Partial exclusion of precipitation: throughfall, stemflow and canopy interception in Eucalyptus plantations in southern Brazil

Hydrological behavior in reforested watersheds is different from that under other forms of cover. The variation may be related to aspects intrinsic to species, planting density, physiological maturity, management system and climatic conditions. Periodically, climatic anomalies such as the case of La Ninã are observed, and these are responsible for the alteration of the rainfall regime and consequently generate water deficits in the southern region of Brazil. Water deficit is responsible for reducing growth and productivity for the Eucalyptus genus, in addition to causing changes in hydrological behavior in reforested watersheds. Accordingly, this study compared the partition of rainfall in throughfall, stemflow and canopy interception of eucalyptus trees submitted or not to partial exclusion of precipitation. In the open field, 3 rainfall collectors were installed, and in the stand, for each rain exclusion treatment, 9 throughfall collectors and 9 stemflow collectors were installed. Every two weeks for 12 months, the volume of the collectors was measured. The quantified precipitation was 1627 mm over a year. In the treatment without exclusion, 84.8, 2.9 and 12.3% referred to throughfall, stemflow and canopy interception, respectively, while in the treatment excluding rainfall 80.6, 2.3 and 17.2% referred to throughfall, stemflow and canopy interception. The regression adjustments for throughfall and stemflow showed satisfactory R2 coefficients.


INTRODUCTION
The forestry sector holds a prominent position in the Brazilian economy. Planted forests occupy 7.83 million hectares of the country, and the genus Eucalyptus represents 72% of this area. Even with the large area occupied, industrial forest plantations cover about 0.92% of the Brazilian territory. Paper and cellulose commodities are the second-most exported products of Brazil (IBÁ, 2019).
Extensive areas occupied by forest plantations are able to modify the hydrological balance (Ferraz et al., 2019). In addition to the interception of rainfall through the canopy of trees, the rapid growth of the Eucalyptus genus increases water consumption, losing it to the atmosphere through transpiration (Rodríguez Suarez et al., 2014;Jackson et al., 2005). Comparing two paired watersheds, Reichert et al. (2017) concluded that the interception in a Eucalyptus plantation is greater when compared to an area of pasture or natural grassland.
The partition of the rainfall in the area covered by vegetation is very dynamic. During precipitation, a portion of the water is intercepted by the canopy and immediately evaporated to the atmosphere (Llorens and Domingo, 2007). Part of the precipitation crosses the canopy and drips into the stand, being called "throughfall" (Navar, 2011). A portion flows from the leaves to the branches and trunk reaching the base of the tree, being called stemflow (Zhang et al., 2016;Johnson and Lehmann, 2006). Knowledge of the partition of precipitation is important in studies of modeling the water balance of a watershed (Chaffe et al., 2010), however, the percentage of interception by canopy varies according to characteristics inherent to the species and the site and, in the absence of this information, the interception is often neglected in hydrological-modeling studies (Savenije, 2004).
Considering the species in the present study, Eucalyptus urophylla, Arcova et al. (2018) reported 89.0%, 5.4% and 5.6% for throughfall, stemflow and canopy interception for a stand at 30-year-old, respectively. Souza et al. (2019) for the same species but at 1-year-old found 95.3; 1.3 and 4.3% for treatment with less fertilizer applied and 91.7; 3.2 and 6.2% for treatment with greater amount of fertilizers applied.
The importance of the partition of precipitation is discussed in several studies (Staelens et al., 2006;Fan et al., 2015); however, the difficulty in assessing the stemflow has caused great variations and still needs studies. Although it represents less relative contribution, the stemflow is responsible for supplying water and nutrients directly to the roots of the trees (Momolli et al., 2019a;Zhang et al., 2013;Bouillet et al., 2002;Levia et al., 2010). The precipitation event carries particles suspended in the atmosphere to the ground (Navar et al., 2009). These particles, added to the interaction with the canopy (throughfall and stemflow), leach the different tissues of the plant, increasing the nutritional contribution (Bhat et al., 2011;Schrumpf et al., 2006). In addition, in areas of low natural fertility, nutrient input through precipitation may be the only source of nutrients (Dawoe et al., 2018;Lu et al., 2017). Canopy interception is determined by biotic factors such as population age, morphophysiological aspects, height, contact surface of different plant species (Keim et al., 2006) and by abiotic factors such as intensity and amount of precipitation, drop size, temperature and wind speed (Sulínski et al., 2001;Momolli et al., 2019b;Gerrits and Savenije, 2011;Bulcock and Jewitt, 2012). For Savenije (2004), in warm regions the percentage of water intercepted and returned to the atmosphere is more significant than in other climates. For the authors Crockford and Richardson (2000), high intercept values are the result of long rain events with low intensity, while small intercept values are the result of shorter and more intense rainfall indexes. In addition, the higher temperature favors evaporation and, consequently, the higher interception rates. Moderate wind contributes to the removal of water from the canopy and evaporation to the atmosphere; however, high winds provide the waterfall which initially was in the treetops. Figure 1 shows the scheme of the dynamic partition of rainfall in forested areas. The water regime in southern Brazil is determined by the atmospheric-oceanic phenomenon. When the cooling of ocean waters occurs in the equatorial Pacific, there is the phenomenon of "La Niña" that is responsible for the significant decrease in rainfall in this region. As a consequence, there is a loss of productivity of several crops, including Eucalyptus plantations. The reduction in precipitation can cause even greater impacts on the hydrological cycle of the watersheds considering that a greater part of the precipitation will be intercepted by the canopy without reaching the soil surface. Therefore, the present study evaluated the partition of precipitation in Eucalyptus stands with and without partial exclusion of throughfall.

