Storage of soil samples leads to over-representation of the contribution of nitrate to plant-available nitrogen

Keywords: air-dry, amino acids, ammonium, in situ soil sampling, microdialysis, nitrogen flux, organic nitrogen, refrigerated storage, soil testing.

Nitrogen (N) in soils is a highly dynamic nutrient, rapidly converting between different forms and moving through the soil profile. Nitrogen pool dynamics are driven by the chemical, physical and biological characteristics of the soil, and can drastically change with a stress applied to any of these conditions, such as during waterlogging and anoxia (Kuypers et al. 2018; Wooliver et al. 2019). This presents a considerable challenge for quantification of soil N pools, where disturbance and removal from the in situ conditions are often unavoidable.

Soil N analysis is an essential undertaking of agronomy and field research in agriculture, particularly when investigating nitrogen use efficiency (NUE) in production systems. Typical analyses include measurement of nitrate, ammonium and, increasingly, dissolved organic N (DON), using water or salt extracts (Mulvaney 1996; Rayment and Lyons 2011). In agronomy settings, testing for soil mineral-N (nitrate and ammonium) can be a key determinant of the amount of fertiliser applied for the subsequent crop. However, inconsistency in sampling methods and data interpretation makes test results less reliable for growers (Schwenke et al. 2019). In research trials comparing N fertiliser forms, application rates, management techniques or loss prevention strategies, gaining accurate measurements of nitrate, ammonium and DON is critical to evaluating plant N availability, susceptibility to loss and ultimately NUE (Shaw et al. 2014; Rochester and Bange 2016; Macdonald et al. 2017a; Allen et al. 2019). Ensuring routine sampling and analysis procedures give accurate representations of N form distribution is crucial to both developing effective strategies for improving NUE and informing improved fertiliser management.

Given the dynamic nature of soil N, it is expected that transformations will continue after a sample is removed from the field (Li et al. 2012). To minimise changes in N pools, samples are stored in either refrigerated or air-dried conditions to slow or limit microbial activity (Stenberg et al. 1998; Verchot 1999; Lee et al. 2007; Černohlávková et al.2009). Studies comparing storage treatments have shown a large increase in nitrate with drying and sieving (Jones and Willett 2006; Ros et al. 2011; Inselsbacher 2014; Gütlein et al. 2016). Fewer studies have investigated the effects of refrigerated sample storage, despite often being the recommended alternative to air drying (Jones and Willett 2006; Lee et al. 2007; Černohlávková et al.2009). Even fewer have compared refrigerated samples to in situ analysis. One study covering both these aspects found a large increase in nitrate and decrease in ammonium concentration, however it did not include measurements of organic N (Arnold et al. 2008).

Determining the importance of DON, such as amino acids, in-field settings is hindered by the low concentration and rapid flux through microbial consumption and production (Warren 2014). This effect is exacerbated during soil sampling and extraction. A rapid decline in added amino acids has been observed across a range of soil processing and extracting methods, including sieving and during extraction with water, potassium chloride (KCl) or potassium sufate (K₂SO₄) (Rousk and Jones 2010; Inselsbacher 2014), illustrating the continued cycling of soil N after sampling. Yet, these observations lack comparison to in situ amino acid levels. To circumvent these problems, microdialysis has been developed as a minimally disruptive technique for measuring fluxes (rather than concentration) of small N molecules in soil (reviewed by Buckley et al. 2020).

Microdialysis samples solutes from the soil solution via diffusion across a semi-permeable membrane into a probe inserted into the soil. The probe sampling surface is small (e.g. 30 mm length, 0.5 mm diameter), and the membrane has a specific molecular weight cut-off (e.g. 20 or 100 kDa), so will exclude larger molecules or microorganisms (Inselsbacher et al. 2011; Buckley et al. 2020). Microdialysis is considered to provide a better representation of plant-available N due to being minimally disruptive, measuring a flux rather than pool size (Hobbie and Hobbie 2012), and being analogous to plant roots in the size of the probe and in sampling by diffusion (Brackin et al. 2017). Microdialysis sampling has been used in situ on undisturbed soils, and demonstrated that amino acid-N can comprise a greater proportion of the soluble N pool compared to traditional extracts (Inselsbacher and Näsholm 2012a; Brackin et al. 2015; Leitner et al. 2017). However, microdialysis samples collected from soils that have been stored tend to show a dominance of nitrate or ammonium over amino acids (Shaw et al. 2014). This suggests a need to re-evaluate both soil sampling and storage methods for measurement of small organic N compounds in soils.

