Introduction

Hydrology in forest ecosystems is made up of complex processes in which biotic and abiotic factors interact. Although the interception of rainfall by vegetation is not the main factor in hydrological estimates, it does allow identifying that changes in tree cover affect the hydrological balance of a tree area, since it alters the soil moisture content and surface runoff (^{Sadeghi et al., 2015}).

The canopy regulates the flow following precipitation before reaching the forest floor, because it modifies the trajectory of the fall of the drops dividing it into: direct rain, interfoliar rain and rain of fustal runoff. In the interception process, it is recognized that forest vegetation participates in the spatial distribution of rainfall. Includes its different strata, and once it reaches the forest floor has patterns of temporality; in such a way that it constitutes a process of great complexity, by the structural elements of the forest and its interaction with the environment (^{Kittredge, 1948}).

Eventhough forests have no major influence on precipitation, their participation can not be considered insignificant, particularly when accounting for the amount and distribution of rainfall on the forest floor (^{Kittredge, 1948}). This is one of the main reasons why several studies of forest ecology have been aimed at generating relationships between the composition and spatial variability of the forest structure with the interception of rainfall (^{Flores et al., 2011}).

Precipitation in forest ecosystems is divided into direct precipitation (PD), fustal runoff (F) and loss of interception (I), classification that is important in studies of forest hydrology (^{Marín et al., 2000}; ^{Lida et al., 2005}).

Net precipitation (PN) is the amount of rain that reaches the forest floor through direct precipitation (PD) and fustal runoff (EF) (^{Manfroi et al. 2004}; ^{Levia and Herwitz 2005}; ^{André et al., 2008}). PD refers to the rain that reaches the forest floor, passing between treetops or dripping. EF corresponds to the fraction of the rain that reaches the forest floor, sliding through the branches and branches of the trees, is caused by the precipitation intercepted by the canopy components (^{Staelens et al., 2008}). Finally, rain interception (I) is the part of the precipitation retained by the canopy of canopies that does not reach the forest floor by evaporation (^{Taghi et al., 2013}). It is calculated indirectly by the difference of the gross precipitation (PB) measured over the canopy or in a nearby area, and the sum of PD and EF (^{Aussenac, 1981}):

According to ^{Sadeghi et al. (2015)}, rainfall interception can account for 10-25 % of total rainfall (PB) in deciduous forests, and up to 40 % in perennial forests. ^{Aussenac (1981)} records a relation of intercept quantities for different species of Abies, Picea, Pinus, Pseudotsuga, Fagus and Quercus. From the above, it is inferred that the density or levels of forest cover will have an impact on the interception process, as they affect the soil moisture content and surface runoff of the land.

The interception of rainfall and its subsequent evaporation affects the water yield of watersheds, their study is complex and can be very variable. On a large scale, weather factors such as wind speed and exposure, rainfall intensity, and fog incidence determine the loss of interception. At a small scale, an important control of the interception loss is the density and composition of the vegetation that define the capacity of the crown to temporarily store water. Other factors are also involved, such as crown exposure, canopy roughness, and rainwater penetration (^{Fleischbein, 2005}; ^{Siegert, 2014}).

The various components of the rainfall interception process have been measured and modeled for many types of vegetation, with particular emphasis on forest stands. However, it is important to recognize the importance of rainfall interception models to predict the effects of changes in plant cover on water resources (Muzylo, 2009), in addition to allow the extrapolation of results of measurements, both in space and time, providing information that refers to the mechanisms involved in the interception process by the tree vegetation and of the attributes that control such processes (^{David et al., 2005}; ^{Návar et al., 2008}).

In this context, it is proposed to determine the amount of rainwater retained by the canopy through the rain interception process, and to generate the respective models in three types of high mountain forest in the Texcoco river basin, Mexico.

