D = 1.69% (w/v); indicating a predicted response value of 68.88%. To assess model consistency, these conditions were evaluated experimentally (in triplicate). The average value obtained was 65%, which represents a 5.63% deviation relative to the value predicted by the model. Thus, the quadratic model (Equation (2)) provided a consistent description of the E₂₄ response for isolate BS16B within the evaluated experimental space, allowing the identification of cultivation ranges associated with higher emulsifying activity.

3.3. Preliminary Structural and Spectroscopic Characterization by FTIR-ATR

Spectroscopic assessment of the partially purified biosurfactants, performed by attenuated total reflection infrared spectroscopy (ATR-FTIR), indicated that the three isolates produced compounds with highly similar spectral profiles, compatible with lipopeptide-like compounds (Figure 4). The presence of the peptide moiety was suggested by the detection of absorption bands characteristic of amide bonds, including a broad band in the region of ~3200 cm⁻¹ (N-H stretching, Amide A) and a prominent band at ~1640 cm⁻¹ (C=O stretching, Amide I). The lipid moiety, in turn, was inferred from the presence of peaks in the region of 1394–1461 cm⁻¹, corresponding to deformations of methyl and methylene (C-H) groups. Despite the high overall similarity in band distribution and intensity, minor spectral differences were detected, including a distinct peak at 1602 cm⁻¹ in the biosurfactant produced by isolate BS16B.

4. Discussion

4.1. Taxonomic Diversity and Functional Convergence of the Microbial Consortium of Origin

The taxonomic diversity of the biosurfactant-producing isolates in this study, spanning three distinct genera: Faucicola, Enterobacter, and Pseudomonas, is consistent with the metabolic complexity of the parent consortium. This pattern, where high variability is observed at the genus level but a tendency toward the recurrent emergence of dominant families such as Enterobacteriaceae and Pseudomonadaceae, is similar to the results of community self-organization in synthetic habitats [32]. In these systems, it has been shown that the final species composition can be highly variable, even when communities are generated from identical inocula under controlled conditions [32]. This phenomenon suggests that, rather than selecting a single lineage over nearly 20 years of selective pressure, the hydrocarbon-rich environment may have favored the recurrent emergence of a shared phenotypic trait related to biosurfactant production across different taxonomic lineages. The coexistence of multiple producer genera supports the interpretation that this function is consistent with an adaptive strategy allowing these species to occupy the same functional niche, a fundamental principle of microbial community ecology [32]. Specifically, the production of biosurfactants, amphiphilic molecules that increase the bioavailability of hydrophobic substrates, is critical for the assimilation of complex and insoluble carbon sources such as those present in petroleum waste sludge [33].

The identification of an isolate of the genus Pseudomonas (Pseudomonas sp. strain BS16B) was expected, given that this genus, in addition to being a biosurfactant producer, is frequently found in hydrocarbon-contaminated ecosystems [34,35]. Pseudomonas sp. and Stutzerimonas sp. with significant potential for biodesulfurization and/or degradation of polycyclic aromatic sulfur compounds (PASC) have previously been identified from the same bioreactor source subjected to ALE, the same source from which the isolates in this study were obtained, emphasizing the importance of this genus in soils contaminated with oily sludge [36].

Similarly, the presence of Enterobacter (Enterobacter sp. strain BS14MR) and Faucicola (Faucicola sp. strain BS5C) is consistent, as both genera contain species previously isolated from oil-contaminated soils [37,38,39]. However, the isolation of biosurfactant producers belonging to the genus Faucicola is less common in the literature, which may suggest the potential of long-term enrichment cultures as reservoirs of microbial diversity producing biomolecules of biotechnological interest. It is worth noting that this study presents Faucicola sp. strain BS5C after the reclassification of the genus [40], which was reported in previous studies as belonging to the genus Moraxella [41]. The phylogenetic position of each isolate, close to clades containing other recognized producers (Figure 1), supports the interpretation that the ability to produce these molecules is consistent with an important phenotypic trait in this environment. By mimicking natural selection under controlled conditions, ALE has been shown to facilitate the enrichment of desirable phenotypic characteristics, circumventing the need for prior genetic knowledge of the trait to be improved [15,42]. Recently, the ALE strategy was shown to be efficient for the selection of Burkholderia thailandensis mutants with increased rhamnolipid production, supporting the potential role of this strategy in selecting efficient biosurfactant producers [43]. Thus, the present study serves not only to characterize new isolates, but also supports previous evidence, in a complex and long-term system, that adaptive laboratory evolution (ALE) is a useful approach for the selection of strains exhibiting phenotypes of high biotechnological interest [15,42].

