Critical review of toxic trace elements and human health risk in marine fishes of the Southwestern Atlantic Ocean

1 Introduction

Potentially toxic trace elements (PTEs) enter aquatic ecosystems from both natural sources (weathering of soil and rocks, erosion, forest fires, and volcanic activity) and anthropogenic sources such as domestic and industrial effluents and agricultural, harbor, and industrial activities. Consequently, these elements are widely distributed in marine, estuarine and freshwater environments. PTEs are divided into non-essential and essential trace elements (Wood et al., 2012a,b). Non-essential trace elements such as arsenic (As), cadmium (Cd), mercury (Hg), lead (Pb), rubidium (Rb), strontium (Sr), and vanadium (V) can be very toxic at low concentrations, whereas essential chromium (Cr), copper (Cu), manganese (Mn), molybdenum (Mo), nickel (Ni), selenium (Se) and zinc (Zn) play an important metabolic role at low concentrations in aquatic organisms, but they can produce toxic effects at high levels. Although essential elements play important metabolic roles at low concentrations, both essential and non-essential elements may produce adverse effects when organisms are exposed to elevated concentrations. In fish, PTE exposure has been associated with hematological alterations, cellular toxicity, and effects on cardiovascular, endocrine, reproductive, immune, and nervous systems (Parrino et al., 2018; Fazio et al., 2019; Shahjahan et al., 2022; Taslima et al., 2022). Of particular relevance to human health, PTEs may accumulate in edible fish tissues and therefore constitute an important pathway of dietary exposure through fish consumption.

The Southwestern Atlantic Ocean (SWAO) is a highly productive marine region shared by Argentina, Uruguay, and Brazil, where several demersal and benthopelagic fish species support important coastal fisheries. In recent decades, numerous studies have documented increasing anthropogenic pressure on these coastal zones, particularly near major urban and industrial areas, along the Brazilian and Argentine coasts (Dias et al., 2013; Newton et al., 2020; Strauch and Erzinger, 2021; López and Marcomini, 2011). Elevated concentrations of PTEs in seawater and sediments have been repeatedly reported, especially in regions influenced by large estuaries and ports, with densely populated areas and industrial activities that receive inputs of untreated domestic and industrial sewage (Brazil: Mirlean et al., 2009; Fonseca et al., 2013; Kim et al., 2016; Kütter et al., 2021; Argentina: Villagran et al., 2021; Uruguay: Viana et al., 2005). Among the most polluted sites are: Todos os Santos Bay (de Santana et al., 2017a,b; Silva et al., 2016a,b; Rocha et al., 2014), Santos Bay (Kim et al., 2016), Guanabara Bay (FEEEMA, 1990; Valentin et al., 1999; Jablonski et al., 2006; Seixas et al., 2013; Ferreira et al., 2015), Sepetiba Bay (Lacerda et al., 1987; Souza and Ferreira, 1990; IFIAS, 1998; Kehrig et al., 1998; Bisi et al., 2012; Kütter et al., 2021), Patos-Mirim Lagoon system (Mirlean et al., 2019; da Silva Marques et al., 2019), Rio de la Plata estuary, Samborombón Bay (Marcovecchio and Moreno, 1993; Marcovecchio, 2004; Tatone et al., 2015; Biolé et al., 2024) and Bahia Blanca estuary (Marcovecchio et al., 2016; La Colla et al., 2017, 2018; Villagran et al., 2021). Although there are many commercially important transboundary fish species in this area, mainly Mugil liza, Micropogonias furnieri, Urophycis brasiliensis, Odontesthes argentinensis, and Cynoscion guatucupa, PTE levels in their tissues have been determined for only a few of them, and the risk associated with their consumption has been evaluated for an even smaller number of species.

The maximum limits of some PTEs (i.e., As, Cd, Cu, Hg, Ni, Pb, Se, or Zn) in fish muscle are established by local and regional guidelines [ANVISA: Brazilian National Health Surveillance Agency (Normative Instruction n° 160/2022; 88/2021; Resolution 42/2013 and Decree n° 55.871/1965); Agência Nacional de Vigilância Sanitária, 2013; Agência Nacional de Vigilância Sanitária (ANVISA), 1965, 2021, 2022b Argentinean Food Codex (AFC) (2023, Chapter III); SENASA: National Food Safety and Quality Service (Res. 533/1994); (Servicio Nacional de Sanidad y Calidad Agroalimentaria (SENASA), 1994) MERCOSUR: Southern Common Market (Res. GMC 12/2011); EU: European Union (Commission Regulation European Union (EU) 2023/915); European Commission, 2023)]. ANVISA, AFC, SENASA, and MERCOSUR were considered because they constitute the national or regional regulatory frameworks directly applicable to fish marketed or consumed in the SWAO countries included in this review. In addition to national and regional regulations directly applicable to SWAO countries, some studies used external regulatory frameworks, particularly European Union legislation, or international standards such as Codex Alimentarius, as complementary reference points.

However, comparison with maximum regulatory levels alone does not provide a complete characterization of potential human health risk because it does not account for the amount of fish consumed, exposure frequency, duration of exposure, or the toxicological characteristics of individual elements. Importantly, regulatory maximum levels, toxicological guidance values, and HRA thresholds represent different concepts and should not be used interchangeably. Maximum levels are concentration-based regulatory limits established for contaminants in foods and are primarily used for compliance and food-control purposes.

Dietary exposure can be further assessed through estimated daily or weekly intake (EDI/EWI), which can be compared with toxicological guidance values (PTDI/PTWI) used for food contaminants such as heavy metals with cumulative properties. These guidance values are health-based exposure benchmarks rather than maximum concentrations permitted in fish tissue. They are established by international organizations such as the Joint FAO/WHO Expert Committee on Food Additives (JECFA) and the European Food Safety Authority [FAO/WHO, 1982, 1983, 2011a,b,c, 2022; Alexander et al., 2009; EFSA CONTAM Panel (EFSA Panel on Contaminants in the Food Chain), 2010, 2012, 2014; EFSA, 2018; EFSA CONTAM Panel (EFSA Panel on Contaminants in the Food Chain), 2020; European Food Safety Authority (EFSA) et al., 2021; Halldorsson et al., 2023] and are used to evaluate whether estimated dietary exposure approaches or exceeds a toxicologically derived level.

When sufficient information on fish consumption and exposure frequency is available, potential non-carcinogenic risk can also be characterized using the Target Hazard Quotient (THQ), which relates the estimated exposure dose to the corresponding reference dose, and the Hazard Index (HI), calculated as the sum of individual THQs (Yi et al., 2011; Tao et al., 2012; USEPA, 2026). THQ or HI values above 1 indicate a level of exposure at which potential non-carcinogenic effects cannot be excluded, whereas values below 1 generally indicate that such effects are unlikely under the evaluated exposure scenario (Chien et al., 2002). For elements with appropriate carcinogenic toxicity values, lifetime exposure can additionally be evaluated using carcinogenic risk (CR). A total CR value is considered acceptable when it is below 10⁻⁶, corresponding to a lifetime cancer risk of less than 1 in 1,000,000. CR values between 10⁻⁶ and 10⁻⁴ indicate a potential need for further evaluation, while CR values >10⁻⁴ are considered unacceptable, representing a cancer risk greater than 1 in 10,000 [U.S. Environmental Protection Agency (USEPA), 2010]. Therefore, THQ, HI, and CR are exposure-based risk-characterization metrics and not regulatory maximum levels. These three levels of evaluation—regulatory compliance, comparison with health-based guidance values, and quantitative HRA—answer related but distinct questions and were treated separately throughout this review.

Comparisons among risk assessments conducted in the SWAO remain difficult because studies have applied different analytical approaches, exposure assumptions, fish-consumption rates, and toxicological reference values. This issue is particularly relevant for arsenic, for which risk estimates may differ markedly depending on whether inorganic arsenic (iAs) is directly measured or inferred as a fixed proportion of total arsenic. Consequently, differences in reported risk among species, locations, and countries may reflect not only actual differences in contaminant concentrations but also methodological differences among studies.

Against this background, this study provides a critical review and re-analysis of PTE occurrence and associated human health risks from the consumption of marine teleost fishes in the Southwestern Atlantic Ocean. Specifically, we integrate available information from Brazil, Uruguay, and Argentina to (i) characterize the geographic, taxonomic, and analytical coverage of PTE studies; (ii) compare the approaches and assumptions used to estimate dietary exposure and human health risk assessment (HRA); (iii) re-evaluate selected risk estimates using exposure parameters and updated toxicological reference values for iAs; and (iv) identify major geographic, methodological, and toxicological knowledge gaps. By distinguishing variability associated with contaminant occurrence from that introduced by methodological assumptions, this review aims to provide a more consistent basis for regional seafood-safety assessment, environmental monitoring, and evidence-based fisheries and public-health policies.

2 Methods

2.1 Literature review

A literature search for peer-reviewed journal articles and published reports documenting As, Cd, Cr, Cu, Hg, Mn, Mo, Ni, Pb, Rb, Se, Sr, V, or Zn concentrations in muscle of marine and estuarine-associated teleost fishes collected in the Southwestern Atlantic Ocean (Brazil, Uruguay, and Argentina) was conducted using Google Scholar (https://scholar.google.es) in May 2026. Keywords and concepts used in the bibliographic search were “marine fish,” “coastal fish,” “muscle,” “Uruguay,” “Brazil,” “Argentina,” “Southwestern Atlantic Ocean,” and “the Spanish or Portuguese name of each PTE evaluated.” Connectors used were “AND” and “OR.” Published documents, articles and reports (hereafter collectively referred to as studies) published from 1996 to 2026 were selected. Studies were initially screened based on titles and abstracts, followed by full-text review. Studies that did not provide a traceable source for the fish samples were excluded. Geographically, all reports with fish samples from the coast of the three countries were included (from 0°S to 55°S). In addition, a search was conducted in FishBase (Froese and Pauly, 2026) to determine the valid name of the fish species studied in each study. Sharks and rays were not included in this review. Data were classified by country and coastal location, sampling latitudinal range, fish species and genera, sampling method (source of fish samples), PTEs studied, analytical instrument used, guidelines cited and HRA method used (Supplementary Table S1). Brazilian studies were separated into three sectors: northern (A: 0–17°S), central (B: 17–26°S), and southern (C: 26–33.5°S). These sectors were defined as operational latitudinal units to facilitate geographic comparison of the reviewed literature and should not be interpreted as strict biogeographic or marine-ecoregion boundaries.

