The Influence of Slope Aspect on the Spatial Heterogeneity of Soil Nutrients and Seedling Regeneration in Pinus sylvestris var. mongolica Plantation Forests

Abstract

In the fields of forestry, ecology, and pedology, different slope aspects exhibit significantly different microenvironments and soil conditions, which ultimately lead to disparities in seedling regeneration. Therefore, studying the effects of soil nutrients on seedling regeneration under different microenvironmental conditions can provide critical data for the artificial promotion of natural regeneration. In July 2021, the seedling regeneration status in 900 m² artificial Pinus sylvestris var. mongolica forests with different slope aspects was investigated. Soil nutrient indices were obtained through the collection and measurement of soil samples. Geostatistics were used to quantify the spatial heterogeneity of soil nutrients at a small scale. Soil nutrient information from the seedling growth locations was acquired by combining geographic information system (GIS) technology and laboratory experiments to analyze the effects of soil nutrients on seedling regeneration. The spatial heterogeneity of soil nutrients and their effects on seedling regeneration change with different slope aspects. Even at a small scale (3 m), spatial heterogeneity remains evident. Shaded slopes are more prone to supporting biennial seedlings and older saplings, while seedlings on sunny slopes exhibit superior growth indicators (height and ground diameter). The correlation calculations and redundancy analysis (RDA) of the relationship between soil nutrients and seedling regeneration show that although the soil nutrient content inhibits seedling quantity, they can enhance seedling growth indicators, among which soil organic matter plays the most critical role. Different slope aspects affect soil nutrients and seedling spatial patterns, and increased soil nutrients can promote the natural regeneration of Pinus sylvestris var. mongolica seedlings.

1. Introduction

Soil serves as the foundation for plant growth and development and is the primary source of nutrients and water required by plants [1]. Due to long-term interactions among parent material, topography, climate, and biological factors, soils exhibit a certain degree of spatial heterogeneity, even at very small scales [2]. Geostatistics quantifies the spatial variability and complexity of ecological factors (such as climate, soil, vegetation, etc.) in terms of their distribution through the use of parameters in semivariogram models [3], which is an effective tool for assessing spatial heterogeneity in ecological processes [4]. By understanding the spatial heterogeneity of soil, the soil data for a specific area can be digitized, thereby providing precise information for ecosystem management and crucial data-driven guidance for land managers [5]. In semi-arid regions, where complex vertical ecosystems are often lacking, exploring the horizontal spatial heterogeneity of soil nutrients is particularly important. However, there is a lack of research on the spatial heterogeneity of soil nutrients in key regions for desertification control [5].

The spatial heterogeneity of soil nutrients is a key factor influencing seedling growth and distribution [6]. The seedling stage represents a critical bottleneck for plant nutrient uptake [7,8], a limitation that is particularly pronounced in semi-arid regions [9]. For example, soil pH directly affects seed germination and seedling growth [10], and soil organic matter improves the soil structure and enhances nutrient availability [11,12]. Readily available nutrients, such as alkali-hydrolyzable nitrogen, available phosphorus, and available potassium, directly determine whether seedlings can obtain sufficient nutrition to support their early morphological development [13]. Although the positive effects of soil nutrient spatial heterogeneity on plant performance have been well-documented [14], these findings rarely reflect the responses of seedlings under natural conditions, and few studies have evaluated seedling–environment associations at small scales [15,16,17].

The Hunshandake Sandy Land, located in the monsoon marginal zone, is part of the northern sand control belt within China’s key ecological function zones [18]. With annual precipitation approaching the threshold required by the ecosystem, it is highly sensitive to climate and environmental changes [18]. Pinus sylvestris var. mongolica is the preferred tree species for afforestation in the Hunshandake Sandy Land, due to its excellent cold, drought, and barren soil resistance [19]. In Duolun County alone, on the sandy land’s southeastern edge, over 1000 hectares are covered by Pinus sylvestris var. mongolica forests [20]. As the forest stands mature, the scarcity of older seedlings under the canopy has become increasingly prominent, indicating challenges to the natural regeneration of these plantations [21]. Addressing the issue of natural seedling regeneration is, therefore, critical for sustaining the ecological functions of Pinus sylvestris var. Mongolica plantations in the Hunshandake Sandy Land. This study investigated the spatial distribution and growth of naturally regenerated seedlings within Pinus sylvestris var. mongolica plantations across different slope aspects in the Hunshandake Sandy Land. Through soil sample collection and physicochemical analysis, we used geostatistics to quantify the spatial heterogeneity of soil properties and geographic information system (GIS) technology to obtain soil nutrient data from seedling growth locations.

This study assumes that the spatial heterogeneity of soil nutrients is different on shady and sunny slopes and that it affects the number and growth of Pinus sylvestris var. mongolica seedlings. The microenvironment indirectly regulates the seedling distribution through nutrient dynamics. The objectives include analyzing the heterogeneity of soil nutrients in regard to different slope aspects, quantifying their effects on seedling regeneration, determining the key driving factors, and providing insights for semi-arid plantation management.

