1

Particulate matter fluxes in a Mediterranean mountain forest: interspecific 1

differences between throughfall and stemflow in oak and pine stands 2

Carles Cayuela1*, Delphis F. Levia2, Jérôme Latron1, Pilar Llorens1 3

1 Institute of Environmental Assessment and Water Research (IDAEA-CSIC), Jordi 4

Girona 18-26, 08034 Barcelona, Spain. 5

2 Department of Geography, University of Delaware, Newark, United States 6

*Corresponding author: Carles Cayuela (carles.cayuela@idaea.csic.es) 7

Key points: 8

Atmospheric particulate matter was highly influenced by Saharan dust intrusions 9

despite accounting for only 16% of the events. 10

Canopies enhanced the concentration of particulate matter in throughfall and 11

stemflow, but stemflow had higher enrichment ratios. 12

Leaf presence or absence influenced the size of the particulate matter, although 13

the smallest particulates were measured in stemflow. 14

2

Abstract 15

In forested areas, canopies play an important role in the partitioning of rainfall. During 16

this process there is also a redistribution of particulate matter (PM) that is deposited 17

from the atmosphere on vegetative surfaces and transported to soil layers by throughfall 18

and stemflow. We collected samples of rainfall, throughfall and stemflow from two 19

different forest plots (pine and oak) in a Mediterranean mountainous area and analysed 20

the amount and size distributions of PM (0.45 µm < PM < 500 µm). The exploration of 21

backward trajectories revealed that PM content varied significantly. This depended on 22

the origin of the air mass, with Atlantic fronts transporting less PM in the atmosphere 23

than North African dust intrusions, which added disproportionate inputs of PM. Overall, 24

throughfall provided the largest proportion of incoming PM under trees, but, at the base 25

of each tree, stemflow led to a localized input of water that was more PM-enriched than 26

water through open precipitation or throughfall. Interspecific differences in PM fluxes 27

were noted with pines retaining more PM in their crowns than oaks. Furthermore, the 28

presence of leaves on oak increased the size and the amount of particulates released by 29

throughfall. The PM in stemflow was smaller and rounder than in throughfall. This 30

study adds to our understanding knowledge of the processes that control the deposition 31

and distribution of PM delivered to forest soils, a fraction that is often ignored in studies 32

of nutrient and energy fluxes in ecosystems. 33

3

1. Introduction 34

Atmospheric particulate matter (PM) is a mixture of physically and chemically diverse 35

substances that exist in ambient air as discrete particles of widely differing sizes. Its 36

source is diverse: it can be introduced to the atmosphere from natural sources (e.g. 37

volcanic eruptions, sea salt, soil dust suspension, natural forest fires, biological elements 38

such as pollen, bacteria, fragments of vegetal organisms or animals, etc.) or from 39

multiple anthropogenic activities (e.g. transport, industry, biomass burning, etc.) (Lequy 40

et al., 2013; Weathers & Ponette-González, 2011). Depending on its size, PM can be 41

transported over long distances: while coarse particulates are rapidly removed from the 42

air by sedimentation, fine PM can be easily transported by the wind up to thousands of 43

kilometres from the area where they were formed (Perrino, 2010). Dust storms, 44

originating primarily in drylands, play a particularly important role in PM distribution. 45

They have numerous source areas, but the Sahara is undoubtedly the largest source of 46

atmospheric desert dust (Middleton, 2017). For instance, Saharan dust has affected the 47

nature of soils in the Canary Islands (Castillo et al., 2017; Menéndez et al., 2007; Muhs 48

et al., 2010) and Mount Cameroon (Dia et al., 2006). Saharan micro-nutrients have also 49

been detected as far away as northern Europe (Avila et al., 1997; Franzén et al., 1994; 50

Yaalon & Ganor, 1979), the Caucasus Mountains (Kutuzov et al., 2013), south-west 51

USA (Prospero, 1999), Caribbean islands (Muhs et al., 2007), the Amazon (Swap et al., 52

1992) and the Andes (Boy & Wilcke, 2008). Numerous diameter classifications of PM 53

have been proposed: for example, air quality standards (USEPA, 2004) regulate fine 54

PM (PM2.5: 0 – 2.5 µm aerodynamic diameter) and coarse PM (2.5 – 10 µm) due to its 55

harmful effects on human health. However, less agreement exists on the classification of 56

PM in ecological studies. Nevertheless, so as to be consistent with much of the 57

4

scientific literature, this study focuses on PM diameter distributions of > 0.45 and < 500 58

µm (Levia et al., 2013). 59

Forest canopies play an important role in the removal of PM from the atmosphere 60

(McDonald et al., 2007). PM can be deposited on vegetative surfaces via wet 61

deposition, in the form of rain, snow or mist; via dry deposition, as direct particulates 62

and gases; or via occult deposition, as dissolved material in cloud droplets. 63

Nevertheless, deposition rates respond to atmospheric conditions, vegetation properties 64

and topographic factors (Grantz et al., 2003). In general, higher rainfall results in greater 65

wet deposition (Weathers & Ponette-González, 2011) and high wind speed usually 66

increases dry deposition due to enhanced particle impaction and turbulent transfer of 67

gases and particulates to forest canopies (Fowler et al., 1989). The amount of PM 68

deposited on forest canopies also depends on the structure of the crowns and bark 69

roughness. For example, coniferous species, with needle-shaped leaves, enhance 70

impaction and retention of PM (Beckett et al., 2000; Dzierżanowski et al., 2011; Song et 71

al., 2015). In addition, as they keep their leaves for several years, PM accumulates for 72

longer periods (Beckett et al., 1998). In contrast, deciduous forest canopies generally 73

capture less PM. However, within deciduous broadleaved trees, species with rough leaf 74

surfaces capture PM more effectively than ones with smooth surfaces (Beckett et al., 75

2000). Rainfall dynamics also play an important role; in general, the concentration of 76

suspended and dissolved materials is highest at the onset of a precipitation event and 77

decreases as an event progresses (Lindberg et al., 1986). Long rainfall events, therefore, 78

remove previously accumulated PM on vegetative surfaces much more effectively. In 79

addition, intense rainfall may enhance the washing of PM deposited on leaves, whereas 80

lower intensities may hydrate previously dry-deposited PM and facilitate the foliar 81

uptake of the substances it contains (Lovett & Lindberg, 1984). 82

5

In forests, PM can be a source of essential macro- and micro-nutrients that reach the soil 83

through the redistribution of rainfall as throughfall and stemflow (Weathers & Ponette-84

