Spatial Distribution of Soil Organic Carbon and Nitrogen Across Salinity Gradients in the Yellow River Delta, China

Abstract

Severe soil salinization and low fertility significantly constrain sustainable agricultural development in the Yellow River Delta, one of the three major estuarine deltas in China. Despite their ecological importance, the regional-scale spatial interactions between soil salinity and nutrients, particularly regarding their vertical variability, remain poorly understood. This study analyzed 228 soil samples from 76 sites distributed across a distinct salinity gradient, which was determined by constructing a spatial salinity distribution map after sampling. Samples were collected at three depths (0–15, 15–30, and 30–45 cm) to investigate the spatial distribution of soil organic carbon (SOC), total nitrogen (TN), and the C/N ratio, along with their underlying driving factors. SOC and TN exhibited similar spatial patterns, with higher values distributed along both banks of the Yellow River. Horizontally, SOC and TN in the 0–15 cm layer decreased gradually from west to east, whereas the 15–30 cm and 30–45 cm layers showed an opposite trend, increasing eastward. Vertically, SOC and TN contents declined significantly with soil depth (p < 0.05), although the magnitude of this decline varied regionally: the 0–15 cm layer in the western area contained markedly higher nutrient levels than deeper layers, while vertical variation was less pronounced in the eastern and estuarine regions. Both variables were positively associated with total phosphorus (TP), available potassium (AK), soil moisture content (MC), clay content, and pH, but negatively correlated with electrical conductivity (EC), particularly in the 0–15 cm layer. Our results highlight that soil texture, moisture, and salinity affect the spatial heterogeneity and vertical decline of SOC and TN in the Yellow River Delta. Future research should focus on the long-term temporal distribution of the coupling of multiple elements under changing hydrological and salinity regimes.

1. Introduction

The Yellow River Delta, one of China’s major estuarine deltas, covers approximately 8243 km², with farmland accounting for about 52% of this area [1,2]. Formed by intensive sediment deposition, the delta has experienced distinct developmental stages driven by frequent river channel shifts and massive upstream sediment delivery, creating a clear soil chronosequence from older soils in the south to younger soils in the north and east [3]. Concurrently, interactions between hydroclimatic stressors (shallow groundwater, seawater intrusion) and human activities (agricultural expansion, saline irrigation, intensive groundwater extraction) have made this region one of the most severely affected by soil salinization in China [4,5,6,7,8].

Increasing soil salt concentrations significantly suppress nutrient cycling processes and constrain carbon and nitrogen transformations. Excessive accumulation of salts induces swelling and dispersion of soil particles and disrupts aggregate structure, which ultimately reduces the physical protection of soil organic carbon (SOC), accelerating its decomposition and loss [9]. Increased salinity (to 2.0‰) can decrease SOC in cotton fields by ~17.5% [10]. Beyond direct physical disruption of soil structure, salinity can indirectly affect soil nutrient cycling via plant and microbial pathways [11]. For instance, soil salinization suppresses crop growth and reduces biomass and yield, thereby diminishing soil organic matter inputs and constraining soil carbon sequestration. Conversely, salinity reduction enhances plant-derived carbon inputs and sequestration [12,13]. In addition, diminished microbial biomass constrains the availability of precursors required for stable SOC formation, resulting in a net decline in SOC [14]. High salinity suppresses nitrification processes [15], thereby hindering nitrogen transformation and limiting total nitrogen (TN) availability. The C/N ratio, as a soil fertility indicator closely related to SOC and TN, is regulated by a combination of climatic, hydrological, edaphic, vegetative, and agronomic factors [16,17].

Most existing studies in the Yellow River Delta have focused on surface soils or plot scales, leaving the spatial interactions between soil salinity and nutrient accumulation across larger regional scales insufficiently understood. For instance, some studies examined the spatial distribution of TN and available potassium (AK), but those analyses were restricted to the 0–15 cm layer [18]. Similarly, investigations of the spatial distribution of SOC were also limited to the top layer (0–20 cm) [19,20], thereby overlooking the variations across the soil profile. Previous studies have explored the effects of groundwater level and salinity gradients on SOC and TN accumulation in the estuary; however, these investigations were geographically limited and could not capture regional variability [17,21]. Additionally, some work has examined both horizontal and vertical patterns of soil nutrients, but it did not explicitly address spatial differentiation in vertical variation across the delta [21,22]. Although the spatial heterogeneity and vertical distribution of soil physicochemical properties have been investigated in a coastal reclamation area in Yancheng [23], the corresponding patterns in the Yellow River Delta remain poorly understood. Moreover, although land-use types are widely recognized as critical determinants of soil nutrient accumulation, their specific regulatory effects on nutrient vertical variation distribution under contrasting salinity gradients remain to be clarified.

