Seasonal assessment of microbial indicators, antibiotic resistance, and heavy metal tolerance in water, sediment, and edible fish from an urban tropical coastal lagoon (Jansen Lagoon, São Luís, Brazil)

Urban coastal lagoons frequently receive untreated sewage and stormwater runoff, favoring fecal contamination and the selection of antibiotic-resistant bacteria. This study evaluated seasonal patterns of microbial indicators, antibiotic resistance, and heavy metal tolerance in different environmental compartments of Jansen Lagoon (São Luís, Brazil). Five sampling campaigns were conducted in 2018 during rainy (February–April) and dry (July–November) seasons. Surface water (n = 25), sediment (n = 15), and fish (n = 10) were collected at five sites. Thermotolerant coliforms and Pseudomonas spp. were enumerated by MPN; presumptive Clostridium perfringens was detected on SPS agar; and Enterobacterales were identified by biochemical tests. Antimicrobial susceptibility was assessed by disk diffusion following CLSI guidelines. Heavy metal tolerance (Pb and Cu) and physicochemical parameters were also evaluated. Physicochemical parameters varied seasonally, with higher TDS during the rainy season and consistently alkaline pH (8.4–8.7). Thermotolerant coliforms peaked in rainy-season water samples (up to 1,600 MPN/100 mL), while sediments showed higher counts during the dry season. A total of 130 gram-negative strains (19 species) were isolated, predominantly Escherichia coli. High resistance was observed to cephalothin (70%) and tetracycline (41%), while chloramphenicol susceptibility was high (91%); five species met MDR criteria. Pseudomonas spp. were more abundant in fish and largely susceptible to gentamicin (92.5%). Presumptive C. perfringens was detected only in the dry season. Most isolates tolerated high Pb concentrations (3500 ppm), and Cu tolerance was dose-dependent. Jansen Lagoon shows recurrent fecal contamination and significant levels of antibiotic resistance across environmental compartments, with seasonal variation. These findings highlight potential co-selection mechanisms and reinforce the need for improved sewage management to mitigate human exposure risks within a One Health framework.

Urban aquatic ecosystems frequently receive untreated domestic effluents and stormwater, with direct implications for the microbiological safety of water and fish and for human exposure during recreation and consumption. The associated organic loads transport a broad diversity of fecal microorganisms from humans and other warm-blooded animals (Gonzalez et al., 2010; Pedrosa de Macena et al., 2021; Vashi et al., 2025). In addition to immediate health risks, characterizing microbial communities in urban surface waters—rivers, lakes, estuaries, and lagoons can reveal near-term hazards and the long-term ecological footprint of pollution (McLellan et al., 2015).

In Brazil, microbiological monitoring typically relies on fecal indicator bacteria (FIB), notably thermotolerant coliforms, Enterococcus, and Escherichia coli (CONAMA, 2005). E. coli, within the order Enterobacterales, remains a key indicator of fecal contamination and exhibits metabolic versatility that supports persistence across environmental gradients, including brackish and saline waters of the São Luís port complex (Costa et al., 2014; Araújo et al., 2017). Nevertheless, the performance of a single indicator can vary with climatic and physicochemical conditions; therefore, complementary indicators, such as Salmonella, Vibrio, Clostridium perfringens, Pseudomonas aeruginosa, and Aeromonas, are often employed (Gonzalez et al., 2010).

In parallel, concern over antimicrobial resistance (AMR) is increasing (World Health Organization, 2024). Inputs of antibiotics and resistant bacteria via sewage, coupled with horizontal gene transfer in receiving waters, can select for resistant lineages and disseminate resistance determinants across environmental compartments. In low- and middle-income settings, extensive antibiotic use in human and animal health and in food production further intensifies selection pressure, threatening sustainable development (Vasconcellos et al., 2006; Farias et al., 2010; Shu et al., 2025). From a health perspective, urban coastal lagoons are critical sentinels because they integrate watershed discharges, tidal exchange, and intense human use (Pedrosa de Macena et al., 2021).

However, the maintenance of resistance reservoirs extends beyond antibiotic inputs; persistent heavy metals, particularly copper and lead, create long-term selective windows for bacterial adaptation that outlast degradable drugs (Murray et al., 2024). These metals, ubiquitous in urban runoff and antifouling residues, accumulate to critical concentrations where they act as potent drivers of antibiotic resistance (Seiler and Berendonk, 2012). In this context, accurate detection is paramount; despite the emergence of genotypic and microfluidic technologies, conventional phenotypic methods (e.g., disk diffusion) remain the gold standard for confirming functional resistance and ensuring data comparability in such surveillance programs (Salam et al., 2023).

In Jansen Lagoon, the waterbody and its margins are widely used for informal leisure (bathing, low-complexity water sports, recreational fishing) and for subsistence artisanal fishing, often by families with lower income and limited access to alternative food sources. For these groups, locally caught fish provide affordable protein and sometimes supplemental income, whereas the lagoon itself serves as a free recreational space. Such uses expand exposure pathways, dermal contact and incidental ingestion of water. Furthermore, since the harvested fish constitutes a primary protein source for the community, the assessment of fish tissues is essential to elucidate foodborne transmission risks. This reinforces the relevance of assessing microbiological contamination and AMR profiles as matters of environmental justice and One Health in addition to fish consumption, reinforcing the relevance of assessing microbiological contamination and AMR profiles as matters of environmental justice and One Health, particularly during the rainy season, when diffuse sewage inputs and sediment resuspension tend to increase bacterial loads (Cutrim et al., 2019).

Despite its touristic importance and documented urbanization, the microbiota of Jansen Lagoon (São Luís, Maranhão, Brazil) remains poorly characterized after decades of shoreline occupation and intermittent sewage inputs (Cantanhêde et al., 2014). In particular, an integrated assessment spanning water, sediment, and fish and linking seasonal forcings (rainy vs. dry) with bacterial indicators and resistance profiles is still lacking. Understanding seasonal variability is critical in this tropical system, where rainfall patterns drive contaminant transport via stormwater runoff and sediment resuspension. Furthermore, analyzing correlations between physicochemical parameters and microbial loads is essential to identify the environmental factors that favor bacterial persistence and the potential co-selection of antibiotic resistance by heavy metals.

