1 Introduction

Nitrogen (N) is an essential nutrient for crop growth. It improves crop quality and increases productivity and production of agriculture, and thus, enormous quantities are applied to the soil to supply the world’s agricultural production every year (Belay 2014; Belete et al. 2018; Dorsey 2014; Solomon and Agena 2017). Nitrogen is also considered the most yield-limiting factor in crop production in non-fertilized agriculture (Bultosa 2012). However, despite the positive effect of N fertilizers on crops, there is an indirect negative effect on soil health in the absence of the proper use of N (Singh et al. 2015). Both excessive and underuses of nitrogen fertilizers are serious problems worldwide. In other words, resource-poor countries, for instance, African countries, have been using under-recommended rates, while resource-rich countries, including most developed countries and China, have been applying above the recommended rates leading to environmental pollution. Misuses of any type of fertilizer (organic or inorganic) pollute the environment and affect the health system. Hence, the management of N is an effective way of bringing about its use efficiency and boosting crop yields while reducing N-related adverse effects that have been challenging the world (Sharma and Bali 2018).

As it is with other crops, proper use of nitrogen fertilizer can significantly increase the yield of wheat—a major world’s strategic food crop, rich in carbohydrates (Seleiman et al. 2019; Wazir 2019), and has higher nutritional values than maize and rice (Mandic et al. 2015). Global wheat production covers 237 million hectares (about 20% of the world’s arable land) annually (Oyewole 2016; Tamado et al. 2015; Zhang et al. 2017). Reports show that worldwide wheat production had increased dramatically from the 1960s to 2013 without much change in the total area grown for wheat. This considerable increase in wheat production was mainly because of the adoption of the green revolution technology packages, including fertilizers. Nevertheless, the increases in wheat production due to the green revolution were not the same everywhere. Those countries that adopted the package from the beginning have successfully boosted their yields, while those lagging for various reasons, such as the high cost of fertilizers, have obtained fewer achievements. For example, Jiao et al. (2020) pointed out that because of mineral fertilizer use and policy development in China, grain yield has increased from 1.3 tons per hectare (t ha−1) in 1960 to more than 5 t ha−1 in 2000, while it remained low in Africa due to less consumption of mineral fertilizer in the region. Tadesse and Assefa (2018) also reported that wheat productivity has only slightly increased, from 1.3 t ha−1 in 1970 to 2.1 t ha−1 in 2014, in Sub-Saharan African countries. These studies suggest that the wheat yield gap has decreased in China while remaining high in Africa. In general, yield increase was stable and insignificant in Africa, unlike the experience of other countries during the green revolution era (Abegaz 2011).

Experiences have also shown that fertilizer application and wheat yield positively correlate and that N fertilizer considerably increases wheat yield. Consequently, over-applying N fertilizers to minimize N inadequacy and increase crop yield remains common practice for farmers in many countries. Because China was among the first consumer of N fertilizer, it has also exceeded the world consumption rate by 29%, which is a heavy consumption that has led to nitrate leaching, greenhouse gas emissions, and eutrophication, in addition to N use inefficiency and economic loss (Xu et al.

2 Methods

2.1 Data Collection

The data for this study were collected from peer-reviewed journal articles examining the effects of nitrogen fertilizer on wheat yield in Africa and China. The articles were obtained from Tylor, Google Scholar, and Web of Science databases. The search terms used include “nitrous” OR “nitrogen fertilizer” AND “yield” AND “wheat” AND “Africa” AND “China”. Data in graphical forms were digitalized using the Get Data v2.22 Software. To minimize biases, we screened the articles based on the following criteria. (1) The articles should be based on the data collected from field trials only; i.e., articles based on pot and greenhouse experiments were excluded. (2) The research areas should be within Africa and China. (3) The articles should contain studies on the chemical nitrogen application and no chemical nitrogen fertilizer application or control. (4) The articles were considered only if the yield data were reported. The specific range of study years was not considered as a criterion. Following these criteria, 65 articles (with a total of 665 observations) from the regions of Africa and 154 articles (2085 observations) from China were identified and used for the meta-analysis. Specifically, the articles identified were from studies conducted in nine potentially wheat-producing African countries (Ethiopia, South Africa, Morocco, Sudan, Rwanda, Egypt, Kenya, Tanzania, and Tunisia) and 14 major wheat-producing provinces of China. The distribution of experimental sites of the studies in Africa and China are shown in Fig. 1a, b.

Fig. 1
figure 1

Locations of N fertilizers field experiment sites on wheat crops in Africa (a) and China (b)

Full size image

2.2 Publication Biases Detection

In this meta-analysis, a quantile-quantile (Q-Q) plot was employed to detect the publication biases and distributions of data as it is an effective identifying tool for assessing normality in data and publication biases. It was constructed by plotting the empirical quantiles of the data against corresponding quantiles of the normal distribution of the effect size of the data. If the empirical distribution of the data is approximately normal, the points on the plot fall on a straight line defined by Y = X with the slope equal to unity, where Y is the ordinate and X is the abscissa (Sileshi et al. 2009). The natural variation of effect size is usually expected to be approximately normally distributed. Hence, skewing may be an indication of publication bias. The extent of bias can be estimated by the difference between the mean and mode of the distribution (Wang and Bushman 1998).

2.3 Data Analysis

The dataset organized based on the studies from the literature review was used for the data analysis. The confidence interval at the 95% level was employed to compare the magnitude of the response ratio between fertilized and non-fertilized plots or treatments to understand the confidence interval and evaluate the result of the study. Following Gao et al. (2019), the natural log (ln) of response ratio (R) was calculated as the effect size representing the effect of chemical nitrogen use. To explain the effects of chemical nitrogen, use on grain wheat yield, the results of R were calculated using the following equation.

$$R = \frac{Ye}{Yc}$$
(1)
$${\text{ln}}\;R\;{\text{ln}} \left(\frac{Ye}{Yc}\right)={\text{ln}}\;Ye-{\text{ln}}\;Yc,$$
(2)

where Ye is the mean value of wheat grain yield in fertilized treatments and Yc is the mean value of no-fertilized treatments.

Since most of the studies did not report the coefficient of variation, we adopted the unweighted method. To correct a bias at the 95% confidence interval, SPSS 20 version was used and a bootstrap** approach with 4999 iterations was applied to improve the probability that the confidence interval was calculated around the cumulative mean effect size for each categorical variable. If the 95% confidence intervals do not overlap with zero, there are significant (p < 0.05) positive or negative differences between fertilized and no-fertilized treatments; if not, the treatment effect is not considered statistically significant between methods. Response variables, such as yield increase (YI), partial factor productivity of nitrogen (PFPN), and agronomic efficiency of nitrogen (AEN) as indicators of nitrogen use efficiency (NUE) in wheat crop production, were quantified for both Africa and China. In this study, attainable yield (Ye) was defined as the actual yield derived under the optimal nutrient management practices. Yield response to N fertilizer was defined as the yield gap between attainable yield and the yield from non-fertilized treatments. Percent yield response (%YI) was defined as the yield gap between attainable yield with chemical nitrogen use and the yield from non-fertilized treatments to the ratio of non-fertilized treatments. AEN and PFPN were some of the nutrient efficiency parameters and were calculated as follows:

$$AEN=\frac{Ye-Yc}{N}$$
(3)
$$YI\;(\%)=\frac{Ye-Yc}{Yc} \times 100$$
(4)
$$PFPN=\frac{Ye}{N}$$
(5)

where AEN represents the agronomic efficiency of nitrogen, Ye is the yield of fertilized treatments and Yc is the yield of non-fertilized treatments. N represents the amount of chemical nitrogen applied on the allocated plot. The measurement unit is kilogram per hectare (kg ha−1) for all response variables computed above.

3 Results

3.1 Publication Bias Detection and Variability in Response Ratios of Wheat Yield in Africa and China

As mentioned earlier in the methods section, the normal Q-Q plots were applied to investigate the existence of publication bias and the distribution of normality assumptions. The results showed that the normal Q-Q plot curve in the case of Africa is slightly U-shaped, suggesting that the data are skewed to the right and publication bias exists when the population effect size differs from zero, presents (Fig. 2a). This could happen because studies with statistically significant results were more likely published than studies whose results were not statistically significant (Borenstein et al. 2009). According to Wang and Bushman (1998), if the point of observation falls on the straight line in a normal Q-Q plot and does not exceed the 95% confidence interval band, the estimated effect suggests no publication bias and is normally distributed. In our study, we made many efforts to minimize publication bias, although it cannot be eliminated in meta-analysis. For instance, we employed inclusion criteria to identify the articles that could answer the intended questions, and the articles that did not meet the aim of the study were excluded from the database.

Fig. 2
figure 2

Normal quantile-quantile plots of wheat yield response ratios (Log. treatments/control) in Africa (a) and China (b) for exploring the normality assumption and publication bias. The circles represent individual observations, while the solid line (Y = X) shows the standard normal distribution with 95% confidence interval bands

Full size image

3.2 Effects of Nitrogen Fertilizers on Wheat Yield in Africa and China

We used regimes of N fertilizers to evaluate the effects of N rates on wheat yield in both Africa and China (Fig. 3). In the case of Africa, the mean wheat yields response at the rates of 50–100 kg ha−1 and over 150 kg ha−1 of N applications were, respectively, 1323 kg ha−1 and 3446 kg ha−1 (Fig. 3a). That is, there were statistically significant increases (p < 0.05) in wheat yields (by 46% for the earlier rates of N and by 63% for the latter) compared to the control (i.e., 2007 kg ha−1). In contrast, the wheat yields response obtained in China were 1193 kg ha−1 (i.e., 98% yield increase compared to the control) when less than 100 kg ha−1 of N was applied, 1446 kg ha−1 of wheat for N rates of 100–150 kg ha−1, and 1886 when 150–200 kg ha−1 was used. However excessive use (over 300 kg ha−1) of chemical nitrogen (response mean 2086 kg ha−1) significantly decreased wheat yield (by 29%) as compared to the control yield (Fig. 3b).

Fig. 3
figure 3

Yield response of grain wheat (yield of wheat fertilized treatments – yield of wheat not fertilized treatments, for four levels of chemical nitrogen fertilizer rates) in Africa (a) and China (b). Box circles indicate means, and the horizontal lines on the left and right sides represent the upper and lower limits of the 95% confidence interval. Means (circles) are not significantly different from one another if their 95% confidence intervals (error bars) overlap. The means and 95% confidence intervals of the response ratios are in the bootstrap** method. Each group of data was determined with an F-test (p < 0.05). The numbers in the bracket show the number of observations

Full size image

3.3 Effects of Soil Chemical and Physical Properties (SOM and pH) on Wheat Yield in Africa and China

We also evaluated the effects of soil chemical and physical properties, such as pH and soil organic matter (SOM) on wheat yields, besides nitrogen fertilizer in both Africa and China. The results showed that the effects of soil pH and soil organic matter on wheat yields were higher in Africa than in China. In Africa, a higher yield of wheat was obtained from alkaline soil (pH > 7) where the mean wheat yield was increased by 213%. In the meantime, a lower yield of wheat was obtained from acidic soil (pH < 7) where there was a mean yield increase of 93% (Fig. 4a, above the small dash line). Contrary to the finding from Africa, the data collected from the provinces of China indicated that a higher wheat grain yield (91% mean yield increase) was obtained from acidic soil, and a relatively lower yield of wheat grain (62% mean yield increase) was obtained from an alkaline soil. That is, the mean wheat yield in acidic soil was statistically significantly higher than the yield in alkaline soil in China (Fig. 4a, below the small dash line). The overall effects of the mean values of soil pH were 148% and 70%, respectively, in Africa and China, implying less pronounced effects of pH on wheat yield in China than in Africa.

