Evaluation of Soil Health of Panax notoginseng Forest Plantations Based on Minimum Data Set

Abstract

Healthy soil serves as the fundamental basis for sustainable Panax notoginseng (Burkill) F.H. Chen ex C.Y. Wu & K.M. Feng cultivation in understory systems. Current management practices have raised concerns about potential soil degradation and ecological imbalance. To comprehensively assess the soil health status, this study investigated typical understory P. notoginseng plantations in the subtropical mountain monsoon region of western Yunnan. By analyzing 29 soil physical, chemical, and biological indicators, we constructed a Minimum Data Set (MDS) using Principal Component Analysis to evaluate soil health and identify major constraints. The results showed that the MDS for soil health assessment consisted of 11 key indicators: acid phosphatase, fungal ACE index, organic matter, total nitrogen, sucrase, fungal Simpson index, fine sand, non-capillary porosity, silt content, bulk density, and microbial biomass nitrogen. Using both linear and non-linear scoring functions, the Soil Health Index (SHI) calculated based on the MDS showed a significant positive correlation with the SHI derived from the Total Data Set (TDS) (linear scoring: R² = 0.43, p < 0.001; non-linear scoring: R² = 0.305, p < 0.001). This indicates that the MDS captures a substantial and significant portion of the variation explained by the TDS and can serve as a practical and simplified alternative for soil health evaluation in this cultivation system. Based on the MDS, the SHI values obtained using linear and non-linear scoring functions ranged from 0.53 to 0.72 and 0.48–0.59, with mean values of 0.62 and 0.51, respectively, indicating moderate soil health status in the study area. Significant differences in SHI were observed across planting durations and seasons (p < 0.05), with two-year-old plantations showing notably better soil health indices than three-year-old plantations, particularly during the rainy season. The main constraints identified in understory P. notoginseng plantations included microbial community degradation, nutrient imbalance, and physical structural deterioration. Implementing scientific soil management strategies such as optimized rotation cycles, organic amendment applications, and microbial community regulation can effectively mitigate these soil constraints, enhance soil health, and promote the sustainable development of understory P. notoginseng cultivation.

1. Introduction

Ensuring food security and ecological stability through sustainable agriculture is a global challenge, particularly within ecologically sensitive regions that are also biodiversity hotspots [1,2,3]. Yunnan Province in southwestern China, recognized as one of the world’s 34 biodiversity hotspots [4,5,6], epitomizes this challenge. It must balance the conservation of its immense biological wealth—hosting 56% of China’s higher plant and 51% of its vertebrate species [5,7]—with the economic development needs of its population [8]. This task is complicated by prevalent karst landforms (44% of the area) with an average topsoil thickness of less than 50 cm and a high soil erosion sensitivity index of 0.72 [9,10,11]. In this context, the understorey economy model, which integrates crop cultivation beneath forest canopies, has emerged as a promising strategy for synergistic ecological and economic development [12,13,14]. Supported by provincial initiatives like the Understorey Economy Development Plan (2021–2025) [12], this model has expanded to over 1.2 million hectares in Yunnan, with 38% dedicated to valuable medicinal herbs such as Panax notoginseng [15].

While transferring Panax notoginseng cultivation from arable land to forest understories offers economic benefits, alleviates land-use pressure [14,16], and can produce a premium product, its long-term sustainability is not guaranteed. The traditional cultivation on arable land accounts for over 98% of national production but faces a triple dilemma: prolonged land occupation, increased soil-borne disease incidence (40%–60%) due to continuous cropping obstacles [17,18], and competition with food crops threatening regional food security [19]. Evidence suggests that even understorey cultivation can lead to soil degradation, notably a 12%–15% decline in soil microbial diversity after three consecutive years [20]. This underscores the urgent need for robust soil health assessment tools to guide sustainable management and prevent the accumulation of cultivation obstacles in these fragile ecosystems, thereby establishing a sustainable management paradigm based on soil health thresholds [20].

Traditional soil quality assessments, often reliant solely on physicochemical indicators, are insufficient for this task as they fail to capture critical biological dynamics [18,19,21,22]. A more comprehensive approach is offered by the Soil Health Index (SHI) framework, which integrates physical, chemical, and—crucially—biological properties, providing a more accurate reflection of agroforestry system sustainability [23]. To implement this framework efficiently, the Minimum Data Set (MDS) method provides a powerful tool. The MDS employs multivariate statistics, such as Principal Component Analysis (PCA), to objectively identify the most informative and non-redundant indicators from a large initial dataset [24,25,26,27]. This method’s key advantage lies in its ability to reduce cost and complexity while retaining the essential information required for a meaningful evaluation, making it superior to approaches that either use a limited set of subjective indicators or an unwieldy full dataset.

This study innovates by constructing an MDS that explicitly integrates soil microbial community structure (e.g., diversity indices from high-throughput sequencing) and enzyme activity profiles with conventional soil properties. We hypothesize that this multidimensional MDS will sensitively reveal the impact of understorey Panax notoginseng cultivation on soil health across different planting durations and seasons. Therefore, the objectives of this study are to: (1) quantify the effects of understorey planting on soil physical, chemical, and biological properties; (2) elucidate the regulatory effects of seasonal dynamics and planting duration on soil health; and (3) establish a dynamic SHI to assess the sustainability of this cultivation model and provide a scientific basis for management interventions, including guidance for organic fertilizer application, biological control strategies, and crop rotation cycles [25,28,29].

2. Materials and Methods

2.1. Overview of the Study Area

The study area is located in the Xiaodaohe Forest Farm, Boshang Town, Linxiang District, Lincang City, Yunnan Province, China (23°43′16″ N, 100°7′3″ E). This region is characterized by a subtropical mountain monsoon climate with distinct dry and rainy seasons. The mean annual temperature is 14.68 °C (maximum 37.06 °C, minimum 2.53 °C), and the mean annual precipitation is 1254 mm [7]. Although the regional rainy season typically commences in late May to early June, the soil sampling conducted in early May 2022 captured the critical period at the end of the dry season, immediately before the onset of rains, thereby maximizing the representation of dry season soil characteristics. In contrast, the sampling in October 2022 fully represented post-rainy season soil conditions. During the entire 2022 sampling period, the total precipitation was 986 mm, with average temperature and relative humidity of 18.2 °C and 82%, respectively(Figure 1).

The study was conducted at a uniform altitude of 2210 m. The soil is classified as red loam with an acidic pH (4.23–5.65), containing 11%–21.8% organic matter and moderate mineral nutrient content [9,10]. The total experimental area is 2.1 hectares. The primary forest canopy consists of nearly mature Pinus yunnanensis Franch. (stand density 0.7, mean height 17.5 m, mean diameter at breast height 27 cm). The understory shrub layer is dominated by Hypericum monogynum, Schima wallichii, and Vaccinium bracteatum, while the herbaceous layer primarily comprises Isodon amethystoides, Ageratina adenophora, and Oplismenus compositus.

2.2. Soil Sample Collection

The study area is characterized by a subtropical mountain monsoon climate with distinct dry and rainy seasons. To investigate the seasonal dynamics of soil properties, sampling was conducted during the dry season (May 2022) and the rainy season (October 2022), respectively.

Sampling covered three land-use types: natural forestland (control, CK), 2-year-old Panax notoginseng planting forestland (BP), and 3-year-old Panax notoginseng planting forestland (TP). For each land-use type, three independent sampling plots (10 m × 10 m) were established. Within each plot, soil samples were collected from three randomly selected points at a depth of 0–20 cm using a soil auger, with the sample from each point retained as an individual sample without compositing.

According to this design, three individual samples were obtained from each plot of every land-use type per season, resulting in a total of 27 soil samples per season (3 land-use types × 3 plots × 3 individual samples). Consequently, this study collected and analyzed a grand total of 54 individual soil samples (27 from the dry season and 27 from the rainy season).

2.3. Analysis of Soil Physicochemical Properties and Enzyme Activities

Soil water content was determined using the aluminum box method. Soil bulk density, non-capillary porosity, capillary porosity, total porosity, saturated water holding capacity, field water holding capacity, and capillary water holding capacity were measured by the core method [30].

Particle size distribution, including the contents of clay (<0.002 mm), silt (0.002–0.05 mm), very fine sand (0.05–0.1 mm), fine sand (0.1–0.25 mm), medium sand (0.25–0.50 mm), coarse sand (0.50–1.00 mm), and very coarse sand (1.00–2.00 mm), was determined using a laser particle size analyzer. Soil pH was measured potentiometrically using a laboratory pH meter (ST3100 (Ohaus Instruments (Changzhou) Co., Ltd., Changzhou, China). Soil organic matter was determined by the potassium dichromate external heating method [31]. Available phosphorus was extracted with an ammonium fluoride solution and measured by the molybdenum-antimony anti-colorimetric method. Available potassium was extracted with an ammonium acetate solution and determined using a flame photometer. Soil microbial biomass carbon was determined by chloroform fumigation-extraction followed by potassium dichromate oxidation with external heating [32]. Microbial biomass nitrogen was measured using an ultraviolet spectrophotometer. Total nitrogen and total phosphorus were determined by digestion with concentrated sulfuric acid and hydrogen peroxide, followed by analysis using a continuous flow analyzer.

Soil enzyme activities were assessed as follows [33]: urease activity was measured using the sodium phenolate colorimetric method; acid phosphatase activity was determined by the disodium phenyl phosphate colorimetric method; catalase activity was assessed by the potassium permanganate titration method; and sucrase activity was measured using the 3,5-dinitrosalicylic acid colorimetric method [34,35,36].

2.4. Analysis of Soil Microbial Community Structure

To assess the biological dimension of soil health, the structure of soil microbial communities was analyzed. The procedure encompassed two main phases: experimental processing of samples and subsequent bioinformatic analysis of the sequencing data.

2.4.1. DNA Extraction, Amplification and Sequencing

Total microbial genomic DNA was extracted from soil samples using the QJ Magnetic Beads kit according to the manufacturer’s instructions. The quality of the extracted DNA was verified by 1% agarose gel electrophoresis, and the concentration and purity were determined using a NanoDrop2000 spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA).

The hypervariable V3−V4 region of the bacterial 16S rRNA gene was amplified using the primer pair 338F (5′-ACTCCTACGGGAGGCAGCAG-3′) and 806R (5′-GGACTACHVGGGTWTCTAAT-3′) [37]. The fungal ITS1 region was amplified using the primer pair ITS1F (5′-CTTGGTCATTTAGAGGAAGTAA-3′) and ITS2R (5′-GCTGCGTTCTTCATCGATGC-3′) [38]. The resulting PCR amplicons were purified, quantified, and subjected to high-throughput sequencing on an Illumina platform by Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China).

2.4.2. Bioinformatics Analysis

The processing of 16S rRNA and ITS sequencing data followed the bioinformatics pipeline described below. First, raw sequencing data were subjected to quality control using Fastp (v0.19.6, https://github.com/OpenGene/fastp, accessed on 12 December 2024) [39]. Subsequently, paired-end reads were merged using Flash (v1.2.11, https://ccb.jhu.edu/software/FLASH/, accessed on 12 December 2024) [40]. The merged sequences were clustered and chimeras were removed using the Uparse algorithm in USEARCH11 (v11.0.667, http://www.drive5.com/usearch, accessed on 12 December 2024) [41], generating operational taxonomic units (OTUs) at a 97% similarity threshold.

Taxonomic annotation of OTUs was performed using the RDP classifier (v2.13) [42]. Bacterial OTUs were assigned against the Silva 138/16s_bacteria database (https://www.arb-silva.de, accessed on 12 December 2024), and fungal OTUs were assigned against the UNITE 8.0/its_fungi database (https://unite.ut.ee, accessed on 12 December 2024).

Alpha diversity indices, including Chao1, ACE, Shannon, and Simpson, were calculated using mothur (v1.30.2) [43]. Principal coordinate analysis (PCoA) was performed based on Bray–Curtis distances. Similarities in microbial community structure among samples were tested using analysis of similarities (ANOSIM), and significant differences between sample groups were further assessed using PERMANOVA non-parametric testing.