2. Materials and Methods

2.1. Site Description

The study area is located in an artificial Pinus sylvestris var. mongolica forest (stand age 18 years) in sandy areas within Dabao Ecological Park, Duolun County, Xilingoule League, Inner Mongolia Autonomous Region (42°08′22.55″ N, 116°29′21.03″ E), located at the southeastern edge of the Hunshandake Sandy Land. The study area is situated in a temperate, semi-arid to semi-humid transitional continental climate zone. Its main climatic characteristics are as follows: dry and windy springs, cool summers with frequent thunderstorms, mild autumns, and long, frigid winters. The mean annual temperature is 1.6 °C, with a mean temperature in January of −18.3 °C and a mean temperature in July of 18.7 °C. Historical extreme temperatures have reached −39.8 °C (29 December 1954) and 35.4 °C (23 July 1955). The mean annual precipitation is 385.5 mm, concentrated primarily in July–September. The mean annual number of gale days is 30–65, accompanied by frequent occurrences of sandstorms, blowing sand, and dust-floating events.

The soil types are mainly chestnut soil and sandy soil [21], with the parent material consisting primarily of aeolian sand and alluvial sand. This region is a typical agro-pastoral economic zone within China’s northern agro-pastoral ecotone. The terrain is hilly and sandy, with an altitude of 1280 m. Large areas of the study region are planted with Pinus sylvestris var. mongolica, while the natural vegetation includes Ulmus pumila, Betula platyphylla, and Populus spp. The herbaceous plants in the area include Artemisia frigida, and Leymus chinensis (Trin.) Tzvel., and Setaria viridis. In July 2021, one permanent plot (900 m², 30 m × 30 m) was established on two contrasting slope aspects within the temperate, semi-arid sandy Pinus sylvestris var. mongolica plantation ecosystem (Figure 1). The other site conditions of the two plots were relatively similar, with information on the altitude and slope gradient recorded. The stand and microclimate characteristics of the plots are presented in

Table 1. Characteristics of two stands with different slope aspects.

Stand	Slope Aspect	Area (m2)	Slope (°)	Altitude (m)	Longitude (°)	Latitude (°)	Density (Stems·hm2)	Crown Density	Mean Height (m)	Mean DBH (cm)
I	sunny slope	900	12	1287.519	116.489197 N	42.139612 E	1667	0.6	9.39 ± 1.93	9.97 ± 2.81
II	shady slope	900	11	1298.317	116.490558 N	42.140400 E	1722	0.7	9.37 ± 2.00	10.00 ± 3.07

Note: Data in the table represent the mean ± standard deviation.

and

Table 2. Microclimate of two stands with different slope aspects.

Index	Months	Sunny Slope	Shady Slope
PPFD (μmol/m2∙s)	7	498.47 ± 334.58	359.16 ± 248.19
	8	416.54 ± 292.51	460.87 ± 371.99
	9	225.80 ± 163.83	129.96 ± 120.01
	10	167.45 ± 134	62.00 ± 59.63
	Total	327.07 ± 184.29	253.00 ± 130.04
Air temperature (°C)	7	25.20 ± 0.59	26.73 ± 4.81
	8	23.06 ± 0.31	24.42 ± 0.45
	9	17.74 ± 0.17	17.70 ± 0.16
	10	9.31 ± 0.28	9.00 ± 0.19
	Total	18.83 ± 0.26	19.40 ± 1.21
Relative air humidity (%)	7	49.22 ± 1.19	55.91 ± 63.62
	8	43.8 ± 1.86	39.92 ± 5.30
	9	64.66 ± 18.69	60.14 ± 6.23
	10	25.2 ± 2.86	23.9 ± 0.52
	Total	45.72 ± 4.79	44.97 ± 15.98

Note: Photon flux density (PPFD) refers to the luminous flux density in photosynthetically active radiation, which represents the number of photons incident in the wavelength range of 400–700 nm per unit of time per unit area. Data in the table represent the mean ± standard deviation.

, respectively.

2.2. Sampling Point Setting

Apart from different slope aspects, the two plots had similar altitude and slope gradients, with no significant differences in the stand characteristics. The mother trees of Pinus sylvestris var. mongolica were planted at a spacing of 2 m × 3 m. Each plot was gridded and divided into 100 subplots (3 m × 3 m). From the center of each subplot, soil samples were collected from the 0–20 cm depth layer. Sampling strictly followed the principle of single-point sampling: only one soil sample was collected from the center of each subplot to avoid the loss of micro-scale spatial variation information through the use of composite sampling. A total of 200 soil samples were collected and transported to the laboratory for nutrient analysis. During the transportation process, ensure that the soil samples are in a low-temperature and dry environment. The whole process of sampling, transportation, and laboratory storage is strictly protected from light to minimize the impact of environmental factors on soil nutrient indicators.