Soil sample storage conditions and sampling technique evidently affect observations of soil N pools. Using minimally disruptive soil sampling techniques like microdialysis, it is possible to directly evaluate the deviation from in-field soil N status occurring during sample storage and processing (Inselsbacher and Näsholm 2012a; Brackin et al. 2015; Warren 2018; Buckley et al. 2020). In this study, we consider in situ microdialysis sampling of soil N flux as the measurement most representative of plant-available N (Brackin et al. 2015), and compare how sample storage duration and condition (refrigerated at 4°C or dried) affects levels of available N measured in traditional KCl extracts and microdialysis samples. These sampling procedures were applied in the field and to soil collected from a cotton field trial comparing high and low N fertiliser rates, giving direct relevance to a typical soil sampling scenario. We hypothesise that microdialysis will show a greater availability of amino acids relative to inorganic N than KCl extracts, and that organic N levels will decrease with duration of soil sample storage.

The study site was located at Riverview (28°35′56″S, 149°49′37″E; 192 m asl), near Toobeah, Queensland, Australia. The soil is classified as a Grey Vertosol (Vertisol in the USDA NRCS Soil Taxonomy) with the dominant clay mineral being Montmorillonite (Soil Survey Staff 1999; Isbell 2016). Selected soil properties are summarised in Table 1. Samples were collected from a skip-row furrow-irrigated field trial with two varieties of cotton (Sicot 714 B3F and Sicot 746 B3F) and two N fertiliser rates (172 and 309 kg N ha⁻¹). The field was arranged into plots of 24 rows wide with 1.02 m row spacing and 14 plants m⁻¹. Each fertiliser rate was applied to four replicate plots, arranged in randomised blocks, with the two varieties as subplots (12 rows). All plots received 147 kg N ha⁻¹ as urea before planting, and 25 kg N ha⁻¹ with irrigation 88 days after planting. The high N rate plots received an additional 137 kg N ha⁻¹ 50 days after planting. Samples were collected on 22 January 2019 (mid-growing season, 99 days after planting, 3 days after irrigation).

Table 1.  Properties of soil sampled from a cotton field, near Toobeah, Queensland, Australia. All measurements taken from 0 to 15 cm depth, before planting of the field trial, and reported on dry weight basis. Clay, <0.002 mm; silt, 0.002–0.02 mm, sand, 0.02–2 mm.

Microdialysis sampling was performed using probes with a 30 mm membrane length, 0.5 mm outer diameter and 20 kDa membrane cut-off (CMA 20, CMA Microdialysis). Probes were attached to a four-channel syringe pump (CMA 4004) set to a flow rate of 5 μL min⁻¹ with deionised water as the perfusate (Inselsbacher et al. 2011). All dialysate samples were collected over 60 min for microdialysis sampling in both the field and laboratory microcosms. Probe calibration and recoveries of nitrate, ammonium and amino acids under these sampling conditions can be found in Buckley et al. (2017).

In situ sampling of nitrate, ammonium and amino acid diffusive flux using microdialysis was conducted on the shoulder of the cotton hill. A shovel was used to dig a vertical cut into the shoulder and the microdialysis probe inserted horizontally into the soil face at 100 mm depth. One sampling location was used in each field replicate plot. Four probes were inserted at each sampling location (separated by approximately 100 mm to avoid interference), resulting in four technical replicates which were analysed separately for concentration of each N form, then averaged to form a single sample per field plot. Soil temperature was measured at 100 mm depth by inserting a probe thermometer during sample collection (see Supplementary Table S1, available at the journal website).

