A decision support tool for the selection of ¹⁵ N analysis methods of ammonium and nitrate

Abstract

The stable nitrogen isotope (¹⁵ N) analysis of ammonium (NH₄⁺) and nitrate (NO₃^–) is widely used in ecological research, providing insights into N cycling and its underlying regulating mechanisms in both aquatic and terrestrial ecosystems. To date, a large number of methods have been developed for the preparation and measurement of ¹⁵ N abundance of NH₄⁺ and NO₃^– in liquid environmental samples at either natural abundance or enriched levels. However, these methods are all subject to certain specific advantages and limitations, and ecologists might be looking for an efficient way to select the most suitable methods in face of shifting sampling and analytical conditions. Based on our extensive review of these ¹⁵ N analysis methods we developed a decision support tool (DST) to provide quick and proper guidance for environmental researchers in finding the optimal method for preparing their liquid samples for ¹⁵ N analysis in NH₄⁺ or NO₃^–. The DST is a decision tree based on several key criteria that users need to take into account when choosing the preferred sample preparation method for their samples. The criteria concern: the sample matrix, the ¹⁵ N abundance and the concentration of the target N species, the contamination by other N-containing chemicals, the isotopic fractionation, the availability of equipment, concerns about toxicity of reagents, and the preparation time. This work links field-scale experiments and laboratory ¹⁵ N analysis. Potential applications of our decision trees include ¹⁵ N studies ranging from natural abundance to tracer level in a wide range of terrestrial, freshwater and marine ecosystems.

Introduction

The increasing world-wide concern about the impact of high N inputs in ecosystems and their contribution to climate change has highlighted the need for a deeper understanding of N cycling and its underlying regulating mechanisms in ecosystems. Ammonium (NH₄⁺) and nitrate (NO₃^–) are the two dominant forms of bioavailable N, driving productivity and affecting biodiversity in both terrestrial and aquatic ecosystems (Canfield et al. 2010; Zhu et al. 2015). In addition, they contribute to global N contamination as important forms of anthropogenic N input through fertilizer and atmospheric deposition (Galloway 2005; Xue et al. 2009; Mayer et al. 2013). Due to the crucial roles of NH₄⁺ and NO₃^– in N transformations, the two N species have been widely investigated using stable N isotope techniques to better quantify N dynamics.

Nitrogen is composed of two stable isotopes, the abundant ¹⁴ N and the rare ¹⁵ N. The stable N isotopic composition in a N-compound is usually expressed as either an “absolute value (x¹⁵N)” or a “relative value (δ¹⁵N)” in agricultural and biological research (Fry 2006; Coplen 2011; Mayer et al. 2013; Chalk et al. 2015). The former is preferred in applications where ¹⁵ N is used as a tracer (Hauck and Bremner 1976; Smith and Mullins 2000; Fillery and Recous 2001; Scrimgeour and Robinson 2003). The latter denotes a difference measurement made relative to a standard reference material (usually atmospheric N₂) during the actual analysis, and is more appropriate for samples close to natural abundance (Mariotti 1983; Shearer and Kohl 1993; Peoples et al. 2001; Stewart 2001; Scrimgeour and Robinson 2003; Fry 2006; Michener and Lajtha 2008; Coplen 2011; Kendall and McDonnell 2012; Mayer et al. 2013; Chalk et al. 2015). Measuring ¹⁵ N abundance of important N compounds like NH₄⁺ and NO₃^– can provide valuable information on their sources, flow-paths, transformations and fates, and the N cycle of the ecosystems (Hauck 1982; Shearer and Kohl 1993; Kendall and McDonnell 2012; Ito et al. 2013). The ¹⁵ N analysis of NH₄⁺ and NO₃^– is commonly involved in ¹⁵ N-related ecological investigations such as ¹⁵ N pool dilution (Davidson et al. 1991; Wessel and Tietema 1992; Laine et al. 2018), ¹⁵ N tracing (Mulholland et al. 2000, 2008; Rütting et al. 2011; Björsne et al. 2014) and natural abundance studies (Gathumbi et al. 2002; Houlton et al. 2007; Denk et al. 2017). The ¹⁵ N pool dilution and tracing assays are both based on ¹⁵ N labelling of pools of N species, providing information about gross rates of main N transformation processes, and N sources, influxes and sinks in catchments and food webs (Hauck 1982; Rütting and Müller 2007; Braun et al. 2018). The natural ¹⁵ N abundance approach, on the other hand, makes use of the isotopic discrimination that occurs when nutrients move through pathways, and provides an important tool for identifying and quantifying N sources and fluxes in soil and vegetation on a larger spatial and temporal scale (Owens 1988; Kendall 1998; Robinson 2001; Pörtl et al. 2007; Kendall and McDonnell 2012; Ito et al. 2013; Granger and Wankel 2016; Nikolenko et al. 2018).

In field experiments where the cycling and flows of N are studied using ¹⁵ N techniques, the ¹⁵ N analysis of NH₄⁺ or NO₃^– is commonly conducted in an environmental solution, such as soil extracts, freshwater or seawater samples. To acquire reliable isotopic data it is important to prepare and measure the samples in an appropriate way. A wide range of sample preparation and measurement methods have been developed, and they commonly involve converting NH₄⁺ or NO₃^– to an isotopic representative gas (e.g., N₂ or N₂O) and then quantifying ¹⁵ N abundance of the analyte. The gas conversion is necessary because most ¹⁵ N ratio measurements use either a mass spectrometer, an emission spectrometer, or a laser spectrometer, which accept only simple gas analytes (Bremner 1965a, 1996; Hauck 1982; Preston and Owens 1983; Mulvaney 1993; Brand 1996; Brenna et al. 1997; Mayer et al. 2013). Alternatively, a few sample preparation methods without gas conversion have also been reported. For example, the ¹⁵ N ratio of NH₄⁺ in aquatic samples can be analyzed directly using HPLC (Gardner et al. 1991, 1995, 1997).

With so many ¹⁵ N analysis techniques available, it can be tedious and time-consuming for researchers to select a method in agreement with their analytical needs among a wide range of published methodologies. During past decades there have been great and valuable reviews published regarding the classic sample preparation and ¹⁵ N measurement approaches (Bremner 1965a, 1996; Fiedler and Proksch 1975; Hauck and Bremner 1976; Bremner and Hauck 1982; Hauck 1982; Owens 1988; Mulvaney 1991, 1993, 1996, 2008; Knowles and Blackburn 1993; Shearer and Kohl 1993; Sparks et al. 1996; Kendall 1998; Smith and Mullins 2000; Scrimgeour and Robinson 2003; Chang et al. 2004; Groot 2004). The well-reviewed sample preparation methods such as diffusion, bacterial denitrifier and azide techniques have been widely adopted for ¹⁵ N analysis in NH₄⁺ and NO₃^– in ecological solutions (Christensen and Tiedje 1988; Bremner 1996; Mulvaney 1996; Mulvaney et al. 1997; Sigman et al. 1997, 2001; Holmes et al. 1998; Scrimgeour and Robinson 2003; Chang et al. 2004; McIlvin and Altabet 2005; Xue et al. 2009; Mayer et al. 2013). However, they either require laborious incubation and maintenance of special bacteria cultures, or use highly toxic chemicals. In recent years, a number of novel methods and modifications to traditional methods have been reported (Houlton et al. 2007; Zhang et al. 2007, 2015; Isobe et al. 2009; Xing and Liu 2011; David Felix et al. 2013; Liu et al. 2014; Kiba et al. 2016; Tu et al. 2016; Jin et al. 2020; Gebus-Czupyt et al. 2020; Kawashima et al. 2021; Hilkert et al. 2021), leading to improvements in concentration ranges and analytical precision while reducing time, cost and environmental risk. While they are less well-known and documented, such advances have the potential to promote the more widespread practical application of isotope techniques in the management of nutrient issues in the future.

To assist method users select the appropriate sample preparation approach for the ¹⁵ N analysis of NH₄⁺ and NO₃⁻ in their environmental samples, we developed a “Decision Support Tool (DST)” based on a review and comparison of the well-known “classic” and novel methods in terms of several key criteria that need to be considered when selecting methods for ¹⁵ N analysis in liquid samples. These criteria concern: the composition of the sample matrix, the ¹⁵ N abundance and concentration of the target N, the contamination by other N-containing chemicals, the isotopic fractionation, the availability of equipment, concerns about toxicity of reagents, and the preparation time. We organize the sections below in the following order: an overview of the methods, the development of the DST, and finally a discussion of two application cases and of the benefits and drawbacks of the present DST.

Overview of sample analysis techniques

In this section, we give a brief overview of methods reported in the literature to analyze ¹⁵ N in dissolved NH₄⁺ and NO₃^–, arranged according to their final analytes and instruments. The most popular off-line gas methods are presented first, followed by on-line gas methods, and finally non-gas approaches. A more detailed description of the overall procedures, sample requirements, advantages and limitations of the methods can be found in Tables 1 and 2, and in the supplementary information.

