Introduction

Forests play significant roles in providing fresh O2-rich air and emitting volatile organic compounds, which benefit human health by having antioxidant and anti-inflammatory effects on human airways (Antonelli et al. 2020). The study in Ryukoku forest (dominated by conifer and deciduous trees) in Japan reported that the forest could purify the polluted air via the deposition of air pollutants, NO2 and O3, onto leaf surface. Also, the effect of tree windbreak results in lower air pollutant concentration inside the forest compared to the outside level (Ichikawa et al. 2015). Fowler et al. (1989) demonstrated that forests generally exhibit high deposition rates of gaseous pollutants. Still, the rate for fine dust was low for the elevated forest in the UK (up to 500 m) because the fine dusts are frequently incorporated into cloud droplets (Fowler et al. 1989). Atmospheric dust deposition is also essential for forest sustainability since it contains nutrients and essential elements (Stoorvogel et al. 1997; Weiss et al. 2002). Gross et al. (2016) estimated phosphorous deposition in a tropical rain forest in Panama. They found that long-range transport of dust-attached phosphorus, probably from the Sahara Desert, was equivalent to 10–29% of the phosphorous in litterfall. The coniferous forest has the highest dust removal efficiency, followed by evergreen forest and deciduous forest (Han et al. 2020). The high dust capturing capacity for the coniferous forests is related to the fine leaves, complex foliage structure, and high wind speed, accelerating dust deposition rate (Han et al. 2020). Studies on the roles of forests as the sink of atmospheric dust have been widely conducted, but there is limited quantitative information for the tropical mixed deciduous forests, in which seasonal defoliation could limit dust capturing.

Mixed deciduous forest covers extensive areas of the Mekong Basin and the Ganges Basin of India (Rundel 2009). Its canopy is typically closed, high (30 m or more) and has a relatively open understory (Rundel 2009). Compared with the evergreen forests, deciduous forests exhibit strong seasonality in climate and often experience fires in the dry season. For the deciduous forests in western Thailand, approximately 70% of the leaves are flushed during the dry season, probably associated with changes in photoperiod and timing of rainfall. The duration of deciduousness ranges from 2 to 21 weeks (Williams et al. 2008), typically in March (Rundel 2009). The falling leaves could bring dust to the ground, which is one of the dust deposition pathways of the forest. Leaf area for trees in tropical deciduous forests range from 2.5 to 201 cm2, and leaf lifespan is 4.8–10 months (Chaturvedi et al. 2011).

For particulate matter (PM), accumulation on tree leaves is the primary dust removal mechanism for phytoremediation of polluted air (Wei et al. 2021). The accumulation involves attaching to the surface of leaves and being absorbed into plant tissues. Many studies show that the dust removal efficiency depends on the light intensity, leaf area, leaf surface, leaf texture, and stomatal frequency (Yoon et al. 2009; Ram et al. 2014, 2015). Fowler et al. (1989) also emphasized the roles of a large tree canopy and high airflow in increasing the PM deposition onto leaves.