González, 2011). However, depending on the chemical composition and magnitude of 85

the deposition, PM may affect plants, indirectly alter soil nutrient cycling and inhibit 86

plant nutrient uptake (Grantz et al., 2003). Although stemflow may be a low percentage 87

of open rainfall, it can funnel >20-fold more water to near-stem soils than open rainfall 88

in an equivalent area (Carlyle-Moses et al., 2018; Levia & Germer, 2015). In addition, 89

the longest path that a rain drop travels is as stemflow, which involves a prolonged 90

interaction between intercepted rainfall and canopy surfaces (leaves, branches and 91

stems) that causes a greater exchange of solutes and particulates (Michalzik et al., 92

2016). Thus, soils near the stems receive a supply of water that can be more chemically 93

enriched than water from rainfall or throughfall (Levia & Frost, 2003; Levia & Germer, 94

2015). 95

If the particulate matter fraction (0.45 µm < PM < 500 µm) is not taken into account, it 96

can result in misleading inferences and budgeting gaps when nutrient and energy fluxes 97

in ecosystems are studied (Levia et al., 2013). Until now, only a few studies have 98

focused on PM fluxes and their size distribution below the canopy (e.g. Lequy et al., 99

2014; Levia et al., 2013; Song et al., 2015). As these PM fluxes provide essential 100

information for understanding the dynamics of different forest covers in the removal of 101

PM from the atmosphere, further research is required into biosphere-atmosphere 102

interactions (Levia et al., 2013). In this study, PM fluxes below the canopy (in 103

throughfall and stemflow) and their size distribution for two tree species, downy oak 104

(Quercus pubescens Willd.) and Scots pine (Pinus sylvestris L.), in a Mediterranean 105

mountainous area were measured. Our specific aim was to: (i) analyse how different 106

tree species affect the PM content reaching the forest soil; and (ii) evaluate the 107

6

differences in PM content and size distributions between throughfall and stemflow. This 108

study will help to fill the gap in the understanding of the processes that control the 109

deposition and distribution of PM on forest soils as well as measure their rates. 110

2. Methodology 111

2.1. Study area 112

Forest stands of downy oak and Scots pine located in the Vallcebre research area (NE 113

Spain, 42º 12’N, 1º 49’E) in the eastern Pyrenees were chosen to analyse the deposition 114

of PM. The climate in this mountainous area is Sub-Mediterranean, with a mean annual 115

temperature of 9.1 ± 0.67ºC, mean annual potential evapotranspiration, as calculated by 116

the Hargreaves & Samani (1982) method, of 823 ± 26 mm and mean annual 117

precipitation of 880 mm ± 200 mm (1989-2013). The precipitation regime is seasonal, 118

with autumn and spring generally being the wet seasons, while summer and winter are 119

drier. Summer rainfall is characterized by intense convective events, while winter 120

precipitation is caused by frontal systems, with snowfall accounting for less than 5% of 121

precipitation (Latron, Llorens, et al., 2010; Latron, Soler, et al., 2010; Llorens et al., 122

2018). 123

The selected forest stands are less than 1 km from each other. There are neither high-124

polluting factories nor important motorways in the surroundings of the study area. The 125

pine stand has an area of 900 m2, a tree density of 1,189 trees ha-1 and a basal area of 126

45.1 m2 ha-1, and is oriented northeast at an elevation of 1,200 m. The oak stand has an 127

area of 2,200 m2, a tree density of 518 trees ha-1 and a basal area of 20.1 m2 ha-1, and is 128

oriented southeast, at an elevation of 1,100 m. In the oak stand, leaves appear during the 129

first half of May, quickly reaching the average accumulated Leaf Area Index of 3.35 130

(SD ±0.5), and fall in autumn, with 90% of leaves falling between October to December 131

(Muzylo et al., 2012). 132

7

2.2. Sampling design 133

Prior to the experiment, canopy coverage and DBH distributions were measured in both 134

stands via hemispherical photography and forestry measurements, respectively (Llorens 135

& Gallart, 2000; Molina et al., 2019). A total of 10 throughfall and 4 stemflow 136

collectors were deployed in each forest stand, covering different representative canopy 137

coverages (from 30 to 88% in the pine stand, and from 30 to 95% in the oak stand) and 138

the Diameter at Breast Height (DBH) distributions (~15, 20, 25 and 30 cm) (Cayuela et 139

al., 2018). Throughfall collectors were made of plastic funnels connected to a 1-litre 140

bottle, funnels were held by a stake 0.5 m above the ground and the bottles were 141

covered by an opaque PVC tube to minimize irradiation impact and algae growth. 142

Stemflow collectors were fashioned from plastic funnels sealed to the trees with silicone 143

at breast height and connected to 60-litre opaque polyethylene buckets. Bulk rainfall 144

was collected by means of the same methodology as throughfall, but outside the forest 145

stands. Funnels were equipped with a nylon sieve (1 mm mesh width) to prevent sample 146

contamination with coarse matter. Photographs of the collection methods can be viewed 147

in Figure S1 (supporting information). In addition, each stand was equipped with 148

automatic tipping-bucket gauges to measure rainfall, throughfall and stemflow every 149

five minutes. 150

2.3. Analysis of particulate matter fluxes 151

The sampling was conducted weekly (if rainfall occurred) between July 2015 and July 152

2016. Each sampled event included the total dry and wet deposition of the period 153

between samplings. In total, 36 events were sampled, covering both the growing and the 154

dormant season for oak. Sampling of PM in December was not possible due to the lack 155

of rain. For each sampled event, the origin of its air mass was identified through 5-day 156

backward trajectories calculated using the HYSPLIT model (Rolph et al., 2017; Stein et 157

8

al., 2015) with the one-degree meteorological GDAS (Global Data Assimilation 158

System) dataset, downloaded from the available portal in the HYSPLIT interface. 159

For every rainfall event, single sub-samples of rainfall, throughfall and stemflow were 160

prepared by pooling spatially distributed samples to one volume-weighted sample of161

250 ml. Afterwards, sub-samples were filtered with nitrocellulose filters of 0.45 µm 162

pore size (MF-Millipore HAWP04700) previously dried in a desiccator and weighed. 163

After filtering, the filters were dried again in an oven at low temperature (~30ºC during 164

