Introduction

Air pollution, a major hazard in residential environment, is estimated to cause ~ 9 million premature deaths worldwide annually1. Air pollution caused by a mixture of gases and particulate matters in the air, such as NOx, O3, PM2.5, PM10, pollen and spores2. The interaction among these air pollutants coexisting in the same medium may increase their adverse impacts due to synergistic effects3. Allergenic pollen is the main cause of seasonal allergic reactions, including those caused by hay fever, rhinitis, and asthma, which may be aggravated by air pollution4,5,6. SO2 causes increased airway inflammation and altered allergen expression7. O3 has considerable effects on pollen and its allergenicity8.

Allergic rhinoconjunctivitis (AR) induced by allergenic pollen considerably decreases quality of life, reduces productivity and exacerbates asthma, affecting 30% of the population of which half are misdiagnosed or poorly treated9,10,11,12. Hence, monitoring airborne allergenic pollen is becoming an increasingly important concern for public health. Compared with over 500 airborne pollen monitoring stations owned by the European Aeroallergen Network (EAN) in Europe and 71 pollen monitoring stations established by the National Allergy Bureau (NAB) in the United States13,14, very few stations use Hirst-type volumetric trap have been reported in China15. Moreover, there are very few epidemiological and aerobiological studies in Beijing which supports almost 22 million population.

Almost 100 types of pollen-allergic plants are reportedly distributed within the Fifth Ring in the urban area of Beijing16. Cupressaceae plants which blooms in spring are cultivated as ornamentals in the urban area in Northern China17,18,19. Similar to anemophilous plants, most plants in the Cupressaceae family can produce abundant pollen grains20. Damialis et al. (2011)21 estimated that billions of pollen grains can be produced by a typical cypress tree (Cupressus sempervirens). Moreover, Cupressaceae pollen is small and tends to fracture22, facilitating the release of allergen protein from pollen grains, considerably increasing allergic risk in the surroundings. In many parts of the world, Cupressaceae pollen is considered to be an important component of aeroallergens20,23, posing a potential allergy risk that may be elevated by air pollution24,25,26,27.

The effects of air pollution should be considered when the effect of pollen allergens on the prevalence of allergic diseases is evaluated28,29,30. In Europe, people living in the urban area are more allergic to pollen-induced respiratory allergies compared to those living in rural areas, even though a higher level of allergenic pollen concentration is observed in the rural area31,32. Air pollutants facilitate the release of pollen allergen through thinning and rupturing exine layers, and chemically modify allergenic proteins, act as adjuvants inducing IgE responses33,34,35.

For instance, compared with pollen collected in the rural areas, the proteome of Betula pendulapollen collected in urban area has been modified by air pollution8. NO2 and O3 can facilitate the release of aeroallergen Pla a3 by damaging the membranes of pollen cells in subpollen particles, and largely aggravate pollen-induced pneumonia36. Similarly, such structural changes in Artemisia pollen induced by air pollutants might increase the residence time of pollen allergens in the atmosphere, raising allergic risks in a pollen-allergic population37,38. Some fine particles attach on the pollen grains and allergens39,40, contributing to lung inflammation41.

To the best of the authors’ knowledge, most studies used aerobiological data and pollution data obtained from different monitoring stations, and the aerobiology, pollution and meteorology monitoring stations were several kilometers apart. This induces errors in determining the true variation in high-resolution analysis owning to the spatial heterogeneity of aerosol sources. Particularly, some pollen allergens or air pollutants vary more over smaller spatial scales, yielding more errors42. Considering the error likely caused by this method, three-year synchronous data at four sampling sites are used herein, and pollen, meteorology, and pollution were monitored at the same time and ___location. Furthermore, all the studies used daily pollen concentrations, rather than the finer temporal data, which did not increase data precision2,3,43,44,45,46. It is possible that pollen peaks in diurnal variation during high-pollen episodes are more important than daily averages. Therefore, both daily and hourly values for pollen, pollution, and meteorology were used to increase data precision in the present study. Besides, previous studies on the influence of air pollutants on airborne pollen indicated that pollen response must be examined at the taxon level, rather than the assemblage level45,47.