The previous studies on dust capturing capacity for individual tree leaves focused on the urban trees exposed to high air pollutant loading. Traffic emissions have been dominant sources of very fine aerosols in urban atmospheres. However, seasonal variations of aerosol magnitudes responded primarily to secondary aerosols formed under photochemical reactions and anthropogenic biomass burnings (Wu and Boor 2021). The survey of leaf dust accumulation on plant leaves along a highway in India showed higher accumulation for those close to the highway, and the overall leaf dust capturing capacity ranged from 0.0083 g m−2 in Toona ciliata to 0.0597 g m−2 in Grewia optiva (Kaler et al. 2016). These magnitudes, however, were much lower than those estimated in China. Tan et al. (2022) studied common plants at Xiangnan University and found the leaf dust capturing capacity ranged from 1.005 g m−2 for Magnolia denudate to 3.451 g m−2 for Osmanthus fragrans. The high capturing capacity was associated with gully of the leaf surface and fluff, stomata, and sawtooth structure, whereas the low capturing was associated with smoothness, wax layer, leathery, and gully (Tan et al. 2022). The study for common urban plants in Xi’an, China, reported leaf dust capturing capacities ranged from 0.23 g m−2 for Trifolium repens to 4.51 g m−2 for Pittosporum tobira (Wang et al. 2011). Wang et al. (2011) also noted that the wax tubules on the leaf surface could result in poor dust capturing. Liu et al. (2012) estimated dust capturing capacities for afforested plants in Guangzhou, China. They found the highest amounts of dust captured were found for Mangifera indica, 12.723 g m−2 in an industrial area and 1.482 g m−2 in a commercial area (Liu et al. 2012). The lowest was Bauhinia blakeana, 2.682 and 0.720 g m−2, respectively. The deep grooves and high stoma density favor dust retention, whereas epicuticular wax resulted in poor capacity (Liu et al. 2012). There is limited literature on estimating the dust capturing capacity for trees in forests, especially tropical deciduous forests. Key differences between urban trees and forests are tree heterogeneity, surface roughness, and local meteorology.

Forests are not only the sink for atmospheric dust, but forest fires are the key source of fine dust across the downwind region, such as in the western US (Liu and Peng 2019). Deciduous forests are typically exposed to wildfires in the dry season when the leaves flushing are ongoing. These functions make the forest play both sink and source roles of atmospheric dust. Biomass burning smoke from wildfires can adversely affect public health and generally interferes with heart and lung functions (Fowler 2003). The largest fraction of fine aerosol in forest fire smoke plumes is black carbon, which plays a role in global warming and changes in hydrological processes over large areas (Cochrane 2003). The fires can occur naturally and are induced by climate change. In addition, human disturbance and deforestation can also contribute to fires. The dust sink role of the forests could be offset or outrun by the prominent dust emissions in the dry seasons.

This study aims to assess the roles of the tropical mixed deciduous forest in atmospheric dust sink and source. The dust capturing capacity of individual leaves in the dry season was also estimated, along with assessing the factors that influenced the dust capturing capacity, including climatic variables, trees, and leaf morphological variables. The findings provide quantitative evidence on the roles of the forest in dust capturing by focusing on the critical periods when forests are functioning as both a sink and a source of atmospheric dust. The results can elucidate the forest function on air purification in more detail.

Methodology

Study area

In this study, the tropical mixed deciduous forest is located in the part of Mahidol University Kanchanaburi Campus, Thailand, where the area was reserved for nature conservation. It lies along NW to SE direction of the limestone mountain range at 200–300 m above mean sea level, covering an area of approximately 10.24 km2 (see Fig. 1). There are two sampling sites for ambient air quality: (1) above forest canopy (AFC)—34 m from the ground at 14.13174602 °N and 99.1401076 °E and (2) below forest canopy (BFC)—2 m from the ground at 14.132755 °N and 99.138903 °E. The areas for sampling tree leaves are adjacent to the two air quality sampling sites (distance ~ 500 m for AFC and ~ 330 m for BFC, as illustrated in Fig. 1.

Fig. 1
figure 1

Map of two meteorological sites (AFC and BFC), leaf collecting area, and digital elevation model (DEM) of the studied area

Full size image

During the sampling period, hourly meteorological parameters were collected at the AFC station, and there was no rainfall during the sampling period from November 21, 2021, to January 23, 2022. At the time of sampling, Thailand was influenced by La Niña (Thailand Meteorology Department 2022), resulting in the early onset of the rainy season. The daily minimum and maximum temperatures were 13.8–22.4 °C and 31.1–38.5 °C, respectively. The average absolute pressure was 987.7 hPa, and the average relative humidity was 50%. Average wind speed 0.52 m s−1 with prevailing SE-to-S winds from the Indian Ocean and Myanmar in December 2021. In January 2022, the wind speed slightly declined to 0.40 m s−1, prevailing both the SE-to-S winds and N-to-NE winds from the Southeast Asian Mainland.