Given the critical nature of this review, methodological differences among studies that could affect the comparability of exposure and risk estimates were specifically examined. To this end, the fish consumption rates, exposure frequencies used, population groups (general population and fishing community, referred to as fishermen) or consumption scenarios considered, the toxicological reference values applied, and the indicators used for risk characterization were recorded. Regulatory and toxicological references were classified according to their function. National and regional food regulations directly applicable to the countries included in the review (ANVISA for Brazil, AFC and SENASA for Argentina, and MERCOSUR provisions relevant to the three countries) were distinguished from external regional regulatory frameworks such as the European Union, and international standards or guideline, such as Codex Alimentarius, which were used by the original studies for comparative purposes. Health-based toxicological guidance values issued by JECFA or EFSA and dose-response parameters used in quantitative HRA were recorded separately from maximum contaminant concentrations. This distinction was maintained because regulatory maximum levels, toxicological reference values, and HRA thresholds are not methodologically equivalent. When a reviewed study compared its results with more than one regulatory or guidance framework, all cited frameworks were recorded, but comparisons were interpreted according to their original purpose rather than treated as interchangeable thresholds.

In addition, given the particular relevance of arsenic to the risk estimates reported in the reviewed literature, arsenic assessments were additionally classified according to whether they were based on total As, measured inorganic As (iAs), measured other As species considered potentially toxic [methylarsonic acid (MA) and dimethylarsinic acid (DMA), which are analyzed separately because they should not be assumed to have toxicological potency equivalent to iAs], or iAs estimated as a fixed proportion of total As (proportions of total As assumed to represent iAs were recorded and compared among studies). This distinction was considered toxicologically relevant because arsenic species differ in their toxicity, bioavailability, and metabolism (Zhang et al., 2023; Mukherjee et al., 2024), as discussed in Section 4.4. This classification approach was used to identify differences in risk classification potentially associated with methodological assumptions rather than with differences in measured As concentrations.

2.2 Health risk re-evaluation

To update the toxicological basis of the risk estimates, studies reporting analytically measured arsenic species and sufficient information on exposure parameters were selected for re-evaluation. The primary re-evaluation was based on measured inorganic arsenic [iAs; As(III) + As(V)], using the updated Oral Reference Dose (Rfd) and Cancer Slope Factor (Csf) established by the USEPA Integrated Risk Information System (IRIS) in 2025. The updated toxicological values differed substantially from those previously used: the Rfd for non-carcinogenic effects decreased 5-fold, from 0.0003 to 0.00006 mg kg⁻¹ day⁻¹, whereas the Csf for carcinogenic effects increased approximately 21-fold, from 1.5 to 32 (mg kg⁻¹ day⁻¹). These changes were not assumptions introduced in the present review, but resulted from the updated USEPA-IRIS toxicological assessment of inorganic arsenic (U.S. Environmental Protection Agency (USEPA), 2025). The revised values reflect a re-evaluation of the available epidemiological, mechanistic, and dose–response evidence and therefore represent an updated characterization of the non-carcinogenic and carcinogenic potency of chronic oral iAs exposure. The lower Rfd indicates a more protective estimate of the daily oral exposure considered unlikely to produce appreciable non-carcinogenic effects over a lifetime, whereas the higher Csf reflects the revised quantitative relationship between chronic iAs exposure and cancer risk (USEPA, 2026). Consequently, historical HRA based on the former toxicological values may yield substantially different risk estimates when recalculated using the updated USEPA-IRIS criteria.

USEPA-IRIS values were selected because they provide updated quantitative dose–response parameters (Rfd and Csf) that can be directly incorporated into the THQ and CR equations used in the present re-evaluation (U.S. Environmental Protection Agency (USEPA), 2025). Although recent EFSA assessments [(EFSA CONTAM Panel (EFSA Panel on Contaminants in the Food Chain), 2024a] provide an updated toxicological evaluation of iAs, their benchmark-dose-based approach is not directly interchangeable with the Rfd- and Csf-based framework required for the quantitative recalculations performed here. The two approaches should therefore be considered complementary rather than directly equivalent. Because some studies quantified other arsenic species considered potentially toxic in addition to, or instead of, iAs, an additional conservative sensitivity scenario was also evaluated. In this scenario, the sum of measured potentially toxic arsenic species (Σtoxic As: iAs + MA + DMA) was considered for comparison with the risk estimates based exclusively on measured iAs. This scenario was treated as an upper-bound exploratory assessment and not as toxicologically equivalent to the iAs-based re-evaluation, because MA and DMA should not be assumed to have the same toxicological potency as iAs. Accordingly, the updated Rfd and Csf for iAs were applied primarily to measured iAs concentrations, whereas calculations based on Σtoxic As were used only to evaluate the influence of arsenic-speciation assumptions on risk classification (Table 1).

Table 1. Assumption underlying the Σtoxic As scenario.

These updated values were therefore applied to recalculate the Target Hazard Quotient (THQ) and Carcinogenic Risk (CR) for studies in which measured iAs concentrations and sufficient exposure information were available.

The THQ for iAs (THQiAs) was recalculated according to the equation of Tao et al. (2012) and USEPA (2026).

THQ = [(Ef * Fir * Ed * C)/(Rfd * Wab * Ta)] * 10-3

where Ef is the exposure frequency; Fir is the food ingestion rate; Ed is the exposure duration, and C is the measured iAs concentration in fish muscle; Wab is the average adult body weight; Ta is the average exposure time for non-carcinogens; Rfd is the update value of the Oral Reference Dose and 10⁻³ is a conversion factor. Ef and Fir are generally specified by country for the general population.

The Hazard Index (HI) was recalculated when PTEs other than As were evaluated, as the sum of the THQ values of each PTE.

The CRiAs was recalculated using the following equation

CR = [(Ef * Fir * Ed * C * Csf)/(Wab * Ta)] * 10-3

where Ef, Fir, Ed, Wab and Ta are the same as for THQ, 10⁻³ is a conversion factor and Csf is the update Oral Cancer Slope Factor.

2.3 Data analysis

Data extracted from the selected studies were compiled and analyzed using descriptive and comparative approaches. The number and proportion of studies were calculated according to country and geographic sector, fish species and genera, PTEs analyzed, analytical methods, regulatory or guideline values used, and human health risk assessment (HRA) approaches. Temporal and geographic patterns in the available information were also examined to identify areas, species, and PTEs with comparatively high or low research coverage.

For HRA, studies were classified according to the approach used to characterize dietary exposure and risk, including estimated daily or weekly intake (EDI/EWI), Target Hazard Quotient (THQ), Hazard Index (HI), and carcinogenic risk (CR). The exposure parameters and assumptions used in each study, particularly fish ingestion rate (Fir), exposure frequency (Ef), population group, and toxicological reference values, were compared to evaluate methodological heterogeneity among studies.

For studies in which iAs was analytically determined and sufficient exposure information was available, THQ and CR values were recalculated using the updated USEPA-IRIS toxicological reference values described in Section 2.2. Recalculated values were compared with those originally reported to determine whether the updated toxicological criteria modified risk classification. THQ and HI values were interpreted relative to the reference value of 1, whereas CR values were classified according to the risk ranges adopted in this review.

Because of the substantial heterogeneity among studies in sampling design, analytical methodology, species evaluated, sample size, exposure assumptions, and reporting of PTE concentrations, no meta-analysis of PTE concentrations was performed. Results were therefore synthesized using descriptive statistics, frequencies, proportions, and comparative risk classifications.

3 Results

3.1 General overview

A total of 97 unique studies published between 1996 and 2026 met the inclusion criteria. Three studies included sampling sites from two countries. The geographic distribution of the available evidence was markedly uneven among the three Southwestern Atlantic Ocean (SWAO) countries. Brazil, with more than 7,400 km of marine coastline, accounted for the largest number of studies (81) and also showed the greatest taxonomic coverage, with 152 fish species analyzed.

Argentina, with more than 4,700 km of coastline, was represented by only 16 studies covering 12 species, whereas Uruguay, with approximately 670 km of coastline, was represented by three studies covering seven species.

It is worth noting that more than 98% of the species evaluated in the selected studies are consumed by humans. No robust evidence of human consumption was identified for only five Brazilian species and for Boridia grossidens in Uruguay in the commercial, fisheries, and ethnobiological sources consulted (Supplementary Table S1).

Only three studies included sampling sites from more than one country, in all cases Argentina and Brazil (Avigliano et al., 2019, 2020; Biolé et al., 2023), highlighting the limited number of studies conducted at a transboundary or regional SWAO scale.

The temporal distribution of the available literature also differed among countries. Brazil showed the longest regulatory history regarding PTE concentrations in food among the countries considered, with maximum metal concentrations regulated since 1965 (Decree 55.871/65, published on March 27, 1965). Nevertheless, 63% of the Brazilian studies selected in this review were published between 2011 and 2021. In Argentina, research was even more recent, with 88% of the selected studies published during the last decade (2016–2026) (Figure 1). Moreover, evaluation of HRA was mostly or entirely performed in those periods in each country (78% and 100% in Brazil and Argentina, respectively; Figure 1). In addition to differences in the number and temporal distribution of studies, major geographic gaps were evident. Argentine studies were concentrated primarily in Buenos Aires Province, particularly in Samborombón Bay–Partido de La Costa and the Bahía Blanca estuary, and did not cover the entire Argentine coastline. To the best of our knowledge, no published studies meeting the inclusion criteria reported PTE concentrations in the muscle of coastal marine fishes between approximately 40°S and 55°S (Figure 2).

