Introduction

The rise of antimicrobial resistance (AMR) is a global challenge that threatens the lifesaving role of antibiotics, upon which modern medicine is highly dependent. The World Health Organization (WHO) considers antimicrobial resistance to be one of the most significant public health challenges1. AMR infections are increasing worldwide, and in many cases can only be treated by antimicrobials of last resort2. A recent paper published in The Lancet reported the most comprehensive analysis to date, based on data from 204 countries, and highlighted that there were 1.27 million deaths in 2019 due solely to AMR (> HIV/AIDS or malaria) and that AMR infections played a decisive role in the deaths of 4.95 million individuals worldwide3.

Numerous drivers, such as the unregulated sale of antibiotics through pharmacies; poor hygiene practices in hospitals, food production and community settings; overuse of antibiotics in agriculture and livestock; and use of harsh cleaning products in public settings, have been identified as contributing to AMR selection and transmission4,5,6,7,8,9,10. Qiao et al. (2018) reviewed the current literature and showed that not only pathogens but also antimicrobial resistance genes (ARGs) are present in hospitals, solid waste landfills, sewage plants and rivers11. The selection and transmission of AMR is further contributed to by various vectors, including water sources, surfaces (fomite transmission) and a newly described transmission vector: the air12,13. The respiratory tract is the primary route through which ARGs in bioaerosols enter the human body; these bioaerosols carry both antimicrobial resistant bacteria (ARBs) and ARGs14. This dissemination of ARGs aggravates their transmission in the environment15.

Antimicrobial resistance genes have been found to be widely distributed in water, soil, livestock and compost of farming and urban communities11but until recently their presence in and dissemination through the air has not received as much attention, although aerosolised ARGs are known to be disseminated in the air of farms, hospitals and wastewater treatment plants16. The concentration of ARGs in outdoor aerosols changes depending on the level of PM2.5(particulate matter below 2.5 micrometres in diameter)17, and seasonal climate18. However, studies investigating the impact of temperature and air quality on the dissemination of aerosolised ARGs in urban settings usually consider one of these factors in isolation, missing opportunities to understand how they interact with one another to exacerbate AMR risk19. Moreover, interventions suggested by such studies to decrease the dissemination of ARGs in the air, such as planting more diverse urban forests with trees selected specifically for their ability to filter ARGs19, have not been implemented. It is widely acknowledged that planting trees is an effective intervention to improve air quality20, however, whether such an intervention has positive effects in decreasing the dissemination of ARGs in the air in urban settings remains to be explored. Such an understanding is crucial to developing a better understanding of the phenomena, as this can help to inform urban planning policy in the crucial period to 2050, when the current rapid increases in urban growth are expected to plateau: the next two decades are a crucial period in which good urban design, which leaves space for ecosystem services, can be locked in21,22.

In this study, a real-time sampling protocol was used to investigate the effect of vegetation coverage (by comparing high vegetation coverage sites (HV), such as gardens and parks, with low- or no vegetation sites (LNV), such as playgrounds and sites that are known to emit high levels of PM2.5, and from urban heat islands (UHI) – urban areas where temperatures can be higher than surrounding areas due to high energy use, heat absorbent surfaces or lack of shade). We consider the effects of temperature, humidity, wind speed, bacterial community composition, and atmosphere pollutant levels on the dissemination of aerosolised ARGs. Our study expands the understanding of factors that may aid in decreasing the dissemination of ARGs through air.

Methods

Sample collection

Air samples were collected from high vegetation coverage sites (HV) (North latitude N39°10’74.56”; East longitude E117°18’43.78”), low or no-vegetation sites (LNV) (North latitude N39°10’87.18”; East longitude E117°18’42.18”), and from the outdoor environment of an urban heat island (UHI) known for its of high levels of PM2.523 (North latitude N39°10’75.45”; East longitude E117°18’33.63”) of Tianjin Medical University. The metrics of vegetation coverage are included in the supplementary material (Table S1). Air samples were collected from Oct 2021 to June 2022, including during autumn (Oct 2021, over 3 days), winter (from Nov to Dec 2021, over 7 days), spring (from March to May 2022, over 17 days), and in summer (Jun 2022, over 6 days).