To address the above knowledge gaps, we investigated the spatial (sampling interval <10 km) and vertical (0–45 cm) distributions of SOC, TN, and the C/N ratio across the Yellow River Delta and identified the key environmental factors controlling these patterns. Specifically, the objectives are to: (1) characterize the horizontal and vertical variation distribution of SOC, TN, and the C/N ratio at different soil depths; (2) examine the relationships between soil nutrients and physicochemical properties; and (3) evaluate the influence of land use and environmental factors on nutrient accumulation in saline soils.

2. Materials and Methods

2.1. Study Area

This study is located in the alluvial plain of the middle and lower reaches of the Yellow River, specifically in Shandong Province, China, which features a warm-temperate continental monsoon climate. The average annual temperature is approximately 12.1 °C, and the average annual precipitation is about 551.6 mm [24]. The altitude ranges from −6 m to 110 m. The region has been shaped by the natural evolution of the Yellow River, characterized by periodic sedimentation, frequent channel shifts, and avulsion at the estuary [19,25]. The region features a complex mosaic of hills, gentle slopes, and depressions. The main crops are winter wheat, corn, cotton, rice, and soybeans [26,27,28]. The main soil types in this region are coastal saline fluvo-aquic soil and solonchaks [25]. Silt loam is the dominant soil texture in the area (Figure S1). Seasonal alternations drive the repeated upward movement and leaching of salts, resulting in the widespread occurrence of soil salinization [29].

2.2. Soil Sampling and Analysis

In April 2025, soil samples were collected from 76 sites across the study area using a grid point distribution method adjusted for field accessibility, with an inter-site spacing of 3–10 km (Figure 1). Following the classification criteria for the Yellow River Delta proposed by Ye et al. [30], these sites were categorized by their formation periods, i.e., Ancient (12 sites), Modern (39 sites), and Contemporary (25 sites), as well as their land-use types, including dry land (68 sites), paddy land (5 sites), orchard land (2 sites), and forest land (1 site).

Figure 1. Location of the study area and distribution of sampling sites across different land use types.

At each sampling point, undisturbed soil samples were collected using cutting rings (50 mm in diameter, 50 mm in height, 100 cm³ in volume) from three soil layers: 0–15 cm, 15–30 cm, and 30–45 cm. Soil bulk density was determined by the oven-drying method [31].

At each sampling location, the soil profile was stratified into three layers with 15 cm increments. From each layer, three replicate subsamples were collected from the lateral margins and the central position (The distance between the lateral margins and the center position is 50 cm.), pooled in self-sealing polyethylene bags, and thoroughly mixed to form a composite sample, resulting in a total of 228 samples. After air-drying, the samples were placed in sealed plastic bags and stored at 4 °C to inhibit microbial activity. Prior to analysis, the sample was first coarsely sieved using a 10-mm sieve, then passed through a 2-mm stainless steel sieve for further sieving to remove plant debris and stones. SOC was determined using the dichromate oxidation method [32]. Carbonates were removed by adding 2 mol/L HCl dropwise until no further bubbling was observed. After centrifugation and removal of the supernatant, the residue was washed repeatedly with deionized water (3–5 times) until neutral, then dried at 60–80 °C and weighed. Chloride interference was masked by the addition of silver sulfate during digestion. A correction factor of 1.04 was used to adjust the oxidation results. TN was measured by the Kjeldahl method [33]. Total phosphorus (TP) was determined using the molybdenum-antimony colorimetric method [34]. AK was measured with a flame photometer. Available phosphorus (AP) was extracted with NaHCO₃ and then analyzed using the Mo-Sb colorimetric method with a spectrophotometer (Shimadzu UV2550, Kyoto, Japan) [35]. Soil particle size distribution was determined by laser diffraction using a Mastersizer 2000 (Malvern Instruments, Malvern, UK). Soil pH and electrical conductivity (EC) were measured in a 1:5 soil-to-water (Mass ratio) suspension using a pH meter and the four-electrode method, respectively [36]. MC was determined using the oven-drying method (105 °C).