24 h) to prevent the calcination of organic particulates and then weighed again. Mean 165

particulate matter flux per area (kg ha-1) was calculated for each event by taking the 166

weight of each filter (fi, in g) weighted by the volume of each corresponding sample (wi, 167

in l), the area of the funnels for open rainfall and throughfall (FA, in m2) (Equation 1), 168

and the mean basal area for stemflow (BA, in m2) (Equation 2). 169

,

∑

∑10 (1) 170

∑

∑10 (2) 171

Net deposition (ND) was calculated as the difference between the PM fluxes below the 172

canopy (throughfall (PMT) plus stemflow (PMS)) and the PM fluxes in open rainfall 173

(PMP) (Equation 3). 174

(3) 175

In addition, the flux-based stemflow enrichment ratios compared to rainfall (EP,B) 176

(Equation 4) and throughfall (ET,B) (Equation 5) were calculated following Levia and 177

Germer (2015). This parameter is a flux-based ratio which seeks to quantify the extent 178

to which trees concentrate solutes and particulates at their base as: 179

,

(4) 180

9

,

(5) 181

where CS, CP and CT are PM concentrations in stemflow, open rainfall and throughfall, 182

respectively (g l-1); P and T are depth equivalents of rainfall and throughfall, 183

respectively (mm); and SY is stemflow in volume (l). 184

Finally, particulate matter fluxes for four selected events were analysed sequentially in 185

the pine stand. For each selected event, samples of rainfall and throughfall were taken at 186

every 5 mm of rainfall; and stemflow samples, at approximately every 2 litres. The 187

sampling was carried out sequentially by three automatic water samplers (ISCO 188

3700C). Rainfall and throughfall automatic samplers were connected to a plastic funnel 189

(160 mm diameter) and the stemflow automatic sampler was connected to a stemflow 190

collar. Samples were filtered, dried and weighed as described above. 191

2.4. Analysis of particulate matter size distributions 192

Open rainfall, throughfall and stemflow filters of 7 rainfall events were selected for 193

microscope analysis. The selected events were representative of the range of rainfall 194

events and took into account the leafed and leafless periods of oak. To quantify 195

particulate size, a Zeiss LSM 510 Meta-5 Live Duo confocal microscope was used. 196

Similarly, differential interference contrast microscopy (another optical microscopy 197

method) has been employed to obtain particulate sizes and shape measurements (e.g. 198

Billiones et al., 1999). Each filter was scanned by means of a grid of 9x9 tiles and each 199

individual tile was magnified by a 5x lens with a resolution of 1,536 x 1,536 pixels. 200

Post-processing was conducted with ImageJ 1.51g software. Specifically, tiles were 201

stitched together in a single image, and then individual particulate sizes and shapes were 202

calculated. Maximum PM diameters were calculated as the longest distance between 203

any two points along the identified particles. In addition, the roundness of the PM was 204

calculated as the ratio between the maximum and the minimum feret diameters, which 205

10

ranged between 1 (i.e. round or orthogonal) and infinity (i.e. a very thin wafer). Finally, 206

differences in the size distribution among water fluxes were examined through an 207

analysis of the variance and a Tuckey post-hoc test. The confocal microscopy analyses 208

were conducted at the BioImaging Centre at the University of Delaware. All data used 209

in the analysis can be found in the CSIC’s digital repository 210

http://hdl.handle.net/10261/176564 (Cayuela et al., 2019). 211

3. Results 212

3.1. Particulate matter fluxes in open rainfall, throughfall and stemflow 213

For the 36 events analysed, throughfall represented 82% and 84% and stemflow 214

represented 3.1% and 3.4% of the open rainfall for the pine and oak, respectively. 215

Throughfall and stemflow did not significantly differ for PM concentrations. 216

Nevertheless, in both forest stands, the concentration of PM collected below the forest 217

canopy was higher (median value of 7.8 g l-1 in throughfall and 7.6 g l-1 in stemflow) 218

than the concentration of PM collected in open rainfall (median concentration of 3.4 g l-219

1) (Figure 1). Enrichment ratios revealed that stemflow funnelled larger amounts of PM 220

per unit basal area than open rainfall and throughfall. The median PM stemflow 221

enrichment ratios for pine were 9.5 times and 4.2 times larger than open rainfall and 222

throughfall, respectively. Likewise, the corresponding values for oak were 10.1 and 8.3 223

larger than PM in rainfall and throughfall, respectively (Figure 2). The total annual (July 224

2015-July 2016) amount of PM collected below both tree canopies was higher than the 225

58 kg ha-1 of PM collected in open rainfall. In the pine stand, the annual flux of PM was 226

69 kg ha-1 in throughfall and 2.4 kg ha-1 in stemflow. For oak, the annual flux of PM was 227

84 kg ha-1 in throughfall and 1.5 kg ha-1 in stemflow. The load of PM in rainfall was 228

related to the origin of the air mass (Figure 3a). 229

11

Precipitation originating from air masses coming from the West (Atlantic sea) had the 230

lowest content of PM in open rainfall, with median PM fluxes of 0.13 kg ha-1. On the 231

contrary, rainfall produced during Saharan air mass intrusions had the highest 232

concentrations of PM, with median PM fluxes 25 times higher (4.89 kg ha-1). High PM 233

fluxes were also observed for air mass trajectories from the northeast crossing the 234

European continent, with median PM fluxes of 1.64 kg ha-1 (Figure 3b). Overall, the 235

deposition of PM in the study area was dominated by Saharan dust intrusions. The 6 236

rainfall events that occurred during Saharan dust intrusions accounted for almost 60% 237

of the total annual PM flux in rainfall. 238

The rate of PM was not evenly distributed throughout the year. May was the month with 239

the highest amount of PM per surface area in open rainfall (Figure 4b), coinciding with 240

the rainiest month of the studied period (Figure 4a). However, throughfall in both 241

species followed a similar distribution to open rainfall (Figures 4c). Differences 242

between species were generally small, except during May, when PM for oak was twice 243

as high as for pine. The total contribution of PM in the study area due to stemflow was 244

almost 10-fold lower than PM due to throughfall (Figures 4d). A significant correlation 245

between precipitation and PM in rainfall, throughfall and stemflow for both species (p < 246

0.05) was found. Nevertheless, the goodness of the fit (R2) of these correlations 247

decreased when examined at the weekly scale (Figure S2, supporting information). 248

Overall, net deposition was positive throughout the year. Total net deposition into soil 249

was higher for oak than pine; pine released 12.7 kg ha-1 to the soil layers via throughfall 250

plus stemflow (Figure 5b) while oak released annually 27.4 kg ha-1 (Figure 5c). The 251

greatest differences between species occurred mainly during spring and summer, 252

coinciding with the leafed season for oak. Nevertheless, the highest retention of PM on 253