Characterization of the experimental area
The experimental area is located in a Cfb climate (temperate climate), with an average annual temperature of 18.8ºC and an average annual precipitation of 1,646 mm, according to the Köppen classification (Alvares et al., 2014). According to Flores et al. (2016) the species E. urophylla is classified as having a low climatic suitability for the study region. Figure 2 shows meteorological data for the period from July 2017 to June 2018, obtained from the meteorological station located at the company Klabin SA in Telêmaco Borba -PR -Brazil at 880 m altitude, 24°12'40.6" S and 50°33'29.2" W. The distance between the experimental area and the weather station is approximately 3.4 km in a straight line. In March 2017, dendrometric characterization was performed, measuring the height and diameter at breast height (DBH) of all trees. According to Table 1, the mean DBH and the total height in the Excluding (E) treatment were 16.80 cm and 28.60 m, respectively. For the Without exclusion (WE) treatment, the mean DBH was 17.30 cm and the total height 28.80 m. The volume per hectare in the E and WE treatment was 346.17 and 365.07 m³ ha -1 , respectively.

Experimental design
The study belongs to the TECHS project (Tolerance of Eucalyptus Clones to Hydric, Thermal and Biotic Stresses). The experiment was carried out in a completely randomized design, with 720 m² plots of eight lines with ten plants each in spacing of 3 m x 3 m (1,111 trees ha -1 ). For the hybrid E. urophylla x E. sp. two treatments of the water regime were defined: 5 Partial exclusion of precipitation: throughfall, stemflow … Rev. Ambient. Água vol. 16 n. 5, e2645 -Taubaté 2021 one receiving 100% of the throughfall (WE) and the other receiving only 70% of the throughfall (E). For the treatment (E), that received 70% of the precipitation, a system of partial exclusion of the throughfall was used with plastic gutters that prevent the precipitation reaching the ground. A schematic representation of the precipitation exclusion treatment (E) can be seen in Figure 3. This technique is based on the coverage between the planting lines covering 216 m² of the area of each treatment, being the equivalent to 30% of the plot area (Binkley et al., 2017). The monitoring of the partition of precipitation under the two rainfall regimes occurred from July 2017 to June 2018. The sampling of this experiment was started when the stand was 66 months old. For the (E) treatment, a partial exclusion system of throughfall was used, installed when the stand completed one year of age.

Measurement of rainfall, throughfall and stemflow
To quantify the rainfall (R), three collectors with a 20 cm catchment diameter were installed in an area adjacent to the eucalyptus stand at a height of 1.5 m above ground level. Straps with steel wires were placed in order to prevent birds from using the collectors as perches.
For the evaluation of throughfall (Tf), nine collectors per treatment were installed, with a collection diameter of 20 cm and height of 1 m from the ground level, systematically distributed along the line, between the lines and diagonally between four trees. Collections were started at 66 months of age.
The stemflow quantification (Sf) occurred with the installation of nine sets formed by a plastic hose with a diameter of one inch and a reservoir for water storage. The hose was cut longitudinally and then it was installed in a spiral shape on the tree trunk. This configuration allowed the water to drain through the trunk and be stored in the reservoir. Fortnightly, the values of rainfall (R), throughfall (Tf) and stemflow (Sf) were measured. To obtain the values of precipitation and throughfall in millimeters, the following Equation 1 was used: Where: The rainfall and throughfall collectors were composed of plastic bottles with a capacity of 2 liters, and the collections were carried out every 15 days. The throughfall collectors were arranged on the line, between the lines and diagonally between four trees.

Statistics and Data Analysis
Descriptive statistics are presented as the mean, percentage and coefficient of variation (%) for the quantities in each rainfall regime and each partition of precipitation. Statistical analysis was performed using IBM SPSS 20.0 (IBM, 2011). Regression equations were adjusted for throughfall, stemflow and canopy interception as a function of the incident precipitation variable.
The distribution of the regression residues was then analyzed in order to validate the homogeneity of variance. The residues were presented in graphic form as a function of the variable analyzed.