The resulting taxonomic tables and alpha diversity indices were subsequently used for statistical analysis of soil microbial communities.

2.5. Soil Health Assessment

2.5.1. Minimum Dataset Construction

A Minimum Data Set (MDS) was developed through principal component analysis (PCA) combined with Norm value calculation and correlation analysis to identify the most crucial and non-redundant indicators from the total dataset for land productivity evaluation. The specific procedures are as follows.

Principal components with eigenvalues greater than or equal to 1 were extracted according to the Kaiser criterion, representing the primary directions and structure of variation in the soil property data. For each extracted principal component, soil properties with absolute factor loadings greater than or equal to 0.5 were grouped. If an indicator met this criterion on multiple components, it was assigned to the group where it exhibited the lowest correlation (i.e., the largest difference in factor loadings) to maintain independence between components. Indicators with all loadings below 0.5 were assigned to the component where their absolute loading was highest [44].

The Norm value was calculated for each soil property within its assigned group to assess its relative contribution. The formula is:

$$N_{ik}=\sqrt{{\sum }_{j=1}^k\left(U_{ik}^2{\lambda }_k\right)}$$

(1)

In Equation (1), $N_{ik}$ is the combined loading of the ith soil property indicator on the first k principal components with eigenvalues ≥ 1; $U_{ik}$ is the loading of the ith soil property indicator on the kth principal component; and ${\lambda }_k$ is the eigenvalue of the kth principal component.

Candidate indicators were selected using a group-specific dynamic threshold rather than a universal absolute value or simply the highest value alone. Specifically, within each principal component group, the maximum Norm value ($N_{max}$) was identified. All indicators with Norm values greater than or equal to $0.9\times N_{max}$ (i.e., falling within the top 10% range of the highest Norm value) were selected as candidates from that group. This approach aims to capture all highly representative indicators within the group and avoids the potential bias of selecting only the single highest value.

Finally, all candidate indicators from various components were pooled for Pearson correlation analysis. Pairs of candidate indicators with a correlation coefficient (r) less than 0.5 were considered to represent different aspects of soil properties and were retained in the MDS. If the correlation coefficient was 0.5 or greater, the indicators were considered highly redundant, and only the one with the higher Norm value was included in the MDS.

This systematic procedure effectively screens the most critical indicators for land productivity evaluation from a larger dataset, forming the MDS. The method optimizes the evaluation system by reducing data volume and complexity while ensuring scientific soundness and effectiveness [45].

2.5.2. Construction of Soil Health Evaluation Function

Two scoring functions, linear and non-linear, were used to convert each soil health indicator into a dimensionless number from 0 to 1 [46,47,48]. For the linear scoring function, the ‘bigger is better’ indicator in soil was calculated using the following formula:

$$S_{L1}={\displaystyle \frac{x}{x_{max}}}$$

(2)

The following formula is used to calculate the ‘smaller is better’ indicator for soil:

$$S_{L2}={\displaystyle \frac{x_{min}}{x}}$$

(3)

In Equations (2) and (3), $S_L$ is the soil linear scoring function, $x$ is the measured value of the soil health indicator, and $x_{max}$ and $x_{min}$ are the maximum and minimum values of each soil indicator measured, obtained by averaging the top or bottom 5% of all the corresponding measured indicators.

For the nonlinear scoring function, the formula is as follows:

$$S_{NL}={\displaystyle \frac{a}{1+({x/x_{\mu })}^b}}$$

(4)

where $S_{NL}$ soil nonlinear scoring function, between 0 and 1; $a$ is the highest score reached by the scoring function, which is equal to 1 in this study; $x_{\mu }$ is the average value of each corresponding indicator; $b$ is the slope of the equation, and the scoring function of the indicator that meets the ‘bigger is better’ has a value of −2.5, and the scoring function that meets the ‘smaller is better’ has a value of 2.5 [47,48], and $x$ is the same as the above equation.

2.5.3. Calculation of the Soil Health Index

A weighted composite method was used to calculate the soil health index (SHI) and the weights of each indicator were calculated using principal component analysis [44].

$$SHI={\sum }_{i=1}^n\left(W_i+S_i\right)$$

(5)

where $S_i$ is the score of each indicator using a linear or non-linear scoring function, $n$ is the number of indicators used to calculate the soil health index, and $W_i$ is the corresponding weight value for each soil indicator.

2.6. Statistical Analysis of Data

Data organization was performed using Microsoft Excel 2016. Statistical analyses were conducted using SPSS Statistics (v24.0, IBM Corp., Armonk, NY, USA). One-way analysis of variance (ANOVA) was employed to assess differences between treatment groups, followed by LSD post hoc tests for multiple comparisons when significant differences were detected (p < 0.05). Prior to principal component analysis, the suitability of the dataset for factor analysis was evaluated using the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity. The dataset satisfied the prerequisite checks with a KMO value exceeding 0.6 and Bartlett’s test yielding a statistically significant result (p < 0.05), confirming adequate correlations among variables for subsequent factor analysis [49,50,51]. All figures were prepared using Origin (v2021, OriginLab Corp., Northampton, MA, USA).

2.7. Suitability Test for Multivariate Analysis

In order to comprehensively assess land productivity and construct a land productivity rating system, it is necessary to analyze soil property indicators using statistical methods to ensure that the selected indicators are suitable for multivariate analyses such as factor analysis. In this process, the Kaiser-Meyer-Olkin (KMO) test and Bartlett’s test of sphericity represent two pivotal pre-testing steps employed to evaluate the appropriateness of a dataset for factor analysis [49]. The KMO test quantifies the proportion of variance among variables that might be common variance, thereby assessing the suitability of the dataset for factor analysis. The KMO statistic ranges from 0 to 1. A value closer to 1 indicates smaller partial correlations among the variables relative to the total correlations, suggesting that the data are more suitable for factor analysis. Typically, a KMO value greater than 0.6 is considered moderately to highly suitable. Bartlett’s test of sphericity is used to test the null hypothesis that the variables in the dataset are uncorrelated. If the test is significant (i.e., the p-value is less than a predetermined significance level, commonly set at 0.05), the null hypothesis is rejected, indicating that the variables are correlated to some extent and thus suitable for factor analysis. The test essentially assesses whether the correlation matrix is an identity matrix, which would imply no correlation among the variables [52].

3. Results

3.1. Construction and Testing of a Full Dataset of Panax notoginseng in the Forest

To assess the suitability of the data for factor analysis, a preliminary screening was conducted on the initially selected 35 soil indicators. The examination revealed that the correlation matrix of the initial indicators was non-positive definite, and six indicators (natural water content, medium sand, coarse sand, very coarse sand, available potassium, and total phosphorus) exhibited low communalities. To ensure the validity of the analysis, these six indicators were excluded.

Subsequently, the Kaiser-Meyer-Olkin (KMO) measure of sampling adequacy and Bartlett’s test of sphericity were formally performed on the remaining 29 indicators. The results (

Table 1. KMO and Bartlett’s test.

Bartlett’s Test
KMO	Approximate Chi-Square	Degrees of Freedom	Significance
0.666	1324.508	406	0.000

) showed a KMO statistic of 0.666 (>0.6), and Bartlett’s test of sphericity was significant at the 0.001 level, rejecting the null hypothesis that the variables are uncorrelated. These outcomes collectively confirm adequate common variance among the retained 29 indicators, demonstrating the dataset’s suitability for factor analysis.

3.2. Descriptive Statistics of Soil Health Indicators for Understory Panax notoginseng

The descriptive statistics of the 29 measured soil health indicators are presented in

Table 2. Descriptive Statistics of Soil Health Indicators for Understory Panax notoginseng.

Indicator	Soil Health Indicators	Minimum	Maximum	Mean	Standard Deviation	Coefficient of Variation/%
Physical indicators	Total Porosity/%	69.82	78.00	74.61	2.36	3.2%
	Saturated Water Holding Capacity/%	17.22	23.94	20.99	1.92	9.2%
	Capillary Water Holding Capacity/%	10.56	14.94	13.08	1.30	9.9%
	Field Capacity/%	4.29	8.41	6.49	1.08	16.7%
	Capillary Porosity/%	2.61	4.72	4.04	0.64	15.8%
	Non-capillary Porosity/%	65.37	75.27	70.57	2.37	3.4%
	Bulk Density/mg m−3	0.51	0.73	0.63	0.05	8.6%
	Clay Content/%	3.23	12.99	7.91	2.69	34.0%
	Silt Content/%	53.88	87.11	71.95	7.66	10.7%
	Very Fine Sand Content/%	1.70	8.42	5.13	1.85	36.2%
	Fine Sand Content/%	0.00	4.68	0.78	1.00	128.1%
Chemical indicators	PH/%	4.23	5.65	5.07	0.32	6.4%
	Soil Organic Matter/(mg·kg−1)	11.06	21.81	15.48	3.03	19.6%
	Microbial Biomass Carbon/(mg·kg−1)	0.12	1.58	0.77	0.46	59.8%
	Microbial Biomass Nitrogen/(mg·kg−1)	0.05	0.13	0.08	0.03	32.3%
	Total Nitrogen/(mg·kg)	2.05	6.49	4.37	1.10	25.1%
	Sucrase Activity/(mg·kg)	0.75	5.31	2.28	1.20	52.8%
	Acid Phosphatase Activity/(g·kg)	0.53	1.69	1.03	0.38	37.0%
	Available Phosphorus/(mg·kg−1)	9.37	31.54	20.33	6.27	30.8%
	Urease Activity/(mg·kg)	0.12	1.82	0.84	0.43	50.6%
	Catalase Activity/(μg·kg−1)	9.03	12.07	10.57	0.78	7.3%
Biological indicators	Bacterial Chao1 index	1715.48	3453.06	2570.87	420.09	16.3%
	Fungal Chao1 index	348.33	1411.08	947.41	240.51	25.4%
	Bacterial ACE index	1705.76	3442.75	2600.69	426.54	16.4%
	Fungal ACE index	344.68	1450.40	995.10	263.05	26.4%
	Bacterial Shannon index	4.92	6.10	5.64	0.29	5.1%
	Fungal Shannon index	2.94	4.58	3.82	0.43	11.3%
	Bacterial Simpson index	0.01	0.03	0.02	0.01	34.2%
	Fungal Simpson index	0.02	0.18	0.07	0.03	41.4%

. Overall, all indicators exhibited a certain degree of variation among samples, with coefficients of variation (CV) ranging from 3.2% to 128.1%, reflecting the spatial heterogeneity of soil properties within the study area.

Among the soil physical indicators, structural properties such as Total Porosity (mean 74.61%), Non-capillary Porosity (mean 70.57%), and Bulk Density (mean 0.63 mg m⁻³) showed relatively low variation (CV < 10%), indicating a more uniform distribution. In contrast, texture composition indicators, particularly Fine Sand Content (mean 0.78%, CV = 128.1%) and Clay Content (mean 7.91%, CV = 34.0%), exhibited high variability, suggesting considerable spatial differences in soil texture.

Soil chemical and biochemical indicators generally demonstrated higher variability than physical indicators. Soil Organic Matter (mean 15.48 mg·kg⁻¹) and Total Nitrogen (mean 4.37 mg·kg⁻¹) contents were at moderate levels. Among soil enzyme activities, Sucrase Activity (CV = 52.8%) and Urease Activity (CV = 50.6%) showed high coefficients of variation. Notably, Microbial Biomass Carbon (mean 0.77 mg·kg⁻¹, CV = 59.8%) was one of the most variable indicators measured in this study, revealing a high degree of spatial heterogeneity in soil microbial biomass.

Soil microbial community diversity indices revealed that the richness indices of bacterial communities (e.g., Bacterial Chao1 index, mean 2570.87) were higher than those of fungal communities (e.g., Fungal Chao1 index, mean 947.41). The mean Bacterial Shannon index (5.64) was also higher than the Fungal Shannon index (3.82), indicating greater species diversity in bacterial communities. In terms of variability, the Fungal Simpson index (CV = 41.4%) varied more than the Bacterial Simpson index (CV = 34.2%), while the coefficients of variation for bacterial and fungal ACE and Chao1 indices were similar (approximately 16%–26%), suggesting comparable patterns of richness between the two community types.

In summary, the descriptive statistics delineate the basic ranges and variation characteristics of soil indicators in the understory P. notoginseng system. Indicators such as Fine Sand Content, Microbial Biomass Carbon, and certain enzyme activities demonstrated exceptionally high spatial variability, potentially representing key drivers of differences in soil health.

3.3. Minimum Data Set Establishment for Soil Health Evaluation

3.3.1. Principal Component Analysis and Selection of Candidate Indicators

Factor analysis (principal component analysis, PCA) was performed on the 29 selected soil property indicators to eliminate information redundancy and reduce the number of indicators for evaluating land productivity, thereby constructing the Minimum Data Set (MDS).

Table 3. Principal Component Loading Matrix and Norm Values for Soil Properties Indicators for the total data Set (TDS).

Indicators of Soil Properties	Post-Rotation Factor Loadings							Clusters	Norm	Variance of a Common Factor
	1	2	3	4	5	6	7
Bulk density	−0.001	0.123	−0.011	0.181	0.203	−0.847	−0.084	6	1.192945754	0.813
Total porosity	0.329	0.110	0.133	−0.162	0.824	−0.178	0.219	5	1.530009473	0.923
Capillary porosity	0.829	0.354	−0.042	−0.108	0.244	−0.161	0.091	1	2.567159318	0.921
Non-capillary porosity	0.478	0.064	0.169	0.000	0.693	−0.132	0.152	5	1.702397325	0.781
Saturated water holding capacity	0.380	0.443	−0.575	−0.015	0.009	0.270	−0.257	3	1.955820026	0.810
Capillary water holding capacity	0.837	0.328	−0.037	−0.030	−0.130	0.140	−0.108	1	2.549130242	0.859
Field Capacity	0.826	0.107	0.083	−0.204	0.233	0.009	−0.154	1	2.444988813	0.819
Clay content	0.460	0.292	0.273	−0.483	0.300	0.070	−0.132	1	1.797171955	0.717
Silt content	−0.096	0.235	0.112	0.288	−0.204	0.678	0.231	6	1.13997348	0.714
Very Fine Sand Content	0.006	0.261	−0.137	0.798	−0.061	0.031	0.236	4	1.386406533	0.784
Fine Sand Content	0.266	−0.050	0.275	0.307	0.107	−0.746	−0.069	4	1.350260399	0.816
pH	0.269	0.412	−0.564	0.116	−0.003	0.333	−0.065	2	1.741748821	0.689
Soil Organic Matter	0.210	−0.052	0.909	0.052	0.182	−0.101	0.098	3	1.983138784	0.928
Microbial Biomass Carbon	−0.057	−0.107	−0.133	−0.539	0.262	0.295	−0.556	7	1.180359536	0.788
Microbial Biomass Nitrogen	−0.245	−0.222	0.023	−0.198	0.292	0.429	−0.597	7	1.284072075	0.776
Total nitrogen	0.009	0.086	0.863	−0.100	0.147	0.055	−0.115	3	1.800131052	0.799
Available phosphorus	−0.034	0.364	−0.044	−0.134	0.451	0.392	−0.509	5	1.270095788	0.770
Urease Activity	−0.605	0.031	0.436	0.310	−0.249	0.055	0.183	3	2.052415906	0.752
Sucrase Activity	−0.579	−0.133	0.727	0.160	0.026	0.158	0.073	3	2.279872248	0.938
Acid Phosphatase Activity	0.840	0.383	0.182	0.043	0.120	−0.131	−0.100	1	2.623063376	0.929
Catalase Activity	0.224	−0.033	0.869	0.239	−0.094	0.005	0.103	3	1.934685493	0.883
Bacterial Chao1 Index	0.312	0.814	0.010	0.199	−0.210	0.095	−0.286	2	2.167069447	0.934
Fungal Chao1 Index	0.185	0.848	−0.107	−0.286	0.209	0.135	0.058	2	2.130345807	0.911
Bacterial ACE index	0.349	0.780	−0.017	0.237	−0.156	0.134	−0.241	2	2.14433525	0.888
Fungal ACE index	0.173	0.917	−0.111	−0.140	0.138	0.134	−0.008	2	2.233135243	0.939
Bacterial Simpson Index	−0.150	−0.137	0.212	−0.037	0.335	0.087	0.782	7	1.180094673	0.819
Fungal Simpson Index	−0.161	−0.067	0.172	0.824	−0.091	0.089	−0.219	4	1.391290448	0.803
Bacterial Shannon Index	0.318	0.595	0.021	0.141	−0.242	−0.063	−0.599	2	1.829233859	0.898
Fungal Shannon Index	0.308	0.339	−0.273	−0.727	0.152	0.028	0.108	2	1.715936425	0.849
Eigenvalue	8.328	5.464	4.195	2.197	1.687	1.209	1.166
variance contribution	28.718	18.718	14.467	7.575	5.819	4.168	4.020
Cumulative Variance Contribution Rate (%)	28.718	47.561	62.028	69.603	75.422	79.589	83.609

shows that the cumulative variance contribution rate of the principal components with eigenvalues ≥ 1 was 88.92%, indicating a highly effective analysis suitable for information extraction.

In PC1, soil properties with absolute factor loadings ≥0.4 included capillary porosity, capillary water holding capacity, field water holding capacity, clay content, and acid phosphatase. Among these, acid phosphatase had the highest Norm value (2.623). Soil properties within the 10% range of the highest Norm value were capillary porosity, capillary water holding capacity, field water holding capacity, and acid phosphatase.

In PC2, soil properties with absolute factor loadings ≥0.4 included saturated water holding capacity, pH, bacterial Chao1 index, fungal Chao1 index, bacterial ACE index, fungal ACE index, bacterial Shannon index, and fungal Shannon index. The fungal ACE index exhibited the highest Norm value (2.233). Soil properties within the 10% range of the highest Norm value were the bacterial Chao1 index, fungal Chao1 index, bacterial ACE index, and fungal ACE index.