Here, we (i) characterized physicochemical conditions in situ; (ii) quantified thermotolerant coliforms and enumerated Pseudomonas spp., with presumptive detection of C. perfringens; (iii) isolated and identified Enterobacterales (including E. coli); and (iv) determined disk-diffusion susceptibility to commonly used antimicrobials. We evaluated whether contamination levels and resistance patterns vary seasonally and covary with basic water quality parameters. This matrix-to-matrix approach provides a baseline for risk assessment and management in a heavily used urban coastal lagoon.

Study area

The Jansen Lagoon (Figure 1) is a coastal water body spanning approximately 140 ha with an average depth of approximately 1.5 m and is in the northwestern sector of São Luís Island, northeastern Brazil (02°29′08′′ S; 044°18′02′′ W). It connects to São Marcos Bay via the Ana Jansen tidal channel. Its hydrodynamics are governed by a tropical climate featuring well-defined wet and dry seasons, with the eastern border being delineated by mangrove forest (Cutrim et al., 2019).

Sampling was conducted at five georeferenced stations, three near the margin and two central points. L1 (2°30′06.0′′ S 44°17′53.0′′ W) is a semienclosed, stagnant area with degraded mangrove margins; L2 (2°30′06.0′′ S 44°18′00.1′′ W) is located near the center of the lagoon, representing general water conditions; L4 (2°30′01.5′′ S 44°18′17.7′′ W) is situated in the southwest area and is influenced by the input of saltwater from São Marcos Bay; and, finally, L3 (2°29′56′′ S 44°18′23′′ W) and L5 (2°30′07.7′′ S 44°18′14.9′′ W) are sites significantly impacted by the release of untreated domestic sewage from the local neighborhood.

Sampling and measurement

The collection of samples was undertaken at Laguna da Jansen covering the period from February to December 2018, with the objective of assessing the influence of the wet and dry seasons on the parameters analyzed. Water samples were collected from the subsurface (∼30 cm) in sterile 500 mL bottles. Following collection, all samples were promptly stored in insulated boxes containing crushed ice and transported on the same day to the Laboratory of Food Microbiology and Water Physical Chemistry (PCQA-UFMA) for subsequent processing.

Additionally, sediment samples were procured from three distinct stations—two situated adjacent to urbanized areas (stations 3 and 4) and one central station. The sediment was collected using a metallic dredge (∼500 g per sample), transferred to clean plastic containers, and kept refrigerated until analysis. Regarding fish, two individuals from distinct species were analyzed per sampling campaign (Table 1). The fish were captured and maintained on ice until laboratory processing. The collection of water and fish was conducted under authorization from IBAMA/SISBIO (Permit No. 65972−1) as part of the federal environmental licensing process. The following physicochemical parameters were measured in situ using a HANNA HI-98280 multiparameter probe: pH, total dissolved solids (TDS), conductivity, temperature, and salinity.

Microbiological analysis

Microbiological procedures were conducted in a sterile environment, following which all materials were sterilized by autoclaving (121 °C for 15 min). Sample preparation protocols were distinct for each matrix. For sediment samples, a 10 g aliquot was aseptically weighed and transferred into 90 mL of sterile saline solution (0.85% NaCl), followed by homogenization to establish a 1:10 dilution. Regarding fish samples, specimens were initially scaled, eviscerated, and filleted using sterile utensils to prevent external contamination. Subsequently, 10 g of the muscle tissue was aseptically excised and transferred to 90 mL of the diluent for homogenization at the same 1:10 ratio. Water samples were processed without prior dilution, being inoculated directly onto the culture media.

Thermotolerant coliforms (CTs) were determined using the Multiple-Tube Most Probable Number (MPN) technique. Aliquots of 10.1 and 0.1 mL were inoculated in triplicate (five tubes per dilution) into Lactose Broth (single- and double-strength) containing inverted Durham tubes. After 24–48 h of incubation at 35∘C, positive tubes (turbidity and gas production) were transferred to Escherichia coli (EC) Broth and incubated at 45∘C in a water bath for 24–48 h. Tubes showing turbidity and gas in EC Broth were deemed positive for thermotolerant coliforms, and the results were calculated using the Hoskins table (MPN/g for sediment and fish; MPN/100 mL for water). For the isolation of Enterobacterales, aliquots from positive EC tubes were streaked onto Eosine Methylene Blue (EMB) agar and incubated at 35 °C for 24 h. Suspicious colonies (black with a metallic green sheen) were subcultured onto Tryptone Soya Agar (TSA) and incubated at 37 °C for 24 h for purification and subsequent identification. The identification of suspected strains was performed using conventional biochemical tests, including the IMViC test (Indole, Methyl Red, Voges-Proskauer, and Citrate), motility, lysine decarboxylation, malonate, H 2 S production, and carbohydrate utilization (arabinose, sucrose, xylose, and raffinose), which is based on Bergey’s Manual of Systematic Bacteriology (Bergey et al., 2005) and supplemented by the Bactray I/II system (Laborclin, Paraná, Brazil), which comprises 20 biochemical assays. Detection of Pseudomonas aeruginosa was performed using the multiple-tube technique as described in Standard Methods (Apha, 2022). Aliquots of 10, 1, and 0.1 mL in triplicate were inoculated into Asparagine Broth. After 24–48 h of incubation at 35 °C, tubes exhibiting greenish fluorescence under ultraviolet light (presumptive for Pseudomonas spp.) were transferred to Acetamide Broth and reincubated under the same conditions. Differentiation was based on acetamide deamination: the development of a pink color was recorded as positive for P. aeruginosa, confirming the species and distinguishing it from other fluorescent pseudomonads which are acetamide-negative. For the detection of thermotolerant sulfite-reducing clostridia (SRC), the deep-tube inoculation (overlay) technique with selective SPS Agar was employed, following the APHA methodology (Apha, 2022). Aliquots of 0.1 mL were transferred to Petri dishes and mixed with Sulfite-Polymyxin-Sulfadiazine (SPS) Agar, with a second layer of agar (5–10 mL) added after solidification. The plates were incubated in an anaerobic jar at 46 °C for 24–48 h, and black colonies were registered as positive for SRC (presumptive C. perfringens).

