Introduction

Nitrate (NO3−) leaching from grazed pastures is an environmental concern due to the adverse effects of leached nitrogen (N) on water quality (Cameron et al. 2013). Grazing ruminants excrete 70–95% of their ingested herbage N, mostly in urine (Selbie et al. 2015) and urinary-N concentration is strongly correlated with the amount of N in the diet (Dijkstra et al. 2013). The N loading under a ruminant urine patch can be as high as 1000 kg N ha−1 (Haynes and Williams 1993), often exceeding the requirement of pasture plants by more than 400 kg N ha−1 year−1 (Moir et al. 2011). Nitrate can accumulate in urine patches following the transformation of urea (the predominant form of N in urine) to NH4+, which is subsequently converted to NO3− by nitrifying microorganisms. Nitrate is highly mobile in most soils and leaching from urine patches can account for a significant proportion (average estimated at 20% but with high variability according to season) of total NO3− leaching losses from grazed pastures (Selbie et al. 2015).

Treating grazed pastures with synthetic nitrification inhibitors, such as dicyandiamide (DCD), to maintain urinary-N in the less mobile NH4+ form, is an effective means of reducing NO3− leaching (Monaghan and Barraclough 1992; Venterea et al. 2015). Nitrate leaching from animal urine patches can be reduced by up to 50% when DCD is applied at the recommended rate (Di and Cameron 2016). However, its use in New Zealand has been suspended following the detection of trace amounts of DCD in test samples of dairy products (Astley 2013).

A range of plants have been identified that exude compounds that inhibit nitrification. Tropical pasture grasses (e.g. Brachiaria humidicola), field crops (e.g. Sorghum bicolor), and weeds (e.g. Raphanus raphanistrum) have been shown to reduce soil nitrification through the exudation of biological nitrification inhibitors (BNIs) (O’Sullivan et al. 2017b; Subbarao et al. 2009, 2013). The BNIs isolated from root exudates and plant tissues (root and shoot) have a diverse range of chemical structures including phenolic acids (Rice and Pancholy 1974), fatty acids (Subbarao et al. 2008), phenylpropanoids (Zakir et al. 2008), flavonoids (Subbarao et al. 2013), quinones (Tesfamariam et al. 2014), diterpenoids (Subbarao et al. 2009), monoterpenes (Ward et al. 1997), and isothiocyanates (Brown and Morra 2009).

The genus Plantago (Plantaginaceae) is known to contain a suite of plant secondary metabolites (PSMs) that belong to several of the chemical classes noted above (Goncalves and Romano 2016). In Plantago species, the iridoid (monoterpenoid) glucosides aucubin and catalpol, along with the phenylpropanoid glycoside verbascoside (syn: acteoside), have been identified as biologically active. These compounds have antimicrobial (Rombouts and Links 1956; Rumball et al. 1997), antifungal (Oyourou et al. 2013; Shirley et al. 2015), anti-inflammatory (Park 2013), and anti-oxidative (Yan-Chun and Rong-Liang 1991) properties.

Aucubin and verbascoside occur in ribwort plantain (Plantago lanceolata) at relatively high concentrations (up to 3 and 9%, respectively), depending on cultivar, season, and other factors (Adler et al. 1995; Al-Mamun et al. 2008; Navarrete et al. 2016; Tamura and Nishibe 2002) and there is evidence to suggest that these compounds may act as BNIs in soil. Rauber et al. (2008) observed a decline in N mineralisation in field plots when potatoes were under-sown with P. lanceolata, while soil mesocosms dominated by P. lanceolata had a significantly lower NO3− concentration, mineralisation, and nitrification rates than mesocosms dominated by Anthoxanthum odoratum or Lotus corniculatus (Massaccesi et al. 2015). Additionally, a soil incubation study has demonstrated the suppression of N mineralisation and nitrification following incorporation of plantain leaf material into soil (Dietz et al. 2013).

A domesticated and cultivated line of P. lanceolata (‘Ceres Tonic’, marketed as Tonic) was commercialised in New Zealand over 20 years ago (Stewart 1996) and has become a common forage plant for livestock, particularly sheep, beef cattle, and deer. Its success as a pasture species is due to excellent seasonal production, forage quality, palatability to livestock, and performance in a wide range of environments (Judson et al. 2018). The incorporation of Tonic plantain into the pasture sward has been observed to reduce N-leaching from urine patches (Carlton et al. 2019) with the suggestion that this reduction was partly due to the release of BNIs from the roots of the plant. Additionally, it has been shown that the inclusion of Tonic plantain in the diet of grazing dairy cows resulted in a reduction in total urinary-N output, with average urine N loading rates in the urine patch reduced from 700 to less than 500 kg N ha−1 (Box et al. 2017). An in vitro study suggested that urinary-N output may be reduced due to the influence of the plantain PSMs aucubin and verbascoside on rumen ammonia production (Navarrete et al. 2016).