Soil samples were collected from each field plot as soon as microdialysis sampling ended. Soil was collected to 100 mm depth and adjacent to where microdialysis probes were placed to avoid any possible confounding effects of in situ microdialysis sampling. One soil sample was collected by trowel, placed in a plastic zip-lock bag, and mixed. Sub-samples were used in ‘disturbed microdialysis’ and ‘field extractions’ described below. The remaining soil sample was kept in a portable refrigerator at 4°C while stored overnight and during transport to the laboratory the following day. A second soil sample was collected using a bulk density ring (diameter, 48 mm; length, 38 mm) and placed into a second zip-lock bag. This was immediately weighed on a balance in the field to determine current soil fresh bulk density.

To evaluate the effects of disturbance alone (without storage delays) on soil N measurements, microdialysis samples were taken in the field, immediately after soil sampling to compare to samples collected in situ. Mixed soil was placed into 50 mL microcosms (adjusted method from Inselsbacher et al. 2009) and tapped to the correct bulk density for its plot of origin (soil height, ∼970 mm). Microcosms were returned to the field and buried vertically with the opening level with the soil surface to maintain soil temperature. Microdialysis was then conducted using the setup described earlier, with one microdialysis probe inserted from the top per microcosm.

Soil extracts were conducted in the field using 5 g fresh, homogenised soil mixed with 25 mL of 1 M KCl in a 50 mL centrifuge tube. The tubes were laid horizontally on an orbital shaker and mixed for 30 min, then left to stand until settled. One mL of supernatant was removed by pipette and stored in the portable refrigerator for transport to the laboratory, where it was stored frozen until analysis.

The next day after field sampling, soil samples stored at 4°C overnight were prepared for microdialysis in microcosms and extractions as described above. Soil was returned to the original in-field temperature before microdialysis was performed in order to prevent temperature induced artefacts in diffusion rates (Inselsbacher and Näsholm 2012b).

The remaining soil samples were split into two halves: one half was stored at 4°C; and the other half was air dried at 25°C until constant weight. After 1 month, soil extracts and microdialysis perfusate samples were collected using the same microdialysis microcosms and KCl extract procedures. Air-dried soils were rehydrated to the original moisture content due to the soil moisture limitations of microdialysis sampling and the potential influence on solute recovery (Miró et al. 2010). Immediately before microdialysis and extraction, deionised water was added to return the soil samples to their mass at the time of storage. Soil water content at the time of sampling was estimated by measuring the drained upper limit (DUL) at 100 cm suction (DUL₁₀₀) using a pressure plate, and was equal to 0.43 ± 0.01 g cm⁻³, assuming the soil was near field capacity 3 days after irrigating (Dalgliesh et al. 2016). DUL is the highest water content of a soil after it has been thoroughly wetted and allowed to drain until drainage becomes negligible (Ratliff et al. 1983). When DUL is measured under controlled laboratory conditions using a suction plate, it is taken to be equivalent to soil water retained at a suction of 100 cm and is referred to as DUL₁₀₀ (Marshall 1959).

All KCl extract and microdialysis samples were stored frozen at −20°C until analysis. Samples collected in the field were stored at 4°C during transport to the laboratory, then transferred to a freezer. Samples were analysed for nitrate using the colorimetric vanadium (III) chloride method (as per Miranda et al. 2001), and ammonium and amino acids via ultra-performance liquid chromatography (UPLC) (as per Holst et al. 2012).

The soil properties listed in Table 1 were obtained using standard methods described by Rayment and Lyons (2011): pH and electrical conductivity via 1:5 soil:water suspension; total C, total N and organic C (after pre-treatment with sufurous acid (H₂SO₃) to remove carbonate) using a LECO TruMac CN analyser (LECO Corporation); effective cation exchange capacity as sum of exchangeable bases measured by leaching with alcoholic 1 M ammonium chloride (NH₄Cl) at pH 8.5; and particle size analysis via hydrometer.

