Fertilization Highly Increased the Water Use Efficiency of Spring Maize in Dryland of Northern China: A Meta-Analysis

Abstract

Water and fertilizer play an important role in crop growth in dryland areas. It is a necessity to improve the water use efficiency (WUE) of the crop once the water resource is limited. In northern China, where there is a wide shortage of water resources, it is therefore necessary to investigate how fertilization affects the WUE of spring maize and to quantify the effects. A total of 33 published peer-reviewed papers were collected, and a meta-analysis and random forest model analysis were performed with 364 WUE comparisons, aiming to explore the effects of fertilization on the WUE of spring maize and to clarify the optimal conditions for WUE under fertilizer management. The results showed that fertilization significantly increased the WUE of spring maize by 56.72% (P < 0.01) when compared with non-fertilization. The WUE effect under the organic–inorganic fertilizer combination (MNPK) was approximately twice as high as that under inorganic fertilizer (NPK) or organic fertilizer (M). The greatest increase in WUE occurred at 0–100 kg ha⁻¹ of nitrogen application (NA). Under environmental conditions including 7 ≤ mean annual temperature in the test year (T) ≤ 10 °C, 400 ≤ mean annual precipitation in the test year (P) ≤ 600 mm, and mean altitude (A) > 1500 m, and soil conditions including 10 ≤ soil organic matter content (SOM) ≤ 14 g kg⁻¹ and available phosphorus (AP) < 5 mg kg⁻¹, the fertilization optimally enhanced the WUE of spring maize when the agronomic measures of ridge–furrow planting (RFP) and mulching film (MF) were used. The random forest model analysis indicated that the influence factors (i.e., fertilizer regimes, environmental factors, soil factors, and agronomic measures) caused 65.62% of the variation in spring maize WUE effects, while in all influence factors, fertilizer types related to fertilizer regimes caused the most variation. The initial available potassium (AK) and available nitrogen (AN) of the soil were negatively correlated to the WUE effect, indicating that fertilization imposed a better effect on the WUE of spring maize when the soil was infertile. Fertilization significantly increased the WUE of spring maize, and organic and inorganic fertilizer application provided an effective measure for the sustainable development of spring maize in northern China. After clarifying the required conditions for fertilization increasing WUE, high-efficiency water use may be achieved.

1. Introduction

In China and other countries with water shortages, the amount of water used for agriculture has decreased continuously in recent years. Water resource shortage has become one of the main factors impeding the development of agriculture [1,2,3]. Grain production remains insufficient for the needs of the increasing population and the water resource shortage is becoming more and more serious; these issues are challenging China today. Improving the water use efficiency (WUE) of crops may significantly benefit agricultural production and food security [4].

Crop WUE represents the grain yield produced per unit of evapotranspiration during the growing season [5]. WUE may be influenced by many factors, e.g., climate, management, and soil properties. Fertilization is a widely used management measure to increase crop productivity and is seen in almost all agricultural areas [6]. Fertilizers are classified into two main types according to their properties, including organic ones (e.g., manure, root stubble, and crop residues), and inorganic ones (e.g., nitrogen, phosphorus, potassium, and a combination of nitrogen, phosphorus, potassium) [7,8]. By carrying out an experimental investigation for as long as 12 years in Heyang, Shaanxi Province, northern China, Li et al. (2023) found that balanced fertilization (NPK) could reduce soil bulk density, increase soil porosity, and improve soil structure, leading to an increase in soil water content and crop WUE [9]. Fan et al. (2005) carried out a long-term fertilization experiment in the Loess Plateau of China, and the results showed that the combined application of organic and inorganic fertilizers significantly upgraded soil nutrients, soil fertility, and the water productivity of crops [8]. However, the negative effects of fertilization on soils and crops are also frequently reported. Excessive use of nitrogen fertilizer may cause a loss of soil organic matter (SOM) and a deterioration of soil quality [10,11,12], consequently reducing crop WUE. Excessive application of organic fertilizer induces an accumulation of soil salt, which may make it difficult for crop roots to absorb water, thus resulting in crop ground parts wilting and a reduction in crop yield, suppressing WUE [13]. It has also been revealed that the excessive application of organic fertilizer may strengthen the antagonism between different ions during the process of nutrient element absorption of crops. As a consequence, the absorption of other nutrient ions by crops decreases, thus lowering the quality and yield of crops as well as the WUE [14]. The inconsistency of the results mentioned above somewhat highlights the importance of how fertilization affects WUE which may determine the feasibility of fertilization to improve the crop yield.

Maize is the main cereal crop in China and is mainly planted in northern China. The planting area of maize with one harvest per year is rapidly increasing [15]. Under dryland farming, the input–output ratio of spring maize is relatively high. Traditional irrigated agriculture is facing a great threat due to climate change and the over-consumption of groundwater. Dryland agriculture is becoming promising for agricultural production [3]. Dryland agriculture produces 60% of the world’s food with about 80% of the world’s arable land, and thus dryland agriculture has a significant effect on the world’s food security [16]. The total arable land in China is about 13 million hm² of which more than 60% is only suitable for dryland agriculture and is distributed in northern China [17]. The water consumption of dryland crops during the growth period is high while the WUE remains at a low level. Therefore, revealing the effect of fertilization on dryland spring maize WUE in northern China may provide a basis for the high-efficiency use of the limited water resources.

