Soil and Its Interaction with the Climate Jointly Drive the Change in Basic Soil Productivity under Long-Term Fertilizer Management

Abstract

Basic soil productivity (BSP) is the productive capacity of farmland soils with their own physical and chemical properties during a specific crop season under local field management. Improving BSP as an effective way to increase or maintain crop yield, and researching its changes and potential driving factors under long-term fertilization are crucial for ensuring high and stable crop yields. In this study, the yields of BSP were simulated using the decision support system for agrotechnology transfer (DSSAT) crop model based on three long-term experiments, and its changing characteristics and driving factors were investigated under various fertilization treatments in wheat–maize rotation systems during 1991–2019. Five treatments were included: (1) unfertilized control (CK); (2) balanced mineral fertilization (NPK); (3) NPK plus manure (NPKM); (4) high dose of NPK plus manure (1.5NPKM); and (5) NPK plus crop straw (NPKS). This study found that the BSP of wheat and maize exhibited a fluctuating increase or stable change trend under four fertilization treatments at the Yangling (YL) and Zhengzhou (ZZ) sites, while a fluctuating reduction trend was observed at the Qiyang (QY) site. Compared with CK, NPKM, 1.5NPKM and NPKS significantly (p < 0.05) improved the BSP of wheat and maize at the YL and ZZ sites. The BSP of both maize and wheat under NPKM and 1.5NPKM was significantly (p < 0.05) higher than that under NPK or NPKS, whereas no statistically significant difference was found between NPK and NPKS at the QY site. The contribution rates of basic soil productivity (CBSP) of maize and wheat at the YL (41.5–60.7% and 53.0–64.3%) and ZZ sites (44.4–59.2% and 56.8–66.7%) were overall higher than that at the QY site (25.8–37.6% and 48.3–60.1%). In most cases, the difference in CBSP among different fertilization treatments was consistent with that in BSP. Moreover, a significant positive correlation was recorded between BSP and soil pH (r = 0.73, p < 0.01; r = 0.87, p < 0.01), TN (r = 0.56, p < 0.01; r = 0.62, p < 0.01) and TK content (r = 0.49, p < 0.01; r = 0.58, p < 0.01) in maize and wheat. Soil pH significantly correlated positively with the BSP of maize (R² = 0.54, p < 0.001) and wheat (R² = 0.49, p < 0.001) at the QY site, but negative correlation (R² = 0.20, p < 0.001; R² = 0.30, p < 0.001) was only found in maize at the YL and ZZ sites. The BSP of maize and wheat showed a significant negative linear correlation with MAP (R² = 0.49–0.67, p < 0.001) and MAT (R² = 0.36–0.62, p < 0.001). Random forests (RF) and variance partitioning analysis (VPA) revealed that soil properties and its interaction with the climate showed a higher explanation rate for BSP, indicating that these factors are the key drivers of BSP change. Overall, chemical fertilizers combined with manure can effectively increase BSP, while the effects of fertilizer combined with straw on BSP vary by region. The changes in BSP in wheat–maize cropping were mainly driven by both the soil and its interactions with the climate.

1. Introduction

Global food consumption has rapidly increased as a result of the rise in population over the past 70 years from 2.5 billion to 7.96 billion [1]. As a significant source of carbohydrates, cereal yield is critical to the world’s diet [2]. The increase in cereal yields is of great significance for ensuring global food security and social stability. Wheat–maize crop rotation has been widely employed in production techniques as a method of intensive farmland management in the global agricultural system [3]. Especially in many areas of China, rotation systems of wheat and maize in succession in a single year are geographically located in the Northeastern region, Huang-huai-hai region, Southwestern region and many parts of the Northwestern region [4]. According to the United Nations Food and Agriculture Organization, maize and wheat production amounts to about 364 million tons in China, accounting for 17% of the world total production [5]. Therefore, the productivity of the wheat–maize rotation system is of great significance to ensure food security in China.

Fertilization is a main artificial approach to improve crop productivity and soil fertility, which can increase crop yields by at least 30–50% [6,7]. This is mainly because, on the one hand, fertilization can provide the required nutrients for crop growth, and on the other hand, it can improve the physical, chemical and biological properties of the soil, and further enhance the soil production and ecological capacity [8,9,10]. However, the excessive and irrational application of chemical fertilizers can also have serious negative impacts on agroecosystems, including soil quality degradation, decreased soil productivity and environmental pollution [11]. With the development of agricultural technology, various organic fertilizers are used in agricultural production, containing rich nutrients and organic matter, which has a positive effect on improving soil fertility and fertilizer efficiency compared to chemical fertilizers [12,13]. In addition, the combined application of organic and inorganic fertilizers could increase crop yield or soil quality while reducing the waste of chemical fertilizer resources and the resulting environmental risks [14,15,16]. For this reason, a large number of studies mainly focus on whether the combined organic and inorganic application has a positive effect on improving crop yield or soil quality, ignoring its ability to maintain high soil productivity.