A volume of 600 L air was collected using a Coriolis µ air sampler (Bertin Technologies) at a flow rate of 200 L/min for 3 min at 1.5 m vertical distance from the ground24,25. All samples were collected into 2 ml solution A, part of the genomic DNA extraction kit (Solarbio Technology Ltd, China).

Air quality

Meteorological data and air pollution levels were obtained from the air quality institutes of China (https://www.aqistudy.cn/) (Table S2). Dali Road was the nearest monitoring site (distance:1.8 km) to Tianjin Medical University. Meteorological and air pollution levels were obtained from the monitoring site located on Dali Road at the time the aerosolised samples were collected.

DNA extraction

DNA of aerosolised samples was extracted using a genomic DNA extraction kit according to the manufacturer’s guidelines (Solarbio Technology Ltd, China). An ultrafine spectrophotometer (Nanodrop2000, Thermo, USA) and ultraviolet spectrophotometry method were used to measure the concentration of DNA in each sample. Genomic DNA was subsequently diluted in double distilled water to a final concentration of 200 ng/ml.

Quantification PCR

All samples were tested for nine ARGs that have previously been reported to have clinical importance26. These included aac(6’)-Ib (quinolone), ermB (macrolide), ermC (macrolide), tetM(tetracycline), tetW (tetracycline), sul1 (sulfonamides), sul3 (sulfonamides) blaTEM (β-lactams) and mecA (β-lactams). 16S rRNA gene was used as a reference for results obtained by the quantification PCR (qPCR). All reactions were performed in triplicate. Antimicrobial resistance genes were amplified by Roche LightCycler480II. SYBRTM Green PCR Master Mix (Novo Start® SYBR qPCR Super Mix Plus, China) was used in the qPCR reaction system of a total 10 µL, including 5 µL of 2×Novo Start® SYBR qPCR Super Mix Plus, 3.5 µL of ddH2O, 0.5 µL of each primer and 0.5 µL of templates. Melting curve analyses were automatically generated by the Light Cycler R 480 Software 1.2.1.62. 2−△CTmethod was used for the ARGs relative abundance assessment27 (Table S3 and S4). ΔCT=CT(ARG)-CT(16 S rRNA), where CT,(ARG) and CT,(16s rRNA) were the amplification threshold cycles, respectively. 16S rRNA was used for reference purposes. The relative abundance of ARGs was determined as the number of copies of genes normalised to the copy number of 16 S rRNA genes.

Identification of bacterial communities

Three replicates of aerosol DNA samples were extracted and mixed, and the V3-V4 hypervariable region of the bacteria 16 S rRNA gene was sequenced by the Illumina Miseq™/Hiseq™ platform, and the image data file was converted into the sequences through base calling (Sangon Biotech, Shanghai, China). Reads were assembled by PEAR (version 0.9.8)28, and the sequencing quality control was assessed using PRINSEQ (version 0.20.4)29. Chimeric sequences were determined and filtered by UPARSE-OTU algorithm command to achieve 97% Operational Taxonomic Units (OTU) clustering (cluster_aggd command (drive5.com)). Taxonomic classification was identified by RDP 16 S database. OTU were clustered at the 97% similarity level by Usearch, and OUT was annotated by the Basic Local Alignment Search Tool (BLAST®).

Statistical analysis

The relative abundance of ARGs conformed to a normal distribution by logarithmic transformation. One-way analysis of variance (ANOVA) tests and independent t-test was conducted by R 4.1.0. software, and P < 0.05 suggested that the difference has statistical significance. The “ggcor” package was used to analyse the correlation between the relative abundance of aerosolised ARGs and the various influencing factors. Average relative abundance was used to represent the average amount of ARGs determined in different locations, seasons and haze levels.