The 15 tree species

Leaves from 15 tree species were included in this study. These trees were selected from their important values (IVI)—estimated from relative density, relative frequency, and relative dominance—as assessed by the author’s experiment (Prasopsin, S.). The top 6 tree species with the highest importance values are Grewia eriocarpa (IVI = 35.83), Millettia brandisiana (18.1), Vitex Scabra (15.48), Sterculia Pexa (15.36), Garuga pinnata (15.30), and Bridelia ovata (11.30). A list of the trees and their leaf/leaflet morphology is in Table 1.

Table 1 Fifteen tree species and their macro-morphology
Full size table

Field measurement

Leaf/leaflet: Thirty leaves for each of 15 tree species were collected weekly at approximately 3 m height using an extendable tree pole pruner. The leaves were kept in ziplock bags, 5–10 per bag, and carried gently to the laboratory, 2.8 km away.

Particulate matters: Ambient concentrations of particulate matter with the size less than 1 mm (PM1), 2.5 mm (PM2.5), and 10 mm (PM10) were measured using a handheld low-cost AIRBEAM2 sensor (from HabitatMap, USA). The measurement was conducted weekly at two stations (AFC and BFC, see Fig. 1) three times (8.00–9.00 am, Noon–1.00 pm, and 4.00–5.00 pm). The AIRBEAM2 uses a light scattering method to measure particulate matter. It was claimed to have comparable quality with the referencing technology (TSI DustTrak DRX Aerosol Monitor 8533), with R2 = 0.89 for PM2.5 and R2 = 0.88 for PM1 (Michael and Lim 2018). The AIRBEAM2 also provides the measurement of PM10. The precision for PM10 could be low, R2 of 0.41 with the referencing method (Michael and Lim 2018).

Meteorological parameters: Hourly measurement of meteorological variables was collected using a mobile weather station (Weatherwise Instruments Model WS-1090-Solar), installed at the AFC site (Fig. 1). The parameters included air temperature, atmospheric pressure, relative humidity, wind speed, and direction. There was no rain during the study period.

Tree canopy cover: The tree no-canopy cover was measured under the leaf-collected trees. The measurement was done by using a spherical crown densitometer and counting the number of squares without shading. It is noted that the measurement was influenced by the canopy cover of nearby trees.

Tree height: The height of the leaf-collected trees was estimated by using a clinometer.

Dust capturing capacity

After the leaves were collected from the forest, the leaves were divided into six groups, five leaves each. One group was analyzed for leaf morphology, including leaf arrangement, leaf surface, leaf texture, and leaf area (see Table 1). The leaf surface was identified through an optical microscope. The other five groups were for estimating dust capturing capacity with five replicas per tree species. Average values from the five replicas were reported for each week. For estimating dust capturing capacity, leaf area was estimated using LeafByte App, a free and open-source mobile app for measuring leaf area developed by Adam Campbell. The leaf was later washed using distilled water with gentle rubbing to ensure the dust was washed off. Wearing non-powder gloves was required. The washing water was filtered through Whatman™ Glass Microfiber filters (GF/CTM) having a diameter of 47 mm under applied pressure. The filtered filters were desiccated for 24 h before being weighed using a digital balance with 0.1 mg resolution. The ratio of the dust mass to the leaf area is the estimate of the dust capturing capacity (g m−2). This method of dust extraction may not include the dust absorbed into the leaf tissue.

Assessing roles of the forest canopy in dust sink and source

The roles of forest canopy on dust sink and source were assessed by subtraction of the above-canopy dust concentration with the below-canopy dust concentration. A positive value suggests the dominant dust sink role of the forest, in which the above-canopy dust was largely trapped by the forest canopy. A negative value suggests the prominent source role, in which more dust was from beneath the canopy, probably from forest fires or soil dust resuspension. Dividing the subtraction by the above-canopy concentration provides dust retention efficiency of the forest canopy.

