Environmental background in Amazonian rivers near the industrial pole, northern Brazil

Abstract

The lack of a robust long-term environmental baseline has hindered risk management in the industrial complex region of Abaetetuba and Barcarena, Eastern Amazon. This study fills that gap by analyzing a 41-year time series (1980–2021; N = 19,292) and applying statistical tools such as the Median Absolute Deviation (MAD) to establish the geochemical background. Results for key physicochemical parameters indicate typical tropical conditions (temperature near 30 °C; slightly acidic pH: 6.8 ± 1.0) and low buffering capacity (alkalinity < 30 mg l⁻¹). Nutrient data revealed clear anthropogenic enrichment, with total phosphorus (TP) occurring in excess (0.2 ± 1.9 mg l⁻¹) and a low N/P ratio (2.5). Four metallic elements exceeded legal thresholds: aluminum (Al, 0.0009–1.31 mg l⁻¹), iron (Fe, 0.002–1.14 mg l⁻¹), cadmium (Cd, 0.0001–0.001 mg l⁻¹), and lead (Pb, 0.0001–0.019 mg l⁻¹). While Al and Fe concentrations are linked to regional geology and exhibited negative temporal trends, Cd and Pb require particular attention due to their association with rising anthropogenic pressures. Overall, the study successfully identified environmental thresholds, generating the first 41-year background reference for the region. This contribution is crucial for distinguishing natural variability from long-term contamination and significantly strengthens environmental monitoring and impact assessment in Amazonian aquatic ecosystems.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-44852-3.

Introduction

Establishing geochemical background (GB) values is a fundamental requirement for any environmental assessment or monitoring program. These values define the natural concentration range of chemical elements in water, sediment, or soil and are therefore essential for distinguishing natural geochemical variability from anthropogenic inputs¹. This distinction is particularly important in ecosystems of high ecological sensitivity, such as the Amazon, where efficient management strategies depend on well-defined environmental baselines capable of supporting contamination identification, regulatory decision-making, and long-term risk mitigation.

Over the past four decades, the Amazonian region of northern Brazil has undergone sustained population growth and rapid industrial expansion. The metropolitan mesoregion of Belém—particularly the municipalities of Barcarena and Abaetetuba—hosts a dense industrial hub that includes a major port, metallurgical and chemical facilities, fertilizer plants, refineries, food industries, and extensive supporting infrastructure. This industrial pressure, combined with the region’s limited sanitary sewage treatment, accelerated urbanization, and improper disposal of urban solid waste near the Murucupi River², poses significant environmental contamination risks. At the same time, large-scale environmental stressors, such as projected shifts in hydrological dynamics and sea-level rise, may modify sediment transport and estuarine processes in the region^3–6. Understanding how these natural and anthropogenic pressures reshape aquatic geochemistry remains a pressing scientific challenge.

Despite the strategic importance of this industrial corridor, there is still no consolidated long-term geochemical baseline capable of defining what constitutes “natural” water quality in the Barcarena–Abaetetuba system. Existing studies have focused on isolated parameters, specific contamination events, or short-term monitoring programs, leaving a critical knowledge gap regarding long-term trends and natural variability. The absence of a robust, multi-decadal dataset prevents reliable differentiation between background concentrations and sustained anthropogenic enrichment.

This study addresses this gap by establishing the first long-term environmental background for the Barcarena and Abaetetuba rivers through the analysis of a 41-year historical dataset (1980–2021). Using robust, non-parametric statistics—including frequency histograms, the median absolute deviation (MAD), and percentile-based environmental thresholds—we identify stable geochemical ranges and long-term trends for major physicochemical parameters, nutrients, and metals. We hypothesize that this extended time-series approach can reliably distinguish natural hydrogeochemical signatures from persistent anthropogenic inputs, offering an unprecedented baseline for environmental governance in for northern Amazonia, contributing to improved environmental management. This framework provides regulators and stakeholders with a scientifically defensible reference for monitoring, impact assessment, and the future management of cumulative industrial and urban pressures on Amazonian aquatic environments.

Data sources and methods

Study area

The study area encompasses the Belém Metropolitan Mesoregion, with a central focus on the municipalities of Barcarena and Abaetetuba (Fig. 1). Located in Pará State, northern Brazil, these municipalities are part of the Metropolitan Region of Belém, the state capital. They share similar demographic characteristics, including comparable land area, population size, and population density. Sanitation infrastructure is insufficient, with a low proportion of treated sewage^7,8 (Table 1). Additionally, urban solid waste is disposed of in a municipal dump near the Murucupi River, where soil heavy metal concentrations such as Cu, Hg, Pb, and Zn already exceed precautionary thresholds².

Fig. 1.

Study area and sampling sites for chemical parameters in Amazonian rivers. The circular shaded area indicates the location of the industrial hub. The map was generated using ArcGIS Desktop software, ArcMap 10.8 (Esri, Redlands, CA, USA; https://www.esri.com/en-us/arcgis/products/arcgis-desktop/overview). All shapefiles and geographic data were obtained from National Water Agency-ANA and Brazilian Institute of Geography and Statistics-IBGE with open access databases.

Table 1.

Environmental characteristics of study area in Pará state.

Sites	Municipality area1 (km2)	Sewage treatment2 (%)	Population3 (inhabitants)	Density3 (pop km2)	Fluvial discharge4 (m3 s−1)	Main type of soil5 (%)
Abaetetuba	1,610	16.5	158,188	98.21	11,000	Gleissols(17) Latosols (50) Alluvial (33)
Barcarena	1,310	27.8	126,650	96.65	6,500	Latosols (77) Gleissols (5) Alluvial (18)
Total	2,920	–	184,838	–	17,500	–

¹IBGE⁸; ²ANA⁷ (www.ana.gov.br); ³IBGE⁹ (www.ibge.gov.br/cidades). Density obtained for the basin population/basin area. ⁴Prestes et al.¹². ⁵Santos et al.¹³.

