A bibliometric evaluation was performed using VOSviewer (v. 1.6.20) based on extracted keywords. The dataset was derived from Scopus metadata exported as a Tables–delimited text file and processed to map co-occurrence relationships among author and indexed keywords.

The dataset was obtained from Scopus using the following query string:

TITLE-ABS-KEY (“microbiota” OR “bacteria” OR “microorganism*”) AND TITLE-ABS-KEY (“heavy metals” OR cadmium OR lead OR chromium OR nickel OR mercury OR arsenic OR copper OR zinc) AND PUBYEAR > 2,000 for a first search, and then restricted to 2022–2025.

A threshold of minimum 50 occurrences per term was applied to ensure statistical robustness. The software generated a network visualization where node size corresponds to keyword frequency, link thickness to co-occurrence strength, and color coding to cluster affiliation.

This systematic review was conducted according to the PRISMA 2020 guidelines (Page et al., 2021) to ensure transparency, reproducibility, and methodological rigor. The protocol was defined a priori and included a description of the research question, databases, search strategy, inclusion/exclusion criteria, data-extraction procedures, and synthesis methods; the checklist is available in the Supplementary Table 1. The primary aim was to identify, analyze, and compare recent studies (2022–2025) addressing bacterial bioremediation of toxic heavy metals across environmental matrices. The central research questions were:

Three major databases, Scopus, Web of Science (WoS), and PubMed, were systematically queried to identify peer-reviewed research articles and reviews published between January 2022 and June 2025. Searches were limited to English-language publications and subject categories related to environmental science, biotechnology, and microbiology. Equivalent Boolean search strings were adapted for each database to ensure semantic and structural consistency. Minor syntactic differences were due to database-specific query syntax.

Before defining the final search strings, several preliminary and more permissive query combinations were tested across the three databases. These exploratory strings included broader taxonomic and functional terms (e.g., microorganism, biodegradation, pollutant removal, metalloid detoxification) to evaluate retrieval coverage and background noise. After iterative refinement, the most specific and reproducible queries were selected and applied in the final systematic search.

The final search strings, which generated the dataset used in this review (6^th text string), is reported in Supplementary Table 2. The last search was performed in September 2025. Reference lists of retrieved papers were manually screened to capture additional relevant studies not indexed under the same keywords.

Studies were considered eligible for inclusion if they met all the following criteria.

First, only original experimental research published in peer-reviewed international journals from January 2022 onward and written in English was included. Furthermore, the following relevant parameter were used as including markers:

Studies were excluded if they met one or more of the following conditions:

A detailed summary of all inclusion and exclusion criteria applied in this review is provided in Supplementary Table 3.

All records retrieved from Scopus, Web of Science, and PubMed were exported to Microsoft Excel for centralized management. Duplicate entries were identified and removed through automated matching of DOI, title, and first author fields, followed by manual verification.

Title and abstract screening was independently performed by two reviewers according to pre-established eligibility criteria. Each record was classified as Include, Maybe, or Exclude. Disagreements were resolved through discussion until a consensus was reached. Studies marked as Maybe were retained for full-text evaluation. The entire screening workflow and decision log were managed in a structured Excel matrix to ensure transparency and reproducibility.

Inclusion criteria required that (i) bioremediation represented the primary focus of the study, (ii) the biological system consisted of bacterial isolates or microbial consortia, (iii) at least one of the target metals was Cd, Pb, Cr, Ni, Cu, Zn, Hg, or As, (iv) quantitative removal data were reported (e.g., initial and residual concentrations, percentage removal, adsorption capacity, or kinetic parameters), (v) robust analytical methods were applied (ICP-MS, AAS, ICP-OES, SEM-EDS, FTIR, or XRD), and (vi) taxonomic identification was provided at least to the 16S rRNA gene level.

Studies were excluded if they focused exclusively on phytoremediation or non-metal pollutants (e.g., dyes, hydrocarbons, pesticides), employed aquatic or wastewater systems without soil relevance, lacked quantitative outcomes, failed to identify microorganisms, or represented reviews, commentaries, or duplicated datasets.

All records (188) retrieved from Scopus (n = 98), Web of Science (n = 86), and PubMed (n = 4) were imported into Microsoft Excel for centralized management and deduplication. Duplicate entries were identified through automated DOI matching and manually verified; 28 duplicates were removed, resulting in 160 unique studies eligible for screening.

