Yafei Wang · Zhuobiao Ni · Di Wu · Chen Fan ·Jiaqi Lu , · Beicheng Xia

Abstract Negative air ions (NAIs) benefit the mental and physical health of humans, but rapid urbanization can decrease the abundance of NAIs. Quantifying the spatial and seasonal distribution of NAIs and determining the factors that inf luence the concentration during urbanization is thus essential. In the present study of a typical developing urban district in southern China, negative air ion concentrations (NAICs) in 60 forests sites and 30 urban green spaces were quantif ied on seven consecutive days in each of the four seasons. Large seasonal variations in NAIC were revealed in forests and urban green spaces with trough values in summer. NAIC progressively decreased from forests to urban green spaces and was inf luenced by local land morphology,vegetation characteristics, and climatic factors. The vast,heavily vegetated northeastern region was the richest area for NAIs, whereas the narrow central region (urbanized area)was the poorest, implying dramatic impacts of urbanization on the spatial distribution of NAIs. The relationship between air temperature and NAIC was better f itted with a quadratic equation than a linear equation. Moreover, the NAIC was more sensitive to local morphology in urban green spaces than in urban forests, indicating the vulnerability of NAIs in urbanized areas. Therefore, the appropriate design of local urban morphology is critical.

Keywords Forests · Negative a ir i on c oncentration ·Urban g reen s paces · Shenzhen

Introduction

Negative air ions (NAIs) can absorb dust particles (Daniell et al. 1991), reduce fog and smoke (Fletcher et al. 2007;Sawant et al. 2012), and decrease bacteria and pollen(Kawamoto et al. 2006; Fletcher et al. 2007). Importantly,several studies have shown that exposure to negative air ions can prevent and cure respiratory disease (Alexander et al. 2013), strengthen physical functions (Bowers et al.2018), enhance the immune system (Shal’nova 1994;Kondrashove et al. 2000), improve sleep quality (Perez et al. 2013), and relieve fatigue (Perez et al. 2013), thereby enhancing human mental and physical health. The process of forming NAIs is called air ionization, which can be caused by cosmic radiation, ultraviolet rays, radioactive materials, coronal discharge or lightning (Krueger and Reed 1976; Yates et al. 1986; Iwama 2004). In addition,vegetation can generate NAIs through the point discharge and photoelectric effect during photosynthesis in plant cells and respiration in plant roots (Liang et al. 2014; Li et al. 2015) and increase the concentration of NAIs by releasing volatile substances (Huang et al. 2004)). High levels of NAIs are also found near waterfalls, lakes, and streams (Pino and Ragione 2013), and anthropogenic sources such as fuel combustion, high-voltage discharge,and hot surfaces also produce NAIs (Fialkov 1997; Maricq 2006; Shahsiah et al. 2008; Ling et al. 2010).

Rapid urbanization often leads to the degradation of urban environmental quality and impacts the production and concentration of NAIs (NAIC). Over the past decades, numerous investigations have been carried out on the spatial and seasonal distribution of NAIs (Kosenko et al. 1997; Liang et al. 2014; Yan et al. 2015) and inf luential factors, including natural factors (e.g., temperature,humidity, wind, weather, vegetation coverage, biodiversity, water) (Reiter 1985; K?hler et al. 2001; Yan et al.2016; Miao et al. 2018) and human factors (e.g., people and vehicles, building materials, industrial pollution)(Tikhonov et al. 2004; Wang and Zhuang 2012). However,these studies mainly concentrated on a single region, for example, forest (Liang et al. 2014), health resort (Li et al.2013), rural area (Pawar et al. 2010), or urban area (Yan et al. 2015), while only a few focused on variations in NAIs among multiple land-use types (Pawar et al. 2012;Li et al. 2015), and the conclusions regarding the effects of the factors were contradictory (H?rrak et al. 2003; Xu et al. 2008; Pawar et al. 2010; Li et al. 2015), so comprehensive data from different land-use types and seasons has been needed.

