Evaluation of Variable Application Rate of Fertilizers Based on Site-Specific Management Zones for Winter Wheat in Small-Scale Farming

Abstract

China is currently experiencing a severe issue of excessive fertilization. Variable rate fertilization (VRF) technology is key to solving this issue in precision agriculture, and one way to implement VRF is through management zone (MZ) delineation. This study is aimed at evaluating the feasibility and potential benefits of VRF based on site-specific MZs in smallholder farm fields. This study determined the amounts of basal and top-dressing fertilizers in different spatial units, based on soil nutrient MZs and crop growth MZs, respectively. The potential agronomic, economic, and environmental advantages of spatial variable rate fertilization were further assessed by comparing the farmer’s treatment, the expert’s treatment, and the variable rate fertilization treatment based on management zones (VR-MZ). The results showed that VR-MZ reduced the use of nitrogen (N), phosphorus (P), and potassium (K) fertilizers by 22.90–43.95%, 59.11–100%, and 8.21–100%, respectively, and it also increased the use efficiency of N, P, and K by 12.27–28.71, 89.64–176.85, and 5.48–266.89 kg/kg, respectively, without yield loss. The net incomes of VR-MZ were 15.5–449.61 USD ha⁻¹ higher than that of traditional spatially uniform rate fertilization. Meanwhile, less nitrous oxide emission (23.50–45.81%), ammonia volatilization (19.38–51.60%), and nitrate ion leaching amounts (28.77–53.98%) were found in VR-MZ compared to those in uniform fertilization. The results suggest that the VR-MZ has great potential for saving fertilizers, significantly increasing farmers’ net income, reducing environmental pollution, and promoting the sustainable use of resources. This study provides a theoretical basis and technical support for exploring a VRF suitable for village-scale farming.

1. Introduction

China’s total consumption of agricultural chemical fertilizers reached 59.91 million tons in 2021 [1], with nitrogen (N), phosphorus (P), and potassium (K) fertilizers accounting for 23%, 21%, and 26% of world consumption, respectively [2]. A large amount of these fertilizers was consumed by small-scale farmers, who constitute more than 70% of China’s farming systems [3]. They are agricultural micro-subjects that employ the family as the unit and integrate production and consumption [4]. Chemical fertilizer application as a management practice carried out by smallholder farmers has partially led to the high spatial variability of soil nutrients. Unlike large-scale precision agricultural systems, where it is needed to manage both among-field and within-field spatial variability, small-scale systems are characterized by greater challenges caused by unreasonable and unscientific fertilization by smallholders. This fertilization resulted in increasing levels of potentially toxic elements in most ecosystems on earth, which led to a series of environmental issues, such as soil acidification, serious water pollution, and excessive greenhouse gas emissions [5,6,7]. These have seriously affected China’s ecological environment, food security, and human health and are not favorable to sustainable agriculture development.

Variable rate fertilization (VRF), different from traditional uniform rate fertilization (URF), has been widely used in developed countries to achieve a high yield, high quality, and environmental protection [8]. According to previous studies, the application of VRF has exceeded 60% based on the development of an extensive field management information system and service information network [9]. The VRF technology involves a thorough analysis of various factors, such as the physical and chemical properties of the soil, climatic conditions, and historical yield data. By utilizing expert systems for crop growth and plant nutrition, it is able to determine the optimal amounts of fertilizers needed for different areas within a field. The implementation of VRF enables a precise and scientific approach to fertilization, resulting in a more efficient use of resources and improved crop yields [10]. Therefore, the application of VRF may help to address several issues that China is presently facing, including air and water pollution, climate change, excessive harmful substances, and low nitrogen use efficiency [6].

The first step in implementing VRF is to find units that are homogeneous within the field [11]. Management zones (MZs) are sub-areas that exhibit homogeneous attributes of landscape and soil conditions [12]. MZs enable the site-specific nutrient management and fertilization recommendations for fields [13]. In early research, farmers’ management experience and topography were used to delineate the MZs to propose spatially variable rate fertilization programs [14]. However, it is not scientific to rely merely on empirical delineation. Soil heterogeneity should be identified more objectively, and fields should be divided into roughly homogeneous areas. Currently, the most commonly used data for management zone delineation are soil chemical properties and physical properties, historical yield data, and leaf area index (LAI) [15,16,17].

