A Systematic Review on Plant-Atmosphere Synergy: Dual Purification Strategies for PM_2.5 and O₃ Pollution

Abstract

Globally, the combined pollution of fine particulate matter (PM_2.5) and ground-level ozone (O₃) poses severe challenges to public health and sustainable urban development. Recent data indicate that the annual average PM_2.5 concentration in the vast majority of cities worldwide fails to meet World Health Organization safety standards, with air pollution causing millions of premature deaths annually. As a nature-based solution, the purification efficacy of vegetation remains poorly quantified due to unclear coupling mechanisms with local meteorological conditions. This study systematically reviewed and synthesized 229 empirical studies published between 2000 and 2025 from Web of Science and China National Knowledge Infrastructure (CNKI), aiming to clarify the quantitative relationships and regulatory mechanisms of plant–meteorological synergistic purification of PM_2.5–O₃. Following double-blind independent screening (κ = 0.85) and data extraction, a quantitative minimal feasible synthesis approach was adopted due to high data heterogeneity. The results indicated the following. (1) The median canopy purification efficiency of urban vegetation for PM_2.5 was 18.2% (IQR: 12.5–30.1%, n = 17), with a median dry deposition velocity (Vd–PM) of 0.05 cm s⁻¹ (0.02–30 cm s⁻¹, n = 15). The median dry deposition velocity (Vd–O₃) for O₃ was 0.55 cm s⁻¹ (0.12–1.82 cm s⁻¹, n = 8), with non-stomatal deposition contributing approximately 35%. (2) Meteorological factors exhibit nonlinear regulation: relative humidity (RH) > 70% significantly enhances PM_2.5 adsorption, wind speeds of 1.5–3.0 m s⁻¹ are optimal for PM_2.5 deposition, and temperatures > 30 °C generally inhibit plant uptake of both pollutants (n = 7). (3) Functional traits strongly correlate with purification efficacy: species with high leaf roughness (R² = 0.8), high stomatal conductance, and low BVOC emissions (e.g., Ginkgo biloba, Platycladus orientalis) exhibit optimal synergistic purification potential. Species with high BVOC emissions (Populus przewalskii, Eucalyptus robusta) can increase daily net O₃ pollution equivalents by up to 86 g and must be strictly avoided. Based on quantitative evidence, a green space planning decision matrix indexed by climate zone and pollution type was developed, specifying vegetation configuration patterns, functional group selection, and key design parameters (canopy closure, green belt width, etc.) for different scenarios. This study provides an actionable scientific basis for precision planning and climate-adaptive management of urban green infrastructure.

1. Introduction

The urbanization process, while driving socioeconomic development, has also led to regional atmospheric composite pollution characterized by fine particulate matter (PM_2.5) and ground-level ozone (O₃) [1,2,3]. As typical secondary pollutants, both are regulated by common precursors (NO_X, VOCs) and meteorological conditions [4,5,6]. Frequent synergistic pollution events pose serious threats to public health and ecosystems [7,8,9]. In heavily polluted regions like eastern China, synchronous exceedance of PM_2.5 and O₃ concentrations during summer synergistic pollution events occurs at rates exceeding 30% [10,11], posing severe threats to public health and ecosystems. The Global Burden of Disease (GBD) study indicates that in 2019 alone, combined exposure to environmental PM_2.5 and O₃ contributed to approximately 4.5 million premature deaths [12], while also causing yield reductions of 5–15% in major crops such as wheat and corn [13]. Increasing urban vegetation represents a key nature-based solution for mitigating air pollution [14,15,16]. Vegetation directly removes atmospheric pollutants through physical interception (e.g., leaf adsorption, canopy trapping) and physiological processes (e.g., stomatal uptake) [17,18,19]. Numerous studies have examined the effects of plant functional traits (e.g., leaf micromorphology, stomatal conductance) and community structure on purification efficiency for individual pollutants [17,18,19,20,21]. However, purification efficacy is regulated by key meteorological factors, including temperature, humidity, wind speed, and solar radiation [20,21,22]. Meteorological conditions influence pollution levels by controlling boundary layer structure, chemical reaction rates, and pollutant diffusion capacity [23,24,25]. Simultaneously, they directly regulate plant physiological activities (e.g., stomatal opening/closing, metabolic intensity), thereby affecting purification functions [26,27]. This complex joining among “plants–weather–pollutants” presents a scientific challenge requiring systematic analysis.

Current research faces three critical knowledge gaps that limit its precise application in urban composite pollution management. First, existing evidence remains fragmented, with most studies focusing on single pollutants or meteorological factors, lacking systematic analysis of the “plant–weather–composite pollution” system. Second, quantification standards lack uniformity. Core metrics such as dry deposition velocity (Vd) and concentration reduction rates are inconsistently defined, with uncertainties regarding thresholds and “optimal ranges” remaining unresolved, making cross-study comparisons challenging. Third, practical translation is insufficient. Existing conclusions rarely target specific urban/suburban scenarios, failing to develop actionable planning tools adaptable to diverse climatic zones and pollution types. To clarify this review’s quantitative analytical framework, its core metrics are defined as follows. Vegetation’s “purification efficiency” denotes the relative percentage reduction in pollutant concentration, serving as an endpoint metric for assessing net environmental benefits. “Dry deposition velocity (Vd)” characterizes the process rate parameter for pollutant migration to the surface (cm s⁻¹), while “flux” (e.g., stomatal ozone absorption flux PODᵧ) denotes the absolute exchange rate per unit time and area. Distinguishing between the aggregate outcome of “concentration change” and process–mechanism parameters like “deposition/flux” is fundamental for systematic analysis and comparison. To address these gaps, this study centered on urban sustainability within an urban/suburban context. Through a systematic review integrating 229 global empirical studies, it aimed to: (1) establish a unified quantitative indicator system defining core parameters (Vd, PODᵧ, etc.) and meteorological control thresholds for plant purification of PM_2.5 and O₃; (2) decode synergistic purification mechanisms linking vegetation functional traits, canopy structure, and meteorological conditions; and (3) develop a green space planning decision framework indexed by climate zones and management objectives, providing scientific support for precision planning and climate-resilient management of urban green infrastructure.