Data analysis was performed using R software (R Core Team 2020). Data for amino acid-N, ammonium-N and nitrate-N concentrations in KCl extracts and flux in microdialysis samples were analysed separately. Samples from plots growing the two cotton varieties were used as replicates, giving eight replicate samples of each N fertiliser rate. All data were log₁₀ transformed to meet the model assumptions of normally distributed residuals. The effect of storage treatment and N fertiliser rate on N species concentration or flux was analysed using a mixed-effects linear model using the nlme package (Pinheiro et al.2020). Field plot was included as a random effect for the slope in the model. Tukey pairwise comparisons and P-values were determined using the emmeans package (Lenth et al.2020).

Data from KCl extracts and microdialysis samples cannot be directly compared since the former measured a concentration (mg N kg soil⁻¹) while the latter measured a flux (nmol N cm⁻² h⁻¹) of N compounds. To allow for comparison, the proportion of N in each form was calculated by dividing the concentration or flux of amino acid-N, ammonium-N and nitrate-N by the sum of N in all three forms (total measured-N) in each sample. This was analysed for the effect of sampling method and storage treatment using a mixed-model using the nlme package, including an interaction term for the two fixed effects. Proportional data were averaged across N fertiliser rates and cotton varieties, as no significant effect was found (P > 0.05), and plot was included as random slope in the model. The emmeans package was used to calculate P-values for Tukey pairwise comparisons across sampling method and storage treatments.

Overall, there was no difference between the 172 (low) and 309 (high) kg N ha⁻¹ fertiliser rates in soil ammonium-N and nitrate-N concentration measured in 1 M KCl extracts (Fig. 1b, c). In contrast, amino acid-N concentration was higher in extracts from the low N rate plots (2.7 ± 0.3 mg N kg⁻¹; mean ± s.e.) than the high (1.8 ± 0.1 mg N kg⁻¹) (P = 0.007; Fig. 1a). No interaction between N rate and storage treatment was found in either the KCl extract or microdialysis data (P > 0.05).

Fig. 1.  Concentration of amino acid-N (a), NH₄–N (b) and NO₃–N (c) (mg N kg⁻¹ dry soil) in soil collected from two N fertiliser rates (172 and 309 kg N ha⁻¹) of a cotton field trial, measured using a 1 M KCl extract either in the field, the day after sampling, or after 1 month of refrigerated or air-dried storage. All samples were returned to field moisture before extraction. Values are means ± s.e. (n = 8). Letters show significant differences within N forms (amino acids, bold italics; nitrate, UPPERCASE).

Within each N rate, there was no statistically significant change in the extracted amino acid-N or ammonium-N concentration between the storage treatments (Fig. 1a, b). In soil from the low N rate plots, amino acid-N slightly decreased from soil extracted in the field to stored soils, although this effect was not significant. In soil from the high N rate plots, ammonium-N was similar between the field and air-dried treatments, and slightly lower in the next day and refrigerated soils, but no statistically significant differences were found.

Nitrate-N concentration measured using KCl extracts showed the greatest difference between storage treatments, with a similar pattern across the two N rates (P < 0.001; Fig. 1c). Nitrate-N content was 2 and 3.5 times higher in soil extracted the next day than in the field, in both low and high N rate plots, respectively; although there was high variation in the soils processed the next day and therefore the difference was not statistically significant (P = 0.49). Soil samples stored for 1 month, either refrigerated or air-dried, had nitrate levels significantly higher than soil extracted in the field or the next day. Nitrate in refrigerated soil was 6.2 and 4.0 (low and high N rate) times higher than field extracted (P < 0.001); air-dried soil was 4.4 and 3.7 (low and high N rate) times higher than soil extracted in the field (P < 0.001). Soil samples stored for 1 month showed no difference in nitrate levels measured in KCl extract between refrigerated or air-dried treatments (P = 0.48). The increase in nitrate with storage contributed an overall increase in the cumulative measured-N detected in the KCl extracts (data not shown).