Researchers in developed countries have studied MZ-based fertilization and set up a relatively systematic variable rate fertilization method. Some scholars divided the research area into three MZs (high, medium, and low vegetation vigor), combined with the existing fertilization model; determined the fertilizer rate for each MZ; and used machinery to implement the spatially variable rate fertilization [18,19]. MZ-based VRF can produce greater benefits, such as increased fertilizer uptake and farmers’ income, and it makes it easy to obtain the data source of management delineation; these facts promote the implementation of spatially variable rate fertilization [20,21]. In addition, MZs contribute to environmentally friendly and sustainable development because of fertilizer input optimization and profit maximization [20].

On the other hand, previous MZ-based VRF studies have only defined MZs [22] and mostly guided seasonal fertilization based on soil and yield data, ignoring the impact of seasonal climatic conditions and farmers’ management measures on crop growth [23]. Secondly, scholars are more inclined to study spatially variable rate N fertilization [8,24,25], although P and K fertilizers also include necessary nutrients for crop growth, and the scientific ratio of N, P, and K plays a very important role in improving crop yield. Thirdly, the growth difference caused by other factors during wheat growth and development was not studied, so this gap did not promote real-time top-dressing fertilization [26]. In addition, the existing literature pays little attention to the potential agronomic, economic, and environmental benefits of spatially variable rate fertilization.

Therefore, the objective of this study was to evaluate the feasibility and potential benefits of MZ-based VRF. This study mainly constructed a basal and top-dressing fertilization model based on before planting MZs and in-season MZs. Before planting, MZs were delineated based on soil nutrients and target yields, using the nutrient balance method, in order to guide the recommendation of basal fertilizer rates. In-season MZs were delineated based on LAI method in order to aid in the recommendation of top-dressing fertilizer rates. In addition, this study assessed the agronomic, economic, and environmental benefits of MZ-based VRF by comparing the fertilization practices of farmers and experts. Two underlying hypotheses are adopted in light of the potential of VRF: (1) MZ-based VRF can decrease the quantity of fertilizer used for wheat and enhance fertilizer efficiency; and (2) compared to URF, MZ-based VRF has more potential agronomical, economic and environment benefits. This study offers scientific guidance and technical support for developing an appropriate village-level VRF based on MZs, while providing fundamental concepts and a basis for large-scale farm management in China.

2. Materials and Methods

2.1. Experiment Site and Design

The study area is located in Zhaoji Village, Jiangsu Province, China (117°56′24″ E, 33°54′36″ N). There were 458 fields, totaling 209 ha of crop land. Each field has an area of 0.45 ha ca. Based on target yield and soil nutrients, the study area was firstly divided into two MZs, which are high-yield (MZ 1) and low-yield zones (MZ 2) [17]. Three fields were selected from each MZ for the verification test of spatially variable rate fertilization, and the three fields were distributed as evenly as possible, as shown in Figure 1a. In addition, in order to explore the feasibility of the application and promotion of this MZ-based variable rate fertilization, the verification test of this study was carried out, and the verified field was approximately 2336.8 m². In the experimental area, the main local cultivar of winter wheat, Xu Mai 33, was mechanically sown on 13 October 2020, with a row spacing of 20 cm, and harvested on 30 May 2021. After analyzing the chemical and physical properties of the soil, we determined that the soil pH ranges from 7.51 to 8.46, soil total nitrogen (TN) ranges from 0.59 to 1.88, soil available phosphorus (AP) ranges from 5.85 to 261.77, and soil available potassium (AK) ranges from 89.00 to 323.15.

Three types of fertilization treatments were assessed at the field level: (1) The farmer program (UR-FP) calculated and used the farmers’ average fertilizer amount based on the survey of landowners in the study area. This treatment is for uniform fertilization; N, P, and K application amounts are 187.5, 112.5, and 56.25 kg ha⁻¹, respectively. (2) The expert program (UR-EP) followed the fertilization treatment promoted by the local Agriculture and Rural Bureau. UR-EP is also for uniform fertilization; the N, P, and K application amounts are 129.98, 101.25, and 50.63 kg ha⁻¹, respectively. (3) The variable rate fertilization program based on management zone (VR-MZ) calculated the basal fertilizer amount based on different target yields and soil nutrients contents, using the nutrient balance method; and the top-dressing amount, using the leaf area index method.

Two replicates were set up for each fertilization treatment, and the design of the protocol is shown in Figure 1b. The fertilization in the experimental area was consistent with that of the local area: P fertilizer was applied once, while N and K fertilizers were applied twice, as both a basal fertilizer and top-dressing one. The basal fertilizer was applied when sowing the winter wheat, while the top dressing was applied during the sprouting of this crop. The other management practices were consistent with those of local farmers.

2.2. Remote-Sensing Data Acquisition

PlanetScope CubeSat constellation was used to obtain 3–5 m spatial resolution satellite images of high quality, with high frequency and global coverage [27]. For this study, 3B-level data products with 3 m spatial resolution before and after winter wheat sprouting in 2018, 2019, and 2020 between late March and early April were obtained from the official website of Planet Lab (https://www.planet.com/ (accessed on 1 January 2021)). These images were used to calculate the normalized difference vegetation index (NDVI) [28] and the optimized soil adjusted vegetation index (OSAVI) [29,30] by using Equations (1) and (2). As Planet’s level, 3B data products are sensor-calibrated, radiometric, orthorectified, and atmospheric-calibrated, and the image quality is exceptional, allowing it to be used directly for relevant research purposes.

$$\mathrm{N}\mathrm{D}\mathrm{V}\mathrm{I}=\frac{R_{NIR}-R_{Red}}{R_{NIR}+R_{Red}}$$

(1)

$$\mathrm{O}\mathrm{S}\mathrm{A}\mathrm{V}\mathrm{I}=\frac{(1+0.16)(R_{NIR}-R_{Red})}{(R_{NIR}+R_{Red}+0.16)}$$

(2)

2.3. Data Processing

2.3.1. Management Zone Delineation

Both basal fertilization and the top-dressing one were carried out in two critical periods for winter wheat. The MZ delineation (before planting) based on soil properties and yield in the years 2018–2020 (Figure 2) guided the recommendation of basal fertilizer rates. Moreover, the growth process of winter wheat is affected by many other factors. Timely monitoring and understanding of the growth information, as well as timely fertilization regulation, are of great significance for increasing the yield. LAI is an important parameter reflecting crop growth and, therefore, plays an important role in crop growth monitoring, yield estimation, and fertilizer regulation [31]. The MZ delineation (in-season) based on LAI guided the recommendation of top-dressing fertilizer rates. In this study, the PlanetScope satellite images of winter wheat at the early jointing stage (4 March) were obtained, so that the enhanced vegetation index (EVI) was calculated and the LAI of the study area was inverted according to the leaf area estimation model proposed by Li et al. [32]. The specific method of MZ delineation was the same one used by Yuan et al. [17].

2.3.2. Recommended Fertilization Algorithm

The principle of nutrient balance (target yield method) is to balance the nutrient supply and demand between the crop and the soil. Based on the difference between fertilizer demand by the crop and fertilizer supply by the soil, the fertilizer amount to be applied to the crop was calculated by using Equation (3) [33].

$$t=\frac{Y_t\times N_u-S_f}{N_f\times t^{\prime }}$$

(3)

where t is the fertilizer amount to be applied (kg ha⁻¹), Y_t is the target yield (kg ha⁻¹), N_u is the nutrient amount (kg) absorbed by 100 kg of grain (fertilizer demand by crop), S_f is the fertilizer supply by soil (kg ha⁻¹), N_f is the nutrient content inside fertilizer (%), and t′ is the fertilizer recovery efficiency (%).

In the above equation, the target crop yield was determined by an average 10% increase in crop yields in the study area over the past three years; the nutrient amount absorbed by 100 kg of grain was determined by referencing the published literature, so that the N, P, and K contents required for 100 kg of grain were 2.8, 1.1, and 2.7 kg, respectively [34,35]. Soil fertilizer supply was determined by laboratory analysis. The nutrient content of fertilizers was specifically deduced from the type of used fertilizer. The soil nutrient correction coefficient was determined according to the local soil nutrient content and the abundance–deficiency index. The correction coefficients for wheat N, P, and K in the study area were 1.2, 1.5, and 0.3, respectively; the fertilizer recovery efficiency (32, 28, and 42% for N, P, and K, respectively) was calculated based on local historical data by using Equation (4).

$$t^{\prime }\left(\%\right)=\frac{u_1-u_0}{T}\times 100$$

(4)

where t′ is the fertilizer recovery efficiency in the season (%), u₁ is the crop absorption in the fertilization zone (kg), u₀ is the crop absorption in the blank area (kg), and T is the total amount of fertilizer containing the nutrient (kg).