The broader geographic coverage of studies in Brazil allowed their classification into three latitudinal sectors (Figure 2). Because some studies included sampling sites located in more than one Brazilian sector, they were assigned to each corresponding sector for the geographic analysis; consequently, the sum of sector-specific study counts exceeds the total number of individual Brazilian studies. The northern sector (Sector A: 0–17°S) included 14 studies and 59 fish species belonging to 46 genera (Table 2), with the highest concentration of studies in Todos os Santos Bay (Salvador, Bahia State). Fish specimens were obtained through direct field sampling using different fishing nets and trawling surveys, collaboration with local fishermen, and collection or purchase of fresh fish from local markets.

Table 2. Summarized information of selected studies per country and sector.

The central sector (Sector B: 17–26°S; Figure 2) showed the highest research effort, with 59 studies covering 106 fish species belonging to 79 genera (Table 2). Studies were concentrated mainly in several coastal areas of Rio de Janeiro State, including Guanabara Bay, Sepetiba Bay, Santos Bay, and Cabo Frio, as well as in the Cananéia-Iguape estuarine system in São Paulo State. Fish specimens were obtained through field sampling using gillnets and different types of trawling gear, including bottom and otter trawls, collaboration with local fishermen (fishing community), and samples obtained from local ports.

The southern sector of Brazil (Sector C: 26–33.5°S; Figure 2) showed research coverage similar to that of the northern sector, with 10 studies evaluating 47 fish species belonging to 40 genera (Table 2). Sampling locations were concentrated mainly in Rio Grande do Sul State, particularly in the Patos-Mirim Lagoon system. Samples were obtained from fishery landings, artisanal and commercial fisheries, and local markets.

Only three studies were identified in Uruguay, near Montevideo city (Rio de la Plata estuary; Figure 2). A total of seven species belonging to seven genera were investigated (Table 2). Fish specimens were obtained through field sampling.

In Argentina, 16 studies were concentrated primarily in two areas of Buenos Aires Province: Samborombón Bay–Partido de La Costa and the Bahía Blanca estuary. A total of 12 species belonging to 10 genera were investigated (Table 2). As indicated above, no published studies meeting the inclusion criteria were identified for PTE concentrations in the muscle of coastal marine fishes south of approximately 40°S. Fish specimens analyzed in Argentine studies were obtained through direct field sampling using different types of fishing nets and pair trawling, artisanal and commercial fisheries, and collaboration with local fishermen.

Overall, fish samples across the reviewed literature were therefore obtained from a wide range of sources, including direct scientific sampling, artisanal and commercial fisheries, fishery landings, collaboration with local fishermen, ports, and local markets. This diversity in sample origin reflects differences in study objectives and sampling logistics and constitutes an additional source of methodological heterogeneity that should be considered when comparing PTE concentrations among species, locations, and studies.

Table 2 summarizes the fish species recorded in at least 10% of the studies within each country and Supplementary Table S2 shows the complete list of fish species studied per country. Despite the limited number of studies conducted across national boundaries, several commercially important fish species were investigated in more than one country. Species such as Mugil liza, Micropogonias furnieri, Urophycis brasiliensis, Odontesthes argentinensis, and Cynoscion guatucupa occur along the coastal waters of two or more SWAO countries and can therefore be considered shared transboundary fishery resources. However, in spite of their transboundary distribution, available information on PTE concentrations and associated HRA for these shared resources remained geographically fragmented.

Considerable methodological diversity was also observed in the analytical techniques used to quantify PTEs during the three decades covered by this review. Atomic absorption spectrometry (AAS) was used in different configurations, including portable systems, flame, graphite furnace, cold vapor for Hg determination, electrothermal atomization, and thermal decomposition-amalgamation approaches. X-ray fluorescence (XRF) was also applied using different configurations, including energy-dispersive and synchrotron-based approaches. Inductively coupled plasma (ICP)-based techniques included mass spectrometry (ICP-MS) and optical emission spectrometry (ICP-OES). Other analytical approaches included instrumental neutron activation analysis (INAA) and direct mercury analysis (DMA). Less frequently used techniques included atomic fluorescence spectrometry, gas chromatography, irradiation-based approaches, spectrofluorometry, and particle-induced X-ray emission.

AAS and ICP-based techniques were the most frequently used analytical approaches during the three decades considered, although their relative contribution differed among countries and geographic sectors. AAS and ICP-OES predominated in northern Brazil (Sector A), each accounting for 50% of studies. In central Brazil (Sector B), AAS was used in 58% of studies and ICP-OES in 19%, respectively. In southern Brazil (Sector C), AAS and ICP-MS accounted for 50% and 40% of studies, respectively. In Argentina, ICP-OES and ICP-MS predominated, accounting for 44% and 31% of studies, respectively. Notably, HPLC-ICP-MS was used in all six studies from Brazil and Argentina in which inorganic arsenic (iAs) was analytically quantified.

The number of PTEs simultaneously determined within individual studies was also highly variable. Studies analyzing fewer than four PTEs represented 43%, 59%, and 60% of the literature from Brazilian sectors A, B, and C, respectively, 100% of the studies from Uruguay, and 62% of those from Argentina (Figure 3). In contrast, studies simultaneously reporting concentrations of between 9 and 14 of the PTEs considered in this review were comparatively uncommon, accounting for 29%, 5%, and 30% of studies in Brazilian sectors A, B, and C, respectively, and 26% of studies in Argentina (Figure 3).

The frequency with which individual PTEs were investigated also varied among countries and geographic sectors (Figure 4). Among non-essential PTEs, As, Cd, Hg, and Pb were, on average, the most frequently evaluated elements, each being analyzed in more than 45% of the studies. These elements are also among those for which maximum levels or guideline values are established in the national or regional regulations considered in this review (ANVISA, AFC, and MERCOSUR). In contrast, Rb, Sr, and V were the least frequently investigated non-essential PTEs, occurring in fewer than 20% of the studies. Among essential PTEs, Cu and Zn were the most frequently evaluated; guideline values for these elements were identified in the Argentine regulations considered (AFC and SENASA), whereas Mo was the least frequently investigated essential element.

Overall, the available literature revealed a strongly uneven geographic, temporal, taxonomic, and analytical coverage of PTEs in marine fishes across the SWAO. Research was concentrated in a relatively small number of coastal areas, particularly in Brazil, whereas extensive portions of the region (most notably the southern Argentine coast) and several shared transboundary fishery resources remained poorly represented. In addition, substantial differences were observed in sample origin, analytical methodology, and the number and type of PTEs investigated. This heterogeneity constitutes an important limitation for direct comparisons among countries, species, and geographic areas and provides the context for the subsequent critical evaluation of dietary exposure and HRA.

3.2 Human risk analysis of PTE intake by fish consumption

The approaches used to evaluate the potential implications of PTE concentrations in fish muscle for human health differed substantially among studies and countries. Two main approaches were identified: (i) comparison of measured PTE concentrations with maximum levels established by national or international regulations, and (ii) quantitative assessment of dietary exposure and human health risk. Importantly, these approaches provide different types of information: compliance with regulatory concentration limits indicates whether PTE levels in fish muscle exceed established maximum values, whereas HRA incorporates exposure-related parameters, such as fish consumption rate and exposure frequency, to estimate the potential risk associated with consumption.

Comparison with regulatory limits was the most frequently applied approach. Most studies evaluated PTE concentrations against national or regional food regulations, including ANVISA in Brazil, the Argentine Food Code (AFC) in Argentina, and MERCOSUR or national regulations in Uruguay. In Brazil, such comparisons were performed in 71%, 68%, and 50% of the studies from sectors A, B, and C, respectively. The corresponding proportions were 67% in Uruguay and 81% in Argentina (Figure 5A). A smaller number of Brazilian studies additionally compared PTE concentrations with international standards or guideline values, particularly those established by EFSA or CODEX. These international references were used as complementary benchmarks in the original studies and were not considered equivalent to the national or regional regulatory requirements applicable in each SWAO country. For inorganic arsenic (iAs), some studies also used regulatory values from Australia–New Zealand and China (Zmozinski et al., 2015; Ignacio et al., 2025a,b).

In contrast, quantitative assessment of dietary exposure and human health risk was considerably less frequent (Figure 5A). HRA approaches included estimated daily or weekly intake (EDI/EWI), Target Hazard Quotient (THQ), Hazard Index (HI), and carcinogenic risk (CR). In Brazil, at least one of these approaches was applied in 36%, 31%, and 40% of the studies from sectors A, B, and C, respectively. In Argentina, HRA was performed in 69% of the studies, whereas none of the studies identified for Uruguay included a quantitative HRA (Figure 5B). The absolute number of studies reporting each type of HRA for Brazil and Argentina is shown in Figure 5B.

These results reveal an important methodological gap in the available SWAO literature. Although comparison with maximum permissible concentrations was relatively common, substantially fewer studies incorporated exposure parameters into the assessment of potential human health effects. Consequently, the available evidence on PTE occurrence in edible fish tissues is considerably more extensive than the evidence available for characterizing the actual dietary exposure and associated human health risk.