Antimicrobial susceptibility

The isolates were activated in Brain Heart Infusion (BHI) broth for 24 h at 37 °C, standardized, and then seeded onto Mueller–Hinton agar. Antimicrobial agents were selected to represent distinct chemical classes widely used in clinical and veterinary settings in Brazil, serving as key indicators of anthropogenic contamination (Böger et al., 2021). Commercial discs (Laborclin, Brazil) were applied. For Enterobacteriales, the set included representatives of tetracyclines (tetracycline, 30 µg), cephalosporins (cephalothin, 30 µg), and phenicols (chloramphenicol, 30 µg). For Pseudomonas spp., the agents tested were penicillins (penicillin G, 10 µg), aminoglycosides (gentamicin, 30 µg), and sulfonamides (sulfamethoxazole–trimethoprim, 25 µg). After 24 h at 37 °C, the diameters of the inhibition zones (mm) were measured. Interpretation was performed according to the zone-diameter interpretive criteria detailed in Table 2, established in accordance with the manufacturer’s instructions (Laborclin) and CLSI (Clinical and Laboratory Standards Institute) guidelines (Clinical and Laboratory Standards Institute, 2015). The categories were defined as follows: Susceptible (S) implies that isolates are inhibited by the usually achievable concentrations of the antimicrobial agent when standard dosages are used; Intermediate (I) represents a category where response rates may be lower, and Resistant (R) indicates that usually achievable concentrations do not inhibit isolates.

Exposure of Enterobacteriales and Pseudomonas spp. to different heavy metal concentrations

Tolerance to copper (CuSO 4) and lead [Pb(NO 3) 2 ] was assessed by agar microdilution at concentrations of 500, 1500, and 3500 ppm based in the methodology adapted from (Ngwewa et al., 2022). Isolates maintained on Triptone Soya Agar (TSA) were precultured in nutrient broth (Himedia; 24 h at 37 °C). Subsequently, 10 µL of each culture was inoculated (spot) onto nutrient agar supplemented with metallic salts and incubated at 35 °C for 24 h. Growth was interpreted as tolerant; absence or marked reduction in growth was classified as susceptible. To investigate the influence of medium pH on metal solubility, a subset of nine representative strains was selected (three E. coli, three C. diversus, and three P. aeruginosa from water, sediment, and fish). The assay was performed in sterile microcentrifuge tubes containing 1 mL of Nutrient Broth supplemented with copper (Cu 2+) or lead (Pb 2+) at concentrations of 3500 ppm. Each tube was inoculated with 10 µL of a 24-h active culture and incubated at 37 °C for 48 h. Controls included uninoculated medium with metal (negative control) and inoculated medium without metal (positive growth control). After incubation, bacterial growth was initially confirmed by visual turbidity. To further validate cell viability and confirm pure culture recovery, 10 µL aliquots from each tube were subcultured onto Plate Count Agar (PCA) and incubated at 37 °C for 24 h. Subsequently, the final pH of the supernatant was assessed using universal pH indicator strips (Merck KGaA, Darmstadt, Germany) to detect significant alkalinization events capable of inducing metal precipitation.

Statistical analyses

All the data were log-transformed and tested by nonparametric statistics. A Pearson matrix was used to determine the correlations between microbiological and environmental measurements. Correlations were considered statistically significant when a significance level of 95% (p < 0.05) was found. The relationships among all the variables were assessed by principal component analysis (Hammer et al., 2001).

Análise de Dados

All data processing, statistical analysis, and graphical visualization were performed using Python programming language (version 3.x) (Van Rossum and Drake, 2009). The raw dataset was curated to standardize taxonomic nomenclature. The analysis focused on isolates belonging to the order Enterobacteriales, the genus Pseudomonas and unidentified isolates from the specific diversity calculations. The relative frequency of bacterial species was calculated for each environmental matrix (water, sediment, and fish) based on the total number of isolates recovered from each source (n water =82; n sediment =40; n fish =46). To improve visual clarity in the abundance analysis, species with a frequency lower than 2% across all sources simultaneously were filtered out. The remaining taxa were ranked by total abundance, and a heatmap was generated using the Seaborn library (Waskom, 2021) with a sequential color scale to represent relative frequencies.

Principal Component Analysis (PCA) was employed to evaluate the global structure of water quality data and identify environmental drivers, using scikit-learn (Pedregosa et al., 2011). Factor loadings and eigenvalues were used to determine the contribution of each physicochemical parameter to the seasonal variance. For temporal and spatial comparisons of microbial indicators and environmental parameters, data are expressed as mean values ± standard deviation. Statistically significant differences between sampling months and seasons were determined using the non-parametric Kruskal-Wallis test (p<0.05). Venn diagrams were constructed to evaluate the ecological overlap and exclusivity of bacterial communities among the three compartments. Two distinct scenarios were modeled: (i) total species diversity, considering the presence/absence of all identified taxa; and (ii) antimicrobial resistance profiles, specifically filtering for species exhibiting resistance to at least one antibiotic (R ≥ 1) and multidrug resistance to three or more antibiotics (R ≥ 3).

Water physical–chemical parameters

To evaluate the environmental drivers influencing microbial persistence and contaminant transport, we first characterized the region’s hydrological regimes. Based on the climatological series (INPE, 2018), precipitation between February and April 2018 characterized the wet season (≈280–537 mm), whereas the period from July to November corresponded to the dry season (≤94 mm). Air temperatures ranged from 24.8 to 27.0 °C, with the highest values recorded from August to December. Consequently, sampling campaigns were classified as wet (Feb, Apr) or dry (Jul, Sep, Nov).

Regarding pH (Figure 2A), the profiles generally evidenced a high buffering capacity with alkaline tendencies. During the wet season, mean pH values ranged from 8.36 (S4) to 8.57 (S3). In the dry season, values remained alkaline (mean 7.95–8.18) but exhibited increased variability due to a marked acidification event in November (pH ≈ 6.5). These values frequently exceed the upper limits recommended by Conama Resolution 357/2005 (6.5–8.5) for Class I–II brackish waters (CONAMA, 2005). This prevailing alkalinity is consistent with historical data for Jansen Lagoon (Cantanhêde et al., 2014; Cutrim et al., 2019) and likely reflects the shift in the carbonate-bicarbonate equilibrium driven by intense photosynthetic uptake of CO 2 (Malone et al., 2012). Conversely, the shift towards acidic conditions in the late dry season indicates the dominance of heterotrophic processes (respiration/decomposition). This quantitative acidification aligns with field observations of putrefactive odors and massive phytoplankton senescence (algal die-offs). Although local sanitation authorities reported mitigation of point-source sewage inputs during this period, the hydrochemical signature suggests that CO 2 production from the decomposition of accumulated autochthonous biomass was the primary driver of the observed pH reduction.