Vertical variation distribution was calculated by ΔEC, ΔSOC, ΔTN, and ΔC/N, which denote the changes in EC, SOC, TN, and C/N ratios between different soil layers, respectively. The calculation formulas are as follows:

$$∆\mathit{EC}={(\mathit{EC}}_D-{\mathit{EC}}_u)/{\mathit{EC}}_u\times 100$$

(1)

$$∆\mathit{SOC}=({\mathit{SOC}}_D-{\mathit{SOC}}_u)/{\mathit{SOC}}_u\times 100$$

(2)

$$∆\mathit{TN}={(\mathit{TN}}_D-{\mathit{TN}}_u)/{\mathit{TN}}_u\times 100$$

(3)

$$∆C/N={(C/N}_D-{C/N}_u)/{C/N}_u\times 100$$

(4)

where EC_D and EC_u refer to the EC values in the deeper and upper soil layers, respectively. Given the three soil layers in this study, three ΔEC values were computed: ΔEC between layers of 0–15 cm and 15–30 cm, 15–30 cm and 30–45 cm, and 0–15 cm and 30–45 cm. ΔSOC, ΔTN, and ΔC/N were calculated in the same way.

2.3. Statistical Analysis

Statistical analyses were carried out using SPSS version 27.0.1 and R software (version 4.6.1). The coefficient of variation (CV) was classified according to the criteria of Nielsen et al. [37]: CV ≤ 10% indicates low variability, 10–100% indicates moderate variability, and ≥ 100% indicates high variability. Pearson correlation coefficients were calculated to determine the correlations among soil variables. To compare soil physicochemical properties among the three depth intervals (0–15, 15–30, and 30–45 cm), one-way analysis of variance (ANOVA) was performed, followed by Tukey’s honestly significant difference (HSD) test at the 0.05 significance level. Spatial interpolations of EC and nutrient contents were performed using Empirical Bayesian Kriging (EBK) within ArcGIS Pro (version 3.0.2) (WGS 1984 geographic coordinate system; no data transformation; power semivariogram model). The cross-validation statistics for the EBK procedure are provided in Table S1.

3. Results

3.1. Spatial Distribution of SOC, TN, and C/N Ratio

3.1.1. Descriptive Statistics

Descriptive statistics for EC, pH, SOC, TN, and the C/N ratio across the three soil layers are summarized in Table 1. EC has CV values exceeding 100% in all layers, indicating spatial heterogeneity and increasing with depth from 112.83% in the 0–15 cm layer to 130.80% in the 30–45 cm layer. Mean EC values decreased slightly with depth, from 454 ± 513 μS cm⁻¹ in the 0–15 cm layer to 364 ± 476 μS cm⁻¹ at 30–45 cm. In contrast, soil pH exhibited very low variability (CV = 3.51–4.32%) and increased slightly with depth, ranging from 8.19 ± 0.29 to 8.38 ± 0.36.

Table 1. Statistical summary of soil salinity and nutrients in the study region.

Variable	Depth (cm)	Max	Min	Mean	Median	SD	CV (%)	p
EC (μS cm−1)	0–15	2728	103	454 a	415	513	112.83	0.478 1
	15–30	2611	90	374 a	361	461	123.35	0.497 2
	30–45	2308	66	364 a	352	476	130.80	0.999 3
pH	0–15	8.71	7.39	8.19 a	8.23	0.29	3.51	0.087 1
	15–30	9.27	7.19	8.30 a	8.30	0.35	4.17	0.001 2
	30–45	9.48	7.45	8.38 b	8.37	0.36	4.32	0.338 3
SOC (g kg−1)	0–15	15.93	2.61	7.96 a	8.39	2.93	36.82	0.001 1
	15–30	13.74	1.29	5.64 b	6.26	2.45	43.44	0.001 2
	30–45	9.49	1.06	3.21 c	3.00	2.04	63.56	0.001 3
TN (g kg−1)	0–15	1.50	0.24	0.82 a	0.82	0.29	35.70	0.001 1
	15–30	1.28	0.17	0.57 b	0.60	0.23	40.84	0.001 2
	30–45	1.19	0.12	0.34 c	0.31	0.22	66.68	0.001 3
C/N	0–15	13.65	7.95	9.73 a	9.41	1.22	12.58	0.888 1
	15–30	13.99	7.56	9.81 a	9.60	1.34	13.66	0.610 2
	30–45	13.07	6.92	9.53 a	9.46	1.21	12.69	0.336 3

Notes: SOC: soil organic carbon; TN: total nitrogen; EC: electrical conductivity. SD, standard deviation; CV, coefficient of variance. Lowercase letters (^a, ^b, ^c) denote significant differences (p < 0.05) between layers for each nutrient. The sample size was 76 in each soil layer. Superscripts ¹, ², and ³ indicate significant differences between soil layer pairs: ¹ for 0–15 vs. 15–30 cm, ² for 15–30 vs. 30–45 cm, and ³ for 0–15 vs. 30–45 cm.