The key objectives of the current research were to untangle the relationships between the major allergenic pollen (Cupressaceae) and air pollutants during spring in the urban and suburban areas of Beijing, and discuss the influences of meteorological variables on allergenic pollen and air pollutants at different timescales. This is the first study aiming at the interaction between air pollutants and Cupressaceae pollen which is the major cause for spring pollen allergies, and their association with meteorological variables in Beijing city.

Methods

Research area

Beijing is the largest city in Northern China with a population of 21.88 million. It is located in an area with continental monsoon climate which is also typical in other regions of Northern China. During 2016–2018, the average temperature was 26.9 °C in July and − 3.2 °C in January, and the average annual precipitation was 550.3 mm in the monitoring stations. In this research, four monitoring stations were utilized, from the north to the south (Fig. 1). These include the Mangshan (MS) station (40.27°N, 116.28°E, Mangshan National Forest Park), which is located in the northern suburban area, Research Center for Eco-Environmental Sciences (RCEES) (40.01°N, 116.34°E) and Haidian Park (HP) (39.99°N, 116.29°E) stations that are localized within the Fifth Ring in the northern urban area, and Beijing Teaching Botanical Garden (BTBG) station (39.88°N, 116.43°E) that is located in the Second Ring in the southern urban area (Fig. 1). Figure 1 was generated by Google Earth 7.1.8.3036 (32-bit).

Fig. 1
figure 1

Location of pollen monitoring station (red point), meteorology monitoring station (yellow box) and pollutant monitoring station (green box) in the MS, RCEES, BTBG and HP station in Beijing, China.

Full size image

Data collection and process

Airborne pollen data

A Burkard seven-day volumetric spore trap was used at each station to monitor daily and hourly airborne pollen concentrations during 2016–2018. The Burkard spore trap is a wind-oriented, time-discriminated Hirst-type sampler. Air is sucked in through a 2 mm × 14 mm orifice at a rate of 10 L min−1. Airborne particles are deposited on a tape mounted on a drum. The drum is driven by a clockwork mechanism, enabling it to rotate 2 mm every hour, thus ensuring continuous air sampling and the provision of both hourly and daily data for an entire week.

For daily pollen data, 24-hour segment (length: 48 mm) was cut out from the seven-day samples. Each segment was mounted on a glass slide dripped with glycerin and then covered by a cover slip. Then five continuous horizontal sweeps over the whole slide were selected, and pollen grains were identified at 40 × 10 magnification using the Leica optical light microscope. Daily pollen concentration (grains m−3 d−1) was calculated using the Eq. (1).

$$\:PC\:\left(grains\:{m}^{-3}\:{d}^{-1}\right)=\:\frac{n\:\times\:\:14}{5\:\times\:\:14.4\:\:\times\:\:0.61}$$
(1)

where the n is the number of pollen grains counted, 14 (mm) is the width of the orifice, 5 is the number of sweeps, 14.4 (m3) is the volumes of air for a day, and 0.61 (mm) is the diameter of the field of vision in this study.

For hourly pollen data, 24 vertical sweeps (width: 2 mm for 1-hour segment) over 1-day sample were selected, then pollen grains for each hour were identified at 40 × 10 magnification. Hourly pollen concentration (grains m−3 h−1) was calculated using the Eq. (2).

$$\:PC\:\left(grains\:{m}^{-3}\:{h}^{-1}\right)=\:\frac{n\:\times\:\:2}{1\:\times\:\:0.6\:\:\times\:\:0.61}$$
(2)

where n is the number of pollen grains counted, 2 (mm) is the width of each segment, 1 is the number of sweeps, 0.6 (m3) is the volumes of air for one hour, and 0.61 (mm) is the diameter of the field of vision.