The LAI method is often used for spatially variable rate fertilization, which can reduce excessive N use, while increasing the farmer income. The principle hypothesis of the method is as follows: if the current crop LAI is higher than the target value, the amount of fertilizer to be applied is lower than the standard amount, or no fertilization is required, while the opposite is true if the current LAI is lower than the target value, where the mass to be applied is equal to the product of the N required per LAI unit and the increase or decrease in LAI [36]. The LAI was derived from the remote-sensing images based on the following calculation steps (Figure 3): (1) In Step 1, remote-sensing images of wheat before sprouting were obtained, and the LAI values of each MZ were calculated by using the LAI inversion model [32] to set different LAI target values for different MZs. (2) In Step 3, this study was based on expert criteria [35,37], and the crop canopy N requirement was set to 30 kg ha⁻¹ N for LAI unit. (3) In Step 5, due to the difficulty of soil sampling in the reproductive stage, scholars prefer to replace the soil fertility level of the entire growth period with soil nutrients content before crop planting and then guide spatially variable rate fertilization, still obtaining considerable benefits [36,38,39]. Therefore, soil samples were obtained before winter wheat planting, and the soil nutrient content was obtained through laboratory analysis. (4) In Step 6, since it has been reported that the average N recovery rate of wheat on yellow-brown soil is 44.9% under the application of chemical fertilizer [40,41], the crop would absorb 60 kg ha⁻¹ N from the applied 133.33 kg ha⁻¹ N.

2.3.3. Partial Factor Productivity Assessment

The partial factor productivity (PFP), that is, the ratio between crop yield and the applied amount of a specific fertilizer, is an important indicator reflecting the comprehensive effect of soil fertility level and applied rate of the fertilizer in the study area. Therefore, the PFP was calculated to assess the fertilizer application through the fertilization protocol (Equation (5)):

$$\mathrm{P}\mathrm{F}\mathrm{P}=\frac{Grain yield}{Amount of nutrient input}$$

(5)

2.3.4. Economic Benefits Assessment

In this study, the economic benefits of spatially variable rate fertilization were evaluated based on net income, calculated by using Equation (6):

$${\mathit{N}}_{\mathit{I}}={\mathit{W}}_{\mathit{Y}}\times {\mathit{W}}_{\mathit{p}}-\mathit{t}\times {\mathit{F}}_{\mathit{p}}-\mathit{O}\mathit{C}$$

(6)

where N_I is net income, W_Y is wheat grain yield, W_P is grain price, t is fertilizer applied amount, F_P is fertilizer price, and OC refers to other costs.

In this study, the market purchase price of wheat grain was 0.38 USD kg⁻¹ (based on the average exchange rate of 6.45 for 2021, the USD to CNY is calculated), and the unit price of N, P, and K fertilizer was 0.76, 0.74, and 1.08 USD kg⁻¹ (based on pure N, P, and K), respectively. Other production costs included wheat seeds, tillage, pest control, and harvest.

2.3.5. Environmental Benefits Assessment

The empirical models of N₂O emissions, NH₃ volatilization, and NO₃⁻ leaching [34] were used to evaluate the environmental effects of different fertilization treatments by using Equations (7)–(9).

$$Y=0.26e^{0.0045x}(R^2={0.19}^{*})$$

(7)

$$Y=3.21+0.068x (R^2={0.17}^{*})$$

(8)

$$Y=4.93e^{0.0057x}(R^2={0.50}^{**})$$

(9)

where * indicates significance difference at p < 0.05; ** indicates significance difference at p < 0.01.

3. Results

3.1. Basal Fertilization Recommendation Based on Management Zones

After taking into account the MZs as units, the average outputs of each zone for three consecutive years were recorded, and the target output for each MZ was set by increasing the average production by 10%. Therefore, the target output set for the high- and low-yield zones was 8920.75 and 7996.30 kg ha⁻¹, respectively. Combined with the principle of nutrient balance and its parameter settings, fertilization was recommended on a per-field basis. Moreover, ArcGIS software (version 10.3, ESRI Inc., Redlands, CA, USA) was further used to visualize the amount of fertilization and obtain a spatially variable rate fertilization map for N, P, and K, as shown in Figure 4. The recommended amounts of N, P, and K fertilizers in the study area resulted in being highly spatially variable. Among them, the applied rate of P and K was 0 kg ha⁻¹, indicating that their contents in the soil were sufficient to meet the growing needs of the crop and that no P or K fertilizer was required. The basal fertilizer amount to be applied in each field was determined based on the spatially variable rate fertilization map and the survey data of farmers, as shown in

Table 1. Nitrogen, phosphorus, and potassium fertilizer applied in the experimental field.

Fields	Basal Fertilization Rate (kg ha−1)			Top-Dressing Fertilization Rate (kg ha−1)
VR-MZ	N	P	K	N	K
1	79.66	31.70	32.97	80.22	32.97
2	82.69	0	12.50	83.52	12.50
3	65.89	0	0	83.20	0
4	100.22	149.98	75.75	100.22	75.75
5	80.93	41.40	24.70	96.58	24.70
6	166.18	0	51.63	123.98	51.63
UR-FP	146.2	112.5	56.25	142.62	56.25
UR-EP	129.98	101.25	50.63	129.98	50.63

Notes: VR-MZ indicates variable rate fertilization based on MZs; UR-FP indicates uniform-rate–farmer-program fertilization; UR-EP indicates uniform-rate–expert-program fertilization.