At present, fertilization has reached a high level, and increasing the amount of fertilizer to enhance crop yield is no longer an effective way; the exploitation of farmland production potential has become an important way to increase production. The approaches to improve soil productivity include rational water and nutrient input, optimized field management and the increase in basic soil productivity (BSP) [17]. BSP has been defined as the productive capacity of farmland soils with their own physical and chemical properties during a specific crop season under local field management [18]. Based on this concept, there are two main contributions to crop yield under the same climatic conditions, irrigation and cultivation methods, namely, external fertilizer application and BSP [17,19]. Therefore, crop yield mainly depends on the contribution of BSP to yield under the same climate conditions, water and fertilizer input and management conditions (e.g., irrigation, tillage and cultivation practices). A number of studies have explored changes in crop yield without fertilization based on long-term experiments, with varying results. A gradual decline trend was recorded in wheat yields without fertilizer treatment based on a 50-year long-term experiment in the Caslav Crop Rotation Experiment in the Czech Republic [20]. Similarly, the yield of wheat with no fertilization treatment showed a declining trend in four long-term wheat–corn rotation systems in China from 1991 to 2005 [21,22]. In contrast, Han et al. (2020) [23] indicated that both wheat and rice yields increased when treated without fertilizer under a 34-year long-term rotation system. The yields of wheat have remained stable or increased for more than 50 years under high natural fertility conditions even without fertilization [24], which suggested that BSP could indicate soil fertility and maintain crop yield. BSP was generally expressed by a comprehensive index of the yield by BSP, and characterized by the yield of no fertilization treatment or the contribution percentage of BSP (percentage of yield without fertilization to yield of corresponding fertilization treatment, %) in most previous studies [25,26,27]. However, there was an obvious problem in long-term experiments; that is, soil nutrients were constantly depleted without fertilizer treatment, and the yield without fertilization could not reflect the actual change in BSP under fertilization. Therefore, it is necessary to further explore the evolution of BSP under long-term fertilization through more scientific indication methods.

A crop model to simulate crop yields has been proposed in previous studies as an effective method to solve the problem of BSP under different fertilization treatments [18]. The decision support system for agrotechnology transfer (DSSAT) could accurately predict crop growth and production, nutrient uptake, and scientifically assess the effect of agronomic measures (such as tillage, irrigation and fertilization) on crop yield [28,29,30,31,32,33,34]. Some studies have used the DSSAT model to explore the difference in BSP under different fertilization and the influence of main soil nutrient indexes (such as SOC, total and available nitrogen, phosphorus and potassium content) on BSP, which shows a good simulation effect and can more truly reflect the basic soil productivity of farmland [17,18,19]. Although these existing studies have a certain understanding of the differential characteristics and influencing factors of BSP under different nutrient management strategies, they mainly focus on a single test site with similar climatic conditions and field management, and the differences and key driving factors of BSP under different fertilization conditions in various regions are still unclear. This study used the DSSAT model to simulate the yield of BSP, and explore the change characteristics and key driving factors of BSP under different nutrient management strategies based on multi-site long-term fertilization experiments. We hypothesize that long-term combined organic and inorganic application could effectively increase the BSP of farmland in the wheat–maize planting pattern, and the changes in BSP in different regions are jointly driven by the climate, fertilization and soil properties. Therefore, the main objectives of this study are as follows: (i) to investigate the dynamic changes in BSP under different long-term fertilization treatments in wheat–maize cropping; (ii) to evaluate the effects of the climate and soil properties on BSP; and (iii) to clarify the contribution of the driving factors affecting BSP.

2. Materials and Methods

2.1. Long-Term Experimental Design and Management

In order to evaluate the effects of different long-term nutrient management strategies on BSP, three typical wheat–maize rotation systems were established in 1990’s based on the China Soil Fertility and Fertilizer Effect Long-term Monitoring Network [35]. From north to south, the sites are Yangling (YL) in Shaanxi Province, Zhengzhou (ZZ) in Henan Province and Qiyang (QY) in Hunan Province. Their geographical location and climatic conditions are presented in Table S1, and the soil types of the three sites are Calcaric Regosol, Calcaric Cambisol and Ferralic Cambisol, based on the soil classification of FAO. The initial soil profile properties at each site are shown in

Table 1. Initial chemical and physical properties of soils in various depths of soil profiles at the three experimental sites before the treatments were carried out.