2.3. Vegetation Investigation

A comprehensive plant survey was conducted. For all the mother trees within each subplot, the height, age, canopy diameter, diameter at breast height (DBH), and location were measured and recorded. For biennial seedlings and older saplings (>2 years old) of Pinus sylvestris var. mongolica, the seedling height (SH), ground diameter (GD), and location were measured and recorded. Seedling age was determined using lateral branch whorls, with older saplings defined as those older than two years. Considering the vulnerability of annual seedlings, measuring their caliper diameter and seedling height could potentially cause damage to them. Therefore, only the number of annual seedlings in each subplot was recorded; their caliper diameter and seedling height were not assessed. Finally, a total of 17,264 seedlings were counted, and the height and ground diameter of 2578 seedlings (biennial and older) were measured.

All the vegetation surveys were completed based on the divided independent subplots. As the study area was sandy before the artificial planting of Pinus sylvestris var. mongolica, the sample area contained only Pinus sylvestris var. mongolica mother trees, regenerating seedlings, and a small number of herbaceous plants. This simple species composition facilitated our targeted study of the impact of soil on seedling regeneration.

2.4. Determination of Environmental Factors

During the growing season (July to October) in 2021, six sunny days per month were selected. At the center of each subplot, the photosynthetic photon flux density (PPFD), air temperature, and air humidity were measured using a quantum/illuminance sensor (Apogee Instruments Inc., Logan, UT, USA) and a digital thermo-hygrometer (TES Electrical Electronic Corp., Taiwan, China). Measurements were taken daily between 09:00 and 11:00 h and completed within a 2 h window. The average value for each subplot was calculated and used to represent its microclimate conditions. Simultaneously, the soil volumetric water content (VWC) at 0–20 cm depth was measured directly at the center of each subplot, using a portable time domain reflectometer (TDR 200, Spectrum Technologies Inc., Plainfield, IL, USA). Each sampling point was measured five times, and the average value was recorded as the monthly surface soil (0–20 cm) VWC for that subplot.

2.5. Determination of Soil Factors

The samples were air dried in a ventilated laboratory under dark conditions. During the drying period of the soil, turn the sample every day to avoid mildew. After the manual removal of stones and visible plant residues, the air-dried samples were ground and sieved to <2 mm for the soil pH determination (using a Rex PHS-25 pH meter, Shanghai, China). The soil pH was measured in a 1:2.5 soil/water suspension. Some of the soil was ground and passed through a 0.25 mm sieve. The prepared soil samples were used for soil nutrient analysis [22]. The soil organic matter (OM) and total nitrogen (TN) were determined using an elemental analyzer (Vario EL III, Elementar, Langenselbold, Germany). The total phosphorus (TP) was determined using the sulfuric acid–perchloric acid digestion–molybdenum antimony colorimetric method. The available phosphorus (AP) was determined using the 0.5 mol/L sodium bicarbonate extraction method. The available potassium (AK) was determined using the ammonium acetate extraction–flame photometric method.

2.6. Data Analysis

The descriptive statistics for the soil nutrients and other indicators (OM, TN, TP, AP, AK, pH, VWC) were calculated using SPSS 27.0 for Windows. Calculate the minimum, maximum, mean, standard deviation (SD), and coefficient of variation (CV) of the original data. In classical statistics, the coefficient of variation is an important parameter for the analysis of data variability. The variation is considered weak when the CV ≤ 0.1, moderate when 0.1 < CV < 1, and strong when the CV ≥ 1. Geostatistics requires the assumption of intrinsic stationarity. The normality was assessed to minimize the influence of outliers on semivariogram modeling. The normality was evaluated using the Kolmogorov–Smirnov (K-S) test and by examining the skewness and kurtosis values. Logarithmic conversion was performed on the data that did not obey the normal distribution. The semivariogram is the primary tool for analyzing spatial heterogeneity in geostatistics [23]. The semivariance of regionalized variables $\gamma (h)$ can be calculated using the following Equation (1):

$$\gamma (h)={\displaystyle \frac{1}{2N(h)}}{\displaystyle \sum _{i=1}^{N(h)}{[Z(x_i)-Z(x_i+h)]}^2}$$

(1)

Among them, $N\left(h\right)$ is the number of samples at the step size h, and $Z\left(x_i\right)$ and $Z(x_i+h)$ are the measured values of the regionalized variable $Z\left(x\right)$ at the spatial positions $x_i$ and $x_i+h$, respectively. After calculating the experimental semivariogram, the theoretical models (e.g., Gaussian, exponential, spherical) were fitted.