2. Methodology

This study was conducted in strict adherence to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA 2020) guidelines (Supplementary Materials), ensuring transparency, reproducibility, and completeness of reporting throughout the process.

2.1. Literature Search Strategy

To ensure the comprehensiveness and reproducibility of the literature search, we conducted a systematic search on the two Chinese and English electronic databases of the Web of Science Core Collection and China National Knowledge Infrastructure (CNKI). The search timeframe was set from 1 January 2000 to 20 October 2025. This timeframe was selected based on the following considerations. First, PM_2.5 emerged as a key fine particulate matter indicator with widely recognized health effects internationally from the late 20th to early 21st century, with related research proliferating since 2000. Second, studies on the composite pollution mechanisms and synergistic purification of PM_2.5 and O₃ at the urban scale have become a prominent frontier field over the last two decades, driven by rapid urbanization. Finally, the selected primary databases achieved near-complete coverage of the core literature in this field after 2000, ensuring comprehensive retrieval. The search strategy constructed queries based on four core concepts: pollutants (PM_2.5, O₃), purification agents (plants, vegetation), processes (removal, deposition, absorption), and regulatory factors (meteorology, climate). To precisely balance recall and precision, Boolean operators (AND, OR) were used to combine concepts. The specific strategy involved linking synonyms within the same concept group with “OR,” then connecting different concept groups with “AND” to ensure search results encompassed all core elements simultaneously. Queries were adjusted according to each database’s syntax rules. Using Web of Science as an example, the core query was as follows:

(TS = ((“PM2.5” OR “fine particulate matter” OR “particle matter 2.5”) AND (“O3” OR “tropospheric ozone”) AND (“plant*” OR “vegetation” OR “urban forest*” OR “green space*” OR “tree*” OR “shrub*”) AND (“removal” OR “deposition” OR “uptake” OR “adsorption” OR “dry deposition” OR “purif*” OR “filter*”) AND (“meteorolog*” OR “climate” OR “temperature” OR “humidit*” OR “relative humidity” OR “wind” OR “precipitation” OR “solar radiation” OR “sunlight”)))

The search query was adapted according to the syntax rules of different databases. Additionally, a retrospective search was conducted on the reference lists of included studies to supplement potentially overlooked literature.

2.2. Inclusion and Exclusion Criteria

Inclusion and exclusion criteria form the foundation of ensuring the relevance and rigor of a systematic review. Based on the methodological framework of the PRISMA 2020 guideline for systematic reviews and focusing on the core issue of “plant–climate–composite pollution” mechanisms, we established the criteria shown in

Table 1. Inclusion and exclusion criteria (developed according to the PRISMA 2020 guideline).

Dimension	Inclusion Criteria	Exclusion Criteria
Study Subjects	Terrestrial vascular plants (trees, shrubs, herbaceous plants) and plant communities situated in urban or peri-urban environments	Studies conducted in non-urban/peri-urban settings such as farmland, natural forests, wetlands; research involving indoor foliage plants.
Intervention/Exposure	Atmospheric PM2.5 and/or O3 pollution	Studies primarily involving other atmospheric pollutants (e.g., SO2, NOₓ).
Control/Comparison	Provides comparisons of pollutant concentrations, fluxes, or removal rates with/without vegetation, across different vegetation types, or under varying meteorological conditions.	Studies lacking a control group.
Outcome Measures	Must report quantifiable indicators of phytopurification effects (e.g., dry deposition velocity Vd, purification efficiency, adsorption per unit leaf area, ozone deposition flux PODᵧ), with quantitative analysis of at least one meteorological factor (temperature, humidity, wind speed, etc.) regulating these effects (statistical measures required, e.g., correlation coefficients, regression coefficients).	Studies that only qualitatively describe meteorological influences; studies that only report relationships between pollutant concentrations and meteorological factors without linking them to the phytoremediation process.
Study type	Empirical studies (observational studies, controlled experiments) and mechanistic modeling studies.	Pure commentary, opinion pieces, theoretical articles; reviews that do not provide original data.

.

2.3. Literature Screening and Data Extraction

Literature screening was conducted independently by Qinling Wang and Yutong Bai. The process is illustrated in Figure 1 (PRISMA flow diagram): (1) merging search results and removing duplicate records using EndNote software, (2) initial screening based on titles and abstracts, and (3) full-text retrieval and rescreening. Any discrepancies during screening were resolved through discussion or arbitration by a third senior researcher, Shaowei Lu. Inter-rater agreement at the initial screening stage was assessed using Cohen’s kappa coefficient, yielding κ = 0.85, indicating excellent agreement.

Information was extracted from the final included studies using a predesigned standardized data extraction form (Microsoft Excel). Extracted fields included the following.

Study identification: extracted ID (e.g., S001), first author, publication year, DOI.

Study context: country, city, climate zone (Köppen classification), environmental type (e.g., roadside, park, residential area).

Methodological characteristics: study type (observational/experimental/modeling), study duration, spatial scale.

Plant information: species, life form, community structure.

Purification metric data: pollutant type, specific metric (e.g., Vd, PODᵧ), numerical values (mean, median, range, standard deviation), units.

Meteorological correlation data: analyzed meteorological factors, their quantitative relationship with purification metrics (direction, correlation coefficient r, regression slope, p-value).

2.4. Assessment of Bias Risk and Research Quality Evaluation

To evaluate the internal validity and potential bias of included studies, we employed the following tools. For observational studies, evaluation was conducted using the modified Newcastle–Ottawa Scale (NOS) [28]. Scores were assigned across three dimensions: subject selection (4 points), comparability between groups (2 points), and outcome assessment (3 points). Studies with a total score ≥ 6 were classified as high quality. For modeling studies, core assessment elements included clear descriptions of atmospheric chemistry mechanisms, stomatal conductance models, and dry deposition parameterization schemes, reasonableness of boundary and initial condition settings, and use of independent observational data for validation (reporting statistical metrics, such as R², NMB, and NME).

The results of the quality assessment were primarily used for descriptive analysis to characterize the overall methodological rigor of the included evidence pool. Given the extreme heterogeneity of the data and the absence of clear systematic patterns in core composite measures (e.g., Vd, PODᵧ) across studies with different quality scores, we did not adjust study weights based on quality scores in the quantitative minimal feasible synthesis. When calculating the weighted median, all included studies were assigned equal weight. The primary value of this assessment lies in identifying potential sources of bias and serving as a reference for interpreting heterogeneity and limitations in the Discussion section (see Section 4.4).

2.5. Data Synthesis and Analytical Methods

Given the high heterogeneity among included studies in plant species, experimental design, measurement methods, and reported metrics, traditional meta-analysis to generate a single pooled effect estimate was infeasible. We therefore employed a strategy combining narrative synthesis with a quantitative minimal feasible synthesis [29].

Data Standardization and Classification: First, all purification metrics reported across studies were standardized to core parameters, including Vd (cm s⁻¹), PODᵧ (mmol·m⁻²), concentration reduction rate (%), and BVOC emission flux (μg·g⁻¹·h⁻¹). To enhance evidence clarity, all extracted quantitative data points were classified by source: (1) observed data—measured values from field measurements, controlled experiments, or flux observation towers; (2) modeled output data—results from atmospheric chemistry models, deposition models, or scenario simulations. This classification underpinned subsequent stratified analysis and visualization, enabling clear differentiation between evidence types.

Data Processing and Weighting Scheme: To address data non-independence and establish clear weighting rules, a two-step approach was implemented. First was aggregation at the study level. For multiple values reported under different conditions within the same study, we calculated the study median as the representative value for that parameter. During synthesis, each independent study (represented by its study median) was assigned equal weight. We report the weighted median and its interquartile range under this “equal-weight per study” approach to describe the central tendency and dispersion of the data distribution. The PM_2.5 concentration-weighted Vd–PM value mentioned in the abstract represents a specific scenario analysis. Its methodology draws from studies like S010 and is used solely when examining pollution level effects. It is not the default comprehensive weighting method employed in this paper.

Hierarchical Quantitative Synthesis and Uncertainty Characterization: Data were hierarchically synthesized by climate zone (temperate, subtropical, tropical, etc.), season (spring/summer/autumn/winter), canopy structure (multi-layered/single-layered), plant functional group (evergreen/deciduous, coniferous/broadleaf), and data source (observational/modeled). The weighted median, interquartile range (IQR), and full range for each layer were calculated to fully capture the data variability, including potential extremes. When analyzing Vd–PM, we explicitly define its synthesis range and discuss in the text that extreme high values primarily stem from specific modeling scenarios (see Section 3.2.1), thereby distinguishing contributions from different evidence types to the statistics.

Sensitivity Analysis: To assess the robustness of the integrated results, we conducted sensitivity analyses, primarily including: (1) stratification by data source—calculating statistics separately for those based solely on observational data and those incorporating modeled data; and (2) stratification by study quality—comparing integrated results from high-rated versus low-rated study subsets. The analysis indicated that the direction of core conclusions remained stable, but the inclusion of modeled data significantly affected numerical ranges (e.g., upper bounds). Relevant details are described in the corresponding sections of the results.

Evidence visualization and decision tool development: Box plots illustrate the distribution of key continuous variables, clearly distinguishing observed and modeled data points. Decision matrices and conceptual decision trees synthesize condition-dependent evidence to provide clear, actionable scientific support for green space planning across diverse urban contexts.

2.6. Data Analysis and Visualization Software

The following software and online platforms were used for data management, analysis, and visualization in this study.

Literature Management and Screening: EndNote 20.

Data Extraction and Management: Microsoft Excel 365.

Statistical Calculation and Primary Visualization: OriginPro 2024.

3. Results

3.1. Study Overview and Screening Process

Through systematic retrieval, 1382 records were initially identified, with 1267 remaining after removing duplicates. Following title/abstract screening and full-text review, 229 studies were ultimately included in the qualitative and quantitative synthesis (Figure 1). Among these, 122 studies (53.3%) primarily focused on plant purification of PM_2.5, 42 (18.3%) focused on O₃ purification, and 65 (28.4%) addressed PM_2.5–O₃ composite pollution or interactions. Among the combined pollution studies, only 16 (7.0% of the total) directly quantified the synergistic purification effects of plants on both pollutants, forming the core evidence base for subsequent synergistic/trade-off analyses. Analysis of publication years revealed a clear temporal evolution trend in this field Research began emerging in the early 21st century and entered a rapid growth phase after 2015. The period from 2019 to 2023 marked a sustained high-output phase, with annual publications consistently exceeding 15 articles, peaking in 2023 (n = 37). Although data for 2024–2025 may be incomplete due to publication lags, the overall trend indicates that research on PM_2.5–O₃ plant–meteorological synergistic purification has established itself as a distinct and increasingly active research frontier over the last decade.

3.2. Synthesis of Quantitative Evidence on Plant Purification of PM_2.5

3.2.1. Canopy-Scale Purification Efficiency and Dry Deposition Rate

A synthesis of 17 studies (n = 17) indicates that urban vegetation exhibits a median average purification efficiency of 18.2% (IQR: 12.5–30.1%) for surrounding PM_2.5 concentrations [30,31,32]. Community vertical structure is a key determinant of this efficiency. The purification efficiency of multi-layered tree–shrub–grass structures (median: 22.4%) was significantly higher than that of single-grass lawns (2–5%) or pure tree forests [33,34]. Significant seasonal variations in purification efficacy were observed across different vegetation configurations (Figure 2). As shown in Figure 2, multi-layered structures like tree–shrub–grass exhibited higher PM_2.5 reduction rates in spring, summer, and winter, owing to the coupled effects of moderate tree canopy permeability and interception by understory shrubs and grasses (n = 9) [35,36,37,38]. In contrast, single-grassland types exhibited significantly reduced purification capacity during spring and winter when vegetation cover decreased and wind speeds increased (Figure 2). The weighted median Vd–PM for PM_2.5 across 15 studies (n = 15) was 0.05 cm s⁻¹, though reported values spanned an extremely wide range (0.02–30 cm s⁻¹) [39,40,41]. Verification of data sources revealed that 14 studies provided observational data, while one study (S98) reported model-based estimates (mean across 10 cities: 0.65 cm s⁻¹). Extreme values exceeding 10 cm s⁻¹ originated from specific high-turbulence model scenarios. To clarify the influence of different evidence sources, analysis was conducted according to the methodology in Section 2.5. The pooled results from 14 observational studies showed a median Vd–PM of 0.05 cm s⁻¹ (IQR: 0.02–0.12 cm s⁻¹), with values predominantly distributed within the range of 0.02–2.0 cm s⁻¹. This indicates that in typical urban observational studies, the dry deposition rate of PM_2.5 by vegetation is generally low. The model-derived estimate reported in the S98 study (0.65 cm s⁻¹), a significant high outlier, reflects the theoretical potential under specific parameterization schemes, but does not alter the lower baseline level established by extensive observational data. This contrast underscores the importance of distinguishing data sources in synthesis studies and explains why the median value (0.05 cm s⁻¹) in this synthesis is lower than some higher individual values cited in the literature. This distinction is explicitly maintained throughout this study’s synthesis visualizations and decision-making applications. Vd–PM exhibits pronounced seasonality, typically higher in winter than summer, primarily driven by variations in pollution emission intensity and atmospheric stability [42,43,44,45].

