Importance of Fog and Cloud Water Contributions to Soil Moisture in the Andean P á ramo

: P á ramos are particular ecosystems of the Tropical Andes, where fog and low-intensity rainfall such as drizzle are commonly frequent—but the contribution of these water sources to soil water replenishment and discharge is not yet clear, mainly because the development of techniques for separating fog from drizzle and wind-driven rainfall has been challenging. Fog was measured with a cylindrical Juvik gauge and types of precipitation other than fog with a high-resolution disdrometer. Soil moisture was measured at 100 mm depth by means of Water Content Reﬂectometers, then Effective precipitation (EP) was calculated. We categorized events as two types: fog only (FO) and cloud water (CW). We found that in the case of FO events, only small amounts reached the soil (EP ranged between 0.1 and 0.2 mm); in contrast, greater amounts of EP originated from CW events (maximum value of 4.3 mm). Although we found that FO events are negligible for stream water contribution; they are ecologically important for maintaining high relative humidity, low net radiation, and consequently low evapotranspiration rates. Our research provides new insights into the hydrological role of fog, enabling us to better understand to what extent its input inﬂuences the water resources of the Andean p á ramo. Author Contributions: Conceptualization, G.B., P.C. R.C.; methodology, G.B., P.C. A.O.-S.; software, G.B.; formal analysis, G.B.; data curation, G.B.; writing—original draft preparation, G.B.; writing—review and editing, P.C., A.O.-S., B.P.W. and R.C.; supervision, P.C. and R.C.; funding


Introduction
The tropical Andes extend over approximately 1.5 million km 2 , running from 11 • N to 23 • S, and include peaks topping 3500 m.a.s.l, steep slopes, valleys, and plateaus [1]. They encompass a variety of ecosystems-montane forests, páramo grasslands, wetlands, glaciers [2]-that exhibit complex interactions among species and between species and their environments [3]. As a result of the advection of clouds, these mountains are commonly covered in fog [4], the intensity and frequency of which affect water inputs and outputs [5], thus playing an important ecological and hydrological role. At high elevations, fog transported by wind deposits liquid water droplets on plant canopies that then seep down to the soil surface [6,7]. Herein, we define as fog-only (FO) events those that are 100% fog; and as cloud-water (CW) events those consisting of fog combined with drizzle and/or light rain, but fog is the major contributor (percentages between 50% and 99%) [8].
Montane grasslands and shrublands known as páramos, found above the tree line in the upper region of the northern Andes, are characterized by low temperatures and high radiation [9]. They are home to distinct plant communities [1], dominated by tussock grasses and cushion plants with forest patches (Polylepis sp.) growing along water streams. The main soil types are Andosols and Histosols. The wet and cold climate favor organic The study site is located within the Zhurucay Ecohydrological Observatory, on the western side of the Andes in Southern Ecuador (Figure 1a,b). Zhurucay has a drainage area of 7.53 km 2 and elevations ranging from 3200 to 3900 m.a.s.l. The climate is influenced by the Pacific regime and the predominant continental air masses of the Amazon Basin, resulting in conditions that are frequently cloudy and foggy with minor seasonal differences during the year. The climate belongs to the Cfc class (warm temperate climate, fully humid, with cool summer and cold winter) according to Köppen-Geiger classification [29]. Mean annual precipitation is 1210 ± 101 mm year −1 [30], some 50% of which occurs at intensities Hydrology 2022, 9,54 3 of 18 of 2 mm h −1 or less [31]. On average, cloud water flux is about 340.1 mm, and fog occurs 68% of the time [8]. Although fog can be present at any time of the day in the Andean páramo, it is most common in the early morning and at night [8]. The mean annual temperature is 6.1 • C, mean relative humidity is 93.6%, and mean daily total solar radiation is 14 MJ m −2 . Wind blows predominantly from the east or northeast and wind speeds vary seasonally, with a monthly mean of 3.21 m s −1 from October to April and 4.77 m s −1 from May to September. The mean daily actual evapotranspiration is 1.7 mm day −1 , with a cumulative annual value of 622 mm [30,32]. Vegetation at the study site is dominated (>80% coverage) by tussock grasses (genera Calamagrostis and Festuca), commonly known as "pajonal", which are perennial plants that reach heights between 30 and 80 cm. Soils correspond mainly to volcanic origin Andosols [33], which are black loamy soils with a high organic matter content and low bulk density (0.2-0.8 g cm 3 ) [10,34]; their depth ranges from 0.5 m at the hillslope-top to 1.1 m at hillslope-bottom. The surface horizon (Ah) has a clay-loam texture with a granular structure without rocks; organic matter is 57.41% with a density of 0.40 g cm −3 . This horizon has a dark color and a high-water retention capacity: Field capacity is 0.72 cm 3 cm −3 (pF = 2.52), wilting point is 0.5 cm 3 cm −3 (pF = 4.2), and saturation point is 0.8 cm 3 cm −3 (pF = 0) [35]. Below the Ah horizon there is a mineral horizon (C), with a low content of organic matter (6.52%). Field capacity is 0.4 cm 3 cm −3 , wilting point is 0.2 cm 3 cm −3 , and saturation point is 0.65 cm 3 cm −3 [35]. The horizons are acid with a pH ranging between 4 and 6 [10]. The water table is between 2 to 5 m deep, therefore the topsoil is not supplied with groundwater.