3.2.1 Estimated consumption of fish

The variables Fir and Ef determined the average fish consumption rate per capita per day (Fir) and per year (Fir x Ef), which are needed to estimate HRA. These values differed between Argentina and Brazil (no information on the Uruguayan population was available in the selected studies). In Argentina, all studies established a Fir value of 13 g person⁻¹ day⁻¹ with an Ef of 365 days year⁻¹ for the general population (or Ef: 12 days year⁻¹ × Fir: 400 g person⁻¹ day⁻¹) which results in an average consumption of 4.8 kg person⁻¹ year⁻¹ (FAO, 2016; the latest information reported by FAO for Argentina; thereafter, FAO reported a single value for all Latin America). On the other hand, in Brazil those values are somewhat dissimilar among studies; Fir has been set as low as 15 g person⁻¹ day⁻¹ [Reis et al., 2025 based on Töre et al. (2021)] up to 36 g person⁻¹ day⁻¹ (da Silva et al., 2019, 2021; da Luz et al., 2026 based on the U.S. Environmental Protection Agency (USEPA), 1989 recommendation), with several studies proposed “consumption scenarios” that set this value up to 300 g person⁻¹ day⁻¹ [20-300 g day⁻¹ (Kasper et al., 2007); 227 g person⁻¹ day⁻¹ (Trevizani et al., 2019; Gusso-Choueri et al., 2018); 300 g person⁻¹ day⁻¹ (Almeida do Carmo et al., 2012)]. These “consumption scenarios” consider dissimilar values of Ef for the Fir evaluated (12, 52,156 or 365 days year⁻¹; Almeida do Carmo et al., 2012; Gusso-Choueri et al., 2018; Trevizani et al., 2019). Considering the higher value of Ef (365 days year⁻¹) yearly consumption estimates range from 5.5 to 13 kg person⁻¹ year⁻¹, with “consumption scenarios” between 7 and 109 kg person⁻¹ year⁻¹. It is worth noting that these “consumption scenarios” of extreme values are not based on Brazilian consumer information. In addition, there are regional differences in consumption in Brazil [Instituto Brasileiro de Geografia e Estatística (IBGE), 2010]; seafood consumption in northern and northeastern Brazil is up to 104 and 40 g person⁻¹ day⁻¹, respectively, while seafood consumption in the South and Southeast is only up to 8.6 and 14.9 g person⁻¹ day⁻¹, respectively, so the highest values of seafood consumption are in Amazonia (36 kg person⁻¹ year⁻¹) [Conselho Nacional de Aqüicultura e Pesca (CONAPE), 2007; Instituto Brasileiro de Geografia e Estatística (IBGE), 2011; Grotto et al., 2012; Cardoso et al., 2019]. Moreover, several official studies report an average consumption for general population in Brazil of about 9–11 kg of fish per person annually, which is around 25–30 g person⁻¹ day⁻¹ [Conselho Nacional de Aqüicultura e Pesca (CONAPE), 2007; Instituto Brasileiro de Geografia e Estatística (IBGE), 2010, 2011; Ministério de Pesca e Aquicultura (MPA), 2009, 2013]. This information has been used in several of the selected studies here (Ferreira et al., 2015; Avigliano et al., 2019, 2020; de Oliveira and Poletto, 2020; Kütter et al., 2021; Biolé et al., 2023; Ferreira et al., 2024; Rodrigues et al., 2026). It is worth noting that the official fish consumption statistics used in the selected studies for Argentina and Brazil are, at best, 10–15 years old.

Although the importance of fish in the diet of fishermen population, estimations of health risk has been really few, in Argentina it was estimated as four times the consumption of fish for general population [19.2 kg person⁻¹ year⁻¹, Ignacio et al., 2025a,b, based on Griboff et al. (2017)] and in Brazilian reports indicated that fishermen consumption of fish is about 70 kg person⁻¹ year⁻¹ [Kütter et al., 2021 based on Instituto Brasileiro de Geografia e Estatística (IBGE), 2010].

3.2.2 Estimated daily/weekly intake

The proportions of studies that evaluated EDI or EWI and compared them with the international guidelines of PTDI or PTWI were below 20% in each country (and sector; Figure 5B). The main results for each country and sector are summarized below.

In sector A of Brazil, EWI for As (100%) and EDI for As (100%), Cd, and Zn were evaluated by Ferreira et al. (2024) and de Santana et al. (2017a), respectively. Ferreira et al. (2024) found no risk when intake was compared with PTWI but determined the presence of risk when compared with the maximum safe consuming quantity (MSCQ; Ferreira et al., 2023), while de Santana et al. (2017a) found a risk for Cd in one location when compared with maximum safe daily consumption rate.

Sector B of Brazil showed the higher number of studies evaluating estimated intake (Figure 5B). There were two types of studies: (i) studies that analyzed only one PTE and (ii) studies that analyses several PTEs. In the first group, there were four studies that evaluated Hg on a weekly intake basis and compared it with PTWI, Bauer et al. (2021) and Di Beneditto et al. (2021) conclude that intake for the species studied was below PTWI, Galvão et al. (2021) using the highest concentration of Hg for all the fish species studied, determined that intake was above PTWI, while Ferreira et al. (2012) using the “consumption scenario” concluded that intake was over PTWI for consumptions of 100 and 300 g day⁻¹. The other PTE evaluated individually was As based on a daily (Gao et al., 2018) or weekly (Angeli et al., 2013) intake. Gao et al. (2018) using measured inorganic and toxic As species, estimated intake below PTDI, while Angeli et al. (2013) using total As, detected intake above PTWI for Cathorops spixii, although they acknowledged the limitation of using total As.

In the second group, which evaluated several PTEs, two studies determined that intake was below reference values; Medeiros et al. (2014) determined for non-essential PTEs As (10%), Cd and Pb the weekly intake and compared it with PTWI and the daily intake for the essential PTEs Cu, Mn, Zn, and Se and compared it with recommended dietary allowance [Agência Nacional de Vigilância Sanitária (ANVISA), 2005], while Bauer et al. (2023) estimated the daily intake for As (3%), Cu, Hg(MeHg), Ni, Sr, and Zn and compared it with the Maximum Safe Consumption. On the other hand, two studies found some health risk from fish intake; Gusso-Choueri et al. (2018) estimated daily intake of iAs (3%), Cd, Pb, and Zn and compared it with PTDI, PTWI and PTMI, they found that intake of Cd and Pb from C. spixii was above the limits in all samples, and found some risk for As (only in 1 of 5 places, and only in 1 of 3 seasons evaluated). Finally, Reis et al. (2025) estimated daily intake of As (100%), Cd, Cr, Cu, Ni, Pb, Se, and Zn and compared them with Rfd and determine a higher intake of As (total As) from Cynoscion acoupa.

In sector C of Brazil, only two studies were found (Figure 5B). Both concluded that intake was below reference values used for the species studied. Bauer et al. (2023), as previously described in sector B, estimate the daily intake for As (3%) Cu Hg (MeHg) Ni Sr Zn and compared it with the Maximum Safe Consumption, while Grotto et al. (2012) evaluated non-essential PTEs [As (100%), Cd, Hg, Pb, Sr, V] and compared them with Rfd [except for Sr and V without Rfd values at time of the study; currently, both PTEs have Rfd values USEPA, 2026] and essential PTEs (Cu, Mn, Mo, Se, Zn) and compare with the dietary reference intakes [DRIs; Institute of Medicine (IOM), 2002].

In Argentina, three studies deal with intake. La Colla et al. (2018) determined EDI for Cd, Cu, Ni, and Zn and found values below Rfd, while Ignacio et al. (2025a) and Ignacio et al. (2025b) determine EDI for total As and As toxic species and made conservative comparisons with Rfd of iAs. In both studies Ignacio et al. (2025a) and Ignacio et al. (2025b) found that the results did not exceed the daily allowed intake limit for all age groups and children in the general and fishermen populations in all fish species when toxic As species were considered. However, Ignacio et al. (2025a) found that considering total As, the daily allowed intake limit (Rfd) was exceeded for all ages and exposure groups in more than 80% of the cases.

Overall, these studies showed that conclusions regarding dietary exposure were strongly influenced not only by PTE concentrations, but also by the exposure scenarios, reference values, and chemical forms considered in the assessment. This was particularly evident for As; studies based on measured toxic or inorganic As species generally produced different conclusions from those in which total As was used or a fixed proportion of total As was assumed to represent iAs. Thus, methodological differences in As treatment constitute a major source of variability when comparing dietary risk estimates among studies.

3.2.3 Target hazard quotients

The proportion of studies estimating THQ (and HI) varied considerably between Brazil and Argentina (Figure 5B), from 17% in sector B of Brazil to 63% in Argentina. In addition, although THQ (and HI) can be estimated for the 14 PTEs evaluated here, the only five PTEs with potential non-carcinogenic risk were commonly considered in the studies (As, Cd, Cr, Hg, and Pb). The main results for each country and sector are summarized below.

In Sector A of Brazil, THQ_Hg showed values over 1 for some of the fish species studied (Centropomus undecimalis, Mycteroperca interstitialis and Seriola spp., da Silva et al., 2019, 2021) while THQiAs in Sardinella brasiliensis showed dissimilar results; according to Ferreira et al. (2024), who assumed 100% of total As as iAs, the THQiAs > 1 while for da Silva et al. (2019) and da Silva et al. (2021) which assume 3–10% of total As as iAs, the THQiAs < 1.

In Sector B of Brazil, of the 10 studies that evaluated THQ of PTEs (Figure 5B) eight found values above 1; for As [Lopholatilus villarii; Medeiros et al., 2014 (assuming iAs to be 10% of total As); Genidens barbus; Avigliano et al., 2019 (only when assuming 20% iAs, but not when assuming 1% iAs); Cynoscion acoupa; Reis et al., 2025 (using total As for THQ estimation), and for Isopisthus parvipinnis, Paralonchurus brasiliensis and Stellifer rastrifer Trevizani et al., 2019 (assuming 3% iAs), for Hg [Euthynnus alletteratus; Rodrigues et al., 2026, Micropogonias furnieri and Trichiurus lepturus; Ferreira et al., 2015 (for a high-consumption scenario)], for Cr (Mugil curema; Almeida do Carmo et al., 2012, and Paralonchurus brasiliensis Trevizani et al., 2019), for Pb (Cathorops spixii; Gusso-Choueri et al., 2018 and Stellifer rastrifer, Trevizani et al., 2019) and for Cd (Cathorops spixii; Gusso-Choueri et al., 2018). Although PTE concentrations in these 11 species have been reported in several studies from Brazilian Sector B (Supplementary Table S1), additional THQ assessments were generally unavailable. The exceptions were Cathorops spixii and Genidens barbus. For C. spixii, Kütter et al. (2021) reported THQ < 1, although at a different location. For G. barbus, Avigliano et al. (2020) directly measured toxic and inorganic As forms, which represented less than 1% of total As, and estimated THQiAs < 1. This result is consistent with the THQiAs < 1 previously obtained by Avigliano et al. (2019) when 1% of total As was assumed to represent iAs, whereas THQiAs >1 was obtained when a 20% iAs fraction was assumed. Additional information was available for some species, although exposure was assessed through intake estimates rather than THQ. Isopisthus parvipinnis was evaluated for As by Gao et al. (2018), whereas Micropogonias furnieri and Trichiurus lepturus were evaluated for Hg by Ferreira et al. (2012). For the other six species, C. acoupa, E. alletteratus, L. villarii, M. curema, P. brasiliensis and S. rastrifer no other reports of health risk assessment estimates for the PTEs of concern in each case were identified.