Analyses of both cattle and goat urine have detected PSM-derived urinary metabolites with excretion profiles that are dependent on the species of plant ingested (Keir et al. 2001; Lane et al. 2006). Therefore, when plantain PSMs such as aucubin, catalpol, and verbascoside are consumed and digested by a ruminant, they and/or their derivatives may accumulate in ruminant urine. Given the BNI potential of these compounds, it is possible that an inhibitory effect within urine patches may be observed, reducing the rate of nitrification and ultimately the leaching of NO3−. However, the fate and transformations of these PSMs during ruminant digestion are not well understood, and the forms and concentrations of breakdown products excreted in ruminant urine are as yet unknown.

Reducing the environmental impact of agricultural livestock production in New Zealand is a key focus in the use of Tonic plantain in pasture systems (Minnée et al. 2019). In this study, sheep were used as the model ruminant to determine if metabolites excreted in the urine after the consumption of plantain can inhibit soil nitrification. We examined multiple plantain cultivars for their urine-excreted BNI potential, with the hypothesis that the degree of nitrification inhibition conferred by urinary metabolites, derived from the consumption of plantain by ruminants, is cultivar-dependent.

Methods and materials

Source of urine

The urine samples used in this study were obtained from sheep participating in trials carried out in the spring and autumn to compare effects of plantain-based versus ryegrass-based diets on animal performance. As there was only a single plot of each cultivar, significances are based on between-animal variation within the cultivars. The trials were conducted at the Agricom Kimihia Research Centre, in Canterbury, New Zealand.

In a spring animal performance trial, 40 fifteen-month-old Romney cross ewes (45 kg live weight average) that had previously been grazing on ryegrass/clover pasture were transferred onto plots (10 sheep per 1.5-ha plot) with different plantain cultivars (P. lanceolata cv. Tonic, ‘Agritonic’, or WDA+) or perennial ryegrass (L. perenne ‘Samson’). Tonic is an upright, winter-active plantain and ‘Agritonic’ was bred from a Tonic population and selected for improved agronomic performance. WDA+ is a breeding line from a gene pool that is winter-dormant with high aucubin concentrations relative to Tonic and ‘Agritonic’. Concentrations of aucubin, catalpol, and verbascoside for these cultivars are shown in Table 1. Aucubin concentrations ranged from 7 to 19 mg g−1, depending on cultivar and season, whereas catalpol was low (≤ 0.5 mg g−1) except in WDA+, which contained up to 6 mg g−1. Concentrations of verbascoside (~ 20–50 mg g−1) were greater than those of aucubin or catalpol, with WDA+ exhibiting the highest concentrations. The diet on offer in the plantain treatments comprised at least 90% plantain, with the balance being weed grasses. Urine was collected manually in a sheep handler at 48 h and 2 weeks after commencement of grazing.

Table 1 Concentrations of aucubin, catalpol, and verbascoside in the three cultivars of plantain used in the grazing trials (adapted from Box and Judson (2018))
Full size table

In an autumn animal performance trial, 20 five-month-old Romney ewes (35 kg live weight average) that had previously been grazing on ryegrass/clover pasture were transferred onto plots (10 sheep per 1.5-ha plot) of either ‘Agritonic’ plantain (> 90% plantain) or ‘Samson’ ryegrass. Urine was collected manually in a sheep handler 2 weeks after commencement of grazing.