For exploring the changes in NAIC and distribution of NAIs under urbanization in the present study, we selected a typical developing urban district in southern China(Dapeng New District, Shenzhen) that is experiencing thriving urbanization and encompasses a large area of well-protected forestry land. Using on-site measurements at 60 sites in forests and 30 sites in urban green spaces during 1 week for each season, we compared the spatial and seasonal distribution of NAIs between the forestry land and the urban green spaces and quantif ied the impacts of urbanization on NAIC. Moreover, the relationships between the inf luencing factors and NAIC were explored by studying vegetation characteristics, climate factors and local morphology of the study sites to better understand the capacity of different land-use types to produce NAIs.

Materials a nd methods

Study ar ea

The Dapeng Peninsula, an official New District in southeastern Shenzhen, China, is a green belt planned in 2011 as an expansion of Shenzhen (latitude 22°38′32.31″, longitude 114°24′40.87″). With a subtropical maritime climate,it has wet, scorching summers and mild, sunny winters. The average air temperature f luctuates between 18 and 22.4 °C.Sunshine is plentiful all year; the mean annual solar radiation is approximately 5225 MJ m -2 , with 11—13 h per day.The subtropical monsoon brings abundant rainfall with a long rainy season from April to September (Meteorological Bureau of Shenzhen Municipality 2015).

With the sea on three sides, Dapeng Peninsula covers a land area of 294 km 2 with a high altitude to the southeast and low altitude to the northwest. Its three main subdistricts(Dapeng, Nan’ao, Kwai Chung) comprise 25 neighborhoods,but 73% of the total study area is forestry land (approximately 214 km 2 ) with 202 km 2 of trees. Shrub lands comprise approximately 11 km 2 and other forest lands comprise 0.4 km 2 . Shenzhen-Dapeng Peninsula National Geological Park also lies within the district (Statistics Bureau of Shenzhen City 2015).

Measurement me thodology and ins trumentation

In total, 90 sites were selected for the NAIC measurements:60 in forests and 30 in urban green spaces (Fig. 1). The sites in forests were evenly distributed in f ive subdistricts including Qi’niang Mountain, Central Corridor, Paiya Mountain,Bijia Mountain and Honghua Mountain, while the sites in green spaces comprised eight from Kwai Chung subdistrict and 11 each from Dapeng and Nan’ao subdistricts. At each site, the NAIs were measured with a portable NAI detector between 10:00 and 16:00 h for seven consecutive days in the spring (10—16 March 2016), summer (20—26 June 2016),autumn (19—25 September 2016) and winter (24—30 December 2015). Five groups of people were set up to do the measurements; each group was responsible for one subdistrict.

NAIC was measured with a portable air negative oxygen ion concentration detector (KEC900 + Air Ion Tester,Dongguan NAPUI Electronic Technology Co., Ltd., Guangdong) equipped with a shielding plate to effectively prevent interference with electrostatic and atmospheric f low. The instrument has a dynamic range of 10—1.999 × 10 6 ions cm -3 with three gears: 10—1.999 × 10 4 , 1000—1.999 × 10 6 ,100—1.999 × 10 5 . Its accuracy is ± 25% and resolution 10 ions cm -3 , when the ambient temperature is between 5 and 45 °C and moisture content is less than 95%. Air Ion Tester was positioned on a tripod 1.5 m above ground, the average level where people breathe. When the NAIC was beingrecorded, air temperature, relative humidity, wind speed,and air pressure were measured using a Heat Index WBGT Meter (WBGT-2010SD, Lutron Electronic Enterprise Co.,Ltd., Taipei), Hot Wire Anemometer-Thermometer (HHFSD1, Omega, Stamford, CT, USA) and humidity/barometer (MHB-382SD, Lutron Electronic Enterprise Co., Ltd.,Taipei).

Fig. 1 Locations of the 90 sites where negative air ions were measured. Triangles: forest sites; circles: urban green spaces

Data anal yses

We f irst tested whether the NAIC differed among seasons and land-use types. Since the data were not normally distributed, so a nonparametric Kruskal—WallisH-test was applied to analyze differences in NAIC among the four seasons and the 90 forest and urban green space sites. To predict the spatial variations of NAIC in the whole Dapeng Peninsula and to interpolate a surface from discrete data points, we used a linear interpolator, the Kriging technique, and a Geographic Information System (ArcGIS, ESRI, Redlands, CA, USA).The Gaussian model was selected as the semi-variation function of Kriging interpolation in this study.