3.2.2. Leaf-Scale Adsorption Capacity and Life-Form Variation

Based on data from 42 studies, the median PM_2.5 adsorption capacity per unit leaf area was 25.4 μg·cm⁻², with significant interspecies variation (range: 0.17–573 μg·cm⁻²) [46,47,48,49,50,51]. Leaf microstructure emerged as a key determinant, with leaf surface roughness exhibiting the strongest positive correlation with adsorption capacity (coefficient of determination R² reaching up to 0.8, n = 22) [52,53,54,55,56]. From a life-form perspective, distinct plant life forms exhibit significant variations in PM_2.5 adsorption capacity (Figure 3). Trees dominate annual particulate retention [57,58], yet shrubs often demonstrate superior retention per unit leaf area, primarily attributed to deeper leaf grooves and higher stomatal and trichome densities [59,60]. Evergreen shrubs, with their stable year-round leaf area and low canopy height, can continuously and effectively intercept near-surface particulate matter [61], whereas herbaceous plants contribute a relatively minor proportion to annual particulate matter retention [62,63] (Figure 3).

3.3. Synthesis of Quantitative Evidence for Plant Purification of O₃

3.3.1. Dry Deposition Rate and Non-Stomatal Contributions

The purification of O₃ by plants is primarily achieved through stomatal uptake. A synthesis of eight flux observation studies indicates that the weighted median ozone dry deposition velocity (Vd-O₃) for typical urban vegetation is 0.55 cm s⁻¹ (range: 0.12–1.82 cm s⁻¹) [64,65,66,67]. Stomatal absorption contributes approximately 65% of total deposition on average, while non-stomatal deposition (e.g., leaf surface, soil chemical reactions) accounts for a median contribution of about 35% (range: 15–65%, n = 4) [64,65]. Non-stomatal contributions are significantly influenced by leaf wetness, surface roughness, and atmospheric stability [68,69].

3.3.2. Variation in Stomatal Absorption Flux and Species Sensitivity

PODᵧ serves as a critical cumulative dose indicator for assessing O₃ uptake and risk in plants. Based on existing data (n = 6), the average PODᵧ value for temperate forest canopies typically ranges around 4.55 mmol·m⁻², while seedling experiments under high O₃ exposure have recorded PODᵧ values as high as 70 O₃ mmol·m⁻² [64,70,71]. Species sensitivity to O₃ significantly influences absorption capacity and stomatal response strategies, with marked variations in parameters such as stomatal conductance response and O₃ uptake rate across different tree species (Figure 4). As shown in Figure 4, sensitive species like Populus deltoides and Fraxinus chinensis exhibit substantial stomatal conductance reductions (>50%) under elevated O₃ concentrations, limiting their uptake capacity [72]. In contrast, tolerant species such as Platycladus orientalis and Quercus mongolica maintain higher stomatal conductance and sustained uptake capacity [73]. Furthermore, O₃ deposition rates in mature trees generally correlate positively with canopy height, while absorption rates among young tree species also exhibit considerable variation (Figure 4).

3.4. Synergy and Trade-Offs Under PM_2.5–O₃ Composite Pollution

3.4.1. Background on Pollutant Concentration Correlation

The spatiotemporal correlation between PM_2.5 and O₃ concentrations forms the basis for composite pollution events. A synthesis of 12 macroscale observational studies (n = 12) revealed significant variation in their correlation coefficients (r), ranging from −0.69 to +0.79 [74,75,76]. The direction of this correlation is primarily regulated by temperature: during the cold season (November–March), it predominantly exhibits a negative relationship, whereas during the warm season (especially summer), when photochemical activity is high, it often shifts to a positive correlation, leading to frequent synergistic pollution events [77,78].

3.4.2. Interactions of Plant Purification Functions

A synthesis of 16 direct studies (n = 16) revealed distinct interactive patterns of plant purification functions under composite pollution (

Table 2. Synergy and trade-offs in plant purification functions under PM_2.5–O₃ composite pollution.

Interaction Type	Key Conditions/Thresholds	Effect Description and Intensity	Number of Supporting Studies	References
Meteorological Synergy	T: 20–30 °C; RH: 50–70%	Synchronized enhancement of plant purification efficiency, exhibiting positive synergy	15	[79,80]
Meteorological Inhibition	T > 30 °C; RH < 40%	Widespread suppression of plant purification functions	7	[81,82,83,84]
BVOC Trade-Off	Planting species with high isoprene emissions	Offsets PM2.5 adsorption benefits, increasing net O3 pollution risk	9	[81,85,86,87,88,89]