Fluxes of N as amino acids, ammonium or nitrate measured using microdialysis were not statistically different between the high and low N fertiliser rates (Fig. 2a–c). The storage treatments had differing effects on the flux of each N form. In situ microdialysis showed a higher flux of amino acid-N than all storage treatments, except 1 month refrigerated (Fig. 2a). Amino acid-N flux decreased between disturbed-field and next day stored soil, then increased in the soil samples refrigerated for 1 month.

Fig. 2.  Flux of amino acid-N (a), NH₄–N (b) and NO₃–N (c) (nmol N cm⁻² h⁻¹) in soil from two N fertiliser rates (172 and 309 kg N ha⁻¹) of a cotton field trial, measured using microdialysis either in situ, or with excavated soil samples in the field, the day after sampling, or after 1 month of refrigerated or air-dried storage. All samples were returned to field moisture and bulk density before microdialysis. Values are means ± s.e. (n = 8). Letter codes show significant differences within N forms (amino acids, bold italics; ammonium, lowercase; nitrate, UPPERCASE).

Ammonium-N flux was higher than amino acid-N flux. There was no difference between soil sampled in situ, disturbed-field or stored until the next day, but ammonium-N flux was significantly higher after 1 month of storage both refrigerated and air-dried (P < 0.05; Fig. 2b).

Nitrate-N flux in all disturbed and stored soil samples was at least double in situ flux and increased with time in storage (P < 0.01; Fig. 2c). Soil refrigerated for 1 month had the greatest nitrate-N flux (average 306.6 ± 37.4 nmol-N cm⁻² h⁻¹) and was significantly higher than in situ measurement, and from disturbed-field and air-dried stored soils. Nitrate-N flux was on average 5.6-times higher than in situ microdialysis in soil stored for 1 day (next day), and 8.2-times higher in soil refrigerated for 1 month.

No samples taken from disturbed soil (field, next day, fridge and air-dry), with either microdialysis or KCl extract, had the same relative proportions of amino acid-N, ammonium-N and nitrate-N as measured with in situ microdialysis. The in situ samples had the highest proportion of N as ammonium (51%) and the lowest as nitrate (22%; Fig. 3a). Across microdialysis measurements, the proportion of amino acid-N was immediately lower in the disturbed-field soil (P < 0.01; Fig. 3b), then showed no difference in all other storage treatments (Fig. 3c–e). The proportion of amino acid-N in extracts conducted in the field (Fig. 3b) was similar to in situ microdialysis and declined after 1 month of storage (Fig. 3d, e). The proportion of nitrate-N follows an increasing trend with delay in soil processing. It is the dominant N form measured with KCl extracts and microdialysis from the next day after field sampling (Fig. 3c), with only a slight increase after 1 month (Fig. 3d).

Fig. 3.  Proportion of total measured-N (sum of NO₃–N, NH₄–N and amino acid-N) from each individual N form in soil microdialysis (MD) or 1 M KCl extracts (Extract) samples. Samples were collected from a cotton field trial using microdialysis in situ (a), and both microdialysis and KCl extracts with excavated soil samples in the field (b), the day after sampling (c), or after 1 month of refrigerated (d) or air-dried (e) storage. All samples were returned to field moisture and bulk density before sampling. Values are means ± s.e. (n = 16). Letter codes show significant differences within N forms (amino acids, bold italics; ammonium, lowercase; nitrate, UPPERCASE).

Comparing KCl extracts to microdialysis (Fig. 3b–e) shows a different distribution of N in the field and next day stored soils, and a more similar distribution in the 1 month refrigerated and air-dried soils. Microdialysis generally showed a lower proportion of amino acid-N, but this was only statistically significant in the samples taken in the field and the next day (P < 0.01; Fig. 3b, c). The proportion of ammonium-N was not significantly different between the two sampling techniques in all storage treatments, although it was slightly higher in the microdialysis samples taken from 1 month refrigerated and air-dried soils (Fig. 3d, e). The proportion of nitrate-N was higher in microdialysis samples taken from disturbed-field and the next day treatments (Fig. 3b, c) and was not different between KCl and microdialysis in the 1 month stored soil.