In Sector C of Brazil, there were only two reports; Biolé et al. (2023) and da Luz et al. (2026) covering only two species, Urophycis brasiliensis and Mugil liza, respectively. In both reports, all THQ were below 1.

In Argentina, none of the 10 studies evaluating THQ reported THQ>1 for the general population. Considering the fishermen population, Ignacio et al. (2024) reported THQ>1 for As (assuming 5% of iAs) and Hg for Micropogonias furnieri and Urophycis brasiliensis, while Ignacio et al. (2026) reported THQ>1 for Hg for some specimens of Cynoscion guatucupa and Percophis brasiliensis.

HI values were reported only when two or more PTEs were evaluated simultaneously (Supplementary Table S1). In the selected studies, HI did not modify the interpretation derived from individual THQs; no study reported a case in which all individual THQ values were < 1, but their sum resulted in a HI >1.

3.2.4 Carcinogenic risk

Carcinogenic risk (CR) was less frequently assessed than non-carcinogenic risk in the reviewed literature, with only six studies identified in Brazil and seven in Argentina (Figure 5B). With the exception of Ignacio et al. (2026), which evaluated Cd, Pb, Cr, and inorganic As (iAs), all studies estimated carcinogenic risk exclusively for inorganic As (CRiAs). The main results for each country and Brazilian sector are summarized below.

In Sector A of Brazil, Ferreira et al. (2024) concluded that consumption of Sardinella brasiliensis could be associated with carcinogenic risk due to As exposure. However, this assessment was based on the conservative assumption that 100% of the total As concentration corresponded to iAs.

In Sector B of Brazil, Avigliano et al. (2019); and Avigliano et al. (2020) estimated CRiAs values for Genidens barbus within the re-evaluation range. Gusso-Choueri et al. (2018), for Cathorops spixii, reported CRiAs values within the acceptable range for adults, whereas estimates for children fell within the re-evaluation range. Gao et al. (2018), for the three species evaluated, reported either CR values within the acceptable range, indicating no carcinogenic concern according to the criteria used in this review, or values within the re-evaluation range when a high-consumption scenario of 54 g day⁻¹ was considered.

In Sector C of Brazil, Biolé et al. (2023) estimated CRiAs values within the acceptable range for Urophycis brasiliensis for both the general and fishermen populations. These estimates were based on the assumption that 10% of total As corresponded to iAs.

In Argentina, Biolé et al. (2023) also determined CRiAs values within the acceptable range for Urophycis brasiliensis for both the general and fishermen populations, while Biolé et al. (2024) reported CRiAs values within the acceptable range for Odontesthes argentinensis for the general population, whereas estimates for the fisher population fell within the re-evaluation range. These estimates were based on the assumption that 10% of total As corresponded to iAs. Ignacio et al. (2026) evaluated carcinogenic risk for Cd, Pb, Cr, and iAs associated with the consumption of Cynoscion guatucupa, Percophis brasiliensis, Mugil liza, and Urophycis brasiliensis. For both population groups considered, CR values remained either within the acceptable range or within the re-evaluation range. Ignacio et al. (2025a,b), considering these four species together with Micropogonias furnieri, used analytically measured iAs concentrations to estimate CRiAs. In these studies, CRiAs values did not exceed 10⁻⁴ in any case. Finally, Avigliano et al. (2019, 2020) estimated CRiAs values for Genidens barbus within the re-evaluation range, with values approaching the upper limit of 10⁻⁴.

Overall, the available CR assessments were strongly focused on As, with only one study considering additional carcinogenic PTEs. The reported risk classifications varied according to species, population group, and consumption scenario. For As, interpretation was additionally influenced by whether iAs was analytically measured or inferred from total As. Studies based on measured iAs generally reported CR values below 10⁻⁴, whereas the assessment reporting carcinogenic concern for Sardinella brasiliensis assumed that 100% of total As was present as iAs. These methodological differences should therefore be considered when comparing carcinogenic risk estimates among studies.

3.3 Arsenic: health risk re-evaluation

As previously described the methodological treatment of As differed markedly among countries and sectors (Figure 6). Whereas, total As was widely determined, direct measurement of iAs and other As species was much less frequent, and several studies instead estimated iAs by applying fixed theoretical proportions of total As (Figure 6). Among the selected studies, 12 estimated inorganic arsenic (iAs) concentrations by assuming theoretical proportions ranging from 1% to 100% of total As: 1% and 5% (Ignacio et al., 2024), 1% and 20% (Avigliano et al., 2019), 2% and 30% (Kütter et al., 2021), 3% (Gusso-Choueri et al., 2018; Trevizani et al., 2019; da Silva et al., 2021; Bauer et al., 2023), 10% (Medeiros et al., 2014; da Silva et al., 2019; Biolé et al., 2023, 2024), and 100% (Ferreira et al., 2024). However, the relative contribution of As species considered toxic to human health varies significantly among fish species and, even within the same species, among sampling sites (Ignacio et al., 2025b; Table 3). Therefore, the use of fixed theoretical proportions of iAs may introduce substantial uncertainty into HRA.

Table 3. Review of studies that measured arsenic species in muscle of marine fish species in different countries.

Concentrations of inorganic arsenic [iAs: As (III) +As(V)] and toxic arsenic (iAs + MA + DMA) are shown as mean values or mean ± standard deviation or mean value (minimum-maximum range).

^*Change in risk classification: (NC, No change; I, Increased, but remained within same category; C, Changed from intermediate to unacceptable risk; AI, Changed from acceptable to intermediate risk. THQ is also expressed as mean value (minimum-maximum range). For re-evaluated CR ranges, the maximum value is shown. NA, not available. Bold values: unacceptable cancer risk.

To minimize this source of uncertainty, the HRA re-evaluation was restricted to studies reporting measured concentrations of potentially toxic As species (iAs, MA, and/or DMA). Six studies met this criterion (Zmozinski et al., 2015; Gao et al., 2018; Avigliano et al., 2020; Ignacio et al., 2025a,b, 2026). However, Zmozinski et al. (2015), who evaluated four fish species (Supplementary Table S1) and detected the toxic organic As species (MA and DMA) but not iAs (below limit of detection) could not be included in the quantitative re-evaluation because As species concentrations were reported on a dry-weight basis and no specific conversion factor to wet weight was available; the authors only reported a broad moisture range of 45–94%. Consequently, five studies, encompassing eight fish species from Argentina and Brazil, were quantitatively re-evaluated (Table 3). Among them, only Ignacio et al. (2026) incorporated PTEs other than As into the HRA, and allowed HI to be calculated.

The THQ (and HI) re-evaluation yielded risk interpretations consistent with those of the original studies in all cases, regardless of whether As was assessed individually or together with other PTEs (Table 3). In contrast, the re-evaluation of carcinogenic risk (CRiAs) substantially changed risk classification. Using the updated cancer slope factor (Csf), approximately 21-fold higher than the value previously applied, the proportion of exposure scenarios with CR > 10⁻⁴ increased markedly, especially with the higher values of FirxEf (Table 3). Whereas, only 2% of the exposure scenarios evaluated in Table 3 showed CR values exceeding 10⁻⁴, 46% exceeded this threshold after re-evaluation (Table 3).

Except for Gao et al. (2018), which did not discriminate among the measured toxic As forms, the CR re-evaluation was based on the measured proportion of iAs and the updated Csf. If all measured As species considered potentially toxic, including the organic forms MA and DMA, were incorporated into the CR calculation following the approach used by Gao et al. (2018), CR values could exceed 10⁻⁴ in 100% of the exposure scenarios (Table 3). However, this should be regarded as a conservative exploratory scenario because the Csf used in the calculation has been established only for iAs (Table 1). This result highlights the strong influence of both As speciation and toxicological assumptions on carcinogenic risk estimates and emphasizes the need to distinguish measured iAs from other As species when comparing HRA results among studies.

4 Discussion

This critical review provides an integrated regional assessment of potentially toxic trace elements (PTEs) in marine fishes from the Southwestern Atlantic Ocean (SWAO) and shows that the principal limitation of the available evidence is not simply the number of studies, but their pronounced geographic, taxonomic, analytical, and toxicological heterogeneity. More importantly, differences among published human health risk assessments (HRA) cannot be interpreted exclusively as reflecting differences in contaminant concentrations among species or sampling areas. Fish-consumption assumptions, exposure frequency, population group, analytical strategy, toxicological reference values, and particularly the approach used to estimate inorganic arsenic (iAs) exposure differed substantially among studies. Consequently, part of the variability in reported health-risk classifications is methodological rather than necessarily environmental. This distinction is particularly relevant in the SWAO, where coastal environments are exposed to multiple anthropogenic pressures associated with urbanization, industrial activities, agriculture, ports, and wastewater discharges (Dias et al., 2013; Newton et al., 2020; Strauch and Erzinger, 2021).

4.1 Three decades of PTE research in SWAO marine fishes: advances, trends, and remaining gaps

Over the last three decades, research on PTEs in marine fishes of the SWAO has expanded substantially, but unevenly across time and space. Most Brazilian studies were published between 2011 and 2021, whereas research in Argentina intensified mainly during the last decade. Despite this growth, Brazil clearly dominates the available evidence, while Uruguay and large portions of the Argentine coast remain poorly represented. Moreover, multinational studies remain exceptional despite the occurrence of several commercially important transboundary fishery resources (Avigliano et al., 2019, 2020; Biolé et al., 2023). Thus, increased research effort has primarily strengthened localized knowledge rather than producing a coordinated regional assessment of shared fishery resources.