The sum of daily pollen concentration during the entire pollen season is defined as the seasonal pollen index (SPI). The days exceeding the 80th percentile of daily Cupressaceae pollen concentration within the whole season of 2018 were selected to analyze the Cupressaceae pollen diurnal distribution. Hourly Cupressaceae pollen concentrations were standardized within 24 h. Both the peak (maximum value) and secondary peak which refers to the value exceeding 90% of the maximum value, were used as the occurrence of hourly peaks to estimate the frequency of hourly peaks over 24 h.

Meteorology and air pollution data

Local meteorological variables, including temperature (T), relative humidity (RH), rainfall (RF), wind speed (WS), and wind direction (WD), were collected at each monitoring station. The Beijing Urban Ecosystem Research Station provided the meteorology data at the MS, RCEES, and BTBG stations and the air pollution data (NOX, O3, PM2.5) at the MS and BTBG stations, while the Meteorological Bureau of Haidian District collected the meteorology data at the HP station. At each monitoring station, pollen data, meteorology or/and pollution data were collected at the same ___location.

Vegetation data

Field survey was conducted in the surroundings with a 1000-m radius (MS: 2000 m) at each sampling site. Trees were recorded in a 10 m × 10 m plot. For each plot, the number of individuals, diameter at breast height and canopy diameter were measured for each species. For trees, 250, 308, 184, and 69 plots were recorded at the BTBG, HP, RCEES and MS station, respectively. The field survey was only conducted in the east and south as a cliff is present on northern and western edges on the top of Mangshan mountain. The coverage (%) map for Cupressaceae plants at each site was generated in 10 m × 10 m.

Modeling

Backward trajectory analysis

The days exceeding the 80 th percentile for daily Cupressaceae concentration within the whole season during 2016–2018 were selected for backward trajectory calculation. The HYSPLIT model was used to generate air back-trajectories48. Herein, 48-h back-trajectories arriving at the urban (BTBG) and suburban (MS) stations were generated at UTC 00:00, 04:00, 08:00, 12:00, 16:00 and 20:00 (height: 500 m above ground level).

Concentration weighted trajectory (CWT) model

The CWT value for a grid cell is calculated using the daily pollen data weighted by all the daily concentrations corresponding to the trajectory points passing through this grid cell. It is represented as

$$\:{C}_{ij}=\frac{1}{\sum\:_{k=1}^{N}{n}_{ijk}}\:\cdot\:\:\sum\:_{k=1}^{N}{C}_{k}\cdot\:{n}_{ijk},$$

where \(\:{C}_{ij}\) represents the CWT value in the ijth grid cell, k represents the serial number of the trajectory, N represents the number of trajectories, \(\:{C}_{k}\) represents the true daily pollen concentration recorded on the day when trajectory k arrives, and \(\:{n}_{ijk}\) represents the number of trajectory points within trajectory k passing through the ijth grid cell. The TrajStat software was used to compute CWT49. The CWT values were multiplied with the weight function W(i, j), shown as

$$\:\text{W}\left(i,\:j\right)\left\{\begin{array}{c}1.00,\:\:\:\:{n}_{ijk}>80\\\:0.70,\:\:\:20<{n}_{ijk}\le\:80\\\:0.42,\:\:\:10<{n}_{ijk}\le\:20\\\:0.05,\:\:\:{n}_{ijk}\le\:10\end{array}\right.$$

Results and discussion

Cupressaceae pollen season and high pollen-induced allergic risk in Beijing

SPI is an important parameter that quantifies pollen production. Furthermore, it is an indicator representing the intensity of an allergenic pollen season4,50. In the urban area, Cupressaceae plants, represented by Juniperus chinensis (L.), were mainly distributed in the dominant wind directions that prevailed during the pollen season (March), except for the RCEES station (Fig. 2). The annual variation of the wind frequencies of dominant directions in three urban stations coincided well with the annual variation in SPI (Table S1), which was consistent with the findings on other pollen taxa reported by Rojo et al. (2015)51. Particularly for trees, it is reported that allergenic tree species contribute to a majority of the allergenic pollen in the residential green space of urban areas52,53. Therefore, greenness exposure in residential areas is a rising risk for AR54.

