This study was conducted in the Qilian Mountains (Qifeng, Sidalong, and Xishui Forest Farms), Sunan Yugur Autonomous County (38°49′18.7″N 99°36′09.8″E), Zhangyi city, Gansu Province, northwest China (Figure 1). The elevation ranges from 2,450 to 3,300 m. The climate is arid and semi-arid, with mean annual precipitation of 410–540 mm, which increases with rising elevation (Wang et al., 2003). Most precipitation (86%) falls from May to September (Zhang et al., 2011). The mean annual air temperature varies from −0.3 to 8.1°C and decreases with rising elevation (Wang et al., 2009). The annual accumulated frost-free period is 127 days. About 60% of the daily photoperiodic is full sunshine, and the yearly pan evaporation is 1,785 mm (Yuan and Hou, 2015). The primary soil parent materials are slope deposits and residuals from limestone, shale, and conglomerate. The degree of downwards minerals transition was mostly indistinct/vague and obvious. The compactness was usually moderate, slightly loose, while less soils were compact and firmly loose. The soil structure was frequently massive or bulky and granular. The soil texture was commonly loam, sandy loam, and clay loam, while soil moisture was mostly moist/damp, but some soils were dry in the PF and NF.

Most of the forests belong to the natural forests of Qinghai spruce (Picea crassifolia), which grow on the shady and semi-shady slopes within the elevation range of 1,750–3,550 m (but concentrated within 2,200–3,200 m), and also the natural forests of Qilian juniper (Sabina przewalskii), which grow on the dry and sunny slopes within the elevation range of 2,600–4,000 m (Rong et al., 2019). At the same time, the adjacent PF was also dominated by Qinghai spruce and Qilian junipers. Both PF and NF belong to the middle-aged groups. There are rich ground mosses on the floor of spruce forests due to their unique habitat. The dominant shrubs and herbs species were Berberis diaphana, Caragana jubata, Salix sylvestris, Sabina procumbens, Dasiphora fruticose, Carex oshimensis, Potentilla fruticose, Polygonum viviparum, and Agropyron cristatum under both forest types. The average tree density of the surveyed plots was 1,163 and 1,332 trees ha⁻¹ for PF and NF, respectively. The average tree height was 5.8 and 7.82 m, and tree DBH was 16.7 and 24.7 cm for PF and NF, respectively. Similarly, the average height of dominant trees was 8.77 and 12.5 m, and the canopy density was 42% and 39% for PF and NF, respectively (Table 1).

For both PF and NF, 46 sample plots were selected to investigate the physicochemical soil properties. The plot size was 20 m × 20 m, and a pit of 1 m² was dug out at the forest floor and replicated three times for each soil profile. Each soil profile was classified into O, A, AB, B, BC, C, and other horizons, and a total of 230 soil samples (115 of each forest type) were collected. For physical analysis (soil bulk density, BD), the soil samples were collected at each soil depth using stainless-steel augers and oven-dried at 104°C for 48 h. The soil samples for chemical analysis were collected from the soil profiles using a multi-point mixed-method at the soil depths of 0–20, 20–40, 40–60, 60–80, and 80–100 cm. The litter, leaves, roots, insects, and larger residues were separated and removed. All soil samples were screened through a 2-mm sieve mesh. The soil samples were air-dried under the ceiling fan and stored at room temperature to determine physicochemical soil parameters. The stored soil samples were then analyzed with standard methods to determine the pH value, soil organic carbon (SOC), total N (TN), and total P (TP) contents and transformed into stocks (SOCs, TNs, TPs) and soil stoichiometry (C:N:P ratios) under the soil depth (SD) and forest types (FT).

The soil organic carbon content was determined by the K₂Cr₂O₇–H₂SO₄ digestion method (Walkley, 1947). The content of total soil P was measured by dissolved nitric acid (NH₃), perchloric acid digestion (HClO₄), and hydrofluoric acid (HF) using an automated discontinuous analyzer. The content of total soil N was measured with a fully automatic Kjeldahl analyzer (dissolved by H₂SO₄ along with a catalyst). The soil pH value was measured with a pH-meter at 1:5 soil water suspension (Mclean, 1982).

The SOC stock at each soil depth was calculated as follows (Hu et al., 2018): S O C s ( M g   h a - 1 ) = S O C   c o n t e n t s   ( g   k g − 1 ) · B D ( g   c m − 3 ) · S T ( c m ) / 10 .(1)

Similarly, the soil stocks of total N and P (TNs and TPs) were calculated using the above equation. Here, ST denotes the soil layer thickness (cm).

The data analysis was performed using one-way ANOVA (Analysis of Variance) to determine the impacts of FT and SD, respectively, on the physicochemical soil properties. Two-factor ANOVAs were conducted to investigate the impact of SD, FT, and their interactions on soil properties. The Tukey HSD all-pairwise comparison (least significant difference) test at p < 0.05 was used to compare the means. Origin Lab-2019 was used to evaluate the correlation between physicochemical soil parameters and perform the PCA (principal component analysis) for each forest type.

The soil BD and pH values differed among soil depth and forest types significantly (p ≤ 0.05), showing an increase with rising soil depth for both forest types (Figure 2A). The mean BD of PF and NF increased from the lowest values of 0.833 and 0.813 at the topsoil depth (0–20 cm) to the highest values of 1.37 and 1.10 g cm⁻³ at the deepest soil depth (80–100 cm). Similarly, the pH values increased from 8.16 and 7.80 at the topsoil depth for PF and NF to 8.41 and 8.23 at the deepest soil depth (Figure 2B). The BD at all soil depths of PF was higher than those of NF, and this difference increased with rising soil depth, from 0.02 g cm⁻³ at the topsoil depth (0–20 cm) to 0.27 g cm⁻³ (based on the data pair) at the soil depth of 80–100 cm (Figure 2A).

Similarly, the soil pH values at all soil depths of PF were higher than those of NF, but this difference decreased with rising soil depth, from 0.36 at the topsoil depth to 0.18 (based on the data pair) at the deepest soil depth (Figure 2B). The mean soil BD of 0–100 cm soil depth was 1.03 g cm⁻³ for PF, higher than the corresponding value of 0.97 g cm⁻³ for NF. Similarly, the mean soil pH value of 8.30 for 0–100 cm soil depth for PF was higher than the pH value of 8.10 for NF (Table 2).

The contents of SOC, TN, and TP differed among soil depths (SD) and forest types (FT) significantly (p ≤ 0.05), showing a decrease with rising soil depth for both PF and NF. The SOC and TN contents at all soil depths of PF were lower than those of NF (Figures 2C, D). The difference in SOC content decreased from 12.7 g kg⁻¹ (based on the data pair of 49.9 and 62.6 g kg⁻¹) at the topsoil depth to 12.3 g kg⁻¹ (based on the data pair of 19.1 and 31.4 g kg⁻¹) at the deepest soil depth (Figure 2C). The mean SOC content of 0–100 cm soil depth for NF (44.3 g kg⁻¹) was significantly (p ≤ 0.05) higher than that of PF (32.0 g kg⁻¹) (Table 2). The difference in TN content decreased from 1.05 g kg⁻¹ (based on the data pair of 3.53 and 4.58 g kg⁻¹) at the topsoil depth to 1.00 g kg⁻¹ (based on the data pair of 1.18 and 2.18 g kg⁻¹) at the deepest soil depth (Figure 2D). The mean TN content of 0–100 cm soil depth for NF (3.07 g kg⁻¹) was significantly higher (p ≤ 0.05) than that of PF (2.15 g kg⁻¹) (Table 2).

The mean TP contents of PF and NF generally decreased from the highest values of 0.72 in both PF and NF at the topsoil depth to the lowest values of 0.55 and 0.66 at the deepest soil depth (Figure 2E). However, the variation pattern of TP content along soil depth differed from the continuously decreasing pattern of SOC and TN. For NF, the TP content declined non-significantly and leveled-off within the soil depth of 0–100 cm. For PF, the TP content decreased insignificantly within 0–40 cm and obviously within the depth of 60–100 cm; however, a significant difference existed only between 0–20 cm and 80–100 cm. The mean TP content of 0–100 cm for NF (0.67 g kg⁻¹) was significantly higher than those of PF (0.63 g kg⁻¹) (Table 2). At each soil depth, the percent contribution of overall average contents of SOC, TN, and TP was presented in Figure 4, showing a decrease in soil depth for PF and NF.