Analytical and risk-assessment approaches have also diversified. AAS remains widely used, particularly in Brazil, whereas ICP-OES and ICP-MS are proportionally more prominent in the Argentine literature. More recently, HPLC-ICP-MS has enabled direct determination of individual As species, representing an important advance over assessments based exclusively on total As or assumed iAs fractions (Maher et al., 2015; Luvonga et al., 2020b). In parallel, seafood-safety assessment has gradually expanded from comparison with regulatory maximum levels toward quantitative exposure and risk approaches, including EDI/EWI, THQ, HI, and CR (Yi et al., 2011; Tao et al., 2012; USEPA, 2026). Nevertheless, this transition remains incomplete, and information on PTE occurrence is still considerably more extensive than evidence on actual dietary exposure and health risk.

Important gaps therefore persist in geographic coverage, transboundary resources, high-consumption and vulnerable populations, updated fish-consumption information, analytical harmonization, and direct As speciation. These gaps are interconnected because geographic fragmentation limits spatial interpretation, non-local or outdated consumption parameters increase uncertainty in exposure estimates, and inadequate As speciation can substantially alter toxicological interpretation. Future SWAO research should therefore prioritize regional representativeness and methodological comparability rather than simply increasing the number of isolated studies.

4.2 Geographic and taxonomic gaps in SWAO PTE research

The marked geographic imbalance identified in this review limits the regional representativeness of the available evidence. Research effort is concentrated in a relatively small number of Brazilian coastal and estuarine systems and, to a lesser extent, in a few areas of Buenos Aires Province, whereas Uruguay and the Argentine coast south of approximately 40°S remain poorly represented. Consequently, existing information cannot be assumed to represent the SWAO as a whole.

This geographic fragmentation is particularly relevant for transboundary species such as Mugil liza, Micropogonias furnieri, Urophycis brasiliensis, Odontesthes argentinensis, and Cynoscion guatucupa. Although these resources occur in two or more SWAO countries, multinational assessments remain exceptional (Avigliano et al., 2019, 2020; Biolé et al., 2023). Consequently, available evidence largely represents local conditions rather than contaminant exposure throughout the distribution range of shared fishery resources.

This limitation has implications beyond environmental monitoring because fish harvested in one geographic area may enter domestic or international commercial chains and be consumed far from the sampling locality. PTE concentrations measured at a particular site should therefore not automatically be extrapolated to an entire species or distributional range. Coordinated studies covering multiple areas and countries would help distinguish spatial variation in environmental exposure from species-specific accumulation patterns and would provide a stronger basis for regional seafood-safety assessment.

The dietary relevance of the species investigated should also be considered when defining monitoring priorities. The great majority of species included in the reviewed literature have documented use for human consumption, whereas robust evidence of regular consumption could not be identified for only a small number of taxa. Future HRA should therefore explicitly establish whether the species and size classes analyzed effectively enter local, subsistence, domestic-commercial, or export food chains before applying human-consumption scenarios.

4.3 Analytical heterogeneity and PTE coverage

Analytical heterogeneity represents another important source of uncertainty in between-study comparisons. Differences in instrumentation alone do not necessarily invalidate comparisons when validated methods and adequate quality-assurance and quality-control procedures are used. However, comparability becomes increasingly uncertain when instrumental differences are combined with variation in sample preparation, digestion procedures, tissue selection, moisture basis, detection and quantification limits, analytical recovery, calibration procedures, certified reference materials, and reporting units (El Hosry et al., 2023).

The analytical platforms identified in this review differ in sensitivity, selectivity, multi-element capability, and susceptibility to interferences. In general, ICP-MS provides lower detection limits and broader multi-element capability than ICP-OES or conventional AAS, whereas element-specific AAS methods, including graphite-furnace and cold-vapor approaches, can provide adequate sensitivity for targeted analyses. Nevertheless, analytical performance depends not only on the instrumental platform but also on the analyte, sample preparation, calibration strategy, digestion efficiency, and matrix composition (El Hosry et al., 2023). ICP-based techniques may be affected by spectral, isobaric, and polyatomic interferences, whereas AAS may be influenced by chemical and background-absorption effects; complex fish-tissue matrices may additionally cause signal suppression or enhancement. Appropriate digestion, calibration controls, blanks, recovery assessment, and certified reference materials are therefore essential for distinguishing analytical variability from genuine biological or geographic differences.

For As, analytical selectivity is particularly relevant because total-As determination and chemical speciation provide fundamentally different toxicological information. Reliable speciation requires efficient extraction and preservation of the original chemical forms, chromatographic separation, appropriate standards, and element-specific detection. HPLC-ICP-MS enables direct distinction and quantification of iAs and organoarsenic species, but introduces additional uncertainty associated with extraction efficiency, species transformation during preparation, chromatographic recovery, and matrix effects (Maher et al., 2015; Wolle and Conklin, 2018; Luvonga et al., 2020b).

The restricted number of PTEs simultaneously analyzed in many studies represents an additional limitation. Although the predominance of As, Cd, Hg, and Pb is understandable because of their toxicological and regulatory relevance, absence of reported risk for another element cannot be interpreted as absence of exposure when that element was not analyzed. This becomes particularly important for cumulative non-carcinogenic assessment because HI represents the sum of the individual THQs included in the calculation.

Consequently, an HI based on a restricted analytical panel characterizes only the fraction of the contaminant burden represented by the PTEs actually measured.

Overall, differences in analytical procedures, PTE coverage, reporting formats, and Quality Assurance/Quality Control (QA/QC) can generate between-study variability independently of genuine biological or geographic differences. Lower instrumental detection limits therefore do not necessarily imply greater data quality or comparability; robust interpretation requires validated methods, transparent reporting of LOD/LOQ and recoveries, control of matrix effects and interferences, and consistent expression of concentrations on a comparable mass basis. These sources of heterogeneity justify the descriptive and comparative synthesis adopted here rather than a formal meta-analysis and support the use of harmonized analytical protocols, explicit QA/QC reporting, certified reference materials, wet-weight reporting for food-safety purposes, and sufficiently broad multi-element panels in future regional assessments.

4.3.1 Regulatory compliance and HRA provide complementary information

Regulatory compliance and quantitative HRA should be regarded as complementary rather than interchangeable components of seafood-safety evaluation. Maximum contaminant concentrations are essential tools for food control and risk management, but they do not by themselves characterize dietary risk because exposure additionally depends on fish-consumption rate and frequency, exposure duration, body weight, characteristics of the consumer population, and contaminant-specific toxicity.

Quantitative HRA therefore addresses a different question by integrating contaminant occurrence with exposure and toxicological parameters to estimate whether the dose associated with a defined consumption scenario may represent a potential health concern [USEPA, 2026]. Studies specifically combining contaminant measurements in fish and shellfish with dietary risk assessment have similarly shown that concentration-based regulatory evaluation and exposure-based risk characterization provide related but distinct information (Angeli et al., 2013; Ferreira et al., 2015; da Araújo et al., 2016; da Silva et al., 2019, 2021; Kütter et al., 2021; Biolé et al., 2023, among others).

This distinction becomes especially relevant when contaminant concentrations approach regulatory thresholds or when high-consumption and vulnerable populations are considered. A concentration below a regulatory maximum does not necessarily imply equivalent exposure across all consumer groups, whereas exceedance of a maximum concentration does not itself quantify the magnitude of health risk associated with actual consumption. The comparatively limited use of EDI/EWI, THQ, HI, and CR in the SWAO literature therefore indicates that knowledge of contaminant occurrence has advanced more rapidly than characterization of actual consumer exposure.

4.3.2 Strengths and limitations of current HRA approaches

The HRA approaches identified in the reviewed literature provide complementary but conceptually different information and should not be interpreted as equivalent measures of risk. EDI and EWI estimate dietary exposure and allow comparison with health-based guidance values such as PTDI or PTWI, but do not themselves quantify the probability or severity of an adverse effect. THQ compares estimated exposure with a contaminant-specific reference dose and is therefore a screening metric for potential non-carcinogenic effects rather than a probability of toxicity; consequently, THQ >1 indicates the need for further evaluation rather than evidence that adverse effects will necessarily occur (Yi et al., 2011; Tao et al., 2012; USEPA, 2026).

HI extends non-carcinogenic assessment by summing individual THQs, but its interpretation depends on the PTEs actually measured and assumes simplified additivity. An HI derived from a restricted analytical panel may therefore characterize only part of the potential contaminant burden and does not explicitly account for differences in toxicological mechanisms, target organs, contaminant interactions, or the evidence underlying individual reference doses. CR addresses lifetime carcinogenic risk using a contaminant-specific cancer slope factor and is consequently highly sensitive to the toxicological potency value applied, as demonstrated by the substantial change in iAs risk classification obtained in the present re-evaluation after updating the Csf.

More generally, all HRA estimates depend strongly on their input parameters, including contaminant concentration, fish ingestion rate (Fir), exposure frequency (Ef), exposure duration (Ed), body weight (BW), and the population considered. Thus, apparently precise numerical estimates may conceal substantial uncertainty when national average consumption rates or default exposure parameters are transferred to fishing communities, children, or other high-consumption or vulnerable populations. This dependence on exposure assumptions is especially relevant in the SWAO, where consumption parameters varied considerably among the selected studies.

Most HRAs identified here were deterministic and based on single-point estimates or predefined exposure scenarios. These approaches are transparent, reproducible, and useful for screening, but they do not fully characterize population variability or the combined uncertainty of multiple input parameters. When sufficiently representative empirical data are available, deterministic assessments could be complemented by probabilistic approaches, including Monte Carlo simulation and sensitivity analysis, to characterize exposure distributions, high-exposure percentiles, and the relative contribution of individual parameters to overall uncertainty [USEPA, 2001]. However, such models should only be applied when the underlying distributions of contaminant concentrations, consumption, body weight, exposure frequency, and other relevant variables are supported by adequate empirical data.