Analysis of urine

Immediately following collection, urine samples were analysed for total N after persulfate oxidation (Cabrera and Beare 1993) with concentrations of NH4+-N and NO3−-N in each sample determined concurrently with total N on a Lachat Instruments QuikChem 8500 Series 2 Flow Injection Analysis System (Loveland, CO, USA). A subsample of each urine was set aside for mass spectrometry–based untargeted metabolomics analysis. Ultra-high-performance LC–MS analysis was performed on a Thermo Scientific™ (San Jose, CA, USA) Q Exactive™ Plus Orbitrap (HR/AM) LC–MS/MS coupled with a Vanquish™ UHPLC system (Binary Pump H, Split Sampler HT, Dual Oven). The samples were diluted 1:1 with cold acetonitrile within a Single Step® vial 0.22-μm PVDF (Thompson™ Part No. 65531–200) filter and a 2-μL aliquot of each sample (plus standards of aucubin, catalpol, and verbascoside) was separated by reverse phase chromatography (Accucore™ Vanquish™ C18 1.5 µm, 100 mm × 2.1 mm) maintained at 40 °C, flow rate of 400 µL min−1; mobile phase of 0.1% formic acid in type 1 water (A) and 0.1% formic acid in acetonitrile (B), gradient 0–1 min/0% B, linear increase to 7 min/50% B, linear increase to 8 min/98% B, isocratic to 11 min/98% B, equilibration 11–12 min/0% B, isocratic to end 17 min/0% B. The eluent was scanned from 1 to 11 min by API-MS with electrospray ionisation in negative and positive ion modes. Data were acquired for precursor masses from m/z 110–1200 amu at 70 K resolution with data-dependent ms/ms product ions generated by normalised collision energy (NCE: 30) at 17.5 K resolution. Mass features identified include both the charged molecular ions of the parent molecule and any quasimolecular ions including adducts that might be formed during ionisation, e.g. formic acid adducts. Data were processed with the aid of Xcalibur® 4.0 and Compound Discoverer 2.1 (Thermo Electron Corporation).

Soil incubations with added urine

Soil collection and characterisation

A Templeton silt loam (Udic Haplustept (USDA)) from a site under low-input, long-term un-grazed Italian ryegrass (Lolium multiflorum ‘Andy’) was used in a series of incubation experiments with applied urine. The soil was sampled from the Plant & Food Research farm in Lincoln, Canterbury, New Zealand (43°38′17.8″ S 172°28′27.4″ E). Multiple soil cores were taken to a depth of 5 cm, mixed thoroughly into one composite sample, passed through a 4-mm sieve, and dried in a drying cupboard for 7 days at 20 ± 2 °C. Soil properties are listed in Table 2.

Table 2 Properties of Templeton silt loam soil used in laboratory incubations
Full size table

The potential of this soil to nitrify added NH4+-N was determined by the method of Belser and Mays (1980). Briefly, a soil slurry containing 140 mg N kg−1 (added as (NH4)2SO4) was shaken for 24 h and subsamples obtained at intervals over this time period were analysed for NO3−-N. From a linear regression of the NO3−-N concentration over time, the potential nitrification rate was determined.

Incubation details

The urines collected from individual sheep were combined to provide sufficient volume for all treatments and the composite urines analysed for total N using the method above. To ensure that urinary N inputs were standardised across treatments, the concentration of N was adjusted to that of the lowest N (composite) urine by addition of deionized water.

Samples of soil (20 g air-dry) were weighed into plastic containers and 6.3 mL of urine was added dropwise using an electronic pipette in titrate mode; this volume of urine brought the soil to field capacity (defined as the soil water content at − 10 kPa; actual water content of 32%). Urea (> 99.5%, UNIVAR®, Ajax Finechem) was applied in solution at the same N rate as the urines as a positive control, while a negative control received distilled water with no added N.

The microcosms were covered with Parafilm®, which was pierced to allow for rapid gas diffusion in and out of the microcosms. Microcosms were placed in 5.5-L boxes which were positioned in a split-plot arrangement in an incubator set at 20 ± 2 °C. During the incubation, the soils were maintained at − 10 kPa by gravimetric addition of water, if required. Experimental replicates of each treatment were destructively sampled to determine the net nitrification rate at 1, 7, 14, 21, 28, and 35 days. At each time point, NO3−-N and NH4+-N were determined by 1-h extraction of 5 g soil with 25 mL of 2 M KCl and subsequent analysis of the filtered extract for NO3−-N and NH4+-N on a Lachat Instruments QuikChem 8500 Series 2 Flow Injection Analysis System (Loveland, CO, USA). A further 5 g of soil was used to measure water-extractable C (WEC; 1:6 soil-to-water ratio, shaking time 30 min at 20 °C). The organic C content of the WEC extract was determined using a TOC-VCSH analyser (Shimadzu Corporation, Kyoto, Japan). The phenolic content of the water extracts was determined using Folin-Ciocalteu’s reagent with gallic acid as a standard (Ohno and First 1998).

