Co-occurrence of biofilm formation, acid tolerance, and antibiotic resistance in environmental Escherichia coli associated with lettuce Authors: Saja Karanbash, Claude Daou, Christelle F. Iskandar Published: 2026 ABSTRACT BackgroundEnvironmental niches represent important reservoirs of Escherichia coli with stress-adaptation traits that support persistence outside the host. Contaminated irrigation water and soils can facilitate transfer to fresh produce, where bacterial survival may reduce the effectiveness of downstream control measures. This study investigated the co-occurrence of biofilm formation, acid tolerance, and antibiotic resistance (AR) in environmental E. coli and their contribution to persistence along the farm-to-produce continuum.ResultsEighteen E. coli isolates recovered from irrigation water, soil, and lettuce were characterized using phenotypic assays and genome-based analyses. Most isolates remained susceptible to the majority of tested antibiotics, with multidrug resistance observed in only 11.1% of isolates. In contrast, moderate-to-strong biofilm formation was widespread (83.3%), and several isolates exhibited reduced susceptibility to acetic acid at concentrations relevant to household washing practices. Genotypic screening revealed a broad distribution of adhesion, iron acquisition, biofilm-associated, and plasmid-borne resistance determinants, indicating substantial functional diversity. Significant positive associations were observed between acid tolerance, biofilm formation, and antibiotic resistance, suggesting co-occurrence of stress-adaptation phenotypes rather than definitive evolutionary convergence. While antibiotic resistance phenotypes showed strong concordance with corresponding resistance genes, biofilm formation and acid tolerance were not associated with specific genetic determinants, supporting a multifactorial basis of these traits.ConclusionsThese findings demonstrate that environmental E. coli can combine multiple stress-adaptation mechanisms that enhance persistence across agricultural and food-associated environments, even in the absence of high-risk resistance profiles. The observed co-occurrence of phenotypic traits highlights the potential for co-selection under environmental pressures and underscores the limitations of relying solely on downstream decontamination strategies. Effective risk mitigation requires integrated, preventive approaches targeting pre-harvest contamination and environmental reservoirs.Supplementary InformationThe online version contains supplementary material available at https://doi.org/10.1186/s12866-026-05311-1. Introduction Escherichia coli naturally inhabits the intestinal tract of humans and warm-blooded animals. While most strains are harmless commensals, some pathogenic variants, including enterohemorrhagic E. coli (EHEC) and enteropathogenic E. coli (EPEC), are responsible for foodborne illnesses such as diarrhea, hemorrhagic colitis, and hemolytic uremic syndrome [1]. Beyond its pathogenic potential, E. coli is widely used as an indicator of fecal contamination in water, soil, and agricultural produce [2]. Agricultural environments serve as critical reservoirs of E. coli, with the bacterium frequently detected in soil, irrigation water, and fresh produce, particularly leafy greens, which are often consumed raw [3, 4]. Contaminated water and organic amendments represent important routes through which E. coli can be introduced into environmental and agricultural systems [5]. Once introduced into soils or onto plant surfaces, E. coli can persist for extended periods depending on environmental conditions, increasing the likelihood of transfer to crops and subsequent exposure to consumers [6]. Contaminated irrigation water is widely recognized as a major vehicle for introducing bacterial pathogens into fresh produce systems, with fecal indicators such as E. coli commonly used to assess microbiological water quality and contamination risk [7, 8]. A major factor contributing to the persistence of E. coli in agricultural environments is its capacity to form biofilms. Biofilm formation is a widespread survival strategy that enables bacteria to persist across diverse environmental niches, including abiotic surfaces and host-associated environments. Biofilms are structured microbial communities embedded in a self-produced extracellular matrix that enhances tolerance to antibiotics, disinfectants, and environmental stresses [9]. On leafy greens, biofilm-associated E. coli are particularly difficult to eliminate, as attachment to plant surfaces reduces susceptibility to conventional sanitization and prolongs survival during post-harvest handling [10]. Although antibiotic resistance (AR) is often studied in clinical contexts, agricultural environments also serve as important reservoirs of resistant bacteria and mobile genetic elements [11]. The dissemination of resistant E. coli from irrigation water, soil, and fresh produce into the food chain raises concerns, particularly in regions where wastewater is frequently used for irrigation [12]. Environmental exposure to diverse stressors, including sublethal disinfectants, acidic conditions, and nutrient limitation, may further promote the co-selection of resistance and persistence traits. At the household level, vegetables are commonly washed with salt or vinegar (acetic acid) solutions due to their accessibility and perceived natural safety. However, the efficacy of these practices against stress-adapted environmental E. coli remains poorly characterized [13]. Acid tolerance responses in E. coli can enhance survival under moderately acidic conditions and may also confer cross-protection against other environmental stresses, reflecting broader stress-adaptation mechanisms that may contribute to environmental persistence and cross-protection against multiple stressors. Together, biofilm formation, tolerance to acidic stress, and antibiotic resistance represent interconnected stress-adaptation mechanisms that may co-select in agricultural environments and contribute to the persistence of E. coli along the food chain. Within this context, the present study used a wastewater-impacted system as a representative model to investigate the persistence and stress-adaptation potential of E. coli across environmental and food-associated niches. Specifically, the objectives were to (i) characterize antibiotic resistance profiles of E. coli isolated from irrigation water, soil, and lettuce; (ii) quantify biofilm-forming capacity; (iii) assess tolerance to acetic acid at concentrations relevant to household washing and mild post-harvest treatments; and (iv) determine the distribution of virulence- and stress-associated genes using genome-based screening. By integrating phenotypic and genotypic data, this study aims to provide insight into how environmental E. coli combine multiple stress-adaptation traits that may facilitate persistence and dissemination across environmental and food-associated niches. Methods Bacterial isolates Six agricultural demonstration plots were selected across the Litani River Basin (LRB), a highly contaminated area in Lebanon. Soil, irrigation water, and lettuce samples were collected at pre-, mid-, and post-harvesting periods to isolate E. coli. Sampling, bacterial isolation, purification, and confirmation of E. coli isolates were conducted within the framework of a previously published study [14], coordinated by a governmental organization. Access to sampling sites was granted by farmers managing the fields, and collections were carried out with their knowledge and agreement. For