12

leaves and stems of pine and oak occurred during Saharan dust events, which were 254

associated with low rainfall intensities (Figure 5a). 255

3.2. Within-storm particulate matter fluxes 256

The analysis of intra-event fluxes in pine showed that particulate matter fluxes varied 257

not only between events, but also within the event. Overall, stemflow enrichment ratios 258

of PM were higher when compared to rainfall (EPB) than when compared to throughfall 259

(ETB). Rainfall intensity controlled enrichment ratios in different ways. For example, 260

during low-intensity events (Figure 6a), enrichment ratios increased throughout the 261

event and differences between EPB and ETB decreased, whereas high-intensity rainfall 262

events resulted in peaks of enrichment ratios, mainly for EPB, although this was also 263

observable in ETB. This trend was clearly seen in Figure 6c, when peak rainfall intensity 264

coincided with maximum PM enrichment. Nonetheless, this tendency was also observed 265

in events of varying intensity (Figures 6b and d). 266

3.3. Particulate matter size distributions 267

The maximum diameters of 672,906 individual particulates were analysed to examine 268

differences between open precipitation, throughfall and stemflow for pine and oak 269

(Figure S3, supporting information). In general, no significant differences were found 270

among the size distribution classes of water fluxes (F4,10 = 0.005, p > 0.05). 271

Nevertheless, maximum median throughfall PM diameters were slightly higher than 272

open rainfall ones. The largest diameters were observed for oak during the leafed period 273

when the maximum median PM size increased by 2 µm. The maximum PM diameters 274

in open rainfall and in throughfall in oak during the leafless season were very similar. In 275

contrast, maximum stemflow PM diameters tended to be smaller than open rainfall 276

ones, except for oak during the leafed period. Although the lower quantile of particulate 277

13

diameters was similar among water fluxes, there was greater variability among the 278

higher quantiles (Table 1). 279

All particulate diameter frequency distributions were skewed to the right (Figure 7). 280

Within each water flux, significant differences were found in the particulate count per 281

size group distribution (0.45-2.5, 2.5-10, and 10-500µm) (F2,12 = 83.15, p < 0.01); 282

however, no significant differences were observed in the number of particulates among 283

water fluxes (F4,10 = 0.051, p > 0.05). Nevertheless, some trends were observed, the 284

highest number of particulates was found in throughfall and stemflow for pine fluxes. In 285

addition, stemflow had a higher proportion of small particulates (0.45 < PM < 2.5 µm) 286

than throughfall, but a lower proportion of coarse PM for both pine and oak (Figure 8). 287

Results showed that, in general, PM tended to be round; the less round particulates were 288

found in stemflow for pine and in throughfall for oak during the leafed period. A 289

negative exponential relationship was observed between roundness and PM size (r2 = 290

0.70, p < 0.05); in general, the coarse PM of water fluxes tended to be less round than 291

fine PM. 292

4. Discussion 293

4.1. Spatio-temporal variations of particulate matter fluxes 294

The load of particulate matter arriving at the study site varied significantly, depending 295

on the origin of its air mass. Like Castillo et al. (2017), we found that rain-laden 296

Atlantic advections cleaned the atmosphere, leading to the lowest content of PM in open 297

rainfall. On the contrary, the load of PM increased for rainfall coming from the 298

Mediterranean basin or from other regions of the European continent. These air masses 299

would have been more affected by pollutants that build up during long dry spells. 300

Particularly important were the air masses coming from North Africa. They represented 301

more than half of the total PM reaching the study area, even though they only accounted 302

14

for 16% of the events. On the Iberian peninsula, African intrusions usually occur in 303

spring-summer (Castillo et al., 2017), but, as observed here, sporadic intrusions may 304

occur throughout the year, leading to disproportionate outbursts of PM on specific days 305

with significant inputs of minerals such as phyllosilicates, quartz or calcite (Lequy et 306

al., 2018) and nutrients, particularly phosphorous, which in some ecosystems could be a 307

limiting factor (Morales-Baquero & Pérez-Martínez, 2016). Total PM inputs increased 308

with rainfall due to enhanced wet deposition and the redistribution of previously dry-309

deposited particulate material (Grantz et al., 2003). When analysed at the weekly scale, 310

the goodness of the fit was lower, whereas at the monthly scale the fit was stronger. 311

These differences were related to the observed Saharan dust outbursts that sometimes 312

occurred during events of low rainfall depth. 313

4.2. Particulate matter fluxes below the canopy 314

Throughfall is the dominant flux of water below trees, which is why studies analysing 315

PM fluxes in forested areas have often focused on it (e.g. Cape, 2008; Lindberg et al., 316

1986; Lovett & Lindberg, 1984). When expressed by area, throughfall PM inputs were 317

1.2- and 1.4-fold greater than open rainfall and 28- and 56-fold greater than stemflow in 318

pine and oak, which is similar to the findings of Lequy et al. (2014). However, despite 319

representing only a small proportion of rainfall, the higher enrichment ratios of 320

stemflow (compared to rainfall and throughfall) underscore its importance as a localized 321

input source of water and PM to soil near the trunks. These findings corroborate 322

previous work on stemflow (e.g. Carlyle-Moses et al., 2018; Levia & Germer, 2015; 323

Michalzik et al., 2016), which highlighted the importance of stemflow as a preferential 324

flow path of chemically enriched water to the soil. Higher stemflow enrichment ratios 325

are most probably due to the longer path that stemflow has to take before reaching the 326

soil. The increase of the contact time of water with leaves and stems may enhance the 327

15

enrichment of water with PM. This is especially true during higher rainfall intensities 328

when the mobilization of previously dry-deposited PM and greater scouring of the bark 329

surface increase input of particulates to the soil, causing short-term changes in the rates 330

of nutrient cycling due to the prompt availability to PM (Lovett & Ruesink, 1995). 331

Canopy, bark structure and leaf characteristics partly explained the differences between 332

the species studied. Like other researchers (Beckett et al., 1998; Grantz et al., 2003; 333

Sæbø et al., 2012), we found that, overall, pine accumulated more particulates than 334

broadleaved species, although this did depend strongly on rainfall amount and intensity. 335

Low rainfall and low-intensity events were less effective when removing PM previously 336

deposited on leaves. In addition, higher enrichment ratios in stemflow for oak suggested 337

a higher mobilization of PM on its stems, which increased the availability of 338

particulates and nutrients at the base of oak trees. On the contrary, as pine retained more 339