4.2. Cultivation Conditions and Strain-Specific Differences in Biosurfactant Production Robustness

The distinct responses of the three isolates to the culture conditions in this study (Table 1), particularly pH, reflect the complexity of secondary metabolism regulation. While many studies report higher biosurfactant production at neutral or acidic pH [44,45], the association with higher production at alkaline pH (8.5–9.5) observed for Faucicola sp. strain BS5C and Enterobacter sp. strain BS14MR is also supported by the literature [46,47]. One hypothesis for this phenomenon is that, at the phenotypic level, as bacterial fermentation tends to acidify the medium through the production of acidic metabolites [47], initiating cultivation under alkaline conditions may function as a buffering strategy, maintaining the pH within a productive range for longer. Additionally, alkaline pH stress has been reported to act as a trigger for the overproduction of secondary metabolites in other microorganisms [48], which is consistent with previous observations of increased biosurfactant production at pH 9 [46].

In contrast, the strong negative effect of high dextrose concentrations observed in all three isolates is consistent with carbon catabolite repression (CCR). This phenomenon is particularly well understood in bacteria with versatile metabolisms such as the genus Pseudomonas, one of the genera isolated in this study. In these organisms, the presence of a preferred carbon source, such as glucose, activates a well-characterized regulatory cascade. According to the literature, this cascade, mediated by elements such as the RNA-binding protein Crc and small RNAs (sRNAs) such as CrcZ, acts to repress the expression of genes essential for the maintenance of secondary metabolic pathways, as well as genes associated with the catabolism of less favorable substrates [49,50]. However, although this CCR mechanism is widespread, the specific preference for certain substrates when it comes to biosurfactant production is largely strain-dependent. This fact is well documented, with studies reporting higher biosurfactant yields when glucose is used compared to other complex substrates [51,52]. Thus, the behavior observed in the isolates evaluated in the present study can be interpreted as reflecting both this regulatory control and the metabolic specificity presented by each strain. In addition, this regulatory behavior can be interpreted in light of the ecological origin of the isolates. In oily sludge–contaminated environments, readily metabolizable carbon sources such as simple sugars are scarce, whereas complex hydrocarbons predominate. Under such conditions, long-term selective pressure may favor regulatory systems optimized for the utilization of complex and hydrophobic substrates rather than simple carbohydrates. Consequently, the observed repression of biosurfactant production at higher dextrose concentrations may reflect an adaptive regulatory legacy, in which glucose acts as a non-preferred or even disruptive carbon source for secondary metabolism.

The contrast observed in the modeling, in which a statistically adequate model was obtained for Pseudomonas sp. strain BS16B, compared with the lack of fit for Faucicola sp. strain BS5C and Enterobacter sp. strain BS14MR, highlights the complex interaction between the adaptive history of the strains and the predictability of the bioprocess. Before attributing this divergence solely to the intrinsic behavior of the strains, it is important to consider the methodological approach of this study. Part of the variability in the data that led to the lack of fit can be attributed to the response methodology. The emulsification index ($E_{24}$), although effective for screening, has different analytical resolution than methods such as surface tension determination, frequently used in modeling studies reporting higher coefficients of determination [53,54]. Furthermore, the determination of $E_{24}$ is sensitive to factors not controlled in this study, such as the pH and final salinity of the supernatant [55]. A second possibility is that the quadratic model proposed in the response surface methodology (CCRD) may not have fully captured key factors affecting the biological response of microorganisms.

However, although methodological and mathematical aspects may influence overall variability, they alone do not explain the high dispersion observed between replicates of the same experimental site (Supplementary Tables S5 and S6). This suggests that the high variability of the isolates Faucicola sp. strain BS5C and Enterobacter sp. strain BS14MR can be interpreted as a phenotypic consequence of their adaptive background. Therefore, it is hypothesized that the unpredictability of these two isolates may be associated with phenotypic heterogeneity, a phenomenon in which subpopulations with different metabolic profiles coexist within a genetically homogeneous culture [56]. Exposure to nearly two decades of selective pressure may have favored lineages through bet-hedging strategies, in which phenotypic diversity acts as a safety strategy for population survival in the face of unpredictable challenges [56,57]. Therefore, the limited fit of these strains to the proposed statistical model highlights the challenge of increasing microbial robustness under conditions where strains possess adaptive phenotypic traits.

The findings of this study have direct implications for biotechnology, particularly by highlighting the challenge of scaling up biosurfactant production using new microbial strains. The literature already demonstrates that laboratory-scale performance does not match that of commercial bioreactors, a challenge that increases the financial and technical risk of bioprocess implementation [16]. Our research suggests that one of the contributing factors of this discrepancy is associated with the evolutionary history of these strains. Traditional microorganisms adapted to laboratory conditions are selected over decades to possess stable and predictable behaviors under controlled conditions [58]. However, the isolates in this study are the product of nearly 20 years of adaptation to a highly selective laboratory environment, a process that promotes selection within a heterogeneous population [13]. As a result, the isolates possess a biotechnologically relevant phenotype but exhibit population heterogeneity that challenges the predictability of the bioprocess. Therefore, these data suggest that optimizing biosurfactants for large-scale production and increasing robustness and productivity may require beyond modeling physicochemical parameters and incorporate phenotypic instability as a key variable. To this end, recent approaches such as the analysis of single-cell lifelines and the application of systems biology to understand physiology under industrially relevant conditions may be crucial to accelerating large-scale production [16,59].