Studies on spring maize WUE in dryland areas have been widely carried out. However, most of the published studies focused on the experimental investigation of a single site, and comprehensive effects analyses based on test data collected from multiple sites are seldom performed. Meta-analysis provides a possible way to perform an integrated analysis of the test data from multiple independent experiments which have the same research purpose [18]. So far, the effects of tillage methods on maize WUE [19], the effects of drip irrigation on crop WUE [20], the effects of farmland mulching on crop WUE [21,22], and the effects of organic fertilizer on wheat WUE [23] all have been explored. However, studies investigating the effects of fertilization on the WUE of dryland spring maize and how these effects have been influenced in the farmland systems in northern China have not been widely performed. In this study, a meta-analysis and a random forest model were used to quantify the effects of fertilization on dryland spring maize WUE in northern China. Specifically, the objectives of this study were (1) to quantify the effects of fertilization on spring maize WUE in northern China; (2) to reveal how the effects of fertilization on WUE vary with environmental conditions and management practices; (3) to clarify the optimal conditions for spring maize WUE under fertilization, and to provide possible ways to further improve spring maize WUE.

2. Materials and Methods

2.1. Data Search and Collection

The published papers have been collected from several databases, i.e., Web of Science (http://apps.webofknowledge.com/) (accessed on 10 November 2022), China National Knowledge Infrastructure (http://www.cnki.net/) (accessed on 10 November 2022), and Baidu Academic (http://xueshu.baidu.com/) (accessed on 10 November 2022), and were identified using a combination of keywords (fertilization, maize, water use efficiency, or water use), in order to investigate the effects of fertilization on spring maize WUE. The year of the publications ranges from 1990 to 2022. Non-English published journals were excluded except for the journals in Chinese. Each paper must meet the following criteria: (1) The experiment was carried out in Chinese fields, including fertilized treatment and non-fertilized control, and no irrigation was used during the growing season. (2) The means of spring maize WUE, as well as standard deviation (standard error) and sample sizes, could be extracted directly from the papers or calculated indirectly; if the standard deviation was not reported in the literature, SD = 5% *Mean. (3) The mean annual precipitation (P), mean annual temperature (T), mean altitude (A), planting pattern, mulching management, initial soil (0–20 cm) organic matter (SOM), available phosphorus (AP), available potassium (AK), available nitrogen (AN), and fertilizer application amounts (NA) were recorded in the test year. In case of a lack of reported climatic information in the paper, this information was obtained from the nearest weather station (China Weather Data Network, http://data.cma.cn/) (accessed on 5 December 2022). (4) For the experimental data published several times, the paper with more complete data was selected. After the screening, 33 publications met the criteria, including 4 in English and 29 in Chinese. If the data were in the form of pictures, they were extracted with GetData Graph Digitizer software (Russian Federation). A total of 364 sets of data came from 18 points, and the specific point distribution and experimental data are shown in Figure 1 and in the Supplementary Materials. The dominant climate type in northern China is a monsoon climate, mainly including semi-arid zones and temperate zones. The P ranges from 249.3–718.5 mm and the T ranges from 1.5–13.0 °C. Spring maize is usually sown in April and harvested in September, and the commonly used agronomic measures include MF and RFP.

2.2. Data Classification

To investigate the effects of different factors on the WUE of spring maize with fertilization, factors with a sufficient number of samples (≥5) were selected from the established database and then they were categorized into 4 groups, i.e., fertilization regimes, environmental factors, soil factors, and agronomic measures [24]. The fertilization regimes included NA and fertilizer types. The NA was divided into four ranges, i.e., NA = 0 kg ha⁻¹, 0 < NA ≤ 100 kg ha⁻¹, 100 < NA ≤ 200 kg ha⁻¹, and NA > 200 kg ha⁻¹ [23,25]. There were 16 fertilizer types in total while only 8 representative types with a sufficient number of samples were considered during analysis. Fertilizer types included nitrogen only (N), nitrogen and phosphorus (NP), nitrogen and phosphorus and potassium (NPK), organic fertilizer only (M), organic fertilizer plus inorganic nitrogen (MN), organic fertilizer plus nitrogen and phosphorus (MNP), and organic fertilizer plus nitrogen, phosphorus, and potassium (MNPK), as well as straw return plus nitrogen and phosphorus (SNP). Environmental factors included the regions, T, P, and A for the test year. T, P, and A were classified following the suggestions in published papers [25,26] and then made consistent with the database in this study. Three main regions were included in the analysis, i.e., northwest (NW), northeast (NE), and north China (NC). P included 3 ranges, i.e., P < 400 mm, 400 mm ≤ P ≤ 600 mm, and P > 600 mm. T also had 3 ranges of T < 7 °C, 7 °C ≤ T ≤ 10 °C, and T > 10 °C, as did A, i.e., A < 1000 m, 1000 m ≤ A ≤ 1500 m, and A > 1500 m. Soil factors included SOM, AP, AK, and AN. SOM and AP were classified into 3 levels according to the second soil survey in China and previous publications [27,28]. Regression analysis was used to further illustrate the relationship between AN, AK, and WUE. SOM was deemed to be low when it was less than 10 g kg⁻¹, and when SOM was greater than 14 g kg⁻¹ then the soil was considered as fertile, while a moderate SOM ranged between 10 and 14 g kg⁻¹. AP was divided into 4 main levels, low (i.e., AP < 5 mg kg⁻¹), moderate (i.e., 5 ≤ AP ≤ 10 mg kg⁻¹), high (i.e., 10 < AP ≤ 20 mg kg⁻¹), and very high (i.e., AP > 20 mg kg⁻¹). Agronomic measures included planting patterns and mulching management. The planting pattern further included ridge–furrow planting (RFP) and traditional flat planting (TFP) while the mulching management included mulching film (MF) and no mulching film (NMF). Specifically, MF represented that all or part of the soil surface was covered by film while NMF represented no soil surface covered by film. The main data from the 33 published papers are shown in