SOC and TN displayed moderate to high variability, and both declined significantly with increasing soil depth (p < 0.05). SOC decreased from 7.96 ± 2.93 g kg⁻¹ in the 0–15 cm layer to 3.21 ± 2.04 g kg⁻¹ at 30–45 cm, while TN decreased from 0.82 ± 0.29 g kg⁻¹ to 0.34 ± 0.22 g kg⁻¹. Their spatial variability also increased with depth, as indicated by the rising CV values. In contrast, the soil C/N ratio remained relatively stable across the soil profile, with moderate variability (CV = 12.49–13.54%) and mean values ranging from 9.61 to 9.89. One-way ANOVA indicated significant differences in SOC and TN among soil layers (p < 0.05), whereas the C/N ratio did not differ significantly with depth.

3.1.2. Horizontal Spatial Patterns

The spatial distributions of EC, SOC, and TN across the three soil layers are shown in Figure 2. EC exhibited a clear west–east gradient in the 0–15 cm layer, with lower values in the western region (200–300 μS cm⁻¹) and markedly higher values in the eastern coastal areas (700–1400 μS cm⁻¹). In deeper layers (15–30 cm and 30–45 cm), high EC values were primarily concentrated along both banks of the Yellow River and in coastal areas, indicating stronger salt accumulation in these regions.

Figure 2. Spatial distributions of soil EC (a–c), SOC (d–f), and TN (g–i) across different depths. Panels (a,d,g), (b,e,h), and (c,f,i) represent soil depths of 0–15, 15–30, and 30–45 cm, respectively.

SOC and TN displayed similar spatial patterns across the study area. In the 0–15 cm layer, higher nutrient concentrations were mainly distributed along both banks of the Yellow River, with values gradually decreasing from west to east. Western regions generally exhibited SOC values of 10–12 g kg⁻¹ and TN values of 1.00–1.20 g kg⁻¹, whereas eastern areas showed lower concentrations. In the 15–30 cm layer, high SOC and TN contents remained concentrated along the Yellow River corridor and near the estuary. In the 30–45 cm layer, nutrient contents decreased overall but relatively higher values were still observed in the western region and around the estuary.

Spatially, the distribution of SOC closely mirrored that of TN across all layers, indicating a strong coupling between SOC and TN in the region. In contrast, the spatial distribution of the C/N ratio showed relatively weak variability. Although some regional differences were observed, particularly near the estuary and coastal areas, the C/N ratio across the entire study area remained within a relatively narrow range of 8.5–12.5 for all soil layers (Figure S2).

3.1.3. Vertical Variation

The spatial patterns of vertical variation in EC, SOC, and TN were presented in Figure 3. EC exhibited clear depth-related changes, particularly in coastal regions. Pronounced vertical differences between the 0–15 cm and 15–30 cm layers were mainly observed near the Yellow River estuary, whereas inland areas showed relatively small changes with depth. It is worth noting that in the central part of the study area, the salinity in the 15–30 cm layer was even higher than that in the 0–15 cm layer. In most regions, EC in the 30–45 cm layer exceeded that in the 15–30 cm layer, indicating salt accumulation in deeper soil layers.

Figure 3. Vertical variation in EC (a–c), SOC (d–f), and TN (g–i) across different soil layers. Panels (a,d,g), (b,e,h), and (c,f,i) represent soil layers between 0–15 cm and 15–30 cm, 15–30 cm and 30–45 cm, and 0–15 cm and 30–45 cm, respectively.

Although descriptive statistics indicated an overall decrease in SOC and TN with depth, their vertical variation spatial patterns varied regionally. The difference in SOC between the 15–30 cm layer and the 0–15 cm layer decreased from west to east. In the western region, the SOC content in the 15–30 cm layer was 25% to 50% lower than that in the 0–15 cm layer, whereas in the eastern and coastal areas, the difference in SOC content between the two layers became smaller. In some localized areas near the estuary, the SOC content in the 15–30 cm layer even exceeded that in the 0–15 cm layer. For the 30–45 cm layer, SOC content was consistently lower than that in the overlying layers across most of the study area. However, the difference in SOC between the 30–45 cm layer and the 0–15 cm layer gradually decreased from west to east, indicating weaker vertical variation in coastal regions. The vertical variation distribution patterns of TN were largely consistent with those of SOC.