The dynamic nature of N in soils creates a considerable challenge for gaining accurate measurement of N pools, particularly plant-available forms of nitrate, ammonium and DON, such as amino acids and peptides. Correctly determining how much N is in a soil sample and the forms that it exists in is central to outcomes for both agricultural research and efficient fertiliser use in production systems.

The present study used an established cotton N fertiliser rate trial to determine the effect of soil sampling technique and handling processes on outcomes of soil N measurement. Despite the considerable gap between the N rates (172 and 309 kg-N ha⁻¹), there were no significant differences in the concentration or flux of ammonium or nitrate at the time of sampling. This was likely due to high N losses immediately after fertiliser application and furrow-irrigation (Macdonald et al. 2017a, 2017b). Contrary to trends in inorganic N forms, amino acid-N concentrations measured using KCl extracts were significantly higher in the low N rate, while the flux showed no difference between fertiliser rates. This inconsistency between extracts and microdialysis is not unexpected since amino acids may be displaced from exchange sites or released from disrupted microbial cells during extraction (Jones and Willett 2006; Rousk and Jones 2010). The lower amino acid level in the high fertiliser rate could be related to suppression of microbial activity and lower microbial biomass (Zhang et al. 2018). The similarity in inorganic N availability, but disparity in amino acid levels between the low and high N rates highlights the need to better understand soil N dynamics for improved N management, including changes in availability over the season and effects of N fertiliser additions. Ensuring soil N measurement protocols yield data which accurately reflects the in situ soil N pools is pivotal to research in this area.

It is known that soil handling processes affect N pools between sampling and analysis, but little work has evaluated current soil handling practices and how this influences interpretation of N availability. Our findings show an increase in the amount of nitrate-N within 1 day of sampling and a continued increase after 1 month of refrigerated or air-dried storage measured by both extraction and microdialysis (Figs 1c and 2c). Surprisingly, this did not appear to be related to changes in ammonium- or amino acid-N levels, and suggests that ammonification and nitrification were not adequately inhibited by storage at 4°C. This is consistent with the findings of Arnold et al. (2008), and contradicts the notion that chilled transportation and storage is effective for preserving the state of the soil as it was sampled in the field (Rayment and Lyons 2011). Freezing soil samples to preserve inorganic N pools has been tested in previous studies; however, variability in the duration of thawing, temperature of freezing, and between soils has given inconsistent conclusions for the effectiveness of frozen storage (Nelson and Bremner 1972; Crouse et al. 1994; Ma et al. 2005). Further, the effect of freezing on soil organic N pools has received little investigation, but changes in microbial activity, particularly N mineralisation, after frozen storage suggest that organic N levels would likely change during thawing (Stenberg et al. 1998). Performing soil extracts in the field, or as close to sampling as possible, should be adopted to avoid inflated measurements of nitrate-N.

Comparing observations from microdialysis and KCl extracts allows separation of the differences due to disturbance, storage, and the extraction process. More pronounced differences between storage treatments were seen in samples collected via microdialysis, possibly indicating confounding N transformations during extraction (Fig. 2). Further, extract samples may be expected to have a higher proportion of ammonium, since microdialysis does not include the fixed pool (Inselsbacher et al. 2011; Brackin et al. 2015). However, this was not observed, suggesting that ammonium may have been nitrified during extraction or is not effectively exchanged from the clay interlayer in this vertic soil (Fig. 3; Leinweber et al. 2013). The immediate decrease in amino acids and increase in nitrate between in situ and disturbed microdialysis is further evidence that the disturbance caused by sampling accelerates mineralisation and nitrification. Other studies examining N recoveries have demonstrated exacerbated loss of amino acids and concomitant increase in inorganic N due to disturbance caused by sieving, and a rapid decline in added isotope labelled amino acids within 1 h of extracting (Rousk and Jones 2010; Inselsbacher 2014).