Results

Aerosolised samples

A total of 65 aerosolised samples were collected between Oct 2021 and Jun 2022. This included 31 samples collected from the LNV sites, 16 samples collected from the outdoor environment of the UHI site and 18 samples from the HV sites. The DNA concentrations that were extracted from aerosolised samples ranged from 0.2 to 35.4 ng/µL (Table S5).

The range of relative amount of aerosolised ARGs

Relative abundance ranges of each aerosolised ARGs subtypes were as follows: blaTEM (0.0007–28.5408 (ARGs/16S rRNA)), mecA (0.0039–4.4449 (ARGs/16S rRNA)), sul3 (0.0002–3.9770 (ARGs/16S rRNA)), ermB (0.0022–0.0498 (ARGs/16S rRNA)), ermC (0.0004–0.1228 (ARGs/16S rRNA)), aac(6’)-Ib (8.0476 × 10−5−0.0491 (ARGs/16S rRNA)), tetM (0.0006–0.0232 (ARGs/16S rRNA)), tetW (9.9893 × 10−5−0.1205 (ARGs/16S rRNA)), sul1 (0.0007–0.0046 (ARGs/16S rRNA)) Moreover, the relative abundance of blaTEM and mecA genes were significantly higher than that of other ARGs (F = 6.126, P < 0.001) (Fig. 1, Table S5).

Fig. 1
figure 1

Statistical analysis of the relative abundance of each ARG subtypes.

Full size image

Aerosolised ARGs in different locations

We further measured the average relative abundance for each of the aerosolised ARGs in seasons requiring heating (Nov 2021-Mar 2022) and those that do not (Apr-Oct 2022) (Table 1). In the season that required heating, the average relative abundance of nine aerosolised ARGs was as follows: UHI site (0.0280 (ARGs/16S rRNA)), LNV site (0.0213 (ARGs/16S rRNA)) and HV site (0.0140 (ARGs/16S rRNA)). After central heating was switched off, the average relative abundance of 9 ARGs was as follows: LNV (0.0135 (ARGs/16S rRNA)), UHI (0.0083 (ARGs/16S rRNA)) and the HV (0.0083 (ARGs/16S rRNA)). The relative abundance of ARGs in the HV areas was slightly lower than around the UHI and LNV playground sites. However, no statistical significance was observed (F = 1.27, P = 0.29). Moreover, the relative abundance of nine aerosolised ARGs outside the UHI site was slightly higher than in the LNV and the HV sites in the heating season, but not of statistical significance (F = 0.77, P = 0.46). (Table 2; Fig. 2).

Table 1 Average relative abundance of each aerosolised ARGs in different seasons, locations and haze levels.
Full size table
Table 2 Statistical analysis of the average relative abundance of nine ARGs in different locations, seasons and haze levels.
Full size table
Fig. 2
figure 2

The relative abundance of ARGs in different sites during the heating and non-heating season. UHI: urban heat island, HV: high vegetation coverage (garden), LNV: Low or no vegetation (playground).

Full size image

Aerosolised ARGs in different seasons

The average relative abundance of eight aerosolised ARGs in autumn, winter, spring and summer were 0.0028 (ARGs/16S rRNA), 0.0098 (ARGs/16S rRNA), 0.0161 (ARGs/16S rRNA) and 0.0097 (ARGs/16S rRNA) respectively. The relative abundance of aerosolised ARGs was found to be different during all four seasons (F = 7.3, P = 0.001). The relative abundance of aerosolised ARGs in spring exhibited significantly higher relative abundance than in autumn (t = 6.79, P < 0.001) and winter (t = 3.73, P = 0.003). Moreover, the relative abundance of aerosolised ARGs around the UHI site significantly decreased compared with the heating season (which usually runs from November to March; the non-heating season usually runs between April to October) (t = 2.06, P = 0.04). In addition, no significant reduction of ARGs was observed in the aerosolised samples collected from the LNV (t = 0.02, P = 0.99) and HV sites (t = 1.07, P = 0.29) between the heating season and the non-heating season (Table 2).

