Study site and selection of rock outcrops
The study was conducted in a typical semi-humid karst grassland in Shilin County (24°38′ to 24°58′N, 103°11′ to 103°29′E; 1,776 to 1,789 m a. s. l.), Yunnan Province, Southwest China (Fig. 1a, b). The area climate is subtropical plateau monsoon with a mean annual temperature of 16.2°C, which fluctuates from a mean maximum of 20.7°C (July) to a mean minimum of 8.2°C (January). The mean annual rainfall is 967.9 mm, with 80–88% falling between May and October. The zonal vegetation is semi-humid evergreen broadleaved forest. Because trees and shrubs were removed by human activity, rocky desertification characterizes the study site. The site included perennial herbs, primarily Imperata cylindrica and Heteropogon contortus, as well as annual herbs, such as Arthraxon hispidus. Shrubs and small trees seldom occurred in the area. The karst landforms are primarily composed of Carboniferous, Devonian, and Permian carbonate rocks (Sebela et al. 2004). The soils in the region are shallow, and ROCs rise above the ground along with the surrounding soil patches to form a mosaic structure.
Thirty isolated ROCs were randomly selected. To minimize the possibility of interaction between ROCs, a distance greater than or equal to 6 m separated them. The coordinates and morphological parameters of each ROC were recorded at the beginning of the dry season. Five sampling areas were established based on their distance to each ROC (Fig. 1c): A, 0 to 30 cm; B, 31 to 60 cm; C, 61 to 90 cm; D, 91 to 190 cm; and E, 191 to 290 cm. Soil physical and chemical properties and aboveground and belowground plant biomass were determined in the areas.
Morphological parameters of rock outcrops
A self-made profilometer was fixed along the vertical direction of each ROC to measure the surface roughness (Du and Ge, 1999). When a contactor of the profilometer moved, a 20-cm sampling length curve was drawn on a paper. The difference between the highest point of the curve and the reference plane was the roughness value of the sample. Three replicates were taken on the surface of each ROC along horizontal and vertical directions, and their average value was the final roughness value of an ROC. Because there were no significant differences in the roughness of the ROCs (F = 0.162, P < 0.01), the physical and chemical properties of the ROCs in the study area were assumed to be similar.
A leveling staff was used to measure the height of each ROC, and a gradiometer was used to measure their slopes. A Nikon D7200 digital camera was fixed above a sampling subplot at a height of approximately 3.5 m to obtain a photograph. A ruler was placed on the ground to provide a reference scale. Then, the coverage area and perimeter of the rocks were calculated using Image J software.
Rainfall and rock outcrop runoff collection and chemical analyses
The collection system for ROC runoff water was a PVC plastic frame approximately 1 m2 in coverage and approximately 6 cm in height that was inserted into grooves cut by cutting machines on the surface of the rocks and fixed with glue, following Wang et al. (2016a). Rock runoff was the water that collected in the frame and drained into a barrel through a plastic pipe. A 22-cm diameter funnel connected to a barrel was placed next to each rock to collect the precipitation received by a ROC. The water in the barrels was collected and measured monthly.
Water samples were collected from ROC runoff and precipitation in June, August, and October 2019. The samples were analyzed for pH and total N (TN), total P (TP), K+, Ca2+, and total organic carbon (TOC) concentrations to determine their values at different times in the rainy season. The pH was measured using a pH meter (FE28, Mettler Toledo, Shanghai, China). Total N was determined via the alkaline potassium persulfate digestion ultraviolet (UV) spectrophotometric method, and TP was determined using the ammonium molybdate spectrophotometric method. A Shimadzu UV–visible spectrophotometer (UV-2450, Shimadzu Corporation, Tokyo, Japan) was used in the analyses. The K+ and Ca2+ concentrations were determined using inductively coupled plasma atomic emission spectroscopy (iCAP 6300, Thermo Electron Corporation, Waltham, MA, USA).