.

3.2. In-Season Management Zone Delineation Based on Leaf Area Index

The map of the winter wheat LAI before planting is shown in Figure 5a: it ranged from 1.53 to 3.07, indicating that the growth of winter wheat is highly spatially variable. Specifically, the growth of winter wheat in the northern–western area was the best, with a high LAI, while winter wheat growth in the eastern area was slightly worse. Growth was the worst in the southern–western area, with an LAI within one. Using LAI as the input variable of MZs, the fuzzy C-means clustering algorithm was used to delineate the crop-growth MZs, and the results are shown in Figure 5b. When the number of delineated MZs was two, the fuzzy performance index (FPI) and normalized classification entropy (NCE) reached the minimum at the same time, so the optimal number of delineated MZs was two, and the results are shown in Figure 5c. The first MZ is mainly distributed in the fields northwest and east of the village, while the second MZ is distributed in the fields south and northeast of the village itself. In this study, the rationality of the delineation results was further evaluated using analysis of variance (ANOVA), and the results are shown in Figure 5d. The LAI differed significantly between the two MZs. Winter wheat grew relatively well in MZ 1, with an average LAI of 2.87. In contrast, the overall growth of winter wheat in MZ 2 was poor, with an average LAI of 2.13.

In addition, the division into in-season high- and low-yield MZs dramatically altered the MZs before planting them, as can be seen by comparing Figure 1a and Figure 5c. The majority of locations presented a high yield (in-season MZ 1), instead of a low yield (before planting).

3.3. Top-Dressing Fertilization Recommendation Based on in-Season Management Zones

Based on the results of the division into MZs (Figure 5c), the average LAI for each MZ was 2.87 and 2.04, respectively, and there were significant differences in wheat growth between the two MZs. The maximum LAI values of the high yield level and medium–low yield level were taken as the LAI target values of the two MZs, respectively. Thus, the target LAI value is 6.5 for MZ 1 and 6 for MZ 2. The top-dressing fertilizer amount was calculated by using the LAI method and the results are shown in Figure 6. The N demand of winter wheat at the sprouting stage resulted in being highly spatially variable, and the top-dressing N fertilizer amount was 53.36–167.14 kg ha⁻¹. Specifically, the winter wheat in Fields 1–3 grew well, and the N demand was relatively small. However, the wheat in Fields 4–6 had a high N demand, because two of them required more than 100 kg ha⁻¹ of N fertilizer (Fields 4 and 6).

The top-dressing fertilizer amount to be applied in each field was determined based on the top-dressing application map and the farmer survey data, and the results are shown in Table 1. N and K fertilizers for the farmer treatment and the expert treatment were applied with rates equal to 50% of the total amount, and the same fertilizer amount was applied in each experimental area.

3.4. Potential for Reducing Fertilizer Application Rate and Improving Use Efficiency

Compared to the fertilizer application amount recommended by farmers and experts, the variable rate fertilization treatment has a lower fertilizer amount and, therefore, a higher fertilizer-saving potential (

Table 2. Fertilizer reduction potential of variable rate fertilization compared to farmer and expert fertilization treatments.

Fields	Fertilizer Rate (kg ha−1)			Fertilizer-Saving Potential			Fertilizer-Saving Potential
	VR-MZ			(UR-FP*) (%)			(UR-EP*) (%)
	N	P	K	N	P	K	N	P	K
1	159.88	31.70	65.94	43.95	71.83	41.39	38.50	68.69	34.87
2	166.21	0	24.99	41.73	100	77.79	36.60	100	75.32
3	149.09	0	0	47.73	100	100	42.65	100	100
4	200.43	149.98	151.50	29.73	−33.32	−34.67	22.90	−48.13	−49.63
5	177.51	41.40	49.39	37.77	63.20	56.10	31.71	59.11	51.22
6	290.16	0	103.26	−1.73	100	8.21	−11.60	100	−1.99

Notes: VR-MZ indicates variable rate fertilization based on MZs; UR-FP* indicates fertilizer-saving potential of VR-MZ compared to UR-FP; UR-EP* indicates fertilizer-saving potential of VR-MZ compared to UR-EP.

). It can be found that the reduction potential of P and K fertilizers is high and even equal to 100%. According to the survey data of farmers, the amounts of P and K applied by farmers can reach 112.5 kg ha⁻¹ (Table 2), resulting in a serious waste of P and K fertilizers. Hence, P and K reduction is the key issue of spatially variable rate fertilization. In addition, the N, P, and K fertilizers in Fields 1, 2, and 3 had higher application reduction potentials than those in Fields 4, 5, and 6.

