Biological and synthetic approaches to inhibiting nitrification in non-tilled Mediterranean soils

The increasing demand for food production has led to a tenfold increase in nitrogen (N) fertilizer use since the Green Revolution. Nowadays, agricultural soils have been turned into high-nitrifying environments that increase N pollution. To decrease N losses, synthetic nitrification inhibitors (SNIs) such as 3,4-dimethylpyrazole phosphate (DMPP) have been developed. However, SNIs are not widely adopted by farmers due to their biologically limited stability and soil mobility. On the other hand, allelopathic substances from root exudates from crops such as sorghum are known for their activity as biological nitrification inhibitors (BNIs). These substances are released directly into the rhizosphere. Nevertheless, BNI exudation could be modified or even suppressed if crop development is affected. In this work, we compare the performance of biological (sorghum crop) and synthetic (DMPP) nitrification inhibitors in field conditions. Sorghum crop BNIs and DMPP prevented an increase in the abundance of ammonia-oxidizing bacteria (AOB) without affecting the total bacterial abundance. Both nitrification inhibitors maintained similar soil NH4+ content, but at 30 days post-fertilization (DPF), the sorghum BNIs resulted in higher soil NO3− content than DMPP. Even so, these inhibitors managed to reduce 64% and 96%, respectively, of the NO3−-N/NH4+-N ratio compared to the control treatment. Similar to soil mineral N, there were no differences in leaf δ15N values between the two nitrification inhibitors, yet at 30 DPF, δ15N values from sorghum BNI were more positive than those of DMPP. N2O emissions from DMPP-treated soil were low throughout the experiment. Nevertheless, while sorghum BNIs also maintained low N2O emissions, they were associated with a substantial N2O emission peak at 3 DPF that lasted until 7 DPF. Our results indicate that while sorghum root exudates can reduce nitrification in field soil, even at the same efficiency as DMPP for a certain amount of time, they are not able to prevent the N pollution derived from N fertilization as DMPP does during the entire experiment.

beverages from consumers who cannot eat products containing wheat, barley or rye [7]. Besides becoming a safe food product, sorghum is also known for the allelopathy of its root exudates. The molecules that have been characterized in sorghum exudates are sorgoleone [8] and 3,4-hydroxyphenyl propionate (MHPP) [9]. Sorgoleone was initially used for its substantial weed-suppressing capability [10], but later its potential as a biological nitrification inhibitor (BNI) was discovered. More than 80% of the hydrophobic component of sorghum-exuded BNIs is sorgoleone [11], while MHPP comprises the hydrophilic component. Both molecules can reduce nitrification in soil [12,13]. Subbarao et al. [14] characterize the BNI activity from MHPP and sorgoleone in the soil. Once these molecules are released from sorghum roots, their nitrification inhibitory action seems to be relatively stable over a pH range of 3.0 to 9.0. This is in contrast to BNIs release from other plants, such as Brachiaria humidicola, whose molecules are reported to have a total loss of inhibitory function at pH ≥ 8.0. Moreover, the inhibitory effect on soil nitrification of MHPP and sorgoleone appears to be stable in the temperature range of 20 to 30 °C.
The use of crops that produce BNIs could be the first step towards a low-nitrifying agronomic environment in agricultural systems [15]. The tenfold increase in the use of nitrogen (N) fertilizers since the Green Revolution [16] has greatly augmented food production, but it has turned agricultural soils into high-nitrifying environments with significant environmental costs [17]. Nitrification is the biological transformation of N in the form of ammonium (NH 4 + ) into oxidized N. Chemolithoautotrophic ammonia-oxidizing bacteria (AOB) and archaea (AOA) oxidize NH 4 + to hydroxylamine (NH 2 OH) through the ammonium monooxygenase enzyme (AMO) encoded by the amoA gene [18]. NH 2 OH is converted to nitrite (NO 2 − ) and then, nitrite-oxidizing bacteria (NOB) oxidize it to nitrate (NO 3 − ) [19]. MHPP blocks the AMO enzyme, while sorgoleone blocks both the AMO and the HAO enzyme [14]. The chemical structure of sorgoleone, which has a hydroquinone head and a fatty acid tail with a terminal double bond, has the potential to disrupt the electron flow between AMO and HAO enzymes and hence the nitrification activity [20]. Following with the final product of nitrification, as NO 3 − is a negatively charged anion, it is repelled by negatively charged soil colloids and is thus lost through leaching, causing eutrophication and contamination of groundwater supplies [21]. Furthermore, in anoxic conditions, NO 3 − is the substrate for a denitrification process that releases N gases such as nitrous oxide (N 2 O) [22]. The global warming potential (GWP) of N 2 O is 265 to 298 times higher than CO 2 over a 100-year time horizon [23]. Therefore, it is necessary to reduce the pollution originating from the application of N fertilizer, particularly as its use is expected to double by 2050 [24].