4.3. Spectroscopic Evidence of Phenotypic Convergence Toward Lipopeptide-like Biosurfactants

Spectroscopic analysis by ATR-FTIR provided supportive evidence for the hypothesis of functional convergence discussed above. Despite the taxonomic distance, all three isolates produced compounds compatible with lipopeptide-like biosurfactants, suggesting that this molecule may represent a functionally relevant strategy in their ecological niche of origin. This convergence toward the production of biosurfactants of the same class is ecologically significant, since lipopeptides (such as surfactin) are known for their well-documented ability to reduce surface and interfacial tension, properties essential for increasing the bioavailability of hydrophobic substrates such as those present in oily sludge [60]. Evidence supporting this shared classification is provided by the similarity of the FT-IR spectra (Figure 4), which presented absorption bands characteristic of both peptide bonds (~3200 cm⁻¹, ~1640 cm⁻¹) and lipid moieties (methyl or methylene of lipids, ~1400 cm⁻¹). The identification of the bands was consistent with spectral profiles of lipopeptides reported in the literature, such as those produced by Bacillus and Azotobacter [61,62].

Furthermore, the fingerprint region (1500 cm⁻¹ to 600 cm⁻¹) presented very similar peaks for the three isolates, indicating stable patterns [63]. However, beyond the observed similarities, the analysis revealed a subtle structural variation in the spectrum of the biosurfactant produced by Pseudomonas sp. strain BS16B, with a unique peak in the 1602 cm⁻¹ region, a region commonly associated with aromatic structures, suggesting possible differences in the amino acid composition of the peptide chain, such as the presence of aromatic residues [64]. Although it is a discreet structural variation, this variation may imply differences in the physicochemical and functional properties of each biosurfactant.

Lipopeptides are already recognized in the literature for their efficiency in various applications [65,66,67]. In contexts of environmental contamination, for example, the diffusion, desorption, and dissolution characteristics of contaminants by lipopeptide micelles facilitate their access for microbial bioremediation, improve the interaction of organic contaminants with the cell surface, and, furthermore, the presence of charges on the residues of the functional groups of lipopeptide molecules gives these molecules a surface charge, making them potent chelating agents [68]. Thus, the functional convergence for lipopeptide production by the three isolates can be associated with the selective pressures to which the strains were subjected since their ecological niche of origin, contaminated with oily sludge. This finding supports the importance of isolating microorganisms in contaminated environments and the possibility of potential application of these lipopeptides in bioremediation of organic compounds.

Thus, the results indicate that the strains belong to distinct genera and exhibit functional convergence toward the production of lipopeptide biosurfactants with high biotechnological potential. However, the studied strains exhibited significant differences in responsiveness to critical cultivation variables, despite having evolved together in the same bioreactor under the same stress conditions. Our results suggest that the unpredictability of biosurfactant production by these strains may be largely associated with phenotypic heterogeneity, a legacy of the strains’ adaptive history, and expose the challenges associated with metabolic robustness and stability in bioprocesses.

5. Conclusions

Despite their phylogenetic distance, bacterial isolates derived from a long-term oil sludge-adapted consortium exhibited comparable phenotypic outcomes associated with biosurfactant production. Spectroscopic analysis by ATR-FTIR indicated that the biosurfactants produced by the different isolates share comparable spectral features consistent with lipopeptide-like compounds, suggesting a common functional class rather than definitive structural identity. These findings reinforce the ecological relevance of lipopeptide-like biosurfactants in hydrocarbon-contaminated environments, where increasing the bioavailability of hydrophobic substrates represents a key selective advantage.

Beyond strain identification, the main contribution of this work lies in highlighting how the adaptive history of environmentally selected microorganisms influences the predictability and robustness of biosurfactant production processes. The contrasting modeling outcomes observed among the isolates suggest a potential trade-off between the high metabolic potential often associated with long-term adapted strains and the process stability required for industrial scale-up. In this context, phenotypic heterogeneity (likely shaped by prolonged exposure to selective pressures) emerges as a relevant factor affecting reproducibility and productivity, even under controlled cultivation conditions.

Overall, these results suggest that the development of scalable biosurfactant bioprocesses may need to extend beyond productivity-driven screening and physicochemical optimization. Instead, strain selection strategies should explicitly consider phenotypic stability and process robustness as key parameters. Incorporating these aspects at early stages of bioprocess development may improve the translation of promising environmental isolates into reliable industrial applications. Future studies integrating surface tension measurements, single-cell analyses, or molecular approaches may further elucidate the regulatory basis of the observed phenotypic heterogeneity.