PM and keep their needles for several years, the recycling of PM accumulated on their 340

needles is slower (Dzierżanowski et al., 2011). 341

4.3. Particulate matter size distributions 342

As in other studies (Levia et al., 2013; Song et al., 2015), PM size distributions were 343

skewed to the right, with a larger amount of fine particulates (< 6 µm), confirming that 344

our study site corresponded to a rural area far from busy roads, where the proportion of 345

coarse PM retained in leaves would be higher (Beckett et al., 2000). Seasonality was 346

also found to influence particulate matter size distribution between species. In general, 347

maximum PM diameters were higher for throughfall in pine and oak during the leafed 348

period, whereas, during the leafless period, the size of PM in oak become closer to that 349

in open rainfall, indicating that leaves were able to retain and enhance the interaction 350

and aggregation between PM. This aggregation could happen between the plant surfaces 351

and the wax layer (Dzierżanowski et al., 2011). The smaller maximum PM diameter in 352

16

stemflow than in throughfall may be the result of its scouring with the bark during its 353

pathway to the soil. Levia et al. (2013) also found higher maximum particulate 354

diameters in throughfall than stemflow. Since large quantities of particulate sulfate and 355

nitrate can be found on particulates in the 0.1 – 1.0 µm size range (Grantz et al., 2003), 356

the high volume of water enriched in fine PM that reaches the base of trees as stemflow 357

could enhance the near-trunk input of soil nutrients, energy flows and the spatial 358

patterning of biogeochemical processes (Lovett & Ruesink, 1995; Michalzik et al., 359

2016). Of course, the larger overall deposition of PM via throughfall, despite it being 360

less enriched in PM than stemflow per unit trunk basal area, is of importance since it is 361

also a source of sulfates, nitrates, base cations, and heavy metals to forested ecosystems 362

(Lindberg et al., 1982, 1986; Lovett & Lindberg, 1984). 363

5. Conclusions 364

Particulate matter content below the canopy of a Scots pine and a downy oak stand was 365

almost 1.5 times higher than in open rainfall. Overall, the content of PM in rain, 366

throughfall and stemflow correlated with rainfall amount, although Saharan dust events 367

increased PM content disproportionately: only 16% of rainfall events occurred during 368

Saharan dust intrusions, but these represented almost 60% of the total PM in the study 369

area. Overall, the concentration of PM was similar between throughfall and stemflow, 370

yet the higher flux-based stemflow enrichment ratios confirmed its importance as the 371

preferential flow path of chemically enriched water to the soil in the near-trunk zone. 372

Within an event, rainfall intensity enhanced the mobilization of PM to the soil. Further, 373

the interaction between PM and vegetative surfaces was found to be a key factor 374

determining the amount and size of PM. The presence of leaves on oak increased the 375

diameter and the content of PM released by throughfall. On the other hand, the diameter 376

of PM in stemflow was smaller than in open rainfall and throughfall, indicating a 377

17

possible scouring of the particulates by the bark during their transport to the soil. This 378

study highlights the importance of considering stemflow in nutrient and energy flux 379

studies, as it is the major source of PM at the base of trees. Future research will need to 380

examine the elemental analysis of PM content, as this will provide valuable information 381

about nutrients and pollutants reaching the forest floor. 382

6. Acknowledgements 383

This research was supported by the projects TransHyMed (CGL2016-75957-R 384

AEI/FEDER, UE) and MASCC-DYNAMITE (PCIN-2017-061/AEI) funded by the 385

‘‘Agencia Estatal de Investigación”. C. Cayuela was the beneficiary of a pre-doctoral 386

FPI grant (BES-2014-070609) and a pre-doctoral mobility grant (EEBB-I-16-11510), 387

both funded by the Spanish Ministry of Economy and Competitiveness. P. Llorens was 388

the beneficiary of (PRX15/00326) funding from the Spanish Ministry of Education, 389

Culture and Sport for professors and senior researchers working in foreign universities 390

and research centres. Support provided by the members of the Bioimaging centre of the 391

University of Delaware is gratefully acknowledged. We are grateful too to the members 392

of the Surface Hydrology and Erosion group of the IDAEA-CSIC, especially Elisenda 393

Sánchez-Costa for her help during fieldwork and Jordi Bellés for his assistance in the 394

laboratory. Finally, we want to thank Michael Eaude for reviewing the English. Data 395

can be download from the DIGITAL.CSIC repository 396

http://hdl.handle.net/10261/176564 in Cayuela et al. (2019). 397

7. References 398

Avila, A., Queralt-Mitjans, I., & Alarcón, M. (1997). Mineralogical composition of 399

African dust delivered by red rains over northeastern Spain. Journal of Geophisical 400

Research, 102(D18), 977–996. https://doi.org/https://doi.org/10.1029/97JD00485 401

Beckett, P., Freer-Smith, P. H., & Taylor, G. (1998). Urban woodlands: Their role in 402

18

reducing the effects of particulate pollution. Environmental Pollution, 99(3), 347–403

360. https://doi.org/10.1016/S0269-7491(98)00016-5 404

Beckett, P., Freer-Smith, P. H., & Taylor, G. (2000). Effective Tree Species for Local 405

Air-Quality Management. Journal of Arboriculture, 26(1), 12–19. 406

Billiones, R. G., Tackx, M. L., & Daro, M. H. (1999). The geometric features, shape 407

factors and fractal dimensions of suspended particulate matter in the Scheldt 408

estuary (Belgium). Estuarine, Coastal and Shelf Science, 48(3), 293–305. 409

https://doi.org/10.1006/ecss.1999.0420 410

Boy, J., & Wilcke, W. (2008). Tropical Andean forest derives calcium and magnesium 411

from Saharan dust. Global Biogeochemical Cycles, 22(1), 1–11. 412

https://doi.org/10.1029/2007GB002960 413

Cape, J. N. (2008). Interactions of forests with secondary air pollutants: Some 414

challenges for future research. Environmental Pollution, 155(3), 391–397. 415

https://doi.org/10.1016/j.envpol.2008.01.038 416

Carlyle-Moses, D. E., Iida, S., Germer, S., Llorens, P., Michalzik, B., Nanko, K., et al. 417

(2018). Expressing Stemflow Commensurate with its Ecohydrological Importance. 418

Advances in Water Resources, 121, 472–479. 419

https://doi.org/10.1016/J.ADVWATRES.2018.08.015 420

Castillo, S., Alastuey, A., Cuevas, E., Querol, X., & Avila, A. (2017). Quantifying dry 421

and wet deposition fluxes in two regions of contrasting African influence: The NE 422