Synthetic nitrification inhibitors (SNIs) have been developed to decrease N losses by suppressing soilnitrifier activity. The dimethylpyrazole-based SNI DMPP (3,4-dimethylpyrazole phosphate) is able to reduce AOB abundance, delaying the oxidation of NH 4 + while diminishing N 2 O emissions [25,26]. However, the use of SNIs is not widely adopted by farmers [27]. The inhibitory effect does not last more than a few weeks, they have biologically limited stability, and SNI mobility could prevent these molecules from acting on the sites of nitrification [15,28]. On the other hand, BNIs are exuded directly into the rhizosphere, which is the main site of nitrification due to the great abundance of AOB and AOA [29]. Moreover, sorghum BNIs are known for being stable across a wide range of soil pH and temperature [14]. In addition, BNIs from sorghum can be released until close to physiological maturity of the crop [14], which would ensure the presence of nitrification inhibitors during all the stages of crop development. Nevertheless, BNI exudation is related to the physiological state and development of the plant [30], so biotic or abiotic stresses that affect crop growth might modify the rate of BNI exudation or even prevent its release.
Sorghum allelopathy is highlighted in the framework of sustainable agriculture and its use could drive cultivation systems towards environmentally friendly agronomic practices that allow us to meet global food demand while reducing the loss of reactive N into the environment. For this reason, the aim of this work was to compare under field conditions the performance of biological and synthetic nitrification inhibitors in retaining NH 4 + in soil for longer periods and reducing AOB growth and its effect on N 2 O emissions.

Experimental design
This work was conducted in Pamplona, northern Spain (42° 47ʹ N, 1° 37ʹ W, 450 m above sea level), from May to October 2019. Table 1 describes the soil characteristics of the upper horizon. Daily precipitation and mean temperatures are shown in Additional file 1: Fig. S1. Sorghum (Sorghum bicolor L. var. PR88P68 Pioneer Corteva Agriscience ® ) was sown under no-tillage at a rate of 15 kg seeds ha −1 on 15th May 2019 after a previous hairy vetch (Vicia villosa L.) winter cover crop. The vetch was terminated with 1.5 kg ha −1 dose of glyphosate on 29th April, rate that is routinely applied in no-till system from this region, and left on the soil surface. This experiment consisted of three randomized N treatments with four replications (5 m × 5 m plots). The N treatments were (1) sorghum without fertilizer (Control); (2) sorghum fertilized with ammonium sulfate (AS) and (3) sorghum fertilized with ammonium sulfate combined with DMPP (AS + D). The fertilizer application rate was 150 kg N ha −1 in the form of ammonium sulfate 21%, with the fertilizer hand broadcast on 7th July 2019 in a single application at the beginning of stem elongation (Z30) according to the Zadoks scale [31]. Fertilizer combined with DMPP inhibitor was provided by EuroChem Agro Iberia S.L. The DMPP rate was 0.8% of the NH 4 + -N applied with the fertilizer. As the purpose of this experiment was only to measure the effects of the sorghum crop on N losses, sorghum was not harvested and it was terminated on 14th October 2019 and left on the soil surface according to usual management practices.

