**3. Results**

#### *3.1. Variation in Soil Moisture and Fertility*

Figure 2 shows the changes of soil moisture along the elevation gradient. The soil moisture was higher on the shade slopes than on the sunny slopes at all elevations in the dry season (Figure 2a), and it was much higher in the rainy season than at the end of the dry season at a given elevation (Figure 2b). The soil moisture increased slightly, but not significantly, with increasing elevation, except the shade slope at 1940 m asl (*p* < 0.05) and, at elevations ( ≤1520 m), with an average of 0.055, did not differ significantly among elevations.

**Figure 2.** Changes in soil moisture (**a**) along the elevation gradient for shade and sunny sites and (**b**) between seasons at each elevation. Values are mean ± SD. Note that the values on the y-axis differ greatly between the two graphs. Values labeled with different lowercase letters differ significantly (*LSD*, *p* < 0.05) between elevations for a given site category (sunny and rainy); values labeled with the same capital letters did not differ significantly between elevations for a given site category (shade and dry).

There was no significant difference in soil fertility (based on the holistic index of soil fertility) among the elevations (Figure 3). The soil fertility reached its maximum value (49.07) at the lowest elevation (1380 m) on the sunny slope, and its minimum value (19.57) at 1520 m on the shade slope.

**Figure 3.** Changes in soil fertility (based on the holistic index of soil fertility; Equation (1)) with increasing elevation for the aspects. Values are mean ± SD. Values labeled with same lowercase letters differ not significantly (*LSD*, *p* >0.05) between elevations for the shade slope; values labeled with the same capital letters did not differ significantly between elevations on the sunny slope.

#### *3.2. Changes in Vegetation Biomass as a Function of Elevation and Aspect*

Table 1 summarizes the changes in the characteristics of the vegetation as a function of elevation. For *D. viscosa*, the density and crown width increased with increasing elevation, reaching a significantly higher maximum at 1520 m, then decreased thereafter. Diameter at breast height did not differ significantly among elevations, except for a significant decrease at 1940 m. Height was significantly higher at the two lowest elevations than at the highest elevations. All *D. viscosa* parameters presented their lowest values at 1940 m. For *P. yunnanensis*, density increased with increasing elevation, and the differences were generally significant. Diameter at breast height, crown width, and height showed a similar pattern of increase, but with a maximum at 1640 m. Vegetation cover showed an inconsistent pattern, but reached a significantly higher maximum value at 1520 m. The density of the *P. yunnanensis* increased from 0.1 to 13.0 per 100 m2, and the average diameter, crown width, and height at 1940 increased to approximately 4 to 10 times the values at 1380 m.



a Cumulative diameter of all stems for *D. viscosa*. Values labeled with same lowercase letters differ not significantly (*LSD*, *p* > 0.05) between elevations for *D. viscosa*; values labeled with the same capital letters did not differ significantly between elevations for *P. yunnanensis*.

Figure 4 presents the aboveground biomass of the two species. The biomass of *D. viscosa* first increased, and then decreased with increasing elevation on the shade slopes, with a maximum of 0.74 kg/m<sup>2</sup> at 1440 m; by contrast, it decreased with increasing elevation on the sunny slopes, with a maximum value of 0.75 kg/m<sup>2</sup> at 1380 m (Figure 4a). The biomass of *P. yunnanensis* biomass first increased and then decreased with increasing elevation on the shade slope, with a maximum value

of 5.87 kg/m<sup>2</sup> at 1520 m, but increased with increasing elevation on the sunny slopes after it first appeared at an elevation of 1520 m (Figure 4b).

**Figure 4.** Changes in the aboveground biomass of (**a**) *D. viscosa* and (**b**) *P. yunnanensis* as a function of elevation. Values are mean ± SD. Note that the values on the y-axis differ greatly between the two graphs. Values labeled with different lowercase letters differ significantly (*LSD*, *p* < 0.05) between elevations for a given site category (shade slope); values labeled with same capital letters did not differ significantly between elevations for the sunny slope).