Subsequently, the effects of vegetation characteristics(i.e., normalized difference vegetation index [NDVI], average height and crown width of the vegetation) and climate factors (i.e., air temperature, wind speed, relative humidity and air pressure) on NAIC were analyzed with a multiple linear regression. The Kruska—WallisH-test was used to evaluate differences in NAIC among the vegetation types,i.e., broad-leaved forest, coniferous forest, planted forest,and shrub. Although the vegetation species varied with landuse types, the forests and urban green spaces have similar dominant plant species. Within Dapeng Peninsula, the dominant tree species for the broad-leaved forest areFicus microcarpa,Sterculia lanceolata,andSchefflera octophylla. The main coniferous tree species arePinus thunbergii,Araucaria cunninghamii.Acacia confusa,Acacia mangiumandAcacia auriculiformisare the dominant tree species in the planted forest. In addition, Dapeng Peninsula has many shrub species withRapanea neriifoliaandSinosideroxylon wightianumas the two dominant species.

Using the proximity analysis tool in SPSS version 22, we determined the presence of water bodies, main roads and residential areas within a 500-m radius of all study sites. A nonparametric two-sample Mann—WhitneyU-test was then performed to determine whether the NAIC differed due to the presence of these elements.

Data were analyzed with SPSS version 22 (IBM, Armonk,NY, USA). All analyses were based on a 95% conf idence interval at a signif icance level of 0.05, using data expressed as mean ± standard deviation (SD).

Results

Seasonal and s patial v ariations in N AIC

Seasonal v ariations

Fig. 2 Mean (± SD) negative air ion concentration (NAIC) by season in forests and urban green space

The seasonal patterns of NAIC were similar for forests and urban green spaces, although the observed values and the seasonal variations of NAIC for the two land areas were quite distinct (Fig. 2). The Kruskal—WallisH-test revealed signif icant differences in the NAIC among the four seasons for forests and urban green spaces (p＜ 0.0005 for each). On measurement days, the mean seasonal NAIC for forests among the sites in the spring, summer, autumn and winter was 6424 ± 569, 3870 ± 477, 6018 ± 561 and 6765 ± 799 ions cm -3 , respectively. The NAIC in forests peaked at 8470 ions cm -3 in the winter, while the minimum was 3050 ions cm -3 in the summer. The values of NAIC for urban green spaces tended to be much lower than in forests, with mean 1205 ± 224 ions cm -3 in the spring,713 ± 169 ions cm -3 in the summer, 1808 ± 793 ions cm -3 in the autumn, and 1263 ± 214 ions cm-3in the winter. The maximum was 3000 ions cm -3 in the autumn, and the minimum was 480 ions cm -3 in the summer. The average daily f luctuations in NAIC were greater in the winter and spring than in the summer and autumn for forests, but smaller in the summer than in the other three seasons for urban green spaces. In general, NAIC was lowest in the summer for both forests and urban green spaces and appeared to be highest in the winter for forests and in the autumn for urban green spaces.

Spatial v ariations

Spatial variations in the NAICs on the Dapeng Peninsula are shown in Fig. 3. A signif icant difference in the NAICs between forests and urban green spaces was revealed by the Mann—WhitneyU-test (p＜ 0.0005) for all four seasons. As expected, the NAICs progressively decreased from the forests to the urban green spaces. Yet, the difference between forests and urban green spaces was especially obvious in the autumn. Moreover, the northeastern region and the central region were, respectively, the richest and poorest areas of NAI. In general, the NAICs in forests were almost 5 times higher than in urban green spaces. The results suggested that vegetation coverage has a great inf luence on the NAIs and that urbanization dramatically affects the NAIC.