Water temperature exhibited high spatial thermal homogeneity and low temporal amplitude, ranging from mean values of 26.9 ± 0.5 °C (Station S1, Dry) to 29.6 ± 2.2 °C (Station S2, Dry) (Figure 2B). From a kinetic perspective, these elevated temperatures accelerate microbial metabolism and organic matter mineralization (Scofield et al., 2015), while simultaneously reducing gas solubility, potentially impacting the dissolved oxygen budget (Rojas and Costa Neto, 2014).

The hydrodynamic behavior of salinity, conductivity, and TDS revealed a decoupling from the typical estuarine dilution gradient, driven by anthropic interventions (Figures 2C–E). The wet season displayed the highest hydrochemical instability, with salinity ranging from 17.8‰ (S4) to 19.9‰ (S3). This variability reflects two contrasting hydrodynamic forcing mechanisms: (i) intense marine intrusion events in February (euhaline conditions, >30‰) facilitated by the opening of sluice gates for flood control; and (ii) efficient flushing and dilution in April (oligohaline conditions, <5‰) driven by peak precipitation.

In the dry season, the system exhibited greater stability but maintained polyhaline to euhaline characteristics. Salinity averaged 19.4–19.5‰, with conductivity ranging from 31.2 to 31.4 mS cm - ¹. Spatially, station S3 acted as a distinctive hydrological niche; while the main lagoon body followed the general saline trend, S3 frequently exhibited lower conductivity. Global mean salinity values observed in this study (Wet: 18.5‰; Dry: 19.4‰) align with historical patterns reported for the Jansen Lagoon (Cantanhêde et al., 2014; Cutrim et al., 2019). However, relying solely on global averages obscures the system’s spatial dynamics. Unlike previous studies that focused on general trends, our point-specific analysis reveals that these averages result from the interaction between intense marine intrusion (Station S4) and localized freshwater runoff (Station S3). Thus, the lagoon functions not as a homogeneous body, but as a complex mosaic driven by the interplay between sluice gate management and urban drainage.

Principal Component Analysis (PCA) was performed to evaluate the global structure of water quality data (Figure 3). The first two components (eigenvalues > 1.0) accounted for 90.9% of the total variance (Supplementary Table 1). The first axis (PC1), explaining 66.7% of the variation, represented the ionic gradient. It showed strong positive loadings for conductivity (0.59), salinity (0.58), and TDS (0.54) (Supplementary Table 2), clearly distinguishing samples with high marine influence (February/September) from those diluted by rainfall (April).

The second axis (PC2) explained 24.2% of the variance and was strongly and positively correlated with pH (0.98). This indicates that variations in alkalinity/acidity occur independently of the salinity gradient. This axis was crucial in differentiating the dry season samples, particularly highlighting the drop in pH observed in November (positioned negatively on PC2), likely associated with organic matter decomposition and restricted water exchange, contrasting with the alkaline conditions of other months. The observed alkalinity is consistent with previous studies in Jansen Lagoon (Cantanhêde et al., 2014; Cutrim et al., 2019) and may reflect the accumulation of bicarbonates/carbonates, elevated photosynthetic activity on sunny days (Malone et al., 2012), or the influence of spring tides, which are known to increase pH in estuaries (Wanderley et al., 2020; Santos et al., 2023).

Seasonal variation in microbial abundance in water, sediment and fish

The enumeration of microbial indicators revealed distinct contamination gradients across the sampling stations. In the water column (Figure 4A), high bacterial loads were pervasive, but spatial heterogeneity was evident. Station S4 (sluice gate) and Station S3 (near restaurants) consistently presented the highest contamination levels. For Total Coliforms (TC), Station S4 frequently reached the upper quantification limit, with concentrations ranging from 3.6×10 2 to 1.5×10 3 MPN 100 mL - ¹ (corresponding to 2.56 to 3.20 log 10 units), maintaining these high values in February, April, and July. According to the Brazilian Conama Resolution 357/2005, 44% of the samples exceeded the threshold for Class 2 waters (3.0 log 10 MPN 100 mL - ¹, equivalent to 1,000 MPN 100 mL - ¹). However, the annual mean remained within the established criteria (<2,500 MPN 100 mL - ¹, or <3.40 Log 10 units) (CONAMA, 2005). The growth profile of P. aeruginosa followed a similar trend to TC, with mean concentrations ranging from 3.9 to 2.4×10 3 CFU 100 mL - ¹ (0.59 to 3.38 log 10 units).

Similarly, Pseudomonas spp. counts (Figure 4D) at S4 were not ably high, reaching 2.4×10 3 CFU 100 mL - ¹ (3.38 log 10 units) during the rainy month of April and remaining elevated (4.5×10 2 CFU 100 mL - ¹; 2.66 log 10 units) even in September. In contrast, Station S1 exhibited the lowest baseline values during the dry season, with TC counts dropping to 7 MPN 100 mL - ¹ (0.85 log 10 units) in July, suggesting a greater influence of marine dilution or distance from point sources. However, during peak rainfall (April), spatial differences were homogenized, with all stations (S1–S5) exhibiting maxima for both TC (≈ 1.5×10 3 MPN 100 mL - ¹) and Pseudomonas spp. ≈ 2.4×10 3 CFU 100 mL - ¹).

Similar seasonal elevations in coastal areas of São Luís have been attributed to domestic sewage runoff and increased beach usage during the rainy season (Santos et al., 2021), as well as to the dilution/resuspension processes that accompany rainfall events (Rojas and Costa Neto, 2014). Furthermore, UV radiation during the dry period can reduce E. coli populations in coastal waters (Burkhardt, 2000; Rozen and Belkin, 2001), which helps explain the relative decrease observed outside the peak periods.