the present study, no additional environmental sampling or bacterial isolation was performed. Instead, previously recovered and confirmed E. coli isolates were retrieved from the culture collection of the Food Safety Facility at the American University of Beirut, where they had been stored at − 80 °C following purification. A total of 18 E. coli isolates were recovered and stored at − 80 °C after purification in the Food Safety Facility at the American University of Beirut. The isolates were distributed as follows: 3 irrigation water isolates (E1, E8, E16), 8 soil isolates (E6, E7, E9, E10, E14, E15, E17, E18), and 7 lettuce isolates (E2, E3, E4, E5, E11, E12, E13). Due to the exploratory nature of the study, all E. coli isolates recovered were included in subsequent analyses. Phenotypic analysis Antimicrobial susceptibility was assessed using the disk diffusion method according to the Clinical and Laboratory Standards Institute (CLSI) guidelines [15]. Briefly, bacterial suspensions equivalent to a 0.5 McFarland Unit (MFU) (approximately 1.5 × 108 Colony-forming-Unit per Milliliter CFU/mL) were inoculated onto Mueller–Hinton agar plates, and antibiotic disks (Bio-Rad, USA) representing major antibiotic classes were applied: amoxicillin (AMX, 20 µg), amoxicillin-clavulanate (AMC, 30 µg), piperacillin-tazobactam (TZP, 110 µg), cefuroxime (CXM, 30 µg), cefoxitin (FOX, 30 µg), cefotaxime (CTX, 30 µg), ceftriaxone (CRO, 30 µg), ceftazidime (CAZ, 30 µg), cefepime (FEP, 30 µg), aztreonam (ATM, 30 µg), meropenem (MEM, 10 µg), amikacin (AMK, 30 µg), gentamicin (GEN, 10 µg), streptomycin (STR, 10 µg), azithromycin (AZM, 15 µg), ciprofloxacin (CIP, 5 µg), levofloxacin (LVX, 5 µg), chloramphenicol (CHL, 30 µg), nitrofurantoin (NIT, 300 µg), fosfomycin (FOS, 200 µg), trimethoprim–sulfamethoxazole (SXT, 25 µg), and tetracycline (TET, 30 µg). After incubation at 37 °C for 18–24 h, inhibition zone diameters were measured and interpreted as susceptible, intermediate, or resistant based on CLSI breakpoints. Isolates were classified as multidrug-resistant (MDR) when resistant to at least one agent in three or more antibiotic classes. For extended-spectrum β-lactamase (ESBL), standard laboratory detection follows a two-step CLSI-recommended process: initial screening for reduced susceptibility to indicator cephalosporins, followed by confirmatory testing via synergy with a β-lactamase inhibitor [15]. For azithromycin, as CLSI does not provide specific breakpoints against E. coli, inhibition zone diameters were interpreted using the EUCAST epidemiological cut-off (ECOFF) for E. coli, allowing classification of isolates as wild-type (WT ≥ 13 mm) or non-wild-type (NWT) [16, 17]. Biofilm-forming capacity Biofilm-forming capacity was quantified as attached biofilm biomass using the standard crystal violet microtiter plate assay, adapted from Djordjevic et al., [10] and Stepanović et al., [9]. The assay quantifies total surface-associated biomass and is commonly used as a proxy for biofilm-forming capacity. This assay was performed under standardized laboratory conditions to enable comparison of intrinsic biofilm-forming capacity among isolates. Overnight cultures were adjusted to 0.5 MFU, inoculated into 96-well polystyrene microplates, and incubated statically at 37 °C for 24 h. Following incubation, wells were washed to remove planktonic cells, and adherent biofilms were stained with crystal violet. After solubilization, biofilm biomass was quantified by measuring absorbance at 570 nm. Biofilm formation capacity was quantified using the Biofilm Formation Index (BFI), calculated according to the following formula: Equation 1. Biofilm Formation Index (BFI) formula [18]\documentclass[12pt]{minimal} \usepackage{amsmath} \usepackage{wasysym} \usepackage{amsfonts} \usepackage{amssymb} \usepackage{amsbsy} \usepackage{mathrsfs} \usepackage{upgreek} \setlength{\oddsidemargin}{-69pt} \begin{document}$$\:BFI=\frac{\left(AB-CW\right)}{G}$$\end{document} Where AB represents the mean OD₅₇₀ values of the stained biofilm for each isolate measured in triplicates, CW represents the mean OD₅₇₀ of the stained control wells without bacterial inoculation, and G represents the OD₆₃₀ of the bacterial growth in the suspended culture. Based on the calculated BFI values, the E. coli isolates were classified into four categories: non-biofilm producers (BFI < 0.35), weak biofilm producers (0.35 ≤ BFI ≤ 0.69), moderate biofilm producers (0.70 ≤ BFI ≤ 1.09), and strong biofilm producers (BFI > 1.10) [18]. All assays were performed in triplicate in at least two independent experiments. Acetic acid tolerance assay Acetic acid concentrations were selected to reflect ranges commonly used in household produce washing and mild post-harvest disinfection practices. The tolerance of E. coli isolates to acetic acid was evaluated following [19] with modifications. Briefly, standardized bacterial suspensions at 0.5MFU were diluted to reach 10⁶ CFU/mL, then inoculated into LB broth containing acetic acid at final concentrations of 0.1%, 0.2%, 0.3%, 0.4%, and 0.5% (v/v). Non-acidified LB broth served as the control and was used as the reference for log₁₀ reduction calculations. Cultures were incubated statically at 37 °C for 24 h. The minimal inhibitory concentration (MIC) was defined as the lowest acetic acid concentration tube showing no visible growth after 24 h. After incubation, samples were serially diluted in BPW and plated in duplicate onto LB agar. Plates were incubated at 37 °C for 24 h, and viable counts were expressed as Log₁₀ CFU/mL. Log₁₀ reductions relative to untreated controls were calculated to quantify tolerance. The minimal bactericidal concentration (MBC) was defined as the lowest acetic acid concentration that resulted in < 25 colonies on agar plates after subculture. Isolates were categorized based on Log₁₀ reduction values as resistant (< 1 log₁₀ reduction), intermediate (1–5 log₁₀ reduction), or sensitive (> 5 log₁₀ reduction).Following the determination of the MIC and MBC for acetic acid, the MBC/MIC ratio was calculated to assess the antimicrobial effect of each treatment. An MBC/MIC ratio ≤ 4 was interpreted as indicating a bactericidal effect, reflecting complete bacterial killing, whereas a ratio > 4 but < 16 was considered bacteriostatic, indicating inhibition of growth without total elimination of the bacteria [20]. Genotypic analysis DNA extraction A fresh culture was prepared from the stocked glycerol by plating on MacConkey agar plates and incubating at 37 °C overnight. A few colonies of the same morphology were streaked onto a second MacConkey agar plate and incubated at 37 °C overnight. The colonies were then collected to extract the DNA. Bacterial DNA was extracted from bacterial strains cultured on MacConkey agar using the Quick-DNA™ Fungal/Bacterial Miniprep kit (Zymo Research, Irvine, CA). Samples were lysed via mechanical disruption in ZR BashingBead™ Lysis Tubes, followed by DNA purification using spin-column technology. DNA was purified using the Genomic DNA Clean & Concentrator™ kit (Zymo Research) according to the manufacturer’s protocols and quantified on a NanoPhotometer N60 spectrophotometer (Implen, Munich, Germany) before downstream applications. Library preparation and sequencing The Illumina DNA prep kit (Illumina, San Diego, CA) was used to prepare DNA libraries as per the manufacturer’s protocol. Following a tagmentation step via bead-linked transposons and post-tag-mentation clean-up, the IDT for Illumina DNA Unique Dual (UD) indexes were added. Then, the enhanced polymerase chain reaction mix (EPM) and index adapters were added, and DNA amplification was carried out. After library clean-up and elution, library concentrations were measured on a Qubit 4 fluorometer using the Qubit dsDNA High Sensitivity assay kit (Thermofisher Scientific, Waltham, MA). DNA libraries were pooled, denatured, and diluted to 12 pM. PhiX Control v3 (Illumina) was added then the pooled library was sequenced on an Illumina MiSeq platform using a MiSeq V2 Reagent Kit on (Illumina) for 250 × 2 cycles for a coverage of 50x. Bioinformatic analysis Raw reads in FASTQ format were assembled using Unicycler [21]. Assembled reads in FASTA format were then used for downstream analyses. Assembly quality was assessed by determining the total assembly size, number of contigs, N50 value, largest contig length, and GC content. Assembly statistics for all isolates are provided in Supplementary Table S1. As the assemblies were generated using Illumina short-read sequencing data only, all genomes remained in draft form and complete chromosome closure was not achieved. Determination of the virulence genes: AMRFinderPlus [22] and CARD [23] were used to identify antimicrobial resistance genes, while VirulenceFinder was used to detect virulence genes VirulenceFinder [24]. PlasmidFinder [25] was used to look for potential plasmids while only plasmid replicon hits exhibiting ≥ 95% sequence identity and ≥ 95% coverage were retained for downstream analyses. The virulence genes and plasmids information are described in Supplementary Tables S2 and S3. The whole-genome sequences generated in this study have been deposited in the NCBI database under BioProject accession number PRJNA1427110 and BioSample accession numbers from SAMN55903715 to SAMN55903732. Data analysis Data were analyzed using Excel (Microsoft) and R Software (v4.5.1). For biofilm-forming capacity assays, raw absorbance values were corrected by subtracting negative controls. Biofilm Formation Index (BFI) values were calculated using the defined formula, and isolates were categorized accordingly. All calculations and descriptive statistics (means and standard deviations) were performed in Excel. For the acid stress survival assay, bacterial counts (CFU/mL) were obtained from duplicate plates. Logarithmic reductions in viable counts at each concentration were calculated relative to the control (non-acidified LB broth) using R (version 4.5.1) for statistical computing. Means and standard deviations were computed in R software. Associations between isolate source (environmental vs. produce) and phenotypic traits (biofilm capacity, antibiotic resistance, and acid tolerance at 0.2%) were evaluated using Fisher’s exact test due to small sample size and sparse contingency tables. Statistical significance was defined as p < 0.05. Given the limited sample size (n = 18) and the relatively large number of pairwise comparisons, results were interpreted as exploratory and hypothesis-generating, and no formal correction for multiple testing was applied; p-values are therefore reported descriptively and should be interpreted with caution. Associations between phenotypic traits and functional genetic features (including antimicrobial resistance genes, virulence-associated genes, and plasmid presence) were evaluated using the Chi-square test of independence, and Fisher’s exact test when expected counts were < 5. The strength of associations was estimated using Cramér’s V, while Spearman’s rank correlation was applied to test relationships among ordinal phenotypic variables (e.g., AR vs. acid resistance vs. biofilm formation). Analyses were performed using Python (v3.12) with the SciPy (v1.13) and Pandas (v2.2) libraries. Statistical significance was defined as p < 0.05. Results Phenotypic analysis of the E. coli isolates Antibiotic Resistance (AR) The antimicrobial susceptibility profiles of the 18 E. coli isolates are summarized in Table 1. Overall, the isolates displayed low levels of phenotypic resistance across the tested antibiotic classes. All β-lactams, including penicillins combined with β-lactamase inhibitors, second- to fourth-generation cephalosporins, carbapenems, and the monobactam aztreonam, demonstrated complete susceptibility. No ESBL phenotype was detected. Table 1Phenotypic analysis of E. coli isolatesBeta-lactamsAminoglycosidesMacrolidesFluoroquinolonesSulfonamidesTetracyclinesPhenicolsPhosphonic acidNitrofuransBiofilm FormationAcid resistancePenicillinCephalosporinsMonobactamsCarbapenems2e G3e G4e GIsolateAMX20AMC30TZP110CXM30FOX30CTX30CRO30CAZ30FEP30ATM30MEM10GEN10AMK30STR10AZM15CIP5LVX5SXT25TET30CHL30FOS200NIT300BFIMIC %MBC %Antimicrobial ratioE1SSSSSSSSSSSSS I WTSSSSSSSNB0.10.22E2SSSSSSSSSSSSS I NWT SS R SSSSSt0.10.33E3SSSSSSSSSSSSS I WTSSSSSSSSt0.10.22E4SSSSSSSSSSSSS I WTSSSSSSSSt0.10.33E5 R SSSSSSSSSSSS R WTSS R I SSSSt0.10.22E6SSSSSSSSSSS R S R NWT SS R R I SSSt0.10.22E7SSSSSSSSSSSSS I WTSSSSSSSM0.10.33E8SSSSSSSSSSSSS I WTSSSSSSSNB0.10.22E9SSSSSSSSSSSSS I WTSSSSSSSM0.10.22E10SSSSSSSSSSSSS I WTSSSSSSSM0.10.33E11SSSSSSSSSSSSSSWTSSSSSSSM0.10.33E12SSSSSSSSSSSSS I WTSSSSSSSW0.10.33E13SSSSSSSSSSSSSSWTSSSSSSSM0.10.33E14SSSSSSSSSSSSS I WTSSSSSS I St0.10.22E15SSSSSSSSSSSSS I WTSSSSSSSSt0.10.22E16SSSSSSSSSSSSS I WTSSSSSSSSt0.10.22E17SSSSSSSSSSSSS I WTSSSSSSSSt0.10.22E18SSSSSSSSSSSSS R WTSSSSSSSM0.10.22BFI Biofilm index, MIC Minimal inhibitory concentration, MBC Minimal Bactericidal concentration, S Sensitive, I Intermediate, R Resistance, NB No biofilm, W Weak, M Moderate, St Strong. The bold font highlights the resistance pofile.The MBC/MIC ratio is calculated to determine the nature of the antimicrobial effect: Bactericidal Effect: MBC/MIC ratio ≤ 4 ; Bacteriostatic Effect: MBC/MIC ratio > 4 Table 1Phenotypic analysis of E. coli isolatesBeta-lactamsAminoglycosidesMacrolidesFluoroquinolonesSulfonamidesTetracyclinesPhenicolsPhosphonic acidNitrofuransBiofilm FormationAcid resistancePenicillinCephalosporinsMonobactamsCarbapenems2e G3e G4e GIsolateAMX20AMC30TZP110CXM30FOX30CTX30CRO30CAZ30FEP30ATM30MEM10GEN10AMK30STR10AZM15CIP5LVX5SXT25TET30CHL30FOS200NIT300BFIMIC %MBC %Antimicrobial ratioE1SSSSSSSSSSSSS I WTSSSSSSSNB0.10.22E2SSSSSSSSSSSSS I NWT SS R SSSSSt0.10.33E3SSSSSSSSSSSSS I WTSSSSSSSSt0.10.22E4SSSSSSSSSSSSS I WTSSSSSSSSt0.10.33E5 R SSSSSSSSSSSS R WTSS R I SSSSt0.10.22E6SSSSSSSSSSS R S R NWT SS R R I SSSt0.10.22E7SSSSSSSSSSSSS I WTSSSSSSSM0.10.33E8SSSSSSSSSSSSS I WTSSSSSSSNB0.10.22E9SSSSSSSSSSSSS I WTSSSSSSSM0.10.22E10SSSSSSSSSSSSS I WTSSSSSSSM0.10.33E11SSSSSSSSSSSSSSWTSSSSSSSM0.10.33E12SSSSSSSSSSSSS I WTSSSSSSSW0.10.33E13SSSSSSSSSSSSSSWTSSSSSSSM0.10.33E14SSSSSSSSSSSSS I WTSSSSSS I St0.10.22E15SSSSSSSSSSSSS I WTSSSSSSSSt0.10.22E16SSSSSSSSSSSSS I WTSSSSSSSSt0.10.22E17SSSSSSSSSSSSS I WTSSSSSSSSt0.10.22E18SSSSSSSSSSSSS R WTSSSSSSSM0.10.22BFI Biofilm index, MIC Minimal inhibitory concentration, MBC Minimal Bactericidal concentration, S Sensitive, I Intermediate, R Resistance, NB No biofilm, W Weak, M Moderate, St Strong. The bold font highlights the resistance pofile.The MBC/MIC ratio is calculated to determine the nature of the antimicrobial effect: Bactericidal Effect: MBC/MIC ratio ≤ 4 ; Bacteriostatic Effect: MBC/MIC ratio > 4 Phenotypic analysis of E. coli isolates BFI Biofilm index, MIC Minimal inhibitory concentration, MBC Minimal Bactericidal concentration, S Sensitive, I Intermediate, R Resistance, NB No biofilm, W Weak, M Moderate, St Strong. The bold font highlights the resistance pofile. The MBC/MIC ratio is calculated to determine the nature of the antimicrobial effect: Bactericidal Effect: MBC/MIC ratio ≤ 4 ; Bacteriostatic Effect: MBC/MIC ratio > 4 Among non-β-lactam antibiotics, most