Instrumentation
The Zhurucay Ecohydrological Observatory encompasses a monitoring super-site located in a hillslope at 3770 m.a.s.l. (Figure 1c) equipped with a variety of instrumentation, including (1) a meteorological station that records variables such as solar radiation and long-and short-wave net radiation (Campbell Scientific CS300 Apogee

Instrumentation
The Zhurucay Ecohydrological Observatory encompasses a monitoring super-site located in a hillslope at 3770 m.a.s.l. (Figure 1c) equipped with a variety of instrumentation, including (1) a meteorological station that records variables such as solar radiation and long-and short-wave net radiation (Campbell Scientific CS300 Apogee pyranometer), wind velocity and direction (CNR2-Kipp and Zonen Met-One 034B Windset anemometer), atmospheric pressure (Vaisala barometer PTB110), and air temperature and relative humidity (Vaisala thermometer/hygrometer HMP155 with a radiation shield); (2) CS616 Campbell Scientific Water Content Reflectometers (WCRs) installed horizontally at the hillslope-top, at a soil depth of 10 cm and 4 m apart, to measure soil water content. There is no lateral flow at the measurement site. WCRs accuracy is 2.5% with standard calibration, whereas resolution and precision are better than 0.1%; (3) a laser disdrometer (Thies Clima Laser Precipitation Monitor [LPM] 5.4110.00.000 V2.4 × STD, with a resolution of 0.01 mm), which measures the size and fall velocity of drops and is well suited for the detection of different types of precipitation (e.g., drizzle, rain, and mixed precipitation); (4) a tipping-bucket rain gauge (RG) (Texas Electronics tipping-bucket TE525MM, with a resolution of 0.1 mm); (5) a cylindrical modified Juvik fog gauge (FG) (dimensions 40.5 cm high, 14 cm outer diameter, 567 cm 2 cross-sectional area) having a louvered aluminum shade screen collection surface with a 50-cm-diameter protective cover and two plastic funnels-one underneath the screen and one underneath the tipping-bucket rain gauge, which in turn is connected to a bucket via a plastic pipe to drain the collected water; the tipping-bucket rain gauge and the bucket were covered with plastic to avoid vertical precipitation (Figure 1d). To estimate the cloud water flux, the FG effective capture area was defined as the projected capture area (cylinder diameter × height).
In a prior study [8], we measured cloud water interception with different passive fog collectors-two cylindrical types (Juvik and Wire Harp) and two flat-screen types-oriented to the two main wind directions. Our results showed that for our study site, cylindrical Juvik type FG performed the best, yielding better fog and cloud water estimations. For this reason, all the analyses for the current study were done with this FG. All of the variables (e.g., fog, rainfall, volumetric water content, meteorological variables) were continuously recorded at 5-min intervals from September 2017 to December 2019. Weather conditions during the study period are shown in Figure 2.