Materials and Methods

In 2009, a research line was initiated in the Campo Experimental Valle de México de INIFAP (Valle de México Experimental Station from INIFAP) to study the interception of rainwater by arboreal vegetation in the area of forest hydrology in the Texcoco river basin. Coatepec, Santa Mónica, San Bernardino, Chapingo, Coxcacoaco, Xalapango and Papalotla rivers are part of the Texcoco hydrological basin in the State of Mexico (Semarnat, 2015).

Three types of high mountain forest were selected, which were evaluated in three different years: Abies religiosa (Kunth) Schltdl. et Cham. (2009), located at 3 000 masl; Quercus spp. (2010) at 2 900 masl., and Pinus hartwegii Lindl. (2011) at 3 650 masl (Figure 1). The vegetation in each type of forest, arboreal stratum and undergrowth, was identified by botanical collections. The design of the modules consisted of 1.0 ha (100 x 100 m) plots where only dominant tree species were considered. The variables included were: diameter at breast height measured with diametric tape (Ben Meadows), total height and clean stem height measured with Suunto clinometer; location, within each plot, of trees and direct reading pluviometers (Tru-Chek^®), using Cartesian coordinates. Additionally, the unevenness of the terrain was measured with a 20 m hose level to generate its physiography.

Figure 1 Location of the research modules within the San Pablo Ixayoc forest ejido, Texcoco municipality, State of Mexico. 

The amount of total annual precipitation and its distribution for each forest condition, was determined from measurements of daily precipitation under open conditions. For the measurement of the amount of intercepted rainfall per module in each module, 75 Tru-Check^® (direct reading) rain gauges under the canopy, with a precision of 0.1 mm, were placed at random at 0.9 m above the floor level to avoid the splashed effect (^{Prasad, 2016}).

The procedure to express the amount of rainfall intercepted was by the difference between the total precipitation measured under open conditions and the overall average of the water volume of the collectors under the vegetation canopy, for each rainfall event (^{Xiao et al., 2000 Sadeghi et al., 2015}).

The study began in 2009, in the fir plot, continued in 2010, with oak and concluded in 201 with pine. This allowed to generate values of interception by species, precipitation event and per year. The fustal runoff was not quantified.

Based on the dispersion of points, the ratio of total precipitation against the percentage of interception was analyzed. In order to statistically correlate the precipitation and intercept values for each module, logarithmic and exponential non-linear models were fitted by regression analysis (^{Hosseini et al., 2012}; ^{Sadeghi et al., 2015}).

The selection criteria of the best adjusted models included the coefficient of determination and significance of the regression parameters. Statistical analysis was performed in the SAS program (^{SAS, 2007}).

Results

Annual Average Precipitation

The general characteristics regarding the location of the three study sites are summarized in Table 1. In terms of altitude, the pine forest is the one located in the highest part of the basin, followed by the fir forest and in the lower part, the oak forest.

Table 1 Location of research modules and year of establishment. 

Table 2 shows some of the population and dasometric characteristics of the dominant vegetation in each research module.

Table 2 Dominant vegetation of each research module. 

¹ = Taxonomic identification performed by Dr. Enrique Guizar Nolazco from the. Herbario of the División de Ciencias Forestales of the Universidad Autónoma Chapingo.

For Pinus, Abies and Quercus, 72, 74 and 63 precipitation events were measured, respectively. Results indicate that for the Pinus hartwegii forest, a cumulative annual rainfall of 680.20 mm, in the Abies religiosa forest, 707.70 mm and for the Quercus spp. Forest, 503.00 mm (Table 3). The intervals of variation per event were 0.2-23.5 mm for Pinus, 0.2-33 mm for Abies and 0.2-38 mm for Quercus. Pinus and Abies recorded atypical data, the first of 82 mm and the second of 62 mm.

Table 3 Results of the interception measurements (%) in the three research modules. 

Periods for data taking: ¹ = May 22^nd to September 30^th; ² = May 27^th to October 30^th; ³ = June 24^th to September 30^th.