Fig. 4
figure 4

Yield response to the power of pH (a) and soil organic matter in Africa and China. Box circles indicate means. Error bars represent 95% confidence intervals. Means (circles) are not significantly different from one another if their 95% confidence intervals (error bars) overlap. The means and 95% confidence intervals of the response ratios are in the bootstrap** method. The numbers in brackets indicate the number of observations

Full size image

Similarly, the effect of soil organic matter (SOM) on wheat yield was also assessed and the results demonstrated a positive yield response to SOM in Africa. However, there was no statistically significant difference between the SOM groups (< 4 and > 4 g kg−1). The evidence obtained from the Africa dataset indicated that the availability of SOM in the soil was very low, indicating a high depletion of nutrients from the soil (Fig. 4b, above the small dashed line). In the case of China, the highest yield of wheat was obtained within the range of 15–20 g kg−1 (mean yield increase by 63%) than within 10 g kg−1 of SOM. The results indicated that the concentration of SOM fell at a very high level in China (Fig. 4b, below the small dashed line).

3.4 The Effects of Precipitation on Wheat Yield

The overall means of wheat yield to rainfall response ratio were 325, 143, and 93%, respectively, for < 600 mm, 600–900 mm, and > 900 mm of rainfall during the growing seasons in Africa. The overall means of wheat yield response rations in China were 39 and 25% for < 600 mm and 600–900 mm of rainfall, respectively. The results illustrated that the average wheat yield response was higher in Africa than in China although the amount of rainfall in the growing season was the same; suggesting that rainfall was less important in China compared to Africa. In Africa, the maximum wheat yield response was found during the growing season of < 600 mm rainfall, indicating that this amount of rainfall was optimal for wheat production (Fig. 5).

Fig. 5
figure 5

Effects of precipitation on wheat yield in China (a) and Africa (b). The box circle indicates means. Error bars represent 95% confidence intervals. Means (circles) are not significantly different from one another if their 95% confidence intervals (error bars) overlap. The means and 95% confidence intervals of the response ratios are in the bootstrap** method. The numbers in brackets indicate the number of observations

Full size image

3.5 Measurement of N fertilizer Use Efficiency

The results showed that the mean values of partial factor productivity of nitrogen (PFPN) fertilizers were 58 for Africa and 36 for China (Fig. 6a). That is, the mean value of PFPN fertilizer for Africa was significantly higher than that of China. In other words, the overuse of N fertilizer significantly reduced the N use efficiency in the case of China. The box plot (Fig. 6a) indicated a significant and negative relationship between PFPN fertilizer and wheat yield both in Africa and China. The result also showed that a higher mean value of agronomic use efficiency of N was found in the case of Africa (Fig. 6b).

Fig. 6
figure 6

Relationships between fertilizer application of nitrogen fertilizer (N) and its partial factor productivity of nitrogen (a and b) and Agronomic use efficiency in kg ha−1 (c and d) for wheat crop in China and Africa. In the regression model, Y is partial factor productivity of nitrogen (a and b) and agronomic use efficiency (c and d), X is application rates of N. Significance of regression coefficients and intercepts was determined with a F-test (p < 0.05). The number of data points was given in each equation

Full size image

4 Discussion

4.1 Wheat Yield Response to Chemical N Rates in Africa and China

In this meta-analysis, chemical nitrogen application rates and other related information were collected from different literature, a dataset was developed, and the data were then organized into different regimes of fertilizer application rates. The findings indicated that different application rates of N fertilizer significantly affected the mean yield of wheat grain both in Africa and China. There were also variations in the magnitude of effects among the rates of N fertilizers applied in the two regions. In the case of Africa, there was a positive and linear correlation between the N rate(s) and the yield of wheat. These findings comply with the findings of Abebe (2016), Dugassa et al. (2019), and Tilahun and Tamado (2019). In particular, Abebe (2016) reported that wheat grain yield was highly influenced by the N fertilizer rate and that the proper rate and time of application are critical for meeting crop needs and attaining higher grain yields.