Table 3. Statistical results of yield components and fertilizer partial factor productivity (PFP) in experimental fields.

Fields	Treatment	Yield and Yield Composition				PFP (kg kg−1)			$\Delta \mathbf{PFP}$ (kg kg−1)
		Number of Spikes (104 ha−1)	Spike Grains (Panicle−1)	1000 Grain Weight (g)	Yield (t ha−1)	N	P	K	N	P	K
1	UR-FP	453.02 a *	36.31 a	48.55 a	7.75 a	27.17	68.89	68.89	-	-	-
	UR-EP	445.00 a	36.38 a	48.06 a	7.67 a	29.51	75.75	75.75	-	-	-
	VR-MZ	465.00 a	36.93 a	47.25 a	7.79 a	39.77	245.74	118.14	12.6	176.85	49.25
2	UR-FP	477.18 b	36.29 a	45.34 a	7.91 b	27.31	69.24	69.24	-	-	-
	UR-EP	481.44 ab	35.71 a	48.93 a	8.26 ab	31.62	81.19	81.19	-	-	-
	VR-MZ	510.40 a	34.15 a	48.71 a	8.33 a	50.54	-	336.13	23.23	-	266.89
3	UR-FP	461.72 a	36.84 a	48.11 a	8.15 a	28.57	72.44	72.44	-	-	-
	UR-EP	477.80 a	39.13 a	47.30 a	8.51 a	32.74	84.05	84.05	-	-	-
	VR-MZ	478.80 a	39.88 a	46.04 a	8.54 a	57.28	-	-	28.71	-	-
4	UR-FP	442.73 a	35.75 b	49.05 a	6.97 b	22.12	56.09	56.09	-	-	-
	UR-EP	440.73 a	33.34 ab	47.07 a	6.55 b	25.20	64.69	64.69	-	-	-
	VR-MZ	453.13 a	35.75 a	49.05 a	7.04 a	35.12	46.94	46.47	13.00	−9.15	−9.62
5	UR-FP	386.25 a	35.14 a	46.5 a	6.03 a	21.14	53.60	53.60	-	-	-
	UR-EP	401.88 a	35.93 a	48.7 a	6.34 a	25.39	65.19	65.19	-	-	-
	VR-MZ	403.75 a	36.18 a	44.56 a	6.27 a	33.41	143.24	120.06	12.27	89.64	66.46
6	UR-FP	434.38 a	37.09 a	47.95 a	6.77 a	23.74	60.18	60.18	-	-	-
	UR-EP	451.88 a	34.48 a	46.51 a	6.58 a	25.24	64.79	64.79	-	-	-
	VR-MZ	468.13 a	37.34 a	45.49 a	6.78 a	23.37	-	65.66	−0.37	-	5.48

Notes: PFP indicates partial factor productivity; UR-FP indicates farmer program fertilization; UR-EP indicates expert program fertilization; VR-MZ indicates variable rate fertilization based on MZs; * the different letters indicate that there are significant differences at p < 0.05.

compares the effects of different fertilization treatments on yield and yield components and their effects. Overall, the yield and yield components obtained through variable rate fertilization were not significantly different from those of the farmer and expert treatments. Specifically, in Fields 1, 3, 5, and 6, the yield obtained by variable rate fertilization was slightly higher than that of the other two fields, but the difference was not significant. In Field 2, the yield obtained by variable rate fertilization was 8.33 t ha⁻¹, which was significantly higher than that obtained by the farmer treatment and not significantly different from that of the expert treatment. In Field 4, variable rate fertilization achieved a wheat yield of 7.04 t ha⁻¹, which was significantly higher than that of the farmer and expert treatments. In addition, the results of ANOVA (Figure 7b) showed that there was no significant difference in yields among all three fertilization treatments within the same MZ. However, the yields in the high-yield zones were significantly higher than those in the low-yield zones.

As can be seen from Table 2 and Table 3, variable rate fertilization generally yields higher fertilizer PFPs. In Field 6, the N partial productivity (PFPN) obtained by variable fertilization was reduced by 0.37 kg kg⁻¹ compared to farmer treatment. In Filed 4, the P partial productivity (PFPP) obtained by variable fertilization was 9.15 kg kg⁻¹ lower than that of the farmer treatment. In Field 4, variable rate fertilization reduced K partial productivity (PFPK) by 9.62 kg kg⁻¹ compared to the farmer treatment. In summary, the highest advantage of variable rate fertilization is not to significantly increase the yield but to highly improve the fertilizer application under the premise of ensuring yield.