Selection of Events and Determination of Effective Precipitation
For our analysis, we selected FO and CW events lasting longer than 10 min and separated by 4 h (according to Ochoa-Sánchez et al. [34], this time gap allows for the assumption that tussock grass leaves have dried before the next event).
As noted earlier, we characterized each event as one of two types: (i) Fog-only (FO) events, defined as those captured by the FG but not by either the LPM or the RG (i.e., the event consisted of 100% fog); and (ii) cloud-water (CW) events, defined as those for which the water collected by the FG was greater than the amount recorded by the LPM and the RG (i.e., the fog proportion of the total collected was between 50% and 99%). In addition, to reduce the bias due to wind-driven precipitation, all the CW events with wind speeds higher than 4 m s −1 were excluded from the study [8].
To measure volumetric water content (VWC) in cubic meter per cubic meter with the CS616 WCR output period (P) in microseconds, a calibration curve shown in Equation (1) was developed for this site, as described in Ochoa-Sánchez et al. [34]: For a more detailed information about the calibration procedure, the reader is referred to supplementary material in Ochoa-Sánchez et al. [34]. In order to obtain a data set that shows which events reached the soil, we included only events that met the following criteria: (1) VWC remained at a minimum (0.01 m 3 m 3 ) or did not change for at least 2 h before the event; (2) VWC increased during the event, caused by a water input from drizzle, light rain, and/or high-intensity fog; and (3) the change in VWC during the event had a value higher than the precision and resolution of the WCR (0.1% VWC). These criteria rule out interference from other processes, such as lateral flows.
a plastic pipe to drain the collected water; the tipping-bucket rain gauge and the bucket were covered with plastic to avoid vertical precipitation (Figure 1d). To estimate the cloud water flux, the FG effective capture area was defined as the projected capture area (cylinder diameter × height).
In a prior study [8], we measured cloud water interception with different passive fog collectors-two cylindrical types (Juvik and Wire Harp) and two flat-screen typesoriented to the two main wind directions. Our results showed that for our study site, cylindrical Juvik type FG performed the best, yielding better fog and cloud water estimations. For this reason, all the analyses for the current study were done with this FG. All of the variables (e.g., fog, rainfall, volumetric water content, meteorological variables) were continuously recorded at 5-min intervals from September 2017 to December 2019. Weather conditions during the study period are shown in Figure 2. To calculate the change in soil water storage (SW ch ), we multiplied the increase in VWC (recorded by the average of two WCRs) over the course of the event by the installation depth of the sensor (10 cm). For the purposes of this study, we use the term effective precipitation (EP) to represent the amount of increase in SW ch . Thus, each event was further classified according to EP: Class I for FO and CW events with EP (i.e., SW ch > 0 mm) and Class II for FO and CW events without EP, (i.e., SW ch = 0 mm). This classification yielded 30 Class I events (9 FO and 21 CW) and 52 Class II events.

Identification of Events Consisting of Fog and Rainfall Combined
To determine what other type(s) of precipitation (drizzle, light rain, rain, heavy rain) occurred in combination with fog during the CW events, we used the LPM, which is highly sensitive to small drops. Its data telegram generates a matrix that provides the number of drops detected at 22 different drop sizes (from 0.125 to 8 mm) and at 20 different velocities (from 0 to 10 m s −1 ). First, we obtained the number concentration of raindrops per unit volume per unit size-N(D)-from the LPM counts using Equation (2), where N(Di) is the number concentration in ith size class (mm −3 mm −1 ), Di is the mid-size diameter of the ith class (mm), A is the cross-sectional area of the sensor (m 2 ), t is the measuring time (seconds), nij is the number of drops within the ith size and jth velocity class, and Vj is the fall speed of the jth velocity class (m s −1 ).
Then, we calculated the mean volume diameter Dm (mm) at 5-min intervals from a normalized distribution. In Equation (3), N(D) is the drop size distribution, D is the particle diameter, and D 3 and D 4 are the third and fourth moments calculated for each spectrum.
According to Orellana-Alvear et al. [36], working at the same study site, the 0.1 < Dm [mm] ≤ 0.5 corresponds to drizzle, the 0.5 < Dm [mm] ≤ 1.0 corresponds to light rain, the 1.0 < Dm [mm] ≤ 2.0 corresponds to rain, and the Dm [mm] > 2.0 is defined as heavy rain. On that basis, we calculated the amounts of different kinds of precipitation for each of the CW events (independently of whether they had been classified as Class I or Class II). To identify what type of precipitation commonly reaches the soil, we used a bar plot for the Class I events; and to observe how fog and cloud water depth relate to EP, we calculated the percentages of fog and of other kinds of precipitation for each of the 82 events, then used them to construct a scatter plot.