Site	Soil Depth (cm)	Organic Matter	Total N	Total P	Total K	Available N	Available P	Available K	pH	Bulk Density (g cm−3)
		(g kg−1)				(mg kg−1)
YL	0–17	10.92	0.83	1.39	2.28	61.3	9.57	191.0	8.62	1.30
	17–38	8.87	0.71	1.31	2.32	52.3	4.97	156.0	8.51	1.56
	38–57	5.95	0.56	1.11	2.21	32.5	1.96	143.0	8.54	1.46
	57–77	5.98	0.52	0.74	2.32	32.2	1.35	120.0	8.49	1.43
ZZ	0–28	10.60	1.01	0.65	1.69	76.6	6.50	71.7	8.10	/
	28–52	5.00	0.40	0.51	1.73	54.1	3.80	59.2	8.40	/
	52–87	4.00	0.37	0.50	1.73	48.7	2.20	62.3	8.30	/
	87–120	3.60	0.37	0.50	1.86	65.8	2.00	54.5	8.20	/
QY	0–20	15.15	0.85	0.71	13.29	67.9	10.80	109.2	5.39	1.19
	20–40	8.79	0.62	0.46	15.29	37.4	8.10	63.2	5.79	1.48
	40–60	3.08	0.39	0.43	13.56	12.4	2.70	46.3	5.83	1.52
	60–80	4.11	0.35	0.44	13.38	18.3	2.40	31.4	5.29	1.58

Note: YL, ZZ and QY indicate Yangling, Zhengzhou and Qiyang, respectively; / indicates that no data are available.

. Winter wheat–summer maize rotation is a common farming system in the region, yielding two crops a year.

Three long-term fertilization experiments were conducted, and detailed information about the design of the experiment has been reported by previous studies [21,22,36]. The three sites were designed as a complete trial that included the same number of treatments at each site, i.e., repeated treatments at three sites but not at a single site, which is limited by the size of the available fields and the requirements for large treatment plots. Five different fertilization treatments were implemented in this study, including (1) unfertilized control (CK); (2) a balanced application of mineral nitrogen, phosphorus, and potassium fertilizers (NPK); (3) NPK combined with manure (NPKM); (4) a high dose (1.5 times) of NPK combined with manure (1.5NPKM); and (5) NPK combined with crop straw (NPKS). At each site, all chemical nitrogen fertilizers were in the form of urea (N), P fertilizers in the form of super phosphate (P₂O₅), and K fertilizers in the form of potassium sulphate (K₂O), respectively. Except at QY, an equivalent amount of N was applied to all treatments, whereas extra N (from wheat straw) was included in NPKS. Due to the complications in balancing all nutrient applications, only total N (not P and K) in the manure source was considered, as the amount of manure applied at each site depends on its annual nutrient (mainly, total N) content. Horse manure was applied in the ZZ site from 1990 to 1998 and cow manure from 1999 to 2019; cow manure and pig manure were applied, respectively, in the YL and QY sites from 1990 to 2019. The ratio of organic N to inorganic N is 7:3 for NPKM, and no topdressing was applied at any site. All manure and straw were applied once before wheat seeding. The application amounts of N, P, and K nutrients in the wheat and maize seasons under each treatment were presented in a previous study [37], and are shown in Table S2.

At the ZZ and YL sites, maize was grown between mid-June and late September, while wheat was grown between mid-October and early June. At the QY site, maize was sown between the wheat strips in early April and harvested in July, while winter wheat was sown in early November and harvested in early May in the following year. All above-ground crops were harvested with a harvester and removed, and no straw was returned into the soil except NPKS treatment. Before subsequent maize and wheat planting, maize or wheat stubble and roots were incorporated into the soil with a plow. No irrigation was given to crops at the QY site, but wheat was irrigated 2–3 times and maize was irrigated once (about 75 mm each time) at the ZZ and YL sites, depending on precipitation. Except for the fertilizer treatment, all other agronomic management measures was the same in fertilized and unfertilized plots. Additionally, the mean annual precipitation (MAP) and mean annual temperature (MAT) at the three sites were also recorded from 1990 to 2019, and are presented in Figure 1.

2.2. Sample Collection and Determination

Crop samples were harvested manually, and grain yield was determined and recorded from the whole plot during the harvest season every year. Topsoil samples (0–20 cm) were collected annually about 15 days after maize harvest. To collect representative samples, a total of 20 soil cores from each plot were collected using an auger with an internal diameter of 5 cm, and evenly mixed to produce five composite subsamples. All collected soil samples were air-dried and screened with 2.0 mm stainless steel sieve for further analysis. The bulk density (BD) was also measured every year by classical analytical methods, and soil types were classified based on the United Nations Food Agriculture Organization (FAO) soil taxonomy system [38]. The soil pH was measured using a pH meter (Metro pH 320; Mettler-Toledo IL., Shanghai, China) in a 1:2.5 soil/water (H₂O) suspension, while the organic matter/carbon (SOM/SOC) was determined using the potassium dichromate volumetric method [39]. The total N (TN) was determined using the semimicro Kjeldahl method (K9840, Shanghai Lijing Scientific IL., Shanghai, China), the total P (TP) was determined by sodium hydroxide melting-molybdenum antimony antichromic (U3900, Hitachi, TKY, Japan), and the total K (TK) was measured using sodium hydroxide melt-flame spectrophotometry (M410, Sherwood, CA, UK) [40]. The available N (AN) was determined using a micro-diffusion method after alkaline hydrolysis, the available P (AP) was extracted with 0.5 mol L⁻¹ NaHCO₃ (soil:solution = 1:20) and measured with the Olsen method, and the available K (AK) was extracted from the soil with 1 mol L⁻¹ NH₄Ac (soil:solution = 1:10) and measured with flame photometry (M410, Sherwood, CA, UK) [40].

2.3. Simulation of Basic Soil Productivity (BSP)

2.3.1. Simulation Method

The CERES-Maize and CERES-Wheat modules in DSSAT (Version 4.8) were used to simulate yield by basic soil productivity in this study. The simulation process was referenced based on previous research [17] and is as follows: (a) The climate, soil, and field management parameters required for model operation were input; (b) the genetic parameters of summer maize and winter wheat treated with NPK were calibrated; (c) the simulated yield of the other four fertilization treatments was evaluated; and (d) under the same parameter set, the method of no fertilization in the current season and fixed fertilization in other years was used to simulate the current yield of BSP.

2.3.2. Input Data and Parameters Calibration

Input data includes field management, initial soil profile data, weather data and crop cultivar parameters, which are required for this model. The initial soil profile data and field management were obtained from the National Soil Fertility and Fertilizer Effects Long-term Monitoring Experiment Station (Yangling, Zhengzhou and Qiyang sites). The initial soil profile data are presented in Table 1, and these data were saved in the DSSAT soil file. The field management data must be entered when creating the trial file, including the crop planting date, planting density and depth, row spacing, tillage method, tillage depth, fertilizer application and rate, irrigation, harvest date, etc.

In addition, the minimum weather data sets required by DSSAT include daily minimum temperature (TMIN, ℃), daily maximum temperature (TMAX, ℃), daily precipitation (RAIN, mm), and daily solar radiation (SRAD, (MJ m⁻²), which was obtained from the China Meteorological Data Sharing Service System from 1990 to 2019. The latitude and longitude of each site were also combined to build a weather station of the model. Because solar radiation cannot be directly measured, the daily solar radiation quantity (Q) was calculated using the number of hours of daily sunshine [41].

$$\mathrm{Q}={\mathrm{Q}}_0\times (\mathrm{a}+\mathrm{b}\times \frac{\mathrm{S}}{{\mathrm{S}}_0})$$

(1)

where Q₀ represents astronomical radiation, S₀ and S are the available sunshine hours and measured sunshine hours, respectively, and a and b are functions of S and S₀.

Finally, a “trial and error” method was used to calibrate the genetic parameter of crop variety [31], based on the yield data of wheat and maize under NPK treatment. The normalized root mean square error (NRMSE) was used to find the best matching parameter (i.e., the minimum NRMSE). After the calibration of the variety parameters, the matching genetic parameters were obtained by the least differences of NRMSE between the measured and simulated values, and are presented in Table S3.

2.3.3. Model Evaluation

In order to accurately evaluate the performance of the DSSAT model, the normalized root mean square error (NRMSE) and index of agreement (D) [42] were used to check the match between the measured values and the simulated values, and test the goodness of fit between them. The calculation formula is shown below:

$$\mathrm{N}\mathrm{R}\mathrm{M}\mathrm{S}\mathrm{E}=\frac{100}{\mathrm{A}}\times \sqrt{\frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{({\mathrm{S}}_{\mathrm{i}}-{\mathrm{M}}_{\mathrm{i}})}^2}{\mathrm{n}}}$$

(2)

$$\mathrm{D}=1-\left[\frac{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{({\mathrm{S}}_{\mathrm{i}}-{\mathrm{M}}_{\mathrm{i}})}^2}{{\sum }_{\mathrm{i}=1}^{\mathrm{n}}{(\left|{\mathrm{S}}_{\mathrm{i}}^{\prime }\right|-\left|{\mathrm{M}}_{\mathrm{i}}^{\prime }\right|)}^2}\right]$$