Table 1 Methods reported for ¹⁵ N analysis of NH₄⁺ in liquid samples in case studies

Table 2 Methods reported for ¹⁵ N analysis of NO₃^– in liquid samples in case studies

Methods producing N₂ as the analyte (N₂ methods)

In the gas methods, the NH₄⁺ or NO₃^– is converted into analytical gases such as N₂, N₂O, NO or volatile derivatives. Most gas methods involve isolating the NH₄⁺ or NO₃^– from the liquid sample, converting it into N₂, and determining the ¹⁵ N abundance of the produced N₂ by an isotope ratio mass spectrometer (IRMS) (Hauck and Bremner 1976; Hauck 1982; Mulvaney 1993, 2008; Shearer and Kohl 1993; Bremner 1996; Smith and Mullins 2000; Groot 2004; Michener and Lajtha 2008; Cook et al. 2017). Previously, the conversion into N₂ and subsequent purification were achieved using either hypobromite oxidation (i.e., the Rittenberg technique) or quartz-tube Dumas combustion, and was determined via a dual-inlet isotope ratio mass spectrometer (DI-IRMS) or an emission spectrometer (Fiedler and Proksch 1975; McInteer et al. 1981; Bremner and Hauck 1982; Fisher and Morrissey 1985; Mulvaney et al. 1990; Kendall and Grim 1998; Mulvaney and Liu 1991; Liu and Mulvaney 1992a; Mulvaney 1993; Shearer and Kohl 1993; Bremner 1996). Révész et al (1997) also reported an off-line combustion technique using catalyzed graphite to convert potassium nitrate (KNO₃) into CO₂, K₂CO₃ and N₂ for simultaneous N and O isotope analysis. The Rittenberg oxidation was later automated by coupling a Rittenberg apparatus to the IRMS (Griffiths et al. 1981; McInteer et al. 1981, 1984; Mulvaney et al. 1990; Mulvaney 1991; Mulvaney and Liu 1991; Liu and Mulvaney 1992b), but this is no longer commonly used. The workhorse in most modern isotope ratio laboratories is the combination of an elemental analyzer (EA) with a continuous-flow IRMS (EA-CF-IRMS, EA-IRMS), in which isolated N-bearing salts from various sample preparation procedures are sealed into tin capsules and loaded into an autosampler, followed by online combustion in the EA and subsequent sweeping of the produced N₂ into the IRMS with an He carrier gas (Preston and Owens 1983; Barrie and Workman 1984; Marshall and Whiteway 1985; Owens 1988; Barrie et al. 1989; Owens and Rees 1989; Egsgaard et al. 1989; Craswell and Eskew 1991; Mulvaney 1993; Barrie and Prosser 1996; Brand 1996; Bremner 1996; Chang et al. 2004; Horita and Kendall 2004).

Prior to N₂ conversion and isotope measurements, the NH₄⁺ or NO₃^– in samples can be concentrated and isolated from solutions by a number of sample preparation approaches. The target N species can be distilled as ammonium salts (Bremner and Edwards 1965; Bremner and Keeney 1965; Cline and Kaplan 1975; Hauck and Bremner 1976; Hauck 1982; Bremner 1996; Mulvaney 1996; Feast and Dennis 1996; Mulvaney and Khan 1999), diffused onto acidified filters (Conway 1947; Brooks et al. 1989; Sørensen and Jensen 1991; Liu and Mulvaney 1992b; Saghir et al. 1993b; Lory and Russelle 1994; Herman et al. 1995; Mulvaney 1996; Sparks et al. 1996; Stark and Hart 1996; Sigman et al. 1997; Khan et al. 1998; Holmes et al. 1998; Sebilo et al. 2004; Cao et al. 2018) or into acid solutions (Saghir et al. 1993a, b; Khan et al. 1997; Mulvaney et al. 1997; Mulvaney and Khan 1999), precipitated as salts (Fisher and Morrissey 1985; Huber et al. 2011, 2012; Stock et al. 2019), concentrated into zeolites (Velinsky et al. 1989; Burke et al. 1990; Böhlke et al. 2006; Frey et al. 2014) or ion-exchange resins (Chang et al. 1999; Silva et al. 2000; Lehmann et al. 2001; Xing and Liu 2011; Li et al. 2015), or derivatized as an azo dye (Preston et al. 1996, 1998; Clark et al. 2006, 2007; Ward 2011). The isolated N-containing materials can then be dried and loaded into an EA-IRMS, or oxidized/combusted off-line to produce N₂ (Fiedler and Proksch 1975; Hauck 1982; Velinsky et al. 1989; Kendall and Grim 1998; Knowles and Blackburn 1993; Mulvaney 1993, 1996). In the derivatization method, the isolated azo dye can be further converted into a volatile derivative, which can then be analyzed using a GC–MS (or GC-QMS, gas chromatography–quadrupole mass spectrometry) (Preston et al. 1996, 1998; Clark et al. 2006, 2007). Apart from these isolation methods, it is also possible to directly convert the target N species (e.g., NH₄⁺) in water samples into N₂ gas and analyze the ¹⁵ N abundance directly using a membrane-inlet mass spectrometer (MIMS) (Yin et al. 2014).

If the initial NH₄⁺ or NO₃^– concentration is too low for these sample preparation methods, samples may be first concentrated by evaporation, freeze-drying, ion-exchange, or by diffusing or distilling multiple sample aliquots (Fiedler and Proksch 1975; Hauck and Bremner 1976; Hauck 1982; Owens 1988; Mulvaney 1993, 1996; Bremner 1996; Smith and Mullins 2000; Scrimgeour and Robinson 2003; Chang et al. 2004; Horita and Kendall 2004; Michener and Lajtha 2008). Alternatively, samples can be spiked with standards with known N masses and ¹⁵ N abundances (Hauck 1982; Glibert and Capone 1993; Mulvaney 1993; Højberg et al. 1994; Stephan and Kavanagh 2009; Griesheim and Mulvaney 2019).

Methods producing N₂O as the analyte (N₂O methods)

While classic sample preparation methods rely on the conversion of NH₄⁺ or NO₃^– to N₂ for ¹⁵ N analysis, contemporary methods primarily favor off-line conversion of NH₄⁺ or NO₃^– into N₂O and subsequent IRMS/laser analysis. In earlier studies the N₂O is always reduced to N₂ by Dumas combustion for ¹⁵ N analysis (Bremner 1965a; Fiedler and Proksch 1975; Bremner and Hauck 1982; Kendall and Grim 1998; Mulvaney 1993; Preston 1993; Shearer and Kohl 1993; Chang et al. 2004). However, recent developments in instrumentation and analytical techniques have allowed to enter N₂O directly into the IRMS or the laser spectrometer for ¹⁵ N analysis, making it possible to obtain both N and O isotope ratios simultaneously (Dore et al. 1998; Sigman et al. 2001; Toyoda and Yoshida 2004; Groot 2004; McIlvin and Altabet 2005; Kendall et al. 2007; Michener and Lajtha 2008; Altabet et al. 2019). The N₂O derived from NO₃^– or NH₄⁺ can be extracted and purified off-line (Sigman et al. 2001; Soto et al. 2015; Wassenaar et al. 2018), or be extracted on-line using a purge and trap system (e.g., Finnigan GasBench II system) coupled to a continuous flow-IRMS with gas chromatograph interfaces (PT-CF-IRMS, GC-IRMS, or PT-IRMS). Because N₂O can be easily trapped and concentrated cryogenically into a small volume of helium carrier gas without significant contamination from background N₂O, a large fraction of the sample can be entered into the IRMS as opposed to being lost to waste at the open split, yielding a high sensitivity (Stevens et al. 1993; Stevens and Laughlin 1994; Dore et al. 1998; Sigman et al. 2001; Weigand et al. 2016). GC–MS was also reported for quantifying ¹⁵ N abundance in N₂O, but it is limited to ¹⁵ N-enriched samples due to low instrument precision (Russow and Förstel 1993; Isobe et al. 2009).

For atmospherically-derived NO₃^– (e.g. in rainwater), using N₂O as the analyte for IRMS analysis may result in a slight overestimation of δ¹⁵N values by 1 − 2‰ due to mass-independent ¹⁷O fraction (Knowles and Blackburn 1993; Shearer and Kohl 1993; Sigman et al. 2001; Michalski et al. 2002; Ohte et al. 2013). In the mass spectrometer, measurements of m/z 45 and m/z 44 allow for δ¹⁵N calculation after correcting for the contribution of ¹⁴N¹⁴N¹⁷O. For most environmental samples, this correction can be performed routinely by most IRMS data software, which calculates ¹⁷O contribution from ¹⁸O abundance by assuming a mass-dependent relationship (Böhlke et al. 2003; McIlvin and Altabet 2005). However, atmospheric NO₃^– is enriched in ¹⁷O due to large mass-independent ¹⁷O fractionation from ozone formation, resulting in an underestimation of the m/z 45 contribution of ¹⁷O (Michalski et al. 2002, 2003; Böhlke et al. 2003; Soto et al. 2015). This error can be minimized by using a denitrifying strain that primarily produce oxygen in N₂O from H₂O (Hastings et al. 2003; Coplen et al. 2004). Alternatively, the N₂O can be analyzed by coupling PT-IRMS to a thermal decomposition system (Brand 1995; Kaiser et al. 2007; Komatsu et al. 2008; Smirnoff et al. 2012; Hattori et al. 2016) or directly by laser spectroscopy (Soto et al. 2015; Wassenaar et al. 2018; Altabet et al. 2019; Harris et al. 2020) to avoid mass-overlap correction and help discern atmospheric N sources (See 3.1.6 Availability of equipment).

Both microbial and chemical methods have been reported for converting dissolved NO₃^– to N₂O gas for ¹⁵ N analysis. The “denitrifier method” is a well-established technique, in which NO₃^– is reduced to NO₂^– and finally to N₂O by denitrifying bacteria (Christensen and Tiedje 1988; Sigman et al. 2001; Casciotti et al. 2002; Hastings et al. 2003; Coplen et al. 2004; Mørkved et al. 2007; Michener and Lajtha 2008; Isobe et al. 2009; McIlvin and Casciotti 2011; Weigand et al. 2016). A number of chemical reducing methods are also gaining ground, using cadmium (Cd) or vanadium chloride (VCl₃) to reduce NO₃^– to NO₂^– and using azide or hydroxylamine (NH₂OH) as catalyst for further reduction of NO₂^– to N₂O (McIlvin and Altabet 2005; Lachouani et al. 2010; Ohte et al. 2013; Tu et al. 2016; Wassenaar et al. 2018; Zhao et al. 2019; Jin et al. 2020). Recently, Altabet et al. (2019) presented an one-step chemical method in which NO₃^– is reduced to N₂O using Ti(III) chloride. In all these microbial and chemical methods, the NO₂^– present in the sample is reduced alongside the NO₃^–, and the measured ¹⁵ N abundance of the N₂O is a composition of both the NO₂^– and the NO₃^–. When preexisting NO₂^– concentration is high and only the ¹⁵ N abundance of NO₃^– is required, NO₂^– can be removed using sulfamic acid (Norman and Stucki 1981; Ward et al. 1984; Mulvaney 1996; Granger et al. 2006).