The local economies are driven primarily by tourism and a rapidly expanding industrial sector that hosts several large corporations. In 2021, Barcarena had 1,659 registered business units, while Abaetetuba had 1,172: representing a 56% increase since 2006 for Barcarena and a 27% increase for Abaetetuba. These figures underscore substantial industrial growth in the region over the past 15 years⁹. Industrial activities span a wide range of sectors, including fertilizer production, metallurgy, chemicals, textiles, and food manufacturing.

The industrial complex is surrounded by the Baía do Capim and Acará River basins (Fig. 1). The region has a humid equatorial climate (Köppen Af;¹⁰), with annual rainfall ranging from 2,385 to 4,500 mm. Mean River discharge for the Tocantins-Pará and Acará Rivers is approximately 11,000 m³ s⁻¹ and 6,500 m³ s⁻¹, respectively¹¹.

Geologically, the region is dominated by Cenozoic deposits composed mainly of quartz-rich sediments-such as gravels, sands, siltstones, and clays-derived from fluvial and alluvial processes. The predominant geological unit is the Barreiras Group, characterized by sandstones, siltstones, and lateritic claystones. Its chemical composition is variable but typically dominated by SiO₂ (~ 70%), with < 15% Fe and Al. The soils are mainly classified as red-yellow latosols and red-yellow gleisols (Table 1).

Cations (Ca²⁺, Mg²⁺, K⁺, Na⁺) and anions (Cl⁻, SO₄²⁻, HCO₃⁻) are key indicators of baseline limnological conditions. Their analysis is essential for the sustainable management of freshwater ecosystems and for monitoring potential impacts arising from large-scale development or climate change. This study also estimates total nitrogen, total phosphorus, and metal concentrations-parameters for which reference values are not yet established in this classification. These estimates consider inputs from soil runoff and atmospheric deposition and are based on precipitation patterns, watershed runoff, and soil characteristics (Supplementary Material, Section S1).

Literature review

Database

The literature review compiled all technical and scientific publications produced between 1980 and 2021 (Fig. 1), focusing primarily on the municipalities of Barcarena and Abaetetuba. The search strategy combined both targeted and broad approaches. The targeted component involved examining technical documents produced by private companies and public agencies, including Environmental Impact Studies (EIS) and Environmental Impact Reports (EIR), as well as academic grey literature (e.g., undergraduate theses, specialization monographs, master’s dissertations, and doctoral theses). For a broader scope, peer-reviewed scientific articles were retrieved from the Web of Science database (Thomson Reuters).

To establish a robust and comprehensive environmental baseline and identify major data gaps, the scope of this review was expanded to include not only scientific articles but also unpublished academic works and technical reports. All sources, categorized by document type and the presence or absence of analyzed physicochemical and heavy metal parameters, are summarized in Supplementary Material Table S1. This consolidation is essential for generating long-term time series and for contextualizing environmental risk assessments.

Data management

All secondary data-including information from scientific journals, grey literature, and technical reports—were organized using a structured data management workflow aligned with the study’s analytical objectives. These data formed the basis for building a regional database encompassing surface-water hydrogeochemical parameters throughout the area delineated in Fig. 1. The assembled database includes multiple hydrogeochemical variables grouped by parameter name, unit, sample size, and period of data availability, as shown in Table 2.

Table 2.

Hydrogeochemical parameters grouped by name, unit, sample size, and period of available data.

Parameter	Unit	Sample size (N)	Sample period
Physicochemical parameters
Water temperature	ºC	2,173	1997–2021
pH	units	1,919	1980–2021
Electrical conductivity (EC)	μS cm−1	1,393	1980–2021
Dissolved oxygen (DO)	mg l−1	2,008	1997–2021
Total alkalinity (TA)	mg l−1	847	2009–2021
Total Dissolved Solids (TDS)	mg l−1	1,696	2000–2021
Major ions
Chloride (Cl−)	mg l−1	793	2009–2021
Sodium (Na+)	mg l−1	674	1998–2021
Potassium (K+)	mg l−1	8	2011–2021
Sulfate (SO42−)	mg l−1	1,546	1998–2021
Calcium (Ca2+)	mg l−1	668	1998–2021
Bicarbonate (HCO3−)	mg l−1	847	2009–2021
Magnesium (Mg2+)	mg l−1	652	1998–2021
Nutrients
Total Phosphorus (TP)	mg l−1	570	2004–2021
Total Nitrogen (TN)	mg l−1	606	2004–2021
Metallic elements
Aluminum (Al)	mg l−1	1,468	1998–2021
Iron (Fe)	mg l−1	1,570	1980–2021
Manganese (Mn)	mg l−1	636	2004–2021
Copper (Cu)	mg l−1	448	2007–2021
Chromium (Cr)	mg l−1	176	2007–2021
Cadmium (Cd)	mg l−1	149	2007–2021
Lead (Pb)	mg l−1	242	2007–2021
Mercury (Hg)	mg l−1	39	2011–2021
Nickel (Ni)	mg l−1	235	2007–2021
Zinc (Zn)	mg l−1	346	2007–2021

Data quality and standardization

To ensure the reliability of the long-term analysis, a rigorous data quality and standardization procedure was applied to the heterogeneous secondary dataset (1980–2021). Only measurements obtained using standardized and widely accepted analytical techniques were included. All units were converted to the International System of Units (SI).