An identical set of microcosms, prepared as described above, were used for determination of CO2 production. Each microcosm was placed in a 500-mL air-tight jar (fitted with a rubber septum to facilitate gas sampling) and incubated at 20 °C. Production of CO2 was assessed by periodically sampling the headspace of microcosm-containing jars using a gas-tight Hamilton syringe, fitted with a non-coring needle. At each sampling, the headspace was mixed by pumping the syringe three times before a 20-mL sample was withdrawn and the CO2 concentration determined by infrared gas analysis (LI-COR, Lincoln, NE, USA). After sampling, the jars were opened and flushed with fresh air to return CO2 concentrations to ambient before returning them to the incubator.

Nitrification inhibition bioassay

Nitrification inhibition by the urine organic fraction (UOF) was assessed by a bioassay in which the Griess reaction was used to detect NO2− produced by the nitrifying bacterium Nitrosospira multiformis (ATCC® 25198™) as it metabolises NH4+ (O’Sullivan et al. 2017a). To remove inorganic salts (including urea), 30 mL of urine was washed through a 20-mL Phenomenex StrataX C18 SPE column (Torrance, CA, USA) with three column volumes of type 1 water (Purite Select Fusion 40, Suez Water, Thame, UK). The adsorbed organic fraction was eluted with 30 mL of 90:10 v/v methanol:water, after which the methanol was driven off by Speedvac-assisted evaporation (Labconco, Kansas City, KS, USA) and the remaining fraction retained for use in the bioassay.

A cell culture of N. multiformis was revived and maintained in Nitrosolobus medium (ATCC medium 929, pH 7.8). Cultures were maintained in a shaking (70 rpm) incubator at 28 °C and sub-cultured weekly by aseptically transferring 1 mL of the culture into 100 mL of fresh, sterile medium. Prior to use in the bioassay, 500-mL suspensions of the cell culture were centrifuged at 8470 × g for 10 min and washed (3 ×) in N-free medium to ensure removal of residual NH4+ and NO2−. After washing, the pelleted cells were re-suspended in 6 mL of fresh Nitrosolobus medium and 1-mL aliquots were stored at 4 °C until required. These aliquots were centrifuged at 16,000 × g for 5 min and re-suspended in 6.5 mL of fresh medium immediately prior to their addition to the bioassay.

In a 96-well plate, 50 µL of the cell culture was assayed with an aliquot of the UOF, with water added to bring the total well volume to 335 µL. Triplicate wells per UOF were included to enable the time course of NO2− production to be determined. Blank wells contained an aliquot of the UOF without cells. An inhibition control of allylthiourea (ATU), a synthetic nitrification inhibitor, was included at a concentration of 10 µM (determined, in our laboratory, as the minimum inhibitory concentration for our strain of N. multiformis). Assays were performed in triplicate and the production of NO2− was followed spectrophotometrically over a 100-min period at 28 °C in a temperature-controlled CLARIOstar microplate reader (BMG LABTECH GmbH, Ortenberg, Germany). At intervals of 20 min, 15 µL of the Griess colour reagent (sulfanilamide 16.2 g L−1, N-1-naphthylethylenediamine dihydrochloride 0.8 g L−1, concentrated phosphoric acid 41 mL L−1) was added to a set of wells and the colour change reaction allowed to proceed for 5 min before the absorbance was read at 548 nm. Absorbance values were converted to NO2− concentrations using the standard curve (0 to 2.5 mg L−1 NO2−-N) and nitrification rates were calculated from a linear regression of NO2− formation over time. The BNI capacity of UOFs was calculated from the decrease in nitrification rate relative to the uninhibited controls.

Antibacterial/bacteriostatic effects of UOF were assessed using a bacterial growth assay. An aliquot of the UOF was combined with tryptic soy broth in a covered 48-well plate and the growth of an Acidovorax isolate, a facultative anaerobe (Garrity et al. 2005) isolated from soil at Plant & Food Research (Lincoln, New Zealand), was monitored by following the increase in optical density at 600 nm (OD600) at 28 °C in the microplate reader described above. The UOF was tested (in triplicate) over a range of concentrations. A cell-only treatment was used as a positive control, while chloramphenicol antibiotic treatment was included as a negative control.

Statistical analysis

A differential analysis of the UOF constituent profiles was performed using Compound Discoverer™ 2.1 (Thermo Electron Corporation).