Methods for generating N₂O gas from dissolved NH₄⁺ usually involve the conversion of NH₄⁺ to NO₃^– (or NO₂^–) and the reduction of NO₃^– (or NO₂^–) to N₂O, with or without pre-isolation of NH₄⁺ from solutions. The isolation of NH₄⁺ by diffusion (Holmes et al. 1998; Koba et al. 2010; Lachouani et al. 2010; Zhang et al. 2015) or ion-exchange resins (Kawashima et al. 2021) removes pre-existing NO₃^–, NO₂^– and other interfering chemicals in the matrix. The conversion of NH₄⁺ to NO₃^– or NO₂^– can be achieved by persulfate oxidation (PO) (Houlton et al. 2007; Isobe et al. 2009; Lachouani et al. 2010) or hypobromite oxidation (BO) (Stevens et al. 1993; Stevens and Laughlin 1994; Zhang et al. 2007, 2015; David Felix et al. 2013; Liu et al. 2014), respectively. The final conversion of NO₃^– (or NO₂^–) to N₂O can be achieved using bacterial (Houlton et al. 2007; Isobe et al. 2009; Kawashima et al. 2021), azide (Lachouani et al. 2010), or NH₂OH reduction (Zhang et al. 2018), similar to the ¹⁵ N analysis methods for NO₃^–.

On-line gas methods

In all of the foregoing preparation methods, the isolation and/or conversion of the target NH₄⁺ or NO₃^– is partially performed off-line. Instead, the SPIN (sample preparation unit for inorganic nitrogen) method allows for complete on-line sample preparation and ¹⁵ N measurement in ¹⁵ N-enriched samples. In this approach, NO₃^– is reduced to NO or N₂O, while NH₄⁺ is oxidized to N₂. All the reactions are automated and regulated in the SPIN reaction vessel. The analytical gases are subsequently transferred to a quadrupole mass spectrometer (QMS) with continuous-flow (CF-QMS) or membrane-inlet (MIMS) mode for simultaneous determination of ¹⁵ N abundance and N concentration (Stange et al. 2007; Eschenbach et al. 2017, 2018).

Non-gas methods

For ¹⁵ N-enriched samples, the ¹⁵ N abundance and the concentration of NH₄⁺ or NO₃^–can also be analyzed simultaneously in filtered solutions as dissolved NH₄⁺ using the ammonium retention time shift-high performance liquid chromatography (AIRTS-HPLC) (Gardner et al. 1991, 1995; Lu et al. 2019). This is based on the difference in retention time between ¹⁵NH₄⁺ and ¹⁴NH₄⁺ as they pass through HPLC columns. The NH₄⁺ in filtered water samples can be analyzed directly by HPLC, while NO₃^– is first reduced to NH₄⁺ by zinc reduction under acidic conditions. To calculate the accurate ¹⁵ N of NO₃^–, the ¹⁵ N abundance and concentration of initial NH₄⁺ must be pre-measured.

Development of the decision support tool (DST)

To assist ecological researchers in finding a suitable method for their specific research needs we developed a Decision Support Tool (Figs. 1 and 2), including all currently available methods, as summarized in the overview in the previous section. The users can arrive at the best solution by systematically selecting preferences based on key criteria. The aim was to provide a streamlined, visualized, and action-oriented decision-supporting tool for preparing liquid environmental samples for ¹⁵ N analysis of NH₄⁺ and NO₃^–, which is easier to use for a researcher than a review that merely sums up all the details of the various methods.

Fig. 1

Decision support tool for ¹⁵ N analysis of NH₄⁺ in liquid samples. The figure is created with Microsoft® PowerPoint® version 2012 and Adobe Illustrator CC 2018 version 22.1

Fig. 2

Decision support tool for ¹⁵ N analysis of NO₃^– in liquid samples. The figure is created by Microsoft® PowerPoint® version 2012 and Adobe Illustrator CC 2018 version 22.1

We organize the DST as well as the descriptions of the methods in the supplementary information by presenting NH₄⁺ and NO₃^– methods separately, although there are unavoidable overlaps since many methods can be used for both species. To optimize the procedure we provide only key information in the DST, but present a detailed comparison of all the methods in Tables 1 and 2, where the characteristics, advantages, and disadvantages can be easily found. Furthermore, a comprehensive and up-to-date description of the published approaches for both NH₄⁺ and NO₃^– preparation is provided as supplement. When navigating by the DST, users can and should always refer to the original literature before proceeding.

Criteria for method selection

Among the numerous ¹⁵ N analysis approaches that have been reported, each method employs a set of specific procedures, materials, and instruments to deliver optimal results under certain conditions. In this section, we discuss the criteria that should be generally considered when determining the appropriate method for ¹⁵ N analysis of NH₄⁺ and NO₃^– in liquid samples. Note that these criteria are interrelated and should be considered together when evaluating methods.

Sample matrix

In ecological research, the liquid samples containing the ¹⁵NH₄⁺ or ¹⁵NO₃^– to be determined can stem from a range of environmental sources (e.g., freshwater, seawater and soil extracts), and, as a consequence, can vary strongly in composition. Salts, dissolved organic matter, pH, hazardous chemicals, and other nitrogen-containing substances can all contribute to isotopic fractionation and contamination, reducing the precision and accuracy of ¹⁵ N measurements.

Sample salinity is one of the most important factors limiting the choice of appropriate methods. Salts (e.g., Mg²⁺, K⁺, SO₄^2–, Cl^–) and dissolved organic carbon (DOC) can act as competitive chemicals in exchange methods, reducing recovery of target N components (Chang et al. 1999; Silva et al. 2000; Böhlke et al. 2006; Li et al. 2015). Salts may also reduce recovery by changing the kinetics of certain reactions. The conversion of NO₃^– into N₂O by Ti(III) reduction, for example, is significantly impeded by SO₄^2– in the solutions (Altabet et al. 2019). Poor recovery causes isotopic fractionation, resulting in erroneous depletion or enrichment of ¹⁵ N in the final product (see 3.1.5 Isotopic fractionation). Because salt removal and calibration by references with an identical matrix are often impractical, such salinity-sensitive methods are most suitable for use of freshwater samples.

Distillation and diffusion are used for a wide range of matrices because they isolate target N species from solutions. They can be used to directly produce N-containing salts for off-line combustion or EA-IRMS analysis (Bremner and Edwards 1965; Keeney and Bremner 1966; Fiedler and Proksch 1975; Hauck 1982; Keeney and Nelson 1983; Brooks et al. 1989; Sørensen and Jensen 1991; Mulvaney 1993, 1996; Bremner 1996; Holmes et al. 1998), or they can be used in front of ion exchange or chemical conversions to pre-isolate N from high salinity solutions (Velinsky et al. 1989; Isobe et al. 2009; Lachouani et al. 2010; Zhang et al. 2015). The two methods, especially distillation, are more appropriate for samples containing low amounts of dissolved organic N (DON), because liberation of NH₄⁺ by DON hydrolysis can cause errors to ¹⁵ N abundance of NH₄⁺ and NO₃^– being measured. More detailed discussion of this issue can be found in Sect. 3.1.4 (Contamination by other N-containing chemicals).

Other matrix characteristics can also limit choices of sample preparation methods. Methods involving derivatization, MIMS and HPLC techniques, are appropriate for both freshwater and marine samples from tracer studies, but whether they are appropriate for soil extracts has not yet been examined. Toxic compounds and soil extractants (e.g., KCl) can suppress bacterial activity in the denitrifier method, resulting in low recovery and fractionation (Sigman et al. 2001; Casciotti et al. 2002). The presence of NO₂^– might cause errors in methods such as derivatization, denitrifier and azide conversions where it occurs as an intermediate product in the N transformations (Sigman et al. 2001; McIlvin and Altabet 2005; Zhang et al. 2007; Lachouani et al. 2010; Ward 2011; David Felix et al. 2013; Altabet et al. 2019), but it can be measured using a colorimetric method and removed using sulfamic acid (Granger et al. 2006; Ward 2011).

¹⁵ N abundances of NH₄ ⁺/NO₃ ^– in samples

Samples for ¹⁵ N analysis can be at natural ¹⁵ N abundance or ¹⁵ N-enriched level (in case of pool dilution and ¹⁵ N-tracing experiments), and researchers must choose a method that is sufficiently precise and accurate to measure the large or small ¹⁵ N variations in their samples.

Natural abundance studies require excellent precision because the δ¹⁵N abundances of most natural ecological samples fall within the narrow range of –20‰ to + 30‰ (x(¹⁵ N) = 0.362 – 0.377%) (Kendall 1998; Robinson 2001; Fry 2006; Mulvaney 2008; Michener and Lajtha 2008). Only IRMS and laser spectroscopy provide this level of precision, and only sample preparation methods employing these two instruments are appropriate for natural δ¹⁵N abundances. Modern N₂O methods, in particular, are gaining popularity in natural abundance studies due to their high sensitivity and capability for dual O/N isotope analysis (Sigman et al. 2001; Casciotti et al. 2002; McIlvin and Altabet 2005; Altabet et al. 2019).

The ¹⁵ N abundance of target N species are typically > 0.5% (x(¹⁵ N)) in tracer experiments. The analysis of such samples also requires precision, but the limiting factors are often the highly variable ¹⁵ N abundances and the low N concentrations in the samples (Knowles and Blackburn 1993; Fillery and Recous 2001). Researchers need to be aware of these factors at initial planning, and calculate the level of ¹⁵ N addition based on the system, the frequency of tracer addition, the extent of the change in ¹⁵ N abundance during incubation, as well as the analytical constraints (Edwards 1978; Fillery and Recous 2001; Lipschultz 2008; Mayer et al. 2013). IRMS is usually used to measure ¹⁵ N abundances from natural abundance to low enrichment (e.g. x(¹⁵ N) < 10%), and measurements beyond this range become less precise and accurate because of nonlinear amplification of ion currents and memory effects (Mulvaney 1993; Werner and Brand 2001). Samples with higher ¹⁵ N enrichments can be spiked with natural abundance materials to reach the ideal working range for IRMS analysis, although this introduces analytical errors, which increase with increased ¹⁵ N enrichment (Mulvaney 1993; Mulholland et al. 2004; Griesheim and Mulvaney 2019). It is also possible to measure ¹⁵ N enriched samples using ES (Fiedler and Proksch 1975; Craswell and Eskew 1991; Hoult and Preston 1992; Preston 1993), GC–MS (Russow and Förstel 1993; Clark et al. 2006, 2007), QMS (Stange et al. 2007), MIMS (Yin et al. 2014; Eschenbach et al. 2017, 2018), or HPLC (Gardner et al. 1991, 1995) with acceptable precision (SD < 0.03%, SD: standard error).