Assessing factors governing the dust capture capacity

Spearman’s rank correlation coefficient (r) was used to assess the degree of a monotonic relationship between leaf dust capturing capacity and environment factors, which are continuous variables. The factors include climate variables (temperature, humidity, air pressure, wind speed in both daytime from 6.00 am to 5.00 pm and nighttime from 6.00 pm to 5.00 am) and tree morphology (% no-canopy cover and tree height).

Chi-square test of independence was performed to assess whether the leaf dust capturing capacity is related to the tree or leaf/leaflet morphology (see Table 1). Variables included tree family, tree crown shape, leaf arrangement, leaf surface, leaf texture, and leaf size (small size with leaf area < 100 cm2, medium size with leaf area 100 to 200 cm2, and large size with leaf area > 200 cm2). To perform the test, the dust capturing capacity was categorized into five percentile levels: < 20th, 20th–40th, 40th–60th, 60th–80th, and 80th–100th.

Result and discussion

Roles of the tropical mixed deciduous forest canopy in dust sink and source

Diurnal variation

Figure 2 shows the concentration of dust in the above-forest canopy subtracted by the below-canopy concentration. A positive value suggests a downward flux of dust and net dust sink into the forest, whereas a negative value indicates upward flux and net dust source from the forest to the atmosphere. A net downward flux was predominant at the midday for PM1 and PM10. The average PM1 and PM10 concentration trapped by the forest canopy during the midday dry period was 0.79 and 2.24 µg m−3, respectively, accounting for 4.51 and 7.45% dust retention efficiency, respectively. The midday dust sink role of the forest canopy could be enhanced by turbulent exchange generated by solar and ground radiation (Han et al. 2020). In this study, strong air-mass turbulence could significantly enhance the dust sink role of the forest and increase the deposition velocity of the dust (Petroff et al. 2008). As shown in Fig. 3, the midday meteorological condition generally exhibited comparatively high wind speed. In addition, the highest forest dust capture was 8.5 µg m−3 (the above–canopy level minus the below-canopy level) for PM1, 5.92 µg m−3 for PM2.5, and 16.11 µg m−3 for PM10 on January 10th, 2022, when high wind speed was experienced (Fig. 3).

Fig. 2
figure 2

Above-canopy concentrations (upper), below-canopy concentrations (lower), and the upper minus the lower for PM1, PM2.5, and PM10 for different times of day

Full size image
Fig. 3
figure 3

Hourly meteorological parameters measured weekly from November 27, 2021, to January 30, 2022

Full size image

The forest as the dust source was predominant in the morning and the late afternoon. As shown in Fig. 2, net forest dust concentrations trapped by the canopy in the morning were − 0.22 µg m−3 (− 1.21% dust retention efficient and negative value represents upward flux) for PM1, − 0.08 µg m−3 (− 0.33%) for PM2.5 and − 0.52 µg m−3 (− 1.62%) for PM10 and in the late afternoon, the concentrations were − 0.37 µg m−3 (− 1.69%), − 0.22 µg m−3 (− 12.85%), and −0.84 µg m−3 (− 12.54%), respectively. Atmospheric stability, inducing low planetary boundary layer height and accumulating dust concentration near the ground, could explain the dominant upward dust flux in the morning and the late afternoon (Chiesa et al. 2019).

Another form of aerosol contributed from the forest was haze-fog, induced by high emissions of air pollutants, probably from forest fires and humidity. The contribution was predominant in the morning when relative humidity was usually high, and the air temperature was low. As shown in Fig. 2, the upward fluxes of aerosol were markedly elevated in the morning of December 26th, 2021, after the day of high aerosol concentration on December 25th, 2021. This corresponded with comparatively high relative humidity (see Fig. 3). Similar evidence was also found on January 22nd, 2022, with both the aerosol source role of the forest and the high humidity in concordance (Figs. 2 and 3). Achtemeier (2006, 2008) took several measurements during biomass burnings and confirmed that the burning temperature could induce more moisture to be released at night. The smoke was mixed with ambient air and moisture and later developed into a larger fog, which was 17 times larger than the natural fog on the following day (Achtemeier 2008). The haze-fog pollution has typically been pronounced in urban cities. Wang et al. (2016) showed that the secondary aerosol formation (sulfate) was feasible on fine aerosols with high relative humidity. Zhao et al. (2013) also affirmed that the increased relative humidity on haze days was favorable for the oxidation of SO2 and formation of secondary aerosols.