In the sediment (Figures 4B, E), contamination was pronounced at Station S3 and Station S4, where TC counts consistently exceeded those in the water column up to 2.4×10 3 MPN g - ¹ (3.38 log 10 units) in November, reinforcing the role of sediment as a reservoir and source of recontamination via resuspension (Karbasdehi et al., 2017). The growth behavior of P. aeruginosa remained similar to that detected in the water, with mean values ranging from 0 to 2.4×10 3 CFU g - ¹ (0 to 3.38 log 10 units). The literature indicates that bacterial density increases with increasing grain size and organic matter accumulation (Santos et al., 2008), characteristics already described for the silts of Jansen Lagoon (Rojas and Costa Neto, 2014). In April, the water exceeded the sediment, a pattern consistent with the export of organic matter/feces to the water column during intense rainfall.

In fish (Figures 4G, H), TC counts were generally low (<91 MPN g - ¹; <1.96 log 10 units), except for Albula vulpes in July (2.4×10 3 MPN g - ¹; 3.38 log 10 units), possibly because of their carnivorous habit and ingestion of sediment/benthic items (Lopes, 1999; Barboza et al., 2014). For regulatory purposes, it is important to note that the current Brazilian legislation does not establish a limit for thermotolerant coliforms in fresh fish, only limits for E. coli (1.0×10 2 CFU g - ¹; 2 log 10 units) (Brazil, 2022). In contrast, the growth of Pseudomonas spp. remained high from February to September (2.4×10 3 CFU g - ¹; 3.38 log 10 units), with the exception of November (12 CFU g - ¹; 1.08 log 10 units). The recurrence of high counts in fish is compatible with the adaptation of this bacterium to the gastrointestinal tract/tissues of marine and estuarine species (Egerton et al., 2018).

The seasonal distribution of Clostridium spp. (Figures 4G, H) in Jansen Lagoon presented the greatest environmental distinction, being exclusively detected in water and sediment during the dry season. The genus was not detected in the rainiest months (February and April), which contrasts with the maximum load of other contamination indicators. Conversely, the maximum detection of Clostridium spp. occurred in July, September, and November, which are periods typically associated with the dry season. These data demonstrate that the detection of Clostridium spp. is inhibited by conditions of high dilution and greater water column oxygenation during the rainy season. In contrast, its presence in the dry season can be explained by the role of the sediment as a reservoir of spores, and temperature is a representative indicator of the anoxic or hypoxic conditions at the lagoon bottom created during water column stability rather than surface contamination. However, in this study, Clostridium spp. was not detected in the muscle tissue of any fish sample. Although Clostridium is a common gut symbiont (Egerton et al., 2018), our analysis focused on muscle tissue (edible part). The absence of Clostridium confirms that the gut barrier effectively prevented translocation, indicating the fish were healthy and the muscle was free contamination.

The Pearson correlation matrix revealed a direct relationship between the physical-chemical variables and microbial contamination in the water of Jansen Lagoon (Figure 5). The water quality variables salinity and total dissolved solids (TDS) exhibited the strongest positive correlation (ρ=0.99), confirming that saline intrusion is the primary regulator of the ionic content in the lagoon. Notably, precipitation presented the most intense correlation in the entire matrix with pH (ρ=−0.84). This suggests that intense rainfall events act as the main modulator of water acidity, likely by diluting the carbonate-bicarbonate buffering capacity of the system and introducing slightly acidic runoff from the urban drainage.

In terms of the bacterial indicators, the abundance of Pseudomonas spp. exhibited a strong positive correlation with total coliforms (TCs) (ρ=0.58), suggesting that these bacterial groups, which are frequently associated with effluents and organic matter, are introduced into the system through the same pathways, primarily the continuous discharge of untreated domestic effluents and organic matter accumulation. Pseudomonas spp. has halotolerance in coastal environments, where salinity does not impose a strong limitation, which explains its presence in the water samples from Jansen Lagoon (Rodriguez et al., 2015; Karbasdehi et al., 2017).

The genus Clostridium was also positively correlated with TC (ρ=0.23) and turbidity (ρ=0.22). Notably, the concentrations of Pseudomonas and Clostridium did not present strong correlations with the main dilution factors (salinity and TDS), indicating that their population dynamics in the lagoon may be more influenced by continuous effluent discharge and particulate matter than by seasonal fluctuations in salinity or precipitation.

Diversity of Enterobacteriales and Pseudomonas spp. in Jansen Lagoon

Phenotypic characterization allowed for the recovery of a total of 168 bacterial isolates from water, sediment, and fish samples. The analysis of the isolation frequency revealed that Escherichia coli, a traditional indicator of fecal contamination, was the most prevalent species within the global sample set. A total of 65 isolates of E. coli were obtained, corresponding to 44.6% of the total recovered microorganisms.

In comparison, isolates belonging to the genus Pseudomonas represented the second most abundant group, with 38 isolates recovered (22.6% of the total). Figure 6 presents a heatmap of the percentage distribution of species. While E. coli showed a consistent presence across all matrices, the data reveals a distinct ecological pattern for Pseudomonas spp., which exhibited its highest prevalence in fish samples (n=19 isolates), surpassing E. coli (n=15 isolates) in this specific matrix.

The quantitative prevalence of E. coli in the total sampling suggests a significant influence of fecal contamination sources in the environment studied. However, the substantial recovery of Pseudomonas spp., particularly in the ichthyofauna, highlights the competitive fitness of this group. Pseudomonas species are ubiquitous and metabolically versatile, traits that favor their colonization of aquatic organisms and persistence in the environment. Unlike E. coli, which may decline outside the host or enter a viable but non-culturable (VBNC) state, Pseudomonas spp. naturally thrives in water and sediment matrices.

Therefore, the results indicate a dual profile of the area studied, a widespread fecal impact indicated by the high overall frequency of E. coli and a co-occurring with a robust population of opportunistic pathogens like Pseudomonas, whose isolation success was particularly evident in the biotic (fish) matrix.

The water column exhibited the highest diversity of Enterobacteriales, including exclusive genera such as Hafnia, Providencia, and specific Klebsiella species that were not detected in the sediment or fish. Conversely, the sediment compartment showed a significant presence of Citrobacter freundii (15.0%), which was less frequent in water (2.4%) and fish (4.3%).