Whether samples are at natural abundance or have high ¹⁵ N enrichments, reference materials that bracket the entire expected ¹⁵ N abundances should always be processed concurrently with the unknown samples to guarantee accurate and reliable results (Klesta et al. 1996; Werner and Brand 2001; Brand and Coplen 2012; Carter and Fry 2013; Brand et al. 2014; Meier-Augenstein and Schimmelmann 2019; Mohn et al. 2022). If possible, the reference materials should be subjected to the same procedures and have the same matrices as those of the samples (i.e. the identical treatment principle). This strategy minimizes systematic errors, such as the aforementioned fractionation and memory effects in the IRMS system.

Concentration of NH₄ ⁺/NO₃ ^– in samples

The concentrations of target N in samples should be within the working range of the employed sample preparation and analysis method in order to achieve optimal precision and accuracy. A method can vary in its appropriateness for large samples volumes or smaller sample N amounts to accommodate samples with low N concentrations that require special preparation or analysis to obtain high quality data (Hauck and Bremner 1976; Owens 1988; Velinsky et al. 1989; Knowles and Blackburn 1993; Mulvaney 1993, 1996; Shearer and Kohl 1993; Bremner 1996; Holmes et al. 1998; Smith and Mullins 2000; Robinson 2001; Scrimgeour and Robinson 2003; Chang et al. 2004; Groot 2004; Cook et al. 2017). The NH₄⁺ and NO₃^– concentrations (or total amounts) as well as sample volumes required in reported methods are shown in Tables 1 and 2.

Methods that involve transformation into N₂ and its ¹⁵ N analysis by IRMS require 2–400 µmol of target N, resulting in an ideal concentration of > 50 µM in a typical volume of ~ 50 ml. Lower NH₄⁺ and NO₃⁻ concentrations (e.g. 5 − 10 µM) can, however, occur in many aquatic environments (Holmes et al. 1998; Teece and Fogel 2004) as well as groundwater-paddy soil agroecosystems (Soldatova et al. 2021). In traditional N₂ methods this is resolved by using large volumes of solutions, and then concentrating the N in some way. This processing increases not only the time and effort for sample preparation, but also errors of isotopic fractionation and contamination by the large amount of reagent N (See 3.1.4 Contamination by other N-containing chemicals and 3.1.5 Isotopic fractionation). As an alternative, it is possible to add a known quantity of a standard of known ¹⁵ N content into samples and calibrate the results by an isotope dilution equation, but such an addition introduces analytical errors, which increase with increased ¹⁵ N enrichment (Hauck 1982; Glibert and Capone 1993; Mulvaney 1993; Stephan and Kavanagh 2009). Consequently, methods developed for small N amounts and volumes are more appropriate alternatives. For example, the N₂O methods and non-gas AIRTS-HPLC methods require nanomole levels of target N and allow for samples containing micromolar amounts of NH₄⁺ or NO₃^– in milliliters of solutions. With advances in sample preparation, analysis and calibration (Toyoda et al. 2017; Wassenaar et al. 2018; Yu et al. 2020; Harris et al. 2020; Mohn et al. 2022), chemical and biological N₂O methods are increasingly used not only for low N samples but also for large N samples (e.g. > 50 µM) following dilution. This delivers more accurate and consistent results for natural environmental samples varying largely in their concentrations, as well as facilitates lab compatibility and accelerates progress in large-scale isotopic studies. When diluting samples, however, one must pay attention to the water chemistry of the dilution solution, and an accurate blank correction (see 3.1.4 Contamination by other N-containing chemicals) is necessary (Lachouani et al. 2010; Altabet et al. 2019).

Contamination by other N-containing chemicals

When using certain methods, the isotopic composition of non-target N compounds in solutions can bias the ¹⁵ N measurement of the target N (Hauck 1982; Jensen 1991; Liu and Mulvaney 1992b; Herman et al. 1995; Mulvaney 1996; Stark and Hart 1996; Sigman et al. 1997; Holmes et al. 1998; Mulvaney and Khan 1999; Robinson 2001; Stephan and Kavanagh 2009). Distillation and diffusion of NO₃^– are particularly prone to such contamination, because reagents such as Devarda’s alloy and soil KCl extractants, as well as residual NH₄⁺ can all result in erroneously dilution or enrichment of ¹⁵ N abundances of the target NO₃^–. This is of course affected by the purity and quantity of the contaminates, as well as the ¹⁵ N differences between the N impurities and the target N. When diffusing 50 µg NO₃^–, Stephan and Kavanagh (2009) found that reagent N depleted by 10‰ relative to the target δ¹⁵N resulted in underestimating the target δ¹⁵N by 0.8 to 1.6‰. Such an error increases with an increasing amount of reagent N, is difficult to quantify, and can cause significant inaccuracies in samples with low NO₃^– contents (Liu and Mulvaney 1992b; Herman et al. 1995; Sigman et al. 1997; Robinson 2001; Stephan and Kavanagh 2009). Residual ¹⁵ N-labeled NH₄⁺ can also lead to substantial overestimation in successive ¹⁵ N analysis of unenriched NO₃^–, even at trace levels. Although an isotope-dilution technique or a cleaning procedure can reduce (but not completely eliminate) such an inaccuracy, it can still be undesirable for samples with low NO₃^– contents (e.g. 50 µg) and/or large volumes (e.g. > 30 ml) (Hauck 1982; Liu and Mulvaney 1992b; Saghir et al. 1993b; Mulvaney et al. 1994; Herman et al. 1995; Mulvaney 2008; Griesheim and Mulvaney 2019).

Non-target N compounds such as DON in samples may be converted to NH₄⁺ due to chemical hydrolysis under prolonged diffusion and distillation, contributing to excess N and dilute ¹⁵ N values in the product (Hauck 1982; Velinsky et al. 1989; Liu and Mulvaney 1992b; Mulvaney 1993; Mulvaney et al. 1994, 1997; Khan et al. 1997; Sigman et al. 1997; Mulvaney and Khan 1999; Chang et al. 2004). This contamination was found to be significant for distillation and diffusion at > 50 ml, and is prompted by a higher temperature, pH and the use of Devarda’s alloy reagent (Khan et al. 1997; Mulvaney et al. 1997; Sigman et al. 1997; Holmes et al. 1998; Mulvaney and Khan 1999). For organic-rich samples with low target N (especially NO₃^–) concentrations, such as extracts from organic horizons of forest soils, it is preferable to use methods in which DON is removed. The acetone method, for example, isolates NO₃^– from organic matter in solutions and has been used for soils with organic amendments (Reichel et al. 2018), although the sample N requirement is large (4–9 µmol) (Huber et al. 2011, 2012). In the case of NH₄⁺, many N₂ and N₂O methods require isolating NH₄⁺ from soil extracts by diffusion and are inherently susceptible to DON contamination (Lachouani et al. 2010; Zhang et al. 2015). This risk can be mitigated by a lower temperature, a shorter diffusion period, no vigorous shaking, and gas-phase trapping (Mulvaney 2008; Cao et al. 2018).

In contrast to traditional N₂ methods, chemical and biological N₂O conversion methods generally report constant and small contamination (only nanomoles of N) for natural abundance samples with low N amount (Sigman et al. 2001; McIlvin and Altabet 2005; Isobe et al. 2009; Tu et al. 2016; Zhao et al. 2019; Jin et al. 2020). In particular, the interferences by DON to target δ¹⁵N is significantly reduced, because NO₃^– (and sometimes also NH₄⁺) in matrix can be converted selectively (Sigman et al. 2001; McIlvin and Altabet 2005; Zhang et al. 2007; Liu et al. 2014). The N contribution of non-target chemicals with the denitrifier and azide methods are reported as 2.5–5 and 0.1–10%, respectively (Sigman et al. 2001; McIlvin and Altabet 2005). Since such “method blanks” are often consistent and reproducible, they can be calibrated by using reference materials in N-free water with a similar matrix as the samples. Both reagents and water used for blank correction and sample dilution should be purified (e.g. by boiling or combusting) and, if necessary, calibrated for their N amount and isotopic composition (Sigman et al. 2001; McIlvin and Altabet 2005; Lachouani et al. 2010; Altabet et al. 2019).

Isotopic fractionation

Isotopic fractionation may arise from any stage of the sample preparation and instrument analysis, lowering the precision and accuracy of ¹⁵ N measurements. It is the result of incomplete transformation or recovery of the target N, as the rate of such a process usually differs between ¹⁴ N and ¹⁵ N. To avoid fractionation, a complete recovery is required, but this can be challenging when processing large volumes of samples by distillation/diffusion (Fiedler and Proksch 1975; Hauck 1982; Velinsky et al. 1989; Liu and Mulvaney 1992b; Mulvaney 1993; Holmes et al. 1998; Chang et al. 2004; Stephan and Kavanagh 2009) and freeze-drying techniques (Hauck 1982; Stock et al. 2019), as well as when there are large sample matrix effects (Chang et al. 1999; Silva et al. 2000; Lehmann et al. 2001; Altabet et al. 2019). For example, Holmes et al. (1998) found that fractionation of NH₄⁺ diffusion increased from 0.2 to 10‰ in 200 ml to 3 l samples. Stock et al. (2019), observed a fractionation of up to 10‰ when freeze-drying samples from 500 ml to ≤ 5 ml to obtain enough N for the use of the TPB precipitation method. If not accounted for, such an error can be problematic for natural abundance studies, as the δ¹⁵N in most ecological samples varies only by < 20‰ (Robinson 2001). Fractionation can also occur when samples are entered into a mass spectrometer (Mulvaney 1993; Dawson and Brooks 2001).