(3)

where S_i indicates the simulated value, M_i indicates the measured value, A indicates the average measured value, S_i′ = S_i − A, M_i′ = M_i − A, and n indicates the number of simulated value samples. An NRMSE value ≤ 50% is generally considered acceptable model performance [43]. The index of agreement (D) (0 ≤ D ≤ 1) is a descriptive measure that is both a relative measure and a bounded measure [42]. When D ≥ 0.9, this indicates “excellent” agreement; when 0.8 ≤ D < 0.9, this is “good” agreement; when 0.7 ≤ D < 0.8, this is “moderate” agreement; and when D < 0.7, this is “poor” agreement between the measured and predicted values [44]. The present study showed that the NRMSE values (3.60–25.12%) between the measured value and simulated value of wheat and maize yield under all treatments were less than 30%, and the values of D ranged from 0.73 to 0.85 (Table S5), indicating that the DSSAT model was suitable for simulating crop yield and BSP.

2.4. Contribution Rate of Basic Soil Productivity (CBSP)

In order to compare the yield benefits brought by basic soil productivity under different fertilization strategies, the contribution rate of basic soil productivity (CBSP) was calculated based on the following formula:

$$\mathrm{C}\mathrm{B}\mathrm{S}\mathrm{P}(\%)=\frac{{\mathrm{Y}\mathrm{B}\mathrm{S}\mathrm{P}}_{\mathrm{i}}}{{\mathrm{Y}\mathrm{M}}_{\mathrm{i}}}\times 100$$

(4)

where YBSP_i and YM_i are the yield of BSP and the measured yield in the corresponding fertilization treatment, respectively.

2.5. Statistical Analysis

The CERES-Maize and CERES-Wheat modules in DSSAT (Version 4.8) were used to simulate wheat and maize yields and BSP under each fertilization treatment. One-way variance analysis of BSP and its contribution percentage under different fertilization treatments was conducted using SPSS 23.0 (SPSS Inc., Chicago, IL, USA), and all bars were created with Origin 2019b (OriginLab., Northampton, MA, USA). The relationships among BSP, climate and soil properties were analyzed using linear regression and correlation analysis. The relative importance of fertilization, climate condition and soil properties on BSP was analyzed using the random forest (RF) model and variance partitioning analysis (VPA) based on the ‘vegan’ package of R (version 4.1.2). Unless otherwise stated, the least significant difference (LSD) test was used to compare mean values at p < 0.05 levels. All data of this study are based on dry weight.

3. Results

3.1. Effect of Long-Term Fertilization on BSP

Whether it is maize or wheat season, BSP under fertilizer treatment at the YL and ZZ sites was overall higher than that at QY during 29 fertilization treatment periods (Figure 2). Compared with CK treatment, the three treatments combining organic fertilizers with chemical fertilizers obviously (p < 0.05) improved the BSP of wheat and maize at the YL and ZZ sites, whereas the NPK treatment only significantly (p < 0.05) improved BSP in the wheat season. The BSP of two crop seasons under the three treatments combining organic and chemical fertilizers was significantly higher than that under NPK at YL, and the three treatments combining organic and chemical fertilizers significantly improved the BSP of maize at the ZZ site, but the BSP of wheat was only improved under the 1.5NPKM treatment. At the QY site, the BSP of both maize and wheat under the chemical fertilizers combined with manure (NPKM and 1.5NPKM) was significantly higher than that under the chemical fertilizers alone (NPK) or combined with straw (NPKS), and no significant difference was recorded between NPK and NPKS. Compared with CK, the four fertilization treatments significantly increased the BSP of the maize season at the QY site, but there was no significant difference with the NPK treatment of the wheat season.

From 1991 to 2019, BSP showed a similar zigzag fluctuation trend under different fertilization treatments in both maize and wheat seasons at the three sites (Figure S1). At the YL and ZZ sites, the BSP of wheat and maize under the four fertilization treatments showed a fluctuating increase or stable change trend, while a fluctuating decrease trend was observed at the QY site.

3.2. Contribution Rate of BSP

Figure 3 shows the effects of different fertilization treatments at the three sites on the contribution rate of BSP (CBSP) of maize and wheat. Similar to BSP, CBSP under different fertilization treatments at the YL and ZZ sites was also overall higher than that at the QY site. At the YL site, the CBSP of maize and wheat ranged from 41.5 to 60.7% and from 53.0 to 64.3%, respectively, and showed the order of 1.5NPKM > NPKM > NPKS > NPK in both two seasons; the difference was significant (p < 0.05) between each two treatments. At the ZZ site, the CBSP of maize and wheat ranged from 44.4 to 59.2% and from 56.8 to 66.7%, respectively, and compared with the NPK treatment, NPKM, 1.5NPKM and NPKS significantly (p < 0.05) improved the CBSP of maize, but only 1.5NPKM and NPKS significantly improved the CBSP of wheat. At the QY site, the CBSP of maize and wheat ranged from 25.8 to 37.6% and from 48.3 to 60.1%, respectively, and showed the order of 1.5NPKM > NPKM > NPK > NPKS in maize seasons; the difference was significant (p < 0.05) between treatments. However, the CBSP of wheat under 1.5NPKM and NPKM was obviously (p < 0.05) higher than that under NPK and NPKS at the QY site, and there was no significant difference between NPK and NPKS.