Table 1. Main data from 33 published papers.

Number	Reference	Province	Experiment Year	Sample Sizes	Fertilizer Types	NA (kg ha−1)
1	Zhang et al., 2001	Liaoning	1997–1998	3	CK, NP, MP, MNP,	0, 300
2	Zhang et al., 1992	Liaoning	1988–1990	3	CK, NP, MP, MNP	0, 150, 300, 450
3	Chen et al., 2021	Gansu	2018–2019	3	CK, NPK, MNP, MNPK	0, 200
4	Zhang et al., 2012	Shanxi	2010	4	CK, NPK	0, 150, 300, 450
5	Zhang et al., 2013	Shanxi	2010	3	CK, NPK	0, 150, 300, 450
6	Yan et al., 2022	Gansu	2018–2019	3	CK, NP, MNP, MP, SP	0, 200
7	Dang et al., 2015	Gansu	2013	3	CK, N, MN	0, 180
8	Guo et al., 2015	Gansu	2012–2014	3	CK, NP, NPK	0, 150, 225
9	Zhang et al., 2015	Gansu	2011–2013	3	CK, NP, NPK	0, 150, 225
10	Gao et al., 2015	Jilin	2012	3	CK, NPK, SNPK, MNPK	0, 165, 112
11	Chen et al., 2012	Shanxi	2010	3	CK, N, NPK, MN, MNPK	0, 240
12	Zou et al., 2009	Heilongjiang	2007	4	CK, NP, MNP	0, 150
13	Zou et al., 2012	Heilongjiang	2010	4	CK, NP, MNP	0, 150
14	Meng et al., 2005	Heilongjiang	1994/1997/2000	3	CK, NP, MNP	0, 96, 138
15	Shang et al., 2010	Shaanxi	2008	3	CK, NP, NPK	0, 225, 150
16	Xu et al., 2014	Gansu	2011–2012	3	CK, NP	0, 120, 150, 180, 210, 240
17	Wang et al., 2009	Shaanxi	2007–2008	3	CK, M	0
18	Zhang et al., 2011	Gansu	2008–2010	3	CK, NP	0, 180
19	Li et al., 2022	Gansu	2020	3	CK, NP	0, 120, 150, 180, 210, 240
20	Wang et al., 2017	Shaanxi	2013–2015	3	CK, NP, NPK	0, 75, 225, 150
21	Han et al., 2020	Gansu	2018	3	CK, PK, NPK, NK, NP	0, 138, 276, 414
22	Wang et al., 1994	Gansu	1992	2	CK, P, N, NP, M, PM, MN, MNP	0, 240
23	Zhao et al., 2018	Shanxi	2017	3	CK, NP, NPK	0, 80, 355
24	Huang et al. 2013	Shaanxi	2007–2008	3	CK, NP, M	0, 230
25	Wang et al., 2010	Gansu	1979/1980/1985/1986/1991/1992/2005/2006	3	CK, N, NP, SNP, M, MNP	0, 90
26	Pang et al., 2021	Ningxia	2012–2016	3	CK, NP	0, 150, 300, 450
27	Bai et al., 2018	Liaoning	2014–2015	3	CK, SPK, SNPK, NPK	0, 113, 225, 338
28	Zhou et al., 2004	Shanxi	1992–2001	10	CK, NP, MNP, M	0, 60, 120, 180, 240
29	Zhang et al., 2019	Gansu	2016–2018	3	CK, NP, SP, SNP	0, 225, 180, 135
30	Li et al., 2018	Ningxia	2015–2016	3	CK, NP	0, 100, 200, 300
31	Zhang et al., 2020	Ningxia	2014–2017	3	CK, NP	0, 117, 173, 229, 285
32	Wu et al., 2022	Shaanxi	2019–2020	3	CK, NP	0, 225
33	Liu et al., 2007	Shaanxi	1997–1998	3	CK, MNP	0, 120, 240

and basic information on regional distribution is shown in

Table 2. Basic information on regional distribution.

Regions	P (mm)	T (°C)	A (m)	Fertilizer Types	Planting Pattern	Mulching Management
NW	249.3–643.1	6.1–13.0	521–2000	CK, N, P, NP, PK, NPK, M, MN, MP, MNP, MNPK, SP, SNP	TFP and RFP	NMF and MF
NE	303.7–718.5	1.5–6.6	106–316	CK, NP, NPK, MP, MNP, MNPK, SPK, SNPK,	TFP	NMF
NC	438.5–479.0	6.0–9.1	858–1500	CK, N, NP, NPK, M, MN, MNP, MNPK	TFP	NMF and MF

.