Fractionation during sample preparation and measurement can be largely calibrated by subjecting reference materials and unknowns to the same sample preparation and analysis pathway (Klesta et al. 1996; Holmes et al. 1998; Werner and Brand 2001; Gröning 2004; Stephan and Kavanagh 2009; Brand and Coplen 2012; Carter and Fry 2013; Altabet et al. 2019; Stock et al. 2019). Ideally, one needs to match samples and standards with respect to chemical matrix and sample N concentration, as well as the range of ¹⁵ N abundance, however finding ideal references may be a difficult task. If there are substantial interferences by reagents and matrix, such as in distillation and diffusion of soil NO₃^–, the calibration becomes more tedious and inaccurate (Liu and Mulvaney 1992b; Herman et al. 1995; Mulvaney et al. 1997; Holmes et al. 1998; Khan et al. 2000b; Stephan and Kavanagh 2009). All of these issues compromise the ultimately achievable accuracy. Whether this error is acceptable is determined by the accuracy necessary for observable changes among samples, and there may be a trade-off between analytical accuracy and the time and effort required for complete recovery and calibration (Mulvaney 1993; Holmes et al. 1998; Silva et al. 2000; Dawson and Brooks 2001; Benson et al. 2006; Stock et al. 2019).

Given the difficulty of complete recovery and accurate calibration when isolating and freeze-drying low N samples, methods adapted for natural abundance, and small N amounts and volumes, are preferred in such instances. For example, the denitrifier (Christensen and Tiedje 1988; David Felix et al. 2013), azide (McIlvin and Altabet 2005; Zhang et al. 2007; Ryabenko et al. 2009; Tu et al. 2016; Zhao et al. 2019), and NH₂OH procedures (Liu et al. 2014; Zhang et al. 2015; Jin et al. 2020) all reported satisfying recovery and negligible fractionation when producing sub-micromoles of N₂O from milliliters of solutions.

Availability of equipment

The availability of equipment, such as special materials and apparatus for sample preparation, as well as analytical instruments, limits the variety of methods that can be used. For example, the distillation method requires a, commercially available, distillation apparatus; the ion-exchange method entails resins and reagents that are expensive for large batches; the denitrifier method relies on anaerobic denitrifying bacteria strains that are difficult to obtain and maintain in non-microbiology laboratories; the SPIN method uses a non-commercially available reaction unit that has been custom built in only a few laboratories worldwide (Silva et al. 2000; Sigman et al. 2001; Chang et al. 2004; Stange et al. 2007). When such materials and equipment are not readily available, preparation methods using off-the-shelf chemicals are the only alternatives, such as the diffusion, acetone extraction and chemical-based N₂O methods (Brooks et al. 1989; Holmes et al. 1998; McIlvin and Altabet 2005; Zhang et al. 2007; Lachouani et al. 2010; Huber et al. 2011, 2012; Liu et al. 2014).

The most popular instrument for ¹⁵ N analysis is IRMS (Brenna et al. 1997; Scrimgeour and Robinson 2003; Meier-Augenstein 2004), but there are also alternatives such as LS (laser spectroscopy) (Mohn et al. 2014; Soto et al. 2015; Wassenaar et al. 2018; Ji and Grundle 2019), QMS (quadrupole mass spectrometry) (Eschenbach et al., 2017, 2018; Stange et al., 2007), ES (emission spectrometry) (Fiedler and Proksch 1975; Knowles and Blackburn 1993; Preston 1993; Heiling et al. 2006), and HPLC (Gardner et al. 1991, 1995; Lu et al. 2019).

Due to their high precision (a RSD of  < ± 0.2 ‰ and < ± 1‰, respectively, RSD: relative standard deviation), the IRMS and laser analysis following N₂/N₂O-based sample preparation methods are currently the only candidates for quantifying ¹⁵ N abundance in natural abundance studies (Mulvaney 1993; Robinson 2001; Scrimgeour and Robinson 2003; Soto et al. 2015; Wassenaar et al. 2018). Off-line Rittenberg oxidation or Dumas combustion and ¹⁵ N analysis by DI-IRMS was laborious and prone to atmospheric contamination, and have been widely replaced by the commercially available continuous flow-IRMS in routine ecological ¹⁵ N research (Bremner 1965a; Hauck and Bremner 1976; Hauck 1982; Kendall and Grim 1990; Knowles and Blackburn 1993; Mulvaney 1993, 1996; Preston and Slater 1994; Feast and Dennis 1996; Chang et al. 2004). Such development reduces systematic errors associated with off-line N₂ production steps and significantly improves precision, sensitivity and productivity (Preston and Owens 1983; Barrie and Workman 1984; Marshall and Whiteway 1985; Barrie et al. 1989; Harris and Paul 1989; Egsgaard et al. 1989; Craswell and Eskew 1991; Jensen 1991; Smith and Mullins 2000; Mulvaney 2008). The EA-IRMS allows for online combustion of samples containing micromoles N (Knowles and Blackburn 1993; Mulvaney 1993; Barrie and Prosser 1996; Brand 1996; Boutton 1996; Horita and Kendall 2004), while the PT-IRMS is capable of ¹⁵ N analysis in nanomole N₂O gas (Casciotti et al. 2002; Chang et al. 2004; McIlvin and Altabet 2005; Lachouani et al. 2010; Toyoda et al. 2017). In addition, by coupling an on-line decomposition system such as a gold-furnace (Brand 1995; Kaiser et al. 2007; Komatsu et al. 2008; Smirnoff et al. 2012) or a microwave discharge unit (Hattori et al. 2016, 2019) to a modified PT-IRMS, the N₂O can be decomposed to N₂ and O₂ prior to isotopic analysis, preventing the interference from ¹⁷O with δ¹⁵ N results. More recently, laser spectroscopy is emerging as an alternative technology to IRMS for simultaneous determination of N₂O concentration and isotopic abundance at natural abundance, possessing benefits such as real-time analysis, lower operating costs, and potential field applicability. In addition, laser spectroscopy measurements are not affected by differences in ¹⁷O abundance, thus also eliminating any potential overestimation of ¹⁵ N in samples containing mass-independent ¹⁷O variations. However, the reliability and accuracy of laser-based analysis substantially depend on the interplay between matrix and trace gas effects, spectral interferences, concentration dependencies of isotopic signals, and challenges in calibrating the instruments, making current applications difficult and limited (Köster et al. 2013; Mohn et al. 2014; Soto et al. 2015; Wassenaar et al. 2018; Ji and Grundle 2019; Yu et al. 2020; Jung et al. 2020; Harris et al. 2020).

¹⁵ N dilution and tracing studies do not necessitate such high precision as can be achieved by IRMS and laser techniques. For such studies equipment may be used that is much less expensive and more readily available, such as ES, GC–MS, MIMS and HPLC. These instruments can serve as alternatives for laboratories where ¹⁵ N abundances are not routinely determined, requiring much smaller amounts of N but their precision may be 10 to 100 times less than IRMS analysis (See 3.1.9 Precision and accuracy) (Knowles and Blackburn 1993; Russow et al. 1995; Lu et al. 2019). Although not commonly used, it is possible to connect a QMS or a MIMS to a specialized automated apparatus (e.g., an elemental analyser, or a SPIN unit) to achieve on-line gas production (Russow et al. 1995; Russow and Goetz 1998; Stange et al. 2007; Eschenbach et al. 2017, 2018).

When the instrument needed for the method chosen is not readily available, a final sample that is appropriate for transport and storage is necessary. In such cases, freshwater and marine samples can be extracted into ion-exchange resins and distilled into zeolites, respectively (Velinsky et al. 1989; Lehmann et al. 2001; Li et al. 2015). Alternatively, samples can be prepared using modern N₂O methods, as the produced N₂O gas can be stored in gas-tight containers for years (Sigman et al. 2001; McIlvin and Altabet 2005; Lachouani et al. 2010).

Toxicity of reagents

Several of the methods described involve the use of toxic reagents, and this aspect may also affect the suitability of these methods (Fisher and Morrissey 1985; Dudek et al. 1986; McIlvin and Altabet 2005). The mercury precipitation method has been rarely used since its development, due to its extreme toxicity. Some modern N₂O methods involve the use of an azide buffer to convert NH₄⁺ or NO₃^– to N₂O (McIlvin and Altabet 2005; Zhang et al. 2007; Lachouani et al. 2010). For non-microbiology laboratories, these are alternatives to the denitrifier methods. Under the required acidic conditions, the azide reagent is highly toxic, volatile and explosive, necessitating high safety precautions in handling and disposal of the chemicals. Another example is the derivatization method, in which the caustic and toxic phenol is used (Selmer and Sörensson 1986; Dudek et al. 1986; Kator et al. 1992; Preston et al. 1996, 1998; Johnston et al. 1999; Köster and Jüttner 1999; Clark et al. 2006, 2007). Ecologists generally are not trained as chemists to use such dangerous chemicals, so the handling of these chemicals would require experienced technicians.

New developments, especially in chemical N₂O methods, have reduced the environmental and health risks brought by toxic reagents. For example, a modification to the azide method has substantially reduced the dose of the azide reagent used for preparing ¹⁵ N analysis of NO₃^– (Tu et al. 2016). A safer alternative to the azide buffer is hydroxylamine hydrochloride (NH₂OH∙HCl), which has been used in recent sample preparation methods (Liu et al. 2014; Zhang et al. 2015; Jin et al. 2020). Another chemical method is the use of Ti(III) chloride to convert NO₃^– directly to N₂O (Altabet et al. 2019). These advances give a comparable precision and accuracy as their predecessors while being more user-friendly, providing practical options for users without assistance from a chemist or microbiologist.

Preparation time

Sample preparation methods for ¹⁵ N analysis in NH₄⁺ and NO₃^– require hours to days for complete recovery, depending on sample N concentrations and the batch size. The N₂ methods with EA-IRMS analysis (e.g., diffusion, ion-exchange, precipitation and acetone extraction) requires processing large volumes of liquids for low N samples (e.g., < 5 µM), which considerably increases the preparation time, up to several days. To reduce sample preparation time, contemporary N₂O methods, which were developed for samples with low N concentrations and small volumes, are more appropriate.

Some sample preparation methods are fast for a small number of samples, but inefficient for batch processing of large numbers of samples. For example, the distillation method takes only minutes for one sample, but the procedures to reduce cross-contamination to successive samples and to dry samples for combustion can be very cumbersome (Bremner 1965a, b, 1996; Bremner and Edwards 1965; Hauck 1982; Mulvaney 1986, 1993, 1996; Chang et al. 2004). In contrast, the diffusion and ion-exchange method require days for complete recovery, but a batch of samples can be processed simultaneously (Tables 1 and 2).