3.3. Effects of Soil Properties on BSP

A significant positive correlation was found between BSP and soil pH (r = 0.73, p < 0.01; r = 0.87, p < 0.01), TN (r = 0.56, p < 0.01; r = 0.62, p < 0.01) and TK content (r = 0.49, p < 0.01; r = 0.58, p < 0.01) in maize and wheat seasons (Figure 4), which suggested that BSP is closely related to pH, TN and TK content in soil. Moreover, the linear regression analysis showed that TN was significantly positively correlated with the BSP of maize (R² = 0.20, p < 0.001; R² = 0.26, p < 0.001; R² = 0.22, p < 0.001) and wheat (R² = 0.12, p < 0.001; R² = 0.18, p < 0.001; R² = 0.07, p < 0.01) at all three sites, and TK showed a significant positive linear relationship (R² = 0.23, p < 0.001 for maize; R² = 0.33, p < 0.001 for wheat) in the overall analysis of the three sites (Figure S2A–D). The soil pH showed a significant positive correlation with the BSP of maize (R² = 0.54, p < 0.001) and wheat (R² = 0.49, p < 0.001) at the QY site, but a negative correlation (R² = 0.20, p < 0.001; R² = 0.30, p < 0.001) only in the maize season at the YL and ZZ sites (Figure S2E,F).

3.4. Effects of MAT and MAP on BSP

The BSP of maize and wheat under all fertilization treatments showed a significant negative linear correlation with MAP (R² = 0.49–0.67, p < 0.001) and MAT (R² = 0.36–0.62, p < 0.001) (Figure 5), which also indicated that BSP gradually decreased with the increase of MAP and MAT.

3.5. Proportional Contributions of Environmental Factors to BSP

Random forests (RF) were performed to analyze the relative importance of fertilization, soil properties and climate factors on the BSP of maize and wheat (Figure 6). Across all fertilization treatments and test sites, 81.2% and 88.5% of the total variability in the BSP of maize and wheat could be explained by fertilization, soil and climate factors and their interactions. Among these factors, the pH value has the highest explanation rate for the BSP variation in maize and wheat, which is 11.5% and 13.9%, respectively. Meanwhile, MAT (10.0% and 13.0%), TN (9.4% and 9.5%) and TP content (9.2% and 9.5%) in soil were also main variables affecting the BSP of maize and wheat. The explanation rate of BSP for maize and wheat by fertilization was the lowest, only 5.6% and 6.9%. In addition, the influencing factors are divided into three categories and their contribution to BSP was analyzed using VPA (Figure 7). Soil, fertilization and climate factors and their interactions explain 73.9% and 83.1% of the BSP variation in maize and wheat, respectively. In summer maize, soil, fertilization, and climate factors explained 13.8%, 1.1% and 2.1% of the BSP variation, respectively, while the interactions between soil and climate factors and fertilization factors explained 45.6% and 11.9% of BSP variation, respectively (Figure 7A). In winter wheat, soil, fertilization, and climate factors, respectively, explained 17.7%, 0.2% and 0.4% of the BSP variation, while the interactions between soil and climate factors and fertilization factors explained 63.1% and 2.3% of BSP variation, respectively. The interaction between the climate and fertilization had a slight effect on BSP in both maize and wheat (Figure 7). Overall, soil properties and its interaction with climatic factors are the key drivers of BSP change in wheat–maize rotation systems.