The semivariogram mainly judges the spatial heterogeneity of the index using three important parameters: nugget value (C₀), sill value (C₀ + C), and range (A₀). The block basis ratio C₀/(C₀ + C) is used to judge the degree of spatial autocorrelation of the variables. If C₀/(C₀ + C) > 75%, it means that the spatial autocorrelation is weak. If C₀/(C₀ + C) is 25%–75%, the spatial autocorrelation is moderate; if C₀/(C₀ + C) is less than 25%, the spatial autocorrelation is strong. The range (A₀) is a parameter that measures the size of the spatial autocorrelation range of variables. When the spatial distance between two points is greater than A₀, the spatial autocorrelation will disappear.

The geostatistical analysis was performed using GS+9.0 software (Gamma Design Software, Plainwell, MI, USA). Semivariogram model selection was performed automatically in GS+9.0 by comparing the goodness-of-fit statistics. For each variable, the model with the highest coefficient of determination (R²) and the lowest residual sum of squares (RSS) was selected. ArcGIS 10.3 (Esri, Redlands, CA, USA) was used to perform ordinary kriging interpolation on the soil variables and to map their spatial distribution, along with seedling locations. The kriging predictions were back-transformed to the original units for mapping and interpretation. The soil properties at seedling growth locations were extracted using the GIS spatial analyst ‘Extract Values to Points’ tool (ArcGIS 10.3), based on the Kriged soil surfaces. Spearman correlation analysis was used to test the correlation between the soil indexes (OM, TN, TP, AP, AK, pH, VWC) and the seedling regeneration characteristics, such as the number, seedling height (SH), and ground diameter (GD). Origin 2024 (OriginLab Corporation, Northampton, MA, USA) was used to generate correlation heatmaps and redundancy analysis (RDA) ordination diagrams.

3. Results

3.1. Microenvironment and Soil in Regard to Different Slope Aspects

As shown in Table 1, there was no significant difference in the stand characteristics between the different slope aspects; however, the difference in the microclimate caused by different slope aspects was significant (Table 2). The PPFD was significantly higher on sunny slopes than on shady slopes, but the air temperature of the shady slope was slightly higher than that of the sunny slope. The descriptive soil statistics for the two different slope aspects are presented in

Table 3. Descriptive statistics of soil in regard to two stands with different slope aspects.

Slope Aspect	Items	Min.	Max.	Mean	SD	CV (%)	Kurtosis	Skewness	K-S D	K-S P
Sunny slope	OM (g/kg)	3.50	21.24	8.14	4.02	49%	0.46	0.93	0.12	0.001
	TN (g/kg)	0.67	1.47	0.97	0.18	18%	−0.29	0.58	0.10	0.010
	TP (mg/kg)	58.55	222.49	103.25	22.59	22%	10.60	2.35	0.15	0.000
	AP (mg/kg)	14.77	62.43	31.59	7.48	24%	1.72	0.82	0.08	0.117
	AK (mg/kg)	2.47	27.22	9.13	4.32	47%	2.18	1.09	0.18	0.000
	pH	5.80	6.22	6.01	0.09	1%	−0.11	−0.02	0.07	0.200 *
	VWC (%)	4.80	9.11	6.38	0.82	13%	0.60	0.64	0.11	0.009
Shady slope	OM (g/kg)	3.41	27.24	9.05	4.65	51%	2.68	1.52	0.16	0.000
	TN (g/kg)	0.49	1.65	0.97	0.24	25%	0.38	0.52	0.08	0.102
	TP (mg/kg)	83.90	139.28	101.46	9.31	9%	3.05	1.12	0.09	0.039
	AP (mg/kg)	17.00	62.56	33.55	8.45	25%	1.20	0.91	0.13	0.000
	AK (mg/kg)	2.47	32.17	10.34	5.69	55%	1.51	1.06	0.17	0.000
	pH	5.73	6.18	5.95	0.09	2%	−0.30	0.02	0.06	0.200 *
	VWC (%)	4.87	8.58	6.57	0.87	13%	−0.85	0.24	0.10	0.015

Note: The sample size of each soil item was 100. Soil volumetric water content is the average value from July to October. The “*” representative data conforms to the normal distribution after being tested.

. The soil nutrient levels were generally higher on shady slopes, except for the soil total phosphorus and soil pH. The coefficient of variation of the soil organic matter and soil available potassium in regard to different slope directions exceeded 45%, showing moderate variation. For the other soil properties, the CV ranged from 9% to 25%, which indicates the heterogeneous distribution of soil nutrients. However, the coefficient of variation of the soil volumetric water content in regard to the different slope directions is less than 15%, and the coefficient of variation of the soil pH is less than 2%. Therefore, the spatial dispersion of the soil water content and soil pH is small.

In this study, a total of 17,264 seedlings were counted, and the seedling height and ground diameter of 2578 seedlings were measured (

Table 4. Seedling density and growth in regard to stands with different slope aspects.