Soil samples at different distances from rock outcrops and their physical and chemical properties
Six ROCs were randomly selected for soil sampling. Excavations 0 to 30-cm deep were made next to each of the six ROCs, and bulk density was determined with a cutting ring. A soil drill was taken for each sample, and soil was sealed in bags and returned to the laboratory for physical and chemical analyses. The methods described by Danielson and Sutherland (1986) were used to determine soil density, total porosity, initial gravimetric water content, capillary porosity, noncapillary porosity, and capillary holding capacity. The following formulas were used to determine the properties.
![](https://myfiles.space/user_files/83064_0857a92044b57365/83064_custom_files/img1624335202.png)
![](https://myfiles.space/user_files/83064_0857a92044b57365/83064_custom_files/img1624335223.png)
![](https://myfiles.space/user_files/83064_0857a92044b57365/83064_custom_files/img1624335245.png)
![](https://myfiles.space/user_files/83064_0857a92044b57365/83064_custom_files/img1624335285.png)
![](https://myfiles.space/user_files/83064_0857a92044b57365/83064_custom_files/img1624335307.png)
![](https://myfiles.space/user_files/83064_0857a92044b57365/83064_custom_files/img1624335339.png)
where WCR (g) is the weight of the cutting ring (inner diameter 50.46 mm; height 50.0 mm; volume 100 cm3), WCRWS (g) is the weight of the cutting ring filled with natural wet soil, WWD2h (g) is the weight of the cutting ring saturated with distilled water for 24 h and then drained by gravity for 2 h, and WCRDS (g) is the weight of the cutting ring oven-dried at 105°C for 24 h. To calculate total soil porosity, the particle density of the soil was 2.65 g/cm− 3.
The Biogeochemistry Laboratory at the Xishuangbanna Tropical Botanical Garden measured the soil chemical properties. The soil pH values were measured at 1:5 soil:deionized water. The TOC was measured using the potassium dichromate oxidation heating method, the TN by an elemental analyzer (Vario MAX CN, Elementar Analysensysteme GmbH, Hanau, Germany), and the TP, total potassium (TK), available K (AK), and Ca by inductively coupled plasma atomic emission spectrometry (iCAP 7400, Thermo Fisher Scientific, Waltham, MA, US) after digestion in HClO4–HF. The available N (AN) was determined using the alkaline hydrolysis diffusion method, and the available P (AP) was determined using colorimetric analysis.
Aboveground and belowground plant biomass
To obtain biomass samples, a 30 cm × 30 cm frame was used to sample areas A, B, and C at each of the 30 ROCs, and a 100 cm × 100 cm frame was used to sample areas E and F (Fig. 1c). All plants in the framed area were cut and oven-dried at 80°C to a constant weight to determine aboveground biomass (AGB). Drills, 6.8 cm in diameter and 10.0 cm in height, were collected at three layers: shallow (0 to 10 cm), middle (10 to 20 cm), and deep (20 to 30 cm). Roots were carefully washed and collected on a 0.25-mm sieve. After removal of debris, roots were oven-dried at 80°C to a constant weight to determine belowground biomass (BGB). There were three replicates for each layer at the same site.
Statistical analyses
All data were tested for normality and homogeneity of variances. Paired t-tests were used to compare the chemical properties between ROC runoff and rainfall. A one-way ANOVA followed by multiple comparisons was used to test differences between biomass and soil properties among different sampling areas. Correlation analyses were used to test for correlations between the morphological parameters of the ROCs and plant biomass.
Regressions were used to test relations between AGB or BGB and the distance to an ROC. Then, the critical value of the 95% confidence lower limit of the fitted curve to the lower limit of the function was taken as the maximum distance of the effect of an ROC on biomass (A). Logistic regression was used to test relations between AGB or BGB and the coverage area of ROCs. Then, the maximal and average size of an ROC (C) that affected biomass was calculated.