The composition is typical of impacted surface waters (Nascimento et al., 2016) and is consistent with regional findings in the São Luís Port Complex (Costa et al., 2014) and in aquatic and river systems in Northeast Brazil (Parente et al., 2011). The prolonged viability of enterobacteria in seawater, even with salinity, has been attributed to latency and to physical factors such as UV radiation and organic matter (Rozen and Belkin, 2001). Stenotrophomonas maltophilia and Klebsiella spp., when present, are opportunistic pathogens associated with respiratory and urinary infections, particularly in immunocompromised individuals (Rodrigues et al., 2011; Brooke, 2012).

Antimicrobial susceptibility of Enterobacteriales and Pseudomonas spp.

The antibacterial activities of three commercial antibiotics against Enterobacteriales isolated from Laguna da Jansen are summarized in the clustered heatmap (Figure 7A). Among Enterobacteriales, higher resistance was observed to cephalothin (70%), a first-generation β-lactam antibiotic, followed by tetracycline (41%), whereas a marked level of susceptibility was recorded to chloramphenicol (73%). Similar patterns of elevated resistance to tetracyclines and sulfonamides among environmental E. coli have been reported in Brazil (Vasconcelos et al., 2010; Beleza et al., 2019) and in estuarine systems in Ceará, including tetracycline-resistant Salmonella strains (Figueredo et al., 2015).

The hierarchical clustering heatmap highlights phenotypic relatedness among species, revealing that Citrobacter, Klebsiella and Enterobacter spp. grouped tightly and formed the major branch of multidrug-resistant taxa. This clustering reinforces the emergence of resistance within clinically relevant genera commonly associated with opportunistic and community-acquired infections. In contrast, species exhibiting lower resistance frequencies, such as Escherichia fergusonii and Hafnia alvei, segregated into peripheral branches, reflecting reduced phenotypic similarity. The gradient distribution further indicates that cephalothin resistance was pervasive, whereas chloramphenicol susceptibility remained relatively conserved.

The Venn diagram (Figure 7B) depicts the distribution of Gram-negative isolates across water, sediment and fish, highlighting both unique and shared taxa. When filtering the community for resistance traits, the distribution pattern of species resistant to at least one antibiotic (R ≥1) closely mirrored the total diversity profile. The core resistant community is shared among all sources comprised of five species (E. coli, Pseudomonas spp., Citrobacter diversus, Citrobacter freundii and Citrobacter amalonaticus). This similarity indicates that antimicrobial resistance is a widespread trait within the dominant bacterial populations rather than being restricted to specific or rare strains.

Notably, while Salmonella choleraesuis was part of the shared core in the total diversity analysis, resistant strains of this species were not recovered simultaneously from all compartments, suggesting differential selective pressures or distinct contamination sources. Multidrug resistance (≥ 3 antibiotics) was a rare event in this dataset, observed exclusively in E. coli isolates from the sediment, highlighting the benthic zone as a potential reservoir for the accumulation of multidrug-resistant bacteria.

These observations raise concern from a public health perspective, since high resistance within Enterobacteriales circulating in urban coastal systems increases the likelihood of transfer to human-associated settings, especially via recreational water use, consumption of aquatic organisms, or wastewater exposure. The similarity of resistance phenotypes detected in our isolates and those reported in clinical environments suggests that coastal waters may act as reservoirs or dispersal interfaces for multidrug-resistant Enterobacteriales. Supplementary Table 5 provides the absolute and relative frequency distribution of resistance for each species.

For Pseudomonas spp., sulfamethoxazole–trimethoprim and penicillin G exhibited intrinsic resistance, consistent with the well-established lack of activity of these agents against this genus; therefore, resistance was 100%. Consequently, only gentamicin produced interpretable inhibition patterns according to CLSI criteria. Gentamicin resistance was low, with no resistant isolates recovered from water or sediment and only 5% resistance among fish isolates, corresponding to 92.5% susceptibility overall (n= 37/40). This pattern reinforces the intrinsic antimicrobial resistance profile of Pseudomonas spp. and indicates that measurable variability was restricted to aminoglycoside susceptibility. High gentamicin susceptibility among fish isolates has also been reported elsewhere (Lamari et al., 2017), whereas urban and hospital effluents are known to select for gentamicin-resistant Pseudomonas in surface waters (Fuentefria et al., 2008).

In the context of Laguna da Jansen, across the interface environments of water, sediment, and fish, the presence of these MDR strains signifies that the environment may act as a reservoir and vector for the dissemination of resistance genes to human and animal microbiota, which could compromise the efficacy of treatments, such as those used in cholera cases (Figueredo et al., 2015).

The low susceptibility to β-lactam antibiotics, such as cephalothin, is often associated with the inherent low permeability of Gram-negative outer membranes to this class (Rodrigues et al., 2011). Tetracyclines are among the most frequently detected antibiotics in lake environments and strongly adsorb to sediments (Yang et al., 2018), which reinforces the ecological plausibility of the phenotypic resistance patterns observed. In aquatic ecosystems, inputs from domestic, agricultural, and aquaculture effluents create persistent selective pressure favoring resistant Gram-negative bacteria (Fuentefria et al., 2008; Vasconcelos et al., 2010). Water environments have been highlighted as key environmental reservoirs for resistant bacteria such as E. coli, Klebsiella pneumoniae and Pseudomonas spp., which may spread to humans through consumption or exposure pathways (Meradji et al., 2025).

The presence of multidrug-resistant (MDR) isolates (≥3 drug classes), including E. coli, E. fergusonii, K. ornithinolytica, Kluyvera cryocrescens, and Providencia spp., is alarming, as the WHO classifies Enterobacterales and Pseudomonas spp. among high-priority pathogens for new drug development (World Health Organization, 2024). Although this study relied on phenotypic testing rather than genomic confirmation, the detection of MDR phenotypes in environmental matrices aligns with findings that environmental resistomes, including sewage, harbor multiple resistance mechanisms supporting survival under antimicrobial stress and diverse MDR gene repertoires (Seiler and Berendonk, 2012; Ramos et al., 2024).