Fig. 2
figure 2

Left panel shows vegetation coverage (%) distribution map of Cupressaceae plants within 500 m at the (a) BTBG, (c) HP, (e) RCEES stations and 1000 m at the (g) MS station. Right panel shows wind rose charts for direction and speed at the (b) BTBG, (d) HP, (f) RCEES and (h) MS stations during Cupressaceae pollen season (March) of 2016–2018.

Full size image

Apart from meteorological factors, the amount of arboreal pollen is largely determined by local tree distribution and the distance from the source plants to the receptor site55,56,57. The results revealed that local vegetation distribution, coupled with local wind direction may be associated with temporal changes in SPIs, particularly when source plants are distributed along the dominant wind directions during the whole pollen season. However, this finding implied that high annual pollen levels may be largely attributed to an inappropriate plantation of allergenic trees in urban areas, posing a potential health risk for allergy sufferers living in Beijing city.

In terms of epidemiological studies, daily pollen concentrations monitored at each site are commonly used to characterize pollen exposure. The severity of allergic symptoms induced by pollen allergens is strongly associated with allergenic pollen concentration, mainly depending on the specific threshold of allergic response in a susceptible individual58,59. Herein, the mean daily concentrations and peak values for Cupressaceae pollen varied from 19 to 102 grains m−3, and from 331 to 892 grains m−3, respectively (Table S1). Yao and Zhang (2009)60 reported that Cupressaceae pollen concentration in Beijing was related to the severity of allergic symptoms (r = 0.804, p < 0.01), which was classified into four levels, ranging from 0 to III (Table S2). Patients under level II and level III exhibit obvious symptoms adversely affecting normal life (16–66 grains m−3) and display serious allergic symptoms clearly influencing normal life and sleep (≥ 67 grains m−3), respectively. In Beijing, in level II and level III, 12 days and 17 days were observed on average in the urban and suburban areas (Table S2), accounting for 38.1% and 53.1% of the pollen season length, respectively (Fig. S1). Notably, Cupressaceae pollen in Beijing has become an important biological constituent in the atmosphere, posing an urgent health threat to over 20 million inhabitants in Beijing.

Correlation between concentrations of Cupressaceae pollen and NOx, O3 and PM2.5

Considerable differences were observed in NOx and O3 between the urban (BTBG) and suburban (MS) stations. Compared with the suburban station, urban station exhibited higher concentrations of NOx and lower concentrations of O3 (Fig. S2). It is reported that relatively high NOxconcentration was observed in the urban area mainly due to intensive industrial and traffic emissions61,62. To understand the relationship between the variation of Cupressaceae pollen and air pollutants at a daily timescale, NOx, O3 and PM2.5 daily concentrations were revalued according to four allergic levels for daily Cupressaceae pollen concentration (Fig. S2). Concentration of PM2.5 at the urban station displayed an obvious enhancement when pollen concentrations were increased, whereas PM2.5 concentrations at the suburban station showed no significant changes when pollen concentration rose (Fig. S2d). In the suburban area, daily O3 level presented an increasing trend as pollen concentrations were increased (Fig. S2c). Spearman correlation analysis indicated that daily concentration for Cupressaceae pollen was found to be positively related to NOx (r = 0.32, p < 0.01) and PM2.5 (r = 0.29, p < 0.01) at the urban sampling site as well as O3 (r = 0.20, p < 0.05) at the suburban site (Fig. 3a). Airborne pollen is positively associated with atmospheric pollutants, especially CO and O3, while is negatively correlated with NO22,63. PM2.5 is reported to be significantly associated with daily pollen concentration2,45.