Relationship between Effective Precipitation and Meteorological Variables
To determine how EP amounts might be influenced by climatic variables and by the various types of precipitation, we plotted the following variables using scatter plots and heat maps: EP amounts; fog depth, drizzle, light rain, and rain; intensity of events; duration of events; mean wind speed, relative humidity, radiation, temperature, dew temperature, and vapor pressure deficit. A linear model and Pearson correlations were performed to analyze the independence of EP and meteorological variables. All these analyses were coded in R x86_64-pc-linux version 3.6.3 software using the lubridate, dplyr, ggplot2, stringr, reshape, psych, corrplot, and heatmap packages.

Selection of Events and Determination of Effective Precipitation
From the dataset, 82 events were selected for analysis: 45 FO events and 37 CW events. Of these, 30 were designated Class I (effective precipitation > 0 mm) and 52 were designated Class II (effective precipitation = 0 mm). Figure 3 shows an example of each classification for each event type, four hours pre event and after event were plotted to observe that changes in volumetric water content are due to the water input during the event and not for other prior events nor by lateral flow processes. Table 1 summarizes the main characteristics of the FO and CW events and the calculated effective precipitation for each. Even though the FO events outnumbered the CW events, their total duration was around three times shorter. The combined total depth from the FO and CW events was 146.6 mm, of which 21.5 mm (15%) reached the soil as EP. Total EP consisted of 57% from the CW events but only 20% from the FO events. The composition of the CW events' total depth was 69% fog, 17% light rain, and 14% drizzle.     Figure 4 shows the total depth for the 30 events that reached the soil, as well as the composition of the CW events. The Class I FO events yielded between 0.1 and 3.1 mm of fog water, whereas the yield of the Class I CW events included between 0.3 and 11.3 mm of water from fog. Drizzle and/or light rain were the most common types of precipitation that contributed to soil moisture increase: half the CW events contained between 5 mm and 16.8 mm, while the other half contained less than 5 mm.  Figure 4 shows the total depth for the 30 events that reached the soil, as well as the composition of the CW events. The Class I FO events yielded between 0.1 and 3.1 mm of fog water, whereas the yield of the Class I CW events included between 0.3 and 11.3 mm of water from fog. Drizzle and/or light rain were the most common types of precipitation that contributed to soil moisture increase: half the CW events contained between 5 mm and 16.8 mm, while the other half contained less than 5 mm. Effective precipitation, measured as an increase in VWC, ranged from 0.1 to 0.2 mm for all Class I FO events. Similarly, for Class I CW events, EP ranged from 0.1 to 0.3 mm, with fog accounting for over 90% of the total ( Figure 5). For cases in which the fog proportion was lower (70%-80%) and the drizzle and light rain proportion was higher, EP was also higher-0.3-2.2 mm; and for cases in which the fog proportion was lower still (50%-70%) with even higher amounts of drizzle and light rain, EP rose as high as 4.3 mm. In other words, the CW events with relatively higher EP (over 1 mm) were those having relatively lower percentages of fog and a higher total depth (over 5 mm). We can infer, then, that EP increases with total amount per event and with greater proportions of drizzle and light rain rather than fog. Effective precipitation, measured as an increase in VWC, ranged from 0.1 to 0.2 mm for all Class I FO events. Similarly, for Class I CW events, EP ranged from 0.1 to 0.3 mm, with fog accounting for over 90% of the total ( Figure 5). For cases in which the fog proportion was lower (70%-80%) and the drizzle and light rain proportion was higher, EP was also higher-0.3-2.2 mm; and for cases in which the fog proportion was lower still (50%-70%) with even higher amounts of drizzle and light rain, EP rose as high as 4.3 mm. In other words, the CW events with relatively higher EP (over 1 mm) were those having relatively lower percentages of fog and a higher total depth (over 5 mm). We can infer, then, that EP increases with total amount per event and with greater proportions of drizzle and light rain rather than fog.

Relationship between Effective Precipitation and Meteorological Variables
The FO and CW events were plotted on heat-maps in ascending order of their EP amounts, together with the mean values of total depth, duration, and intensity of the events, as well as the available meteorological variables (wind speed, air temperature, dew temperature, relative humidity, radiation, and vapor pressure deficit, see Appendix A). For both FO and CW events, duration, intensity, and total event depth were directly related with increasing EP. Figure 6 shows the linear regression for EP and meteorological variables, the coefficient of determination and p-value from the regression indicate a poor fit of the model; values of R 2 < 0.1 were obtained for all the meteorological variables assessed. In addition, Pearson correlation for the same variables showed low coefficients < 0.2 ( Figure 7); which demonstrates that EP and the considered meteorological variables are not linearly related.

Relationship between Effective Precipitation and Meteorological Variables
The FO and CW events were plotted on heat-maps in ascending order of their EP amounts, together with the mean values of total depth, duration, and intensity of the events, as well as the available meteorological variables (wind speed, air temperature, dew temperature, relative humidity, radiation, and vapor pressure deficit, see Appendix A). For both FO and CW events, duration, intensity, and total event depth were directly related with increasing EP. Figure 6 shows the linear regression for EP and meteorological variables, the coefficient of determination and p-value from the regression indicate a poor fit of the model; values of R 2 < 0.1 were obtained for all the meteorological variables assessed. In addition, Pearson correlation for the same variables showed low coefficients < 0.2 ( Figure 7); which demonstrates that EP and the considered meteorological variables are not linearly related.   For both FO and CW events, the most important event character were duration and depth; these are shown in Figure 8, plotted against E of which 44% lasted less than 1 h and 56% lasted from 1 to 8 h-prod the Class II FO events duration was highly variable and did not show Class I FO events, at least 1 h was needed for water to reach the soil. T duced considerably more EP-as high as 4.25 mm for an event of 14 h depth of over 10 mm. Even though the Class I CW events showed high all, they varied widely in both duration and depth. In the páramo, e For both FO and CW events, the most important event characteristics related to EP were duration and depth; these are shown in Figure 8, plotted against EP. The FO events-of which 44% lasted less than 1 h and 56% lasted from 1 to 8 h-produced little EP. For the Class II FO events duration was highly variable and did not show any pattern; for the Class I FO events, at least 1 h was needed for water to reach the soil. The CW events produced considerably more EP-as high as 4.25 mm for an event of 14 h' duration and depth of over 10 mm. Even though the Class I CW events showed higher EP values overall, they varied widely in both duration and depth. In the páramo, events lasting more than 13 h and yielding more than 10 mm are potentially important in the hydrological cycle, because of their higher EP. The four events in this category during our study consisted mainly of fog combined with drizzle and light rain (Figure 8), in the following amounts as measured by the instrumentation: fog = 11.3 mm, 11.2 mm, 6.6 mm, and 5.5 mm; drizzle = 1.9 mm, 0.3 mm, 3.5 mm, and 2.8 mm; and light rain = 1 mm, 5.2 mm, 0.4 mm, and 1.9 mm.

Quantification of Effective Precipitation
As shown in Table 1, of the 146.6-mm total water yield from the 82 FO and CW events we analyzed, 15% became EP (added to soil moisture). In Table 2 we compare our results with findings from other ecosystems (even though the methodologies used for these other studies varied, the estimations obtained are useful for understanding the importance of fog-and cloud-water contributions to soil moisture). During our 26-month study, only a small percentage of the total water collected from the FO events ended up reaching the soil (1.1 mm, or 4.5%)-suggesting that in the Andean páramo, fog water contribution to the net hydrological input is negligible. Fleischbein et al. [37], working in montane forests in Ecuador, reported also negligible fog water inputs. In contrast, the studies of Clark et al. [38] and Liu et al. [39] for cloud and rain forests respectively, found higher values-the latter showing an average annual fog contribution amounting to about 5% of annual rainfall at a site in southwest China.