Plant analysis
The N isotopic composition in sorghum leaves was determined with an elemental analyser (FlashEA1112 Ther-moFinnigan) coupled to a mass spectrometer (DELTA plus Finnigan MAT) in the Unidade de Técnicas Instrumentais de Análise, Servizos de Apoio á Investigación (SAI), Universidade da Coruña. To do so, one sorghum plant per plot was taken randomly at 10, 20, 30, and 60 days post-fertilization and dried at 80 °C in a circulation oven for 72 h until a constant dry weight was reached. Later, dry plants were ground with a ball miller (Retsch MM 500) at a frequency of 27 s −1 for 2 min. The values of the isotopic ratio of 100 mg of ground material were expressed as δ 15 N, in parts per thousand (‰) relative to atmospheric N 2 . The isotope composition values δ (‰) were obtained with the following equation: where R sample is the 15 N/ 14 N ratio of the plant sample and R standard is the 15 N/ 14 N ratio of the atmospheric N 2 .

Soil analysis
Soil N 2 O emissions were measured using the closed chamber method [36]. Gas samples were taken over 60 days post-fertilization at decreasing sampling frequency from three times per week over 2 weeks to twice per week in the subsequent 2 weeks and, finally, once per week until the end of measuring time. N 2 O samples were measured as detailed in [37].
Soil mineral N was determined based on the soil NH 4 + and NO 3 − contents. Three soil subsamples (3 cm diameter × 0.3 m depth) per plot were taken the day before the treatment application, and later at 10, 20, 30 and 60 days post-fertilization. Then, soil subsamples were homogenized with rocks and roots being removed. The

NH 4
+ and NO 3 − contents were determined as described in [37]. Each day of soil and/or gas measurement two additional soil subsamples (3 cm diameter × 0.3 m depth) were taken randomly from the field to determine soil water content. After removing rocks and roots, they were placed into a circulation oven at 80 °C for 72 h until a constant dry weight was reached. Following [38], soil water content was described as the percentage of waterfilled pore space (WFPS): where 2.65 Mg m −3 was used as particle density. The density of the bulk soil was measured at the beginning of the experiment resulting in 1.0 Mg m −3 .
Soil samples from mineral N determinations at 20 days post-fertilization were used to quantify the abundance of total bacteria (16s rRNA), and nitrifying (bacterial amoA) and denitrifying (nirK) populations. Quantification was done using quantitative polymerase chain reaction (qPCR) in a StepOne PlusTM Real-Time PCR System. Soil DNA isolation and gene amplification were carried out as explained in [39].

Statistical analysis
Data obtained in this experiment were statistically analysed with one-way ANOVA followed by Duncan's multiple range tests for separation of means between treatments using SPSS statistical software (IBM Corp. Released 2016. IBM SPSS Statistics for Windows, Version 24.0. Armonk, NY: IBM Corp). Significant differences are expressed at p < 0.05.

Results
Fertilizer treatments did not have any effect on total bacterial abundance (Fig. 1a). Based on the 16S rRNA gene copy number, bacterial abundance ranged from 1.00·10 9 to 1.10·10 9 . Nitrifying bacteria were also not affected by N treatments, having an abundance of between 9.31·10 6 and 1.01·10 7 amoA gene copy numbers g −1 dry soil (Fig. 1b). Alike AOB, denitrifying microorganisms neither were affected by addition of fertilizer. They showed an abundance that varied from 7.31·10 5 to 8.12·10 5 nirK gene copy numbers g −1 dry soil (Fig. 1c).