Nevertheless, few studies discussed the correlation between hourly concentrations for pollen and air pollutants. In the present study, at an hourly timescale, pollen concentration was positively correlated to the concentrations of NOx and PM2.5 at both the urban and suburban stations. At the suburban station, there was positive correlation between hourly pollen concentrations and hourly O3 concentration, which was consistent with the relationship at a daily timescale (Fig. 3b). O3 can change allergenicity of allergen pollen due to its promotion on nitration of allergen proteins and post-translational modification64,65,66. O3 is found to be a key factor resulting in enhanced allergenicity of Betula pollen, and co-exposures of O3 and pollen can make increasing IgE responses triggered in allergic population32,67,68.

Fig. 3
figure 3

Spearman correlation coefficients among (a) daily and (b) hourly concentrations of Cupressaceae pollen, air pollutants (NOx, O3, PM2.5) and meteorological elements (mean air temperature, relative humidity and wind speed) in the urban (upper triangular matrix) and suburban (lower triangular matrix) area. *, p < 0.05; **, p < 0.01.

Full size image

It is reported that compared with rural areas, pollen sensitization rate is higher in urban areas, especially in the residential areas along the main roads, even though lower pollen concentration is observed in the highly urbanized area28,29,69. Beyond that, compared to rural pollen extracts, chemotactic activity on human neutrophils of the pollen extract in the urbanized area is higher70. Some researches indicate that pollen grains can be influenced by air pollutants, including the structure, stability and allergenicity of allergen proteins, ultimately its allergic effect5,6,71,72,73,74. Recent findings show that inflammatory induced by pollen allergen and cellular immune responses could be enhanced, and protein expression could be altered by co-exposure of pollen allergens and gaseous pollutants75,76. Hence, the impact of atmospheric pollutants on allergenic pollen must be considered in urban areas.

Influence of meteorological factors on pollen season

In Beijing, Cupressaceae pollen season mainly started in the middle of March and ended in early April, maximum daily concentration appeared in late March or early April (Table S1). Our results showed that the pollen season start (PSS) varied from year to year. Such variability, largely depends on meteorological conditions before pollen season, especially the temperature influencing flowering timing and intensity77,78,79,80,81.

We found that an earlier start and a longer duration of Cupressaceae pollen season in 2017 when compared with 2016 and 2018, which may be due to warmer spring (February) in 2017 (urban: 2.2 ~ 3.6 ℃, suburban: −0.8 ℃), and colder spring in both 2016 (urban: 1.3 ~ 2.4 ℃, suburban: −2.6 ℃) and 2018 (urban: −0.8 ~ 0.4 ℃, suburban: −4.2 ℃) (Fig. S3, Table S1). These facts are in accordance with previous reports which indicate that enhancement in temperature frequently results in acceleration of plant development, leading to an early start of flowering phenology82,83, consequently an early arrival of pollen season84,85.

Moreover, suburban-urban and intra-urban differences in pollen season were also identified in this study. Compared with the suburban station, start dates of pollen season at the urban stations appeared earlier. It was probably attributed to earlier flowering timing of Cupressaceae plants caused by suburban-urban temperature variations. Mean temperatures of the urban area in the month (February) prior to the pollen season (February: −0.7 ~ 3.6 ℃) were evidently higher than those in the suburban area (February: −4.2 ~ −0.8 ℃) in Beijing (Fig. S3, Table S1). Beyond that, intra-urban variation of March temperatures in the occurrence of pollen peak were observed within a metropolitan area, even though the difference was no more than 1 °C (Table S1).