). Weather-driven synergistic window: When meteorological conditions fall within the range of temperature 20–30 °C and relative humidity 50–70%, plant physiological activity is vigorous and atmospheric dispersion conditions are favorable, leading to synergistic enhancement effects in the purification of PM_2.5 and O₃ (75% of studies reported synergism, n = 9) [79,80]. Meteorologically driven inhibition window: Under conditions of temperature > 30 °C and relative humidity < 40% (high temperature and low humidity), plants suffer from water and oxidative stress, leading to stomatal closure and metabolic suppression. This significantly reduces purification capacity for both pollutants (n = 7) [81,82,83,84]. Key trade-offs induced by BVOC emissions: High-BVOC-emitting tree species (e.g., certain poplars and eucalyptus) trigger significant negative trade-offs. Their emitted VOCs, such as isoprene, O₃ precursors, may offset or even reverse the PM_2.5 purification benefits achieved through leaf adsorption by these species. Studies indicate that high-emitting tree species can contribute up to 86 g of net daily O₃ pollution equivalents per tree (n = 9) [85,86,87,88,89]. Therefore, “low BVOC emissions” is a core criterion for selecting tree species for synergistic purification. It should be noted that the quantitative evidence (n = 9) supporting the synergistic potential and trade-off risks discussed in this section is highly geographically concentrated. Geocoding analysis of relevant studies revealed for Ginkgo biloba and Platycladus orientalis that evidence supporting their synergistic purification potential almost exclusively originates from observational studies in roadside and park green spaces of major Chinese cities like Beijing and Shanghai. Evidence for Populus spp. and Eucalyptus spp. as species posing trade-off risks due to high BVOC emissions primarily stems from controlled experiments and scenario simulations conducted in the North China Plain and Yangtze River Delta regions. This spatial enrichment phenomenon (see Section 3.8.1) implies that current quantitative understanding of the relationship between functional traits and purification efficacy in these tree species primarily originates from specific urban environments within East Asia’s monsoon climate zone. Therefore, when applying related tree species selection recommendations (e.g., prioritizing Ginkgo biloba and Platycladus orientalis) to other climatic zones (e.g., Mediterranean, tropical), one must recognize the limitations of the evidence base and prioritize empirical data based on local native tree species. Calculation note: This value represents a theoretical maximum estimate intended to point to the potential risks of high-BVOC-emitting tree species. Calculations are based on the following assumptions and parameters: (1) tree species—represented by high-isoprene-emitting species (e.g., Populus przewalskii); (2) emission rate: utilizes the highest observed emission flux reported in the literature, i.e., XX μg·g⁻¹·h⁻¹; (3) ozone generation potential—employs the maximum increment reaction activity coefficient (MIR) value of YY g O₃/g BVOC; (4) tree biomass—assumed to be mature trees with leaf dry weight ZZ kg; and (5) meteorological conditions—assumed to be photochemically active conditions of high temperature and strong light. The calculation formula is: net O₃ generation equivalent = Σ(emission rate × MIR × leaf dry weight × time). This calculation does not deduct the tree’s own O₃ absorption, aiming to present the worst-case scenario.

3.5. Quantification of Regulatory Effects of Key Meteorological Factors

Meteorological factors exhibit nonlinear and threshold-dependent regulatory influences on the purification process (synthesized from 58 studies) (

Table 3. Summary of regulatory effects of key meteorological factors on the plant purification process.

Meteorological Factors	Impact on PM2.5 Purification	Impact on O3 Purification	Net Effect on Composite Pollution	Key Threshold/Effect Size	Number of Supporting Studies	References
Temperature (T)	Weak negative correlation (r ≈ −0.2)	Strong positive correlation (O3 formation)	Bimodal: synergistic at moderate temperatures, trade-off at high temperatures	T = 20–30 °C synergistic window	37	[75,97,98]
Relative humidity (RH)	Positive correlation (significant at RH > 70%)	Strong negative correlation (inhibition at RH > 60%)	Synergistic at moderate humidity (50–70%), trade-off at high humidity	RH = 55% inflection point	35	[90,91]
Wind speed (WS)	Bidirectional: weak winds accumulate, moderate winds promote deposition, strong winds resuspension	Weak positive correlation (enhances transport), strong winds dilute	1.5–3.0 m·s−1 optimal range	WS < 1.5 m·s−1 high risk of stagnation	19	[94,95,96]

). Relative humidity (RH) shows a positive correlation with PM_2.5 adsorption: when RH > 70%, leaf wetting significantly promotes particulate matter attachment (n = 35) [90,91] while exhibiting a strong negative correlation with O₃. High humidity (RH > 60%) inhibits photochemical reactions and enhances wet removal of O₃, potentially affecting stomatal behavior [92,93]. Wind speed has a distinct optimal range for PM_2.5 deposition. Wind speeds between 1.5 and 3.0 m·s⁻¹ most effectively enhance canopy turbulent exchange and promote deposition: velocities < 1.5 m·s⁻¹ tend to cause stagnant accumulation, and velocities > 8 m·s⁻¹ may trigger surface dust resuspension, leading to concentration rebound (n = 19) [94,95,96]. Temperature is the primary driver of O₃ photochemical production, showing a strong positive correlation with near-surface O₃ concentrations (sensitivity ~+1.1 ppb K⁻¹, n = 12) [75,97,98]. However, for plant uptake, optimal temperatures (20–30 °C) promote stomatal opening, while temperatures > 30 °C often cause stomatal closure, inhibiting uptake [99]. Solar radiation—directly driving photochemical reactions—is the essential energy source for O₃ formation and shows a strong positive correlation with O₃ concentrations [100,101].

3.6. Visualization of Key Purification Parameter Distributions

To visually demonstrate the core quantitative indicators of plant purification capacity, this study further analyzed the data distribution characteristics of Vd and PODᵧ (Figure 5 and Figure 6). The results revealed the central tendency and range of variation for key parameters under different conditions.

3.6.1. Distribution Characteristics of Dry Deposition Velocity (Vd)

Analysis indicates (Figure 5) that plant Vd–PM data for PM_2.5 exhibit high heterogeneity. A synthesis of 15 studies (n = 15) yielded a weighted median of 0.05 cm s⁻¹ [39,40,41]. Data verification revealed that 14 studies provided observational data, yielding a pooled median of 0.05 cm s⁻¹ (IQR: 0.02–0.12 cm s⁻¹), primarily distributed within the range of 0.02 to 2.0 cm s⁻¹. One additional study (S98) reported 0.65 cm s⁻¹ as a model-derived estimate, serving as the primary outlier extending the upper range of data. This composition indicates that existing observational evidence supports a lower sedimentation rate baseline, while occasionally cited higher values in the literature may stem from individual model projections or specific scenarios. Comparing findings across life forms (based on all data), the median Vd–PM values for trees and shrubs are relatively close (0.61 and 0.72 cm s⁻¹, respectively), but both groups exhibit substantial internal variability. Given the low median of the main observational dataset (0.05 cm s⁻¹), the higher medians observed for these habitat types are likely significantly influenced by a small number of outliers, including modeled estimates from S98. This further underscores that the apparent effects of habitat type may be modulated by complex factors such as outliers, data sources, and canopy microenvironments, requiring caution in interpretation. Compared to PM_2.5, high-resolution continuous direct observations of O₃ dry deposition velocity (Vd–O₃) are even scarcer (two studies provided such datasets [64,65]). A synthesis of all eight studies yielded a median Vd–O₃ of approximately 0.55 cm s⁻¹, preliminarily indicating a similar magnitude to Vd–PM.