isolates remained susceptible to aminoglycosides, fluoroquinolones, tetracycline, chloramphenicol, fosfomycin, and nitrofurantoin. However, resistance was observed in specific classes. Streptomycin exhibited reduced effectiveness, with 16.7% (3/18) of isolates resistant and 72.2% (13/18) classified as intermediate. Resistance to sulfamethoxazole/trimethoprim was detected in 16.7% (3/18) of isolates. Reduced susceptibility to azithromycin, as defined by EUCAST epidemiological cut-offs, was observed in 2 isolates (11.1%). Two isolates (11.1%) met the criteria for multidrug resistance (resistance to ≥ 1 agent in ≥ 3 antibiotic classes). One lettuce isolate (E5) exhibited resistance to amoxicillin, streptomycin, and sulfamethoxazole/trimethoprim, while one soil isolate (E6) exhibited resistance to gentamicin, streptomycin, azithromycin, sulfamethoxazole/trimethoprim, and tetracycline. When grouped by source, antibiotic resistance (resistant or intermediate categories) was more frequent among environmental isolates than produce isolates; however, this trend did not reach statistical significance (Fisher’s exact test, p = 0.137). Given the limited sample size, these findings should be interpreted as exploratory. Biofilm-formation capacity A wide range of biofilm-forming capacities was observed among the E. coli isolates, from no detectable biofilm to strong biofilm formation (Table 1, Supplementary Table S4). Overall, 83.3% (15/18) of isolates exhibited moderate to strong biofilm-forming capacity, indicating a high prevalence of surface-associated growth phenotypes across agricultural and produce-associated samples. Strong biofilm formation was observed in nine isolates (50.0%), while six isolates (33.3%) were classified as moderate biofilm producers. Only one isolate (5.6%) exhibited weak biofilm formation, and two water-derived isolates (11.1%) showed no detectable biofilm production. Strong biofilm producers were identified among isolates from lettuce, soil, and irrigation water, demonstrating that enhanced biofilm formation was not restricted to a single source. When isolates were grouped by origin (environmental vs. produce), no significant difference in the prevalence of moderate-to-strong biofilm formation was detected (Fisher’s exact test, p = 1.00). Given the limited sample size, these analyses were considered exploratory. These findings indicate that biofilm formation represents a common persistence strategy among agricultural E. coli. Acid tolerance assay All isolates were inhibited at 0.1% acetic acid (MIC), while MBC ranged from 0.2% to 0.3% (Table 1). For most isolates (11/18; 61.1%), complete killing occurred at 0.2%, whereas seven isolates (38.9%) required 0.3% acetic acid to achieve bactericidal activity. Log₁₀ reduction analysis confirmed a concentration-dependent antimicrobial effect (Fig. 1, Supplementary Table S5). At 0.1% acetic acid, reductions ranged from 1 to 5 log₁₀, indicating partial tolerance. At 0.2%, 61.1% of isolates exhibited > 5 log₁₀ reductions, while at 0.3% acetic acid, all isolates demonstrated 8–9 log₁₀ reductions, corresponding to complete elimination. Across all isolates, MBC/MIC ratios were ≤ 3, indicating bactericidal activity at effective concentrations. When grouped by origin, environmental isolates (water and soil) showed significantly higher acid tolerance at 0.2% acetic acid compared with produce isolates (9/11 vs. 2/7; Fisher’s exact test, p = 0.049). Given the limited sample size, these findings should be interpreted as exploratory. Fig. 1Log reduction of E. coli isolates exposed to different concentrations of acetic acid. The horizontal yellow lines represent log reductions of 1 and 5, serving as reference points to categorize the isolates’ susceptibility: Below the lower yellow line (log reduction < 1): Resistant/low susceptibility; Between the yellow lines (log reduction between 1 and 5): Intermediate Resistance; Above the upper yellow line (log reduction > 5): Sensitive. Full Log10reduction data and standard deviations are included in Supplementary Table S5 Log reduction of E. coli isolates exposed to different concentrations of acetic acid. The horizontal yellow lines represent log reductions of 1 and 5, serving as reference points to categorize the isolates’ susceptibility: Below the lower yellow line (log reduction < 1): Resistant/low susceptibility; Between the yellow lines (log reduction between 1 and 5): Intermediate Resistance; Above the upper yellow line (log reduction > 5): Sensitive. Full Log10reduction data and standard deviations are included in Supplementary Table S5 Genotypic analysis of the E. coli isolates Genome assembly statistics Assembly statistics for all sequenced isolates are summarized in Supplementary Table S1. Draft genome assemblies ranged from 4.58 to 5.12 Mb in size and consisted of 79–274 contigs (mean = 156 contigs per assembly). N50 values ranged from 83.1 to 271.9 kb, while the largest contigs ranged from 264 to 770 kb. The average GC content was 50.67%, consistent with previously reported Escherichia coli genomes. As expected for assemblies generated exclusively from Illumina short-read sequencing data, complete chromosome closure was not achieved and all genomes remained in draft form. Virulence of the isolates Genomic screening revealed a broad distribution of adhesion, iron acquisition, biofilm-associated, toxin, and stress-related genes among the E. coli isolates (Table 2; Supplementary Table S2 and S3), indicating substantial functional heterogeneity across environmental and produce-associated strains. Table 2Virulence gene profiles of E. coli isolates: distribution of adhesion, invasion, siderophore, serum survival, biofilm, toxin, and enzyme-associated genesIsolateAdhesionInvasionSiderophoresSerum survivalBiofilmToxinsEnzymesPlasmidsE1 fdeC espX1 IncI1IncFIB1IncFIAE2fdeC, iha, lpfA espX1 iucABCD,sslE, sigA capU IncB/O/K/Z iutA IncFIIE3fdeC, lpfA espX1 iss sslE IncI1Col156E4 fdeC IncHI1AIncHI1B(R27)IncX3IncYE5fdeC, lpfA espX1 iroBCDEN, iucABCD, iutA iss sslE cvaBC IncFIIIncFIBE6fdeC, lpfA sslE astA IncYIncI1E7fdeC, lpfA espX1 iss sslE IncFIBE8 fdeC espX1 E9fdeC, lpfA espX1 sslE E10fdeC, lpfA espX1 sslE astA IncI1E11fdeC, lpfA espX1 sslE astA epeA IncFIBE12 fdeC espX1 E13fdeC, lpfA espX1 sslE astA epeA IncFIBE14fdeC, lpfA espX1 ireA, iroBCDEN iss sslE IncFIBIncFIAIncFIIE15fdeC, lpfA espX1 sslE astA IncB/O/K/ZIncI1E16 fdeC espX1 ybtPQ astA E17fdeC, lpfA espX1 sslE astA IncX1IncI1E18fdeC, lpfA espX1 sslE IncFIB Table 2Virulence gene profiles of E. coli isolates: distribution of adhesion, invasion, siderophore, serum survival, biofilm, toxin, and enzyme-associated genesIsolateAdhesionInvasionSiderophoresSerum survivalBiofilmToxinsEnzymesPlasmidsE1 fdeC espX1 IncI1IncFIB1IncFIAE2fdeC, iha, lpfA espX1 iucABCD,sslE, sigA capU IncB/O/K/Z iutA IncFIIE3fdeC, lpfA espX1 iss sslE IncI1Col156E4 fdeC IncHI1AIncHI1B(R27)IncX3IncYE5fdeC, lpfA espX1 iroBCDEN, iucABCD, iutA iss sslE cvaBC IncFIIIncFIBE6fdeC, lpfA sslE astA IncYIncI1E7fdeC, lpfA espX1 iss sslE IncFIBE8 fdeC espX1 E9fdeC, lpfA espX1 sslE E10fdeC, lpfA espX1 sslE astA IncI1E11fdeC, lpfA espX1 sslE astA epeA IncFIBE12 fdeC espX1 E13fdeC, lpfA espX1 sslE astA epeA IncFIBE14fdeC, lpfA espX1 ireA, iroBCDEN iss sslE IncFIBIncFIAIncFIIE15fdeC, lpfA espX1 sslE astA IncB/O/K/ZIncI1E16 fdeC espX1 ybtPQ astA E17fdeC, lpfA espX1 sslE astA IncX1IncI1E18fdeC, lpfA espX1 sslE IncFIB Virulence gene profiles of E. coli isolates: distribution of adhesion, invasion, siderophore, serum survival, biofilm, toxin, and enzyme-associated genes Adhesion-associated genes were highly