2.3. Data Analysis

2.3.1. Meta-Analysis

The WUE of spring maize which has a unit of kg m⁻³ was calculated using [29]

$$WUE=\frac{Y}{10ET}$$

(1)

$$ET=Pg+S{W}_s-S{W}_h$$

(2)

where Y represents spring maize grain yield which has a unit of kg ha⁻¹, while ET represents crop water consumption during the growing period which has a unit of mm. In Equation (2), $Pg$ is the precipitation during the growing period which has a unit of mm, while SW_s and SW_h are, respectively, the soil water storage before sowing and after harvesting.

The natural logarithm of the response ratio, i.e., LnRR, a quantitative analysis method in the meta-analysis, was adopted to estimate the effect of fertilization on spring maize WUE in this study. This method standardizes the treatment group/control group of each independent experiment so that each result can be quantified. The traditional analysis methods can only reach qualitative conclusions. The effect size of each study can be compared in a meta-analysis, which improves the accuracy and has more practical significance.

The calculation formula Is shown by Equation (3) [18].

$$LnRR=Ln\left(\frac{X_t}{X_c}\right)=Ln{X}_t-Ln{X}_c$$

(3)

In Equation (3), X_t represents the mean WUE of spring maize under fertilization treatment, while X_c represents the mean WUE of spring maize under non-fertilization obtained from control groups.

The weighted mean method was used to improve the precision as it may mitigate the possible statistical differences in each independent study [30]. Each independent experiment was assigned a weight of ω_i (inverse of the variance) as shown in Equation (5), and the response ratio of the weights, i.e., LnRR₊₊, was calculated using Equation (6). The standard error of LnRR₊₊ and the 95% confidence interval (i.e., 95% CI) were calculated using Equations (7) and (8), respectively.

$$\nu =\frac{S_t{}^2}{n_t{X}_t^2}+\frac{S_c{}^2}{n_c{X}_c^2}$$

(4)

In Equation (4), S_t and S_c, respectively, represent the standard deviations of spring maize WUE in the groups with fertilization treatment and control groups in which the fertilization was absent. Here, n_t and n_c represent the numbers of replications of treatment groups and control groups; $\nu$ indicates the variance of the LnRR, and v_i in Equation (5) indicates the variance of the LnRR of each independent study.

$${\omega }_i=\frac{1}{{\nu }_i}$$

(5)

$$LnR{R}_{++}=\frac{{{\displaystyle \sum }}_{i=1}^k{\omega }_{i}LnR{R}_i}{{{\displaystyle \sum }}_{i=1}^k{\omega }_i}$$

(6)

$$S\left(LnR{R}_{++}\right)=\sqrt{\frac{1}{{{\displaystyle \sum }}_{i=1}^k{\omega }_i}}$$

(7)

$$95\%CI=LnR{R}_{++}\pm 1.96S\left(LnR{R}_{++}\right)$$

(8)

The results were converted into a percentage using Equation (9).

$$Z=\left(e^{LnR{R}_{++}}-1\right)\times 100$$

(9)

If there was no overlap between the 95% CI and 0, fertilization was then considered as having a significant effect on the WUE of spring maize. A Z greater than 0 indicated that fertilization increased the WUE, while a Z less than 0 indicated the opposite.

$$Q={\displaystyle \sum }_{i=1}^k{\omega }_i{(LnR{R}_i)}^2-\frac{{\displaystyle {{\displaystyle \sum }}_{i=1}^k}{\omega }_i{(LnR{R}_i)}^2}{{\displaystyle {{\displaystyle \sum }}_{i=1}^k}{\omega }_i}$$

(10)

The Q statistic test was adopted to examine whether the independent studies were heterogeneous. The examined statistic obeyed the chi-square distribution with freedom of K-1. Therefore, after the calculation of Q, the probability (P) could be obtained from the chi-square distribution. When P < 0.05, the random effects model was used; otherwise, the fixed effects model was adopted. If P < 0.05 occurred among subgroups, it indicated that the mean effect between subgroups was significantly varied. Statistical software Metawin2.1 (Sinaure Associates Inc., Sunderland, MA, USA) was used to achieve the above process with 4999 iterations [31,32,33].

2.3.2. Random Forest Model

The random forest model, a nonlinear non-parametric machine learning regression algorithms method, was used to build a forest randomly [34]. The built forest was composed of multiple decision trees, and each decision tree did not correlate. As a machine model for combinatorial decision-making, the random forest has the following advantages [35,36]: (1) it has strong anti-noise ability and model generalization ability; (2) the importance of evaluating indicators in training and testing can effectively overcome local overfitting; and (3) it has strong adaptability to data sets and is able to process high-dimensional and multi-class data. WUE includes a vast system of farmland which is affected by environmental conditions, soil properties, fertilization regimes, and agronomic practices. The random forest model was suitable to evaluate the importance of influencing factors (the influential factors are detailed in Section 2.2) on WUE effects. This study was based on the random forest package in R software, and the tidyverse function was used for the data processing. IBM SPSS Statistics 26 (Chicago, IL, USA) was used for the significance analysis of data.