3.5. Economic and Environmental Benefits

The economic benefits of different fertilization treatments are shown in Figure 7c,d. Compared to the income obtained by farmer treatment, variable rate fertilization can increase income by 15.5–449.61 USD ha⁻¹, and the income increase potential is 1.04–20.09%. Compared to the benefits obtained by expert treatment, variable rate fertilization can increase the income by 139.53–279.07 USD ha⁻¹, and the income increase potential can reach 9.42–11.79% (Figure 7d). It can be seen that variable rate fertilization has a great potential to increase net income. In Fields 1, 2, and 3 (high-yield zones), ANOVA showed that variable rate fertilization significantly increased the net income, compared to the other two fertilization treatments, but there was no significant difference in the net income obtained by farmer and expert treatments. In Fields 4, 5, and 6 (low-yield zones), the net income of variable rate fertilization was slightly higher than that of the other two treatments but not at a significant level, thus indicating that variable rate fertilization produced greater economic benefits.

The N₂O emission, NH₃ volatilization, NO₃⁻, and total N pollution by different fertilization treatments are shown in Figure 8. It can be seen that, compared to the fertilization amounts of farmer and expert treatments, variable rate fertilization has a higher emission reduction potential, and the results are shown in Figure 9. Compared to farmer treatment, variable rate fertilization can reduce N₂O emission by 31.72–45.81%, NH₃ volatilization by 32.40–51.06%, and NO₃⁻ leaching by 38.33–53.98%. Compared to the expert treatment, variable rate fertilization reduced the total N pollution by 24.15–41.52%. In Field 6, variable rate fertilization increased the total N pollution by 1.48–18.79% compared with the other two treatments. In summary, variable fertilization can effectively reduce N pollution, and, therefore, it contributes to environmental protection and sustainable development.

4. Discussion

4.1. Factors of Spatial Variability in Crop Growth

Crop growth is affected by a combination of factors, such as the farmers’ management measures (applied amounts of fertilizer and irrigation), soil fertility level, soil type, and climatic conditions, resulting in high growth differences, both temporally and spatially [42]. The spatial variability of LAI in the early stage of winter wheat sprouting (Figure 5a) showed that the crop growth in the high-yield zone was better than that in the low-yield zone, indicating that the soil fertility level was the prerequisite for the growth and development of winter wheat, in agreement with previous studies [43,44,45].

However, there were many fields with poor growth in the high-yield zones, mainly in the southern–western and northern–eastern areas. According to the surveys of farmers, irrigation was not carried out before and after wheat planting, due to the lack of ditches and ponds for irrigation in the eastern and southern parts of the study area. Coupled with low rainfall during the season, the wheat in these fields suffered water shortage. A previous study simulated crop growth dynamics and crop yield under different water conditions through crop models and found that water scarcity does not promote nutrient uptake, thus negatively affecting crop growth and grain yield [46]. Therefore, in the high-yield zone of this study, the amount of irrigation water is the main factor affecting crop growth.

Moreover, soil properties have an important impact on nutrient leaching and, therefore, nutrient retention [47]. In the Eastern part of the study area, the soil was mainly sandy, with poor water and nutrient retention capacity. The soil cannot retain enough water and nutrients to meet the needs of crop growth, resulting in slow crop growth in the eastern area. It can be seen that farmers’ management measures, especially irrigation measures, as well as soil type, have a great impact on crop growth.

In addition, there were some fields where a good growth was achieved in the low-yield zones (Fields 4 and 6). This result could be mainly explained by VRF based on before planting MZs. It may also be partially explained by the fact that other management practices, e.g., sowing rates and time, still have an impact on winter wheat growth [48].

4.2. Variable Rate Fertilization Techniques Based on Management Zone

Understanding soil nutrient spatial variability and crop growth differences is a prerequisite for decision making on variable rate fertilization [42]. Previous studies have proved that the MZ method is an important and effective means to guide VRF [9,20]. In this study, MZs delineated before wheat sowing and during the wheat growing period (before sprouting) for basal fertilizer and top-dressing recommendations, respectively. The in-season MZs were delineated based on the real-time growth status of the crops, reflecting the actual growth status of crop. The detected similarities and certain differences detected indicated that crop growth is not only affected by soil fertility levels but also by external factors such as weather conditions and management measures [49].