prevalent. The fdeC gene was detected in all isolates, while lpfA was present in most strains, supporting widespread potential for surface attachment. The invasion-associated gene espX1 was also frequently detected, suggesting retention of host-interaction traits among agricultural isolates. Multiple iron acquisition systems were identified, including aerobactin (iucABCD/iutA), salmochelin (iroBCDEN), yersiniabactin (ybtPQ), and ireA, reflecting diverse strategies for survival under iron-limited environmental conditions. The serum survival gene iss was detected in several isolates, indicating potential resistance to host defense mechanisms. Biofilm-associated and stress-related genes were widely distributed. The extracellular metalloprotease gene sslE was detected in the majority of isolates, while the stress regulator ariR was present in all strains, indicating a conserved genetic capacity for biofilm development and environmental stress adaptation. Additional virulence-associated genes, including astA, sigA, epeA, capU, and cvaBC, were detected sporadically, highlighting isolate-specific pathogenic potential. Plasmid profiling revealed a heterogeneous distribution of high-confidence plasmid replicons among the E. coli isolates. IncI1 and IncFIB were the most frequently detected incompatibility groups, each identified in six isolates, followed by IncFII (3 isolates) and IncFIA (2 isolates). Less common replicon types included IncB/O/K/Z (2 isolates), IncY (2 isolates), and IncHI1A, IncHI1B(R27), IncX3, IncX1, and Col156, which were each detected in a single isolate. Several isolates harbored multiple IncF replicons, which is consistent with the multireplicon nature of IncF plasmids. Four isolates lacked detectable high-confidence plasmid replicons. Overall, the presence of diverse plasmid incompatibility groups highlights the potential role of plasmids in the dissemination of adaptive traits and genetic exchange within agricultural environments (Table 2, Supplementary material S3). Collectively, the widespread presence of adhesion factors, siderophore systems, biofilm-associated genes, and mobile genetic elements suggests that these isolates possess a modular genetic toolkit supporting persistence in produce-associated ecosystems. Resistance to antibiotics Antimicrobial resistance genes (ARGs) were unevenly distributed among isolates (Table 3). The intrinsic β-lactamase gene blaEC was detected in all strains. Acquired resistance determinants were primarily confined to a subset of isolates and corresponded well with observed phenotypic resistance patterns. Table 3Distribution of ARG among E. coli isolatesIsolateClass of AntibioticsBeta - lactamMacrolideSulfonamideTrimethoprimQuinoloneAminoglycosideTetracyclineChloramphenicolFosfomycinNitrofurantoin# Efflux pumpsE1 blaEC 40E2 blaEC mph(A) sul1 dfrA5 gyrA_S83L 41E3 blaEC 40E4 blaEC qnrB7 40E5 blaEC sul2 aph(3”)-Ib, aph(6)-Id tet(A) 40E6 blaEC mph(A) sul3 dfrA14 aac(3), aadA1, aadA2, aph(3”), aph(4)-Ia, aph(6)-Id- aph(3’)-la tet(A) cmlA1 43E7 blaEC 40E8 blaEC 41E9 blaEC 40E10 blaEC 39E11 blaEC 40E12 blaEC 41E13 blaEC 40E14 blaEC 41E15 blaEC 39E16 blaEC 41E17 blaEC 39E18 blaEC 39For each antibiotic class, only detected genes are listed. Empty cells represent the absence of any corresponding ARG in that isolate Table 3Distribution of ARG among E. coli isolatesIsolateClass of AntibioticsBeta - lactamMacrolideSulfonamideTrimethoprimQuinoloneAminoglycosideTetracyclineChloramphenicolFosfomycinNitrofurantoin# Efflux pumpsE1 blaEC 40E2 blaEC mph(A) sul1 dfrA5 gyrA_S83L 41E3 blaEC 40E4 blaEC qnrB7 40E5 blaEC sul2 aph(3”)-Ib, aph(6)-Id tet(A) 40E6 blaEC mph(A) sul3 dfrA14 aac(3), aadA1, aadA2, aph(3”), aph(4)-Ia, aph(6)-Id- aph(3’)-la tet(A) cmlA1 43E7 blaEC 40E8 blaEC 41E9 blaEC 40E10 blaEC 39E11 blaEC 40E12 blaEC 41E13 blaEC 40E14 blaEC 41E15 blaEC 39E16 blaEC 41E17 blaEC 39E18 blaEC 39For each antibiotic class, only detected genes are listed. Empty cells represent the absence of any corresponding ARG in that isolate Distribution of ARG among E. coli isolates aph(3”)-Ib, aph(6)-Id For each antibiotic class, only detected genes are listed. Empty cells represent the absence of any corresponding ARG in that isolate Macrolide resistance (mph(A)) was detected in two isolates, while sulfonamide (sul1, sul2, sul3) and trimethoprim (dfrA5, dfrA14) resistance genes were identified in isolates exhibiting reduced susceptibility to sulfamethoxazole/trimethoprim. Aminoglycoside resistance genes were concentrated in two isolates, with one strain harboring multiple determinants (aac(3), aadA1, aadA2, aph variants). Tetracycline resistance (tet(A)) and chloramphenicol resistance (cmlA1) were detected in the same multidrug-resistant isolates. No resistance genes associated with fosfomycin or nitrofurantoin were identified. Efflux pump genes were detected across all isolates, suggesting a baseline capacity for reduced intracellular antimicrobial accumulation. Overall, ARG carriage was limited to a small number of strains but frequently co-occurred with plasmids and biofilm-associated genes, supporting the integration of resistance traits within broader stress-adaptation frameworks. Functional gene–phenotype convergence underlying stress adaptation Correlation analyses revealed significant associations among phenotypic traits (Table 4-A), with acid tolerance positively correlated with AR (Spearman’s ρ = 0.58, p = 0.012) and biofilm-forming capacity (ρ = 0.48, p = 0.043), while biofilm formation also correlated with AR (ρ = 0.50, p = 0.033), indicating partial convergence of stress-adaptation phenotypes. Table 4Functional convergence of stress-adaptation traits in agricultural E. coliA- Phenotype–phenotype correlationsComparisonTestStatisticp-valueInterpretationAR × Acid toleranceSpearmanρ = 0.580.012SignificantBiofilm × ARSpearmanρ = 0.500.033SignificantBiofilm × Acid toleranceSpearmanρ = 0.480.043SignificantB- Functional gene / mobile element vs. phenotype (2 × 2 Fisher)FeaturePhenotype tested+/++/–p-value - InterpretationARG presenceAR3/41/140.019 - Significant sslE Strong biofilm7/132/51 – Not SignificantSiderophoresAcid tolerance (0.2%)1/46/141– Not SignificantPlasmidsAcid tolerance (0.2%)6/151/31– Not SignificantPlasmidsAR4/150/31– Not SignificantNot significant (p > 0.05) Table 4Functional convergence of stress-adaptation traits in agricultural E. coliA- Phenotype–phenotype correlationsComparisonTestStatisticp-valueInterpretationAR × Acid toleranceSpearmanρ = 0.580.012SignificantBiofilm × ARSpearmanρ = 0.500.033SignificantBiofilm × Acid toleranceSpearmanρ = 0.480.043SignificantB- Functional gene / mobile element vs. phenotype (2 × 2 Fisher)FeaturePhenotype tested+/++/–p-value - InterpretationARG presenceAR3/41/140.019 - Significant sslE Strong biofilm7/132/51 – Not SignificantSiderophoresAcid tolerance (0.2%)1/46/141– Not SignificantPlasmidsAcid tolerance (0.2%)6/151/31– Not SignificantPlasmidsAR4/150/31– Not SignificantNot significant (p > 0.05) Functional convergence of stress-adaptation traits in agricultural E. coli Not significant (p > 0.05) To explore genetic contributors to these patterns, isolates were grouped based on the presence of functional determinants (Table 4-B). Strong biofilm formation occurred in both sslE-positive (7/13) and sslE-negative (2/5) isolates, indicating that enhanced surface attachment was not solely explained by sslE carriage. Similarly, siderophore-positive isolates did not exhibit higher acetic acid tolerance at 0.2% compared with siderophore-negative strains (1/4 vs. 6/14; Fisher’s exact test, p = 1.00). In contrast, AR phenotypes showed strong