Statistical tests on the soil incubation data were performed using Genstat 64-bit Release 19.1 (VSN International Ltd, Hemel Hempstead, UK). Differences between treatment means were tested by one-way analysis of variance (ANOVA) and, if treatment effects were significant at p < 0.05, least significant difference (LSD5%) was used to separate means.

Results and discussion

Nitrogen concentrations in sheep urine

The total N concentration in plantain-fed sheep urine was generally less than half of that in urine of sheep fed with ryegrass (Table 3). This is consistent with results obtained in other studies (Box et al. 2017). The reduction in total N has been attributed to a diuretic effect (O’Connell et al. 2016) or to changes in ruminal fermentation (Navarrete et al. 2016), both thought to be caused by compounds in plantain leaves or compounds derived from a precursor contained in the leaves. Additionally, a number of studies have shown that N concentration was lower in urine of dairy cows grazing on “diverse” pastures that included a plantain component than in urine of cows grazing on grass-clover pasture (Bryant et al. 2017; Edwards et al. 2015; Minnée et al. 2017; Totty et al. 2013; Woodward et al. 2012).

Table 3 Total N concentrations of composite sheep urines used in the laboratory soil incubation experiments
Full size table

The variability in urine-N concentration between individual sheep was large, with total N concentrations across the three experiments ranging from 1.3 to 17.4 g N L−1 in ryegrass-fed sheep and from 0.9 to 6.1 g N L−1 in sheep fed with plantain. This variability is consistent with other studies, with Hoogendoorn et al. (2010) observing a range of 1.4 to 17.8 g N L−1 in urine of sheep grazing a ryegrass/cocksfoot/white clover pasture while Bristow et al. (1992) reported a range of 3.0 to 13.7 g N L−1 for sheep on a ryegrass/white clover diet.

Nitrate–N was not detected in any of the urine samples. Ammonium-N comprised between 0.6 and 13% of urine-N with concentrations of NH4+-N usually greater in plantain-fed sheep urine than in urine of ryegrass-fed sheep (Table 4).

Table 4 Ammonium-N concentration in the composite sheep urines used in laboratory incubation experiments
Full size table

Organic constituent profiles in urine of sheep fed with plantain or ryegrass

Analysis of the metabolite profile of the whole urine samples (collected in the autumn animal performance trial) separated by reverse phase LC–MS identified almost 3000 mass features. Of these mass features, 94 were present in greater concentrations than creatinine, a biomarker commonly used to normalise other urinary metabolites. This normalisation was to account for differences in the time of urine collection, urine concentration, urine flow rate (Warrack et al. 2009), and in the type and availability of forage consumed by the animal (David et al. 2015; Ma et al. 2014). Differential analysis (p ≤ 0.005, log2 fold ≥ 2) showed there were more than 800 metabolites that were either unique to, or present in greater concentrations, in the plantain urine than in the ryegrass urine (designated in light grey in Fig. 1). Based on predicted formulae, N-containing mass features, defined as “non-urea nitrogenous components” (NUNCs), were almost twice as numerous in plantain urine as in the ryegrass urine (524 and 283, respectively).

Fig. 1
figure 1

Volcano plot of the mass feature profiles (separated by reverse phase chromatography) of whole ryegrass urine and whole plantain urine collected in the autumn animal performance trial. The x-axis specifies the fold-change in concentration and the y-axis specifies the negative logarithm (base 10) of the t-test p-values. Compound related mass features unique to, or present in much greater concentrations, in the plantain urine than in the ryegrass urine are indicated on the positive side of x-axis. Compound mass features designated in light grey are those that are either unique to, or present in greater concentrations, in the plantain urine than in the ryegrass urine (p ≤ 0.005, log2 fold ≥ 2)

Full size image

Interrogation of the mass spectral data of the major metabolites detected in the urine suggested that many of these were sulfated, methylated, and glycine conjugates. A number of the metabolites were identified based on known ovine metabolites (http://lmdb.ca/metabolites). Hippuric acid, phenylacetylglycine, p-cresol sulfate, N-cinnamoyl glycine, and related ovine metabolites exhibited similar profiles regardless of whether the urine came from animals fed ryegrass or plantain diets.

The plantain PSMs aucubin, catalpol, and verbascoside were not identified in the urine samples. However, based on the mass spectral fragmentation patterns, metabolic derivatives of these compounds were present (Ghisalberti 1998; Hattori et al. 1990; Qi et al. 2013; Visioli et al. 2003).