Statistical robustness was ensured using non-parametric and robust statistical measures, specifically the Median and the Median Absolute Deviation (MAD), which are minimally affected by the strong asymmetry and high-value outliers common in environmental datasets. The geochemical background (GB) was conservatively defined as GB = Median ± 2 × MAD^14,15.

Anomalous concentrations (outliers) were identified using Tukey’s Inner Fence (TIF) method, complemented by the 95th and 98th percentiles. This process transformed the historical dataset into a coherent and high-quality long-term time series.

Geochemical background

Geochemical background (GB) values must reflect the natural variability of the environment while remaining statistically robust against outliers arising from localized contamination or mineralized geological formations. We determined GB values using the equation GB = Median ± 2 × MAD^14,15.

The use of the Median and MAD is justified because these non-parametric measures are resistant to the strong right-skewness and extreme values frequently observed in environmental concentration data. This approach produces a conservative and reliable estimate of natural concentration ranges and is therefore superior to traditional parametric methods for defining background levels.

Environmental thresholds

Environmental thresholds are essential for identifying samples with unusually high concentrations that may warrant further investigation. Thresholds were established using Tukey’s Inner Fence (TIF) and the 95^th and 98^th percentiles.

• Tukey’s Inner Fence (TIF):

Defined as Q3 + 1.5 × IQR, this non-parametric method is widely used for detecting geochemical outliers due to its inherent resistance to the extreme values it is designed to flag.

• Percentiles (95^th and 98^th):

High-order percentiles¹ serve as practical environmental reference limits. The 98^th percentile is frequently used in regulatory frameworks and is statistically comparable to the parametric Mean + 2 SD threshold, providing a reliable basis for identifying potentially anomalous concentrations^1,15.

Time series and trend analysis

The compiled dataset was evaluated using methodologies established in previous studies and based on official analytical techniques and then analyzed through descriptive dispersion statistics. All measurement units were standardized to the International System of Units (SI).

Final datasets for selected parameters were characterized using distribution histograms and boxplots to determine the mean, median, standard deviation, and percentile values. Long-term trends (1980–2021) were assessed using the Mann–Kendall test (α = 0.05). This non-parametric test is particularly suitable for environmental time series because it does not assume normality or linearity and is robust to data noise and missing values-characteristics common in historical water-quality records. It is thus the most appropriate method for detecting monotonic (increasing or decreasing) trends over multi-decadal periods.

Additionally, the non-parametric Mann–Whitney test was applied to compare two independent datasets (e.g., seasonal differences), consistent with the non-normal distribution of environmental parameters. All statistical procedures, including descriptive analyses, were performed using PAST® 4.03 software¹⁶.

Results and discussion

Water budget

The study area experiences a humid equatorial climate (Köppen Af¹⁰), with high year-round rainfall (> 60 mm month⁻¹) and correspondingly elevated runoff. Quantifying the precipitation–runoff relationship is essential for interpreting climatic controls on solute chemistry.

Seasonal rainfall patterns show consistently high precipitation (> 100 mm month⁻¹ on average) during the wet season (January–June), whereas evapotranspiration peaks during the drier months of September–November (Fig. 2a).

Fig. 2.

Climatological rainfall normal for the 1981–2021 historical series (a); rainfall and evaporation time series in the study region (b); relationship between runoff and rainfall for the study period.

Long-term analysis (1981–2021) reveals a modest but statistically significant increases in monthly rainfall (Mann–Kendall test, p = 0.006, Sen’s slope =  + 0.004 mm month⁻¹ year⁻¹).

In tropical environments, intense precipitation typically generates substantial runoff. Here, runoff and precipitation are strongly correlated (R² = 0.72; Fig. 2b), as expected for humid tropical catchments. Runoff (R) can be approximated as R = P − ET, where P is precipitation and ET is evapotranspiration. Although both tropical and temperate watersheds receive high P and produce high R, temperate systems generally exhibit higher R/P ratios because of lower ET losses¹⁷. In Fig. 2b, the 1:1 line represents direct equivalence between P and R; most data points fall below this line, confirming substantial water loss through evapotranspiration characteristic of tropical settings.

Physic-chemical parameters in surface water (indicators of stress and variability—pH, DO, and Ions)

Temporal trends and dispersion statistics are summarized in Table 3.

Table 3.

Temporal trends and dispersion statistics of physic-chemical parameters. Values in bold indicate statistical significance for α = 0.05.

Parameters	Mean ± SD	Trend (Mann–Kendall)	Statistical significance
Temperature (ºC)	29.0 ± 0.2	Not significant	p = 0.1800
pH	6.8 ± 1.0	Positive	p = 0.0002
EC (μS cm−1)	86.0 ± 184.0	Not significant	p = 0.2300
DO (mg l−1)	6.4 ± 1.7	Negative	p = 0.0010
TDS (mg l−1)	41.0 ± 67.0	Negative	p = 0.0020

Significant values are in bold.

Analysis of core physicochemical parameters reveals both relative stability and emerging vulnerabilities driven by anthropogenic pressure.

Water temperature remained remarkably stable at 29.0 ± 0.2 °C, with no significant temporal trend. Mean pH was slightly acidic (6.8 ± 1.0; range 6.5–7.5), typical of Amazonian “white-water” rivers¹⁸, and displayed a statistically significant increasing trend (Fig. 3a–b,Tables 3, 4; Figs. S1–S3; Supplementary Material). All values fell within the Brazilian regulatory range of pH 6–9 for freshwater¹⁹,Fig. 3b).

Fig. 3.

Temporal series of the physic-chemical parameters in the study region. Water temperature (a), pH (b), electric conductivity-EC (c), dissolved oxygen-DO (d) and TDS (e). EC and TDS in logarithmic scale. Dashed black lines indicate environmental legislation limit. Dashed red line indicates trend slope.

Table 4.

Average monthly values for physic-chemical parameters in the study region.

Month	Water temperature (ºC) (1997–2021)	pH (units) (1980–2021)	Electric conductivity (μS cm−1) (1980–2016)	Dissolved oxygen (mg l−1) (1997–2021)	TDS (mg l−1) (2000–2021)
January	28.9	6.7	153	6.5	70.0
February	29.0	6.9	84	5.8	32.0
March	29.2	6.9	63	6.7	38.0
April	28.8	6.5	60	6.0	27.0
May	29.5	7.1	49	6.6	31.0
June	29.6	6.8	89	6.0	38.0
July	28.7	6.7	63	6.5	33.0
August	29.3	6.6	70	5.8	37.0
September	29.4	7.2	113	6.5	46.0
October	29.5	6.6	52	6.2	44.0
November	29.5	6.9	117	7.3	36.0
December	29.5	7.1	91	6.9	45.0

Electrical conductivity (EC; mean 86 ± 184 µS cm⁻¹) showed no significant trend, whereas total dissolved solids (TDS; mean 41 ± 67 mg l⁻¹, median 30 mg l⁻¹) exhibited a significant declining trend (p = 0.002; Fig. 3c, e). These low values confirm the dilute, high-discharge character of Amazonian white-water systems, with EC (40–760 µS cm⁻¹) markedly higher than in regional black-water (< 28 µS cm⁻¹) or clear-water (< 50 µS cm⁻¹) rivers¹⁸. Both parameters remained well below regulatory limits (TDS < 500 mg l⁻^{1 19}), typically ranging 30–70 mg l⁻¹ (Fig. 3e,Table 3, 4; Figs. S5–S6, S9–S10; Supplementary Material).

The most concerning signal is the strong declining trend in dissolved oxygen (DO; p = 0.001; Fig. 3d; Table 3). Although mean DO (6.4 ± 1.7 mg l⁻¹) still exceeds the legal minimum of 5.0 mg l⁻¹ for Class II freshwaters¹⁹, hypoxic episodes (< 2 mg l⁻¹) have become more frequent over the past decade. This pattern points to steadily increasing organic loading from untreated domestic and industrial wastewater, such as from slaughterhouse and meat processing plants tanneries, dairy industries, fish and seafodd processors, beverage producers, and sugar-ethanol mills. Monthly DO typically ranged between 5.8 and 7.3 mg l⁻¹ (Table 4; Figs. S7–S8; Supplementary Material), corresponding to ~ 65% saturation at the observed temperature (~ 29 °C) and negligible salinity (~ 0.1 psu), a level considered saturated under these conditions²⁰.

The contrasting stability of TDS and EC versus the progressive decline in DO indicates that current management priorities should focus on reducing organic and nutrient inputs rather than salinity or suspended-sediment loads. If the observed DO trend continues, regulatory thresholds for aquatic life protection will soon be breached, posing an immediate risk to ecosystem integrity.

Nutrients (nitrogen and phosphorus)

Long-term trends in nitrogen (TN) and phosphorus (TP) reveal a clear intensification of anthropogenic pressure in the study area. TN showed a significant positive trend between 2004 and 2021 (p = 0.001; slope =  + 0.0002), with an average concentration of 0.5 ± 0.5 mg l⁻¹ (Fig. 4a). TP displayed persistent enrichment well above the freshwater limit of 0.1 mg P l⁻^{1 19}, increasing significantly throughout the series (p = 0.0001,slope =  + 0.0001), with a mean of 0.2 ± 1.9 mg l⁻¹ (Fig. 4b).

Fig. 4.

Temporal series in the study region of TN (a) and TP (b). The red dashed line indicates slope of Mann–Kendall test (α = 0.05). The dashed green line indicates the maximum limit for P. The Y axes are on a logarithmic scale.

The extremely low N/P ratio (mean = 2.5) confirms strong P dominance in the system. This deviates sharply from the canonical Redfield ratio (16:1) and even from ratios typically observed in urban wastewater (~ 12:1 ²¹) or tropical rivers (~ 10:1 ²²). These stoichiometric patterns provide strong evidence of disproportionate phosphorus loading over time.

Nutrient source apportionment (Table 5) showed that anthropogenic inputs account for 85% of TN and 84% of TP, with domestic wastewater representing the single largest contributor, supplying 40% of total P. Soil runoff accounted for 14% of P and 11% of N, indicating that land-use pressures amplify the existing nutrient imbalance. These findings align with regional sanitation data showing that < 15% of sewage is treated⁷, and with persistent population growth⁸.

Table 5.

Estimates of N and P emissions (t yr⁻¹) from natural (green color) and anthropogenic sources (red color). The relative contribution (%) of each individual source is in parenthesis.

Sources	N (t yr−1)	P (t yr−1)
Soil runoff	267 (11)	73 (14)
Atmospheric deposition	103 (4)	7 (1)
Wastewater	1,040 (42)	208 (40)
Urban runoff	236 (10)	50 (10)
Solid waste	222 (9)	44 (8)
Industrial	155 (6)	34 (7)
Livestock	112 (5)	50 (10)
Agriculture	346 (14)	49 (9)
Total	2,481	515
Total natural	370 (15)	80 (16)
Total anthropogenic	2,109 (85)	435 (84)

The long-term increase in wastewater-derived nutrient emissions was statistically significant (Mann–Kendall, p < 0.05; Fig. 5). Combined, these trends demonstrate a structural vulnerability: rapid urban expansion and insufficient sanitation capacity are overwhelming the natural assimilative potential of the aquatic system. Accordingly, phosphorus remains the primary limiting factor for water-quality improvement, and regional management should prioritize expanded wastewater treatment and the control of diffuse runoff sources.