For both PF and NF, stocks of SOC and TN decreased significantly (p ≤ 0.05) in each soil depth, while that of TP increased (Figures 3A–C; Table 5). The highest SOC stocks, 82.6 and 96.8 Mg ha⁻¹ for PF and NF, were at the topsoil depth (0–20 cm), and the lowest values, 48.4 and 66.7 Mg ha⁻¹, were at the deepest soil depth (80–100 cm). Similarly, the maximum TN stocks of 5.83 and 7.19 Mg ha⁻¹ for PF and NF were at the topsoil depth, while the lowest values of 3.02 and 4.73 Mg ha⁻¹ were at the deepest soil depth. The TP stock was highest (1.47 Mg ha⁻¹) for both PF and NF at the deepest soil depth, while the lowest values (1.19 Mg ha⁻¹) for both PF and NF were at the topsoil depth. The SOC and TN stocks of NF were higher than PF at all soil depths. The SOC and TN stock differences between PF and NF increased with soil depth, from 14.2 and 1.36 Mg ha⁻¹ in the topsoil depth to 18.3 and 1.71 Mg ha⁻¹ (based on the data pair) at the deepest soil depth (Figures 3A, B). The average SOC stocks, 64.3 Mg ha⁻¹ and 80.6 Mg ha⁻¹ for PF and NF at soil depth (0–100 cm), showed that NF was significantly higher (p ≤ 0.05) (Table 2). Similarly, the average TN stocks, 4.32 Mg ha⁻¹ and 5.72 Mg ha⁻¹ for PF and NF at soil depth (0–100 cm), showed NF was significantly higher (p ≤ 0.05), while no significant difference existed for the TP stocks 1.35 Mg ha⁻¹ and 1.33 Mg ha⁻¹ for PF and NF at soil depth (0–100 cm) (Table 2). The percent contribution of SOC and TN stocks of each soil depth to their total stocks over 0–100 cm soil depth decreased with rising soil depth for both PF and NF, but increased in the case of TP stock (Figures 4B, E).

The soil C:P and N:P ratios of PF and NF decreased significantly (p ≤ 0.05) with soil depth. The C:P ratios decreased from 73.3 and 89.5 for PF and NF at topsoil depth to 34.7 and 47.7 at the deepest soil depth. The N:P ratios decreased from 5.05 and 6.45 for PF and NF at the topsoil depth to 2.13 and 3.34 at the deepest soil depth (Figures 3E, F). The C:N ratios were similar among all soil depths for both PF and NF, although the variation range for PF was bigger than that for NF (Figure 3D).

The C:P and N:P ratios for NF were significantly higher than those for PF at all soil depths. The differences in C:P and N:P ratios decreased with rising soil depth, from 16.2 to 13.0 for C:P ratios and from 1.40 to 1.21 for N:P ratios, respectively (based on the data pairs in Figures 3E, F). The difference in the C:N ratio between PF and NF increased and then decreased with rising soil depth but was non-significant at all soil depths (Figure 3D). The average C:N ratios were 16.7 and 14.5 for PF and NF at soil depth (0–100 cm), where PF was significantly higher, while the C:P and N:P ratios for PF (51.5 and 4.38) were significantly lower than those for NF (66.1 and 5.85) (Table 2). The percent contributions of each soil depth to the overall stoichiometry of C:P and N:P of soil depth 0–100 cm were calculated for PF (Figure 4C) and NF (Figure 4F), showing decreasing percent contribution of each soil property with rising soil depth for both PF and NF, while the C:N ratios in PF showed an increase up to 40–60 cm and thereafter decreased, and were rather similar in NF.

The soil depth and forest type intimately affected the physicochemical soil properties, while their interaction showed non-significant effects except for BD and TP content (Table 3). The soil BD, pH value, and the contents of SOC and TN were highly significantly correlated with both FT and SD. The TP content was positively associated with SD (p ≤ 0.001) and significantly correlated with FT (p ≤ 0.05). The stocks of SOC and TN were highly significantly correlated with both FT and SD (p ≤ 0.001), but the TP stock was only positively associated with SD (p ≤ 0.001) and non-significantly correlated with FT. The C:N ratio was somewhat significantly correlated with FT (p ≤ 0.05) but insignificantly correlated with SD. The C:P and N:P ratios were highly correlated (p ≤ 0.001) with FT and SD. Overall, most physicochemical soil properties were highly affected by SD and FT (except TP stock and C:N ratio).