concordance with acquired ARGs. Phenotypic resistance to at least one antibiotic class was observed in 3/4 ARG-positive isolates compared with 1/14 ARG-negative isolates (Fisher’s exact test, p = 0.019), supporting a clear genotype–phenotype relationship for AR. Plasmids were detected in 15/18 isolates, including all AR-positive strains (4/4). However, plasmids were also common among AR-negative isolates (11/14), and no statistically significant association between plasmid carriage and AR phenotype was detected (Fisher’s exact test, p = 1.00). Likewise, plasmid presence was not associated with increased acetic acid tolerance at 0.2% (6/15 plasmid-positive vs. 1/3 plasmid-negative isolates; Fisher’s exact test, p = 1.00). Together, these findings indicate that while antibiotic resistance is largely explained by corresponding resistance determinants, biofilm formation and acid tolerance are not driven by single virulence or stress-associated genes in this dataset, suggesting multifactorial regulation of persistence trait. Given the limited number of isolates, these analyses are exploratory and hypothesis-generating. Discussion Agricultural environments are increasingly recognized as reservoirs of E. coli carrying functional traits that enhance survival outside the host and facilitate persistence along the food chain. Irrigation water, soil, and fresh produce surfaces represent dynamic ecological niches in which fecal indicator bacteria encounter fluctuating nutrient availability, desiccation, and diverse anthropogenic and environmental stresses [3, 4]. Under such conditions, selection is expected to favor strains with adaptive traits that promote environmental fitness rather than classical virulence alone. The virulence gene profiles observed in this study indicate that the environmental E. coli isolates circulating in the agricultural system possess harbor multiple virulence-associated determinants. The presence of fdeC (adherence factor to the host epithelial cells) in all isolates is particularly noteworthy [26]. Its ubiquity indicates that environmental strains retain conserved adhesion-associated functions, consistent with findings from other contaminated agricultural settings [27]. A substantial proportion of isolates also carried lpfA, encoding long polar fimbriae, and espX1, a type III secretion effector. These genes are typically associated with attaching-and-effacing or invasive pathotypes and are widely distributed among strains adapted to persist in both environmental and host-associated niches [27]. Their high prevalence suggests that these isolates retain adaptive traits that may support persistence on plant-associated surfaces and, potentially, opportunistic host colonization. Beyond adhesion and invasion, the detection of multiple iron acquisition systems across the isolates provides strong evidence of enhanced ecological fitness and potential virulence. Siderophore systems such as aerobactin (iucABCD/iutA), salmochelin (iroBCDEN), yersiniabactin (ybtPQ), and ireA play essential roles in enabling E. coli to grow under iron-restricted conditions typical of soil, plant surfaces, and host tissues [28, 29]. In parallel, the detection of the serum-survival gene iss (E3, E5, E7, E14) indicates that several isolates can resist complement-mediated killing, a trait strongly associated with increased systemic virulence [30]. Taken together, the co-occurrence of adhesion, invasion, and serum resistance mechanisms demonstrates that the agricultural E. coli population analyzed here harbors a distributed and modular pathogenic toolkit capable of supporting survival across environmental and host-associated contexts. Plasmid profiling based on stringent coverage criteria revealed a heterogeneous distribution of high-confidence incompatibility plasmid replicons among the E. coli isolates. IncI1 and IncFIB were the most frequently detected types, followed by IncFII and IncFIA, while IncB/O/K/Z, IncHI1A, IncHI1B(R27), IncX3, IncX1, IncY, and Col156 were confined to individual or small subsets of isolates. Inc plasmids are well known for carrying and disseminating virulence and ARGs [31]. IncI1 plasmids, frequently observed in Enterobacteriaceae, often encode multiple resistance and virulence determinants [32]. Likewise, IncFIB and IncFII plasmids identified in our isolates are associated with both classical ExPEC strains and emerging multidrug-resistant lineages [33]. Recent genomic analysis of multidrug-resistant E. coli isolates revealed that IncFIB replicons are frequently linked with pathogenicity islands (PAIs) and enable stable co-transfer of multiple resistance and virulence genes [33]. Consequently, the high prevalence of IncI1, IncFIB, and IncFII plasmids in our agricultural isolates indicates that the environment serves as a nexus for horizontal gene transfer, facilitating the dissemination and recombination of resistance and virulence traits across distantly related lineages. In the present study, the widespread detection of adhesion-associated determinants and stress-related genes supports this ecological interpretation. Adhesion is a critical first step in colonization of plant surfaces and subsequent biofilm development, enabling persistence during pre-harvest growth and post-harvest handling. Similar findings have been reported in produce-associated E. coli, where surface attachment enhances survival and reduces susceptibility to washing and sanitization [34–36]. A key finding of this study was the high prevalence (83%) of moderate-to-strong biofilm formation among isolates. Biofilms confer protection against environmental stressors by promoting structured community growth and altered metabolic states, reducing susceptibility to antimicrobial agents and disinfectants [37]. In food systems, biofilm formation on fresh produce has been shown to decrease the efficacy of decontamination treatments, allowing attached cells to survive mild sanitization [38, 39]. The high proportion of strong biofilm producers observed here may therefore contribute to persistence in food-associated environments. This observation is consistent with previous studies reporting strong biofilm phenotypes among E. coli isolated from irrigation water, soil, and fresh produce, in which fluctuating nutrients, moisture, and surface contact favor an attached lifestyle [3, 5]. Similar biofilm prevalences have been described for produce-associated E. coli, particularly on leafy greens, where biofilm formation enhances survival on plant surfaces and limits removal during post-harvest handling [40]. The predominance of biofilm-forming isolates observed here appears higher than that reported in some clinical collections, where planktonic growth is more common, underscoring the role of environmental selection pressures in shaping persistence traits outside the host. Anderson et al. [41] demonstrated that environmental E. coli frequently form stronger biofilms than laboratory or clinical strains, supporting the notion that biofilm development is a key ecological adaptation rather than a lineage-specific virulence trait. In agreement with this, biofilm strength in the present study varied widely even among isolates sharing similar genotypic backgrounds, indicating that phenotypic expression is strongly influenced by regulatory and environmental factors rather than the mere presence of adhesion genes. From a food safety perspective, the high prevalence of biofilm-forming organisms may partially explain the reduced efficacy of mild decontamination treatments observed in previous studies. Biofilms can shield E. coli from acidic and osmotic