Similarly, in PC3, soil properties with absolute factor loadings ≥0.4 included organic matter, total nitrogen, urease, sucrase, and catalase. Sucrase possessed the highest Norm value (2.280). Consequently, sucrase, organic matter, and total nitrogen, falling within the 10% range of the highest Norm value, were selected as candidate indicators.

In PC4, soil properties with absolute factor loadings ≥0.4 included very fine sand, fine sand, and fungal Simpson index. The fungal Simpson index had the highest Norm value (1.391). Soil properties within the 10% range of the highest Norm value were very fine sand, fine sand, and the fungal Simpson index.

In PC5, soil properties with absolute factor loadings ≥0.4 included total porosity, non-capillary porosity, and available phosphorus. Non-capillary porosity demonstrated the highest Norm value (1.702) and was the sole property within the 10% range of this value, leading to its selection.

In PC6, soil properties with absolute factor loadings ≥0.4 included silt content and bulk density. Bulk density exhibited the highest Norm value (1.193). Both silt content and bulk density were within the 10% range of the highest Norm value.

In PC7, soil properties with absolute factor loadings ≥0.4 included microbial biomass carbon, microbial biomass nitrogen, and the bacterial Simpson index. Microbial biomass nitrogen showed the highest Norm value (1.284). Soil properties within the 10% range of the highest Norm value were microbial biomass carbon, microbial biomass nitrogen, and the bacterial Simpson index.

Through this process, a total of 18 candidate indicators were preliminarily selected from the seven principal components.

3.3.2. Correlation Analysis and Finalization of the Minimum Data Set

To eliminate information redundancy and ensure the independence of indicators in the MDS, Pearson correlation analysis was conducted on all 18 candidate indicators (Figure 2). According to the established criterion (r ≥ 0.5 indicates high redundancy), pairs of highly correlated indicators were further screened.

The correlation matrix revealed strong relationships among several candidate indicators. For instance, microbial biomass carbon and microbial biomass nitrogen exhibited a significant positive correlation, indicating substantial overlap in representing soil microbial biomass. Therefore, only the indicator with the higher Norm value was retained. Similarly, strong correlations were observed between bacterial Chao1 and bacterial ACE indices, as well as between fungal Chao1 and fungal ACE indices. In each pair, only one representative indicator (fungal ACE index due to its higher Norm value) was included in the MDS to avoid redundancy. Other notable correlations, such as between total porosity and non-capillary porosity and between sucrase and soil organic matter, further justified the necessity of this step.

After applying the correlation filter, the final MDS comprises the following 11 key indicators: acid phosphatase, fungal ACE index, organic matter, total nitrogen, sucrase, fungal Simpson index, fine sand, non-capillary porosity, silt content, bulk density, and microbial biomass nitrogen.

3.4. Soil Health Assessment Based on Minimum and Total Datasets

By performing principal component analysis for the full dataset containing 29 indicators and the minimum dataset filtered to contain 11 indicators, the common factor variance and weights of each soil property indicator were obtained as shown in

Table 4. Common factor variance and weights of land productivity evaluation indicators.

Indicators of Soil Properties	TDS	TDS	MDS	MDS
	Variance of a Common Factor	Weights	Variance of a Common Factor	Weights
Non-capillary Porosity	0.781	0.032	0.703	0.083
Soil Organic Matter	0.928	0.038	0.909	0.108
Microbial Biomass Nitrogen	0.776	0.032	0.716	0.085
Sucrase Activity	0.938	0.039	0.909	0.108
Acid Phosphatase Activity	0.929	0.038	0.912	0.108
Bulk Density	0.813	0.034	0.648	0.077
Silt Content	0.714	0.029	0.677	0.080
Fine Sand Content	0.816	0.034	0.844	0.100
Total Nitrogen	0.799	0.033	0.803	0.095
Fungal ACE Index	0.939	0.039	0.713	0.085
Fungal Simpson Index	0.803	0.033	0.598	0.071
Total Porosity	0.923	0.038
Capillary Porosity	0.921	0.038
Saturated Water Holding Capacity	0.810	0.033
Capillary Water Holding Capacity	0.859	0.035
Field Capacity	0.819	0.034
Aggregate	0.717	0.030
Very Fine Sand Content	0.784	0.032
pH	0.689	0.028
Microbial Biomass Carbon	0.788	0.033
Available Phosphorus	0.770	0.032
Urease Activity	0.752	0.031
Catalase Activity	0.883	0.036
Bacterial Chao1 Index	0.934	0.039
Fungal Chao1 Index	0.911	0.038
Bacterial ACE Index	0.888	0.037
Bacterial Simpson Index	0.819	0.034
Bacterial Shannon Index	0.898	0.037
Fungal Shannon Index	0.849	0.035

.

The Soil Health Index (SHI) was calculated using both linear and non-linear scoring functions based on the Total Data Set (TDS, 29 indicators) and the Minimum Data Set (MDS, 11 indicators). The SHI values derived from the MDS showed a significant positive correlation with those from the TDS, validating the representativeness of the selected MDS indicators.

The overall soil health in the study area was at a moderate level. Using the linear scoring function, the mean SHI was 0.62 (range: 0.53–0.72) for the MDS and 0.68 (range: 0.61–0.76) for the TDS. The non-linear function yielded more conservative estimates, with mean SHI values of 0.51 (range: 0.48–0.59) and 0.50 (range: 0.44–0.59) for the MDS and TDS, respectively (Figure 3). The coefficient of variation for SHI was consistently larger when calculated from the MDS compared to the TDS, indicating that the simplified dataset captured a wider range of soil health variability.

Significant differences in SHI were observed across different planting durations and seasons (Figure 4). Regardless of the dataset or scoring function used, a consistent trend emerged: two-year-old plantations exhibited higher SHI values than three-year-old plantations. This decline in soil health with extended cultivation was particularly pronounced during the rainy season, where three-year-old plots recorded the lowest SHI values across all treatments. Conversely, soil health indices were generally higher during the dry season compared to the rainy season. The natural forest control (CK) consistently maintained the highest or among the highest SHI values, underscoring the impact of P. notoginseng cultivation on soil properties.