3.6.2. Distribution of Cumulative Ozone Absorption Dose (PODᵧ)

The physiological effects of ozone on plants depend on the cumulative dose entering leaves through stomata (PODᵧ). As shown in Figure 6, existing studies report a wide range of PODᵧ values [64,70,102]. During the typical growing season in temperate forests, the canopy-scale PODᵧ typically reaches approximately 4.55 mmol·m⁻² (n = 6). In controlled experiments, however, seedlings exposed to high O₃ concentrations can absorb doses as high as 70 O₃ mmol·m⁻². This magnitude difference clearly indicates that environmental O₃ concentration and exposure duration are key drivers determining the actual absorbed dose in plants. Although high-quality PODᵧ observation datasets remain limited, their wide range underscores the importance of considering specific exposure scenarios when assessing ozone risks and screening tree species. It should be noted that the core studies (n = 4) providing the aforementioned PODᵧ data exhibit geographical limitations. Consistent with the overall spatial pattern of the literature included in this study (see Section 3.8.1), these data primarily originate from forest ecosystems or controlled experiments in temperate regions of the Northern Hemisphere (e.g., China, Europe). This implies that current understanding of PODᵧ dose–response relationships remains particularly weak in different climatic zones, such as tropical and arid regions, and in studies of typical urban shrub or herbaceous plant communities. Therefore, when applying these PODᵧ range values to global ozone ecological risk assessments across diverse regions, differences in local climatic conditions, dominant tree species, and exposure histories should be fully considered.

3.7. Correlation Between Plant Functional Traits and Purification Capacity

Leaf and canopy traits form the intrinsic basis for determining purification functions (

Table 4. Representative evidence linking plant functional traits to purification capacity.

Trait Category	Effect on PM2.5 Retention	Effect on O3 Uptake	Recommended Trait Range/Type	Number of Supporting Studies	References
Leaf surface roughness	Strong positive correlation (R2 up to 0.8)	Weak positive correlation (non-stomatal deposition)	High roughness (SEM quantification)	22	[48,52,53]
Stomatal density	Positive correlation (r = 0.51–0.96)	Strong positive correlation (primary uptake pathway)	High density (e.g., >200 mm−2)	14	[104]
BVOC emission potential	Indirect negative effect (SOA formation)	Direct negative effect (O3 precursor)	Low-emission tree species prioritized	6	[83,85,86]

). Leaf surface roughness exhibits the strongest correlation with PM_2.5 adsorption capacity [103]. Stomatal density and conductance directly determine the stomatal absorption flux of O₃. BVOC emission potential is positively correlated with O₃ formation risk [104]. Consequently, when selecting tree species, one should pursue trait combinations featuring high leaf roughness and stomatal conductance coupled with low BVOC emissions to achieve synergistic purification while avoiding trade-offs.

3.8. Chronological and Geographical Characteristics of Research Activities

3.8.1. Geographic Distribution and Spatial Enrichment

Based on the geocoding of the literature (n = 226), approximately 85% of the studies (n = 192) were concentrated in China, particularly in eastern urban clusters such as the Beijing–Tianjin–Hebei region, the Yangtze River Delta, and the Pearl River Delta. These three regions accounted for over 60% of the research sites within China (Figure 7). The remaining studies (n = 34) were scattered across regions, including Europe and North America. This highly uneven spatial distribution directly correlates with the urbanization processes, regional air pollution pressures, and research investment intensity of major global economies. It directly leads to a critical methodological limitation: the vast majority (>85%) of existing quantitative understanding regarding the mechanisms and efficacy of plant–meteorological synergistic purification is based on urban environmental observations in the East Asian monsoon region (temperate to subtropical). Evidence for tropical, arid, and Southern Hemisphere urban ecological systems remains extremely scarce. Therefore, when extrapolating the core conclusions of this review to these climatic zones, the limitations of the geographical evidence base must be acknowledged, and priority should be given to localized studies.

3.8.2. Chronological Evolution of Research Activities

Research activities in this field exhibit three distinct phases (Figure 8): a slow start-up period (2000–2014), with annual publications generally below 5, a rapid growth phase (2015–2018), where annual publications increased to 5–15, and a high-plateau phase (2019–2023), with annual publications consistently exceeding 15 and peaking in 2023 (n = 37). This growth trajectory, particularly the acceleration after 2015, exhibits high temporal synchrony with two key drivers: globally, a surge in academic interest in nature-based solutions (NBSs) for air pollution governance, and at the national level, robust environmental policies such as China’s Action Plan for Air Pollution Prevention and Control (implemented in 2013) spurred extensive empirical research on the purification effects of urban greening in local contexts. Despite publication lags affecting recent data, this field has emerged as a distinct and dynamic research frontier.

3.8.3. Geographic Representativeness and Limitations of Tree Species

The geographic origins of evidence for the frequently studied tree species summarized in Section 3.3.2 (such as Platanus spp., Ginkgo biloba, and Platycladus orientalis) align perfectly with the aforementioned spatial enrichment patterns. For instance, empirical studies on the purification performance of plane trees and ginkgo trees were almost exclusively conducted in roadside or park green spaces of major eastern Chinese cities like Beijing and Shanghai. This provides solid localized evidence for the application of species like ginkgo and arborvitae in similar environments across eastern China. However, this high degree of geographic concentration exposes three critical gaps in the current knowledge base: (1) lack of data on the purification efficacy and adaptability of these “recommended tree species” in other climatic zones (e.g., Mediterranean, tropical); (2) virtually no comparative studies on their performance versus native tree species in urban ecosystems worldwide; and (3) uncertainty regarding whether their functional traits (e.g., stomatal conductance response to O₃) are influenced by geographic provenance. Therefore, the tree species selection recommendations proposed in this review are primarily applicable to cities in East Asia’s monsoon climate zone. When planning for other regions, two critical steps must be undertaken: First, carefully assess the extrapolation risks of this review’s conclusions, and second, prioritize initiating baseline studies on the purification efficacy and BVOC emission characteristics of locally dominant tree species to establish a localized evidence base.