Pollen peak appeared earlier at the RCEES station than at the BTBG and HP stations, likely due to their differences in March mean temperatures (RCEES: 9.3–10.1 ℃, BTBG: 8.8–9.9 ℃, HP: 8.3–9.1 ℃). This was in line with the findings reported by Mimet et al. (2009)86and Katz et al. (2019)13 who demonstrate that inter-cities variations in temperature can cause differences of flowering phenology and pollen release timing in inter-neighborhood regions. This study further confirmed that mismatches in flowering phenology lead to systematic heterogeneity in airborne pollen counts within urban areas probably due to intra-urban temperature variations.

Impacts of meteorological factors on daily and hourly concentrations of Cupressaceae pollen

Temperature is the most critical meteorological variables in determining pollen concentrations by the way of governing the phenology and strength of flowering87,88. In this study, daily variation in pollen concentration corresponded well with thermal conditions (Fig. S4). Urban pollen levels displayed a tendency to increase when temperature increased from a low level. Pollen peak emerged before the appearance of maximum temperature of a day. Pollen concentrations dramatically decreased in response to decreases in temperatures after the maximum temperature, and dropped to a lower level when temperature fell from the peak. Compared with three urban stations, an elongated pollen season together with increased pollen concentration were monitored at the suburban station when encountering a secondary warming period. This indicated that temperature exerted a strong control on airborne pollen concentration by influencing pollen release and re-suspension mechanisms beyond the peak-day87.

Relative humidity reduces the buoyancy of atmospheric pollen grains by increasing their weight, thereby favoring their agglutination and deposition89,90. When high-humidity conditions occur, pollen grains would become heavier and inflate, and finally fall down due to their harmomegate characteristic91. High-level pollen concentrations (109 ~ 687 grain m−3) corresponded to a dry condition coupled with low relative humidity (25.8 ~ 41.8%), and low-level pollen concentrations (1 ~ 22 grain m−3) corresponded to higher relative humidity (46.1 ~ 52.5%) (Fig. S4). In our study, light rainfall (0.1 mm per day) occurring in the peak day exhibited no significant effects on daily pollen concentrations (Fig. S4). Although the anthers must be shaken to release pollen grains by wind forces92, high wind speed can dilute particles floating in the air87,93. Our data showed that the highest wind speed corresponded to very low pollen concentrations (Fig. S4). When pollen concentrations maintained at high levels, wind speeds were low (Fig. S4).

Diurnal analysis showed that hourly Cupressaceae pollen concentration peaked during 9:00 ~ 11:00 at the BTBG, RCEES and MS stations (Figs. S5a, 5c, 5 d), and during 5:00 ~ 7:00 at the HP station (Fig. S5b). Moreover, high frequencies of peaks and secondary peaks fell during 8:00 ~ 11:00 at both the urban and suburban stations (Fig. S5e). These are in accordance with other authors who observe that the start of pollen emission from the anthers always varies in the first three hours after sunrise94, thereby pollen concentration usually sharply increases between 8:00 and 11:0090,95,96.

Variation in diurnal peak of pollen concentration varies due to considerable changes in weather conditions within the course of a single day. We observed that hourly pollen concentration increased when temperature was increasing and relative humidity was decreasing after the sunrise, and reached the diurnal peak in the first a few hours after sunrise (8:00 ~ 11:00) (Fig. 4). This is consistent with previous studies that have indicated that raise of temperature and decrease of relative humidity are two important requirements for anther opening (dehiscence) as well as the key step for pollen release process97,98. Cupressaceae pollen can be produced by pollen sacs covered by the male cone scales22. Cone scales interlock closely with each other under humid conditions, pollen release under dry conditions when cone scales are separated99. An external force is required to extract pollen grains from adhesion to anther wall so as to transport pollen grains upwards once anthers are open100,101. Therefore, very low wind speed in the morning may result in a delay of pollen peak101. However, pollen concentrations were dramatically reduced to low levels when wind speed increased (Fig. 4).

Fig. 4
figure 4

Variation of hourly Cupressaceae pollen concentration and meteorological factors (T, RH, WS, RF) at the (a) BTBG, (b) HP, (c) RCEES and (d) MS stations during the pollen season in 2018.