Iberian Peninsula and the Canary Islands. Atmosphere, 8(5). 423

https://doi.org/10.3390/atmos8050086 424

Cayuela, C., Llorens, P., Sánchez-Costa, E., & Latron, J. (2018). Modification of the 425

isotopic composition of rainfall by throughfall and stemflow: the case of Scots pine 426

and Downy oak forests under Mediterranean conditions. Ecohydrology, 11(e2025). 427

19

https://doi.org/https://doi.org/10.1002/eco.2025 428

Cayuela, C., Levia, D. F., Latron, J., & Llorens, P. (2019). Throughfall and stemflow 429

particulate matter fluxes and size distribution [Dataset]. DIGITAL.CSIC. 430

https://doi.org/http://dx.doi.org/10.20350/digitalCSIC/8613 431

Dia, A., Chauvel, C., Bulourde, M., & Gérard, M. (2006). Eolian contribution to soils 432

on Mount Cameroon: Isotopic and trace element records. Chemical Geology, 433

226(3–4), 232–252. https://doi.org/10.1016/j.chemgeo.2005.09.022 434

Dzierżanowski, K., Popek, R., Gawrónska, H., Sæbø, A., & Gawrónski, S. W. (2011). 435

Deposition of Particulate Matter of Different Size Fractions on Leaf Surfaces and 436

in Waxes of Urban Forest Species. International Journal of Phytoremediation, 437

13(10), 1037–1046. https://doi.org/10.1080/15226514.2011.552929 438

Fowler, D., Cape, J. N., Unsworth, M. H., Mayer, H., Crowther, J. M., Jarvis, P. G., et 439

al. (1989). Deposition of Atmospheric Pollutants on Forests. Philosophical 440

Transactions of the Royal Society B: Biological Sciences, 324(1223), 247–265. 441

https://doi.org/10.1098/rstb.1989.0047 442

Franzén, L. G., Hjelmroos, M., Kållberg, P., Brorström-Lundén, E., Juntto, S., & 443

Savolainen, A.-L. (1994). Tihe ‘Yellow Snow’ Episode of Northern Fennoscandia, 444

March 1991 a Case Study of Long-Distance Transport of Soil, Pollen and Stable 445

Organic Compounds. Atmospheric Environment, 28(22), 3587–3604. 446

Grantz, D. A., Garner, J. H. B., & Johnson, D. W. (2003). Ecological effects of 447

particulate matter. Environment International, 29(2–3), 213–239. 448

https://doi.org/10.1016/S0160-4120(02)00181-2 449

Hargreaves, G. H., & Samani, Z. A. (1982). Estimating potential evapo-transpiration. 450

Journal of the Irrigation and Drainage Division-Asce, 108, 225–230. 451

Kutuzov, S., Shahgedanova, M., Mikhalenko, V., Ginot, P., Lavrentiev, I., & Kemp, S. 452

20

(2013). High-resolution provenance of desert dust deposited on Mt. Elbrus, 453

Caucasus in 2009-2012 using snow pit and firn core records. Cryosphere, 7(5), 454

1481–1498. https://doi.org/10.5194/tc-7-1481-2013 455

Latron, J., Llorens, P., Soler, M., Poyatos, R., Rubio, C., Muzylo, A., et al. (2010). 456

Hydrology in a Mediterranean mountain environment – the Vallcebre research 457

basins (northeastern Spain). I. 20 years of investigation of hydrological dynamics. 458

IAHS-AISH Publication, 336, 38–43. 459

Latron, J., Soler, M., Llorens, P., Nord, G., & Gallart, F. (2010). Hydrology in a 460

Mediterranean mountain environment – the Vallcebre research basins (northeastern 461

Spain). II. Rainfall runoff relationships and runoff processes. IAHS-AISH 462

Publication, 336, 151–156. 463

Lequy, É., Nicolas, M., Conil, S., & Turpault, M. P. (2013). Relationship between 464

atmospheric dissolved deposition and mineral dust deposition in French forests. 465

Water, Air, and Soil Pollution, 224(9). https://doi.org/10.1007/s11270-013-1680-4 466

Lequy, É., Calvaruso, C., Conil, S., & Turpault, M. P. (2014). Atmospheric particulate 467

deposition in temperate deciduous forest ecosystems: Interactions with the canopy 468

and nutrient inputs in two beech stands of Northeastern France. Science of the 469

Total Environment, 487(1), 206–215. 470

https://doi.org/10.1016/j.scitotenv.2014.04.028 471

Lequy, É., Avila, A., Boudiaf Nait Kaci, M., & Turpault, M. P. (2018). Atmospheric 472

deposition of particulate matter between Algeria and France: Contribution of long 473

and short-term sources. Atmospheric Environment, 191(February), 181–193. 474

https://doi.org/10.1016/j.atmosenv.2018.08.013 475

Levia, D. F., & Frost, E. E. (2003). A review and evaluation of stemflow literature in 476

the hydrologic and biogeochemical cycles of forested and agricultural ecosystems. 477

21

Journal of Hydrology, 274(1–4), 1–29. https://doi.org/10.1016/S0022-478

1694(02)00399-2 479

Levia, D. F., & Germer, S. (2015). A review of stemflow generation dynamics and 480

stemflow-environment interactions in forests and shrublands. Reviews of 481

Geophysics, 53(3), 673–714. https://doi.org/10.1002/2015RG000479 482

Levia, D. F., Michalzik, B., Bischoff, S., Näthe, K., Legates, D. R., Gruselle, M. C., & 483

Richter, S. (2013). Measurement and modeling of diameter distributions of 484

particulate matter in terrestrial solutions. Geophysical Research Letters, 40(7), 485

1317–1321. https://doi.org/10.1002/grl.50305 486

Lindberg, S. E., Harriss, R., & Turner, R. (1982). Atmospheric deposition of metals to 487

forest vegetation. Science, 215, 1609–1611. 488

https://doi.org/10.1126/science.215.4540.1609 489

Lindberg, S. E., Lovett, G. M., Richter, D. D., & Johnson, D. W. (1986). Atmospheric 490

Deposition and Canopy Interactions of Major Ions in a Forest. Science, (5), 491

231:141. 492

Llorens, P., & Gallart, F. (2000). A simplified method for forest water storage capacity 493

measurement. Journal of Hydrology, 240(1), 131–144. 494

https://doi.org/10.1016/S0022-1694(00)00339-5 495

Llorens, P., Gallart, F., Cayuela, C., Roig-Planasdemunt, M., Casellas, E., Molina, A. J., 496

et al. (2018). What have we learnt about Mediterranean catchment hydrology? 30 497

years observing hydrological processes in the Vallcebre research catchments. 498

Geographical Research Letters, 44(2), 475–502. 499

https://doi.org/http://doi.org/10.18172/cig.3432 500

Lovett, G. M., & Lindberg, S. E. (1984). Dry Deposition and Canopy Exchange in a 501