4. Discussion

4.1. Quantitative Benchmarks and Practical Value of Multi-Scale Purification Efficiency

Through quantitative synthesis, this study provides a multi-scale quantitative reference benchmark for the PM_2.5 and O₃ purification services of urban vegetation. The median purification efficiency of 18.2% at the canopy scale (n = 17) [30,31,42,105] (Table 2) offers a critical reference for assessing and predicting the ecological functions of green spaces. More importantly, the data reveal potential for performance enhancement through optimized design: employing multi-layered tree–shrub–grass structures can systematically increase efficiency to over 22% (n = 9) [106,107]. This provides concrete quantitative support for planning concepts such as “vertical greening” and “forest cities.” At the leaf scale, substantial interspecific variation in leaf area-specific adsorption capacity (26.1 μg·cm⁻²) (n = 42) [108,109,110,111] underscores the importance of precise tree species selection based on leaf micromorphology, rather than relying solely on empirical life-form classifications.

4.2. Identification of Meteorological Synergy Windows and Adaptive Management Transformation

The meteorological synergistic window (T: 20–30 °C, RH: 50–70%) and suppression window (T > 30 °C, RH < 40%) identified in this review translate the complex effects of meteorological factors into specific, manageable temporal nodes and condition combinations (n = 21, suppression window n = 12) [112,113,114,115]. This finding necessitates a shift in urban green space management from traditional static, uniform maintenance models towards dynamic, weather-alert-based adaptive management approaches. During synergistic windows, management should prioritize plant health to maximize natural purification services. Conversely, in high-temperature, low-humidity suppression windows, measures such as irrigation must mitigate stress, shifting the core objective to ensuring plant survival and long-term health. This refined management strategy is pivotal for enhancing urban green spaces’ climate resilience and service efficacy [116,117].

4.3. Addressing Composite Pollution: From Species Selection to Systematic Green Space Design

In response to the synergistic pollution of PM_2.5 and O₃, the core insight of this study is the imperative to evolve from a simplistic “tree selection” mindset to a “systemic design” approach. The primary principle is to prioritize “low BVOC emissions” as a key criterion for species selection, thereby minimizing the risk of mutual cancellation in purification functions. Building upon this foundation, a spatial differentiation and functionally complementary green space configuration model should be implemented (

Table 5. Mapping matrix of meteorological combinations and composite pollution event types.

Pollution Event Type	Key Meteorological Driver Combination	Typical Meteorological Parameter Range	Potential Dominant Mechanism	Associated Phytoremediation Strategy	Evidence Basis, Source	Remarks (Evidence Type, Description)
PM2.5 stagnant accumulation type: low temperature + high humidity + calm winds	Temperature: <10 °C Relative Humidity: >75% Wind Speed: <1.5 m/s Low-level mixing layer	Enhanced liquid-phase secondary aerosol formation; severely limited vertical and horizontal dispersion capacity	Activate “interception-deposition” mode: Increase irrigation and humidification of existing green spaces (maintain leaf surface moisture, elevate Vd). Avoid operations generating dust through disturbance (e.g., pruning, sweeping).	Activate “interception-deposition” mode: 1. Enhance irrigation and humidification of existing green spaces (maintain leaf surface moisture, increase Vd). 2. Avoid operations generating dust through disturbance (e.g., pruning, sweeping).	RH > 75% enhances adsorption: (n = 5) [S030, S045, S112] Low wind speed (<1.5 m/s) causes accumulation: (n = 3) [S091, S092]	Quantitative evidence + expert inference High humidity enhances adsorption with quantitative evidence; specific management measures (irrigation, work restrictions) are mechanism-based expert recommendations
O3 photochemical formation	High temperature + strong radiation + low humidity	Temperature: > 30 °C Solar radiation: > 600 W/m2 Relative humidity: <40% Wind speed: 1.5–3.0 m/s (facilitates precursor transport)	Maximizes photochemical reaction rates (ROx cycle); increased plant BVOC emissions supply precursors.	Initiate “low absorption-promote diffusion” mode: Implement misting for cooling and humidification on low vegetation in densely populated areas to temporarily suppress O3 formation. Strictly prohibit activities during this period that may damage foliage or induce BVOC emissions.	High temperature and low humidity suppress plant functions (n = 7) [S078–S081] High BVOC tree species increase net O3 risk: (n = 9) [S082–S086, S148, S149, S187]	Quantitative Evidence Synthesis Strong evidence exists for the O3 generation potential of BVOCs; spray measures represent evidence-based emergency management inferences
Composite pollution synchronous phase	Moderate temperature + moderate humidity + stable atmospheric conditions	Temperature: 20–28 °C Relative humidity: 50–70% Wind speed: <1.5 m/s Radiation: Moderate	Conditions for secondary PM2.5 formation and photochemical O3 generation concurrently met; pollutant accumulation and chemical generation mutually reinforce	Activate “spatial differentiation-synergy” mode: 1. Strengthen management of spatially differentiated zones to ensure normal functioning of ground-level shrub/grass layers and upper tree canopy layers. 2. Precisely align with “meteorological windows” (i.e., current conditions) to maximise overall plant physiological activity (e.g., through moderate irrigation).	Climate-Adaptive Window (T: 20–30 °C, RH: 50–70%): (n = 9) [S076, S077] Multi-Layer Structure High-Efficiency Purification: (n = 9) [S030–S034, S103–S104]	Quantitative Evidence Synthesis Both the “synergy window” and multi-layer structure efficiency are supported by quantitative evidence
Dust-particulate dominant	Dry + high wind	Relative humidity: <30% Wind speed: >4.0 m/s (dust-lifting threshold) Vegetation cover: low	Dominant primary particulate matter (coarse/fine) transport; potential entrainment of local PM2.5	Activate “intercept–fix” mode: 1. Urgently activate sprinkler systems in frontline shrub-grass buffer zones to precipitate near-surface particulates. 2. Inspect and reinforce exposed ground cover within green spaces.	Windbreak belts reduce wind speed and particulate matter concentration: (n = 3) [S055, S098, S114] Sand-lifting threshold at wind speeds > 4.0 m/s: fluid mechanics principles/expert inference	Evidence synthesis + expert inference Evidence supports windbreak function; specific wind speed thresholds incorporate expert judgment.
Post-rain cleansing type	Precipitation + moderate to strong winds	Precipitation: >5 mm, wind speed: 2.0–4.0 m/s Enhanced radiation post-precipitation	Wet deposition (precipitation washout) dominates removal process; good atmospheric transparency, favorable dispersion conditions	Initiate “physiological recovery and monitoring” mode: 1. Conduct plant health inspections during favorable air quality conditions. 2. Light irrigation may be applied to rinse residual particulate matter from foliage, promote restoration of leaf stomatal function, and prepare for the next pollution episode.	Precipitation wet removal effect: (n = 4) [S095, S096, S118, S119] Irrigation washing of leaf surface particles: (n = 3) [S030, S045, S145]	Quantitative evidence + expert inference Wet removal has observational evidence; irrigation promoting recovery is based on plant physiological inference.