Importantly, environmental settings themselves may facilitate persistence and selection of resistance beyond antibiotic exposure, since metals, disinfectants and other contaminants can co-select for antibiotic resistance through cross-resistance and co-regulation pathways and heavy-metal-impacted aquatic systems have been proposed as hotspots for such co-selection (Su et al., 2017; Quadra et al., 2023). Moreover, global assessments predict that untreated effluents, aquaculture areas and urban water bodies are among the environments of greatest risk for sustaining resistant bacterial populations (Quadra et al., 2023).

Within the context of Laguna da Jansen, MDR isolates across water, sediment and fish reinforce that this system may function as a reservoir and dissemination interface for resistant Enterobacterales. While this study did not evaluate genetic determinants, the phenotypic detection of multidrug resistance indicates public-health relevance, since environmental exposure of humans to resistant bacteria, through water contact, seafood consumption or recreational activities, is a documented transmission pathway (Meradji et al., 2025). These findings underscore the need for surveillance in tourist-impacted coastal systems and strengthen global One-Health concerns that environmental AMR reservoirs contribute to human exposure, treatment failure and increased disease burden, especially in low- and middle-income regions (Al-Khalaifah et al., 2025).

Effects of heavy metals on the growth of Enterobacteriales and Pseudomonas sp.

The hierarchical clustering heatmap (Figure 8A) further illustrates phenotypic differentiation at the metal tolerance level, revealing that Klebsiella, Enterobacter and Citrobacter spp. consistently grouped as highly tolerant taxa following 24 h exposure to Cu 2+ and Pb 2+. This clustering aligns with their resistance patterns observed in antibiotic assays, suggesting that members of these genera may possess co-selection mechanisms or adaptive traits favoring survival under chemically stressed conditions. In contrast, isolates such as Hafnia alvei and Enterobacter aerogenes segregated into peripheral branches, reflecting reduced growth under metal challenge and reinforcing their lower phenotypic robustness. The gradient distribution indicates that tolerance to elevated Cu 2+ concentrations was particularly widespread, whereas Pb tolerance appeared more variable across taxa.

The highest rate of resistance to copper was observed at the lowest concentration (500 ppm), where most species from all sources exhibited 100% growth. However, increasing the concentration to 1500 ppm and, especially, to 3500 ppm revealed selectivity. Water-isolated species, such as E. coli (35 strains) and Citrobacter diversus (9 strains), were the most tolerant, with a significant portion of their populations resisting 3500 ppm copper. Conversely, tolerance to 3500 ppm copper was low or null for almost all strains isolated from fish and sediment, except for Citrobacter freundii and Citrobacter kasei in the sediment. These findings suggest that the water column, characterized by a higher taxonomic diversity of tolerant isolates, may act as a significant reservoir for copper-resistance traits in Jansen Lagoon. However, further studies are required to confirm if this environmental compartment serves as the primary driver for the horizontal transfer of resistance genes across the lagoon’s biota.

Regarding lead (Pb 2+) exposure, a distinct phenotypic pattern was observed compared to copper. Bacterial proliferation was ubiquitous across all matrices, occurring even at the highest nominal concentration of 3500 mg·L −1 (3500 ppm). While such elevated concentrations inherently challenge the concept of bioavailability due to solubility limits, raising valid concerns regarding precipitation, our data elucidates that this survival is not merely an artifact but a result of biologically mediated environmental modulation. As detailed in Figure 9, isolates exposed to Pb 2+ induced a marked alkalinization of the medium, shifting the pH from an initial ~6.0–7.0 to approximately 8.0 in several strains.

This pH elevation appears to be the deterministic factor for the observed resilience, allowing these isolates to exceed tolerance thresholds typically reported in recent literature. For instance, (Kumar and Singh, 2025) characterized bacteria from mining tailings with tolerances up to 300 mg·L −1 for Cu/Fe solutions, while Kumar and Singh (2025) identified specific strains of Microbacterium, Serratia, and Klebsiella pneumoniae capable of tolerating up to 800 mg·L −1 of lead. Corroborating the taxonomic data from (Kumar and Singh, 2025), our hierarchical clustering (Figure 8A) also identified Klebsiella and Enterobacter spp. as highly tolerant taxa.

Furthermore, a distinct visual change was observed in colonies exposed to Lead, which developed a characteristic brown pigmentation surrounding the growth area (Figure 8A-II). This chromatic alteration is consistent with the extracellular sequestration and precipitation of lead into insoluble complexes, a well-recognized mechanism for heavy metal detoxification. Similar phenotypic observations were reported by (Naik et al., 2013), who demonstrated that lead-resistant bacteria can bioprecipitate soluble Pb 2+ as insoluble brown compounds (such as lead phosphates) through enzymatic activity, effectively reducing the metal’s bioavailability. This phenotypic evidence, combined with the observed medium alkalinization, reinforces that these isolates from Jansen Lagoon employ active biochemical strategies to mitigate lead toxicity, effectively reducing its bioavailability through environmental modulation. However, a crucial mechanistic distinction exists, while previous studies often focus on removal efficiency at controlled neutral pH, the isolates in the present study actively induced an alkaline shift, permitting survival at nearly four-fold the concentration reported by Kumar and Singh (Kumar and Singh, 2025).

Furthermore, Abbas et al. (2025) documented Pb tolerance up to 2100 mg·L - ¹ in industrial isolates, attributing resistance primarily to biosorption. The persistence of growth at 3500 mg·L - ¹ in our isolates suggests that, while intracellular resistance and biosorption remain effective at moderate concentrations, extreme tolerance is achieved through pH-driven phosphate precipitation. As described by McBride et al. (2019), organic ligands may enhance Pb solubility under acidic conditions, but this effect is strongly suppressed above pH 6.0. Under alkaline conditions (pH ≈ 8.0), phosphate precipitation dominates, immobilizing Pb into thermodynamically stable mineral phases such as hydroxypyromorphite. Consequently, widespread growth at 3500 ppm should be interpreted as an active process of niche construction rather than direct tolerance to bioavailable Pb² +.