4. Discussion

4.1. Evolution Characteristics of BSP under Fertilization

As a comprehensive index for evaluating soil basic fertility, BSP improvement is another effective way to increase crop yield in addition to fertilization [19]. Therefore, studying the effects of different long-term fertilization treatments on BSP is crucial to ensure food security. This study showed that the combination of organic and inorganic fertilizers could improve BSP in different degrees at the three test sites, especially in the 1.5NPKM treatment (Figure 2). Compared with NPK alone, chemical fertilizers combined with manure and straw significantly increased BSP at the YL and ZZ sites. These results were attributed to the combined application of chemical fertilizers and organic fertilizers which increased the soil nutrient contents (Table S4), and further increased soil fertility and basic soil productivity. Several studies have also indicated that the incorporation of organic amendments into chemical fertilizers can obviously improve the physical and chemical properties of the soil, further improve the quality of the soil and crop yield [45,46,47]. Similarly, previous research suggested that chemical fertilizers combined with manure or straw significantly increased BSP compared to chemical fertilizers applied alone in black soil and fluvo-aquic soil tests [17,18]. Unfortunately, chemical fertilizers combined with straw returning did not significantly increase the BSP of maize and wheat at the QY site compared to chemical fertilizers alone, which was primarily attributed to the fact that the chemical fertilizers alone or combined with straw returning accelerated soil acidification [48], reducing BSP at the QY site. This result was also confirmed by a significant decrease in pH when NPK and NPKS were used, compared to treatment without fertilization (Table S4). Guo et al. (2010) [49] have found that the long-term application of N fertilizers leads to soil acidification and a sharp decrease in yield in southern China. From 1991 to 2019, the BSP of maize and wheat showed a fluctuating increase or stable trend at the YL and ZZ sites, while a fluctuating decrease trend was observed at the QY site. A previous study also found that the BSP of spring maize under NPK, NPKM, 1.5NPKM, and NPKS showed a similar increasing trend in the black soil area of northeast China from 1990 to 2011 [18]. Although there was no decrease in the soil pH under NPKM and 1.5NPKM at the QY site (Table S4), a decreasing trend was recorded in the BSP of wheat (Figure S1). This result may be due to the overall low pH in the soil at the QY site (Table S4).

In addition, CBSP at the YL and ZZ sites was generally higher than that at the QY site (Figure 3), which may be due to the fact that the basic soil fertility at the YL and ZZ sites was higher than that at the QY site in the wheat–maize rotation system, and the lower CBSP at the QY site was affected by soil acidification. A previous study found that the average percentage contribution of BSP under fertilization was 74.4–84.7% [18], which is higher than the CBSP at the three sites of this study. However, the percentage contribution of BSP under fertilization treatments varied from 42.5 to 59.9% for winter wheat and from 57.4 to 68.7% for summer maize in a fluvo-aquic soil area of north China [17], which was comparable with that at the YL and ZZ sites, but higher than that at the QY site. In most cases, the NPKM, 1.5NPKM and NPKS treatments increased the CBSP of maize and wheat at the YL and ZZ sites, while NPKS did not significantly increase CBSP compared with the chemical fertilizers alone (Figure 3). This was also explained for reasons consistent with the above BSP. In summary, the long-term application of chemical fertilizers combined with manure can effectively increase BSP in the three experimental sites of China, but the effect of combined straw application on BSP varies with the region.

4.2. The Relationship between BSP and Soil Properties

The difference in BSP mainly depends on soil properties under different fertilization management strategies [18]. This study found that the BSP of maize and wheat was significantly positively correlated with the soil pH, TN and TK at the three test sites (Figure 4 and Figure S2), which indicated that the change in BSP was mainly affected by the soil pH, TN and TK content at the three study areas. Similarly, the random forest (RF) model results also show that the pH, TN and TP were key limiting factors leading to BSP in the three regions (Figure 6). These results may be due to the large pH difference in the three sites which are located across China from north to south, especially the low pH of the soil at the QY site and the serious soil acidification (Table S4), which led to a decline in soil fertility and further reduced the BSP of the farmland. Qiyang is located in the red soil region of southern China, and due to high precipitation, low cation exchange capacity and the increased nitrification of chemical fertilizers, the pH has been decreasing since 1990 [22,50]. However, the pH of the YL and ZZ sites in northern China did not change much (from 7.98 to 8.33), and the above significant positive correlation with BSP may be caused by the larger pH threshold at the QY site (from 4.60 to 5.94) (Table S4). Similar to the general conclusion, BSP increased linearly with the increase of pH in the acidic soil of the QY site, but decreased with the increase of pH in the alkaline soil of the YL and ZZ sites (Figure S2). Several studies also found that with the extension of the years of application of organic fertilizers, both alkaline soil and acidic soil gradually tend to become neutral, which is conducive to the improvement of soil fertility and crop growth [51,52]. As expected, the TN, TP and TK contents were the main factors affecting regional differences in BSP, and could directly reflect the soil fertility and its productive capacity. The NPKM and 1.5NPKM treatments obviously improved the content of total nitrogen, phosphorus and potassium in the course of 29 years (Table S4). Thus, the increase in TN, TP and TK content in the soil caused by the application of organic fertilizers was the main reason for BSP increasing. Different from this study, previous studies have confirmed that soil organic matter or organic carbon content (SOM or SOC) was the main factor affecting BSP change in black soil and fluvo-aquic soil areas of China [17,18], which could be attributed to the fact that the study area was primarily a single test site, and organic matter or organic carbon content is the key limiting factor of soil fertility in this region. However, this study has the advantage of crossing climate zones from north to south, so the key influencing factors of BSP were mainly reflected in the soil properties with large regional variations such as pH and TN. Numerous previous studies have also indicated that the main factors affecting the change in farmland soil fertility vary with the study area [53,54,55,56,57].