Dust size dependence

In forests, organic PM1 is primarily from biomass burning and photochemical processes of volatile organic compounds released from forest trees (Masalaite et al. 2017; Kourtchev et al. 2005). The PM1 is in accumulate-mode size and the deposition and dispersion rates of PM1 should be driven by molecular diffusion, thermal convection, and surface roughness-driven air–mass turbulent (Chiou et al. 2010; Guha and Samanta 2014; Petroff et al. 2008). Consistently, Chiesa et al. (2019) reported the increases in downward deposition velocity of the PM1 in the broadleaved deciduous forest under unstable atmosphere during the period with leaves. The PM10 is considered coarse-size dust. The deposition velocity of the PM10 was primarily governed by gravitational settling, whereas its resuspension was exclusively under strong eddy (Chiou et al. 2010).

In this study, the tropical mixed deciduous forest exhibited average PM1 concentrations at the above canopy of 17.02, 13.79, and 11.84 µg m−3 for 8.00–9.00 am, noon–1.00 pm, and 4.00–5.00 pm, respectively, and the below-canopy levels were 17.24, 13.00, and 12.21 µg m−3 for 8–9 am, at noon–1 pm, and 4–5 pm, respectively. The average PM10 concentrations at the above forest canopy were 32.38, 24.00, and 21.15 µg m−3 for 8–9 am, noon–1 pm, and 4–5 pm, respectively, and the below-canopy levels were 32.90, 21.77, and 21.98 µg m−3, respectively. Comparing the dust concentrations for the above and the below canopy, similar diurnal patterns, upward in the morning and the late afternoon and downward in the midday, were found for the PM1 and PM10, as reported in the earlier section.

In this study, the dry-season upward fluxes of PM2.5 to the atmosphere were often greater than the downward flux (see Fig. 2). Thus, the forest as PM2.5 sources to the atmosphere could outrun the PM2.5 sink role of the forest canopy for entire of the dry season. Average PM2.5 concentrations at the above-forest canopy were 21.27, 17.71, and 15.18 µg m−3 for 8.00–9.00 am, noon–1.00 pm, and 4.00–5.00 pm, respectively, and the below-canopy levels were 21.35, 17.73, and 15.40 µg m−3, respectively.

We hypothesized that forest fire should be one of the dust sources as the upward flux was usually marked when the below-canopy dust concentrations were high. The forest fires, probably arson fires, have regularly occurred in this season for land clearance and ease of harvesting forest products. The fires could also be induced by agricultural burnings in perimeter fields. See et al. (2006) concordantly reported that aerosol particles in Southeast Asia during hazy days, attributed to long-range transport of forest fires, had mass-size distribution peaks at 0.5–0.6 and 3.2 μm. Furthermore, Chomanee et al. (2020) reported that the PM1 was predominant (54.6–59.1%) in southern Thailand during the 2015 Southeast Asian haze episode. We speculate that the fires could also induce coarse-size dust resuspension due to thermal convection, but the impact may be limited to adjacent areas.