Qualitative observations under UV light (365 nm) revealed that metal exposure influenced the production of fluorescent secondary metabolites in the isolates. Specifically, strains inoculated in media containing lead (Pb) exhibited a marked emission of fluorescence (Figure 10), which is descriptively consistent with the induction of pyoverdine-like siderophores as a response to metal stress (Ramalho et al., 2002; Kothari et al., 2022; Venil et al., 2020; Abdelaziz et al., 2023). Notably, copper exposure suppressed pigment production, likely due to oxidative stress or thiol-binding toxicity, lead induced siderophore synthesis, suggesting a metal-specific regulatory response (Lear et al., 2022). This response aligns with the mechanistic framework proposed by (Fatima et al., 2025), who demonstrated that Pseudomonas aeruginosa employs siderophores, particularly pyoverdine and pyochelin, to chelate lead ions as a detoxification strategy. In the isolates from Jansen Lagoon, pyoverdine thus appears to fulfill a dual function, mediating both iron acquisition and heavy metal sequestration.

Taken together, the secretion of chelating agents and the induction of alkaline conditions represent a multilayered resistance strategy. These findings support previous calls for mechanistic approaches to bioremediation (Kumar and Singh, 2025), demonstrating that under extreme metal stress, the synergistic interaction between environmental alkalinization (precipitation) and siderophore-mediated chelation is critical for overcoming metal toxicity by limiting bioavailability.

The Venn diagram analysis (Figure 8B) further revealed that most isolated species were capable of growth in the presence of at least one tested metal, irrespective of their matrix of origin. A core group of six species, E. coli, Pseudomonas sp., Citrobacter freundii, C. diversus, C. amalonaticus, and Salmonella choleraesuis, exhibited shared tolerance across water, sediment, and fish. Notably, the water column harbored the highest diversity of tolerant organisms, including eight exclusive species such as Klebsiella pneumoniae, Hafnia alvei, and Stenotrophomonas maltophilia.

Species unique to sediment (C. kosei, Enterobacter agglomerans) and fish (Kluyvera cryocrescens) also demonstrated metal tolerance, indicating that this trait is broadly distributed across the cultivable bacterial community rather than confined to specific genera. The coexistence of shared and niche-specific tolerant taxa suggests that metal resistance contributes both to habitat specialization and to cross-matrix dissemination, underscoring its ecological relevance in structuring bacterial assemblages within the lagoon.

Comparable patterns of heavy metal- and antibiotic-resistant E. coli have been reported from coastal environments in São Luís, Maranhão, a known source of water input to Jansen Lagoon (Raposo Barros et al., 2014). These observations reinforce evidence that metals such as copper exert strong selective pressure on microbial communities (Ngwewa et al., 2022; Nnaji et al., 2024), in contrast to lead, which primarily induces indirect tolerance mechanisms. While certain bacteria can transform metallic ions via volatile metabolic products to mitigate toxicity (Essa et al., 2018), lead resistance is more commonly associated with phosphate precipitation, intracellular sequestration, and efflux systems (Ngwewa et al., 2022), explaining the broad tolerance observed among Enterobacteriales and Pseudomonas spp.

Finally, resistance to heavy metals, particularly copper, may act as a co-selective force for antibiotic resistance genes (Shu et al., 2025). In this study, E. coli and C. diversus exhibited both high copper tolerance and elevated resistance to tetracycline and cephalothin. This co-occurrence suggests that chronic metal contamination in Jansen Lagoon may favor the persistence of multidrug-resistant bacteria, posing potential risks to environmental and public health.

Principal component analysis (PCA) was used to explore the correlation between antibiotic resistance and heavy metal resistance. As shown in Figure 11, in the PCA for isolates belonging to Enterobacterales, the first two principal components (PC1 and PC2) explained 57.7% of the total data variance. PC1 (34.5%) was predominantly characterized by resistance to copper at the 3500 ppm concentration (Cu 3.5, loading of +0.682), whereas PC2 (23.2%) was characterized by resistance to antibiotics, with a focus on chloramphenicol (A3, loading of +0.748). Copper and antibiotic resistance were inversely proportional, demonstrating that copper does not exert selective pressure on the resistance of Enterobacterales to commercial antibiotics. Furthermore, isolates from water and fish clustered on PC1, indicating a tendency toward greater copper resistance, whereas those from fish presented greater resistance to antibiotics.

In the specific PCA for Pseudomonas, which explained 63.9% of the total variance across the first two axes, correlations between antibiotic resistance and heavy metal resistance became evident and highly environmentally specific (Figure 12). Axis 1 (PC1, 39.1%) was strongly defined by resistance to copper at the 3500 ppm concentration (loading of +0.824), whereas Axis 2 (PC2, 24.8%) was dominated by resistance to the antibiotic sulfamethoxazole–trimethoprim (A2, loading of +0.783). These results show clear coresistance among the antibiotic sulfamethoxazole–trimethoprim and the metals copper at 1500 ppm (Cu 1.5) and lead at 3500 ppm (Pb 3.5), whose vectors are located in the same quadrant. This multiple resistance profile was primarily concentrated in the fish isolates, indicating that cross-selective pressure resulted in the manifestation of combined resistance. In contrast, the majority of Pseudomonas isolates from water and sediment projected onto the low-resistance side for copper (3500 ppm) and sulfamethoxazole–trimethoprim, suggesting that the coselection or pressure for the development of resistance to these variables is less intense in those environments.

This integrated study of Laguna da Jansen revealed a chronic and complex contamination risk, despite the water body partially meeting local regulatory standards. Critically, while elevated thermotolerant coliforms are periodically detected in the water, the consistently high counts in the sediment establish it as a persistent pathogen reservoir, where contaminant resuspension by rainfall poses an enduring threat to public health and recreational safety. The detection of nineteen Enterobacterales species, including multidrug-resistant (MDR) phenotypes of E. coli, Escherichia fergusonii, Klebisiella ornithinolytica, Kluyvera cryocrescens and Providencia spp., is particularly concerning, underscoring the role of the lagoon as a hotspot for antimicrobial resistance (AMR). This environmental MDR burden is likely driven by phenotypic co-selection, as evidenced by the high bacterial tolerance to copper, which correlates with resistance to clinically important drugs such as tetracycline and cephalothin. Furthermore, the superior growth of presumptive Pseudomonas spp. in fish suggests adaptation to aquatic biota, linking environmental contamination to the food web. In summary, Laguna da Jansen is an ecosystem where persistent pollution acts as a selective environment for microbial adaptation, necessitating urgent mitigation of effluent sources and mandatory AMR surveillance to safeguard human and animal health.