3. Results

3.1. Overview of Dryland Spring Maize WUE

The spring maize WUE in fertilized and unfertilized conditions followed the normal distribution, as shown in Figure 2. When fertilizer was applied, the maize WUE of northern China ranged from 0.2 to 5.28 kg m⁻³, with an average of 1.98 kg m⁻³, while for the non-fertilization condition, the WUE ranged from 0.17 to 4.83 kg m⁻³, having an average of 1.37 kg m⁻³.

WUE significantly varied in different regions (Figure 3a), and the WUE with fertilization remained considerably greater than that with non-fertilization. For the fertilization condition, WUE was minimal in the NE where the mean WUE was 1.25 kg m⁻³. The mean WUE in the NW and NC were, respectively, 2.14 kg m⁻³ and 1.97 kg m⁻³, and the difference between them was insignificant. A similar trend was observed in the non-fertilization condition. The WUE of spring maize in northern China also varied with different fertilizer types (Figure 3b), in which the WUE was the largest under MNPK application, approaching 2.96 kg m⁻³, being twice the WUE under N fertilizer or organic fertilizer alone.

3.2. Effects of Fertilization on Dryland Spring Maize WUE Using Meta-Analysis

3.2.1. Overall Effects

The dataset of spring maize WUE with fertilization had high heterogeneity (P_Q < 0.01), and consequently, the random effect model was used (

Table 3. Overall effects of fertilization on spring maize WUE and publication bias.

Item	Model	Increase Rate (%)	95% Confidence Interval (%)	Q	PQ	B	PB
Water use efficiency	Random effect model	56.72	50.97–62.68	433.95	0.006	0.3316	0.0805

Q represents the statistics of heterogeneity; P_Q represents the significance of Q; B represents the statistics of publication bias; and P_B represents the significance of B.

). When compared with no fertilization, WUE increased by 56.72% once the fertilizer was applied. The lower and upper limits of the 95% CI were, respectively, 50.97% and 62.68%. As P_B > 0.05 as given in Table 3, there was no publication bias for WUE, and thus it remained consistent with the normal distribution shown in Figure 2.

3.2.2. Fertilization Regimes

WUE was improved by four different NA (Figure 4). The increase rate of WUE was 39.93, being the lowest when the NA was absent. When the NA ranged from 0 to 100 kg ha⁻¹, WUE had the greatest increases, approaching 73.83%. Fertilizer types (Figure 4) had a great effect on WUE, which increased by 25.93%, 61.67%, and 46.18%, respectively, when only the chemical fertilizer of N, NP, and NPK was applied. When only the organic fertilizer, i.e., M, was applied, the increase in WUE was 56.48%. The beneficial effects of chemical fertilizer alone and organic fertilizer alone remained much lower than the application of combined organic and inorganic fertilizers, e.g., the application of MNPK led to an increase in WUE of 116.28%.

3.2.3. Environmental Factors

All environmental factors imposed a significant effect on WUE under fertilization management conditions, usually in a beneficial way (Figure 5). When compared with no fertilizer application, WUE increased the most in NC, i.e., 69.06%. The effect value of WUE increased first and then decreased with the rise in precipitation (P). The greatest increase in WUE, i.e., 61.95%, was observed when P ranged between 400 and 600 mm. The effect value of WUE seemed to be stable during the initial variation in temperature (T), while a rapid decrease occurred along with the following increase in T. At a temperature of 7–10 °C, the largest positive effect of fertilization on WUE was attained, causing an increase in WUE of 62.86%. With the increase in altitude (A), WUE continuously increased, reaching a maximum effect value of 81.54%. When samples were compared between environmental factors, the management practice (i.e., fertilization regimes or agronomic measures) was not the same.

3.2.4. Soil Factors

SOM significantly affected the spring maize WUE in northern China (Figure 6). Fertilization increased WUE by 56.44%, being the greatest when the SOM was 10–14 g kg⁻¹. The WUE effect significantly decreased and then became stable with the increase in AP. When AP was less than 5 mg kg⁻¹, the maximum increase in WUE was attained to be 91.36%. The WUE effects of spring maize showed a negative linear correlation with both the initial AN and AK of the soil significantly (Figure 7). It further indicated that the response of spring maize WUE to fertilization was greater when the initial soil fertility was low.

3.2.5. Agronomic Measures

The effects of fertilization on spring maize WUE varied when different mulching management and planting patterns were involved (Figure 8). Planting patterns had a significant effect on WUE improvement, causing an increase of 70.57% in WUE under the RFP in comparison with non-fertilization, and an increase of 20% when compared with TFP (P < 0.01). The positive effect of MF on WUE was greater than that of NMF. The increase in the WUE of spring maize with fertilization under MF was 64.07%.

3.2.6. Importance Analysis of Factors

A random forest model was adopted to examine the importance of the factors influencing the WUE effects of dryland spring maize under fertilizer management (Figure 9). This model was able to explain 65.62% of the variation in the WUE effect values of spring maize. The environmental factors, soil factors, fertilization regimes, and agronomic measures, respectively, explained 32.61%, 39.63%, 16.27%, and 11.49% of the variation in WUE effects. In all factors, fertilizer types seemed to be more important than other factors, and factors related to soils made the most cumulative contribution to the WUE effects variation.

