Soil contaminants at the environment-health interface require function-centered risk assessment

1 Introduction

Soil contamination includes metals and metalloids, petroleum hydrocarbons, pesticides, pharmaceuticals, persistent organic pollutants, industrial chemicals, and particulate materials. Agricultural inputs, mining, industry, waste disposal, spills, and atmospheric deposition produce different combinations of these stressors (1). Minerals, organic matter, water, roots, and microorganisms continually modify contaminants’ chemical forms and transport. Accordingly, an identical total concentration can coexist with different soluble, sorbed, precipitated, or organism-accessible pools. Total-concentration measurements remain indispensable for inventories and regulatory screening, but cannot alone determine receptor-specific exposure (2).

Mixtures here mean co-occurring substances within and across contaminant classes, including relevant transformation products; different fractions of one substance describe its partitioning, not a chemical mixture. Silva et al. (3) detected pesticide residues in 83% of 317 European agricultural topsoils, with two or more residues in 58%. Köninger et al. (4) examined 63 pesticides across 373 European sites, detected residues at 70% of sites, and identified associations with soil biodiversity and functional attributes. These surveys establish widespread occurrence and motivate biological investigation, but do not make every detection an adverse-effect threshold or demonstrate human-health risk.

The monitoring gap concerns the limited integration of soil-contamination measurements with exposure, effects, and recovery assessment. Occurrence surveys, extraction tests, organism responses, and soil-process measurements often answer different questions. Established ecological risk assessment already integrates exposure and effects through problem formulation, analysis, and risk characterization (5). Our proposal therefore strengthens implementation by making soil-function endpoints and repeated recovery measurements explicit, rather than claiming to replace existing risk assessment. We hypothesize that combining receptor-relevant exposure, soil-state variables, and functional indicators will improve prediction of impairment and recovery relative to total concentrations alone.

2 Soil state governs contaminant fate and accessible exposure

Terminology must specify the receptor, exposure route, timescale, and measurement method. In this framework, the biologically accessible fraction is an operational umbrella for contaminant mass that can reach or contact a specified receptor during the assessment period; it is not a universally measurable soil property or a synonym for absorbed dose. Bioaccessibility describes release from the matrix under a defined test or biological setting. In the oral-exposure context discussed by Ruby et al. (6), it concerns release into gastrointestinal fluids and is distinguished from bioavailability, which involves absorption. An in vitro bioaccessible fraction predicts oral bioavailability only where an appropriate relationship has been validated. Ecological usage may encompass material available now or becoming available over time (7). We therefore use receptor-specific exposure in decision-making and retain assay names when reporting extracted fractions.

For cationic metals such as cadmium and zinc, acidification often increases solubility by weakening sorption and dissolving reactive mineral phases. Clay exchange sites, iron and manganese oxides, carbonates, and organic functional groups control retention through adsorption, complexation, precipitation, and coprecipitation (8). Metalloids cannot simply be assigned the same pH response: arsenic behavior depends on oxidation state, oxide binding, and competing ions. Reductive dissolution of iron oxides can release associated arsenic, whereas other reducing conditions may favor retention in sulfide phases (2). Treatment effects must therefore be evaluated for each contaminant and soil state.

Organic amendments provide a practical example of these competing mechanisms. Compost, manure-derived materials, and biochar can alter pH, sorption capacity, microbial substrates, and DOM. Solid organic matter may retain contaminants, while soluble ligands can mobilize metals. Lian et al. (9) demonstrated DOM-dependent mobilization and complexation of zinc, cadmium, and lead in rhizosphere solutions. Dissolved concentration therefore does not necessarily equal free-ion activity, uptake, or toxicity. Amendment composition, dose, and ageing require evaluation alongside pore-water chemistry and receptor responses.

For organic chemicals, hydrophobic partitioning into soil organic matter, sorption to mineral surfaces, ionization, and diffusion into less accessible domains affect exposure. Ageing may reduce rapid desorption, while degradation can reduce parent compounds without eliminating hazardous transformation products (2, 7). Antibiotics additionally differ in charge and mineral interactions; residue-associated community changes do not establish a universal concentration threshold for resistance selection. Zhao et al. (10) reported total concentrations of 26 antibiotic residues ranging from 83.24 to 4,237.93 µg kg^-1 in greenhouse soils, alongside microbial-community responses to residue gradients. These observational relationships require exposure refinement and do not establish causality or a transferable effect threshold.