Compared with SOC and TN, the C/N ratio showed relatively minor vertical variation (Figure S2). Although localized increases or decreases were observed in some regions, the overall differences among soil layers were small.

3.2. Correlation Analysis of SOC, TN, and C/N Ratio

3.2.1. Factors Controlling Horizontal Spatial Patterns

Correlation analysis between soil nutrients and physicochemical properties is presented in Table 2. SOC and TN were strongly and positively correlated with each other across all soil layers (p < 0.01), indicating close coupling between SOC and TN. In the 0–15 cm layer, neither SOC nor TN showed significant correlations with the C/N ratio, whereas in deeper layers (15–30 cm and 30–45 cm) SOC and TN exhibited positive relationships with C/N.

Table 2. Pearson correlation coefficients of soil SOC, TN, and C/N ratios across soil layers in the study region.

	0–15 cm			15–30 cm			30–45 cm
	SOC	TN	C/N	SOC	TN	C/N	SOC	TN	C/N
SOC (g kg−1)	1	0.931 **	0.185	1	0.934 **	0.337 **	1	0.968 **	−0.057
TN (g kg−1)	0.931 **	1	−0.166	0.934 **	1	0.007	0.968 **	1	−0.264 *
C/N	0.185	−0.166	1	0.337 **	0.007	1	−0.057	−0.264 *	1
TP (g kg−1)	0.655 **	0.681 **	−0.061	0.448 **	0.537 **	−0.119	0.524 **	0.610 **	−0.231 *
AK (mg kg−1)	0.468 **	0.456 **	0.041	0.589 **	0.565 **	0.152	0.729 **	0.731 **	−0.132
AP (mg kg−1)	0.273 *	0.290 *	−0.029	0.372 **	0.411 **	0.009	0.392 **	0.485 **	−0.182
MC (%)	0.314 **	0.303 **	0.097	0.303 **	0.287 *	0.090	0.295 **	0.278 *	−0.006
pH	0.241 *	0.137	0.218	−0.203	−0.241 *	0.022	−0.233 *	−0.253 *	0.054
EC (μS cm−1)	−0.341 **	−0.359 **	0.129	0.138	0.099	0.128	0.162	0.140	0.031
Sand (%)	0.030	0.002	0.072	−0.040	−0.070	0.068	−0.317 **	−0.326 **	0.118
Silty (%)	−0.167	−0.145	−0.070	−0.272 *	−0.269 *	−0.085	0.052	0.061	−0.070
Clay (%)	0.205	0.223	−0.028	0.487 **	0.540 **	0.004	0.700 **	0.703 **	−0.158
BD (g cm−3)	−0.213	−0.185	−0.001	−0.090	−0.046	−0.170	−0.308 **	−0.283 *	0.008

Notes: SOC: soil organic carbon; TN: soil total nitrogen; TP: soil total phosphorus; AK: soil available potassium; AP: soil available phosphorus; MC: soil moisture content; EC: electrical conductivity; BD: soil bulk density. * Correlation is significant at p = 0.05 level; ** correlation is significant at p = 0.01 level.

SOC and TN showed significant positive correlations with TP, AK, and AP across most soil layers. MC was also positively correlated with SOC and TN, with particularly strong relationships in the 0–15 cm layer. By contrast, EC was negatively correlated with SOC and TN only in the 0–15 cm layer; in deeper layers, the correlations were positive but not significant. Unlike EC, pH exhibited a positive correlation with SOC and TN in the 0–15 cm layer, whereas negative correlations were observed in deeper layers.

The influence of soil texture became more pronounced with increasing depth. In the 0–15 cm layer, soil texture showed no significant relationships with SOC or TN. However, in deeper layers (15–30 cm and 30–45 cm), clay content was strongly positively correlated with SOC and TN, while sand content showed significant negative correlations. These results suggest that finer soil textures may favor the accumulation of SOC and TN in deeper soil layers.

3.2.2. Factors Influencing Vertical Variation

The relationships between vertical variation in soil nutrients and other physicochemical factors are summarized in Table 3. Across all depth intervals, changes in SOC and TN were highly positively correlated with each other (p < 0.01), indicating consistent vertical responses of SOC and TN.