As shown in Figure 4a,c, In the full dataset, the soil health index of Panax notoginseng under forestry conditions in the rainy season exhibited a slight increase following planting, reaching a higher level than that of the CK group after planting. Concurrently, the soil health index of Panax notoginseng after two years was higher than that of Panax notoginseng after three years. The Land Health Index demonstrated higher values than those observed for the rainy season as a whole during the dry season. The index for forested land devoid of cultivation exhibited the highest values, which corresponded to the results obtained during the rainy season. This analysis was conducted for the purpose of comparing two-year-old and three-year-old Panax notoginseng. As demonstrated in Figure 4b,d, the trend of the soil health index in the minimum dataset exhibited a higher value during the dry season compared to the rainy season. However, analysis of the soil health index within the minimum dataset demonstrated a decreasing trend with increasing planting year. Furthermore, the health index of natural forest land without Panax notoginseng planting was found to exceed that of the planted land.

3.5. Validation of Soil Health Evaluation Method for Understory Panax notoginseng Based on Minimum Data Set

The results of linear regression analyses of the soil health indices calculated for the full and minimum data sets showed a significant positive correlation between the soil health indices of the full and minimum data sets for both scoring functions (Figure 5a,b). Also, there was a significant positive correlation between the two linear and non-linear methods (Figure 5c,d). It can be seen that the minimum dataset can be used instead of the full dataset for evaluating the effect of understory Panax notoginseng cultivation on soil health in this study.

4. Discussion

4.1. Construction of the Minimum Dataset for Soil Health Evaluation of Understory Panax notoginseng and Analysis of the Role of Key Indicators

The minimum dataset for soil health evaluation of understory Panax notoginseng established in this study incorporates eleven key indicators: acid phosphatase, fungal ACE index, organic matter, total nitrogen, sucrase, fungal Simpson index, fine sand content, non-capillary porosity, silt content, bulk density, and microbial biomass nitrogen. This dataset was constructed through principal component analysis. Soil health indices were calculated using both linear and nonlinear scoring functions based on the total dataset and the minimum dataset, respectively. Linear regression analysis confirmed significant positive correlations among the four resulting indices, thereby validating the effectiveness of the minimum dataset approach. This finding is consistent with results from similar studies conducted under different cropping systems and management regimes, demonstrating the feasibility and robustness of integrating the minimum dataset framework with scoring functions for assessing soil health in this cultivation system [45,53,54].

Soil organic matter serves as the cornerstone of the constructed minimum dataset, a common feature in related studies, likely due to its direct and multifaceted role in regulating soil biological activity and overall productivity [55]. As evidenced in the literature, organic matter is a critical reservoir of nutrients essential for maintaining soil structure and promoting the retention of water and nutrients [56,57]. A higher organic matter content is typically associated with optimized soil structure, enhanced water retention capacity, and increased nutrient availability. Acid phosphatase, a ubiquitous soil enzyme, is another vital component. Its activity level is directly linked to the bioavailability of phosphorus in soil, serving as a key indicator for assessing soil biological activity and phosphorus availability. The critical role of acid phosphatase lies in facilitating the timely cycling of phosphorus, a key nutrient for plant growth. Sucrase, a key soil enzyme, was also selected [58]. Its activity level indirectly reflects soil fertility status and the transformation efficiency of organic matter, which is important for understanding the soil carbon cycle and fertility levels [59]. Related research indicates that practices such as straw returning can increase sucrase content and activity. Elevated sucrase activity signifies a healthy and active microbial community, which is crucial for maintaining soil structure and driving nutrient cycling processes. Microbial biomass nitrogen, as a highly active component of the soil nitrogen pool, directly reflects microbial biomass and the level of soil biological activity within the soil micro-ecosystem [60]. It acts as a sensitive quality indicator for the soil nitrogen pool and holds significant importance for evaluating soil organic matter dynamics and overall soil health [61]. Relevant studies note that microbial biomass nitrogen is highly susceptible to changes in organic carbon and seasonal turnover, being inherently less stable and sensitive to environmental disturbances [62,63]. Therefore, monitoring microbial biomass nitrogen can facilitate timely adjustments to soil nutrient management strategies. Non-capillary porosity, defined as the percentage of soil volume occupied by pores with a diameter greater than 0.1 mm, was identified as a key indicator of soil physical condition and a critical factor affecting soil ecological functions and overall productivity [64]. Improving non-capillary porosity through optimized management is a primary mechanism for enhancing soil aeration and drainage, thereby supporting soil health.

Crucially, the indicators within the minimum dataset do not operate in isolation but are interconnected through a network of biogeochemical processes, forming synergistic or antagonistic relationships that collectively determine soil health status. For instance, soil organic matter acts as a fundamental driver, significantly influencing microbial biomass nitrogen and sucrase activity. This is because soil organic matter provides the essential carbon and energy sources necessary to stimulate microbial proliferation and enhance their metabolic activity. However, changes in microbial activity and community composition, in turn, regulate the decomposition and stabilization of organic matter. Furthermore, selected microbial indicators such as the fungal ACE index and fungal Simpson index reflect the structural state of the microbial community. Shifts in this structure, potentially driven by root exudates over consecutive cultivation years, can influence the functional output of the microbiome, including the activity of enzymes like acid phosphatase, as different microbial taxa contribute variably to phosphorus cycling. Simultaneously, soil physical properties impose critical constraints. Deterioration of physical structure, indicated by increased bulk density and decreased non-capillary porosity, restricts oxygen diffusion and root exploration. This physically degraded environment can suppress the very microbial and enzymatic activities that depend on adequate aeration, creating a negative feedback loop where poor physical conditions inhibit biological function, which may further exacerbate physical degradation. The inclusion of fine sand and silt content addresses textural influences on water and nutrient holding capacity, which modulate the habitat and resource availability for soil biota [65].