Mixed Oak Forest as Determined by Analysis of Throughfall. Journal of Applied 502

22

Ecology, 21(3), 1013–1027. 503

Lovett, G. M., & Ruesink, A. E. (1995). Carbon and nitrogen mineralization from 504

decomposing gypsy moth frass. Oecologia, 104(2), 133–138. 505

https://doi.org/10.1007/BF00328577 506

McDonald, A. G., Bealey, W. J., Fowler, D., Dragosits, U., Skiba, U., Smith, R. I., et al. 507

(2007). Quantifying the effect of urban tree planting on concentrations and 508

depositions of PM10 in two UK conurbations. Atmospheric Environment, 41(38), 509

8455–8467. https://doi.org/10.1016/j.atmosenv.2007.07.025 510

Menéndez, I., Díaz-Hernández, J. L., Mangas, J., Alonso, I., & Sánchez-Soto, P. J. 511

(2007). Airborne dust accumulation and soil development in the North-East sector 512

of Gran Canaria (Canary Islands, Spain). Journal of Arid Environments, 71(1), 57–513

81. https://doi.org/10.1016/j.jaridenv.2007.03.011 514

Michalzik, B., Levia, D. F., Bischoff, S., Näthe, K., & Richter, S. (2016). Effects of 515

aphid infestation on the biogeochemistry of the water routed through European 516

beech (Fagus sylvatica L.) saplings. Biogeochemistry, 129(1–2), 197–214. 517

https://doi.org/10.1007/s10533-016-0228-2 518

Middleton, N. J. (2017). Desert dust hazards: A global review. Aeolian Research, 24, 519

53–63. https://doi.org/10.1016/j.aeolia.2016.12.001 520

Molina, A. J., Llorens, P., Garcia-Estringana, P., Moreno de las Heras, M., Cayuela, C., 521

Gallart, F., & Latron, J. (2019). Contributions of throughfall, forest and soil 522

characteristics to near-surface soil water-content variability at the plot scale in a 523

mountainous Mediterranean area. Science of the Total Environment, 647, 1421–524

1432. https://doi.org/10.1016/j.scitotenv.2018.08.020 525

Morales-Baquero, R., & Pérez-Martínez, C. (2016). Saharan versus local influence on 526

atmospheric aerosol deposition in the southern Iberian Peninsula: Significance for 527

23

N and P inputs. Global Biogeochemical Cycles, 30, 501–513. 528

https://doi.org/10.1002/ 2015GB005254 529

Muhs, D. R., Budahn, J. R., Prospero, J. M., & Carey, S. N. (2007). Geochemical 530

evidence for African dust inputs to soils of western Atlantic islands: Barbados, the 531

Bahamas, and Florida. Journal of Geophysical Research: Earth Surface, 112(2), 532

1–26. https://doi.org/10.1029/2005JF000445 533

Muhs, D. R., Budahn, J. R., Skipp, G., Prospero, J. M., Patterson, D., & Bettis III, E. A. 534

(2010). Geochemical and mineralogical evidence for Sahara and Sahel dust 535

additions to Quaternary soils on Lanzarote, eastern Canary Islands, Spain. Terra 536

Nova, 22, 399–410. https://doi.org/10.1111/j.1365-3121.2010.00949.x 537

Muzylo, A., Llorens, P., & Domingo, F. (2012). Rainfall partitioning in a deciduous 538

forest plot in leafed and leafless periods. Ecohydrology, 5(6), 759–767. 539

https://doi.org/10.1002/eco.266 540

Perrino, C. (2010). Atmospheric particulate matter. Biophysics and Bioengineering 541

Letters, 3(1), 35–43. 542

Prospero, J. M. (1999). Long-term measurements of the transport of African mineral 543

dust to the southeastern United States: Implications for regional air quality. 544

Journal of Geophysical Research Atmospheres, 104(D13), 15917–15927. 545

https://doi.org/10.1029/1999JD900072 546

Rolph, G., Stein, A. F., & Stunder, B. (2017). Real-time Environmental Applications 547

and Display sYstem: READY. Environmental Modelling and Software, 95, 210–548

228. https://doi.org/10.1016/j.envsoft.2017.06.025 549

Sæbø, A., Popek, R., Nawrot, B., Hanslin, H. M., Gawrónska, H., & Gawronski, S. W. 550

(2012). Plant species differences in particulate matter accumulation on leaf 551

surfaces. Science of the Total Environment, 427–428, 347–354. 552

24

https://doi.org/10.1016/j.scitotenv.2012.03.084 553

Song, Y., Maher, B. A., Li, F., Wang, X., Sun, X., & Zhang, H. (2015). Particulate 554

matter deposited on leaf of five evergreen species in Beijing, China: Source 555

identification and size distribution. Atmospheric Environment, 105, 53–60. 556

https://doi.org/10.1016/j.atmosenv.2015.01.032 557

Stein, A. F., Draxler, R. R., Rolph, G., Stunder, B., Cohen, M. D., & Ngan, F. (2015). 558

Noaa’s hysplit atmospheric transport and dispersion modeling system. Bulletin of 559

the American Meteorological Society, 96(12), 2059–2077. 560

https://doi.org/10.1175/BAMS-D-14-00110.1 561

Swap, R., Garstang, M., Greco, S., Talbot, R., & Kållberg, P. (1992). Saharan dust in 562

the Amazon Basin. Tellus B: Chemical and Physical Meteorology, 44(2), 133–149. 563

https://doi.org/10.3402/tellusb.v44i2.15434 564

USEPA. (2004). Air Quality Criteria for Particulate Matter. Air Quality Criteria for 565

Particulate Matter, I(October), 900. https://doi.org/EPA 600/P-99/002aF-bF 566

Weathers, K. C., & Ponette-González, A. G. (2011). Atmospheric Deposition. In D. F. 567

Levia, D. Carlyle-Moses, & T. Tanaka (Eds.), Forest Hydrology and 568

Biogeochemistry (Vol. 216, pp. 357–370). Springer. https://doi.org/10.1007/978-569

94-007-1363-5 570

Yaalon, D., & Ganor, E. (1979). East Mediterranean trajectories of dust-carrying storms 571

from the Sahara and Sinai. In G. Morales (Ed.), Saharan Dust: Mobilisation, 572

Transport and Deposition (pp. 187–193). John Wiley, New York. 573

574

575

55

5

5

5

5

5

5

5

Tabl578 stem579

Open r

Stemfl

Pine

Oak le

Oak le

Throug

Pine

Oak le

Oak le

579

Figu580

581

Figu585

and 586

corre587

1.5 ti588

le 1. Statistimflow and th

M

rainfall

low

eafed

eafless

ghfall

eafed

eafless

ure captions

ure 1. Boxp

stemflow

espond to th

imes the int

ical summarhroughfall in

Maximu

edian Low

quant

7.2

6.7

8.1

5.4

8.1

9.4

7.2

s:

plots of ann

in pine an

he 25th and 7

terquartile r

ry of maximn pine and o

um Particulate der tile

Higher quantile

4.2 13.6

4.1 11.4

4.5 13.0

3.6 9.8

4.5 13.6

5.0 17.0

4.2 12.8

nual PM con

nd oak. Th

75th percent

range.