After fertilizer application, the soil NH 4 + content increased in AS and AS + D treatments maintaining higher values during the first 30 days post-fertilization (DPF) (Fig. 2a). At 30 DPF, the soil NH 4 + content of fertilized treatments decreased to levels that were similar to the control treatment. However, it was observed that WFPS = soil gravimetric water content × bulk density the soil NH 4 + content from the AS and AS + D treatments increased at 60 DPF, which might have been a consequence of mineralization. On the other hand, the AS treatment showed the highest soil NO 3 − content during the 60 days of measurement (Fig. 2b). Although the AS + D treatment showed constant soil NO 3 − content during the first 20 DPF, it was able to diminish its formation to levels of unfertilized treatment until 60 DPF. Control treatment also showed low soil NO 3 − content. Indeed, the highest NO 3 − -N/NH 4 + -N ratio throughout the experiment was observed in the control treatment (Fig. 2c). The AS and AS + D treatments had equally low ratios during the first 20 DPF because there were no differences between them in terms of soil NH 4 + and NO 3 − content. Nevertheless, because the AS treatment did not decrease soil NO 3 − levels at 30 DPF, the NO 3 − -N/NH 4 + -N ratio showed a sixfold increase compared to the AS + D treatment. Still, at 30 DPF both treatments were able to reduce the NO 3 − -N/NH 4 + -N ratio in AS and AS + D by 64% and 96%, respectively, compared to the control treatment.
Sorghum leaves from the control treatment showed the least negative δ 15 N values (Fig. 3a). The similarity in leaf δ 15 N values between the AS and AS + D treatments until 20 DPF indicated no effect from DMPP up to this point. However, at 30 DPF, both fertilizer treatments showed an increase in δ 15 N values that was greater in the AS treatment than the AS + D treatment, and the same δ 15 N values were maintained until 60 DPF. As expected, the unfertilized treatment possessed the lowest leaf N content (Fig. 3b). Fertilized treatments showed higher N contents that declined throughout the experiment until they reached similar N values to the control treatment at 60 DPF. In this case, there were no differences between the AS and AS + D treatments.
The treatment fertilized with AS showed a substantial N 2 O emission peak at 3 DPF with an emission of 38.7 g N 2 O-N ha −1 d −1 (Fig. 4a). Nevertheless, the peak was quickly reduced from 7 DPF, with N 2 O emissions in the AS treatment maintained between 3.79 and 0.74 g N 2 O-N ha −1 d −1 . In contrast, the N 2 O emissions from the control and AS + D treatments were both low throughout the experiment, ranging from 4.23 to 0.67 g N 2 O-N ha −1 d −1 for the control treatment and from 4.01 to 0.31 g N 2 O-N ha −1 d −1 for the AS + D treatment. Although N 2 O emissions from AS treatment were reduced to levels similar to the control and AS + D treatments, it had the highest total cumulative N 2 O emissions due to the short emission peak (Fig. 4b).
There were no differences between control and AS + D treatments in total cumulative N 2 O emissions, with reductions of 54% and 59%, respectively, compared to the AS treatment.

Discussion
During the last few decades, root exudates from sorghum have been well studied due to the presence of allelopathic substances [40]. Lately, investigations have focused on their ability to inhibit the nitrification pathway. There are several greenhouse and microcosm studies where molecules such as sorgoleone and MHPP have been characterized and their potential as BNIs investigated [8, 9, 12,   14]. One of the main aspects to consider in the attempt to improve agricultural sustainability is whether these substances could have a negative effect on soil health. Encouragingly, while MHPP molecules reduce AOB abundance, they do not exert a general negative impact on the soil bacterial community, as indicated by maintenance of 16S rRNA gene abundance in microcosm experiments [12]. In the same manner, no effects on total bacterial abundance have been observed in soil from potgrown sorghum that release different quantities of sorgoleone [13]. Here, we corroborate that sorghum plants do not alter total bacterial abundance under field conditions (Fig. 1a). This is a confirmation that both sorghum root exudates and the synthetic nitrification inhibitor (SNI) DMPP do not produce general deleterious effects because DMPP also had no effect on 16S rRNA abundance, as reported here (Fig. 1a) and in several other studies [41,42]. Nevertheless, previous work with DMP-based nitrification inhibitors and with sorgoleone has shown that there are some shifts in non-target bacterial abundance, even when the total bacterial abundance is not altered, with SNIs associated with decreases in bacterial diversity [43] and BNIs associated with changes in bacterial networks (BNIs) [44]. These studies are still preliminary, so further work should expand these analyses to determine exactly what effects are exerted by these compounds on the soil microbiota.