Aerosolised ARGs in different temperatures and humidity

We found that the relative amount of blaTEM (r=−0.35, P<0.05) and ermC (r=−0.49, P<0.01) significantly decreased when the temperature increased (Table 3). Moreover, the relative abundance of aac(6’)-Ib(r=−0.13, P > 0.05), ermB(r=−0.03, P > 0.05), sul1(r=−0.09, P > 0.05), tetM(r=−0.26, P > 0.05), tetW(r=−0.14, P > 0.05) and mecA(r=−0.01, P > 0.05) were negatively correlated with temperature. In contrast, sul3(r = 0.38, P < 0.05) was positively correlated with temperature.

Table 3 Correlation matrix of the analysed parameters.
Full size table

We found that the relative amount of blaTEM and humidity showed a significantly positive correlation (r = 0.4, P < 0.05) (Table 3, Figure S1). In contrast, ermB (r=−0.11, P > 0.05), tetM(r=−0.31, P > 0.05), tetW(r=−0.02, P > 0.05) and mecA(r=−0.04, P > 0.05) decreased with increased humidity. Moreover, aac(6’)-Ib(r = 0.03, P > 0.05), ermC(r = 0.23, P > 0.05), sul1(r = 0.29, P > 0.05), and sul3(r = 0.18, P > 0.05) increased with increasing humidity.

Aerosolised ARGs on haze and non-haze days

According to a previous report13, haze days are defined as PM2.5 ≥ 75 µg/m3, while non-haze days are defined as PM2.5 < 75 µg/m3. The average relative abundance of nine aerosolised ARGs on haze days (0.0097 (ARGs/16S rRNA)) was lower than that on non-haze days (0.0142 (ARGs/16S rRNA)). We found that ermB(r = 0.13), ermC(r = 0.11) and sul1(r = 0.05) were positively correlated with the concentration of PM2.5, but no statistical significance was observed (P > 0.05). In contrast, we observed that aac(6’)-Ib(r=−0.05), sul3(r=−0.20), tetM(r=−0.30), tetW(r=−0.11), blaTEM(r=−0.08) and mecA(r=−0.02) were negatively correlated with PM2.5 concentration, but no statistical significance was identified (t = 0.80, P = 0.42) (Table 2 and S6, Figure S2).

Aerosolised ARGs in different wind speed

We found that the relative amount of aac(6’)-Ib(r=−0.19, P > 0.05), ermB(r=−0.02, P > 0.05), ermC(r=−0.27, P > 0.05), sul1(r=−0.32, P > 0.05), sul3(r=−0.10, P > 0.05), tetM(r=−0.10, P > 0.05), tetW(r=−0.15, P > 0.05) and blaTEM(r=−0.16, P > 0.05) slightly decreased with increased wind speed (Table 3). Moreover, the relative abundance of the mecA (r = 0.18, P > 0.05) gene was increased with elevated wind speed.