A high grain yield production has been achieved with increased usage of N fertilizer in developed countries even though it remains a concern in low-income countries because of fertilizers’ high prices (Bhandari et al. 2020). As of the present meta-analysis, the regime of N fertilizer that resulted in the highest wheat yield increase has partially agreed with the N fertilizer rates (30 to 138 kg ha−1) recommended by Yohalashet et al. (2017). However, these N rates could be lower than developed countries’ N consumption levels. In African countries, particularly in Ethiopia, the consumption level of fertilizer is considerably below the recommended rates because of its high prices, and less than 40% of farmers use fertilizers in the region (Agbahey et al. 2015). In nutshell, the amount of fertilizers used in low-income countries is minimal and inadequate to compensate for the nutrients removed by harvested crops.

Nutrient use efficiency is a critically important concept in the evaluation of crop production systems. Nutrient use efficiency, which affects the performance of the crop** system, is in turn greatly affected by fertilizer management. Nitrogen fertilizer management, in terms of using the right source, the right rate, in the right place, and at the right time, is reported to be poor in the cases of African countries. This poor management of fertilizers has led to poor nitrogen use efficiency and has caused a low crop yield response. Underrated consumption of chemical fertilizer is found to be the primary factor that leads to a significant reduction in yield response in Africa. Farmers are using a small amount of fertilizer that does not meet crop requirements. This may result in poor nitrogen use efficiency and wheat production performance. To address this challenge, nitrogen management should receive due attention in addition to the crop** system. Hence, maximizing the usage of N fertilizers to the level of crop requirements is one of the ways to enhance NUE and wheat productivity in Africa. Because Africa is a resource-poor country and farmers are unable to afford and apply sufficient fertilizers that can meet the crops’ requirements, the governments of African countries should explore the appropriate approach that subsidizes the smallholder farmers to increase the consumption level of fertilizers that improve nitrogen use efficiency, thereby increase yield response. Increasing nitrogen usage up to an optimum nitrogen fertilizer rate increases nitrogen use efficiency and yield response.

The result found in the case of China was different from that of Africa. The pattern showed that the wheat yield was increasing with the N fertilizer rate up to a certain level, and then started decreasing. This implies that there was excessive usage of N fertilizer that led to a reduction in wheat yield. The use of N fertilizer in the case of China was more than the crop requirement. A similar finding was also reported by Habbib et al. (2017) that the overuse of fertilizers has been practiced to achieve a high wheat yield in many countries. However, the unreasonable application of fertilizers has led to N accumulation in soil and later would result in yield reduction as well as environmental pollution. Similarly, our study witnessed N accumulation in many provinces of China. A related study (Wang et al. 2012) also discussed that an increase in application rates of N fertilizer can increase the accumulation of N in the soil, especially when over 221 kg ha−1 of N fertilizer is used. This could result in severe adverse effects on soil properties (Liu et al. 2018). In general, our results showed that there was a significant non-linear negative relationship between N application rates and PFPN, and AEN, indicating that PFPN and AEN decreased as an increasing amount of fertilizer was applied. Therefore, to improve NUE and maximize wheat production, environmentally friendly fertilization regimes based on soil tests are strongly recommended. Similarly, to achieve high PFPN and AEN, managing the N supply from the soil and other indigenous source is as important as maximizing nitrogen efficiency.

5 Conclusion

The study analyzed different methods that could improve wheat production and how efficiently it uses nitrogen. The research found that using N fertilizer to grow wheat resulted in better yields in China compared to Africa. This means that there is more potential to increase the productivity of wheat in Africa than in China. Our finding shows that wheat production in Africa had slightly better nitrogen use efficiency compared to China. The impact of Ph and soil organic matter on wheat production was stronger in Africa than in China. Rainfall during crop growing season has a lesser impact in China compared to Africa. This suggests that it is not possible to grow wheat in China without using irrigation. The study highlighted that a high yield response to chemical N use in Africa was accompanied by N depletion, while in China it was associated with a high N surplus. To control the use of nitrogen in the future, it is important to have guidelines that farmers can easily follow. These guidelines should be specific to nitrogen use, the type of soil, and the type of crops being grown. These guidelines would assist farmers in understating what they need to measure and where they should focus on making improvements. The guidelines will explain why changes in management are necessary and how they will affect the performance of the system.