Table 3. Pearson correlation coefficients of ΔSOC, ΔTN, and ΔC/N ratios across soil layers in the study region.

	15–30 vs. 0–15 cm			30–45 vs. 15–30 cm			30–45 vs. 0–15 cm
	ΔSOC	ΔTN	ΔC/N	ΔSOC	ΔTN	ΔC/N	ΔSOC	ΔTN	ΔC/N
ΔSOC (g kg−1)	1	0.952 **	0.318 **	1	0.977 **	0.212	1	0.977 **	0.018
ΔTN (g kg−1)	0.952 **	1	0.031	0.977 **	1	0.028	0.977 **	1	−0.167
ΔC/N	0.318 **	0.031	1	0.212	0.028	1	0.018	−0.167	1
ΔTP (g kg−1)	0.701 **	0.674 **	0.225	0.570 **	0.588 **	0.115	0.616 **	0.638 **	−0.031
ΔAK (mg kg−1)	0.747 **	0.793 **	0.048	0.740 **	0.766 **	0.033	0.733 **	0.756 **	−0.131
ΔAP (mg kg−1)	0.683 **	0.700 **	0.089	0.885 **	0.849 **	0.114	0.750 **	0.785 **	−0.127
ΔMC (%)	0.041	0.050	−0.009	−0.046	−0.052	−0.001	0.065	0.073	−0.117
ΔpH	−0.219	−0.219	−0.039	−0.301 **	−0.304 **	−0.042	−0.301 **	−0.293 *	0.013
ΔEC (μS cm−1)	0.566 **	0.647 **	−0.098	0.313 **	0.326 **	0.051	0.588 **	0.601 **	−0.069
ΔSand (%)	−0.095	−0.088	−0.060	−0.084	−0.058	−0.124	−0.205	−0.167	−0.118
ΔSilty (%)	−0.028	−0.004	−0.024	0.023	0.037	−0.085	0.112	0.086	0.104
ΔClay (%)	0.496 **	0.460 **	0.247 *	0.552 **	0.558 **	0.069	0.545 **	0.513 **	0.026
ΔBD (g cm−3)	−0.105	−0.166	0.116	−0.052	−0.014	−0.152	−0.092	−0.012	−0.066

Notes: ΔSOC: variation in soil organic carbon between soil layers; ΔTN: variation in total nitrogen between soil layers; ΔTP: variation in total phosphorus between soil layers; ΔAK: variation in available potassium between soil layers; ΔAP: variation in available phosphorus between soil layers; ΔMC: variation in moisture content between soil layers; ΔEC: variation in electrical conductivity between soil layers; ΔBD: variation in bulk density between soil layers; * correlation is significant at p = 0.05 level; ** correlation is significant at p = 0.01 level.

Vertical variation in SOC and TN was also significantly associated with changes in TP, AK, AP, EC, and clay content (p < 0.01). In contrast, vertical variation in pH was negatively correlated with SOC and TN, with the strength of this relationship increasing with depth. For the C/N ratio, significant correlations with environmental factors were relatively limited, with a positive relationship with clay content observed only in the 15–30 cm vs. 0–15 cm interval.

3.3. Effects of Formation Period and Land Use Type

Across the three formation periods (ancient, modern, and contemporary), Figure 4a shows that EC consistently followed the order contemporary > modern > ancient, with significant differences between contemporary and both modern and ancient periods (p < 0.05), but these significant differences were observed only in the 0–15 cm layer. Within each period, EC decreased gradually with increasing soil depth. Figure 4b indicates that SOC did not differ significantly among periods within any soil layer; however, for all periods, SOC decreased significantly with depth (p < 0.05). Figure 4c reveals that TN did not differ significantly among periods within any soil layer, but within each period TN decreased significantly with depth (p < 0.05). Figure S3a demonstrates that the C/N ratio exhibited no significant differences either across periods within the same soil layer or across soil layers within the same period.

Figure 4. Variations in EC, SOC, and TN across different formation periods and soil depths. (a) Variations in EC among different soil layers during different formation periods; (b) Variations in SOC among different soil layers during different formation periods; (c) Variations in TN among different soil layers during different formation periods. Uppercase letters (A, B) denote variations across different formation periods within the same soil layer, whereas lowercase letters (a, b, c) denote variations across different soil layers within the same formation period. Ancient (12 sites), Modern (39 sites), and Contemporary (25 sites).