Slope Aspect	Seedling Type	Number	Density (Plant/m2)	Height of Seedlings (cm)	Ground Diameter of Seedlings (mm)
Sunny slope	Annual seedlings	7308	8.12	no data	no data
	Biennial seedlings	556	0.62	4.45 ± 2.55	0.72 ± 0.31
	Older saplings	34	0.04	13.62 ± 7.68	2.14 ± 0.82
Shady slope	Annual seedlings	7378	8.20	no data	no data
	Biennial seedlings	1910	2.12	3.8 ± 1.1	0.64 ± 0.13
	Older saplings	78	0.09	6.69 ± 3.1	0.96 ± 0.38

). These seedlings are the type we would prefer to see because they represent the potential for more superior natural regeneration. According to the age of the seedlings, the seedlings were divided into annual seedlings, biennial seedlings, and older saplings. There was no significant difference in the number of annual seedlings in regard to different slope directions. The number of biennial seedlings on the shady slope was significantly higher than that of the sunny slope, and the number of older saplings on the shady slope was more than twice that of the sunny slope. However, the seedling height and ground diameter of the regenerated seedlings on the sunny slope were significantly higher than those on the shady slope, especially the ground diameter of the older saplings, which was almost twice that of the shady slope.

3.2. Spatial Heterogeneity of Soil Nutrients in Regard to Different Slope Aspects

The semivariogram model parameters of the soil nutrients in regard to different slope directions are shown in

Table 5. Parameters of semivariogram models of two stands with different slope aspects.

Slope Aspect	Soil Property	C0	Sill (C0 + C)	A0 (m)	C0/(C0 + C)	Theoretical Model	R2	RSS
Sunny slope	OM	0.0122	0.1156	55.69	10.55%	Gaussian	0.989	3.27 × 10−5
	TN	0.00192	0.01012	35.47	18.97%	Gaussian	0.981	8.99 × 10−7
	TP	0.00559	0.01198	170.82	46.66%	Exponential	0.683	1.89 × 10−6
	AP	0.00605	0.029	70.17	20.86%	Gaussian	0.932	5.40 × 10−6
	AK	0.036	0.0906	213.00	39.74%	Exponential	0.830	4.30 × 10−5
	pH	0.00497	0.01563	163.38	31.80%	Exponential	0.715	4.91 × 10−6
	VWC	0.00045	0.00692	60.02	6.50%	Spherical	0.975	3.18 × 10−7
Shady slope	OM	0.01386	0.04052	27.46	34.21%	Spherical	0.931	3.49 × 10−5
	TN	0.00423	0.02866	46.83	14.76%	Gaussian	0.98	5.24 × 10−6
	TP	0.00093	0.00545	80.40	17.06%	Gaussian	0.973	5.09 × 10−8
	AP	0.0039	0.0127	23.57	30.71%	Spherical	0.926	3.82 × 10−6
	AK	0.0054	0.066	6.39	8.18%	Exponential	0.262	2.53 × 10−4
	pH	0.00292	0.02084	45.29	14.01%	Gaussian	0.997	5.20 × 10−7
	VWC	0.00001	0.00389	19.69	0.26%	Spherical	0.967	3.05 × 10−7

Note: The sample size of each soil item was 100.

. The range (A₀) ranged from 6.39 m to 213 m, which is greater than the distance between the different sampling points. Except for the R² of AK in regard to the shady slope, which was 0.262, the R² values approached 1, indicating that the fitted model accurately captured the spatial structure characteristics of the soil factors. The Gaussian model, exponential model, and spherical model are the best-fitting models of the TN, AK, and VWC in regard to the different slope directions, respectively. The other soil indexes have different best-fitting models in regard to the different slope directions. The nugget ratios of the OM, TN, AP, and VWC in regard to the sunny slope and the TN, TP, AK, pH, and VWC in regard to the shady slope were all less than 25%, which means that they have strong spatial dependence, indicating that structural factors played a dominant role in the spatial structure of these indicators. The nugget ratio of the TP and AK on the sunny slope is much larger than that of the shady slope, and shows moderate spatial dependence. This is because the existence of random factors in the field leads to a decrease in their spatial dependence. The nugget ratio of the TP in regard to the sunny slope is the largest (46.66%) of all the indicators, but it is still far less than 75%, so the soil indicators of the different slope aspects have certain spatial autocorrelation characteristics.

According to the concept and model of the semivariogram, the spatial distribution of the soil total nitrogen and soil volumetric water content with strong spatial autocorrelation in different slope directions was interpolated using kriging (Figure 2). The cross-validation diagnostics confirmed the reliability of the kriging interpolation models (

Table 6. Cross-validation diagnostic results from ordinary kriging model.