Rainfall partitioning
According to the results obtained, the total rainfall (R), during the study period, was  Where: ± variation coefficients (%); R = rainfall, Sf WE = throughfall in the treatment without exclusion; Sf E = throughfall in the treatment with partial exclusion; Tf WE = stemflow in the treatment without exclusion; Tf E = stemflow in the treatment with partial exclusion; I WE = canopy interception in treatment without exclusion; I E = canopy interception in treatment with partial exclusion.
Regarding throughfall, the WE treatment showed 1379 mm, equivalent to 84.8% of the precipitation. In contrast, the E treatment showed a lower value (1311 mm), corresponding to 80.6% of the precipitation. The volume collected for the Sf WE treatment was 48.40 mm, representing 2.9% of the rainfall. For the Sf E treatment, 37 mm was accumulated, corresponding to 2.3% of the rainfall. In the canopy interception, the lowest observed value was 200 mm (12.3%) for WE treatment, while in treatment E the value was 280 mm (17.2%).
The regression analyses for throughfall, stemflow and canopy interception as a function of precipitation in the WE treatment showed an adjustment R = 0.97; 0.73 and 0.34, respectively (Figure 4). For E treatment, the adjustment was R = 0.96; 0.67 and 0.59, for throughfall, stemflow and canopy interception, respectively.
Evaluating forest formations of Eucalyptus cloeziana, Pinus sp. and Seasonal Semideciduous Forest in Iperó, SP, Gasparoto et al. (2014) found that throughfall represents 76.2%, 85.1% and 84% of precipitation, respectively. Almeida et al. (2013), evaluating an E. grandis x E. urophylla stand in Minas Gerais, found that the canopy interception was 9%, this value being lower than that found in this work. Momolli et al. (2019b) studied the distribution of rainfall in an E. dunnii Maiden stand, aged 72 months in Alegrete, RS, and found an average canopy interception of 8.9% in one year of evaluation (1903 mm). Supangat et al. (2012) evaluating the behavior of rainwater after interaction with the canopy of six-year-old Eucalyptus pellita found an interception of 15.4%.  Corrêa et al. (2019) evaluating a stand of Eucalyptus dunnii in Alegrete, RS at 16.5 months of age observed that the precipitation was 1,586 mm, with 7% being intercepted by the canopy of the stand. The throughfall corresponded to 98% of the effective precipitation, the remaining 2% referring to stemflow. For the same species, Dick et al. (2018) studied a four-year-old stand and for the first year of assessment found that the throughfall was 91.08% (1,033.29 mm) and the stemflow 0.91% (10.40 mm). In the second year, the throughfall corresponded to 91.48% (1,497.46 mm) and the stemflow 1.28% (20.98 mm). The authors associated these results with the leaf area index of the stand, since the canopy cover allowed greater passage of rainfall.
For the same species, Dick et al. (2018) studying a four-year-old stand, the throughfall was 91.08% (1033 mm) and the stemflow was 0.91% (10.4 mm). In the second year, the same authors found that the throughfall corresponded to 91.48% (1497 mm) and the stemflow of 1.28% (21 mm). The authors associated these results with the leaf area index of the stand. The lower density of the canopy allowed the precipitation to reach the forest floor.
Other works developed with the genus Eucalyptus sp. has shown high correlations between variables. In a stand of Eucalyptus dunni aged between 1.4 and 2.4 years, Corrêa et al. (2019) found correlations between rainfall and throughfall, stemflow and canopy interception in the order of 0.99, 0.83 and 0.53%, respectively.
In a stand of Eucalyptus grandis, Balieiro et al. (2007) found a correlation of R = 0.99 for throughfall and R = 0.93 for stemflow. Supangat et al. (2012) found E. pellita coefficients of determination (R²) of 0.99; 0.77 and 0.29 for throughfall, stemflow and canopy interception, respectively. Gasparoto et al. (2014) evaluated Eucalyptus cloeziana and found a linear adjustment of R = 0.90 in the relationship between rainfall and throughfall. Studying the partition of rainfall in a stand of E. dunni at 6 years of age, Momolli (2019b) found adjustments of 0.99; 0.90 and 0.60 for throughfall, stemflow and canopy interception. Many factors are involved in the variability of the data, including the spacing of the trees, the degree of canopy closure and the spatial distribution of the collectors as possible factors (Xiao and McPherson 2011).
According to Momolli (2019b) the volume of precipitation is an important factor of variability. In the precipitation, the variation coefficient was 16% and 5% in the precipitation volumes below 20 mm and above 180 mm. For stemflow, the increase in the volume of precipitation from 0.5 mm to 1.7 mm resulted in a decrease from 60% to 20% in the variation coefficient. The author concludes that the increase in precipitation volumes decreases CV%.

CONCLUSION
The linear regression of the throughfall and stemflow as a function of the incident precipitation, presented adjustments of 97% and 72% without exclusion treatment and 97% and 55% in exclusion treatment. In addition, the analysis of the graphical distribution of the residues showed homogeneous behavior around the mean zero and without bias.
The regression model of throughfall and stemflow as a function of precipitation presented satisfactory adjustments. The higher leaf area index in the treatment excluding partial precipitation explains the greater canopy interception.
The treatment in which there is a 30% reduction in precipitation caused an increase from 12.3% to 17.2% in the interception of trees. This is attributed to the water-holding capacity of the canopy. This information is important because it can be used in projections of hydrological models during periods of the La Ninã phenomenon or even due to climate changes.

ACKNOWLEDGEMENTS
This study was financed in part by the Coordenação de Aperfeiçoamento de Pessoal de Nível Superior -Brasil (CAPES) -Finance Code 001.