4.3.3 Fish consumption is a major source of uncertainty in health risk assessment

Fish-consumption parameters were among the most heterogeneous components of the reviewed HRA. Because estimated dietary exposure is directly dependent on the amount and frequency of consumption, comparable PTE concentrations may produce markedly different THQ or CR estimates when different ingestion rates, exposure frequencies, or population characteristics are applied.

The use of national average consumption rates or generic default values is therefore particularly problematic where strong geographic, socioeconomic, and cultural differences in fish consumption occur. This is well-documented in Brazil, where consumption may be substantially higher in Amazonian and fish-dependent communities than suggested by national averages, directly affecting estimated contaminant exposure (Hacon et al., 2020; Ignacio et al., 2025a, 2026, Table 3).

Children, pregnant women, women of childbearing age, and other potentially vulnerable groups require specific consideration because exposure and susceptibility differ across life stages, and lower body weight can increase the estimated dose for a given intake. This is particularly relevant for methylmercury (MeHg), for which fetal and early-childhood development represent sensitive exposure windows. Epidemiological studies have associated prenatal MeHg exposure with effects on several neurodevelopmental domains, including attention, language, memory, visuospatial function, and cognitive performance, although the magnitude and consistency of these associations vary among cohorts and exposure contexts [Grandjean et al., 1997; Myers et al., 2003; EFSA CONTAM Panel (EFSA Panel on Contaminants in the Food Chain), 2012]. Studies in Amazonian communities similarly support separate assessment of children and adolescents (Mendes et al., 2020). Importantly, these findings should not be interpreted as evidence that fish consumption itself is detrimental: fish provides high-quality protein, long-chain n-3 polyunsaturated fatty acids, selenium, iodine, and other nutrients important for fetal and child development. Thus, risk-benefit interpretation should focus on contaminant exposure associated with particular species, consumption patterns, and consumer groups rather than generalized reductions in fish intake.

Inorganic arsenic (iAs) also warrants consideration during pregnancy and childhood, although epidemiological evidence for developmental effects is less consistent than for MeHg. Interpretation is complicated by co-exposure, nutritional and socioeconomic factors, and differences in exposure assessment; moreover, most evidence derives from drinking water or mixed dietary exposure rather than specifically from marine fish consumption. This distinction is critical because As in seafood is frequently dominated by organoarsenic compounds of much lower toxicological relevance than iAs [Luvonga et al., 2020a,b; EFSA CONTAM Panel (EFSA Panel on Contaminants in the Food Chain), 2024a]. Consequently, epidemiological evidence based on drinking-water or total-As exposure should not be extrapolated directly to marine-fish consumption without information on As speciation.

Overall, transferring national averages or exposure parameters derived from unrelated populations may underestimate exposure in high-consumption communities while overestimating it among less frequent consumers, and this uncertainty is compounded by the age of much of the consumption information used in the reviewed HRA. Changes in dietary habits, fish availability, prices, commercial distribution, and consumer preferences make periodically updated and geographically resolved dietary surveys essential. This is particularly important because pregnant women and children were rarely evaluated in the selected studies (Joyeux et al., 2004; Almeida do Carmo et al., 2012; de Santana et al., 2017a; Gusso-Choueri et al., 2018; Cardoso et al., 2019; Trevizani et al., 2019; Galvão et al., 2021; Biolé et al., 2024; Ignacio et al., 2025a). Future HRA should therefore use recent, geographically and culturally representative consumption data and explicitly distinguish average consumers from high-consumption populations, fishing and traditional communities, children, pregnant women, women of childbearing age, and other potentially vulnerable groups.

4.4 Arsenic speciation is the principal methodological determinant of As-related risk

Among all methodological sources of variability identified in this review, the treatment of As had the greatest potential to modify risk classification. The theoretical fractions of total As assumed to represent iAs ranged from 1% to 100%, making direct comparisons among HRAs intrinsically problematic. This limitation is particularly relevant in seafood because total As is an inadequate proxy for toxicological risk owing to the diversity of As compounds and their markedly different toxicological properties (Luvonga et al., 2020a,b).

The contrasting risk classifications reported for S. brasiliensis and G. barbus demonstrate how assumptions concerning the iAs fraction can generate apparently contradictory HRA outcomes even without evidence of corresponding differences in environmental exposure.

This interpretation is consistent with the broader seafood literature. Arsenobetaine (AsB) is commonly a major As species in marine fish and is generally considered of low toxicological concern, whereas iAs frequently represents only a small proportion of total As. At the same time, the toxicological information available for other organoarsenic compounds remains incomplete, and they should not automatically be assumed to be equivalent either to AsB or to iAs (Luvonga et al., 2020a,b). Emerging mechanistic evidence from mouse models suggests that AsB, although generally regarded as a low-toxicity and rapidly excreted arsenic species, may undergo partial biotransformation in the gastrointestinal tract, with gut microbiota potentially influencing the resulting arsenic-speciation profile (Zhang et al., 2023; Mukherjee et al., 2024). However, these findings derive from experimental animal models and should not be extrapolated directly to quantitative human seafood-risk assessment. They do not demonstrate toxicological equivalence between AsB and inorganic arsenic (iAs), nor do they justify applying the oral Reference Dose (Rfd) or Cancer Slope Factor (Csf) established for iAs to AsB. Accordingly, these studies do not modify the THQ or CR calculations performed in the present review. Rather, they highlight that the metabolism and toxicological relevance of organoarsenic compounds are more complex than previously assumed and remain an important source of uncertainty in seafood-related arsenic risk assessment. This further reinforces the need for species-specific arsenic determination and for avoiding the use of total arsenic or undifferentiated organic arsenic as direct surrogates for toxicologically relevant iAs.

Direct speciation studies further demonstrate that the relative contribution of individual As species can vary among fish species and among sampling sites for the same species (Zmozinski et al., 2015; Gao et al., 2018; Avigliano et al., 2020; Ignacio et al., 2025a,b, 2026). A universal conversion factor from total As to iAs is therefore unlikely to adequately represent the variability occurring in marine fishes. Analytical reviews likewise emphasize that reliable dietary exposure assessment requires direct identification and quantification of individual As species rather than extrapolation from total As alone (Wolle and Conklin, 2018; Luvonga et al., 2020b; Chung, 2025). However, routine implementation of arsenic speciation is more analytically demanding and costly than total-As determination because it requires appropriate extraction procedures, preservation of chemical species, chromatographic separation, element-specific detection, specialized instrumentation such as HPLC-ICP-MS, and trained personnel. These constraints may be particularly relevant for routine monitoring programs with limited analytical capacity and further support a tiered strategy in which total As is used for initial screening followed by confirmatory speciation of samples of concern.

The intake assessments reviewed here reinforce this interpretation: conclusions based on total As differed from those based on analytically measured inorganic or potentially toxic As species. Consequently, total As should primarily be regarded as a screening parameter when HRA is the objective. Elevated total As should trigger confirmatory speciation rather than automatic conversion to iAs through a fixed theoretical percentage. This tiered approach would reduce the likelihood of both overestimating and underestimating risk.

4.5 Non-carcinogenic risk is scenario-specific

The occurrence of THQ >1 in the reviewed literature was restricted to particular combinations of contaminant, species, geographic area, and consumption scenario rather than representing a generalized pattern across the SWAO. Similar context-dependent variation has been reported in other marine regions, where PTE accumulation and consumer-risk estimates differed according to species, location, environmental conditions, and consumption pattern (Renieri et al., 2019; Prabakaran et al., 2024).

Importantly, several exceedances identified in the present review depended on high-consumption or fishing-community scenarios. Risk classification should therefore be interpreted at the level of a clearly defined exposure scenario rather than generalized to an entire species, sampling area, or country. Contaminant concentration alone is insufficient to identify the populations at greatest potential risk because estimated exposure also depends on consumption rate and frequency, body weight, and contaminant-specific toxicological parameters.

Likewise, THQ >1 under a particular exposure scenario should not be interpreted as evidence that marine fish consumption in the SWAO is generally hazardous. Current international risk-benefit approaches emphasize that fish provides important nutritional benefits and recommendations should consider both contaminant exposure and the nutritional value of seafood, together with the characteristics of the species and consumer population (FAO/WHO, 2024).

Risk-management and communication strategies should therefore target the specific combinations of contaminant, species, geographic area, and consumer group associated with increased exposure. This would allow consumer-protection measures to be better targeted while avoiding unnecessarily broad recommendations that could discourage fish consumption.

4.6 Updated toxicological criteria substantially modify carcinogenic risk interpretation

The re-evaluation using updated USEPA-IRIS toxicological values represents one of the principal findings of this review. Although the 5-fold reduction in the oral Rfd for iAs did not modify the qualitative non-carcinogenic risk classification of the studies that could be re-evaluated, the approximately 21-fold increase in the Csf substantially changed the interpretation of carcinogenic risk. The proportion of re-evaluated cases with CR >10⁻⁴ increased from 2% to 46%.

This result is particularly informative because neither measured contaminant concentrations nor consumption assumptions were modified. The change in classification resulted from updating the toxicological potency factor. HRA results should therefore not be regarded as static : interpretation of historical contaminant data may change when new epidemiological, mechanistic, and dose-response evidence leads regulatory agencies to revise toxicological reference values.

Recent reassessments of iAs illustrate this process. The substantial changes in the USEPA-IRIS Rfd and Csf reflect an updated toxicological assessment rather than an arbitrary change in risk-assessment parameters. The revised values incorporate newer epidemiological, mechanistic, and dose–response evidence and therefore modify the quantitative interpretation of chronic oral iAs exposure (U.S. Environmental Protection Agency (USEPA), 2025). Independently, the EFSA CONTAM Panel reassessed chronic dietary exposure to iAs and identified carcinogenic effects as critical for risk characterization [EFSA CONTAM Panel (EFSA Panel on Contaminants in the Food Chain), 2024a]. These independent reassessments reinforce the need to periodically revisit HRA when the toxicological values on which they were originally based change substantially.