Slope Aspect	Items	ME	RMSE	MSE	SA	NRMSE
Sunny slope	OM	0.14	2.35	2.61	0.07	0.99
	TN	0	0.11	0.11	0.06	0.98
	TP	0	22.8	19.97	−0.02	1.1
	AP	−0.13	6.28	6.02	−0.03	1.03
	AK	0.02	4.22	4.63	−0.02	0.98
	pH	0	0.07	0.07	−0.01	0.92
	VWC	0	0.31	0.4	0.01	0.75
Shady slope	OM	−0.02	3.25	3.52	0.02	0.89
	TN	0	0.16	0.16	0.03	0.96
	TP	0.19	8.3	8.1	0.02	1.05
	AP	−0.03	6.78	6.64	0	0.99
	AK	0.14	5.05	6.57	0.02	0.78
	pH	0	0.06	0.06	0	1.1
	VWC	0	0.21	0.31	0.01	0.69

Note: The sample size of each soil item was 100. ME, mean error; RMSE, root mean squared error; MSE, mean standard error; SA, standard average; NRMSE, normalized root mean square error.

). The soil total nitrogen showed a clear zonal distribution on the sunny slope, and a localized low-value center occurred centrally in the sample plot. Biennial seedlings predominated at 0.81–1.02 g/kg TN, while the older saplings were concentrated in the southwest region, with a relatively high soil total nitrogen, and the soil total nitrogen in this region was in the range of 1.02 g/kg–1.31 g/kg. The distribution of the soil total nitrogen in regard to the shady slope was patchy, and there were obvious high-value and low-value centers in the central and northeastern corners of the southern margin. Different from the sunny slope, the biennial seedlings and older saplings on the shady slope were concentrated in the range of 0.92 g/kg–1.11 g/kg of the soil total nitrogen, and there was no dense seedling distribution in the areas with low and high soil total nitrogen content.

The soil volumetric water content on the sunny slope showed an obvious patchy distribution. The area with higher water content was located on the west side of the plot, and the low numerical center was located in the central area of the plot, which accounted for a large proportion of the total. Most of the biennial seedlings are distributed in the area with low soil moisture content, and the older saplings are mainly distributed in the southwest corner with a high soil volumetric moisture content. The specific values are between 7.23% and 9.11%. The soil volumetric water content of the shady slope also shows a patchy distribution, but the specific spatial distribution characteristics are more complex. The areas with higher water content are distributed in the northwest and southeast corners of the plot, and there are three high numerical centers. The distribution of older seedlings on the shady slope is dense and concentrated, mainly distributed in the middle area with a low soil moisture content, and the specific value in this area is between 4.87% and 6.31%.

3.3. Correlation Analysis Between Soil Nutrients and Seedling Regeneration

It can be seen from Table 4 that the number of seedlings and the regeneration in terms of the seedling height and ground diameter in regard to different slope directions are completely different. Therefore, the results of this study divide the correlation between the soil nutrients and seedling regeneration into two parts: seedling number and seedling development (seedling height and ground diameter). The Spearman correlation between the soil nutrients and seedling regeneration in regard to different slope aspects is shown in Figure 3.

On the sunny slope, the pH increases the biennial seedling density, which is contrary to the results of the shady slope. The soil organic matter and soil available phosphorus will inhibit the number of seedlings in regard to different slope directions.

The effects of the soil nutrients in regard to different slope directions on seedling development were complex. The seedling height of the biennial seedlings on the sunny slope was significantly positively correlated with the soil organic matter, soil total nitrogen, and soil volumetric water content. The ground diameter was significantly negatively correlated with the soil total phosphorus and soil pH.

The soil nutrients in regard to the shady slope could promote the development of biennial seedlings, and only the soil available phosphorus and soil pH were significantly negatively correlated with seedling development.

Compared with the biennial seedlings, the soil total phosphorus and soil pH on the sunny slope were significantly positively correlated with the seedling height and ground diameter of older saplings. The soil available phosphorus on the shady slope was significantly negatively correlated with the seedling height of older saplings. The soil total phosphorus was significantly positively correlated with the seedling height of older saplings. The organic matter and pH showed significant positive correlations with older sapling growth.

Through the use of redundancy analysis (Figure 4), we tried to find the most significant factor for seedling regeneration among all the soil factors, and the results were consistent. Except that the effect of the soil organic matter on seedling development on the sunny slope was second only to the soil moisture content, the soil organic matter was the most important factor in regard to different slope directions. However, the role of soil organic matter is not the same in both conditions, mainly in that the soil organic matter will inhibit the number of seedlings, but can promote the development of seedling height and the ground diameter of older saplings, which is similar to the role of other soil nutrients.

4. Discussion

In this study, we investigated the effects of spatial heterogeneity of soil nutrients in regard to different slope aspects on the seedling regeneration of Pinus sylvestris var. mongolica. Undoubtedly, soil nutrients affect the regeneration of seedlings, but the nutrient effects varied with the slope aspect in terms of seedling regeneration. This reflects a trade-off between plant survival and growth under different microenvironments and soil conditions [21]. Our results support the perspective that seedling regeneration depends on the combined effects of environmental factors, which includes climatic factors closely related to seedling growth [24] and complex biological processes [25].