Full size image

Effects of meteorological factors on daily and hourly variability of air pollutants

In addition to emission conditions for air pollutants, meteorological factors significantly affect air pollutants by the means of physical and chemical processes along with chemical conversion and dry and wet depositions102,103,104. In order to investigate the effects of meteorological variables upon NOx, O3 and PM2.5, boxplots of air pollutants within 5 ℃ in air temperature and 10% in relative humidity during the pollen season were performed at the urban and suburban stations separately (Figs. S6, S7). Daily and hourly concentrations of NOx and PM2.5obviously increased with an increase of 10% in relative humidity at both the urban and suburban stations (Figs. S6f, S7f). It is reported that high concentrations of particle matter are associated with high relative humidity in Beijing102,105. A rising trend in NOx was found at both the urban and suburban stations as the relative humidity rose (Figs. S6d, S7 d). Spearman correlation analysis indicated that relative humidity was positively correlated with daily and hourly concentrations for PM2.5 and NOx (Fig. 3), which are in accordance with previous reports62,102. Among the meteorological variables except for relative humidity, temperature and other variables displayed little influence on PM2.5concentrations in Beijing102. Dissolution of gaseous precursors in wet aerosols and aqueous reactions were facilitated by high relative humidity, which largely accelerates the evolution of particulate matter pollution106.

The formation of O3 largely depends on solar radiation and temperature. Our results indicated that O3 level corresponded with a higher temperature at both the urban and suburban stations (Figs. S6b, S7b). Significant positive correlation was identified between temperature and daily and hourly concentrations of O3 at the urban (r = 0.26, p < 0.05; r = 0.75, p < 0.01) and suburban (r = 0.19, p < 0.05) stations (Fig. 3). Moreover, hourly concentration of O3 at the urban station evidently decreased with increasing relative humidity (Fig. S6e). And urban O3 was negatively related to relative humidity at daily (r = − 0.25, p < 0.05) and hourly timescales (r = − 0.76, p < 0.01) (Fig. 3). These findings are in line with previous studies researches61,107. Temperature have significant effects not only on chemical reaction rates which are linked to O3 generation, but also on emission rates of its precursors, including biogenic VOCs108,109. In addition, water vapor reacts with O3 in the atmosphere, leading to loss of O3110,111. High humid conditions slow the photochemical process, leading to reduction of cloud abundance and solar radiation, thereby inhibiting photochemical reactions in the course of O3 formation61,112. Furthermore, hourly concentration of O3 increased with increasing wind speed (Fig. S7 h), and a positive correlation (r = 0.54, p < 0.01) was found between them O3 (Fig. 3b). This may be due to the negative correlation (r = − 0.34, p < 0.01) between wind speed and NOx which is an important precursor of O3, as there was less NOx titration to reduce O3 under higher wind speed61.

Source areas of Cupressaceae pollen and air pollutants

In order to compare the local source area of allergenic pollen and air pollutants, more precise data monitored at the same station during 24 hours in the days exceeding the threshold of daily pollen concentration was used to evaluate the influence of local wind direction on the dispersion of allergenic pollen and air pollutants at an hourly timescale. At the urban station, southeasterly winds prevailed within the high-risk pollen-allergy episode with a frequency of 56% (Fig. 5). Nevertheless, easterly and southeasterly wind directions were associated with high concentrations of Cupressaceae pollen. Furthermore, as shown in Fig. 2, Cupressaceae plants at the BTBG site were distributed in the eastern and southeastern areas within a 500-m radial circle surrounding the receptor site. Differences in the amounts of pollen for any given geographical ___location are possibly due to the distances between receptor site and pollen source plants, together with wind directions13,51,113. The results of this research are in accordance with previous reports which emphasized the major impacts of local pollen sources on airborne pollen counts114,115,116.