Quantification of Effective Precipitation
As shown in Table 1, of the 146.6-mm total water yield from the 82 FO and CW events we analyzed, 15% became EP (added to soil moisture). In Table 2 we compare our results with findings from other ecosystems (even though the methodologies used for these other studies varied, the estimations obtained are useful for understanding the importance of fog-and cloud-water contributions to soil moisture). During our 26-month study, only a small percentage of the total water collected from the FO events ended up reaching the soil (1.1 mm, or 4.5%)-suggesting that in the Andean páramo, fog water contribution to the net hydrological input is negligible. Fleischbein et al. [37], working in montane forests in Ecuador, reported also negligible fog water inputs. In contrast, the studies of Clark et al. [38] and Liu et al. [39] for cloud and rain forests respectively, found higher values-the latter showing an average annual fog contribution amounting to about 5% of annual rainfall at a site in southwest China. Soil moisture is measurably higher beneath tree canopies [43], but fog inputs also depend on the density and frequency of fog [25]. Despite the high frequency of fog in the Andean páramo, days with FO and no precipitation are rare [8], and most of the FO events that reached the soil occurred in the early morning, when evaporation is low. In these grasslands, fog water is intercepted by the tussock leaves, diminishing water input to the ground, although the water storage capacity of tussock grasses (approximately 2 mm) [34] is smaller than that of forests, which have more complex canopies [44]. Variations in fog drip are affected by the type, size, location, and density of the foliage [45]. In the páramo, the low intensity of fog is not sufficient to saturate the tussock grasses leaves; causing no drip. On the contrary, in forests, their denser foliage allows capturing more fog water.
The amount of water collected from CW events was higher than that from FO events; however, the EP was also low (Table 1), accounting for only 17% (20.4 mm) of the total depth (122.2 mm). Our findings are in agreement with those of other studies, e.g., Herckes et al. [42]; Gomez Peralta et al. [40], working in cloud forests below 2500 m, where CW input might be insignificant; and Giambelluca et al. [46], working in a dry cloud forest where the majority of intercepted water was re-evaporated from the wet vegetation and never reached the ground.
In páramo, soil is constantly saturated, which means soil water content is normally above 80% [47]. To study the gain in soil moisture from FO and CW we needed events when soil is not saturated and for this reason, we excluded from our analysis events with soil water content ≥ 80%. Cárdenas et al. [17] reported gains in VWC in Colombian páramo soils under field capacity conditions, following large occult precipitation events; however, it should be noted that in this study, the contribution of occult precipitation to soil moisture was not separated from that of rainfall.

The Impact of Cloud-Water Deposition on Effective Precipitation
The high-resolution laser disdrometer, with its ability to measure small drops, proved to be an effective instrument for monitoring the other types of precipitation that accompany fog during CW events. We determined that (as shown in Figure 4) the CW events that resulted in EP consisted mainly of fog combined with drizzle and/or light rain and had a total amount per event ranging from 0.3 to 11.3 mm. In addition, all the events occurred in the early morning; and wind speeds during the events varied between 1.5 and 3.4 m s −1 . Since fog water contribution is also influenced by fog density and frequency as well as wind speed [25], it is plausible that the CW events with low total depth had high fog density and occurred at lower wind speeds.
At higher wind speeds, deposition via impaction increases; for instance, Liu et al. [39] attributed low fog-drip to the low wind speeds (<2 m s −1 ), and Holwerda et al. [48] reported low amounts of CW in a secondary forest and in a mature cloud forest (about 6% and 8%, respectively, of dry season rainfall) due to low wind speeds. As noted earlier, all the events included in our analysis occurred at wind speeds below 4 m s −1 therefore deposition on vegetation by these events was also low despite differences in plant height and complexity of plant leaves between tussock grasses and forests. This hypothesis is supported by previous findings that the most important factors affecting fog deposition rate are fog droplet size distribution [41], the topography of the field site [49], fog duration, surface area, and geometry of the vegetation [18].
The CW events having higher percentages of drizzle and light rain recorded the higher amounts of EP ( Figure 5), suggesting that EP is related to a major presence of drizzle and light rain rather than fog. Since drizzle and light rain droplets are bigger than fog droplets, and their fall velocity is higher, during these larger events it is likely that the tiny droplets of fog were more susceptible to being blown away and dried before deposition on tussock grass leaves. In addition, from field observations we know that fog occurs only intermittently during the day and most of the time is of brief duration, while drizzle occurs frequently enough in the páramo that only 12% of the days are completely dry [31].