Fertilizer stimulates root development, changes the soil pH and increases the availability of nutrients for microorganisms and, consequently, the soil bacterial consortia Control a a a a are greatly influenced [45][46][47]. Specifically, the AOB population exhibits a strong response to N fertilizers [48]. Shortly after the fertilizer application, soils tend to show a large increase in amoA abundance [39,49]. This growth can be avoided by applying SNIs such as DMPP, and the AOB population size is maintained at the level of unfertilized soils (Fig. 1b) [50,51]. Furthermore, the objective of BNIs is to suppress soil nitrification by decreasing the ammonia-oxidizing microorganisms' populations [52]. Interestingly, in the absence of DMPP (AS treatment) we also observed no increases in amoA abundance in soils.
The lack of AOB growth may be associated with the use of glyphosate as an herbicide to terminate winter vetch crop. Several studies have examined the effect of glyphosate on soil microbiology, but the results are highly variable. While some authors described that the use of glyphosate cause negative impacts on microbial community structure [53,54], others affirmed that glyphosate is able to increase soil microbial biomass and respiration [55,56] or, at least, have no significant impact at all [57,58]. Nevertheless, it has been reported that the application of glyphosate at higher rates than in this experiment (1.5 kg ha −1 ) had no effect on AOB and AOA abundances [59,60]. Moreover, glyphosate is routinely applied in no-till systems from this region. Therefore, it is reasonable that we could conclude that the inhibition of AOB growth was due to the action of BNI molecules present in sorghum root exudates. This is the first field demonstration that sorghum can avert AOB growth with the same efficiency as SNIs. Soil mineral N is a useful tool to monitor the activity of AOB based on the oxidation of NH 4 The use of SNIs such as DMPP maintains soil NH 4 + for longer periods due to a delay in NH 4 + oxidation as a consequence of AOB inhibition [61]. When ammoniumbased fertilizers are applied without nitrification inhibitors, soil NH 4 + increases substantially followed by a rapid decrease and the appearance of NO 3 − [62,63]. In our work, AS treatment kept soil NH 4 + content in parallel with the AS + D treatment (Fig. 2a). This could be a consequence of BNIs released by sorghum, demonstrating the ability to maintain NH 4 + content at the same level as DMPP, which aligns with the equal AOB populations in both soils (Fig. 1b). Although 20% of total N losses from field-applied N occur through volatilization of ammonia, the great majority of N losses occur after microbial reactions transform NH 4 + in soils into NO 3 − [64]. Therefore, it seems that the use of BNI could be a good option to reduce soil N losses due to its ability to withhold NH 4 + oxidation derived from the inhibition of AOB growth. Nonetheless, the capacity of BNIs to diminish nitrification seemed to decline over time, as suggested by the daily evolution of soil mineral N (Fig. 2a, b). Although soil NH 4 + and NO 3 − contents were equivalent between AS-and AS + D-treated soils, differences arose after 20 DPF. The NO 3 − -N/NH 4 + -N ratio under AS treatment increased relative to the AS + D treatment (Fig. 2c), which may indicate that the efficiency of BNIs in reducing NH 4 + oxidation only lasted until 20 DPF. It would be interesting to track AOB abundance over time for longer in further studies to examine the effect of this possible decline in BNI activity on AOB growth. These differences in the NO 3 − -N/NH 4 + -N ratio were also associated with an effect on sorghum leaf δ 15  variation in the heavy N isotope ( 15 N) has now being used with increasing frequency in physiological studies related to N metabolism [65][66][67]. For example, N fixation processes, both free and biological, tend to impoverish δ 15 N because the value of the atmospheric N 2 is zero. Thus, because the sorghum crop was sown after winter vetch, which is an N-fixing crop, the leaf δ 15 N values of the sorghum control treatment were below zero (Fig. 3). In addition, [66] noted that plants grown with NH 4 + as the sole source of N demonstrated a decrease in δ 15 N. Therefore, as suggested by [68], δ 15 N can be used as an indicator of the origin of the main N source available to the plant during its development. The lower δ 15 N values of the AS and AS + D treatments indicated that the sorghum crops were exposed to a dominating ammonium nutrition for a longer period than the control