stresses and promote prolonged survival on produce surfaces, thereby increasing the likelihood of consumer exposure. Together, these findings reinforce that biofilm formation is a critical barrier to effective control of E. coli contamination and highlight its role in contributing to environmental persistence at the pre-harvest and early post-harvest stages of the food production chain. Nevertheless, the interpretation of in vitro biofilm assays should be approached cautiously when extrapolating persistence under field conditions. The microtiter plate assay used in the present study provides a standardized and reproducible measure of surface-associated growth potential; but it does not fully reproduce the physicochemical and ecological complexity of the phyllosphere. Plant surfaces are characterized by fluctuating humidity, nutrient limitation, UV exposure, microbial competition, and heterogeneous surface topography, all of which can substantially influence bacterial attachment and persistence [42]. Recent work by Schlechter et al. showed that in vitro biofilm formation capacity was not directly correlated with the persistence of antibiotic-resistant E. coli on gnotobiotic lamb’s lettuce leaves, highlighting the context dependence of biofilm phenotypes in plants [43]. Similar observations have been reported in studies demonstrating that regulatory pathways involved in biofilm formation under laboratory conditions may differ substantially from those activated during plant colonization and environmental persistence, particularly under nutrient-limited and desiccation-associated conditions [42, 44]. Therefore, the strong biofilm phenotypes observed in the present study should be interpreted as indicators of intrinsic surface-associated growth potential rather than direct predictors of persistence on leafy greens under natural agricultural conditions. The E. coli isolates exhibited heterogeneous tolerance to acetic acid at concentrations relevant to household and mild processing decontamination practices (0.2%–0.3%), with several isolates retaining viability under acidic stress. The heterogeneity observed in acetic acid tolerance further supports the presence of adaptive stress responses. Comparable heterogeneity in acid tolerance has been described in food-associated E. coli, where prior environmental exposure significantly modulates survival under acetic acid stress [45]. Recent work further demonstrates that mild acid treatments, including vinegar-based washes, produce variable reductions and are markedly less effective against attached or biofilm-embedded cells [46]. E. coli possesses multiple acid resistance systems that enable survival under moderately acidic conditions, including acid tolerance responses induced by prior environmental exposure [47]. Acid adaptation has also been associated with cross-protection against additional stressors, suggesting that exposure to sublethal environmental challenges may enrich strains with broader stress resilience [48, 49]. The observed association between acid tolerance and biofilm formation is therefore biologically plausible and consistent with the concept of multifactorial stress adaptation. From a consumer perspective, these findings align with reports that household washing practices using vinegar or similar mild acid treatments produce variable reductions in bacterial counts and may not eliminate surface-attached cells [50]. The present results suggest that mild acid exposure may not fully eliminate stress-adapted E. coli. The E. coli isolates exhibited generally low levels of antibiotic resistance, with most remaining susceptible to clinically important antibiotic classes and multidrug resistance restricted to a small subset of isolates. No extended-spectrum β-lactamase or carbapenemase phenotypes were detected, indicating that the isolates do not currently represent high-risk clinical resistance profiles. Similar observations have been reported in agricultural and environmental contexts where antibiotic selection pressure is lower than in clinical or intensive livestock settings [11, 51]. Despite this low phenotypic resistance burden, acquired resistance genes were detected and showed strong genotype–phenotype concordance. Resistance to sulfamethoxazole/trimethoprim and reduced susceptibility to azithromycin were confined to isolates carrying sul/dfrA and mph(A) genes, respectively, consistent with previous studies identifying environmental E. coli as reservoirs of mobilizable resistance determinants relevant to environmental dissemination [12, 17, 52]. These findings suggest that even when resistance levels are modest, agricultural environments can harbor genetic traits with the potential for dissemination along the food chain. Importantly, antibiotic resistance frequently co-occurred with biofilm formation and acid tolerance, indicating that resistance traits are embedded within broader stress-adaptation strategies rather than being isolated phenotypes. This co-occurrence supports the concept of co-selection in pre-harvest environments, where exposure to multiple non-antibiotic stressors may favor the persistence of bacteria combining resistance, stress tolerance, and surface-associated survival traits [51, 53]. At the same time, the relatively low antibiotic resistance burden observed in the present isolates should not obscure the broader role of agricultural environments as reservoirs and dissemination pathways for clinically relevant resistance determinants. Several environmental surveillance studies have reported high prevalences of multidrug-resistant and ESBL-producing Enterobacteriaceae in irrigation water, rivers, and fresh produce production systems, even in the absence of clearly identifiable local antibiotic selection pressure [54–56]. For example, irrigation water has been recognized as an important transmission route for resistant E. coli and mobile resistance elements across the farm-to-food continuum [54]. More recent metagenomic investigations further demonstrated that agricultural and aquatic environments can sustain complex environmental resistomes shaped not only by antibiotic exposure but also by heavy metals, disinfectants, wastewater contamination, and horizontal gene transfer dynamics [55, 57]. Similar trends have also been described in surface waters impacted by urban and agricultural runoff, where resistant Enterobacteriaceae persist despite limited evidence of direct antimicrobial application [56]. In this context, the comparatively low resistance frequencies observed in the present study may reflect local watershed characteristics, contamination sources, and agricultural practices rather than the absence of environmental selective pressures. Collectively, these findings reinforce the importance of considering environmental compartments as dynamic reservoirs capable of maintaining and disseminating resistance determinants even under apparently low antibiotic pressure conditions and support the view that environmental compartments are active ecological interfaces where mobile genetic elements, stress adaptation, and bacterial persistence can interact to shape the environmental resistome. Taken together, these findings indicate that biofilm formation, acid tolerance, and antibiotic resistance represent interconnected stress-adaptation traits rather than independent characteristics. Biofilm formation was highly prevalent among isolates, and acid tolerance further contributed to environmental persistence, with several isolates surviving acetic acid concentrations relevant to household washing practices. Although overall