Land use types influenced soil EC (Figure 5a). Paddy land exhibited the highest EC across all soil layers, whereas forest land consistently exhibited the lowest. (Note: Because forest land was limited to a single sampling site (n = 1), all observations for this category throughout this study are considered preliminary and preclude robust statistical comparisons.) Drylands maintained moderate and stable EC levels, while orchards displayed an anomalous EC peak in the 15–30 cm layer. Notably, while EC varied significantly among land use types, differences across depths within each specific land use were not significant.

Figure 5. Variations in EC, SOC, and TN across different land use types and soil depths. (a) Variations in EC across different soil layers under different land use types; (b) Variations in SOC across different soil layers under different land use types; (c) Variations in TN across different soil layers under different land use types. Lowercase letters (a, b, c) denote variations across different soil layers within the same land use type. Dry land (68 sites), paddy land (5 sites), orchard land (2 sites), and forest land (1 site). Owing to the large sample size imbalances across land use types, ANOVA was not conducted.

Regarding soil nutrients, forest land recorded the highest SOC across all layers (Figure 5b). Dryland SOC decreased significantly with depth; paddy land and orchards had relatively low SOC with no significant differences between the 15–30 cm and 30–45 cm layers. Similarly, forest land showed the highest TN content with a slight decline with depth (Figure 5c). Dryland TN decreased significantly among all three layers; paddy land showed no significant depth differences, while orchard TN gradually decreased from the 0–15 cm layer. Finally, Figure S3b demonstrates that no significant differences in soil C/N ratio were found either across soil horizons within the same land use type or across land use types within the same soil horizon.

4. Discussion

4.1. Horizontal Distribution and Influencing Factors of Spatial Patterns

The higher SOC and TN concentrations observed along the Yellow River’s banks can be explained by the relationships shown in Table 2. In the surface layer (0–15 cm), nutrient contents were significantly negatively correlated with EC (p < 0.01) and positively correlated with MC (p < 0.01). Freshwater from the river tends to lower soil salinity and maintain higher moisture levels in adjacent areas, and these correlations imply that freshwater supply is a plausible factor contributing to elevated nutrient levels. Moreover, this pattern may result from a combination of natural processes, including upstream erosion and sediment transport, estuarine accumulation, and agricultural runoff. Specifically, erosion materials from upstream and downstream areas, including plant and animal residues, are transported by water flow to the estuary [38] and deposited under elevated salinity and altered hydrodynamic regimes caused by seawater intrusion [39], thereby providing a sustained external nutrient supply to the estuarine zone [25,38]. In addition to these natural processes, abundant water resources along both banks of the Yellow River support relatively intensive agricultural fertilization, which also significantly contributes to nutrient enrichment in the 0–15 cm layer [40].

The elevated SOC and TN in the western and eastern regions relative to the central area likely result from distinct processes. The western region corresponds to the older Yellow River Delta, long-term cultivation and soil amelioration may have historically enhanced nutrient retention. In the eastern estuarine zone, tidal deposition of macroalgae and marine organism debris, combined with seawater intrusion, could supply additional organic matter [41,42]; subsequent evaporation under high temperatures may concentrate salts to supersaturation, triggering surface salt crystallization [43], while elevated salinity might suppress microbial decomposition of residues [44], potentially favoring SOC preservation [22,45]. It should be noted that several potentially influential factors, including salt crystallization dynamics, microbial activity, and organic residue inputs, were not measured in the present study. Consequently, the interpretations offered here are preliminary and require further validation through specifically designed experiments.

4.2. Vertical Variation Distribution and Controlling Factors

The decline in SOC and TN with increasing soil depth is a widespread pattern across farmland in the Yellow River Delta [22,46,47,48]. Meanwhile, a study from Yancheng City in China also showed the similar trend [23]. Yu et al. (2016) [22] reported a continuous decline in SOC and TN contents in the 0–30 cm farmland soil layer between 1855 and 2016. However, our records reveal a distinct reversal of this trend. Specifically, the average SOC content in the 0–30 cm layer increased from 4.76 g kg⁻¹ to 7.39 g kg⁻¹, while TN content rose from 0.51 g kg⁻¹ to 0.76 g kg⁻¹, suggesting a possible recovery in nutrient levels over the past decades.