4.1. Effect of Slope Aspect on Soil Spatial Heterogeneity

The growth response of plants to resources may be an important mechanism affecting species distribution, coexistence, and community structure. In addition to soil nutrients [26], light is also considered to be an important resource for seedling growth. The microclimate conditions, such as light and temperature, on different slopes in semi-arid areas are quite different [27]. The slope aspect determines the soil moisture content, soil temperature, and soil nutrients by affecting solar radiation [28,29]. While the soil nutrients were sampled once in July, their spatial patterns represent seasonally stable gradients in regard to this semi-arid system. The use of growing season mean microclimate data captures the persistent spatial drivers of ecological processes [30]. This approach is common in studies coupling stable soil properties with dynamic climate factors [31].

Zhou et al. explored the effect of grazing on the spatial heterogeneity of the soil total nitrogen in a semi-arid grassland area (block basis ratio 5%–19%) [30], and the results were slightly lower than the block-to-base ratio of the soil total nitrogen in the Pinus sylvestris var. mongolica plantation (block basis ratio 15%–19%). The difference in soil spatial heterogeneity in similar research areas is not large, and the soil spatial structure of grassland is slightly stronger than that of woodland. However, nutrient fluctuations exacerbated by strong radiation make the coefficient of variation of organic matter in regard to sunny slopes (50%) far exceed the results recorded by Qin et al. in regard to temperate natural forests (15%–25%) [32].

The soil nutrients (such as TP, AK, and pH) on the sunny slope showed higher spatial randomness (higher C₀/(C₀ + C)) than those on the shady slope, which confirmed that the aspect caused differences in soil spatial heterogeneity by regulating solar radiation [33]. Its direct manifestation is that the soil characteristics in terms of different slope directions do have different best-fitting models [29]. While different models may yield varying range estimates, our model selection, based on R² and RSS, ensures the most accurate spatial structure representation for each soil parameter. Even though the study area is relatively concentrated, the soil volumetric water content in terms of the sunny slope still has a clear patchy distribution, and there are even three high numerical centers, which means that even at the small-scale research level, the soil still has strong spatial heterogeneity [29]. The patchy distribution of the soil revealed in this study relates to a microscale of 3 m. This finding can provide a reference value for the sampling design of related experiments in the future.

4.2. Effects of Slope Aspect on Seedling Regeneration Pattern

Regeneration differed significantly across the slope aspects, with aggregated distributions on shady slopes, which was closely related to the characteristics of Pinus sylvestris var. mongolica species and the heterogeneity of environmental factors such as the microclimate and soil [31]. Light is critical during seedling regeneration, leading to superior regeneration on the sunny slope than on the shady slope. This may be because Pinus sylvestris var. mongolica, a fast-growing light-loving tree species, has a higher demand for light and soil [34,35], and can adapt to sunny slopes with less soil moisture [36].

The density of old seedlings on the shady slope (0.09 plants/m²) was significantly lower than that of natural P. sylvestris forests [31], whereas the GD was nearly double on sunny aspects than shady slopes. The growth advantage brought about by the sunny slope in regard to Pinus sylvestris var. mongolica seedlings exceeded the microclimate control experimental results by Hill et al. in the arid area of the western United States [24], and we found a consistent rule, that is, the ground diameter of Pinus sylvestris var. mongolica seedlings exhibited superior growth in a low density and relatively warm microclimate.

Although the number of seedlings on the shady slope was significantly higher than that of the sunny slope, the quality of seedlings was much lower than that of the sunny slope. In nutrient-deficient environments, a high seedling density may lead to severe interspecific competition, which may lead to serious adverse effects on seedling regeneration [37]. Nutrient patching can also lead to the excessive aggregation of seedlings in eutrophic areas. Plants adopt a ‘diffusion strategy’ in homogeneous environments and an ‘aggregation and competition strategy’ in heterogeneous environments [38]. Janzen-Connell’s hypothesis points out that plants have a stable mechanism to promote species coexistence and maintain diversity, which makes intraspecific competition for resources more intense than interspecific competition [39]. In general, the survival rate of individuals in the same high-density area was significantly lower than that in the same low-density area, which could explain the difference in the regeneration of Pinus sylvestris var. mongolica seedlings in regard to different slope aspects to a certain extent [40].

4.3. Effects of Soil Nutrients on Seedling Regeneration

The soil nutrient heterogeneity varied substantially with the slope aspect. In nutrient-poor plantations, competition increases plant sensitivity to soil heterogeneity [41]. The regeneration of seedlings in the patchy distribution area of soil nutrients is usually better than that in the uniform distribution area [42]. In regard to seedling density, all the significant soil factors showed negative correlations with the seedling number, indicating that the increase in soil nutrients may aggravate the intraspecific competition of Pinus sylvestris var. mongolica seedlings [43].