Dynamics of leaf dust capturing capacity

Dust capturing capacities of individual leaves for the 15 tree species are illustrated in Fig. 4. In general, the maximum capturing capacities for the trees in the tropical mixed deciduous forest ranged from 0.95 to 5.197 g m−2. The three best dust capturing capacities were Homto (5.197 g m−2), followed by Sismu (4.301 g m−2), and Lagve (3.814 g m−2). The four least dust capturing capacities were Garpi (0.95 g m−2), Milbr (1.63 g m−2), Stepe (1.64 g m−2), and Greer (1.65 g m−2). These magnitudes are similar to the estimates for urban deciduous and coniferous trees in China, which are 1.005 g m−2 for Magnolia denudate to 3.451 g m−2 for Osmanthus fragrans in Xiangnan University (Tan et al. 2022), 0.23 g m−2 for Trifolium repens to 4.51 g m−2 for Pittosporum tobira in Xi’an (Wang et al. 2011), and 0.72 g m−2 for Bauhinia blakeana in the commercial area to 12.723 g m−2 for Mangifera indica in an industrial area in Guangzhou (Liu et al. 2012).

Fig. 4
figure 4

Dust capturing capacities of individual leaves for 15 tree species. The error bar represents a standard error of 5 replicas

Full size image

The dust capturing capacities usually increased from the beginning of the dry season to the end of the dry season (see Fig. 4). As shown in Table 2, the leaf dust accumulation rate was estimated to be 13.5 mg m−2 day−1 for Mille to 37.6 mg m−2 day−1 for Sismu. At the end of the sampling period on January 23rd, 2022, 6 of 15 tree species were completely devoid of leaves. The increasing trend could imply that the dust could be accumulated on the leaf surfaces in high amounts and be deposited onto the ground along with the falling leaves. There were some exceptions for Homto, Garpi, and Greer, in which the capturing capacities were high in the early dry season and slightly declined or constant toward the end (− 28.6, − 4.4, − 0.4 mg m−2 day−1, respectively, see Table 2). The various dynamics of the dust capturing capacity trends among tree species imply ecological benefits of increasing overall dust capturing capacity by the forest for the entire dry period. Furthermore, dust deposition is considered an influx of nutrients to the forest (Stoorvogel et al. 1997; Weiss et al. 2002). Therefore, the deposition rate of the nutrient onto the forest should be further investigated to understand forest–dust interaction and the forest nutrient cycle.

Table 2 Dust accumulation rate (slope) and interception estimated by linear regression model
Full size table

Factors that influenced the dust capturing capacity

Significant effects of environmental factors and morphology of leaves and trees are indicated by the Spearman’s rank correlation coefficient in Table 3 and the Chi-square test results in Table 4.

Table 3 Spearman’s rank correlation coefficients between the weekly dust capturing capacity of individual tree leave and environmental factors (climatic and tree properties) at corresponding week
Full size table
Table 4 Chi-square statistic for the dependence test and its p-value between the dust capturing capacity classes and tree characteristics and leaf morphology
Full size table

Climatic factors

Meteorological conditions during the sampling period are detailed in Fig. 3. The unstable condition was more robust in the later period, suggested by the increasing daytime air temperature and decreasing daytime humidity along with pressure from the onset of the dry season toward the end.

As shown in Table 3, high leaf dust capturing capacity was significantly (p < 0.05) associated with low nighttime temperature (r =  − 0.28), low humidity (r =  − 0.32 and − 0.35 for daytime and nighttime, respectively), high atmospheric pressure (r = 0.21 and 0.20, respectively), and high wind speed (r = 0.31 and 0.22, respectively). The meteorological conditions implied that strong atmospheric stability under cold and dry weather could enhance the capturing capacity due to limiting convection of air mass and accumulation of ground-level air pollutants. The strong lateral wind speed could also accelerate dust deposition rate via increasing dust deposition velocity and decreasing PM concentration, as similar to the observed in the coniferous forests (Ichikawa et al. 2015; Han et al. 2020).

Tree morphology

According to the correlation coefficients in Table 3, tree height was not found to be significantly associated with the capturing capacity but the canopy coverage was. The open forest canopy could enhance leaf dust capturing capacity. Such conditions could be prominent toward the late dry season when leaves are mostly flush. Since the leaves were collected at the fixed levels of three meters, more dust could be deposited on the low-rise leaves when the canopy becomes opened and exposed to stronger winds. This finding agrees with the report by Han et al. (2020) that the seasonal highest dust deposition velocity onto the forest canopy was in fall, whereas the lowest velocity was in spring.