Therefore, the constructed minimum dataset captures a series of continuously interacting indicators. The calculated Soil Health Index integrates these physical, chemical, and biological indicators. Within the minimum dataset, indicators such as organic matter, acid phosphatase, sucrase, microbial biomass nitrogen, and non-capillary porosity work in concert through a series of interconnected processes to maintain and promote soil health. However, due to the inherent limitations posed by local environmental conditions and management practices in the study area, the applicability of this specific dataset to other regions of Chinese herbal medicine cultivation requires further verification. Nevertheless, in subsequent studies, priority can be given to measuring these categories of indicators when conducting soil health assessments in other regions. They comprehensively reflect the biochemical and physical conditions of soils and depict the multidimensional characteristics of soil health, which is imperative for evaluating land health, ecological functions, and guiding targeted land management interventions.

4.2. Impact of Forest Panax notoginseng Cultivation on Soil Health

This study analyzed variations in soil health across different cultivation years (two-year-old vs. three-year-old plantations) and seasons (rainy vs. dry). The findings indicated that understory P. notoginseng cultivation negatively affected soil health indices. The most pronounced degradation was observed in three-year-old plantations during the rainy season, which exhibited the lowest Soil Health Index values.

Compared to three-year-old plantations, two-year-old P. notoginseng exhibited a higher soil health index. This phenomenon may be attributed to the accelerated growth stage of two-year-old plants, which enhances their efficiency in utilizing soil resources such as light, water, and nutrients. Secondly, the shorter growth cycle may reduce the accumulation and spread of soil-borne pests and diseases, thereby mitigating negative impacts on plant growth and soil biology. Furthermore, a more developed but less densely packed root system in biennial plants might result in less intense competition for nutrients and water within the rhizosphere, contributing to a more favorable soil health status. Critically, extending the cultivation period to three years appears to trigger a cascade of detrimental interactions. Continuous root exudation and monotrophic nutrient uptake likely drive a shift in the soil microbial community from bacterial to fungal dominance, as reflected in changes to key fungal indices (e.g., fungal ACE and Simpson). This “fungalization” is often associated with continuous cropping obstacles. Concurrently, intensive root growth and management activities may exacerbate soil compaction, leading to increased bulk density and decreased non-capillary porosity—two strongly correlated indicators in our MDS. This deterioration of soil physical structure further impedes water infiltration and gas exchange, fostering a more anaerobic environment. Such conditions can suppress the activity of aerobe-associated enzymes like sucrase and acid phosphatase, establishing a negative feedback loop where declining biological activity and worsening physical structure reinforce each other, ultimately manifesting in the significantly lower Soil Health Index observed in the three-year-old plots [66].

The calculated soil health index was higher during the dry season than the rainy season. This discrepancy may initially be attributed to improved plant water-use efficiency under water stress. Studies have shown significant seasonal variations in water-use efficiency among species [67]. Vegetation adapts its water uptake strategy seasonally, absorbing deeper soil water during the dry season and utilizing near-surface water during the rainy season, aligning with variations in soil moisture and root distribution. Plants may alter root system architecture or adjust physiological mechanisms to adapt to water-stressed environments, strategies that can help maintain land productivity [68]. However, the lower SHI in the rainy season is likely the result of synergistic pressures from leaching losses and altered biological activity. Heavy rainfall first leads to the loss of available nutrients (e.g., nitrogen, phosphorus) via surface runoff and deep percolation, directly reducing chemical fertility indicators like total nitroge [69]. Secondly, prolonged soil saturation and potential transient anaerobic conditions alter microbial metabolic pathways, potentially suppressing the activity of aerobic enzymes like acid phosphatase while shifting the fungal-to-bacterial ratio (reflected in changes in fungal community indices) [70]. Notably, while higher soil moisture in the rainy season improves physical indicators like field capacity, excess water fills soil pores—particularly non-capillary pores—thereby reducing soil aeration. This dual stress from deteriorating physical conditions and changed chemical environment collectively suppresses soil biological activity, resulting in an overall lower multi-dimensional Soil Health Index during the rainy season compared to the dry season [71]. The slightly higher SHI observed in P. notoginseng plantations compared to natural forestland during the rainy season may be attributed to the canopy cover effect of P. notoginseng, which reduces the direct impact of raindrops, minimizes soil crusting, and may enhance infiltration rates, partially mitigating the adverse effects of heavy rainfall [72].

5. Conclusions

This study developed a Minimum Data Set (MDS) incorporating microbial community indicators to systematically evaluate soil health in the understory Panax notoginseng cultivation system. Based on the findings, the following specific management strategies and intervention thresholds are proposed:

• Control of Planting Duration: Given the significant decline in the Soil Health Index (SHI) observed after the third year of cultivation, it is recommended that the monoculture period be strictly limited to two years. Subsequently, crop rotation with nitrogen-fixing plants or non-medicinal crops should be implemented to disrupt the development of continuous cropping obstacles.
• Seasonal Precision Management: In response to the consistent pattern of superior soil health during the dry season compared to the rainy season, soil conservation measures should be enhanced prior to the onset of the rainy season. Specific recommendations include using mulches (e.g., straw or plastic film) to reduce nutrient leaching and applying organic amendments to improve soil buffering capacity and resilience.
• Microbial Community Regulation: The inclusion of key microbial indicators in the MDS highlights that soil biological degradation is a core driver of health decline. In practice, the application of microbial inoculants or functional organic fertilizers can be employed to directionally modulate the soil microbiome and counteract the functional degradation associated with deleterious shifts in fungal community structure.
• Establishment of an SHI-based Early Warning and Intervention Mechanism: The SHI values obtained using linear and non-linear scoring functions ranged from 0.53 to 0.72 (mean 0.62) and 0.48–0.59 (mean 0.51), respectively, establishing the baseline soil health status for the region. Based on the more conservative non-linear scoring results, it is proposed to set an SHI value of 0.50 as the early warning threshold, triggering intensified monitoring and management optimization when values fall below this level. An SHI value of 0.48 should be established as the mandatory intervention threshold, necessitating practices such as crop rotation, fallowing, or comprehensive soil remediation when breached.

This research translates a theoretical evaluation framework into a tiered management plan featuring concrete operations, timing schedules, and quantitative thresholds. It provides a directly executable decision-making tool for the precision management and sustainable development of understory P. notoginseng cultivation.