Title, abstract, and keyword screening was independently performed by two reviewers according to the inclusion and exclusion criteria reported in Supplementary Table 3. After this phase, 68 studies were considered potentially relevant and retrieved for full-text assessment. Following detailed evaluation, 41 studies met all eligibility requirements and were included in the systematic review. The main reasons for exclusion were: (i) absence of quantitative data on heavy-metal removal; (ii) focus on phytoremediation or non-metal pollutants; (iii) lack of taxonomic identification; and (iv) studies not performed on soil or soil-related matrices. The complete identification, screening, and inclusion process is summarized in Figure 1, which follows the PRISMA 2020 flow-diagram structure. The corresponding PRISMA 2020 checklist is provided in the Supplementary materials.

For each included article, a structured data-extraction form was completed to capture, when applicable:

Data were independently extracted by two reviewers and cross-checked for consistency. Discrepancies were resolved through consensus.

The methodological quality of the included studies was evaluated using a modified Newcastle–Ottawa Scale (NOS) adapted for environmental and microbiological research. Each study was assessed on the following domains, as reported in Table 1.

A maximum score of 10 indicated high methodological quality. Studies scoring ≥7 were retained for full synthesis; those below were discussed qualitatively. To strengthen the methodological robustness of the synthesis, the quality assessment based on NOS was also analyzed according to bacterial genus and target metal. For each genus–metal combination represented by at least three studies, mean NOS scores, standard deviations, and score ranges were calculated. The results are in the Supplementary Tables 4, 5.

A mixed-methods synthesis approach was applied:

Statistical summaries (mean ± SD, range, median) and visualizations (boxplots, heatmaps) were generated in Statistica v.14 (Statsoft, Tulsa, OK).

Results were organized thematically into four domains:

The bibliometric co-occurrence analysis conducted via VOSviewer revealed a structured network of research themes within the field of microbial heavy-metal bioremediation (Figure 2). Four major clusters emerged: (1) “bioremediation/heavy metals/removal” (red), characterized by high frequencies of nodes such as “bioremediation,” “adsorption,” “chromium,” “biosorption”; (2) “soil pollution/microbial community” (green), containing keywords like “soil,” “microbial community,” “plant growth,” “soil pollutants”; (3) “genetics/metabolism/bacterial taxonomy” (yellow), including “metabolism,” “genetics,” “Pseudomonas,” “metal tolerance”; and (4) “environmental engineering/water/sediment” (blue), encompassing “water pollution,” “sediment,” “ecosystem restoration.” The size of nodes (frequency of keyword occurrences) and thickness of links (co-occurrence strength) indicate that the red and green clusters are the dominant thematic domains, with “bioremediation” and “soil pollution” serving as central hubs. The dense interconnections between clusters highlight that research on microbial remediation of heavy metals is inherently multidisciplinary, bridging microbial physiology, soil science, pollutant removal technologies and ecosystem restoration. Notably, keywords associated with the genus Pseudomonas and “metal tolerance” are centrally located within the yellow cluster, thus supporting the decision to treat Pseudomonas as a benchmark genus in this review. The complete list of keywords ranked by frequency and link strength is provided in Supplementary Table 6.

The 41 studies reveal a consistent convergence toward six dominant functional pathways (Figure 3) governing bacterial heavy-metal detoxification: EPS-mediated biosorption, biofilm-driven community resilience, enzymatic and redox transformations, biomineralization and MICP, and genetic/efflux-based regulation. These mechanisms frequently co-occur within the same strain or consortium, underscoring the multi-layered adaptability of environmental bacteria under metal stress (Manikandan and Nair, 2023; Yan et al., 2024).

Extracellular polymeric substances-mediated biosorption represented the most widespread process, documented in 65.9% of the studies. EPS secreted by Bacillus, Enterobacter, and other species provided abundant binding sites capable of complexing Pb²⁺, Cd²⁺, and Ni²⁺ ions (Chen et al., 2025; Mondal et al., 2025; Wang et al., 2025). FTIR and SEM-EDS analyses consistently identified these functional groups (such as carboxyl, amino, and phosphate groups) and confirmed their participation in surface complexation (Yasmin et al., 2022; Yan et al., 2024; Mitra et al., 2025). The strong correlation between EPS density and removal efficiency was associated with the high performance of Bacillus in direct biosorption assays (with efficiencies up to 99.8% for Pb) (Shaaban Emara et al., 2023; Bhandari et al., 2024).

Biofilm formation and community-level resilience appeared in roughly 41.5% of the dataset, mainly among Pseudomonas, Bacillus, and mixed consortia, specifically noted in studies on fungal-bacterial and B. subtilis biofilms (Henagamage et al., 2022; Ghosh et al., 2024). In rhizospheric systems, microbial consortia stabilized metal ions at the root interface, reducing plant uptake by 50%–80% (Zhang et al., 2023; Gao et al., 2024). The co-occurrence of EPS production and biofilm formation was frequently reported in studies describing the persistence of microbial populations under chronic metal exposure.