Generation of Interception Models

According to the point dispersal from the relation of total precipitation against interception percentage, and in order to statistically, correlate them, a non-linear regression was made using the PROC GLM of SAS procedure. Exponential and logarithmic models were tested (^{Scatena, 1990}; ^{Hosseini et al., 2012}; ^{Sadeghi et al., 2015}) (Table 4):

y = a LnX + b

y = a X-b 

Where:

• y = Rain interception under the canopy in percentage

• X = Total precipitation (under open sky), measured in mm

• Ln = Natural logarithm

• a, b = Coefficients of regression

Table 4 Logarithmic and exponential type models. 

The behavior of the three generated models is graphially shown in Figure 2.

Negative exponential for Abies religiosa (Kunth) Schltdl. et Cham.; negative logarithmic for Pinus hartwegii Lindl. and Quercus spp., from the interception percentage ratio as a function of total precipitation.

Figure 2 Graphic behavior of the regression models from species under study. 

The graphic shows the behavior of rainfall interception which reflects a part of the condition of the forest in terms of the mensuration characteristics of the species under study. Abies tends to form more dense masses than Pinus and Quercus and from their habit of keeping a great numberof leaves along the stem, a a result, the amounts of rain water retention are greater than those from pine and oaks. The former explains the reason why interception in this species is higher, even in low values of total precipitation.

Even though the behavior of interception is similar in the oak and pine forests, it is evident that the interception curve in the pine forest is greater, which can be attributed to a larger density of the basal area, as well as height and average diameter. This statement is reinforced with the oberved behavior of the Abies forest in regard to the mensuration variables (Table 2).

When analyzing the rainfall interception percentages en the three cases, it is worth noting that they are consistent with the value obtained in different temperate forests (^{Aussenac, 1981}). Interception values of 21.5 % have been documented for Pinus halepensis Miller and of 20.4 and 24.9 % for Quercus sessiliflora Salisb. (^{Aussenac, 1981}), which are consistent with those recorded in the actual study, except for Abies. ^{Cantú and González (2001)} quote 13.5 % for Quercus in forests of Linares, Nuevo León State, Mexico, which differ from those indicated in this study, which is understood from the differnt climatic conditions.

According to more recent research studies, the results are consistent in terms of the behavior of the interception models (^{Hosseini et al., 2012}; ^{Yáñez-Díaz et al. 2014}; Sadegi et al., 2015). It is important to highlight that the percentages here obtained suggest that other variables (^{Torres-Rojo, 2014}), such as density, crown architecture and leaf morphology, among other things that were not analyzed, also participate in the rain interception process, and thus, need to be assessed.

Conclusions

Three models were obtained that adequately describe the values and behavior of rain interception for each dominant species considered. The evaluated characteristics allow, through the proposed models, the prognosis of the effects of the vegetal cover on the input of rainwater in the study area.

The three modules show different characteristics, in terms of species, altitude, exposure and micro-environmental conditions. Based on the differences in interception models, it is possible that other climatic, site or vegetation variables (allometric) need to be incorporated into the modeling of the interception process of rainfall by temperate forests.

Conflict of interests

The authors declare no conflict of interests

Contributions by author

Eulogio Flores Ayala: design and planning of the research, supervision of field work, data taking at the field, analysis of the application of results and writing of the document; Vidal Guerra De la Cruz: research support, field data analysis, structuring, presentation and writing of results in the document; Gerardo H. Terrazas González: statistical analysis and fitting of the generated models; Fernando Carrillo Anzures: field data taking, assessment of the aplicability of the information taken in the field and forestry interpretation; Fabián Islas Gutiérrez: assessment of the aplicability of the information taken in the field and review of the fit of the equations; Miguel Acosta Mireles: data taking at the field, research support, field data analysis, interpretation of the fit to the generated models, structuring, presentation and writing of results in the document; Enrique Buendía Rodríguez: field data review and depuration and presentation of results in the document.