Particulate contaminants require additional descriptors. For microplastics, polymer, particle size, shape, surface weathering, and physical incorporation into aggregates influence organism contact and soil structure; a dissolved-fraction assay cannot capture all particle-mediated effects. Experimental evidence demonstrates changes in soil properties and plant performance that vary among plastic treatments (11). These examples are illustrative rather than exhaustive, and do not justify applying one extraction test across metals, organic chemicals, and particles. Figure 1 locates these exposure processes within the proposed adaptive assessment cycle.

3 Ecotoxicological responses must be connected to ecosystem function

Biological responses are relevant at every level of organization. Their interpretation depends on the protected endpoint: a molecular stress response, altered community composition, and impaired nutrient cycling are related but distinct observations. Biomarkers can provide early warning without demonstrating population decline, and community turnover may occur while some process rates remain stable. Conversely, unchanged richness can conceal loss of sensitive taxa or specialized functions. Ecological assessment should combine suitable organism tests with direct process measurements rather than require ecosystem damage before recognizing a biological warning (5).

Functional endpoints can include decomposition, nitrogen transformation, soil respiration, aggregate stability, and plant productivity. Their interpretation requires matched reference soils and adjustment for moisture, pH, substrate supply, and land management. Increased respiration after adding compost may reflect the new carbon substrate rather than recovery; reduced respiration can reflect either stress or altered substrate availability. Shu et al. (12) demonstrated that soil pH moderated resistance and resilience of carbon- and nitrogen-cycling functions under stress. We propose measuring recovery relative to a defined reference range, with repeated observations after relevant seasonal changes or disturbances.

Co-occurring metals, metalloids, organic chemicals, and particles may alter one another’s exposure or effects. Consequently, single-substance screening should be complemented by tests of the relevant contaminated soil or mixture where feasible. Neither additive effects nor synergy should be assumed without an appropriate model or evidence. Table 1 distinguishes conventional screening metrics from the additional exposure and function evidence needed for this interpretation.

Table 1. Conventional screening metrics and complementary exposure and function evidence.

4 Predictive technologies and remediation should be tied to decision endpoints

Targeted chemistry can be extended with suspect screening, while eDNA and omics can characterize biological composition and functional potential. However, gene presence does not establish current process activity. Spatial models can priorities sampling but require independent validation and explicit uncertainty; SoilGrids 2.0 illustrates uncertainty-aware prediction of soil properties, not direct validation of contaminant-risk predictions (16). These tools should answer a defined exposure or management question and supplement field measurements.

Remediation options include source removal or containment, soil washing, thermal treatment, chemical immobilization or transformation, microbial degradation, plant-assisted remediation, and mycorrhizal support, among others. Selection depends on contaminant chemistry, exposure route, land use, treatment depth, and feasible management of residuals (1). Organic amendments include compost, appropriately characterized manure-derived materials, and biochar; mineral amendments include lime, clays, and reactive oxides. These materials are not interchangeable, and their own contaminant content and effects on salinity, nutrients, and DOM require checking. In a pot experiment, modified biochar reduced metal uptake by ryegrass, but this does not establish long-term field safety (17).

Soil washing separates contaminants into fractions or treatment liquids that still require management. Thermal treatment can remove or transform suitable contaminants while changing soil organic matter and other properties (18). Immobilization leaves the contaminant inventory in place, so lower extractability must be checked against leaching, uptake, and changing soil conditions. Biodegradation requires confirmation of transformation products and residual effects. Across technologies, the proposed evaluation is the same: acceptable residual exposure, recovery of target functions, and no unacceptable transfer to water, dust, crops, or other receptors.

5 Discussion and integration with established risk assessment

Geochemical screening and exposure-based risk assessment serve different purposes. The geoaccumulation index compares a concentration with background after an adjustment factor; EF normalizes to a reference element to assess relative enrichment. Both depend strongly on appropriate background and reference choices and neither measures toxicity directly (13). The potential ecological risk index combines contamination factors with toxic-response weights; its original sediment-based formulation should not be transferred to soils or new contaminant classes without justification (14). Table 1 summarizes their roles alongside hazard quotients.