Precision and accuracy

The sample preparation and analysis method chosen must offer a precision and accuracy capable of detecting the level of variation among samples. As previously stated, in order to acquire optimal results it is important that the methods and instruments are used under their required matrix, concentration and ¹⁵ N enrichment, with interferences from fractionation and ¹⁵ N dilution being minimized. In addition, the samples must be subject to quality control with the use of reliable RMs (reference materials), and detailed guidelines can be found in literature (Klesta et al. 1996; Dawson and Brooks 2001; Werner and Brand 2001; Brand and Coplen 2012; Carter and Fry 2013; Brand et al. 2014; Skrzypek 2019; Meier-Augenstein and Schimmelmann 2019).

The precision and accuracy of reported methods have been often assessed using reference materials of different x(¹⁵ N)/ δ¹⁵N ranges, and have been frequently expressed in different ways (e.g., relatively or absolutely), making comparisons difficult. Traditional distillation and diffusion methods and IRMS analysis were usually assessed at x(¹⁵ N) < 10% because most ¹⁵ N-enriched studies are conducted at this range. Methods with GC–MS, MIMS and HPLC were assessed at a higher ¹⁵ N enrichment. Natural abundance methods were often evaluated in original papers over a narrow range of δ¹⁵N values, but they may be appropriate for wider applications as well (Michener and Lajtha 2008). We present the general range of precision and accuracy of different methods given by case papers, but they have not been included as criteria in the DST. It is worth noting that the DST is intended to be used as a navigation tool and it does not include the plethora of details for each method. Users should consult the original papers to get thorough information and verify the methods under their own scenarios.

Precision and accuracy of instruments The analysis of natural abundance requires considerable precision, which can currently only be achieved by IRMS and laser spectroscopy. Precision of DI-IRMS is usually < 0.1‰ δ¹⁵N with a N amount of 40 – 400 µmol N for manual N₂ preparation and ¹⁵ N analysis. When coupled to an automatic analyzer (i.e. EA-IRMS) or a purge and trap system (i.e. PT-IRMS), precision of CF-IRMS is usually < 0.2‰ δ¹⁵N from natural abundance to tracer level (< 5%) with a N mass of 2–10 µmol N or 20 − 60 nmol N₂O, respectively (Bremner and Hauck 1982; Dawson and Brooks 2001; Sigman et al. 2001; McIlvin and Altabet 2005; Michener and Lajtha 2008). Precision may be lower when analyzing samples with sub-micromoles of N or higher ¹⁵ N enrichment (Mulvaney 1993).The recently emerged laser spectroscopy yields a precision of < 0.5‰ with down to 0.5 nmol N₂O at near natural abundances (Mohn et al. 2014; Soto et al. 2015; Wassenaar et al. 2018; Ji and Grundle 2019).

Precision for instruments other than IRMS and laser spectrometers is typically between 1 − 5% RSD in a wide range of ¹⁵ N-enriched levels, which precludes their use for ¹⁵ N-depleted or natural abundance samples. Optimal emission spectrometers and GC–MS yield a similar precision and accuracy of < 3% (RSD and percent error) with down to 1 − 10 µg N at tracer levels (e.g. x(¹⁵ N) > 0.5%) (Fiedler and Proksch 1975; Bremner and Hauck 1982; Craswell and Eskew 1991; Hoult and Preston 1992; Preston 1993; Heiling et al. 2006). For a similar enrichment range, GC–MS yields a precision of 1 − 3% RSD for both N₂ (> 30 nmol) and N₂O (> 200 pmol) (Russow and Förstel 1993; Isobe et al. 2009). Precision for HPLC and MIMS is typically < 5% RSD, but the latter can measure a wider range of x(¹⁵ N) (25 − 75% vs. 0.5–100%) (Gardner et al. 1991, 1995; Yin et al. 2014).

Precision and accuracy of sample preparation methods Tracer studies are usually conducted at x(¹⁵ N) > 0.5%, and sample preparation methods developed for these studies generally showed standard deviations ranging from 0.02 to 5% (x(¹⁵ N)) and accuracies within ± 5% of the true values, depending on the ¹⁵ N abundances, N amounts, and procedures and equipment used. Previous evaluations have found that with optimal concentrations (~ 140–360 µmol and ~ 4–11 µmol N, respectively), distillations and diffusions yield comparable precision (SD < 0.05%) and accuracy (percent error < 5.3%) for tracer samples at x(¹⁵ N) < 5% (Liu and Mulvaney 1992b; Lory and Russelle 1994; Mulvaney et al. 1994, 1997; Høgh-Jensen and Schjoerring 1994; Khan et al. 1997, 1998; Sigman et al. 1997; Holmes et al. 1998; Sebilo et al. 2004; Diaconu et al. 2005; Stephan and Kavanagh 2009). However, caution should be given to sequential distillation and diffusion, because cross-contamination by highly ¹⁵ N-enriched NH₄⁺ can cause enormous errors to natural ¹⁵ N analysis of NO₃^– (Liu and Mulvaney 1992b; Mulvaney et al. 1994; Herman et al. 1995; Griesheim and Mulvaney 2019). At lower N masses and the whole tracer ¹⁵ N range, the denitrifier methods with GC–MS analysis achieved a precision and accuracy of 0.02–0.08% and 0.02% x(¹⁵ N), respectively (Christensen and Tiedje 1988; Højberg et al. 1994), better than those of SPIN methods (SD and percent error of 1–3% and 0.6%) (Russow 1999; Stange et al. 2007; Eschenbach et al. 2017) and derivatization methods (SD and percent error of 0.2–0.7% and 0.7%) (Selmer and Sörensson 1986; Dudek et al. 1986; Kator et al. 1992; Kanda 1995; Clark et al. 2006, 2007).

The analysis of natural ¹⁵ N variances necessitates considerably higher precision and methods usually offer comparable performance at their working range (Owens 1988; Shearer and Kohl 1993; Fry 2006; Mulvaney 2008). Several modifications have expanded the use of distillation and diffusion methods to natural δ¹⁵N range. For example, the modification to distillation by Velinsky et al. (1989) have enabled analysis of δ¹⁵N of NH₄⁺ with an overall precision of < 0.5‰ (SD) and a relative error of within ± 4% of the true δ¹⁵N at low (< 5 µM) concentrations. At a similar working range, modifications to diffusion by Sigman et al. (1997) and Holmes et al. (1998) have allowed δ¹⁵N analysis of NO₃^– and NH₄⁺ in water samples to be within 0.3‰ and 0.36–0.73‰ δ¹⁵N (after fractionation correction) of the true values, respectively, and with an overall precision of < 0.3‰ for both N species. Stephan and Kavanagh (2009) found that the diffusion method can be applied to KCl extracts at natural ¹⁵ N abundance if a precision or accuracy of 1.3‰ is not required. Other methods for natural δ¹⁵N analysis, such as ion-change (Chang et al. 1999; Silva et al. 2000; Lehmann et al. 2001; Fukada et al. 2003; Xing and Liu 2011; Gebus-Czupyt et al. 2020), acetone extraction (Huber et al. 2011, 2012), denitrifier (Sigman et al. 2001; David Felix et al. 2013), as well as azide (McIlvin and Altabet 2005; Zhang et al. 2007; Tu et al. 2016), NH₂OH (Liu et al. 2014; Zhang et al. 2015; Jin et al. 2020) and Ti(III) reduction (Altabet et al. 2019) methods, achieved comparable precision ranging from 0.2 to 1‰ δ¹⁵N. The δ¹⁵N values from these methods were generally within 5‰ of the true values, which were verified by a multi-point calibration using international or in-house isotopic standards following the identical treatment principle (Werner and Brand 2001; Meier-Augenstein and Schimmelmann 2019). The accurate δ¹⁵N values of the samples were calibrated using the linear regression established by the measured δ¹⁵N values of the standards against their assigned δ¹⁵N values (Klesta et al. 1996; Werner and Brand 2001; Brand and Coplen 2012; Brand et al. 2014; Meier-Augenstein and Schimmelmann 2019).

Integrating ¹⁵ N analysis methods and criteria into the DST

The criteria discussed are not all given the same importance in the DST. The ¹⁵ N abundance, the sample N concentration, the instrument availability, and the sample matrix are given priority, because they always limit the preparation approaches available. The ¹⁵ N abundance and N concentration of samples determine the acceptable levels of ¹⁵ N dilution and isotopic fractionation. Sometimes the researchers must make trade-offs between the different criteria. The toxicity of reagents and the sample preparation time are of course more malleable, as relaxing the requirements based on these criteria does not compromise the result of the measurement, but simply requires more investments. All criteria are organized in the DST in a way to streamline the decision-making steps and minimize repetitions while matching the most common situations, so that the users and laboratories of diverse interests and disciplines can quickly find the suitable approaches by following the criteria.

In the DST, the “¹⁵ N abundance” is set as the first criterion, dividing the decision trees into two parts: methods for ¹⁵ N-enriched samples (the left branch) and methods for natural ¹⁵ N abundance samples (the right branch) (Figs. 1 and 2). The “Sample N” is organized as the third criterion after the choice of “Instruments”, followed by “Sample matrix” (including salinity and DON content), “Toxicity”, “Time” and “Fractionation”. In the following sections we guide the readers through the tree and explain how we split sample preparation methods into small groups based on these criteria. Because their structures are similar, we discuss the DSTs for NH₄⁺ and NO₃^– at the same time.

Non-IRMS/laser methods for ¹⁵ N-enriched samples in the DST

The ¹⁵ N-enriched samples can be prepared either by methods suitable for precise IRMS/Laser analysis, or by methods developed for other instruments such as ES, GC–MS, QMS/MIMS, or HPLC. If researchers decide to use the latter, they will be guided to the left branch of the decision tree, where they have to make a choice at the node “Sample N”, based on the N amounts and volumes of their samples (Figs. 1 and 2). For ¹⁵ N-enriched samples of sufficiently large volumes and containing sufficient N amounts (e.g. > 30 ml and > 4 µmol), the NH₄⁺ or NO₃^– can be isolated by diffusion or distillation from various kinds of sample matrices, followed by off-line Dumas combustion and optimal emission analysis (Fiedler and Proksch 1975; Hauck 1982; Mulvaney et al. 1990; Preston 1993; Mulvaney 1996). These methods are easy to set up, but they can be tedious and error-prone for large batches of low-N samples.