Urine-induced nitrification inhibition in soil microcosms

The soil used in the incubation studies was a slightly acidic silt loam with an organic carbon content of 29 g kg−1 and nitrification potential of 9 mg kg−1 day−1 (Table 2). This soil type is representative of many sheep-grazed pastoral soils in the South Island of New Zealand.

In the first incubation, urine collected from sheep grazing (in the spring) on plantain cultivars (Tonic or WDA+) and Samson ryegrass for 48 h was applied to the soil at a rate of 511 µg N g−1 soil. The NO3−-N production in the plantain cultivar (Tonic or WDA+) urine treatments was similar to that from the ryegrass urine treatment (Fig. 2A). Over the first 21 days, significantly less NO3−-N accumulated in the urine-treated soils than in the soil receiving urea (229, 236, and 223 µg NO3−-N−1 g soil for Tonic, WDA+, and ryegrass urine, respectively, versus 279 µg NO3−-N g−1 soil for urea; LSD5% = 12). The synthetic inhibitor DCD (at the recommended rate, equivalent to 30 kg ha−1) was effective in suppressing nitrification in urea-treated soil and less NO3−-N was produced in the urea + DCD treatment compared to the control receiving no N (31 and 52 µg NO3−-N g−1 soil over the 35 days, respectively; LSD5% = 10).

Fig. 2
figure 2

Concentrations of nitrate–N (NO3−-N) in soil microcosms during a 35-day incubation following the application of sheep urine collected after (A) 48 h or (B) 2 weeks of grazing on plantain cultivars (Tonic or WDA+) compared with NO3−-N concentrations in microcosms treated with ryegrass urine, urea or urea + DCD. The error bars represent LSDs (p < 0.05) at each incubation time

Full size image

Tonic plantain has been shown to induce diuresis in sheep after only a 24-h grazing period (O’Connell et al. 2016). While diuresis was evident by way of a reduction in total N concentration (Table 3), the inhibitory effect on nitrification was not obvious. This finding is echoed in the work of Podolyan et al. (2019) and suggests that grazing plantain for 48 h is not adequate to confer a sufficient concentration of the plantain PSMs responsible for nitrification inhibition to the urine.

In the second incubation (Fig. 2B), the urines collected after 2 weeks of grazing (in spring) were applied at a slightly higher N application rate (773 µg N g−1 soil). Significantly less NO3−-N was produced in the Tonic treatment than in the control or urea treatments at intervals up to 17 days. The cumulative amount of NO3−-N produced in 17 days in the Tonic treatment was 146 µg NO3−-N g−1 cf. 241 µg NO3−-N g−1 (LSD5% = 15) when ryegrass urine was applied. There was no significant difference in NO3−-N production between the WDA+ and ryegrass treatments at any point during the incubation. Nitrate–N production in the urine-treated microcosms converged at approximately 25 days, although NO3−-N produced at this point remained less than in the urea treatment (~ 320 µg versus ~ 390 µg NO3−-N g−1 soil). There was no significant difference in the treatments at 28 days, and after 35 days, NO3−-N production in the Tonic microcosms was significantly greater than in the other treatments. The results from these two incubations suggest that compounds in the urine of sheep grazing on plantain have nitrification inhibition capability when applied to soil but the degree of inhibition depends on the length of the grazing period prior to urination and on the cultivar.

In a third incubation, urine collected from sheep grazing ‘Agritonic’ plantain or ‘Samson’ ryegrass for 2 weeks (in autumn) was applied at a rate of 802 µg g−1 soil). In this incubation, the urine of sheep grazing ‘Agritonic’ plantain had a particularly strong inhibitory effect on the production of NO3−-N (Fig. 3A). After 14 days of incubation, NO3−-N produced in microcosms treated with urine of ‘Agritonic’-fed sheep (44 µg g−1) was significantly less than that produced in the negative control receiving no N (61 µg g−1), and was ~ 5.5 times less than that produced in microcosms treated with urine of ryegrass-fed sheep. Throughout the incubation, there was no discernible difference in NO3−-N production between the ryegrass urine and urea treatments. Until day 28, NO3−-N production in the plantain urine microcosms was significantly less than in the ryegrass and urea treatments, but the difference diminished with time and was no longer statistically significant by day 35. These results indicate that the inhibitory effect of the plantain urine declined with time, presumably due to degradation and/or inactivation of the active compounds.