The PC1 and PC2 define physiochemical soil parameters for PF and NF (Figures 5A, C). The sum of PC1 and PC2 explains 80.3% and 80.6% variations in physiochemical soil properties of PF and NF. The cumulative contribution rate of PC1 (59.7%) mainly reflects the soil BD, SOC, and TN for PF, while the contribution rate (56.6%) for NF mainly reflects the pH values, SOC, and TN. Similarly, the cumulative contribution rate of PC2 (20.6%) reflects the significant contribution of pH values for PF, while the contribution rate (24.0%) for NF defined the significant variability in soil BD and TP. The score plot for soil depth in each forest type was also developed (Figures 5B, D). The physicochemical soil properties under each soil depth occupied different sections of the diagrams. The soil depths (60–80 and 80–100 cm) were close for PF and NF and were in the analogous ordination space, resulting in the least significant differences. As a result, the correlation coefficient of soil parameters at these depths should be high. The soil properties occupying different positions in the diagram, such as the opposite location, reveal significant variation and quantitative changes between the soil properties. The soil depths (0–20 and 20–40 cm) for PF and NF were also lying closely, indicating minimal differences, though the overall differences for a soil depth of 0–20 cm were higher than those at 20–40.

The Pearson’s correlation analysis revealed a strong correlation among the physicochemical soil properties under PF and NF at 0–100 cm soil depth (Figure 6). The SOC was positively correlated with TN, and TP, while negatively correlated with pH (r = 0.78, 0.41, and −0.30, respectively) under PF (Figure 6A). Similarly, the SOC was positively correlated with TN and slightly correlated with TP, while a negative correlation was observed with pH for NF (r = 0.83, 0.17, and -0.66, respectively) (Figure 6B). The TN was positively correlated with TP (r = 0.54), while a negative correlation was found with pH for PF (r = −0.04). For NF, the TN was also positively correlated with TP (r = 0.30), and a negative correlation has existed with soil pH values (r = −0.61). The TP has negatively correlated with soil pH and BD for PF (r = −0.04, −0.51), while only negatively correlated with pH for NF (r = −0.16). The soil BD was strongly negatively correlated with SOC, and TN for PF and NF, while positively correlated with soil pH values (r = −0.78, −0.75, −0.64, −0.45, 0.28, and 0.30, respectively).

The soil BD was significantly affected by soil depth and forest type (Figure 2A; Tables 3, 4). The increase of BD with rising soil depth is an integrated result of changing soil texture (Landsberg, 2003; Ali et al., 2019) and decreasing soil organic matter input from the humus layer and tree root litter due to the limited gaseous exchange (Tokunaga, 2006). However, the higher BD in PF than NF might be attributed to the poor growth and associated lower stand biomass, but unlikely due to mechanical compaction on the exposed surface soil by management activities (such as heavy harvest and human trampling) because the forest management for timber production has been prohibited in the natural reservation area for more than 4 decades. Of course, the natural forests were excluded from commercial use in this natural reservation area (Liao et al., 2012).

The increased pH value with soil depth could be associated with less input of humus material (with organic acids) into the deeper soil depths (Table 4). A similar study (Balstrøm et al., 2013) reported that the organic acids present in the topsoil tend to break down before they reach the deeper soil depths, leading to higher pH values in deeper soil depths. The observed lower soil pH value under NF than PF confirmed the results of other studies (Chen et al., 2004). Thus, it may also be explained by the higher SOC content (SOM, correspondingly higher organic acid production) in NF than PF and the much longer time of forest cover and humus decomposition (Chen et al., 2016) in NF than in PF, which ultimately reduced the pH values of soil in a natural forest. Moreover, the better forest stand structure of NF than PF is also an influencing factor (Kafle, 2019).