4. Discussion

A systematic quantitative analysis of the effects of fertilization on spring maize WUE in dryland areas in northern China was conducted based on a dataset collected in published papers containing independent experiments. Fertilization had a mean increase of 56.72% in spring maize WUE. Therefore, fertilization was considered to be an important method to improve WUE, while the WUE effects were also affected by environmental factors, soil factors, fertilization regimes, and agronomic measures. There were many factors affecting spring maize WUE, and the influencing factors selected in this paper explained 65.62% of the variation in the effects of fertilization on spring maize WUE, indicating that the effects of fertilization on spring maize WUE were influenced by a coupling of multiple factors rather than a single factor. The unexplained causes may be controlled by other alternative factors. Some studies have shown that crop WUE was also influenced by factors such as soil bulk density, soil texture, and planting density [23,37,38]. In this study, however, those factors were not analyzed.

4.1. Effects of Fertilization on Dryland Spring Maize WUE Based on Fertilization Regimes

N is an essential element for crop growth. The application of N fertilizer may affect crop yield and WUE by regulating soil C/N [39,40]. N fertilizer can significantly increase soil nitrogen content and alleviate the adverse effects of N deficiency on crop growth. In addition, N application can promote the growth of crop roots, improving soil water utilization, and thus increasing crop yield as well as the WUE [41]. The response of fertilization on WUE was the greatest when the NA was less than 100 kg ha⁻¹ (Figure 4), which was consistent with the results reported by Yin et al. (2022) [25]. In northern China, water plays an important role in affecting the fertilizer uptake by crops. A proper amount of N fertilizer supply can enhance the high-efficiency use of the limited water resources by crops. Excessive N fertilizer supply may, however, inhibit the assimilation products from transferring and distributing to crop roots, and thus reduce soil water uptake by roots. The carboxylase activity in crop photosynthesis may also be suppressed by the excessive N fertilizer, which may reduce the assimilation capacity of crops and decrease the yields, imposing an adverse effect on the improvement in WUE. The general NA of dryland spring maize (Zhongdan No. 2) in northern China is 180 kg ha⁻¹ [8], but NA is affected by many factors, e.g., maize variety, planting density, soil fertility, and management practices. Dang et al. (2006) concluded that in northwest China, maize had the maximum WUE at an N application of 120 kg ha⁻¹ [42], although an NA greater than 100 kg ha⁻¹ improves maize WUE more or less in this study. Qin et al. (2021) revealed that the utilization rate of N when the application of N fertilizer is greater than 150 kg ha⁻¹ remains lower than it may be when the NA is less than 150 kg ha⁻¹ [26]. Some studies also showed that an NA more than 120 kg ha⁻¹ would lead to the redundancy of soil nitrogen and an imbalance in nutrient content, which further caused a drop of the nitrogen-related invertase activity and the prevention of the assimilation transfer in crops, thus reducing the fertilization effects and suppressing the increase in crop yield and WUE [43,44,45]. In addition, fertilizer application may include organic fertilizers, which can provide N for maize growth [46]. In this study, it was observed that when the NA was less than 100 kg ha⁻¹, fertilizer application not only greatly improved the WUE of spring maize, but also reduced N fertilizer losses, which is consistent with the result reported by Mo et al. (2022) [47]. When the NA is in the range of 100–200 kg ha⁻¹ or greater than 200 kg ha⁻¹, soil nutrients significantly accumulate, causing a high concentration, which may induce an antagonistic effect between ions and reduce nutrient uptake by crop roots. As a result, the benefits of fertilization on crop WUE are reduced. Therefore, too much N application is unbeneficial to the development of sustainable agriculture.

Fertilization improves crop yield and WUE. However, an application of organic fertilizer together with inorganic fertilizer provides a much better effect than a single application of chemical fertilizer or organic fertilizer. Excessive use of chemical fertilizer may destroy the ecological environment, and it may also cause an enrichment of NO₃⁻ in soils as well as eutrophication in the surface water and underground water [12,23,48,49,50]. As the nutrient release rate of organic fertilizer is slower than that of chemical fertilizer, using only organic fertilizer may not provide the sufficient nutrients needed by maize, resulting in a slower increase in WUE [51]. Therefore, the combined application of organic and inorganic fertilizers has become a promising management measure for sustainable agricultural development [7,8,52,53,54]. This is consistent with the research of Xie et al. (2019) [55], who demonstrated that 37.5% and 50% substitution of chemical fertilizer by organic fertilizer increased the WUE of spring maize the most.

4.2. Effects of Fertilization on Dryland Spring Maize WUE Based on Environmental Factors

The effects of fertilization on the WUE of spring maize are also influenced by the regions, T, P, and A. Fertilization improves soil fertility and alleviates the conflict between soil nutrient supply and nutrient demand during maize growth and development, which improves the ability of maize to use soil water [41]. The difference in the increase in WUE between different regions is insignificant. However, there are considerable differences in WUE between subgroups under other factors. The absence of significant differences in WUE effects between different regions may be sourced from the differences in fertilizer types between different regions (Table 2).