treatment. Nevertheless, at 30 DPF, the NO 3 − -N/NH 4 + -N ratio in soils from the AS treatment showed an increase compared to the AS + D treatment (Fig. 2c), probably due to the aforementioned decline in BNI effectivity. This means that sorghum plants under AS had greater access to NO 3 − than sorghum under the AS + D treatment. As a consequence, the δ 15 N values of the AS treatment were less negative than those of AS + D at 30 DPF (Fig. 3a). In addition, δ 15 N values of the fertilized treatments did not attain the values of the control without fertilization. This indicates that while BNIs are less efficient than the synthetic inhibitor DMPP, they still promote a certain amount of plant NH 4 + nutrition. The main process responsible for N 2 O production in soils is denitrification [69]. N 2 O emissions are related to soil water content [70], but the humid Mediterranean climate is characterized by a hot and dry summer. The threshold between water-limited and aeration-limited microbial processes is supposed to occur at soil moisture levels of 60% WFPS [70]. Therefore, since the soil WFPS of the present study did not exceed 25% most of the time (Additional file 2: Fig. S2), it is possible that denitrifiers were not responsible for N 2 O emissions. This may have been due to the lack of variation between treatments in the abundance of the nirK gene at 20 DPF (Fig. 1c). On the other hand, nitrifying microorganisms can also produce N 2 O via nitrifiers' denitrification processes [71]. Nevertheless, although nitrifying populations might have been responsible for the emission peak in the AS treatment at 3 DPF (Fig. 4a), at the same time there were no differences in AOB abundance at 20 DPF between AS and AS + D (Fig. 1b). We hypothesized that the presence of the N 2 O emission peak at 3 DPF and the inhibition of AOB growth at 20 DPF is related to the amount of BNIs released by the sorghum roots. The release of BNIs is influenced by soil NH 4 + content, which at higher concentrations has been shown to stimulate greater BNI release in sorghum roots [72]. Therefore, the limited soil NH 4 + content before the N 2 O measurements did not promote the release of enough BNIs to reduce nitrification in the first few days after fertilizer application. Nonetheless, BNI release increased in sorghum after the addition of ammonium-based fertilizer, inhibiting AOB growth at 20 DPF, and this occurred despite a lack of complete inhibition of nitrification during the first 7 DPF. The fact that the N 2 O emission peak in the AS treatment was reduced before 7 DPF, while the N 2 O emission peaks of fertilized treatments without nitrification inhibitors lasted more than 15 days [51,73] is in line with this hypothesis. During the rest of the experiment, sorghum exudates were able to maintain low N 2 O emissions similar to the AS + D treatment, which showed great efficiency in reducing them, as described in other studies [74][75][76]. Nonetheless, this indicates that even though BNIs have a similar efficiency to SNIs in reducing N 2 O emissions, the delay in BNI release due to the absence of high soil NH 4 + content does not prevent N 2 O emissions in the short-term.

Conclusions
The use of allelopathic substances from plants to reduce nitrification in the soil is a topic of increasing interest. BNI inhibition could be a nature-based solution to diminish N losses, avoiding reliance on new technologies that are not widely adopted. BNIs from sorghum were able to prevent an increase in amoA after N fertilization with the same efficiency as DMPP. Moreover, total bacterial abundance was not affected by either the presence of sorghum roots exudates or by DMPP. In addition, both BNIs and SNIs maintained similar soil NH 4 + contents throughout the experiment. However, sorghum root exudates could not prevent the appearance of soil NO 3 − after 20 DPF, which might indicate that the BNI effect decreases in efficiency after a certain amount of time. While DMPP maintained low N 2 O emissions throughout the experiment, the AS treatment presented one peak at 3 DPF that lasted until 7 DPF. Since the release of BNIs is related to the soil NH 4 + concentration, we hypothesize that the limited soil NH 4 + concentration before the N 2 O measurements did not allow release of enough BNIs to avoid this emission peak. Therefore, although sorghum root exudates can reduce nitrification in field soil, even with the same efficiency as DMPP for a certain amount of time, they are not able to prevent the N pollution derived from N fertilization as DMPP does during the entire experiment.