Comparisons among multiple subdistricts showed signif icant differences in NAIC for forests in the spring, summer,and winter withp＜ 0.05 for all, whereas the values of NAIC in the autumn were not signif icantly different among the subdistricts withp＞ 0.05. Table S1 in the Supplementary Material summarizes the maximum, minimum, mean ± SD and range of NAICs among the subdistricts in the four seasons.During the year, the largest spatial difference in the NAIC was detected in the spring, with an average 1530 ions cm -3 difference between the richest and the poorest NAI area. The spatial differences in NAIC were lowest later in the autumn,when the difference between the maximum and minimum NAIC was 911 ions cm-3. Moreover, we found that the SD for the mean NAIC for all subdistricts followed a similar seasonal character. The highest and lowest SD for mean NAIC among subdistricts, 468 and 204 ions cm -3 , occurred in the spring and autumn, respectively. The richest area of NAIs, on average, was located on Paiya Mountain, the highest NAIC of 7429 ions cm-3in winter day at site 28, which is generally covered with broad-leaved trees. The poorest area for NAIC was found in the Central Corridor; the lowest mean value (3359 ions cm -3 ) was found at a conifercovered location, site 55, on a summer day. Moreover, the largest ranges for NAICs were also found on Paiya Mountain, followed by Qi’niang Mountain, Honghua Mountain,Bijia Mountain, and f inally, the Central Corridor. For better comparison of the NAI distribution among the multiple subdistricts, the mean, maximum, minimum and range of NAIC were normalized to the same scale. Figure 4 a shows the normalized variables among the forest subdistricts.

Fig. 3 Interpolated results for negative air ions (NAIs) for the entire Dapeng Peninsula for the four seasons. See color key for NAI levels(ions cm -3 )

Regarding the urban green spaces, the differences in NAIC among the subdistricts were signif icant during all four seasons,withp＜ 0.05 for all. The spatial NAIC variation was largest toward the autumn and smallest in the summer, with corresponding values of 2049 and 673 ions cm -3 . The SD of NAIC reached the highest of 767 ions cm -3 in the autumn and lowest of 166 ions cm -3 in the summer. The richest and poorest areas of average NAIs were located in Kwai Chung and Nan’ao.The maximum NAIC of 2683 ions cm -3 was measured in the autumn at site 67 (Kwai Chung) where broad-leaved trees are dominant, the minimum NAIC (480 ions/cm 3 ) at site 74(Nan’ao), mainly covered by shrubs, in the summer. Subdistrict Kwai Chung had the largest range in NAIC, followed by Nan’ao and Dapeng (see Fig. 4 b).

Factors i nf luencing NAIC

Vegetation c haracteristics

Fig. 4 Normalized mean, maximum, minimum and range of negative air ion concentration (NAIC) among multiple subdistricts in a forests and b urban green spaces

The Kruskal—WallisH-test indicated that no signif icant differences in NAIC among the vegetation types (broad-leaved forest, coniferous forest, planted forest, and shrubs) for the forests during the summer and autumn (p＜ 0.05). However,during the spring and winter, the sites located at coniferous forests had the signif icantly highest NAIC, followed by broad-leaved forests and then planted forests, and f inally by shrub covered areas (p＜ 0.05 for all). In urban green spaces, there was no signif icant difference among the vegetation types (broad-leaved forest, shrub and grass) in the summer (p＜ 0.05). During the spring, autumn and winter,broad-leaved forests had the signif icantly highest NAIC, followed by shrub covered areas, and f inally by grass cover areas (p＜ 0.05 for all). Figure 5 shows the NAICs for the different vegetation types for forests and urban green spaces during the four seasons.

Besides the type of vegetation, other vegetation characteristics (i.e., NDVI, average height and crown width of the vegetation) were analyzed for an effect on NAIC using a multiple linear regression analysis (see Table 1). For the forests, the three variables signif icantly affected the NAIC only in the summer (p＜ 0.05). However, contrary to expectation,NDVI and average crown width of the vegetation negatively affected the NAIC (regression coefficients of - 1550.01 and- 47.27, respectively). For urban green spaces, the expected positive effects of NAIC and crown width on NAIC werefound in both spring and winter, withp＜ 0.05 for all. However, NDVI also negatively affected NAIC in the summer and autumn (p＜ 0.05; regression coefficients of - 691.95 and - 6046.60, respectively).

