1. Introduction
Surface runoff is a multi-factor combined process that happens in a complex underlying surface (Xu et al., 2019; Maria et al., 2018; Ochoa et al., 2016). Vegetation and topography are the basic elements that make up the underlying surface, as well as the main factors that affect surface runoff (El et al., 2013). Studies have demonstrated that the effects of vegetation and topography on surface runoff are manifested by jointly changing the upslope inflow of a certain slope unit, and these two factors are usually inseparable (Qin et al., 2015). Specifically, aboveground traits of vegetation constitute the main drivers for generating hydraulic roughness (Kervroëdan et al., 2019), while the topography is the carrier for the occurrence and development of surface runoff generation (Sabzevari and Talebi, 2019).
Regarding on vegetation, topography, and surface runoff, most studies tend to analyze the impact and contribution of a single factor to the surface runoff process (Li et al., 2008; Bal et al., 2011; Bennett and Bridge, 2010). Related studies have shown that the surface runoff coefficient has a negative correlation as well as logarithmic function relationship with surface vegetation coverage, has a negative correlation as well as power function relationship with the thickness of the litter layer, and has a positive correlation as well as exponential function relationship with slope (Zhang et al., 2006; He et al., 2014; Duan et al., 2016). Some scholars have also paid attention to the coupling effects of vegetation and topography on surface runoff, but theoretical research is relatively few, mostly involved in other research topics. For example, Zhang (2018) studied the variation process of water erosion dynamics under different vegetation spatial configuration through indoor runoff scouring experiments and proposed that when grass strip was planted at about 80% of the slope length, it could better exert soil and water conservation function. Cao et al. (2017) analyzed the response of erosion and sediment yield characteristics to the interaction of slope and vegetation cover based on the field simulation test and proposed that the effect of vegetation coverage on surface runoff gradually weakened with the increase of slope, while the effect of slope on surface runoff gradually increased, and finally became the dominant factor affecting soil erosion. Ren et al. (2018) explored the response of water erosion dynamics to vegetation cover and slope allocation based on the WEPP (Water Erosion Prediction Project) model and proposed that the increase of vegetation canopy cover could significantly reduce soil loss and sediment yield, and the least soil loss occurred when vegetation was distributed in the lower slope conditions.
The main purpose of the above research was to determine the optimal configuration of vegetation under typical topography conditions. However, these studies had certain limitations in revealing the coupling effects of vegetation and topography on surface runoff. On the one hand, most studies had adopted simulation methods and defaulted the slope condition to a homogeneous slope, artificially setting or simplifying the combination of vegetation and topography features, ignoring the fact that the formation of the overlapping distribution of vegetation and topography was the result of their long-term interaction and co-evolution (Saco and Mariano, 2013; Kim and Kupfer, 2016). On the other hand, when it came to the coupling effects of vegetation and topography on surface runoff, most of the research objects focused on herbs or shrubs, rarely involved the spatial distribution of tree species.
When studying the coupling effect of complex topography and the spatial distribution of tree species on surface runoff generation, the causal relationship between multiple factors is involved. Structural equation modeling (SEM) is a statistical method based on confirmatory factor analysis and path analysis to reveal the structural theory of a certain phenomenon, which has advantages in simulating and verifying the complex relationship between multiple factors and has been widely used in the environmental and ecological research. For example, Xi et al. (2018) built an SEM of three potential variables including terrain, stand structure, and soil characteristics to study the multi-factor coupling relationships between typical Robinia pseudoacacia L. and Pinus tabulaeformis Carr. mixed plantations. Hou et al. (2020) used SEM to reveal the relationships between vegetation coverage and the reduction rate of the runoff coefficient and the reduction rate of the sediment yield. Yang et al. (2018) used SEM to test the impacts of abiotic and biotic driving factors on plant biomass and root/shoot ratios.
In this study, cypress forests on a steep slope in southwestern China was taken as the research object, the SEM was used to simulate the causal relationship between the spatial distribution of cypress, topography, and surface runoff, and the response surface method (RSM) was used to further analyze the response of surface runoff to the spatial distribution of cypress and topography. Finally, the strategies of stand structure adjustment were proposed to reduce surface runoff, which would provide theoretical and technical support for the improvement of water conservation capacity of cypress forests on steep slopes in southwestern China.
2. Materials and methods
2.1. Site description
The study was carried out on a steep slope (slope angle>30°) in Huaying County (30°25′21″N, 106°50′2″E), Sichuan Province, with an annual average precipitation of 1200 mm and an annual average temperature of 18 degrees Celsius. Most of the rainfall occurred between May and August, accounting for 70% of the annual precipitation. The slope was a bedding slope and the soil was limestone yellow soil. The vegetation conditions on the slope were composed of cypress and sparse weeds. The cypress on the slope originated from the Grain-for-Green Project at the beginning of the 21st century. Aerial-seeding afforestation was carried out on the degraded slope. After two decades of succession of vegetation communities, the slope had developed into an open-canopied cypress forest, with significant differences in stand density and distribution patterns.
2.2. Experiment design
2.2.1. Runoff plot setting
Twelve runoff plots (5 m×10 m) were built at the same slope position, and the relative height difference of each runoff plots was similar so that the influence of topography on surface runoff only depended on the internal topographic characteristics. The basic characteristics including the stand density of cypress, surface vegetation coverage, and the relative height difference of each runoff plot were shown in Tab.1. The statistical results showed that the standard deviations of the surface vegetation coverage and the relative height difference of the 12 runoff plots were 2.25% and 0.13m, respectively, which were not of the same order of magnitude as the sample data. Therefore, the interference of surface vegetation coverage and the relative height difference to this study could be eliminated.
2.2.2. Data collection
In each runoff plot, observation data including topography, the spatial distribution of cypress, rainfall, and surface runoff was collected. Specifically, Real-time kinematic GPS (RTK-GPS) was used to measure topography, while the spatial location of cypress was also determined. In the process of topographic measurement, spatial point data was measured at 0.2m intervals, and when encountering areas with large terrain variability, an intensive measurement was performed at 0.1m intervals. After each rainfall, the surface runoff was measured by the water-level gauge in the runoff storage pond, and the rainfall data monitored by the small weather station was recorded. No less than 20 rainfall and runoff data were observed during the rainy season.
2.3. Data processing
2.3.1. Determination and calculation of factors that characterize the spatial distribution of cypress
The Ripley’s K index (Ripley, 1977), the contagion index (Pommerening, 2002; Aguirre et al., 2003), and the stand density of cypress were used to reflect the spatial distribution of cypress in each runoff plot. In this study, Ripley’s K index described the number of individual plants in a circle with a point as the center and r was the radius, which was typically used to compare a given point distribution with a random distribution: