Introduction

Agriculture accounted for 11% of world greenhouse gas (GHG) emissions in 2019, with 10.3% of these emissions being derived from the use of synthetic fertilizers (FAOSTAT 2019a). The application of nitrogen (N)-based fertilizers leads to the production of nitrous oxide (N2O), a potent GHG. Furthermore, low nutrient use efficiency may result in ammonia (NH3) volatilization, nitrate leaching (NO3−), and further environmental degradation (i.e., groundwater contamination and indirect N2O emissions) (Carlos et al. 2022; de Paulo et al. 2021; Simon et al. 2020). Therefore, mitigation measures, such as nitrification inhibitors (NIs), have been developed to reduce NO3− leaching and increase N use efficiency.

The oxidation of ammonium (NH4+) to nitrite (NO2−) can be delayed by applying NIs, which act to inhibit ammonia-oxidizing bacteria (AOB) activity (Ruser and Schulz 2015; Hayden et al. 2021). Dicyandiamide (DCD), 3,4-Dimethylpyrazole phosphate (DMPP), and 2-chloro-6-(trichloromethyl) pyridine (Nitrapyrin) are the most widely investigated and commercially utilized NIs (Zerulla et al. 2001; Wolt 2004; Yang et al. 2016). Guo et al. (2022) indicated that DCD and DMPP can reduce N2O emissions by up to 85% and 99%, respectively, under a range of temperature and moisture levels. Conversely, Mazzetto et al. (2015) and Nauer et al. (2018) reported that N2O emissions from cattle urine and urea applications were not reduced by DCD and DMPP, respectively. Such a wide range of different results are commonly found in the literature because NI efficiency relies on numerous factors such as temperature, moisture, pH, texture, organic carbon content, tillage, the choice of NI, and fertilizer types and rates (Ekwunife et al. 2022; Guo et al. 2022). Therefore, researchers have conducted a series of meta-analyses to quantitatively pool the available information about NI efficiency in contrasting environments, agricultural practices and soil conditions, aiming to identify the most important influencing factors and optimize NI use (Abalos et al. 2014; Ekwunife et al. 2022; Linquist et al. 2013; Thapa et al. 2016; Yang et al. 2016).

Ekwunife et al. (2022) showed that NIs can reduce N2O over-winter emissions by 23% in temperate regions with soils subjected to freezing–thawing. Abalos et al. (2014) indicated that NIs perform better under conditions that favor high drainage and when high inputs of N fertilizer are applied, increasing crop productivity. Thapa et al. (2016) also found that NIs mitigate N2O release, and this effect was more prominent in neutral, coarse-textured, and irrigated soils, although no effect on crop yield was reported. Considering only field studies, Yang et al. (2016) compiled 81 studies across the globe and observed that both DCD and DMPP were similarly effective in regulating N soil transformations and controlling N2O emissions, but DCD showed the best performance in increasing crop yield. However, most of the studies described above were carried out in subtropical and temperate climate zones, and tropical regions are under-represented in this field of research. Nevertheless, tropical agriculture plays an important role in global food production, fertilizer use, and GHG emissions. For instance, Brazil, a country containing subtropical and tropical climates, was the fourth highest consumer and the second greatest importer of N-based fertilizers worldwide in 2019, accounting for 4.5% of N2O emissions derived from the agricultural use of synthetic fertilizers (FAOSTAT 2019b).

The use of NIs could be an important tool for climate change mitigation; however, they are not universally efficient. Correspondingly, we investigated the key factors contributing to increasing or decreasing the efficiency of NIs in Brazilian subtropical and tropical agriculture. We hypothesized that NI performance is weaker under tropical than subtropical and/or temperate conditions due to the interaction of factors such as warm temperatures and more frequent soil wetting and drying cycles, which may influence NI degradation and stimulate the abundance and activity of AOB (Mazzetto et al. 2015).

We utilized a meta-analytic approach to identify the impact of NIs (DMPP, DCD, and DCD + NBPT) on (i) crop yield, (ii) N2O emissions, (iii) soil NH4+ and NO3− concentrations, and (iv) NH3 volatilization in Brazilian subtropical and tropical regions. The environmental, soil, and agricultural management factors driving N losses were also evaluated. To the best of our knowledge, this is the first meta-analysis to focus on determining whether NIs increase crop yield and reduce N losses in tropical and subtropical climates.

Material and methods

Data search, selection criteria, and investigated variables

Our data survey considered only literature published before January 2022 and it was performed using the ISI-Web of Science and Google Scholar databases through the following query: (nitrification inhibitor* OR DCD OR DMPP OR Nitrapyrin) AND Bra*il. Articles were then selected according to the following criteria: (1) the study was carried out in Brazil and was published in a peer reviewed journal; (2) the experimental design included at least one treatment with a N-based fertilizer combined with DMPP, DCD, or Nitrapyrin and a control treatment with the equivalent N fertilizer; and (3) crop yield, soil NH4+ and NO3−, N2O emissions, or NH3 volatilization were reported. Subsequently, a total of 50 articles (supplementary material Table S1) were included in the data set for further analysis. In addition to extracting the results of the response variables, we also extracted supplementary information including location, climate zone (supplementary Fig. 1), cultivated crops, pH, soil organic carbon (SOC), texture, temperature, soil moisture, experimental method, soil management, irrigation, fertilizer source, rate and type, NI type and rate. When data were included only as figures, we utilized the WebPlotDigitizer software (Rohatgi 2021) to extract means and standard deviations (SD). When SD or standard error (SE) were not present in the original articles, we conservatively estimated the SD based on the mean of the observation and 150% of the mean variance across the data set (Ros et al. 2020).

Controlling factors

Data were grouped according to a range of controlling factors to maximize subgroup homogenization and identify the conditions that enhance or curb NI efficiency. Firstly, studies were grouped into tropical (between latitudes 23.5° N and 23.5° S) or subtropical (between latitudes 30° N and 30° S) climate zones (Albanito et al. 2017) (supplementary Fig. 1). In addition, studies were grouped according to the following factors: mean temperature: < 20 °C or 20–25 °C; experimental method: field or controlled conditions (greenhouse or incubation); soil management: till or no-till; irrigation: irrigated or rainfed; cultivated crops: cereals (maize, wheat and rice), forage, or vegetables/industrial (sugarcane, cotton, eucalyptus, pineapple, potato and lettuce); texture: coarse, medium, or fine; SOC: < 1%, 1%–2%, or > 2%; pH: < 6 or 6–8; fertilizer source: ammonium sulfate, urea, or organic; N fertilizer rates: ≤ 150, 150–300, or ≥ 300 kg ha−1; NI type: DCD, DCD + NBPT, or DMPP. N-(n-butyl) thiophosphoric triamide (NBPT) is a urease inhibitor (UI) and can be applied in combination with an NI to prevent unintended NH3 volatilization from stabilized NH4+-N and to further minimize N losses. As only one study included Nitrapyrin (Martins et al. 2017), it was not possible to include this NI in the subgroup analysis. In addition, DCD was the only NI that was applied at variable rates, and thus we also grouped studies according to DCD rate as follows: < 5% and ≥ 5% of the applied NH4+-N or urea-N. Soil moisture data were not included in this analysis because of the high temporal variability within the studies making it impractical to extract average values. In addition, we did not include these data because of the difficulty in homogenizing values reported as a proportion of different indicators (field capacity, water-holding capacity, or water-filled pore space). The abovementioned subgroups were chosen to allow comparison with previous meta-analyses and because they comprise the known factors that are most likely to affect N losses (Abalos et al. 2014; Thapa et al. 2016; Wu et al. 2021; Yang et al. 2016). Further information about the configuration of each subgroup, the approaches taken to homogenize data that were expressed or measured differently between studies, and descriptive statistics of the influencing factors and calculated indicators can be found in the supplementary material.

Data analysis

The effect size of NI application on the studied variables was estimated according to the natural logarithm of the response ratio (RR) as follows (Hedges et al. 1999):

$$\ln \left( {{\text{RR}}} \right) = \ln \left( {\frac{{\overline{xe} }}{{\overline{xc} }}} \right)$$
(1)

where \(\overline{xe}\) is the mean of the treatment with N fertilizer and NI and \(\overline{xc}\) is the mean of the control treatment with the analogous fertilizer. Mean effect sizes were considered statistically significant if the 95% confidence interval (CI) did not overlap with zero, and the difference among subgroups was significant if their 95% CIs did not overlap with each other (Hedges et al. 1999). To facilitate data visualization, the values were back-transformed to give the percentage change of the “fertilizer with NI” treatment in relation to the “fertilizer-only” control treatment: ((RR − 1) × 100%).

We performed the meta-analysis with the R package metafor, in which we calculated mean effect sizes using random/mixed-effects models and applied the restricted maximum-likelihood estimation (R Development Core Team 2022; Viechtbauer 2010). The effect sizes were weighted by the inverse of their variance and study identification was incorporated in the model as an independent factor. This step was needed because more than one observation was extracted from most studies, which would otherwise lead to non-independent comparisons (Crystal-Ornelas 2022).

The fail-safe number technique at p ≤ 0.05 was used to assess publication bias in the data set (Rosenthal 1979). This approach estimates the number of studies reporting null results (effect size equal to 0) that are needed to change a significant result to a non-significant result. The analysis indicated that 646–81,959 studies with null effect size would be necessary to cause non-significant results (depending on the variable), while the critical value ranged from 270 to 595 (5n + 10). This indicates that it is unlikely that any publication bias present in the considered literature would be sufficient to interfere with the overall results presented in this study.

Results

Data set overview

The created data set compiled a total of 50 studies, including observations from a range of sites embracing 10 different Brazilian states and for various different seasons (supplementary Fig. 1). Such studies were carried out under tropical (n = 34) and subtropical (n = 17) climates; field (n = 36) and controlled conditions (n = 15) (considering controlled conditions experiments; cultivated soils: n = 10, and non-cultivated soils: n = 5); no-till (n = 23) and tilled (n = 8) management; irrigated (n = 25) and rainfed (n = 17) areas. Of all studied NIs, DCD (alone or combined with NBPT) was most represented. It was used predominantly in acidic as well as fine and medium-textured soils, all within a similar range of SOC contents, N sources, and crop types. The data set for crop yield mainly comprised cereal crops (n = 23), such as maize, wheat and rice. In addition, the majority of studies focused on low (< 150 kg ha−1) and medium (150–300 kg ha−1) fertilizer N rates. Nitrogenous fertilizer rates of around 150 kg N ha−1 are typical for cereals in Brazil; however, the application rate can vary considerably in accordance with factors such as soil, region, technology utilized by farmers, and crops or cultivars. The mean temperature subgroups (< 20 °C and 20–25 °C) did not affect the studied variables, which is likely to be because of the high variability of maximum and minimum temperatures among experiments even within the same subgroup (data not shown).

Crop yield

Considering 113 observations from 33 studies, NIs slightly increased crop yield by 3% (0.1–5%), whereas DCD use, tilled and medium-textured soils enhanced plant productivity (Fig. 1). Moreover, this increase in crop yield was not consistent within the data set once a null response was verified for several controlling factors. Therefore, the agronomic advantage of using NIs was not proportional to the relatively high mitigation potential of N losses (Figs. 1–3), even when NIs and UIs were applied together. It should be noted that crop yield in this study refers to grain production for cereals, biomass for forage, and various indicators for vegetable/industrial according to the crop type (i.e., biomass, fruit or tuber yield). Moreover, the measured indicator of plant productivity (i.e., grains, biomass, fruits, tubers) did not change the effect of NI addition.

Fig. 1
figure 1

Effect of nitrification inhibitors (NIs) on A crop yield and B N2O emissions under controlling factors: climate zone, experimental method, soil tillage, irrigation, crop, texture, soil organic carbon (SOC), pH, N source, N rate, NI type and DCD rate. Horizontal bars represent 95% confidence intervals (CIs) and the number of observations and studies per subgroup are between parentheses. The red line indicates the threshold for null response. An asterisk (*) represents significant difference among controlling factors within the same subgroup according to the 95% CIs

Full size image

N2O emissions

Addition of DMPP, DCD, or DCD + NBPT decreased N2O emissions by 85, 82, and 53%, respectively (62% overall reduction). Moreover, NI addition was found to be efficient in terms of N2O reduction regardless of the climate zone, experimental method, soil tillage, irrigation, SOC, pH, and NI type (Fig. 1). Conversely, forage and cereal crops, medium-textured soils, organic and urea N sources, and high N rates associated with low DCD addition rates showed variable results, with no overall mitigation of N2O emissions. Trials under controlled conditions (− 86%), which included greenhouse and incubation studies, showed greater reduction of N2O release than field experiments (− 45%). Vegetable/industrial crops (− 76%), fine-textured soils (− 73%), ammonium sulfate (− 98%), high DCD addition rates (− 80%), and low (− 64%) and medium (− 72%) N-addition rates favored NI effectiveness relative to the other subgroups within the same controlling factor.

Soil NH4 + and NO3 − concentrations

The soil NH4+ and NO3− concentrations were based on the average values presented in the original articles. Thus, the impact of NIs on slowing down the conversion of NH4+ to NO3− was possibly weakened in long-term studies and field studies because NIs have a relatively short-term effect (Fig. 2). In spite of this, NIs significantly increased soil NH4+ and decreased NO3− concentrations by 60 and 31%, respectively (Fig. 2). In general, these results reflected the pattern found for N2O emissions and confirmed that NIs were effective under a broad range of controlling factors. Fine texture (+ 183%), high SOC (+ 145%), and a urea fertilizer source (+ 260%) increased NH4+ concentration, and subtropical climate, field experiments, and rainfed areas did not change NO3− concentration as a result of NI application. Finally, NIs showed a lower efficiency under cereal crops, medium-textured soils, organic fertilizers, and high N and low DCD rates. Among the studies included in the data set, Cantú et al. (2017) and de Paulo et al. (2021) respectively reported that DCD and DMPP efficiently reduced NO3− leaching.

Fig. 2
figure 2

Effect of nitrification inhibitors (NIs) on soil concentration of A NH4+ and B NO3− under controlling factors: climate zone, experimental method, soil tillage, irrigation, crop, texture, soil organic carbon (SOC), pH, N source, N rate, NI type and DCD rate. Horizontal bars represent 95% confidence intervals and the number of observations and studies per subgroup are between parentheses. The red line indicates the threshold for null response. An asterisk (*) represents significant difference among controlling factors within the same subgroup according to the 95% CIs

Full size image

NH3 volatilization

The application of NIs did not increase NH3 volatilization in any of the considered controlling factors; however, application of an NI with a UI significantly reduced N losses by 39% (Fig. 3). Moreover, a subtropical climate (− 41%), irrigation (− 38%), medium-textured soils (− 29%), high SOC concentration (− 50%), and low N rates (− 42%) were shown to reduce NH3 volatilization.

Fig. 3
figure 3

Effect of nitrification inhibitors (NIs) on NH3 emissions under controlling factors: climate zone, experimental method, soil tillage, irrigation, crop, texture, soil organic carbon (SOC), pH, N source, N rate, NI type and DCD rate. Horizontal bars represent 95% confidence intervals and the number of observations and studies per subgroup are between parentheses. The red line indicates the threshold for null response. An asterisk (*) represents significant difference among controlling factors within the same subgroup according to the 95% CIs

Full size image

Discussion

Crop yield

Overall, NI application slightly stimulated crop yield, which is likely to be due to the increased N use efficiency and reduced N losses (Fig. 1). Previous meta-analyses, which did not consider or underrepresented tropical sites, also reported slight increases in plant productivity due to NI application as follows: 7.5% (Abalos et al. 2014), 7% (Thapa et al. 2016), 11% (Qiao et al. 2015), 4.4% (Feng et al. 2016), and 6.5% and 1.2% for DCD and DMPP, respectively (Yang et al. 2016). Some further discussion focusing on why NI application led to only a slight increase in yield, despite a substantial reduction in N losses, is presented as supplementary material.

N2O emissions and soil NH4 + and NO3 − concentrations

Our meta-analysis clearly showed that N2O emissions were mitigated by NIs (Fig. 1). This result was observed in both subtropical and tropical climates. Considering only field experiments, N2O release was decreased by 45% (CI: 21–62%). This reduction in N2O emissions was comparable to values previously published meta-analyses as follows: 41% (CI: 29–53%) (Gilsanz et al. 2016); 38% (CI: 31–44%) (Akiyama et al. 2010); 44% (CI: 39–48%) (Qiao et al. 2015); 30% (Feng et al. 2016); 38% (CI: 33–44%) (Thapa et al. 2016); and 45 and 48% for DCD and DMPP, respectively (Yang et al. 2016). This suggests that NI application could be as advantageous to agriculture in tropical zones as it is in subtropical and temperate zones.

The reduction in N2O emissions following NI application was likely to be due to the ability of NIs capacity to chelate Cu2+ ions. These function as co-factors of the ammonia monooxygenase (AMO) enzyme, delaying the first step of nitrification and resulting in higher NH4+ concentrations (+ 60%) and lower NO3− concentrations (− 31%) (Fig. 2) (Corrochano-Monsalve et al. 2021c; Ruser and Schulz 2015). Furthermore, de Paulo et al. (2021) indicated that DMPP can retain N as NH4+ and reduces NO3− losses without increasing NH4+ leaching. They also observed enhanced 15N recovery in soil and in cotton plants. Qiao et al. (2015) and Yang et al. (2016) also indicated that NIs can efficiently delay nitrification, which is supported by our results.

NH3 volatilization

Losses of NH3 were not significantly affected by NIs when applied without UI addition (Fig. 3). In contrast, Qiao et al. (2015) and Wu et al. (2021) reported that NI application increased the overall NH3 volatilization by 21% and 36%, respectively, as a consequence of retaining a high soil NH4+ concentration for longer periods. Our results suggest practices such as irrigation could mitigate NH3 losses when combined with NI and UI application, particularly in soils with high SOC levels, a medium texture, and N application rates < 150 kg ha−1. Furthermore, neutral to alkaline soils showed reduction in NH3 volatilization relative to acidic soils, which, is contrary to the typical understanding that neutral–alkaline soils trigger NH3 volatilization. This inconsistency may have arisen because the associated use of NIs and UIs was preferably studied under high-pH soils and thus the effect was more closely tied to UI application than soil pH. Kim et al. (2012) found NIs can increase, decrease, or even not affect NH3 volatilization depending on soil pH, temperature, moisture, and cation-exchange capacity. This suggests that NIs can increase NH3 volatilization if favorable conditions are present at the moment of the application (high soil pH, high temperature and moisture, and low cation-exchange capacity), and the combined use of NIs and UIs should be adopted in such situations.

Effect of climate zone

The use of NIs under a subtropical climate resulted in higher soil NH4+ concentration and lower NH3 volatilization than in tropical areas (Figs. 2, 3). This might be explained by warmer temperatures in tropical areas, which may stimulate volatilization and nitrification. Surprisingly, NIs reduced soil NO3− concentration only in tropical sites. It should be noted that the soil NH4+ and NO3− concentrations presented in Fig. 2 are based on the average values from original articles with different experimental durations and therefore cannot demonstrate the temporary effect of NIs on preventing nitrification that was shown in some studies (Ibarr et al. 2021).

The overall increase in NH4+ and decrease in NO3− soil concentration resulted in lower N2O emissions from NI-treated fertilizers (Fig. 1). Feng et al. (2016) found NIs and UIs to be less effective in humid than in arid regions because rainfall can promote the spatial separation of the NI and soil NH4+ and cause leaching. Nonetheless, NI addition significantly increased crop yields and mitigated N2O release in humid regions. In addition, Abalos et al. (2014) found no effect of climate on the effectiveness of NIs. However, the data set did not include studies in tropical areas. Overall, our data set demonstrated NIs to be as effective in tropical zones as they are in subtropical zones, with similar results to previously published meta-analyses that did not consider or underrepresented tropical areas (Abalos et al. 2014; Ekwunife et al. 2022; Gilsanz et al. 2016; Thapa et al. 2016; Yang et al. 2016).

Effect of soil tillage and experimental method

Experiments under field conditions confirmed a more modest NI performance (Fig. 1b) because of the broad range of interfering environmental conditions. For instance, Kelliher et al. (2014) concluded that DCD half-life was significantly shorter in soils under field conditions than in samples incubated under an analogous temperature because of enhanced soil microbial activity associated with the influence of plants in the field. The application of NIs in no-till and tilled soils showed a similar response in terms of N2O and NH3 release as well as soil inorganic N concentrations. However, NI application increased crop yield only in tilled soils. In contrast, Thapa et al. (2016) found that NIs promoted crop yield in both no-till and tilled soils. The impact of no-till on NI effectiveness and N2O emissions can be widely variable. On the one hand, no-till can mitigate N2O emissions by lowering soil temperature and increasing water drainage as a result of the enhanced vertical pore connectivity (Corrochano-Monsalve et al. 2021b). Enhanced drainage is also a factor that may promote NIs persistence and effectiveness in soil. On the other hand, no-till management can increase soil moisture content and bulk density, which favor anaerobic conditions and N2O emissions (Ekwunife et al. 2022).

Effect of irrigation

The decrease in soil NO3− following NI addition in irrigated soils (Fig. 2) might be associated with a better incorporation of the NIs in soil and, consequently, their protection against UV radiation and wide temperature variations that commonly occur at the soil surface and may accelerate NI degradation (Byrne et al. 2020). Thapa et al. (2016) found that NI application increased crop yield only in irrigated areas; however, their data set did not show the effect of irrigation on the mitigation of N2O emissions. Irrigated fields tend to favor NO3− leaching and N2O emissions because of the higher soil water-filled pore space over the cropping season (Guo et al. 2022). Nevertheless, irrigation (5–10 mm) after urea application moves the urea molecule down to the soil profile (approximately 20 mm below soil surface), which helps to reduce the NH4+ concentration in surface soil (0–20 mm) and thereby NH3 emissions. Moreover, high soil water content results in slow NH3 diffusion and, consequently, minimal NH3 emissions (Carlos et al. 2022). Therefore, the combination of NI application with irrigation is likely to be an advantageous strategy to promote crop yield and fertilization efficiency.

Effect of crop type

The NI effectiveness was high under vegetable/industrial subgroup, which mainly comprised studies investigating sugarcane cultivation (88% of the observations). The null response of cereal crops for N2O emissions and soil inorganic N concentration occurred as a result of the variable response among studies (Figs. 1, 2). Martins et al. (2017) found urea treated with Nitrapyrin increased maize yield and reduced N losses in comparison to unamended urea. Nevertheless, inconsistent results were reported for DCD and DCD + NBPT application in maize and wheat fertilized with swine slurry under different application methods and cropping seasons (Aita et al. 2014, 2015, 2019). For instance, Aita et al. (2015) found that DCD reduced cumulative N2O emissions for wheat but not for maize, which was explained by background soil NO3− availability as a result of previous incorporation of perennial grasses.

In addition, most studies in the forage subgroup (Fig. 1) compared DCD and control emissions from cattle urine patches in grasslands and did not show a significant reduction in N2O emissions. Warmer temperatures associated with the enhanced microbial activity and high root density in grasslands (Gilsanz et al. 2016) can probably lead to DCD inefficacy. Moreover, the high N input from urine depositions (> 1,000 kg N ha−1) and relatively low DCD rate (10 kg ha−1) (Mazzetto et al. 2015; Simon et al. 2020, 2018) possibly contributed to the null effect under warm climatic conditions. A good efficiency of NIs in grasslands has been reported in previous meta-analyses focusing mainly on colder regions, whereas the results in tropical pasturelands are variable. For example, Moir et al. (2007) and Monaghan et al. (2013) reported that DCD significantly reduced N2O emissions from cattle urine in grasslands in New Zealand under a temperate climate. However, Mazzetto et al. (2015) and Barneze et al. (2015) indicated that the same inhibitor at the same rate did not mitigate emissions from an analogous N source in Brazil and in the United Kingdom in summer. Simon et al. (2018) also indicated DCD was more efficient in reducing N2O release from cattle urine in colder rather than under warmer seasons in the Brazilian subtropical zone.

Effect of soil texture

The NIs were found to be highly effective in reducing N2O emissions and in delaying nitrification under fine-textured soils (Figs. 1, 2). Clay particles can promote higher adsorption of NIs and decrease their bioactivity (Barth et al. 2019); however, DMPP is less mobile in soil than DCD and it is more likely to sorb to clay particles (Byrne et al. 2020). Consequently, as DCD was the NI utilized in 70 and 86% of the studies measuring N2O emissions and soil NH4+ and NO3− concentrations in our data set, respectively, the fine-textured soils probably promoted DCD effectiveness by reducing leaching and increasing its persistence in soil, explaining the enhanced reduction of both N2O emissions and nitrification process.

Conversely, NIs showed no effect on delaying nitrification and reducing N2O release under medium-textured soils possibly as a result of the opposite situation, in which spatial separation of DCD from NH4+ was facilitated. It should be noted that studies with very high N rates from cattle urine and relatively low DCD rates (Mazzetto et al. 2015) and with variable results from pig slurry applications (Aita et al. 2015) were grouped into this category, which probably contributed to higher within-subgroup variability and to the null effect on N2O emissions and soil NH4+ and NO3− concentrations. When considering crop yield, a significant NI effectiveness was observed only under medium-textured soils, which is likely to be because they are more vulnerable to NO3− leaching than fine-textured soils. Therefore, as NH4+ is less susceptible to leaching than NO3−, delaying nitrification probably has a greater impact in soils vulnerable to nutrient leaching. The NI effect shown under medium-textured soils was not observed in coarse-textured soils probably because of the relatively low number of studies (n = 4) associated with substantial variability within studies in this subgroup. Thapa et al. (2016) found that NIs were equally effective in decreasing N2O emissions in all soil textures. Abalos et al. (2014) found NIs were less effective in increasing crop yield in fine-textured soils.

Effect of SOC and pH

The SOC and soil pH did not show a consistent influence on NIs effect. Higher SOC contents may stimulate the biodegradability of NIs through the higher microbial activity (Adhikari et al. 2021), although this effect was not verified in our study. The NI efficiency in high-pH soils might be reduced because of enhanced N losses from NH3 volatilization (Thapa et al. 2016; Wu et al. 2021). Nevertheless, our data set included only acidic and neutral soils, which led to similar results in all studied controlling factors. In contrast, Abalos et al. (2014) showed NI efficiency in increasing crop yield and N uptake decreased in neutral–alkaline soils. Yang et al. (2016) found no effect of soil pH on NI performance, which was explained by the absence of an observed increase in NH3 volatilization following NI application, which was also observed in our study (Fig. 3).

Effect of N source

The data set for N2O emissions from urea included six articles, all of which reported significant reduction of N2O emissions. However, 50% of these articles did not report any data variation indicator (SD or SE) (Cantú et al. 2017; Ibarr et al. 2021; Lourenço et al. 2021; Martins et al. 2017; Soares et al. 2015, 2016). It is therefore likely that the conservative estimation of the SD adopted in this meta-analysis (150% of the variance in the data set) associated with the relatively low number of observations within this subgroup resulted in a wide confidence interval and led to the non-significant effect of NIs on N2O emissions (Fig. 1). In contrast to urea, a variable effect of NIs on emissions from cattle urine (Mazzetto et al. 2015) and swine slurry (Aita et al. 2014, 2015, 2019) was reported. The N2O emissions from swine slurry may be mitigated by DCD; however, its performance is variable according to the season and fertilizer application method (Aita et al. 2014, 2019). In addition, elevated NO3− availability relating to previous management and crops might interfere with NI effectiveness (Aita et al. 2015). Akiyama et al. (2010), Qiao et al. (2015), Yang et al. (2016), and Gilsanz et al. (2016) indicated that DCD and DMPP were effective for mitigating N2O losses from mineral and organic N sources.

Effect of N addition rate, NI type, and DCD rates

Our data set revealed that NIs were not efficient under high N rates (Figs. 1b, 4), which is likely to be because the increase in N input was not coupled with an increase in NI rate. Studies with high N addition rates were conducted with DCD, the only NI that was used under substantially variable rates in the analyzed studies. Abalos et al. (2014) suggested that the NI application rate could be lowered if the degradation kinetics of a given inhibitors is low, such as under low temperature and restricted water supply. Nevertheless, in warmer climates with available water for cropping throughout rainfall or irrigation, reduced DCD rates are likely to lead to a null effect. This may be particularly true for DCD because of its higher mobility compared with DMPP (Adhikari et al. 2021; Byrne et al. 2020).

Fig. 4
figure 4

Effect of DCD rates on the effect sizes of A N2O emissions, soil concentration of B NH4+ and C NO3−. The red line indicates the threshold for null response

Full size image

The high DCD soil mobility is ascribed to the polar nature of this compound, while DMPP is positively charged and less mobile as a result of binding with clay and silt particles (Zerulla et al. 2001; Byrne et al. 2020). These characteristics may indicate that NH4+ and DMPP have a similar distribution in the soil profile, which could result in greater effectiveness of this inhibitor. However, our data set did not show differences between the effects of DCD and DMPP. In addition, both inhibitors are water soluble, have a relatively low vapor pressure, and are susceptible to biodegradation (Zerulla et al. 2001). For instance, in an incubation study with grassland soils amended with urine at 15 °C, Chibuike et al. (2022) found that DCD and DMPP have half-life values of 20–33 days and 12–17 days, respectively, whereas temperature, clay, and SOC contents may stimulate microbe activity and NI biodegradation. In contrast to DCD, DMPP and Nitrapyrin are expected to efficiently inhibit nitrification at low rates of approximately 1% of the applied NH4+-N or urea-N. Accordingly, these were the rates utilized in the studies included in our data set.

For DCD, a rate of at least 5% of the applied NH4+-N or urea-N is required, as indicated by the significant correlation between DCD rate and effect sizes of N2O emissions (p ≤ 0.05) or soil NH4+ concentration (p ≤ 0.01) (Fig. 4). Conversely, Adhikari et al. (2021) reported no relation between DCD rate and reduction in N2O release from N-urine patches. However, Gilsanz et al. (2016) found that higher NI rates decreased the emission factor from NI-treated fertilizers. Increasing DCD rates can be a limiting factor as low levels of DCD residues were detected in New Zealand dairy milk powder as a result of 10 kg ha−1 application rates in pastures (Adhikari et al. 2021; Byrne et al. 2020). In addition, in Europe, DCD rates are limited to the minimum and maximum rates of 2.3 and 4.5% (w/w) of the total N in the fertilizer, respectively (European Commission 2014).

Future research perspectives

Data focused on the effectiveness of DMPP and other modern NIs remain insufficient for tropical climates. This is particularly important as excessive application of DCD (Adhikari et al. 2021) and DMPP (Rodrigues et al. 2018, 2019) may present a risk of entrance into the food chain and effects on non-target soil microbes have been reported (Corrochano-Monsalve et al. 2021a). Therefore, life-cycle assessments of NIs are still required to ensure their safe use.

Furthermore, there is evidence that higher temperature can accelerate NI degradation (Mazzetto et al. 2015). Therefore, split application of fertilizers treated with NIs and/or UIs and reduced N rates (supplementary material) should be further investigated in different scenarios, especially in tropical and subtropical regions. For instance, Martins et al. (2017) found that split application of urea-N with Nitrapyrin and NBPT resulted in 23% higher maize grain yield under a tropical climate. Thus, since warm temperatures may decrease NI persistence in soil, it is possible that split N application with reduced application rates could maximize NI performance, but this premise should be confirmed by further studies.

Conclusions

The hypothesis that NI performance is inferior in tropical climates was not supported by our findings. Conversely, the reduction in N2O emissions and soil NO3− concentration was similar among tropical and subtropical sites, and these values were also comparable to those previously reported in meta-analyses that did not consider or underrepresented tropical areas. Moreover, NI effectiveness was higher in irrigated fields, fine-textured soils, and mineral N sources, whereas NIs did not increase NH3 volatilization since conditions were unfavorable for this N-loss path.

Low DCD rates in relation to the amount of applied NH4+-N or urea-N under warm temperatures were considered primarily responsible for the reported failure of NIs to control N losses.

This meta-analysis confirmed that fertilizers treated with NIs or NIs and UIs can reduce N losses in tropical sites. Furthermore, as the increase in crop yield tends to be small, future research should investigate the feasibility of reducing N rates to reduce costs and stimulate the adoption of this technology by farmers. Overall, NI use was found to be an efficient strategy to delay nitrification and mitigate N2O emissions in tropical agroecosystems.