Aerosolised ARGs in different concentrations of ozone, NO2 and SO2

aac(6’)-Ib(r=−0.03, P > 0.05), sul3(r=−0.40, P < 0.05), tetM(r=−0.21, P > 0.05), tetW(r=−0.17, P > 0.05), blaTEM(r=−0.24, P > 0.05) and mecA(r=−0.16, P > 0.05) were negatively correlated with NO2. In addition, sul1(r=−0.12), sul3(r=−0.22), tetM(r=−0.18), blaTEM(r=−0.05) and mecA(r=−0.19) were negatively correlated with SO2, However, the ermB(r = 0.20, P > 0.05) and ermC(r = 0.22, P > 0.05) were positively correlated with NO2, and aac(6’)-Ib(r = 0.18, P > 0.05), ermB(r = 0.01, P > 0.05), ermC(r = 0.06, P > 0.05) and tetW(r = 0.10, P > 0.05) were positively correlated with SO2. In addition, the relative abundance of ermB(r=−0.13, P > 0.05), ermC(r=−0.46, P<0.01), tetM(r=−0.18, P > 0.05) and blaTEM(r=−0.07, P > 0.05) were negatively correlated with ozone concentration (data obtained from Deli Road). In contrast, the relative abundance of aac(6’)-Ib(r = 0.14, P > 0.05), sul1(r = 0.01 P > 0.05), sul3(r = 0.42, P<0.05), tetW(r = 0.31 P > 0.05) and mecA(r = 0.14, P > 0.05) were all positively correlated with ozone concentration.

Microbial taxa

The quantification of bacterial communities for different locations and seasons resulted in identification of 52 bacterial genera (Table S7). The dominant genera identified in the HV sites were Lactobacillus and Lactococcus. The main genera identified in spring and summer were Candidatus and Vibrio, whereas Lactobacillus and Lactococcus were the main species detected in autumn. Moreover, Lactobacillus, Staphylococcus and Lactococcus showed the highest relative abundance in winter (Fig. 3, Table S7).

Fig. 3
figure 3

Diversity and Relative abundance of bacteria in each sample. Note: A1 playground: non-green space (playground) LNV (Spring); A2 playground (summer); A3 urban heating island: UHI (summer); A4 high vegetation coverage (garden): HV (summer); A5 playground: (Autumn); A6 playground (Winter); A7 Intra Hospital.

Full size image

Discussion

In this study, the varied relative abundance of aerosolised antimicrobial resistance genes was determined in different types of ___location (LNV, HV and UHI), seasons, temperature, humidity, haze levels, wind speeds, and concentrations of PM2.5, ozone, NO2 and SO2 etc. We found that the relative abundance of aerosolised ARGs was different at different vegetation coverage sites. Seasonal variation of relative abundance of ARGs in air was determined and, we discussed the impact of various factors on the relative abundance of ARGs in air, showing that the relative abundance is multifactorial.

The relative abundance of the aerosolised ARGs during the study period ranged from 1.3413 × 10−4−47.57 (ARGs/16S rRNA)/m3. The lowest amount of ARGs was lower than that previously reported by Zhang et al. (2023) (3.85 × 10−3 (ARGs/16S rRNA)/m3)17 but in contrast, the highest amount of ARGs was higher than in Zhang’s report (0.014315 (ARGs/16S rRNA)/m3)17. We therefore hypothesise that Coriolis air sampler, used in our study but not in Zhang’s, is more sensitive for the detection of ARGs in air samples. This may well be due to the fact that it uses a cyclonic approach to collect and concentrate biological particles into a liquid. We showed that the relative abundance of blaTEM and mecA genes was significantly higher than that of other ARGs. Our findings were consistent with previous studies that reported the blaTEMgene to be the dominant ARG in the urban air of Tianjin and Beijing30,31. Moreover, blaTEM has also determined to be the most dominant ARG identified in the air of the inner hospital environment32. Although Adams et al. (2020) reported that the mecAgene could spread via air in the hospital environment33, no relative abundance of the mecA gene in aerosol in the urban air has been reported previously. To our knowledge, this is the first study to report the relative abundance of the mecA gene in urban air. The potential risk of dissemination of ARGs from hospitals to the surrounding areas by air is apparent due to blaTEM and mecAgenes being identified both in hospital and urban areas32,33. In this study, the average relative abundance of the eight ARGs detected in the playground was significantly elevated in the spring, surpassing levels observed in both autumn and winter. In addition, the average relative abundance of nine ARGs in the vicinity of the UHI site exhibited a significant decline during the non-heating season compared to the heating season. The relative abundance of blaTEM was significantly reduced with rising temperatures and increased humidity.