Interpretation nevertheless remains dependent on which As species are included in the exposure calculation. In the sensitivity scenario, in which all measured As species considered potentially toxic, including MA and DMA, were incorporated following the approach of Gao et al. (2018), CR values exceeded 10⁻⁴ in all evaluated cases. However, this finding should not be interpreted as evidence that MA, DMA, and iAs have equivalent carcinogenic potency. Arsenic species differ in their toxicokinetics and toxicity, and recent assessments have evaluated small organoarsenic species separately rather than assigning them the same toxicological potency as iAs [EFSA CONTAM Panel (EFSA Panel on Contaminants in the Food Chain), 2024b]. The CONTAM Panel highlighted major knowledge gaps for MA and DMA, including the lack of robust and validated analytical methods and certified reference materials for organoarsenic species in foods such as rice and seafood, as well as limited information on their metabolism and health effects in humans following ingestion. No studies are currently available on adverse health effects in humans associated with dietary exposure to MA or DMA, and human toxicokinetic data are also limited. Although there is convincing evidence that DMA is carcinogenic in rodents, additional toxicological studies, particularly on developmental and neurotoxic effects, are needed for both MA and DMA. Given these substantial data gaps, the CONTAM Panel concluded that it is currently not appropriate to establish health-based guidance values for either MA or DMA [EFSA CONTAM Panel (EFSA Panel on Contaminants in the Food Chain), 2024b].

The sensitivity scenario should therefore be interpreted as evidence of the strong dependence of risk estimates on both chemical speciation and the toxicological assumptions applied. The chemical species included in the exposure estimate and the toxicological parameters assigned to each species are inseparable components of As risk characterization.

The small number of studies suitable for quantitative re-evaluation further highlights the importance of standardized analytical and reporting procedures. Reliable As speciation requires adequate extraction, chromatographic separation, species preservation, analytical quality control, and explicit reporting of the concentration basis (Wolle and Conklin, 2018; Luvonga et al., 2020b). Reporting individual As species, detection and quantification limits, analytical methodology, wet- or dry-weight basis, and exposure parameters would not only improve comparisons among contemporary studies but would also allow existing datasets to be re-used when toxicological reference values are subsequently revised.

4.7 Implications for regional monitoring, regulatory updating, and public policy

The evidence synthesized in this review supports greater regional coordination of seafood-safety monitoring, particularly for transboundary fishery resources whose contaminant profiles may vary geographically while products circulate through domestic and international markets. Brazil, Uruguay, and Argentina would benefit from common minimum monitoring requirements covering sampling design, edible tissue, fish size, sampling location and date, wet-weight reporting, analytical QA/QC, and limits of detection and quantification. Such harmonization would not require identical national programs but would improve comparability among datasets and is consistent with the standardization principles of MERCOSUR and Codex contaminant-management frameworks (MERCOSUR/GMC, 2011; Codex Alimentarius Commission, 2025).

A second priority is updated and geographically resolved fish-consumption information. National surveys such as Brazil's Pesquisa de Orçamentos Familiares (POF, IBGE) and Argentina's Encuesta Nacional de Gastos de los Hogares (ENGHo, INDEC) provide potential platforms for periodically distinguishing the general population from fishing and traditional communities, high-consumption groups, children, and other vulnerable consumers. Incorporating such population-specific information into HRA would reduce reliance on historical national averages and improve the correspondence between estimated and actual exposure.

For As, monitoring should adopt a tiered approach in which total As is retained as an initial screening measurement and samples of concern undergo confirmatory speciation, rather than converting total As to iAs using a fixed theoretical percentage. MERCOSUR GMC Resolution No. 18/21 provides a relevant regulatory precedent by introducing an iAs-specific criterion for rice, demonstrating recognition that chemical speciation becomes important when different forms of an element differ substantially in toxicological relevance (MERCOSUR/GMC, 2021). Although this criterion does not apply to fish, it supports periodic reassessment of contaminant limits as analytical capacity, toxicological evidence, and dietary-exposure information evolve. Such updating is feasible within existing frameworks: MERCOSUR/GMC Resolution No. 12/11 established harmonized limits for inorganic contaminants in foods; Brazil subsequently consolidated its framework through RDC No. 722/2022 and IN No. 160/2022; Argentina incorporated GMC Resolution No. 18/21 and updated Article 156 of the Argentine Food Code through Joint Resolution No. 4/2026; and Uruguay incorporated the corresponding MERCOSUR update through Decree No. 187/024 [Agência Nacional de Vigilância Sanitária (ANVISA), 2022a; Administración Nacional De Medicamentos, Alimentos y Tecnología Médica - Secretaría de Agricultura, Ganadería y Pesca (ANMAT-SAGyP), 2026; Uruguay, 2024]. Future reviews should therefore consider whether limits applicable to fish adequately reflect chemical speciation, toxicological potency, dietary exposure, and high-consumption or vulnerable populations.

Regulatory and monitoring agencies should also periodically re-evaluate existing HRA when major toxicological reference values change. Maintaining sufficiently detailed and standardized historical datasets would allow recalculation of risk using updated criteria without necessarily repeating complete environmental sampling campaigns. Monitoring priorities should therefore integrate hazard, exposure, and dietary relevance rather than contaminant concentration alone. Species should be prioritized according to contaminant concentration and chemical form together with their importance in the human diet, consumption frequency, occurrence of high-consumption populations, and geographic extent of commercial distribution; thus, a moderately contaminated but widely consumed species may warrant greater surveillance than a more contaminated species that is rarely consumed (Codex Alimentarius Commission, 2025). Geographic gaps should likewise guide monitoring effort, with Uruguay and the Argentine coast south of approximately 40 °S representing priorities for baseline information, while continued surveillance remains necessary in highly urbanized and industrialized Brazilian coastal systems.

Regional integration would be strengthened by a shared or interoperable database containing standardized information on species, sampling location and date, fish size, edible tissue, concentration basis, analytical methodology, PTE concentrations, As speciation when available, consumption parameters, population group, and calculated EDI/EWI, THQ, HI, and CR. Such a system would facilitate comparison of transboundary resources, detection of spatial and temporal trends and knowledge gaps, and rapid re-evaluation of historical data when toxicological reference values or regulatory limits are revised, while also providing a scientific basis for future updates of the MERCOSUR Technical Regulation on inorganic contaminants. As an initial practical step, a regional technical workshop involving food-safety, fisheries, environmental-monitoring, and public-health authorities from Brazil, Uruguay, and Argentina, potentially within existing MERCOSUR cooperation mechanisms, could define a minimum harmonized dataset and analytical criteria for monitoring shared fishery resources.

Risk communication should nevertheless remain proportional to the evidence. The present review does not support characterizing marine fish consumption in the SWAO as generally hazardous; potential health concerns were associated with specific combinations of contaminant, species, geographic area, and consumption pattern. Public-health recommendations should therefore be species-, area-, and population-specific, protecting consumers with greater potential exposure while avoiding unnecessarily broad advice that could discourage fish consumption and compromise its nutritional and socioeconomic benefits.

Several limitations should be considered when interpreting these recommendations. The available literature is geographically and taxonomically uneven and includes heterogeneous sampling designs and sample sources, analytical methods, concentration bases, QA/QC practices, PTE coverage, fish-consumption parameters, population groups, exposure scenarios, regulatory frameworks, toxicological reference values, and HRA assumptions. The review was also restricted to the predefined inclusion criteria and published literature identified through the search strategy; unpublished or inaccessible data may therefore be absent. This heterogeneity precluded a formal meta-analysis and means that between-study differences should not automatically be interpreted as geographic differences in environmental contamination or consumer risk, because part of the observed variability may derive from methodological and analytical differences.

5 Conclusions and regional perspective

Overall, the synthesis presented here demonstrates that the principal challenge for SWAO PTE research is not simply the number of available studies, but the uneven geographic, methodological, and toxicological structure of the evidence base. PTE research is concentrated in a limited number of coastal areas, HRA is much less frequent than comparison with regulatory limits, consumption parameters are heterogeneous and often based on relatively old information, and only a small proportion of studies have directly measured inorganic As, and even fewer have measured MA and DMA.

Most importantly, this review demonstrates that variability among published HRA reflects both actual differences in contaminant occurrence and variability introduced by methodological assumptions. Arsenic provides the clearest example: theoretical iAs fractions (ranging from 1% to 100%) can fundamentally alter risk classification, while direct speciation demonstrates that the toxic fraction varies among species and locations (Avigliano et al., 2020; Ignacio et al., 2025b). Furthermore, application of the updated USEPA-IRIS Csf increased the proportion of re-evaluated cases with CR >10⁻⁴ from 2% to 46%, demonstrating the importance of periodically revisiting HRA when toxicological criteria change.

Future SWAO seafood-safety research should therefore prioritize broader geographic coverage, direct As speciation, updated and population-specific fish-consumption information, multi-element assessment, standardized reporting, and coordinated investigation of transboundary fishery resources. Incorporating these priorities into public-health, fisheries, and environmental-monitoring policies would allow Brazil, Uruguay, and Argentina to distinguish more reliably between genuine geographic differences in contaminant exposure and differences created by methodological assumptions, providing a stronger scientific basis for seafood safety and evidence-based regional decision-making.

Generative AI statement

The author(s) declared that Generative AI was used in the creation of this manuscript. ChatGPT (OpenAI) was used solely for English-language editing and grammatical correction. All scientific content was developed, reviewed, and verified by the author(s).

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher's note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Supplementary material

The Supplementary Material for this article can be found online at: https://www.frontiersin.org/articles/10.3389/focsu.2026.1971894/full#supplementary-material

The supplementary Excel file, ‘Supplementary Table S1.xlsx' contains the full references to the 97 selected studies published during the past 30 years (1996–2026) and included in this review, and ‘Supplementary Table S2.xlsx' contains the full list of fish species included in the selected studies for each country, according to currently accepted names in FishBase (Froese and Pauly, 2026). We retain full responsibility for all errors of fact, omission, and interpretation.

Supplementary material Supplementary material at the source

Supplementary material Supplementary material at the source

Footnotes

Edited by: Vinicius Tavares Kutter, Federal University of Pará, Brazil

Reviewed by: David Johane Machate, Federal Institute of São Paulo, Brazil

Sami Mili, University of Carthage, Tunisia