25

mum particuoak during l

diameter (µm)

Mean

Sd

6 12.7

4 10.0

0 10.7

8 8.2

6 11.8

0 15.6

8 11.0

ncentration

he horizont

tiles and wh

ulate diametleafed and l

Standard deviation

Mf

20.9

13.2

12.0

9.8

15.6

22.9

15.0

s in g l-1 fo

tal line ind

hiskers repr

ters for openeafless peri

Max/Min feret ratio

S

1.66

1.70

1.69

1.60

1.65

1.75

1.63

or open rain

dicates the

resent value

n rainfall, iods.

Skewness

8.1

7.2

8.9

7.1

8.3

6.3

7.5

nfall, throug

median, b

es that fall w

ghfall

boxes

within

5

5

5

5

5

5

5

5

5

5

586

Figu591

ratio 592

throu593

medi594

fall w595

592

Figu595

with 596

load 597

ure 2. Boxp

compared

ughfall flux

ian, boxes c

within 1.5 ti

ure 3. (a) 5-

the HYSP

of the PM f

plots of stem

d to precip

x. Note that

correspond t

imes the int

-day backw

LIT dispers

flux. (b) Bo

mflow enric

pitation flu

t the y-axis

to the 25th a

erquartile ra

ard trajecto

sion model

oxplots of th

26

chment ratio

ux, and ET

is in log s

and 75th per

ange and ci

ories ending

for the 36

he PM load

os for pine

TB is enric

scale. The h

rcentiles, wh

rcles repres

g at the Vall

6 events sel

for events g

and oak. E

chment rati

horizontal li

hiskers repr

sent outliers

lcebre Rese

lected. Colo

grouped by

EPB is enrich

io compare

ine indicate

resent value

s.

earch catchm

our indicate

origin: We

hment

ed to

es the

es that

ments

es the

st (16

5

5

5

5

5

6

6

6

even599

boxp600

perce601

and c602

600

Figu603

throu604

for e605

nts, 243.5 m

plot, the hor

entiles, whi

circles repre

ure 4. Mon

ughfall for p

ach month

mm), East (1

rizontal line

iskers repre

esent outlier

nthly distrib

pine and oa

is shown by

14 events, 5

e indicates t

esent values

rs.

bution of (a

ak, (d) PM

y vertical lin

27

572 mm) an

the median,

s that fall w

a) Rainfall,

in stemflow

nes. Saharan

nd South (6

, boxes corr

within 1.5 ti

, (b) PM in

w for pine a

n dust even

6 events, 20

respond to

imes the int

n open rain

and oak. Sta

ts occurred

05 mm). In

the 25th and

nterquartile

nfall, (c) P

andard devi

during Feb

n each

d 75th

range

PM in

iation

bruary

6

6

6

6

6

(1 ev605

May 606

606

Figu608

oak. 609

vent), May (

and Leafle

ure 5. Time

Colours ind

(2 events), J

ss period on

-series of (

dicate the lo

June (2 eve

n October.

a) daily rai

oad of the P

28

ents) and Au

infall and ev

PM in rainfa

ugust (1 eve

vent net de

all for each e

ent). Leafed

position in

event.

d period star

(b) pine an

rts on

nd (c)

6

6

6

609

Figu611

(EPB)612

ure 6. Intra

) and throug

-event time

ghfall (ETB)

e-series of

in the pine

29

stemflow e

e stand.

enrichment ratios comppared to ra

ainfall

6

6

6

6

6

6

6

6

612

Figu617

throu618

relati619

illust620

diam621

618

Figu619

ure 7. Par

ughfall and

ive to the

tration, ther

meter was fo

ure 8. Relati

rticulate m

stemflow i

total num

re are only 1

ound.

ive number

atter diam

in pine and

mber of par

19 classes r

of particula

30

meter freque

d oak for se

articulates i

represented

ates by flux

ency distri

even rainfa

in each siz

in the Figu

and diamet

ibutions of

ll events. T

ze class. F

re, but PM

ter class.

f open rai

The frequen

For purpose

of up to 50

infall,

ncy is

es of

00 µm

6

66

6

6

6

6

Supp620

621 Figu625

used 626

used 627

conn628

626

porting info

ure S1. (a) T

to sample

to sequen

nected to a s

ormation:

Throughfall

rainfall and

ntially sam

stemflow rin

l collector, (

d throughfal

mple stemflo

ng.

31

(b) stemflow

ll every 5 m

ow, in tha

w collector

mm of rainf

at case the

and (c) seq

fall. A simil

e sequentia

quential coll

lar approach

al collector

lector

h was

r was

66

6

6

6

6

627 Figu631

for o632

Corre633

throu634

632

ure S2. (Lef

open rainfal

elations bet

ughfall and

ft) Correlati

ll, throughf

tween week

stemflow, i

ions betwee

fall and stem

kly rainfall

in the pine a

32

en monthly

mflow, in

and weekl

and oak fore

rainfall and

the pine an

ly particula

est plots.

d monthly p

nd oak fore

ate matter fo

particulate m

est plots. (R

for open rai

matter

Right)

infall,

66

6

6

6

6

633 Figu637

Meta638

name639

pines640

638

ure S3. Part

a-5 Live Du

es at the top

s (Sf P), Thr

iculate matt

uo confocal

p refer to O

roughfall in

ter filters fo

microscop

pen rainfall

n oaks (Th O

33

or the 7 eve

pe. Number

l (Rain), Th

O) and stem

ents examin

in the left

hroughfall in

mflow in oak

ned with the

side indicat

n pines (Th

ks (Sf O).

e Zeiss LSM

te the event

h P), Stemflo

M 510

t, and

ow in