Fig. 3
figure 3

NO3−-N (A) and NH4+-N (B) concentrations and cumulative CO2-C production (C) during a 35-day incubation for soil treated with urea or urine of sheep grazing on either ‘Agritonic’ plantain or ryegrass (N addition rate 802 µg N g−1). The error bars represent LSDs (p < 0.05) at each incubation time

Full size image

Measurements of NH4+-N during the incubation showed that hydrolysis of urea in both the plantain and ryegrass urine treatments was rapid, with near-complete conversion to NH4+-N within 1 day (Fig. 3B). In contrast, less than half the urea-N was in the NH4+-N form after 1 day. The rapid hydrolysis of urine-urea may have been caused by contamination of the urine by faecal ureases, or because of intrinsic ureases produced by the urinary microbiome (Bao et al. 2017; Whiteside et al. 2015). However, in the case of the urea treatment, hydrolysis was reliant on ureases produced by the soil microbiota. In both the urea and ryegrass urine treatments, NH4+-N gradually decreased between days 7 and 35, with little difference between the two treatments for the remainder of the incubation.

In the plantain urine treatment, there was a significant increase (166 µg g−1 soil) in NH4+-N between days 1 and 14, which was not accounted for by urea-hydrolysed N. The probable source of this NH4+-N was organic matter mineralisation, brought on by the large increase in dissolved organic matter observed following application of plantain urine (discussed below). Additionally, the mass spectroscopy data suggested that ‘Agritonic’ urine was rich in nitrogenous compounds and these compounds may have contributed to the greater mineralisation of N in microcosms treated with ‘Agritonic’ urine. Between days 14 and 35, NH4+-N in the soil decreased, but concentrations remained greater than in the urea and ryegrass urine treatments (by ~ 250 µg N g−1 at day 35).

In the first 7 days of incubation, there was a pulse of respiration (CO2 production), particularly where plantain urine was applied (plantain urine > ryegrass urine > urea > control) (Fig. 3C). In the urea treatment, a substantial part (~ 57%) of the pulse can be attributed to CO2-C released by hydrolysis of urea (estimated at 344 µg C g−1), with the remainder being due to mineralisation of soil organic matter solubilised as a result of the urea-induced pH increase (from pH 5.6 to 6.9, data not shown). Between 7 and 35 days, there was only a gradual increase in the production of CO2 that was similar in all treatments, including the control. The additional quantity of CO2-C generated in the plantain urine treatment relative to the urea treatment (i.e., CO2-C released by mineralisation of UOF and/or soil organic matter) in the first 14 days amounted to ~ 1400 µg C g−1. This quantum of C mineralisation is compatible with the additional N mineralised (~ 160 µg g−1) in the first 14 days after application of plantain urine, assuming a ratio of ~ 10:1 between mineralised C and mineralised N. Although addition of ryegrass urine also increased CO2-C production, there was no evidence of an associated increase in N mineralisation (total mineral N did not differ significantly from that in the urea treatment between days 7 and 35).

The respiration pulse following plantain urine addition was associated with a large increase in water-extractable organic C (WEC) in the soil (Fig. 4A). Within 1 day of incubation, WEC had increased from ~ 100 µg g−1 in the control to 1570 µg g−1 in soil treated with plantain urine, before declining to ~ 300 µg g−1 on day 7. This rapid decline in WEC was presumably due to its decomposition by soil microorganisms. The WEC was probably the source of a large part of the CO2-C produced from the plantain urine treatment during the early part of the incubation. An increase in WEC was also observed following addition of ryegrass urine and urea, but it was a much smaller amount than in the plantain urine treatment. The WEC in the plantain urine treatment initially (day 1) had a large proportion (ca. 20%) of phenolic-C (Fig. 4B), the source of which was presumably phenolic compounds in the plantain urine. Phenolic compounds are known to be inhibitory to nitrification (Castaldi et al. 2009; Kholdebarin and Oertli 1994; Lodhi and Killingbeck 1980; Rice and Pancholy 1974) but a causal link between specific urinary phenolic metabolites and nitrification inhibition remains to be established.

Fig. 4
figure 4

Water-extractable C (A) and the proportion of phenolic-C (B) over 35 days of incubation for soil treated with urea or urine of sheep grazing on either ‘Agritonic’ plantain or ryegrass. The N addition rate was 802 µg N g−1. The error bars represent LSD (p < 0.05) at each incubation time

Full size image

A secondary finding in this incubation study was that while urea solution may closely mimic urine from a ryegrass system, it is not necessarily a good proxy for urine from animals fed other forages. Indeed, “artificial” urine has previously been shown to have different N dynamics compared to “natural” urine (Kool et al. 2006), and these results further support why natural urine has previously been recommended for experimental use (de Klein et al. 2003).