Genetic and efflux-based mechanisms (31.7%) were confirmed through the detection of specific resistance genes (Table 2). The most clearly identified mechanisms in the dataset relate to chromium (Cr) and arsenic (As). For instance, in P. putida, the chromate reductase (ChrR) gene was explicitly isolated and transferred to E. coli to confirm its responsibility for Cr(VI) reduction (Sundarraj et al., 2023). Further genomic analysis on a Pseudomonas/Rhodococcus consortium (Su et al., 2022) detailed the Cr(VI) tolerance mechanism, showing it was mediated by the efflux of chromate via the chrA and copZ genes, supplemented by DNA-repaired proteases. For arsenic detoxification, increased expression of the arsenite oxidase enzyme was observed in Pseudomonas spp. AK9 and was associated with higher transformation of As(III) (Satyapal et al., 2024).

Redox and enzymatic transformations were identified, respectively in 34.1% and 2.8% of the papers, particularly in Pseudomonas and Bacillus isolates (Ali et al., 2022; Sundarraj et al., 2023). Cr(VI) reduction to Cr(III) was the most frequently observed process, mediated by reductase enzymes that generate insoluble Cr(OH)3 or mixed chromite phases (Monga et al., 2022; Ye et al., 2024). Additional redox reactions included transformations between arsenic species (Ghosh et al., 2024; Satyapal et al., 2024). Enzymatic activities such as dehydrogenase, peroxidase, and superoxide dismutase were reported in 25% of studies, suggesting a coupling effect between biochemical and physicochemical detoxification (Yan et al., 2024; Wang et al., 2025).

The systematic assessment of 41 studies published between 2022 and 2025 revealed a pronounced taxonomic convergence in heavy-metal bioremediation, with members of Bacillus, Pseudomonas, Enterobacter, and Lysinibacillus dominating across environmental matrices. These genera consistently exhibited high removal efficiencies for cadmium, lead, and chromium, which aligns with the robust structural and physiological traits characteristic of Gram-positive and metabolically versatile Gram-negative bacteria.

Bacillus species showed exceptional biosorption capacities, often exceeding 100 mg g^–1 for Pb²⁺ and Cd²⁺ (Ye et al., 2024; Wang et al., 2025), attributable to their thick peptidoglycan layers, teichoic acids, and ureolytic biomineralization potential. Conversely, Pseudomonas strains, particularly P. aeruginosa, P. putida, and P. fluorescens, displayed remarkable multi-metal tolerance, frequently sustaining growth under 500–700 mg L^–1 of Cd or Cr while maintaining removal rates above 70%. These outcomes are consistent with a broad detoxification spectrum in Pseudomonas, particularly in terms of mechanistic diversity and multi-metal tolerance, rather than uniformly higher removal efficiencies across all contexts. The predominance of these taxa corroborates prior evidence indicating that bacterial consortia combining Bacillus, as high-affinity sorbents, and Pseudomonas, as redox transformers and EPS producers, outperform single strains in multi-metal matrices (Peng et al., 2023; Cai et al., 2025).

Mixed consortia exhibited synergistic mechanisms in which Bacillus contributed carbonate precipitation and surface complexation, whereas Pseudomonas enhanced redox transformations through periplasmic reductases and metal-chelating siderophores. The co-occurrence of these functions yields a sequential detoxification process, whereby soluble ions are first immobilized extracellularly and subsequently converted to less mobile species within biofilms. This synergy underlines the ecological advantage of cooperative systems in natural and engineered environments. Three dominant pathways emerged: (i) passive biosorption through carboxyl, hydroxyl, and phosphoryl moieties within the cell wall; (ii) bioaccumulation and redox transformation mediated by enzymatic systems such as chromate reductases and arsenate reductases; and (iii) biomineralization processes involving carbonate and phosphate precipitation.

Studies employing advanced spectroscopy (FTIR, XRD, SEM–EDS) confirmed the formation of CdCO₃, Pb₃(PO₄)₂, and Cr(OH)₃ as common end-products, providing molecular validation for immobilization mechanisms (Kaur et al., 2024; Zhang et al., 2024). Together, these findings support the interpretation that microbial metal removal is not a single process but a sequence of coupled reactions integrating physicochemical adsorption with biologically driven transformations.