Spheroidal tree crown shape was prevalent in this study but not associated with the leaf dust capturing capacity (see Fig. 5). Irregular tree crown shape was often found with high dust capturing capacity, whereas umbrella tree crown was more prominent with low dust capturing capacity. Based on this observation, irregular shaped trees induced more canopy movement responding to the wind speed above the forest canopy. On the other hand, umbrella-shaped trees tended to be more wind resistant, resulting in lower dust deposition velocity induced by surface wind turbulence.

Fig. 5
figure 5

Frequency matrix between leaf dust capturing capacity and tree family (a), tree crown shape (b), and leaf/leaflet morphological factors (c to f)

Full size image

Figure 5 shows distributions of the observations between leaf dust capturing capacity and tree family. Salicaceae and Sapindaceae were often found with high dust capturing capacity. Leaf or leaflets of Salicaceae and Sapindaceae trees can be found in medium to large sizes (greater than 100 cm2, see Table 1). Salicaceae is the willow family of flowering plants, in which the leaves are usually stipulate, a vein traverses into the tooth apex, and a rounded gland or a stout trichome (Simpson 2019). In this study, trees in Salicaceae are Casgr and Homto. In Fig. 4, high dust capturing capacity for the Salicaceae could be found since the early dry season with the estimates of 1.646 g m−2 for Casgr and 4.195 g m2 for Homto (see Table 2 for interception). Sapindaceae is also a family of flowering plants. Unlike the Salicaceae, trees in Sapindaceae are usually exstipulate (Simpson 2019). In this study, trees in Sapindaceae are Schol and Sismu. As shown in Table 2, Schol and Sismu exhibited low dust capturing capacity in the early state (0.609 and 0.862 g m−2, respectively). Still, they increased the capacity with the accumulation rate of 28.7 and 37.6 mg m−2 day−1, respectively, toward the end.

Fabaceae and Malvaceae were often found with low dust capturing capacity (see Fig. 5). The Fabaceae or Leguminosae is the third largest flowering plant family globally (Tekdal 2021), and this study includes Milbr and Mille. Both exhibited comparatively small leaf/leaflet areas. The low dust accumulation rate on the leaf was 19.8 mg m−2 day−1 for Milbr and 13.5 mg m−2 day−1 for Mille. The Malvaceae trees, or the mallows, often have stellate trichomes (Simpson 2019). The Mavaceae, including Stepe, exhibited low dust capturing capacity at the early state (0.401 g m−2 see Table 2), and Greer showed a low dust accumulation rate (− 0.4 mg m−3 day−1 see Table 2).

Leaf morphology

The results obtained from the Chi-square test for independence in Table 4 showed that leaf morphology was significantly associated with the dust capturing capacity of the individual leaf. The morphology includes leaf texture, indumentum top and bottom, and leaf/leaflet size, but not much significance was associated with leaf arrangement.

Leaves with high dust capturing capacity often exhibited large size (> 200 cm2) and either thick and leathery texture (Coriaceous) or thin and semi-translucent, membrane-like texture (Membranaceous). Indumentum top could be short, stiff trichomes (Scabrous). The bottom surface was densely short, with soft trichomes (Tomentose). Some of these traits were found in leaf/leaflet of the trees with high dust capturing capacity, which was Homto, Sismu, and Lagve (see Table 1).

  • Homto: Major traits of its leaves that promote high dust capturing capacity in the early dry season were its Scabrous indumentum top, Tomemtose indumentum bottom, and large leaf surface area. Nonetheless, the characteristics did not promote dust accumulation in the long run since the capturing capacity was peak in the onset and decreasing toward the end of the dry season, as described earlier.

  • Sismu: Major traits of its leaves that promote high dust capturing capacity and high dust accumulation rate were its Puberulous indumentum top and bottom, Coriaceous texture, and medium leaf size.

