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. 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 mg m-3 (the above-canopy level minus the below-canopy level) for PM1, 5.92 mg m-3 for PM2.5, and 16.11 mg m-3 for PM10 on January 10th, 2022, when high wind speed was experienced (Fig. 3).
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 mg m-3 (1.21% dust retention efficient) for PM1, 0.08 mg m-3 (0.33%) for PM2.5 and 0.52 mg m-3 (1.62%) for PM10 and in the late afternoon, the concentrations were 0.37 mg m-3 (1.69%) for PM1, 0.22 mg m-3 (12.85%) for PM2.5, and 0.84 mg m-3 (12.54%) for PM10. Forest fire should be one of the dust sources as the upward flux was usually marked when the below-canopy fine dust concentrations were high. Such a fire-contributed dust episode was found on January 17th, 2022, at noon to 1.00 pm for PM1 and PM2.5, but not for PM2.5 (see Fig. 2). 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.
Another form of dust 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 dust were markedly elevated in the morning of December 26th, 2021, after the day of high dust 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 dust 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 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). In the tropical mixed deciduous forest, average PM1 concentrations at the above canopy were 17.02 mg m-3, 13.79 mg m-3, and 11.84 mg m-3 for 8.00–9.00 am, noon–1.00 pm, and 4.00–5.00 pm, respectively; and the below canopy were 17.24 mg m-3, 13.00 mg m-3, and 12.21 mg m-3 for 8–9 am, at noon–1 pm, and 4–5 pm, respectively. The PM10 is considered coarse-size dust. Excluding desert dust, which is not the dust source in the studied forest, the coarse dust in the forest could be bioaerosol particles, sea salt particles from long-range transport, and biomass burning (Moran-Zuloaga et al. 2018). The deposition velocity of the PM10 was primarily governed by gravitational settling, whereas its resuspension was exclusively under strong eddy (Chiou et al. 2010). The average PM10 concentrations at the above forest canopy were 32.38 mg m-3, 24.00 mg m-3, and 21.15 mg m-3 for 8–9 am, noon–1 pm, and 4–5 pm, respectively; and the below-canopy levels were 32.90 mg m-3, 21.77 mg m-3, and 21.98 mg m-3, respectively. Comparing the dust concentrations for the above and the below canopy, similar diurnal patterns were found for the PM1 and PM10, as discussed earlier. This suggests that molecular diffusion (for very fine dust) and gravitation settling (for coarse dust) play roles. Tree leaves play a crucial role in dust capturing, and the capturing efficiency depends on wind speed, particle size, and physical and biological characteristics of the tree canopy (Wei et al. 2020).
The seasonal upward fluxes of PM2.5 to the atmosphere were greater than the downward flux (see Fig. 2). Average PM2.5 concentrations at the above-forest canopy were 21.27 mg m-3, 17.71 mg m-3, and 15.18 mg m-3 for 8.00–9.00 am, noon-1.00 pm, and 4.00–5.00 pm, respectively; and the below-canopy were 21.35 mg m-3, 17.73 mg m-3, and 15.40 mg m-3, respectively. As shown in Fig. 2, the downward fluxes of PM2.5 were dominant most of the time at the midday, similar to those found for PM1 and PM10. On January 17th, 2022, the surface PM2.5 and PM1 concentrations were significantly higher than the above-canopy concentration, and smoke from forest fires should be responsible for these high dust episodes. The forest fires as PM2.5 sources to the atmosphere could outrun the PM2.5 sink role of the forest canopy for entire of the dry season.
(Add Fig. 2 here)
(Add Fig. 3 here)
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 g m-2 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 green trees in China, which are 1.005 g m-2 for Magnolia denudate Desr. to 3.451 g m-2 for Osmanthus fragrans Lour 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 Linn in an industrial area in Guangzhou (Liu et al. 2012).
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 m2 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 (Stooryogel 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
Tree species | Dust accumulation rate, mg m− 2 day− 1 | Interception, g m− 2 |
Casgr | 22.1 | 1.646 |
Hubce | 25.7 | 0.623 |
Milbr | 19.8 | 0.170 |
Homto | -28.6 | 4.195 |
Ferad | 33.8 | 0.655 |
Schol | 28.7 | 0.609 |
Sismu | 37.6 | 0.862 |
Garpi | -4.4 | 0.765 |
Stepe | 27.3 | 0.401 |
Greer | -0.4 | 0.995 |
Crope | 15.4 | 1.167 |
Briov | 23.0 | 0.249 |
Mille | 13.5 | 0.682 |
Lagve | 21.0 | 2.135 |
Vitsc | 24.1 | 0.474 |
(Add Fig. 4 here)
(Add Table 2 here)
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.
Environmental factors | Spearman’s rank correlation coefficient, r |
Climatic Factora | |
Daytime temperature | -0.05 |
Nighttime temperature | -0.28* |
Daytime humidity | -0.32* |
Nighttime humidity | -0.35* |
Daytime atmospheric pressure | 0.21* |
Nighttime atmospheric pressure | 0.20* |
Daytime wind speed | 0.31* |
Nighttime wind speed | 0.22* |
Tree characteristics | |
% No-canopy cover | 0.48* |
Tree height | 0.17 |
aDaytime: 6.00 am to 5.00 pm; Nighttime: 6.00 pm to 5.00 am |
*p < 0.05 |
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
Tree/Leaf characteristics | Chi-square statistic | p-value |
Tree family | 77.9281 | < 0.001 |
Tree crown shape | 40.1119 | < 0.001 |
Leaf/Leaflet arrangement | 18.5882 | 0.099 |
Leaf surface (top) | 36.3145 | 0.002 |
Leaf surface (bottom) | 32.8605 | 0.001 |
Leaf texture | 34.1218 | < 0.001 |
Leaf size | 17.7377 | < 0.007 |
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 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. 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.
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 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) or without trichomes (Glabrous). 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.
(Add Table 3 here)
(Add Table 4 here)
(Add Fig. 5 here)