In case samples have low N amounts and small volumes (e.g., < 1 µmol and < 20 ml), one should pick the other branch of the node “Sample N”, and then make decisions based on the kind of sample matrix, the availability of materials, and concerns about toxicity and time. In terms of sample matrix, both soil extracts and water samples can be prepared using the denitrifier method and then measured using a GC–MS. If NH₄⁺ is the target N compound (Fig. 1), a hybrid denitrifier method can be used, which involves isolating NH₄⁺ by diffusion, converting NH₄⁺ to NO₂⁻ by hypobromite oxidation, and converting NO₂⁻ to N₂O (Diffusion-BO-denitrifier). Alternatively, liquid samples of diverse matrices can be prepared using the SPIN-QMS/MIMS method. In that case the liquid samples are entered by means of an autosampler into a special sample preparation unit, in which different reagents are automatically dosed, to convert NH₄⁺ and NO₃⁻ to a gaseous N-product (e.g., N₂, N₂O, or NO). The N-gases are then fed into the coupled mass spectrometer. It is the only fully on-line method to analyze the isotopic composition of NH₄⁺ and NO₃⁻ (and NO₂⁻) in liquid samples. However, this specialized SPIN apparatus is probably not available in most laboratories.

Other reported non-IRMS/Laser methods are limited to freshwater and seawater samples, and vary in the level of toxicity and sample preparation time. The Hg precipitation is salinity-sensitive and extremely toxic, and is rarely used today (Fig. 1). The derivatization methods also use hazardous chemicals. However, adaptations to derivatization using GC–MS make it possible to analyze samples containing only nanomoles of target N, which can be of interest in case of pool dilution experiments in oligotrophic ocean environments. Methods without the use of highly toxic reagents include OX (oxidation)-MIMS (only for NH₄⁺) (Fig. 1) and AIRTS-HPLC (for both NH₄⁺ and NO₃⁻) (Figs. 1 and 2), both of which allow for the analysis of small volume samples with low N concentrations (Tables 1 and 2). In the OX-MIMS method, NH₄⁺ in solution is directly converted to N₂ using hypobromite iodine oxidation. This procedure is prone to N contamination and has low precision (< 5%), making it ideal for studying small and highly dynamic natural water NH₄⁺ pools, such as those in sediment oxygen-transition zones (Yin et al. 2014). In the AIRTS-HPLC method, the conversion into N₂ gas is not needed, and the isotopic ratio of NH₄⁺ is measured based on a retention-time shift of the combined peak of the ¹⁵NH₄⁺ and ¹⁴NH₄⁺. The shift is relative to the percentage of ¹⁵NH₄⁺ and the best relationship occurs between 25–75% x(¹⁵ N). As a result, the AIRTS-HPLC method is suitable for pool dilution experiments with variations in ¹⁵ N abundances within this range (Gardner et al. 1991, 1995, 1997; Lu et al. 2019). Because the individual NH₄⁺ peak is isolated from organic-N peaks on the HPLC column, it is advantageous for applications such as analyzing the degradation of peptide and amino acid ¹⁵ N to ¹⁵NH₄⁺ (Gardner et al. 1995; Yin et al. 2014).

IRMS/Laser Methods in the DST

Both ¹⁵ N-enriched and natural ¹⁵ N abundance samples can be processed using sample preparation methods that produce N₂ or N₂O for IRMS/laser analysis, but only the latter necessitates such highly precise methods and instruments. IRMS and laser methods can be found on the right branch of both decision trees and are split up at the node “Instrument”, and the node “Sample N” (Figs. 1 and 2). There are two major categories of methods in this part: samples are prepared in one category as solids for on-line combustion into N₂ and ¹⁵ N analysis by EA-IRMS (i.e., N₂ methods), and in the other as N₂O gas for ¹⁵ N analysis by PT-IRMS or laser spectroscopy (i.e., N₂O methods). They vary in their requirements of sample N amounts and volumes. The first category of methods include distillation, diffusion, derivatization, ion-exchange, TPB precipitation (only for NH₄⁺) and acetone extraction (only for NO₃⁻), all of which are appropriate for samples containing large N amounts (e.g., > 1 µmol). For samples with a low N concentration, the concentration procedure is tedious and prone to errors from fractionation and ¹⁵ N dilution by reagents. In the second group of methods N₂O is produced as the analyte by means of microbes or chemicals. Because the ¹⁵ N analysis of N₂O has a high sensitivity, these methods allow samples containing smaller N amounts (20–60 nmol) and volumes (< 20 ml). Because the ¹⁵ N analysis of N₂O has a high sensitivity, these methods allow samples containing smaller N amounts (20–60 nmol) and volumes (< 20 ml). Samples with larger N amount can also be prepared using N₂O methods after dilution, yielding more compatible and precise results for natural abundance studies in environmental samples.

N₂ methods for samples with large N amounts (micromoles of N) After choosing a N₂ or N₂O method based on sample N amount and volume, researchers must check the salinity level and DON content of the matrix. Among N₂ methods, the distillation, diffusion, and NaOH-acetone extraction are salinity-insensitive and hence appropriate for both soil extracts and water samples. In particular, the NaOH-acetone method extracts NO₃^– into an acetone solvent, isolating it from other salts and organic matter, which is advantageous for samples containing large quantities of DON (e.g., several times that of NO₃^–). However, the freeze-drying process to obtain large N amounts (4–9 µmol) increases preparation time. The diffusion also requires days of preparation, but many samples can be processed simultaneously. In contrast, the distillation method is efficient when preparing small batches of samples but can be tedious and contamination-prone for large batches. It is used primarily for ¹⁵ N-enriched studies in contemporary ecological research. Another salinity-insensitive method is derivatization, but the use of toxic chemicals and the need for complex procedures render it impractical.

Methods that are salinity-sensitive are only appropriate for freshwater samples. Both NH₄⁺ and NO₃^– can be prepared using the ion-exchange methods, which are simple to set up, and allow for field processing and long-term storage with minimal fractionation. However, the costs of resin and chemicals (e.g., silver oxide for NO₃^– elution) can be an issue for large numbers of samples. The TPB precipitation and the Ba(NO₃)₂-acetone techniques, on the other hand, use less expensive materials and are appropriate for preparing NH₄⁺ and NO₃^–, respectively. They both use freeze-drying to concentrate solutions, which means that fractionation must be minimized in that process or accounted for.

N₂O methods for samples with small N amounts (nanomoles of N) For samples with low N content and small volumes, the N₂O methods with IRMS/laser analysis are more appropriate as they eliminate time-consuming concentration processes and reduce potential fractionation and contamination. In addition, they produce N₂O gas in gas-tight vials, which can be easily transported and preserved. These methods are located on the right side of the decision trees (Figs. 1 and 2). In these methods, NH₄⁺ is usually first isolated from solution using diffusion or ion-exchange resins, and then converted to NO₃^– or NO₂^– by persulfate oxidation (PO) or hypobromite oxidation (BO), respectively. The NO₃^– or NO₂^– is then converted to N₂O by denitrifying bacteria, azide or NH₂OH techniques (Fig. 1). If NO₃^– is the target N, it can be converted to NO₂^– and then to N₂O by denitrifying bacteria, Cd-azide, VCl₃-azide, Cd-NH₂OH, or Ti(III) reduction methods (Fig. 2).

The sample matrix and the N species being studied determine the time and procedures required for N₂O methods. NH₄⁺ in freshwater and seawater samples can usually be prepared within 2–3 days using hypobromite oxidation coupled to denitrifier, azide or hydroxylamine reduction (BO-denitrifier, BO-azide and BO-NH₂OH), although the pre-incubation of denitrifying bacteria may take 10–12 days. On the other hand, NH₄⁺ in soil extracts is usually first isolated by diffusion, due to interferences from soil extractants (e.g., KCl), which unavoidably increases the preparation time by 3–5 days (Fig. 1). If NO₃^– is of interest, both soil extracts and water samples can be prepared within a few days using one of the above-mentioned conversions.

The N₂O methods are usually similar in performance, but the chemicals involved differ in toxicity. Many N₂O techniques involve the use of the highly toxic azide reagent, which necessitates safety precautions and chemical guidance (Figs. 1 and 2). As alternatives, less-toxic NH₂OH methods or non-toxic bacterial methods can be used. The denitrifier method is the safest, but the difficulties in getting denitrifying bacteria strains, as well as the need for long-term incubation and particular care of microbial cultures, make it less practical than hydroxylamine methods for most laboratories. Another toxin-free option for NO₃^– preparation is the Ti(III) method, but it needs a strict calibration strategy to account for isotopic fractionation, owing to the fact that the recovery is relatively low and declines significantly with increasing salt concentration (Fig. 2). The other N₂O methods usually have very small isotopic fractionation and ¹⁵ N dilution by reagents at their working range (Tables 1 and 2).

Discussion

Application cases

In this section, we present two examples to give a better understanding of the use of the DST.

Case 1: Measuring natural ¹⁵  N abundance of NH ₄ ⁺ and NO ₃ ^– in agricultural soils

The variations in natural ¹⁵ N abundance (δ¹⁵N) of N in soils have been used to identify N contamination sources and transformation processes (Robinson 2001). Agricultural soil extracts typically contain high concentrations of NH₄⁺ and NO₃^– (e.g., > 50 µM) and DON due to high nutrient input from fertilizers. We assume that IRMS is available in this case.

To find the ¹⁵ N analysis method for NH₄⁺ and NO₃^–, we first select “Natural abundance” for “¹⁵ N abundance” in the DST (Figs. 1 and 2). If a PT-IRMS or a laser spectrometer is available, the N₂O methods are recommended, which are more specific and less prone to contamination and fractionation. Therefore, we choose “Small” for “Sample N” and “Soil extracts” for “Sample matrix” for both NH₄⁺ (Fig. 1) and NO₃^– (Fig. 2). In the case of NH₄⁺ (Fig. 1), alkaline soil extracts can be analyzed using the BO-NH₂OH method within 2–3 days; otherwise, the NH₄⁺ must first be isolated using diffusion, resulting in a longer preparation time. Following persulfate oxidation or hypobromite oxidation, the denitrifier techniques (Diffusion-PO/BO denitrifier) enable an efficient and eco-friendly sample preparation. On the other hand, azide or NH₂OH techniques (Diffusion-PO-azide, Diffusion-BO-azide, and Diffusion-BO-NH₂OH) can be used in non-microbiology laboratories, and the latter employs less hazardous reagents. In the case of NO₃^– (Fig. 2), similarly, the final choice depends on toxicity concern and strain availability.