Many scholars have demonstrated the potential of a single factor to guide variable rate fertilization [38,50]. However, there are some fields (Fields 4 and 6) with good growth in the low-yield zone (Figure 5a). It is believed that the soil base fertility and real-time crop growth should be comprehensively considered for decision making on variable rate fertilization to better meet the nutrient requirements of crops on different spatial units. VRF, which integrates soil information and crop growth, determines higher yields and N use than a single-factor approach [21], in agreement with this study.

Scholars have guided variable rate fertilization by combining the results of delineating MZs with the target yield method or using accurate fertilization algorithms based on active canopy sensor/remote sensing, and they have achieved considerable benefits. However, these studies usually set spatially uniform target yield values for MZs [21]. In this study, it was considered that different target values should be set for different MZs, because yield levels vary widely among MZs. This phenomenon may be influenced by management measures or determined by the internal structure of the soil. Short-term fertilization regulation can hardly change the internal structure of the soil and, in turn, narrow productivity differences in different MZs. Even short-term excessive fertilization may not effectively increase crop yields but may cause a waste of fertilizer. Fertilization based on variables with different yield targets would obtain higher fertilizer use efficiency and economic benefits than those based on the same yield targets [20], in agreement with this study. In addition, studies have shown that variable rate fertilization algorithms based on MZs can reduce the difference in rice yield by an average of 3.7% and the gap in N fertilizer use efficiency by an average of 63.5% [51]. It was shown that the variable rate fertilization algorithm based on the MZs could effectively narrow the difference between crop yield and N fertilizer applied amount in the MZ and make the crop growth and yield in the MZ prone to being spatially uniform. However, there is still a need to develop a more universally recommended fertilization algorithm than the classic one, considering various factors affecting crop growth and development, such as climate conditions, varieties, and crop management measures.

4.3. Potential Benefits of Variable Rate Fertilization

Variable rate fertilization is a practice that implies that different amounts of nutrients are required by crops on different spatial units; thus, this fertilization method has the potential to highly reduce fertilizer use. The nutrient reduction potential of VR-MZ indicated that the management experience of smallholder farmers has led to an excessive use of fertilizer, which would lead to soil acidification, alkalinization, and the accumulation of harmful substances, which hinders soil sustainability [52]. However, not all high-yield zones showed a positive fertilizer reduction potential. Negative fertilizer reduction potential indicates that the amount of applied fertilizer in the variable rate fertilization treatment is even higher than that used by farmers. This is due to the serious lack of nutrients in the soil of some fields, thus requiring a high amount of fertilizer to maintain basic crop growth.

This study found that there was no significant difference in yield between VR-MZ and the other two treatments, but the nutrient use efficiency was highly improved. It shows that the ultimate goal of variable rate fertilization is not to significantly increase yield or blindly reduce the applied fertilizer amount but to scientifically allocate fertilizer according to the actual needs of crops to maximize its use efficiency, in agreement with many other studies [10,53]. In addition, variable rate fertilization can effectively improve the economic return of farmers, while reducing environmental pollution and, therefore, contributing to environmentally sustainable development [38]. Moreover, this study found that the fertilizer reduction potential, the benefit of income increase, and the emission reduction potential of high-yield zones were higher than those in low-yield zones. This is because farmers easily identify the location of low-yield fields based on years of planting and management experience and focus on high-yield fields that may bring higher economic returns and ensure high crop yields by giving enough or even excessive fertilizers. This is the reason why high-yield zones are usually the hotspots of nutrient accumulation. Nonetheless, situations that arise in low-yield zones cannot be disregarded. In Field 6, for example, the recommended fertilizer amount under the VR-MZ treatment resulted in being increased despite a decreased N emission (Figure 8 and Figure 9). This could be attributed to the fact that the target yield in low-yield zones was uniformly set at 7.99 t ha⁻¹, disregarding the individual differences in Field 6. Consequently, the target yield set for Field 6 was too high, resulting in N loss. Furthermore, one of the causes of the N loss was the relatively high LAI target value (6.5) that was selected for the top-dressing fertilization based on in-season MZs. This indicates that the field management practices should be taken into account when delineating MZs in future studies on MZ-based VRF.

5. Conclusions

This study demonstrates the potential of spatially variable rate fertilization for winter wheat crops. By taking into account the spatial variability in crop growth status across the research area, it was possible to delineate two distinct management zones and use the LAI as a guide to determine the optimal amount of top-dressing fertilizer for each zone. The results indicated that this approach could lead to a significant reduction in fertilizer usage without any negative impact on crop yield. Additionally, the improved fertilizer use efficiency can have positive economic and environmental effects, such as the reduction of N pollution and promotion of sustainable agricultural practices. The issue of how MZ-based VRF affects soil health is an intriguing one which could be usefully explored in future research.