4.3. The Relationship between BSP and Climate Factors

Climatic conditions (e.g., temperature and precipitation) generally strongly affect soil properties, such as the soil pH and carbon, nitrogen and phosphorus nutrient contents, and its stoichiometry, which further affects soil fertility and productivity [50,58,59,60,61]. Similarly, the results of this study also show that in the three study areas from north to south in China, the BSP of farmland was significantly negatively correlated with MAP and MAT in all fertilization treatments (Figure 5). These results also indicate that the BSP of farmland showed a gradual decline trend with the increase of MAT and MAP. On the one hand, the increase in temperature can greatly improve the decomposition of soil nutrients by changing microbial community structure, biomass, and activity [62,63,64], which may lead to a decline in farmland BSP. On the other hand, the increased temperature may enhance the absorption of nutrients by soil microorganisms and plants, and reduce the content of soil residual nutrients [65]. Moreover, Zhao et al. (2017) [66] found that the mineralization and decomposition of soil carbon, nitrogen and phosphorus are affected by precipitation and its effect on soil moisture. Abundant precipitation may have accelerated the loss of these nutrients, thus reducing the BSP of farmland accordingly. It has also been reported that a higher mean annual precipitation may lead to soil acidification and further stimulate the mineralization of available soil nutrients [67,68], which was also confirmed in this study. Overall, MAT and MAP have been confirmed to be important factors affecting soil fertility and BSP in this study.

4.4. Contribution of Fertilization, Climate and Soil to BSP

Generally, fertilization, climate and soil factors jointly affected the change in BSP in wheat–maize rotation in different long-term fertilization experiments. The present study showed that soil properties such as the pH, TN, TP and TK were the main factors affecting BSP (Figure 6), and their explanation rate for BSP was much higher than that of climate and fertilization (Figure 6 and Figure 7). This result is also consistent with the general conclusion that the BSP is also the embodiment of soil fertility level, which mainly depends on the nature of the soil itself. Fertilization showed the lowest explanation rate for BSP, which may be due to the fact that fertilization indirectly affects BSP in farmland by affecting soil nutrients, physical properties, and the structure and diversity of the soil microbial population. The interesting finding in this study is that the interaction between the climate and soil factors presented a higher explanation rate for BSP, suggesting that the changes of BSP in wheat–maize cropping were primarily driven by both the soil and its interactions with the climate.

5. Conclusions

During the 29-year experiment, the BSP of wheat and maize showed a fluctuating increase or stable change trend at the YL and ZZ sites in north China, while the BSP of the four fertilization treatments showed a fluctuating decrease trend at the QY site in south China. In most cases, BSP and CBSP showed roughly consistent patterns, and under fertilization treatments at the YL and ZZ sites, the values of both BSP and CBSP were overall higher than those at the QY site. Chemical fertilizers combined with manure or straw could effectively increase the BSP of wheat and maize at the YL and ZZ sites. Compared with NPK, the BSP of wheat and maize was improved by chemical fertilizers combined with manure and straw at the YL site, and the three treatments combining organic and chemical fertilizers significantly improved the BSP of maize at the ZZ site, but the BSP of wheat was only improved under the 1.5NPKM treatment. Low and high quantitative chemical fertilizers combined with manure could improve the BSP of two crop seasons at the QY site, but there was no significant difference between chemical fertilizers combined with straw and chemical fertilizers alone. In addition, the BSP of the three sites decreased with the increase of MAT and MAP, and the soil pH, TN and TK were the main soil factors affecting BSP. The BSP of acidic soil increased linearly with the increase in pH, but the BSP of alkaline soil decreased linearly with the increase in pH. Among three categories of factors, the soil properties and its interaction with the climate presented a higher explanation rate for BSP. Overall, chemical fertilizers combined with manure and straw can effectively increase BSP in alkaline soil, while their combination with straw did not improve the BSP in acidic soil. The changes in BSP in wheat–maize cropping were mainly driven by both the soil and its interactions with the climate. Therefore, the appropriate organic fertilizer types and application methods should be adopted according to local conditions in agricultural production practice, which could effectively improve the BSP and provide technical support for sustainable agricultural development.