If only a EA-IRMS is available, alternatively, we choose “Large” for “Sample N” and “Soil extracts” for “Sample matrix” in the DST for ¹⁵ N analysis of NH₄⁺, which brings us to the node “Time” (Fig. 1). The distillation method provides field applicability with a commercially available apparatus, but it may lead to large inaccuracies in natural δ¹⁵N results due to the high risk of fractionation and cross-contamination. In contrast, the diffusion method eliminates cross-contamination, and is more suitable for batch-processing without requiring specialized apparatus. In the DST for NO₃^– (Fig. 2), we choose, similarly, “Natural abundance”, “Large sample N”, and “Soil extracts”. In the current case, the soil extracts are enriched with a high level of DON that is likely to break down and cause large contamination during sample preparation. Therefore, we prefer to use the NaOH-acetone method due to its DON-removal capability.

Case 2: Estimating gross N transformations in a pool dilution experiment in seawater

Gross NH₄⁺ and NO₃^– transformation rates in aquatic and terrestrial ecosystems can be estimated using the ¹⁵ N pool dilution technique, whereby small amounts of ¹⁵NH₄⁺ or ¹⁵NO₃^– are added, and the rates of individual N transformation processes can be estimated from changes in size and ¹⁵ N abundance of the mineral N pools during incubation (Kirkham and Bartholomew 1954; Tietema and Wessel 1992; Laine et al. 2018). The pool dilution technique has been widely used to measure N dynamics such as NH₄⁺ regeneration and nitrification in benthic and pelagic freshwater and marine ecosystems (Gardner et al. 1991; Ward 2011). Here we consider a case of a pool-dilution experiment in seawater. In such a case, samples are generally characterized by low NH₄⁺ and NO₃^– concentrations (e.g., around 10 µM NH₄⁺ or NO₃^–) and a high salinity. Furthermore, we assume that samples are small in volume (e.g., 20 ml), and that either the IRMS or the LS, and also denitrifier cultures are not readily available. The lack of such instruments is often the case in laboratories where the ¹⁵ N abundance is not routinely measured.

In the DST for ¹⁵NH₄⁺ analysis, we choose “¹⁵ N enriched” for “¹⁵ N abundance”, “Others” for “Instrument”, and “Small” for “Sample N” (Fig. 1). We can then choose both “Water samples only” and “Soil extracts and water samples” for “Sample matrix”. Among methods appropriate for both soil extracts and water samples, the SPIN method is rapid but it requires a higher NH₄⁺ concentration (> 70 µM) and special reaction apparatus. The diffusion-BO-denitrifier method, on the other hand, allows for the preparation of samples with lower NH₄⁺ concentrations (> 1 µM), but it needs a long preparation time and the use of special denitrifying bacteria. Alternatively, we can choose sample preparation methods that are “Water samples only” and with low “Salinity sensitivity”. The derivatization method uses hazardous reagents and necessitates a sample volume of > 50 mL (Table 1), which is inappropriate in this case. As a result, we can choose between “AIRTS-HPLC” and “OX-MIMS”, both of which work for small sample N and volumes. The AIRTS-HPLC method is capable of determining the concentration and the ¹⁵ N abundance of NH₄⁺ simultaneously, but it requires more time than the OX-MIMS method (1 h vs. 24 h for a set of ~ 10 samples) (Table 1).

With regard to the ¹⁵NO₃^– analysis (Fig. 2), we similarly choose “¹⁵ N enriched” for “¹⁵ N abundance”, “Small” for “Sample N” and “Water samples” for “Sample matrix”. Due to the toxicity and large volume requirement of the derivatization method, we prefer other methods such as the “AIRTS-HPLC”, “SPIN-QMS/MIMS” and “denitrifier” methods. If it is possible to obtain the SPIN apparatus or the cultures of denitrifying bacteria, the latter two methods are fast and simple choices. On the other hand, the AIRTS-HPLC method has a low throughput while using more readily available reagents and equipment. As NO₃^– is reduced to NH₄⁺ as the analyte, the ¹⁵ N abundance of ¹⁵NO₃^– should be calculated based on the NH₄⁺ results. Therefore, if this method is used, the NH₄⁺ should be prepared and measured prior to the NO₃^–.

Pros and cons of the present DST

In this work, we focus on the ¹⁵ N analysis of NH₄⁺ and NO₃^– in liquid samples from various environmental matrices, as they are the two key mineral N species in ecosystems and primarily used in environmental ¹⁵ N studies. While some simple and cheap methods such as the OX-MIMS and Ti(III) reduction method target only a single N species, most other methods are in principle applicable to both N species following an initial N conversion procedure. Other N-bearing compounds, such as NO₂^– and organic N compounds, are discussed as interfering chemicals in this work. However, the ¹⁵ N analysis of such species in environmental samples can also be of great interest to ecological researchers. In neutral or alkaline soils where NH₄⁺ or NH₄⁺ forming fertilizers are applied, the amount of NO₂^– can be significant (Burns et al., 1995; Russow et al., 2009; Ward, 2011). Many NH₄⁺ and NO₃^– methods convert target N to NO₂^– and further to N₂O, and they can be modified to convert only NO₂^– by controlling the reaction procedures (Sigman et al. 2001; McIlvin and Altabet 2005; Zhang et al. 2007; Ward 2011; David Felix et al. 2013; Liu et al. 2014). On the other hand, to prepare organic N for ¹⁵ N analysis, it is usually first separated from inorganic N via diverse techniques. The isolated organic N is then converted to NH₄⁺ or NO₃^–, or is combusted directly (Bronk and Glibert 1991; Feuerstein et al. 1997; Chang et al. 2004; Knapp et al. 2005; Toshihiro et al. 2005; Mulvaney 2008; Tsunogai et al. 2008; Johnson et al. 2013; Lu et al. 2019; Cao et al. 2021). The ¹⁵ N abundances of particular types of DON, such as free amino acids, can now be measured using GC–MS, CE-MS (capillary electrophoresis-mass spectrometry), GC-C-IRMS (gas chromatography-combustion-isotope ratio mass spectrometry), and high-resolution-MS (Mawhinney et al. 1986; Hofmann et al. 2003; Meier-Augenstein 2004; Wanek et al. 2010; Warren 2012, 2018, 2019; Andresen et al. 2015). Such techniques in combination with ¹⁵ N pool dilution have been developed recently to quantify gross rates of amino acid transformations, facilitating a better understanding of protein depolymerization and soil N dynamics (Wanek et al. 2010; Andresen et al. 2015). The integration of these important N compounds into the present DST can make it applicable to a wider range of ¹⁵ N-related studies.

The present DST highlights the knowledge gaps in the application of certain published methods to other matrices or samples. Most non-IRMS/LS methods, such as derivatization, HPLC, and OX/MIMS methods, have been developed for freshwater and marine samples, but may be applied also to soil samples if interfering chemicals are removed by additional procedures (Gardner et al. 1991, 1995; Stephan and Kavanagh 2009; Yin et al. 2014). Such adaptations and developments offer more options to different laboratories and are attractive to those subject to equipment and budget limitations. Likewise, the NH₂OH method has just been reported and tested in a few studies, but with additional purification procedures it may become suitable for samples with a more complex matrix (Liu et al. 2014; Zhu et al. 2015; Jin et al. 2020). Because NH₂OH is more environmentally friendly than the azide reagent, such techniques may rise in favor over the azide-based methods in the future due to rising environmental and health concerns, as well as risk-control regulations in many laboratories and nations.

On the other hand, some promising ¹⁵ N analysis methods, have received less widespread implementation due to the lack of access to specialized equipment, or challenges with compatible and accurate isotopic analysis (Mohn et al. 2014; Harris et al. 2020). For example, the SPIN unit and the PT-IRMS equipped with a thermal decomposition system are only available in specialized laboratories (Stange et al. 2007; Smirnoff et al. 2012; Mohn et al. 2014; Hattori et al. 2016; Eschenbach et al. 2017). The laser-based methods have just been developed for ¹⁵ N analysis of NO₃^– following azide and bacterial methods (Soto et al. 2015; Wassenaar et al. 2018) and the accuracy is limited by instrumental precision, drift, matrix effects and spectral interferences (Mohn et al. 2014; Harris et al. 2020). Recent developments in calibration protocols and reference materials (Harris et al. 2020; Mohn et al. 2022) will encourage the routine use of laser- and N₂O-based techniques to a wider range of environmental ¹⁵ N studies. Another promising novel approach is to measure diverse isotopologues of environmental NO₃^– by LC-ESI-Orbitrap-MS (liquid chromatography mass spectrometry with electrospray ionization Orbitrap) after sample preparation by ion-exchange and gradual elution, yielding high precision and accuracy at natural abundance level (Hilkert et al. 2021). This technique allows for the calculation of mass-independent O isotope variations, as well as the exploration of non-random isotopic distributions. Advances in sample preparation, analysis, and data interpretation procedures may prompt the application of the ESI-Orbitrap technique, allowing researchers to gain insights into multidimensional isotopic fingerprints in diverse organic and inorganic solutes.

Finally, it is also important to note that the DST is intended to be used as a navigation tool, but it does not offer comprehensive details about all the methods. Users are always advised to check the original papers and evaluate the performance of the method in their own laboratories before routine use is undertaken.

Conclusion

In this paper, we developed a decision support tool based on the criteria that are primarily considered when selecting methods for ¹⁵ N analysis of NH₄⁺ or NO₃^– in liquid samples. This tool is straightforward, visually appealing, and user-friendly, with the potential to be applied to a broader spectrum of ¹⁵ N-related environmental research. Integration of other N species and sample matrices into the current tool, along with advancements in current preparation approaches, can further improve the applicability of the DST in the future.