Here, we show that KCl extracts and microdialysis have complementary characteristics for measuring soil N pools. While KCl extracts are a highly disruptive, but industry standard technique, microdialysis provides a direct and minimally invasive measure of native N pools. We show that microdialysis has a greater sensitivity for determining changes in amino acid levels, such as during soil storage, than traditional extracts. The necessity for such sensitivity is evident in Leitner et al. (2017), in which microdialysis sampling detected temporal changes in in situ N flux while extracts failed to show any significant differences between sampling times over a 20 h period. Overall, prompt processing of soil for N measurement appears more influential than the sampling method used for preserving N pools (Fig. 3).

Techniques for measuring amino acids in soils have received considerable research attention, but it is still uncertain which is most representative of plant-available N. Microdialysis is considered a more appropriate technique for measuring soil amino acids because absolute concentrations in the soil solution are low, but the pool turnover is high. Its strengths include being minimally invasive and capable of high spatial and temporal resolution, and measuring a flux rather than pool size (Warren 2014, 2018; Buckley et al. 2020). In contrast, extracts using a salt solution disintegrates the soil structure and collects amino acids that were protected in soil aggregates, bound to exchange sites, or contained within microbial cells (Jones and Willett 2006; Rousk and Jones 2010). While this does not exactly reflect the quantity of organic N immediately available for plant uptake, it provides broader measurement of potentially available N. In situ microdialysis often shows a much higher proportion of amino acid-N when compared to soil extracts (Inselsbacher and Näsholm 2012a; Brackin et al. 2015; Buckley et al. 2017). Our results provide an interesting caveat: extracts performed immediately after soil sampling in the field had a higher proportion of amino acid-N than in situ microdialysis. This is likely due to displacement from exchange sites or disruption of microsites with increased microbial activity in extraction, and the inherent bias of microdialysis toward nitrate due to faster diffusion (Inselsbacher et al. 2011; Hill et al. 2019). Contrarily, microdialysis of a disturbed soil sample in the field gave lower relative proportion of amino acid-N and higher nitrate-N; this pattern persisted to the day after field sampling. An alternative explanation to the typically higher amino acids seen in microdialysis samples could be that delays before soil extracts increase soil nitrate levels, overshadowing the amount of amino acid-N. Considering the shortfalls of either extracts or microdialysis discussed, in situ soil solution collection such as with small tension lysimeters could provide an intermediate approach. Use of lysimeters is less destructive than extracts, capable of in situ deployment and measures the N pools immediately available to the roots. Previous studies tend to show relatively more nitrate and ammonium than amino acids in soil solution samples collected via lysimeter, suggesting it may not reflect flux of N compounds well (Andersson and Berggren 2005; Jämtgård et al. 2010; Inselsbacher et al. 2011). Comparing microdialysis to extracts highlights the need to consider the combined effects of soil handling, storage conditions and measurement techniques on perceived importance of soil N pools, and to ensure that the sampling technique used matches the goals of N pool interpretation.

In situ microdialysis has value for directly measuring a root’s perspective of N in soil, so may be a better indicator of plant-available N (Inselsbacher and Näsholm 2012a; Brackin et al. 2017; Randewig et al. 2019). In the present study, we have considered in situ microdialysis as the ‘gold standard’ of available soil N sampling techniques because we were specifically interested in how other methods deviated. We could conclude that most N available to the crop was as ammonium (Fig. 3a). However, microdialysis has considerable uncertainties for making generalised statements about the status of soil N. Some of these have already been alluded to, including the small volume of soil sampled and collection of only free solutes. It remains unclear how well microdialysis can be scaled to larger areas of soil, or the impact of factors such as soil moisture, mass flow of soil water, soil texture or temperature on recovery of N molecules (Buckley et al. 2020).