Although the coefficient of variation of many indicators (such as pH and VWC) in this study was low, their effects were pronounced in different microenvironments and presented a complex pattern. The soil pH is considered to be the control variable of soil chemistry, which has a profound impact on numerous chemical reactions involving essential nutrients for plants [44]. The SH and GD of older saplings on the sunny slope were significantly positively correlated with soil pH, while the ground diameter of biennial seedlings on the shady slope was negatively correlated with soil pH. This finding supports the view that pH indirectly affects seedling growth by regulating nutrient availability and plant physiological processes [10].

Even though the nutrient requirement of Pinus sylvestris var. mongolica seedlings is low in the early stage of regeneration [45], the availability of nutrients improved by soil organic matter is still crucial for seedling development [46]. Soil organic matter, as the most important influencing factor, is consistent with the law of action of other soil nutrients, that is, regulating seedling regeneration through density-dependent effects. The decrease in seedlings caused by the increase in organic matter was 2.5 times that in the humid area, which confirmed that the nutrient competition in the semi-arid area was more intense [6]. High-nutrient patches inhibit the number of seedlings, but promote the morphological development of surviving individuals through nutrient release. This also means that the ‘quantity–quality trade-off’ is an adaptive strategy for Pinus sylvestris var. mongolica in order for them to cope with heterogeneous environments. Plants can adjust their growth strategies to adapt to different climates and soil conditions [47], which explains the negative effect of soil nutrients on the number of seedlings to a certain extent. Here again, we explore the changes in seedling regeneration strategies [48,49], and how to balance quantity and quality during seedling regeneration [50], which is an interesting question.

4.4. Research Limitations and Management Implications

Our research is always focused on promoting the natural regeneration of Pinus sylvestris var. mongolica seedlings. Based on the identified mechanisms, we recommend the adoption of aspect-specific management practices. High seedling density was maintained on the shady slope, but the intraspecific competition was alleviated by thinning after 3 years of settlement. In regard to the sunny slope, the patchy nutrients were improved through micro-topography transformation to promote seedling growth. At the same time, according to the spatial heterogeneity of soil, precise nutrient regulation is carried out. Biochar fertilizer is applied to promote the carbon and nitrogen balance in areas with low organic matter content and straw is laid in areas with low water content to reduce evaporation to realize sustainable renewal in terms of ‘ensuring the number of seedlings on shady slopes and promoting individual quality on sunny slopes’.

In this study, the stand characteristics, climate, and soil factors can only explain a small part of seedling regeneration, which highlights the complexity of multi-scale ecological processes. Although different regeneration strategies (a high number of seedlings on the shady slope and a high quality of seedlings on the sunny slope) were revealed, the underlying mechanism of this difference was not quantified. We did not monitor the biological interaction process; biological factors may mask the direct effects of soil nutrients. At the same time, the point data from a single time point cannot be used to analyze the key ecological processes, which limits the assessment of the sustainability of the updated strategy.

Future research should continue to track seedling population dynamics to obtain the birth rate and mortality of seedlings. It is necessary to carry out a multi-dimensional analysis of the relevant biological factors, such as soil microbiome sequencing. Another important are of work is to track the source and transformation of organic matter. Of course, future studies should develop cross-scale models that embed small-scale heterogeneity in regard to larger landscape scales to enhance the predictions of regional ecological functions and provide managers with more suggestions and measures.

5. Conclusions

In semi-arid areas, even on a small scale, soil factors still have obvious spatial heterogeneity, and the spatial heterogeneity of soil nutrients in regard to different slope aspects is very different. The spatial autocorrelation of the soil nutrients (such as total nitrogen and organic matter) on the shady slope was generally stronger than that of the sunny slope, while the soil total phosphorus and other indicators on the sunny slope showed higher randomness than the shady slope. Among them, the soil total nitrogen and soil volumetric water content have strong spatial autocorrelation in regard to different slope directions. The two factors mainly showed patchy or zonal distributions in regard to different slope directions, with obvious high-value and low-value centers. The shady slope was characterized by high seedling density, but the individual growth index (seedling height/ground diameter) was low, which may be related to the intensification of intraspecific competition in nutrient-rich areas. The density of the seedlings on the sunny slope was low, but the growth of the surviving biennial seedlings and older saplings was significantly better than that of the shady slope, which reflected the ‘quality first’ strategy. The increase in soil organic matter and available phosphorus may inhibit the number of seedlings, which supports the density-dependent effect hypothesis. Soil nutrients have a significant promoting effect on the morphological development of older saplings, among which soil organic matter plays the most critical role, but the slope aspect regulates its action pathway. The effect of soil nutrients on natural regeneration needs to be viewed dialectically in combination with the slope aspect. Although a high amount of nutrients promote individual growth, it may be at the expense of the number of seedlings.