). For instance, downwind of primary pollution sources (e.g., major thoroughfares and industrial zones), wide-span, multi-layered protective forest belts centered on evergreen conifers (e.g., cedar and white pine) with high PM_2.5 adsorption and low BVOC emissions can be planned. These trees leverage stable physical deposition and year-round protective functions [118]. In densely populated residential areas or parks, sparse woodlands with grasslands dominated by broadleaf trees (e.g., ginkgo and golden rain trees) with high O₃ absorption and low BVOC emissions can be configured. These areas provide ecosystem services such as air purification, shade, and cooling [119]. Therefore, advancing a “zone-specific, function-oriented” design paradigm will be a crucial direction for achieving synergistic purification of multiple pollutants and enhancing the comprehensive benefits of urban spaces in the future.

4.4. Limitations and Future Research Directions

This study has certain limitations. Firstly, the geographical distribution of included research is uneven, predominantly concentrated in temperate and subtropical cities of the Northern Hemisphere, with insufficient evidence coverage for tropical, arid regions and the Southern Hemisphere, potentially affecting the generalisability of conclusions. Secondly, due to the relatively scarce literature reporting advanced parameters such as PODᵧ and BVOC fluxes in original studies, the sample size for some quantitative syntheses is limited. High-quality flux observational evidence concerning O₃ purification remains scarce. Key parameters characterizing purification efficiency, such as the vertical deposition rate (Vd-O₃) and non-stomatal deposition contribution, are based on only eight and four studies, respectively, predominantly from forest ecosystems, with a lack of long-term observations for typical urban vegetation configurations. Finally, heterogeneity in measurement methods, indicator definitions, and reporting formats across studies remains a primary challenge for data integration and comparison. Furthermore, at the methodological level, although we conducted risk-of-bias assessments and quality evaluations for included studies, the high heterogeneity of evidence made it challenging to establish systematic calibration relationships between quality scores and effect measures (e.g., Vd) in quantitative syntheses. Consequently, quality assessment results were primarily used for descriptive analysis rather than directly applied to adjust statistical weights or perform subgroup analyses, thereby limiting the depth of differentiated integration across evidence of varying quality.

Future research should prioritize the following directions: establishing long-term, multi-site urban ecological observation networks to simultaneously monitor key plant purification parameters (Vd, PODᵧ, BVOC), pollutant concentrations, and meteorological factors, thereby accumulating standardized time-series datasets; developing and validating high-resolution mechanistic models capable of coupling plant physiology, canopy microclimate, urban wind fields, and atmospheric chemistry for scenario simulation and planning optimization; and promoting standardized protocols for measuring and reporting plant purification traits while enhancing the screening and evaluation of pollution-tolerant, low-emission plant germplasm resources for urban applications.

5. Conclusions and Sustainability Implications

Urban vegetation exhibits purifying capacity for PM_2.5 and O₃, though its efficacy undergoes profound nonlinear modulation by meteorological conditions. Median canopy efficiency for PM_2.5 purification stands at approximately 18%, with median dry deposition velocities around 0.65 cm s⁻¹. Median dry deposition velocities for O₃ purification are approximately 0.55 cm s⁻¹. Among meteorological factors, suitable humidity promotes PM_2.5 removal, while moderate temperatures (20–30 °C) and humidity (50–70%) optimize synergistic purification of both pollutants. Vigilance is required regarding potential trade-off effects from tree species with high BVOC emissions.

To achieve synergistic enhancement between “blue skies” and “green spaces” and advance sustainable urban living environments, the following recommendations are proposed.

Promote planning paradigm shifts: Integrate quantitative evidence from this review (e.g., purification efficiency benchmarks, synergistic meteorological windows, tree species trait thresholds) into urban green space system planning, ecological design of renewal projects, and green infrastructure standard-setting. Transition from “empirical greening” to “evidence-based precision greening.”

Implement climate-adaptive management: Establish dynamic urban green space maintenance mechanisms linked to meteorological alerts. Prioritize plant health during synergistic windows to optimize purification services while implementing resilience-enhancing measures during stress windows. In urban wind corridor planning, employ simulations to optimize wind speeds within corridors to the beneficial range of 1.5–3.0 m·s⁻¹.

Innovative policies and assessment tools: Explore establishing ecological compensation mechanisms incentivizing the adoption of low-emission, high-purification native tree species. Integrate composite pollution synergistic purification efficacy into urban green space quality evaluations and gross ecosystem product (GEP) accounting systems, positioning it as a key indicator for measuring urban ecological well-being and sustainable development levels.

Through interdisciplinary integration of quantitative knowledge and innovative planning management practices, fully leveraging the synergistic purification effects of vegetation and meteorology will provide core ecological support for building a more resilient, healthy, and sustainable urban future.