  • Lagve: Interestingly, Lagve’s leaf exhibited Glabrous indumentum top and bottom and small size. These traits may not promote high dust capturing capacity. Nonetheless, the only key feature of Lagve playing a role was the Membranaceous texture.

Low dust capturing capacity often associated with leaves having small size (< 100 cm2), being thin and dry texture (Scarious), and the presence of star-shaped trichomes (Stellate) in indumentum top—commonly found in Malvaceae. Stellate in the indumentum bottom is also pronounced for the low dust capturing capacity. Garpi exhibited the least dust capturing capacity (see Fig. 5). The key traits resulting in the poor dust capturing were hairless leaf surface (Glabrous), Chartaceous leaf texture, and small size (< 100 cm2).

Based on the observations in this study, indumentum top and bottom, leaf texture, and leaf size greatly influenced the dust capturing capacity for tree leaves in the mixed deciduous forest. Zhang et al. (2017) also agreed that the dust adsorptive capacities of the broad‐leaved tree were more or less influenced by the micro-roughness of the leaf surface. Furthermore, Zhang et al. (2017) also discussed that the main factors affecting the adsorptive capacities of the leaves in coniferous trees are the number of stomata, the amount of epicuticular wax, and the properties of the cuticle. The stomata type, size, and number were not included in the scope of this study. However, since the stomata could be less open in the dry season, when plant transpiration is limited, its effect on the capturing capacity should be further studied in detail.

Conclusion

In the dry season, tree leaves in the tropical mixed deciduous forest play a role in dust capturing, whereas the forest fire could offset the dust sink role. This study aimed to assess the functions of the forest in atmospheric dust sink and source. The capturing capacity of individual leaves in the dry season was also estimated, along with evaluating the factors influencing dust capturing capacity, including climatic, tree, and leaf morphological variables. The studied site was the conservation forest in Mahidol University Kanchanaburi Campus, Thailand, and the measurement was weekly during the local dry season.

The result showed that the downward flux and the forest dust sink role were predominant in the midday for PM1 and PM10, but low retention efficiency (4.51 and 7.45%, respectively). For the PM2.5, the forest fires as PM2.5 sources to the atmosphere could outrun the PM2.5 sink role of the forest canopy for the entire dry season.

Considering individual tree leaves, maximum dust capturing capacities for the forest trees ranged from 0.95 to 5.197 g m−2. The three best dust capturing capacities were for Homalium tomentosum (5.197 g m−2), followed by Sisyrolepis muricata (4.301 g m−2) and Lagerstroemia venusta (3.814 g m−2). The lowest dust capturing capacities were Garuga pinnata (0.95 g m−2), followed by Millettia brandisiana (1.63 g m−2), Sterculia pexa (1.64 g m−2), and Grewia eriocarpa (1.65 g m−2). The dust capturing capacities usually increased from the beginning of the dry season to the end of the dry season. There were some exceptions for Homalium tomentosum, Garuga pinnata, and Grewia eriocarpa. The capturing capacities were high in the early dry season and slightly declined or remained constant toward the end. These various dust capturing dynamics among tree species could benefit dust capturing for the forests for the entire dry season.

Leaf dust capturing capacity was enhanced under cold and dry weather, strong lateral winds, defoliated and irregular-shaped trees, and leaves/leaflets having at least one of the following characteristics; large size (> 200 cm2), either thick and leathery texture (Coriaceous) or thin and semi-translucent, membrane-like texture (Membranaceous), indumentum top could be short, stiff trichomes (Scabrous), and indumentum bottom surface was densely short, soft trichomes (Tomentose). The key traits exhibiting poor dust capturing were hairless leaf surface (Glabrous), Chartaceous leaf texture, and small leaf size (< 100 cm2). The finding helps clarify mechanisms of forest dust capture, in which tree leaves play an important role. Further studies should focus on analyzing deposition fluxes of nutrient and toxic elements onto the forest and leaf surface to fulfill an understanding of the balances of nutrient and harmful elements in the forest ecosystem.