For non-cancer human-health screening, an oral HQ is estimated exposure dose divided by the relevant reference dose; an ecological HQ commonly compares exposure with an appropriate ecological benchmark. Values above one identify potential concern, not a probability or certainty of harm, and values below one do not resolve unassessed pathways or substances (5, 15). Exposure and benchmark must use compatible routes, durations, and concentration bases. Bioaccessibility must not be inserted as an arbitrary multiplier: any adjustment requires receptor-specific validation and applicable assessment guidance (6). Cancer risks and particulate or biological hazards require their own suitable methods.

The proposed contribution is an explicit link between these screening results, receptor-specific exposure, and measured functional condition. High enrichment with low measured exposure supports further evaluation of stability and future land use rather than automatic declarations of safety. Conversely, a low index with impaired function requires examination of unmeasured contaminants and nonchemical stressors. Evidence should be integrated transparently rather than compressed into an unvalidated composite score. Laboratory-to-field extrapolation, confounding in observational studies, assay comparability, mixture uncertainty, and resource constraints remain limitations. Multi-site comparisons are needed to test whether the proposed additions improve prediction and management decisions.

6 Practical implications and an ordered assessment process

Step 1 define the problem and protection goals

Document sources, land-use history, migration pathways, and current and plausible future receptors. Specify the soil functions to protect, such as nitrogen transformation or vegetation establishment, and the separate human-health pathways requiring assessment. Agree on decision criteria and control continuing inputs where feasible.

Step 2 establish the baseline and reference conditions

Design spatially representative sampling with suitable depths, seasons, matched reference soils, and quality controls. Measure priority contaminant concentrations, relevant transformation products, pH, organic carbon, texture, moisture, and redox conditions. Record analytical recovery, detection limits, dry-mass basis, and particle-size coverage where relevant.

Step 3 screen and prioritize

Compare concentrations with applicable screening values and calculate justified indices or hazard quotients from Table 1. State background values, toxicity factors, exposure assumptions, and uncertainty. Prioritize substances and pathways for refinement; retain statutory requirements and do not interpret screening exceedance as confirmed ecological damage.

Step 4 refine receptor-specific exposure

Choose measurements suited to the contaminant and pathway: pore water or passive sampling, speciation, validated oral-bioaccessibility assays, uptake measurements, or particle characterization. Report assay conditions and limitations. Recalculate exposure estimates only when the refinement is compatible with the toxicity benchmark and recognized assessment method.

Step 5 measure biological effects and soil functions

Pair relevant organism or community endpoints with direct measurements of the selected functions. Include untreated contaminated soil, matched reference soil, and treatment controls where feasible. Distinguish treatment-induced stimulation from recovery, and test whether effects track measured exposure after accounting for soil-state differences.

Step 6 characterize risk and uncertainty

Integrate chemical, exposure, biological, and functional evidence for each protection goal. Explain conflicting findings and identify missing pathways, mixture uncertainties, and alternative causes of impairment. Decide whether evidence supports acceptable conditions, further investigation, or intervention; absence of data is not evidence of negligible risk.

Step 7 select and evaluate intervention

Compare feasible technologies against exposure reduction, functional recovery, durability, secondary wastes, and cross-media transfer. Pilot the preferred option where necessary and set measurable post-treatment targets relative to baseline and reference conditions. Maintain containment or land-use controls when residual contamination precludes unrestricted use.

Step 8 verify and adapt

Repeat exposure and function measurements across relevant seasons and after plausible disturbances, including flooding or drying where applicable. Confirm compliance and acceptable residual risks alongside recovery toward the agreed functional range. Rebound, persistent impairment, or new transfer pathways trigger renewed investigation and management; monitoring frequency follows site-specific risk.

7 Conclusion

Widespread contaminant occurrence identifies where investigation is needed; it does not reveal absorbed dose, ecotoxicological effects, loss of soil function, or human-health risk. Biodiversity associations add biological evidence but do not remove the need to assess exposure, confounding, and protected endpoints. Function-centered assessment therefore complements established screening and risk characterization with receptor-specific measurements, direct soil-process endpoints, and verification over time. The proposed workflow supports monitoring, risk assessment, and remediation evaluation without assuming that a reduction in contaminant concentration alone establishes ecological recovery.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

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.

Footnotes

Edited by: Humberto Aponte, Universidad de O’Higgins, Chile

Reviewed by: Sajjad Ahmad, Technological University Dublin, Ireland

Footnotes

DOM, dissolved organic matter; eDNA, environmental DNA; EF, enrichment factor; HQ, hazard quotient; Igeo, geoaccumulation index; RI, potential ecological risk index.