antibiotic resistance levels were low and no high-risk phenotypes, such as ESBL production, were detected, resistance determinants were present and frequently co-occurred with biofilm formation and acid tolerance, supporting the concept of co-selection under non-antibiotic environmental pressures that may undermine downstream control measures. The strongest genotype–phenotype concordance observed in this study was between acquired ARGs and corresponding resistance phenotypes. This agreement supports the reliability of genomic screening tools for detecting resistance determinants and confirms that even limited ARG carriage in agricultural isolates can translate into measurable phenotypic resistance. A notable exception to this overall genotype–phenotype agreement was the chromosomally encoded blaEC gene, which was detected in all isolates despite the absence of phenotypic resistance to β-lactam antibiotics and the lack of ESBL production. This observation highlights an important limitation of genome-based resistance prediction: the mere presence of a resistance-associated gene does not necessarily imply functional expression or clinically relevant resistance. Gene expression levels, promoter activity, regulatory pathways, gene truncation, and other physiological factors may all influence whether a detectable resistance phenotype is ultimately expressed. Similar genotype–phenotype discrepancies have previously been described in environmental and commensal Enterobacteriaceae, particularly for intrinsic or weakly expressed β-lactamase genes and silent resistance determinants [58, 59]. These findings emphasize the importance of interpreting genomic resistance data in conjunction with phenotypic susceptibility testing, especially when evaluating environmental isolates where resistance determinants may remain silent or weakly expressed under standard laboratory conditions. In contrast, no single virulence- or stress-associated gene explained variation in biofilm formation or acid tolerance. This observation is consistent with the understanding that such phenotypes are governed by complex regulatory networks rather than single-gene presence/absence patterns [60]. Environmental persistence is likely influenced by global stress regulators, metabolic state, and gene expression dynamics that are not captured by simple genomic detection [61]. The positive correlations identified among antibiotic resistance, acid tolerance, and biofilm formation suggest partial phenotypic co-occurrence that may arise through co-selection processes, where exposure to environmental stressors, including disinfectants, metals, or agricultural antibiotics, can indirectly select for bacteria harboring multiple adaptive traits [62, 63]. The coexistence of ARGs and stress-tolerance traits across isolates mirrors findings from multiple studies reporting positive associations between biofilm formation and antibiotic resistance across different contexts [64–66]. Conversely, other investigations have documented weak or inconsistent correlations, particularly among environmental or non-clinical isolates [67, 68], and some have emphasized that such associations depend on specific antibiotic classes and genetic backgrounds [69]. Plasmids may contribute to the co-occurrence of these adaptive traits by facilitating co-localization of resistance genes and accessory fitness determinants, even when overt multidrug resistance remains uncommon [70]. Overall, the trait combinations observed in this study reflect the multifaceted survival strategies of environmental E. coli, where AR, biofilm formation, and acid tolerance interact to support persistence across environmental and food-associated niches. These observations support the concept that environmental persistence in E. coli is driven by the coordinated action of multiple stress-adaptation mechanisms rather than by a single genetic determinant. Although the present study provides an integrated phenotypic and genomic characterization of environmental E. coli, several limitations should be acknowledged. The relatively small number of isolates limits statistical power and restricts the ability to infer broad ecological or evolutionary trends across agricultural environments. In addition, multiple comparisons were performed across phenotypic and genomic variables, and therefore the observed associations should be interpreted cautiously and considered exploratory rather than definitive. Consequently, the relationships identified among biofilm formation, acid tolerance, plasmid carriage, and antibiotic resistance should be viewed as indicative patterns that warrant confirmation in larger and geographically diverse isolate collections. Nevertheless, the integrated approach used here provides valuable insight into the coordinated stress-adaptation traits that may contribute to the persistence of environmental E. coli within agricultural systems. Although this study was not designed to evaluate intervention strategies, the observed co-occurrence of biofilm formation, acid tolerance, and antibiotic resistance highlights the capacity of environmental E. coli to persist across diverse niches. From a One Health perspective, such coordinated stress-adaptation traits may facilitate the circulation of bacteria and mobile genetic elements between environmental, food-associated, and host-associated systems. These findings emphasize the importance of integrating phenotypic, ecological, and genomic perspectives when assessing the persistence, adaptation, and dissemination potential of environmental E. coli within agricultural systems. Conclusion In conclusion, this study demonstrates the convergence of multiple stress-adaptation traits, including biofilm formation, acid tolerance, and antibiotic resistance, among environmental Escherichia coli, even in the absence of high-risk clinical resistance phenotypes. The observed convergence indicates that environmental selection pressures promote multifunctional persistence strategies rather than isolated phenotypic characteristics. Although overall antibiotic resistance levels were low, acquired resistance determinants were present and showed strong concordance with corresponding phenotypes, highlighting the role of environmental E. coli as reservoirs of mobilizable resistance genes. In contrast, biofilm formation and acid tolerance were not explained by single genetic determinants, supporting a multifactorial model of stress adaptation driven by regulatory and ecological factors. Together, these findings contribute to a better understanding of how environmental E. coli combine multiple stress-adaptation mechanisms to persist across diverse niches and emphasize the importance of considering trait convergence when evaluating bacterial survival and dissemination. Supplementary Information Supplementary Material 1: Table S1: Genome Assembly Statistics. Table S2: Virulence and Antimicrobial Resistance Genes (ARGs) information. Table S3: Plasmids information. Table S4: Absorbance, Growth, and calculated Biofilm Formation Index (BFI) for 18 E. coli isolates. Table S5: Log reduction of 18 E. coli isolates exposed to acetic acid at varying concentrations (0.1–0.5%), with standard deviations. Supplementary Material 1: Table S1: Genome Assembly Statistics. Table S2: Virulence and Antimicrobial Resistance Genes (ARGs) information. Table S3: Plasmids information. Table S4: Absorbance, Growth, and calculated Biofilm Formation Index (BFI) for 18 E. coli isolates. Table S5: Log reduction of 18 E. coli isolates exposed to acetic acid at varying concentrations (0.1–0.5%), with standard deviations.