Fig. 5 Box plot of negative air ion concentrations (NAICs) by vegetation type for forests and urban green spaces in four seasons. Small circles: outlier values stars: far outliers or extreme values. The bars represent the range within 1.5 interquartilerange (IQR)

Table 1 Results of regressions of the normalized difference vegetation index (NDVI), mean height and crown width of the vegetation on negative air ion concentration

Climatic f actors

Table 2 shows the results of regressions for air temperature,wind speed, relative humidity and air pressure on NAIC. For the forests, NAIC can be best explained by air temperature,wind speed, relative humidity and air pressure (p＜ 0.05 for all). Air temperature and relative humidity were negatively correlated with NAIC, whereas wind speed and air pressure were positively correlated with NAIC. Air temperature,wind speed and relative humidity were the dominant factors affecting NAIC in urban green spaces (p＜ 0.05 for all). Air temperature and relative humidity were determined to have negative effects on NAIC, and wind speed was positively correlated with NAIC.

Although a negative relationship between air temperature and NAIC was detected for forests, the residual suggested a curvilinear relationship. Therefore, this relationship was further analyzed to choose the best-f it equation. The air temperature was f irst divided into 26 bins with an increment of 1 °C, then the mean NAIC was calculated for the corresponding bin. We found that a quadratic equation with intercept had a better f it (adjustedR2 = 0.71) than the linear equation with intercept (adjustedR2 = 0.47). The scatter plots in Fig. 6 show the relation between air temperature and mean NAIC. The quadratic equation with 95% conf idencelimits is expressed as Mean NAIC = - 14.693Ta2 + 553.21 Ta+ 1713.5 (adjustedR2 = 0.7119,p＜ 0.0005).

Table 2 Results of multiple regression of air temperature,wind speed, relative humidity and air pressure on negative air ion concentration

Fig. 6 The relation between air temperature and mean negative air concentration (NAIC)

The vertex of the quadratic equation stands for the maximum value of NAIC (6921 ions/ cm -3 ), when Ta was equal to 18.8 °C. In other words, the value of NAIC gradually increased as the temperature increased, then decreased gradually after the peak.

Local land mor phology

The effects of water bodies, main roads and residential areas on NAIC clearly varied among land areas and seasons. In the Mann—WhitneyU-test, the effects of water bodies in a 500-m radius on NAIC were not signif icant for the forests during all four seasons (p＜ 0.05 for all). However, for the urban green spaces, the differences in NAIC were signif icant among the sites with and without water bodies in a 500-m radius during the summer and autumn (p＜ 0.0005 for both). The comparison of mean NAIC between the sites with and without water bodies, main road and residential area in 500-m radius for forests and urban green spaces during four seasons is shown in Fig. 7. The sites adjacent to the water bodies had higher NAICs than those without.For the effects of the main roads, no signif icant differences in NAIC were found between the areas with and without main roads for the forests during all four seasons (p＜ 0.05 for all). Nevertheless, the main roads had more pronounced impacts on the NAIC in urban green spaces. The presence of main roads within a 500-m radius was a signif icant factor in explaining the differences in NAIC during the summer and autumn (p＜ 0.05 for both). The NAIC in the areas with main roads was relatively lower than that in the area without.However, during the spring and winter, the effects of the main roads were not signif icant. In terms of the effects of human habitats, statistically signif icant differences in NAIC were found in the summer for the forests (p= 0.001) and in the autumn for urban green spaces (p＜ 0.0005) due to the presence of residential areas within a 500-m radius. The sites surrounded with residential areas within a 500-m radius had lower NAIC compared to those without. In general, NAICs in the urban green spaces were more sensitive to changes in the local morphology (including water bodies, main roads and residential areas) than those in the forests.

Discussion

Fig. 7 Comparison of mean negative air concentration (NAIC) between sites with and without water bodies, main road and residential area within a 500-m radius for forests and urban green spaces during four seasons. Error bars represent the standard deviation