P also has a significant effect on WUE. Under fertilization, the increase in WUE when P ranged from 400 to 600 mm was significantly greater than that when P < 400 mm and P > 600 mm (Figure 5), which remained consistent with Sun’s (2017) study [56]. In addition, some studies have found that too much or too little precipitation reduces crop yield and leads to a decrease in WUE [57]. With less precipitation, the water deficit may limit nutrient transport and transformation and thus decrease the effectiveness of nutrients. The water deficit may also restrain the development of crop roots which may inhibit the uptake of mineral nutrients and resultant in a decrease in crop yield and quality. When the precipitation is greater, i.e., P > 600 mm, the fertilization effect may be suppressed in the later growth period of maize due to the heavy rain, low temperature, and low light, causing crop yield and WUE to not steadily increase with the rise in the amount of fertilizer application. As a consequence, the fertilization effect when P > 600 mm remained close to that when P < 400 mm. In addition, excessive rainfall causes NO₃⁻ and P nutrients in fertilizer to flow as runoff through the surface and underground water, which prevents the maize from absorbing the required nutrients [58]. In addition, maize is a type of deep-rooted water-consuming crop with an average annual water requirement of 400–450 mm. Therefore, fertilization within a P of 400–600 mm maximizes water–fertilizer coupling, which facilitates the utilization of natural precipitation resources, and thus increases WUE.

The effects of fertilization on spring maize WUE was greatest when T was 7–10 °C (Figure 5). Both too high and too low temperatures suppress the fertilization effects. A high temperature strengthens the volatilization and thus quickens the nutrient loss, which is likely to cause crop yield reduction. A high temperature also increases evaporation and thus increases soil water consumption, causing the WUE response to fertilization to become smaller. Temperature is one of the key factors affecting nitrogen transformation. At lower temperatures, the activity of nitrogen-related transformation enzymes in the soil is low, which is not favorable for nitrogen decomposition and transformation, and which then reduces crop absorption, resulting in lower yields and lower WUE. Fertilization in areas with T of 7–10 °C is beneficial to the water utilization of crops, which is consistent with the findings of Ma et al. (2020) [59].

With the increase in A, spring maize WUE increased significantly. The maximum fertilization effect was 81.54%, which occurred when A > 1500 m (Figure 5). At high altitudes, a large temperature variation between day and night frequently exists. With higher temperatures and stronger light in the daytime, strong plant photosynthesis occurs, which enhances the absorption of nutrients from fertilizers. However, the temperature at night drops significantly and the maize’s cellular activity level becomes low and it remains difficult to consume nutrients. As a result, fertilizer application further improves the accumulation of dry matter in grains, resulting in a greater increase in maize yield and WUE [60]. Besides, in high-altitude regions, soils are usually infertile and fertilizer application can compensate for the shortage caused by soil nutrient deficit, and thus increase maize WUE. The growth period of maize is longer in high-altitude regions than in low-altitude regions, which leads to the nutrients from fertilizers being well absorbed by maize, which also increases WUE. This result was consistent with that of the potato studied by Li et al. (2021) [61] and the maize studied by Yin et al. (2017) [62].

4.3. Effects of Fertilization on Dryland Spring Maize WUE Based on Soil Factors

SOM content is an important indicator to characterize the fertility of farmland [11,63]. Several studies indicated that fertilization increased SOM content, and thus improved soil fertility [8,64,65]. The results of this study were consistent with the findings of Wang et al. (2020) [23]. The results showed that WUE increased significantly when 10 g kg⁻¹ ≤ SOM ≤ 14 g kg⁻¹ under fertilization management. However, the increase in the WUE of spring maize in this study was greater than the increase in the WUE of wheat in Wang et al.’s study [23]. The difference may be caused by the different growing seasons of spring maize and wheat. Besides, maize is a C4 plant. It converts nutrients at a higher rate than wheat, and therefore, the effect of fertilization on maize WUE is greater than that on wheat when the SOM content remains the same. In addition, when 10 g kg⁻¹ ≤ SOM ≤ 14 g kg⁻¹, the soil fertility is at a low to moderate level, and fertilizer application increases the soil water storage and soil fertility [8,46], which improves crop productivity and increases maize yield and WUE [65].

Soil AP content plays an important role in the growth and development of crops. It has been shown that crop growth is affected by phosphorus limitation [66]. With the increase in AP content, the effect of fertilizer application on WUE showed a significant decrease (Figure 6). AP was an important factor affecting spring maize WUE in dryland areas, explaining 11.60% of the variation in WUE (Figure 9), which was consistent with the findings of Hou et al. (2020) [66]. The effect of fertilization on the WUE of spring maize was greatest (91.36%) when the soil phosphorus content was low (AP < 5 mg kg⁻¹). With low soil AP content, maize growth and development did not receive sufficient nutrients. Organic or chemical fertilizer application increases the nutrient content of the soil, and it also enhances the drought resistance of maize. Maize growth is promoted and thus the WUE is increased. Fan et al. (2005) concluded that M, NP, SNP, and MNP treatments, respectively, increased the AP by 0.19, 0.25, 0.38, and 0.67 mg kg⁻¹ per year [8]. However, after continuous fertilization for several years, fertilizer application led to an accumulation of soil AP. The effect of fertilization on WUE decreases once the accumulation of AP exceeds the requirement of a specific plant. The soil AP requirement of dryland spring maize is around 10 mg kg⁻¹. When the initial soil AP is at the levels of 5–10, 10–20, and >20 mg kg⁻¹, the initial AP in the soil reaches a medium or high level. The content of the initial soil AP roughly satisfies the demand for maize growth, and therefore the effect of fertilization on WUE becomes less significant than that when AP is less than 5 mg kg⁻¹.