These findings illustrate that common steps in soil analysis procedures, principally the delay between soil sampling and processing for analysis, can cause an overstated representation of nitrate in soil. This has likely contributed to the historically nitrate-centric plant nutrition models and prevailing dominance of nitrate measurements in agronomic fertiliser recommendations (Robinson et al. 2011; Rochester and Bange 2016; Schwenke et al. 2019). For research, this raises a need to improve soil sampling practices to better characterise soil N, particularly if concerned with field-relevant measurements. In-field extracts with refrigeration (or freezing) of extract samples are evidently the best practice, but time constraints and additional equipment make this difficult. Taking intact cores may reduce disruption, but prolonged storage is still not advisable (Arnold et al. 2008). Further research should evaluate if the effects of storage observed here are universal across soils of contrasting characteristics, including land use, texture, pH and carbon content. At a minimum, the duration of soil storage should be reported, and interpretations of N data weighted appropriately.

The over-representation of nitrate in soil samples does not necessarily undermine its historic use as a soil test in agronomy. High nitrate levels in stored samples are indicative of the mineralisation and nitrification rates. This is consistent with the observed correlation between pre-planting nitrate and crop N uptake, and the large supply of N from mineralisation (Rochester and Bange 2016; Macdonald et al. 2017a; Brackin et al. 2019). Even though a soil test may not reflect in-field nitrate levels, it may still indicate overall N availability. This suggests that formalised assessment of mineralisable N, as has been expanding in other cropping industries such as sugarcane, may have greater value for predicting fertiliser requirements (Allen et al. 2019; Brackin et al. 2019). As with research applications, details of soil sample handling and storage, such as time and temperature, should be considered when interpreting commercial soil test results.

The work reported here did not capture the various other forms dissolved N exist as, nor the myriad of N transformation process in soils (Kuypers et al. 2018). It is difficult to confidently describe changes to N conversion processes from looking at the size of pools and fluxes of amino acids, ammonium and nitrate. For instance, the lower nitrate flux measured by microdialysis in the air-dried samples (but not KCl extracts) is possibly caused by changes in N transformations on rewetting, such as denitrification or rapid immobilisation of nitrate (Fig. 2c; Leitner et al. 2017). Given the low temperature used for air-drying, N transformation are expected to have continued during drying, also limiting the assessment of this method. Further investigation of changes to N transformation process using ¹⁵N pool dilution techniques (Barraclough and Puri 1995) would determine if nitrate accumulates at a normal rate, or if it is enhanced during soil sampling and storage.

In this study, we have used both conventional KCl soil extracts and an in situ microdialysis technique to examine the influence of soil disturbance during sampling, soil sample storage conditions and storage duration on the measurement of plant-available N forms. In-field extracts showed a distribution of nitrate, ammonium and amino acids most similar to the in situ N flux measured with microdialysis. Delaying processing of soil samples, even until the following day, tended to disproportionately increase measured nitrate, biasing the perceived importance of this pool of plant-available N. Storage of soil samples before analysis should be avoided or minimised where possible, instead favouring in-field extracts immediately after soil collection. In situ sampling techniques such as microdialysis or soil solution sampling can provide a valuable tool for minimally disruptive sampling with samples bearing direct relevance to plant-available N. However, applying these techniques outside of research in commercial soil testing requires evaluation for how to translate results to field-scale fertiliser applications. Conventional soil sampling and extracts are currently the best option for measuring available N in agronomic settings, However, there needs to be details of the soil handling recorded to aid interpretation of measured N levels, particularly high levels of nitrate.

Supplementary material is available online.

Data and code available on request.

MF is an Associate Editor for Soil Research. He had no editorial involvement with this paper, which was handled independently by other members of the Editorial Board. The authors declare no other conflicts of interest.

This work was funded by the Australian Cotton Research and Development Corporation (CRDC) (Grant number CSP1904 awarded to BM and MF at CSIRO, and RB and NR from UQ). TB’s PhD is supported by an Australian Government RTP scholarship and a CSIRO Postgraduate Scholarship.