Relationship between Effective Precipitation and Meteorological Variables
None of the meteorological variables analyzed (relative humidity, vapor pressure deficit, wind speed, air temperature, dew temperature, and solar radiation) showed any relationship with EP, for either FO or CW events (Figures 6 and 7). All the FO events that resulted in EP occurred at relative humidities over 90%, and all the CW events that resulted in EP occurred at humidities higher than 98%. While fog formation generally takes place under high relative humidity-conditions ranging from undersaturated to slightly supersaturated [50]-relative humidity was not a determinant of EP; FO events occurring at humidities between 80% and 100% resulted in no EP. Similarly, we found no relationship between EP and vapor pressure deficit, dew temperature, air temperature, wind speed or solar radiation, suggesting that meteorological variables are not drivers for EP.
For both FO and CW events, three characteristics did show a direct relationship with EP: duration, total amount per event and intensity (see Appendix A). The duration of FO events varied between 1 h and 8 h (Figure 8), whereas CW events were longer and with intensities > 0.5 mm h −1 , which produced higher amounts of EP. Fog density information was beyond the scope and possibilities of this current study, then we assume that FO events were dense enough to cause fog drip to the ground and resulted in EP despite small total depth.
In the Andean páramo, hydrological regulation capacity might be related to the high frequency of fog and low-intensity rainfall (drizzle). But although it is widely acknowledged that fog plays an important hydrological role in different ecosystems, from our findings we can suggest that fog has an important ecological role in the páramo mainly due to its availability for vegetation-reducing CO 2 uptake, transpiration, and evaporative losses [47]; however, its contribution to streamflow is negligible. From a hydrological point of view, inputs from CW are more important. For this reason, water balance studies should focus on observations of low-intensity rainfall rather than fog.

Conclusions
This study was initiated to assess the importance of fog-only (FO) and cloud-water (CW) events in contributing to soil moisture in the Andean páramo. The two types of events were divided into two classes according to whether or not they produced effective precipitation (EP), i.e., increases in soil water storage: Class I = EP > 0 mm; Class II = EP of 0 mm. Hence, we focused on the analysis of Class I events.
The total depth from all the FO events was 24.4 mm, only 4.5% of which reached the soil, whereas for CW events 16.7% of the total depth (122.2 mm) turned into EP. The water yield for the FO events was less than 5 mm per event, the maximum duration was 8 h, and at least 1 h was needed for water to reach the soil. Although FO events make a negligible contribution to streamwater, their presence contributes to the maintenance of high relative humidity and low net radiation rates translating to decreased evapotranspiration rates. In contrast, CW events are potentially important for both soil moisture and streamflow in the Andean páramo, because of their long duration (averaging over 13 h) and high total depth per-event (over 10 mm)-owing to the large proportion of drizzle and light rain.
We found no relationship between EP and meteorological variables (wind speed, air temperature, dew temperature, vapor pressure deficit, solar radiation, and relative humidity) since low coefficients of determination and correlation were obtained. While the contribution of fog water to soil moisture is related to fog deposition on vegetation, these meteorological variables are not determinant factors in the estimation of EP amounts.
Because our study was based on rainfall data from a high-resolution disdrometer as well as fog data from equipment tested for monitoring fog in the Andean páramo, we were able to carry out a comprehensive study of various types of precipitation. We believe that our findings significantly advance our understanding of the hydrological role of fog and cloud water in these environments. Funding: This study was executed in the framework of the project "A research network for the resilience of headwater systems and water availability for downstream communities across the Americas" and was funded by the Vice-rectorate for Research of University of Cuenca.
Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: All data, models, and code that support the findings of this study are available from the corresponding author upon reasonable request. Figure A1. Heat-map for fog-only events as they relate to effective precipitation, the event characteristics, and the meteorological variables (wind, relative humidity, vapor pressure deficit, temperature, dew temperature, and radiation) analyzed to find an influence on soil moisture gain. Each bar represents a FO event and they are organized by effective precipitation increasing values. Figure A2. Heat-map for cloud-water events as they relate to effective precipitation, the event characteristics, and the meteorological variables (wind, relative humidity, vapor pressure deficit, temperature, dew temperature, and radiation) analyzed to find an influence on soil moisture gain. Each bar represents a CW event and they are organized by effective precipitation increasing values. Figure A2. Heat-map for cloud-water events as they relate to effective precipitation, the event characteristics, and the meteorological variables (wind, relative humidity, vapor pressure deficit, temperature, dew temperature, and radiation) analyzed to find an influence on soil moisture gain. Each bar represents a CW event and they are organized by effective precipitation increasing values.