Bioremediation performance was highly dependent on environmental conditions, particularly pH, temperature, and substrate composition. Optimal pH values clustered between 6.5 and 8.0, where deprotonation of cell-wall ligands enhances cation binding and carbonate precipitation. Acidic conditions (pH < 5) generally reduced sorption efficiency by promoting metal solubility and proton competition, although acid-tolerant Pseudomonas and Rhodococcus bacteria maintained activity in soils contaminated with mixed metals. Temperature affected both adsorption kinetics and metabolic rates, with optimal ranges between 28 °C and 35 °C across most isolates. Extreme thermophiles or psychrotolerant strains were rare, confirming that mesophilic systems dominate soil remediation under temperate climates.

Matrix effects were also significant; in fact, synthetic solutions frequently yielded higher removal percentages than soil or sludge matrices due to reduced ionic competition. However, field-relevant matrices better captured the complexity of environmental constraints and revealed a systematic reduction in remediation performance that is not evident in laboratory batch systems. In particular, studies employing real contaminated soils demonstrated that bacterial immobilization efficiency decreases by 10%–25% relative to aqueous systems, highlighting a substantial laboratory-to-field performance gap primarily due to metal–clay interactions and organic matter interference (Lin et al., 2022; Zhang et al., 2024). Nonetheless, the persistence of biofilm-forming strains may partially mitigate these losses by maintaining localized microenvironments conducive to sorption and precipitation, although it does not fully offset the observed laboratory-to-field performance gap.

Nutrient availability and carbon sources further modulated bioremediation kinetics. Simple sugars and organic acids facilitated EPS synthesis, enhancing sorption sites and metal chelation. The inclusion of urea or ammonium salts in media triggered MICP, increasing immobilization through CaCO₃ lattice formation. Such findings suggest that bioremediation efficiency arises from a balance between microbial physiology and physicochemical context, where even minor shifts in pH or nutrient status can redefine metal speciation and bioavailability.

While Bacillus species remain among the most efficient biosorbents due to their robust cell walls and sporulation capacity, recent evidence highlights the broader ecological and mechanistic versatility of Pseudomonas spp. as model organisms and reference systems for heavy-metal bioremediation. However, the application of Pseudomonas in field-scale remediation is constrained by several operational limitations. As non-spore-forming bacteria, Pseudomonas strains generally display reduced long-term persistence under desiccation, UV exposure, nutrient limitation, and fluctuating redox conditions when compared with spore-forming genera such as Bacillus. These constraints often necessitate repeated inoculation, immobilization on carriers (e.g., biochar, alginate, mineral matrices), or integration into biofilm-supported systems to ensure sustained activity in soil environments. Comparative analyses indicate that Pseudomonas can operate effectively across a wide spectrum of environmental conditions, including acidic and multi-metal-contaminated soils, where Bacillus may exhibit reduced metabolic activity, depending on strain-specific and matrix-related factors. Genomically, Pseudomonas is shown in the dataset to harbor a well-characterized set of resistance operons, such as chromate efflux systems (e.g., chrA) and specific reductases (Su et al., 2022; Sundarraj et al., 2023), along with stress-response regulators under quorum-sensing control (Lu et al., 2022), which may confer rapid adaptability and multi-metal tolerance. From an applicative standpoint, non-spore-forming Pseudomonas strains develop resilient biofilms capable of maintaining metabolic activity under fluctuating redox and moisture conditions, making them suitable for continuous bioreactors and biosensor interfaces. Moreover, the genus is one of the most genomically annotated among soil bacteria, providing a reproducible framework for cross-study comparison and systems-biology modeling. Overall, these observations indicate that removal efficiencies reported under controlled laboratory conditions tend to overestimate field performance, underscoring the need for cautious extrapolation and for validation under realistic soil and environmental conditions.

Within the analyzed dataset, Pseudomonas spp. consistently represented the most functionally diverse genus. Crucially, it also stands out as one of the most mechanistically and genetically well-characterized genera in the context of these studies, providing a clearer and more reproducible framework for its application compared to many other taxa.

Genome-level investigations and specific gene analyses, which were reported for Pseudomonas more frequently than for other genera, have begun to elucidate its extensive repertoire of resistance determinants. For example, specific genomic analyses in the dataset linked Cr(VI) tolerance directly to the function of efflux genes chrA and copZ (Su et al., 2022). This was further supported by studies isolating the specific chromate reductase (ChrR) gene from P. putida, supporting its role in Cr(VI) reduction (Sundarraj et al., 2023). Similarly, arsenic detoxification was explicitly correlated with the “increased expression of arsenite oxidase” in Pseudomonas sp. (Satyapal et al., 2024).