#### *3.3. Relationship between Vegetation and Environment*

We used Pearson's correlation coefficient (*r*) to reveal the significant (*p* < 0.05) relationships between the vegetation and environmental factors (Supplemental Table S2). For *D. viscosa*, biomass was positively correlated with branches, density, diameter at breast height, height, and hydrolyzable nitrogen, but negatively correlated with elevation and distance of the quadrat from the river. Branches were positively correlated with density. Diameter at breast height was positively correlated with crown width and height, but negatively correlated with elevation, distance of quadrat from the river, and soil moisture. Crown width was positively correlated with height. Height was negatively correlated with elevation, distance of quadrat from the river, and soil moisture. For *P. yunnanensis*, biomass was positively correlated with density, diameter at breast height, crown width, height, elevation, and distance of the quadrat from the river. Density was positively correlated with height, elevation, distance of the quadrat from the river, soil moisture, and organic matter. Diameter at breast height was positively correlated with crown width, height, and aspects. Crown width was positively correlated with height. Height was positively correlated with elevation, distance of the quadrat from the river, and soil moisture. Vegetation cover was positively correlated with the topographic wetness index.

Among the significant topographic factors, elevation was positively correlated with the distance of the quadrat from the river and soil moisture; and distance of the quadrat from the river was positively correlated with soil moisture and organic matter. Among the significant soil factors, total nitrogen was positively correlated with available nitrogen and soil fertility, and the available nitrogen and phosphorus were both positively correlated with soil fertility.

In addition, we used redundancy analysis to clarify the relationships among the soil properties and topographic parameters for the two species. Figures 5 and 6 present the results for *D. viscosa* and *P. yunnanensis*, respectively. When the vegetation parameters were used as response variables, and the soil properties and topographic parameters were used as independent variables, RDA analysis (Figure 5) showed that the constrained variables explained 64% of the total variance for *D. viscosa* along all RDA axes. The distance from quadrats to the river, soil moisture, and elevation were positively correlated and contributed strongly to RDA axis 1; soil moisture was mainly affected by elevation and distance from quadrats to the river, which agrees with the correlation results (Table S2). The biomass, height, crown width, and diameter at breast height were negatively correlated with soil moisture, whereas the slope and total nitrogen were positively correlated with *D. viscosa* biomass, and the topographic wetness index and total potassium significantly affected the density and vegetation cover of *D. viscosa*. We divided the environmental variables into two independent variables: soil

properties and topographic conditions. The soil properties explained 30% of the total variation of *D. viscosa*, and topographic conditions explained 51%.

The RDA analysis for *P. yunnanensis* (Figure 6) showed that the constrained variables explained 69% of the total variance along all RDA axes. The distance from quadrats to the river, soil moisture, and elevation were positively correlated and contributed strongly to the first RDA axis. The vegetation parameters were positively correlated with soil moisture. In addition, the slope aspect and available phosphorus were positively correlated with the biomass of *P. yunnanensis*, while the total nitrogen was negatively correlated with the vegetation characteristics. The RDA analysis showed that soil properties explained 31% of the variation and topographic variables explained 53%.

**Figure 5.** Results of the redundancy analysis for the relationships between the vegetation parameters for *D. viscosa*, soil properties, and topographic conditions. Variable names: AP, available phosphorus; Bio, aboveground biomass; Bra, branches; Cov, vegetation cover; Cro, crown width; DBH, diameter at breast height; Den, plant density; DQR, distance from quadrats to the river; Ele, elevation; Hei, height; HN, hydrolyzable nitrogen; ASP, aspect; OM, soil organic matter; Slo, slope; SM, soil moisture; TK, total potassium; TN, total nitrogen; TWI, topographic wetness index.

**Figure 6.** Results of the redundancy analysis for the relationships between the vegetation parameters for *P. yunnanensis* and the soil properties and topographic conditions (the abbreviations in Figure 6 was the same as them in Figure 5).