Interestingly, NOx and PM2.5 may have the same local source area with Cupressaceae pollen at the urban station (Figs. 5a, c, g). Unlike the urban station, local source areas for pollen at the suburban station were different from the source for air pollutants. Southwesterly and westerly wind directions were dominant with a frequency of 28% and 23%, respectively, which were also corresponded to high concentrations of NOx, O3 and PM2.5 (Figs. 5d, f, h). Southwesterly and westerly wind directions may significantly influence local pollen dispersion, which led to pollen peak within a day (Fig. 5b). We found that high emission of O3 was associated with the dominant wind directions at both the urban and suburban stations (Figs. 5e, f).

Fig. 5
figure 5

Hourly concentration rose chart for (a, b) Cupressaceae pollen, (c, d) NOx, (e, f) O3, (g, h) PM2.5 during the peak days in the pollen season of 2018 in the urban (a, c, e, g) and suburban (b, d, f, h) areas.

Full size image

Cupressaceae, all species are wind-pollinated with a high ability for pollen production. Cupressaceae pollen in North American regions have been cited as an evidence for long-distance transport117. In order to identify the regional source areas of Cupressaceae pollen and air pollutants during the pollen peak days in the urban and suburban areas, the concentration-weighted trajectories were performed using the 48-hour backward trajectories during the peak days which exceeded the threshold of daily pollen concentration (Fig. 6). In the urban area, a total of 144 and 162 back trajectories was calculated for CWT at the urban and suburban stations, respectively. In the urban area, major potential source area for allergenic pollen and O3 both included western and northern regions of Hebei Province and middle region of Inner Mongolia (Figs. 6a, e). Moreover, potential source areas for NOx and PM2.5 both contained eastern region of Hebei Province and middle region of Shandong Province (Figs. 6c, g). In the suburban area, southern region of Hebei Province and northwestern region of Shandong Province were major source areas for Cupressaceae pollen (Fig. 6b). However, major potential sources for air pollutants were different from those for Cupressaceae pollen, including eastern regions of Hebei Province and Tianjin city (Figs. 6b, d, f, h).

Fig. 6
figure 6

Concentration weighted trajectory (CWT) results of concentrations for (a, b) Cupressaceae pollen, (c, d) NOx, (e, f) O3 and (g, h) PM2.5 using 48-hour back-trajectories at the (a, c, e, g) urban and (b, d, f, h) suburban station. Darker colors represent greater potential source.

Full size image

Mechanism of interactions between Cupressaceae pollen and air pollutants

This study elucidates the dual mechanisms underlying the correlations between Cupressaceae pollen and air pollutants (NOx, O3, PM2.5) through analyses of transport pathways and biogenic emissions.

In the urban area, significant positive correlations were observed between the concentrations for Cupressaceae pollen and NOx (p < 0.05) and PM2.5 (p < 0.01) at both daily and hourly scales (Fig. 3). These findings suggest a local-scale transport mechanism, where southerly, southeasterly, and easterly winds (Fig. 5a, c,g) facilitate the co-dispersion of pollen grains and pollutants along shared atmospheric pathways.

Conversely, in the suburban area, Cupressaceae pollen exhibited stronger positive correlations with O3 (p < 0.05) at both temporal scales (Fig. 3). This pattern is attributable to biogenic emission mechanisms. In the suburban area, dense Cupressaceae plantations and higher biogenic volatile organic compound (VOC) were emitted from woody vegetation, which undergo photochemical reactions to form O3118,119. Elevated O3 concentrations in the suburban area3 (Fig. S2) further supported the hypothesis of synchronized emissions of pollen and O3. Notably, lack of dense traffic networks in the suburban area, largely reduced NOx emission and highlighted the dominance of biogenic-O3 formation.

There are some other possible relationships between allergenic pollen and air pollutants. It is reported that some fine particles attached on the pollen grains and allergens modifying their allergenicity39,40, leading to anomalies and altering morphological structure in pollen grains120,121, such interactions do not directly influence pollen concentration levels122.