Our results showed that after the central heating was switched off, the relative abundance of aerosolised ARGs around the UHI site significantly decreased. In contrast, no significant reduction of the ARGs was observed in the aerosolised samples collected from the LNV and HV sites between the heating season (Nov to March) and non-heating season (April to Oct). Therefore, UHI sites may be a potential reservoir of ARGs in urban air during the heating season. Previous studies have shown that methicillin-resistant Staphylococcus aureus(MRSA) outbreaks in hospitals in the UK often occurred in October, when the central heating was switched on and hot air currents disseminated bacteria living in heating vents/behind radiators34; the findings of this study suggest this effect may also be happening at a citywide level. In addition during this time, released smoke dust may well act as carriers for ARGs14, and thus contribute to the increased amount of ARGs.

It has been reported that trees can reduce human exposure to particulate matter (PM) from traffic35,36. Although the relative abundance of ARGs in the HV was slightly lower than the UHI and LNV, the difference was not statistically significant; however, the reduction of ARGs in HVs may reduce human exposure to ARGs. Thus, planting trees, bushes and hedgerows may be an effective way to reduce the dissemination of aerosolised ARGs, though non-allergenic vegetation must be chosen, as the impact of allergens including pollen on human health need to be considered. We hypothesize that such phenomenon may be due to the ability of plants reduce the particle pollution in the air37, consequently reducing ARGs that were carried by dust in that air. We found that the Operational Taxonomic Unit‌ (OTU) abundance of Lactobacillus and Lactococcus accounted for more than 50% of the total bacteria in the HV site, which was higher than that identified in the UHI site and LNV site (Table S7). Lactobacillusis plant-associated lactic acid bacteria and is essential for the tolerance of plants against biotic and abiotic stress38. Moreover, it has previously been reported that Lactobacillus and Lactococcus spp.showed antibacterial properties against antimicrobial-resistant bacteria (ARB)39,40. Therefore, the dominance of Lactobacillus and Lactococcus in the air may inhibit the survival and propagation of ARB and thus contribute to the lowest relative abundance of ARGs in the HV areas.

The dominant ARGs subtype varied in different seasons. Quinolone (aac(6’)-Ib), macrolide (ermB, ermC), tetracycline (tetM, tetW), sulfonamides (sul1, sul3), and β-lactams (blaTEM, mecA) are antimicrobial drugs that are commonly used in the clinic and thus the presence of these free antimicrobial resistance genes aggravated the horizontal antibiotic-resistant gene transfer in the environment15. The ermB gene was the dominant ARGs in autumn, blaTEM was the most dominant gene in winter and spring, and sul3was the main ARGs that was detected in summer. This is consistent with previous studies, in which seasonal variations of aerosolised ARGs have also been observed18,41. Cyanobacteria, Candidatus and Vibiro accounted for 66% and 55% of total bacteria found in aerosolised samples of spring and summer respectively, however, the relative amount of Cyanobacteria, Candidatus and Vibiro reduced dramatically in autumn and winter samples. Candidatus and Vibirowere known to be carriers of multiple ARGs42,43. In contrast, Lactobacillus and Lactococcus were barely found in aerosolised samples of spring and summer, and the relative abundance of Lactobacillus and Lactococcus increased up to 26.5% and 21% respectively of the total bacteria detected in the aerosol samples of autumn and winter (Table S7). Increased Lactobacillus and Lactococcusin autumn and winter might be due to the changes of plant fermentation processes in autumn and winter44. The relative abundance of dominant bacterial species varied between seasons, consistent with previously published studies18 but the bacterial species found to be dominant during each season were different from those reported previously; this may be associated with the sampling procedure, i.e. the height and methods used.

In this study, blaTEM and ermCwere negatively correlated to temperature. The findings were consistent with previously published works which have also reported a negative correlation between ARGs and temperature45. Li et al. also demonstrated that temperature rise could affect the growth of bacteria, and thus result in a decrease in ARGs46. In this study, sul3 was significantly positively correlated with temperature, and blaTEM and humidity showed a significantly positive correlation. In general, when the temperature was high and humidity was low, the relative abundance of aerosolised ARGs was low but when the temperature and humidity were low, the relative abundance of ARGs gradually increased. When the humidity was high and the temperature was low, the relative abundance of ARGs increased dramatically. When the temperature was high and humidity was low, the relative abundance of the total ARGs gradually decreased. Therefore, attention should be paid to the monitoring of aerosolised ARGs in low temperature and high-humidity environments. The relationship of these climate conditions to the levels of ARGs present in the air should also be considered alongside emerging novel understandings of disease risk under different climate conditions, for example, high temperature and humidity are correlated with increased risk of septicaemia, anthrax, and black quarter outbreaks in livestock12. The levels of ARGs in the air may be influencing the levels of disease risk observed.

In contrast with a previously published study in which the authors demonstrated that wind speed reduced the relative abundance of ARGs45, no significant decrease of aac(6’)-Ib, ermB, ermC, sul1, sul3, tetM, tetW and blaTEMwas observed with increased wind speed. Bai et al. reported that wind promotes the transmission of aerosolised ARGs from hotspots to the surrounding environment47. Therefore, the increased relative abundance of the mecA gene in this study might be due to the wind-associated transmission from the hotspot. An air filter system for hot areas might be an effective way to eliminate the emission of ARGs from the hotspot into the surrounding environment.

A number of previously published studies reported that atmospheric particulate matter (PM) can be a carrier of aerosolised ARGs, and thus promote ARGs dissemination14,30,48. Moreover, it has been reported that PM2.5 can impact the dissemination of ARGs due to low levels of integron on hazy days49, which may explain the negative correlation between ARGs and PM2.5. However, no statistically significant association between PM2.5 and ARGs was observed.

It has been reported that ARGs may be destroyed by NO3−, and SO42–49, while other studies report that antimicrobial-resistant bacteria might acquire their nutrients from SO2 and NO2 30. In this study, we found that sul3 was significantly decreased with increased concentration of NO2. However, no statistically significant association between NO2 and other ARGs was observed. The oxidation effect of O3 might reduce ARGs in the environment50. In contrast with a previous study50, we found that sul3 significantly decreased with a reduced concentration of ozone. However, no statistically significant association between ozone and other ARGs was observed. Therefore, the influence of air pollutants on the dissemination of aerosolised ARGs was complex due to multiple factors that affected the relative abundance of aerosolised ARGs in combination as well as individually. Further studies are warranted to fully investigate the impact of these factors and their intersection on the dissemination of aerosolised ARGs (Table 3).

Although we observed that HV could reduce the amount of ARGs, further study is necessary to confirm this result due to the small sample size of this study. Moreover, the collected air samples in this study were approximately equal to the amount of air inhaled by an individual in a single breath51. In addition, we found an even higher relative abundance of ARGs/m3 compared with the traditional methods (i.e. PM sampler), which suggested that the Coriolis air sampler may well be a more effective method for studying ARGs in the air.

Conclusion

In conclusion, compared with autumn and winter, the average relative abundance of the ARGs detected in the playground was significantly elevated in spring. Moreover, we observed a reduction of ARGs in high vegetation coverage sites. Therefore, planting non-allergenic trees, bushes, and hedgerows may be an alternative way to reduce the dissemination of ARGs through the air. We also identified that the relative abundance of aerosolised ARGs around the UHI site was significantly reduced when central heating was switched off.The relative abundance of blaTEM demonstrated a significant reduction with rising temperatures, yet experienced an increase with the elevation of humidity. The consideration of the impact of different environmental and climate conditions, and the use of climate change risk mapping tools to predict increased risk during specific weather conditions, could help public health offices to plan communication campaigns and predict when increased cases of resistant infections are likely to occur.