Our results demonstrated large seasonal differences in NAIC of both forests and urban green spaces on the Dapeng Peninsula. NAICs were lowest in the summer for both forests and urban green spaces, while the highest were observed in the winter for forests and in the autumn for urban green spaces. This f inding contrasts with some previous studies in temperate regions that reported the highest NAIC in the summer and the lowest in the winter (Li et al. 2013; Liang et al. 2014). This difference can be explained by the fact that,in the temperate regions, the short days in the winter reduce the rate and duration of photosynthesis, leading to a decrease in NAIC. In contrast, the strong short-wave ultraviolet rays in the summer promote a point discharge and photoelectric effect of leaves, thereby increasing the NAIC (Liang et al.2014). Moreover, deciduous trees lose most of their leaves during winter in the temperate regions and resulted in much lower NAIC. However, our study area, Dapeng Peninsula,mainly has a humid subtropical monsoon climate, with relatively scorching summers and sunny winters. Winter and summer are fairly similar in receiving about 11 and 13 h of sun per day, respectively. Moreover, previous studies proved that high temperature inhibits photosynthesis by reducing stomatal conductance and net photosynthetic rate (Sharkey et al. 1998; Wang 2004; Zhang et al. 2007). Hence, the high temperature during the summertime may also lead to a reduction in NAI production in our study area. Overall,the humid subtropical monsoon climate could remarkably inf luence the production and existence of NAIs in the limited surface area of the Dapeng Peninsula and likely be responsible for the discrepancy in the measured NAIC between our study and others.

NAICs were much higher in forests than in urban green spaces in line with previous studies in different regions and at different temporal scales (Liang et al. 2014; Li et al.2015). The highest NAIs were found in the northeastern region with vast and heavily forested land and the lowest in the Central Corridor of Dapeng Peninsula.

During the spring and winter, the sites located in coniferous forests had the signifi cantly highest NAIC,followed by broad-leaved forests and then planted forests, and f inally by shrub-covered areas. These f indings agree with those of earlier studies by Wang ( 2004) and Li et al. ( 2013). The effects of NDVI, average height and crown width of the vegetation on NAIC were signif icant for both forests and urban green spaces in certain seasons.However, contradictory effects (i.e., positive and negative effects) of NDVI and average crown width on NAIC were observed in the different seasons. Hence, NDVI and average crown width cannot be considered as signif icant factors in explaining differences in NAICs. A plausible explanation is that, in this study, the ranges of NDVI and average crown width were relatively small (forests: NDVI 0.1—0.5, average crown width 1—9 m; urban green spaces:NDVI 0.1—0.4, average crown width 2—5 m). In addition, the role of different plant species in producing NAIs should be addressed in the future.

The multiple linear regression analysis showed that air temperature, wind speed, and relative humidity signif icantly inf luenced NAICs, as reported in numerous NAI studies.However, the conclusions from these studies are often contradictory. For example, some studies demonstrated that NAIC has a signif icantly negative correlation with temperature (Pawar et al. 2010), others have shown a signif icant positive correlation between NAIC and temperature (H?rrak et al. 2003; Xu et al. 2008; Li et al. 2015). In our study, the residual also suggested a curvilinear relationship between air temperature and NAIC. We found that a quadratic regression with intercept provided the best f it. The peak value of NAIC and the corresponding air temperature were derived by solving the equation. The assumption that the initial increase in NAIC as temperature increasing temperature then gradually decrease after it peaked was caused by an inhibitory effect of high temperature on photosynthesis, could explain the discrepancies among the various studies and the lowest NAICs in the summer.

In addition, the NAIC in urban green spaces was more sensitive to changes in the local morphology (the presence of water bodies, main roads, and residential areas within 500-m radius) than in the forests. Hence, appropriate urban designs to maximize NAICs are critical.

Conclusion

The main f indings of our study on the spatial and seasonal distribution of NAIs in forests and urban green spaces on the Dapeng Peninsula, a typical developing urban district in southern China are the following:

Large seasonal differences in NAIC characterized forests and urban green spaces. The subtropical maritime climate in the coastal area of southern China remarkably inf luences the abundance of NAIs, especially on a peninsula with the limited surface area. NAIC f irst increased as the temperature increased, peaked, then gradually decreased. High summer temperatures presumably reduce NAI production by inhibiting photosynthesis. The effects of the urbanization on NAIC were dramatic. The NAIC progressively decreased from the forests to urban green spaces and was highest in the large, heavily vegetated northeastern region and lowest in the narrow urbanized central region. Vegetation type and characteristics (NDVI,average height, crown width), climate (air temperature,wind speed, relative humidity), and local morphology(water bodies, main roads and residential areas within a 500-m radius) inf luenced the NAIC. In urban green spaces,the NAIC was more sensitive to the urban morphology than in the forests.

AcknowledgementsThe authors acknowledge the volunteers who helped with the f ield investigations.