Soil depth is an important factor influencing the contents of SOC, TN, and TP (Berger et al., 2002; Olorunfemi et al., 2018). Our study showed a decreasing tendency in SOC, TN, and TP contents (Yang and Liu, 2019); this could be explained by the decreasing inputs of SOM with rising soil depth (Table 4) from both above and below-ground litterfall (Ngaba et al., 2019), especially by the decreasing influence of the humus layer covering the soil surface (Sperfeld et al., 2017; Zhang et al., 2018; Xu et al., 2019; Yang and Liu, 2019) and the decreasing activities of soil animals and microorganisms (Barreto et al., 2011). With rising soil depth, the soil permeability decreases, and the BD increases, limiting the input of dissolved SOM through leaching (Berger et al., 2002; Clemmensen et al., 2013; Rahman et al., 2022).

The lower contents of SOC, TN, and TP in PF compared with NF (Table 4) could be attributed to the lower accumulation of litterfall and the faster mineralization of SOM due to the more open stand structure, which leads to a higher soil temperature (Liao et al., 2010; Zhang et al., 2016; Wang et al., 2018). Furthermore, the harvesting of biomass and logging residues during site preparations for afforestation, the forest managing measures (e.g., weeding and thinning), and the quicker absorption of nutrients of trees in the fast-growing stage of a plantation may also contribute to the lower contents of SOC, TN, and TP in the soils of PF than NF (Paul et al., 2002; Jobbágy and Jackson, 2003; Berthrong et al., 2009; Liao et al., 2012).

The stocks of SOC and TN significantly decreased with soil depth irrespective of the forest types of PF or NF (Wang et al., 2016; Li et al., 2019; Ngaba et al., 2020) except for the TP stocks (Table 5). Our study showed that SOC and TN stocks firstly declined significantly with rising soil depth within the range of 0–60 cm and then stabilized or leveled-off thereafter within the soil depth range of 60–100 cm (Gao et al., 2020). The most important reason might be the highest biological activity in the topsoil, characterized by the highest SOM content and associated nutrient accumulation (Ngaba et al., 2020). Higher SOC and TN stocks in NF than PF in this study were comparable with previous studies (Gong et al., 2011; Wang et al., 2018), thought to be due to the more significant inputs of soil nutrients to the mineral soils in NF from their higher accumulated humus and litterfall compared with PF (Gong et al., 2011; Zhang et al., 2016). The TP stock likely increased with soil depth due to deep phosphorous parent materials or compounds down the soil profile (Xu et al., 2019). The similar stocks of TP (for PF and NF) in this study (Table 2) are in line with other studies (Hu et al., 2018), which indicated that forest conversion increases the TP stocks in the middle and deep soil depths (20–100 cm). This result is possibly due to the higher P input into the deeper soil depth because of the accelerated humus mineralization and leaching and the reduced P uptake by the young trees in the earlier growth stage of plantations (Xu et al., 2018).

The differences in C:P and N:P ratios among soil depths were significant in this study, as reported in many studies (Liu et al., 2017; Xu et al., 2019; Liu et al., 2021), while it showed a non-significant difference in the C:N ratios among soil depths (Figure 3D; Table 6). A study in the Loess Plateau of China showed a similar result that the C:N ratios remained unchanged with soil depth (Deng et al., 2013). The C:N ratios of forest soils are negatively associated with the mineralization rate of SOM (Stevenson and Cole, 1999; Hu et al., 2019). Therefore, the lower C:N ratios in NF might be explained by the mineralization stage of SOM in the undisturbed NF. The C:P and N:P ratios could be used as indicators of the saturation or limitation of P and N (Sterner and Elser, 2002) and to assess the structure, health, and function conditions of forests that influence the quality and quantity of litter/humus as well as their mineralization (Wang et al., 2014; Yang et al., 2018). The observed variations of C:N:P ratios with soil depth and forest type in our study are consistent with other investigations (Gong et al., 2011; Zhang et al., 2016; Luo et al., 2019). Based on our results, the soil C:P and N:P ratios for NF were higher than those for PF, but the soil C:N ratios for NF were lower than that of PF, which could be associated with the element content changes, and litter decomposition rate (Yu et al., 2010; Zhang et al., 2016; Wang et al., 2018; Yang et al., 2018). Thus, we concluded that the NF present higher soil C:P and N:P ratios but lower C:N ratios than the PF with the same age group, tree species composition, and similar site conditions.