When soil AN content is high, fertilization increases soil N content, possibly causing N redundancy, which inhibits maize growth and reduces yield [41]. Therefore, fertilization at low soil AN content was more beneficial to increasing the WUE of spring maize.

Potassium (K) plays an important role in many important physiological processes in plants by improving product quality and making stems strong enough to survive in unfavorable conditions [67]. In fertilized conditions, the K requirements of crops decrease with increasing grain yield [68]. Su et al. (2001) concluded that K was a rare limiting factor to affect the growth of dryland spring maize in northern China [69]. This indicated that the AK content in the soil was high and additional fertilizer application caused soil AK accumulation and reduced the effect of fertilization on WUE. Therefore, fertilization at low potassium levels benefited the WUE of spring maize more.

4.4. Effects of Fertilization on Dryland Spring Maize WUE Based on Agronomic Measures

In all trials, all fertilizers except for N fertilizer were applied to the soil as the basal fertilizers. In most of the studies, N fertilizer was applied to the soil one time while a few other studies applied partial N fertilizer as the basal fertilizer, and the remaining fertilizer was used as the top application before sowing. After applying the basal fertilizers, fields were ploughed to ridge and furrow, and the maize was then sown in the furrow which was followed by the mulching, or the mulch was covered first and the sowing followed.

The WUE effect of spring maize using RFP responded to fertilization increased by 20% (P < 0.01) when compared to that using TFP (Figure 8), which might be because most fertilizers have been applied as basal fertilizers before using RFP and TFP. The ridge is usually 20 cm high and RFP suppresses the migration of fertilizers to the deeper soil more than TFP does, and, resultantly, the profile distribution of fertilizer is consistent with the distribution of crop roots, which improves the nutrition condition of the crop, leading to an Increase in yield and in the WUE of spring maize. Using RFP is helpful to collect rainwater and thus maintain soil moisture [70]. Sufficient moisture in soils is beneficial to the maize roots, allowing them to better absorb nutrients from fertilizer and thus increase WUE. Farmland mulching film is a widely used agronomic management measure [71,72]. MF can increase soil temperature, enhance soil-related conversion enzyme activity, improve soil organic carbon conversion and nutrient cycling, enhance the mineralization of nutrients under fertilization conditions, and promote plant uptake, thus increasing maize yield and WUE. MF usually has good airtightness; covering the mulch on the soil with fertilizer application can reduce the evaporation of nutrients and soil water, benefiting the absorption by the roots of nutrients from the fertilizer [73,74]. Here, the WUE of spring maize with MF under fertilization was increased by 13% in comparison with NMF under fertilization (P < 0.05) (Figure 8).

5. Conclusions

Based on a meta-analysis and a random forest model, the effects of fertilization on the spring maize WUE in northern China were investigated. To achieve this, 364 groups of data were collected from 33 publications. Several conclusions were drawn as follows.

Fertilization increased spring maize WUE by 56.72% (P < 0.01). The effect of the application of organic fertilizer together with inorganic fertilizer (i.e., MNPK and MNP) yielded a much better effect on the increase in WUE than the application of chemical fertilizers alone (i.e., N, NP, and NPK) or organic fertilizers alone (i.e., M). The effect of fertilization on WUE was greatest when the NA was 0–100 kg ha⁻¹. The greatest effect of fertilization on the WUE of spring maize was attained under specific environmental conditions, which involved 400 ≤ P ≤ 600 mm, 7 ≤ T ≤ 10 °C, and A > 1500 m, and soil fertility conditions with soil AP < 5 mg kg⁻¹ and 10 ≤ SOM ≤ 14 g kg⁻¹. Both initial AN and AK of soil negatively correlated to WUE. Fertilization affected the WUE of spring maize more when the initial fertility of the soil was low (low AN, AP, SOM, and AK). Fertilization together with RFP and MF was the agronomic measure that significantly increased the WUE of spring maize. Fertilization regimes, environmental factors, soil factors, and agronomic measures accounted for 65.26% of the variation in spring maize WUE effects. Fertilizer type related to fertilization regime was the most important factor causing the variation.

Water shortage is becoming serious in northern China where irrigation is difficult to be widely used and natural precipitation is relied on significantly in agricultural development. Therefore, improving WUE plays a crucial role in ensuring grain supply and efficient use of water resources in northern China. The combined application of organic and chemical fertilizers will be a promising measure for the sustainable development of dryland spring maize in northern China. Under certain environmental conditions, soil fertility conditions, and agronomic management measures, fertilization is beneficial to the efficient use of water resources in dryland crops.