Nitrification inhibition bioassay

With the incubation results demonstrating that the urine of plantain-fed sheep contained compounds capable of inhibiting nitrification in soil microcosms, the ability of the organic fraction of urine (UOF; derived from the urines sourced from the autumn animal performance trial) to inhibit nitrification was tested in vitro. Figure 5A shows that the addition of a small (1.4 µL) aliquot of ‘Agritonic’ plantain UOF substantially reduced the amount of NO2− produced by N. multiformis in a nitrification inhibition bioassay. The UOF from ryegrass urine produced no inhibition when added at doses up to 50 µL (maximum tested). The antibacterial/bacteriostatic effects of the UOF were assessed using a bacterial growth assay (Fig. 5B). The same aliquot of plantain UOF tested in the bioassay caused a slight delay in the growth of the Acidovorax isolate relative to the control. However, this delay was less than that caused by the synthetic inhibitor allylthiourea (ATU) and by the ryegrass UOF.

Fig. 5
figure 5

The effect of plantain- and ryegrass-derived urine organic fraction (UOF) and allylthiourea (ATU) on (A) the production of nitrite (NO2−) by Nitrosospira multiformis and (B) the growth curve for Acidovorax isolate as monitored by an increase in optical density at 600 nm (OD600). The error bars represent the LSD5%

Full size image

To further explore the inhibitory effect of ‘Agritonic’ plantain UOF, a preparation that had been freeze-dried and reconstituted to a known concentration was added in increasing quantities (up to 180 µg) to the bioassay (Fig. 6). The degree of nitrification inhibition increased asymptotically as the UOF rate increased, with 50% inhibition at the 25-µg addition rate and near-complete nitrification inhibition at 150-µg addition, indicating that compounds in plantain urine were strongly inhibitory.

Fig. 6
figure 6

Nitrification inhibition (% NO2−-N production inhibited relative to uninhibited control) curve for Nitrosospira multiformis challenged with increasing concentrations of an ‘Agritonic’ plantain-derived urine organic fraction (UOF) in a bioassay volume of 335 µL. The error bars represent the LSD5%

Full size image

These results suggest that the urine of ‘Agritonic’ plantain-fed sheep contained compounds that were not overtly bactericidal but were inhibitory toward ammonia monooxygenase (the enzyme that catalyses the conversion of NH4+ to NO2−); the amount of ‘Agritonic’ plantain UOF that caused a significant inhibition of nitrification relative to the control caused only a small and brief retardation of bacterial growth.

In this bioassay, a single bacterial species was challenged with the UOF, so caution is required when interpreting these results. Soils contain a consortium of nitrifying organisms that may differ in their sensitivity to the UOF. It has been shown that the response of nitrifying microorganisms (NH4+-oxidising archaea and bacteria) to synthetic nitrification inhibitors varies considerably (Lehtovirta-Morley et al. 2013; Shen et al. 2013). Therefore, it is essential to confirm that inhibition occurs under field conditions.

Conclusions

Soil incubation studies showed that net nitrification after the application of plantain urine was reduced compared to ryegrass urine, with the urine of sheep fed with the plantain cultivar ‘Agritonic’ producing the strongest reduction in NO3− produced. Mass spectroscopic analyses showed that plantain urine had many more N-containing compounds (excluding urea) than ryegrass urine; these included metabolic derivatives of the plantain bioactive compounds aucubin, verbascoside, and catalpol. In vitro assays suggested that the reduction in NO3− production by compounds excreted in plantain urine was due to the inhibition of nitrification, with ammonia monooxygenase being the target enzyme. In the future, targeted metabolomic analyses combined with bioassay-directed fractionation may be useful in identifying which organic metabolites have a role in suppressing nitrification and in providing targets for breeding programmes to develop cultivars that confer superior urinary-excreted BNI activity. However, the effect of these metabolites on other N transformations in the soil that reduce NO3− concentrations cannot be discounted and studies that consider gross N transformations should be considered. The activity of urinary BNI metabolites in soil may be affected by a multitude of factors and synergistic effects will likely influence the magnitude of their effect in mitigating NO3− leaching losses from agricultural systems.