Fig. 5.

Time series (1980–2021) of population growth in the cities of Abaetetuba and Barcarena (squares and black line); N load (t yr⁻¹) via urban sewage (circle and red line); P load (t yr⁻¹) via urban sewage (circle and green line), N load (t yr⁻¹) via natural sources (circle and blue line) and P load (t yr⁻¹) via natural sources (circle and orange line) in the study region.

Total alkalinity, anions and cations

The inorganic chemistry of the study area, characterized by total alkalinity (TA) and major ions (Ca²⁺, Mg²⁺, Na⁺, K⁺, Cl⁻, HCO₃⁻, SO₄²⁻), exhibits overall stability but reveals a critical vulnerability due to low buffering capacity (Figs. 6, 7, 8).

Fig. 6.

Temporal series in the study region of total alkalinity. The red dashed line indicates slope of Mann–Kendall test (α = 0.05). The Y axes are on a logarithmic scale.

Fig. 7.

Major anions: chloride-Cl⁻ (a), bicarbonate-HCO₃⁻ (b), and sulphate-SO₄²⁻ (c), in surface waters from the study region. The dashed line indicates the slope of the time series. The Y axes are on a logarithmic scale.

Fig. 8.

Major cations: calcium (a), magnesium (b), sodium (c), and potassium (d), in the historical series in the study region. The dashed line indicates the slope of the time series. Mg and Na in the Y axes are on a logarithmic scale.

The most important management implication is the persistently low TA. Mean TA was 27.7 ± 3.0 mg l⁻¹ (median 18.0 mg l⁻¹), with many values < 30 mg l⁻¹, indicating limited acid-neutralizing capacity (Fig. 6; Figs. S15–S16; Supplementary Material). No significant long-term trend was detected (Mann–Kendall, p = 0.17, slope =  + 0.0003).

Bicarbonate accounted for > 90% of TA, confirming heavy reliance on the carbonate system for pH buffering (Fig. 7b; Figs. S19–S20; Supplementary Material). Estimated water hardness was 75.8 mg CaCO₃ l⁻¹, consistent with moderate buffering capacity. This inherent fragility renders the system highly sensitive to acidification from atmospheric deposition or uncontrolled industrial discharges, which could trigger pH drops, heavy-metal mobilization, and ecological damage. Strict regulation of acid-forming emissions remains essential. It should be noted, however, that part of this variability may also be associated with natural conditions. The study area encompasses different aquatic systems, including igarapés characterized by naturally acidic surface waters, as well as estuarine regions influenced by the intrusion of oceanic waters, which can significantly affect TA. As shown in Fig. 1, the collected data cover the entire mouth of the Tocantins/Pará River basin up to the ocean, reflecting a broad environmental gradient that also contributes to the variations observed in the physicochemical parameters. 

Although major-ion concentrations were generally low and typical of dilute freshwaters, several ions displayed significant temporal trends linked to population growth and industrial activity (Figs. 7, 8).

HCO₃⁻ showed a significant increasing trend (p = 0.04, slope =  + 0.0006), whereas Cl⁻ and SO₄²⁻ exhibited significant decreasing trends (p < 0.05). Chloride, despite a slight negative trend (Fig. 7a), remains a sensitive tracer of urban and domestic pressure. Mean Cl⁻ concentration (7.3 ± 13.0 mg l⁻¹) was far below the regulatory limit of 250 mg l⁻^{1 19} and typical background levels in pristine rivers (< 100 mg l⁻¹) (Figs. S17–S18; Supplementary Material). Nevertheless, continued urban expansion necessitates ongoing Cl⁻ monitoring to detect emerging diffuse pollution.

Both HCO₃⁻ and SO₄²⁻ remained well within legal and natural Amazonian ranges. The significant declining trend in SO₄²⁻ (p < 0.05) and low concentrations (Fig. 7c; Figs. S21–S22; Supplementary Material) suggest that industrial sulfate inputs, if present, are effectively controlled or rapidly diluted by high river discharge.

Major cations reflected underlying chemical stability alongside specific vulnerabilities (Fig. 8). Ca²⁺ and Mg²⁺ concentrations were within typical freshwater ranges but lower than many Amazonian “white-water” rivers (Fig. 8a–b; Figs. S23–S26; Supplementary Material). The resulting hardness of 75.8 mg CaCO₃ l⁻¹ classifies the water as moderately hard²³ and reinforces the limited buffering capacity noted earlier. Ca²⁺ displayed a significant decreasing trend (p = 0.0004, slope = − 0.00008), offset by an opposing increasing trend in Mg²⁺ (p = 0.00009, slope =  + 0.000008).

Sodium emerged as a robust indicator of anthropogenic influence, exhibiting a significant positive trend (p = 0.007, slope =  + 0.001) (Fig. 8c). Mean Na⁺ (8.2 ± 8.9 mg l⁻¹) falls within the range reported for Amazonian white waters and aligns with values of 0.7–8.0 mg l⁻¹ documented by Souto et al.²⁴ (Figs. S27–S28; Supplementary Material). The upward trajectory likely reflects growing urban runoff and untreated sewage associated with regional population growth, making Na⁺ a valuable long-term tracer of human impact.