The west-to-east reduction in nutrient differences between the 0–15 cm and 15–30 cm layers may be closely linked to salinization processes. Figure 3a reveals a gradual decline in the vertical EC difference from inland to coastal sites. Consistent with this pattern, Table 3 shows a significant positive correlation between vertical differences in EC and those in nutrient content. As the EC contrast narrows westward to eastward, reaching as low as 50% to 25% near the estuary, the corresponding nutrient contrast also diminishes; when the EC difference is negative, the overall distinction between the two layers is further reduced. Moreover, earlier studies suggest that high salinity in coastal zones may influence organic matter by modulating the flocculation–dispersion behavior of soil particles. Concurrently, salt ion toxicity inhibits microbial activity and plant productivity, slowing organic matter decomposition and thereby weakening vertical nutrient variation between soil layers [49,50]. In contrast, the vertical variation in nutrient content in the 15–30 cm and 30–45 cm layers were significantly lower than that in the upper layer. Based on the trends in EC variation between layers shown in Table 1 and the positive correlation between EC and nutrient variation shown in Table 3, it can be concluded that the reduction in vertical EC variation largely explains the decrease in nutrient differences between these two soil layers.

The evaluated forest site consistently stored the highest SOC and TN among the evaluated land uses. This occurs because continuous organic inputs from litter fall and root turnover are protected from rapid decomposition by a thick litter layer and minimal anthropogenic disturbance. However, this nutrient pool is sensitive to land-use conversion; for example, forest clearing can reduce surface (0–10 cm) carbon stocks by up to 63%, and native forests store 26.4% to 84.4% more SOC and 27.1% to 82.4% more TN than bare soils, croplands, and plantations [51,52]. Among the remaining land uses, dryland SOC and TN in the 0–15 cm layer exceed those in paddy land. This disparity arises because drylands receive substantial organic matter inputs from straw incorporation, whereas paddy lands experience salinity levels far exceed the critical threshold (3.95 g kg⁻¹) that triggering SOC decline [53,54]. While these underlying mechanisms align with established ecological principles, the reliance on a single forest plot implies that these peak nutrient levels should be treated as a site-specific observation pending broader regional verification.

Soil C/N ratios in the study area ranged narrowly from 8 to 12 across all layers, reflecting low spatial variability and a stable local equilibrium between SOC and TN. While broader studies have documented marked spatial and vertical heterogeneity, including latitudinal gradients [55], depth-dependent fluctuations driven by surface accumulation and downward migration [17], and significant land use effects [56], the constrained range observed here indicates that carbon and nitrogen pools maintain a relatively stable spatial balance at the local landscape scale.

4.3. Limitations

Despite the valuable insights provided by this study regarding the spatial distribution of soil nutrients and their controlling factors in the Yellow River Delta, several limitations should be acknowledged. First, the data were collected from a single sampling campaign and therefore do not capture seasonal or interannual variability. Future work should include repeated sampling across multiple seasons and years to assess the temporal stability of the observed patterns. Second, sample sizes were highly unbalanced across land-use types (e.g., farmland n = 68, forested land n = 1), which reduces the statistical power and reliability of cross-group comparisons. Third, spatial autocorrelation among sampling points may exist, as neighboring samples are likely to be more similar than distant ones. Although the linear mixed-effects model partially accounts for this by treating sites as random effects, residual spatial dependence may not have been fully addressed. Fourth, spatial interpolation in areas with low sampling density is subject to considerable uncertainty, and the resulting spatial patterns should be interpreted with caution. Finally, we lacked detailed information on site-specific management history (e.g., fertilization records, cultivation duration, and irrigation practices), making it difficult to disentangle the effects of land use from those of management intensity. Collectively, these constraints limit the strength of causal inference. Future research should address these issues through multi-season and multi-year sampling, more balanced experimental designs, collection of management records, and explicit modeling of spatial structure.

5. Conclusions

This study conducted a comprehensive assessment of the horizontal and vertical variation distribution of soil salinity and nutrients in the Yellow River Delta. Salinity does not simply inhibit nutrient accumulation; rather, it interacts with soil moisture to shape the surface enrichment and deep layer migration of nutrients, resulting in contrasting spatial distribution patterns between the western and eastern parts of the delta. This finding suggests that soil fertility management in the region should account for the east–west contrast: in western agricultural areas, priority should be given to maintaining surface organic matter to prevent carbon and nitrogen losses under progressive salinization; in eastern coastal wetlands, more attention should be paid to the availability of nutrients in deeper layers to mitigate environmental risks from leaching. Future research should incorporate dynamic water salt transport models to further explore how seasonal salinity fluctuations affect nutrient turnover, thereby offering more practical guidance for the ecological restoration of saline–alkali lands.