This depth of available genetic information—spanning efflux systems, enzymatic reduction, and stress-response enzymes—is complemented by its proven metabolic versatility. Mechanistically, Pseudomonas exhibits a dual detoxification strategy that combines:

While other genera, particularly Bacillus, often demonstrate exceptionally high removal efficiencies through passive biosorption, biomineralization, and long-term environmental persistence, Pseudomonas spp. are better interpreted as a mechanistic benchmark rather than a universally superior remediation agent. The value of Pseudomonas lies primarily in the depth of available genetic, transcriptomic, and biochemical characterization, which enables detailed dissection of efflux systems, redox enzymes, and regulatory networks that are less frequently resolved in other taxa.

The combination of its broad metabolic capabilities (including redox transformations often lacking in other biosorbents), its detailed genomic annotation, and its robust stress-response mechanisms makes it a particularly well suited and adaptable candidate for bioaugmentation, depending on the environmental context and remediation objectives.

The evidence underscores that microbial bioremediation is a multi-layered process integrating community ecology, metal chemistry, and environmental dynamics. In terrestrial matrices, microbial assemblages restructure in response to chronic metal exposure, selecting for resistant phenotypes with enhanced EPS and biofilm production. This adaptive remodeling not only supports metal immobilization but also contributes to soil stabilization and nutrient cycling. The co-occurrence of plant growth promoting traits, such as phosphate solubilization, indole-acetic-acid synthesis, and siderophore production, demonstrates that many remediation-competent strains simultaneously enhance soil fertility, bridging the gap between bioremediation and agroecological restoration (Gao et al., 2024; Ghosh et al., 2024). From an applied perspective, the convergence of adsorption, biomineralization, and redox conversion mechanisms across different taxa suggests that integrated consortia or sequential treatment systems can achieve superior outcomes in heterogeneous contaminated soils. Field-scale trials incorporating Pseudomonas–Bacillus co-inocula or biochar-supported bacterial systems have demonstrated stable performance under fluctuating pH and moisture. However, ensuring ecological compatibility and long-term persistence remains essential to avoid unintended perturbations of indigenous microbiota. At the same time, the environmental sustainability of microbial remediation extends beyond removal efficiency. The re-establishment of soil microbial diversity and enzymatic functionality is crucial to restoring ecological services after decontamination. The capacity of Pseudomonas to reinstate biogeochemical cycling reinforces the dual role of these bacteria as both detoxifiers and ecological engineers.

Although the present synthesis provides robust evidence on the efficiency and mechanisms of bacterial heavy-metal remediation, several methodological limitations persist. The heterogeneity of experimental conditions, particularly regarding initial metal concentrations, exposure times, and analytical protocols, still hampers quantitative comparability among studies. Many works rely on batch experiments with synthetic solutions, which tend to overestimate removal rates compared with complex soil matrices. Moreover, the limited number of field-scale investigations constrains the extrapolation of laboratory findings to real environments.

Despite the rigorous application of PRISMA inclusion criteria and NOS quality assessment, several potential sources of bias remain inherent to the analyzed literature. These include selection bias arising from the restriction to English-language publications and recent years (2022–2025), experimental and analytical heterogeneity related to differences in matrix type, initial metal concentrations, exposure duration, and analytical techniques (AAS, ICP-OES, ICP-MS), as well as matrix bias associated with the frequent use of synthetic batch systems that tend to overestimate removal efficiencies relative to field conditions. Additional reporting bias may result from the inconsistent disclosure of replicate numbers, standard deviations, or negative outcomes, while interpretation bias can occur when the presence of resistance genes (e.g., chrA, copZ, arsC) is equated with functional detoxification in the absence of enzymatic or transcriptional validation.

A synopsis of the potential sources of bias is in the Supplementary Table 7, and according to authors’ point of view there are at least five different sources/kinds of bias (matrix, performance, analytical, reporting, and mechanistic), each of them characterized by low, medium, and high levels of risk.

Addressing these limitations will require a shift toward integrated, functionally validated research frameworks. Future studies should prioritize whole-genome and transcriptomic approaches to resolve regulatory networks underlying metal resistance and to support predictive modeling of microbial community dynamics. In parallel, hybrid biotechnological systems combining microbial consortia with immobilization strategies (e.g., biochar, alginate carriers, or plant-assisted platforms) should be further explored to enhance stability, persistence, and ecological compatibility under field conditions. Emerging tools, including AI-assisted strain selection and CRISPR-based gene editing, offer promising avenues to validate and fine-tune key efflux and redox pathways, improving both specificity and kinetics of metal transformation. Long-term, field-based monitoring integrated with digital environmental sensing will be essential to assess persistence, horizontal gene transfer, and ecosystem resilience, thereby enabling the transition from laboratory-scale performance to scalable, data-driven, and environmentally sustainable bioremediation systems.