Potassium also showed an increasing tendency (slope =  + 0.0006), although limited sampling (n = 8) prevented formal statistical testing. Mean K⁺ concentration (1.2 ± 0.8 mg l⁻¹) remained within typical freshwater ranges (Fig. 8d; Figs. S29–S30; Supplementary Material).

Metals in surface waters–trends

The dataset includes dissolved Fe and Al and total Mn, Cu, Zn, Ni, Pb, Cr, Cd, and Hg, enabling evaluation of both natural geochemical signatures and anthropogenic pressures (Figs. 9, 10, 11; Table 6; Figs. S31–S40; supplementary material). While statistical thresholds help identify anomalous concentrations, interpretation of origin is guided by temporal behavior and known geochemistry^1,15.

Fig. 9.

Time series of metallic elements in the surface waters of the study region. Dissolved aluminum-Al (a), dissolved iron-Fe (b), and total manganese-Mn (c). The blue dashed line indicates the slope of the series. The maximum limit allowed by environmental legislation¹⁹ is indicated by the green dashed line or the green text on the graph.

Fig. 10.

Time series of metallic elements in the surface waters of the study region. Dissolved copper-Cu (a), total zinc-Zn (b), total nickel-Ni (c). The blue dashed line indicates the slope of the series. The maximum limit allowed by environmental legislation¹⁹ is indicated by the green dashed line or the green text on the graph.

Fig. 11.

Time series of metallic elements in the surface waters of the study region. Lead-Pb (a), chromium-Cr (b), cadmium-Cd (c), and mercury-Hg (d). The blue dashed line indicates the slope of the series. The maximum limit allowed by environmental legislation¹⁹ is indicated by the green dashed line or the green text on the graph.

Table 6.

Concentrations of metals elements in surface waters of the study region.

Elements [mg l−1]	Na	Min–Max	Mean ± SD	CV [%]	GB	Trend	Stat. sign. [p < 0.05]
Fe	1,451	0.0020–1.1400	0.37 ± 0.240	67	0.30 ± 0.500	Negative	0.02
Al	1,300	0.0009–1.3100	0.29 ± 0.300	106	0.20 ± 0.200	Negative	0.0001
Mn	573	0.0005–0.0320	0.01 ± 0.007	109	0.01 ± 0.007	Positive	0.0009
Cu	423	0.00056–0.0083	0.002 ± 0.001	66	0.002 ± 0.002	Positive	0.07
Zn	303	0.0007–0.0600	0.014 ± 0.013	93	0.01 ± 0.010	Negative	0.003
Ni	224	0.0001–0.0130	0.004 ± 0.003	83	0.003 ± 0.004	Negative	0.06
Pb	201	0.0001–0.0190	0.003 ± 0.003	114	0.001 ± 0.002	Negative	0.0001
Cr	147	0.00052–0.0030	0.001 ± 0.000	40	0.001 ± 0.000	Positive	0.0019
Cd	117	0.0001–0.0010	0.0009 ± 0.000	32	0.001 ± 0.000	Negative	0.95
Hg	31	0.000185–0.0002	0.0002 ± 0.000	7	0.0002 ± 0.000	Positive	0.65

N: total; Min–Max: minimum and maximum values; SD: standard deviation; CV: coefficient of variation; GB: geochemical background. Trend obtained using the Mann–Kendall test for α = 0.05. Stat. sign. indicates statistical significance for p < 0.05.

^a: after outlier elimination.

Significant values are in bold.

Geogenic metals

Fe and Al were the most abundant elements, consistent with the mineralogy of the Barreiras Formation^14,25. Both exhibited significant negative trends (p < 0.05), suggesting reduced mobilization, likely linked to declining soil erosion. Although average Fe slightly exceeded the legal limit (0.3 mg l⁻^{1 19}), the decreasing trend and narrow range (0.0–0.5 mg l⁻¹) suggest gradual improvement. Mn remained well below its threshold but showed a significant positive trend; Cr presented a low, non-significant increase. (Fig. 9c; Fig. 11b).

Micronutrient metals with diffuse inputs

Cu, Zn, and Ni occurred at low concentrations typical of tropical rivers²⁴ and remained far below regulatory limits¹⁹. Zn showed a significant negative trend, while Cu and Ni exhibited non-significant temporal changes (Fig. 10a–c, Table 6). Their behavior is consistent with strong organic-matter complexation and dilution by high discharge²⁶.

Anthropogenic metals of concern

Pb, Cd, and Hg showed patterns indicative of episodic or year-round anthropogenic inputs (Fig. 11). Pb exhibited a significant negative trend but had high variability (CV = 117%), suggesting possible past high-magnitude domestic contamination events,as observed by², who demonstrated the influence of the open-air landfill on the contribution of Soil Quality Indices (SQIs), such as Cu, Hg, Pb, and Zn, to the soil in the Barcarena area.

Cd (0.0009 ± 0.0002 mg l⁻¹) remained near its limit (0.001 mg l⁻¹), and Hg (0.0002 ± 0.00001 mg l⁻¹) approached the threshold for Class 2 waters¹⁹. Both elements are closely associated with industrial and mining activities in tropical basins^20,26. The main anthropogenic sources of Hg include industrial processes such as artisanal gold mining, cement production, coal combustion, waste incineration, and other activities that release the metal either intentionally or as an operational by-product. Cd is generated primarily as a by-product of refining Zn sulphide ores and is also released through metallurgical activities and various industrial processes. Although their long-term trends were not statistically significant, their proximity to regulatory limits highlights a need for intensified, high-frequency monitoring.

Taken together, most metals remained within acceptable ranges; however, Cd, Hg, and Pb represent the primary concerns, reflecting the cumulative influence of multiple anthropogenic activities occurring within the watershed, such as industrial operations and the severe lack of sanitation. This has been documented by² and others, who investiged the isotopic signature of Pb in the so-called ‘red mud’, in bottom sediments of the Murucupi River, and in domestic sewage. Their findings demonstrated that the Pb identified in the Murucuri River does not originate from bauxite residue but rather from domestic wastewater, corroborating the results reportes by².  These findings emphasize the importance of continuous monitoring and targeted regulatory oversight.

Geochemical background: concentrations of potentially toxic elements

Based on the collected surface water data, we determined the variation ranges (i.e., environmental background values) for each parameter. This range, encompassing minimum, maximum, mean, standard deviation, median, and MAD values, establishes the geochemical background for the Barcarena and Abaetetuba region (Table 6).

Descriptive statistics and estimated Geochemical Background (GB) revealed high heterogeneity for Al, Mn, and Pb. Fe, Al, and Mn were the most abundant elements in the study’s surface waters (Table 6). This finding aligns with the region’s dense Cenozoic sedimentary complex, primarily formed by exposed Tertiary sediments (Barreiras Formation) in the study municipalities²⁵. Fe, Al, and Mn oxides and hydroxides are commonly found in the Barreiras Group’s sediments, typical of tropical climates.

Further supporting this, a study on Pará state soils identified Al, Fe, and Mn as the most abundant elements, with wide concentration ranges¹⁴. Their prevalence in regional freshwaters is partly explained by their abundance in the Barreiras Formation rocks and soils. High concentrations of Al, Mn, and Barium (Ba) in the region’s water bodies are also associated with intense leaching during the rainy season²⁵.

The elements with the highest coefficients of variation in this study (Al, Mn, Pb; Table 6) are consistent with other research. For instance, average Al concentrations in the Pará River (0.1 to 0.3 mg l⁻¹) reported by Santos et al.²⁵ are very similar to our estimated background values (Table 6).

Seasonal distribution of metallic elements

The analysis of the seasonal distribution of metallic elements against the historical rainfall series (1981–2021) is critical for deciphering element mobility and identifying sources, thereby informing targeted management strategies.

Hydrology-driven mobilization (geogenic background)

The mobility of metals predominantly derived from geological sources is strongly linked to seasonal river discharge.

Fe concentrations showed a slight association with river discharge early in the wet season (Fig. 12a), while Al concentrations showed a clearer seasonal association (Fig. 12b). These elements, which are highly abundant in the Barreiras Formation, are mobilized through soil runoff and leaching during periods of high flow. Mn concentrations peaked during the dry season (September, October, and November; Fig. 12c), consistent with release through natural weathering of regional minerals and rocks or resuspension from sediments during low flow²⁰. Despite these processes, observed Mn concentrations remained below the legal limit.

Fig. 12.

Climatological series of metallic elements in the surface waters of the study region. The blue line indicates the monthly concentration over the year at Fe (a), Al (b) and Mn (c). The red (Tocantins River) and green lines (Guamá River) indicate the fluvial discharge. The black line indicates monthly rainfall in the study region. Element concentrations are on a logarithmic scale (Blue color).

Cu concentrations, often released by rock weathering, also peaked in the dry season/wet-dry transition (Fig. S41a; Supplementary Material), similar to Mn^20,27. Pb showed a slight association with discharge, peaking in November (Fig. 13a). The mobilization of Pb is favored by high acidity soils and rapid organic matter decomposition in the region, suggesting that controlling soil erosion remains a crucial environmental management implication even for naturally dominant elements¹⁴.

Fig. 13.

Climatological series of metallic elements in the surface waters of the study region. The blue line indicates the monthly concentration over the year at Pb (a), Cr (b), Cd (c) and Hg (d). The red (Tocantins River) and green lines (Guamá River) indicate the fluvial discharge. The black line indicates monthly rainfall in the study region. Element concentrations are on a logarithmic scale (Blue color).

Metals indicating constant anthropogenic inputs

The elements Cd, Cr, and Hg exhibited behaviors suggesting inputs decoupled from natural dilution processes, strongly pointing toward continuous anthropogenic effluents irrespective of rainfall (Fig. 13).

Both Cd (Fig. 13c) and Hg (Fig. 13d) showed no clear association with river discharge in the seasonal analysis. In fact, Cd showed an inverse relationship with rainfall and discharge, indicating that its source (e.g., industrial, chemical, and mining waste) is constant throughout the year and its concentration is only diluted during periods of high flow^20,26. This sustained input of high-risk metals, especially Hg nearing the legal threshold, serves as a critical warning for the industrial hub and necessitates urgent, high-frequency effluent monitoring.

Cr similarly showed no association with river discharge (Fig. 13b). Given that elevated Cr levels are frequently linked to industrial activities (e.g., electroplating), its decoupling from natural flow further suggests an anthropogenic contribution that, while currently low, must be factored into industrial discharge regulations²⁸.

Metals with stable behavior

Zinc (Zn) (Fig. S41b; Supplementary Material) and Nickel (Ni) (Fig. S41c; Supplementary Material) showed no significant seasonal association with river discharge, with concentrations remaining below permissible limits. This stability is likely due to their strong affinity for adsorption onto organic substances, which minimizes their leaching potential, or is a result of effective dilution by the high river flows^26,29. Their presence in Pará state soils may be strongly related to shared geological events, such as the formation of lateritic zones common beneath the deep Ferralsols or Acrisols across western and southwestern Pará³⁰. This indicates that, for Zn and Ni, management efforts can focus primarily on preventing future major contamination events rather than mitigating constant seasonal flux. Soils near the Barcarena open-air dump already exhibit heavy metal concentrations, such as Cu, Hg, Pb, and Zn, that exceed precautionary threshold values [2]

Environmental thresholds

Environmental thresholds were established using the Tukey Inner Fence (TIF) and the 95^th and 98^th percentiles, which are widely adopted for identifying anomalous concentrations in heterogeneous environmental datasets^1,14. These non-parametric approaches require no assumptions regarding data distribution and are therefore well suited for Amazonian waters, where geochemical variability and skewed concentration profiles are common³¹.

The thresholds derived here do not define pollution per se; instead, they indicate concentrations that are statistically unusual relative to the natural variability of the basin. This concept is especially relevant in Amazonian rivers, where elements such as Fe, Al, and Mn naturally show broad ranges due to weathering of the Barreiras Formation and lateritic soils^14,25.

Using TIF as an upper boundary, Fe and Al emerged as the dominant elements, reflecting their geological abundance. Similarly, the 95^th percentile identified Al, Fe, and Zn as the main contributors to the upper concentration range, whereas the 98^th percentile included Al, Fe, Zn, and Mn (Table 7). These results establish a statistically robust set of environmental thresholds that highlight the parameters most likely to deviate from natural baselines, thereby guiding regulatory monitoring and the early detection of unusual geochemical signatures.

Table 7.

Environmental thresholds for potentially toxic elements (PTEs) and comparison with regulatory limits¹⁹.

Elements	TIF	95th	98th	Legislation19
Fe	0.900	0.900	1.000	0.300
Al	0.900	1.000	1.200	0.100
Mn	0.020	0.025	0.030	0.100
Cu	0.009	0.007	0.007	0.009
Zn	0.040	0.050	0.060	0.180
Ni	0.011	0.010	0.012	0.025
Pb	0.011	0.011	0.015	0.010
Cr	0.001	0.002	0.003	0.050
Cd	0.001	0.001	0.001	0.001
Hg	0.0002	0.0002	0.0002	0.0002

Final considerations

The environmental background values established for the 41-year time series reveal that several key parameters-particularly total phosphorus (TP), aluminum (Al), and iron (Fe) exceeded current legislative limits¹⁹. While Al and Fe exhibited negative temporal trends and may gradually approach permissible levels, TP showed persistent exceedance driven by multiple sources. As shown in Table 5, soil runoff contributed 14% of TP loads, while domestic sewage accounted for 40% and urban runoff 10%; these proportions are consistent with the region’s low sewage treatment capacity (< 15%)⁷ and rapid demographic expansion⁸.

The long-term geochemical background values (Table 6; Table 7) reveal substantial shifts in Al, Fe, and P concentrations over the 1980–2021 period, reflecting both natural hydrological variability and escalating anthropogenic pressure. The presence of potentially toxic elements, particulary Hg, Cd, and Pb, demands caution, as their concentrations reflect not only industrial inputs but also the persistent lack of adequate sanitation infrastructure in the region. These elements remain close to, or in some cases approach, regulatory thresholds, as documented by ^2,20,26.

Finally, the study identifies a structural challenge in historical data availability. Sparse and discontinuous records for Hg, Cd, and Pb limit the reliability of long-term assessments, underscoring the need for improved environmental surveillance. Strengthening regional monitoring networks, especially in high-risk zones near the industrial complex, will be essential for detecting emerging contamination patterns and supporting evidence-based management.

Conclusions

This study establishes the first statistically robust, long-term environmental baseline for the Barcarena–Abaetetuba region using a 41-year dataset (1980–2021). The findings confirm the expected dilute, slightly acidic character of Amazonian white-water rivers but identify two critical vulnerabilities: persistently low total alkalinity, conferring limited acid-buffering capacity, and chronic phosphorus enrichment driven by untreated domestic sewage, agricultural runoff, and soil erosion. Trace-metal concentrations were generally low and dominated by natural sources (Al, Fe), yet Cd, Pb, and Hg frequently exceeded regional natural background levels and exhibited spatial–temporal patterns indicative of multiple anthropogenic pressures consistent with observations by ^2,20,26.

The geochemical baselines and environmental thresholds derived here provide an objective framework for distinguishing natural variability from anthropogenic perturbation in these fluvial systems. They also supply regulatory agencies with regionally appropriate reference values, particularly for Fe and Al, in this heavily industrialized lower Amazon river basin.

Despite these advances, important limitations remain. Measurements of Cd, Pb, and Hg were sparse throughout the 41-year record, limiting the power of trend analyses. Additionally, the dataset is restricted to surface water-column samples and therefore cannot address metal accumulation or remobilization from bed sediments. To strengthen future monitoring, we recommend implementing high-frequency sampling of priority toxic metals and conducting targeted sediment-core studies near the industrial hub to quantify long-term storage, flux, and ecological risk.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

M.R.: Conceptualization, Investigation, Methodology, writing—review & editing; C.N.: Conceptualization, Investigation, Methodology, and Writing—review & editing; S.M.: Conceptualization, Investigation, Methodology and Writing; M.C.: Writing—review & editing.; R.A.: Writing—review & editing.; A.M.: Writing—review & editing.; G.P.: Investigation, Writing—review & editing. ; I.B.: Investigation, Writing—review & editing.; N.M.: Investigation, Writing—review & editing.

Data availability

The datasets generated and analyzed in this study are available from the corresponding author, Marcelo Rollnic, upon reasonable request.

Declarations

Competing interests

The authors declare no competing interests.